Evolution of inflammation in nonalcoholic fatty liver disease: The multiple parallel hits hypothesis


  • Herbert Tilg,

    Corresponding author
    1. Christian Doppler Research Laboratory for Gut Inflammation, Medical University Innsbruck, Innsbruck, Austria
    • Christian Doppler Research Laboratory for Gut Inflammation, Medical University Innsbruck, Anichstrasse 35, 6020 Innsbruck, Austria
    Search for more papers by this author
    • fax: +43 512 504 67 23374

  • Alexander R. Moschen

    1. Christian Doppler Research Laboratory for Gut Inflammation, Medical University Innsbruck, Innsbruck, Austria
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


Whereas in most cases a fatty liver remains free of inflammation, 10%-20% of patients who have fatty liver develop inflammation and fibrosis (nonalcoholic steatohepatitis [NASH]). Inflammation may precede steatosis in certain instances. Therefore, NASH could reflect a disease where inflammation is followed by steatosis. In contrast, NASH subsequent to simple steatosis may be the consequence of a failure of antilipotoxic protection. In both situations, many parallel hits derived from the gut and/or the adipose tissue may promote liver inflammation. Endoplasmic reticulum stress and related signaling networks, (adipo)cytokines, and innate immunity are emerging as central pathways that regulate key features of NASH. (HEPATOLOGY 2010;52:1836-1846)

Nonalcoholic fatty liver disease (NAFLD) includes a disease spectrum ranging from simple steatosis to nonalcoholic steatohepatitis (NASH), liver fibrosis, cirrhosis, and hepatocellular carcinoma.1 The majority of patients with NAFLD are obese or even morbidly obese and have accompanying insulin resistance.2-4 The proportion of patients with NAFLD who have NASH is still not entirely clear but might range from 10%-20%. This is relevant because inflammation and/or fibrosis determine the long-term prognosis of this disease, whereas steatosis per se might not adversely affect outcome.5-8 Most studies indicate that 1%-3% of the Western population might have NASH. The natural history of NAFLD is still poorly understood, and in particular, it is not known why certain patients progress toward inflammation, fibrosis, and cirrhosis and why others do not. One of the burning questions in NAFLD remains which factors could be the driving forces toward a more progressive, inflammatory disease phenotype. Day and colleagues presented more than a decade ago the so-called “two-hit” model, suggesting that after a first hit (i.e., hepatic steatosis) another hit (e.g., gut-derived endotoxin) is needed to develop NASH.9

Because simple hepatic steatosis is a benign process in the majority of patients, NASH might be a separate disease with a different pathogenesis. Here, we propose a new model suggesting that many hits may act in parallel, finally resulting in liver inflammation and that especially gut-derived and adipose tissue–derived factors may play a central role. Inflammation may precede steatosis in NASH, as inflammatory events may lead to subsequent steatosis. Furthermore, we want to highlight the potential importance of endoplasmic reticulum (ER) stress in various aspects of this disease.


AhR, aryl hydrocarbon receptor; ATF-6, activating transcription factor 6; ChREBP, carbohydrate response element-binding protein; DGAT, diacylglycerol acyltransferase; DNL, de novo lipogenesis; ER, endoplasmic reticulum; IKKβ, inhibitor of nuclear factor-κB kinase-β; Gpr, G protein–coupled receptor; IL, interleukin; IRE1, inositol-requiring enzyme 1; JNK1, c-jun N-terminal protein kinase 1; LPS, lipopolysaccharide; mRNA, messenger RNA; PERK, pancreatic ER kinase; PI3K, phosphatidyl inositol 3-kinase; patatin-like phospholipase 3 PNPLA3; PPARγ, peroxisome proliferator-activated receptor-gamma; ROS, reactive oxygen species; SCFA, short chain fatty acid; SOCS3, suppressor of cytokine signaling 3; SREBP, sterol regulatory element-binding protein; TLR, toll-like receptor; TNF, tumor necrosis factor; UDCA, ursodeoxycholic acid; UPR, unfolded protein response; XBP1, X-box binding protein 1.

