FNDC5 Alleviates Hepatosteatosis by Restoring AMPK/Mtor-Mediated Autophagy, Fatty Acid Oxidation, and Lipogenesis in Mice

3262 Diabetes Volume 65, November 2016

Tong-Yan Liu,1 Xiao-Qing Xiong,1 Xing-Sheng Ren,1 Ming-Xia Zhao,1 Chang-Xiang Shi,1 Jue-Jin Wang,1 Ye-Bo Zhou,1 Feng Zhang,1 Ying Han,1 Xing-Ya Gao,1 Qi Chen,2 Yue-Hua Li,2 Yu-Ming Kang,3 and Guo-Qing Zhu1,2

FNDC5 Alleviates Hepatosteatosis by Restoring AMPK/mTOR-Mediated Autophagy, Fatty Acid Oxidation, and Lipogenesis in Mice

Diabetes 2016;65:3262–3275 | DOI: 10.2337/db16-0356

Fibronectin type III domain-containing 5 (FNDC5) protein Nonalcoholic fatty liver disease (NAFLD) is characterized induces browning of subcutaneous fat and mediates the by triacylglycerol (TG) accumulation within hepatocytes beneficial effects of exercise on metabolism. However, (1). Hepatosteatosis is strongly associated with obesity whether FNDC5 is associated with hepatic steatosis, auto- and may progress to steatohepatitis and even to end-stage phagy, fatty acid oxidation (FAO), and lipogenesis remains liver disease, including liver cirrhosis and hepatocellular unknown. Herein, we show the roles and mechanisms of carcinoma (2,3). Fatty acid b-oxidation (FAO) in mitochon- FNDC5 in hepatic steatosis, autophagy, and lipid metabo- dria is a process to shorten the fatty acids into acetyl-CoA, 2/2 lism. Fasted FNDC5 mice exhibited severe steatosis, which can be converted into ketone bodies or incorporated reduced autophagy, and FAO, and enhanced lipogenesis in into the tricarboxylic acid cycle for full oxidation (4). Accu- – the liver compared with wild-type mice. Energy deprivation mulation of lipid in the liver can be traced by the impaired induced autophagy, FAO, and AMPK activity were attenu- FAO and increased de novo lipogenesis (5). ated in FNDC52/2 hepatocytes, which were restored by METABOLISM Autophagy is a mechanism involved in cellular homeo- activating AMPK with 5-aminoimidazole-4-carboxamide ri- stasis delivering cytoplasmic content to the lysosomes for bonucleotide (AICAR). Inhibition of mammalian target of degradation to macronutrients (6). Defects in autophagy rapamycin (mTOR) complex 1 with rapamycin enhanced play a major role in metabolic dysregulation (7). Although autophagy and FAO and attenuated lipogenesis and steatosis in FNDC52/2 livers. FNDC5 deficiency exacer- some studies showed the lipogenic role of autophagy, bated hyperlipemia, hepatic FAO and autophagy impairment, most experiments supported autophagy as a lipolytic hepatic lipogenesis, and lipid accumulation in obese mice. mechanism (6). Reduced autophagic function promotes Exogenous FNDC5 stimulated autophagy and FAO gene the initial development of hepatic steatosis and progres- expression in hepatocytes and repaired the attenuated auto- sion of steatosis to liver injury, and agents to augment phagy and palmitate-induced steatosis in FNDC52/2 hepa- hepatic autophagy may have therapeutic potential in non- tocytes. FNDC5 overexpression prevented hyperlipemia, alcoholic steatohepatitis (8–10). hepatic FAO and autophagy impairment, hepatic lipogene- Fibronectin type III domain containing 5 (FNDC5) is a sis, and lipid accumulation in obese mice. These results in- type I membrane protein that has 209 amino acid dicate that FNDC5 deficiency impairs autophagy and FAO residues. FNDC5 induces browning of subcutaneous and enhances lipogenesis via the AMPK/mTOR pathway. adipocytes and mediates the beneficial effect of exercise FNDC5 deficiency aggravates whereas FNDC5 overexpres- on metabolism (11). Irisin, a cleaved and secreted frag- sion prevents the HFD-induced hyperlipemia, hepatic lipid ment of FNDC5, acts on white adipose cells to induce a accumulation, and impaired FAO and autophagy in the liver. broad program of brown fat–like development (11). Our

1Key Laboratory of Cardiovascular Disease and Molecular Intervention, Depart- This article contains Supplementary Data online at http://diabetes ment of Physiology, Nanjing Medical University, Nanjing, Jiangsu, China .diabetesjournals.org/lookup/suppl/doi:10.2337/db16-0356/-/DC1. 2 Department of Pathophysiology, Nanjing Medical University, Nanjing, Jiangsu, © 2016 by the American Diabetes Association. Readers may use this article as China long as the work is properly cited, the use is educational and not for proﬁt, and the 3 Department of Physiology and Pathophysiology, Cardiovascular Research Cen- work is not altered. More information is available at http://www.diabetesjournals ’ ’ ter, Xi an Jiaotong University School of Medicine, Xi an, Shaanxi, China .org/content/license. Corresponding author: Guo-Qing Zhu, [email protected]. Received 17 March 2016 and accepted 21 July 2016. diabetes.diabetesjournals.org Liu and Associates 3263 recent studies have shown that FNDC5 overexpression Institute, Nanjing University, Nanjing, China) were used ameliorates hyperlipemia and enhances lipolysis in adi- in the experiments. In HFD-induced obesity models, pose tissues in obese mice (12) and that irisin inhibits mice at the age of 4 weeks began to receive the HFD hepatic gluconeogenesis and increases glycogen synthesis (21.8 kJ/g, 60% of energy as fat) for 12 weeks. Normal in type 2 diabetic mice and hepatocytes (13). However, chow diet (14.7 kJ/g, 13% of energy as fat) was used as whether FNDC5 could improve hepatosteatosis, autophagy, the control (12,17). Procedures were approved by the and FAO remains unknown. Nutrient deprivation is known Nanjing Medical University Experimental Animal Care to activate AMPK, resulting in the inhibition of mammalian and Use Committee and conformed to the Guide for the target of rapamycin (mTOR) complex 1 (mTORC1), which Care and Use of Laboratory Animals (NIH publication, 8th regulates lipid metabolism, cellular proliferation, and edition, 2011). Mice were caged in an environment with autophagy (14,15). The mTORC1 inhibits peroxisome controlled temperature and humidity with free access proliferator–activated receptor (PPAR)-a activity, which to water and food under a 12-h light/dark cycle. At the regulates mitochondrial functions and FAO (16). Interest- end of experiments, mice were fasted overnight and then ingly, PPAR-a acts downstream of FNDC5 (11). The current euthanized with an overdose of pentobarbital sodium study is designed to investigate the roles and underlying (150 mg/kg, i.v.). mechanisms of FNDC5 in hepatic steatosis, autophagy, 2/2 FAO, and lipogenesis in FNDC5 mice, high-fat diet FNDC5 Overexpression in Mice – (HFD) induced obese mice, and primary hepatocytes. More- Mice at the age of 4 weeks began to receive the control over, the therapeutic effects of FNDC5 were investigated. diet or the HFD for 12 weeks. One intravenous injection of recombinant lentivirus (1 3 108 transuding units/mL, RESEARCH DESIGN AND METHODS 100 mL) expressing FNDC5 or EGFP vector was adminis- FNDC52/2 Mice and HFD-Induced Obese Mice tered at the end week 6 after the diet application (12). 2 2 Male C57BL/6 wild-type (WT) and FNDC5 / mice on Acute experiments were performed 6 weeks after the len- a C57BL/6 background (Nanjing BioMedical Research tivirus introduction.

