Skip to main content
NCBI home page
Cellular Oncology logoLink to Cellular Oncology
. 2022 Mar 31;45(2):275–291. doi: 10.1007/s13402-022-00668-7

Soluble TGFBI aggravates the malignancy of cholangiocarcinoma through activation of the ITGB1 dependent PPARγ signalling pathway

Jungwhoi Lee 1,✉,#, Jungsul Lee 2,#, Woogwang Sim 3,#, Jae-Hoon Kim 1,4,✉
PMCID: PMC12978060  PMID: 35357655

Abstract

Background

Cholangiocarcinoma is a devastating cancer with a poor prognosis. Previous reports have presented conflicting results on the role of transforming growth factor-β-induced protein (TGFBI) in malignant cancers. Currently, our understanding of the role of TGFBI in cholangiocarcinoma is ambiguous. The aim of the present study was to investigate the role of TGFBI in human cholangiocarcinoma.

Methods

Iterative patient partitioning (IPP) scoring and consecutive elimination methods were used to select prognostic biomarkers. mRNA and protein expression levels were determined using Gene Expression Omnibus (GEO), Western blot and ELISA analyses. Biological activities of selected biomarkers were examined using both in vitro and in vivo assays. Prognostic values were assessed using Kaplan–Meier and Liptak’s z score analyses.

Results

TGFBI was selected as a candidate cholangiocarcinoma biomarker. GEO database analysis revealed significantly higher TGFBI mRNA expression levels in cholangiocarcinoma tissues compared to matched normal tissues. TGFBI protein was specifically detected in a soluble form in vitro and in vivo. TGFBI silencing evoked significant anti-cancer effects in vitro. Soluble TGFBI treatment aggravated the malignancy of cholangiocarcinoma cells both in vitro and in vivo through activation of the integrin beta-1 (ITGB1) dependent PPARγ signalling pathway. High TGFBI expression was associated with a poor prognosis in patients with cholangiocarcinoma.

Conclusions

Our data suggest that TGFBI may serve as a promising prognostic biomarker and therapeutic target for cholangiocarcinoma.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13402-022-00668-7.

Keywords: Cholangiocarcinoma, TGFBI, ITGβ1, Prognosis, Liptak’s z value, PPARγ

Introduction

Cholangiocarcinoma is a devastating cancer associated with bile duct epithelia and accounts for 2% of all cancer-related deaths worldwide annually [1]. Despite its rarity, the global incidence of cholangiocarcinoma is rapidly rising [2]. As yet, the molecular mechanisms underlying cholangiocarcinoma are still unclear and efficacious therapies to overcome this aggressive neoplasm have yet to be developed. Even though innovative therapies have been proposed for targeting cholangiocarcinoma, so far none have been found to be overly efficacious and clinical outcomes continue to worsen [3]. Due to the unresolved pathogenesis and the lack of effective therapies, the quest for biomarkers and targets for the prognosis and therapy of cholangiocarcinoma has become increasingly pressing.

Transforming growth factor-β-induced protein (TGFBI), also known as βig-H3, RGD-CAP or kerato-epithelin, is a released extracellular matrix (ECM) protein carrying a N-terminal secretory signal, four FAS1 internal domains and a RGD domain at its C-terminus [4]. TGFBI is secreted from fibroblasts, chondrocytes and smooth muscle cells, and binds to various types of collagen including laminin, fibronectin and secreted protein acidic and rich in cysteine (SPARC) [5]. It has been disputed whether TGFBI in malignant cancers acts as a cell or anti-cell adhesive protein [6–8]. In addition, TGFBI has previously been reported to supress lung and ovarian cancer growth [9, 10], which is in contrast with its recognized role as an oncogene in colon carcinoma, oesophageal squamous cell carcinoma, melanoma and renal cell cancer [11, 12]. Moreover, several conflicting reports suggest either a negative or a positive correlation of TGFBI expression with clinical outcome [13, 14]. A few studies have even reported that TGFBI may serve as a diagnostic biomarker in colorectal and gastrointestinal cancers [15, 16]. To date, the role of TGFBI in cholangiocarcinoma remains ambiguous and, as indicated above, the overall role of TGFBI in cancer progression continues to be unclear.

The peroxisome proliferator-activated receptor gamma (PPARγ) transcription factor has been reported to act as a potentially critical regulator of adipocyte differentiation, insulin sensitivity [17] and urothelial terminal differentiation [18]. In addition, PPARγ has been reported to exert counteractive functions, either as a tumour suppressor or as a tumour activator, in various malignant cancers [19–22]. So far, data are too sparse to conclude whether PPARγ acts as a positive or a negative regulator in cholangiocarcinoma.

Considering the pleiotropic features of TGFBI, the biological functions evoked by secreted TGFBI should be examined not only with respect to its putative function as a molecular target, but also to its potential as a soluble biomarker in cholangiocarcinoma. In this study, we present data indicating that soluble TGFBI aggravates the malignancy of cholangiocarcinoma through activation of the ITGB1-dependent PPARγ signalling pathway, and that enhanced TGFBI expression may be associated with a poor prognosis, suggesting that TGFBI may serve as a promising prognostic biomarker and therapeutic target in cholangiocarcinoma.

Materials and methods

Data collection and gene expression analysis

mRNA expression profiles were obtained from the Gene Expression Omnibus (GEO) public microarray database at NCBI (https://www.ncbi.nlm.nih.gov/geo/), ArrayExpress (https://www.ebi.ac.uk/arrayexpress/), cBioportal (https://www.cbioportal.org/) and ICGC (https://icgc.org/). We integrated data sets independently obtained from several research groups using the absolute normalization method SCAN.UPC [23]. We restricted the integration to the Affymetrix Human Genome U133 Plus 2.0 Array platform (GPL570) because the normalization method is dependent on the total number of probes, and GPL570 has more probes than GPL96 and GPL97. All data were normalized using the default option of SCAN.UPC. TGFBI mRNA expression in cholangiocarcinoma and various normal tissues was also analysed in the Oncopression database (https://www.oncopression.com) as described previously [24].

Iterative patient partitioning (IPP) calculation

To calculate IPP scores, GSE89748, GSE89747 and CHOL-TCGA-V1 datasets were used as reported previously [25].

Consecutive gene eliminating method

To select candidate cholangiocarcinoma biomarkers, a total of 11,574 genes from the GSE-26566 data set was used as reported previously [26].

Cells and culture conditions

Hepatocellular carcinoma (Hep3B and Huh7), pancreatic cancer (Panc-1 and SNU-213), cholangiocarcinoma (SNU-308, SNU-478, SNU-869, SNU-1079 and SNU-1196) and Detroit-551 cells were purchased from the Korean Cell Line Bank (Seoul, Korea). HUVECs were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). The cells were cultured in DMEM (Gibco-BRL, Gaithersburg, MD, USA; Hep3B, Huh7, Panc-1, SNU-308, SNU-478, SNU-869, SNU-1079, SNU-1196 and Detroit-551) or RPMI-1640 medium (Gibco-BRL; SNU-213) supplemented with 10% foetal bovine serum (FBS, Gibco-BRL), 1 × 105 unit/L penicillin and 100 mg/L streptomycin (Gibco-BRL) at 37 °C in a humidified atmosphere containing 5% CO2. HUVECs were grown in EGM-2 BulletKit™ medium (Lonza, Hopkinton, MA, USA) at 37 °C in a humidified atmosphere containing 5% CO2. All experiments were performed using HUVECs within 3–6 passages.

