Fresh surgical resection specimens of HCC and adjacent normal tissues were collected from 80 patients with HCC at The Central Hospital of Enshi Autonomous Prefecture (Enshi, China) from January 2012 to January 2013. Curative resection was defined as the removal of all identifiable tumor tissue with a clear microscopic margin. None of the patients received any preoperative therapy, for example, transcatheter arterial chemoembolization, radiofrequency ablation or percutaneous ethanol injection. The patients included 63 men and 17 women, ranging in age between 33 and 76 years, with a mean age of 56 years. The clinicopathological variables, including the size of the primary tumor, vascular invasion, intrahepatic metastasis, serum α-fetoprotein (AFP) levels, liver cirrhosis, differentiation and tumor-necrosis-metastasis (TNM) stage, are listed in Table I. Follow-up data following surgery were obtained from all patients by measurement of the AFP levels and ultrasound or computed tomography at least every 3 months. All patients provided written informed consent to participate in the present study, and the study protocol was approved by the Ethics Committee of The Central Hospital of Enshi Autonomous Prefecture.

The HCC tissues and the adjacent normal tissues were fixed in 10% formalin and embedded in paraffin. Serial 4-µm sections were cut from each block. Following deparaffinization and rehydration, heat-induced antigen retrieval by autoclave pretreatment (120°C for 10 min) in citrate buffer solution (pH 6.0) was performed. Endogenous peroxidase activity was deactivated by soaking the sections in absolute methanol solution containing 3% H₂O₂ for 5 min at room temperature. Subsequently, the sections were treated with 5% bovine serum albumin (Beyotime Institute of Biotechnology, Shanghai, China) for 30 min to block non-specific reactions, and were then incubated with anti-ZEB1 antibody (cat. no. ab180905; 1:150 dilution; Abcam, Cambridge, MA, USA) or anti-VIM antibody (cat. no. sc-80975; 1:50 dilution; Santa Cruz Biotechnology, Inc., TX, USA) at 4°C overnight and washed with phosphate-buffered saline (PBS). Subsequently, the biotin-labeled secondary antibody (cat. no. ab6789; 1:200 dilution; Abcam) was added, and the sections were incubated at room temperature for 1 h. Following the addition of horseradish peroxidase (HRP)-conjugated streptavidin working solution and washing with PBS, the sections were stained with 3,3′-diaminobenzidine and counterstained with hematoxylin for 15 min, followed by successive steeping in an ethanol solution with hydrochloric acid and dilute ammonia for several minutes. Finally, the sections were dehydrated through gradient alcohol solutions and mounted in neutral balsam. The immunohistochemical staining was evaluated by two experienced pathologists using an inverted microscope. The results were assessed using a double-blind method and were repeated at least three times.

Two human HCC cell lines (Hep3B and Huh-7) and a hepatoblastoma cell line (HepG2) (12) were obtained from American Type Tissue Collection (Manassas, VA, USA). The HCCLM9 cell line was obtained from the Liver Cancer Institute, Fudan University (Shanghai, China) (13). All cells were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (Gibco; Thermo Fisher Scientific, Inc., Waltham, MA, USA) and 1% penicillin and streptomycin in a humidified atmosphere of 5% CO₂ at 37°C.

Proteins (20 µg) from the clinical HCC specimens and cell lines were extracted with lysate buffer (Beyotime Institute of Biotechnology), separated by 7.5% (for ZEB1) and 10% (for VIM) SDS-PAGE and transferred onto PVDF membranes; protein concentration was determined via a BCA kit (Beyotime Institute of Biotechnology). Subsequently, the PVDF membrane was blocked with 5% skimmed milk powder in TBST [10 mM Tris-HCl (pH 7.5), 150 mM NaCl and 0.1% Tween-20] at room temperature for 1 h. The anti-ZEB1 antibody (cat. no. sc-515797; 1:200 dilution) and the anti-VIM antibody (cat. no. sc-66001; 1:200 dilution; both Santa Cruz Biotechnology, Inc.) were added and incubated overnight at 4°C. Following washing with TBST buffer, the HRP-labeled goat secondary antibody (cat. no. ab6789; 1:2,000 dilution; Abcam) was added and the membrane was incubated at room temperature for 1 h. The proteins were visualized by autoradiography using the ECL chemiluminescence reagent (PeptBio, Wuhan, China). The relative expression of the protein of interest was represented as the grayscale ratio of the protein to GAPDH, and the results were analyzed by GraphPad Prism 5.0 software (GraphPad Software, Inc., La Jolla, CA, USA).

