Between July 2014 and June 2017, 83 patients with unifocal primary ESCC (histologically diagnosed) were recruited at the Department of Oncologic Surgery of the First Affiliated Hospital of Xi'an Jiaotong University (Xi'an, Shaanxi, China). During the operation, tumor and corresponding normal tissues (>5 cm from the edge of the tumor) were collected and immediately stored at −80°C until use. All patients underwent margin-negative complete tumor resection (R0 resection) and extensive lymphadenectomy. Patients with advanced-stage cancer received 5-FU/leucovorin-based postoperative chemotherapy and radiotherapy. Tumor stage and histological grade were recorded according to the 7th edition of the classification guidelines of the American Joint Committee on Cancer (AJCC) (19). After surgery, the patients were followed routinely for more than 36 months or until death. The follow-up was censored in February 2019. The overall survival (OS) and disease-free survival (DFS) rates were calculated with the Kaplan-Meier method. The study was approved by the Ethics Committee at the First Affiliated Hospital of Xi'an Jiaotong University. Written informed consent was obtained from all patients.

Six human ESCC cell lines (EC9706, EC109, KYSE150, KYSE450, TE1, and TE10) were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The human esophageal epithelial cell line (HET1A) was obtained from the American Type Culture Collection (ATCC). EC9706, EC109, KYSE150, KYSE450, TE1, and TE10 cells were cultured in RPMI-1640 medium (HyClone; GE Healthcare) containing 10% fetal bovine serum (FSB; HyClone; GE Healthcare) and 1% penicillin/streptomycin (Sigma-Aldrich; Merck KGaA). The HET1A cells were cultured in Dulbecco's modified Eagle's medium (DMEM; HyClone; GE Healthcare) supplemented with 10% FBS and 1% penicillin/streptomycin. All cells were grown in a humidified atmosphere with 5% CO₂ at 37°C.

In order to overexpress KIAA0101, the full-length KIAA0101 cDNA was cloned into the mammalian expression vector p-EGFP (RiboBio, China) to generate the KIAA0101-overexpressing plasmid. The empty p-EGFP plasmid was used as a control. The plasmids were transfected into the cell lines using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions. At 48 h after transfection, G418-sulfate (600 ng/ml for EC9706, and 800 ng/ml for TE1; Sigma-Aldrich; Merck KGaA) was added to the cell culture medium throughout the experimental periods, as previously described (7). In order to knock down KIAA0101, the ESCC cells were transfected with the pSIREN-Shuttle vector (RiboBio) encoding the KIAA0101-targeted shRNA (5′-GCAACCTGATCACACAAATGA-3′) using Xtreme HP (Roche), according to the manufacturer's instructions. For the transfection of small RNA molecules, ESCC cells were cultured in 6-well plates (2×10⁵ cells per well), and the miR-216a-5p mimic (30 nM), or the miR-216a-5p inhibitor (300 nM), or the negative controls (all were from RiboBio) were transfected into cells using Lipofectamine RNAiMAX (Life Technologies; Thermo Fisher Scientific, Inc.). The cells were used at 48 h after transfection.

Total RNA was extracted from the cultured cells and tissues using TRIzol (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocols. RNA concentrations were measured using a NanoDrop1000 spectrophotometer (NanoDrop Technologies). RNA integrity was assessed by gel electrophoresis. To detect the expression levels of miR-216a-5p, the miR-216a-5p-specific primers (RiboBio) were used for the reverse transcription of the total RNA. To detect the expression levels of KIAA0101, total RNA (1 µg each sample) was used to synthesize cDNA using the PrimeScript^® RT Master Mix Perfect Real-Time Reagent Kit (Takara Bio Inc.). Quantitative reverse transcription PCR (RT-qPCR) for miR-216a-5p and KIAA0101 was performed using a standard protocol based on the SYBR Green PCR kit (Toyobo) on an iQ5 system (Bio-Rad Laboratories, Inc.). The relative expression of miR-216a-5p was normalized against U6, while that of KIAA0101 was normalized to β-actin, both using the 2^−ΔΔCq method (20). When comparing the ΔCq between tumor and normal tissue, 2^−ΔCq was used. When normal cells or tissue were used as references, 2^−ΔΔCq was compared. The 2^−ΔΔCq of normal tissue was 1. Each sample was run in triplicates, and the experiment was repeated at least twice.