Development of Hepatic Steatosis

A fatty liver is the result of the accumulation of various lipids.10 Several mechanisms may lead to a fatty liver: (1) increased free fatty acids supply due to increased lipolysis from both visceral/subcutaneous adipose tissue and/or increased intake of dietary fat; (2) decreased free fatty oxidation oxidation; (3) increased de novo hepatic lipogenesis (DNL) and (4) decreased hepatic very low density lipoprotein–triglyceride secretion.11 Free fatty acid delivery to the liver accounts for almost two-thirds of its lipid accumulation.12 Elevated peripheral fatty acids and DNL therefore predominantly contribute to the accumulation of hepatic fat in NAFLD. Besides the well-established lipogenesis-controlling factors such as sterol regulatory element-binding protein (SREBP) or carbohydrate response element-binding protein (ChREBP), X-box binding protein 1 (XBP1), known as a key regulator of the unfolded protein response (UPR) secondary to ER stress, is a only recently characterized regulator of hepatic lipogenesis.13

Triglycerides are the main lipids stored in the liver of patients with NAFLD. Although large epidemiological studies suggest triglyceride-mediated pathways might negatively affect disease,14 recent evidence indicates that trigylcerides might exert protective functions. Diacylglycerol acyltransferase 1 and 2 (DGAT1/2) catalyze the final step in triglyceride synthesis. In a model of diet-induced obesity, mice with overexpression of DGAT1 in adipocytes and macrophages are protected from macrophage activation and their accumulation in white adipose tissue, from systemic inflammation and insulin resistance.15 Inhibition of triglyceride synthesis via DGAT2 antisense oligonucleotides improves liver steatosis but worsens liver damage, also suggesting that accumulation of liver triglycerides could be a protective mechanism.16 Hepatic steatosis (i.e., triglyceride accumulation) is dissociated from insulin resistance in patients with familial hypobetalipoproteinemia, providing further evidence that increased intrahepatic triglyceride content might be more a marker rather than a cause of insulin resistance.17 In summary, triglyceride synthesis seems to be an adaptive, beneficial response in situations where hepatocytes are exposed to potentially toxic triglyceride metabolites. Thus, evidence is increasing that accumulation of fat in the liver in many instances cannot be regarded as a pathology or disease, but rather as a physiologic response to increased caloric consumption.18

Free fatty acids and cholesterol, especially when accumulated in mitochondria, are considered the “aggressive” lipids leading to tumor necrosis factor alpha (TNFα)-mediated liver damage and reactive oxygen species (ROS) formation.19, 20 These lipids could also be present in a nonsteatotic liver and act as early “inflammatory” hits leading to the whole spectrum of NAFLD pathologies. The concept of lipotoxicity and involved lipid species has been introduced and discussed in several excellent review articles.21, 22

Inflammation Preceding Steatosis.

Simple hepatic steatosis, which is benign and nonprogressive in the majority of patients, and NASH may reflect different disease entities. Inflammation results in a stress response of hepatocytes, may lead to lipid accumulation, and therefore could precede steatosis in NASH. Such a cascade is supported by various studies. Patients with NASH may present without any or much steatosis, suggesting that inflammation could take place first.1 Anti-TNF antibody treatment and metformin, an antidiabetic drug that inhibits hepatic TNFα expression, improve steatosis in ob/ob mice.23, 24 Other proinflammatory mediators might also contribute to the development of steatosis because in some studies hepatic steatosis was not dependent on TNFα.25, 26 In patients with severe alcoholic hepatitis, treatment with infliximab, an anti-TNF antibody, primarily improves hepatic steatosis.27 Loss of Kupffer cells also leads to hepatic steatosis probably via decreased interleukin-10 (IL-10) release from Kupffer cells.28 Other cell types might also promote hepatic steatosis because obesity leads to the hepatic recruitment of a myeloid cell population that further promotes hepatic lipid storage.29 In all these situations, hepatic steatosis may be considered as “bystander phenomenon” subsequent to inflammatory attacks. Very diverse processes including toxic lipids, nutrients, and other gut-derived and adipose-derived signals (as discussed later) may represent such inflammatory insults.

Certain Dietary Factors: A Direct Roadmap to Lipotoxicity? The consumption of trans-fatty acids has increased dramatically in the last decades and mice fed trans-fatty acids develop larger livers with NASH-like lesions and insulin resistance.30 Although virtually absent from our diet in the past, fructose has now become a major constituent of modern diet. When obese subjects consumed glucose- or fructose-sweetened beverages for 10 weeks, fasting plasma glucose and insulin levels increased and insulin sensitivity decreased in subjects consuming fructose but not in those consuming glucose.31 Daily fructose consumption is associated with increased hepatic inflammation and fibrosis in humans.32 The aryl hydrocarbon receptor (AhR) is a ligand-activated transcription factor sensing xenotoxicants such as dioxin. This pathway may play a major role in inflammatory processes.33 Many AhR agonists are present in the diet such as indolo-(3,2-b)-carbazole and 3,3′-diindolylmethane (metabolized from indole 3-carbinol), or flavonoids. Transgenic mice with constitutively activated AhR develop spontaneous hepatic steatosis and increased hepatic oxidative stress.34 It remains to be identified how certain nutrients might directly lead to liver inflammation.

