Rosiglitazone metformin adduct inhibits hepatocellular carcinoma proliferation via activation of AMPK/p21 pathway

Rosiglitazone, metformin and compound C were purchased from Gao Meng Chemical Co., Ltd. (Beijing, China), Zhong Xin Pharmaceutical Co., Ltd. (Tianjin, China) and Apexbio Co., Ltd (USA), respectively. All these agents were dissolved in dimethyl sulfoxide (DMSO, Sigma, USA) and then diluted to indicated concentrations by phosphate-buffered saline (PBS, final concentration of DMSO < 0.1%) in vitro and in vivo experiments.

Mortars and pestles were previously washed and placed in a dryer for 30 min. Then, each compound (1 mg) was ground into powder with dried KBr (50 mg) in a mortar. The mixture was pressed into a piece of slice and tested in an infrared spectrometer (Nicolet, USA).

Human hepatoma cell lines HepG2, SK-hep1 and immortalized normal human liver cell line (L02) were obtained from the Chinese Academy of Sciences Cell Bank. Cells were cultured in DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin under 37 °C and 5% CO₂ humidified condition. Logarithmically growing cells were treated with indicated concentrations of each compound for 48 h and then replaced with regular culture medium.

The effect on cellular proliferation was evaluated using Cell Titer96^® Aqueous One Solution Cell Proliferation Assay Kit (Promega, USA). This kit contains an MTS reagent, which was composed of [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] and phenazinemethosulfate (PMS). 2 × 10³ cells were plated into each well of 96-well plates and incubated in regular culture medium for 6 h. Then, cells were treated with indicated doses of the agents for 48 h. 20 μL MTS reagents were added into each well and incubated at 37 °C for 2 h. The absorbance was measured at 490 nm. Experiments were performed in triplicate.

5 × 10² cells were seeded into six-well plates with indicated compounds treatment for 48 h and replaced with regular culture to continuously incubate for 10 days. Colonies were washed and fixed with 4% paraformaldehyde for 30 min. Then, colonies were stained with 0.1% crystal violet for 30 min and counted.

Cells were plated at 5 × 10⁴ per well in 24-well plates. After treated for 48 h, cells were washed twice with PBS, fixed with 4% paraformaldehyde for 30 min and stained with 10 μg/mL hoechst reagent (Hoechst 33342, Beytime, China) for 15 min. Cells were then washed thrice for scanning at 200× by fluorescent microscope.

Total RNA was extracted from cells using Trizol reagent (Life Technologies Corporation, USA). RNA reverse transcription was processed following the manufacturer’s instructions by using a Reverse Transcription Kit (Takara, Japan). Then, RT-qPCR was performed using Faststart Essential DNA Green Master (Roche, Indianapolis, IN, USA). GAPDH was used as an internal control. All results expressed as the mean ± SEM were performed at least three independent experiments. Comparative quantification was determined using the 2^−ΔΔCt method. The premier sequences were as follows: p53 (forward): 5′-GGAAATTTGCGTGTGGAGTATTT-3′, (reverse): 5′-GTTGTAGTGGATGGTGGTACAG-3′; CCND1 (forward): 5′-CCTCGGTGTCCTACTTCAAATG-3′, (reverse): 5′-CACTTCTGTTCCTCGCAGAC-3′; p21 (forward): 5′-GTCACTGTCTTGTACCCTTGTG-3′, (reverse): 5′-TTTCTACCACTCCAAACGCC-3′; CDK2 (forward): 5′-AGATGGACGGAGCTTGTTATC-3′, (reverse): 5′-CTTGGTCACATCCTGGAAGAA-3′; CDK6 (forward): 5′-TCACGAACAG ACAGAGAAACC-3′, (reverse): 5′-CTCCAGGCTCTGGAACTTTATC-3′; EGFR (forward): 5′-G GAAGTACAAAGAGGAGGAAGAG-3′, (reverse): 5′-GGGAAGATGCCAGGGATAAA-3′; GADD45A (forward): 5′-GGAGAGCAGAAGACCGAAAG-3′, (reverse): 5′-GATCAGGGTAGTGGATCTG-3′; NANOG (forward): 5′-TCCTGAACCTCAGCTACAAAC-3′, (reverse): 5′-GCGTCACACCATTGCTATTC-3′; MYC (forward): 5′-GCTGCTTAGACGCTGGATTT-3′, (reverse): 5′ GAGTCGTAGTCGAGGTCATAGTT-3′; GAPDH (forward): 5′-CGGAGTCAACGGATTTGGTCGTAT-3′, (reverse): 5′-AGCCTTCTCCATGGTGGTGAAGAC-3′.