Total protein was extracted from cultured cells or tissues using the radioimmunoprecipitation assay lysis buffer (RIPA; Beyotime Institute of Biotechnology). Protein concentration was measured using the BCA kit (Beyotime Institute of Biotechnologya). Western blotting was performed as previously described (7). The primary antibodies included KIAA0101 (dilution 1:500, 15 kDa; cat. no. H0009768-M01; Abnova) and β-actin (dilution 1:1,500, 42 kDa; cat. no. sc-47778; Santa Cruz Biotechnology, Inc.). The secondary antibody was goat anti-mouse antibodies conjugated to horseradish peroxidase (HRP; dilution 1:1,000; cat. no. 31430; Pierce Biotechnology; Thermo Fisher Scientific, Inc.). Proteins of interest were visualized using an enhanced chemiluminescence kit (EMD Millipore). The band intensities were quantified by densitometry using ImageJ software (version 1.8.0; National Institutes of Health, Bethesda, MD, USA). Each experiment was repeated at least three times.

Immunohistochemistry was carried out as previously described (21), using the primary antibody for KIAA0101 (dilution 1:300; Abnova) and a secondary antibody (goat anti-mouse IgG; Pierce Biotechnology; Thermo Fisher Scientific, Inc.). By calculating the percentage of KIAA0101-positive cancer cells as previously described (7), all samples were classified into two groups: Negative or weak KIAA0101 expression (<15% or 16–30%) and positive or strong KIAA0101 expression (31–60% or >60%). All slides were independently assessed by at least two pathologists who were unaware of the clinicopathological features of the tumors.

The KIAA0101 3′-UTR containing the binding sites of miR-216a-5p was amplified by PCR and inserted into the firefly luciferase reporter vector pmiR-RB-REPORT (RiboBio). Following the manual of Gene Tailor Site-Directed Mutagenesis System (Invitrogen; Thermo Fisher Scientific, Inc.), the construct with mutations in the miR-216a-5p seed sequence (the sequence of TGAGATT at 357–363 bp was mutated to TCACAAT) was produced by the mutagenic oligonucleotide primers (forward: 5′-GTTGTATTTCACAATTGCTTAGATTGTTGTAC-3′ and reverse: 5′-CTAAGCAATTGTGAAATACAACATACTTGTAA-3′). EC9706 and TE1 cells were co-transfected with 30 ng of pmiR-RB-KIAA0101 3′-UTR or their mutant constructs, 50 ng luciferase reporter, and 10 ng Renilla luciferase reporter using Lipofectamine 2000. At 24 h after transfection, the firefly and Renilla luciferase activities were measured using the dual-luciferase reporter assay system (Promega Corp.). For each analysis, the Renilla luciferase signal was standardized to the firefly luciferase signal.

At 24 h before transfection, the EC9706 and TE1 cells were seeded into 96-well plates at 5,000 cells/well. At 48 h after transfection with the KIAA0101-expressing or KIAA0101-specific shRNA expressing plasmids, the cells were assessed using the Cell Counting Kit 8 reagent (CCK-8, Sigma-Aldrich; Merck KGaA) for proliferation, following the manufacturer's instructions. Cell viability was quantified by measuring absorbance at 450 nm on a microplate reader. The optical density (OD) values of the samples in the experimental groups were normalized to that of the groups with control transfection at each time point. Each experiment was repeated at least three times.

EC9706 and TE1 cells were trypsinized and incubated in 6-well plates (1,000 cells/well) in a humidified 5% CO₂ atmosphere for 10 days. The colonies were fixed with 70% ethanol for 15 min at room temperature and stained with 2% crystal violet. The positive colonies (more than 50 cells or 3 mm² per colony) were counted. Images of stained sections were obtained using an Olympus BX51 optical microscope (Olympus Corp.) equipped with a digital camera (PD71; Olympus). The experiments were repeated at least three times.

