Thiazolidinediones inhibit hepatocarcinogenesis in hepatitis B virus–transgenic mice by peroxisome proliferator-activated receptor γ–independent regulation of nucleophosmin


  • Potential conflict of interest: Nothing to report.


Antidiabetic thiazolidinediones (TZD) have in vitro antiproliferative effect in epithelial cancers, including hepatocellular carcinoma (HCC). The effective anticancer properties and the underlying molecular mechanisms of these drugs in vivo remain unclear. In addition, the primary biological target of TZD, the ligand-dependent transcription factor peroxisome proliferator-activated receptor γ (PPARγ), is up-regulated in HCC and seems to provide tumor-promoting responses. The aim of our study was to evaluate whether chronic administration of TZD may affect hepatic carcinogenesis in vivo in relation to PPARγ expression and activity. The effect of TZD oral administration for 26 weeks was tested on tumor formation in PPARγ-expressing and PPARγ-deficient mouse models of hepatic carcinogenesis. Proteomic analysis was performed in freshly isolated hepatocytes by differential in gel electrophoresis and mass spectrometry analysis. Identified TZD targets were confirmed in cultured PPARγ-deficient hepatocytes. TZD administration in hepatitis B virus (HBV)–transgenic mice (TgN[Alb1HBV]44Bri) reduced tumor incidence in the liver, inhibiting hepatocyte proliferation and increasing apoptosis. PPARγ deletion in hepatocytes of HBV-transgenic mice (Tg[HBV]CreKOγ) did not modify hepatic carcinogenesis but increased the TZD antitumorigenic effect. Proteomic analysis identified nucleophosmin (NPM) as a TZD target in PPARγ-deficient hepatocytes. TZD inhibited NPM expression at protein and messenger RNA levels and decreased NPM promoter activity. TZD inhibition of NPM was associated with the induction of p53 phosphorylation and p21 expression. Conclusion: These findings suggest that chronic administration of TZD has anticancer activity in the liver via inhibition of NPM expression and indicate that these drugs might be useful for HCC chemoprevention and treatment. HEPATOLOGY 2010

Hepatocellular carcinoma (HCC) is the most frequent solid tumor of the liver. Half a million cases occur annually, making it the fifth most common malignancy in men and the ninth most common in women.1

Hepatocarcinogenesis is a multistep process involving genetic and epigenetic events that accumulate during chronic liver diseases. The extent of hepatic dysfunction limits therapeutic options for HCC and survival of patients with this tumor remains dismal, as the average survival from time of diagnosis of unresectable HCC is measured in months.2 In this scenario HCC is therefore an attractive target for identification of potential chemopreventive drugs.

Thiazolidinediones (TZD) are a class of antidiabetic drugs which attenuate insulin resistance and impaired glucose tolerance in humans as well as in several animal models of non–insulin-dependent diabetes mellitus.3 The mechanisms of TZD action are still being investigated but it has been clearly demonstrated that some of their effects are mediated through activation of the peroxisome proliferator-activated receptor-γ (PPARγ), a member of the nuclear receptor superfamily of ligand-dependent transcription factors predominantly expressed in adipocytes but also in other normal and transformed cells.4

Beyond the metabolic actions, several studies indicate that TZD may have also anticancer properties in a variety of different epithelial malignancies. Indeed TZD treatment of cancer cells cultured in vitro or implanted in nude mice causes reduction of growth rate, cell differentiation and apoptosis.5

Despite the suggestions that TZD might favor cancer remission, there are conflicting data on whether PPARγ activation promote or suppress tumorigenesis when applied in animal model of cancer.6 Studies in colon and breast carcinogenesis have shown that TZD-dependent activation of PPARγ leads to an increase of tumor formation.7, 8 In addition, PPARγ is overexpressed in many epithelial tumor cells and regulates the production of hepatocyte growth factor which can favor tumor growth, suggesting that this nuclear receptor might represent a prosurvival factor.9

It has been previously shown that in human HCC, cancer cells express PPARγ and treatment with troglitazone, the first TZD initially approved for clinical use, induces a dose dependent reduction of cell proliferation, and a significant increase of apoptosis by a mechanism involving the induction of the cell cycle inhibitor p27.10 Conversely, recent results indicate that specific PPARγ inhibitors prevent adhesion to extracellular matrix and induce anoikis, causing a more effective cell death than TZD.11

Given these apparently discrepant observations on whether PPARγ activation could be growth-inhibitory or tumor-promoting in hepatic cancer cells, this study was designed to analyze the potential in vivo anticancer effect of chronic oral administration of TZD in a HBV-related model of hepatocarcinogenesis and to define the correlation between TZD actions and PPARγ expression and transcriptional activity in hepatocytes. Here we show that TZD treatment inhibits tumor formation in HBV transgenic mice with a significant reduction of hepatocyte proliferation and increased apoptosis independently of PPARγ expression. Proteomic analysis identifies nucleophosmin (NPM), a nucleolar protein initially characterized in the process of ribosomal RNA assembly and transport, as master coordinator of TZD antineoplastic action in hepatocytes.


