A total of 33 pairs of eligible esophageal mucosa samples from patients with ESCC were collected from the First Affiliated Hospital of Soochow University, Suzhou, China between July 2011 and April 2013. Each patient provided written informed consent for their tissue samples to be used for research purposes. The present study was approved by the Ethics Committee of Soochow University and the Scientific Advisory Panel of our institute.

Human esophageal epithelial cells (HEECs) and ESCC cell lines (EC109, TE-1, EC9706 and KYSE150) were obtained from the Chinese Academy of Sciences (Shanghai, China) and cultured in RPMI-1640 medium (Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin (Invitrogen). All cells were cultured in a humidified chamber containing 5% CO₂ at 37°C.

TargetScan software (http://www.targetscan.org), miDRB software (http://mirdb.org/cgi-bin/search.cgi) and miRecords software (http://www.mirbase.org) use two ways to search for predicted miRNA targets. One is searching the name of the miRNA (enter the name of the required miRNA and view its predicted targets). Another is searching by gene target information (enter the GenBank Accession No., NCBI Gene ID or Gene Symbol and view the miRNAs which target the gene of interest.

To construct a BMI1 overexpression plasmid, the BMI1 expression construct was generated by PCR to amplify a 2293-bp fragment encoding the BMI1 cDNA (without 3′-UTR) which was obtained by reverse transcription-polymerase chain reaction (RT-PCR) using RNA from the EC109 cells. The sense primer (5′-CGCGGATCCATGAGAGGCAGAGATCGGGG-3′) contains a BamHI restriction site and the antisense primer (5′-CCGGAATTCAGGTCGAACCAGTTGGGAGA-3′) contains an EcoRI restriction site. The PCR products were purified and then cloned into pcDNA3.1(+) (Invitrogen). The construct was verified by DNA sequencing.

To construct a luciferase reporter plasmid containing the BMI1 3′-UTR fused to the 3′ end of a luciferase reporter gene, the psiCHECK-2 dual luciferase vector (Promega, Madison, WI, USA) was used. Briefly, a 388-bp fragment containing 2 predicted miR-218 target sites (position 1470–1477 and position 1751–1758) was amplified by PCR using the following primers: forward, 5′-CCGCTCGAGTGTTCATCACCCATCAGTTATT-3′ (underlined letters indicate the XhoI site) and reverse, 5′-ATAAGAATGCGGCCGCAGCAATGTATTCTCTTTAACG-3′ (underlined letters indicate the NotI site). The XhoI and NotI-digested PCR product was then cloned into a psiCHECK-2 vector digested with XhoI and NotI to generate the psiCHECK-2-BMI1-3′-UTR-wild type. To prepare mutants, 4 bases in the predicted miR-218 target sites were changed (Fig. 3D). The mutated fragment was directly synthesized (Genewiz, Beijing, China), digested with XhoI and NotI, and cloned into the psiCHECK-2 vector to generate the psiCHECK-2-BMI1-3′-UTR-mutant. Subsequently, the EC109 cells were plated in a 24-well plate and co-transfected with 50 ng of psiCHECK-2-BMI1-3′-UTR-wild type or psiCHECK-2-BMI1-3′-UTR-mutant type and with 20 nM of either miR-218 mimics (5′-UUGUGCUUGAUCUAACCAUGU-3′) or the miR negative control (miR-NC). A scrambled sequence (5′-UUCUCCGAACGUGUCACGUTT-3′) was used as the miR-NC. At 48 h post-transfection, the EC109 were collected, and the luciferase activities were measured using the Dual-Luciferase Reporter assay kit (Promega) on a TD-20/20 Luminometer (Turner Designs, Sunnyvale, CA, USA).

Each experiment was carried out in triplicate. The results were expressed as relative Renilla luciferase activities, which were obtained by normalization to Firefly luciferase activities. All the transient transfections, including transfection with anti-miR-218 (5′-ACAU GGUUAGAUCAAGCACAA-3′) and anti-miR-NC, were performed using Lipofectamine 2000 (Life Technologies, Carlsbad, CA, USA). The scrambled sequence (5′-CAGUACUUUUGUGUAGUACAA-3′) was used as the anti-miR-NC. The knockdown of BMI1 was performed by using BMI-siRNA (CAAGCAGAAAUGCAUCGAATT) (Genepharma, Shanghai, China). A scrambled sequence (5′-UUCUCCGAACGUGUCACGUTT-3′) was used as the control.

