Inhibition of ATG3 ameliorates liver steatosis by increasing mitochondrial function

Inhibition of ATG3 ameliorates liver steatosis by increasing mitochondrial function Graphical abstract Steatosis Ameliorated steatosis Mitochondrial function ATG3 ATG3 JNK SIRT1 CPT1 Highlights ATG3 levels are elevated in the livers of mice and patients with NAFLD. Inhibition of ATG3 in human hepatocytes and in mouse liver ameliorates steatosis by stimulating SIRT1, CPT1a, and mitochondrial function. Hepatic knockdown of SIRT1 or CPT1a blunts the ability of ATG3 inhibition to increase mitochondrial activity and to alleviate steatosis. Authors Natália da Silva Lima, Marcos F. Fondevila, Eva Nóvoa, ., Agueda Gonzalez-Rodriguez, Patricia Aspichueta, Ruben Nogueiras Correspondence ruben.nogueiras@usc. es (R. Nogueiras). Lay summary We show that autophagy-related gene 3 (ATG3) contributes to the progression of non-alcoholic fatty liver disease in humans and mice. Hepatic knockdown of ATG3 ameliorates the development of NAFLD by stimulating mitochondrial function. Thus, ATG3 is an important factor implicated in steatosis. https://doi.org/10.1016/j.jhep.2021.09.008 © 2021 The Author(s). Published by Elsevier B.V. on behalf of European Association for the Study of the Liver. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). J. Hepatol. 2022, 76, 11–24 Research Article NAFLD and Alcohol-Related Liver Diseases Inhibition of ATG3 ameliorates liver steatosis by increasing mitochondrial function Natália da Silva Lima 1,† , Marcos F. Fondevila 1,2,† , Eva Nóvoa 1,† , Xabier Buqué 3,4 , Maria Mercado-Gómez 5 , Sarah Gallet 6 , Maria J. González-Rellan 1 , Uxia Fernandez 1 , Anne Loyens 6 , Maria Garcia-Vence 7 , Maria del Pilar Chantada-Vazquez 7 , Susana B. Bravo 7 , Patricia Marañon 8 , Ana Senra 1 , Adriana Escudero 1 , Magdalena Leiva 9 , Diana Guallar 10 , Miguel Fidalgo 1 , Pedro Gomes 11,12,13 , Marc Claret 14,15 , Guadalupe Sabio 9 , Marta Varela-Rey 5,16 , Teresa C. Delgado 5 , Rocio Montero-Vallejo 17 , Javier Ampuero 17 , Miguel López 1,2 , Carlos Diéguez 1,2 , Laura Herrero 2,18 , Dolors Serra 2,18 , Markus Schwaninger 19 , Vincent Prevot 6 , Rocio Gallego-Duran 14,20 , Manuel Romero-Gomez 17,20 , Paula Iruzubieta 21 , Javier Crespo 21 , Maria L. Martinez-Chantar 5,20 , Carmelo Garcia-Monzon 8,20 , Agueda Gonzalez-Rodriguez 8,22 , Patricia Aspichueta 3,4,20 , Ruben Nogueiras 1,2, * 1 Department of Physiology, CIMUS, University of Santiago de Compostela-Instituto de Investigación Sanitaria, Santiago de Compostela, Spain; 2 CIBER Fisiopatologia de la Obesidad y Nutrición (CIBERobn), Spain; 3 Department of Physiology, University of the Basque Country UPV/EHU, Spain; 4 Biocruces Bizkaia Health Research Institute, Spain; 5 Liver Disease Lab, Center for Cooperative Research in Biosciences (CIC bioGUNE), Basque Research and Technology Alliance (BRTA), 48160 Derio, Bizkaia, Spain; 6 Univ. Lille, Inserm, CHU Lille, Laboratory of Development and Plasticity of the Neuroendocrine Brain, Lille Neuroscience & Cognition, UMR-S 1172, European Genomic Institute for Diabetes (EGID), F-59000 Lille, France; 7 Proteomic Unit, Health Research Institute of Santiago de Compostela (IDIS), Santiago de Compostela, 15705 A Coruña, Spain; 8 LiverResearchUnit, Santa Cristina University Hospital, Instituto de Investigación Sanitaria Princesa, Madrid, Spain; 9 Centro Nacional de Investigaciones Cardiovasculares (CNIC), Madrid, Spain; 10 Department of Biochemistry, CIMUS, University of Santiago de Compostela- Instituto de Investigación Sanitaria, Santiago de Compostela, Spain; 11 Department of Biomedicine, Unit of Pharmacology and Therapeutics, Faculty of Medicine, University of Porto, Porto, Portugal; 12 Coimbra Institute for Clinical and Biomedical Research (iCBR), Faculty of Medicine, University of Coimbra, Coimbra, Portugal; 13 Institute of Pharmacology and Experimental Therapeutics, Coimbra Institute for Clinical and Biomedical Research(iCBR), Faculty of Medicine, University of Coimbra, Coimbra, Portugal; 14 Neuronal Control of Metabolism (NeuCoMe) Laboratory, Institut d’Investigacions Biomèdiques August Pi iSunyer (IDIBAPS), 08036, Barcelona, Spain; 15 CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), 08036, Barcelona, Spain; 16 Gene Regulatory Control in Disease, CIMUS, University of Santiago de Compostela, Santiago de Compostela, Spain; 17 UGC Aparato Digestivo, Instituto de Biomedicina de Sevilla. Hospital Universitario Virgen del Rocío, Universidad de Sevilla, Sevilla, Spain; 18 Department of Biochemistry and Physiology, School of Pharmacy and Food Sciences, Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain; 