Figure 1—FNDC5 deﬁciency causes lipid accumulation in liver after fasting. WT and FNDC52/2 mice (2 months old) were fasted for 16 h. A: Oil Red O staining showing lipid droplets in the liver sections. B: Liver TG and NEFA levels. C: Serum TG and NEFA levels. D: Serum FNDC5 levels determined with ELISA method. E: Liver FNDC5 protein expression determined with Western blot. F: Liver FNDC5 mRNA. n =6.*P < 0.05 vs. WT; †P < 0.05 vs. fed. 3264 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016

Knockdown of AMPK or Atg5 by Small Interfering RNA Quantitative Real-Time PCR Analysis in Hepatocytes RNA extracted from livers or hepatocytes was subjected to Primary hepatocytes were transfected with small interfer- reverse transcription, and quantitative real-time PCR was ing RNA (siRNA) for knockdown of AMPK or autophagy performed using the StepOnePlus Real-Time PCR System protein 5 (Atg5). Scramble siRNA was used as the control. (Applied Biosystems, Foster City, CA). Expression levels of The sequences of siRNA were as follows: AMPK: sense all genes were normalized by GAPDH levels. The sequences CGGGAUCCAUCAGCAACUATT, antisense UAGUUGCU of primers are listed Supplementary Table 1. GAUGGAUCCCGAT (18). Atg5: CCGGCCTTGGAACATCA CAGTACATCTCGAGATGTACTGTGATGTTCCAAGGTTTTTG. Measurement of Lipids and Markers of Hepatocyte Injury Enzymatic methods were used to evaluate the levels of Primary Hepatocyte Isolation and Cell Culture nonestesterified fatty acid (NEFA), TG, and cholesterol Primary hepatocytes were isolated and cultured as pre- and the activity of alanine aminotransferase (ALT) and viously described (13,19). Briefly, mice were anesthetized aspartate aminotransferase (AST) in serum or the liver. The with pentobarbital (50 mg/kg, i.p.). HEPES buffer con- kits for serum NEFA serum were bought from Wako Pure taining collagenase II (0.66 mg/mL) was perfused via Chemical Industries, Ltd. (Osaka, Japan); for serum TG, the portal vein. Livers were removed and excised asepti- AST, ALT, and hepatic NEFA from Jiancheng Bioengineering cally. Cells were dispersed and filtrated. Hepatocyte sus- Institute (Nanjing, China); and for hepatic TG and choles- pensions were purified by centrifugation in Percoll adjusted terol from Applygen Technologies Inc. (Beijing, China). to a density of 1.065 g/mL for 10 min at 50g to reduce the amount of nonparenchymal cells. With this method, the Assessment of FAO Rate In Vitro 2/2 nonparenchymal cells are less than 1% (20). Cell viability Hepatocytes isolated from WT and FNDC5 mice was determined with trypan blue dye. Plates with cell via- were incubated with 0.05 mmol/L carnitine and 0.25 14 bility greater than 95% were used for experiments. The he- mCi [ C]palmitate (GE Healthcare Life Sciences, Pittsburgh, patocytes were maintained in low glucose DMEM containing PA) for 24 h. A total of 100 mLofthemediumwasusedfor 10% FBS with penicillin (100 units/mL) and streptomycin measuring acid-soluble metabolites with scintillation coun- (100 mg/mL) at 37°C in a 5% CO2 atmosphere. ter, and 800 mL of medium was harvested on ice and mixed with ice-cold perchloric acid (70%, 200 mL) to precipitate Autophagy Monitoring BSA-fatty acid complexes. The samples were centrifuged for fl Cells were transfected with tandem GFP-red uorescent 10 min at 14,000g, and the radioactivity of the supernatant protein (RFP)-LC3 adenovirus (Hanbio, Shanghai, China) 14 was evaluated by liquid scintillation as captured CO2 (18). for 24 h according to the instructions. Cells were treated with amino acid starvation, rapamycin, or chloroquine Chemicals for 2 h to observe the autophagy flux. When autophagy FNDC5, lipopolysaccharide (LPS), palmitate, WY14643, inducts, GFP and RFP are both expressed as yellow dots carnitine, and rapamycin were bought from Sigma-Aldrich representing autophagosomes after the images emerge. Inc. (St. Louis, MO), and 5-aminoimidazole-4-carboxamide When autophagosomes fuse with lysosomes and form ribonucleotide (AICAR) was bought from Beyotime Bio- autolysosomes, the GFP degrades in an acid environment, technology Inc. (Shanghai, China). but RFP-LC3 maintains, showing as red dots (21). Statistics Oil Red O Staining and Immunohistochemistry Data are presented as mean 6 SEM. A value of P , 0.05 Livers were fixed in 4% neutral-buffered formalin phos- was considered statistically significant. Two-tailed un- phate and were embedded in paraffin or optimal cutting paired Student t tests were used to compare two treat- temperature compound, respectively. The tissues were ment groups. One-way and two-way ANOVA were used subsequently sliced into 5-mm sections. Oil Red O stain- for data analysis of more than two groups, followed by ing was used to detect the lipid content in the livers. For Bonferroni post hoc analysis. immunohistochemistry evaluation, liver sections were incubated with anti-p62 antibody (Abcam Ltd., Cambridge, RESULTS U.K.) or anti-LC3B antibody (Cell Signaling Technology, Danvers, MA). The anti-LC3B antibody showed stronger Fasting Causes Severe Lipid Accumulation in FNDC5- Deleted Mice reactivity with LC3BII, according to the manufacturer’s Fasting causes lipid mobilization from peripheral depots instructions. into the liver (22). To determine the role of FNDC5 in Western Blot hepatic lipid accumulation, the responses of lipids to fast- 2 2 Protein extracts were electrophoresed, blotted, and then ing were compared between FNDC5 / mice and WT incubated with antibodies against FNDC5, S6, phosphory- mice. Lipid accumulation in livers was increased in 2 2 lated (P)-S6, AMPKa,P-AMPKa,LC3B,Raptor,P-Raptor, FNDC5 / mice and became severe after fasting for 16 h ULK1, P-ULK1, and GAPDH (Cell Signaling Technology) compared with WT mice (Fig. 1A and B). Fasting caused 2 2 and P62 (Abcam Ltd.) with appropriate secondary horserad- more increases in serum NEFA and TG levels in FNDC5 / ish peroxidase–conjugated antibodies, and then developed. mice than in WT mice (Fig. 1C). The efficiency of FNDC5 diabetes.diabetesjournals.org Liu and Associates 3265 gene knockout was confirmed by serum FNDC5 levels and levels