Role of adipose triglyceride lipase (PNPLA2) in protection from hepatic inflammation in mouse models of steatohepatitis and endotoxemia

Authors

  • Pooja Jha,

    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    2. Laboratory of Experimental and Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine, Medical University of Graz, Graz, Austria
    3. Institute of Pathology, Medical University of Graz, Graz, Austria
    Search for more papers by this author
  • Thierry Claudel,

    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Anna Baghdasaryan,

    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Michaela Mueller,

    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    2. Laboratory of Experimental and Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine, Medical University of Graz, Graz, Austria
    Search for more papers by this author
  • Emina Halilbasic,

    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    Search for more papers by this author
  • Suman K. Das,

    1. Institute of Pathology, Medical University of Graz, Graz, Austria
    Search for more papers by this author
  • Achim Lass,

    1. Institute of Molecular Biosciences, University of Graz, Graz, Austria
    Search for more papers by this author
  • Robert Zimmermann,

    1. Institute of Molecular Biosciences, University of Graz, Graz, Austria
    Search for more papers by this author
  • Rudolf Zechner,

    1. Institute of Molecular Biosciences, University of Graz, Graz, Austria
    Search for more papers by this author
  • Gerald Hoefler,

    1. Institute of Pathology, Medical University of Graz, Graz, Austria
    Search for more papers by this author
    • These two authors contributed equally to this work.

  • Michael Trauner

    Corresponding author
    1. Hans Popper Laboratory of Molecular Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Vienna, Austria
    • Address reprint requests to: Michael Trauner, M.D., Professor of Medicine and Chair of Gastroenterology and Hepatology, Division of Gastroenterology and Hepatology, Department of Internal Medicine III, Medical University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria. E-mail: michael.trauner@meduniwien.ac.at; fax: +43 1 40 400 4735.

    Search for more papers by this author
    • These two authors contributed equally to this work.


  • Potential conflict of interest: Prof. Trauner consults for, is on the speakers' bureau for, and received grants from Falk. He is on the speakers' bureau for and received grants from MSD. He consults for Phenex. He is on the speakers' bureau for Gilead and Roche. He received grants from Intercept.

  • Supported by grants F3008 (to MT) and F3002 (R. Zechner) from the Austrian Science Foundation and European Community's Seventh Framework Program (FP7/2007-2013) under grant agreement HEALTH-F2-2009-241762 for the project FLIP (to MT). P. Jha was also supported by the PhD program of the Medical University of Graz.

Abstract

Hepatic inflammation is a key feature of progressive liver disease. Alterations of fatty acid (FA) metabolism and signaling may play an important role in the pathogenesis of nonalcoholic fatty liver disease (NAFLD) and its progression to nonalcoholic steatohepatitis (NASH). Moreover, FAs activate peroxisome proliferator-activated receptor α (PPARα) as a key transcriptional regulator of hepatic FA metabolism and inflammation. Since adipose triglyceride lipase (ATGL/PNPLA2) is the key enzyme for intracellular hydrolysis of stored triglycerides and determines FA signaling through PPARα, we explored the role of ATGL in hepatic inflammation in mouse models of NASH and endotoxemia. Mice lacking ATGL or hormone-sensitive lipase (HSL) were challenged with a methionine-choline-deficient (MCD) diet as a nutritional model of NASH or lipopolysaccharide (LPS) as a model of acute hepatic inflammation. We further tested whether a PPARα agonist (fenofibrate) treatment improves the hepatic phenotype in MCD- or LPS-challenged ATGL-knockout (KO) mice. MCD-fed ATGL-KO mice, although partially protected from peripheral lipolysis, showed exacerbated hepatic steatosis and inflammation. Moreover, ATGL-KO mice challenged by LPS showed enhanced hepatic inflammation, increased mortality, and torpor, findings which were attributed to impaired PPARα DNA binding activity due to reduced FABP1 protein levels, resulting in impaired nuclear FA import. Notably, liganding PPARα through fenofibrate attenuated hepatic inflammation in both MCD-fed and LPS-treated ATGL-KO mice. In contrast, mice lacking HSL had a phenotype similar to the WT mice on MCD and LPS challenge. Conclusion: These findings unravel a novel protective role of ATGL against hepatic inflammation which could have important implications for metabolic and inflammatory liver diseases. (Hepatology 2014;59:858–869)

Abbreviations
ATGL

adipose triglyceride lipase

DBA

DNA binding activity

FA

fatty acid

FF

fenofibrate

HSL

hormone sensitive lipase

IHC

immunohistochemistry

KO

knockout

LPS

lipopolysaccharide (endotoxin)