Reagents

Polyclonal antibodies directed against TGFBI were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA, sc-28660 and sc-14742,) and a monoclonal antibody directed against TGFBI from Abcam (Cambridge, UK, ab242889). Antibodies directed against phospho-FAK (Y397), FAK, phospho-AKT (S473), AKT, phospho-ERK1/2 (Thr202/204), ERK1/2, BCL-x, ERBb1, enolase 2, osteopontin, PPARγ, ITGB1 and GAPDH were purchased from Cell Signalling Technology (Beverly, MA, USA). Recombinant TGFBI and anti-LaminB1 and anti-phosphor ITGB1 (Y783) antibodies were obtained from Abcam (Cambridge, UK). U0126, U0124, FAK inhibitor 14 and LY294002 were purchased from Calbiochem (La Jolla, CA, USA).

Cell viability measurement

To evaluate the viability of human cholangiocarcinoma cells under exogenous recombinant TGFBI treatment conditions, WST-1 (Abcam, Cambridge, UK) was used as described previously [26].

Small interfering RNA (siRNA) transfection

Transfections of siRNAs were performed using Effectene (Qiagen, Hilden, Germany) as reported previously [27]. Oligonucleotides specific for TGFBI (sc-43123 and 7045–1) and PPARγ (sc-29455 and 5468–1) were purchased from Santa Cruz Biotechnology and Bioneer (Daejeon, Korea), respectively. Scrambled siRNA (sc-37007) was obtained from Santa Cruz Biotechnology.

Cell migration assay

Cell migration was assessed using an 8.0 μm pore size Transwell® system (Corning Costar, Lowell, MA, USA) as described previously [28].

Cell invasion assay

Cell invasion was assessed using 24-Transwell plates coated with growth factor-reduced Matrigel (BD Biosciences, San Diego, CA, USA) as described previously [29]. Briefly, cells were layered in the upper chamber containing serum-free RPMI-1640 medium for 20 h after which the cells that had migrated through the Matrigel were stained. The eluted stain was measured at 560 nm using an ELISA reader (Bio-Rad, Richmond, CA, USA).

Cell fractionation

A Nuclear/Cytosol Fractionation Kit (Bio Vision, Mountain View, CA, USA) was used according to the manufacturer’s instructions. Cytoplasmic fractions were probed with GAPDH and nuclear fractions with Lamin B1.

Western blot analysis

To assess endogenous TGFBI protein levels in various cancer and normal cells, Western blotting was performed as described previously [23]. To assess soluble TGFBI levels, immunoprecipitation was performed using sc-14742 for immunoprecipitation or sc-28660 for Western blotting with 10 μl supernatant of cells that were incubated for 72 h. Bands were measured by densitometry using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Immunoprecipitation (IP) assay

SNU-1079 and SNU-1196 cells were seeded in 60-mm dishes (SPL, Daejeon, Korea). After 3 days, cell lysates were prepared and aliquots were incubated with anti-TGFBI or normal IgG antibodies, followed by the addition of Protein G agarose (Amersham Bioscience, Little Chalfont, UK). Bound proteins were subjected to Western blotting and probed with an anti-ITGB1 antibody.

PPARγ reporter assay

Cells were transfected with PPARγ luciferase reporter plasmids containing the Renilla luciferase gene using effectene to normalize the transfection efficiency (Qiagen). After 48 h, the cells were treated with rTGFBI (20 µg/ml), denatured rTGFBI (20 µg/ml), U0126 (1 µM), U0124 (1 µM), FAK inhibitor 14 (1 µM) or LY294002 (1 µM) for 24 h, and lysed 72 h post-transfection. Luciferase activity was determined using a dual luciferase assay system (Promega, Madison, WI, USA) according to the manufacturer’s instructions. Luminescence was measured using a Wallac multilabel counter (Perkin-Elmer, Gaithersburg, MD, USA).

Enrichment analysis of selected signature genes

Cholangiocarcinoma patients in GSE89747 were stratified into TGFBI high or low expression groups by median threshold. Enrichment and protein–protein interaction (PPI) network analyses were performed using the Reactome Functional Interaction (FI) plugin app of Cytoscape [30]. All enrichment results from the Gene Ontology and the Reactome databases with a FDR < 0.05 were filtered and presented as top 5th results in order of the number of genes (Reactome FI Network Version: 2019, FI Network Construction Parameters: Fetch FI annotations).

Whole blood sample collections

Different whole blood samples from healthy donors (n = 16), cholangiocarcinoma patients (n = 4), pancreatic cancer patients (n = 6) and liver cancer patients (n = 6) were purchased from Innovative Research (Novi, MI, USA).

Enzyme-linked immunosorbent assay (ELISA)

To measure soluble TGFBI in whole blood samples, 96-well plates (SPL) were coated with a goat anti-TGFBI polyclonal antibody (Santa Cruz Biotechnology, sc-14742) at 1 μg/ml as described previously [31]. Bound soluble TGFBI was detected using a rabbit anti-TGFBI polyclonal antibody (Santa Cruz Biotechnology, sc-28660) or a rabbit anti-TGFBI monoclonal antibody (Abcam, ab242889). Absorbance was measured at 450 nm using a microplate reader (Bio-Rad, Richmond, CA, USA).

Animal models

BALB/c nude mice were purchased from Orient (Seongnam, Korea) at 6–8 weeks of age. SNU-1079 (1 × 107) and SNU-1196 (1 × 107) cells were injected subcutaneously into the left side of the flanks as described previously [32]. Once the tumours reached sizes of approximately 60 mm3 or 100 mm3, the mice were randomised into two experimental groups: phosphate-buffered saline (PBS) and recombinant TGFBI (10 mg/kg). Tumour volumes (V) were evaluated using the formula 0.523 LW2 (L = length, W = width). PPARγ reporter assays using SNU-1079 and SNU-1196 cells from the xenograft models and primary cultures were performed as previously described [33, 34]. Animal care and experiments were carried out in accordance with guidelines approved by the animal bioethics committee of Jeju National University (2020–0005).

GSE dataset analysis

For assessment of the prognostic value of a gene, a log-rank test was performed using Graphpad Prism (version 5), and several log-rank p-values from the data sets were integrated into a single p-value using the Liptak's weighted z-score method as described previously [25].

Statistical analyses

All data are presented as mean ± standard deviation. Significant differences between two independent samples were determined using Student’s t-test. One-way ANOVA with Tukey’s post hoc test was performed for difference analyses among independent groups (SPSS 12.0 K for Windows; SPSS Inc., Chicago, IL, USA).

Results

Transforming growth factor-β-induced protein (TGFBI) expression in human cholangiocarcinoma

To select a specific candidate affecting cholangiocarcinoma prognosis, we calculated IPP scores from three different cholangiocarcinoma datasets, GSE89748, GSE89747 and CHOL_TCGA_V1 (Fig. 1a). To further select a specific candidate for cholangiocarcinoma, we analysed a total of 11,574 genes from the GSE26566 dataset (Fig. 1b). Based on intersections between the IPP calculation and consecutive gene eliminating methods, TGFBI was selected as a final biomarker candidate gene for cholangiocarcinoma.

Fig. 1.

