This project was authorized by permission of the Ethics Committee of the Affiliated Hospital of Nantong University (Nantong, China), and signed informed-consent forms were collected from all patients. Relevant clinical data on each patient were collected from his or her medical records. The ESCC tissues and paired normal adjacent mucosa tissues were derived from 187 ESCC patients who had undergone surgery at the Affiliated Hospital of Nantong University between January 2008 and December 2014. The mean age of patients at the time of surgery was 53.81 years (range, 21–88 years). The percentage of male or female patients accounted for 67.91 and 32.09% of the ESCC patients, respectively. Histological classifications and stages per the TNM Classification of Malignant Tumors (TNM stages) (https://www.uicc.org/resources/tnm-classification-malignant-tumours-8th-edition) were determined independently by two senior pathologists according to the classification criteria of the American Joint Committee on Cancer. Resected samples were snap-frozen in liquid nitrogen until RNA and protein extraction could be performed. After the enrolled ESCC patients underwent surgery, they were treated in strict accordance with therapeutic schedule on the basis of ESCC treatment guidelines. The ESCC patients who were diagnosed with pathologically confirmed lymph node metastases were able to select treatment with chemotherapy of 2–4 cycles. Chemoradiotherapy using albumin bound paclitaxel plus platinum antineoplastic agent was used in the treatment of the patients with ESCC.

We purchased the normal esophageal epithelium cell line Het-1a and ESCC cell lines (TE-10, ECA-109 and KYSE-150) from the Shanghai Institute for Biological Sciences, Chinese Academy of Sciences. Het-1a, TE-10, ECA-109 and KYSE-150 cells were maintained in RPMI-1640 or DMEM, supplemented with 10% fetal bovine serum (FBS), ampicillin (50 U/ml) and streptomycin (50 µg/ml), in a humidified chamber at 37°C with 5% CO₂. Notably, TE-10 and ECA-109 cells were transfected with plasmids encoding RASSF10, or shRNA against RASSF10, along with a vector control. The cDNA encoding full-length human RASSF10 was cloned into the PCDH vector. The shRNA targeting sequence for RASSF10 was 5′-GGAAAUUCCUUGAAUGUUAAU-3′. The expression constructs were verified by DNA sequencing. RASSF10 expression in the transfected cells was validated by western blot analysis.

Total RNA isolation and qPCR analyses of RASSF10 expression were performed as previously reported (15). The relative mRNA expression levels of RASSF10 were calculated in human tissues using the 2^−ΔΔCq method (16). The primer sequences of RASSF10 were designed as follows: forward primer (5′-CAGAGCAGGGTGAGGGTAGA-3′) and reverse primer (5′-GCTCGGACCACGTCAATTTC-3′). The primer sequences of GAPDH were designed as follows: forward primer (5′-CAGGAGGCATTGCTGATGAT-3′) and reverse primer (5′-GAAGGCTGGGGCTCATTT-3′). All experiments were repeated three times.

For the streptavidin-peroxidase (S-P) IHC studies, the primary antibodies were as follows: anti-RASSF10 (1:200; cat. no. Ab113105; Abcam), anti-E-cadherin (1:500; cat. no. ab40772; Abcam) and anti-vimentin (1:500; cat. no. ab8978; Abcam). Staining intensity and the fraction of positive cells were scored according to 10 randomly selected visual fields. The immunoreactive scores (IRS) for the proportion of positive cells and the staining grade were calculated for each specimen.

The percentages of positive cells were calculated as follows: 0 (≤5%), 1 (>5–25%), 2 (>25–50%) and 3 (>50%). Staining intensity was quantified according to four classification: 0 (negative), 1 (weak positive), 2 (moderate positive), and 3 (strong positive). According to the scoring criteria above, the expression levels of RASSF10 or E-cadherin or vimentin in the ESCC tissues were classified into two groups: samples with IRS >3 defined as high expression and the samples with IRS ≤3 defined as low expression.