The cultured EC9706 and TE1 cells were digested with trypsin, washed with phosphate-buffered solution (PBS), and fixed with 70% ethanol overnight at 4°C. The cells were incubated with DNase-free RNase (100 µg/ml; Thermo Fisher Scientific, Inc.) for 20 min at 37°C, and stained with propidium iodide (PI) (50 µg/ml; Sigma-Aldrich; Merck KGaA). The DNA contents were determined using the FACSCalibur™ platform (BD Biosciences), following the manufacturer's instructions. The data were generated from 20,000 events (cells) using the ModFit LT software (Verity Software House, USA).

The migration and invasion assays were carried out using Transwells with 6.5-mm polycarbonate membranes and 8.0-µm pores (Corning, Inc.) with and without Matrigel (BD Biosciences). A total of 1×10⁵ EC9706 or TE1 cells were seeded in the upper chambers. In the migration and invasion experiments, we used a serum-free RPMI-1640 medium in the upper chamber, and a medium containing 10% FBS in the lower chamber to avoid the influence of cell proliferation. We determined the time of observation and counting according to the proliferation cycle and growth of the two types of cells and avoided the effects of cell proliferation as much as possible. After incubation at 37°C for 24 h, the cells in the chambers were stained with 0.1% crystal violet, observed under a light microscope (IX71; Olympus), and photographed at ×200 magnification.

Data are presented as mean ± standard error of the mean (SEM). The differences between ESCC and corresponding normal tissues were compared using the Wilcoxon t-test. The correlations between miR-216a-5p expression and the clinicopathological features of the patients with ESCC were evaluated using the Chi-square test. The independent risk factors associated with the survival of the patients with ESCC were identified by Cox proportional hazards regression analysis. The survival curves of the patients with ESCC were obtained using the Kaplan-Meier method, and the survival rates were compared using the log-rank test. The significance level was set to P<0.05. Data analysis was performed using SPSS 20.0 (IBM, Corp.) and Graph Pad Prism 5 (GraphPad Software, Inc.).

To explore the potential roles of miR-216a-5p in the pathogenesis and chemoresistance of ESCC, we first evaluated the binding site of miR-216a-5p to the 3′UTR of KIAA0101. We found that miR-216a-5p could bind the 3′UTR of KIAA0101 (Fig. 1A), which suggests that miR-216a-5p directly targets the KIAA0101 mRNA. Then, the expression levels of miR-216a-5p were examined in 29 ESCC specimens. RT-qPCR analysis demonstrated that compared with the adjacent normal tissues, the tumor tissues had significantly lower expression of miR-216a-5p (Fig. 1B). We next examined the miR-216a-5p expression in 83 human ESCC specimens to investigate the relationship between miR-216a-5p expression and the prognosis of patients with ESCC. The patients with ESCC were categorized into high- or low-expression groups using a cut-off value according to the median miR-216a-5p expression level. The miR-216a-5p expression levels in ESCC were not correlated with sex, age, pTNM stage, or preoperative CEA levels, but were correlated with histology (P=0.022), tumor depth (P=0.006), and lymph node metastasis (P=0.018; Table I). In addition, the Cox regression analysis (Table II) revealed that the lymph node metastasis (P=0.001), pTNM stage (P<0.001), and miR-216a-5p levels (P<0.001) were independent risk factors for shorter overall survival in patients with ESCC. Furthermore, patients with ESCC and lower miR-216a-5p expression displayed reduced DFS (P<0.0001; Fig. 1C) and OS (P=0.0073; Fig. 1D). Immunohistochemistry was carried out to examine the expression of KIAA0101 in ESCC specimens (Fig. 1E). Lower miR-216a-5p expression in ESCC specimens was accompanied by higher KIAA0101 protein expression (P<0.001, Fig. 1E and F). Taken together, miR-216a-5p expression was low in ESCC specimens, and its expression was inversely correlated with KIAA0101 protein expression.

We next examined the miR-216a-5p expression levels in ESCC cell lines. The six common ESCC cell lines, including EC109, TE1, TE10, KYSE150, KYSE450, and EC9706, demonstrated a marked reduction in miR-216a-5p expression (Fig. 2A) and significant increase in KIAA0101 protein expression (Fig. 2B) compared with the normal esophageal epithelial HET1A cells.

