Exendin-4 alleviates hepatic steatosis by lowering FABP1 and raising FOXA1 expression via the Wnt/-catenin signaling pathway

Non-alcoholic fatty liver disease (NAFLD) is the leading chronic liver disease worldwide. Agonists of the glucagon-like peptide-1 receptor (GLP-1R), currently approved to treat type 2 diabetes, hold promise to improve steatosis and even steatohepatitis. However, due to their pleiotropic effects, the mechanisms underlying their protective effect on NAFLD remain elusive. We aimed to investigate these mechanisms using an in vitro model of steatosis treated with the GLP-1R agonist Exendin-4 (Ex-4). We established steatotic HepG2 cells by incubating HepG2 cells with 400 µM oleic acid (OA) overnight. Further treatment with 200nM Ex-4 for 3 hours signicantly reduced the OA-induced lipid accumulation (p < 0.05). Concomitantly, Ex-4 substantially reduced the expression levels of Fatty Acid-Binding Protein 1 (FABP1) and its primary activator, Forkhead box protein A1 (FOXA1). Interestingly, the silencing of β-catenin with siRNA abolished the effect of Ex-4 on these genes, suggesting dependency on the Wnt/ β-catenin pathway. Furthermore, after β-catenin silencing, OA treatment signicantly increased the expression of nuclear transcription factors SREBP-1 and TCF4, whereas Ex-4 signicantly decreased this upregulation. Our ndings suggest that direct activation of GLP-1R by Ex-4 reduces OA-induced steatosis in HepG2 cells by reducing fatty acid uptake via FABP1 downregulation.


Introduction
Non-alcoholic fatty liver disease (NAFLD), de ned as the excessive accumulation of lipids in the liver, is the most common cause of chronic liver disease in industrialized nations [1] and the most frequent indication for liver transplantation [2,3]. NAFLD refers to a group of liver diseases that includes simple steatosis (benign fatty in ltration), non-alcoholic steatohepatitis (NASH) (fatty in ltration plus in ammation), brosis, and cirrhosis, which occasionally progresses to hepatocellular carcinoma [4]. NAFLD is associated with several comorbidities, including type 2 diabetes (T2D), cardiovascular diseases (CVD), and chronic kidney disease (CKD) [5]. The mechanisms underlying the above associations remain elusive. However, given the liver's crucial role in many aspects of the metabolism of lipids, carbohydrates, and proteins, it is appreciated that any injury to the liver will potentially impact several organs [6]. NAFLD's etiology is not fully elucidated. However, it is accepted that visceral adiposity, insulin resistance, T2D, hypertension, and dyslipidemia are signi cant contributors to NAFLD development [7].
There is currently no approved pharmacotherapy for NAFLD. Hitherto, weight loss is the only intervention proven to be signi cantly bene cial for NAFLD patients [8]. Losing 5% of one's bodyweight improves abnormal liver tests and reduces liver fat [9], whereas losing 7-10% of one's body weight appears to reduce in ammation and injury to liver cells and may even reverse some brosis damage [10]. Unfortunately, most people nd it di cult to lose the weight they need to improve NAFLD and much more challenging to keep it off. Hence, there is an urgent need for novel therapeutic approaches to improve NAFLD independently of weight loss.
Agonists of the glucagon-like peptide-1 receptor (GLP-1R) have recently been investigated to treat NAFLD due to their bodyweight-lowering effects [11]. GLP-1 is a multifaceted hormone secreted by the L cells of the intestine [12]. Among other things, GLP1 regulates blood glucose levels by stimulating glucose-dependent insulin release and decreasing glucagon secretion, promotes proliferation of pancreatic b-cells, slows gastric emptying and inhibits satiety and food intake through effects on central nervous system centers [13,14]. This pleiotropic effect is due to the expression of the GLP-1 receptor by various organs such as the pancreas, brain, kidney, gut, lung, heart, muscle, and liver [15]. Some GLP-1R agonists, like Liraglutide (taken once daily) or Dulaglutide (taken once weekly), are already licensed for T2D and obesity management in humans due to their ability to mimic the effects of GLP-1 [16-18].
Given their weight loss-inducing effect, by reducing satiety and food intake, the impact of the GLP-1R agonists on liver fat content has been investigated in numerous in vivo studies and yielded promising results [19-22, 23, 24-35]. As a result, these drugs are suggested as potential options for treating and slowing the progression of NAFLD.
Nonetheless, it is unclear whether the protective effect of GLP-1R agonists on fat content stems from weight loss, which, among other things, increases insulin sensitivity and improves glycemia and lipid pro le, or from direct activation of the hepatic GLP-1R. Gupta and colleagues [36] were the rst to report GLP-1 receptor expression in human hepatocytes and proposed that they play a direct role in reducing hepatic steatosis in vitro through modulation of effectors of the insulin signaling pathway. Recently, Seo and coworkers [36] suggested that the GLP-1R agonist Exendin-4 (Ex-4) reduces fat content in an in vitro cell model of steatosis by inhibiting hepatic lipogenesis through activation of β-catenin signaling and modulation of the expression of several lipogenesis genes. β-catenin was also suggested to mediate the effect of GLP-1 receptor agonist Exenatide on ameliorating hepatic steatosis induced by a high fructose diet in rats [37]. The β-catenin is an intracellular signal transducer in the Wnt signaling pathway, which is involved in maintaining hepatic homeostasis and contributes to speci c hepatic characteristics, including liver metabolism [38] and metabolic zonation regeneration [39].
Hepatic lipid content and homeostasis are determined by: (a) circulating free fatty acid uptake, (b) hepatic de novo lipogenesis, (c) hepatic β-oxidation, and (d) hepatic lipid export via very-low-density lipoprotein (VLDL) [40,41]. We used HepG2 cells treated with Oleic Acid (OA) as a model of hepatic steatosis in this study to see if direct activation of the GLP-1R with Ex-4 affects any of the four processes listed above and thus improves steatosis.