Fig. 1

Open in a new tab

Schematic representation of cholangiocarcinoma biomarker selection using IPP score and consecutive gene elimination methods. a IPP scores from three different cholangiocarcinoma datasets, GSE89748, GSE89747 and CHOL_TCGA_V1. b In total 11,574 genes from the GSE26566 dataset were used for selecting potential cholangiocarcinoma biomarkers

Next, TGFBI expression profiles were obtained from the public microarray database GEO. We found that the levels of TGFBI mRNA expressed in cholangiocarcinoma were significantly higher than those in the normal biliary tract and surrounding liver tissues (Fig. 2a). Compared to the mRNA levels in bone marrow and whole blood, cholangiocarcinoma exhibited markedly higher TGFBI mRNA levels (Fig. 2b). To validate in vitro expression of TGFBI, we analysed the expression of its protein in various cancer and normal cells using Western blotting. We found that the expression of TGFBI was relatively high or similar to that in cancer cells of the liver, pancreas and cholangiocarcinoma compared to Detroit-551 normal fibroblasts and HUVECs (Fig. 2c). Since TGFBI is secreted by tumours [16], we also evaluated its level in culture supernatants. TGFBI was specifically detected in the culture supernatants of cholangiocarcinoma cells, SNU-1079 and SNU-1196, but not in those of liver and pancreatic cancer cells or normal cells (Fig. 2d). To further evaluate the expression of TGFBI in cholangiocarcinoma cells, we analysed its expression in five cholangiocarcinoma cell lines using Western blotting. We found that the expression of TGFBI was significantly higher in cholangiocarcinoma cells compared to HUVECs, and that soluble TGFBI was functionally released from the carcinoma cells (Fig. 2e-f). We next explored whether TGFBI could be detected as a soluble form in whole blood samples derived from healthy donors and patients with cholangiocarcinoma, as well as from pancreatic or liver cancers. Using sandwich ELISA, we found that soluble TGFBI could specifically be detected in whole blood samples of cholangiocarcinoma patients compared to those of healthy donors and patients with pancreatic and liver cancer (Fig. 2g). Similar results were obtained using Western blot analysis (Fig. 2h).

Fig. 2.

Fig. 2

Open in a new tab

Expression of TGFBI in cholangiocarcinoma. a Transcriptional levels of TGFBI in normal bile duct, cholangiocarcinoma and surrounding normal liver tissues analysed using Gene Expression Omnibus (GEO) databases (p-value evaluated with Student’s t test, ***p < 0.001, a.u. indicates arbitrary unit using the UPCs method). b Transcriptional levels of TGFBI in cholangiocarcinoma, bone marrow and whole blood analysed using the ArrayExpress (AE), Gene Expression Omnibus (GEO) and Oncopression databases (p-value evaluated with Student’s t test, a.u. indicates arbitrary unit using the UPCs method). c TGFBI expression in liver cancer, pancreatic cancer, cholangiocarcinoma and normal cells evaluated by Western blot analysis. Data are representative of three independent experiments. Relative pixel intensities were measured using ImageJ software. Brackets indicate variance. d TGFBI in culture supernatants from liver cancer, pancreatic cancer, cholangiocarcinoma and normal cells detected by Western blot analysis. Data are representative of three individual experiments. e TGFBI expression in five cholangiocarcinoma cell lines and HUVECs evaluated by Western blot analysis. Data are representative of three independent experiments. Relative pixel intensities were measured using ImageJ software. Brackets indicate variance. f TGFBI in culture supernatants from five cholangiocarcinoma cell lines and HUVECs detected by Western blot analysis. Data are representative of three individual experiments. Relative pixel intensities were measured using ImageJ software. g Concentrations of soluble TGFBI analysed by sandwich ELISA using whole blood samples of healthy donors and patients with cholangiocarcinoma, pancreatic and liver cancer. Data represent the sum of three independent experiments. h Soluble TGFBI selected by Western blot analysis using whole blood samples from healthy donors, and patients with cholangiocarcinoma and liver cancer. Data are representative of three individual experiments

TGFBI silencing affects viability, migration, invasion and intracellular signalling in human cholangiocarcinoma

To investigate to what extent silencing of TGFBI expression by siRNA transfection affects human cholangiocarcinoma, we transfected si-TGFBI into both SNU-1079 and SNU-1196 cells. We found that cells transfected with TGFBI-specific siRNA showed a significantly decreased expression of TGFBI protein compared to those transfected with scrambled siRNA (Fig. 3a). TGFBI silencing significantly decreased the viability of both cell types under serum-starved cultured conditions (Fig. 3b). Conversely, we found that TGFBI silencing led to only a weak or no decrease in the viability of both cell types under serum-containing culture conditions (data not shown). To ascertain the loss of viability effect of TGFBI silencing, we investigated caspase-3-mediated apoptosis in SNU-1079 and SNU-1196 cells transfected with si-TGFBI. We found that TGFBI silencing led to an increased cleavage of caspase-3 in these cells (Fig. 3c). In addition, we found that TGFBI silencing markedly decreased the migration and invasion capacities of both cell types (Fig. 3d). To validate the anti-cholangiocarcinoma effects of TGFBI silencing, we analysed its associated intracellular sig

Fig. 3.

Fig. 3

Open in a new tab

In vitro effects of TGFBI silencing on human cholangiocarcinoma. a SNU-1079 and SNU-1196 cells were transfected with scrambled or si-TGFBI (#1) for 72 h, after which TGFBI levels were analysed by Western blotting. GAPDH was used as a loading control. The data are representative of three independent experiments. b SNU-1079 and SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA (#1) and incubated for an additional 72 h under serum-starved culture conditions. Viabilities were measured using a WST-1 viability assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences using ANOVA, p < 0.0001; asterisks indicate a significant difference compared to 0% inhibition, **p < 0.01, ***p < 0.001). c SNU-1079 and SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA (#1) and incubated for an additional 72 h under serum-starved culture conditions, after which caspase-3 and cleaved caspase-3 protein levels were analysed by Western blotting. GAPDH was used as loading control. The data are representative of three individual experiments. Brackets indicate variance. d Left: SNU-1079 and SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA (#1). After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, migrated cells were evaluated using a Transwell-migration assay. Right: SNU-1079 and SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA (#1). After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, invasive cells were evaluated using a Transwell-invasion assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, **p < 0.01, ***p < 0.001). e SNU-1079 and SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA (#1) for 48 h, after which cell lysates were subjected to Western blot analysis using antibodies specific for TGFBI, p-FAK (Y397), FAK, p-AKT, AKT, p-ERK1/2, ERK1/2 and GAPDH. Relative pixel intensities were measured by densitometry using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance

nalling mechanisms in SNU-1079 and SNU-1196 cells. siRNA-mediated TGFBI silencing decreased the expression of phospho-FAK (Y397), phospho-AKT and phospho-ERK1/2 in SNU-1079 and SNU-1196 cells compared to that in control si-RNA transfected cells (Fig. 3e). Similar results related to cell viability, migration, invasion and intracellular signalling were obtained via another TGFBI si-RNA transfection of cholangiocarcinoma cells (Supplementary Fig. 1a-e).