of autophagy genes, including UNC51-like kinase liver FNDC5 expressions (Fig. 1D–F). These results indicate 1 (Ulk1), Ulk2, Atg5, Atg8, and Atg10 were downregulated 2 2 that FNDC5 prevents excessive lipid accumulation in livers. in FNDC5 / mouse livers (Fig. 4A). Fasting reduced p62 FNDC5 Deficiency Causes Defects in AMPK/ expression and increased LC3BII/LC3BI in WT livers but 2/2 fi PPAR-a–Mediated FAO not in FNDC5 livers (Fig. 4B), which was con rmed by Reduced FAO increases hepatic lipid accumulation, and chronic the liver immunohistochemistry (Supplementary Fig. 4). starvation increases FAO gene expressions via transcriptional Moreover, amino acid deprivation caused a greater enhancement in autophagy flux in WT hepatocytes than in mechanisms (5). Activation of AMPK stimulates FAO via 2 2 FNDC5 / hepatocytes (Fig. 4C). These results indicate PPAR-a signaling (18). Thus, the roles of FNDC5 in regulating 2/2 FAO and its downstream pathway were investigated. Basal and that autophagy in the liver is reduced in FNDC5 mice fasting-induced FAO gene expressions (Ppara, Hmgcs2, Cpt1, and that FNDC5 is required for the fasting-induced auto- Acox1, Sirt3,andCyp4a10)(Fig.2A) and AMPK phosphoryla- phagy response. 2 2 tion (Supplementary Fig. 1) were reduced in FNDC5 / mice Nutrient deprivation activates AMPK, which phos- livers. Activation of AMPK with AICAR increased FAO gene phorylates Raptor (an essential component of mTORC1), 2 2 expressions (Fig. 2B) and reduced TG levels in FNDC5 / resulting in net repression of mTORC1 signaling (26,27). mice livers (Fig. 2D). Furthermore, AICAR stimulated liver AMPK stimulates autophagy through direct ULK1 phos- AMPK activation and the subsequent mTORC1 inhibition in phorylation (28). Metformin, a biguanide antihyperglyce- 2 2 WT and FNDC5 / mice (Supplementary Fig. 2). Knockdown mic agent, causes carbohydrate starvation and AMPK of AMPK with siRNA increased TG contents in WT and activation (29). We found that either amino acid starva- 2 2 FNDC5 / hepatocytes (Fig. 2E). WY14643, a PPAR-a ago- tion (Fig. 4D) or metformin (Supplementary Fig. 5) stimulated AMPK, Raptor, and ULK1 phosphorylation in WT nist, caused a greater increase in the expressions of PPAR-a 2/2 target genes (Hmgcs2, Cpt1, Acox1, Ehhadh, Acsl1, Peci, hepatocytes but not in FNDC5 hepatocytes. Activation 2/2 of AMPK with AICAR restored the reduced autophagy Cyp4a10,andCyp4a12) in WT hepatocytes than in FNDC5 2/2 2/2 hepatocytes (Fig. 2C). WY14643 inhibited palmitate-induced in FNDC5 hepatocytes (Fig. 4E) and FNDC5 livers 2/2 (Supplementary Fig. 6). Knockdown of AMPK with siRNA lipid accumulation in WT and FNDC5 hepatocytes (Fig. 2/2 2F and Supplementary Fig. 3). It induced a greater increase attenuated autophagy in WT and FNDC5 hepatocytes 2 2 fi in FAO rate in WT hepatocytes than in FNDC5 / hepato- (Fig. 4F). These ndings indicate that AMPK is an essential cytes (Fig. 2F). These findings indicate that FAO is reduced in downstream effector of FNDC5 in mediating its effect on 2 2 FNDC5 / mice livers and that FNDC5 is important for fast- autophagy. ing-induced FAO gene expressions, which are mediated by the AMPK pathway. FNDC5 deficiency causes impairment in Rapamycin Repairs the Attenuated Autophagy fi AMPK/PPAR-a–mediated FAO. in FNDC5 De ciency Activation of mTORC1 is required for hepatic lipid Rapamycin Prevents the Impaired FAO and Increased accumulation (30). Reduced Raptor phosphorylation in fi 2 2 Lipogenesis in FNDC5 De ciency FNDC5 / livers (Fig. 4D) suggests a possibility that – Because nutrient deprivation induced AMPK represses the increased mTORC1 might be involved in the signaling mTORC1 activity (16), we further investigated whether pathway of impaired autophagy. Immunohistochemistry anal- mTORC1 is involved in the effects of FNDC5 on FAO. ysis showed that inhibition of mTORC1 with rapamycin pre- fi FNDC5 de ciency caused an increase in liver ribosomal S6 pro- vented the increased p62 expression and the reduced LC3B 2 2 tein phosphorylation, a marker of mTORC1 activity (23), which expression in FNDC5 / livers (Fig. 5A). Consistently, West- was suppressed by rapamycin, an mTORC1 inhibitor (Fig. 3A). ern blot analysis showed that rapamycin promoted autophagy It reduced the increased liver TG contents in both fed and 2/2 2/2 2/2 in WT and FNDC5 livers (Fig. 5B)andinFNDC5 fasted states in FNDC5 mice (Fig. 3D). Rapamycin restored hepatocytes (Fig. 5D), and increased autophagy flux in WT the reduced FAO gene expressions (Ppara, Hmgcs2, Cpt1,and 2/2 2/2 and FNDC5 hepatocytes. Chloroquine, an autophagy in- Sirt3)inFNDC5 mice under fed and fasting conditions (Fig. hibitor, blocked autophagosomes fusing with lysosome to 3B). The results indicate that increased mTORC1 activity con- form autolysosomes, as showed by increased autophagosome fi tributes to the reduced FAO in mice with FNDC5 de ciency. accumulation and decreased autolysosomes in both WT and 2 2 Lipogenesis is a factor contributing to hepatic lipid accumulation FNDC5 / hepatocytes (Fig. 5E). These results indicate that (24). The mRNA levels of lipogenic genes (Srebp1c, Dgat1, Fasn, 2 2 increased mTORC1 activity contributes to the attenuated / 2 2 and Scd1) were raised in FNDC5 livers under fed and fasted autophagy in FNDC5 / livers. Moreover, the liver function conditions, which were attenuated by rapamycin (Fig. 3C). markers serum ALT and AST levels were increased in fi 2 2 These results indicate that FNDC5 de ciency potentiates FNDC5 / mice, which were restored by rapamycin (Fig. lipogenesis via increased mTORC1 activity. 5C). Palmitate-induced lipid accumulation is generally used FNDC5 Deficiency Causes Attenuation in Autophagy as a cellular steatosis model (31). To ascertain whether 2 2 and AMPK Activity reduced autophagy in FNDC5 / mice is involved in lipid Reduced p62 expression and increased LC3BII generation accumulation, the effects of rapamycin and Atg5 siRNA on from LC3BI are markers of autophagy (25). The mRNA palmitate-induced lipid accumulation in primary hepatocytes 3266 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016