19 University of Lübeck, Institute for Experimental and Clinical Pharmacology and Toxicology, Lübeck, Germany; 20 CIBER Enfermedades Hepáticas y Digestivas (CIBERehd), Spain; 21 Gastroenterology and Hepatology Department, Marqués de Valdecilla University Hospital. Clinical and Translational Digestive Research Group, IDIVAL, Santander, Spain; 22 CIBER de Diabetes y Enfermedades Metabólicas Asociadas (CIBERdem), Spain Background & Aims: Autophagy-related gene 3 (ATG3) is an enzyme mainly known for its actions in the LC3 lipidation process, which is essential for autophagy. Whether ATG3 plays a role in lipid metabolism or contributes to non-alcoholic fatty liver disease (NAFLD) remains unknown. Methods: By performing proteomic analysis on livers from mice with genetic manipulation of hepatic p63, a regulator of fatty acid metabolism, we identified ATG3 as a new target downstream of p63. ATG3 was evaluated in liver samples from patients with NAFLD. Further, genetic manipulation of ATG3 was performed in human hepatocyte cell lines, primary hepatocytes and in the livers of mice. Results: ATG3 expression is induced in the liver of animal models and patients with NAFLD (both steatosis and nonalcoholic steatohepatitis) compared with those without liver disease. Moreover, genetic knockdown of ATG3 in mice and human hepatocytes ameliorates p63- and diet-induced steatosis, while its overexpression increases the lipid load in hepatocytes. The inhibition of hepatic ATG3 improves fatty acid metabolism by reducing c-Jun N-terminal protein kinase 1 (JNK1), which increases sirtuin 1 (SIRT1), carnitine palmitoyltransferase 1a (CPT1a), and mitochondrial function. Hepatic knockdown of SIRT1 and CPT1a blunts the effects of ATG3 on mitochondrial activity. Unexpectedly, these effects are independent of an autophagic action. Conclusions: Collectively, these findings indicate that ATG3 is a novel protein implicated in the development of steatosis. Lay summary: We show that autophagy-related gene 3 (ATG3) contributes to the progression of non-alcoholic fatty liver disease Keywords: ATG3; sirtuin 1; lipid metabolism; NAFLD; NASH; mitochondria. Received 18 March 2021; received in revised form 3 August 2021; accepted 13 September 2021; available online 21 September 2021 *Corresponding author. Address: Department of Physiology, Research Centre of Molecular Medicine and Chronic Diseases (CIMUS), Instituto de Investigación Sanitaria de Santiago de Compostela, Universidadde Santiago de Compostela (USC), Santiago de Compostela, Spain. E-mail address: [email protected] (R. Nogueiras). † Contributed equally https://doi.org/10.1016/j.jhep.2021.09.008 Journal of Hepatology 2022 vol. 76 j11–24 Research Article NAFLD and Alcohol-Related Liver Diseases in humans and mice. Hepatic knockdown of ATG3 ameliorates the development of NAFLD by stimulating mitochondrial function. Thus, ATG3 is an important factor implicated in steatosis. © 2021 The Author(s). Published by Elsevier B.V. on behalf of European Association for the Study of the Liver. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/). Introduction Non-alcoholic fatty liver disease (NAFLD) is a major health threat in both developed and developing countries and is a precursor of the more advanced liver diseases, including non-alcoholic steatohepatitis (NASH), cirrhosis, and liver cancer. Currently, understanding the multiple and complex molecular pathways implicated in NAFLD onset and progression is a major priority. The transcription factor p63, which belongs to a family comprising p53, p63, and p73, 1 is one of many factors that contributes to the development of liver steatosis. The role of p63 as a tumor suppressor and in cell maintenance and renewal is well studied, but we have recently reported that it is also relevant in the control of lipid metabolism. 2 p63 encodes multiple isoforms that can be grouped into 2 categories; isoforms with an acidic transactivation domain (TA) and those without this domain (domain negative). The TAp63 a isoform is elevated in the liver of animal models of NAFLD as well as in liver biopsies from obese patients with NAFLD. Furthermore, downregulation of p63 a in the liver attenuates liver steatosis in diet-induced obese (DIO) mice, while the activation of TAp63 a increases hepatic fat content, mediated by the activation of IKKband endoplasmic reticulum stress. 2 A specialized form of autophagy that degrades lipid droplets, termed “lipophagy”, is a major pathway of lipid mobilization in hepatocytes. Lipophagy is elevated in hepatoma cells upon exposure to free fatty acids, 3 and reduces the fatty acid load in mouse hepatocytes. 