All The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/) data and figures were accessed, analyzed and generated using the Ovarian Serous Cystadenocarcinoma (TCGA, Provisional) database from cBio Cancer Genomics Portal (http://cbioportal.org) (20). All data included in this manuscript are in agreement with the TCGA publication guidelines. The present study utilized datasets (including Bonome Ovarian, Bittner Ovarian 2005, Badea Pancreas 2008, DErrico Gastric 2009, Gumz Renal 2007, He Thyroid 2005, Korkola Seminoma 2006 and TCGA Colorectal 2011) from the Oncomine database (http://www.oncomine.org), an online microarray database and web-based data-mining platform that comprises transcriptome data, to compare the FXYD5 mRNA expression differences between normal and tumor tissues of multiple types of human cancer. The Kaplan-Meier Plotter (http://kmplot.com/analysis/) was utilized to calculate the probability of disease progression, and the analysis included 1,648 patients with ovarian cancer with a mean follow-up period of 40 months (21). Using TFSEARCH (http://www.cbrc.jp/research/db/TFSEARCH), four putative transcriptional factor binding sites, including SMAD-binding element (SBE)1, SBE2, SBE3 and SBE4, which were located in the -1439/+4 FXYD5 promoter region, were identified.

The present study included 58 patients who were diagnosed with pathologically high-grade and stage III serous ovarian cancer and 22 patients who were diagnosed with a benign ovarian tumor or other benign uterine lesions and underwent prophylactic adnexectomy between March 2015 and October 2016 at the Obstetrics and Gynecology Hospital of Fudan University (Shanghai, China). All patients were female, and aged between 27 and 67 years (average age, 41.8 years). Two experienced and independent pathologists from the Pathology Department at the Obstetrics and Gynecology Hospital of Fudan University (Fudan, China) verified the diagnoses. Approval was obtained from the Human Research Ethics Committee of the Obstetrics and Gynecology Hospital of Fudan University for the use of all samples by using a protocol that conforms to the provisions of the Declaration of Helsinki (as revised in Seoul, 2008; reference no. [2015] 27).

A total of 50 female BALB/c nude mice (age, 4-6 weeks; weight, 13-15 g; n=6-10 mice/group) were used in the present study according to a standard protocol that was approved by the Institutional Animal Care and Use Committee of Fudan University. The mice were purchased from SLAC Laboratory Animal Co., Ltd. (SCXK-2007-004) and maintained at 22±2°C under a 12-h light/dark cycle in a pathogen-free environment. All mice were freely accessed autoclaved standard food and water. For nude mouse xenograft assays, SKOV3 cells (3×10⁶ per mouse) transfected with the FXYD5 overexpression lentivirus or control plasmid vector were suspended in 100 μl PBS and injected subcutaneously into mice on the right side of their backs. The body weight and tumor volume (V) (calculated using the formula V=length x width x thickness in mm³ to estimate the actual volume of the tumors) were monitored 3 times per week. For intraperitoneal metastasis assays, SKOV3-IP cells (7×10⁶/0.2 ml PBS per) transfected with the FXYD5 silencing lentivirus or control plasmid vector were injected intraperitoneally into each mouse. After 3-4 weeks, tumors were surgically excised.

Ovarian cancer cell lines, including SKOV3, OVCAR3, OVCA433, A2780, HEY and CAOV3, were obtained from the American Type Culture Collection. The metastatic human serous ovarian cancer cell line, SKOV3-IP, was obtained from the M.D. Anderson Cancer Center (Houston, TX, USA). The 293T cell line, which was authenticated by applying the short tandem repeat profiling method each year, was obtained from the Shanghai Cell Bank, Type Culture Collection Committee of Chinese Academy of Science. All preserved cell lines in the laboratory underwent routine cell quality examinations by HD Biosciences Co., Ltd. to achieve a high-quality cellular standard. The ovarian cancer cell lines were maintained in RPMI-1640 supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.), 100 IU/ml penicillin, and 100 mg/ml streptomycin. 293T cells were grown in DMEM (Corning Inc.) supplemented with 10% FBS (Gibco; Thermo Fisher Scientific, Inc.), 100 IU/ml penicillin, and 100 mg/ml streptomycin. All cells were cultured in an incubator at 37°C with 5% CO₂.