Figure 2—Impaired FAO in livers of FNDC52/2 mice. A: Hepatic FAO gene expression after starvation for 16 h. B: Effects of AICAR (200 mg/kg, i.p.) for 5 days on hepatic FAO gene expression. C: Effects of WY14643 on PPAR-a target gene expression in hepatocytes treated with palmitate (125 mmol/L) and WY14643 (30 mmol/L) for 6 h. n =3inA–C. D: Effects of AICAR (200 mg/kg, i.p.) for 5 days on hepatic TG contents. E: Effects of AMPK-siRNA (5 nmol/L) for 48 h on TG levels in hepatocytes. F: Effects of WY14643 on TG levels and FAO rate in hepatocytes treated with [14C]palmitate (0.25 mCi), carnitine (0.05 mmol/L), and WY14643 (30 mmol/L) for 24 h (n =6inD–F). *P < 0.05 vs. WT; †P < 0.05 vs. fed; ‡P < 0.05 vs. saline or DMSO; #P < 0.05 vs. control siRNA.

were investigated. Palmitate caused more lipid accumula- FNDC5 Deﬁciency Aggravates Hepatosteatosis, Lipid 2 2 tion in FNDC5 / hepatocytes than in WT hepatocytes, Metabolic Disturbance, and Impairment of Autophagy which was inhibited by rapamycin. Inhibition of auto- in HFD-Induced Obese Mice phagy by knockdown of Atg5, an essential autophagy An HFD is generally used to induce obesity in rodents ﬁ gene, with Atg5 siRNA attenuated the role of rapamycin (12,32). We investigated whether FNDC5 de ciency 2 2 in reducing lipid accumulation in FNDC5 / hepatocytes causes more severe hepatosteatosis, lipid metabolic distur- (Fig. 5F). Moreover, rapamycin reduces TG content in bance, and impairment of autophagy in mice with obesity 2 2 hepatocytes in FNDC5 / mice (Supplementary Fig. 7). induced by an HFD for 12 weeks. Liver weight and the liver 2 2 These results indicate that the ability of rapamycin to re- weight–to–body weight ratio were greater in FNDC5 / / 2 2 duce lipid accumulation in FNDC5 / hepatocytes is largely HFD mice than in WT/HFD mice, but the differences dependent on autophagy. in body weight and food intake between WT/HFD and diabetes.diabetesjournals.org Liu and Associates 3267

Figure 3—Rapamycin attenuates the impaired FAO and enhanced lipogenesis in the liver of FNDC52/2 mice. Mice were treated with rapamycin (5 mg/kg) for 3 days, followed by fasting for 16 h. A: S6 phosphorylation. B: FAO-related gene expressions. C: Lipogenic gene expressions. n =3inA–C. D: TG levels (n = 6). *P < 0.05 vs. WT; †P < 0.05 vs. saline; ‡P < 0.05 vs. fed.