4 Its impairment has been associated with the development of fatty liver and insulin resistance 3,5 ; in contrast, the autophagic flux is increased during the activation of hepatic stellate cells. 6 In the present study, we used an unbiased proteomics approach to gain insight into novel proteins modulating lipid metabolism in the liver of mice with genetic knockdown or overexpression of TAp63 a . We found that autophagy-related gene 3 (ATG3) was upregulated by TAp63 a activation and downregulated after p63 a inhibition. ATG3 is elevated in several animal models of NAFLD and in the liver of patients with NAFLD. Genetic overexpression of ATG3 increased the lipid load in hepatocytes, while its repression alleviated TAp63 a - and diet-induced steatosis. ATG3 exerted its role in lipid metabolism by regulating SIRT1 and mitochondrial function. Collectively, these findings identify ATG3 as a novel factor implicated in the development of steatosis. Materials and methods Animals and diets Mouse experiments were conducted to the standard protocols approved by the Animal Ethics Committee at the University of Santiago de Compostela, and mice received the highest levels of human care. C57BL/6J (8-week-old male) mice were housed in rooms under constant temperature (22C), a 12:12 h light/dark cycle and with ad libitum access to standard diet (SD) (U8200G10R, SAFE), a choline-deficient high-fat diet (CDHFD) (D05010402; 45 kcal% fat, Research Diets), a methionine- and choline-deficient diet (A02082002BR, Research Diets), or a highfat diet (HFD) (D12492; 60 kcal% fat, Research Diets) used during specified times and experiments. Food intake and body weight were measured weekly during all experimental phases. Cohort of patients with NAFLD The study population included 55 patients with biopsy-proven NAFLD (25 with non-alcoholic fatty liver [NAFL], and 30 with NASH), who underwent a liver biopsy during bariatric surgery or with a diagnostic purpose at the Santa Cristina University Hospital (Madrid, Spain) and at the Marqués de Valdecilla University Hospital (Santander, Spain) (Table S1). Liver samples with histologically normal liver obtained from 32 individuals during programmed cholecystectomy without steatosis were selected as healthy controls. Patients consumed <20 g of alcohol/ day, did not take potentially hepatotoxic drugs, had no analytical evidence of iron overload, and were seronegative for autoantibodies and for hepatitis B/C viruses and human immunodeficiency virus. Hepatic histopathological analysis was performed according to the scoring system of Kleiner et al. 7 Minimal criteria for NASH included the combined presence of grade 1 steatosis, hepatocellular ballooning, and lobular inflammation with or without fibrosis. This study was performed in agreement with the Declaration of Helsinki and with local and national laws. The Human Ethics Committees of the 2 hospitals approved the study procedures, and all participants voluntarily signed an informed written consent before inclusion in the study. Results Hepatic ATG3 is regulated by p63 and increased in dietinduced animal models of steatosis Hepatic overexpression of TAp63 a isoform induces steatosis, while inhibition of p63 a ameliorates diet-induced steatosis. 2 Proteomic analyses in the liver of these 2 animal models identified novel pathways and regulators of hepatic lipid accumulation (Fig. 1A-1D). Volcano plots show that multiple changes in hepatic protein levels were triggered by the manipulation of p63 (Fig. 1A-B). We specifically searched for proteins oppositely expressed in mice following the overexpression or knockdown of p63; we found that p63 positively regulated 43 proteins. Analyzing this protein set revealed a significant overrepresentation (44%) of metabolism-related proteins, with mitochondria identified as the most affected cell component (Fig. 1C). A heatmap showed several proteins involved in lipid metabolism with a strong positive regulation by p63 overexpression, including ATG3 (Fig. 1D). Given the relevant role of lipophagy in regulating fatty acid metabolism, and that ATG3 has not previously been related to liver function, we selected this candidate for further investigation. We corroborated that ATG3 mRNA expression and protein levels were upregulated when TAp63 a was induced in the liver, while it was reduced when hepatic p63 a was inhibited in DIO mice (Fig. 1E). Next, we evaluated the expression of ATG3 in livers from mouse models of diet-induced NAFLD and NASH, including mice fed a HFD or a CDHFD, and found that ATG3 mRNA was elevated (Fig. 