TGF-β (0-10 ng/ml; cat. no. AF-100-21C; PeproTech, Inc.) was used to activate the TGF-β pathway in control SKOV3 cells and in SKOV3 cells overexpressing FXYD5. Cyclohexane (CHX; cat. no. 5087390001; Sigma-Aldrich; Merck KGaA) was used to inhibit protein synthesis and MG132 (10 μM; cat. no. M8699-1MG; Sigma-Aldrich; Merck KGaA) was used to inhibit protein degradation through the proteasome pathway in control and FXYD5-overexpressing SKOV3 cells.

Ovarian cancer and normal ovarian tissues were fixed in 10% formalin at room temperature for 24 h, dehydrated in an ascending series of alcohol (70, 85, 95 and 100%) and xylene, embedded in paraffin and sliced into 3-μm sections. Subsequently, sections of paraffin-embedded tissues (3-μm thick) were deparaffinized in dimethylbenzene and rehydrated in a descending series of alcohol (100, 95, 85 and 70%) and distilled water. Endogenous peroxidase activity was quenched using 10-30% hydrogen peroxide in methanol at room temperature for 15 min. Subsequently, antigen retrieval was conducted using citric acid buffer (pH 6.0) at 99°C for 30 min followed by cooling for 20 min, and blocking in 5% BSA (cat. no. SW3015; Beijing Solarbio Science & Technology Co., Ltd.) for 1 h at room temperature. The tissue was incubated in primary anti-FXYD5 antibody (1:500; cat. no. 12166-1-AP; ProteinTech Group, Inc.), anti-β-catenin (1:400; cat. no. 8480; Cell Signaling Technology, Inc.) and vimentin (1:500; cat. no. 5741; Cell Signaling Technology, Inc.) overnight at 4°C. Incubation with the biotinylated secondary antibody (rabbit IgG; 1:1,000-3,000; cat. no. 7074; Cell Signaling Technology, Inc.) was followed by the addition of horseradish peroxidase. Counterstaining was performed using hematoxylin at room temperature for 1 min. Images were acquired using a Leica TCS SP2 confocal laser-scanning microscope (Leica Microsystems, Inc.). For quantification, overall immunostaining scores were calculated using the H-Score system (22). The same procedure was applied for tissues from the mouse in situ and metastatic tumors.

Human FXYD5 cDNA was subcloned from the SKOV3 ovarian cancer cell line into the pCDH-CMV-MCS-EF1-Puro lentiviral vector. The cloned primer sequence is shown in Table SI. Human FXYD5 shRNA and the negative control, which were expressed in the GV248 backbone, were obtained from GeneChem, Inc. Among five identified shRNAs, the two most effective were used for further experiments, and the target sequences are presented in Table SI.

Total RNA was extracted from cultured cells using the TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and RT-qPCR was performed with GAPDH as an internal control. Total RNA was then reverse transcribed into cDNA using the Prime-Script RT Reagent kit (Takara Bio, Inc.). RT was performed using 2.0 μl 5X gDNA Eraser Buffer, 1.0 μl gDNA Eraser, 1 μg Total RNA and RNase Free dH₂O, to a total volume of 10 μl. The volume was maintained at 42°C for 2 min and was then rapidly cooled to 4°C. Subsequently, the aforementioned 10.0 μl reaction solution was mixed with 1.0 μl PrimeScript RT Enzyme Mix 1, 1.0 μl RT Primer Mix, 4.0 μl 5X PrimeScript Buffer 2 and 4.0 μl RNase Free dH₂O. The reaction conditions were as follows: 37°C for 15 min, 85°C for 5 sec, followed by cooling to 4°C and storage at -20°C. qPCR was then performed using SYBR Premix Ex Taq (Takara Bio, Inc.) according to the manufacturer's instructions. The thermocycling conditions were as follows: Denaturation for 10 sec at 95°C, followed by 40 cycles of amplification at 95°C for 15 sec and 60 sec at 34°C, and a final extension step at 72°C for 1 min. qPCR was performed on the ABI Prism 7500 instrument (Applied Biosystems; Thermo Fisher Scientific, Inc.). A fluorescence-based qPCR method was performed using 2 μl cDNA, 10 μl SYBR Green, 0.6 μl PCR forward primer (10 μM), 0.6 μl PCR reverse primer (10 μM) and 6.8 μl dH₂O in a 20-μl PCR reaction volume. GAPDH was used as a reference gene, and the data were normalized using the standard comparative C_q method (23). The specific primers that were used in the present study are listed in Table SI.