2 2 FNDC5 / /HFD mice were not significant (Fig. 6A). reached their maximal at the concentration of 100 nmol/L FNDC5 deficiency aggravated the HFD-induced increases (Fig. 7B). Exogenous FNDC5 attenuated TG accumula- 2 2 in NEFA, TG, and cholesterol levels in serum and livers tion in FNDC5 / hepatocytes (Supplementary Fig. 9). (Fig. 6B). FAO gene expressions (Ppara, Hmgcs2, Cpt1, LPS is known to stimulate autophagy in hepatocytes Acox1,andSirt3) in livers were downregulated (Fig. 6E), (33); thus, we compared the role of FNDC5 with LPS whereas lipogenic gene expressions (Srebp1c, Dgat1, Fasn, in stimulating autophagy. FNDC5 reduced p62 and in- and Scd1) were upregulated (Supplementary Fig. 8) in creased LC3BII levels, similar to the effects of LPS (Fig. 2 2 FNDC5 / /HFD mice compared with WT/HFD mice. Hep- 7C). Importantly, palmitate-induced lipid accumulation 2 2 2 2 atosteatosis was more severe in FNDC5 / /HFD mice in primary FNDC5 / hepatocytes was prevented by than in WT/HFD mice (Fig. 6D). In HFD mice, p62 protein FNDC5 (Fig. 7D),andFNDC5potentiatedautophagy 2 2 2 2 expression was increased in FNDC5 / livers compared in FNDC5 / hepatocytes (Fig. 7E). In addition, exoge- with WT livers (Fig. 6F), and serum ALT and AST levels nous FNDC5 attenuated the AMPK inhibition and mTOR 2 2 2 2 were higher in FNDC5 / mice than in WT mice (Fig. 6C). activation in FNDC5 / hepatocytes (Fig. 7F). These Moreover, deletion of the FNDC5 gene exacerbated AMPK results indicate that exogenous FNDC5 promotes FAO inhibition and enhanced mTOR activation in livers from andautophagyandpreventstheFNDC5deficiency– HFD mice (Fig. 6G). These findings indicate that FNDC5 de- induced lipid accumulation and autophagy impairment ficiency causes more severe hepatosteatosis, lipid metabolic in vitro. disturbance, and impairment of autophagy in obese mice. FNDC5 Overexpression Attenuates Hepatosteatosis Exogenous FNDC5 Enhances FAO and Autophagy and the FAO and Autophagy Impairment in In Vitro HFD-Induced Obese Mice Primary WT hepatocytes incubated with FNDC5 (100 nmol/L) The effects of lentiviral vector–mediated FNDC5 overex- for 12 h or 24 h increased FAO gene expressions (Ppara, pression in HFD-induced obese mice were investigated to Hmgcs2, Cpt1,andAcox1)invitro(Fig.7A). The effects determine whether FNDC5 could be used as a therapeutic of FNDC5 for 24 h on these FAO gene expressions almost strategy for hepatosteatosis and impairment of FAO and 3268 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016

Figure 4—Reduced autophagy in the livers of FNDC52/2 mice. A: Autophagy gene expressions in the liver. B: Expressions of p62 and LC3B in the liver, with or without fasting for 16 h. C: Images show LC3 staining in GFP-RFP-LC3 adenovirus–infected hepatocytes, with or without amino acid deprivation for 2 h. Green dots, autophagosomes; red dots, autolysosomes; yellow dots, autophagosomes. D: Expressions of FNDC5, P-AMPK, Raptor, and ULK1 in hepatocytes, which were subjected to serum starvation for 8 h, followed by amino acid starvation. E: Effects of AICAR (1 mmol/L) for 24 h on P-AMPK, p62, and LC3B expressions in hepatocytes. F: Effects of AMPK siRNA (5 nmol/L) for 48 h on P-AMPK, p62, and LC3B expressions in hepatocytes. n =4.*P < 0.05 vs. WT; †P < 0.05 vs fed or saline or control-siRNA. diabetes.diabetesjournals.org Liu and Associates 3269

Figure 5—Rapamycin repairs the attenuated autophagy in livers of FNDC52/2 mice. A–C: Mice treated with rapamycin (5 mg/kg) for 3 days, followed by fasting for 16 h. A: Hepatic immunohistochemistry for p62 and LC3B. B: Hepatic p62 and LC3B protein expressions. C: Serum ALT and AST. D: Effects of rapamycin (5 mmol/L for 2 h) on p62 and LC3B expressions in hepatocytes. E: Images show the effects of rapamycin (50 nmol/L) or chloroquine (10 mmol/L) on autophagy in GFP-RFP-LC3 adenovirus–infected hepatocytes. Green dots, autophagosomes; red dots, autolysosomes; yellow dots, autophagosomes. F: Oil Red O staining shows lipid accumulation. Hepatocytes were incubated with control siRNA and Atg5 siRNA, followed by treatment with palmitate (250 mmol/L), with or without rapamycin (5 mmol/L) for 16 h. n =3.*P < 0.05 vs. WT; †P < 0.05 vs. saline. 3270 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016 autophagy in obesity. FNDC5 overexpression attenuated Hepatic steatosis is linked to being obese or over- the HFD-induced increases in NEFA, TG, and cholesterol weight in most cases (39). We found that FNDC5 defi- levels in serum and the liver (Fig. 8A) and reduced the ciency deteriorated hepatosteatosis accompanying FAO HFD-induced lipid accumulation in the liver (Fig. 8B). It and autophagy impairment in HFD-induced obesity, restored the reduced FAO gene expression (Fig. 8C) whereas FNDC5 overexpression alleviated hepatosteatosis and the increased lipogenic gene expression (Fig. 8D)in and repaired the reduced FAO and autophagy in HFD- HFD-induced obese mouse livers. Immunohistochemistry induced obese mice. Exogenous FNDC5 stimulated FAO showed that the increased p62 expression and reduced and autophagy and attenuated the palmitate-induced he- LC3B expression in the liver were prevented by FNDC5 patic lipid accumulation in primary hepatocytes. These overexpression (Fig. 8E), which were further confirmed by findings indicate the importance of FNDC5 in attenuating Western blot analysis (Fig. 8F). FNDC5 overexpression hepatic steatosis. Administration of FNDC5 or increased attenuated the AMPK inhibition and mTOR activation FNDC5 production is expected to be a therapeutic regi- in livers from HFD mice (Fig. 8G). The effectiveness of men for preventing hepatosteatosis, FAO, and autophagy FNDC5 overexpression in the experiments was confirmed impairment in obesity. by the changes of serum FNDC5 levels (Supplementary Most fatty acids in the liver are metabolized by FAO Fig. 10). These findings indicate that long-term increased (40). AMPK represses mTORC1 activity (16) and main- FNDC5 effectively prevents hepatosteatosis and attenu- tains energy homeostasis via suppressing cellular ATP- ates the FAO and autophagy impairment in HFD-induced consuming processes and stimulating ATP-producing obese mice. catabolic pathways, including FAO (41). AMPK inhibits protein synthesis and mTOR signaling (42), whereas DISCUSSION mTORC1 inhibits PPAR-a expression and function (16). Hepatic steatosis is generally regarded as the hepatic PPAR-a stimulates the expression of FAO genes and is manifestation of the metabolic syndrome in diabetes and implicated in nonalcoholic steatohepatitis (43). We found obesity and is thought to be the initial stage in NAFLD. that FNDC5 deficiency reduced AMPK, Raptor, and ULK1 NAFLD is characterized by the accumulation of TG in lipid phosphorylation. Activating AMPK or PPAR-a partially droplets within hepatocytes (1,2,34). The primary novel restored the downregulation of PPAR-a and FAO gene findings in the current study are that FNDC5 plays critical expressions. Furthermore, AMPK activation was found 2 2 roles in reducing hepatic lipid accumulation by restoring in association with TG levels in FNDC5 / hepatocytes. AMPK/mTOR-mediated autophagy, FAO, and lipogenesis. The PPAR-a agonist WY14643 increased the FAO rate FNDC5 deficiency deteriorated hepatosteatosis, FAO, and and reduced lipid accumulation in hepatocytes with autophagy impairment in obesity, whereas FNDC5 over- FNDC5 deficiency. Inhibition of mTORC1 attenuated expression alleviated hepatosteatosis and improved FAO the increased mTORC1 activity and TG levels and partially and autophagy in obesity. restored the attenuated Ppara and FAO gene expressions 2 2 Hepatic FAO is increased in response to energy in FNDC5 / livers. These data indicate that FNDC5 de- demand in the fasted state (24). Fasting upregulates TG ficiency reduces AMPK phosphorylation, which subse- hydrolysis to supply NFFA for oxidation to meet cellular quently causes mTORC1 activation, and thus, inhibits energy needs (35). An alternative energy source with re- PPAR-a target gene expressions and FAO. It has been spect to energy deprivation is provided by the breakdown shown that ghrelin upregulates autophagy via AMPK/ of cellular components by autophagy (35,36). Induction mTOR restoration (44). Blockage of the mTOR pathway of autophagy corrects hepatic lipid over-accumulation restores endoplasmic reticulum stress–induced autophagy (37). Defective autophagy is involved in NAFLD (38). (45). We found that the FNDC5 deficiency–induced defect We found that hepatic FAO in the fed state was reduced in autophagy was prevented by AMPK activation and de- 2 2 in FNDC5 / mice and that the fasting-induced FAO teriorated by AMPK suppression. Inhibition of mTORC1 2 2 enhancement was much weaker in FNDC5 / mice than in restored autophagy impairment and attenuated the liver 2 2 WT mice. Consistently, FNDC5 deficiency caused a mild he- injury in FNDC5 / mice. Furthermore, rapamycin allevi- 2 2 patic lipid accumulation in the fed state but a severe hepatic ated palmitate-induced lipid accumulation in FNDC5 / lipid accumulation in the fasted state. Palmitate induced more hepatocytes, which was abolished by the knockdown of 2 2 lipid accumulation in FNDC5 / hepatocytes than in WT the essential autophagy gene Atg5. These results reveal hepatocytes in vitro. Although FNDC5 deficiency had no that FNDC5 deficiency causes AMPK inhibition, mTORC1 significant effect on hepatic autophagy in the fed state, activation, and autophagy defects. The beneficial effect of the fasting-induced autophagy enhancement response in mTORC1 inhibition was largely dependent on restoration 2 2 WT mice almost disappeared in FNDC5 / mice. Inhibi- of autophagy, further suggesting that the autophagic defect tion of autophagy increased lipid accumulation in WT and in FNDC5 deficiency is partially responsible for hepatic 2 2 FNDC5 / hepatocytes. These findings indicate that steatosis. It is noted that rapamycin treatment strongly FNDC5 is strongly associated with hepatic FAO and auto- suppressed the liver S6 phosphorylation but has modest phagy, which at least partially contribute to the reduction roles in increasing FAO and reducing lipogenic gene expres- of hepatic lipid accumulation, particularly in the fasting state. sions in fasting FNDC5-knockout mice. This discrepancy diabetes.diabetesjournals.org Liu and Associates 3271

