Terminal fucose mediates progression of human cholangiocarcinoma through EGF/EGFR activation and the Akt/Erk signaling pathway

Aberrant glycosylation is recognized as a cancer hallmark that is associated with cancer development and progression. In this study, the clinical relevance and significance of terminal fucose (TFG), by fucosyltransferase-1 (FUT1) in carcinogenesis and progression of cholangiocarcinoma (CCA) were demonstrated. TFG expression in human and hamster CCA tissues were determined using Ulex europaeus agglutinin-I (UEA-I) histochemistry. Normal bile ducts rarely expressed TFG while 47% of CCA human tissues had high TFG expression and was correlated with shorter survival of patients. In the CCA-hamster model, TFG was elevated in hyperproliferative bile ducts and gradually increased until CCA was developed. This evidence indicates the involvement of TFG in carcinogenesis and progression of CCA. The mechanistic insight was performed in 2 CCA cell lines. Suppression of TFG expression using siFUT1 or neutralizing the surface TFG with UEA-I significantly reduced migration, invasion and adhesion of CCA cells in correlation with the reduction of Akt/Erk signaling and epithelial-mesenchymal transition. A short pulse of EGF could stimulate Akt/Erk signaling via activation of EGF-EGFR cascade, however, decreasing TFG using siFUT1 or UEA-I treatment reduced the EGF-EGFR activation and Akt/Erk signaling. This evidence provides important insight into the relevant role and molecular mechanism of TFG in progression of CCA.


Results
tfG was over-expressed in hyperplastic bile duct epithelia and ccA tissues. TFG in 79 paraffin-embedded human CCA tissues was determined using UEA-I histochemistry. As shown in Fig. 1Aa, normal bile duct epithelia and hepatocytes were mostly negative for TFG. In contrast, the positive TFG signal with suprabasal staining in cytoplasm and stratification of nuclei were observed in the hyperplastic/dysplastic bile ducts (Fig. 1Ab). For cancerous tissues, CCA of the papillary type displayed cytoplasmic focal apical staining of TFG and a variation of nuclear feathers (Fig. 1Ac), whereas CCA of the non-papillary type mostly showed cytoplasmic staining of TFG with irregular nuclei (Fig. 1Ad). In addition, Kuffer's cells, red blood cells and endothelial cells were also positive for TFG. A high tfG score was an independent prognostic factor for poor survival of ccA patients. To signify the association between TFG expression and clinical features of CCA patients, a univariate analysis was applied. CCA patients were divided according to the median of TFG scores into low (TFG = 0-4, n = 43) and high (TFG = 6-12, n = 37). The stratification did not lead to any significant correlation of TFG expression with clinical features of age, gender, histotype, tumor size, regional lymph node metastasis and staging of the patients (Table S1). To further explore the clinical relevance of TFG on survival of CCA patients, a Kaplan-Meier plot and Log rank test were performed. The median survival time of CCA patients with a high TFG score was 167 days (95% confidence interval, CI; 132-202 days) and those with a low TFG was 302 days (95% CI; 247-357 days). As shown in Fig. 1D, CCA patients with high a TFG had a significantly shorter survival than those with a low TFG (P = 0.024). In addition, multivariate analysis using Cox-proportional hazard regression suggested the significance of TFG (P = 0.044) and histological types (P = 0.008) as independent prognostic factors for poor survival time of CCA patients (Table 1).
High expression of tfG related to carcinogenesis of ccA in hamster model. The Ov + NDMA-induced CCA hamster model closely mimics the pathogenesis of human CCA 13 and hence was used to study the association of TFG with carcinogenesis of CCA in this study. NDMA 14 is demethylated by hepatic microsomes to yield genotoxic reactive intermediates. Alkylation of DNA by these reactive metabolites results in a change in base sequence and deletion of one or more base pairs leading to the NDMA-induced carcinogenesis 14 . The UEA-I histochemistry of liver tissues from 4 groups of hamsters including untreated, Ov infected, NMDA treated and Ov + NDMA treated (CCA) groups at 1, 3 and 6 months post-treatment was performed. All treated groups exhibited bile duct epithelia with hyperplasia/dysplasia after treatment. In the NDMA treated group, CCA was found in 3/5 of the hamsters at 6-months post-treatment. In the Ov + NDMA treated group there was no CCA developed at 1 month, however, 4/5 hamsters in 3-months and 5/5 in 6-months post-treatment group developed CCA. No inflammation was found in the untreated-control group; however, the infiltrating cells were found in all treated groups but had no association with TFG scores. Normal bile ducts of the non-treated group were rarely positive for TFG whereas all hyperplastic/dysplastic bile ducts and the CCA ducts from the treated group were positive for TFG with differential signals (Fig. 2A). TFG reactivity was detected in the bile duct epithelia from all treated groups as early as 1-month post-treatment and the signal increased progressively with time of treatment. The TFG signals of each histological subtype of bile duct epithelia from all treated groups were quantitatively analyzed as shown in Fig. 2B.