1F). Together, these results indicate that ATG3 expression is consistently increased in preclinical models of steatosis and steatohepatitis. ATG3 is increased in the livers of patients with NAFLD ATG3 expression was evaluated in liver biopsies from patients with NAFLD or healthy livers (Table S1). ATG3 mRNA expression 12 Journal of Hepatology 2022 vol. 76 j11–24 Research Article NAFLD and Alcohol-Related Liver Diseases was significantly higher in the livers of patients with NAFLD (both NAFL and NASH) than in those from patients without the disease (Fig. 2A). ATG3 positively correlated with steatosis grade (Fig. 2B), NAFLD activity score (Fig. 2C), and fatty liver index (Fig. 2D). ATG3 levels were increased in obese (BMI>30) vs. nonobese (BMI<30) patients (Fig. 2E), and positively correlated with BMI (Fig. 2F). Altogether, these results indicate that higher ATG3 levels are present in patients with NAFLD. ATG3 increases lipid content and reduces b-oxidation in human hepatocytes ATG3 mRNA expression (Fig. 3A) and lipid content (Fig. 3B) were increased in human hepatic THLE2 cells treated with oleic acid (OA). However, small-interfering (si)RNA-mediated ATG3 silencing reduced OA-induced lipid storage (Fig. 3Band Fig. S1A). As previously reported, 2 TAp63 a overexpression increased the lipid droplets in hepatocytes, but this effect was blunted when ATG3 was silenced (Fig. 3C). The transfection efficiency of the plasmid for p63, and siRNA for ATG3, were corroborated (Fig. S1B). ATG3 overexpression (Fig. S1C) increased the lipid content in THLE2 cells (Fig. 3D). Importantly, similar loss- and gain-of-function experiments in another human hepatic cell line, HepG2, provided the same results: the OA-induced lipid content was reduced when ATG3 was silenced (Fig. 3E), while ATG3 overexpression augmented the lipid load (Fig. S1D and Fig. 3F). The rate of fatty acid boxidation, assessed using 14 C-palmitate, was decreased upon ATG3 overexpression (Fig. 3G).Finally,weisolatedhepatocytes from mice and transfected them with siATG3 and phosphorylated ATG3 (Fig. S1E-S1F) and detected an increase and reduction of fatty acid b-oxidation, respectively, in hepatocytes previously incubated with OA (Fig. 3H-3I). ATG3 regulates mitochondrial activity in hepatocytes The overexpression of ATG3 in HepG2 cells decreased oxygen consumption rate (Fig. 4A). More specifically, basal respiration, ATP-linked respiration, proton leak, maximal respiration, spare capacity, and 24 h non-mitochondrial oxygen consumption were decreased after ATG3 overexpression (Fig. 4A). These results were consistent with reduced protein levels of the mitochondrial complexes II (II-SDHB), III, and V (Fig. 4B). No changes were observed in protein levels of fatty acid synthase (FAS), acetyl CoA carboxylase, or carnitine palmitoyltransferase 1a (CPT1a) (Fig. 4C). However, the induction of ATG3 decreased peroxisome proliferator-activated receptor gamma coactivator 1 alpha (PGC1 a ) and sirtuin 1 (SIRT1) (Fig. 4D), 2 proteins that promote mitochondrial function and fatty acid oxidation. 8 These results were corroborated in isolated hepatocytes, where the overexpression of ATG3 (Fig. S1E) also reduced protein levels of SIRT1 and PGC1 a (Fig. 4E). ATG3 inhibition in HepG2 cells increased the oxygen consumption rate, with significant stimulation of basal respiration, ATP-linked respiration, proton leak, maximal respiration, and non-mitochondrial oxygen consumption (Fig. 4F). This was in agreement with increased levels of mitochondrial complexes I, II (II-SDHB), and IV (IV-COX II) (Fig. 4G), and the upregulation of CPT1a, PGC1 a and SIRT1 levels after silencing ATG3 (Fig. 4H, 4I). The inhibition of ATG3 (Fig. S1F) also increased protein levels of SIRT1 and PGC1 a in primary mouse hepatocytes (Fig. 4J). All results obtained in HepG2 cells were also corroborated in THLE2 cells (Fig. S2). Inhibition of hepatic ATG3 ameliorates CDHFD- induced steatosis We next investigated the in vivo relevance of these findings. Mice were fed a CDHFD for 16 weeks; at week 8, mice were injected in the tail vein with a lentivirus encoding for a scrambled short-hairpin (sh)RNA or shRNA against ATG3. No differences in body weight or feeding were observed (Fig. 5A- 5B). Hepatic ATG3 was decreased after the lentiviral injection (Fig. 5C). Although the liver mass and circulating triglycerides and non-esterified fatty acids were not affected by ATG3- knockdown (Fig. 5DandTable S2), circulating aspartate aminotransferase (AST) and alanine aminotransferase (ALT) were significantly lower (Fig. 5E). Consistent with this, hepatic triglyceride content and Oil Red O