Figure 6—FNDC5 deﬁciency exacerbates lipid accumulation and attenuated FAO and autophagy in the liver caused by 12 weeks of the HFD in mice. A: Body weight (BW), liver weight (LW), LW-to-BW ratio, and average food intake. B: NEFA, TG, and cholesterol (CHO) levels in serum and liver. C: Serum ALT and AST levels (n =7inA–C). D: Oil Red O staining shows lipid droplets in the liver sections. E: FAO- related gene expression in livers. F: Expression of p62 protein. G: Phosphorylation of AMPK, Raptor, and S6. n =3inD–G.*P < 0.05 vs. WT; †P < 0.05 vs. control.

2 2 suggests a possibility that some other signal pathways may lipogenic gene expressions were increased in FNDC5 / be involved in regulating FAO and lipogenesis. mice compared with WT mice. FNDC5 overexpression Lipogenesis is another important factor contributing prevented the increased lipogenic gene expressions in to lipid accumulation in the liver (24). Lipogenic gene HFD-induced obese mice. On the one hand, these results expressions resulting from mTORC1 inhibition were de- indicate that FNDC5 deﬁciency stimulates lipogenesis via 2 2 creased in FNDC5 / mice. In HFD-induced obese mice, the mTOR pathway, which is involved in lipid accumulation 3272 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016

Figure 7—Exogenous FNDC5 enhances FAO and autophagy in hepatocytes. A: Time effects of FNDC5 (100 nmol/L) on FAO-related gene expressions in WT hepatocytes. B: Dose effects of FNDC5 (20, 100, and 200 nmol/L) on FAO-related gene expressions in WT hepatocytes. C: Effects of LPS (100 ng/mL) or FNDC5 (100 nmol/L) on p62 and LC3B expressions in WT hepatocytes. D: Oil Red O staining shows that FNDC5 (100 nmol/L) prevented palmitate-induced (250 mmol/L) lipid accumulation (red color) in WT and FNDC52/2 hepatocytes. E: Effects of FNDC5 (100 nmol/L) on p62 and LC3B expressions in FNDC52/2 hepatocytes. F: Effects of FNDC5 (100 nmol/L) on the phosphorylation of AMPK, Raptor, and S6 in WT and FNDC52/2 hepatocytes. n =3.*P < 0.05 vs. PBS; †P < 0.05 vs. WT.

in the liver of HFD-induced obesity. On the other hand, which was further conﬁrmed in the current study. These 2 2 serum TG and NEFA levels were increased in FNDC5 / results revealed that FNDC5 plays a critical role in prevent- mice, particularly in the fasting state. FNDC5 deﬁciency ing hyperlipemia. deteriorated hyperlipemia in HFD-induced obese mice. Previous study has showed that FNDC5 overexpres- Our previous study showed that FNDC5 overexpression sion in HFD-induced obese mice increases energy expen- prevented hyperlipemia in HFD-induced obese mice (12), diture, attenuates hyperglycemia and insulin resistance, diabetes.diabetesjournals.org Liu and Associates 3273