Protein isolation and western blot analysis were performed as previously reported (17). The following commercial antibodies against RASSF10 (1:500; cat. no. Ab113105), E-cadherin (1:20,000; cat. no. Ab40772), vimentin (1:2,000; cat. no. Ab8978), β-catenin (1:3,000; cat. no. Ab16051), GAPDH (1:3,000; cat. no. Ab9485) and Lamin A (1:5,000; cat. no. Ab226198) followed by secondary antibody antibodies (1:5,000; cat. no. Ab6728; 1:10,000; cat. no. Ab6728) purchased from Abcam were used. All experiments were repeated three times.

Cell proliferation was monitored using the Cell Counting Kit-8 (CCK-8) based on the manufacturer's protocol (Dojindo). The transfected ESCC cell lines (TE-10 and ECA-109) were incubated at a density of 3,000 cells per well in a 96-well plate. Cell viability was determined at daily intervals (every 24 h) following the absorbance measured at 450 nm. All experiments were repeated three times.

For the colony formation assay, the transfected ESCC lines (TE-10 and ECA-109) in the logarithmic growth phase were plated into 6-well plates at a density of 300 cells per well and maintained for two weeks in medium containing G418 (1,000 µg/ml). The colonies were washed with PBS three times, and then fixed with paraformaldehyde and stained with crystal violet (0.1%) at 25°C for 10 min. The colony formation was measured by photographing the stained colonies under a microscope (magnification, ×5; Olympus Corp.). Colony formation was assessed using MShot Image Analysis System (Guangzhou Mingmei Photoelectric Technology Co., Ltd.). All experiments were repeated three times.

Immunofluorescence staining analysis was performed as previously reported using a confocal laser-scanning microscope (Carl Zeiss) (18). To assess the protein subcellular localization, the data were analyzed with Adobe Photoshop 7.0 software.

TOPflash/FOPflash reporters (Millipore) were used in this assay. The cells were transiently transfected with 1 mg of either TOPFlash or FOPFlash, and 0.1 mg pRLTK (Renilla-TK-luciferase vector, Promega Corp.) following the manufacturer's protocol. Firefly and Renilla luciferase activities were conducted employing the Dual-Luciferase Reporter Assay System (Promega Corp.) after a 48-h transfection. The activity of firefly luciferase was normalized to that of Renilla luciferase as a transfection control. All experiments were repeated three times.

Cell motility was determined by cell invasion assays using 8-µm Transwell filters (BD Biosciences) with Matrigel (BD Biosciences). The treated ECA109 and TE-10 cells at a density of 5×10⁴ were seeded into the upper chamber with medium that did not contain serum; medium containing serum was placed into the lower chambers as a chemoattractant. After incubation for 48 h, the treated cells on the upper surface of the membrane were stained with crystal violet (0.1%) at 25°C for 10 min followed by manual counting under a microscope (magnification, ×20; Olympus Corp.) in three randomly selected areas. The Transwell assays were assessed using MShot Image Analysis System (Guangzhou Mingmei Photoelectric Technology Co., Ltd.). All experiments were repeated three times.

TE-10 and ECA-109 cells were plated onto 6-well plates to encourage the formation of a monolayer of cells. A 200-µl pipette tip was employed to vertically scratch the cells in the middle of the well. The floating debris was then washed away. Fresh medium containing 0.1% FBS was supplemented and images of the cells were taken with a light microscope (magnification, ×100; Olympus Corp.) at 0 and 48 h to determine cell migration. The wound healing assays were assessed using MShot Image Analysis System (Guangzhou Mingmei Photoelectric Technology Co., Ltd.). All experiments were repeated three times.

The Wnt inhibitor IWR-1 was purchased from ENZO. Approximately 24 h before transfection, the treated ECA109 cells were plated in 6-well plates at 35–55% confluence. Cells were then transfected with the Wnt inhibitor IWR-1 at a working concentration of 10 µM, respectively, using Lipofectamine 2000 (Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. After 48–72 h, the cells were collected for subsequent experiments.

Statistical analysis was carried out using SPSS software 22.0 (SPSS, Inc.) and GraphPad Prism 5.0 software (GraphPad Software, Inc.). The Chi-squared test or Student's t-test was employed to compare the differences between two groups. Spearman rank correlations test was used to detect the relationship between expression of two genes. Other data were evaluated employing the Mann-Whitney U test as applicable to determine significant differences. Univariate and multivariate Cox regression model analyses were used to determine the prognostic relevant factors. Kaplan-Meier method was used to calculate the survival function and draw the survival curve. A P-value <0.05 was considered statistically significant.

