Human hepatoma SMMC-7721 and human embryonic kidney HEK293T cells were provided by the Institute of Cell and Biochemistry, Chinese Academy of Sciences (Shanghai, China) and cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS) (both from PAA Laboratories, Pasching, Austria) at 37°C under a mixture of 95% air and 5% CO₂.

Formalin-fixed paraffin-embedded (FFPE) tissue blocks of tumor and corresponding adjacent non-tumorous liver tissues from 39 cases of HCC patients were obtained from the Department of Pathology, and 14 cases of fresh frozen clinical specimens were collected from HCC patients who underwent hepatectomy in the Department of Surgery of The First Affiliated Hospital of Xi'an Jiaotong University. No patients received local or systemic therapies before surgery and both tumor and matched adjacent non-tumor tissue were histologically confirmed. The present study was conducted under a protocol approved by the Hospital Ethics Committee and informed consent was obtained from each patient.

Total RNA from cells and frozen tissues was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA), and from FFPE clinical samples with RecoverAll™ Total Nucleic Acid Isolation kit (Ambion, Austin, TX, USA) following the manufacturer's protocol. For detection the expression level of miR-20a-5p and RUNX3, 500 ng RNA was reverse transcribed using PrimeScript RT Master Mix (Takara, Otsu, Japan). The primers are listed in Table I. Subsequently the cDNA was amplified by SYBR Premix Ex Taq™ II (Takara). U6 snRNA and GAPDH were used as controls for miRNA and mRNA level, respectively. Relative quantitation was calculated using the 2^−ΔΔCt method. All experiments were performed in triplicates.

For immunohistochemical detection of RUNX3 expression in tumor and corresponding non-tumorous tissues from HCC patients, these tissues were cut at the thickness of 5-µm. The sections were collected and processed for RUNX3 immunohistochemistry with the avidin-biotin peroxidase complex (ABC) method following the manufacturer's protocol.

Human miR-20a precursor (pre-miR-20) was synthesized by chemical method by Shanghai Sangon Biological Engineering Technology and Services Co. Ltd. (Shanghai, China). Then, pre-miR-20 was subcloned between the EcoRI and HindIII sites of the pcDNA6.2-GW/EmGFP vector (Invitrogen). The sequences of constructed plasmids were confirmed by DNA sequencing (Sangon Biotech, Shanghai, China).

Anti-miR-20a-5p inhibitor and scrambled control fragment were synthesized by GenePharma Company (Shanghai, China). The sequence information of anti-miR-20a-5p inhibitor is presented in Table I. SMMC-7721 cells at 70–80% confluency, were transfected with pre-miR-20a-pcDNA6.2-GW (0.2–2 µg) or anti-miR-20a-5p inhibitor (16–160 pmol) using Lipofectamine 2000 (Invitrogen) method following the manufacturer's protocol. The control vector pcDNA6.2-GW (0.2–2 µg) or scrambled control RNA (16–160 pmol) were also transfected as negative controls.

SMMC-7721 cell proliferation was measured by CCK-8 assay. Briefly, cells were seeded into 96-well culture plates at a density of 6×10³ cells/well. Transfection was performed the next day at the concentration described above. The in vitro proliferation ability of SMMC-7721 cells were measured over 24 and 48 h using the CCK-8 kit (Boster, Wuhan, China) assay according to the manufacturer's instructions. Absorbance was measured at a wavelength of 450 nm in a microplate spectrophotometer.

SMMC-7721 cell migration ability was measured by a wound-healing assay. Full confluent cells were seeded into 24-well plates. A cellular area was created by scraping using a pipette tip. Wound closure was measured at 12, 24, 36 and 48 h intervals.