In vitro effects of exogenous recombinant TGFBI in cholangiocarcinoma

To evaluate the role of exogenous TGFBI in human cholangiocarcinoma, we treated SNU-1079 and SNU-1196 cells with recombinant TGFBI (rTGFBI). We found that treatment with rTGFBI significantly increased the viability of SNU-1079 and SNU-1196 cells in a dose-dependent manner, whereas denatured rTGFBI had no effect on the viability of these cells (Fig. 4a). rTGFBI treatment led to significantly enhanced effects on the migration and invasion of SNU-1079 cells. Treatment with the maximal dose of rTGFBI (20 µg/ml) increased the migration and invasion of SNU-1079 cells by approximately 1.25- and 1.38-fold compared to control cells, respectively. In contrast, we found that exposure to denatured rTGFBI (20 µg/ml) had no effect on migration or invasion (Fig. 4b). Similar results were obtained using SNU-1196 cells (Fig. 4c). In addition, we analysed signal transduction pathways using an oncology array kit to elucidate the mechanism by which soluble TGFBI aggravates the malignancy of human cholangiocarcinoma cells. Exogenous rTGFBI treatment led to a significant increase in the expression of various oncogenic molecules such as BCL-x, ERBb1, enolase 2 and osteopontin in SNU-1196 cells compared to controls (Fig. 4d). To further analyse the effects of rTGFBI treatment, the levels of BCL-x, ERBb1, enolase2 and osteopontin were investigated at different doses of rTGFBI in both SNU-1079 and SNU-1196 cells. We found that exogenous rTGFBI treatment increased the expression of BCL-x, ERBb1, enolase2 and osteopontin in a dose-dependent manner (Fig. 4e). In addition, we found that exogenous rTGFBI treatment stimulated the expression of phospho-FAK (Y397), phospho-AKT and phospho-ERK1/2 (Fig. 4f).

Fig. 4.

Fig. 4

Open in a new tab

In vitro effects of exogenous recombinant TGFBI on cholangiocarcinoma. a SNU-1079 and SNU-1196 cells were incubated with varying concentrations (0, 5, 10 and 20 µg/ml) of rTGFBI or denatured rTGFB (20 µg/ml). Viabilities were measured using a WST-1 assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate a significant difference compared to 0% inhibition, *p < 0.05, **p < 0.01, ***p < 0.001). b Left: SNU-1079 cells were incubated with varying concentrations (0, 5, 10 and 20 µg/ml) of rTGFBI or denatured rTGFB (20 µg/ml). Migration was evaluated using a Transwell migration assay and invasion was evaluated using a Transwell invasion assay Right: SNU-1079 cells were transfected with scrambled or TGFBI-specific siRNA. After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, invasive cells were evaluated using a Transwell invasion assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, *p < 0.05, **p < 0.01, ***p < 0.001). c Left: SNU-1196 cells were incubated with varying concentrations (0, 5, 10 and 20 µg/ml) of rTGFBI or denatured rTGFB (20 µg/ml). Migration was evaluated using a Transwell migration assay and invasion was evaluated using a Transwell invasion assay. Right: SNU-1196 cells were transfected with scrambled or TGFBI-specific siRNA. After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, invasive cells were evaluated using a Transwell invasion assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, *p < 0.05, **p < 0.01). d SNU-1196 cells were incubated with 20 µg/ml rTGFBI, after which an oncology array was used to determine phosphorylation differences. Relative pixel intensities were measured by densitometry using ImageJ software. The data are representative of two individual experiments. Brackets indicate variance. e SNU-1079 and SNU-1196 cells were incubated at 20 µg/ml rTGFBI in a dose-dependent manner, after which cell lysates were subjected to Western blot analysis using antibodies specific for BCL-x, ERBb1, enolase2 and osteopontin. GAPDH was used as loading control. Relative pixel intensities were measured by densitometry using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance. f SNU1079 and SNU-1196 cells were incubated with 20 µg/ml rTGFBI, after which cell lysates were subjected to Western blot analysis using antibodies specific for p-FAK (Y397), FAK, p-AKT, AKT, p-ERK1/2, ERK1/2 and GAPDH. Relative pixel intensities were measured by densitometry using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance

PPARγ acts as a molecular partner of TGFBI in human cholangiocarcinoma

To investigate potential molecular partners of TGFBI in cholangiocarcinoma, we screened differentially expressed signature genes between TGFBI patient groups using data from cholangiocarcinoma patients (GSE89747) (Fig. 5a). TGFBI expression was presupposed to be involved in ECM organization, integrin signaling, focal adhesion and beta1 integrin interaction according to enrichment analyses of signature genes of a high TGFBI expression group (Fig. 5b). Subsequently, we revealed a molecular interaction between TGFBI and ITGB1 using IP analysis following exogenous rTGFBI treatment in SNU-1079 and SNU-1196 cells. ITGB1 was precipitated by exogenously rTGFBI treated SNU-1079 and SNU-1196 cells compared to those treated with a normal mouse IgG antibody (nmIgG) (Fig. 5c). In addition, exogenous treatment with rTGFBI stimulated the phosphorylation of ITGB1(Y783) and nuclear translocation of phospho-FAK (Y397), phospho-AKT and phospho-ERK1/2 (Fig. 5d-e).

Fig. 5.

Fig. 5

Open in a new tab

Identification of a molecular interaction partner of rTGFBI in cholangiocarcinoma. a Gene volcano plot between TGFBI high or low groups from the GSE89747 dataset. b Enrichment analysis involving high expression of TGFBI in cholangiocarcinoma. c ITGB1 and exogenously administered rTGFBI were precipitated using polyclonal TGFBI and normal rabbit IgG antibodies in SNU-1079 and SNU-1196 cells. Bound ITGB1 was subjected to Western blot analysis using a monoclonal antibody specific for ITGB1. Antibody heavy chain (HC) was used as control. d SNU-1079 and SNU-1196 cells were incubated at 20 µg/ml rTGFBI, after which cell lysates were subjected to Western blot analysis using antibodies specific for phospho-ITGB1 and ITGB1. GAPDH was used as loading control. Relative pixel intensities were measured using ImageJ analysis software. The data are representative of three individual experiments. Brackets indicate variance. e SNU-1079 and SNU-1196 cells were incubated with 20 µg/ml rTGFBI, after which fractionated lysates (cytosol and nuclear fractions) were subjected to Western blot analysis using antibodies specific for p-FAK (Y397), FAK, p-AKT, AKT, p-ERK1/2, and ERK1/2 and GAPDH. CTR indicates loading controls for cytosolic (GAPDH) or nuclear (Lamin B1) fractions. Relative pixel intensities were measured by densitometry using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance. f SNU-1079 cells were incubated with 20 µg/ml rTGFBI, denatured rTGFB (20 µg/ml) or various pharmaceutical inhibitors. PPARγ activity was analysed using a PPARγ luciferase reporter assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, *p < 0.05, **p < 0.01, ** p < 0.001, n.s., non-significant). g SNU-1079 cells were transfected with scrambled or si-PPARγ (#1 or #2) for 72 h, after which PPARγ levels were analysed by Western blotting. GAPDH was used as loading control. The data are representative of three independent experiments. h SNU-1079 cells were transfected with scrambled or PPARγ-specific siRNA (#1 or #2) and incubated for an additional 72 h. Viability was measured using a WST-1 viability assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences using ANOVA, p < 0.0001; asterisks indicate a significant difference compared to 0% inhibition, *p < 0.05). i Left: SNU-1079 cells were transfected with scrambled or PPARγ-specific siRNA (#1 or #2). After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, migrated cells were evaluated using a Transwell migration assay. Right: SNU-1079 cells were transfected with scrambled or TGFBI-specific siRNA (#1 or #2). After 48 h of transfection, the cells were exposed to serum-starved conditions. After 18 h of serum-starvation, invasive cells were evaluated using a Transwell invasion assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, ***p < 0.001)

Previously, transcriptional factor peroxisome proliferator-activated receptor gamma (PPARγ) has been reported to be functionally correlated with ITGB1 in vitro and in vivo [35, 36]. Based on our results and those of others, we set out to investigate PPARγ activity following treatment with rTGFBI or various pharmacological inhibitors of FAK, AKT or ERK1/2. Exogenous treatment with rTGFBI significantly induced PPARγ activity in SNU-1079 cells, whereas the denatured protein had no effect on PPARγ activity. In particular, inhibiting ERK1/2 using U0126 suppressed the rTGFBI-induced PPARγ activity, while the inactive structural analog U0124, FAK inhibitor 14 or LY294002 had no suppressive effects on rTGFBI-induced PPARγ activity (Fig. 5f). In addition, we found that inhibition of PPARγ by different si-RNA transfections significantly decreased viability, migration and invasion in SNU-1079 and SNU-1196 cells (Fig. 5g-i and Supplementary Fig. 2a-1c).

