Aberrant FGF15/FGFR4 Signaling Worsens Nonalcoholic Steatohepatitis Severity in FGF21KO Mice

Pharmacological application of the endocrine broblast growth factor15/19 (FGF15/19) and FGF21 holds great promise to be effective therapeutic agents for treating nonalcoholic steatohepatitis (NASH), which is the most severe form of non-alcoholic fatty liver disease (NAFLD) and accepted as a potential precursor of hepatocellular carcinoma (HCC). In our previous studies, we found that FGF21 played a key role in preventing the development of NASH, however, the FGF15/19 mediated-FGFR4 signaling worsened NASH and even contributed to the NASH-HCC transition.

FGF21 played a key role in preventing the development of the major characteristics of NASH: steatosis, in ammation, and metabolic syndrome, however, the FGF15/19 mediated-FGFR4 signaling worsened NASH and even contributed to the NASH-HCC transition.
Fibroblast growth factors family composed of a group of structurally related polypeptides which involved in various biological processes including development, differentiation, neuronal functions, and metabolism [10]. There are three endocrine FGFs-FGF15/19, FGF21, and FGF23-identi ed in mouse/human, and human FGF19 is the orthologous gene of mouse FGF15 [11]. FGF21 is predominantly produced by hepatocytes, while FGF15/19 is mostly secreted from the ileum but targeted to liver. Hepatic FGF21 elicits its metabolic bene ts under the regulation of peroxisome proliferator-activated receptor α (PPARα), acting on the distal adipose tissue adipocytes, through the transmembrane receptor FGFR1coreceptor β-Klotho complex [12]. This major endocrine actions of FGF21 in terms of lipid metabolic bene ts include control of lipolysis, clearance of excessive FFAs, enhancing mitochondrial oxidation and expenditure of the stored lipid energy, therefore negatively regulating hepatic or tissue steatosis, and adiposity [13][14][15]. FGF21, as an endocrine hormone, also takes part in the regulation of glucose and lipid metabolism [16], while pharmacological application has shown that FGF21 can be a promising and effective therapeutic agent for treating NASH, obesity and diabetes [17][18][19]. FGF15/FGF19 has been also reported to prevent NASH [20][21][22]. Although FGF15/19 upon the mitogenic and cytoprotective effects is critical in protection of hetaptocyte from lipid-mediated cellular stress and injury [20], the carcinogenetic role of FGF15/19/FGFR4 signaling has been well studied in various malignances, including breast cancer, gastric cancer, lung cancer, prostate cancer, nasopharyngeal carcinoma and liver cancer [23]. FGF15/19 on glucose metabolism and its crucial regulatory role in bile acid (BA) homeostasis have endorsed FGF15/19's metabolic bene ts for whole body lipid metabolism [24]. However, it is not known whether FGF19 signaling in the liver is indispensable against hepatic lipid accumulation to substitute for FGF21 to compensate the metabolic bene ts when FGF21 protein is compromised in liver.
In this study, early and advanced NASH models were established in FGF21KO mice fed with high fat methionine-choline de cient (HFMCD) diet to investigate FGF15/FGFR4 signaling during NASH development. We sought to determine whether FGF15/FGFR4 signaling could alleviate or aggravate NASH in FGF21KO mice challenged with HFMCD.