Transfection of HCC cells with shRNA was carried out using Lipofectamine 2000 (Invitrogen, USA), according to the manufacturer’s instructions. Cells were grown to 50–70% density in 6-well plates before transfection. 2.5 μg sh-p21 or shNC was incubated for 6 h plus 5 µL lipofectamine reagent in Opti-MEM medium. Then, transfection medium was replaced by fresh regular culture medium. After 48 h of incubation, transfection efficacy was assessed by RT-qPCR and western blot analysis. The sequences of the RNA used in transfection were as follows: sh-p21 (forward): 5′-GATCCGGCTGATCTTCTCCAAGAGGACTTCCTGTCAGATCCTCTTGGAGAAGATCAGCCTTTTTG-3′, (reverse): 5′-AATTCAAAAAGGCTGATCTTCTCCAAGAGGATCTGACAGGAAGTCCTCTTGGAGAAGATCAGCCG-3′. sh-NC (forward): 5′-GATCCGCCACTTTGAAGAACCCAATCCTTCCTGTCAGAGATTGGGTTCTTCAAAGTGGCTTTTTG-3′, (reverse): 5′-AATTCAAAAAGCCACTTTGAAGAACCCAATCTCTGACAGGAAGGATTGGGTTCTTCAAAGTGGCG -3′.

Four week-old male BALB/c-nude mice (purchased from Laboratory Animal Services Center of Chongqing Medical University) were randomly divided into three groups: RZM-treated group, Met-treated group and NC group. Mice were maintained under pathogen free conditions. All procedures for the mouse experiments were approved by the Animal Care Committee of Chongqing Medical College. HepG2 cells (1 × 10⁷, 200 μL) were subcutaneously injected into the left hind leg of the nude mice. Treatment was started when subcutaneous lesions were macroscopically apparent, reaching 30–50 mm³. Mice were treated with the agents by intraperitoneal injection (125 mg/kg), or sterilized saline as control every 4 days. Meanwhile, tumor size and mice weight were evaluated. Tumor volumes (expressed in mm³) were calculated using the following formula: 0.5 × (length × width²). Mice were sacrificed after 4 weeks of treatment. Subcutaneous tumor tissues were harvested and frozen in liquid nitrogen for protein analysis.

Rosiglitazone sodium (6 mmol), metformin (6 mmol) and absolute ethanol (120 mL) were collected in a round bottom flask. Then, the mixture was heated for 10 min under the condition of continuously whisking. After filtration, the liquid was distillated for removal of ethanol with a rotating evaporator and cooled down. At last, rosiglitazone metformin adduct (RZM) was obtained after recrystallized from absolute ethanol (2.26 g, yield 77%, melting point: 182–184 °C) [19]. The structure of Ros, Met, the mixture and RZM were identified with IR spectrum analysis and the data for characteristic absorption peaks were shown in Table 1. Ros possessed C=O bonds in thiolactone and lactam, aromatic oxide, =C–H bonds in benzene and pyridine ring and N–H bond (absorption peaks at 1738.37, 1694.06, 1246.73, 812.24, 763.19, 3416.43 cm⁻¹) (Fig. 1a). Met possessed C=NH, N–H bond, N–H bending vibration, dimethyl anti-symmetric vibration, dimethyl symmetric vibration and NH₂ bond (absorption peaks at 1622.80, 3392.57, 1583.69, 1568.35, 1487.91, 1475.28, 1448.73, 1418.94, 3294.66 cm⁻¹) (Fig. 1b). Because Met has less mass fraction of the mixture at an equal molar mass ratio [MW: 166/(166 + 357)], a few peaks of Met were not obviously found in spectrum of the mixture. The results showed that absorption peaks of Ros and Met did not move and no new characteristic peaks were generated in IR spectrum of the mixture (Fig. 1c). However, the IR spectrum of RZM (Fig. 1d) indicated that N–H, C=O bonds of Ros, and NH₂ of Met moved toward low-frequency vibration in RZM and generated sharp peaks at 3386.60, 1676.56, 1643.85, 3185.41 cm⁻¹. These data confirmed that new hydrogen bonds were produced. At the same time, peaks of C=NH, N–H bending vibration, aromatic oxide, and =C–H bonds did not shift, suggesting that these functional groups did not participate in hydrogen bond formation or other chemical changes.