In vivo growth and signalling effects of rTGFBI

To validate the effect of rTGFBI in vivo, we prepared xenograft models using SNU-1079 and SNU-1196 cells. Tumour-bearing mice were injected intraperitoneally with either PBS or rTGFBI (10 mg/kg) when the tumours reached an average size of approximately 60 and 100 mm3 in SNU-1079 and SNU-1196 xenograft models, respectively. PBS-treated SNU-1079 xenograft tumours grew to an average size of 152.62 ± 10.98 mm3 by 40 days after transplantation, while rTGFBI-treated SNU-1079 xenograft tumours grew to an average size of 278.80 ± 106.90 mm3 during the same time period (Fig. 6a). No significant weight loss was observed in either the control or rTGFBI-treated xenografted mice. In addition, we found that PBS-treated SNU-1196 xenograft tumours grew to an average size of 123.80 ± 8.25 mm3 by 40 days after transplantation, while rTGFBI-treated SNU-1196 xenograft tumours increased to an average size of 171.12 ± 14.02 mm3 in the same time (Fig. 6b). No significant weight loss occurred in either the control or rTGFBI-treated xenografted mice. To ascertain the growth stimulating effects of rTGFBI at the molecular level, we determined the levels of oncogenic and intracellular signalling molecules. We found that treatment with 10 mg/kg rTGFBI up-regulated the levels of BCL-x, ERBb1, enolase2 and osteopontin compared to the PBS-treated group (Fig. 6c). The levels of p-ITGB1, p-FAK, p-AKT and p-ERK1/2 were also increased in the 10 mg/kg rTGFBI-treated group compared to the PBS-treated group (Fig. 6d). To evaluate PPARγ activity evoked by rTGFBI stimulation in vivo, SNU-1079 and SNU-1196 cells were isolated from their respective xenograft models (Fig. 6e). We found that primary cultured SNU-1079 and SNU-1196 cells from the rTGFBI-treated xenograft models showed a higher PPARγ activity compared to the PBS-treated controls (Fig. 6f).

Fig. 6.

Fig. 6

Open in a new tab

In vivo effects of rTGFBI. a-b Effects of rTGFBI on SNU-1079 and SNU-1196 xenograft model growth measured for 40 days using the formula: V = 0.523 LW2 (L = length, W = width), respectively. Bold arrows indicate the time of rTGFBI (10 mg/kg) injection (Tukey’s post-hoc test was used to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to control and rTGFBI treatment, *p < 0.05, **p < 0.01, n.s.: non-significant). c Western blot analysis of control and rTGFBI treated tumor lysates using antibodies specific for BCL-x, ERBb1, enolase2 and osteopontin. GAPDH was used as loading control. Relative pixel intensities were measured using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance. d Western blot analysis of control and rTGFBI treated tumor lysates using antibodies specific for p-ITGB1, ITGB1 p-FAK (Y397), FAK, p-AKT, AKT, p-ERK1/2 and ERK1/2. GAPDH was used as loading control. Relative pixel intensities were measured using ImageJ software. The data are representative of three individual experiments. Brackets indicate variance. e–f PPARγ activity of cultured SNU1079 and SNU-1196 cells derived from control or rTGFBI adminisered xenograft models analysed by PPARγ luciferase reporter assay (n = 3; Tukey’s post-hoc test was applied to detect significant differences in ANOVA, p < 0.0001; asterisks indicate significant difference compared to 0% inhibition, **p < 0.01, ***p < 0.001)

Prognostic value of TGFBI expression in cholangiocarcinoma patients

To assess the role of TGFBI expression in the prognosis of cholangiocarcinoma patients, we analysed a GEO dataset including overall survival (OS) information. We found that high levels of TGFBI expression significantly correlated with a shorter median survival time (MS, 448 days) compared with low levels of TGFBI (MS, 2304 days) in patients with cholangiocarcinoma (Supplementary Fig. 3). The prognostic significance of TGFBI expression in cholangiocarcinoma patients was determined using Liptak’s Z-value based on three cholangiocarcinoma datasets (GSE89748, GSE89747 and CHOL_TCGA_V1). Elevated TGFBI expression correlated significantly with an adverse prognosis in cholangiocarcinoma patients (Fig. 7a; Liptak’s Z-value -2.0437, p value 0.0204). Conversely, we found that a high TGFBI expression in liver cancer patients correlated with a favourable clinical outcome in seven liver cancer datasets (LIRI-JP-RIKEN, GSE54236, GSE4024, GSE14520.HT, GSE14520.HG, GSE10141 and E-TABM-36), although the difference was not statistically significant (Liptak’s Z-value + 0.4536; p value 0.3250) (Supplementary Fig. 4). To further assess the involvement of TGFBI expression in cholangiocarcinoma prognosis, TGFBI, ITGB1 and PPARγ mRNA expression levels were investigated in cholangiocarcinoma patients with either a good or a poor prognoses. We found that the TGFBI mRNA expression level was particularly upregulated in poor prognosis cholangiocarcinoma patients, markedly more so than either ITGB1 or PPARγ (Fig. 7b-d).

Fig. 7.

Fig. 7

Open in a new tab

Prognostic value of TGFBI expression in cholangiocarcinoma patients. a Prognostic significance of TGFBI expression in cholangiocarcinoma patients analysed using Liptak’s Z value based on three cholangiocarcinoma datasets (GSE89748, GSE89747 and CHOL_TCGA_V1). b-d TGFBI, ITGB1 and PPARγ transcriptional levels in good and poor prognosis groups in the GSE89747 dataset (a.u. indicates arbitrary unit using the UPCs method, p-value evaluated with Student’s t test). e Graphical scheme of the putative role of soluble TGFBI as a prognostic biomarker and therapeutic target in human cholangiocarcinoma

Overall, we found that TGFBI was specifically expressed in a soluble form by cholangiocarcinoma cells, which affects the malignancy and prognosis of cholangiocarcinoma through activation of the ITGB1 dependent PPARγ signalling pathway (Fig. 7e).

Discussion

Cholangiocarcinoma is the second most common primary tumour of the liver after hepatocellular carcinoma, but cholangiocarcinoma can randomly occur at any site of the biliary tract [1]. The development of cholangiocarcinoma and its subsequent malignant transformation have been partly attributed to local inflammation, parasitic infestation, hepatitis C virus infection and long-term biliary inflammation [37]. However, the underlying mechanisms of cholangiocarcinoma pathogenesis remain poorly defined, even now its incidence rapidly increases globally [3]. The insidious presentation of cholangiocarcinoma combined with its aggressive malignant nature and obstinate refractoriness to chemotherapy leads to a notorious high mortality rate, representing up to 2% of all cancer-related deaths worldwide yearly [2, 3], which prompted us to exploit an effective soluble-biomarker for diagnosis and prognosis. We found that cholangiocarcinoma cells specifically secrete a soluble form of TGFBI compared to the insoluble form produced by other malignant cancers such as hepatocellular and pancreatic cancer. TGFBI expression silencing led to significant anti-cancer effects in vitro. In contrast, we found that exogenous treatment with soluble TGFBI induced malignancy in cholangiocarcinoma cells through activation of the ITGB1-dependent PPARγ signalling pathway both in vitro and in vivo. Of particular note, enhanced TGFBI expression was found to be associated with a poor prognosis in cholangiocarcinoma patients.