Total RNA was extracted from the ESCC tissue samples and adjacent non-tumor tissues using TRIzol reagent (Invitrogen, Oslo, Norway) according to the manufacturer’s instructions. The amount of RNA was measured on a NanoDrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). The synthesis of cDNA with reverse transcriptase (RT) was performed using a M-MLV First Strand kit (Life Technologies). The concept of a stem-loop RT primer was used to design the RT primer for mature miR-218. The primer sequences for miR-218 and U6 detection are listed in Table I. To analyze the expression of miRNA, quantitative PCR (qPCR) was performed using the Platinum SYBR-Green qPCR SuperMix-UDG (Invitrogen) and an ABI 7500 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). GAPDH mRNA and U6 small nuclear RNA (U6 snRNA) were used as endogenous controls to normalize BMI1 and miR-218 input. Relative expression was calculated using the ΔΔC_t method.

Total protein was extracted using RIPA buffer (Cell Signaling Technology, Danvers, MA, USA) containing protease inhibitor and phosphatase inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA) at 72 h post-transfection and the protein concentration was measured using the BCA Protein Assay kit (Beyotime, Haimen, China). Protein (50 µg) was separated on SDS-PAGE and transferred onto PVDF membranes. The membranes were blocked by 1% BSA for 1 h and incubated overnight at 4°C with the primary antibody (anti-BMI1, #5856S; Cell Signaling Technology; and anti-GAPDH, AP0063; Bioworld Technology, St. Louis Park, MN, USA). The membranes were washed 4 times for 15 min with TBST (Tris-buffered saline and Tween-20). The membranes were then incubated with anti-rabbit secondary antibodies (goat anti-rabbit, sc-2004; Santa Cruz Biotechnology, Santa Cruz, CA, USA) at room temperature for 2 h and washed again. Finally, the proteins were visualized using an ECL detection system (Pierce, Rockford, IL, USA). The band density was quantified using Quantity One 4.6 software (Bio-Rad Laboratories, Hercules, CA, USA).

A total of 400 EC109 cells were seeded in a dish with a diameter of 6 cm and transfected with miR-218 mimics, anti-miR-218 or the negative control(NC), using Lipofectamine 2000 (Life Technologies). The cells were collected and seeded in a 6-well plate in triplicate at 24 h post-transfection. FBS then (0.2 ml) was added to each well after 5 days. Following incubation for another 9–10 days, the plates were gently washed with PBS. Colonies were visualized by staining with 0.1% crystal violet and the colony number was counted (>50 cells).

Cells at the logarithmic phase were seeded into 96-well plates at a density of 4×10⁴ cells/well. Cell proliferation was determined by MTT assay and the 5-ethynyl-2′-deoxyuridine (EdU) (Cell Light EdU DNA imaging kit; Guangzhou RiboBio Co., Ltd., Guangzhou, China). MTT assay was used to evaluate cell proliferation using the CCK-8 kit (Beyotime, Haimen, China) to calculate the cell absorbance value. The optical density (OD) was measured daily over 4 consecutive days at a wavelength of 450 nm (OD450) to estimate the viable cell numbers.

Another method for the determination of cell proliferation was the labeling of the cells undergoing DNA replication with EdU. Cell proliferation was quantified by measurement of the uptake of EdU. At 48 h after transfection, EdU was discerned in the DNA synthesis (S phase) of the cell cycle during 2 h of incubation, and medium containing EdU and 4% paraformaldehyde was then added to fix cells at room temperature for 30 min. The fixed cells were incubated with glycine (2 mg/ml) for 5 min in a shaker, and washed with PBS for 5 min. Following permeabilzation with 0.5% Trion X-100 for 10 min, the cells were washed with PBS for 5 min, followed by the addition of 1X Apollo dyeing reaction buffer to each well for 30 min while protecting the cells from light in a shaker at room temperature. DNA was stained with 1X Hoechst 33342 for 30 min and washed with PBS 3 times. The nucleated cells incorporated with EdU were observed by under a fluorescence microscope (Olympus IX71; Olympus, Tokyo, Japan) and the proportion of EdU uptake was determined.

A wound healing assay was used to assess cell migration. The transfected cells were plated in 6-well culture plates to form cell monolayer (approximately 80% confluence). Following serum starvation for 6 h, a wound was made using a sterile 200 µl pipette tip to scrape off the cells followed by washing twice with PBS to remove the detached cells. The cells were then incubated with 1% FBS. Cell migration was monitored at 0 and 48 h after the wound was made. The distances between the migration fronts were measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA). The assessment of the migration rate by the percentage wound closure was performed as previously described (14).

