Pairs of cancerous and normal tissues diagnosed by a pathologist were collected from sixteen patients with MF type ICC who underwent radical or palliative surgical procedures at Gyeongsang National University Hospital (GNUH) during the period 2006–2009. The clinical data collected included age, gender, differentiation, operation type and TNM stage. None of the patients received preoperative chemotherapy or radiotherapy. Informed patient consent was obtained for all of the cases. The use of the tissue samples for this project was approved by the GNUH Institutional Review Board. Cancerous lesions and the corresponding normal bile duct tissues were excised during surgery and stored at −70°C for proteomic analysis and further study.

Tissue samples (150–200 mg) were homogenized on ice in 1 ml homogenization buffer (50 mM Tris-HCl, pH 7.2) containing a protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA). The homogenate was centrifuged, followed by TCA precipitation. The protein precipitate was resuspended in lysis buffer (8 M urea, 4% CHAPS, 40 mM Tris-base, 100 mM DTT and 2% (w/v) ampholyte) and the protein concentration was measured using Bradford method with Protein assay kit (Bio-Rad Laboratories, Hercules, CA). Equal amounts of protein extracts of each tissue samples (50 μg for silver-stained gels or up to 0.5 mg for Coomassie-stained gels) were subjected to isoelectric focusing (pH 4.0 to 7.0) and sequentially a gradient SDS-polyacrylamide gel (7.5–17.5%) electrophoresis as previously described (15).

Separated two-dimensional gels were fixed and then pretreated in a solution of 0.02% sodium thiosulfate pentahydrate. After washing twice with deionized water, the gels were impregnated in a solution of 0.2% silver nitrate and 0.075% (v/v) formaldehyde, transferred to a developing solution [0.06% (v/v) formaldehyde, 2% sodium bicarbonate, and 0.0004% sodium sulfoxide]. For Coomassie staining, the gels were fixed and rinsed twice in 2% phosphoric acid. The gels were placed in equilibration (18% ethanol, 2% phosphoric acid and 15% ammonium sulfate) followed by overnight staining with Coomassie Brilliant Blue G-250 solution. The gels were scanned using a high resolution scanner (GS-800 Calibrated Imaging Densitometer, Bio-Rad Laboratories) and analyzed using PDQuest™ version 8.0. software (Bio-Rad Laboratories). The intensity of each spot was quantified by calculating the spot volume after normalization of the gel image.

Selected protein spots were excised from 2-DE gels that had been stained with Coomassie blue G-250. Gel pieces first were destained with water/acetonitrile (1:1 v/v) then dehydrated in acetonitrile followed by rehydration in 10 mM DTT/0.1 M ammonium bicarbonate for 45 min at 56°C. After cooling the tubes to room temperature and removing the liquid, 55 mM iodoacetamide in 0.1 M ammonium bicarbonate was added and the tubes were incubated for 30 min at room temperature in the dark. The iodoacetamide solution was removed and the gel particles were washed with 0.1 M ammonium bicarbonate and acetonitrile and then dried in a vacuum centrifuge. The dried gel pieces were incubated in freshly prepared digestion buffer containing 50 mM ammonium bicarbonate, 5 mM CaCl₂, and 12.5 ng/μl trypsin overnight at 37°C. The supernatant was then collected and the digested peptides extracted three times in 5% formic acid/acetonitrile (1:1 v/v).

The digested peptides were redissolved in a solution containing water, acetonitrile, and trifluoroacetic acid (93:5:2 by volume) and sonicated for 5 min in a bath sonicator. Target preparation was carried out by the ‘solution phase nitrocellulose method’ described by Landry et al (16). Briefly, a saturated solution of α-cyano-4-hydroxycinnamic acid (CHCA; ~40 mg/ml) and nitrocellulose solution (20 mg/ml) were prepared separately in acetone and mixed with 2-propanol at a ratio of 2:1:1, respectively. Internal calibrants (50–200 fmole each), des-Arg-bradykinin (monoisotopic mass = 904.4681), and neurotensin (monoisotopic mass = 1672.9715) were added to the mixture. The matrix solution was mixed with the sample at a 1:1 ratio, and 1 μl of mixed sample was spotted onto a MALDI plate. The dried spots were analyzed with a Voyager-DE STR MALDI-TOF mass spectrometer (Applied Biosystems, Foster City, CA). The identity of the proteins was assigned by comparing the observed mass fingerprint with the predicted mass fingerprint of proteins in the SWISS-Prot and NCBI databases using the MS-Fit search program (17).