Briefly, the 293T cells (cells growing to 80% confluence) were co-transfected with lentiviral vectors (PCDH or GV248) and the packaging vectors (psPAX2 and pMD2G), as described previously (24).

The SKOV3 and SKOV3-IP cells (cells growing to 30-50% confluence) were transfected with 50 nM small interfering RNA (siRNA) that targeted corresponding genes or 50 nM scrambled negative control (Shanghai GenePharma Co., Ltd.) using the HiPerFect Transfection reagent (Qiagen GmbH) according to the manufacturer's instructions. Following 48 h of transfection, the cells were collected for use in further experiments. The targeting sequences of the siRNAs and negative control are listed in Table SI.

Total RNA was extracted from FXYD5-overexpressing and control SKOV3 cells using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). Quantified total RNA was further purified using the RNeasy Micro kit (cat. no. 74004; Qiagen GmbH) and RNase-Free DNase Set (cat. no. 79254; Qiagen GmbH) and then used for Solexa/Illumina sequencing (Shanghai Biotechnology Corp.). Hg38 RefSeq (RNA sequences, GRCh38) was downloaded from the UCSC Genome Browser (http://genome.ucsc.edu). Kyoto Encyclopedia of Genes and Genomes and Gene Ontology enrichment analyses were performed on the differentially expressed genes, which were defined as genes with changes in expression of >2 or <0.5 and a false discovery rate Q value (adjusted P-value) <0.05. Data analysis was performed utilizing FunRich (version 3), an open access standalone functional enrichment and interaction network analysis tool (25).

Briefly, whole cell extracts were prepared in chilled RIPA lysis buffer (Beyotime Institute of Biotechnology) containing protease inhibitor cocktail (Roche Diagnostics), as well as phosphatase inhibitors (Sangon Biotech Co., Ltd.). The protein concentration of the supernatants was then measured using a bicinchoninic acid (BCA) assay reagent kit (Thermo Fisher Scientific, Inc.). For western blot analysis, 40 μg total cell lysate, either from ovarian cancer cell culture or from ovarian tumor tissues, was subjected to SDS-PAGE (8-12% gels) and transferred to 0.45 or 0.22-μm PVDF membranes (EMD Millipore). The membranes were blocked with 5% milk or 5% BSA at room temperature for 60 min or at 4°C overnight, followed by incubation with the indicated primary antibodies overnight at 4°C. The membranes were then incubated with the appropriate horseradish peroxidase-conjugated secondary antibodies (Cell Signaling Technology, Inc.) at room temperature for 2 h, and finally identified using an ECL Western Blotting system (Pierce; Thermo Fisher Scientific, Inc.). Densitometric analyses of the immunoblots were conducted using ImageJ software (version 1.50i; National Institutes of Health). All antibodies used in the present study are presented in Table SII.

Briefly, 4×10⁴ ovarian cancer cells were plated in the top chamber with the non-coated membrane (8-μm pore size; Corning Life Sciences) for migration assays and with Matrigel-coated membrane (8-μm pore size; BD Biosciences) for invasion assays in 200 μl serum-free DMEM. DMEM containing 20% FBS was used as a chemo-attractant in the lower chamber. Following incubation for 10-24 h at 37°C, the cells which did not traverse through the pores in the top layer were removed with a sterile cotton swab, whereas the remaining cells on the lower surface of the membrane were fixed with methanol and stained with 0.1% crystal violet (Sigma-Aldrich; Merck KGaA) at room temperature for 30 min. The number of migratory and invasive cells was counted in five randomly selected fields of each chamber under a light microscope (Nikon Corporation; magnification, ×200), and an average number of cells was then calculated.

After 7 days of growth, cell survival was evaluated by the addition of 500 cells/well to 6-well plates, fixing the cells with methanol and staining of the cells with 1% crystal violet at room temperature for 30 min. Colonies were recorded using ImageJ software (version 1.50i; National Institutes of Health).