staining in liver sections of mice after ATG3-knockdown showed a lower amount of lipids, with no differences in fibrotic content (Sirius Red staining) (Fig. 5F). ATG3 inhibition increased CPT1a, PGC1 a ,andSIRT1 (Fig. 5G-5H), as well as the mitochondrial complexes I (I- NDUFB8)andII(II-SDHB)(Fig. 5I) and the activity of mitochondrial complex II (Fig. 5J). Functional assays revealed an increased b-oxidation rate in livers of mice with ATG3- knockdown (Fig. 5K). These effects were independent of cellular proliferation since the expression of cyclin D1 and cyclin D2 and the immunostaining of Ki67 were unchanged between the groups (Fig. S3A-S3B). The knockdown of ATG3 did not affect mRNA expression of fibrotic markers (Fig. S4A). Inhibition of hepatic ATG3 ameliorates TAp63 a - induced steatosis We next evaluated the effects of ATG3 inhibition on TAp63 a - induced steatosis in mice. For this, mice fed a SD were injected via the tail vein with adeno-associated virus (AAV)8-TAp63 a or AAV8-GFP; and then, after 8 weeks, they were injected again with either shATG3 (to knockdown ATG3) or scrambled shRNA (as a control) and maintained on the same diet for another 8 weeks. Body weight (Fig. 6A) and food intake (Fig. 6B) remained unchanged. Administration of AAV8-TAp63 a or lentivirusshATG3 caused an increase in p63 and a decrease in ATG3, respectively (Fig. 6C). Even though liver mass was unchanged (Fig. 6D), higher circulating AST and ALT levels were observed after TAp63 a overexpression; ATG3 inhibition reduced AST and ALT levels (ALT was not statistically significant; p= 0.0513) (Fig. 6E). Serum triglycerides and non-esterified fatty acids were similar among all groups (Table S3). As previously reported, 2 the overexpression of TAp63 a caused liver steatosis, an effect that was blunted by ATG3-knockdown (Fig. 6F). Mechanistically, AAV8-TAp63 a -treated mice had elevated levels of the FAS protein and reduced CPT1a, PGC1 a , and SIRT1 (Fig. 6G-H). ATG3 inhibition restored the levels of CPT1a, PGC1 a , and SIRT1 to those of the control group (Fig. 6G-6H). Hepatic overexpression of TAp63 a decreased the levels of all mitochondrial complexes, while ATG3 inhibition normalized the liver levels of complexes I and II (Fig. 6I). In these groups, mRNA expression of fibrotic markers such as collagen alpha 2, collagen III or a SMA remained unaltered, although collagen alpha 1 was Journal of Hepatology 2022 vol. 76 j11–24 13 Log 2 (fold change) Log 2 (fold change) -Log 10 (p value) -Log 10 (p value) -5 -4 -3 -2 -1 0 1 2 3 4 5 0 2 4 6 8 -4 -3 -2 -1 0 1 2 3 4 0 2 4 6 8 10 12 shRNA p63α vs. sh scrambled AAV8-TAp63α WT Lentiv-shp63α WT AAV8-TAp63α vs. AAV8-GFP Liver protein levels (a.u.) 0 25,000 50,000 75,000 100,000 ATG3 *** Chow diet CDHFD 0 102030405060 Peroxisome Golgi aparatus Endoplasmic reticulum Nucleus Mitochondrion Proteins (%) Cellular component AAV8-GFP AAV8- TAp63α ATG3 GAPDH sh scrambled shp63 ATG3 GAPDH shp63α sh scrambled 0 4,000 8,000 12,000 16,000 ATG3 Liver protein levels (a.u.) ** AAV8-TAp63α AAV8-GFP HFD 6 weeks 0.0 0.5 1.0 1.5 2.0 2.5 *** ATG3 mRNA levels (fold change) HFD 52 weeks 0 1 2 3 4 5** ATG3 mRNA levels (fold change) CDHFD 52 weeks ATG3 mRNA levels (fold change) 0 1 2 3 4 * CDHFD 6 weeks ATG3 mRNA levels (fold change) 0.0 0.5 1.0 1.5 2.0 2.5 * ACSM1 UD11 CTNA1 PDIA5 RMD3 ARP3 MAVS ATG3 PICAL KBL HUTH SPYA AASS NAGS PAPS2 STS ACSM3 CP254 CY250 PCCB ACBP DHB11 ACD11 ADX CP4FE MUTA RETST GLRX5 LPPRC CMLO1 CNPY2 GLU2B LONM TCPG SRPRA ES8L2 COX5A QCR6 QCR8 IDH3A SUCA IF4G1 OCAD2 0 0.51.01.52.02.5 AAV8- TAp63α AAV8- GFP sh scrambled shRNA p63α Lipids Amino acids and derivates Carbohydrates Metabolism 0 5 10 15 20 0123 Cell cycle DNA repair DNA replication Extracellular matrix organization Programmed cell death Signal transduction Inmune system Metabolism of proteins Metabolism of RNA Autophagy Proten localization Cell-cell communication Metabolism -Log10 (FDR) Number of proteins Pathway name p63 mRNA levels (fold change) 0 2 4 6 8 10 * AAV8-GFP AAV8-TAp63α ATG3 mRNA levels (fold change) 0 2 4 6 8 * AAV8-GFP AAV8-TAp63α ATG3 protein levels (% of control) 0 1 2 3 *** AAV8-GFP AAV8-TAp63α p63 mRNA levels (fold change) 0 1 2 3 * sh scrambled shp63α ATG3 mRNA levels (fold change) 0 1 2 3 * sh scrambled shp63α ATG3 protein levels (% of control) 0.0 0.5 1.0 1.5 * sh scrambled shp63α SD diet HFD SD diet HFD SD diet CDHFD SD diet CDHFD AB C E D F Fig. 1. Hepatic ATG3 is upregulated in different mouse models of NAFLD. (A, B) Volcano plots of hepatic protein expression determined by LC-MS/MS proteomics of (A) mice fed a SD, with TAp63 a overexpressed specifically in liver