Gut-Derived Signals Beyond Endotoxin May Promote Liver Inflammation

Conventionalization of germ-free mice with a normal microbiota leads to weight gain, obesity, and insulin resistance, which suggests that the microbiota and/or microbiota-regulated host factors might influence energy absorption, adiposity, systemic inflammation, and development of insulin resistance.35, 36

Endotoxin and Its Role in Obesity.

Endotoxin (lipopolysaccharide [LPS]), a key constituent of many bacteria present in our microbiota, plays a central role in innate immune responses and has been considered the so-called “second hit” in previous NASH models.9 Manipulation at the gut surface, including dietary ingredients, may affect LPS metabolism and result in increased circulating plasma levels. It has been demonstrated that intake of a high-fat or a high-carbohydrate diet in humans over only 3 days leads to an increase in circulating LPS concentrations (i.e., “second hit”).37 Endotoxemia, however, might not only lead to systemic inflammation but might also worsen obesity itself.38 When endotoxemia was induced for 4 weeks in lean mice, liver and adipose tissue weight gain were increased similarly as after a high-fat diet. This weight gain was paralleled by hepatic insulin resistance, and could be prevented by antibiotic therapy. Patients with NAFLD demonstrate increased gut permeability, which importantly has been associated with the severity of liver steatosis but not with the degree of inflammation (NASH).39 This study therefore suggests that gut-derived factors/signals such as endotoxin might also affect accumulation of hepatic fat.

Intestinal Epithelium: Linking Nutrients to Metabolic Diseases.

Our microbiota might influence systemic immune responses. Such an effect might take place via their capacity to digest dietary fiber resulting in the production of short-chain fatty acids (SCFA). SCFAs have anti-inflammatory functions in various models of colitis and human ulcerative colitis probably via interaction with its receptor, the G protein–coupled receptor 43 (Gpr43).40 Gpr43−/− mice show systemic inflammation in various tissues,41 similar to germ-free wild-type mice devoid of bacterial fermenting capacity and hence with almost absent SCFAs in the gut. Various other pathways (i.e., fasting-induced adipose factor; Gpr41) have been characterized that might interfere with metabolism/adiposity, highlighting how the intestinal microbiota and its products might directly regulate host gene expression and affect systemic inflammation.42-45 These pathways involve the intestinal epithelium as “sensor” of the microbiota, implicating a major role for the intestinal epithelium in determining systemic metabolic functions (for details, see Fig. 1). Interference with our microbiota via probiotics or prebiotics might therefore be beneficial and improve systemic inflammation/metabolic function. So far, only a few animal studies have been performed that suggest that this might indeed be the case.23, 46, 47

Figure 1.

