Gastric specimens (n=28) were obtained from the tumor and an adjacent non-cancerous area, ≥6 cm from the tumor tissues of gastric carcinoma patients at the First Affiliated Hospital of Kunming Medical College (Kunming, China). The mean age of the patients at diagnosis was 56 years. The non-neoplastic tissue was confirmed to lack tumor cell infiltration using histological analysis. The tissues were immediately placed in liquid nitrogen and stored at −80°C until use. A gastric cancer tissue microarray representing 110 types of gastric cancer with their non-neoplastic resection margins was constructed (18) at the Shanghai Outdo Biochip Center (Shanghai, China). Human samples were used in accordance with the requirements of the Ethical Committee of the Kunming Institute of Zoology, the Chinese Academy of Sciences, under the guidelines of the World Medical Assembly (Declaration of Helsinki). Written informed consent was obtained from the patient’s families.

RNA extraction and first-strand cDNA synthesis were performed as previously described (19). For semi-quantitative reverse transcription PCR (RT-PCR) and quantitative PCR (qPCR), the following primers were used: Forward, 5′-CTGCTTCTCCAACTTCATCT-3′ and reverse, 5′-CTTAGTAATGGCAGTCTTCC-3′ for TFF2 (74-bp product); and forward, 5′-ATGGGGAAGGTGAAGGTCG-3′ and reverse, 5′-GGGGTCATTGATGGCAACAATA-3′ for glyceraldehyde 3-phosphate dehydrogenase (GAPDH; 107-bp product). GAPDH was used as an internal control. Following RT-PCR, the amplicons were separated by electrophoresis in a 2% agarose gel that was stained with ethidium bromide and viewed under ultraviolet illumination. qPCR was performed using a continuous fluorescence detector (Opticon Monitor; Bio-Rad, Hercules, CA, USA) and PCR was performed using an SYBR Green real-time PCR kit (Takara Bio, Inc., Dalian, China) with the following reaction conditions: Initial denaturation at 95°C for 1 min followed by 40 cycles at 95°C for 15 sec, 60°C for 15 sec and 72°C for 20 sec. Each sample was run three times. No-template controls (no cDNA in the PCR) were run to detect non-specific or genomic amplification and primer dimerization. Fluorescence curve analysis was performed using Opticon Monitor software. The relative quantitative evaluation of TFF2 levels was performed using the E-method (20) and expressed as a ratio of the TFF2 to GAPDH transcripts in the tumor tissue divided by that ratio in the non-neoplastic tissue of the same patient. The identities of RT-PCR and qPCR products were confirmed by DNA sequencing.

AGS and N87 human gastric cancer cells were obtained from the American Type Culture Collection (Manassas, VA, USA). AGS cells were cultured in a 1:1 mixture of Dulbecco’s modified Eagle’s medium and Ham’s media. N87 cells were cultured in RPMI-1640 media containing 10% fetal calf serum, 100 U/ml penicillin and 100 mg/ml streptomycin. The cells were grown in a humidified atmosphere containing 5% CO₂ at 37°C. The cells were seeded at a density of 1×10⁶ cells/ml in a 60 mm dish and treated with 10 mM 5-Aza-2′-deoxycytidine (5-Aza-2′-dC; Sigma-Aldrich, St. Louis, MO, USA). DMSO was used as a control. The cells were collected after 3 days and subjected to RT-PCR, qPCR and western blot analysis.

Tissue and cell samples were homogenized in radioimmunoprecipitation assay buffer containing a protease inhibitor cocktail (Sigma-Aldrich). The protein concentration was determined using a protein assay kit (Bio-Rad). Samples (containing 50 μg of protein) were loaded into a sodium dodecyl sulfate-polyacrylamide gel electrophoresis gel, electrophoresed and then electro-transferred onto a polyvinylidene fluoride membrane. The membrane was subsequently blocked with 3% bovine serum albumin and incubated with an anti-human TFF2 polyclonal antibody (Protein Tech, Chicago, IL, USA) and a horseradish peroxidase-conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Protein bands were visualized using Super Signal reagents (Thermo Fisher Scientific, Inc., Rockford, IL, USA).