Figure 8—FNDC5 overexpression repairs attenuated FAO and autophagy in livers in HFD-induced obese mice. A: NEFA, TG, and cholesterol (CHO) levels in serum and liver (n = 6). B: Oil Red O staining shows the lipid accumulation in liver. C: Hepatic FAO gene expressions. D: Hepatic lipogenic gene expressions. E: Immunohistochemistry of liver sections for p62 and LC3B. F: Hepatic p62 and LC3B protein expressions. G: Phosphorylation of AMPK, Raptor, and S6 (n =3inB–G). *P < 0.05 vs. vector; †P < 0.05 vs. control. and activates lipolysis in adipose tissues (12). Irisin, a (13). Strong irisin immunoreactivity has been found cleaved and secreted fragment of FNDC5, reduces gluco- in the liver (46,47). Serum irisin concentrations were in- neogenesis, increases glycogenesis, and improves insulin versely associated with liver TG contents in the liver in resistance in streptozotocin/HFD-induced type 2 diabetes obese adults (48). A limitation in the current study is that 3274 FNDC5 Attenuates Hepatic Steatosis Diabetes Volume 65, November 2016 we did not investigate whether the effects of FNDC5 are 15. Zoncu R, Efeyan A, Sabatini DM. mTOR: from growth signal integration to caused by its cleaved fragment irisin. cancer, diabetes and ageing. Nat Rev Mol Cell Biol 2011;12:21–35 In summary, FNDC5 reduces hepatic lipid accumulation 16. Sengupta S, Peterson TR, Laplante M, Oh S, Sabatini DM. mTORC1 controls via AMPK/mTOR-mediated autophagy and FAO enhance- fasting-induced ketogenesis and its modulation by ageing. Nature 2010;468: 1100–1104 ment and de novo lipogenesis reduction. FNDC5 deficiency 17. Francés DE, Motiño O, Agrá N, et al. Hepatic cyclooxygenase-2 expression aggravates whereas FNDC5 overexpression prevents he- protects against diet-induced steatosis, obesity, and insulin resistance. Diabetes patic steatosis, hyperlipemia, impaired FAO, and auto- 2015;64:1522–1531 phagy, and enhanced lipogenesis in obesity (Supplementary 18. Tateya S, Rizzo-De Leon N, Handa P, et al. VASP increases hepatic fatty Fig. 11). FNDC5 can be used as a therapeutic regimen for acid oxidation by activating AMPK in mice. Diabetes 2013;62:1913–1922 preventing hepatosteatosis and impairment of FAO and 19. Bamji-Mirza M, Zhang W, Yao Z. Expression of human hepatic lipase autophagy in obesity. negatively impacts apolipoprotein A-I production in primary hepatocytes from Lipc-null mice. J Biomed Res 2014;28:201–212 20. Klingmüller U, Bauer A, Bohl S, et al. Primary mouse hepatocytes for Acknowledgments. The authors thank the generous support of the systems biology approaches: a standardized in vitro system for modelling of Collaborative Innovation Center for Cardiovascular Disease Translational signal transduction pathways. Syst Biol (Stevenage) 2006;153:433–447 Medicine. 21. Liu Y, Palanivel R, Rai E, et al. Adiponectin stimulates autophagy and re- Funding. This study was supported by National Natural Science Foundation of duces oxidative stress to enhance insulin sensitivity during high-fat diet feeding China (31271213, 31571167, and 91439120). in mice. Diabetes 2015;64:36–48 Duality of Interest. No potential conflicts of interest relevant to this article 22. McCue MD. Starvation physiology: reviewing the different strategies animals were reported. use to survive a common challenge. Comp Biochem Physiol A Mol Integr Physiol Author Contributions. T.-Y.L., X.-Q.X., X.-S.R., M.-X.Z., C.-X.S., J.-J.W., 2010;156:1–18 Y.-B.Z., and F.Z. performed the experiments. T.-Y.L., X.-Q.X., Y.H., X.-Y.G., and 23. Morran DC, Wu J, Jamieson NB, et al.; Australian Pancreatic Cancer Genome G.-Q.Z. analyzed the data. T.-Y.L., X.-Y.G., Q.C., Y.-H.L., Y.-M.K., and G.-Q.Z. Initiative (APGI). Targeting mTOR dependency in pancreatic cancer. Gut 2014;63: were involved in study design. T.-Y.-L., X.Y-.G., Q.C., Y.-H.L., Y.-M.K., and G.-Q.Z. 1481–1489 edited the manuscript. T.-Y.L. and G.-Q.Z. wrote the manuscript. G.-Q.Z. is the 24. Bechmann LP, Hannivoort RA, Gerken G, Hotamisligil GS, Trauner M, guarantor of this work and, as such, had full access to all the data in the study and Canbay A. The interaction of hepatic lipid and glucose metabolism in liver dis- takes responsibility for the integrity of the data and the accuracy of the data analysis. eases. J Hepatol 2012;56:952–964 References 25. Kroemer G, Mariño G, Levine B. Autophagy and the integrated stress response. Mol Cell 2010;40:280–293 1. Adams LA, Ratziu V. Non-alcoholic fatty liver - perhaps not so benign. J 26. Hales EC, Taub JW, Matherly LH. New insights into Notch1 regulation of the Hepatol 2015;62:1002–1004 PI3K-AKT-mTOR1 signaling axis: targeted therapy of g-secretase inhibitor re- 2. Byrne CD, Targher G. NAFLD: a multisystem disease. J Hepatol 2015;62 sistant T-cell acute lymphoblastic leukemia. Cell Signal 2014;26:149–161 (Suppl.):S47–S64 3. Marra F, Lotersztajn S. Pathophysiology of NASH: perspectives for a tar- 27. Zhang MZ, Wang Y, Paueksakon P, Harris RC. Epidermal growth factor geted treatment. Curr Pharm Des 2013;19:5250–5269 receptor inhibition slows progression of diabetic nephropathy in association with 4. Eaton S, Bartlett K, Pourfarzam M. Mammalian mitochondrial beta-oxidation. a decrease in endoplasmic reticulum stress and an increase in autophagy. Di- – Biochem J 1996;320:345–357 abetes 2014;63:2063 2072 5. Koo SH. Nonalcoholic fatty liver disease: molecular mechanisms for the 28. Egan DF, Shackelford DB, Mihaylova MM, et al. Phosphorylation of ULK1 hepatic steatosis. Clin Mol Hepatol 2013;19:210–215 (hATG1) by AMP-activated protein kinase connects energy sensing to mitophagy. – 6. Kwanten WJ, Martinet W, Michielsen PP, Francque SM. Role of autophagy Science 2011;331:456 461 in the pathophysiology of nonalcoholic fatty liver disease: a controversial issue. 29. Loos JA, Cumino AC. In vitro anti-echinococcal and metabolic effects of World J Gastroenterol 2014;20:7325–7338 metformin involve activation of AMP-activated protein kinase in larval stages of 7. Codogno P, Lotersztajn S. When autophagy chaperones liver metabolism. Echinococcus granulosus. PLoS One 2015;10:e0126009 Cell Metab 2014;20:392–393 30. Sapp V, Gaffney L, EauClaire SF, Matthews RP. Fructose leads to hepatic 8. Amir M, Czaja MJ. Autophagy in nonalcoholic steatohepatitis. Expert Rev steatosis in zebrafish that is reversed by mechanistic target of rapamycin (mTOR) Gastroenterol Hepatol 2011;5:159–166 inhibition. Hepatology 2014;60:1581–1592 9. Codogno P, Meijer AJ. Autophagy in the liver. J Hepatol 2013;59:389–391 31. Park JY, Kim Y, Im JA, Lee H. Oligonol suppresses lipid accumulation and 10. Xiao Y, Liu H, Yu J, et al. Activation of ERK1/2 ameliorates liver steatosis in improves insulin resistance in a palmitate-induced in HepG2 hepatocytes as a leptin receptor-deficient (db/db) mice via stimulating ATG7-Dependent auto- cellular steatosis model. BMC Complement Altern Med 2015;15:185 phagy. Diabetes 2016;65:393–405 32. Aoun M, Fouret G, Michel F, et al. Dietary fatty acids modulate liver mi- 11. Boström P, Wu J, Jedrychowski MP, et al. A PGC1-a-dependent myokine tochondrial cardiolipin content and its fatty acid composition in rats with non that drives brown-fat-like development of white fat and thermogenesis. Nature alcoholic fatty liver disease. J Bioenerg Biomembr 2012;44:439–452 2012;481:463–468 33. Chen C, Deng M, Sun Q, Loughran P, Billiar TR, Scott MJ. Lipopolysaccharide 12. Xiong XQ, Chen D, Sun HJ, et al. FNDC5 overexpression and irisin ame- stimulates p62-dependent autophagy-like aggregate clearance in hepatocytes. liorates glucose/lipid metabolic derangements and enhances lipolysis in obesity. Biomed Res Int 2014;2014:267350 Biochim Biophys Acta 2015;1852:1867–1875 34. Marchesini G, Mazzotti A. NAFLD incidence and remission: only a matter of 13. Liu TY, Shi CX, Gao R, et al. Irisin inhibits hepatic gluconeogenesis and weight gain and weight loss? J Hepatol 2015;62:15–17 increases glycogen synthesis via the PI3K/Akt pathway in type 2 diabetic mice 35. Mizushima N, Levine B, Cuervo AM, Klionsky DJ. Autophagy fights disease and hepatocytes. Clin Sci (Lond) 2015;129:839–850 through cellular self-digestion. Nature 2008;451:1069–1075 14. Russo GL, Russo M, Ungaro P. AMP-activated protein kinase: a target for old 36. Komatsu M, Waguri S, Ueno T, et al. Impairment of starvation-induced and drugs against diabetes and cancer. Biochem Pharmacol 2013;86:339–350 constitutive autophagy in Atg7-deficient mice. J Cell Biol 2005;169:425–434 diabetes.diabetesjournals.org Liu and Associates 3275