Materials And Methods
Establishing NASH models FGF21 Knockout (FGF21KO) mice with C57 BL/6J background were generously granted by Dr. Steve Kliewer (University of Texas Southwestern Medical Center). Wild-type (WT) C57 BL/6J mice were obtained from Jackson Laboratory (Bar Harbor, ME). Six-weeks old male mice were fed with Rodent Diets, HFMCD (L-amino acid diet with 60 kcal% fat, 0.1% methionine and no added choline, A06071302, Research Diets, Inc., New Brunswick, NJ) to induce NASH. Rodent Diet (CD, 10% kcal% fat, D12450B, Research Diets, Inc., New Brunswick, NJ) was used as control diet (CD). Both FGF21KO and WT mice with respective diets were assigned randomly into the groups: WT-CD; WT-HFMCD; FGF21KO-CD; FGF21KO-HFMCD. The mice were sacri ced at week 2 for early NASH model and 3 months for advanced NASH model according to previous report [25]. Body weight, liver weight and gross anatomy of liver lobes were determined and evaluated. At respective time points, animals were sacri ced to collect serum and hepatic tissues for further biochemical analysis. Alanine aminotransferase (ALT) was measured using an ALT in nity enzymatic assay kit (ThermoFisher Scienti c Inc., Waltham, MA), while triglyceride (TG) was determined in serum and hepatic tissues using a mouse TG assay kit (Cayman Chemical Company, CA). The animal procedures were approved by the Institutional Animal Care and Use Committee of University of Louisville, which is certi ed by the American Association for Accreditation of Laboratory Animal Care.

Gross anatomy, histopathological examination and NASH scoring
The whole liver was isolated, weighted, and examined macroscopically for each animal. The harvested tissues were xed in either 10% neutral phosphate buffered formalin for para n embedding or directly embedded in Optimal Cutting Temperature medium (OCT) and frozen by liquid nitrogen. The formalin xed tissues were further embedded in para n and sectioned to a thickness of 5 µm for histological and immunohistochemical examinations. To detect lipid accumulation in the liver tissues, Oil Red O staining for was performed in OCT-embedded tissue. Hematoxylin-and-eosin (H&E) staining for histological evaluation was performed in para n-embedded frozen tissue and the images were reviewed and analyzed microscopically for determination of NASH. The Histological Scoring System for NASH is For dual-IHC staining, in brief, endogenous peroxidase was blocked with 3% hydrogen peroxide, and then with 5% BSA for 30 min to block non-speci c reaction. These tissue sections were incubated with the rstprimary antibodies (see antibody list in supplemental) over night. Tissue sections were incubated with AP-conjugated polymer (1: 300-400 dilutions with PBS) for 1 hour in room temperature, and then incubated with mixture of AP-substrate, AP-activator and AP-chromogen to develop pink color. Tissue sections were then incubated with the second-primary antibodies (see antibody list in supplemental Table 1) for 2 hours in room temperature. Tissue sections were incubated with HRP-conjugated polymer (1: 300-400 dilutions with PBS) for 1 hour in room temperature. Hematoxylin staining was performed before emerald-chromogen staining. Tissue sections and then incubated with emerald-chromogen to develop green color. Digital images were acquired with the Olympus 1×51 microscope (Olympus, Pittsburgh, PA) at 10x magni cation using the Olympus DP72 digital camera and the length of scratchwound was measured via the cellSens Dimention imaging system. Computer image analysis was performed to acquire color images from the immunohistochemical staining and de ne a standard threshold of positive staining according to the software speci cation. The computer program then quanti ed the threshold area which represented the positive staining.
Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay Using an Apop-Tag Peroxidase In Situ Apoptosis Detection Kit (Chemicon, Billerica, CA), a TUNEL assay was performed, in brief, the tissue sections were treated with proteinase K (20 mg/L) for 15 min, and then incubated with terminal deoxynucleotidyl transferase (TdT) and digoxigenin-11-dUTP for 1 hour at 37 °C.
Then, the anti-digoxigenin antibody conjugated with horseradish peroxidase (HRP) was applied and HRP substrate (DAB-H 2 O 2 ) was applied to develop brow color for visualization. Apoptotic cells were quantitatively analyzed by counting the TUNEL positive cells in tissues for each section at 20X magni cation. The apoptotic index was calculated based on the TUNEL positive cells per 100 cells.

