Surgical resection samples were obtained from 47 patients (including 9 females and 38 males) diagnosed with HCC at the First Affiliated Hospital, Medical College of Xi’an Jiaotong University, Xi’an, China from 2001 to 2003. The clinicopathological data of these patients, including the tumor stage, grade, differentiation, and survival time, were collected, and the follow-up data were updated through June 2006. A total of 9 match-paired HCC tissues and adjacent non-tumor tissues were collected from the Department of Hepatobiliary Surgery, First Affiliated Hospital, Medical College of Xi’an Jiaotong University. All of the tissue samples were obtained from untreated patients who were undergoing surgery. The study was approved by the Medical Ethics Committee of the First Affiliated Hospital, Medical College of Xi’an Jiaotong University, and all the patients formally consented to be a part of the study to the best of their understanding.

A standard immunostaining procedure was performed using mouse monoclonal antibodies against Bmi1 (Millipore; Boston, MA, USA), proliferating cell nuclear antigen (PCNA; Maixin Bio, Fujian, China), Ki-67 (Maixin Bio) or an isotype-matched control antibody. The immunoreactivity and subcellular localization of Bmi1 were evaluated independently by 3 investigators.

Bmi1 staining was classified into 2 groups, negative or positive, based on the percentage of positive cells and the staining intensity. The percentage of positive cells was divided into 4 ranks of scores: <10% (1), 0–25% (2), 25–50% (3) and >50% (4). The intensity of staining was also divided into 4 ranks of scores: no staining (1), light brown (2), brown (3) and dark brown (4). The positivity of Bmi1 staining was determined by the following formula: immunohistochemistry score = percentage score × intensity score. An overall score of ≤8 was defined as negative and >8 as positive.

The HCC-derived cell lines, Bel-7402 (CCTCC GDC035), SMMC-7721 (CCTCC GDC064) and HepG2 (CCTCC GDC024), were all purchased from the China Center for Type Culture Collection (CCTCC; Wuhan, China). The cells were maintained in RPMI-1640 medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (FBS, Invitrogen) in a humidified atmosphere at 37°C with 5% CO₂. The MHCC97 cells (Cell Bank of Chung Shan Hospital, Shanghai, China) were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Invitrogen) supplemented with 10% FBS in a humidified atmosphere at 37°C with 5% CO₂.

The oligonucleotide insert for the hairpin siRNA targeting the Bmi1 mRNA sequences was GGAGGAACCTTTAAAGGA TTA. The oligonucleotide sequences were 5′-CACCGGAG GAACCTTTAAAGGATTATTCAAGAGATAATCCTTTAA AGGTTCCTCCTTTTTTG-3′ and 5′-GATCCAAAAAAGG AGGAACCTTTAAAGGATTATCTCTTGAATAATCCTTT AAAGGTTCCTCC-3′. The synthesized oligonucleotide (GenePharma, Shanghai, China) inserts were annealed and cloned into the PGPU6/GFP/neo-shRNA expression vector (GenePharma) to generate PGPU6/GFP/neo-shBmi1. The plasmid PGPU6/GFP/neo-shControl (GenePharma) was used as the negative control and encoded a hairpin siRNA with a nonsense sequence.

For stable cell line generation, the transfection was performed using Lipofectamine 2000 (Invitrogen) following the manufacturer’s instructions. Bmi1 stable-knockdown cells and the control cells were selected using 0.8 mg/ml G418 (Calbiochem, La Jolla, CA, USA).

Western blot analyses were performed as previously described using cell lysates (21). The crude proteins were then subjected to SDS-PAGE and then transferred onto a PVDF membrane. After blocking, the membrane was incubated with the appropriate antibody against Bmi1 (Millipore) or β-actin (Santa Cruz Biotechnology, Santa Cruz, CA, USA) at 4°C overnight, followed by a horseradish peroxidase-labeled secondary antibody. The blots were developed using a chemiluminescent detection system (Amersham Life Science, Buckinghamshire, UK).

Cells (1×10⁶) were cultured in 6-well plates for 24 h and then harvested and washed with PBS, followed by fixation with 70% ethanol overnight at 4°C. After washing with PBS twice, the cells were stained in PBS with 50 μg/ml propidium iodide (PI; Sigma, St. Louis, MO, USA) and 10 μg/ml RNase A (Sigma) at room temperature in the dark. The cell cycle was assessed by flow cytometry (FACSCalibur; BD Biosciences, Franklin Lakes, NJ, USA), and the data were analyzed with the FACSCalibur flow cytometer using ModFit LT software.