Total RNA was extracted utilizing TRIzol reagent (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. The RNA was reverse transcribed to cDNA using a Takara RNA PCR kit (Takara Bio, Inc, Otsu, Japan). The qPCR procedure was performed with iQ SYBR-Green Supermix (Bio-Rad Laboratories, Inc., Hercules, CA, USA). A reaction volume of 20 µl containing 1 µl of forward and reverse primers each, 2 µl cDNA, 6 µl ddH₂O and 10 µl Supermix. All reactions were run in triplicate on the iCycler IQ multi-color detection system (Bio-Rad Laboratories, Inc.). The amplification profile was denaturation at 95°C for 5 min, followed by 40 cycles of denaturation at 95°C for 20 sec, annealing at 60°C for 20 sec and extension at 72°C for 20 sec. The PCR primers used were as follows: ZEB1 forward, 5′-TGCACTGAGTGTGGAAAAGC-3′ and reverse, 5′-TGGTGATGCTGAAAGAGACG-3′; VIM forward, 5′-AAAACACCCTGCAATCTTTCAGA-3′, and reverse, 5′-GATTCCACTTTGCGTTCAAGGT-3′; β-actin forward, 5′-AGTTGCGTTACACCCTTTCTTGAC-3′, and reverse, 5′-GCTCGCTCCAACCGACTGC-3′. The comparative quantification cycle (Cq) method (14) was applied to quantify the expression levels of mRNA. The relative quantity was calculated using the equation 2^−∆Cq method where ΔCq = (Cq_{mRNA of interest}-Cq_β-actin). For the conventional PCR, the relative amount was represented as the grayscale ratio of mRNA of interest to β-actin.

ZEB1 siRNA (siRNA-ZEB1) and negative control siRNA (siRNA-NC) were purchased from Santa Cruz Biotechnology, Inc. and were transfected into Huh-7 cells during the logarithmic growth phase using Lipofectamine 2000 liposome (Invitrogen; Thermo Fisher Scientific, Inc.) following the manufacturer's protocol. The cells were incubated with the siRNA transfection complex for 12 h at 37°C, and were then harvested for mRNA and protein expression changes, which were assayed via RT-qPCR and western blot analyses, at 12 h.

Using a 96-well plate, a total of 8×10³ Huh-7 cells were seeded into each well and incubated for 18 h. WST-8 (10 µl) from the Cell Counting kit-8 (CCK-8; Boster Biological Technology, Wuhan, China) was then added to each well. Following incubation at 37°C in 5% CO₂ for 0.5, 1, 2, 3 and 4 h, the absorbance of each sample was measured at a wavelength of 450 nm.

The Huh-7 cells were seeded on a slide and incubated in culture medium with or without siRNA-ZEB1 transfection for 1 day. BrdU (10 µM; Sigma-Aldrich; Merck KgaA, Darmstadt, Germany) was added to the culture for 1 h. The cells were fixed with 4% paraformaldehyde and were then exposed to ice-cold methanol for 2 min, incubated in freshly prepared 2 M HCl for 1 h at room temperature, and then incubated in 0.1 M sodium borate for 2 min. Finally, immunocytochemistry for the BrdU-labeling of proliferative Huh-7 cells was performed, and images were captured using a Zeiss epifluorescence microscope with a charge-coupled device camera and processed using Axiovert software (v 4.9.1; both Zeiss AG, Oberkochen, Germany).

Transwell chambers precoated with Matrigel (BD Biosciences, San Jose, CA, USA) were used to perform the invasion assay. The Huh-7 cells were cultured in serum-free medium in the upper chambers of a Transwell plate (Corning Inc., Corning, NY, USA), separated from the lower chambers with permeable 8.0-µm polycarbonate membranes. Medium supplemented with 10% fetal bovine serum was added into the lower chambers as a chemoattractant. After 24 h of incubation at 37°C, those cells which had invaded through the membrane were fixed with 75% ethanol and stained with 0.1% crystal violet solution. The numbers of stained cells were manually counted under an inverted microscope in five different fields of each filter. At least three independent experiments were performed.

The Huh-7 cells were cultured in 6-well plates (seeded at a density of 1×10⁶ cells/well). A wound was created in the confluent cell monolayer using a sterile 10-µl pipette tip and incubated in serum-free medium for 60 h. The migration of cells into the wound and recovery of the monolayer were assessed and images were captured every 12 h up to 60 h using a phase contrast microscope.