Lysates (40 μg) from ICC and its corresponding normal bile duct tissues from 16 patients were separated by on a 15% SDS-polyacrylamide gel. The separated proteins were transferred to a PVDF membrane by electroblotting. The membrane was blocked for 1 h in TBST solution (Tris-buffed saline, 20 mM Tris-HCl, pH 7.6, 137 mM NaCl, and 0.1% Tween-20) containing 5% skim milk. After blocking, the membranes were washed in TBST twice for 10 min each and incubated with anti-FABP5 rabbit polyclonal antibody for 1 h at room temperature. After additional washes with TBST, the membranes were incubated with secondary anti-rabbit IgG (1:10,000) for 1 h. After final washes, proteins were visualized by the enhanced chemiluminescence method (Pierce, Rockford, IL). β-actin was used as a loading control in the stripped blot.

Formalin-fixed, paraffin-embedded tissue blocks from 43 patients with MF-ICC were collected, sectioned at 4 μm and mounted on slides. The slides were deparaffinized and rehydrated in a graded alcohol series, and pretreated with 10 mM sodium citrate. Immunostaining was performed with a rabbit polyclonal FABP5 antibody (1:500) using an LSAB kit (Dako, Glostrup, Denmark). Antigen retrieval was facilitated by microwaving the tissue for 15 min, and the staining procedure was carried out according to the standard avidin-biotin-peroxidase complex system previously described (18). Immunohistochemical staining was interpretated and categorized by an arbitrary semi-quantitative scale as 0 (negative staining), +1 (0~20% positive), +2 (20~40% positive), or +3 (>40% positive) by pathologists. The staining intensity was also simply categorized into ‘weak’ (0 to +1) and ‘strong’ (+2 to +3) groups. Tumor size, degree of differentiation, lymph node metastasis, vascular invasion, perineural invasion and multiplicity were included in the clinicopathological information (Table III).

HuCCT1 cell line was purchased from the Health Science Research Resources Bank (Osaka, Japan). HuCCT1 cells were cultured with RPMI-1640, 10% fetal bovine serum (FBS) and 1× penicillin/streptomycin at 37°C in 5% CO₂ incubator. Human hepatoma cell line, Hep3B were maintained in DMEM supplemented with 10% FBS and 1× antibiotics in a humidified atmosphere containing 5% CO₂ and 95% air at 37°C. Human FABP5 cDNA was cloned into pcDNA3.1 using standard RT-PCR procedures. The empty vector was used as a control for possible effects caused by the transfected plasmid. The constructs were transfected into Hep3B cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol and stable FABP5-expressing cell lines were selected by G418 (2 mg/ml).

The MISSION shRNA bacterial glycerol stock containing 5 anti-FABP5 shRNA sequences and the SHC002 shRNA Control were purchased from Sigma-Aldrich. The shRNA constructs were transfected into HuCCT1 cells using a reagent (Lipofectamine 2000; Invitrogen) according to the manufacturer’s instructions. Stable shRNA-expressing cell lines were established and selected by puromycin (2 μg/ml). Results from shRNA TRCN0000059720 are presented in this report.

For proliferation assay, cells were placed in a 6-well plate at a concentration of 1×10⁵ cells/well. After incubation for 1–3 days, the viable cells were counted with a hemocytometer after trypan blue staining. Invasion assays were performed on 24-well transwells (Costar, Cambridge, MA, USA) with polycarbonate filters (8 μm pore size). The transwells for invation assays were coated with a uniform layer of BD Matrigel™ Basement Membrane Matrix (BD Biosciences, Bedford, MA, USA). The stable cell lines were resuspended in RPMI-1640 or DMEM containing 10% FBS and seeded into the upper wells (1×10⁵ cells/well) and incubated at 37°C for 24 h. Invaded cells were fixed in 4% PFA, stained with DAPI, and counted under the fluorescent microscope at ×100 magnification for 5 random fields.

Statistical analysis was performed by the Chi-squared test using the SPSS program (version 11.0). All P-values were based on a two-tailed statistical analysis, and P-values of <0.05 were considered statistically significant.

We analyzed the proteomic profiles of the ICC tissue and corresponding normal bile duct tissues from 16 patients to establish a tumor specific protein expression profile. The upregulated or downregulated protein expressions between the ICC tissue samples and their healthy counterparts were evaluated using PDQuest™ software. By making comparison with PDQuest software quantification, the 151 protein spots that showed a statistically meaningful expression difference (P<0.05) were selected. The spots that showed more than a 3-fold increase or decrease of intensity were defined as being the up- or downregulated proteins, respectively. According to this definition, among the 67 candidate protein spots, 50 proteins were upregulated and 17 proteins were downregulated in ICC tissue compared to the non-cancerous bile duct tissue. The protein spots were identified using MALDI-TOF-MS and the MS-FIT search program. The identities of the proteins were confirmed by comprehensively comparing the corresponding experimental value of isoelectric point (pI), molecular weight (MW), the number of matched peptides, and the sequence coverage to known proteins in the SWISS-PROT and NCBI databases. The lists of proteins that were found to be up- or downregulated in ICC tissue are shown in Tables I and II, respectively.