Western blot assay
The protein levels in the tissue samples analyzed by Western blot as described previously [25]. In brief, electrophoresis was performed on 12% SDS-PAGE gel to separate proteins, and then the separated proteins were transferred to nitrocellulose membrane. The protein-loading membranes were incubated with the primary antibodies (see Table 1 antibody list in supplemental le) overnight at 4℃ and then were incubated with secondary antibody for 1 hour at room temperature. The targeted protein complexes were then visualized using ECL kit (Amersham, Piscataway, NJ). The visualized protein bands were quanti ed by densitometry analysis.

Real-Time RT-PCR (qPCR)
Total RNA was extracted from tissues using the TRIzol reagent (Invitrogen). First-strand complimentary DNA (cDNA) was synthesized from total RNA, according to manufacturer's protocol of the cDNA transcription kit (Promega, Madison, WI, USA). Quantitative PCR was carried out using the ABI 7300 realtime PCR system (Applied Biosystems, Carlsbad, CA). The primers are listed in supplemental le Table 2. The expression of targeting mRNA was quanti ed and β-actin was used as an endogenous reference. Results were expressed as fold change in gene expression.

Cell lines and in vitro study
The cells for in vitro study include a mouse hepatic cell line, FL83B (ATCC® CRL-2390), human HCC cell lines, HepG2 (ATCC® HB-8065) and Hep3B (ATCC® HB-8064), and a human colorectal adenocarcinoma cell line, Caco-2 (ATCC® HTB-37). The cells were cultured, FL83B cells in the F12K medium (ATCC), HepG2 and Hep3B cells in DMEM medium, and Caco-2 cells in EMEM medium respectively, with 10-15% fetal bovine serum. To study the effects of FFA on the cell lines regarding the FGF15/FGFR4 signaling, palmitate (PA) media (Sigma, P9767), recombinant mouse (rm) FGF-15 protein (Abcam, ab125734), recombinant human (rh) FGF21 protein (Abcam, ab217404) and BLU9931 (MedChemExpress, HY-12823), and rhFGF19 (R&D, 969-FG) were used to treat the cells. PA media was made by dissolving 2% bovine serum albumin (BSA, US Biologicals, A1311) in cell culture medium and the 100uM PA working solution was prepared from a high concentration (20 mM) stock PA solution made by dH2O heated to 70° C. Based on the previous reports, rmFGF15 protein was applied at 100 ng/ml [27], rhFGF21 protein was applied at 1.1 µg/ml [28], BLU9931 was applied at the concentration of 100nM [29], for up to 24 hours. rhFGF19 was applied at the concentration of 100 ng/ml [27], for up to 24 hours. Caco-2 cells were cocultured with HepG2 or Hep3B cells for up to 24 hours to study the FGFR4 signaling based cross-talk between enterocyte and hepatocytes. Immuno uorescent staining was performed in the cells using FITCconjugated or PE-conjugated IgG (see Table 1 antibodies listed in supplemental le) and DAPI for counterstaining.

Statistical analysis
Data being collected from the repeated experiments were presented as mean ± SD. Statistical analysis was performed by using SPSS V.17.0. Statistical signi cance for study groups was determined by ANOVA. The post hoc Tukey's test was used for analysis of any differences between groups. Group difference was considered as statistical signi cance for p < 0.05 (*), p < 0.01(**).

Lack of FGF21 worsens the HFMCD-induced NASH in mice
Based on the previous reports [25], we established an early NSAH model in FGF21KO mice with 2-weeks HFMCD feeding. The gross appearance of NASH liver, unlike the normal liver lobes with red-velvet color, showed diffusely pale-yellow-tan color lobes and increased liver weight along with increased body weight and serum ALT level in FGF21KO-HFMCD mice (Fig. 1S). Steatohepatitis was de ned in mice, as evident by the histology and con rmed by NAFLD Active Score (NAS) system which was accepted as a surrogate of histologic diagnosis of NASH. In a previous study of NAFLD patients, the score of ≥ 5 was strongly correlated with the pathological diagnosis of "de nite NASH", whereas the score ≤ 3 was designated "not NASH" [26]. All the mice with 2-weeks HFMCD feeding were found with NAS of > 5 and diagnosed as steatohepatitis. The highest NAS was found in the group of FGF21KO-HFMCD mice, with statistical signi cance compared to all other groups (Fig. 1A). Consistent to NAS, highest protein level of FGF15 by IHC was found in the group of FGF21KO-HFMCD mice, with statistical signi cance compared to all other groups (Fig. 1B). Western blot analysis con rmed the IHC results (Fig. 1C). The results indicated that FGF15 protein was signi cantly increased the liver tissues of FGF21KO mice, however the increased FGF15 protein did not show protection against the HFMCD induced steatohepatitis in FGF21KO mice.