MCD

methionine-choline-deficient

NAFLD

nonalcoholic fatty liver disease

NASH

nonalcoholic steatohepatitis

NS

nonsignificant

PPARα

peroxisome proliferator-activated receptor α

SD

standard deviation

TG

triglyceride

WAT

white adipose tissue

WT

wild-type

Hepatic inflammation is an integral feature for progression of liver disease with development of fibrosis, cirrhosis, and cancer.[1] Due to the emerging epidemic of obesity and diabetes, the factors determining progression of nonalcoholic fatty liver disease (NAFLD) represent a major clinical challenge.[2] Fatty liver may progress to nonalcoholic steatohepatitis (NASH) when adaptive mechanisms involved in lipid partitioning and catabolism that protect hepatocytes from lipotoxicity of excess fatty acids (FAs) and other lipids are overwhelmed, resulting in inflammation and fibrosis.[2, 3] Moreover, hepatic inflammation may alter hepatic lipid metabolism as well as signaling and vice versa.[4, 5]

Flux of FAs from white adipose tissue (WAT) to the liver and their release from hepatocellular stores may critically determine their availability for metabolism and signaling in the liver. Adipose triglyceride lipase (ATGL) is the first and rate-limiting enzyme for intracellular triglyceride (TG) hydrolysis, followed by hormone sensitive lipase (HSL) for diacylglycerols and monoglyceride lipase (MGL) for monoacylglycerols,[6, 7] with each enzymatic step releasing FAs. ATGL may thus regulate TG turnover and expression of peroxisome proliferator-activated receptor α (PPARα) target genes in the liver, possibly through release of FAs as natural ligands of PPARα.[8-11] Furthermore, we recently showed that functional PPARα signaling in the heart requires ATGL-catalyzed TG hydrolysis, suggesting that FAs first need to be esterified to TGs and rehydrolyzed by ATGL before they can become active signaling lipids.[10] In liver, PPARα regulates FA synthesis, oxidation, gluconeogenesis, ketogenesis, lipoprotein assembly, and represses inflammation.[5, 12] Additionally, lack of functional PPARα contributes to increased susceptibility to the development of NAFLD/NASH, hepatic inflammation, and acute phase response (APR).[13-16] Notably, genetic variants of adiponutrin (PNPLA3), the closest homolog of ATGL (PNPLA2), have recently emerged as key determinants for pathogenesis and progression of NAFLD, and also show a robust association with alcoholic liver disease and chronic hepatitis C in humans.[2, 6, 17] Moreover, mutations of ATGL or its coactivator CGI-58 can result in hepatic steatosis as an important feature of neutral lipid storage disorders.[6, 18] Importantly, patients with NAFLD show reduced ATGL expression, although the mechanisms have remained unclear.[19]

We therefore hypothesized that ATGL may critically determine hepatic inflammation due to its key role in lipid partitioning and PPARα signaling. We herein demonstrate that lack of ATGL (but not HSL) increases the susceptibility to methionine-choline-deficient (MCD)- and lipopolysaccharide (LPS)-induced inflammation in mouse liver, uncovering a novel protective role of ATGL against hepatic inflammation.

Materials and Methods

Mice

ATGL-deficient and HSL-deficient mice were generated on a mixed genetic background (50% C57BL/6 and 50% 129/Ola) and backcrossed onto the C57BL/6 background strain for more than 10 generations.[7, 20] Animals were housed under a 12:12-hour light/dark cycle and permitted ad libitum consumption of water and diet. All experimental protocols were approved by the local Animal Care and Use Committees according to criteria outlined in Guide for the Care and Use of Laboratory Animals prepared by the U.S. National Academy of Sciences (National Institutes of Health publication 86-23, revised 1985).

Feeding and Injection Protocols

The 7 to 8-week-old wild-type (WT), ATGL-KO, and HSL-KO mice were fed an MCD or control diet for 2 weeks. The composition of the MCD and its control diet was the same except that the MCD diet was depleted in methionine and choline. Mice were harvested between 1-4 pm without fasting. The 8 to 10-week-old WT, ATGL-KO, and HSL-KO mice were injected with a single dose of LPS (4 μg/g body weight) intraperitoneally for 12 hours (4 am to 4 pm). Control mice were injected with saline. Both groups were fasted after injection, since LPS is known to induce anorexia.[21] In order to avoid the effects of fasting prior to LPS, mice were injected in the morning (4 am) when they were fully fed. For survival rate analysis, male mice were injected with a single dose of 5 μg/g LPS. For fenofibrate (FF) treatment, 8 to 9-week-old WT and ATGL KO mice were fed with an MCD diet supplemented with 0.1% FF for 2 weeks and compared to mice receiving the MCD diet alone. Additionally, 8 to 9-week-old WT and ATGL-KO mice received a 0.2% FF-enriched diet for 6 days followed by a 12-hour LPS injection on the seventh day to the FF-fed or chow-fed mice. Unless otherwise indicated, all experiments were performed on female mice. All data are representative of 2-4 independent experiments with 4-7 animals per group.