Preparation of oleic acid
We prepared the oleic acid solution as in [42]. Brie y, we dissolved the powder OA (O-1008 Sigma-Aldrich, Germany) at a nal concentration of 12 mM in phosphate-buffered saline (PBS; 137 mM NaCl, 10 mM phosphate, 2.7 mM KCl, and pH 7.4) that contained 11% fatty acid-free bovine serum albumin (FFA-BSA; 0215240110, MP Biomedicals, Santa Ana, CA, USA). The solution was then sonicated and shaken at 37°C overnight using an OM10 Orbital Shaking Incubator (Ratek Instruments Pty, Ltd., Boronia, Australia). The OA solution was then ltered using a 0.22 µm lter, aliquoted, and stored at 4°C. We used a fresh aliquot for each experiment.

Induction of steatosis
To establish the steatosis cell model, we rst determined the optimal concentration of OA needed to obtain saturating levels of triglycerides (TGs). To this aim, we cultured HepG2 cells in 6-well plates at a density of 4×10 5 cells/well until 70% con uence. We then starved the cells for 6 hours in DMEM containing 1% fatty-acid-free bovine serum albumin. Following the starvation, a 16-hour incubation in DMEM containing increasing concentrations of OA (0-500 µM) at 37°C was performed, and steatosis was quanti ed (Fig. 1A).

Treatment with Exendin-4
After steatosis induction, the cells were washed and incubated in fresh DMEM containing 400 µM OA in the absence or presence of Ex-4 (E7144-.1MG, Tocris, Minneapolis, Minnesota). To determine the optimal concentration of Ex-4, we treated the steatotic cells with increasing concentrations of Ex-4 from 0 to 1 mM and with different incubation periods (3, 6, 12, and 24 hours). We then quanti ed the TG content as above. We used a fresh aliquot of EX-4 for each experiment.

Quanti cation of steatosis
We used three methods to quantify steatosis in HepG2 cells:

1) Quanti cation of Triglycerides
We measured total TGs levels using a commercial uorometric assay kit (Abcam TG quanti cation assay kit, ab65336) and a microplate reader (In nite F200 Pro; Tecan, Switzerland). The kit converts triglycerides to free fatty acids and glycerol. Glycerol is then oxidized to generate a product that reacts with a probe to generate uorescence when excited at 535 nm. The emitted uorescence is collected at 587 nm. We calculated the TGs content from a standard curve prepared for each assay using known TGs concentrations. We normalized the data to total cellular protein content.
Brie y, we grew HepG2 on 12 mm coverslips until 70% con uence, starved them, and then treated them with OA and Ex-4 as needed. After a quick wash, we xed the cells with 4% paraformaldehyde for 7 min, washed them with PBS, and then incubated them for 10 minutes with 0.2 µM BODIPY 493/503. We further labeled the nuclei by incubating the cells with 1µM DAPI for 1 min. After a nal wash with PBS, we mounted the coverslips on microscope slides used for imaging on a Zeiss LSM 870 confocal microscope, as we reported recently [44]. To analyze the images, we used ImageJ software (version 1.8.0, NIH, USA). The intracellular lipid accumulation was calculated by dividing the BODIPY uorescence intensity by the DAPI uorescence intensity. Two independent researchers analyzed 200 individual cells for each condition (untreated, steatotic, and Ex-4-treated steatotic cells) from three different experiments.