We then selected the EC9706 and TE1 cells since they maintained moderate expression levels of both miR-216a-5p and KIAA0101 molecules among the six cell lines. To ascertain whether miR-216a-5p functionally affects KIAA0101 expression, miR-216a-5p was overexpressed and knocked down by transfecting the ESCC cell lines EC9706 and TE1 with the miR-216a-5p mimics and miR-216a-5p inhibitor. The transfections of the miR-216a-5p mimic and anti-miR-216a-5p were confirmed by RT-PCR (Fig. S1A). When compared to those transfected with the control mimic (Mock) or inhibitor (Anti-NC), the miR-216a-5p-specific mimic or inhibitor (Anti-miR-216a-5p) did not significantly change the mRNA levels of KIAA0101 (Fig. 2C). Transfection of the miR-216a-5p mimics resulted in downregulation of the KIAA0101 protein levels in EC9706 and TE1 cells, and transfection of the miR-216a-5p inhibitor upregulated the KIAA0101 protein levels (Fig. 2D). These results indicated that miR-216a-5p regulated the expression of KIAA0101 in ESCC cells in a post-transcriptional manner. We then co-transfected the luciferase reporter constructs into EC9706 and TE1 cells together with the miR-216a-5p mimics or NC controls. The relative luciferase activity was significantly decreased in the cells transfected with the KIAA0101-WT-3′-UTR vector together with miR-216a-5p mimics, compared with that in the cells transfected with the plasmid of corresponding mutant (Mut) sites and NC controls (Fig. 2E). Therefore, this finding revealed that miR-216a-5p could reduce the expression of KIAA0101 protein by directly binding to the 3′UTR of KIAA0101 and possible translation suppression.

The inverse correlation between miR-216a-5p and KIAA0101 expressions drove us to investigate the possible biological functions of miR-216a-5p in ESCC cells. The CCK-8 assays (Fig. 3A), colony formation assays (Fig. 3B), and flow cytometry assays (Fig. 3C) revealed that miR-216a-5p overexpression through transfection of the miR-216a-5p mimics significantly decreased ESCC cell proliferation, suppressed colony formation, and induced cell cycle arrest at G0/G1 phase in EC9706 and TE1 cells, whereas inhibition of miR-216a-5p through transfection of the miR-216a-5p inhibitor had the opposite effects. In addition, Transwell assays were carried out on the ESCC cells to examine the roles of miR-216a-5p in cell migration and invasion. Overexpression of miR-216a-5p significantly suppressed the ESCC invasion and reduced the ESCC migration. In contrast, the migration and invasion of ESCC cells increased when endogenous miR-216a-5p was inhibited (Fig. 4A and B).

As shown in Fig. S1B, transfection with KIAA0101 alone increased the protein expression of KIAA0101 in EC9706 and TE1 cells, while transfection with the KIAA0101-targeted shRNA decreased the expression of KIAA0101. As shown in Fig. 5A, transfection of KIAA0101 reversed the decreased expression of KIAA0101 that was mediated by the transfection of the miR-216a-5p mimic. The CCK-8 assays (Fig. 5B), colony formation assays (Fig. 5C), and flow cytometry assays (Fig. 5D) also demonstrated that KIAA0101 ectopic expression significantly attenuated the anti-proliferative effects of miR-216a-5p. Meanwhile, additionally forced KIAA0101 expression in ESCC cells restored the migration and invasion inhibition induced by overexpression of miR-216a-5p (Fig. 6A and B).

These results also indicate that KIAA0101 functions as an oncogene in ESCC. Thus, we further evaluated the impact of KIAA0101 overexpression and knockdown on the migration and invasion of EC9706 and TE1 cells. Compared with the control cells without transfection, KIAA0101 overexpression significantly enhanced ESCC cell migration and invasion, while KIAA0101 knockdown remarkably inhibited ESCC cell migration and invasion (Fig. 7). Collectively, overexpression of miR-216a-5p can inhibit the in vitro proliferation, migration, and invasion of ESCC cells through targeting the oncogene KIAA0101.