The TGF-β levels in the conditioned media were measured using the human TGF-β1 ELISA kit (cat. no. ELH-TGFb1-1; RayBiotech, Inc.) according to the manufacturer's protocol.

SKOV3 stably transfected cell lines grown on glass culture slides were fixed with 4% paraformaldehyde for 10 min at room temperature, followed by permeabilization with 0.5% Triton X-100 for 5 min. Subsequently, the cells were blocked with 5% BSA at room temperature for 1 h, and then incubated with primary antibodies against N-cadherin, E-cadherin and vimentin (1:200 dilution; Cell Signaling Technology, Inc.) overnight at 4°C. The slides were incubated with Alexa Fluor 488-conjugated secondary antibody (1:500 dilution; Invitrogen; Thermo Fisher Scientific, Inc.) for 40 min at room temperature. The immunofluorescence of the cytoskeleton was performed by incubation with rhodamine-conjugated phalloidin (Sigma-Aldrich; Merck KGaA) at room temperature for 40 min. Following counterstaining with DAPI for 10 min at room temperature; images were captured under a confocal microscope (Leica TCS SP5 II; Leica Microsystems, Inc.). All antibodies used in the present study are presented in Table SII.

Control and FXYD5-overexpressing SKOV3 cells (4×10⁶ per group) were treated with MG132 (10 μM; cat. no. M8699-1MG; Sigma-Aldrich; Merck KGaA) at room temperature for 4 h to block proteasome activity. They were then lysed with RIPA lysis buffer (Beyotime Institute of Biotechnology) containing protease inhibitor cocktail (Roche Diagnostics) and phosphatase inhibitors (Sangon Biotech Co., Ltd.), and the lysates were centrifuged at 12,000 × g for 10 min at 4°C. The protein concentrations were measured using a bicinchoninic acid assay reagent kit (Thermo Fisher Scientific, Inc.), and equal amounts of the lysates were used for immunoprecipitation. Thereafter, the cell lysates were immunoprecipitated overnight at 4°C with an anti-TβR1 antibody, and the protein A/G beads were mixed with the immunoprecipitates, followed by incubation at 4°C for 3 h. The precipitates were collected by centrifugation at 10,000 × g at 4°C for 2 min and washed three times with washing buffer. The immunoprecipitated protein complex was separated by SDS-PAGE (8-12% gels), followed by immunoblotting overnight at 4°C with anti-ubiquitin antibody to detect polyubiquitinated TβR1 proteins.

The present study utilized co-immunoprecipitation to test the formation of the SMURF2-SMAD7-TβR1 complex. As aforedescribed, cell lysates were incubated overnight at 4°C with an anti-TβR1 antibody, the complex was conjugated to protein A/G sepharose beads, and the beads were collected and washed with lysis buffer and subjected to SDS-PAGE. Where indicated, the cell lysates were immunoprecipitated overnight at 4°C with the anti-SMURF2 and anti-SMAD7 antibodies (Santa Cruz Biotechnology, Inc.). All antibodies used in the present study are presented in Table SII.

The promoter sequence of FXYD5 was synthesized by GeneCopoeia, Inc. A series of truncated FXYD5-promoter luciferase constructs were generated according to the predicted SMAD2-SMAD3-SMAD4 complex binding sites for FXYD5 promoter in Jaspar (http://jaspar.genereg.net). In the luciferase reporter assays, 293T cells were seeded on 24-well plates at a density of 1×10⁵ cells/well and transfected with PGL3-Promoter 0, 1, 2, 3 or 4 (P0, P1, P2, P3 or P4), together with the Renilla pRL-TK plasmid (a normalizing control; Promega Corporation), using FuGENE HD Transfection Reagent (Promega Corporation), according to the manufacturer's protocol. The cells were then treated with or without 10 ng/ml TGF-β for 24 h. Sequences, such as P0, P1, P2, P3 and P4, that contained truncated promoter regions of FXYD5 were amplified and subcloned into the PGL3-Basic Vector (Promega Corporation). After 48 h, cells were harvested, and the luciferase activities were determined using the Dual-Glo^® Luciferase Assay system with a Modulus™ single tube multimode reader (Turner BioSystems; Thermo Fisher Scientific, Inc.) at 595 nm. The relative firefly luciferase activities were obtained by normalizing the firefly luciferase activity level to the Renilla luciferase activity level. Promoter and primer sequences for the construction of the specific plasmid are listed in Tables SIII and SIV.

