Downregulation of Fat Mass and Obesity Associated (FTO) Promotes the Progression of Intrahepatic Cholangiocarcinoma

Intrahepatic cholangiocarcinoma (ICC) ranks as the second most malignant type of primary liver cancer with a high degree of incidence and a very poor prognosis. Fat mass and obesity-associated protein (FTO) functions as an eraser of the RNA m6A modification, but its roles in ICC tumorigenesis and development remain unknown. We showed here that the protein level of FTO was downregulated in clinical ICC samples and cell lines and that FTO expression was inversely correlated with the expression of CA19-9 and micro-vessel density (MVD). A Kaplan-Meier survival analysis showed that a low expression of FTO predicted poor prognosis in ICC. in vitro, decreased endogenous expression of FTO obviously reduced apoptosis of ICC cells. Moreover, FTO suppressed the anchorage-independent growth and mobility of ICC cells. Through mining the database, FTO was found to regulate the integrin signaling pathway, inflammation signaling pathway, epidermal growth factor receptor (EGFR) signaling pathway, angiogenesis, and the pyrimidine metabolism pathway. RNA decay assay showed that oncogene TEAD2 mRNA stability was impaired by FTO. In addition, the overexpression of FTO suppressed tumor growth in vivo. In conclusion, our study demonstrated the critical roles of FTO in ICC.


INTRODUCTION
Intrahepatic cholangiocarcinoma (ICC) is the second most malignant type of primary liver cancer, and its incidence is increasing (1,2). Intrahepatic cholangiocarcinoma (ICC) originates from epithelial cells of the bile duct or hepatic ducts (3). Since some aspects of the clinical features are similar to those of hepatocellular carcinoma (HCC), ICC has insidious symptoms, a high degree of malignancy and very poor prognosis (4). ICC harbors many genetic aberrations, including mutations in isocitrate dehydrogenase1/2 (IDH1/2), epidermal growth factor receptor (EGFR), fibroblast growth factor receptor (FGFR), KRAS, and BRAF (5,6), as well as amplifications of Cyclin D1 (CCND1). Strategies for ICC treatment include surgery, intervention, ablation treatment, radiotherapy, chemotherapy, immune targeting, and other comprehensive treatments (7)(8)(9). In principle, ICC is insensitive to radiotherapy and chemotherapy, and thus, the clinical outcomes are very poor. N 6 -methyladenosine (m 6 A) represents a predominant RNA modification that, not surprisingly, regulates tumorigenesis, and tumor progression (10). The m 6 A RNA modification is a reversible process that is coordinated by methyltransferase (m 6 A "writers"), m 6 A reader proteins and demethylase (m 6 A "erasers") (11). These members cover more than 13 enzymes. The m 6 A "writers" complex consists of METTL3, METTL14, WTAP, CBLL1, RBM15, ZC3H13, and VIRMA and is responsible for methylation of target RNA transcripts (12). Then, m 6 A readers, including YTHDF1-3, YTHDC1, IGF2BPs, and eIF3, discern these m 6 A modifications to direct RNA alternative splicing, translation, localization, and RNA stability, among other processes (13). However, as the m 6 A "erasers, " FTO and ALKBH5 remove m 6 A from the aforementioned target transcripts (14,15). FTO, the first identified m 6 A demethylase, belongs to the AlkB family of Fe(II)/α-ketoglutarate-dependent dioxygenases (16). FTO mediates multiple RNA modifications, including m 6 A and m 6 A m in mRNA and snRNA as well as m 1 A in tRNA (17). Recent studies demonstrated that FTO plays an oncogenic role in cancers (18)(19)(20). For example, in one study, FTO significantly promoted leukemic cell proliferation, transformation, leukemogenesis, curtailed AML cell differentiation and apoptosis by targeting ASB2, and RARA (18). However, this process may be suppressed by R-2-hydroxyglutarate (R-2HG) and IDH mutations, which competitively inhibit FTO (21,22). FTO also removes m 6 A from β-catenin to induce the chemo-radiotherapy resistance in cervical squamous cell carcinoma (CSCC) (20). In addition, the overexpression of FTO can promote breast cancer and gastric cancer progression (19,23). These studies suggest that FTO participates in the regulation of various biological processes in cancer cells through modulation of the RNA transcripts which are critical to the respective pathways.
In the present study, we sought to explore the clinical relevance and biological functions of FTO in ICC.
Quantitative Reverse Transcriptase-PCR RNA was extracted using TRIzol (Invitrogen). Then, a PrimeScript TM RT Reagent Kit with gDNA Eraser was used according to the manufacturer's protocol (TaKaRa). The primer sequences were as follows: FTO, forward primer: 5 ′ -ACTTGGCTCCCTTATCTGACC-3 ′ , reverse primer: 5 ′ -TGTGCAGTGTGAGAAAGGCTT-3 ′ . RT-PCR was performed using a CFX96 Real-Time PCR Detection System (Bio-Rad, Richmond, CA, USA). The 2 − Ct method was used to calculate the expression of FTO. Three replicates were tested for each gene for each sample, and the mean value was calculated. The experiment was repeated three times.

