Blood samples were collected before GC resection from a total of 153 GC patients (112 males and 41 females) who underwent GC resection at the Department of General Surgery at the Fourth Hospital of Hebei Medical University (Shijiazhuang, China) prior to the operation between February 2008 and November 2010. All GC patients were diagnosed preoperatively via histopathological examination (16). The TNM system was used to determine the different lymph node metastasis and clinical stages (17). In addition, blood samples were collected from 233 healthy subjects (169 males, 64 females) with no previous history of cancer. The characteristics of patients and controls were listed in Table I. Patients provided their informed consent for participation in the present study. All procedures were supervised and approved by the Human Tissue Research Committee of the Fourth Hospital of Hebei Medical University (Shijiazhuang, China).

Genomic DNA was extracted using the Wizard Genomic DNA Purification Kit (A1125; Promega Corp., Madison, WI, USA). Briefly, 100 µl blood was added to the cell lysis solution for red blood cells, followed by lysis of the white blood cells and their nuclei in the Nuclei Lysis Solution with RNase. The cellular proteins were then removed by a salt-precipitation step. Finally, the genomic DNA was concentrated and desalted by isopropanol precipitation.

The miRNA SNPs located at the binding site, including RYR3 (rs1044129), C14orf101 (rs4901706), KIAA0423 (rs1053667), GOLGA7 (rs11337) and KRT81 (rs3660), were genotyped using the ligation detection reaction method (18), featuring forward and reverse primers, in order to amplify DNA fragments flanking the SNPs. This procedure was performed according to the SNP database of the National Center for Biotechnology Information (Bethesda, MD, USA; http://www.ncbi.nlm.nih.gov/snp/).

Polymerase chain reaction (PCR) was performed with 50 ng genomic DNA, 1 µl of 10 nm primer pairs, 12.5 µl of Master Mix and distilled water to a final volume of 25 µl using a PCR Master Mix kit (K1081; Promega Corp.), according to the manufacturer's instructions, and for 35 thermal cycles. The primer and probe sequences used in PCR are shown in Table II. Ligation was performed with various probes, which were matched to the SNPs. Subsequently, the ABI PRISM 3730xl DNA Analyzer (Applied Biosystems; Thermo Fisher Scientific, Inc., Foster City, CA, USA) was used to separate the ligated products. SNPs were detected and verified based on differences in the length of ligated products. The experiment was performed once if it was successful.

A total of 4 oligonucleotides were synthesized based on dbSNP, the NCBI database of genetic variation, which consisted of the following parts (from 5′ to 3′): A XhoI sticky end (5 bp), a fragment from the 3′-UTR of the C14orf101 gene containing the GG or AA genotype (rs4901706; 47 bp), and a NotI sticky end (2 bp). The following sequences were used: GG-containing sense, 5′-TCGAGAGTGCTCAGCTACTTCTCCTCCACTTTGAAAGACCCCTCCCAGATCTGC-3′, and antisense, 5′-GGCCGCAGATCTGGGAGGGGTCTTTCAAAGTGGAGGAGAAGTAGCTGAGCACTC-3′; AA-containing sense, 5′-TCGAGAGTGCTCAGCTACTTCTCCTGCACTTTGAAAGACCCCTCCCAGATCTGC-3′, and antisense, 5′-GGCCGCAGATCTGGGAGGGGTCTTTCAAAGTGCAGGAGAAGTAGCTGAGCACTC-3′. The four oligonucleotides were incubated for 5 min with 1X NEBuffer 2 (New England Biolabs, Ipswich, MA, USA) in a heating block at 95°C. Next, the temperature was gradually reduced until it reached room temperature. The psiCheck-2 vector featuring the Renilla/luciferase and controlled firefly luciferase genes was linearized by digestion with NotI and XhoI (New England Biolabs), and the vector was then purified using a 1% agarose gel electrophoresis. The oligonucleotides were ligated in the linearized psiCheck-2 vector (Promega Corp.) into the cloning sites (NotI and XhoI), which were downstream of the Renilla luciferase reporter gene with T4 DNA ligase (Promega Corp.). Subsequently, the ligated vectors were transformed in Escherichia coli-competent cells, as follows: Competent cells were taken from −80°C freezer and thawed on ice for 20 min. 1 µl of ligated vector was mixed into 100 µl of competent cells in a microcentrifuge. The competent cell/DNA mixture was then placed on ice for 20–30 min and each transformation tube was heat shocked by placing the tube into a 42°C water bath for 90 sec. The tubes were placed back on the ice for 2 min and 500 µl of Luria-Bertani (LB) medium was added. The cells were grown at 37°C in a shaking incubator for 45 min. All of the transformations were plated onto a 10 cm LB agar plate and the plates were incubated at 37°C overnight. A total of 3 bacterial colonies were selected and the bacterial culture was incubated at 37°C for 12–18 h in a shaking incubator. The bacterial fluid samples were sequenced by the methods of Sanger dideoxy (Sangon Biotech, Shanghai, China), and sequencing was used to identify and select positive clones.

