Metformin was purchased from Sigma-Aldrich (St. Louis, MO, USA). Compound C, an AMPK inhibitor, was supplied by Calbiochem (La Jolla, CA, USA). Antibodies against AMPK, p-AMPK (Thr172), p-ACC (Ser79), p53, acetyl-p53 (Lys382), p-p53 (Ser15), p21^waf1/cip1, p16INK4A, p-RB (Ser795), PARP and SIRT1 were provided by Cell Signaling Technology (Danvers, MA, USA). The antibody specially recognizing Dec1 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). p53 and control siRNA were purchased from Cell Signaling Technology. Human liver cancer cell lines HepG2 and Bel-7402 (both bearing wild-type p53), a gift from Professor X. Guo (Sun Yat-Sen University, Guangzhou, China) were maintained in RPMI-1640 supplied with 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen, Carlsbad, CA, USA). Cultures were incubated at 37°C and 5% CO₂. Logarithmically growing cells were treated with the indicated concentrations of metformin or compound C for 48 or 24 h and then replaced with regular culture medium.

Hepatoma cells were seeded on 96-well plates at 5×10³ cells per well. After different treatments, 20 μl of MTS reagent was added to each well using a multi-well pipettor and incubated for 3 h. Cell proliferation was assessed by measuring the absorbance at 490 nm using a Bio-Tek Synergy2 microplate reader. Each experiment was repeated 3 times.

In cells treated with metformin, compound C or siRNA, BrdU incorporation was assayed using a commercial kit (Cell Signaling Technology). Briefly, cells were grown on 22-cm coverslips and incubated with anti-BrdU primary antibody at 4°C overnight. The cells were then washed with PBS and processed for immunofluorescence. Cells were mounted with Prolong Gold Antifade Reagent (Invitrogen, Eugene, OR, USA) and visualized under a fluorescence microscope (Leica CTR6000) at ×100 or ×200 magnification.

To detect whether metformin induces cell apoptosis, we stained the treated cells with an Annexin V kit (Sigma-Aldrich). Briefly, 1×10⁶ hepatoma cells were treated with different concentration of metformin for 48 h, stained with Annexin V/propidium iodide (PI) and then subjected to FACSCalibur cytometry (BD Bioscience, Mountain View, CA, USA).

Cell cycle distribution was performed as previously described (12). The cells were harvested after treatment with metformin, then fixed with 70% cold ethanol and stained with 50 mg/ml PI followed by RNase A treatment for 30 min at room temperature. DNA content was analyzed with a FACSCalibur cytometer (BD Bioscience). The population of cells in each phase was determined by using ModFit LT software (BD Bioscience).

SA-β-gal staining was performed as previously described (16). In brief, hepatoma cells were seeded in 6-well plates and fixed with 4% formaldehyde for 5 min at room temperature. The cells were then washed with PBS and incubated with fresh SA-β-gal staining solution containing 1.0 mg/ml X-galactosidase (Stratagene, La Jolla, CA, USA) at 37°C for 16–18 h to visualize SA-β-gal staining.

Cell extracts were prepared by lysing the cells in RIPA buffer and protein concentration was quantified through BCA assay. Protein sample (50 μg) was subjected to 10% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred onto polyvinylidene difluoride membrane (Millipore Corp., Billerica, MA, USA). The membranes were blocked with 5% non-fat milk in PBS containing 0.05% Tween-20 for 1 h at room temperature and then incubated with specific primary antibodies. Detection of specific proteins was performed by enhanced chemiluminescence reagent (Thermo Scientific, Rockford, IL, USA).

Transfection of hepatoma cells with siRNA was carried out using Lipofectamine 2000, according to the manufacturer’s instructions (Invitrogen). Briefly, aliquots of 1.5×10⁴ cells were seeded on 6-well plates and grown to 50–70% confluence before transfection. Cells were then transfected with 60 pmol p53 siRNA for 6 h plus 1 μl Lipofectamine in Opti-MEM^® I medium. Subsequently, the transfection medium was replaced by fresh regular culture medium. After 48 h of incubation, protein levels were detected by western blot analysis.

The data are presented as the mean ± standard deviation and statistical comparisons between groups were done using one-way analysis of variance followed by Student’s t-test. P<0.05 was considered statistically significant.