All above data showed that RZM is a novel adduct consist of Ros and Met, and these two agents were integrated by hydrogen bonds.

MTS assay was performed to investigate the effect of RZM on HCC cells proliferation. We found that RZM inhibited two human hepatoma cells (HepG2 and SK-Hep1 cells) proliferation in a dose-dependent manner. Ros and/or Met also caused a dose-dependent inhibition in HCC cells. However, a significant reduction of proliferation was found in the presence of RZM at 1 mM, but not in other two agents (Fig. 2a, b). No significant difference was observed when normal human liver cell line L02 cells treated with low doses of RZM (Fig. 2c). Colony formation assay supported the results that RZM significantly decreased the percentage of colonies in HepG2 and SK-Hep1 cells compared with Ros and/or Met (Fig. 2d, e). These results indicated that low concentrations of RZM enhanced antiproliferative effects in HCC cells compared with Ros and/or Met, which did not impair the normal human liver cells viability. We further compared the apoptosis effects of RZM and Met on HepG2 cells. Results displayed that 1 mM Met did not cause cellular apoptosis, which is consistent with previous observations [20]. However, cells exposed to RZM at 1 mM for 48 h displayed cellular apoptosis with nuclear condensation by hoechst staining (Additional file 1: Figure S1). Furthermore, flow cytometric analysis was used to determine the effects of the two compounds on cellular apoptosis. Results showed that RZM triggered significantly cellular apoptosis at 1 mM, which was not found in Met-treated group (Fig. 2f, g). Taken together, the above results demonstrated that RZM inhibited HCC cells proliferation and caused apoptosis in vitro.

To explore the molecular mechanisms by which RZM repressed HCC proliferation, several crucial gene expressions relative to hepatocarcinogenesis were detected by RT-qPCR. As a result, the expression of p21 was obviously elevated in RZM-treated (1 mM) group compared with those of control group (Fig. 3a). Then, we constructed specific shRNA inhibiting p21 (sh-p21) and transfected into HepG2 cells. The knockdown efficiencies of p21 were confirmed on both mRNA and protein levels. Furthermore, the effects of p21 knockdown or/and RZM treatment on expressions of p21 were also detected. Results indicated that RZM at 1 mM partly reversed the depressed expression of p21 in sh-p21 transfected cells as well as restored HepG2 cells proliferation (Fig. 3b–d). In short, the above findings revealed that RZM could specifically upregulate the expression of p21.

Previous studies have shown that the antineoplastic effects of Met might be mainly achieved by activation of AMP-activated protein kinase (AMPK) [21, 22]. Therefore, it was reasonable to hypothesize that RZM also activated AMPK. So we detected phosphorylation of AMPKα in HCC cells in response to these three agents by western blotting analysis. Results indicated that RZM enhanced the stimulatory effect on phosphorylation of AMPKα compared with Met and/or Ros (Fig. 3e). Compound C, an AMPK inhibitor, was used to demonstrate the role of AMPK in which RZM elevated the expression of p21. Results showed that compound C effectively alleviated increased expression of p21 induced by RZM, proving that RZM modulated p21 expression via the activation of AMPK (Fig. 3f). In all, the above results demonstrated that RZM enhanced HCC cells proliferation compared with Met via activation of AMPK/p21 pathway.

Next, we investigated whether RZM could affect HCC tumor growth in vivo. HepG2 cells were injected into male nude mice to develop tumors. Treatment with the two compounds at 125 mg/kg or equal volume of sterilized saline was begun when the tumors were visible (the 8 day after implantation). The significant effects of two compounds on tumor growth were observed following 12 days of treatment (20 days after implantation; Fig. 4a). After 4 weeks, nude mice were sacrificed and the tumors in each group were exhibited (Fig. 4b, c). Both of two agents significantly suppressed tumor growth compared with the control group, while smaller tumor volumes and masses were observed in RZM treated mice (Fig. 4d, e). Moreover, the protein levels of p21 and p-AMPKα in tumor tissues were detected. Results showed that p21 and p-AMPKα expressions were obviously increased in RZM-treated group (Fig. 4f). During the course of treatment, the compounds did not cause noticeable side effects or changes in mice body weight (Additional file 2: Figure S2). Thus, we concluded that RZM could induce an additive inhibition on the tumor growth by increasing p21 and p-AMPK expressions in vivo.