Adipocytes promote cholangiocarcinoma metastasis through fatty acid binding protein 4

Human adipose tissues from the breast cancer patients (> 2 cm away from tumors) and the omentum majus adipose from the nontumorous patients were provided by the First Affiliated Hospital of Soochow University less than 1 h after surgery. This study was approved by the Institutional Review Board of Soochow University. Under aseptic conditions, the adipose tissues were washed with icy-cold PBS containing 50 μg/ml gentamicin, cut into small pieces with a diameter of 2 mm, and then centrifuged briefly to remove red blood cells and debris. Adipose tissues then were incubated in DMEM culture medium (80 mg adipose/mL DMEM medium) for 24 h at 37 °C under 5% CO₂. Adipose tissue extracts were obtained by removing lipochondrions via 0.45 μM filter filtration (EMD Millipore, Billerica, MA, USA). After incubation with the extracts for 24 h, the cancer cells were harvested for subsequent experiments.

A differentiation assay was performed as previously described [30]. Briefly, 3 T3-L1 cells were first incubated in differentiation media I consisting of DMEM, 10% fetal bovine serum (FBS), 0.25 μg/mL insulin, 1 μM dexamethasone, and 0.5 mM 3-isobutyl-1-methylxanthine (Sigma-Aldrich, St. Louis, MO, USA). Two days later, the cells were cultured in complete culture medium and 0.25 μg/mL insulin (differentiation media II) until lipid droplets formed. Oil Red O staining (Abcam, Cambridge, MA, USA) was employed to determine differentiation status as previously described [30]. The supernatant was collected prior to lipid droplet staining through centrifugation; the supernatant was used in subsequent experiments.

The human CCA cell lines RBE and Hccc-9810 were maintained in RPMI 1640 containing 10% FBS, L-glutamine (2 mM), non-essential amino acids (2 mM), penicillin (100 U/mL), and streptomycin (100 U/mL) (Gibco, Grand Island, NY, USA) at 37 °C under 5% CO₂ atmosphere. BMS309403 (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in DMSO (Sigma-Aldrich, St. Louis, MO, USA) at 20 mM to prepare a stock solution. In the subsequent studies, 20 μM BMS309403 was used as working solution.

Full-length cDNA of an adenovirus-carrying human FABP4 was constructed for in vivo expression of FABP4 (ViGene, Shandong, China). Sub-confluent cells were mock-infected or infected with adenovirus in complete medium for 12 h at 37 °C followed by incubation in fresh complete medium for additional 24 h to 48 h for subsequent experiments. The infection efficiency was confirmed by detection of FABP4 mRNA by quantitative real-time PCR (qRT-PCR).

3 × 10⁵/mL exponentially grown CCA cells were split and seeded into 6-well tissue culture plates and allowed to form a confluent monolayer. After a corresponding treatment, the monolayer was scratched with the tips of 200 μL sterile pipettes, washed with PBS to remove floating and detached cells, and cultured with fresh medium supplemented with 2% FBS. To assess cell migration, we designated five randomly chosen points for each treatment group for photographing at 0, 12, 24, and 48 h by using a microscope (40×, Olympus, Tokyo, Japan) equipped with a digital camera (Canon, Tokyo, Japan); the images were analyzed using Image J software (National Institutes of Health, Bethesda, MD, USA). Cell migration was presented as means ± sd of wound width, and the values were compared with those at the starting time point in the control group.

The invasion assay was performed using 24-well Matrigel Invasion Chambers (pore size, 8 μm; Corning, Tewksbury, MA, USA). Inserts were pre-coated with 50 μL of Matrigel (1:8 dilution; BD Biosciences, San Jose, CA, USA) and then polymerized at 37 °C for 4 h before the experiment. 5 × 10⁴/100 μL CCA cells were prepared as a single cell suspension in serum-free medium containing 0.2% bovine serum albumin (BSA) and seeded onto the upper chambers. The lower chamber was filled with complete culture medium supplemented with/without adipose tissue extracts or filled with culture supernatant from differentiated 3 T3-L1 cells. After incubation for 24 h, the cells on the surface of the upper chambers were scraped off. The invading cells were fixed with 4% paraformaldehyde, stained with Giemsa staining solution at room temperature for at least 4 h, and photographed under a microscope.

Cells were grown on 35 mm glass bottom culture dishes (Nest Scientific, NJ, USA). After the corresponding treatments, the cells were washed with PBS and fixed with 4% paraformaldehyde, stained with 20 μg/mL Bodipy 493/503 (ThermoFisher Scientific, Grand Island, NY, USA) for 1 h at room temperature, and washed twice with PBS. The nuclei were visualized using Hoechst 33342 (0.5 μg/mL), and the stained cells were observed under a confocal scanning laser microscope (Olympus, Tokyo, Japan). The relative fluorescence intensity was analyzed using Image J software.