By using state-of-the-art bioinformatics techniques, we selected TGFBI as a cholangiocarcinoma biomarker candidate using iterative patient partitioning (IPP) calculation and a consecutive gene eliminating method. TGFBI was found to be strongly expressed in colorectal, lung and gastrointestinal tract cancers compared to matched normal tissues using immunohistochemical techniques [13, 14, 16]. In addition to these previous reports, we found that TGFBI was highly expressed in cholangiocarcinoma using in silico analysis. TGFBI mRNA levels were found to be significantly elevated in cholangiocarcinoma tissues compared to matched normal tissues and surrounding liver tissues. Of particular interest, the TGFBI mRNA level in whole blood samples from healthy donors was found to be markedly down-regulated compared to that in samples from cholangiocarcinoma patients, implying that the basal expression levels of soluble TGFBI originating from cholangiocarcinoma and normal whole blood can be differentiated.

Currently, several invasive imaging modalities are used for the diagnosis of cholangiocarcinoma, such as ultrasonography-equipped CT scanning, percutaneous trans-hepatic cholangiography and endoscopic retrograde cholangiography with fine-needle aspiration [38]. Although such techniques represent functional tools, they are only available to a limited number of patients due to their high cost, inaccuracy and need for highly trained operators. To improve the detection of cholangiocarcinoma, minimally invasive and soluble biomarkers have been in development for decades. Carcinoembryonic antigen (CEA) and CA19-9 are clinically used as soluble biomarkers for cholangiocarcinoma, despite their low sensitivity and specificity [39]. In the present study, we validated TGFBI as a soluble biomarker for cholangiocarcinoma using a relatively accessible sandwich ELISA technique. The potential role of TGFBI as a soluble biomarker has been reported in sera of patients with gastrointestinal tract cancers, including cholangiocarcinoma, liver and gastric cancers [16]. Consistent with a previous report, we found that TGFBI expression can be used a soluble biomarker for cholangiocarcinoma using whole blood samples obtained from healthy donors and patients with cholangiocarcinoma, pancreatic and liver cancer. However, soluble TGFBI could specifically be detected in whole blood samples from cholangiocarcinoma patients, but not in those of healthy donors or patients with pancreatic or liver cancer using either sandwich ELISA or Western blot analysis. Although serum levels of TGFBI have been reported in gastrointestinal tract cancers including cholangiocarcinoma, liver and gastric cancer, we found that TGFBI may serve as a distinct soluble biomarker for cholangiocarcinoma. Further studies with larger numbers of whole blood samples, especially from patients with early-stage cholangiocarcinoma are, however, required to reliably determine the utility of TGFBI as a soluble biomarker. In addition, a differential diagnosis of cholangiocarcinoma should include benign biliary tract diseases such as cirrhosis, gallstone, hepatitis C infection, liver fluke infection, primary sclerosing cholangitis and prolonged or recurrent biliary infection due to the presentation of similar clinical symptoms [37].

The role of TGFBI in different cancers has been attributed to a role as tumour suppressor or as oncogene [9–12]. In the present study, we investigated the role of TGFBI in cholangiocarcinoma through both loss of function by silencing and subsequent gain of function by the administration of soluble TGFBI. TGFBI silencing affected the viability, migration and invasion of cholangiocarcinoma cells by activating caspase-3 mediated apoptosis or inhibition of key intracellular signalling pathways. By contrast, we found that in vitro and in vivo administration of soluble TGFBI increased malignancy in concert with an enhanced expression of oncogenic molecules such as BCL-x, ERBb1, enolase2 and osteopontin in cholangiocarcinoma compared to control cells. Taken together, we propose that soluble TGFBI in patients with cholangiocarcinoma may be oncogenic in nature. In addition, we propose that correlations between BCL-x, ERBb1, enolase2 or osteopontin and TGFBI should be further considered to develop new combined therapies. Understanding these relationships may significantly facilitate the design of therapeutic strategies based on TGFBI and potential rational combination regimens.

Integrins (ITGs) play critical roles in evading apoptosis and maintaining cellular motility. It has been reported that ITGs help tumor cells to gain malignancy through interactions with their corresponding ECM components [40]. Being a pivotal player in aggressive cancer phenotypes, ITGB1 has been reported to play diverse roles in various cancers [41, 42]. Other reports have indicated that TGFBI may cooperate with various integrins to modulate various cellular phenotypes [43, 44]. Based on previous reports, the integrin signalling pathway, especially the one involving beta 1 integrin interaction, was selected for its putative association with TGFBI using volcano plot and PPI network construction analyses using gene expression profiles of cholangiocarcinoma patients. Of interest, we found that administration of soluble TGFBI significantly induced the phosphorylation of ITGB1, which turned out to be a functional molecular partner of TGFBI in cholangiocarcinoma. The TGFBI-ITGB1 axis could, however, not be verified as the underlying mechanism for malignancy in cholangiocarcinoma due to the basal cytotoxicity of ITGB1 blockade in SNU-1079 and SNU-1196 cells, implying that ITGB1 is a critical ECM component potentially defining the malignancy of cholangiocarcinoma. Elucidation of the relationship between TGFBI and ITGB1 may be expected to contribute to rational drug development and the design of combinatorial therapeutic strategies. The potential clinical impact of understanding the unique correlation between TGFBI-ITGB1 warrants further study.

According to previous reports, PPARγ is functionally related with various integrins and ERK1/2 in malignant tumors, including cholangiocarcinoma [45, 46]. Consistent with previous reports, we revealed a soluble TGFBI-ITGB1-ERK1/2-PPARγ signalling pathway to be active in cholangiocarcinoma models and observed an oncogenic role of PPARγ through ERK1/2 phosphorylation in cholangiocarcinoma, both in vitro and in vivo. Of interest, we found that transfection of si-PPARγ in cholangiocarcinoma cells induced significant anti-cancer effects compared to control cells. We also found that PPARγ activity was higher in cultured cells derived from soluble TGFBI administered xenograft models than in those of control models, implying that PPARγ activity is adversely correlated with malignancy in cholangiocarcinoma. The biological meaning of PPARγ blockade in malignant tumors is, however, still ambiguous, and it is currently not known whether PPARγ blackade acts as a tumour suppressor or activator [47–49]. Further research is required to more exactly define the ambiguous functions of PPARγ in tumor progression.