Soft Agar Colony Forming Assay
Cells were seeded in 24-well culture plates (3 × 10 3 cells/well, 4 repetitions per group) with Roswell Park Memorial Institute 1640 (RPMI-1640) containing 20% FBS. Agar was autoclaved and placed in a 42 • C water bath kettle. Lower layer: 0.7% agar was mixed with 2 × RPMI-1640 (containing 2 × antibiotic and 20% FBS) at the ratio of 1:1, and 200 µl was added to each well of the 24-well plate, which was allowed to solidify in a CO 2 incubator for at least 30 min. Upper layer: The 1.2% agar was also mixed with the cell suspension at a ratio of 1:1, and 200 µl was added to each well of the 24-well plate, which was maintained in a CO 2 incubator. The soft agar colony plate was incubated in a 5% CO 2 incubator at 37 • C for 10∼14 days. Eight fields were randomly selected for colony counts (>50 µm diameter) under an inverted microscope (Leica DMI4000B, Germany) at 40 × magnification. The freeware ImageJ (National Institutes of Health, Bethesda, MD, USA) and Adobe Photoshop (Adobe Systems, San Jose, CA, USA) were used in colony diameter measurement. Thirteen colonies were randomly selected for diameter measurement. Colony volume was calculated using the formula: volume = (length × width 2 )/2.

Flow Cytometry
For the apoptosis analysis, 5 × 10 5 cells were digested with trypsin without EDTA and centrifuged (300 g, at 4 • C for 5 min). Cells were washed twice in prechilled PBS. Then, 100 µl of 1 × Binding Buffer was added to resuspend the cells. Then, the cells were incubated with 5 µl Annexin V-FITC and 5 µl propidium iodide (PI) Staining Solution for 10 min at room temperature after gentle mixing. After adding 400 µl of 1 × Binding Buffer and mixing, the samples were examined for 1 h by flow cytometry. The apoptosis detection kit used in this experiment was purchased from BIomake (B32117, Houston, TX, USA).

Transwell Assay
The Transwell chamber (8 µm, Corning #3422, ME, USA) was placed into a 24-well plate. The transfected cells were diluted with 200 µl 0.1% FBS medium to generate a cell suspension with a density of 1 × 10 6 cells/mL and seeded in the upper chamber. Then, 600 µl RPMI-1640 containing 30% FBS was added to the basolateral chamber. Each group was tested in 3 replicates. After culture at 37 • C with 5% CO 2 for 72 h, the Transwell chambers were removed, the cells were rinsed with PBS twice, fixed and stained with 0.3% crystal violet for 15 min, and rinsed in ddH 2 O three times. Cells in the upper chamber were removed. The invasive cells were observed with an inverted microscope (DMI 3000, Leica, IL, USA) (×100), and 5 fields were randomly selected for cell counts.

Immunohistochemistry (IHC)
Paraffin-embedded sections were deparaffinized and rehydrated. Antigen retrieval was performed by boiling the sections in sodium citrate solution in a pressure-cooker for 3 min. Endogenous peroxidase was blocked by 3% H 2 O 2 solution for 10 min at 37 • C and washed with PBST. Then, the paraffin sections were incubated with the primary antibody at 4 • C overnight. The next day, the paraffin sections were rewarmed at 37 • C for 10 min. After two washes in PBS and treatment with an adjuvant for 20 min at 37 • C, the sections were incubated with the second antibody for 30 min at 37 • C. The immunohistochemical reaction was visualized with 3,3,0-diaminobenzidine (DAB) for 3 min. Sections were counterstained for 3 min in hematoxylin (zSGB-Bio, Beijing). Slides were scanned with the digital 3DHISTECH-Pannoramic MIDI (3DHISTECH Ltd, Budapest). The image analysis software ImageJ IHC Profiler was used for staining quantification (24).