Serum Analysis, Chemicals, and Reagents

These are described in detail in the Supporting Methods.

RNA Isolation and Quantitative Real-Time Reverse-Transcription Polymerase Chain Reaction (qRT-PCR) Analysis

Experimental details and primer sequences used are listed in the Supporting Methods. Messenger RNA (mRNA) levels were normalized to 36b4 or 18S as housekeeping genes, which did not vary between groups.

DNA Binding Activity of PPARα

DNA binding activity of PPARα was assayed from nuclear proteins using an enzyme-linked immunosorbent assay-based PPARα Transcription Factor Assay Kit as described in detail in the Supporting Methods.

Western Blot Analysis of Nuclear FABP1, Hepatic TG Determination, and Conventional Histology

Western blot analysis of nuclear FABP1, hepatic TG determination, and conventional histology is described in the Supporting Methods.

CD11b Immunohistochemistry

CD11b immunohistochemistry was performed as described in the Supporting Methods. For quantification of CD11b-positive cells, the slides were scanned at 40× magnification. The positive cells were then manually quantified in 60 power fields.

Statistics

Data are presented as mean ± SD of reported animals in each group (4-7 animals). Statistical analysis was performed using SPSS v. 14.0. Statistical significance was determined by Student unpaired two-tailed t test. Group difference were considered significant for P < 0.05.

Results

Absence of ATGL Increases Susceptibility to MCD-Induced Steatohepatitis

To study the role of ATGL—a key intracellular lipase involved in hydrolysis of stored fat, lipid partitioning, and PPARα signaling—in hepatic inflammation in a mouse model of NASH, we challenged mice lacking ATGL with an MCD diet. MCD feeding was limited to 2 weeks to avoid excessive loss of white adipose tissue (WAT) (required for further analysis). MCD challenge resulted in enhanced hepatic steatosis in ATGL-KO mice (Fig. 1A). At baseline, ATGL-KO mice did not show significant inflammation (Fig. 1B), which is in line with previous findings in liver-specific ATGL-KO.[11] Notably, when challenged with MCD ATGL-KO mice developed more severe hepatic inflammation compared to MCD-fed WT mice as reflected by increased expression of inflammatory markers: tumor-necrosis factor-alpha (TNF-α), inducible nitric oxide synthase (iNOS), monocyte chemotactic protein-1 (MCP-1) and interleukin-1beta (IL1-β) (Fig. 1B). Quantification of CD11b after immunostaining of liver sections showed increased infiltration by inflammatory cells in MCD-fed ATGL-KO mice (Fig. 1C). Additionally, ATGL-KO mice showed increased expression of fibrosis markers such as collagen1α1 (Col1α1), collagen1α2 (Col1α2) and transforming growth factor β1 (TGF-β1) already at baseline followed by a further increase in the MCD diet which remained mild due to the rather short feeding duration (Fig. 1D). In WT MCD-fed mice, we observed a 3-fold increase in PPARα expression and a 4- to 15-fold increase in the expression of its target genes and serum fibroblast growth factor 21 (FGF21) concentration, which was blunted in ATGL-KO mice (Fig. 1E). Unlike previous reports,[8, 11] baseline expression of PPARα and its targets, CYP4A14 and CPT1α, were not affected in ATGL-KO mice because they were harvested in a nonfasted state.

Figure 1.

Genetic ablation of ATGL aggravates MCD-induced NASH and causes PPARα insufficiency. WT and ATGL-KO mice were fed with control or MCD diet for 2 weeks. Mice were harvested in a nonfasted state. (A) Representative hematoxylin and eosin (H&E) staining shows increased steatosis (left) in MCD-fed ATGL-KO mice, confirmed by hepatic TG quantification (right). (B) Hepatic gene expression of proinflammatory markers. (C) Representative CD11b immunohistochemistry (IHC) and its quantification shows increased mononuclear inflammatory cells in MCD-fed ATGL-KO compared to MCD-fed WT mice. CV, central vein. (D) Gene expression of fibrosis markers. (E) Hepatic gene expression of PPARα, its downstream targets, and serum FGF21 concentration. n = 4-7 per group. Bars represent mean ± SD. *P < 0.05 versus WT control diet-fed group. #P < 0.05, versus ATGL-KO control diet-fed group.