DIGE, difference gel electrophoresis; EMSA, electrophoretic mobility shift assays; HCC, hepatocellular carcinoma; NPM, nucleophosmin; PCNA, proliferating cell nuclear antigen; PGZ, pioglitazone; PPARγ, peroxisome proliferator-activated receptor γ; PPRE, peroxisome proliferator response element; RGZ, rosiglitazone; TZD, thiazolidinedione.

Materials and Methods

Transgenic Animals and Experimental Design.

To achieve a selective elimination of PPARγ in the liver of TgN(Alb1HBV)44Bri mice,12 we realized a triple transgenic animal where the liver-specific Cre expression, obtained by placing Cre DNA under the control of albumin promoter, deletes PPARγ in hepatocytes. Parental transgenic mice were obtained from The Jackson Laboratories (Bar Harbor, ME). Breeding details and histopathological diagnoses are specified in the Supporting Information. Nine-month-old male transgenic mice were treated for 26 weeks with daily gavage administration of TZD (3.0 mg/kg/day) (rosiglitazone [RGZ] or pioglitazone [PGZ]) or with the non-TZD PPARγ ligand GW1929 (5.0 mg/kg/day). Control animals were treated with vehicle alone.


The proliferation of hepatic cells was estimated by immunostaining for PCNA and Cyclin D1 whereas apoptosis was detected by staining for activated caspase-3 and caspase-7. PCNA, Cyclin D1, and apoptotic labeling indexes (LI), were semiquantitatively evaluated by counting the percentage of immunoreactive hepatocytes in at least 10 randomly selected fields using the image processing and analysis software Image J.13 (W.S. Rasband, ImageJ, U.S. NIH, Bethesda, MD,, 1997-2009.)

Hepatocyte Isolation and Proliferation Assay.

Hepatocytes were isolated by a two-step collagenase perfusion of the liver through the inferior cava vein.14

Hepatocytes were plated at a density of 0.3 × 106 per 35-mm dish in DMEM/F12 medium. After 4 hours attachment, cells were starved in serum-free media. DNA synthesis in primary hepatocyte cultures was measured by [3H]thymidine incorporation. Additionally, apoptosis was assessed morphologically by Hoechst 33342 staining (Sigma Chemical, Germany) and fluorescence microscopy (Carl Zeiss, Germany).

Isolation of Nuclear Protein Extract and DNA Binding Assay.

Nuclear proteins were extracted from isolated hepatocytes based on a micropreparation method.15 Electrophoretic mobility shift assays (EMSA) were performed by radiolabeling double-stranded oligonucleotides corresponding to the PPAR Response Element (PPRE) ARE7 (5′-TGCACATTT CACCCAGAGAGAAGGGATTGA-3′).

Transfection of Cultured Hepatocytes.

For transfection, the Amaxa nucleofection technology (Lonza AG, Belgium) was employed. Hepatocytes were transfected following the manufacturer's instructions. Briefly, 100 μL of 2 × 106 cell suspension were mixed with 2.5 μg of (ARE7)3-tk-luciferase reporter plasmid or with 2.5 μg NPM promoter construct. pSV2CAT (2 μg) was used as internal control for transfection efficiency. The mixture was nucleofected with the T-028 nucleofector program. After transfection, hepatocytes were transferred into six-well plates and culture medium was replaced 4 hours later. After 48 hours, the cells were harvested for luciferase and CAT assays.15

Western Blot.

Protein extracts were fractionated by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Proteins were detected incubating primary antibodies overnight at 4°C, followed by the appropriate secondary antibody conjugated with horseradish peroxidase (1:1000) 2 hours at room temperature. Blots were developed with the enhanced chemiluminescence detection system ECL plus kit (Pharmacia Biosciences, Piscataway, NJ).

Proteomic Analysis.

Proteomic analysis was performed by difference gel electrophoresis (DIGE). Samples preparation for two-dimensional DIGE and mass spectrometry identification are described in the Supporting Information.

Statistical Analysis.

Results are expressed as mean ± standard deviation (SD) or standard error (SE). Multiple comparisons were performed by one-way analysis of variance with Bonferroni's correction. A P value less than 0.05 was considered statistically significant.