Colony Formation Assay
ICC cells were seeded in 12-well plates at densities of 2 × 10 2 per well and. After 12 days culture, cells were treated with 0, 10 and 20 µM cisplatin. In 4 days later, cells were washed with cold phosphate buffered saline (PBS) and stained with 0.3% crystal violet (containing methanol). Colonies consisting of more than 50 cells were defined as surviving colonies. All viability measurements are normalized with the untreated group. Cell survival curves were fitted by GraphPad software.

Statistical Analyses
In our study, Data is presented as mean ± SD, and analyzed using the two-tailed Student t-test for two groups, ANOVA for multiple groups or Chi-square tests for Table 1. The statistical significance of Kaplan-Meier survival curves was assessed with a Log-rank (Mantel-Cox) Test (25). A value of p < 0.05 was considered as statistically significant. * P < 0.05, * * P < 0.01, * * * P < 0.001. Student's t-tests, two-way ANOVA, curve fitting and box plots calculations were performed with GraphPad software (GraphPad, La Jolla, CA, USA).IBM SPSS Statistics 19 (IBM, Chicago, USA) is used for Chi-square test analysis (26).

FTO Protein Levels Are Decreased in ICC
To understand the regulation of m 6 A modification in ICC, we examined the expression of m 6 A-related enzymes in human paracancerous intrahepatic biliary epithelial cells (para-HIBECS) and ICCs by IHC (data not shown). After analyzing the tissue microarray containing 142 patient specimens, it was found that FTO was downregulated in ICC (Figures 1A-C). Although no correlation was observed between disease recurrence and FTO expression (p = 0.084, Figure 1D), the expression of FTO was positively correlated with tumor differentiation, and a lower protein level was observed in the poorly differentiated ICC clinical samples (p = 0.0029, Figure 1E).

FTO Expression Is Correlated With the Clinicopathologic Features of ICC
In the further analysis, we assessed the association between FTO expression and the clinicopathologic features. No association was observed between FTO expression and clinicopathologic features such as age, gender, cirrhosis, primary tumor diameter, tumor number, necrosis, and p53 or E-cadherin expression. Carbohydrate antigen 19-9 (CA19-9), a mucin-type glycoprotein in the serum, is one of the tumor markers used for ICC diagnosis, classification, prognosis, and treatment guidance (27,28) . A higher CA19-9 level in the serum is associated with poor prognosis. Our data indicated that the expression of FTO was negatively correlated with CA19-9 concentration (p = 0.016, Table 1). CD34, another tumor marker, is involved in angiogenesis and is used as a quantitative indicator of micro-vessel density (MVD) (29). An association was found between low FTO expression and high CD34 expression in ICC (p = 0.014, Table 1). In addition, FTO expression was also significantly related to the number of WBCs (white blood cells, p = 0.023) and RBCs (red blood cells, p < 0.001), HGB levels (hemoglobin, p < 0.001), AFU levels (α-L-Fucosidase, p = 0.01), and ALP level (alkaline phosphatase, p = 0.008) in the serum.

Low Expression of FTO Predicts Poor Prognosis in ICC
Next, we analyzed the correlation between FTO expression and prognosis through the tissue microarray IHC staining described above. ICC patients with low expression of FTO showed both poorer overall survival (p = 0.0077) and poorer relapse-free survival (p = 0.0195) (Figures 2A,B). Moreover, in 88 patients with necrobiotic tumor, those with a lower expression of FTO also showed a poorer overall survival than those with a higher expression of FTO (p = 0.0273, Figure 2C). However, the overall survival of patients without necrosis did not differ between the lower FTO expression group and the higher FTO expression group (p = 0.1051, Figure 2D).

