Sixty-eight cases of GC and corresponding tumor-adjacent specimens were obtained from the Department of Gastrointestinal Colorectal and Anal Surgery, China-Japan Union Hospital, Jilin University. Tissue specimens were conserved in liquid nitrogen for qRT-PCR or 10% formalin for IHC until use. Informed consent from patients was obtained before the use of all samples. All the clinicopathological information of patients is presented in Table I. The Ethics Committee of Jilin University approved the present study according to the Declaration of Helsinki.

GC-derived cell lines (SGC-7901, MGC-803, MKN-28 and BGC-823) and a normal gastric epithelium cell line (GES-1) were obtained from the Cell Bank of the Shanghai Institute of Cell Biology (Chinese Academy of Medical Sciences, Shanghai, China). The cell lines were cultured in Dulbeccos modified Eagles medium (DMEM) with 10% fetal bovine serum (FBS) (Gibco, Grand Island, NY, USA) with antibiotics (Sigma-Aldrich, St. Louis, MO, USA) in a incubator containing a 5% CO₂ humidified atmosphere at 37°C.

Small hairpin RNA (shRNA) targeting SMURF1 and DAB2IP as well as non-targeting (NT) shRNA were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA). pcDNA3.1-SMURF1 and pcDNA3.1-DAB2IP were synthesized and purchased from GeneChem (Shanghai, China). All vectors were transferred into cells using Lipofectamine 2000 (Thermo Fisher Scientific, Waltham, MA, USA) on the basis of the manufacturer's recommendations.

The tissues that were previously formalin-fixed and paraffin-embedded were sliced into 4 µm sections, and underwent deparaffination and then rehydration. Antigen retrieval, suppression of endogenous peroxidase activity and 10% skim milk blocking were performed before primary antibody incubation. The SMURF1 (Abcam, Cambridge, MA, USA) and DAB2IP (Santa Cruz Biotechnology) antibodies were used as primary antibodies overnight at 4°C. The slides were subsequently incubated with a peroxidase-conjugated secondary antibody (ZSGB-BIO, Beijing, China) for 90 min, and a peroxidase-labeled polymer, DAB solution was used for signal development for 5 min. The sections were counterstained with hematoxylin followed by dehydration and mounting. Staining intensity was scored as: no staining, 0; weak staining, 1; moderate staining, 2; and strong staining, 3. Staining quantity was graded as: <25%, 1; 25–75%, 2; and >75%, 3. IHC score was manually confirmed by two independent experienced pathologists using the formula: IHC score = staining intensity × staining quantity. Sections with an IHC score >1 were considered SMURF1- and DAB2IP-positive expression.

Total RNA was drawn using TRIzol (Invitrogen, Carlsbad, CA, USA). The first strand cDNA was compounded using a TIANScript RT kit (Tiangen Biotech, Beijing, China). The expression of SMURF1 mRNA was detected using the ABI 7300 system (Applied Biosystems, Foster City, CA, USA). GAPDH was employed as the internal control. The primers used for target genes were purchased from Sangon Biotech (Shanghai, China).

For cell proliferation, GC cells that were treated with corresponding vectors were seeded into 96-well plates (1.5×10³ cells/well). After transfection at 24, 48, 72 and 96 h, the cell proliferation assay was performed by addition of 10 µl Cell Counting Kit-8 (CCK-8) solution (Beyotime, Shanghai, China) to each well, followed by incubation at 37°C for 2 h. The absorbance was assessed at a wavelength of 490 nm using a microplate reader (FlexStation III ROM V2.1.28; Molecular Devices, Sunnyvale, CA, USA).

GC cells transduced with corresponding vectors were seeded into 6-well plates to form a single confluent cell layer. The wounds were made with 100 µl tips in the confluent cell layer. After wound scratching at 0 and 24 h, the width of the wound was photographed using a phase-contrast microscope.