compared to the control group (n = 3 per group). Upregulated proteins (blue) and downregulated proteins (brown); or (B) mice fed a CDHFD for 12 weeks, with hepatic p63 a downregulated as compared to the control group (n = 3 per group). Upregulated proteins (orange) and downregulated (violet). Values obtained for ATG3 are represented in the bar graph. (C) Reactome pathway classification of 14 Journal of Hepatology 2022 vol. 76 j11–24 Research Article NAFLD and Alcohol-Related Liver Diseases elevated by TAp63 a and reduced when ATG3 was inhibited (Fig. S4B). Knockdown of hepatic ATG3 does not affect fibrosis We next evaluated whether ATG3-knockdown in liver alleviates fibrosis by injecting mice via the tail vein with a lentivirus shATG3 or shRNA-scrambled (as control); mice were fed a SD for 4 week and a methionine- and choline-deficient diet for another 4 weeks to induce fibrosis, and were sacrificed at week 8. 9 Body weight was similar between ATG3-knockdown and control mice (Fig. S5A). As expected, ATG3-knockdown significantly reduced ATG3 mRNA (Fig. S5B). No differences were observed for liver mass (Fig. S5C), serum AST or ALT (Fig. S5D), inflammatory or profibrotic markers (Fig. S5E), or Oil Red O or Sirius Red staining (Fig. S5F). Of note, ATG3 was highly expressed in isolated hepatic stellate cells compared to hepatocytes or Kupffer cells (Fig. S6A) and in LX2 cells (a human hepatic stellate cell line) compared to THLE2 and HepG2 cells (Fig. S6B). However, the treatment of LX2 cells with TGFbdid not change ATG3 expression, and the TGFbinduced expression of fibrotic markers was not affected after silencing ATG3 (Fig. S6C-S6D). ATG3 requires SIRT1 to modulate lipid metabolism and mitochondrial activity We next assessed the functional relevance of SIRT1 as a potential modulator of ATG3. Similar to the results shown in Fig. 3B, silencing ATG3 in HepG2 cells ameliorated OA-induced lipid load (Fig. S7A). However, when SIRT1 was also silenced, siATG3 failed to decrease OA-induced lipid content (Fig. S7A). These results were also corroborated in THLE2 cells co-transfected with siATG3 and siSIRT1 (Fig. S7B). Moreover, the higher oxygen consumption rate induced by siATG3 in THLE2 cells was also lost after SIRT1 silencing (Fig. S7C). In addition, ATG3-induced lipid content was blunted by resveratrol, a pharmacological activator of SIRT1 in both THLE2 cells (Fig. S8A) and primary hepatocytes (Fig. S8B). In isolated hepatocytes, ATG3 also reduced the oxygen consumption rate while resveratrol blunted this effect (Fig. S8C). In line with these results, silencing ATG3 increased SIRT1 activity and reduced acetylated levels of PGC1 a , while its overexpression reduced SIRT1 activity in isolated hepatocytes (Fig. S9A-S9C). Next, we evaluated the in vivo relevance of SIRT1 as a mediator of the effects of ATG3. For this, mice fed a CDHFD were injected with a lentivirus encoding either a scrambled shRNA (as NL NAFL NASH 0 1 2 30.0308 ATG3/36b4 0.0175 0123 0 1 2 3 Steatosis grade ATG3/36b4 10 20 30 40 50 60 70 0 1 2 3r = 0.3332 p = 0.0012 BMI ATG3/36B4 0 25 50 75 100 125 0 1 2 3r = 0.3019 p = 0.0040 FLI ATG3/36B4 ABC DEF 012345 0 1 2 3 r = 0.2872 p = 0.0062 r = 0.2707 p = 0.0067 NAS score ATG3/36B4 BMI <30 BMI >30 0 1 2 30.0013 ATG3/36b4 Fig. 2. ATG3 is increased in the liver of patients with NAFLD. (A) ATG3 mRNA expression in the liver of patients without NAFLD (NL) (n = 32), with NAFL (n = 25) or NASH (n = 30). (B) Correlation between ATG3 and steatosis grade. (C) Correlation between ATG3 and NAS score. (D) Correlation between ATG3 and fatty liver index. (E) ATG3 mRNA expression in the liver of non-obese vs. obese patients with NAFLD. (F) Correlation between ATG3 and body mass index. HPRT was used to normalize mRNA levels. Data are presented as mean ± standard deviation. pvalue using a Student’sttest (B) (C) (D) (E) (F) or one-way ANOVA followed by a Newman-Keuls multiple comparison test (A). FLI, fatty liver index; NAFL, non-alcoholic fatty liver; NAFLD, non-alcoholic fatty liver disease; NAS, NAFLD activity score; NASH, non-alcoholic steatohepatitis; NL, normal liver. = deregulated proteins in liver of mice from (A) and (B), showing the number of proteins included in each category and the associated FDR. The same proteins were also classified according to the cellular component using FunRich tool. (D) Heatmap representation of protein levels differentially expressed in the groups from (A) and (B). (E) mRNA levels of p63 and ATG3, as well as ATG3 protein levels, in mice in the conditions as in (A) and (B). (n = 4 per group). (F) ATG3 mRNA in liver of mice fed a SD, an HFD, or a