Cell proliferation was evaluated on days 1, 3, 5, and 7 after seeding the cells (5×10⁴) in triplicate in 6-well plates. A total of 200 cells plated on 100-mm cell culture dishes in triplicate were cultured for 3 weeks. The cell colonies were stained with Giemsa solution after being fixed in methanol for 15 min at room temperature.

The stable Bmi1 knockdown or control cell line (10⁶ cells) was injected into the subcutaneous tissue in the dorsum of 4–6-week-old male Balb/c-nude mice. Three animals per group were used in each experiment. The tumors were measured weekly using a vernier caliper, and the volume was calculated according to the following formula: length × width²/2. At the end of the experiment, the tumors were dissected, and their net weights were measured. The experimental protocols were evaluated and approved by the Animal Care and Use Committee of the Medical College of Xi’an Jiaotong University.

Sorafenib was dissolved in DMSO (Sigma) and diluted with RPMI-1640 to the desired concentration (5 μM), with a final DMSO concentration of 0.1% for the in vitro studies. DMSO at 0.1% (v/v) was used as a solvent control.

Cells (1×10³) were plated on 100-mm cell culture dishes in triplicate, and cultured for 3 weeks. Sorafenib or the solvent was added at the appropriate concentration after 24 h. The cell colonies were stained with Giemsa solution after being fixed in methanol for 15 min at room temperature.

A total of 5,000 cells were inoculated in 96-well microtiter plates and incubated overnight at 37°C in a humidified incubator with 5% CO₂. The cell viability was quantified every day using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT; Sigma) staining, according to a standard protocol. The cells were incubated with sorafenib at various concentrations for an additional 72 h to evaluate the inhibitory effect of sorafenib on cell proliferation. The number of viable cells was determined by measuring the absorbance at 490 nm.

All the experiments were repeated at least in triplicate. The data from all of the experiments were pooled, and the results are expressed as the means ± SD. The statistical analysis was performed using SPSS 16.0 software (SPSS Inc.; Chicago, IL, USA). The two-tailed χ² test was used to determine the significance of the differences between the co-variates. For 2-group analyses, Student’s t-test was used to determine the statistical significance, whereas Pearson’s linear regression analysis was performed to examine the correlation between 2 quantitative variables. P<0.05 was considered to indicate a statistically significant difference.

To evaluate Bmi1 expression, western blot analysis was performed using 9 pairs of HCC tissues and their corresponding adjacent non-tumor tissues. Bmi1 expression was normalized to β-actin expression for the semi-quantification analyses. As shown in Fig. 1A and B, Bmi1 expression was significantly higher in all 9 of the cancer tissues than in the matched non-tumor tissues (P<0.001). Furthermore, to determine whether Bmi1 overexpression was linked to the clinical progression of HCC, 47 HCC tissues and 10 normal hepatic tissues were characterized for Bmi1 expression by immunohistochemistry. However, none of the 10 normal tissues were found to be positive for Bmi1, whereas Bmi1 expression was detected in 17 of the 47 cases of HCC (36.17%) (Fig. 1C). In addition, there was no significant correlation between Bmi1 expression and clinicopathological features (Table I). Bmi1 expression in 4 HCC cell lines was detected using western blot analysis (Fig. 1D). Bmi1 showed a high level of expression in the HepG2 and MHCC97 cell lines. Taken together, the western blot and IHC semi-quantitative analyses consistently support the notion that Bmi1 upregulation is required for hepatocellular carcinogenesis, indicating that Bmi1 may function as an oncogene in HCC.