Cells were lysed in RIPA buffer (Beyotime Biotech, Jiangsu, China) and protein concentrations were measured by the BCA assay (Pierce Biotechnology, Inc., Rockford, IL, USA). Total protein (10 µg) was separated on a 12% SDS-PAGE and transferred onto polyvinylidene difluoride (PVDF). membranes. The PVDF membranes were subsequently immune-blotted with the appropriate primary antibodies including mouse anti-human RUNX3 monoclonal antibody (Abcam), or mouse anti-human actin monoclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA). After extensive washing, the membranes were incubated with a horseradish peroxidase-conjugated goat anti-mouse secondary antibody (Pierce Biotechnology, Inc.). Signals were detected by an ECL kit (Pierce Biotechnology, Inc.) according to the manufacturer's instructions.

The 3'UTR of RUNX3 (NM-001031680) containing the predicted miR-20a-5p binding site was constructed. The mutant RUNX3 3'UTR was created by mutating multiple nucleotides complementary to the miR-20a-5p seed region. The sequences of constructed plasmids were confirmed by DNA sequencing (Sangon Biotech).

HEK293T cells were cultured into 96-well plates with 50–70% confluency at 24 h before transfection. A mixture of 100 ng pmiR-RB-Report™ h-RUNX3 wild-type (WT) or mutant (Mut) reporter plasmid vector along with 50 nM pre-miR-20a-pcDNA6.2-GW or corresponding control vectors pcDNA6.2-GW were co-transfected. The luciferase activity was measured 48 h post-transfection using Dual-Glo Luciferase Assay System (Promega, Madison, WI, USA) with Renilla (Rluc) luciferase activity as the reporter gene and firefly luciferase (Luc) as the reference gene. Each assay was repeated in triplicate.

Statistical evaluation was performed using SPSS software (version 11.05; SPSS, Inc., Chicago, IL, USA). The statistical analysis between the two groups was conducted with Student's t-test. Correlations between miR-20a-5p clinicopathological features were analyzed using Kaplan-Meier and univariate Cox proportional hazard regression. Moreover, multiple comparisons between the groups were performed using S-N-K method. P<0.05 was considered to indicate a statistically significant result.

miR-20a is among the most frequently altered miRNAs in human cancers. To investigate the expression of miR-20a in liver cancer, we collected the information of miR-20a expression in the large cohorts of HCC patients available from The Cancer Genome Atlas (TCGA) database (Liver hepatocellular carcinoma: BCGSC IlluminaHiSeq miRNASeq). The data are presented in Fig. 1A, miR-20a is significantly increased in HCC tissues compared with normal liver tissues (P<0.001). To validate the expression of miR-20a in liver cancer, we evaluated the expression of miR-20a-5p patterns in 39 pairs of HCC and adjacent non-tumor clinical tissue specimen by qRT-PCR. As shown in Fig. 1B, an aberrant expression phenomenon of miR-20a-5p was observed in HCC tissues compared with the corresponding paired non-tumor tissues. Over 52% of the samples showed a >1.2-fold upregulation for all members of miR-20a-5p; 28% of the samples showed a >1.2-fold downregulation; and 20% of the samples showed no significant changes. These results showed that miR-20a-5p expression was aberrant in human primary HCCs. The relationships between miR-20a-5p expression and clinicopathological features were analyzed using Kaplan-Meier and univariate Cox proportional hazard regression; the data showed that miR-20a-5p expression had no association with clinicopathological information (data not shown).

There are abundant data reported that the expression of RUNX3 was poor in tumor tissues. We analyzed variation of RUNX3 in HCC using the data from TCGA database (Liver hepatocellular carcinoma: UNC IlluminaHiSeq RNASeq, UNC IlluminaHiSeq RNASeqV2). The information showed that RUNX3 is significantly decreased in HCC tissues compared with no-corresponding normal liver tissues (P=0.001) (Fig. 2A). Furthermore, the RUNX3 protein was predicted as a candidate target of miR-20a-5p using bioinformatics analysis. We assayed the correlations between miR-20a levels and RUNX3 protein. The information was from 47 pairs of corresponding data in TCGA datasets. The results showed that the expression level of miR-20a and RUNX3 mRNA presented a trend of negative correlation (Fig. 2B; r=−0.156, P=0.2941).