Ileum FGF15 upregulates hepatic FGFR4-βklotho in FGF21KO-HFMCD mice FGF15/19 is an enterokine and expresses abundantly in the distal small intestine. Upon bile acids stimulation, the enterocyte secreted FGF15/19 was released to the portal blood and reached liver, binding to FGFR4 and co-receptor β-klotho of hepatocytes and triggering a signaling cascade involving hepatic bile acid, lipid and glucose metabolism [30,31]. Pphysiological expression of FGF15/19 in hepatocytes is not detected [32], but pathological FGF19 expression was detected in liver tissues of patients with hepatitis C virus-related cirrhosis or biliary cirrhosis [33]. We further investigated the resource of FGF15 production and hepatic FGFR4/β-klotho expressions. The results indicated that signi cantly increased mRNA expression in the intestinal tissues and increased serum FGF15 protein levels were found in FGF21KO-HFMCD mice, compared to all other groups( Fig. 2A-B). However, the mRNA of FGF15 was not detectable in hepatic tissues of all groups of mice (data not shown). A dual IHC staining for FGFR4 and βklotho was performed in the liver tissues. The results indicated that FGFR4 and β-klotho were coexpressed in the hepatocytes, while FGFR4/β-klotho expressions were signi cantly up-regulated in the FGF21KO-HFMCD mice, compared to all other groups (Fig. 2C). qPCR and Western blot analysis of liver tissues further con rmed the IHC results (Fig. 2D). The results indicated that the hepatic FGFR4/β-klotho signaling was signi cantly upregulated in the FGF21KO-HFMCD mice, implying that severity of NASH might associate to the aberrant FGFR4/β-klotho signaling.
FGF15 is unable to alleviate steatosis but up-regulates FGFR4 in the FGF21KD hepatocytes The binding of FGF15/19 to FGFR4/β-Klotho not only suppresses BA synthesis in hepatocytes via inhibition of cholesterol 7α-hydroxylase 1 (CYP7A1), the rate-limiting step for bile acid synthesis [34], but also activates signaling cascades leading to increased insulin sensitivity, improved glucose metabolism, as well as reduction of body weight [35]. However, it is unknown whether FGF15 can directly alleviate steatosis in hepatocytes. Therefore, we further investigated the effect of FGF15 on steatosis using FL83B cells, a benign mouse cell line of hepatocyte. To mimic the FGF21KO mice, a shRNA assay to was performed to knockdown (KD) FGF21 gene in the FL83B cells. Both FGF21KD (21KD) FL83B cells and the shRNA control (shCT) FL83B cells were challenged with palmitic acid (PA) and treated with rmFGF15 or rhFGF21. Lipid accumulation in hepatocytes were detected by Oil-red O staining. The result indicated that PA challenging signi cantly up-regulated lipid accumulation in both FGF21KD FL83B cells and shCT-FL83B cells, while treatment with rhFGF21 attenuated the up-regulated lipid accumulation. Unlike FGF21, FGF15 treatment did not show the attenuation of up-regulated lipid accumulation in the shRNA control FL83B cells. In contrast, highest level of lipid accumulation was found in the FGF21KD-FL83B cells with FGF15 treatment (Fig. 3A). As a regulatory function of hepatic FGFR4 to promote hepatic TG accumulation has been reported previously [36], we further investigated the FGFR4 levels in hepatocytes. FGF15 treatment signi cantly up-regulated FGFR4 expression in both shCT-FL83B and FGF21KD-FL83B cells challenged with PA, however, FGF21 treatment did not show up-regulation of FGFR4 levels (Fig. 3B). We further investigated the major enzymes for de novo synthesis, FA esteri cation and FA transport. Upregulated FASN, Acc1 and Acc2 (de novo synthesis), Dgat1 and Acat1(esteri cation), and Mttp, Apoα1 and CD36 (transport) were found in and FGF21KD-FL83B cells challenged with PA (Fig. 3C), implying that the increased TG storage in hepatocytes could be either from de novo synthesis or FFAs uptake. Taken together, the bioactivities of FGF15 on hepatocytes, instead of alleviating lipid accumulation, was shown to up-regulate FGFR4, while compromised FGF21 worsened steatosis in hepatocytes.