Despite therapeutic advances that have been made such as chemo-, immune-, photodynamic- and radio-therapies, cholangiocarcinoma still has a poor prognoses [1–3]. Clinical outcomes of TGFBI expression in malignant cancers have turned out to be either adverse or favourable [13, 14]. Also, the importance of TGFBI expression and its prognostic value in cholangiocarcinoma has remained obscure thus far. Our present study strongly supports a clinical significance of TGFBI expression in cholangiocarcinoma. We found that enhanced TGFBI expression significantly aggravated the prognosis of cholangiocarcinoma patients compared to patients with a low TGFBI expression according to Kaplan–Meier curves. Liptak’s Z analysis including a sufficient number of patients (low TGFBI-expressing group, n = 94; high TGFBI-expressing group, n = 50) further showed that enhanced TGFBI expression was significantly correlated with a poor clinical outcome (p = 0.0204), indicating that the prognosis of cholangiocarcinoma patients was adversely affected by increased TGFBI expression. Of interest, enhanced ITGB1 expression indicated a better prognosis in liver cancer patients, although the difference was not statistically significant (Liptak’s Z value: 0.4536, p = 0.325). A previous preclinical study suggesting that TGFBI expression is critical for hepatocellular tumorigenesis does not correlate with our clinical evidence as previously described [16].

Collectively, our present results indicate that the expression of soluble TGFBI affects the malignancy and clinical outcome of cholangiocarcinoma. Our results suggest that soluble TGFBI may serve as a promising prognostic biomarker and therapeutic target for cholangiocarcinoma.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 428 kb) (428.6KB, docx)
Supplementary file2 (DOCX 722 kb) (722.2KB, docx)

Acknowledgements

Not applicable

Abbreviations

TGFBI

Transforming growth factor-β-induced protein

THPA-DB

The human protein atlas database

GEO

Gene expression omnibus

ITGβ1

Integrin β1

PPARγ

Peroxisome proliferator-activated receptor gamma

Author’s contributions

JW Lee conceived and supervised the experiments. JW Lee, JS Lee and W Sim designed and performed the experiments. JS Lee performed the gene expression profile and survival analyses. JW Lee, J Lee, W Sim and JH Kim analysed the data. JH Kim obtained the funding. JW Lee wrote and proofread the manuscript.

Funding

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF), funded by the Ministry of Education (2016R1A6A1A03012862 and 2021R1I1A1A01041462).

Data availability

All data generated or analysed during this study are included in this article and its supplementary information file.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interest

All authors declare no competing financial interests.

Footnotes

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

J Lee, JS Lee and W Sim contributed equally to this work.

Contributor Information

Jungwhoi Lee, Email: sdjd1108@kaist.ac.kr.

Jae-Hoon Kim, Email: kimjh@jejunu.ac.kr.