FTO Promotes Cisplatin -Induced Apoptosis and Inhibits the Migration of ICC Cells in vitro
We next investigated the biological functions of FTO in ICC. We first analyzed FTO mutation status and copy number alterations (CNAs) in cholangiocarcinoma using the cBioPortal for Cancer Genomics (http://www.cbioportal.org). Neither mutations nor copy number alterations in FTO or other m 6 A modification enzymes were found in cholangiocarcinoma ( Figure 3A). The mRNA level of FTO was lower in ICC cell lines (HCCC-9810, HuCC-T1, TFK1 and RBE) compared with normal human intrahepatic biliary epithelial cells (HIBEPIC, Figure 3B). TFK1 and HCCC-9810 cells were used for both gain-and loss-of-function assays (Figures 3C-E). Cisplatin, which is a standard chemotherapy regimen, has been extensively used for treating ICC and other cancers. To examine a potential function in chemosensitivity, we tested FTO knock-down cells in their response to cisplatin through assaying plating efficiency. Downregulating FTO insensitized the ICC cells to the cisplatin treatment (Figures 4A,B). Consistently, decreased endogenous expression of FTO obviously reduced apoptosis in ICC cell lines (Figures 4C,D), and increased expression of FTO synergized cisplatin-induced apoptosis in ICC cell lines (Figures 4E,F).
To examine the effects of FTO expression on the anchorageindependent growth and invasion of ICC cells, we performed soft agar colony assay and Transwell migration assay. Forced expression of FTO dramatically reduced colony numbers and colony size compared with the control (Figures 5A-C). Additionally, ectopic expression of FTO significantly hampered the invasiveness of TFK1 cells (Figures 5D,E), and knockdown of FTO promoted the invasiveness of HCCC-9810 and TFK1 cells (Figures 5F-I).

FTO Regulates ICC Progression Through Multiple Key Oncogenes and Suppressors
To determine the effects of FTO expression on ICC growth in vivo, TFK1 cells (control cells and FTO overexpressing cells) were subcutaneously transplanted into nude mice. Consistently, overexpression of FTO suppressed the growth, volume and weight of the tumors (Figures 6A-C). In addition,  the overexpression of FTO in tumors was further determined by western blot and immunohistochemistry (Figures 6D,E).
To unravel the molecular mechanisms meditating the biological functions of FTO, we analyzed microarray data obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). In all, 1824 differentially expressed genes between FTO-depleted cells and scrambled cells were found by analyzing GSE33870 using GEO2R (p < 0.05, fold change>2, Figure 7A). The analysis of two ICC chips (GSE45001, 10 ICC samples and paired non-tumorous tissues; GSE32225, 149 ICC tissues and 6 non-tumorous tissues), (30, 31) revealed 3,246 differentially expressed genes in GSE45001 and 1580 differentially expressed genes in GSE32225 (p < 0.05, fold change>2, Figure 7A). Then, 35 differentially expressed genes were found to be overlapped from three databases (Figures 7A,B); these genes were determined to be involved in the integrin signaling pathway, inflammation mediated by chemokine and cytokine signaling pathways, the EGFR signaling pathway, angiogenesis and the pyrimidine metabolism pathway (Figure 7C). Further analysis of the expression status of 35 genes demonstrated that FTO knockdown induced the expression of COL8A1, TEAD2, and CMTM4, which were upregulated in ICC ( Table 2) and that FTO knockdown inhibited the expression of HAO2, NR5A2, CCL19, TCF21, APOA2, NTRK2, DPT FGA, and SCML4, which were downregulated in ICC ( Table 2). We further determined their roles in tumors. TEAD2, a transcriptional enhancer factor, contributes to EMT in breast cancer and pancreatic adenocarcinoma (PDAC) by directing its co-factor YAP1 to the nucleus (32,33). CMTM4 has been identified as a regulator of PD-L1 protein stability in tumor cells (34). The tumor suppressor gene HAO2 inhibits hepatocellular carcinoma metastasis and predicts a favorable prognosis (35). Nuclear receptor NR5A2 is associated with PDAC and transcriptionally regulates inflammatory gene expression (36). Chemokine CCL19 inhibits colorectal cancer (CRC) angiogenesis in a CCR7dependent manner (37). TCF21 also acts as a tumor suppressors in multiple tumors, such as renal tumors, ovarian cancer, colorectal cancer, and lung cancer, among others (38)(39)(40). SCML4 was reported to be downregulated in breast cancer stem cells (41). However, APOA2, NTRK2, and FGA are oncogenes in PDAC, glioblastoma, and gastric cancer (42)(43)(44)(45). Thus, FTO may inhibit ICC progression by reducing oncogene expression (TEAD2 and CMTM4) and inducing tumor suppressor expression (HAO2, NR5A2, CCL19, TCF21, NTRK2, and SCML4). Consistent with our analysis, in TFK1 cells, downregulation of FTO increased the mRNA levels of TEAD2 (Figure 7D). Since FTO was demonstrated to regulate RNA stability, we tested the mRNA stability of TEAD2 through RNA decay assay. The result showed that TEAD2 mRNA stability was increased by inhibiting FTO in TFK1 cells ( Figure 7E). Consistently, the mRNA stability of TEAD2 was impaired by over-expressing FTO in TFK1 cells (Figure 7F).

