Study participants were recruited at the Department of Gynecology, Shanghai First Maternity and Infant Hospital, from April 2013 to December 2013. A total of 48 cervical disease patient specimens were acquired at the time of surgery or biopsy. Patients who were pregnant, had known immunosuppressive diseases or received immunosuppressive therapy, chemotherapy, radiotherapy, or other related antitumor therapies were excluded from the study. All diagnoses were confirmed by at least two certified pathologists following the International Union Against Cancer Guidelines. The overall survival (OS) rate was defined as the period from surgery to death from any cause. Disease-free survival (DFS) rate was defined as the interval from surgery to recurrence/metastasis or death. Clinicopathological characteristics of patients are summarized in Table 1. Samples were frozen and stored within 15 min of removal for quantitative real-time polymerase chain reaction (qRT-PCR) and Western blot analyses.

This study was approved by the Scientific and Ethical Committee of the Shanghai First Maternity and Infant Hospital affiliated with Tongji University. Written informed consent was obtained from each participant after they received detailed explanations of the study objectives and procedures. All experiments were performed in accordance with the ethical standards described in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

Recombinant human transforming growth factor-β (TGF-β) was purchased from R&D Systems (Minneapolis, MN, USA). T₄ DNA ligase and restriction enzymes were obtained from New England Biolabs (Beverly, MA); fibronectin (FN, ≥95.0%) was purchased from Sigma-Aldrich (St. Louis, MO, USA); DAPI was purchased from Beyotime Biotechnology (Jiangsu, China); and protein A/G-agarose beads were purchased from Thermo Scientific (Rockford, IL, USA). Oligonucleotide biosynthesis was done by Generay Biotech Co., Ltd. (Shanghai, China). All the other reagents were of analytical grade and were available from commercial sources.

Paraffin-embedded cervical tissue samples were obtained from Superbiotek Company. The treatment of tissue sections were carried out as previously described (5). Primary antibodies against CDH20 (1:100 dilution) or β-catenin (1:500 dilution) were used. All sections were slightly counterstained with hematoxylin. The IHC results were graded according to the staining intensity of immunostaining and the percentage of positive tumor cells, as previously described (17). Specifically, the intensity of staining was estimated on a scale from 0 to 3, where 0 = no staining, 1 = weak, 2 = moderate, and 3 = strong, and the extent score ranged from 0 to 3, where 0 = no positive cells, 1 = positive cells < 10%, 2 = 10–50%, 3 = positive cells >50%. The final score was calculated as the multiplication of the two scores of the CDH20 protein levels, with grading from 0 to 9, where (score = 0–1 means negative, score = 2–4 means weak, and score = 6–9 means strong.

Human cervical cancer cell lines SiHa (ATCC HTB-35), HeLa (ATCC CCL-2), C-33A (ATCC HTB-31), Ect1/E6E7 (ATCC CRL-2614) and Caski (ATCC CRM-CRL-1550), and the 293T (ATCC ACS-4500) cell line were purchased from the American Type Culture Collection (ATCC) and were authenticated by short tandem repeat (STR) tests. SiHa/C-33A, HeLa/293T, Ect1/E6E7, and Caski cells were cultured in MEM, DMEM, EMEM (MEM plus NEAA), and RPMI 1640 medium (HyClone, Logan, UT) containing 10% of fetal bovine serum (FBS, Gibco, NY, USA) and 1% penicillin/streptomycin (Gibco, NY, USA). Cells were grown in a humidified 5% CO₂ incubator at 37°C.

shRNAs (shRNA#1 and #2) against human CDH20 and a negative control (shCtrl) were synthesized by Shanghai GeneChem Co. Ltd. (Shanghai, China) and then annealed and ligated with pGCSIL-GFP. A human CDH20 knockdown stable cell line was constructed according to a previous study (18). The human CDH20 gene was cloned into pGC-FU to generate pGC-FU-CDH20; the human β-catenin gene was cloned into pGC-FU to generate pGC-FU-CTNNB (GeneChem, Shanghai, China).