Suppression of TFG expression significantly decreased progression of CCA cells. The endoge-
nous expression levels of TFG in 4 CCA cell lines were first determined by UEA-I cytofluorescent staining. CCA cell lines differentially expressed TFG (Fig. 3A); KKU-213 and KKU-214 had strong UEA-I signals whereas KKU-055 and KKU-100 showed weak signals. CCA cells with high TFG expression were selected for further elucidating the functional roles of TFG in CCA cells. www.nature.com/scientificreports www.nature.com/scientificreports/ The specific siRNA to FUT1, the enzyme responsible for TFG biosynthesis, was used to suppress TFG expression. siFUT1 effectively suppressed FUT1 mRNA expression of KKU-213 and KKU-214 cells to 30-40% of the scramble controls and the effect lasted until 72 h (Fig. 3B, P < 0.001). Suppression of FUT1 by siRNA significantly reduced TFG expression as shown by the reduction of TFG signals in UEA-I cytofluorescent stained cells (Fig. 3C). The effect of TFG on cell proliferation was first examined. As shown in Fig. 3D, treatment of siFUT1 had no effect on cell proliferation of either KKU-213 or KKU-214 cells as observed for 96 h. Suppression of TFG expression using siFUT1, however, significantly reduced cell migration as shown in the wound scratching assay (P < 0.01) and migration assay (P < 0.05) (Fig. 3E,F). siFUT1 treatment also significantly decreased numbers of invaded and adherent cells to approximately 30-50% of the scramble control cells (Fig. 3G,H). These results suggested the contribution of TFG in progressive phenotypes of CCA cells. cell surface tfG implicated the migration and invasion ability of ccA cells. TFG was shown to locate at the cell membrane as illustrated in UEA-I histochemistry of CCA tissues and UEA-I cytofluorescent stained CCA cells. Inhibition of TFG function by masking cell surface TFG using UEA-I was designed. As lectin can cause cell aggregation, the concentration of UEA-I that did not generate cell aggregation was first optimized. KKU-213 was treated with various concentrations (0.78 µg/ml -100 µg/ml) of UEA-I for 1 h and the formation of aggregated cells was observed under a microscope. As shown in Fig. S1A, aggregated cells were first observed at 12.5 µg/ml UEA-I and no cell aggregations were detected at 6.25 µg/ml UEA-I, therefore, 5 µg/ml UEA-I was used for cell treatment. No effect of cell aggregation was found under this treatment for 24 h (Fig. S1B). Wound scratching assays were performed in the presence or absence of 5 µg/ml UEA-I. As compared to the controls, UEA-I tfG mediated progression of ccA cells via activation of Akt and erk. Epithelial mesenchymal transition (EMT) is a process involved in driving metastatic phenotypes of cancer cells, e.g., by cell migration, and invasion. Whether TFG mediated progression of CCA cells involved EMT was next examined. KKU-213 and KKU-214 cells were treated with siFUT1 for 24 h and the expression of epithelial markers, E-cadherin and claudin-1, and mesenchymal markers, slug and vimentin, were determined using Western blots. As shown in Fig. 5A, E-cadherin and claudin-1 were increased while slug and vimentin were conversely decreased upon siFUT1 treatment. These results indicated TFG mediated migration and invasion of CCA cells via activation of EMT. In addition, expression of nuclear S100A4, a member of small Ca2+-binding proteins that is involved in invasiveness and metastasization of CCA cells 15,16 was also reduced in siFUT1 treated cells.