The multiple parallel hits model. Lipotoxicity: (1) A liver loaded with lipids consisting primarily of trigylcerides might reflect a benign process because trigylcerides might exert mostly protective effects. Furthermore, hyperleptinemia leads to oxidation of hepatic lipids, thereby also protecting this organ from lipotoxicity. When the capacity of peripheral and central organs of detoxifying “aggressive lipids” fails, lipotoxic attack of the liver might begin. Inflammation may precede steatosis in NASH. Gut-derived signals: Many signals beyond endotoxin might affect hepatic steatosis and inflammation. Several pathways have been identified how the gut microbiota might influence host energy metabolism: (2) Absence of the microbiota in germ-free mice correlates with increased activity of phosphorylated AMPK in the liver and the muscle (not shown). (3) Some of the breakdown products of polysaccharides are metabolized to SCFAs. SCFAs such as propionate and acetate are ligands for the G protein–coupled receptors Gpr41 and Gpr43. Shortage of SCFAs might allow the evolution of systemic inflammatory events. Such mechanisms elegantly combine diet, microbiota, and the epithelial cell as “nutrient sensor.” (4) The microbiota decreases epithelial expression of fasting-induced adipocyte factor (Fiaf), which functions as a circulating lipoprotein lipase (LPL) inhibitor and therefore is an important regulator of peripheral fat storage. (5) Several TLRs, such as TLR5 or TLR9, are not only able to affect microbiota but also to regulate metabolism, systemic inflammation, and insulin resistance, thus highlighting the role of the innate immune system in metabolic inflammation as observed in NASH. (6) Various nutrients such as trans fatty acids (TFAs), fructose or aryl hydrocarbon receptor (AhR) ligands such as 2,3,7,8-tetrachlorodibenzodioxin (TCDD) may directly lead to steatosis/liver inflammation. Adipose tissue–derived signals: Signals derived from the adipose tissue beyond toxic lipids might play a central role in NAFLD/NASH. (7) Here, adipocytokines such as adiponectin and leptin, certain proinflammatory cytokines such as TNFα or IL-6, and others (the death receptor Fas, PPARγ) are of key relevance. The cytokine/adipocytokine milieu might be critical because ob/ob-adiponectin tg mice, although becoming severely obese, are not insulin-resistant. This suggests that in the hierarchy of processes soluble mediators play the central role. Adipose-derived mediators might indeed affect target organs such as the liver, because JNK1 adipose-deficient mice are protected from diet-induced obesity, and experiments have demonstrated that this effect is mediated mainly by IL-6 (a cytokine), which is of key importance in human obesity.

Toll-Like Receptors and Role of Innate Immunity in Obesity-Related Inflammation.

Toll-like receptors (TLRs), also expressed on the gut epithelium, can respond to nutritional lipids such as free fatty acids and might thereby have a role in the pathogenesis of obesity-associated inflammation/insulin resistance.48 The recognition of fatty acids by TLR4 can induce the production of proinflammatory cytokines in macrophages and epithelial cells.49 TLR-4–deficient mice are protected from high-fat diet-induced inflammation and insulin resistance.50 It is, however, not universally accepted whether saturated free fatty acids are ligands for certain TLRs because it has been demonstrated that saturated fatty acids might not directly stimulate TLR-dependent signaling.51 Therefore, observed effects in the above discussed in vivo study49 could also be accounted by gut-derived endotoxin or by endotoxin contamination of the lipids employed.

Other TLRs may also be involved in obesity-related inflammation. TLR9 promotes steatohepatitis because TLR9-deficient mice are protected from liver inflammation.52 The importance of the gut as “metabolic organ” has been convincingly demonstrated by a recent report indicating that mice deficient in TLR5 develop all features of metabolic syndrome including hyperphagia, obesity, insulin resistance, pancreatic inflammation, and hepatic steatosis.53 TLR5 deficiency affected the composition of the gut microbiota and, remarkably, transfer of the microbiota from TLR5−/− mice to healthy mice resulted in transfer of disease. There are two major implications of this work: (1) the innate immune system plays a critical role in the development of the metabolic syndrome and (2) transfer of the gut microbiota to wild-type germ-free mice results in several features of de novo disease (i.e., metabolic syndrome), again supporting a major role for our microbiota in metabolic inflammation.

Adipose Tissue-Derived Signals: The Adipose Tissue Attacks the Liver

Adipose tissue has appeared in the last decade as a highly active endocrine and immune organ with the capacity of producing various mediators including adipocytokines and cytokines both in health and disease. The balance/imbalance of an adipose tissue “mediator cocktail” may profoundly affect not only the situation in the adipose tissue but especially in important target organs such as the liver (Fig. 1).

Adiponectin: Prototypic Adipocytokine in Health and Disease.

Adiponectin is an anti-inflammatory adipocytokine that signals through two receptors.54-56 Obesity is associated with hypoadiponectinemia, and adiponectin levels increase after weight loss.55 Adiponectin induces extracellular Ca2+ influx by adiponectin receptor 1, which is necessary for activation of adenosine monophosphate–activated protein kinase (AMPK) and Sirtuin 1 (Sirt1).57 Hepatocyte-specific deletion of Sirt1 leads not only to hepatic steatosis but also to ER stress and liver inflammation.58 Genetically obese leptin-deficient ob/ob mice exhibit a reversal of the diabetic phenotype with normalization of glucose and insulin levels upon transgenic overexpression of the full-length isoform of adiponectin, despite retaining the obese phenotype.59 This report convincingly demonstrates that, despite massive expansion of subcutaneous adipose tissue, high-level expression of adipose tissue adiponectin reduces liver fat content and improves insulin resistance. Therefore, also in humans, a sufficient production of adiponectin might play a central role in establishing a balance where local and systemic/liver inflammation is prevented.60 In the hierarchy of processes in the adipose tissue, soluble mediators such as adiponectin might be the “big players.”