NASH progression is associated to the up-regulated FGFR4 levels in FGF21KO mice
Although the FGFR4 mediated-benign hepatic TG storage might provide protection on hepatocytes, continuously up-regulated FGFR4 signaling could play a deleterious role contributing to cell proliferation and progression of cancers [23]. According to this hypothesis, we further determined the levels of FGF15 as well its receptors FGFR4/β-klotho (KLB) in an advanced NSAH model of FGF21KO mice with HFMCD feeding for 3 months. Unlike the early stage NASH model, the gross appearance of in advanced NASH liver showed diffusely pale-yellow color lobes and increased liver weight along with increased body weight and serum TG level in FGF21KO-HFMCD mice ( Figure S2). Of note, when we analyzed the morphological changes of NASH liver, multiple nodules were detected microscopically in hepatic parenchyma of the liver tissues from FGF21KO + HFMCD mice (Fig. 4A). Histological changes showed severe steatohepatitis, as evident by signi cantly increased NAS, multiple nodules in hepatic parenchyma, signi cantly increased for Kupffer cells/macrophages detected by IHC for F4/80, and signi cantly increased serum ALT level in the FGF21KO-HFMCD mice compared to all other groups (Fig. 4A). As expected, signi cantly increased protein levels of FGF15 and FGFR4/β-klotho and were detected by Western blot analysis in the liver of FGF21KO-HFMCD mice compared to all other groups (Fig. 4B). Overexpressions of FGFR4 and β-klotho in the liver of FGF21KO-HFMCD mice with advanced NASH were con rmed by IHC and qPCR ( Figure S3). Taken together, continuously up-regulated FGFR4 expression in advanced NASH should call attention because hyperactivation of FGFR4 by FGF19 was reported in colon cancer and HCC [37]. However, FGF15/19 was also reported to down-regulate FGFR4 and β-klotho [27]. Therefore, we further studied the FGFR4 signaling in advanced NASH model in regard of the malignant potential.
Up-regulated FGFR4 signaling is coupled to brotic and malignant events in FGF21KO mice Cirrhosis and HCC have become the major liver-related clinical endpoints in NASH, while brosis progression and malignant transformation are driven by repetitive damages/repairs via apoptosis and cell proliferation [2]. To study brosis and the cellular events, we performed Sirius Red staining for brosis, IHC of PCNA for proliferation, and TUNEL assay for apoptosis in the liver tissues from the advanced NASH model with HFMCD feeding for 3 months. Signi cantly increased level of collagen ber, as showing the red color by Sirius Red staining, was found in FGF21KO-HFMCD mice compared to all other groups (Fig. 5A). Consistent to Sirius Red staining, signi cantly increased levels of apoptosis and cell proliferation, indicated by the indexes of positive apoptotic cells and positive PCNA cells, were also found in FGF21KO-HFMCD mice compared to all other groups (Fig. 5A). To evaluate the potential malignant phenotype, the epithelial-mesenchymal transition (EMT) event was investigated by a dual IHC staining for E-cadherin and vimentin in the liver tissues. The results indicated that E-cadherin expression was signi cantly down-regulated but vimentin expression was signi cantly up-regulated in the FGF21KO-HFMCD mice, compared to all other groups (Fig. 5A). Western blot analysis was further performed in the liver tissues and the results indicated that signi cant increases of cyclin D1 and cleaved caspase-3 but signi cant decrease of BCL-2 was found in FGF21KO-HFMCD mice, compared to all other groups (Fig. 5B). Taken together, increased brosis and deleterious molecular and cellular events were found in in FGF21KO-HFMCD mice which with advanced NASH, while up-regulated FGFR4 expression was coupled to these molecular and cellular events.
