From January, 2011 to January, 2012, 115 patients with CRC who underwent radical resection at the Department of Gastrointestinal Surgery, Tumor Hospital of Xin Jiang Medical University (Ürümqi, China), were enrolled in this study. This study was approved by the Ethics Committee of the Tumor Hospital of Xin Jiang Medical University (approval no. 20110110, 10 January, 2011) and all procedures followed the principles of the Declaration of Helsinki. All subjects provided written informed consent prior to participation. None of the patients had received systemic chemotherapy or radiotherapy prior to surgery. All adjuvant treatments after surgery for all patients were identical. All CRC tissues and their corresponding adjacent normal tissues were collected during surgery, and immediately snap frozen in liquid nitrogen. Follow-up visits were performed for survival analysis for up to 75 months after surgery. All clinical features were obtained from medical records.

All cell culture media were purchased from HyClone (GE Healthcare Life Sciences). Three human colon cancer cell (CCC) lines with different invasive potentials (HT29, SW620 and LoVo) and an immortalized human epithelial cell line (HIEC) were donated by the National Laboratory of Xi Jing Digestive Disease Hospital (Fourth Military Medical University, Xi'an). All cell lines were maintained in RPMI-1640 growth medium (Thermo Fisher Scientific, Inc.) supplemented with 10% heat-inactivated fetal bovine serum, 1% penicillin and 1% streptomycin (Thermo Fisher Scientific, Inc.). The HIEC cells were also supplemented with human insulin (0.1 U/ml).

Cancer tissues and normal matched normal tissues (which were 15 cm away from the lesion margin) were obtained during surgery, and stored at −80°C after being snap-frozen. Total RNA was extracted from the tissues and cell lines using TRIzol^® reagent (Thermo Fisher Scientific, Inc.), following the manufacturer's protocol. The concentration and purity of the total RNA were determined by measuring the optical density at 260 and 280 nm. The mRNA extracted from 115 CRC tissue samples and 85 matched normal tissues were included for further analyses.

A total of 21 CRC samples and 10 adjacent normal samples were used to evaluate the methylation levels of TFF1 and TFF3 promoter regions. Fresh tumor tissues were snap-frozen in liquid nitrogen instantly after surgical resection and stored at ‒80°C. All tumor specimens were carefully micro dissected, ensuring that the final tissue samples consisted of at least 85% tumorous cells. Tissue processing and DNA extraction were performed using the QIAamp DNA Mini kit (Qiagen GmbH), according to the manufacturer's protocol.

Genomic DNA was treated with bisulfite using an EpiTect Bisulfite kit (Qiagen AB) and then amplified using PCR Accessory Set (Sequenom). In vitro transcription and T cleavage were performed with Shrimp alkaline phosphatase (Mass CLEAVE kit, Sequenom) before the acquisition of a purified resin-treated product.

The entire RNase A and clean resin treated product were automatically dispensed onto silicon matrix preloaded chips (SpectroCHIP; Sequenom), and the mass spectra were obtained using a Mass ARRAY Compact matrix assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF-MS) analyzer (Sequenom). The mass spectrometry methylation ratios were generated using EpiTYPER (ver. 4.0; Sequenom).

The colon cancer cell lines were harvested separately in ice-cold PBS. Total protein was extracted employing RIPA lysis and extraction buffer (Thermo Fisher Scientific) and quantified by measuring the integrated optical density (IOD) using the BCA Protein Assay kit (25 µl per lane), and separated by 15% SDS PAGE gel electrophoresis, and transferred onto PVDF membranes. The membranes were subsequently blocked with 5% (w/v) dried skimmed milk powder in Tris-buffered saline (blocking solution) for 1 h at room temperature. Subsequently, the membranes were probed with primary antibodies against TFF1 (1:500, ab92377) TFF3 (1:300, ab108599) and Twist1 (1:500, ab50581), E cadherin (1:500, ab40772), Vimentin (1:500, ab92547), Snail (1:500, ab53519), β-actin (1:2,500, ab8226) (all from Abcam) in blocking solution overnight at 4°C. Following a TBS-Tween wash, the membranes were incubated with goat anti-rabbit secondary antibody (HRP; 1:10,000, ab6721) or rabbit anti-goat secondary antibody (HRP; 1:5,000, ab6741) (both from Abcam) in blocking solution for 1 h at room temperature. Immunoreactive proteins were detected using enhanced chemiluminescence kit (Thermo Scientific Pierce ECL Substrate). The intensity of the bands on western blot was evaluated by Quantity One^® 1-D analysis software (version 4.62, Bio-Rad).

The TFF3 expression plasmid was built by cloning TFF3 cDNA into a pIRES2-enhanced green fluorescent protein vector (no. 6029-1, Clontech). The transfection of TFF3 expression or empty vector plasmids (pIRES2-EGFP) into the HT29 cells was performed using Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. The cells were incubated at 37°C in 5% CO₂ incubator for 48 h, and fluorescence photography following transfection was then carried out to evaluate the efficiency of transfection.