Because adipocytes expand with triglycerides, leptin secretion increases proportionally.61 Hyperleptinemia reduces fat content in peripheral organs. Because leptin stimulates fatty acid oxidation, adipocytes would be oxidizing, rather than storing fat if the endogenous leptin they synthesize acts on them.62, 63 Such an autocrine/paracrine relationship between leptin and its secreting cell, the adipocyte, is prevented by a progressive decline of adipocyte leptin receptor expression. It is assumed that leptin's capacity to oxidize lipids is fully operative in the liver, thereby minimizing ectopic lipid accumulation, at least temporarily. Whether such a mechanism is operative in NAFLD is not known.

IL-6 and TNFα: Key (Adipo)cytokines.

Expression of IL-6 and TNFα, two important proinflammatory cytokines, is profoundly increased in human fat cells from obese subjects and patients with insulin resistance.64 IL-6 serum levels are elevated in obese patients and weight loss results in decreased IL-6 serum levels.65, 66 Enhanced TNFα expression in adipose tissue of obese subjects decreases following weight loss.67

Insulin resistance is an important feature of NAFLD and is caused by a variety of factors, including soluble mediators derived from immune cells and/or adipose tissue.68 Insulin resistance may augment inflammation in NASH because patients with type 2 diabetes mellitus are often worse in terms of histopathological changes such as ballooning, apoptosis, and lobular and/or portal inflammation.1 Serine phosphorylation of insulin receptor substrate by inflammatory signal transducers such as c-jun N-terminal protein kinase 1 (JNK1) or inhibitor of nuclear factor-κB kinase-β (IKKβ) is considered one of the key aspects that disrupt insulin signaling. Sabio et al. reported that JNK1 signaling specifically in adipose tissue consequent to a high-fat diet causes hyperinsulinemia, hepatic steatosis, and hepatic insulin resistance.69 Importantly, this distal effect of adipose tissue on the liver was mediated via increased JNK1-dependent IL-6 secretion from adipocytes, proving that adipose tissue–derived IL-6 regulates distal metabolic effects in the liver. It has to be stated that in this and other models, a high-fat diet is a prerequisite to induce “pathology,” telling us that indeed “an inflammatory diet” might exist that drives certain processes including liver inflammation at the end.

We recently demonstrated that such a mechanism as suggested by Sabio et al. might also be operative in human obesity.70 In this study, IL-6 expression has been more than 100-fold higher in adipose tissue (subcutaneous and visceral) compared to its liver expression, suggesting that in severe obesity, the adipose tissue is indeed the major source of IL-6. Weight loss resulted in a dramatic decrease, especially of IL-6 and TNFα expression with subsequent reduced expression of hepatic suppressor of cytokine signaling 3 (SOCS3) expression and improved insulin sensitivity, and hence evidence of hepatic consequences of these alterations in adipose tissue. The liver might be a key target organ for adipose tissue–derived IL-6 and TNFα, because continuous IL-6/TNFα exposure affects hepatic insulin resistance, e.g., via up-regulation of SOCS3.71 Importantly, enhanced expression of proinflammatory cytokines in adipose tissue was observed, although liver inflammation was still absent, suggesting that adipose tissue inflammation could precede liver inflammation.70 Peroxisome proliferator-activated receptor-gamma (PPARγ), a member of the nuclear receptor family, plays a major role in adipogenesis, atherosclerosis, inflammation, and glucose metabolism. Adipose tissue–specific deletion of PPARγ results in diminished weight gain despite hyperphagia, diminished serum concentrations of leptin/adiponectin, and insulin resistance.72, 73 Mice with a deficiency of the death receptor Fas specifically in adipocytes are not only protected from adipose tissue inflammation (induced by a high-fat diet) but also from hepatic steatosis and hepatic insulin resistance.74