Lentiviruses were produced by 293T cells according to the manufacturer's instructions. The empty vector was infected into cells and served as the control. Briefly, 1.5 × 10⁶ 293T cells were seeded onto a 6-cm plate and co-transfected with the recombinant vector (pGC-FU-CDH20 or pGC-FU-CTNNB), and vectors pHelper 1.0 and pHelper 2.0, using Lipofectamine 3000 (Invitrogen, CA, USA). After 16 h, the infection solution was removed, and fresh complete medium was added. The cell supernatants containing lentiviruses were harvested 48–72 h post-transfection. Then, the lentiviruses were used to infect cervical cancer cells with polybrene (8 μg/mL; Sigma-Aldrich). A total of 1 × 10⁵ cells in 2 mL of medium were infected with ~1 mL of lentivirus supernatant. Cells infected with empty vector were used as a control.

Nuclear and cytoplasmic extracts from cervical cancer cells were obtained by using NE-PERTM nuclear and cytoplasmic extraction reagents (Pierce) following the manufacturer's instructions. Similarly, membrane and cytosol extracts were obtained using a Membrane and Cytosol Protein Extraction Kit (Beyotime Biotechnology, Jiangsu, China). Twenty micrograms of total protein from each preparation was separated by SDS-PAGE and then immunoblotted with an anti-β-catenin antibody to detect the expression. GAPDH (Santa Cruz Biotechnology, CA, USA), histone H2A (Bioworld Technology, Nanjing, China), and ATP1A1 (Bioworld Technology, Nanjing, China) were used to normalize the total protein expression, the nuclear, and membrane fractions in the lysates, respectively.

Total RNA was extracted from cultured cells or tissues using TRIzol reagent (Invitrogen, Carlsbad, CA) and was then incubated with DNase I. A 500 ng sample of total RNA was used for cDNA synthesis with the Two-Step PrimeScript miRNA cDNA Synthesis Kit (Yeasen, Shanghai, China) and an ABI 7500 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). qRT-PCR was carried out using a SuperReal PreMix Plus (SYBR Green) Kit (Yeasen, Shanghai, China) with a Light Cycler system. The relative RNA expression level after normalization to GAPDH was calculated in accordance with the change in expression using the equation 2-ΔΔCt. The primers used for qRT-PCR analysis are listed in Table S1.

Tissues and total cell lysates were prepared in cell lysis buffer (for Western and IP) (Beyotime Biotechnology, Jiangsu, China) supplemented with a protease inhibitor cocktail (Roche, Basel, Switzerland) and phenylmethylsulfonyl fluoride (Sigma-Aldrich, St. Louis, MO, USA). The protein concentrations were quantified using a BCA protein assay kit (Beyotime Biotechnology, Jiangsu, China) following the manufacturer's instructions. Equal amounts of protein were separated by SDS-PAGE and blotted onto polyvinylidene difluoride (PVDF) membranes (Millipore, Billerica, MA, USA). The blotted membranes were incubated with the indicated doses of primary antibodies against CDH20, β-catenin, E-cadherin, N-cadherin, Vimentin, Snail, Smad2, Smad3, pSmad2, and pSmad3 (Abcam, London, UK) at 4°C overnight, and further incubated with a goat anti-rabbit IgG secondary antibody (Abcam, London, UK) for 1 h. An anti-GAPDH antibody was used as a loading control. Immunoreactive bands were detected with an enhanced chemiluminescence (ECL) Kit (MilliporeSigma, Burlington, MA, USA) and visualized with a FluorChem E imaging instrument (ProteinSimple, San Jose, CA, USA).

To assess the effects of gene knockdown or overexpression on SiHa or Caski cell proliferation, cells were seeded in a 96-well plate. After attachment, the cells were starved for 8 h and then incubated with complete medium. Cell proliferation was monitored using the Cell Titer 96 AQueous One Solution Cell Proliferation Assay (MTS) (Promega, USA). Briefly, 20 μL MTS was added to each well at 24, 48, or 72 h. Following incubation at 37°C for 3 h, the absorbance at 490 nm was measured with a MRX II microplate reader (Dynex Technologies, Chantilly, VA, USA).