For invasion assay, a Transwell chamber (Corning Costar, Cambridge, MA, USA) coated with Matrigel was used (BD Biosciences, Franklin Lakes, NJ, USA). Cells in serum-free medium (200 µl) were seeded into the top chamber of wells in a 24-Transwell plate with inserts of 8 µm pores (Corning Costar) at a density of 2.5×10⁵ cells/ml, while the bottom chamber was filled with 500 µl medium containing 20% FBS. Following incubation at 37°C in a humidified incubator with 5% CO₂ for 24 h, the cells which had invaded to the bottom chamber were fixed with 4% paraformaldehyde at room temperature for 30 min, stained with 0.1% crystal violet for 15 min and the cell number was counted in 5 randomly selected fields under an inverted microscope (Olympus CKX41; Olympus).

The miR-218-transfected cells were harvested at 48 h post-transfection and the cell density was adjusted to 1×10⁶ cells/ml. Apoptosis was induced by the addition of 200 µM hydrogen peroxide (H₂O₂) to each well for 30 min. The cells were stained using the Annexin V-FITC Apoptosis detection kit (Beyotime, Beijing, China) according to the manufacturer’s instructions. Briefly, the cells were harvested and washed twice with cold PBS, and then resuspended in 195 µl binding buffer. Annexin V-FITC (5 µl) was added to this suspension, followed by incubation for 10 min at room temperature in the dark. After washing and resuspension, the cells were centrifuged at 1000 rpm to pellet the cells. Finally, apoptosis was assessed by flow cytometry.

All the results were performed at least 3 times and the data are presented as the means ± standard deviation (SD). The data were analyzed using the SPSS 17.0 software package and GraphPad Prism 5.02 software (GraphPad Software, San Diego, CA, USA). For the cell lines, differences between 2 groups were assessed by an unpaired t-test (two-tailed). Comparisons between the clinicopathological characteristics and the mRNA expression levels in the ESCC samples were performed using non-parametric tests (Mann-Whitney U test for 2 groups). The correlation between the expression of miR-218 and BMI1 mRNA was evaluated by calculation of Spearman’s correlation co-efficient. A value of P<0.05 was considered to indicate a statistically significant difference.

In order to determine the role of miR-218 in ESCC, the miR-218 expression levels were determined in tissue samples from 33 patient with ESCC and their paired adjacent normal tissue samples. As shown in Fig. 1A, miR-218 expression was significantly decreased in the ESCC tissues when compared with the paired normal tissues. More importantly, the decreased expression of miR-218 was also associated with the clinicopathological stages of ESCC; the miR-218 expression levels were downregulated to a greater extent in the tissue samples from patients with stage III ESCC than those from patients with stageI and II ESCC (Table II). In addition, the decreased miR-218 expression correlated with the ESCC differentiation stage.

Moreover, the relative expression levels of miR-218 were measured in 5 cell lines. Compared with the HEECs, the miR-218 expression levels were markedly lower in the ESCC cell lines, including the EC109, TE-1, EC9706 and KYSE150 cells (Fig. 1B).

Since the decreased expression of miR-218 was detected in the ESCC tissues and cell lines, we hypothesized that miR-218 acts as tumor suppressor gene and regulates tumor formation and development. We then investigated whether treatment with miR-218 exerts an inhibitory effect on the growth of ESCC cells. We found that miR-218 expression in the mimics-transfected group was significantly higher than that in the miR-control-transfected group (Fig. 2A). The corresponding cell growth curves indicated that the OD450 of the miR-218 group was significantly decreased on the fourth day post-transfection in the EC109 cells (Fig. 2B). In addition, we found that the extent of EdU staining was decreased in the EC109 cells transfected with the miR-218 mimics. EdU assay revealed that the number of EC109 cells was significantly decreased in the group transfected with miR-218 mimics (Fig. 2C). Moreover, colony formation assay also validated that transfection with miR-218 mimics was effective. The restoration of miR-218 expression markedly inhibited clonogenic cell growth, since transfection with miR-218 mimics led to the inhibition of colony formation compared to the miR-control-transfected group (P<0.01; Fig. 2D). These results indicate that miR-218 has the ability to effectively suppress ESCC cell growth.

In addition, the effects of miR-218 on cell migration and invasion were detected in the EC109 cells. Our results revealed that the overexpression of miR-218 significantly suppressed cell migration and invasion in the ESCC cells (P<0.05; Fig. 2E and F).

Next, we examined the effects of miR-218 restoration on the induction of apoptosis in EC109 cells. The EC109 cells were transfected with the miR-218 mimics for 72 h. The apoptosis of the miR-218-transfected cells was significantly increased compared with that of the miR-control-transfected cells (P<0.01; Fig. 2G).