ChIP assays were performed using the EZ-Magna ChIPTM Chromatin Immunoprecipitation kit (EMD Millipore) according to the manufacturer's protocol. Briefly, immunoprecipitation was conducted by mixing the samples with the anti-p-SMAD3 and anti-SMAD4 antibodies (Cell Signaling Technology, Inc.) (incubating overnight at 4°C). Rabbit IgG was used as a negative control. Purified ChIP DNA segments were used as templates for qPCR, which was conducted using SYBR Premix Ex Taq (Takara Bio, Inc.) and the ABI Prism 7500 instrument (Applied Biosystems; Thermo Fisher Scientific, Inc.). The RT-qPCR conditions were as follows: Pre-heating for 10 sec at 95°C, followed by 40 cycles at 95°C for 15 sec and 60 sec at 34°C, and final extension at 72°C for 1 min. qPCR was used to amplify the FXYD5 promoter regions, SBE1-4. The specific ChIP primers that were used to measure the enrichment of the putative SMAD3/4 binding sites in the FXYD5 promoter are listed in Table SI. All antibodies used in the present study are presented in Table SII.

Unless otherwise specified, data are presented as the mean ± SD of at least three independent experiments. Statistical analyses were performed using PRISM v6.0 (GraphPad Software, Inc.) and SPSS v16.0 software (SPSS, Inc.). Student's t-test was used to compare quantitative data between two groups, and one-way analysis of ANOVA with Dunnett's or Least-Significant Difference post hoc tests were used to compare the means among multiple groups. Spearman's correlation analysis was used to analyze the correlation between FXYD5 and TGFB1, and FXYD5 and TGF-β-induced transcript 1 (TGFB1I1) expression. The Kaplan-Meier method and log-rank test were used to plot the survival curves. P<0.05 was considered to indicate a statistically significant difference.

First, the present study confirmed that FXYD5 expression was upregulated in the SKOV3-IP cell line, which is an in vivo passaged variant of SKOV3 cells established by Yu et al that exhibits a more malignant phenotype, with higher cell growth and DNA synthesis rates, when compared with its parental SKOV3 cell line (26) (Fig. 1B). Notably, in a subset of patients (stage III EOC, n=58; normal, n=22) treated at the Obstetrics and Gynecology Hospital of Fudan University, who were diagnosed with a benign ovarian tumor or other benign uterine lesions and underwent prophylactic adnexectomy from March 2015 to October 2016. it was observed that 45 (77.6%) of the tumor samples scored as moderate-strong, whereas 3 (13.6%) normal samples scored as moderate-strong for FXYD5 protein expression (Fig. 1C and D).

Among the 591 patients from the TCGA database, 70 cases (12%) presented with FXYD5 alterations, including gene amplification (n=58), mRNA upregulation (n=19), and gene deletions (n=1; Fig. 1E). One mechanism for the induction of high FXYD5 expression in EOC cases could be copy number aberrations (Fig. S1A). Additionally, genomic alterations of FXYD5 were associated with a poor survival rate (Fig. 1E).

Furthermore, large-scale data analysis using the Kaplan-Meier plotter indicated that FXYD5 mRNA upregulation was associated with poor overall, relapse-free and post-progression survival in patients with EOC [hazard ratio (HR)=1.59, 95% CI, 1.26-2.0, P=7.2×10^-5, n=655; HR=1.69, 95% CI, 1.37-2.08, P=5.2×10^-7, n=614; and HR=1.58, 95% CI, 1.24-2.01, P=1.8×10^-4, n=382, respectively; Fig. 1F]. An intrinsic subtype and clinicopathological feature analysis revealed that the effect of FXYD5 mRNA expression on patient survival may be influenced by the histological subtype, clinical stage and debulking surgery (Fig. S1F; Tables SV-SVII). Notably, for patients treated with chemotherapy (single or combination treatment with platinum and Taxol; Tables SV-SVII) or other types of malignant tumors (Fig. S1G), high FXYD5 expression indicated a poor survival.