3) Relative expression of Perilipin genes
Perilipin family proteins, of which there are ve recognized members (PLIN1-5), are found on the surfaces of intracellular lipid droplets [45]. We used qRT-PCR to quantify the relative expression of PLIN1, 2, and 3 and estimate the lipid accumulation in response to OA and EX-4 treatments. The primers we utilized for the genes are listed in Table 1.   Table 1 lists the sequences of the primers we used in this study. We used Primer-BLAST (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) to design speci c primers that met the following criteria: 1) Primer pairs are unique. They will not bind to other locations in the genome except the intended gene or DNA fragment. 2) Primer pairs do not bind to each other (forming primer dimer): self-or hetero-dimer.
3) The possibility of forming the secondary structure of the primers, which may cause di culties for PCR ampli cation, is very low. 4). Tm (temperature of mismatch) of two primers is designed to be close to each other. 5) TA (Annealing temperature) is much lower than Tm. Moreover, our primers were analyzed by 'OligoAnalyzer 3.1' program from IDT company (http://www.idtdna.com/calc/analyzer).
Gene silencing with siRNA

Statistical analysis
We performed all statistical analysis and graphing with GraphPad Prism 9.0 software (GraphPad Prism v9, La Jolla, CA, USA). Data are presented as the mean ± SEM. To assess the signi cance of differences in mean values between experimental groups, we used unpaired one-way ANOVA analysis (ANOVA). Tukey's posthoc test was used to adjust multiple comparisons between experimental groups. When we silenced β-catenin, we used a twoway analysis of variance (ANOVA) to evaluate the signi cance of differences between the mean values of different experimental groups. Unless otherwise speci ed, a p-value of < 0.05 was considered signi cant.

Results
Exendin-4 reduces lipid content in OA-treated HepG2 cells By treating HepG2 cells with increasing OA concentrations for 16 hours and measuring TG accumulation, we determined the optimal concentration of OA required to induce steatosis (Fig. 1A). With 200 mM OA, we obtained a signi cant accumulation of TGs, but with 400 mM, we obtained saturating levels of TGs (p < 0.001, relative to untreated). As a result, we used 400 mM OA to induce steatosis in all our experiments. On the other hand, we found that treating steatotic cells with 200 nM Ex-4 for 3 hours is optimal for reducing lipid accumulation signi cantly (data not shown). We then compared TGs content between untreated cells, steatotic cells, i.e., cells treated with OA alone (400 µM /16h), and steatotic cells treated with Ex-4 (200 nM /3h) in the continuous presence of 400 µM OA (OA + EX-4). Figure 1B shows that in the presence of Ex-4, the TGs content was signi cantly lower than OA alone (p < 0.05), suggesting that Ex-4 reduces the OA-induced lipid accumulation. Furthermore, confocal microscopy analysis of BODIPY-stained untreated, steatotic, and Ex-4-treated steatotic cells showed that Ex-4 signi cantly decreases the number of lipid droplets (Fig. 1C), con rming the signi cant reduction of the OA-induced accumulation of lipids (p < 0.01) (Fig. 1D).
PLIN proteins play a role in forming lipid droplets and regulating lipid storage [46]. PLIN4 is absent in the liver and expressed weakly in the heart and skeletal muscle [47], whereas PLIN5 is expressed at a low level in the liver [48].
Previously, Carr and colleagues [49] reported that PLIN1 and PLIN2 proteins are upregulated in hepatic steatosis and adult NASH. Since PLINs are associated with lipid droplets, their relative expression is proportional to the number of lipid droplets.
We quanti ed gene expression of the lipid droplet binding proteins PLIN1, 2, and 3, and found that OA signi cantly increases the expression of these genes (Fig. 1E), suggesting an increase in the number of lipid droplets. However, in the presence of Ex-4, the expression of PLIN2 and PLIN3, but not PLIN1, was signi cantly lower than OA alone, indicating that Ex-4 reduces the number of lipid droplets, and thus the lipid content.