As E-cadherin, claudin-1, slug and vimentin are effector molecules modulated via activation of Akt, Erk and STAT3 which have been shown to be associated with progressive phenotypes of CCA 17-21 , whether these signal tfG favored eGf-eGfR activation. EGF-EGFR has been documented to be an active signaling pathway in CCA [22][23][24][25] . As Akt and Erk are downstream effectors of EGFR activation 26,27 and contribution of terminal fucose in EGF-EGFR interactions has been demonstrated 28-31 , therefore, TFG may favor EGF-EGFR activation and motivate Akt and Erk action. This postulation was next elucidated. Phosphorylation levels of EGFR at Y1068 as well as its downstream signaling, Akt and Erk, were used as the indicators of EGFR activation. KKU-213 and KKU-214 were treated with siFUT1 or 5 µg/ml UEA-I for 24 h prior to the stimulation of 100 ng/ml EGF in serum free medium for 10 min. The levels of phosphorylated EGFR, Akt and Erk, were determined using Western blotting. EGF treatment effectively activated EGFR action as the level of phosphorylated EGFR in EGF treated cells was higher than those of the untreated-control cells in both KKU-213 and KKU-214 (Fig. 6A). The phosphorylation level of EGFR, however, was reduced in siFUT1 treated cells. These results emphasized the positive involvement of TFG on EGF-EGFR activation. EGF treatment also activated Akt and Erk signaling, as phosphorylation levels of Akt and Erk were also increased in EGF treated cells and were decreased in siFUT1 treated cells. Similar results were observed in UEA-I treated cells (Fig. 6B). These results underscored the function of TFG on EGF-EGFR activation.

Discussion
A body of evidence indicates specific changes of glycan biosynthesis pathways in cancer that may interfere with regulation of cell adhesion, migration and proliferation. Fucosylation and sialylation are the major glycosylation changes of typical terminal sugars that modify proteins and mediate vital biological functions in cancer [32][33][34] . This study reports for the first time, the significance of terminal fucose glycan, TFG, in tumor development and progression of CCA. Patients with high TFG expression in tumor tissues exhibited shorter survival than those with low TFG expression. The involvement of TFG in carcinogenesis was demonstrated in the liver fluke-induced CCA hamster model. The significance of TFG in promoting CCA progression was emphasized in this study. www.nature.com/scientificreports www.nature.com/scientificreports/ Suppression of TFG expression by siFUT1 or diminishing of TFG action using UEA-I significantly reduced CCA cell migration, invasion and adhesion. The mechanism underlying these observations was shown to be in part via EGFR, activation of Akt and Erk that consequently stimulated the EMT process. The results underscore the roles of TFG in progression of CCA.
Lectin-based histochemistry has been used to identify aberrant glycosylation in several tumor tissues. In the present study, we demonstrated the aberrant expression of TFG in hyperplasia/dysplasia of bile duct epithelia and tumor tissues of CCA patients and Ov-induced CCA hamsters using UEA-I histochemistry. The aberrant expression of TFG during tumor development was shown in Ov-induced CCA hamster models and high expression of TFG in bile duct epithelia with hyperplasia/dysplasia and CCA indicate the significance of TFG in CCA development and progression. High expression level of TFG in tumor seems to be a common feature of cancer as the similar observation was also reported for parotid gland mucoepidermoid carcinoma 35 , colon cancer 36 and metastatic adenocarcinomas of lung 37 . Expression level of TFG did not relate with clinical features but had influence on survival of CCA patients. High expression of TFG in CCA tissues was an independent prognostic factor for short survival of patients. This observation is in agreement with the report in ovarian cancer 38 . Moreover, hepatocytes and HCC tissues rarely expressed TFG, thus TFG is possibly a marker for differential diagnosis between CCA and HCC. The UEA-I histochemistry of hamster tissues from the Ov-induced CCA group (Ov + NDMA) revealed similar results as observed in CCA tissues of patients. The involvement of TFG in carcinogenesis of CCA was supported by the fact that TFG was rarely expressed in normal bile ducts but strongly expressed in hyperproliferative bile ducts and CCA at all stages of carcinogenesis. TFG may therefore be associated with pathogenesis and severity of pathogenesis of bile duct epithelia regardless of the etiology as aberrant expression of TFG was observed in hyperplastic/dysplastic bile duct epithelia found in all treated groups. Moreover, the expression level of TFG was gradually increased with time of treatment. Notably, TFG was observed as early as 1-month post-treatment before CCA development suggesting the significant role of TFG in early carcinogenesis and possibly as an early marker of CCA. These findings together support that TFG may play a significant role in pathogenesis of bile duct epithelia and carcinogenesis of CCA. These finding also raise the possibility of using TFG as a potential biomarker for the early detection of CCA. Additionally, aberrant expression of TFG may be a unique feature of CCA as TFG was detected in 84% of hyperplastic/dysplastic bile ducts and 66% in CCA but only 10% in HCC. This discrepancy suggests the possibility of using TFG as a differential marker between CCA and HCC liver cancers.