Blockage of FGFR4 attenuates proliferation and EMT in hepatocytes and HCC cells.
BLU9931 is a highly selective, covalent, small-molecule inhibitor that speci cally targets FGFR4 via binding within the ATP-binding pocket of FGFR4 to form a covalent bond with Cys552, resulting in > 95% proteolysis [29]. We used BLU9931 to study whether blocking FGFR4 could inhibit the cell proliferation and EMT event. With BLU9931 treatment, signi cantly decreased protein level of FGFR4 was found in FGF21KD FL83B cells either with PA challenging or without. Blockage of FGFR4 signaling by BLU9931 signi cant decreased cyclin D1 in FGF21KD FL83B cells either with PA challenging or without (Fig. 6A). Consistently, blockage of FGFR4 signaling alleviated signi cantly the EMT in FGF21KD FL83B cells either with PA challenging or without (Fig. 6B). In addition, we performed a co-cultured study using human hepatocyte cell lines (HepG2 and Hep3B) and an enterocyte cell line (Caco-2) to mimic the enterohepatic circulation in regards of the FGF19/FGFR4 signaling. When Caco-2 cells co-cultured with either HepG2 or Hep3B, signi cantly increased levels of cyclin D1 was detected in both HepG2 cells and Hep3B cells. When co-cultured with Caco-2 cells, cyclin D1 was also signi cantly up-regulated in HepG2 cells and Hep3B cells. BLU9931 treatment attenuated the up-regulated cyclin D1 levels in both HepG2 cells and Hep3B cells co-cultured with Caco-2 cells (Fig. 6C). These results demonstrated that blockage of FGFR4 could attenuate the deleterious cellular and molecular events which could be associated to NASH and HCC progression.

Discussion
As a liver safeguard [38], FGF21 was widely reported to alleviate hepatic fat stress via directly reducing hepatic lipid accumulation in an insulin-independent manner [39]. Similar to FGF21, FGF15/19 was also reported to function in controlling whole body lipid metabolism through increasing energy expenditure, FA oxidation, and decreasing de-novo lipogenesis [20]. However, it was unknown whether FGF15/19 could directly reduce hepatic lipid accumulation, especially when FGF21 was not function well such as FGF21 resistance in obesity [40]. In this study, we investigated the FGF15/FGFR4 signaling in FGF21KO mice during NASH development. Signi cant increase of FGF15 production was found in the NASH-FGF21KO mice, however the increased FGF15 protein was unable to alleviate hepatic lipid accumulation. In contrast, the up-regulated FGFR4 expression was coupled to brosis, hepatocyte injury/repair, and potential malignant events in the FGF21KO mice with advanced NASH.
Both FGF15 and FGF19 function as a negative feedback signal shutting down BA synthesis when BAs levels are high in the intestinal mucosa. Regarding the bioactivity on NASH, studies either from transgenic mice or treatment with FGF19 protein have shown that FGF19 alleviates lipid accumulation in the liver and thereby prevents NASH [20,41,42]. However, the effect of FGF15 on NASH is reportedly contradictory.