CDHFD (n = 9–14 per group). HPRT and GAPDH were used to normalize mRNA and protein levels, respectively. Lines indicate splicing in the same gel. Data are presented as mean ± SEM; *p<0.05, **p<0.01, ***p<0.001, using a Student’sttest. AAV, adeno-associated virus; CDHFD, choline-deficient high-fat diet; FDR, false discovery rate; HFD, high-fat diet; LC-MS/MS, liquid chromatography tandem mass spectrometry; NAFLD, non-alcoholic fatty liver disease; SD, standard diet; sh(RNA), short-hairpin (RNA); WT, wild-type. Journal of Hepatology 2022 vol. 76 j11–24 15 AB C D F E GHI Control siATG3 OA OA + siATG3 THLE2 cells - Oil Red O staining 100μm100μm 100μm 100μm 50μm50μm 50μm50μm 50μm50μm 50μm50μm Control pATG3 100μm 100μm HepG2 cells Oil Red O staining siATG3 pTAp63α + siATG3 OA + siATG3 Control pTAp63α THLE2 cells - Oil Red O staining 50μm50μm Control pATG3 THLE2 cells Oil Red O staining Control siATG3 OA HepG2 cells - Oil Red O staining 0.8 1.0 1.2 1.4 1.6 ** THLE2 ATG3 mRNA levels (fold change) 0 1 2 3 ** HepG2 Oil Red O area % of control (A.U.) 0 10 20 *** *** HepG2 Oil Red O area % of control (A.U.) *** *** *** 0 2 4 6*** THLE2 Oil Red O area % of control (A.U.) 0 40 80 120 *** *** THLE2 Oil Red O area % of control (A.U.) *** *** *** 0.0 0.5 1.0 1.5 2.0 2.5 ** THLE2 Oil Red O Area % of control (A.U.) *** 450 500 550 600 650 700 750 Oxidation rate (pmol palmitate/h*mg prot) p = 0.053 ASM 50 55 60 65 70 Oxidation rate (pmol palmitate/h*mg prot) * CO 2 CO 2 CO 2 3,000 3,500 4,000 4,500 5,000 5,500 Oxidation rate (pmol palmitate/h*mg prot) ASM * 0 20 40 60 80 Oxidation rate (pmol palmitate/h*mg prot) 2,400 2,600 2,800 3,000 3,200 3,400 3,600 Oxidation rate (pmol palmitate/h*mg prot) ASM *** 25 30 35 40 45 50 Oxidation rate (pmol palmitate/h*mg prot) ** HepG2 cells Mouse hepatocytes Mouse hepatocytes OA + siATG3 OA OA + siATG3 OA + siATG3 OA OA pATG3 Control pATG3 Control pATG3 Control pATG3 Control pATG3 Control Control pATG3 Control siATG3 OA + siATG3 OA Control OA Control siATG3 Control siATG3 pTAp63α pTAp63α + siATG3 Fig. 3. ATG3 regulates lipid accumulation in human hepatic cell lines. (A) ATG3 mRNA in THLE2 cells treated with OA (1 mM) for 24 h (n = 6 per group). (B–F) Representative Oil Red O staining of (B) THLE2 cells with siATG3 treated with OA or vehicle (n = 6 per group); (C) TAp63 a -upregulated THLE2 cells after ATG3 silencing (n = 15 per group); and (D) THLE2 cells with overexpression of ATG3 (n = 12 per group). (E, F) Representative Oil Red O staining of HepG2 cells with ATG3 downregulated (E) or ATG3 overexpressed (F), treated with OA or vehicle for 24 h (n = 4–7 per group). (G) Palmitate oxidation rate (partial and complete oxidation) in HepG2 cells overexpressing ATG3. Palmitate oxidation rate in mouse hepatocytes after overexpression (H) and inhibition (I) of ATG3. The n represents the number of independent experiments. Data are presented as mean ± SEM; *p<0.05, **p<0.01, ***p<0.001, using a Student’sttest (A) (D) (F) (G) or oneway ANOVA followed by a Newman-Keuls multiple comparison test (B) (C) (E). ASM, acid soluble metabolites; OA, oleic acid; si(RNA), small-interferingRNA;TA, transactivation domain. 16 Journal of Hepatology 2022 vol. 76 j11–24 Research Article NAFLD and Alcohol-Related Liver Diseases a control) or shRNA-ATG3. At week 4, a second lentiviral vector with shRNA-scrambled or shRNA-SIRT1 was injected; and mice were sacrificed at week 8. Body weight and food intake remained unchanged (Fig. 7A-7B). ATG3 and SIRT1 were reduced in liver after the injections of shRNA-ATG3 and shRNA-SIRT1, respectively (Fig. 7C). Even though liver weight was not affected, AST and ALT were reduced after ATG3-knockdown, and this reduction was abolished when both ATG3 and SIRT1 were silenced G I H F B CD A E J C-I C-II C-III C-IV C-V 0 100 200 300 ** ** ** HepG2 protein levels (% control) pATG3 V-ATP5A IV-COX II III-UQCRC2 II-SDH5 Control pATG3 I-NDUFB8 GAPDH V-ATP5A IV-COX II III-UQCRC2 II-SDH5 Control siATG3 I-NDUFB8 GAPDH 0 102030406080 0 5 10 15 20 HEPG2 control HepG2 pATG3 Oligo FCCP Rot/AA Min OCR (pmol/min/μg prot) *** * ** ** 0 102030406080 0 2 4 6 8 10 HepG2 control HepG2 siATG3 Oligo FCCP Rot/AA Min OCR (pmol/min/ μg prot) **** Basal respiration ATP-linked respiration Proton leak Maximal respiration Spare capacity Non-mitochondrial oxygen consuption 0 2 4 6 8 *** *** ** *** ** OCR (pmol/min/ μg prot) PGC1α SIRT1 0 100 200 300 ** HepG2 protein levels (% control) * FAS ACC p-ACC p-ACC/ACC CPT1a 0 50 100 150 200 250 300 350 ** HepG2 protein levels (% control) HepG2 protein levels (% control) PGC1a SIRT1 0 100 200 ** * HepG2 protein levels (% control) C-I