Fucosylation is an important feature that contributes to features of progressive phenotypes of tumor cells 7 such as increased migration/invasion 39 , decreased adhesion 40 and uncontrolled tumor growth 41 . siFUT1, the enzyme responsible for generating TFG, effectively decreased TFG expression in correspondence with the decrease of cell migration, invasion and adhesion, but not cell proliferation of CCA cells. Suppression of TFG expression by siFUT1 or TFG action by UEA-I treatment suppressed progressive phenotypes of CCA cells in a similar manner. These data provide important insights into the relevant role of TFG in progression of CCA and may explain the poor prognosis of CCA patients with high TFG levels reported in the current study. As the majority of CCA cases in this study were associated with Ov infection, the findings may not fully reflect those associated with other risk factors. The involvement of TFG and tumor metastasis, however, may depend on tumor origins. A positive correlation between TFG expression and lymph node metastases was demonstrated in lung metastatic adenocarcinomas 37 . In contrast, a high expression of TFG was correlated with decreased metastasis in mouth mucosal carcinomas 42 . These observations, however, connect the cell surface TFG and metastatic ability of CCA cells.
Current searches do not reveal reports on the mechanism underlying TFG promoted metastasis. A body of evidence, however, indicated the significance of Akt, Erk and STAT3 in regulation of cell migration and invasion in CCA cells [17][18][19][20][21] . Western blot analysis in the present study indicated that Akt and Erk but not STAT3 was inactivated in siFUT1 treated cells. This finding was confirmed by a UEA-I masking experiment. Epithelial-mesenchymal transition (EMT) is a well-known process related with an increase of cell motility and can be mediated via Akt and Erk activation [43][44][45] . Western blot analysis indicated that suppression of TFG expression reversed cells from mesenchymal to epithelial as the mesenchymal markers slug vimentin and nuclear S100A4 were decreased and the epithelial marker claudin-1 was increased in siFUT1 treated cells. Similar results were obtained in CCA cells treated with the specific inhibitor of Akt and Erk. Decreased phosphorylation of Akt and Erk was corroborated with Western blot analysis of increased claudin-1 and decreased vimentin. These findings together connect TFG to the activation of Akt and Erk and their downstream signal, EMT.
Fucosylation affects the activity of tyrosine kinase receptors, especially EGFR. Activation of EGFR via ligand-induced dimerization leads to signaling cascade to activate Akt/Erk pathway 26,27 . Modification of EGFR by terminal fucose is necessary for ligand binding and intracellular signal transduction [28][29][30][31] . Overexpression of Figure 6. TFG favored EGF-EGRF activation. CCA cell lines were treated with (A) siFUT1 or scramble (sc) and (B) with UEA-I or without UEA-I (ctrl) for 24 h prior to the stimulation of EGF (100 ng/ml) in the serum free medium for 10 min. The levels of EGFR, phosphorylated EGFR and its downstream signaling Akt and Erk were determined using Western blotting. The activation of EGFR, Akt and Erk of the EGF treated cells were compared with those of the untreated cells by assigning the untreated cells = 1. The full-length blots are presented in Supplementary Fig. 3A,B. www.nature.com/scientificreports www.nature.com/scientificreports/ EGFR and its association to tumorigenesis and progression of CCA have been reported [22][23][24][25] . The present study illustrated the implication of TFG in EGFR activation and CCA progression through the stimulation of Akt/Erk and EMT. A short pulse of EGF markedly activated phosphorylation of EGFR and consequently stimulated Akt and Erk signaling. The activation of EGF, however was reduced when TFG expression was suppressed as shown in siFUT1 and UEA-I treated cells, resulting in the reduction of phosphorylated Akt and Erk. It has been suggested that TFG may facilitate EGF-EGFR binding by stabilizing electrostatic interactions 46 leading to conformational changes of EGFR and EGFR autophosphorylation 47 . The overall mechanism of TFG in promoting progression of CCA is proposed as illustrated in Fig. 7.