Transwell migration and invasion assays were implemented using a chamber containing a polycarbonate filter with a pore size of 8.0 μm (24-well insert; Corning, NY, USA). Polycarbonate filters pre-coated with Matrigel matrix (BD Biosciences, CA, USA) were applied for the invasion assay, along with uncoated filters for the migration assay. Cells were serum-starved for 8 h, and approximately 3–5 × 10⁴ cells were resuspended in 300 μL of serum-free medium and added to the upper chamber. The lower chambers were filled with 800 mL of the indicated medium. After incubation for 16 h at 37°C and 5% CO₂, the lower chambers were incubated with calcein-AM (Thermo Fisher Scientific) to stain the cells that invaded through the Matrigel matrix or the uncoated membrane. The stained cells were counted in four randomly selected visual fields using a CCD camera mounted to an inverted microscope running MetaMorph image analysis software (Molecular Devices, Sunnyvale, CA).

Approximately 2–4 × 10⁴ cells were seeded into a 96-well plate and incubated for 6 h at 37°C. After incubation, cells were washed with PBS and fixed with absolute ethanol, followed by staining with crystal violet (0.05%; Fisher Scientific) for 30 min. Then, cells were water-washed and solubilized with 10% acetic acid. Cell adhesion was recorded by reading the absorbance at 570 nm. To estimate fibronectin cell-matrix adhesion, the 96-well plate was pre-coated with 20 μg/mL fibronectin at 4°C for overnight. The wells were blocked with 1% bovine serum albumin (Sigma-Aldrich, St. Louis, MO, USA) for 30 min before seeding cells on it.

Cell suspensions were placed onto a 35-mm glass bottom plate (Nest, China) at ~40% confluent density. After overnight incubation, the cells were washed three times with ice-cold PBS, fixed with 4% paraformaldehyde for 20 min, and then permeabilized with 0.2% Triton X-100 for another 10 min. Subsequently, the cells were blocked with 5% bovine serum albumin for up to 2 h at room temperature, and incubated with appropriate concentrations of primary antibodies against β-catenin, E-cadherin, N-cadherin, ZO-1, Flag (CDH20) (Abcam, London, UK), Smad3, and pSmad3 (1:100 dilution) overnight at 4°C. Then, the cells were incubated with Alexa Fluor® 488/594 conjugated goat anti-rabbit IgG (Abcam, London, UK, 1:200 dilution) or goat anti-mouse IgG (Abcam, London, UK, 1:200 dilution) secondary antibodies for 1 h. Cell nuclei was stained with DAPI. The stained cells were observed under a fluorescence confocal microscope.

Cells were lysed using cell lysis buffer (for Western and IP) plus 1 mM phenylmethanesulfonyl fluoride (PMSF) (Beyotime Biotechnology, Jiangsu, China). Cell lysates were incubated with the primary antibody β-catenin overnight at 4°C. Protein A/G-agarose beads were then added and incubated for 3 h at 4°C. After that, the beads were washed three times with cell lysis buffer and eluted with 5× loading buffer. The proteins were separated by SDS-PAGE and examined by Western blotting.

The graphical data are presented as the means ± SEM of three independent experiments unless otherwise stated. Statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, CA, USA) with one-way ANOVA followed by Bonferroni's post hoc test or using SPSS software (standard version 19.0; IBM) by the Pearson's χ² test. A P < 0.05 compared with the control was considered statistically significant.

Previous high-throughput sequencing results indicated that CDH20 is mutated in cervical cancer tissues and has a potential role in cervical disease progression (15). To explore the exact role of CDH20, we first analyzed the level of CDH20 mRNA in 48 paired cervical cancer and matching non-cancerous adjacent tissue samples. As shown in Figure 1A, a reduced level of CDH20 mRNA was observed in 37 (~77.1%) cervical cancer tissues. Moreover, the level of CDH20 protein was negatively correlated with cervical cancer in both nonmetastatic or lymphatic metastatic tumor samples (Figures 1B,C), suggesting that CDH20 was downregulated in cervical cancer.