The short hairpin RNA (shRNA) sequences against TFF3 and control shRNA were purchased from Thermo Fisher Scientific, Inc. LoVo cells were transfected with 30 nM shRNAs using Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific, Inc.), according to the manufacturer's protocol. Following incubation at 37°C with 5% CO₂ for 48 h, the knockdown of TFF3 was confirmed by both RT-qPCR and western blot analysis.

Cell proliferation rates were measured by CCK 8 cell proliferation assay. The prepared cell suspension (100 µl/well) in a 96-well plate was pre-incubated at 37°C in 5% CO₂ and 10 µl of the CCK 8 (Alpha CP8 Cell Proliferation Assay kit, no. A32001-500T, Alpha Applied Bioscience) solution were added to each well, and incubation was maintained in an incubator for 1-4 h before measuring the absorbance at 450 nm using a microplate reader (Varioskan LUX, Thermo Fisher Scientific).

Scratch wound healing and Transwell assays were performed to evaluate cell migration and invasion. CCCs were grown to 90% confluence before an ~0.6 mm wide wound was created with the help of an insert. The wounds were subsequently imaged at 0, 24 and 48 h to assess the level of motility. The cell migratory ability was quantified by calculating the area between the two leading edges of cells using ImageJ software (version 1.52e, Broken Symmetry Software).

For the Transwell assays, 24-well Transwell units with 8 µm pore-size filters (Corning, Inc.) were precoated with 40 µl Matrigel (for invasion assay) or without Matrigel (for migration assay). A total of 200 µl of a cell suspension containing 5×10⁴ cells/ml in serum free medium were inserted into the upper chamber. In the lower chamber, 300 µl Dulbecco's modified Eagle's medium containing 20% fetal calf serum were added. Following a 24-h incubation at 37°C with 5% CO₂, the cells in the upper chamber were removed with a cotton swab and the traversed cells in the lower chamber were fixed with 4% paraformaldehyde, stained with 0.5% crystal violet (no. 32675, Sigma Aldrich) at room temperature for 10 min and counted in 5 randomly chosen fields under a light microscope (SFC-100 Series, Motic Europe).

The cells were collected and fixed using 2.5% TEM-specific glutaric dialdehyde solution at 48 h following transfection. The cells were then centrifugated at a speed of 2,504 × g at room temperature and fixed in 1% osmium tetroxide at 4°C for 15-30 min before dehydrating in acetone solutions of 50, 70, 90 and 100%. The cells were subsequently embedded in epoxy resin (EPON812), sliced into 50-70 nm ultrathin sections after being fixed on metal mesh and stained with sodium acetate at room temperature for 5 30 min, and observed under a transmission electron microscope (JEM ARM200F NEOARM). Moreover, cell morphology was photographed under an optical microscope (SFC 100 Series, Motic Europe) to observe the changes in cell morphology following the overexpression or silencing of TFF3.

The continuous data are presented as the means ± standard deviation, and compared using a unpaired Student's t-test (as comparison of mRNA levels of TFF3 or TFF1) or Mann Whitney U test (due to the unpaired, non-normal distribution of methylation level of TFF3 or TFF1). One way analysis of variance (ANOVA) with an LSD post hoc test was used to compare the group data. The clinicopathological features were analyzed between the high TFF3 and low TFF3 expression groups using a Pearson χ² test or χ² test with Yates' correction. Pearson's correlation analysis was performed to evaluate the correlation between relevant factors. The cumulative survival rates were determined using the Kaplan Meier method and compared with the log rank test. The Cox proportional hazards test was used for univariate and multivariate survival analysis. All statistical analyses were performed using GraphPad Prism 5 software (GraphPad Software, Inc.) or SPSS 18.0 (SPSS, Inc.). P<0.05 was considered to indicate a statistically significant difference.

The results from the analysis of the mRNA expression of TFF1 or TFF3 in the CRC tissues and matched adjacent normal tissues are presented in Table I and Fig. 1A. The average mRNA levels of TFF1 or TFF3 in the cancer tissue were significantly higher than those in the matched adjacent normal tissue (P=0.034 and P=0.007, respectively). The analysis of correlation between the mRNA expression of TFF1 and TFF3 indicated the existence of a positive association between TFF1 and TFF3 mRNA expression in patients with CRC (r=0.219, P=0.019; Table II).

The association between the expression of two TFF genes and the clinicopathological characteristics of the patients with CRC was further analyzed. The higher expression of TFF3 was significantly associated with histological type, invasion depth (T), lymph node number (N), metastasis at diagnosis (M), overall TNM staging and recurrence or metastasis following treatment. Only the circumferential resection margin (CRM), metastasis at diagnosis and TNM staging were associated with a higher expression of TFF1 (Table III).