conclusions
The overall results depicted the clinical significance of TFG in CCA development and progression. TFG may be an early marker for pathogenesis of the bile duct and a targeted molecule to improve CCA therapy. The findings in this study together signified TFG as a progression promoting factor in CCA. Activations of EGFR/Akt/Erk and induction of EMT were demonstrated to be an underlying mechanism.

Materials and Methods
patient and hamster tissues. The paraffin-embedded liver tissues from 79 histologically proven CCA patients were obtained from the specimen bank of the Cholangiocarcinoma Research Institute, Faculty of Medicine, Khon Kaen University, Thailand. All patients in this study were classified as mass-forming with bile duct type (mucinous) intrahepatic CCA according to Banales et al. 9 . A majority of the cases (52 of 79, 66%) were non-papillary type CCA. All patients underwent curative intent surgery of CCA at Srinagarind hospital, Faculty of Medicine, Khon Kaen University, Thailand. Informed consent was obtained from each subject. The study design was approved by the Human Research Ethics Committee, Khon Kaen University (HE591063) and all research was performed in accordance with relevant guidelines and the Declaration of Helsinki.
The liver tissues from Ov-associated CCA hamsters were obtained as described previously 48 . The experimental hamsters were divided into 4 groups of 5 hamsters per group; the non-treated control, Ov infected, NDMA treated, and Ov + NDMA treated groups. Animals were euthanized at 1, 3, and 6 months post-treatment. UEA-I histochemistry was performed and compared between groups. The Animal Ethics Committee of Khon Kaen University approved this study (AEMDKKU1/2558) and all experiments were performed in accordance with relevant guidelines and regulations. cell culture and treatments. All human CCA cell lines were established from primary tumors of Thai CCA patients with intrahepatic mass forming CCA 49 . KKU-055 was from a mixed histological type CCA; both well and poorly differentiated tubular adenocarcinoma, KKU-100 was from a poorly differentiated tubular adenocarcinoma, whereas KKU-213 and 214 were originated from adenosquamous carcinoma CCA. All cell lines were obtained from the Japanese Collection of Research Bioresource (JCBR) Cell Bank, Osaka, Japan. These cell lines were cultured in F-12 Nutrient Mixture (Ham's F-12) (Gibco/Invitrogen, Calsbald, CA) supplemented with 10% fetal bovine serum (Gibco/Invitrogen), 1% antibiotic-anti-mycotic (Gibco/Invitrogen) and 14 mM NaHCO 3 . Cells were incubated in a humidified incubator at 37 °C, 5% CO 2 . www.nature.com/scientificreports www.nature.com/scientificreports/ CCA cells in 6 cm-dish were treated with Akt inhibitor, 1 µM of MK2206 (Sigma Aldrich, St. Louis, MO) or Erk inhibitor, 10 µM of PD98059 (Cell Signaling Technology, Danvers, MA) for 6 and 12 h. Cells had been washed before cell lysates were prepared for Western blotting analysis. For short-time stimulation of EGF, CCA cells (2 × 10 5 cells/well) were seeded into a 6-well plate. After 24 h treatment of siFUT1 or UEA-I, CCA cells were cultured in a serum-free Ham's F-12 media for 2 h then stimulated with 100 ng/ml of EGF (Peprotech, Rocky Hill, NJ) in serum free media for 10 min. Cells were washed immediately with cold PBS and harvested with lysis buffer for Western blotting analysis.
UEA-I histochemistry. The lectin histochemistry was performed as previously described 11 . Briefly, non-specific background was blocked with 0.5% periodate-treated BSA (tBSA) for 30 min at room temperature. The tissue sections were then incubated with 1:100 biotinylated UEA-I (Vector laboratories, Burlingame, CA) at room temperature for 2 h and with 1:100 streptavidin-conjugated HRP (Invitrogen, Camarillo, CA) for 40 min. The peroxidase activity was developed by diaminobenzidine tetrahydroxychloride solution (DAB; Dako, Glostrup, Denmark) and counterstained with Mayer's hematoxylin (Bio-optica; Milan, Italy). The TFG of each tissue section was scored under light microscopy according to the Allred scoring system 50 . In brief, the percentage of positive stained cells were graded as 0 = negative, 1-25% = 1, 26-50% = 2, 51-75% = 3 and >76% = 4. The intensities of the stained cells were scored as 1 = weak, 2 = moderate, 3 = strong. The TFG score was calculated as frequency x intensity and classified into low and high according to the medians of the TFG scores.