Cells were harvested and lysed in 50 uL of lysis buffer containing protease inhibitor cocktail (Roche Life Science, Indianapolis, IN, USA) for 30 min on ice. Total protein (50 μg) from each lysate was fractionated by 10% SDS-PAGE and transferred onto polyvinylidene fluoride microporous membranes. After blocking with 5% nonfat milk in PBS-Tween-20 for 2 h at room temperature, the membranes were incubated with primary antibody overnight at 4 °C and then incubated with corresponding secondary antibodies. GAPDH was used as loading control.

The primary antibodies used were rabbit anti-FABP4 (EPR3579, Abcam, Cambridge, MA, USA, 1:1000), rabbit anti-claudin 1 (Abcam, Cambridge, MA, USA, 1:1000), rabbit anti-occludin (EPR8208, Abcam, Cambridge, MA, USA, 1:50,000), rabbit anti-E-cadherin (EP700Y, Abcam, Cambridge, MA, USA, 1:10,000), rabbit anti-SNAIL (Abcam, Cambridge, MA, USA, 1:1000), rabbit anti-Smad3 (EP568Y, Abcam, Cambridge, MA, USA, 1:1000), rabbit anti-β-catenin (Cell Signaling Technology, Danvers, MA, USA, 1:1000), rabbit anti-MMP2 (Cell Signaling Technology, Danvers, MA, USA,1:1000), rabbit anti-MMP9 (Cell Signaling Technology, Danvers, MA, USA,1:1000), and anti-β-actin (Beyotime Biotechnology, Haimen, China). The secondary antibodies used were goat anti-mouse and anti-rabbit horseradish peroxidase-conjugated antibodies (1:1000, Beyotime Biotechnology, Haimen, China). Protein expression was quantified using Image J software.

Total RNA from each group of RBE cells was extracted using TRIzol (Invitrogen, Grand Island, NY, USA) according to the standard TRIZOL method. First-strand cDNA was synthesized from 1 μg of RNA per sample by using Transcriptor First Strand cDNA Synthesis Kit (Roche Life Science, Indianapolis, IN, USA). Real-time PCR was performed on an Applied Biosystems ViiA 7 RT-PCR by using SYBR Green (Thermofisher Scientific, Grand Island, NY, USA). Table 1 shows the primers for FABP1–7 and the internal control GAPDH. Gene expression was calculated using the 2^-△△CT method.

Five-week-old female outbred nude mice were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (Shanghai, China) and housed in a pathogen-free facility with constant temperature and humidity control. CCA metastasis mouse models were constructed by injecting 1 × 10⁶/100 μL RBE cells through the tail vein. Briefly, the mice were randomly divided into five groups, which were separately injected with (1) parental RBE cells, (2) RBE cells co-cultured with human omentum adipose for 24 h, (3) RBE co-cultivated with human adipose and FABP4 inhibitor BMS309403 for 24 h, and (4 and 5) RBE cells infected with control adenovirus and FABP4 adenovirus for 24 h, respectively. For short-term imaging assay, all cells were infected with green fluorescent protein (GFP)-tagged adenovirus for additional 24 h, collected and washed twice with cold PBS, and then re-suspended in cold PBS. 24 h after injection, colonization of RBE cells in organs of nude mice was visualized using an in vivo imaging system (Kodak, Effingham, IL, USA). All photos were taken under the same condition. The relative fluorescence intensity of the tissues was calculated using Image J analysis software. For long-term observation of metastasis, all mice were euthanized 1 month after injection through the tail vein, and their livers and lungs were extirpated and used in pathological examination as described below. The design and implementation of this study were approved by the Ethics Committee of Soochow University.

All of the animals were euthanized 1 month after injection through the tail vein. The liver, heart, kidney, and lung of each mouse were excised. Tissues were fixed in 10% neutral-buffered formalin and embedded in paraffin for H & E staining and immunohistochemistry as previously described [31] to determine the metastasis of RBE cells in vivo.

For immunohistochemistry, 4 μm paraffin sections were deparaffinized and heat-treated with citrate buffer (pH 6.0) for 3 min according to an antigen retrieval protocol. Non-specific binding sites were subsequently blocked with 5% BSA for 30 min. The sections were incubated overnight with primary antibodies (1:200 dilution) at 4 °C. Subsequently, the sections were incubated with secondary antibody dilution (Zhongshan Golden Bridge Biotechnology, Beijing, China) at 37 °C for 1 h followed by diaminobenzidine (DAB) substrate detection, washing, hematoxylin staining, dehydration and mounting. Images of the tissue sections were captured using an Olympus optical microscope (Tokyo, Japan).