To explore the role of Bmi1 in the development of HCC, we introduced shRNAs to suppress Bmi1 gene expression in HepG2 and MHCC97 cells (Fig. 2A); we also wished to examine how Bmi1 modulates the proliferation of HCC cells. The HepG2 and MHCC97 cells in which Bmi1 was knocked down (HepG2-shBmi1 and MHCC97-shBmi1) presented significantly lower proliferation rates than the HepG2-shControl and MHCC97-shControl cells. These results showed that the knockdown of Bmi1 expression significantly inhibited the in vitro proliferation of HCC cells (Fig. 2B). The clone formation assay showed that the Bmi1 knockdown cells formed fewer clones on the plates than the controls (P<0.05) (Fig. 2C and D), suggesting that the knockdown of Bmi1 expression significantly inhibited the in vitro proliferation of HCC cells. Recently, a number of studies have indicated the existence of CSCs in HCC, cells that are critical for the maintenance of tumor growth, progression and metastasis (22,23). It has been demonstrated that Bmi1 regulates the proliferation and differentiation of CSCs in other types of cancer in addition to HCC stem cell formation. We used a tumorsphere culture system to investigate the potential role of Bmi1 in tumorsphere formation. Fig. 2E illustrates that the knockdown of Bmi1 expression inhibited tumorsphere formation and growth, as evidenced by the significantly reduced numbers of tumorspheres compared with the numbers in the control cells (P<0.05). Our findings indicate that Bmi1 is a critical regulator of cell proliferation in HCC and tumorsphere-forming CSCs.

To determine whether Bmi1 is involved in the abnormal proliferation of HCC, we examined PCNA staining in HCC pathological specimens. The PCNA labeling index (PCNA-LI) was the cell proliferation index, and we examined the correlation between Bmi1 staining and cell proliferation activity in HCC tissues. The results showed increased cell proliferation activity in the tissues with higher Bmi1 expression (32.76±4.75 vs. 28.97±3.71%, Fig. 1A and B). Changes in cell proliferation are typically associated with cell cycle modulation, and Bmi1 has been reported to promote the proliferation of a cervical cancer cell line by accelerating the cell cycle. To investigate the mechanisms by which Bmi1 regulates HCC cell proliferation, the cell cycle was investigated using flow cytometry after Bmi1 knockdown using shRNA. As shown in Fig. 3C, the percentage of HepG2-shControl cells in the G0/G1 phase was significantly greater (71.52%) than that of HepG2-shBmi1 cells (67.34%). A similar trend was observed for the MHCC97-shControl cells (66.95%) and MHCC97-shBmi1 cells (56.63%), suggesting that the Bmi1 knockdown led to cell cycle arrest (Fig. 3D). Furthermore, there were no significant differences in cell apoptosis between the HepG2-shBmi1- and HepG2-control-transfected cells (data not shown), indicating that silencing Bmi1 inhibited tumorigenicity and was a result of the arrested cell cycle transition and not cell apoptosis.

To validate the role of Bmi1 in tumor formation in vivo further, xenograft assays were performed by injecting Bmi1-silenced HCC cells and control cells into nude mice. Although the HepG2-shControl and HepG2-shBmi1 cells induced palpable tumors by 8 weeks after the injection, the HepG2-shBmi1 cells developed smaller tumors, with an average size of 678 mm³ and an average net weight of 0.77 g, than those derived from the HepG2-shControl cells (1,492 mm³/1.24 g, Fig. 4A–C). Similar data were obtained from the MHCC97-shControl and MHCC97-shBmi1 cells. Collectively, these data demonstrate that the knockdown of Bmi1 inhibits cell proliferation and tumorigenicity in vivo.

We then examined the expression of the Bmi1 and Ki67 proteins by immunohistochemistry in all the xenograft tumor tissues formed by the HepG2-shControl, HepG2-shBmi1, MHCC97-shControl and MHCC97-shBmi1 cells (Fig. 4D). The tumor tissues formed by HepG2-shBmi1 expressed lower levels of Ki67 and Bmi1 than those formed by the HepG2-shControl cells (Fig. 4E). Similar results were obtained for the MHCC97-shControl and MHCC97-shBmi1 cells. These results demonstrate that Bmi1 promotes tumor formation and the development of HCC through accelerated cell proliferation.

Sorafenib is currently the most promising molecular targeted drug for human HCC. To investigate whether Bmi1 affects the sensitivity of HCC cells to sorafenib, we treated HepG2-shBmi1 cells and HepG2-shControl cells with sorafenib at various concentrations. In this figure, the results display the effects of sorafenib at a concentration of 5 μM. As shown in Fig. 5A and B, the number of colonies formed by the HepG2-shBmi1 cells was lower than that formed by the HepG2-shControl cells. In addition to forming the least number of colonies, the HepG2-shBmi1 cells, with or without sorafenib, all showed a significantly lower proliferation rate than the control cells throughout the experimental period, as measured by cell viability assay (MTT assay) (P<0.05; Fig. 5C). We also observed that patients (n=47) who showed negative Bmi1 expression had longer survival times, as revealed by a Kaplan-Meier analysis (P<0.05; Fig. 5D).