Subsequently, to examine whether FXYD5 was upregulated specifically in EOC, an analysis of additional independent datasets in the TCGA and Oncomine platforms was performed, and it was revealed that FXYD5 expression was elevated in various types of human cancer (Fig. S1B-E).

Overall, these data suggested that FXYD5 may serve oncogenic and metastasis-promoting roles in ovarian cancer, making FXYD5 an interesting target for further investigation.

To investigate the role of FXYD5 in tumor migration and metastasis, three EOC cell line models, in which FXYD5 was overexpressed or knocked down, were established for in vitro and in vivo experiments (Fig. 2A).

Notably, in the SKOV3 and CAOV3 cell lines with relatively low levels of endogenous FXYD5 expression (Fig. 1B), FXYD5 overexpression significantly promoted cell migration and invasion in Transwell assays (Fig. 2A and B). Using a complementary inverse approach, FXYD5 deletion suppressed the migratory and invasive abilities of the SKOV3-IP cells in vitro (Fig. 2A and B).

Subsequently, the present study examined the effects of FXYD5 on cancer cell dissemination in vivo. Notably, the FXYD5-overexpressing group formed more disseminated nodules (12 vs. 5 intraperitoneal nodules on average for the test and control groups) than the control group (n=14; Fig. 2C). Additionally, IHC analysis of the intraperitoneal nodules revealed that, compared with the control tumors, the FXYD5 tumors exhibited a robust ectopic FXYD5 overexpression, with lower β-catenin expression and higher vimentin expression (Fig. 2C).

Additionally, colony formation assays demonstrated that FXYD5 overexpression could significantly promote colony formation of SKOV3 cells, whereas the loss of FXYD5 expression inhibited the colony formation of SKOV3-IP cells (Fig. 2D).

Furthermore, the present study established a xenograft model of ovarian cancer to examine the effects of FXYD5 on tumor growth in vivo. The sizes of the tumors, which were dissected from each mouse, were markedly larger in the FXYD5 group than in the control group (Fig. 2E). Consistently, the average tumor weight was considerably greater in the FXYD5 group, with a 2.5-fold change compared with the control group (Fig. 2E). By day 21, the estimated tumor volume in the FXYD5 group was almost 3-fold greater than that in the control group (Fig. 2E). Conversely, FXYD5 silencing inhibited SKOV3-IP cell survival and tumor growth in vivo, as shown in Fig. 2F. Collectively, these data demonstrated that FXYD5 promoted EOC tumor growth and metastasis in vitro and in vivo.

To investigate the specific mechanisms that drive EOC metastasis, high-throughput RNA sequencing (Illumina) was performed, using the SKOV3 and SKOV3-IP cell lines with FXYD5 over-expression and FXYD5 deletion, respectively, in combination with a global transcriptome- and pathway-based analysis. Using a minimum fold-change threshold of >2 [Q (adjusted P-value) <0.05], differentially expressed genes were identified (Tables SVIII and SIX). Gene set enrichment analysis indicated that the TGF-β (P=6.12×10-4; Q=0.015) signaling pathway was the most significantly deregulated pathway (according to the enriched gene number and Q value) (Figs. 3A and S2). Therefore, it was speculated that overexpression of FXYD5 may activate the TGF-β signaling pathway.

Subsequently, it was observed that the upregulation of FXYD5 slightly elevated the protein levels of TβR1, whereas it notably increased the phosphorylation of SMAD2 and SMAD3 in the SKOV3 cell line, with no alterations observed in the protein levels of TβR2, SMAD2, SMAD3 and SMAD4 (Fig. 3C). Conversely, FXYD5 deletion notably decreased the TβR1 protein, and SMAD2 and SMAD3 phosphorylation levels in SKOV3-IP cells (Fig. 3C). Additionally, ectopic FXYD5 expression did not induced TGF-β gene transcription or protein secretion, and no alterations were observed in the transcription levels of these TGF-β signaling pathway components (Fig. 3B). These findings suggested that FXYD5 activated TGF-β signaling by elevating the TβR1 protein level.