Next, we examined the mRNA and protein expression of RUNX3 in clinical specimens of HCC patients using qRT-PCR and IHC analysis. qRT-PCR analysis from 14 paired specimens showed that most HCC patients (12/14) exhibited significantly decreased level of RUNX3 mRNA in hepatoma tissues compared with the paired non-tumor tissues (Fig. 2C). Consistently, IHC staining showed that RUNX3 was expressed in non-tumor tissue and mainly located in the cytoplasm and nucleus, but it was very rare in hepatoma tissues (Fig. 2D).

To investigate the effect of miR-20a on cell proliferation and migration, we performed both overexpression studies using plasmid of miR-20a precursor and inhibition studies using miR-20a-5p-specific antisense oligonucleotide inhibitor (anti-miR-20a-5p). First, we transfected 2 µg of plasmid into SMMC-7721 cells. The efficiency of transfection was confirmed by qRT-PCR. At 48 h post-transfection with miR-20a precursor plasmid (pre-miR-20a-pcDNA6.2-GW), the expression of miR-20a-5p was increased ~3-fold (Fig. 3A). The data from CCK-8 assay showed an increase in proliferation occurred in cells after transfection with miR-20a precursors plasmid (Fig. 3B). The wound healing assay showed that miR-20a overexpression also enhanced cell migration ability in vitro (Fig. 3C). Accordingly, treatment with anti-miR20a-5p inhibitor caused the opposite effects. Compared with the cells which were treated with scrambled control oligonucleotides, anti-miR-20a-5p inhibitor decreased cell proliferation and migration at all-time points studied (Fig. 4).

When performing analysis on cell migration, and proliferation behavior of cells affecting migration, in the present study, we overexpressed or inhibited miR-20a in SMMC-7721 cells, and we observed the change of proliferation and migration from 12 to 48 h. The data showed an increase in proliferation occurred when cells were transfected with pre-miR-20a after 24 h, at the time of 12 h cell proliferation did not significantly change. At the same time (12 h), the migration of cells was influenced significantly by miR-20a. Therefore, we believe that cell migration detected by wound healing assay was not due to the cell proliferation at this time point. These data indicated that HCC cell growth and migration could be modulated by expression of miR-20a-5p.

Further analysis was carried out to determine whether miR-20a-5p regulated the expression of RUNX3. The result from western blotting showed overexpression of miR-20a led to decreased level of RUNX3 protein (Fig. 5A). Accordingly, transfection of the anti-miR-20a-5p inhibitor recovered the level of RUNX3 protein (Fig. 5B). However, treatment of SMMC-7721 cells with miR-20a precursor plasmid or anti-miR-20a-5p inhibitors for 48 h did not affect the level of RUNX3 mRNA as determined by qRT-PCR (Fig. 5C and D). The data suggested that miR-20a-5p reduced RUNX3 expression during protein translation.

To assess whether miR-20a-5p directly alters the expression of RUNX3, a fragment of the 3'UTR of RUNX3 mRNA containing the putative miR-20a-5p-binding sequence, cloned into the pmiRGLO dual-luciferase reporter vector (RUNX3-WT), and a mutations were generated in the RUNX3 sequences by mutating 4 nt for the seed region (RUNX3-MT), as indicated in Fig. 5E. Then, HEK293 cells were co-transfected with miR-20a and RUNX3-WT or RUNX3-MT 3'UTR vector. It was observed that in the presence of miR-20a-5p, these relative luciferase activities were significantly reduced (P=0.013), but miR-20a-5p failed to inhibit the luciferase activity of the RUNX3-MT reporter vector (Fig. 5F), which indicated that miR-20a-5p modulated gene expression directly at the 3'UTR of RUNX3. Thus, these data provide evidence that miR-20a-5p interacts with the 3'UTR of the RUNX3 transcript and regulating its translation.