To explore the clinical significance of RASSF10 in ESCC progression, qPCR was employed to measure the expression of the RASSF10 mRNA level extracted from the collected 20 pairs of ESCC patient tissues and matched adjacent normal tissues. Our results revealed that the RASSF10 mRNA expression level in the ESCC cancerous tissue samples was significantly downregulated compared with that in the corresponding adjacent normal tissue samples (Fig. 1A, P<0.01). TMA-IHC was further conducted to determine the RASSF10 protein status on a cohort of 187 samples of ESCC tissues and 154 peritumoral tissues. The distribution of the different levels of RASSF10 expression in ESCC and peritumoral tissues is shown in Table I. The expression levels of RASSF10 were markedly higher in the peritumoral tissue samples than that in the tumor tissues (P<0.001). Our results indicated that the RASSF10 protein was mainly localized in the cytoplasm of ESCC cells (Fig. 1B). These results demonstrated that RASSF10 is downregulated in ESCC.

Furthermore, we assessed the association between activated cytoplasmic RASSF10 expression and the clinical pathological parameters in ESCC patients. The data demonstrated that RASSF10 was markedly associated with differentiation (P=0.001) (Fig. 1B), primary tumor (P=0.014), lymph node metastasis (P=0.02) and TNM stage (P 0.006) (Table II). As shown in Fig. 1C, the immunoreaction scores of the RASSF10 status in T3-T4 stages was markedly lower in comparison with that of T1-T2 stages (P=0.0179). It was also found that ESCC patients who developed metastasis had markedly lower immunoreaction scores for RASSF10 compared with than those of non-metastatic patients (P=0.0326) (Fig. 1D).

Next, Kaplan-Meier analysis combined with IHC analysis demonstrated that low expression of RASSF10 was correlated with poor overall survival in ESCC patients (P<0.01) (Fig. 1E). As shown in Table III, multivariate Cox regression analysis confirmed that low cytoplasmic RASSF10 expression was a marked independent prognostic factor (P=0.002). Collectively, our results indicated that there is a convincing association between low RASSF10 expression and poor outcome in ESCC patients.

To explore the role of RASSF10 in the progression of ESCC, endogenous expression of RASSF10 was assessed by qPCR in normal esophageal epithelium cell line Het-1a and human ESCC cell lines TE10, ECA109 and KYSE-150. RASSF10 mRNA levels were lower in ESCC cell lines compared with that in the Het-1a cells (Fig. 2A). We then generated stable infected ECA-109 and TE-10 cells with RASSF10 overexpression, and the ECA-109 cell line was selected to silence RASSF10 expression. Western blot analysis was used to verify the transfection efficiency of the treated cell lines (Fig. 2B and C). Increased proliferation is one of most malignant phenotypes of cancer cells and plays an important part in cancer progression (19). Thus, CCK-8 and colony formation assays were used to evaluated the proliferative capabilities of the mentioned cell models. As shown in Fig. 2D and E, ectopic expression of RASSF10 dramatically suppressed the cell growth of the ECA-109 and TE-10 cells compared with the empty vector group, and the number of colonies which had formed in the RASSF10-transfected ECA-109 and TE-10 cells were significantly decreased in keeping with this finding (Fig. 2G and H), In contrast, RASSF10 knockdown resulted in markedly reverse effects (Fig. 2F and I). These results suggest that RASSF10 is crucial for inhibiting ESCC cell growth.

Since increased capability for motility is indispensable for tumor metastasis, wound healing and Transwell invasion assays were performed. Ectopic expression of RASSF10 significantly inhibited cell migration capability in the ECA-109 and TE-10 cells compared with the vector-transfected cells (Fig. 3A and B). The RASSF10-expressing cells exhibited a marked decrease in reduction in wound closure in comparison with the control groups by quantitative analyses at 48 h. By comparison, knockdown of RASSF10 significantly facilitated the migration of ECA-109 cells (Fig. 3C). Transwell invasion assays were also conducted. As expected, it was demonstrated that RASSF10 overexpression markedly impaired the invasive capability of the ECA-109 and TE-10 cells compared with that of the control groups while RASSF10 knockdown increased the invasive ability of the cells (Fig. 3D-F). Taken together, our results indicated the capability of RASSF10 to inhibit ESCC cell migration and invasion.