Notably, enrichment analysis of the data from the RNA-sequencing and TCGA datasets indicated that EMT and EMT-associated processes, including integrin family cell surface interactions, extracellular matrix, intermediate filament and others, were most associated with the FXYD5 alterations (Figs. S3 and 4; Tables SVIII-X). It has been confirmed that TGF-β stimulates EMT, migration, invasion and the metastasis of ovarian cancer cells (Fig. S5A and B) (27). Therefore, it was speculated that FXYD5 may induce EMT and promote EOC metastasis by activating the TGF-β signaling pathway.

Subsequently, through western blotting and immunofluorescence assays, it was demonstrated that the overexpression of FXYD5 resulted in higher N-cadherin and vimentin expression in the SKOV3 cell line (Fig. 4A and B). Among the other EMT biomarkers, β-catenin, another epithelial marker, was downregulated. However, the expression levels of the other mesenchymal markers, including SNAG transcriptional repressor, snail family transcriptional repressor 2, matrix metalloproteinase (MMP)2 and MMP9, were all upregulated, in response to FXYD5 overexpression (Fig. 4A and B). Additionally, in the more mesenchymal-like and metastatic SKOV3-IP cell line, FXYD5 silencing reversed the expression trends of the EMT markers (Fig. 4A) and stimulated mesenchymal-epithelial transition in the cell line. Morphologically, the FXYD5-overexpressing SKOV3 cells became rounded in shape, which is another hallmark of EMT (Fig. 4B).

Additionally, it was revealed that FXYD5 overexpression or treatment of the SKOV3 cells with TGF-β promoted cell migration and invasion, and the ectopic expression of FXYD5 enhanced the effects of TGF-β on cell movements (Fig. S5A and B). Notably, treatment with GW788388, an inhibitor of TGF-β/SMADs signaling (28), reversed the effects of ectopic FXYD5 on cell migration, invasion and EMT (Fig. 4C and D). Overall, these results suggested that FXYD5 potentiated TGF-β/SMADs signaling and TGF-β-induced EMT in ovarian cancer cells.

Notably, TβR1 protein, but not mRNA alterations, were observed in response to FXYD5 alterations (Fig. 3B and C). Therefore, it was proposed that the loss of FXYD5 would predominantly degrade the TβR1 protein in a post-transcriptional manner. To confirm this hypothesis, the SKOV3 cells overexpressing FYXD5 were incubated with CHX, an inhibitor of protein biosynthesis. Compared with the control group (SKOV3-Con), and according to the curve shown in Fig. 5A, TβR1 protein was markedly degraded at a slower rate, and even increased within 6 h of the CHX treatment in the FXYD5-overexpressing group (SKOV3-FXYD5; Fig. 5A). Furthermore, treatment of these cells with MG132, a proteasome inhibitor, increased the stable TβR1 protein level (Fig. 5B). Additionally, in the SKOV3-FXYD5 cell lines, poly-ubiquitination of TβRI was reduced (Fig. 5C), indicating that TβR1 protein degradation is directed by the ubiquitin-proteasome system.

As shown in Fig. 5D, the ectopic expression of FXYD5 markedly decreased SMAD7 expression in the SKOV3 cells. Conversely, FXYD5 silencing led to a marked increase in SMAD7 expression in the SKOV3-IP cells (Fig. 5D). However, FXYD5 silencing had no effect on SMURF2 expression in the SKOV3-IP cell line (Fig. 5D). To further examine whether SMURF2 was involved in this process, an effective small interfering RNA (si-)SMURF2 construct was generated to knockdown SMURF2, and it was revealed that si-SMURF2 significantly reversed the degrading effects of FXYD5 deletion on TβR1 and the derogation of SMAD7 induced by FXYD5 overexpression (Fig. 5E and F), suggesting that SMURF2 was required for FXYD5-mediated TβR1 and SMAD7 alterations. Furthermore, IP assays demonstrated that sh-FXYD5 promoted the binding of SMURF2 and SMAD7 to TβR1, whereas FXYD5 inhibited these interactions (Fig. 5G). These results demonstrated that FXYD5 upregulation dispersed the SMAD7-SMURF2-TβR1 complex, promoted the deubiquitination and stabilization of TβR1, and activated tTGF-β/SMADs signaling. The SMAD7 protein may also be degraded by the ubiquitin-proteasome system in response to FXYD5 upregulation, which requires further investigation.