Since FA flux from WAT is a major determinant for hepatic lipid accumulation,[22] we next analyzed lipolytic activity in WAT of MCD-fed mice. MCD challenge resulted in a ∼30% and ∼23% loss of body weight in WT and ATGL-KO mice, respectively (Supporting Fig. 1A). WAT loss was also higher in MCD-fed WT compared to the MCD-fed ATGL-KO and was independent of food intake (Supporting Fig. 1B,C). In line with the increased WAT lipolysis and peripheral FA flux seen in NAFLD patients,[22] all WAT depots of WT MCD-fed mice showed increased total, ATGL, and HSL lipase activities (Supporting Fig. 1D). As a result of the excessive loss of visceral WAT, activity assays were technically not feasible. Interestingly, despite substantial WAT loss in MCD-fed ATGL-KO, we observed only a marginal increase of total (∼18%) and HSL (∼20%) lipase activity in gonadal WAT, while no increase was observed in perirenal and visceral WAT (Supporting Fig. 1E).

In contrast to the increased TG hydrolase activity in WAT, total TG hydrolase activity in liver was not affected by MCD treatment in both genotypes (Supporting Fig. 1F, left). However, ATGL-KO mice had a lower TG hydrolase activity compared to WT mice under both normal and MCD diet (Supporting Fig. 1F, left). Importantly, MCD feeding even reduced the non-HSL activity (i.e., ATGL+other lipases activity) in both WT and ATGL-KO mice (Supporting Fig. 1F, middle). However, HSL activity accounting for only 5%-15% of total activity in livers of mice and was significantly increased in MCD-fed ATGL-KO only (Supporting Fig. 1F, right). Collectively, these data indicate that the combined effects of increased FA flux from WAT to the liver and reduced non-HSL activity in liver contributes to MCD induced TG accumulation in this NASH model.

ATGL-KO Mice Are More Susceptible to LPS-Induced Acute Hepatic Inflammation

We next used a classic inflammatory model (LPS) to understand how ATGL deficiency contributes to acute inflammatory stress. LPS is known to induce anorexia,[21] therefore all groups were housed without food during the course of treatment. The 12-hour LPS-injected ATGL-KO mice showed marked up-regulation of inflammatory markers in liver compared to the LPS-injected WT mice (Fig. 2A). Additionally, serum TNF-α and IL6 levels after 6 and 12 hours, respectively, were higher in LPS-challenged ATGL-KO compared to the WT mice (Fig. 2B,C) (Serum TNF-α andIL6 levels were not detectable in controls). In line with this, CD11b immunostaining of liver sections showed a marked increase of inflammatory cells in LPS-injected ATGL-KO compared to the WT (Fig. 2D). Furthermore, after a single higher dose of LPS, ATGL-KO mice showed an 80% mortality rate as opposed to 100% survival in WT mice over a period of 5 days (Fig. 2E). Inflammatory gene expression analysis from WAT did not show any significant differences between LPS-injected WT and ATGL-KO mice (Fig. 2F), suggesting a liver-specific antiinflammatory role of ATGL.

Figure 2.

Exaggerated inflammatory response to endotoxin challenge in ATGL-KO mice. Nonfasted WT and ATGL-KO mice were injected with 4 μg/g body weight LPS or saline for 12 hours. All groups were fasted during the course of treatment. (A) mRNA expression of hepatic proinflammatory markers. (B,C) Serum concentration of TNF-α and IL6 after 6 hours (B) and 12 hours (C) LPS injection. (D) Representative CD11b IHC and its quantification shows marked increase in mononuclear inflammatory cells in LPS-injected ATGL-KO mice. (E) Higher mortality rate in ATGL-KO male mice (9-10 weeks old) when injected with 5 μg/g body weight LPS. n = 6 (KO), 11 (WT). (F) Proinflammatory markers in WAT (NS, nonsignificant). n = 5-7 per group. Bars represent mean ± SD. *P < 0.05 versus WT saline-injected group. #P < 0.05, versus ATGL-KO saline-injected group.