Among the identified proteins, the spot 6010 showed markedly increased intensity in all of investigated patients with an average increase of 4.49-fold (Fig. 1A). Statistical analysis of the spots confirmed that spot 6010 was significantly upregulated (P=0.0125) and subsequently identified as FABP5 by MALDI-TOF-MS, with a molecular mass of 15.2 kDa and a pI 6.6. The sequence coverage of protein isolated from peptide mass matching in the program was 43.7% (Table I).

Expression levels of FABP5 were also examined by western blot analysis in paired ICC and normal bile duct tissues obtained from 16 patients. Nine representative samples of western blotting for FABP5 expression are shown in Fig. 1B. Western blot analysis of FABP5 expression in the same sample as those used for 2-DE confirmed that FABP5 was highly expressed in ICC samples, while the normal bile duct from the same patients showed undetectable FABP5 levels. As shown in Fig. 1C, the upregulation ratio of FABP5 in ICC to normal bile duct tissues was about 5-fold. We next compared expression of FABP5 in cholangiocarcinoma cell lines (Fig. 1D). Western blot analysis revealed that all five cholangiocarcinoma cell lines highly expressed FABP5.

FABP5 was primarily observed in the cytoplasm of epithelial cells in both cancerous and non-cancerous tissues (Fig. 2). Immunohistochemical staining was interpretated and categorized by an arbitrary semi-quantitative scale as 0 (negative staining), +1 (0~20% positive), +2 (20~40% positive), and +3 (>40% positive) by pathologists. All normal bile ducts stained weakly for FABP5 (0 or +1), while the strong staining intensity was verified in ICC (+1 to +3).

In order to investigate the clinical significance of FABP5 expression, we performed immunohistochemical analysis on tissue sections from a larger population of ICC patients. A total of 43 patients with MF type ICC who underwent several types of major hepatectomy and lymph node dissection, either alone or in conjunction with extrahepatic bile duct resection were enrolled in this study. There were 33 men and 10 women with a mean age of 63.21 years (range, 47–80 years). If the tumor seemed to invade adjacent organs grossly in the operative field, combined resection was performed to achieve negative resection margins. Clinical data and outcome of the patients were collected from medical records retrospectively. The results of the study revealed weak FABP5 staining (0 or +1) in 18 of the 43 cases, which were classified as the weak positive group. The remaining 25 cases exhibited strong FABP5 staining and were classified as the strong positive group (+2 or +3). The expression levels of FABP5 were compared with the clinical and pathological characteristics of each case of MF type ICC, including size, differentiation, lymph node metastasis, vascular invasion, perineural invasion, multiplicity and stage (Table III). All eight patients with lymph node metastasis showed strong staining intensity against FABP5. Univariate analysis revealed that tumor size (P=0.047), lymph node metastasis (P=0.013), angio-invasion (P=0.032), and staging (P=0.007) significantly correlated with expression levels of FABP5.

Suppression of FABP5 expression in oral squamous cell carcinoma has been reported to inhibit proliferation and invasion (19). Therefore, we first investigated whether FABP5 was critical for the proliferation of cholangiocarcinoma cells. We used a short hairpin RNA (shRNA) to silence the expression of FABP5. HuCCT1 cells were transfected with control shRNA or FABP5 specific shRNA plasmid. After puromycin (2 μg/ml) selection, western blot analysis confirmed that FABP5-shRNA clones (named FABP5-KD #2, FABP5-KD #8, FABP5-KD #11) showed an almost complete silence of FABP5 expression compared to control shRNA clones (Fig. 3A). Depletion of FABP5 expression significantly (P<0.01) inhibited the cell proliferation rate of HuCCT1 cells without significant increase of dead cells (Fig. 3B). To examine the effects of FABP5 depletion on tumor cell invasion in FABP5-depleting HuCCT1 clones, we then performed in vitro invasion assay. FABP5 depletion caused a significant reduction in the invasiveness of FABP5-depleted HuCCT1 cells (Fig. 3C).

To confirm the enhancing effects of FABP5 on cell proliferation and invasion, we also examined the effects of the FABP5 overexpression on the FABP5-negative hepatoma cell line Hep3B by stable transfection with a plasmid encoding the FABP5 protein. After G418 (2 mg/ml) selection, the expression of FABP5 in selected clones (named FABP5 #1, FABP5 #2, FABP5 #3) was determined by western blotting (Fig. 4A). Cell proliferation rate, counted by trypan blue assay (Fig. 4B) and Matrigel invasion assays (Fig. 4C) of FABP5 overexpressing cell clones were significantly increased compared to the control vector expressing clones. Taking these results together, the expression of FABP5 seems to be involved in the increased cell proliferation and invasion of ICC cells.