Exendin-4 counteracts the effect of OA on the expression of lipogenesis genes in HepG2 cells
Compared to untreated HepG2 cells, steatotic cells showed a signi cant upregulation of the lipogenesis genes SREBP-1, PPAR, FAS, CPT1A, SCD1, DGAT1, and DGAT2 ( Fig. 2A and 2B), while ACADL expression was signi cantly downregulated and ACC expression was unaffected. Interestingly, when compared to OA alone, the presence of Ex-4 signi cantly decreased the expression of SREBP-1, PPAR, CPT1A, ACC, DGAT1, and SCD while signi cantly increasing the expression of ACADL. DGAT2 and FAS expression remained unaffected ( Fig. 2A and  2B). Furthermore, while OA treatment did not signi cantly change the expression levels of FABP1 and FOXA1 compared to untreated cells, Ex-4 treatment signi cantly reduced the expression of these genes compared to OA treatment (Fig. 2C). The ApoB expression, on the other hand, was signi cantly increased by OA treatment, but this increase was reversed by Ex-4 treatment (Fig. 2C). Overall, these ndings support the effect of Ex-4-mediated direct activation of GLP-1R on the expression of lipid metabolism genes.
Exendin-4 activates the β-catenin pathway in HepG2 steatotic cells Seo and colleagues [36] previously reported the activation of the β-catenin pathway in response to Ex-4. Here we con rm this activation by testing the effect of Ex-4 on the expression of the nuclear factors SREPB-1 and TCF4, master transcription factors involved in the Wnt/β-catenin signaling, by silencing β-catenin with siRNA. The knockdown e ciency was tested in total mRNA and the nuclear fraction. The silencing e ciency of β-catenin was 70% and 65% for the total RNA and the nuclear fractions, respectively (Fig. 2D). As shown in Figs. 3A and 3B, OA signi cantly upregulates SREPB-1 in the nucleus after β-catenin silencing (p < 0.01), while the presence of Ex-4 strongly reverses this upregulation (p < 0.001). Similar results were observed for TCF4 (Figs. 3A and 3C). The cytoplasmic factor GSK3β is known for its involvement in Wnt-β-catenin signaling [50,51]. Therefore, we investigated Ex-4's effect on the expression of GSK3β and its inactive form, the phosphorylated GSK3β (pGSK3β). Exendin-4 reduces FABP1 and FOXA1 expression through the activation of β-catenin signaling To better understand the potential role of β-catenin as a molecular determinant through which Ex-4 mediates its bene cial effect on steatosis, we quanti ed the expression of FABP1, FOXA1, and ApoB after β-catenin silencing.
The OA signi cantly increased FABP1 expression following β-catenin knockdown (Fig. 2E, p = 0.032), whereas there was no effect without β-catenin silencing (Fig. 2C). The effect of OA on FOXA1 and ApoB expression was unaffected by β-catenin knockdown (Fig. 2E). Interestingly, after β-catenin knockdown, the presence of Ex-4 further increases, albeit not signi cantly, the OA-induced FABP1 and ApoB upregulation (Fig. 2E), in contrast to signi cant downregulation when β-catenin is not silenced (Fig. 2C). The β-catenin silencing also prevented the signi cant reducing effect of Ex-4 on FOXA1 observed in the absence of silencing (Figs. 2C and E). Overall, these ndings suggest that Ex-4's impact on the expression of these genes involves β-catenin signaling activation.