Impaired Hepatic FA Availability May Cause Defective PPARα Signaling and Torpor in LPS-Challenged ATGL-KO Mice

LPS-injected ATGL-KO mice exhibited signs of torpor and showed progressive lowering of their temperature over a period of 12 hours (Fig. 3A). FGF21, an endocrine regulator of ketogenesis and torpor[23] and a downstream PPARα target, showed only marginal expression in ATGL-KO livers compared to WT and was undetectable after LPS injection (Fig. 3B). Consequently, LPS-injected ATGL-KO mice showed low ketone bodies concentration that mirrored the expression pattern of FGF21 (Fig. 3C). However, LPS-injected WT mice showed more than a 2-fold increase in serum FGF21 levels (contribution from WAT, as also shown by a previous report[24]) which was undetectable in ATGL-KO mice (Fig. 3D). In line with this, hepatic PPARα DNA binding activity (DBA) and its mRNA expression were 60% lower at baseline in ATGL-KO and further reduced after LPS injection (Fig. 3E, left). Similarly, PPARα targets such as FA oxidation genes were already reduced in saline-injected ATGL-KO mice and their expression was further down-regulated by LPS injection in both WT and ATGL-KO mice (Fig. 3E, right). (On a cautionary note, baseline PPARα expression was not reduced in ATGL-KO mice in Fig. 1E because these mice were not fasted before harvesting.)

Figure 3.

PPARα insufficiency in LPS-injected ATGL-KO mice results from decreased hepatocellular FA delivery. (A) Rectal temperature of WT and ATGL-KO LPS- or saline-injected mice. (B) Hepatic FGF21 gene expression. (C) Serum ketone body concentration. (D) Serum FGF21 concentration. (E) Hepatic PPARα DNA binding activity (DBA) from 30 μg nuclear extract (left), PPARα gene expression (middle), and gene expression of CYP4A14 and ACO (right). (F) mRNA expression of FATP5 (left), FABP1 (middle), and western blot analysis of nuclear FABP1 (right) in liver. n = 4-5 per group. Bars represent mean ± SD. *P < 0.05 versus WT saline-injected group. #P < 0.05 versus ATGL-KO saline-injected group.

Since compromised PPARα activity in the livers of ATGL-KO mice may be due to a scarcity of the ligands for PPARα activation, we next explored mRNA expression of genes involved in hepatic FA binding and transport pathway. Notably, fatty acid transport protein 5 (FATP5) was lower in ATGL-KO compared to WT at baseline (Fig. 3F, left). Fatty acid binding protein 1 (FABP1), a PPARα target which shuttles FAs from cytosol to the nucleus and directly interacts with PPARα,[25] showed very low expression in ATGL-KO mice at baseline and its expression was further reduced after LPS injection (Fig. 3F, middle). In line with this, protein analysis from the nuclear extract showed ∼50% reduced FABP1 protein content in ATGL-KO compared to WT, which was further reduced after LPS injection (Fig. 3F, right). Taken together, these data suggest that in ATGL-KO mice the delivery of FAs into the hepatocyte nucleus may be compromised, which is further aggravated by LPS-induced impairment of PPARα signaling.

HSL Ablation Does Not Affect MCD- or LPS-Induced Hepatic Inflammation and PPARα Signaling

HSL is the rate-limiting enzyme for hydrolysis of diacylglycerols, the second step in TG hydrolysis.[6] Complementary to our approach in ATGL-KO mice, we next studied the impact of HSL ablation on susceptibility to NASH. MCD-fed HSL-KO mice had no significant differences compared to MCD-fed WT mice with regard to hepatic steatosis, TG levels, and CD11b quantification (Fig. 4A,B). Similarly, expression of hepatic proinflammatory markers, PPARα and its target genes, and serum FGF21 concentration did not show any significant differences between MCD-fed WT and HSL-KO mice (Fig. 4C). Furthermore, in contrast to the partial protection of WAT loss in ATGL-KO, HSL-KO mice showed similar loss of body weight and WAT depots as MCD-fed WT mice (Supporting Fig. 2A,B). Additionally, in contrast to the response of ATGL-KO mice, LPS challenge in HSL-KO mice revealed no differences to WT mice for CD11b-positive cells, mRNA expression of hepatic, and serum proinflammatory markers (Fig. 4D,E). PPARα DBA, its mRNA expression, and serum FGF21 levels were also the same in both WT and HSL-KO LPS-injected mice (Fig. 4F). Taken together, these data further indicate a specific role of the rate-limiting triglyceride lipase ATGL—but not the diglyceride lipase HSL—in protection from acute inflammation.

Figure 4.

