Human GC samples and their corresponding adjacent non-tumor gastric tissues (located ~5 cm away from the tumor) were collected during surgical resection at The First Affiliated Hospital of Guangxi Medical University (Nanning, China) from 2013 to 2015. All samples were immediately frozen on dry ice and preserved in liquid nitrogen. All the patients provided written informed consent. The present study was approved by the Ethics Committee of The First Affiliated Hospital of Guangxi Medical University. Four human GC cell lines (HGC-27, AGS, SGC-7901 and MGC-803) as well as immortalized gastric epithelial GES-1 cells were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Cells were cultured in RPMI-1640 medium (GE Healthcare Life Sciences, South Logan, UT, USA) supplemented with 50 mg/ml penicillin, 100 mg/ml streptomycin, and 10% fetal bovine serum (FBS) at 37.8°C and 5.0% CO₂.

RNAs were extracted from tissue samples and cells using TRIzol (Invitrogen Corporation, Carlsbad, CA, USA). The cDNA was synthesized from 1,000 ng RNA samples using PrimeScript™ RT reagent kit (Takara Bio, Inc., Tokyo, Japan). SYBR^® Premix Ex Taq™ II (Tli RNaseH Plus) and a ROX Plus reagent kit (Takara Bio, Inc.) were used to amplify mRNA or miRNA expression with GAPDH (mRNA) or U6 (miRNA) as internal standards. The mRNA and miRNA expression was analyzed using the 2^−ΔΔCt method. All primer sequences are listed in Table I.

The Transwell insert was coated with 75 µl (1 µg/ml) Matrigel (BD Biosciences, Mountain View, CA, USA) and 5.0×10⁴ cells in 200 µl serum-free media were transferred to the upper chamber (6.5 mm; Corning, New York, NY, USA). The lower compartment was filled with 700 µl medium containing 5% FBS as a chemoattractant. After culturing for 24 h, the cells in the lower chamber were stained with Giemsa and the migrating cells were counted under a microscope from 8 randomly selected fields (magnification, ×200).

The miR-647 mimic and the null vector LV-GFP were purchased from GeneChem (Shanghai, China). Cells were infected with the lentiviral vector at 25 PFU/cell (SGC-7901) or 60 PFU/cell (MGC-803) [multiplicity of infection (MOI), 25 or 60] at a cell confluency of 35%. Six groups of cells were identified as follows. Cells in the 7901-Ctrl (or 803-Ctrl) represent SGC-7901 (or MGC-803) cells without any treatment. Cells in the 7901-NC (or 803-NC) group were SGC-7901 (or MGC-803) cells transfected with the negative control lentiviral vector LV-GFP, and cells in the 7901–647 (or 803–647) group were SGC-7901 (or MGC-803) cells transfected with the recombinant lentivirus vector miR-647 mimic. After 24 h of transfection, the cells were cultured under normal conditions. The cells were then harvested and used.

Cell proliferation was assessed using the Cell Counting Kit-8 (CCK-8; Dojindo, Kumamoto, Japan). Cells in each group were transferred into 96-well plates at a density of 1.5×10³/well and incubated at 37°C and 5% CO₂ for 24, 48, 72 or 96 h followed by the addition of 10 µl of CCK-8 reagent into each well and incubation at 37°C for 4 h. The absorbance was read at 450 nm using a microplate reader.

Flow cytometry was performed to analyze the cell cycle in 1×10⁶ trypsinized cells, fixed with 70% ethanol and stored at 4°C for 12 h. The cells were incubated in a 100 µl solution containing 200 ng/ml RNase at 37°C for 30 min. Propidium iodide (PI) (400 µl) was added. The cells were incubated at ambient temperature for 30 min in the dark. The results were analyzed using flow cytometry (BD Biosciences).

An apoptosis detection kit (BD Biosciences) was used to determine the cellular apoptosis by incubating 2×10⁶ cells with 100 ml 1X binding buffer containing 50 µl/ml Annexin V-PE and 50 µl/ml 7-amino-actinomycin D. After incubation at an ambient temperature in the dark for 30 min, an additional 400 µl 1X binding buffer was added. The results were analyzed by flow cytometry (BD Biosciences).

The ultrastructure of the cells was observed using TEM. The cells were fixed with 2.5% glutaraldehyde at 4°C overnight and post-fixed in 1% osmium tetroxide for 30 min, dehydrated in a graded series of ethanol baths, and embedded in Epon. Ultrathin sections were stained with uranyl acetate and lead citrate, and observed under a JEM-2000EX transmission electron microscope (Jeol, Tokyo, Japan).