For example, a study reported that the FGF15 knockout mice fed a HFD worsened steatosis [20] but another study that also fed with HFD to the FGF15 knockout mice did not show worsened steatosis severity [43]. The protein discrepancy between FGF15 and FGF19 has been identi ed previously, in which they share only 50% sequence homology [30,44] even though they are orthologs and both are considered as endocrine FGFs because they cannot bind heparin sulfate and thus escape from extracellular matrix.
Studies of chimeric immunodepressed mice which transplanted with human hepatocytes further emphasized the bioactive discrepancy between FGF15 and FGF19, in which FGF19 administration reversed the enlarged bile acid pool size [45], but signi cantly elevated FGF15 was unable to suppress hepatic CYP7A1 expression [33]. The increased hepatic BAs level not only induces liver injury [46] but also plays an important role in the regulation of hepatocytes regeneration [47], contributing to NASH development through repetitive injury/repair. BAs reabsorbed in the intestine can increase the FGFR4/βklotho levels in hepatocytes for subsequent FGF15/FGFR4 signaling in liver [48], while hepatic FGFR4 is reported to promote hepatic lipid accumulation by either HFD or healthy diet [24] [36]. In our study, the signi cant increased FGF15 in the NASH-FGF21KO mice was unable to alleviate hepatic lipid accumulation, while up-regulated FGFR4 expression was coupled to brosis and deleterious molecular and cellular events in advanced NASH. This might be explained by, 1) although FGF15 protein was very high in liver, it was unable to suppress hepatic CYP7A1 expression [33] for BA synthesis and BAs could induce liver injury and cell death [46]; 2) the BAs mediated up-regulation of FGFR4 in hepatocytes [48] could not only promote hepatic lipid accumulation [24] [36] but also induce regeneration [47], causing repetitive injury/repair in liver; and 3) lack of FGF21 further worsened steatosis and the NASH progression. Our ndings indicate a dramatically different roles of FGF15 and FGFR4-β-klotho during NASH development in FGF21KO mice.
The liver generally maintains an appropriate size in adults. Loss of hepatocytes because of repetitive injury allows the liver to begin growing which is accepted as a fundamental mechanism(s) in cancer biology. Emerging studies have shown that FGFR4 play important roles in liver regeneration and carcinogenesis. For post-hepatectomy liver regeneration, FGFR4 was found to complex with β-klotho on hepatocyte membrane, upon FGF19/FGF15 binding, to initiate regeneration and orchestrate BAs acid detoxi cation as a protective mechanism in concurrence with the proliferative signaling [49]. In HCC, FGF19 promotes the HCC cell growth and inhibits apoptosis via activating a FGFR4-GSK3β-Nrf2 signaling cascade [50]. The aberrant FGFR4 signaling has been reported to be an oncogenic-driver pathway for HCC development [51]. In our studies, up-regulated FGF15 and FGFR4/β-klotho were found to coupled brosis and deleterious cellular events during the NASH development in FGF21KO mice. These ndings are in accordance with our previous studies in HFD-or diabetes-induced NASH mice [6,8]. This study is mainly designed to investigate FGF15/FGFR4 expression during the NASH development in FGF21KO mice. The major limitation of this study is lack of FGF15KO/overexpressing mice to elucidate the molecular mechanisms of FGF15/FGFR4 signaling pathway contributing to NASH progression and HCC development. Further studies are needed to determine the roles of FGF15/FGFR4/β-klotho as well as the downstream signaling during NASH progression. The FGFR4/β-klotho/BAs associated carcinogenesis via Gut-Liver axis is also an important issues and needs to further study in NASH-HCC transition models.

Conclusion
The increased FGF15 production in NASH-FGF21KO mice could not substitute for FGF21 to compensate its lipid metabolic bene ts thereby to prevent NASH development. The up-regulated FGFR4/β-klotho was coupled to cellular and molecular events which might associate to carcinogenic transformation. This study provided a new insight into FGF15 and FGFR4/β-klotho during NASH development and the potential pharmacological application to target FGFR4 for treatment in advanced NASH.