TZD Inhibits Tumor Formation in HBV Transgenic Mice.

We chose to examine the effect of TZD chronic administration on a mouse model of HBV-related hepatocarcinogenesis. Hepatocytes of transgenic mice TgN(Alb1HBV)44Bri express and accumulate the large HBsAg protein, resulting in severe chronic hepatocellular injury. This condition is constantly followed by the development of dysplastic hepatic lesions that progress after the ninth month of life to hepatocellular adenomas and carcinomas.12 TZD (RGZ or PGZ), or a non-TZD n-aryl tyrosine activator of PPARγ (GW1929) or vehicle alone (CTRL) were administered daily by oral gavage to HBV transgenic mice for 26 weeks starting from the ninth month of life. Four vehicle-treated, one RGZ-treated, three PGZ-treated, and four GW1929-treated animals died during the study and were not included in the effective numbers: the observed deaths were not caused by the treatments but are caused by natural and technical reasons (i.e., the protracted TZD administration by gavage as demonstrated by necroscopy examination). In the control group, 96% of mice developed hepatocellular adenomas and in 42% of them, we found hepatocellular carcinomas after sacrifice (Table 1, Fig. 2A).

Table 1. TZD Oral Administration Inhibits Hepatocarcinogenesis in HBV Transgenic Mice
 Incidence %MultiplicityMaximum tumor sizeLiver Weight
TreatmentsAdenomasCarcinomasMean ± SEMean ± SE (mm)Mean ± SE (g)
  • The values reported were calculated as described in the Supporting Information Materials and Methods. The number of mice with adenomas/carcinomas per total number of mice is given in parentheses.

  • *

    P < 0.05 versus vehicle-treated animals.

Vehicle96.15 (25/26)42.30 (11/26)6.23 ± 0.708.7 ± 0.873.73 ± 0.56
RGZ58.62 (17/29)27.58 (8/29)2.58 ± 0.53*1.4 ± 0.12*1.85 ± 0.29*
PGZ59.25 (16/27)25.92 (7/27)3.11 ± 0.67*1.8 ± 0.17*1.92 ± 0.26*
GW192992.30 (24/26)53.84 (14/26)5.53 ± 0.629.3 ± 0.793.86 ± 0.45

TZD oral administration markedly suppressed the tumorigenic process in treated mice. Of the 56 TZD-treated mice, only three mice had evident hepatocellular nodules larger than 2 mm, and 12 mice were completely devoid of macroscopically visible formations (Fig. 1A). The smaller number and size of neoplastic foci in TZD-treated mice correlated with the smaller liver mass reflecting an apparent difference in the growth rate of preneoplastic and neoplastic lesions as compared with controls. On the contrary treatment with GW1929 exerted no effect on tumor formation in HBV transgenic mice.

Figure 1.

TZD suppresses hepatic tumorigenesis in HBV transgenic mice. (A) Spontaneous liver tumor development is inhibited by 6 months oral administration of RGZ and PGZ, but not of GW1929. Macroscopic and microscopic (original magnification: ×100) of representative livers are shown. (B) Representative immunostaining for activated caspase-3 and caspase-7, PCNA, and cyclin D1 of HBV transgenic mouse liver at the end of treatments. Scale bar = 50 μm.

TZD Inhibits Hepatocyte Proliferation and Induces Apoptosis in HBV Transgenic Mice.

The inhibition of tumor formation by TZD correlated with the reduction of the proliferative activity and increased frequency of apoptosis of liver cells measured by PCNA (Figs. 1B, 2D), cyclin D1 (Figs. 1B, 2C) and activated caspase-3 and caspase-7 immunostaining (Figs. 1B, 2B). At the microscopic level, TZD-treated livers exhibited only mild dysplasia with considerably less cellular and nuclear enlargement and without advanced nuclear atypia when compared with control littermates (Fig. 1A). Although nodular regeneration was significantly reduced in TZD-treated animals, no differences in degenerative alterations, chronic inflammation and liver cell necrosis were documented (Supporting Information Table 1). Serum concentration of alanine aminotransferase (ALT) was not modified by TZD treatment (Fig. 2E) whereas α-fetoprotein, a marker of hepatocellular regeneration and transformation, was drastically reduced in TZD-treated but not in GW1929-treated transgenic mice (Fig. 2F).

Figure 2.