For Oil Red O staining, 10 μm frozen sections made from fresh frozen tissues were air-dried at room temperature, incubated in fresh Oil Red O (Abcam, Cambridge, MA, USA) for 10 min, and rinsed in water. Images of the slides were viewed and captured under an Olympus optical microscope (Tokyo, Japan).

Tumor microenvironment, which consists of tumor-associated extracellular matrix and stromal cells, is important for tumor growth and progression [32]. To determine the effects of adipose tissues, a major component of tumor microenvironment, on CCA invasion and migration, we incubated the human CCA cell line RBE with extracts of adipose tissues obtained from different species or origins. Figure 1a showed that the invasion of RBE cells, cultured in the presence of the four human adipose tissue extracts, was significantly increased compared with those of the control group (P < 0.05), indicating that the acquired invasiveness of CCA cells was due to cultivation with adipose tissue extracts. Moreover, enhanced migration of RBE cells cultured with adipose tissue extracts was also detected by wound healing assay (Fig. 1b), in which the enhanced migration was indicated by obvious shrinking of “wound” width 24 h following scratching compared with that in the control group (P < 0.05). Enhanced invasion and migration were also observed in RBE cells grown in extracts of adipose tissue from C57/BL6J omentum (Fig. 1c and d).

Adipose tissues consist mainly of mature adipocytes, which play potential roles in ovarian cancer and breast cancer progression [18, 32]. In the present study, mouse pre-adipocyte 3 T3-L1, which can differentiate into mature adipocytes [30], was utilized to determine the effects of adipocytes on RBE metastasis in vitro. As shown in Fig. 1e, pre-adipocyte 3 T3-L1 could be induced to differentiate into mature adipocytes, characterized by the presence of typical lipid droplets in the cytoplasm [30]. Compared with RBE cells grown in complete growth medium, RBE cells co-cultured with undifferentiated 3 T3-L1 exhibited a stronger invasive tendency (P < 0.05). Moreover, co-cultivation with fully differentiated 3 T3-L1 cells significantly increased the number of trans-membrane cells, in contrast to pre-adipocyte-co-cultured RBE cells (P < 0.05). Overall, these results above demonstrated for the first time that crosstalk of CCA cells with adipose tissues and mature adipocytes was essential for CCA metastasis in vitro.

EMT is a pivotal mechanism of tumor metastasis in many epithelial malignancies [7, 20]. In the present study, besides enhanced RBE metastasis in vitro, altered morphologies, from cobblestone-like to typical spindle-like, were found in RBE cells cultivated in adipose tissue extracts (Fig. 2a). To determine whether adipose tissues influence CCA metastasis by regulating EMT signal pathways, we detected related protein markers in RBE cells cultured in the presence or absence of adipose tissue extracts via Western blot assay. Figure 2b and c showed the enhanced EMT phenotype markers in RBE cells cultivated in adipose tissue extracts, including upregulation of matrix metalloproteinase (MMP)-2, MMP-9 and SNAIL, as well as downregulation of E-cadherin and β-catenin. These findings are consistent with the results of Cadamuro et al., who reported that CCA bile ducts expressed several EMT phenotype markers [33].

Furthermore, we detected significantly enhanced expression of transforming growth factor β1 (TGF-β1) in RBE cells cultured with extracts of adipose tissues from different origins (Fig. 2b, Additional file 1 Figure S5A). TGF-β1 is known to promote EMT by inducing SNAIL, and is very important in reprogramming of cellular energy in cancers to satisfy the increased energy demand during EMT to sustain enhanced mobility [34]. Triglycerides (one molecule glycerol plus 3 molecules FA), being the major cellular energy source, are mainly stored in mature adipocytes [35‐37]. To investigate whether adipose tissue extracts promoted EMT phenotype of CCA cells by affecting cellular energy homeostasis, we examined extracellular glycerol and nonestesterified fatty acid (NEFA) in RBE cells before and after cultivation with different adipose tissue extracts. As shown in Fig. 2d, e and Additional file 1 Figure S1, significantly increased amounts of glycerol and NEFA were detected in the adipose tissue extract-containing culture medium (P < 0.05), compared with the regular culture medium. In addition, we found that it is NEFA, rather than glycerol, that was significantly reduced after cultivating RBE cells in culture medium containing adipose tissue extracts (Fig. 2d and e, P < 0.05). Moreover, enhanced lipid accumulation was observed in RBE cells cultivated in adipose tissue extracts (Fig. 2f), consistent with findings suggesting that FA plays a key role in certain pathological processes by promoting cancer cell lipogenesis [38, 39]. Overall, the results above indicated that adipose tissue facilitates EMT of CCA cells in vitro, likely by providing NEFAs as energy source.