The section containing the VIM promoter region was obtained by PCR as aforementioned using VIM-promoter forward and VIM-promoter reverse primers (Table II). The putative binding sites for ZEB1 were predicted using the JASPAR database (http://jaspar.genereg.net/) and the results are shown in Table III. The primers were designed to amplify these predicted sites, and the restriction sites for EcoRI and BamHI were added into these sequences (Table II). The thermocycling conditions comprised denaturation at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 30 sec, annealing at 60°C for 30 sec and extension at 72°C for 30 sec. The PCR products were digested by EcoRI and BamHI and ligated into the pEZX-PG02 vector (GeneCopoeia, Inc., Rockville, MD, USA) using T4 DNA ligase (Takara Bio, Inc.). Subsequently, E. coli DH5α cells were transformed with these products and cultured on Luria-Bertani medium containing kanamycin (all Cwbio Biotechnology, Inc., Beijing, China), and the plates were incubated at 37°C for 20 h. Individual colonies were screened and grown in the kanamycin-containing liquid medium overnight. Subsequently, the plasmids were extracted, followed by purification and further sequencing of the inserted sequences. A QuikChange Multi Site-Directed Mutagenesis kit (Stratagene; Agilent Technologies, Inc., La Jolla, CA, USA) was utilized to construct the mutant pEZX-PG02-VIM promoter plasmids prior to the sequencing analysis.

The Huh-7 cells were seeded on 48-well plates (5×10⁴ cells/ml). After 24 h, the cells were treated with Opti-MEM reduced serum medium (Invitrogen; Thermo Fisher Scientific, Inc.) in preparation for transfection. Lipofectamine 2000 reagent was used according to the manufacturer's protocol. After 48 h, the cells were harvested for luciferase detection using the GLuc Assay kit (GeneCopoeia, Inc.). All values were obtained from at least three independent repetitions of the transfection.

Continuous variables were compared using the Kruskal-Wallis test and categorical data were compared using the χ² or Fisher's exact tests. The Kaplan-Meier method was used for survival analysis, and differences in survival were estimated using the log-rank test. Multivariate analysis of prognostic factors for survival was performed using a Cox logistic regression model. All statistical analyses were performed with GraphPad Prism 5.0 software. P<0.05 was considered to indicate a statistically significant difference.

The results of the immunohistochemistry revealed that the protein expression level of the ZEB1 was higher in the tumor tissues compared with that in the adjacent normal tissues and was primarily localized in the nucleus (Fig. 1A and B). This result was confirmed by western blot analysis (Fig. 1C). According to the expression level of ZEB1, the HCC tissue specimens were divided into four groups: Absent (n=42), 1–5% (n=25), 6–10% (n=10) and >10% (n=3). The expression of ZEB1 was defined as high if at least 1% of HCC cells exhibited nuclear staining (n=38; 47.5%), or as low if there were no stained HCC cells (n=42; 52.5%). The expression of VIM was detected in the extracellular matrix and categorized by comparing its level in HCC tissues with that in adjacent normal tissues. The HCC tissues with a higher staining intensity compared with that of adjacent normal tissues were deemed to have high expression (n=33; 41.3%; Fig. 1D), whereas those with a staining intensity equal to or weaker than that in the adjacent normal tissues were deemed to have low expression (n=47; 58.7%; Fig. 1E). The RT-qPCR analysis demonstrated that the expression of ZEB1 was significantly correlated with the expression of VIM (Fig. 1F). The correlations between immunohistochemical expression and clinicopathological variables are shown in Table I. A high expression of ZEB1 was significantly associated with vascular invasion (P=0.008) and TNM stage (P=0.004). Similarly, a high expression of VIM was significantly correlated with intrahepatic metastasis (P=0.003) and TNM stage (P=0.031). The overall survival rates following surgery based on the expression of ZEB1 and VIM are shown in Fig. 1G and H. Kaplan-Meier analysis demonstrated that a high expression of ZEB1 was significantly associated with a poor overall survival rate (log rank=4.425; P=0.035) and the overall survival rate was significantly higher in the low-VIM expression group compared with that in the high-VIM expression group (log rank=4.031; P=0.045). In addition, a high expression of both ZEB1 and VIM was compared with other expression pattern combinations, and the former group exhibited a significantly poorer overall survival rate (log rank=7.352; P=0.007; Fig. 1I).

The western blot analysis revealed that ZEB1 and VIM exhibited the highest expression levels in Huh-7 cells (Fig. 2A). In addition, the RT-qPCR analysis revealed that Huh-7 cells expressed significantly higher levels of ZEB1 and VIM compared with other HCC cell lines (Fig. 2B and C). These data confirmed that the expression trend of VIM was consistent with that of ZEB1 in different liver cancer cell lines.