Cell migration was measured by culturing the GC cells in 6-well plates, and a straight line scratch wound was created with a 200-µl sterile pipette tip once cells attained 100% confluency. The wound conditions were monitored at 0, 24 and 48 h microscopically and the relative motility was calculated using the following formula: Relative motility = (initial distance - a time point distance)/initial distance × 100%.

All animals were provided by the Guangxi Animal Center (Nanning, China), and the animal studies were approved by the Ethics Committee of The First Affiliated Hospital of Guangxi Medical University. In vivo tumor growth was conducted with 2×10⁶ GC cells in each group re-suspended in 100 µl of phosphate-buffered saline (PBS; Beyotime Institute of Biotechnology, Shanghai, China) and the cells were subcutaneously inoculated in the armpit region, respectively, of 5- to 6-week-old BALB/c nude mice (8 mice/group). After 3 days, the tumors increased in size to ~5.0 mm in diameter. The tumor size was measured every 3 days using a Vernier caliper and calculated by the formula: Tumor volume = a × b²/2 (a, length diameter; b, width diameter). The relative tumor volume (RTV) was estimated as follows: RTV = V_t/V₀ (V₀, initial tumor volume; V_t, current tumor volume at the time of measurement). Mice were euthanized 21 days after inoculation of the cells. The xenografts were excised and embedded in paraffin for hematoxylin and eosin staining.

To study metastasis in vivo, 8×10⁴ cells of each group were re-suspended in 100 µl PBS, and intravenously injected into the tail vein of 5- to 6-week-old BALB/c nude mice (10 mice/group). After 36 days, lungs and livers were excised and embedded in paraffin for further histopathological analysis. The number of mice with metastasis was counted.

The microarray data obtained in the present study from cells in the 803-NC and 803–647 groups were deposited in the UniGene database 219. RNA samples were analyzed by microarray expression profiling using the GeneChip Hybridization Wash and Stain kit (Affymetrix, Santa Clara, CA, USA) according to the manufacturer's instructions. All the raw data were normalized by quantile method and analyzed. Genes upregulated and downregulated >1.5-fold in the 803–647 group compared with those in the 803-NC were selected. The target genes of miR-647 were analyzed by TargetScan (http://www.targetscan.org), microRNA.org (http://www.microrna.org/microrna/home.do) and MiRanda (http://www.miranda.com/).

Cell protein lysates were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride (PVDF) membranes. After blocking with 5% non-fat milk, the membranes were incubated with the primary antibody overnight at 4°C and washed with Tris-buffered saline and Tween-20 (TBST). The membranes were incubated with a dilution of infrared-labeled secondary antibody (1:10,000) for 1 h. After washing twice with TBST, the optical density was quantified using software provided with the Odyssey scanner (LI-COR Biosciences, Lincoln, NE, USA).

Data are presented as mean ± SE as analyzed by SPSS 13.0 (SPSS, Inc., Chicago, IL, USA), and deemed to have statistical significance at P<0.05 using the Student's t-test, one-way analysis of variance (ANOVA) or χ² test.

To determine the role of miR-647 in the pathogenesis of GC, we investigated the miR-647 levels in GC tissue specimens and the corresponding non-tumor samples of 70 GC patients and 4 GC cell lines using qRT-PCR. As shown in Fig. 1A, compared with the corresponding non-tumor samples, the miR-647 level in the GC tissue specimens was significantly decreased in 43 of 70 patients (P<0.05). The histogram also demonstrated an average 0.79±0.09-fold downregulation of miR-647 in the GC tissue specimens compared with that noted in the non-tumor samples (P<0.05; Fig. 1B). To address the clinical significance of the downregulation of miR-647 in GC, the relationship between miR-647 expression and clinical pathology was determined (Table II). The GC samples were assigned a cut-off point to distinguish tumors with low-level expression of miR-647 from those with high-level expression of miR-647. Correlation analysis showed that the low-level expression of miR-647 in GC was significantly associated with tumor size, tumor-node-metastasis (TNM) staging and metastasis including lymph node metastasis and distant metastasis that characterize more aggressive tumor phenotypes (P<0.05; Table II).

Since miR-647 was downregulated in GC and associated with tumor growth and metastasis, we speculated that it may play a similar role in GC cells. The expression of miR-647 was also investigated. Consistent with the results in GC patients, the expression of miR-647 in the HGC-27, AGS, SGC-7901 and MGC-803 cell lines was significantly downregulated with a fold-change of 0.58±0.03, 0.57±0.03, 0.32±0.03 and 0.10±0.00, respectively, compared with that noted in the GES-1 cells (P<0.05; Fig. 1C). We next investigated the invasive ability of the GC cell lines using the Transwell assay. The data demonstrated that the GC cell lines MGC-803 and SGC-7901 displayed a stronger invasive ability with lower levels of miR-647 compared with HGC-27 and AGS cell lines (P<0.05; Fig. 1D). Overall, these results suggest that miR-647 expression is frequently downregulated in GC, and is correlated with proliferation and metastasis, suggesting that miR-647 acts as a tumor-suppressor in GC development.