TZD inhibit proliferation and induce apoptosis in hepatocytes of HBV transgenic mice. (A) Graphic representation showing the percentage of RGZ (n = 29), PGZ (n = 27), GW1929 (n = 26), and control (CTRL; n = 26) treated mice with liver tumors. *P < 0.02, **P < 0.01. (B) Histograms show that apoptotic index is significantly higher in TZD than in CTRL and GW1929-treated mice. (C,D) Histograms show a significant inhibition of labeling index of cyclin D1 and PCNA labeling indexes in RGZ-treated and PGZ-treated mice. Values are expressed as mean percentage ± SE (n = 15 in each group of mice) *P < 0.05. (E, F) serum alanine aminotransferase and α-fetoprotein levels were determined in venous blood samples after 6 months of oral drug administration. Data are expressed as mean ± SD. *P < 0.01 versus CTRL or GW1929 treatment.

PPARγ Transcriptional Activity Is Not Involved in TZD Inhibition of Hepatocyte Proliferation.

To determine the effect of TZD and GW1929 on PPARγ transcriptional activity in HBV transgenic mice, the ability of nuclear proteins extracted from isolated hepatocytes to bind a PPRE probe (ARE-7), that has previously been shown to bind preferentially PPARγ over other PPAR isoforms,15 was tested by EMSA. Nuclear extracts from hepatocytes isolated from control transgenic mice contained proteins that retarded the ARE-7 oligonucleotide (Fig. 3A). PPRE binding was increased in extracts from hepatocytes isolated by TZD and GW1929-treated animals suggesting a ligand activation of PPARγ (Fig. 3A, lanes 2-4). The specificity of this band was confirmed by super-shift in extracts incubated with antibody against PPARγ (Fig. 3A, lanes 5-8). The ability of these drugs to modulate PPARγ activation was confirmed by the induced expression of GLUT-2, a PPARγ target gene,16 in hepatocytes isolated from both TZD-treated and GW1929-treated mice (Fig. 3B, lanes 1-4).

Figure 3.

PPARγ transactivation is not required for TZD inhibition of hepatocyte proliferation. (A) PPRE-binding activity was evaluated in nuclear extract from hepatocytes isolated from vehicle or TZD-treated TgN(Alb1HBV)44Bri mice. EMSA was performed with a probe containing a copy of a PPARγ response element (ARE-7) as described in supplemental materials and methods. Where indicated, antibodies against PPARγ (anti-PPARγ) were used to confirm PPARγ presence in the PPRE binding complexes. (B, lane 1-4) TZD effect on the PPARγ target GLUT-2 was evaluate by western blot analysis of protein extracts from hepatocytes isolated from vehicle and TZD-treated (or GW1929-treated) animals and in cultured hepatocytes transfected with the DN-PPARγ or with the pcmx empty vector (B, lane 5-10). Representative data from three independent experiments, each of which was performed in triplicate, are shown. (C) DNA synthesis was measured by [H3]thymidine incorporation into DNA of cultured hepatocytes treated with increased concentration of TZD or GW1929 and in hepatocytes transfected with the PPARγDN or with the pcmx empty vector (E). (D, F) PPARγ transcriptional activity was monitored in hepatocytes treated with the same experimental condition described above and cotransfected with ARE-73-TK-Luc reporter and pSV2-CAT as an internal control for transfection efficiency. Results are expressed as mean ± SD from three experiments done in sextuplicate. *P < 0.05 versus respective controls. **P < 0.02 versus pcmx-transfected cells.

In cultured HBV-derived mouse hepatocytes, TZD and GW1929 similarly induced PPARγ transactivation as monitored by the activity of transfected (ARE-7)3-tk-luc reporter (Fig. 3D) but only TZD were able to induce a dose-dependent inhibition of DNA synthesis (Fig. 3C), thus confirming the direct effect of TZD on hepatocytes proliferation. The inhibition of DNA synthesis by TZD was not modified by transfection of dominant negative PPARγ (DN-PPARγ) (Fig. 3E) that, on the contrary, abolished the ligand-induced reporter activity (Fig. 3F) and GLUT-2 expression (Fig. 3B, lanes 5-10).

Taken together, these results show that PPARγ activation by TZD is not correlated with the ability of these drugs to inhibit hepatocyte DNA synthesis.

Genetic Deficiency of PPARγ Does Not Alter Carcinogenesis in HBV Transgenic Mice and Does Not Influence the Antitumor Effect of TZD.

In consideration that HCC arise from clonal expansion of hepatocytes in TgN(Alb1HBV)44Bri mice,17 we generated a strain of HBV transgenic mice with specific deletion of PPARγ gene in hepatocytes (Supporting Information Fig. 1A-C) in order to clarify the role of PPARγ signaling in tumor development in HBV transgenic mice and evaluate whether the anticancer effects of TZD were mediated by this nuclear receptor in vivo.