We determined the invasion capacities of GC cells using Transwell chambers of pore size 8-µm (Corning Costar, Cambridge, MA, USA). Twenty-four hours after transduction, 5×10⁴ cells were cultured in the 1:9 diluted Matrigel-coated (BD Biosciences, Franklin Lakes, NJ, USA) upper chamber with 250 µl of serum-free DMEM, while 700 µl DMEM with 10% FBS were added in the lower chamber. After 24 h, we fixed the cells with paraformaldehyde, and the cells in upper chamber were removed. Cells in the lower chamber were subsequently stained using 0.1% crystal violet solution and photographed.

The in vivo growth ability of GC cells was examined using the subcutaneous implantation nude mouse model. MGC-803 cells transfected with NT-shRNA or SMURF1-shRNA were subcutaneously injected into the left flank of nude mice. After 4 weeks, the subcutaneous tumors were finally resected and subjected to weight assessments. A liver metastasis assay in nude mice was performed using the subcapsular splenic injection model in which the MGC-803 cells were injected to the spleen subcapsular. Nine weeks after splenic injection, all mice were euthanized, and the livers were obtained. Furthermore, analysis of micrometastasis was assessed on the left lateral lobe of the liver, that was fixed and cut sagittally into 4 parts, paraffin-embedded, sectioned and stained for hematoxylin and eosin (H&E) (16). The protocol for these animal experiments was approved by the Ethics Review Committee of Jilin University.

Cancer cells were dissociated in RIPA lysis buffer (P0013D) and PMSF (ST506) (both from Beyotime, Haimen, China). Split products were centrifuged at 12,000 rpm, and then supernatants were gathered. A Bradford protein assay kit (P0006) (Beyotime) was used to analyze the protein concentration, and proteins were loaded onto 10% SDS-PAGE. Then proteins after separation were transferred onto polyvinylidene fluoride (PVDF) membranes (Sigma-Aldrich). Then, the PVDF membranes were obstructed with 5% skim milk (GuangMing, Shanghai, China) and incubated with SMURF1 (Abcam), DAB2IP (Santa Cruz Biotechnology), p-Akt (Ser473), Akt, c-Myc or ZEB1 (all from Cell Signaling Technology, Beverly, MA, USA) primary antibodies at 4°C overnight. Then, the specimens were incubated with a secondary antibody conjugated with HRP (Cell Signaling Technology). Signals were detected using an HRP chemiluminescent kit (Thermo Fisher Scientific) and an optional CCD camera as well as an image processing system (Bio-Rad, Hercules, CA, USA). GAPDH (G8140; US Biological, Swampscott, MA, USA) was used as a loading control.

Data were presented as the mean ± SEM and analyzed by GraphPad Prism 5 software (GraphPad Software, Inc., San Diego, CA, USA). Chi-squared test was employed to explore the association between two variables. The Student's t-test and ANOVA were carried out to analyze continuous variables. Survival analysis was performed using Kaplan-Meier's method and the log-rank test. A P-value <0.05 was considered to have statistical significance.

Sixty-eight samples of GC and corresponding tumor-adjacent specimens were detected by IHC for SMURF1 staining. Sections with an IHC score >1 were considered as positive expression of SMURF1. GC tissue samples, 70.59% (48/68) exhibited positive staining of SMURF1, while SMURF1 signal were detected in only 47.06% (32/68) samples of tumor-adjacent specimens (P=0.005; Fig. 1). Next, qRT-PCR further demonstrated that the levels of SMURF1 mRNA in GC tissues was upregulated compared with corresponding tumor-adjacent tissues (P<0.05; Fig. 2A). Moreover, the levels of SMURF1 mRNA expression in GC cell lines (SGC-7901, MGC-803, MKN-28 and BGC-823) were significantly increased when compared to the normal gastric epithelium cell line (GES-1) (P<0.05, respectively; Fig. 2B). In addition, positive expression of SMURF1 in GC patients was correlated with large-sized tumors, more lymph node metastasis and distant metastasis as well as tumor-node-metastasis (TNM) grade (P<0.05, respectively; Table I). Survival analysis revealed that GC patients with SMURF1-positive expression exhibited a prominent decreased 5-year overall and disease-free survival (P<0.05, respectively; Fig. 2C and D). Thus, SMURF1 may be a potential prognostic biomarker in GC.