To explore the potential growth-suppressive role of miR-647 in GC, we first established GC cells stably overexpressing miR-647. The miR-647 mimic and LV-GFP were transfected into SGC-7901 or MGC-803 cells that displayed a stronger invasive ability, respectively. The expression of miR-647 in the stably transfected cells was confirmed by qRT-PCR (P<0.05; Fig. 2A). Cell proliferation assays revealed that miR-647 overexpression significantly reduced the growth rates of the SGC-7901 and MGC-803 cells (P<0.05; Fig. 2B). As shown in Fig. 2C, cell cycle analysis also supported the results, as overexpression of miR-647 was related to G0/G1 phase arrest. Cell counts in the G0/G1 phase apparently increased while those in the S phase decreased (P<0.05). Furthermore, we found that miR-647 increased the basal rate of cells undergoing apoptosis (P<0.05, Fig. 2D). As shown in Fig. 2E, typical features of apoptosis such as formation of apoptotic bodies, chromatin fragmentation or concentration and mitochondrial disappearance or crest fracture were observed in the 7901–647 and 803–647 groups under TEM. Consistent with the results obtained in the in vitro growth assays, miR-647 significantly inhibited the xenograft tumor growth in nude mice (P<0.05, Fig. 2F and G). Collectively, these data clearly demonstrate that miR-647 plays a growth-suppressive function in GC.

We confirmed that the ectopic expression of miR-647 is closely associated with GC metastasis including lymph node and distant metastasis; that is to say, miR-647 mediates the processes of metastasis in the development of GC. To determine the role of miR-647 in metastasis, the effect of miR-647 overexpression on cell migration and invasion was investigated using wound-healing and Transwell assays with Matrigel. The results demonstrated that overexpression of miR-647 in the SGC7901 and MGC803 cells substantially impaired cell migration (P<0.05; Fig. 3A). Similarly, miR-647 significantly decreased GC cell invasion in the Matrigel-coated Transwell assay (P<0.05; Fig. 3B). To probe the effect of abnormal expression of miR-647 on tumor metastasis in vivo, controls and miR-647-overexpressing cells were injected into nude mice via the lateral tail vein. As shown in Fig. 3C and Table III, the result was confirmed by histological staining and statistical analysis, demonstrating that miR-647 overexpression strongly inhibited the ability of GC metastases in the liver. These results clearly indicate that miR-647 suppresses GC migration and invasion.

To investigate the molecular mechanisms of miR-647-mediated tumor suppression, potential direct targets for miR-647 were screened by microarray data analysis combined with bioinformatic target prediction. The functional downstream genes for miR-647 were selected via gene expression profiling of 803–647 and 803-NC cells. Of the 31,741 genes analyzed by UinGene Whole Genome DNA array analysis, 113 and 110 were upregulated and downregulated, respectively, which had an expression change >1.5-fold in GC cells of the 803–647 group compared with the 803-NC group. Following suppression of most targets by miRNAs, the downregulation of 110 genes in the 803–647 group was analyzed for potential miR-647 direct targets predicted by at least two of the 3 computational target prediction algorithms (miRanda, TargetScan and microRNA). Four potential miR-647 targets, ANK2, ZNF609, KNDC1 and TRIM4 were identified (Fig. 4A). To examine whether or not the 4 potential targets were regulated by miR-647 in GC cell lines, the miR-647 overexpression of SGC7901 and MGC803 cells was analyzed for transcriptional expression of ANK2, ZNF609, KNDC1 and TRIM4 by qRT-PCR. As shown in Fig. 4B, among the 4 potential targets, ANK2 and KNDC1 mRNA expression was suppressed by miR-647 in both cells. Since only ANK2 plays an oncogenic function in tumorigenesis, it was selected for further evaluation as the miR-647 potential target. Furthermore, suppression of ANK2 expression by miR-647 overexpression was confirmed in the SGC7901 and MGC803 cells (P<0.05; Fig. 4C). As shown in Fig. 4D, ANK2 was a potential target of miR-647 as predicted by TargetScan and miRNA.

To investigate the mechanism by which miR-647 suppresses GC proliferation and metastasis, we utilized qRT-PCR and western blotting to determine the expression of genes that regulate proliferation and metastasis. Our data demonstrated that the expression of FAK, MMP2, MMP12, CD44 and SNAIL1 was downregulated after miR-647 overexpression in both SGC7901 and MGC803 cell lines (P<0.05, Fig. 4E and F).