The 115 patients with CRC were divided into the high or low expression group according to the X-tile algorithm (29), which generated optimal cut off values for the TFF3 and TFF1 mRNA levels in the CRC samples. The 1, 3 and 5 year survival rates of the patients in the high TFF3 expression group were 91, 55.9 and 21.1%, respectively, and in the low TFF3 in the low TFF3 group, they were 100, 85.2 and 53.9%, respectively (data not shown). Survival analysis revealed that a higher expression of TFF3 was associated with a worse overall survival, while TFF1 expression demonstrated no association with patient survival (P<0.001 and P=0.125, respectively; Fig. 1B). Furthermore, predictive factors of patient survival were calculated using a Cox proportional-hazard model. CRM, vascular invasion, lymph node number (N), metastasis at diagnosis (M), TNM staging, recurrence or metastasis following treatment and TFF3 were associated with patient survival by univariate analysis, while invasion depth (T), metastasis at diagnosis (M), TNM staging, recurrence or metastasis following treatment, tumor size and TFF3 were found to be the predictive factors of patient survival by multivariate analysis (Table IV).

To evaluate the potential association between the level of promoter methylation and respective gene expression, the promoter methylation patterns of TFF1 and TFF3 in 21 CRC samples and 10 adjacent normal samples were evaluated using MALDI-TOF-MS. A total of 527 CpG sites were observed and 465 sites (88.2%) were analyzed. A total of 10 promoter region units of TFF1 and 7 units of TFF3 were successfully detected (Fig. 2). The average methylation levels of the TFF1 or TFF3 promoter region in the CRC group were observed to be lower than those in normal group, although these differences were not statistically significantly (P=0.374 and P=0.145, respectively; Table VII). The methylation levels of each promoter unit of TFF1 or TFF3 did not markedly differ between the CRC group and normal group (Tables V and VI). Two way hierarchical cluster analysis also revealed that less dominant clustering was found in the cancer or normal tissues. The hierarchical cluster algorithm could not separate the colon cancer samples from the normal tissue samples (Fig. 3). In addition, following correlation analysis, no correlation was observed between the promoter methylation and expression of TFF1 or TFF3 (Tables VIII and IX, and Fig. 4).

TFF3 expression in 3 human CCC lines (HT29, SW620 and LoVo) and an immortalized human epithelial cell line (HIEC) was examined by both RT-qPCR and western blot analysis. As illustrated in Fig. 5A, TFF3 expression in the HT29 cells was the lowest among the 4 cell lines, while the highest expression of TFF3 was observed in the LoVo cells. Both RT-qPCR and western blot analysis demonstrated that transfection with a TFF3 expression plasmid resulted in an enhanced expression of TFF3 in the HT29 cells (Fig. 5B). The Cell Counting kit-8 (CCK-8) assay revealed an apparently higher proliferation in the pIRES2-TFF3-transfected cells than in the empty vector transfected controls (Fig. 5C). An elevated expression of TFF3 was also found to have increased the invasive and migratory ability of the pIRES2-TFF3-transfected cells (Fig. 5D). Furthermore, wound healing assay revealed that the pIRES2-TFF3-transfected cells migrated at a more rapid rate (as shown by the wound closure speed) than the empty vector transfected cells (Fig. 5E). Lastly, compared with the control group, the pIRES2-TFF3-transfected cells exhibited fewer intercellular junctions and a reduced number of villi and less pseudopodia. An increased number of polygonal or spindle cells in the TFF3 overexpression group was also observed (Fig. 5F).

TFF3 was silenced in the LoVo cells, which expressed the highest levels of TFF3, via shRNA transfection in order to further assess the effects of TFF3 on cancer progression. As shown in Fig. 6A, both RT-qPCR and western blot analysis revealed that TFF3 expression was markedly reduced following transfection with the shRNA sequences. The downregulation of TFF3 notably reduced the proliferation of the LoVo cells, as indicated by the results of CCK-8 assay (Fig. 6B). The invasive and migratory abilities of the LoVo cells were also found to be markedly reduced following the knockdown of TFF3, as shown by the Transwell assay results (Fig. 6C). Lastly, the wound healing assay also demonstrated that the migratory capacity of the LoVo cells was impaired by the silencing of TFF3 (Fig. 6D).

The expression of EMT markers was examined by both western blot analysis and RT-qPCR to confirm whether TFF3 promotes cancer progression through EMT. As illustrated in Fig. 7A and C, the elevated expression of TFF3 in the HT29 cells resulted in the downregulation of E-cadherin, a well-established epithelial marker, at both the protein and mRNA level. TFF3 expression also promoted an increase in the mRNA and protein levels of the mesenchymal markers, Twist1, Snail and Vimentin. Furthermore, the shRNA mediated knockdown of TFF3 upregulated E-cadherin expression, and suppressed the expression of Twist1, Snail and Vimentin in the LoVo cells at both the protein and mRNA level (Fig. 7B and C). The EMT like morphological changes in the cells were also observed under an optical microscope. The shape of the HT29 cells changed from a round or volmer-weber mode into a triangular or spindle form following the overexpression of TFF3. LoVo cells displayed a reversed morphological changes from the scattered cord like forms to the round, rectangular or clustered patterns following the silencing of TFF3 (Fig. 7D).