37. Farah BL, Landau DJ, Sinha RA, et al. Induction of autophagy improves 43. Pawlak M, Lefebvre P, Staels B. Molecular mechanism of PPARa action and hepatic lipid metabolism in glucose-6-phosphatase deficiency. J Hepatol 2016; its impact on lipid metabolism, inflammation and fibrosis in non-alcoholic fatty 64:370–379 liver disease. J Hepatol 2015;62:720–733 38. Singh R, Kaushik S, Wang Y, et al. Autophagy regulates lipid metabolism. 44. Mao Y, Cheng J, Yu F, Li H, Guo C, Fan X. Ghrelin attenuated lipotoxicity via Nature 2009;458:1131–1135 autophagy induction and nuclear factor-kB inhibition. Cell Physiol Biochem 2015; 39. Arslan N. Obesity, fatty liver disease and intestinal microbiota. World J 37:563–576 Gastroenterol 2014;20:16452–16463 45. Huang H, Li X, Zhuang Y, et al. Class A scavenger receptor activation in- 40. Lodhi IJ, Semenkovich CF. Peroxisomes: a nexus for lipid metabolism and hibits endoplasmic reticulum stress-induced autophagy in macrophage. J Bi- cellular signaling. Cell Metab 2014;19:380–392 omed Res 2014;28:213–221 41. Kemmerer M, Finkernagel F, Cavalcante MF, et al. AMP-activated protein 46. Aydin S. Three new players in energy regulation: preptin, adropin and irisin. kinase interacts with the peroxisome proliferator-activated receptor delta to in- Peptides 2014;56:94–110 duce genes affecting fatty acid oxidation in human macrophages. PLoS One 47. Aydin S, Kuloglu T, Aydin S, et al. Cardiac, skeletal muscle and serum irisin 2015;10:e0130893 responses to with or without water exercise in young and old male rats: cardiac 42. Reiter AK, Bolster DR, Crozier SJ, Kimball SR, Jefferson LS. Repression of muscle produces more irisin than skeletal muscle. Peptides 2014;52:68–73 protein synthesis and mTOR signaling in rat liver mediated by the AMPK activator 48. Zhang HJ, Zhang XF, Ma ZM, et al. Irisin is inversely associated with aminoimidazole carboxamide ribonucleoside. Am J Physiol Endocrinol Metab intrahepatic triglyceride contents in obese adults. J Hepatol 2013;59:557– 2005;288:E980–E988 562