Tissue IHC was performed as previously described (21). Briefly, antigen retrieval was performed by heating samples in an autoclave at 121°C for 5 min. Dewaxed sections were pre-incubated with blocking serum and then incubated overnight with an anti-human TFF2 antibody (P-19; Santa Cruz Biotechnology) at 4°C. Specific binding was detected using a streptavidin-biotin-peroxidase assay kit (Maxim, Fujian, China). The section was counterstained with Harris hematoxylin. Direct microscopic micrographs were captured using a Leica DFC320 camera controlled using Leica IM50 software (Leica, Mannheim, Germany). Sections incubated with normal goat IgG served as negative controls, which were devoid of any detectable immunolabeling. The specificity of the anti-TFF2 antibody was confirmed using an overnight preincubation at 4°C with its antigen in a 20-fold molar excess of antigen to antibody. The preincubation with TFF2 antigens resulted in an absence of immunolabeling. Immunohistochemical staining was semi-quantitatively assessed by measuring the intensity of the staining (0, 1, 2 or 3) and the extent of staining (0, 0%; 1, 1–10%; 2, 11–50%; and 3, 51–100%). The scores for the intensity and extent of staining were multiplied to yield a weighted score for each case (maximum possible, 9). For the statistical analysis, the weighted scores were grouped into two categories, in which scores of 0–3 and 4–9 were considered negative and positive, respectively (22).

Genomic DNA from carefully selected 20-μm sections of gastric cancer, non-neoplastic tissues, and AGS and N87 cell lines was isolated using the Universal Genomic DNA Extraction kit (Takara, Bio, Inc.) and bisulfite-converted using the Clontech EpiXplore™ Methyl Detection kit (Takara, Bio, Inc.). TFF2 promoter sequences were amplified from the bisulfite-converted DNA by PCR, purified from agarose gels and subcloned into the pBackZero T Vector (Takara, Bio, Inc.). For each sample, 11 individual clones were sequenced to identify methylated cytosine residues. The PCR primer sequences used were forward, 5′-GGGATTTTTTTATGTTATTTGTTGG-3′ and reverse, 5′-ATAAAAAAACCCTCTCCTTCACTTACAAAA-3′.

All statistical results were analyzed using SPSS 11.0 software (SPSS, Inc., Chicago, IL, USA). Fisher’s exact and χ² tests were used to analyze the correlation between TFF2 expression and clinicopathological parameters (Tables I and II). Differences in the numerical data between the two paired groups were evaluated using the paired Wilcoxon test (Fig. 1). P<0.05 was considered to indicate a statistically significant difference.

TFF2 mRNA expression in gastric cancer tissues was examined using RT-PCR. In total, four pairs of samples were randomly selected from 28 patients and normalized to the GAPDH level. As shown in Fig. 1A, TFF2 mRNA expression was significantly decreased in cancer tissues compared with the associated normal mucosa. To quantify the differences in the expression of TFF2 mRNA, qPCR was performed on 28 gastric tumor tissue samples. TFF2 expression was downregulated in 93% (26 out of 28) of gastric cancer tissue samples compared with the associated non-neoplastic tissues. In addition, the mRNA levels of TFF2/GAPDH in gastric cancer tissues were significantly lower than those in the corresponding non-neoplastic mucosal tissues (mean ± SE, 3.4±2.7 vs. 9.6±5.4, respectively; P=0.046) (Fig. 1B). The clinical significance of the loss of TFF2 expression was further investigated based on the clinical pathological data. As shown in Table I, there were significant differences in TFF2 mRNA expression in well- and moderately differentiated tumors versus poorly differentiated tumors (P=0.019), and in tumors with lymph node invasion versus non-invasive tumors (P=0.026). In detail, TFF2 mRNA was reduced by a fold-change of 15.0±4.1 (mean ± SE) in the 22 poorly differentiated tumors compared with a fold-change of 1.9±0.7 in the six well- and moderately differentiated tumors (P=0.009; paired Wilcoxon test), and a fold-change of 15.6±3.7 in the 23 lymph node invasive tumors compared with a fold-change of 1.1 ± 0.4 in the five non-invasive tumors (P=0.002; paired Wilcoxon test) (Fig. 1C).