C-II C-III C-IV C-V 0 50 100 150 200 ** ** Basal respiration ATP-linked respiration Proton leak Maximal respiration Spare capacity Non-mitochondrial oxygen consuption 0 4 8 12 ** ** *** ** ** OCR (pmol/min/μg prot) HepG2 protein levels (% control) FAS ACC p-ACC p-ACC/ACC CPT1a 0 100 200 GAPDH Control pATG3 PGC1α SIRT1 p-ACC GAPDH Control siATG3 FAS ACC CPT1a GAPDH Control siATG3 PGC1α SIRT1 p-ACC GAPDH Control pATG3 FAS ACC CPT1a SIRT1 PGC1a 0 1 2 3 4 nietor pse tyco t apehyrami rP )lo rtnocfo%(slevel ** Control siATG3 ** siATG3 PGC1α GAPDH Control SIRT1 GAPDH SIRT1 PGC1a 0.0 0.5 1.0 1.5 Primary hepatocytes protein levels (% of control) CTL pATG3 * ** PGC1α CTL GAPDH GAPDH SIRT1 Fig. 4. ATG3 modulates mitochondrial function in hepatocytes. HepG2 cells overexpressing ATG3 were analyzed for (A) OCR, (B) protein levels of OXPHOS complex subunits I–V, (C) lipid metabolism markers, and (D) protein levels of PGC1 a and SIRT1 (n = 6 per group). (E) PGC1 a and SIRT1 protein levels of primary mouse hepatocytes overexpressing ATG3 (n = 5 per group). ATG3-silenced HepG2 cells were analyzed for (F) OCR, (G) protein levels of OXPHOS complex subunits I–V, (H) lipid metabolism markers, and (I) protein levels of PGC1 a and SIRT1 (n = 5–6 per group). (J) PGC1 a and SIRT1 protein levels of ATG3-silenced primary mouse hepatocytes (n = 6 per group). The n represents the number of independent experiments. GAPDH was used to normalize protein levels. Dividing lines indicate splicing in the same gel. Data are presented as mean ± SEM; *p<0.05, **p<0.01, ***p<0.001, using a Student’sttest. OCR, oxygen consumption rate; OXPHOS, oxidative phosphorylation. Journal of Hepatology 2022 vol. 76 j11–24 17 0.0 0.5 1.0 1.5 2.0 Sirius red area % of control(A. U.) Liver protein levels (% of control) C-I C-II C-III C-IV C-V 0.0 1.5 3.0 4.5 * * 0246810121416 0 20 40 60 80 100 120 140 160 shATG3 Weeks Cumulative food intake (g) 0246810121416 0 3 6 9 12 15 18 shATG3 Weeks Body weight change (g) A B D F C CDHFD sh scrambled CDHFD shATG3 50 μm 50 μm 50 µm50 µm H FAS ACC p-ACC CPT1a CDHFD sh scrambled LPL GAPDH CDHFD shATG3 SIRT1 PGC1α GAPDH CDHFD sh scrambled CDHFD shATG3 I Lentiv-shATG3 WT CDHFD G J E K 0 20 40 60 80 * AST (U/L) 0 20 40 60 80 100 ALT (U/L) *** 0.0 0.5 1.0 1.5 2.0 2.5 ** Oil Red O area % of control (A. U.) Liver protein levels (% of control) FAS ACC p-ACC p-ACC/ACC CPT1a LPL 0.0 1.5 3.0 * p = 0.0521 Liver protein levels (% of control) PGC1a SIRT1 0 1 2 3 4 5 ** * 3.0 3.5 4.0 4.5 Liver mass (g/100g of BW) HEOil Red O staining 50 μm50 μm Sirius red ATG3 GAPDH CDHFD sh scrambled CDHFD shATG3 0.0 0.5 1.0 1.5 ATG3 protein levels (% of control) ** 0 3 6 9 Triglycerides (mg TG/g of liver) * Scrambled shATG3 Scrambled shATG3 Complex II activity (SDH units/g tissue) 0.00 0.05 0.10 0.15 * Scrambled shATG3 CDHFD sh scrambled CDHFD shATG3 III-UQCRC2 I-NDUFB8 V-ATP5A IV-COX II II-SDH5 HSP70 Oxidation rate (pmol palmitate/h*g tissue) 0 10,000 20,000 30,000 40,000 50,000 * [14C]-ASM Scrambled shATG3 Oxidation rate (pmol palmitate/h*g tissue) 0 200 400 600 * [14C]-CO2 Scrambled shATG3 Scrambled shATG3 Scrambled shATG3 Scrambled shATG3 Scrambled shATG3 Scrambled shATG3 CDHFD sh scrambled CDHFD shATG3 Fig. 5. Liver-specific downregulation of ATG3 ameliorates NAFLD induced by CDHFD. (A) Body weight change, (B) food intake, (C) hepatic ATG3 protein levels, and (D) liver mass of mice fed CDHFD for 16 weeks, with lentivirus of shRNA ATG3 or shRNA–scrambled delivered via TVI at week 8, as indicated. (E) Serum levels of AST and ALT. (F) Representative microphotographs of H&E (upper panel), Oil Red O (middle panel) and Sirius Red staining (lower panel) of liver sections. Lipids in Oil Red O-stained sections and collagen deposition in Sirius Red stained sections (pink area) were quantified using ImageJ. Hepatic triglyceride content was also directly measured. (G) Hepatic protein levels of lipid metabolism markers. (H) Liver proteins levels of PGC1 a and SIRT1. (I) Liver protein levels of OXPHOS complex subunits I–V. (J) Complex II activity in liver. (K) Hepatic palmitate oxidation rate (partial and complete oxidation). GAPDH and HPRT were used to normalize protein and mRNA levels, respectively. Dividing lines indicate splicing in the same gel. Data are presented as mean ± SEM; *p<0.05, **p<0.01, ***p<0.001, using a Student’sttest (n = 7–15 per group). ASM, acid soluble metabolites; CDHFD, choline-deficient high-fat diet; OXPHOS, oxidative phosphorylation; sh(RNA), shorthairpin (RNA); TVI, tail vein injection. 18 Journal of Hepatology 2022 vol. 76 j11–24 Research Article NAFLD and Alcohol-Related Liver Diseases