To assess the possible effect of ZEB1 on the expression of VIM, siRNAs were used to knockdown ZEB1 in Huh-7 cells, in which the expression of ZEB1 was found to be high, as mentioned above. The Huh-7 cells were divided into three groups: The siRNA-ZEB1 transfected group, the siRNA-NC group and the blank control group. The knockdown of ZEB1 was confirmed by conventional PCR, which revealed that, compared with that in the siRNA-NC and the blank control groups, the mRNA expression level of ZEB1 in the Huh-7 cells of the siRNA-ZEB1-transfected group was significantly decreased (P<0.05; Fig. 2D), indicating that the interference effect of siRNA on ZEB1 was effective and specific. Subsequently, the siRNA-ZEB1-transfected group was found to be associated with a significantly lower mRNA expression of VIM compared with that in the two other groups (P<0.05; Fig. 2E). Furthermore, western blot analysis demonstrated that the successful interference led to marked downregulation of the expression of ZEB1 and VIM (P<0.05; Fig. 2F), suggesting that the expression of VIM was subject to regulation by ZEB1.

To investigate the effect of the suppression of ZEB1 on cell viability, a CCK-8 assay was used over a period of 4 h, and cell viability was found to be significantly lower in the siRNA-ZEB1-transfected group compared with that in the other two groups (P<0.05; Fig. 3A and B), indicating that cell viability was suppressed by the knockdown of ZEB1. In addition, the BrdU-labeling assay revealed that DNA synthesis was inhibited in the siRNA-ZEB1-transfected Huh-7 cells (P<0.05; Fig. 3C and D). These results confirmed that cell proliferation was markedly inhibited by silencing ZEB1.

The Huh-7 cells transfected with siRNA-ZEB1 had an oval shape (Fig. 3E), suggesting that the knockdown of ZEB1 prevented the HCC cells from undergoing EMT. To further verify whether ZEB1 inhibited the invasion and migration of HCC cells, the in vitro invasiveness of Huh-7 cells was detected using the Transwell invasion chamber assay. Crystal violet staining revealed that the number of Huh-7 cells invading through the polycarbonate membrane of the Transwell invasion chamber was significantly lower in the siRNA-ZEB1-transfected group (P<0.05). No significant difference in invasiveness was found between the siRNA-NC and the blank control groups (Fig. 3F). In addition, wound-healing assays were performed to investigate the cell migration ability in the three groups, revealing that the migration of the siRNA-ZEB1-transfected cells was significantly slower compared with that in the other two groups at 48 h post-wounding (Fig. 3G). These results demonstrated that downregulation of the ZEB1 gene suppressed the motility of Huh-7 cells.

According to the results of prediction (Table III), two types of mutational pEZX-PG02-VIM promoter plasmids (pEZX-PG02-VIM-mut1 and pEZX-PG02-VIM-mut2) were constructed. To verify the activity of the pEZX-PG02-VIM promoter vector, the recombinant plasmid was transfected into Huh-7 cells and the luciferase expression was detected using a fluorescent microscope. Compared with the pEZX-PG02-NC group, the cells containing the pEZX-PG02-VIM recombinant plasmid exhibited significantly higher luciferase activity (P<0.05; Fig. 4A), indicating that the VIM promoter sequence can be bound and activated in Huh-7 cells.

To confirm that ZEB1 has the ability to regulate the expression of VIM and simultaneously identify the binding sites in the VIM promoter sequence, six categories of co-transfection were conducted: i) siRNA-ZEB1+pEZX-PG02-VIM; ii) siRNA-NC+pEZX-PG02-VIM; iii) siRNA-ZEB1+pEZX-PG02-VIM-mut1; iv) siRNA-NC+pEZX-PG02-VIM-mut1; v) siRNA-ZEB1+pEZX-PG02-VIM-mut2; vi) siRNA-NC+pEZX-PG02-VIM-mut2; subsequently, the luciferase expression was detected three times in each group (Table IV). The luciferase activity in the siRNA-ZEB1+pEZX-PG02-VIM group was significantly lower than that in the siRNA-NC+pEZX-PG02-VIM group (P<0.0001; Fig. 4B), which demonstrated that, when the Huh-7 cells were transfected with siRNA-ZEB1, the VIM promoter activity markedly decreased. Therefore, the downregulation of ZEB1 markedly reduced the transcription of the VIM gene. Compared with the siRNA-ZEB1+pEZX-PG02-VIM-mut1 group, the siRNA-NC+pEZX-PG02-VIM-mut1 group was associated with significantly higher luciferase activity (P<0.0001; Fig. 4B). More specifically, ZEB1 was able to regulate the VIM-mut1 promoter activity. However, the results revealed no statistically significant difference in luciferase activity between the siRNA-ZEB1+pEZX-PG02-VIM-mut2 and siRNA-NC+pEZX-PG02-VIM-mut2 groups (P>0.05; Fig. 4B), indicating that ZEB1 did not lead to activation of the VIM-mut2 promoter.