No differences in tumor incidence, latency, size, histopathology, and disease progression were observed in animals carrying the PPARγ deletion (Tg[HBV]CreKOγ compared to parental HBV transgenic mice and control Ppargf/f/Tg[HBV]Bri44 mice (Supporting Information Fig. 1D,E).

Five vehicle-treated animals, two RGZ-treated, five PGZ-treated, and two GW1929-treated animals died before the end of the study and were not included in the effective numbers. The effect of TZD administration on incidences, multiplicities, histological features, and size distribution of tumors are summarized in Supporting Information Table 2.

Administration of TZD almost halved the number of hepatic tumors in Tg(HBV)CreKOγ (Fig. 4) and it correlated with a significant increase of apoptosis (Supporting Information Fig. 2) suggesting that the anticancer effect of these drugs is independent of PPARγ expression in hepatocytes.

Figure 4.

TZD inhibit tumor formation independently of PPARγ expression. Tumors were enumerated in mice livers and the multiplicity quantified as described in the Supporting Information Materials and Methods section. *P < 0.05 versus respective controls; #P < 0.03 versus RGZ-treated TgN(Alb1HBV)44Bri mice; P < 0,05 versus PGZ-treated TgN(Alb1HBV)44Bri.

Protein Expression Profiling in Hepatocytes Isolated from PPARγ-Deficient HBV Transgenic Mice After TZD Administration.

To identify novel protein targets that are differentially regulated under chronic oral administration of TZD independently by PPARγ, we performed two-dimensional gel electrophoresis and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry in primary hepatocytes isolated from Tg(HBV)CreKOγ mice.

We used samples from 10 different vehicle-treated and RGZ-treated animals detecting an average of 3527 spots (Fig. 5A). MALDI-TOF peptide fingerprint analysis characterized 26 proteins that were significantly differential expressed; these proteins are listed in Supporting Information Table 3, with their corresponding molecular weight, isoelectric point (pI), and recognized function according to the Swiss-Prot database. The majority of them belong to cytoskeleton, chaperones, and stress/redox regulatory systems. We chose to further investigate nucleophosmin (NPM) because this nucleolar protein, involved in cell growth and transformation,18 was consistently down-regulated at protein (Fig. 5B,D) and messenger RNA (mRNA) levels (Fig. 5C) in hepatocytes of TZD-treated mice, but it was unaffected by GW1929. Moreover, the role of NPM in the development of liver tumors is completely unknown. A dose-dependent reduction on NPM protein and mRNA expression was confirmed by western blot and RT-PCR analysis in PPARγ-deficient hepatocytes cultured in vitro and treated with TZD (Supporting Information Fig. 3A,B). TZD affected NPM expression in hepatocytes at the transcriptional level as demonstrated by the TZD inhibition of NPM promoter activity in transient transfection experiments (Supporting Information Fig. 3C). This effect was not influenced by cotransfection with wtPPARγ or with DN-PPARγ (Supporting Information Fig. 3D).

Figure 5.

Proteome analysis in hepatocytes from Tg(HBV)CreKOγ mice. (A) Two-dimensional DIGE was carried out with 50 μg proteins of hepatocytes isolated from CTRL and RGZ-treated Tg(HBV)CreKOγ mice, and labeled with Cy5 (red) and Cy3 (green), respectively. Gels from three different experiments were scanned using Typhoon TRIO scanner. Identified proteins with at least two-fold change in volume are indicated by arrows. (B) Western blot analysis was carried out with 30 μg of total proteins using specific antibodies against NPM and β-actin. (C) Total RNA was extracted from freshly isolated hepatocytes and reverse transcribed. Real-time PCR was performed with specific primers for NPM and GAPDH (Applied Biosystems Inc., Foster City, CA). (D) Representative immunohistochemistry for NPM of hepatic tissue obtained from vehicle-treated and RGZ-treated Tg(HBV)CreKOγ mice (original magnification: ×1000).

The effect of RGZ on NPM expression was also confirmed both in hepatocytes isolated from TgN(Alb1HBV)44Bri mice cultured in vitro and in human and mice hepatoma cell lines (HuH7 and Hepa 1-6) (Supporting Information Fig. 4). Analysis of the cellular localization of NPM in TZD-treated hepatoma cells showed a significant nucleolar reduction of both NPM and its phosphorylated form (pThr199-NPM) (Supporting Information Fig. 5). Moreover, TZD treatment localized pThr199-NPM in nuclear speckles (Supporting Information Fig. 5, insets), possibly reflecting a reduction in messenger RNA processing.18

AMPK Activation Is Involved in TZD Inhibition of Hepatocyte Proliferation and NPM Expression.