In this study, metformin caused inhibition of cell proliferation in a dose-dependent manner (Fig. 1A), which is consistent with previous observations (11,12). In addition, several studies have shown that metformin has pro-apoptotic effects on hepatoma cells at concentration of ≥5 mM (11,12). In the present study, 10–50 mM metformin caused a significant induction of apoptosis as evidenced by Annexin V staining and accumulation of cleavage PARP protein. However, low dose of metformin (0.01–1 mM) did not (Fig. 1B and C). Interestingly, 1 mM metformin was sufficient to impair HepG2 cell proliferation without apparent induction of apoptosis. Similar results were obtained in another hepatoma cell line Bel-7402 (data not shown) following metformin treatment, indicating that the effect is not cell line specific. On the basis of these findings, we concluded that metformin exerts a multiphase effect on hepatoma cells, with low dose leading to a sustained decrease in proliferation, and triggering of a higher rate of apoptotic cell death.

HepG2 cells exposed to low concentration of metformin were characterized by enlarged and flatted cell phenotype in our initial observation (data not shown), which is likely associated with cellular senescence (17), therefore we sought to determine whether low doses of metformin might induce senescence in hepatoma cells. Indeed, low concentration of metformin induced HepG2 cell senescence in a dose-dependent manner, as evident by increased SA-β-gal activity, a universal marker of cellular senescence (Fig. 2A and B). Moreover, protein expression level of Dec1, one of the established markers for senescence (18) was enhanced in response to metformin compared with controls (Fig. 2C), which is in accordance with SA-β-gal staining results. Additionally, HepG2 cells treated with 1 mM metformin exhibited a marked decrease in BrdU incorporation, a marker of cell proliferation (Fig. 3A and B). Of note, when we treated HepG2 cells with metformin for 48 h and either metformin was withdrawn or continued the treatment for another 72 h, there was no significant difference in cell viability between the persistent metformin-treated cells and the metformin-removed cells (data not shown), suggesting that metformin-induced proliferation arrest is irreversible even when the drug is withdrawn. To further clarify the effect of metformin on cellular senescence, we investigated the cell cycle progression in the absence or presence of metformin by flow cytometric analysis. The results indicated that the percentage of G₀/G₁ phase cells was significantly increased in metformin treated cells and correspondingly, the cell population of G₂/M was declined compared with the controls (Fig. 3C). Similarly, we observed senescence in Bel-7402 cells with metformin treatment (data not shown), indicating that low doses of metformin triggered senescence in hepatoma cells.

Next, we sought to determine the mechanism by which metformin induced senescence in hepatoma cells. Since 1 mM metformin caused significant induction of cellular senescence without apparent induction of apoptosis, we treated HepG2 cells with 1 mM metformin in the subsequent experiments. We detected the expression of phosphorylation of AMPK and its downstream gene ACC through western blot analysis. In metformin-treated senescent HepG2 cells, robust phosphorylation of AMPK and ACC was observed compared with untreated cells (Fig. 4A), indicating that activation of AMPK signaling cascade are relevant to the senescence in HepG2 cells. To further confirm the relationship between HepG2 cell senescence and AMPK activation, we examined the effects of compound C, a specific AMPK inhibitor, on senescence in HepG2 cells. As shown in Fig. 4B–E, compound C reduced metformin-stimulated SA-β-gal-positive cells and restored proliferation. In addition, the protein level of p-AMPK and p-ACC was attenuated compared to untreated cells (Fig. 4A). These results suggest that AMPK pathway plays a crucial role in metformin-induced senescence in hepatoma cells.

It is well known that senescence is mainly manipulated through canonical p53/p21 and p16/pRB signaling pathway (19). Thus, we sought to determine whether activation of AMPK pathway depends on these two pathways. Interestingly, the acetylation of p53 (Ac-p53) and p21 were increased in metformin-treated HepG2 cells, while the protein expression of phosphorylation of p53 at Serine 15, p53, p16 and pRB was unaltered (Fig. 5A). On the other hand, treatment with compound C for 24 h obviously reduced protein expression of Ac-p53 and p21 compared with untreated cells (Fig. 5A). Furthermore, siRNA-mediated knockdown of p53 diminished metformin-induced SA-β-gal positive cells (Fig. 5B and C). These data imply that the activation of AMPK pathway is dependent on the p53.

As AMPK activation promotes acetylation of p53, we investigated whether SIRT1, a NAD-dependent protein deacetylase (20), was involved in the regulation of p53 transcriptional activity. Toward this goal, HepG2 cells were co-transfected with Myc-p53 and Flag-SIRT1 plasmid and then treated with metformin. As shown in Fig. 6, co-expression of SIRT1 and p53 significantly suppressed the level of acetylated p53, but 1 mM metformin treatment partially rescued the effect of SIRT1 on p53 acetylation and increased SA-β-gal staining, suggesting that activating AMPK by metformin treatment might inhibit the SIRT1 deacetylase activity on p53, resulting in promotion of HepG2 cell senescence.