A human GC cell line (SGC 7901) purchased from the Shanghai Institute of Biochemistry and Cell Biology (Shanghai, China) was seeded in 48-well plates and transfected with the modified psiCheck-2 vector (800 ng) using Lipo 2000 (11668-027; Invitrogen, CA, USA) at 37°C for 4–6 h with the GG or AA genotype. At 48 h after transfection, the Renilla luciferase activity was assessed using the Dual-Lucy Assay kit (Vigorous Instrument Co., Ltd., Beijing, China) and a luminometer BioFix Lumi-10 (Macherey-Nagel, Dusseldorf, Germany). The transfection efficiency was normalized against the firefly luciferase activity.

The χ² test was performed to analyze dichotomous variables, including the presence or absence of an individual SNP in the patients and healthy controls. The odds ratio (OR) and 95% confidence interval (CI) were calculated using an unconditional logistic regression model. In addition, Student's t-test was used to compare the differential expression levels between genotypic groups in the Renilla/luciferase reporter assays. SPSS version 18.0 software (SPSS, Inc., Chicago, IL, USA) was used to perform statistical analyses. Statistically significant differences were defined when the P-value was <0.05.

A total of 153 patients and 233 healthy controls were included in the present study. The clinical characteristics of patients including gender, age, tumor size and location, metastasis, clinical stage, differentiation status and pathological types were listed in Table I. The characteristics of the control, including gender and age were also listed in Table I. The distribution frequency was not identified to be different between patients and controls with respect to their age and gender (P=0.885 and 0.209) respectively. In addition, the mean age between patients and controls was not statistically different. An analysis for these patients and the controls was subsequently performed.

In the present study, miRNA binding-site SNPs were genotyped in the patients and healthy controls, including C14orf101 (rs4901706), RYR3 (rs1044129), GOLGA7 (rs11337), KRT81 (rs3660) and KIAA0423 (rs1053667) to evaluate their association with the risk for cancer. The cancer risk association was evaluated for the distribution frequency of rs4901706 (GG vs. AG+AA), rs1044129 (AA vs. AG+GG), rs11337 (GG vs. GT+TT), rs3660 (GG vs. CG+CC) and rs1053667 (TT vs. CT+CC) between patients and controls using the χ² test. As shown in Table III, the rs1044129 AA carriers were 47 in GC patients and 67 in the controls whereas the AG+GG carriers were 106 in GC patients and 166 in the controls; the rs1053667 TT carriers were 113 in GC patients and 161 in the controls whereas the CT+TT carriers were 40 in GC patients and 72 in the controls. In addition, the rs11337 GG carriers were 92 in GC patients and 141 in the controls while the GT+TT carriers were 61 in GC patients and 92 in the controls; Finally, the rs3660 GG carriers were 97 in GC patients and 145 in the controls while CG+CC carriers were 56 in GC patients and 88 in the controls. The rs 1044129 (relative risk, 1.099; 95% CI, 0.704–1.715; P=0.679), rs11337 (relative risk, 0.984; 95% CI, 0.649–1.493; P=0.940), rs3660 (relative risk, 1.023; 95% CI, 0.662–1.583; P=0.917) and rs1053667 (relative risk, 1.263; 95% CI, 0.801–1.992; P=0.314) were not associated with cancer risk by the analysis of the present study. The GG carriers were 100 in GC patients and 125 in the controls whereas the AG+AA were 53 in GC patients and 108 in the controls. It was also noted that the GG genotype was susceptible to GC carcinogenesis (relative risk, 1.630; 95% CI, 1.070–2.483; P=0.022). The data of the present study demonstrated that the rs4901706 SNP of C14orf101 was a predictive marker for GC risk.

A vector named psiCheck-2 was constructed containing the rs4901706 AA or GG genotypes downstream of the Renilla/luciferase reporter gene in order to assess their functional effect on C14orf101 expression. The vector was transfected into the GC cell line SGC 7901. As shown in Fig. 1, a significant reduction of Renilla/luciferase activity was observed in GG genotypes compared with that of the AA genotypes (P=0.007). These data demonstrated that rs4901706 SNP may change the binding affinity between the rs4901706 genotype at the 3′UTR of C14orf101 and the corresponding miRNA.