Recently, it has been demonstrated that TZD suppress growth factors tumor-promoting activity via AMPK activation.20 Inhibition of AMPK activity by the specific AMPK inhibitor, compound C, or the dominant negative AMPKα2(D157A), completely prevented the growth arrest induced by TZD treatment in PPARγ-deficient hepatocytes (Fig. 6A,B). Furthermore, TZD treatment induced phosphorylation of AMPK both in vivo, as documented in freshly-isolated hepatocytes from PPARγ-deficient mice (Fig. 6C) and in vitro, in cultured hepatocytes (Fig. 6D). Consistent with our results, expression of the dominant-negative AMPK reverted the TZD-mediated inhibition of NPM expression (Fig. 6E). These results strongly suggest that TZD inhibit hepatocytes proliferation through AMPK activation.

Figure 6.

TZD inhibit hepatic carcinogenesis by activating AMPK. (A) DNA synthesis was measured by [3H]thymidine incorporation in cultured hepatocytes isolated from Tg(HBV)CreKOγ mice. The effect of compound C was evaluated by preincubating the cells for 30 minutes before TZD (20μM) or vehicle treatment. (B) Inhibition of AMPK activity was obtained by transfection with DN-AMPK which harbors a mutation in catalytic subunit (dnAMPKα2D157A). Control cells were transfected with empty vector. Data from three independent experiments, each of which was performed in triplicate, are shown. Results are expressed as mean ± SD, *P < 0.01 versus respective controls. Phosphorylation of AMPK was detected by western blot in cultured PPARγ-deficient hepatocytes after incubation with RGZ (20 μM) at different time points (D) and in hepatocytes freshly isolated from TZD or GW1929-treated Tg(HBV)CreKOγ mice (C). The effect of AMPK inhibition on TZD-modulated expression of NPM was evaluated by western blot in cultured hepatocytes transfected with either dnAMPKα2D157A or control vector (E).

Nucleophosmin/p53 Interaction Mediates the PPARγ-Independent Inhibition of Hepatocyte Proliferation by TZD Treatment.

In consideration that NPM is involved in cell death and proliferation interacting with the tumor suppressor p53,18 we tested whether NPM overexpression could antagonize TZD effect via p53.

Cultured hepatocytes isolated from Tg(HBV)CreKOγ mice were transfected with vector expressing FLAG-tagged NPM under CMV promoter (WT-NPM) or a mutant variant with a deletion of the 120 c-terminal amino acids of NPM (NPMΔC) required for the binding to p53. High levels of FLAG-tagged NPM or NPM mutant proteins were achieved in the transfected cells, whereas no FLAG-tagged proteins were detected in samples transfected with control vector (Fig. 7A, inset). Increased expression of WT-NPM completely abrogated the growth inhibitory effect of TZD but it was not associated with an increase of either thymidine incorporation or incidence of apoptosis in control cultured hepatocytes. On the contrary, expression of the mutant NPMΔC did not modify the antiproliferative and proapoptotic effects of TZD (Fig. 7A,B) suggesting that these antidiabetic drugs induce cell growth arrest by inhibiting NPM expression and consequently its interaction with p53.

Figure 7.

NPM expression antagonizes TZD-induced hepatocytes apoptosis via p53 activation. (A) DNA synthesis was measured by [H3]thymidine incorporation in cultured hepatocytes transfected with the FLAG-tagged wild type (wt NPM) or with the mutant variant (NPMΔC) NPM. Control cells were transfected with empty vector. Data from three independent experiments, each of which was performed in triplicate, are shown. Results are expressed as mean ± SD, *P < 0.02 versus respective controls. Inset, overexpression of wt NPM and NPMΔC protein in cultured hepatocytes: western blot analysis was performed with anti-FLAG antibody. (B) Apoptotic hepatocytes were detected by staining with the DNA binding fluorescent dye Hoechst 33342. Apoptotic cells were expressed as a percentage of the total number of cell counted. Results are expressed as mean ± SE of four independent experiments, each of which was performed in duplicate, *P < 0.05 versus respective controls. (C) Effect of TZD and GW1929 on p53 activation in cultured or freshly isolated (E) hepatocytes. (D) Effect of wtNPM or NPMΔC expression on p53 activation. Whole-cell lysates were analyzed by immunoblotting with antibodies specific for phospho-p53Ser-15, total p53, p21, and β-actin.