The protein expression levels of TFF2 in normal and gastric cancer tissues were verified using western blot analysis. After the samples were normalized to the β-actin level, a marked reduction or loss of TFF2 protein was observed in four gastric tumor tissue samples compared with the matched non-malignant tissues (Fig. 2). TFF2 protein levels in normal and malignant gastric mucosa were also assessed using an IHC assay. In 110 gastric cancer tissue microarray assays, TFF2 expression was downregulated in 82% (90 out of 110). TFF2 was expressed at high levels in all investigated normal mucosa tissues and staining was identified from the basal-to-middle portions of gastric glands. Furthermore, TFF2 localization was in the cytoplasm and the membrane of gastric normal epithelial cells (Fig. 3A). However, the expression was significantly reduced in well- (Fig. 3B) and moderately (Fig. 3C) differentiated intestinal gastric cancer tissues, while TFF2 expression was almost absent in the poorly differentiated intestinal and diffuse types of gastric cancer (Fig. 3D). Sections incubated with normal goat IgG served as negative controls (Fig. 3E). Analysis of the correlation between TFF2 expression and clinicopathological data showed that decreased TFF2 expression was closely associated with tumor cell differentiation and lymph node invasion (Table II). In detail, TFF2 expression was decreased in 86.9% of poorly differentiated cancers and 65.4% of well- and moderately differentiated cancers (P=0.02; χ² test). TFF2 expression was decreased in 88.8% of positive lymph node invasion and 63.3% of negative lymph node invasion tumors (P=0.004, χ² test) (Table II).

To elucidate the potential molecular mechanisms underlying the process of TFF2 downregulation in the progression of gastric cancer, AGS cells were treated with 10 mM 5-Aza-2′-dC, which is a demethylating agent. RT-PCR analysis showed that TFF2 expression in AGS cells was significantly increased following 5-Aza-2′-dC treatment for 3 days (Fig. 4A). qPCR indicated a 3.89-fold increase in the mRNA expression levels of TFF2 following 5-Aza-2′-dC-treatment, while western blot analysis also indicated that TFF2 protein expression increased in AGS cells treated with 5-Aza-2′-dC (Fig. 4B). The results suggested that the epigenetic alteration may be involved in the downregulation of TFF2 expression in the progression of gastric cancer.

Treatment with 5-Aza-2′-dC induced demethylation and led to the upregulated expression of TFF2 in AGS cells. Therefore, the methylation level of the TFF2 gene promoter was further analyzed in three gastric cancer and non-neoplastic tissue samples, as well as in AGS and N87 gastric cancer cell lines. Using the genomic bisulfite sequencing method, 16 CpG sites were analyzed in a 571-bp region containing part of the TFF2 promoter region. It included six CpGs found after the transcription start site and 10 CpGs located in the ~300-bp 5′-flanking region. The average promoter methylation level of three gastric cancer tissues was 70.4% and the control of non-neoplastic tissues was 41.0%, which showed that gastric cancer tissues with a decreased expression of TFF2 exhibited hypermethylation levels at the 16 CpG sites. In addition, AGS and N87 gastric cancer lines exhibited 85.2 and 93.7% methylation levels at the 16 CpG sites, respectively (Fig. 5). Therefore, these results indicated that promoter hypermethylation may lead to the inhibition of TFF2 transcription in gastric cancer.