UEA-I cytofluorescent staining. CCA cells (5 × 10 4 cell/well) were seeded into a Matrigel-coated slide chamber and incubated at 37 °C, 5% CO 2 , overnight. After transfection with siFUT1, CCA cells were fixed with 4% paraformaldehyde at room temperature for 30 min and the non-specific binding was blocked with 0.5% tBSA. Cells were then incubated with 1:500 biotinylated UEA-I at 4 °C, overnight, and with 1:500 streptavidin-conjugated Alexa Fluor 488 (Invitrogen, Carlsbad, CA) at room temperature in the dark for 1 h. Nuclei were stained with 1:2,000 Hoechst 33342 (Invitrogen, Eugene, OR). The fluorescent images were graded with a fluorescence microscope (ECLIPSE Ni-U; Nikon, Tokyo, Japan).

Cell aggregation assay and UEA-I treatment.
As lectin can cause cell aggregation, the appropriate UEA-I concentration for UEA-I treatment was first optimized. KKU-213 cells (5 × 10 4 cells/well in a 24-well plate) were incubated with a two-fold dilution of 100 µg/ml unconjugated UEA-I (Vector laboratories, Burlingame, CA) in serum-free Ham's F-12, 1 h. The aggregated cells were observed under light microscopy (Fig. S1). The highest UEA-I concentration that did not generate cell aggregation was used for UEA-I treatment. For migration and invasion assays, CCA cells (3 × 10 5 cells/well) were plated into a 6-well plate for 24 h and then treated with 5 µg/ml UEA-I for 24 h before the migration and invasion assays were performed in the presence of UEA-I. For the wound scratching assay, wound closures were determined in the presence of 5 µg/ml UEA-I. Experiments performed in the absence of lectin were used as controls.
proliferation assay. CCA cell lines; KKU-213 (6 × 10 3 cells/well) and KKU-214 (4 × 10 3 cells/well) were seeded in a 96-well plate for 16 h. Cells were transfected with siFUT1 or siControl and cultured for 24, 48 and 96 h. Cell proliferation was measured using the sulforhodamine B (SRB) colorimetric assay 51 . At the end of incubation time, cells were fixed with cold 10% TCA in 1% acetic acid at 4 °C for 30 min. Fixed cells were stained with 0.4% SRB in 1% acetic acid at room temperature for 30 min and destained with 1% acetic acid. The stained cells were dissolved with 10 mM Tris-base solution (pH 10.5) at room temperature for 30 min on a shaker. The optical density was measured at 540 nm using a microplate reader (SUNRISE, Groedig, Austria).
Wound scratching assay. CCA cells (2-2.5 × 10 4 cells/well) were seeded into a 24-well plate and cultured overnight to reach 80-90% confluent cells. The cell monolayer was scratched and washed with PBS twice. The wound closure was detected at 0, 6, 12 and 24 h under a light microscope. The migrated distance was calculated by subtracting the distance between the width of wound at the indicated time from that at time 0 h then divided by the width of wound at time 0 h. Migration and invasion assay. Migrated and invaded cells were determined using Transwell ® Chambers (8.0 µm-pore size; Corning, Lowell, MA) as described previously 52 . The insert was pre-coated with 100 µl of 0.4 mg/ml of Matrigel overnight before seeding cells for the invasion assay. After 24 h of siRNA transfection, cells were resuspended in serum free media and seeded into the upper chamber. Cells (4 × 10 4 cells) were allowed to migrate or invade to the lower chamber containing completed media; 9 h for KKU-213 and 24 h for KKU-214. The migrated and invaded cells were fixed and stained with 0.5% crystal violet as described previously. The crystal violet-stained cells were counted under the light microscope using a 10X objective lens, 9 microscopic fields/ insert. Adhesion assay. A 48-well plate was pre-coated with 50 µg/ml of Matrigel, overnight. After blocking with 2% BSA in PBS for 1 h, siRNA-transfected cells (3 × 10 4 cells/well for KKU-213 and 2 × 10 4 cells/well for KKU-214) were seeded and allowed to adhere for 2 h at 37 °C, 5% CO 2 . Non-adhered cells were carefully removed. The adhered cells were