It has been shown that NPM interacts with p53 and regulates p53 phosphorylation at the Serine-15 residue which is crucial for p53 transactivation and subsequent apoptotic signals transduction.21 We thus determined whether TZD-inhibited NPM expression may affect p53 activity in PPARγ-deficient hepatocytes. As shown in Fig. 7C, TZD induced both P-p53Ser-15 and its target gene cyclin-dependent kinase inhibitor p21WAF1/CIP1. Strikingly, over expression of NPM significantly reduced the TZD-induced P-p53Ser-15 and p21 expression, whereas overexpression of the mutant NPMΔC failed to oppose TZD effect on p53 activation (Fig. 7D). In parallel, induction of P-p53Ser-15 was also demonstrated in freshly isolated hepatocytes from TZD-treated Tg(HBV)CreKOγ mice compared to vehicle-treated animals (Fig. 7E). Taken together these results indicate that the PPARγ-independent antiproliferative effect of TZD is mediated by NMP through a mechanism involving p53.


Our study shows that chronic administration of two different TZD, significantly inhibit tumor formation in a HBV-related mouse model of hepatocarcinogenesis. This effect was correlated by in vivo and in vitro inhibition of hepatocyte proliferation and induction of apoptosis with negligible effects on the degenerative and the inflammatory responses. On the contrary, the non-TZD PPARγ agonist GW1929 had no effect on tumor formation and hepatocyte proliferation although this drug is able to induce PPARγ transactivation and target gene expression in mouse hepatocytes. This suggests that PPARγ activation is unlikely involved in the antitumor effect of TZD in mouse liver.

Previous in vitro evidences suggest that the antiproliferative effect of TZD is independent of PPARγ activation; indeed, troglitazone induced growth arrest by inhibition of translation initiation in PPARγ−/− embryonic stem cells.22 Similarly, we found that in hepatocytes isolated from HBV transgenic mice, the growth inhibitory effect of TZD is dissociated from the ability of these drugs to promote PPARγ transactivation. In fact, ectopic expression of DN-PPARγ was unable to revert the growth inhibitory effect of TZD.

Although PPARγ is clearly recognized as master regulator of lineage-specific cell differentiation that differs according to the cellular type,23 the correlation between PPARγ activation and programmed cell death induced by TZD is doubted. In pancreatic cancer cells, TZD-induced PPAR-dependent growth arrest is primarily mediated by cell differentiation without proapoptotic effects.24 Conversely, TZD analogues, which have a double bond adjoining the terminal thiazolidinedione ring that is responsible for the abrogation of the PPARγ ligand property, retain the ability to induce apoptosis with a potency equal to that of their parental TZD in cancer cell lines,25 suggesting that mechanisms involved in TZD-induced differentiation differ from those mediating apoptosis. The dissociation of TZD effects on apoptosis from their original pharmacological activity (i.e., PPARγ activation), is in line with the observation that sensitivity of cancer cells to TZD-induced growth inhibition does not correlate with the PPARγ expression levels, and there exists a three orders of magnitude discrepancy between the concentration required to produce antitumor effects and that needed to modify insulin action.26

The PPARγ-independent proapoptotic effect of TZD was confirmed in triple transgenic animals Tg(HBV)CreKOγ in which Cre specifically deletes PPARγ in hepatocytes. In this experimental model, genetic deficiency of PPARγ does not modify the process of hepatic carcinogenesis and tumor development when compared to parental HBV transgenic mice. This is in agreement with the demonstration that PPARγ deletion does not alter the development of experimental prostate and breast cancers.27, 28 In addition, when mice expressing a constitutively active form of PPARγ in mammary glands were bred to transgenic mice prone to mammary gland cancer, the resulting offspring develop tumors with greatly accelerated kinetics,8 suggesting a pro-oncogenic role of PPARγ. Moreover, in human pancreatic and ovarian cancers expression profiles indicate a strong overexpression of PPARγ that positively correlate with higher pT stages and higher tumor grade.29, 30

Interestingly, our experiments showed that in Tg(HBV)CreKOγ mice TZD administration inhibits tumor formation with a potency significantly higher than in parental and control mice. Moreover, PPARγ ectopic expression in PPARγ-deficient hepatocytes reduced the antiproliferative effect of TZD (Supporting Information Fig. 6).