HSL-KO mice do not show increased susceptibility to MCD or LPS-induced hepatic inflammation and PPARα insufficiency. (A-C) WT and HSL-KO mice were fed with an MCD diet for 2 weeks and were harvested in the nonfasted state. (A) Representative H&E staining and triglyceride quantification. (B) Representative CD11b IHC and its quantification. (C) Hepatic gene expression of proinflammatory markers, PPARα and its target genes (left), serum FGF21 concentration (right). n = 5-7 per group. Bars represent mean ± SD. (D-F) Nonfasted WT and HSL-KO mice were injected with LPS or saline for 12 hours. Mice were fasted during the course of treatment. (D) Representative CD11b IHC and its quantification (left), mRNA expression of hepatic proinflammatory markers (right). (E) Serum concentration of IL6 and TNF-α. (F) Hepatic PPARα DBA from 30 μg nuclear extract (left), hepatic PPARα gene expression (middle), serum FGF21 concentration (right). n = 4-5. Bars represent mean ± SD.

PPARα Agonist Treatment Partially Improves the Hepatic Phenotype of ATGL-KO Mice

To further explore the role of compromised PPARα signaling in ATGL-KO mice for aggravation of inflammation and steatosis, we fed MCD or 0.1% fenofibrate (FF) in addition to MCD diet to WT and ATGL-KO mice for 2 weeks. FF as a model PPARα ligand completely normalized hepatic steatosis in ATGL-KO mice and reduced CD11b-positive cells in both WT and ATGL-KO mice (Fig. 5A,B). MCP-1 expression was significantly reduced in ATGL-KO mice on FF-supplemented MCD diet (Fig. 5C). However, the expression of other proinflammatory genes such as iNOS, TNF-α, and IL1-β was not significantly reduced to the level of WT mice on FF-supplemented MCD diet (Fig. 5C). Notably, FF restored the expression of PPARα and its target genes in ATGL-KO mice to the level of WT mice (Supporting Fig. 3A). However, mRNA and serum levels of FGF21 did not increase in either genotype (Supporting Fig. 3B). Collectively, these data suggest that PPARα agonism (using FF as a model compound) reverts hepatic steatosis and partially improves the inflammatory response in MCD-fed ATGL-KO mice.

Figure 5.

PPARα agonist fenofibrate partially improves the hepatic inflammatory response of ATGL-KO mice. (A-C) WT and ATGL-KO mice were fed with MCD or MCD supplemented with 0.1% fenofibrate (MCD+FF) for 2 weeks. Mice were harvested in the nonfasted state. (A) Representative H&E staining and TG quantification. (B) Representative CD11b IHC and its quantification showing reduction in inflammatory cells in FF supplemented MCD group. (C) mRNA expression of hepatic proinflammatory markers. (D-F) WT and ATGL-KO mice were fed with 0.2% FF for 6 days followed by LPS injection to chow-fed (LPS) or FF-fed (FF+LPS) mice. Mice were fasted after a single LPS injection. (D) Representative CD11b IHC and its quantification showing reduction of inflammatory cells in FF-fed+LPS-injected groups. (E) mRNA expression (left) and serum concentration (middle) of IL6. mRNA expression of TNF-α, iNOS, and MCP-1 (right). (F) Survival rate of male WT and ATGL-KO mice (9-10 weeks old) prefed with 0.2% FF for 6 days followed by LPS injection (5 μg/g). n = 5-6 per group. For survival rate, n = 7 (KO), 9 (WT). Bars represent mean ± SD. *P < 0.05 versus corresponding MCD-fed or LPS-injected WT group. #P < 0.05, versus corresponding MCD-fed or LPS-injected ATGL-KO group.

We next assessed the influence of PPARα agonist treatment on LPS-induced acute hepatic inflammatory response in WT and ATGL-KO mice. Pretreatment with 0.2% FF for 6 days lowered the CD11b-positive cells in both WT and ATGL-KO LPS-injected mice, although the levels remained significantly higher in ATGL-KO compared to the WT (Fig. 5D). IL6 mRNA expression as well as serum concentrations were also reduced in both WT and ATGL-KO FF-pretreated LPS-injected mice (Fig. 5E). However, other inflammatory markers like TNF-α, iNOS, and MCP-1 were not reduced in WT mice, while we observed a trend towards decreased expression of these inflammatory markers in ATGL-KO mice (Fig. 5E). Moreover, FF-pretreatment lowered the mortality of ATGL-KO mice from 80% (Fig. 2E) to 40% when injected with a single higher dose of 5 μg/g LPS (Fig. 5F). FF-pretreatment restored PPARα DBA, its mRNA expression, and expression of its target genes in ATGL-KO mice to the level of WT FF-pretreated LPS-injected mice (Supporting Fig. 3C,D). However, nuclear FABP1 protein levels remained low in FF-pretreated LPS-injected ATGL-KO mice compared to the corresponding WT mice (Supporting Fig. 3E). Further, FGF21 mRNA expression and its serum levels were not restored, although marginally induced in ATGL-KO mice (Supporting Fig. 3F), suggesting a highly ATGL-dependent regulation of FGF21. Collectively, these results indicate that PPARα agonist treatment can partially improve the hepatic phenotype of LPS-injected ATGL-KO mice.