fixed with 4% paraformaldehyde for 30 min and stained with 0.5% crystal violet. The stained cells were dissolved with 2% SDS and the absorbance was measured at 540 nm. www.nature.com/scientificreports www.nature.com/scientificreports/ LightCycle 480 ® SYBR green I master (Roche Diagnostics, Mannheim, Germany). The conditions of amplifications were setup according to the LightCycle480 ® instrument protocol as 95 °C for 5 min, followed by 35 cycles of 95 °C for 10 sec, 60 °C for 10 sec and 72 °C for 10 sec and finally 40 °C for 10 sec. The expression of FUT1 mRNA was quantitated from the triplicated samples using LightCycle480 ® Relative Quantification Software (Roche Diagnostics). β-actin was used as an internal control. The primer sequences of FUT1 and β-actin were: FUT1 as F: 5′-TGAGGGATCACTGCCAAAATG-3′, R: 5′TCTTGGCAGTTTATGAGCTTTAAAAA-3′ and β-actin as F: 5′TCGTGCGTGACATTAAGGAG-3′, R: 5′GAAGGAAGGCTGGAAGAGTG-3′. Transient knockdown of FUT1 by siRNA. Specific siRNA to FUT1 mRNA was used to suppress FUT1 expression in CCA cells. The sequences of siFUT1 were: sense strand; 5′-GUAAUCUUCUUCCUCCAUA-3′ and anti-sense strand: 5′-UAUGGAGGAAGAAGAUUAC-3′. CCA cells (2 × 10 5 cells/well) cultured in a 6-well plate overnight were transfected with 100 pmole of siFUT1 using 2 µg/µl of Lipofectamine 2000 according to the recommendation of manufacturer (Invitrogen, Carlsbad, CA). Cells were cultured in HAM's F-12 media with 10% FBS and harvested after 24 h. Control experiments were performed using siControl (Negative control siRNA#1027310, Qiagen, Chatsworth, CA).

SDS-pAGe and western blotting analysis.
Cells were lysed with lysis buffer (1% NP-40, 0.5% Na-deoxycholate, 0.1% SDS, 2 mM EDTA, 150 mM NaCl and 50 mM Tris-HCl pH 7.4) containing phosphataseand protease-cocktail inhibitors (Roche, Mannheim, Germany) and incubated at 4 °C for 15 min. Cell lysates were obtained after centrifugation at 12,000 xg, 4 °C for 15 min and total protein was quantified by the Quick Start TM Bradford protein Assay (Bio-Rad, Hercules, CA). Protein lysates (20 µg/well) were separated in a 10% SDS-PAGE 53 and transferred on to a PVDF membrane (Amersham TM , Buckinghamshire, UK) using Bolt and Mahoney transferring buffer 54 . For S100A4, the nuclear fraction was extracted according to Andrews et al. 55 . Nuclear lysate of 20 µg/well was loaded onto a 15% SDS-PAGE, and histone H1 was used as the internal control.
The antibodies used for Western blot were anti-β-actin (A5441) (Sigma Aldrich, St. Louis, MO), anti-Erk1/2, anti-p-Erk1/2, anti-STAT3 (C20), anti-p-STAT3 (Y705), anti-p-STAT3 (S727) (S727-R) (Santa Cruz Biotechnology, Santa Cruz, CA), anti-E-cadherin, anti-Cludin-1, anti-Slug, anti-N-cadherin, anti-Vimentin and anti-EGFR (Y1068; D7A5) (Cell Signaling Technology, Danvers, MA), S100A4 (Abnova, Taipei, Taiwan). The signal was developed by the ECL TM Prime Western Blotting Detection System (Amersham TM , Buckinghamshire, UK). The imaging was obtained using an ImageQuant LAS 4000 mini image analyzer and analyzed by ImageQuant TM TL analysis software (GE Healthcare, Buckinghamshire, UK). Statistical analysis. Statistical analysis was performed using the SPSS version 17.0 (SPSS Inc., Chicago, IL) and GraphPad Prism ® 5.0 software (GraphPad Software, Inc., La Jolla, CA). Survival analysis was performed using Kaplan-Meier statistics with P-values computed by log-rank tests. A subsequent backward stepwise multivariate analysis using a Cox regression model was used to signify the variables that were associated with survivals of patients with CCA. The chi-square test was used to compare the differences in the clinical features of patients. The two-tail Student's t-test was used to compare the differences between treatments and controls. At least two independent experiments were performed. All quantitative data was mean ± SD of the triplicate samples from one of two independent experiments. P < 0.05 was considered as the statistical significance.