Immunohistochemical (IHC) analysis of the paired 48 cases revealed that CDH20 was expressed mainly in the nonmalignant tissues, and the staining was significantly stronger than that in the cervical cancer tissues (Figures 1D–F). Compared with the non-metastatic tumor tissues, lower levels of CDH20 protein were observed in the tumors with lymphatic metastasis (Figure 1F, right panel). Moreover, a lower level of CDH20 protein was associated with poor OS and DFS rates (Figure 1G). Furthermore, lower CDH20 protein levels were observed to negatively correlate with histologic grade, FIGO stage, and lymph node metastasis (Table 1). Altogether, these data suggest that CDH20 was downregulated in cervical cancer and might promote the progression of cervical cancer by enhancing the motility of cancer cells.

Based on the above results, we speculated that loss of CDH20 function is critical for cervical cancer cell metastasis. To test this assumption, we investigated the role of CDH20 in cervical cancer cell lines. First, we detected CDH20 expression in five different cervical cell lines and found that the selected cervical cancer cell lines had lower expression of CDH20 than the cervical epithelial cells (Ect1/E6E7 cells) (Figure 2A). In addition, CDH20 was highly expressed in SiHa and HeLa cells but was expressed at low levels in Caski cells. Therefore, we chose the SiHa and HeLa cell lines and used two specific shRNAs to reduce CDH20 expression. The knockdown effects of shRNAs on CDH20 protein expression were verified by Western blotting (Figures 2B,C). Then, we performed an MTS assay to detect cervical cell proliferation. Representative results showed a slight decrease in cell number in only one CDH20 knockdown group after 72 h, suggesting that CDH20 downregulation has little effect on cervical cell proliferation (Figures 2D,E). Subsequently, the migration and invasion of cervical cells were assessed by transwell assays. It was observed that decreased CDH20 led to an obvious promotion of cell migration and invasion of 2- to 3-fold (Figures 2F–I). In addition, we determined the effect of CDH20 knockdown on SiHa and HeLa cells adhesion. Cell adhesion was first assessed on uncoated tissue culture plates, and a marked reduction in adhesion was observed (Figures 2J,L). Because fibronectin was reported to strengthen cell adhesion (19), we next examined cell adhesion using a fibronectin matrix. As shown in Figures 2K,M, CDH20 knockdown-induced reduction of cell adhesion was reversed to some extent by fibronectin coating. Taken together, these results suggest that CDH20 inhibits migration and invasion by promoting the adhesion of cervical cancer cells.

To verify the important role of CDH20 in cervical cancer cells using another method, we stably overexpressed CDH20 in the Caski cell line (Figure 3A). As shown in Figure 3B, CDH20 upregulation did not affect cervical cell proliferation within the first 48 h; however, the migration and invasion of Caski cells were significantly reduced (Figures 3C,D). Then, we used cell adhesion experiments with uncoated plates or a fibronectin matrix. The overexpression of CDH20 obviously increased cell adhesion, and fibronectin coating further enhanced this effect (Figures 3E,F). These observed results were compatible with those found in CDH20 knockdown cell lines, further supporting the inhibition effect of CDH20 on cell migration and invasion in cervical cells.

TGF-β induced EMT model in human epithelial cells is a commonly used model to dissect the molecular mechanisms of cancer metastasis (20). To examine this effect in cervical cancer, TGF-β was added to SiHa and Caski cell cultures. As shown in Figure 4A, TGF-β notably induced EMT by decreasing E-cadherin and β-catenin and increasing N-cadherin expression in both SiHa and Caski cell lines. Interestingly, the expression of CDH20 was simultaneously reduced. In light of these observations, we hypothesized that CDH20 inhibited EMT, and further examined whether altering the expression of CDH20 could affect the phenotypes of cervical cancer cell lines. The overexpression of CDH20 augmented the levels of E-cadherin and β-catenin and reduced the levels of N-cadherin and vimentin in Caski cells, implying that CDH20 can inhibit the EMT process (Figure 4B). Subsequent confocal immunofluorescence analysis also confirmed the hypothesis (Figure 4C). Moreover, western blot and confocal immunofluorescence analysis suggested that CDH20 knockdown in SiHa cells generated the opposite results (Figures 4B,D).