Our results thus far suggested that the ectopic expression of miR-218 exerted an effect on ESCC cell proliferation, invasion and apoptosis. The upregulation of miR-218 expression has a potentially tumor suppressive funciton for EC development. To determine whether miR-218 plays a negative regulatory role in downregulating proteins and genes, we used TargetScan, miRDB and miRecords software to predict the potential genes inolved. Based on the results of TargetScan, we found 1,067 conserved sites and 279 poorly conserved sites among biological species. Among these genes, we found that BMI1 was previously identifed in glioma stem cells by directly targeting the 3′-UTR of BMI1 (15). Therefore, we selected BMI1 as a determinant factor for BMI1 expression in ESCC cells based on the prediction by a variety of software and as potential target gene of miR-218. We first investigated the protein expression of BMI1 in ESCC cells by western blot analysis, and found that the protein expression of BMI1 was decreased by the restoration of miR-218 expression (Fig. 3A). BMI1 protein expression was also downregulated by transfection with BMI1-siRNA.

The potential binding sites of miR-218 in the 3′-UTR of BMI1 were perdicted using 2 types of bioinformatics software (TargetScan and miDRB). We found there were 2 binding sites, and one of them was a poorly conserved sequence. In order to confirm that miR-218 regulates the expression of BMI1 by directly binding to the putative binding sites in its 3′-UTR, we constructed a luciferase reporter plasmid containing the full-length wild-type BMI1 3′-UTR and mutant-type BMI1 3′-UTR that had 4 point mutations predicted to disrupt miR-218 binding in each of the seed match regions (Fig. 3B and C). Co-transfection of the EC109 cells was performed with miR-218 mimics or miR-control. The luciferase reporter activities of both wild-type BMI1 3′-UTR binding sites were significantly decreased following transfection with miR-218 mimics compared with the miR-control; however, no significant difference was observed in the 2 point mutation groups (BMI1-AB-MT) (Fig. 3D). In conclusion, our results revealed that miR-218 directly binds to the 3′-UTR of BMI1 mRNA and downregulates its expression.

We further investigated the BMI1 mRNA expression levels in 33 pairs of ESCC samples by RT-qPCR. We found that the mRNA expression of BMI1 was significantly higher in the cancer specimens compared with their paired non-tumor specimens (n=33; P<0.01; Fig. 3E). miR-218 expression and the mRNA levels of BMI1 showed a significant inverse correlation in the 33 ESCC specimens by Spearman’s correlation analysis (r=−0.417; P<0.05; Fig. 3F). These data further suggest that the downregulation of miR-218 inversely correlates with the upregulation of BMI1 in ESCC tissues.

To explore the effects of BMI1 on ESCC, we used BMI1-siRNA to knockdown BMI1 expression in the EC109 cells. The protein expression level of BMI1 was significantly reduced in the BMI1-siRNA-transfected EC109 cells compared with the siRNA-control-transfected EC109 cells (Fig. 4A). MTT assay revealed that the growth rate of EC109 cells transfected with BMI1-siRNA was significantly lower than that of the siRNA-control-transfected cells (Fig. 4B). In order to further elucidate the effects of BMI1-siRNA on cancer cell proliferation, we used EdU incorporation assay to determine the effects of BMI1 on ESCC cell proliferation. EdU assay revealed that cell number was significantly reduced in the BMI1-siRNA-transfected EC109 cells (Fig. 4C). Transwell assay revealed that BMI1 knockdown inhibited the invasion of ESCC cells (Fig. 4D). The number of apoptotic cells were significantly increased in the BMI1-siRNA-transfected EC109 cells compared with the siRNA-control-transfected cells (Fig. 4E).

Since our results suggested that miR-218 inhibited cell growth and invasion through teh downregulation of BMI1, we explored whether BMI1 is a direct functional mediator of the inhibitory effects of miR-218. The EC109 cells were transfected with BMI1 plasmid without the 3′-UTR or control plasmid, in combination with transfection with miR-218 or miR-control. Forty hours after the BMI1 plasmid or the control vector were co-transfected with the miR-218 or miR-control into the cells, western blot analysis confirmed that miR-218 downregulated BMI1 expression; however, this effect was reversed by transfection with BMI1 plasmid (Fig. 5A). It was also suggested that the BMI1 construct without the 3′-UTR was insensitive to miR-218-mediated inhibition. MTT assay revealed that the overexpression of BMI1 significantly reversed miR-218-induced cell growth inhibition (Fig. 5B). Furthermore, BMI1 partially reversed the inhibitory effects of miR-218 on EC109 cell migration and invasion (Fig. 5C and D). These results suggested that the overexpression of BMI1 reversed the effects of the aberrant expression of miR-218 on the phenotype of ESCC cells.