The influence of adipose tissues on CCA metastasis was further confirmed in vivo in CCA metastasis Balb/C nude mouse model (Fig. 3a). First, 48 h after inoculation, distribution and accumulation of RBE cells were visualized using an in vivo imaging system. As shown in Fig. 3b and c, significantly increased fluorescence signals were detected in the liver and kidney of mice injected with RBE cells cultivated in adipose tissue extracts, compared with mice injected with RBE cells under normal culture conditions (P < 0.05). However, the chemical compound BMS309403, which could specifically inhibit FA uptake in vitro [40], compromised the adipose-induced CCA metastasis in vivo, especially in the kidney (P < 0.05).

One month after inoculation, 40% (2/5) of the mice injected with RBE cells cultivated in adipose tissue extracts developed liver metastasis, in which the hepatocytes manifested typical pathologic characters, such as irregular shape, variable sizes, hyperchromatic and pleomorphic nuclei, nuclear disintegration and increased proliferation (Fig. 3d). Moreover, no distance metastasis and obviously suppressed cancer proliferation were observed in BMS309403 plus adipose tissue-cultivated RBE injected mice, compared with adipose tissue extract-cultivated RBE group (Fig. 3d). The results above are consistent with the in vitro data showing that adipose tissue can promote CCA metastasis.

Based on the data above, we speculated that FAs released by adipocytes might at least be involved in adipose-induced CCA metastasis. However, free FAs are relatively insoluble and potentially toxic, and they require non-catalytic binding proteins to perform their biological functions [41]. FABPs are a family of intracellular proteins that exhibit high affinity for noncovalent binding to long-chain FAs [40]. In the present study, we detected the expression of FABP1–7 in RBE cultured with/without adipose tissue extracts by qRT-PCR assay. The result revealed that adipose coculture significantly enhanced the expression of FABP5 and FABP4, while decreased FABP1 expression in RBE cells, as compared with cells cultured in normal complete medium (Fig. 4a). We noticed that, of these three FABPs, FABP4 was the most enriched one in RBE cells under normal culture condition. Because FABP4 has been regarded as an essential adipokine mostly derived from adipocytes [41], it was chosen for further study in the present work.

As expected, we demonstrated that FABP4 highly expressed in human omentum adipose and mature adipocytes, but not in human CCA tissues or cell lines (Fig. 4b and Additional file 1 Figure S2). Coincidently, human CCA tissues were shown to containing more adipocytes than non-cancerous tissues (Fig. 4c). Taken together the finding that synthesized EGFP-tagged FABP4 protein could be taken up by CCA cells (Fig. 4d), we speculated that the adipocyte-derived FABP4 might play a role in mediating adipocyte-induced CCA metastasis.

To test our above hypothesis, we used the specific FABP4 inhibitor BMS309403 in the following study (Additional file 1 Figure S3A). BMS309403 effectively inhibited the lipid accumulation in both adipose-cocultured RBE and Hccc-9810 cells (Additional file 1 Figure S6, Fig. 5a, 6a), and significantly suppressed the adipose-induced CCA invasion and migration in vitro. For example, the number of invading RBE cells cultivated in adipose tissue extracts was nearly 2-fold of that in control. This trend can be completely reversed by adding FABP4 inhibitor to the adipose tissue extract-containing medium (Fig. 5b). Moreover, the promoted migration of RBE cells cultured with adipose tissue extracts could be suppressed by BMS309403 treatment (Fig. 5c). A similar tendency was observed in Hccc-9810 cells as well (Fig. 6).

Western blot assay results showed that MMP-2, MMP-9 and the EMT-associated SNAIL transcription factor were upregulated, whereas the tight junction proteins claudin1 and occludin were downregulated. Moreover, E-cadherin and β-catenin were suppressed in RBE cells cultivated in adipose tissue extracts compared with that in parental RBE cells (Fig. 5d). However, the BMS309403 treatment completely reversed the expression patterns of the above mentioned EMT marker proteins.

To further confirm the role of FABP4 in CCA metastasis, FABP4 recombinant adenovirus was used to demonstrate the involvement of FABP4 in adipocyte-modulated CCA progression (Additional file 1: Figure S3). FABP4 overexpression enhanced the intracellular lipid accumulation, increased the number of invading cells and the rate of wound healing both in RBE and Hccc-9810 cells (Figs. 7 and 8), and promoted the early metastasis of RBE cells in vivo (Additional file 1 Figure S4). Expressions of EMT-related protein markers were examined in FABP4-overexpressing RBE cells via Western blot assay. As expected, FABP4 overexpression further boosted the expression levels of MMP-2 and SNAIL, whereas it inhibited the expressions of tight junction protein and E-cadherin/β-catenin proteins, compared with that in adenovirus vector control group (Fig. 7d). However, despite these differences, no obvious variations in Smad3 was detected in all the treatments (Fig. 5d and 7d). These results collectively indicated that FABP4 was involved in adipose tissue-mediated CCA metastasis.