How PPARγ expression limits TZD anticancer activities remains speculative. We hypothesize that the net effect of TZD in cancer cells is the result of the balance of PPARγ-mediated (pro-oncogenic) and PPARγ-independent (anti-oncogenic) mechanisms that depends on different factors including receptor expression levels, phosphorylation status, expression of the heterodimeric partners, and the presence of endogenous ligands.31 This might explain the limited therapeutic efficacy of TZD treatment in oncological trials except for tumor types with reduced levels and possible loss of function of PPARγ such as prostate and thyroid cancers.32, 33

In vitro studies have suggested that TZD mediate antiproliferative effects through a complexity of PPARγ-independent mechanisms. Experimental evidence indicates that troglitazone and ciglitazone block BH3 domain-mediated interactions between Bcl-2 family members and facilitate the degradation of cyclin D1 through proteasome-mediated proteolysis.34, 35 In our study, we identified a novel molecular target by which TZD inhibit hepatocyte proliferation in vivo. Proteomic analysis showed that TZD reduce the expression of NPM, a nucleolar protein characterized as a central regulator of ribosomal RNA processing that has been found to be more abundant in tumor and growing cells than in corresponding normal cells.17

In HCC, NPM overexpression is correlated with clinical parameters, such as serum α-fetoprotein level and tumor pathological grading, suggesting that NPM might serve as a potential marker for HCC.36 In agreement, in our mouse model we found a progressive age-related increase of NPM that parallels the increase of PCNA-LI in hepatocytes (Supporting Information Fig. 7).

TZD inhibited the expression of NPM at protein and mRNA levels in both isolated hepatocytes and hepatoma cell lines, and significantly repressed NPM promoter activity independently of the ectopic expression of wild-type PPARγ or DN-PPARγ. These data are in agreement with the absence of PPRE in the NPM promoter (A. Galli, E. Ceni, L. Cioni, unpublished observation, 2009).

The molecular pathway involved in the PPARγ-independent regulation of gene expression by TZD is scarcely known. TZD have been recently shown to activate AMPK, a cellular sensor of energy status.37 AMPK may suppress tumorigenesis regulating cell growth via inhibition of mammalian target of rapamycin (mTOR) signaling and p53 activation, and it is indicated as a beneficial target for cancer treatment.38 We showed that TZD induced AMPK activation in PPARγ-deficient hepatocytes and that inhibition of AMPK activity completely prevented the TZD-induced growth arrest and NPM expression. These results are in agreement with the observation that TZD specifically inhibit IGF-I tumor-promoting activity in mouse skin through activation of AMPK and subsequent inhibition of mTOR pathway.20 In addition, AMPK activation was demonstrated to induce p53 phosphorylation and p53-dependent apoptotic cell death in response to energetic stress.39

Although there is no evidence for a direct involvement of NPM in the regulation of the apoptotic machinery, NPM might function as an antiapoptotic protein through indirect mechanisms. Interaction with p53 might be an important step by which NPM inhibits programmed cell death. In fact, NPM overexpression protects mouse embryonic fibroblast against hypoxic cell death, but this effect is not observed in cell that lacks p53.40

We showed that in cultured PPARγ-deficient hepatocytes, ectopic expression of wild-type NPM significantly blocked TZD inhibition whereas a mutant variant lacking the p53-interacting domain did not prevent TZD antiproliferative and proapoptotic actions. Similarly, in malignant haematopoietic cells, the same NPM mutant does not prevent apoptosis in response to stress stimuli, unlike the overexpression of wild-type NPM.41 NPM-p53 interaction inhibits p53 phosphorylation at the serine 15, and subsequently represses p53 target genes expression such as the cell cycle inhibitor p21. However, in hepatic cells TZD may promote p53 phosphorylation by inhibiting NPM gene expression. Interestingly, NPM has also been shown to interact with p53 in hypoxic cells and to inhibit hypoxia-induced p53 phosphorylation on the same residue.42 Besides, regulation of p53 expression and activity by TZD has been also demonstrated in human cholangiocarcinoma cells.43 In consideration that the ability of AMPK to induce cell cycle arrest is dependent on p53 phosphorylation at Ser15,39 it might be conceivable that TZD modify p53 phosphorylation status and activity by an AMPK-mediated down-regulation of NPM.

In conclusion, we have shown that chronic administration of TZD inhibits hepatic tumor formation in mice with a PPARγ-independent mechanism. Furthermore, we found that the anticancer activity of these drugs in the liver was mediated, at least in part, by inhibition of NPM expression and p53 activation. Collectively, these observations provide new insight into the molecular mechanisms of hepatic carcinogenesis and emphasize relevant clinical implication. In fact, recent studies have consistently indicated TZD as antifibrotic drugs for treatment of different chronic hepatic diseases,44, 45 and moreover, our data suggest that these molecules or correlated analogues, alone or in combination with other anticancer drugs, may have translational relevance in effective strategies for HCC chemoprevention and treatment.


The authors thank Dr. R. Evans for PPARγ constructs and Dr. Q. Pang for NPM constructs; Dr. K. Guan for dnAMPKα2D157A plasmid and Professor M. Luconi for many helpful comments and suggestions. We also thank Dr. D. Lucci for the statistical support.