A growing body of studies have revealed that EMT plays a crucial part in ESCC invasion and metastasis cascade, and our above-mentioned results suggested that decreased expression of RASSF10 was correlated with tumor metastasis in ESCC patients (20–22). Therefore, we investigated the protein expression levels of EMT markers, E-cadherin (epithelial phenotype) and vimentin (mesenchymal phenotype), using the same TMA samples to further explore the role of RASSF10 in the EMT process in ESCC. IHC analysis demonstrated that tumors with lower RASSF10 expression exhibited lower E-cadherin expression levels and higher vimentin expression levels (Fig. 4A and D). According to the Spearman rank correlations test, a positive correlation was observed between RASSF10 and E-cadherin (R=0.2427, P=0.0008) (Fig. 4B), and a negative correlation was observed between RASSF10 and vimentin (R=−0.2872, P<0.0001) (Fig. 4E). ESCC patients with lower RASSF10 expression had significantly lower E-cadherin expression than those with higher RASSF10 expression (P=0.0295, Fig. 4C). We also found the ESCC patients with RASSF10 low status exhibited a higher vimentin-positive rate compared with that of the patients with high RASSF10 expression (P=0.0003, Fig. 4F).

To further explore the effect of RASSF10 on the EMT process, western blotting was employed to detect the protein levels of EMT biomarkers. The results indicated that the protein level of E-cadherin was markedly upregulated, while Snail, vimentin and N-cadherin levels were significantly downregulated in the RASSF10-overexpressing cells compared with those in the control group (Fig. 5A). In contrast, knockdown of RASSF10 decreased the expression of E-cadherin and increased the levels of Snail, vimentin and N-cadherin in the ECA-109 cells. These findings suggest that RASSF10 may be involved in suppression of the EMT process. Subsequently, our results using immunofluorescence analysis indicated that the RASSF10-overexpressing cells exhibited higher level of E-cadherin and lower level of vimentin when compared with the control groups (Fig. 5B). Knockdown of RASSF10 reversed this phenotype (Fig. 5B). Our results indicate that RASSF10 may play a crucial part in the regulation of tumor EMT and may control tumor invasion and metastasis by inhibiting the EMT process.

Previous studies suggest that RASSF10 suppresses gastric cancer cell progression by inhibiting the Wnt/β-catenin signaling pathway (23). Therefore, we hypothesized that inactivation of the Wnt/β-catenin signaling pathway may be involved in the RASSF10-mediated inhibition of ESCC cell progression. In general, the Wnt/β-catenin pathway can be mediated by stabilized accumulation and relocalization of Wnt/β-catenin. We first verified the activities of the Wnt/β-catenin pathway employing TOPflash/FOPflash luciferase assay in ECA-109 and TE-10 cells overexpressing RASSF10. As shown in Fig. 6A, we observed that relative TOPflash/FOPflash luciferase activity was decreased in the ECA-109 and TE-10 cells overexpressing RASSF10 compared with that in the control group; while knockdown of RASSF10 increased the relative TOPflash/FOPflash luciferase activity, which suggested that the Wnt/β-catenin pathway is inactivated by overexpression of RASSF10. Western blot analysis was employed to evaluate the protein levels of β-catenin to further verify the nuclear translocation of β-catenin by separation of the cytoplasmic and nuclear fractions of transfected ESCC cells. Our results indicated that RASSF10 overexpression markedly decreased β-catenin nuclear accumulation (Fig. 6B). We then treated ECA-109 RASSF10-kncokdown cells with IWR-1 inhibitor and observed that the levels of nuclear β-catenin were decreased, although RASSF10 levels remained unchanged (Fig. 6C). We also found that treatment with IWR-1 inhibited cell growth and the migration activity of the ECA109 RASSF10-knockdown cells (Fig. 6D and E). These data demonstrated that RASSF10 overexpression inactivated the Wnt/β-catenin signaling pathway in ESCC.