DISCUSSION
The first m 6 A eraser, FTO, has long been considered an oncogene in tumors (18)(19)(20). However, in the present study, we showed that FTO functions as a tumor suppressor in ICC, which implies that FTO may be a context-dependent regulator in oncogenesis network. To explore the clinical relevance of the m 6 A modification in ICC, we screened the differentially expressed m 6 A readers, writers and erasers in para-HIBECS and ICC by IHC. FTO was found to be downregulated in ICC tissues and cell lines. In the clinical ICC samples, downregulated FTO was associated with the ICC biomarker CA19-9, which suggested that FTO negatively modulated the progression of ICC. Down-regulated FTO was also correlated with angiogenesis and MVD, which was demonstrated by CD34 expression. A Kaplan-Meier survival analysis showed that low FTO expression predicted poor prognosis in ICC.
These clinical findings suggested that FTO played a suppressive role in ICC. Previous study reported that FTO was highly expressed and functioned as an oncogene in certain subtypes of acute myeloid leukemia (AML) by targeting ASB2 and RARA through mRNA demethylation (18). However, their report also showed that FTO was downregulated in certain subtypes of AML, such as primary AML cases with MLL rearrangements/inv (16) or t (8,21) (18). Shun Zhou et al. also showed that FTO promoted chemoradiotherapy resistance in cutaneous squamous cell carcinoma (CSCC) by targeting β-catenin through mRNA demethylation (20). These observations suggested that FTO might play dual roles in cancer. Furthermore, FTO and other m 6 A modification enzymes are also rarely mutated in cholangiocarcinoma, and their copy number changes are unknown. In the biological function analysis, forced expression of FTO markedly synergized cisplatin-induced apoptosis of ICC cells. Consistently, decreased endogenous expression of FTO reduced apoptosis of ICC cells. An anchorage-independent assay on soft agar demonstrated that FTO repressed the viability and growth of ICC cells. In addition, ectopic expression of FTO significantly hampered the mobility of ICC cells in vitro. Mechanistically, we turned to a GEO2R analysis using the public GEO microarray data. From the overlapped and differentially expressed genes from the three databases (GSE33870, 45001, 32225), we identified 35 differentially expressed genes. These genes broadly participate in multiple aspects of tumorigenesis, which suggests that FTO regulates the progression of ICC possibly through modulating the expression of these genes. Further research is needed to elucidate the mechanisms of FTO in the promotion of ICC. Recently, it has been reported that IDH mutations competitively inhibit RNA demethylation of FTO (21,22). FTO is an α-ketoglutarate (α-KG)-dependent dioxygenase, which is competitively inhibited by R-2-hydroxyglutarate (R-2HG) and the structurally related metabolite D-2-hydorxyglutarate (D2-HG) (21,22). In several studies, the IDH mutations promoted the accumulation of 2-HG in tumors, including ∼20% of AMLs and ∼15% of ICCs (5,46). In ICC, the mutant IDH1/2 proteins convert alphaketoglutarate (αKG) to 2-hydroxyglutarate (2HG), which inhibits the activity of multiple αKG-dependent dioxygenases and results in alterations in cell differentiation and survival (47)(48)(49). These reports suggested that the mutated IDH1/2 promotes ICC by inhibiting the functions of FTO.
In summary, our study provides a novel insight into ICC development and suggests that down-regulation of FTO might establish a gene network that is in favor of ICC progression.

DATA AVAILABILITY
All datasets generated for this study are included in the manuscript.

ETHICS STATEMENT
All experimental protocols were approved by a Central South University institutional committee. Informed consent was obtained from all subjects. The study was reviewed and approved by the China national institutional animal care and use committee.

AUTHOR CONTRIBUTIONS
Z-XR, NL, Y-ZD, and L-QS designed experiments. Z-XR, ZL, and J-JH carried out experiments. Z-XR, L-YL, X-XR, JG, YM, and Y-DG analyzed experimental results and sequencing data. Y-DG and Y-MD assisted Z-XR with IHC CHIP assay. X-PZ, D-XZ, and NL gathered information of all patients. Z-XR wrote the manuscript.