Discussion

This study unravels a hitherto unknown role of ATGL in protection from hepatic inflammation. Our data provide compelling evidence that ATGL—as a rate-limiting enzyme for TG breakdown and a major player in lipid partitioning and signaling by way of PPARα—may play a critical role in protection from hepatic inflammation as an integral feature of steatohepatitis/NASH. We further demonstrate that ATGL deficiency leads to an exaggerated inflammatory response upon endotoxin (LPS) challenge. Importantly, our data indicate that the antiinflammatory role of ATGL may partially depend on PPARα signaling.

MCD is a commonly used rodent model of NASH, resulting in endotoxin translocation, steatosis, and inflammation. This model recapitulates several key histological and molecular alterations, in particular inflammation and fibrosis, found in patients with NASH, despite several limitations including the absence of obesity and insulin resistance phenotype.[26-28] Other models have been described like fast food feeding, intragastric overfeeding, or high-fat diet feeding, which may be more representative of human NAFLD/NASH but require rather long time spans (25-35 weeks) for development of a full metabolic phenotype.[28-30] In NAFLD patients, increased peripheral FA flux from WAT has been shown to contribute ∼59% to the hepatic TG content, while only ∼26% from de novo lipogenesis and ∼15% from diet.[22] Our study now demonstrates that MCD diet as a murine model increases the activities of key lipolytic enzymes such as ATGL and HSL, resulting in increased flux of FAs from WAT to the liver, thereby causing steatosis. Loss of WAT depots in MCD-fed ATGL-KO mice suggests a role of other yet unknown lipase(s) which need to be further explored.

Notably, ATGL-KO mice are insulin-sensitive in contrast to the metabolic phenotype typically associated with human NAFLD.[7] However, the model offers an opportunity to test whether lack of ATGL may increase hepatic inflammation in progression to NASH due to altered lipid partitioning and impaired PPARα signaling. In ATGL-KO mice, the activity of other compensatory lipases in liver accounted for 65%-75% of the total TG hydrolase activity,[6, 9] which, however, was not sufficient for TG hydrolysis, as evident by the presence and MCD-induced aggravation of steatosis in their livers. In MCD-fed ATGL-KO mice, the slightly increased HSL activity may be viewed as a futile attempt to hydrolyse TGs in the absence of ATGL. This may suggest a rather nonspecific role of HSL in liver which could also be responsible for the overall unchanged total lipase activity in liver on MCD feeding. Collectively, these findings strongly suggest that ATGL in liver is critical for TG breakdown and thereby counteracts hepatic steatosis and inflammation. Our findings also support the concept that ATGL-generated FAs may be important for activation of PPARα in liver.[10] Low expression of PPARα in addition to low ligand availability may further compromise its antiinflammatory effects and FA oxidation in ATGL-KO.

To further study the role of ATGL in hepatic inflammation as a universal feature of progressive liver disease (including NASH), we challenged mice with an endotoxin (LPS). Upon LPS challenge, ATGL-KO mice showed increased hepatic inflammation compared to WT mice. Notably, PPARα expression is higher in liver compared to the adipose tissue, possibly explaining why the enhanced inflammatory response was observed specifically in the liver. Since binding of LPS to lipoproteins neutralizes LPS in vitro and also protects animals from LPS-induced fever, hypotension, and death,[31, 32] low serum lipoproteins in ATGL-KO mice[7] could also contribute to the exaggerated acute phase response and high mortality in LPS-injected ATGL-KO mice. Low levels of ketone bodies in serum of ATGL-KO mice indicate impaired ATGL-catalyzed lipolysis and as a result decreased FAs as a substrate for ketone body synthesis in liver. Diminished expression of hepatic FATP5 and FABP1-mediated FA uptake in liver and lower levels of nuclear FABP1 in ATGL-KO livers may further contribute to PPARα insufficiency during acute inflammatory stress in these animals.