Since SMURF1 was overexpressed in GC, we speculated that the biological functions of SMURF1 may participate in controlling cell proliferation and metastasis. To verify this hypothesis, SMURF1 expression was silenced by a specific shRNA in MGC-803 cells (P<0.05; Fig. 3A). CCK-8 assays were used to analyze the effect of SMURF1 silencing on proliferation of GC cells. We found that silencing of SMURF1 suppressed MGC-803 cell proliferation as compared with control cells (P<0.05; Fig. 3B). To disclose the potential role of SMURF1 in the metastasis of GC, we analyzed the migration and invasion of GC cells using wound healing and Transwell invasion assays. The results revealed that SMURF1 knockdown caused a prominent decrease in the migratory and invasive abilities of MGC-803 cells compared to the control cells (P<0.05, respectively; Fig. 3C and D). Next, SGC-7901 cells were transduced with an empty vector (EV) and pcDNA3.1-SMURF1, respectively. SMURF1 overexpression was confirmed by western blotting in SGC-7901 cells (P<0.05; Fig. 4A). Overexpression of SMURF1 notably enhanced the proliferation, migration and invasion of SGC-7901 cells (P<0.05; respectively; Fig. 4B-D). Collectively, all the results demonstrated that SMURF1 can markedly inhibit GC cell proliferation and mobility in vitro.

To further confirm the effects of SMURF1 on the growth and metastasis of GC cells, a mouse experimental growth or liver metastasis model was constructed using MGC-803 cells via subcutaneous or subcapsular splenic injection. Silencing of SMURF1 markedly decreased the subcutaneous growth of GC in nude mice as determined by the tumor weights (P<0.05; Fig. 5A). In addition, liver metastasis experiments revealed that SMURF1 knockdown notably decreased the number of metastatic nodules in the livers of nude mice (P<0.05; Fig. 5B). Altogether, our data revealed that SMURF1 prominently prohibited GC cell growth and metastasis in vivo.

Previous research revealed that SMURF1 promotes cancer cell proliferation and migration by negatively regulating DAB2IP (15). Therefore, we investigated whether DAB2IP was involved in the role of SMURF1 in GC. Notably, we found that the expression of DAB2IP was increased after SMURF1 knockdown in MGC-803 cells (Fig. 6A), while, SMURF1 overexpression led to a decreased level of DAB2IP in SGC-7901 cells (Fig. 6A). Furthermore, DAB2IP staining was performed in GC tissues using IHC. Our data revealed that the positive expression of DAB2IP was observed in 36.76% (25/68) GC tissue samples. Notably, highly-expressing SMURF1 in GC tissues exhibited weak staining of DAB2IP, while strong signals of DAB2IP were observed in tissues with low expression of SMURF1 (Fig. 6B). Next, DAB2IP was markedly overexpressed by a pcDNA3.1-mediated expression plasmid in MGC-803 cells (P<0.05; Fig. 7A). Notably, DAB2IP restoration revealed similar effects to SMURF1 knockdown in MGC-803 cells with decreased cell proliferation, migration and invasion (P<0.05, respectively; Fig. 7B-D). Therefore, these results indicate that SMURF1 plays an oncogenic role possibly by suppressing DAB2IP in GC cells.

DAB2IP has been reported to regulate various signaling pathways including the PI3K/Akt pathway in human cancer (17). Furthermore, the downstream targets of PI3K/Akt pathway such as c-Myc and ZEB1 regulate proliferation, migration and invasion in GC cells (18,19). The results from western blot analysis revealed that SMURF1 knockdown upregulated the level of DAB2IP and subsequently downregulated the expression of phosphorylated Akt as well as c-Myc and ZEB1 in MGC-803 cells (Fig. 8). However, DAB2IP-silencing abrogated the effects of SMURF1 knockdown on the activation of the PI3K/Akt pathway with increased levels of poshorylated Akt, c-Myc and ZEB1 (Fig. 8). Thus, SMURF1 probably suppresses DAB2IP and subsequently enhances the activation of the PI3K/Akt pathway in GC cells.