Notably, positive correlations were identified between the FXYD5 and TGFB1 and the FXYD5 and TGFB1I1 mRNA levels in the patients with ovarian cancer from the two datasets from TCGA (Figs. 6A and S3D; Table SX). Additionally, TGFB1 and TGFB1I1 were also positively co-expressed in the two TCGA datasets (Fig. S5D). As aforementioned, FXYD5 did not induce either TGF-β gene transcription or protein secretion (Fig. 3B). Therefore, it was proposed that activated TGF-β signaling could induce FXYD5 expression. SKOV3 cells were treated with various concentrations of TGF-β (0-10 ng/ml) for 24 h (Fig. S5E). As shown in Figs. S3E and 6B, the TGF-β treatment upregulated the FXYD5 mRNA and protein levels in a dose- and time-dependent manner. Inversely, treatment with GW788388, an inhibitor of TGF-β/Smad3 signaling, abolished the effects of TGF-β on FXYD5 expression (Fig. S5F-H).

To examine whether the SMAD signaling pathway is required for TGF-β-induced FXYD5 upregulation in human ovarian cancer cells, SMAD signaling transduction was blocked by the siRNA-mediated knockdown of common SMAD4. As shown in Fig. 6C, the SMAD4 siRNA significantly downregulated endogenous SMAD4 expression, and the TGF-β-induced elevation of the FXYD5 protein levels was abolished by SMAD4 silencing. To further determine which SMAD was required for TGF-β-induced FXYD5 expression, a specific siRNA was used to knockdown SMAD2 or SMAD3 in SKOV3 cells. As shown in Figs. 6C and S3I, the silencing of SMAD3 alone, but not that of SMAD2, attenuated TGF-β-induced upregulation of the FXYD5 protein levels. Hence, it was hypothesized that the SMAD3/SMAD4 complex mediates FXYD5 transcription by binding to the promoter region of the FXYD5 gene.

To identify the role of activated TGF-β signaling in regulating FXYD5 promoter transcription, a DNA fragment between -1,439 and +227 relative to the FXYD5 transcription start site (TSS) was cloned into a pGL3-Basic plasmid to yield a pGL3-TGF-β recombinant vector that could deliver the FXYD5 promoter (Fig. 6D and E). Sequence constructs P0, P1 and P2 increased the luciferase activity level in the 293T cells; however, sequence constructs P3 and P4 exhibited luciferase activity levels that were equivalent to those of the control (Fig. 6E). This result indicated that activated TGF-β signaling increased the transcriptional activity of the FXYD5 promoter by binding to the region that was located between -1439 and +4 relative to the TSS (Fig. 6D and E). Furthermore, it was observed that the luciferase activity of the P2 promoter was significantly decreased in the 293T cells in which SMAD3 and SMAD4 were silenced and which were treated with TGF-β (Fig. 6F). These results suggested that the TGF-β-activated SMAD3/4 complex may have the capacity to bind to the FXYD5 promoter and initiate FXYD5 transcription.

Furthermore, using TFSEARCH, four putative transcriptional factor binding sites, including SBE1, SBE2, SBE3 and SBE4, that were located in the -1439/+4 FXYD5 promoter region, were identified (Fig. 6G). Subsequently, various luciferase reporter constructs containing the wild-type and mutant forms of the four TF-binding sites were transfected into the 293T cells. Upon TGF-β stimulation, three constructs exhibited significantly decreased luciferase activities compared with those of the wild-type construct; the mutant construct of the SBE2-binding site was the exception (Fig. 6H). As the TGF-β-activated SMAD3/4 complex can be recruited to SBEs (29,30), the findings suggested the possibility that the TGF-β-activated SMAD3/4 complex may positively regulate the transcription of FXYD5 by binding to SBE1, SBE3 and SBE4.

Finally, the ChIP assays revealed that upon TGF-β stimulation, the SMAD-binding elements in the FXYD5 promoter, including SBE1, SBE3 and SBE4, were more enriched in the immunoprecipitates that were obtained using the corresponding antibodies (Fig. 6I).

Collectively, these results indicated that the TGF-β-activated SMAD3/SMAD4 complex was directly recruited to the FXYD5 promoter region and that the complex interacted with specific SBEs, thus promoting FXYD5 transcription.