Cadherins typically form a complex with β-catenin at the plasma membrane as part of the adherens junctions, while dissociated β-catenin is targeted for ubiquitination and subsequent degradation by the β-catenin degradation complex (21). Intriguingly, β-catenin was observed to partially accumulate on the membrane after overexpression of CDH20 (Figure 5A), while CDH20 knockdown reversed the effect (Figure 5B). This observation prompted us to speculate whether CDH20 affects β-catenin membrane expression in cervical cancer cells. To test this hypothesis, we examined the protein expression of β-catenin in CDH20-overexpressing Caski cells. As shown in Figures 5C,E, CDH20-upregulated Caski cells obviously increased both the total cellular and cytoplasmic or cytomembrane β-catenin but decreased nuclear β-catenin, compared with control cells. In contrast, knockdown of CDH20 was able to generated the opposite results in SiHa cells (Figures 5D,F). Furthermore, the staining distribution of β-catenin in cervical cancer tissues with metastasis was partially transferred to the membrane compared with that in adjacent normal tissues (Figure 5G). Therefore, the results suggest that CDH20 affected both expression and membrane distribution of β-catenin in cervical cancer cells.

As is known, β-catenin is an indicator of Wnt/β-catenin pathway activation. Herein, the increased nuclear β-catenin in the control Caski cells (Figure 5E) suggested that CDH20 might act as a negative regulator of the canonical Wnt signaling. To test this possibility, the expression levels of Wnt/β-catenin target genes were examined. As expected, the protein levels of c-Myc and cyclin D1 were observably reduced upon overexpression of CDH20 (Figure 5H), implying that CDH20 might antagonize the canonical Wnt pathway.

To further confirm the importance of CDH20 in the presence of β-catenin, we performed rescue experiments in cervical cancer cells. The β-catenin overexpression plasmid was transfected into CDH20-downregulated SiHa cells, and cell proliferation, migration, and invasion assays were performed. The overexpression effects on β-catenin transcription and protein expression were confirmed by qRT-PCR and Western blotting (Figures 6A,B). The results of the cell proliferation assay showed that increased β-catenin did not have an evident effect on the proliferation of SiHa cells (Figure 6C). In contrast, β-catenin overexpression partially counteracted the promotion effects of CDH20 knockdown on cervical cell migration and invasion (Figures 6D,E).

We next investigated whether the CDH20-mediated synergistic effect with β-catenin involves their physical interaction. First, confocal immunofluorescence analysis was performed. CDH20 was mainly located on the cytoplasm and membrane, while β-catenin was mainly located on the nucleus and membrane. Overexpression of CDH20 enhanced the intensity of these two proteins on the membrane (Figure 6F). To further confirm the interaction between CDH20 and β-catenin in a mammalian system, we co-expressed vectors containing CDH20 and β-catenin in HEK293T cells. The co-immunoprecipitation assays revealed that CDH20 bound to β-catenin (Figure 6G). Thus, the results indicated that CDH20 interacts with β-catenin to inhibit cervical cancer cell migration and invasion.

To address how CDH20/β-catenin suppresses EMT and migration and invasion in cervical cancer cells, we investigated the causal relationship among CDH20, TGF-β, and EMT. As shown in Figure 7A, increased N-cadherin or Snail and decreased E-cadherin or β-catenin expression were observed in both CDH20-overexpressing and vector control Caski cells in response to TGF-β, suggesting that TGF-β distinctly induced EMT. Considering that Snail expression is induced by Smad 2/3-mediated phosphorylation in various cancer cells (11), we then examined the expression level of Smad 2/3. As shown, Smad2/3 maintained a consistent expression level, and pSmad2/3 was observed with the addition of TGF-β (Figure 7B). Notably, overexpression of CDH20 reduced the level of Snail (Figure 7A), which in turn might have suppressed TGF-β induced EMT by upregulating the expression of E-cadherin. Moreover, CDH20 upregulation reduced the phosphorylation of Smad2/3 to some degree and inhibited the nuclear translocation of pSmad2 (Figures 7B,C). Therefore, these findings indicate that CDH20/β-catenin suppresses EMT via TGF-β/Smad2/3/Snail signaling in cervical cancer cells.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.