Discussion
In this study, we investigated the possible mechanisms underlying the protective effect of the GLP-1R agonist Ex-4 on hepatic steatosis in an in vitro cell model. We used the HepG2 cell line treated with oleic acid as a steatosis model and con rmed that Ex-4 signi cantly reduces OA-induced lipid accumulation. GLP-1R agonists have a wide range of complex physiological effects due to the widespread expression of the GLP-1 receptors throughout the body [14]. Because of this pleiotropic effect, distinguishing between direct, i.e., via agonist-receptor interaction, and indirect effects of these agonists in vivo is challenging. Therefore, it remains unclear whether the reduction of steatosis observed in animal and human trials in response to treatment with GLP-1R agonists results from direct activation of hepatic GLP-1R or the indirect impact such as weight loss, increased insulin sensitivity, brain-liver signals such as brain leptin [52], or other hormonal signals that these agonists might trigger [14]. To overcome this challenge, we opted for the in vitro model to ascertain that Ex-4's effect on steatosis results from direct activation of the GLP-1R.
The most important nding of our study is the signi cantly lower expression of FABP1 (also known as liver-type fatty acid-binding protein or L-FABP) in Ex-4-treated cells compared to steatotic cells. Fatty acid-binding proteins (FABPs) are small cytoplasmic proteins involved in intracellular lipid metabolisms such as fatty acid uptake, transport to mitochondria or peroxisome for oxidation, lipid synthesis, storage in lipid droplets, and regulation of nuclear receptors [53]. FABP1 is highly expressed in hepatocytes and is required for FFA uptake and shuttling [54].
Previously, Wolfrum and coworkers [55] elegantly showed that increasing the FABP1 expression by treating HepG2 cells with the potent peroxisome proliferators beza brate and Pirinixic acid leads to increased uptake of radiolabeled oleic acid by 38% and 78%, respectively. Conversely, decreasing FABP1 expression by antisense FABP1 mRNA to one-sixth of its regular expression reduces the ratio-labeled oleic acid uptake rate by 66%. Similar results agonist liraglutide on obesity-induced chronic kidney injury in obese rats showed that the agonist signi cantly reduced the lipid content and, concomitantly, the expression level of FABP1 protein in the obese kidney, relative to untreated rats [59].
In principle, four separate mechanisms may lead to hepatic lipid accumulation: (a) enhanced uptake of circulating free fatty acids, (b) increased hepatic de novo lipogenesis, (c) diminished hepatic β-oxidation, and (d) decreased hepatic lipid export via VLDL [40,41]. Therefore, one explanation for the Ex-4-induced improvement in steatosis observed in our model could be a decreased fatty acid uptake by FABP1. The transcription factor FOXA1 is among the most effective activators of human FABP1 [70]. We show that the presence of Ex-4 signi cantly reduces the FOXA1 expression relative to OA alone, which may, in turn, decrease FABP1 expression. Interestingly, FOXA1 is downregulated in liver samples from humans and rats with simple steatosis [71], probably as a feedback mechanism to reduce FAs uptake by FABP1. Furthermore, FOXA1 promotes fatty acid breakdown by inducing peroxisomal fatty acid b-oxidation. [71]. Nonetheless, given the reduced FOXA1 expression induced by Ex-4 in our study, it is unlikely that the observed Ex-4-induced TG content reduction is due to the stimulation of peroxisomal fatty acid -oxidation. Ex-4 induces a signi cant downregulation of FOXA1 (Fig. 2C) compared to steatotic cells. However, this downregulation is abrogated upon silencing of β-catenin, suggesting a role of the Wnt/β-catenin pathway in this process.
The involvement of the β-catenin signaling in the Ex-4-induced improvement in hepatic steatosis was suggested previously by Seo and coworkers [36], who showed that the β-catenin inhibitor IWR-1 abrogates the protective effect of Ex-4 against palmitate-induced steatosis. Our results also indicate the potential involvement of the βcatenin signaling pathway by showing the impact of Ex-4 on the expression of nuclear transcription factors SREBP-1, a key regulator of lipid metabolism in the liver [72], and TCF4, a central transcription factor in the βcatenin pathway, when β-catenin is silenced. Hence, after β-catenin knockdown, OA treatment signi cantly upregulates both SREBP-1 and TCF4. However, the presence of Ex-4 drastically reduces this upregulation.
Interestingly, in the context of Wnt/β-catenin signaling-dependent liver tumorigenesis, it was suggested that TCF4 might act in concert with the FOXA factors to regulate hepatocellular carcinoma-speci c Wnt target gene expression [73]. Therefore, GLP-1R stimulation may activate the β-catenin pathway, which may result in a concerted action by TCF4 and FOXA1 to regulate the expression of FABP1 and hence prevent the lipid accumulation induced by OA. It is worth noting that FABP1was suggested as a critical driver gene in hepatitis B Xprotein-induced hepatic lipid accumulation [74]. However, further investigations are warranted to decipher the complete mechanism underlying the protective effect of GLP1R agonists against hepatic steatosis.
In conclusion, the present study proposes that the direct activation of GLP-1R by Ex-4 reduces OA-induced steatosis in HepG2 cells by stimulating the Wnt/β-catenin signaling pathway, which reduces FOXA1 expression. FOXA1 downregulation, in turn, reduces FABP1 expression, which ultimately leads to a decrease in FFAs uptake.
Targeting FABP1 expression in the liver could be bene cial as a medical treatment for fatty liver disease.  Supplementary Files