References

  • 1.J.M. Banales, J.J.G. Marin, A. Lamarca, P.M. Rodrigues, S.A. Khan, L.R. Roberts, V. Cardinale, G. Carpino, J.B. Andersen, C. Braconi, D.F. Calvisi, M.J. Perugorria, L. Fabris, L. Boulter, R.I.R. Macias, E. Gaudio, D. Alvaro, S.A. Gradilone, M. Strazzabosco, M. Marzioni, C. Coulouarn, L. Fouassier, C. Raggi, P. Invernizzi, J.C. Mertens, A. Moncsek, S. Rizvi, J. Heimbach, B.G. Koerkamp, J. Bruix, A. Forner, J. Bridgewater, J.W. Valle, G.J. Gores, Nat. Rev. Gastroenterol. Hepatol. 17, 557–588 (2020). 10.1038/s41575-020-0310-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.M.C. Bragazzi, L. Ridola, S. Safarikia, S.D. Matteo, D. Costantini, L. Nevi, V. Cardinale, Ann. Gastroenterol. 31, 42–55 (2018). 10.20524/aog.2017.0209 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.S.A. Khan, S. Emadossadaty, N.G. Ladep, H.C. Thomas, P. Elliott, S.D. Taylor-Robinson, M.B. Toledano, J. Hepatol. 56, 848–854 (2012). 10.1016/j.jhep.2011.11.015 [DOI] [PubMed] [Google Scholar]
  • 4.P. Romero, M. Vogel, J.M. Diaz, M.P. Romero, L. Herrera, Mol. Vis. 14, 829–835 (2008) [PMC free article] [PubMed] [Google Scholar]
  • 5.E. Hanssen, B. Reinboth, M.A. Gibson, J. Biol. Chem. 278, 24334–24341 (2003). 10.1074/jbc.M303455200 [DOI] [PubMed] [Google Scholar]
  • 6.S.E. Klamer, C.G. Kuijk, P.L. Hordijk, C.E. van der Schoot, M. von Lindern, P.B. van Hennik, C. Voermans, Cell Adh. Migr. 7, 434–449 (2013). 10.4161/cam.26596 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.P. Nummela, J. Lammi, J. Soikkeli, O. Saksela, P. Laakkonen, E. Holtta, Am. J. Pathol. 180, 1663–1674 (2012). 10.1016/j.ajpath.2011.12.035 [DOI] [PubMed] [Google Scholar]
  • 8.L. Shelton, J.A. Rada, Invest. Ophthalmol. Vis. Sci. 50, 3542–3552 (2009). 10.1167/iovs.09-3460 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Y.L. Zhao, C.Q. Piao, T.K. Hei, Oncogene 21, 7471–7477 (2002). 10.1038/sj.onc.1205891 [DOI] [PubMed] [Google Scholar]
  • 10.G. Wen, M.A. Partridge, B. Li, M. Hong, W. Liao, S.K. Cheng, Y. Zhao, G.M. Calaf, T. Liu, J. Zhou, Z. Zhang, T.K. Hei, Cancer Lett. 308, 23–32 (2011). 10.1016/j.canlet.2011.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.D. Schneider, J. Kleeff, P.O. Berberat, Z. Zhu, M. Korc, H. Friess, M.W. Buchler, Biochem. Biophys. Acta. 1588, 1–6 (2002) [DOI] [PubMed] [Google Scholar]
  • 12.P. Buckhaults, C. Rago, B. St Croix, K.E. Romans, S. Saha, L. Zhang, B. Vogelstein, K.W. Kinzler, Cancer Res. 61, 6996–7001 (2001) [PubMed]
  • 13.J. Zhu, X. Chen, Z. Liao, C. He, X. Hu, Int. J. Clin. Exp. Pathol. 8, 702–710 (2015) [PMC free article] [PubMed] [Google Scholar]
  • 14.M.J. Pajares, J. Agorreta, E. Salvo, C. Behrens, Wistuba, II, L.M. Montuenga, R. Pio, A. Rouzaut, Brit. J. Cancer 110, 1545–1551 (2014). 10.1038/bjc.2014.33 [DOI] [PMC free article] [PubMed]
  • 15.B.W. Robinson, R.A. Lake, N. Engl. J. Med. 353, 1591–1603 (2005). 10.1056/NEJMra050152 [DOI] [PubMed] [Google Scholar]
  • 16.B. Han, H. Cai, Y. Chen, B. Hu, H. Luo, Y. Wu, J. Wu, Mol. Cancer 14, 64 (2015). 10.1186/s12943-015-0335-z [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.B. Desvergne, W. Wahli, Endocr. Rev. 20, 649–688 (1999). 10.1210/edrv.20.5.0380 [DOI] [PubMed] [Google Scholar]
  • 18.C.L. Varley, J. Stahlschmidt, W.C. Lee, J. Holder, C. Diggle, P.J. Selby, L.K. Trejdosiewicz, J. Southgate, J. Cell Sci. 117, 2029–2036 (2004). 10.1242/jcs.01042 [DOI] [PubMed] [Google Scholar]
  • 19.J. Youssef, M. Badr, Br. J. Pharmacol. 164, 68–82 (2011). 10.1111/j.1476-5381.2011.01383.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.A. Stravodimou, G. Mazzoccoli, I.A. Voutsadakis, PPAR Res. 2012, 367450 (2012). 10.1155/2012/367450 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.T. Takashima, Y. Fujiwara, K. Higuchi, T. Arakawa, Y. Yano, T. Hasuma, S. Otani, Int. J. Oncol. 19, 465–471 (2001) [PubMed] [Google Scholar]
  • 22.T.H. Chang, E. Szabo, Caner Res. 60, 1129–1138 (2000) [PubMed] [Google Scholar]
  • 23.J. Lee, J. Hun Yun, J. Lee, C. Choi, J. Hoon Kim, Sci. Rep. 5, 12296 (2015). 10.1038/srep12296 [DOI] [PMC free article] [PubMed]
  • 24.J. Lee, J. Lee, J.H. Yun, C. Choi, S. Cho, S.J. Kim, J.H. Kim, Sci. Rep. 7, 12760 (2017). 10.1038/s41598-017-13023-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.W. Sim, J. Lee, C. Choi, Sci. Rep. 7, 16926 (2017). 10.1038/s41598-017-17213-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.J. Lee, J. Lee, J.H. Kim, Anticancer Res. 39, 5963–5971 (2019). 10.21873/anticanres.13801 [DOI] [PubMed] [Google Scholar]
  • 27.J. Lee, D.H. Kim, J.H. Kim, Phytomedicine 58, 152762 (2019). 10.1016/j.phymed.2018.11.022 [DOI] [PubMed] [Google Scholar]
  • 28.J. Lee, J. Lee, J.H. Kim, Cell. Oncol. (2020). 10.1007/s13402-020-00527-3 [Google Scholar]
  • 29.J. Lee, J.H. Kim, PLoS ONE 11, e0155264 (2016). 10.1371/journal.pone.0155264 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.B. Jassal, L. Matthews, G. Viteri, C. Gong, P. Lorente, A. Fabregat, K. Sidiropoulos, J. Cook, M. Gillespie, R. Haw, F. Loney, B. May, M. Milacic, K. Rothfels, C. Sevilla, V. Shamovsky, S. Shorser, T. Varusai, J. Weiser, G. Wu, L. Stein, H. Hermjakob, P. D’Eustachio, Nucleic Acids Res. 48, D498–D503 (2020). 10.1093/nar/gkz1031 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.J. Lee, J. Lee, H. Yu, K. Choi, C. Choi, Cancer Lett. 309, 145–150 (2011). 10.1016/j.canlet.2011.05.026 [DOI] [PubMed] [Google Scholar]
  • 32.J. Lee, J. Lee, M. Kim, J.H. Kim, J. Funct. Foods 30, 303–312 (2017). 10.1016/j.jff.2016.12.032 [Google Scholar]
  • 33.I.S. Miller, L.P. Shiels, E. Conroy, K. Connor, P. Dicker, W.M. Gallagher, N.O. Donovan, R.S. Kerbel, J. Crown, A.T. Byrne, Sci. Rep. 9, 9204 (2019). 10.1038/s41598-019-45444-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.T. Kuwata, K. Yanagihara, Y. Iino, T. Komatsu, A. Ochiai, S. Sekine, H. Taniguchi, H. Katai, T. Kinoshita, A. Ohtsu, Cells 8, (2019). 10.3390/cells8060585 [DOI] [PMC free article] [PubMed]
  • 35.Z. Zolota, G. Koliakos, K. Paletas, M. Kaloyianni, Cell Adh. Migr. 5, 258–265 (2011). 10.4161/cam.5.3.14534 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.X. Chen, C. Whiting, C. Borza, W. Hu, S. Mont, N. Bulus, M.Z. Zhang, R.C. Harris, R. Zent, A. Pozzi, Mol. Cell. Biol. 30, 3048–3058 (2010). 10.1128/MCB.00892-09 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.H. Malhi, G.J. Gores, J. Hepatol. 45, 856–867 (2006). 10.1016/j.jhep.2006.09.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.R.I.R. Macias, J.M. Banales, B. Sangro, J. Muntane, M.A. Avila, E. Lozano, M.J. Perugorria, F.J. Padillo, L. Bujanda, J.J.G. Marin, Biochim. Biophys. Acta 1864, 1468–1477 (2018). 10.1016/j.bbadis.2017.08.002 [DOI] [PubMed] [Google Scholar]
  • 39.M. Miwa, G. You, H. Tanaka, S. Taniguchi, T. Fujii, K. Kamemura, M. Suzaki, T. Isono, I. Tooyama, M. Tanaka, P. Srivatanakul, C. Viwatthanasittiphong, S. Sangrajrang, T. Khuhaprema, J. Hepatobiliary Pancreat. Sci. 21, 397–398 (2014). 10.1002/jhbp.69 [DOI] [PubMed] [Google Scholar]
  • 40.S.J. Shattil, C. Kim, M.H. Ginsberg, Nat. Rev. Mol. Cell Biol. 11, 288–300 (2010). 10.1038/nrm2871 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.R. Barrow-McGee, N. Kishi, C. Joffre, L. Menard, A. Hervieu, B.A. Bakhouche, A.J. Noval, A. Mai, C. Guzman, L. Robbez-Masson, X. Iturrioz, J. Hulit, C.H. Brennan, I.R. Hart, P.J. Parker, J. Ivaska, S. Kermorgant, Nat. Commun. 7, 11942 (2016). 10.1038/ncomms11942 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.S. Berrazouane, M. Boisvert, S. Salti, W. Mourad, R. Al-Daccak, F. Barabe, F. Aoudjit, Cell Death Dis. 10, 357 (2019). 10.1038/s41419-019-1593-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.D.A. Tumbarello, J. Temple, J.D. Brenton, Mol. Cancer 11, 36 (2012). 10.1186/1476-4598-11-36 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.A.A. Ahmed, A.D. Mills, A.E. Ibrahim, J. Temple, C. Blenkiron, M. Vias, C.E. Massie, N.G. Iyer, A. McGeoch, R. Crawford, B. Nicke, J. Downward, C. Swanton, S.D. Bell, H.M. Earl, R.A. Laskey, C. Caldas, J.D. Brenton, Cancer Cell 12, 514–527 (2007). 10.1016/j.ccr.2007.11.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.K. Tachibana, D. Yamasaki, K. Ishimoto, T. Doi, PPAR Res. 2008, 102737 (2008). 10.1155/2008/102737 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.F. Chen, M. Wang, J.P. O’Connor, M. He, T. Tripathi, L.E. Harrison, J. Cell. Biochem. 90, 732–744 (2003). 10.1002/jcb.10668 [DOI] [PubMed] [Google Scholar]
  • 47.H.A. Elrod, S.Y. Sun, PPAR Res. 2008, 704165 (2008). 10.1155/2008/704165 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Y. Peng, Q. Zhang, R.M. Zielinski, R.D. Howells, W.J. Welsh, Pharmacol. Res. Perspect. 8, e00693 (2020). 10.1002/prp2.693 [DOI] [PMC free article] [PubMed]
  • 49.D. Panigrahy, S. Huang, M.W. Kieran, A. Kaipainen, Cancer Biol. Ther. 4, 687–693 (2005). 10.4161/cbt.4.7.2014 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary file1 (DOCX 428 kb) (428.6KB, docx)
Supplementary file2 (DOCX 722 kb) (722.2KB, docx)

Data Availability Statement

All data generated or analysed during this study are included in this article and its supplementary information file.

Articles from Cellular Oncology are provided here courtesy of Springer

RESOURCES