In line with a key role of ATGL in PPARα signaling and an antiinflammatory role of PPARα,[5, 6, 10, 14, 33] we used FF as a model PPARα agonist to test whether its activation can ameliorate the hepatic phenotype in ATGL-KO mice. FF treatment was able to partially attenuate MCD-induced NASH and LPS-induced inflammation in ATGL-KO mice. Importantly, IL6 mRNA expression as well as serum concentrations were reduced in both WT and ATGL-KO FF-pretreated LPS-injected mice. This is in line with previous findings which show that IL6 is the principal inducer of acute phase response (APR) gene expression and PPARα activation by FF reduces IL6-mediated APR gene expression.[14, 15] Contrary to our findings, a previous study reported that a PPARα agonist does not restore the expression of PPARα and its target genes to the WT levels in mice injected with ATGL short hairpin RNA (shRNA).[8] This discrepancy could be due to a lower dose of FF and also the overnight fasting in the other study,[8] which further suggests an indispensable role of the lipolytic products of ATGL during stress conditions. Additionally, our data suggest a role of ATGL for regulation of FGF21, in line with its recently uncovered role for PPARα regulation.[10] Untreated and LPS-injected ATGL-KO mice had undetectable serum FGF21 levels that was only marginally induced by FF pretreatment in contrast to the pronounced induction in WT mice upon LPS injection with or without FF-pretreatment. Since FGF21 protects mice from the toxic effects of LPS,[24] the lack of circulating FGF21 in ATGL KO mice could further aggravate the inflammatory response in these mice.

While ATGL ablation enhanced the hepatic inflammatory response in the current study, its absence protected mice from tunicamycin-induced hepatic ER stress.[34] These two separate effects of ATGL may be due to the distinct signaling pathways involved in the modulation of MCD or LPS-induced inflammation versus tunicamycin-mediated ER stress. Also, in contrast to the antiinflammatory effect of ATGL shown in the present study, ATGL ablation in macrophages has been shown to attenuate atherosclerotic lesions.[35] Recent studies have shown amelioration of hepatic steatosis in mice by adenoviral overexpression of ATGL and HSL,[9] while liver-specific ATGL-KO[11] resulted in increased hepatic steatosis. While these studies suggest a liver-specific role of ATGL or HSL in hepatic lipid metabolism and signaling, our data add an important additional aspect with regard to the indispensable role of ATGL in regulating hepatic inflammation. In line with our current observations, mice lacking the ATGL coactivator, CGI-58, in liver also develop steatohepatitis and fibrosis.[36] Notably, mice lacking HSL (the rate-limiting diglyceride lipase) did not show aggravated response to MCD and LPS, which argues against a critical role of the lipolytic products of HSL in modulating the inflammatory response. Figure 6 summarizes the proposed mechanisms which may result in increased susceptibility to hepatic inflammation in the absence of ATGL.

Figure 6.

Proposed mechanism for enhanced susceptibility to hepatic inflammation in ATGL-KO mice upon MCD or LPS challenge. Under normal conditions (WT mice), the classical pathway of lipolysis comprising ATGL, HSL, and MGL in adipose tissue and liver provides FAs to the liver for activation of PPARα along with endogenously synthesized and dietary FAs. In ATGL-KO mice, flux of FAs from this classical pathway of lipolysis in adipose tissue to the liver is decreased and loss of ATGL in liver itself further limits FA release from triglycerides. Hence, during inflammatory stress FAs from other pathways cannot completely compensate for the absence of lipolytic products provided by ATGL for PPARα activation due to the deficiency of FABP1. Consequently, antiinflammatory PPARα signaling is impaired, resulting in enhanced inflammation and APR, while impaired FGF21 signaling leads to decreased ketogenesis and torpor. Decreased FA oxidation further contributes to steatosis and insufficient metabolic fueling to sustain the inflammatory stress. Additionally, ATGL also regulates FGF21 signaling independent of PPARα. Altogether, these series of events render ATGL-KO mice susceptible to hepatic inflammation in mouse models of NASH and endotoxemia, which can be ameliorated by treatment with a model PPARα agonist.

In conclusion, our findings with the MCD and LPS mouse models provide compelling evidence that ATGL may play a key role in protection from hepatic inflammation. Future studies will have to address the potential prognostic and therapeutic relevance of ATGL (PNPLA2) or its coactivator (CGI-58) for progression of human (metabolic) liver disease to potentially lethal consequences such as liver cirrhosis and cancer.

Ancillary