Many human studies suggest that the amount of visceral fat directly correlates with degree of hepatic steatosis and inflammation. Hepatic inflammation and fibrosis correlate with the amount of visceral fat.75 Trunk fat has been shown to be a major factor leading to increased serum alanine aminotransferase levels, which might reflect more advanced disease such as NASH.76 This large clinical study further supports the important association between adipose tissue and liver disease. Besides certain adipocytokines/immune mediators, the cellular infiltrate in the adipose tissue is also of major importance because ablation of adipose macrophages (CD11c+ cells) improves insulin sensitivity and decreases inflammation.77 Importantly, adiponectin and PPARγ promote adipose tissue macrophage polarization toward an alternative/anti-inflammatory phenotype.78, 79 Altogether, our and several other studies80 present evidence that adipose tissue inflammation is a common event in morbid obesity, and this tissue could reflect the major cytokine source in obesity. Adipose tissue–derived mediators might attack the liver, thus promoting liver inflammation.

A Key Role for IL-6 and TNFα in Steatohepatitis and Tumor Formation in Rodents.

Park and colleagues recently demonstrated that these two cytokines play a central role in the promotion of liver inflammation and tumorigenesis in dietary and genetic obesity.81 In their studies, obesity-related liver tumor development was dependent on enhanced production of the tumor-promoting cytokines IL-6 and TNFα which both cause liver inflammation and activation of the oncogenic factor STAT3. IL-6−/− and TNFR1−/− mice are resistant to obesity-related tumor promotion. The absence of either IL-6 or TNF receptor 1 (TNFR1), decreased high-fat diet induced liver lipid accumulation and liver inflammation as assessed by reduced infiltration with macrophages and neutrophils. The role of IL-6, however, is probably more complex because other studies have demonstrated that IL-6 can prevent obesity82 or that IL-6–deficient mice are prone to obesity.83

Previous studies have shown that hepatocyte-specific deletion of the IKK regulatory subunit NF-κB essential modifier (NEMO)/IKKγ results in spontaneous liver damage, hepatosteatosis, liver fibrosis, and tumor development.84, 85 Therefore, many studies support the notion that the cytokine milieu in the liver plays a critical role in the development of many features of human NAFLD including inflammation, fibrosis, and tumor development.

ER Stress: XBP1 as Missing Link Between Steatosis, Insulin Resistance, and Inflammation?

A chronic imbalance between energy supply and demand, as observed in obesity, might expose cells to toxic lipids, thereby activating cellular stress pathways. This type of cellular stress originates from the accumulation of unfolded or misfolded proteins in the ER and usually triggers an adaptive response aimed at resolving ER stress, the UPR.86 The UPR is mediated by at least three different stress-sensing pathways including pancreatic ER kinase (PERK), inositol-requiring enzyme 1 (IRE1), and activating transcription factor 6 (ATF6). IRE1, apart from acting as a kinase, also possesses endoribonuclease activity, thereby excising a 26-nucleotide fragment from XBP1 messenger RNA (mRNA), which results in a frame-shift and consequent translation of the active transcription factor XBP1s. ER stress (XBP1), however, not only regulates lipid synthesis but also interacts with inflammatory cascades at various stages: (1) IRE1-mediated activation of JNK; (2) activation of IKK–NF-κB signaling pathways; and (3) production of ROS87 (Fig. 2). All of those three pathways, i.e., JNK, IKK–NF-κB, and ROS, have been demonstrated to be involved in the regulation of obesity-related insulin resistance and inflammation. Free fatty acids may also induce ER stress,88 whereas certain adipose tissue–derived unsaturated fatty acids such as palmitoleate (i.e., “lipokines”) might inhibit ER stress and related events.89, 90 The liver needs to adapt and to function in obesity-related disorders under this chronic exposure to high energy and nutrient intake. Hepatocytes maintain one of the highest protein synthesis rates in the body and can produce millions of proteins per minute. Therefore, failure to maintain ER integrity may develop in such an organ more easily and lead to other ER-stress–controlled events such as inflammation.

Figure 2.

ER stress/XBP1: a unifying pathway regulating many aspects of experimental and human fatty liver disease. The unfolded protein response comprises at least three different stress-sensing pathways, namely the pancreatic ER kinase (PERK), inositol-requiring enzyme 1 (IRE1), and activating transcription factor 6 (ATF6). (1) IRE1 unconventionally splices XBP1 mRNA, with spliced XBP1s mRNA encoding the active transcription factor. (2) Intact insulin signaling inhibits the formation of heterodimers of the regulatory subunits p85α and p85β of PI3K. p85 monomers promote nuclear translocation of XBP1s. (3) XBP1 not only regulates hepatic de novo lipogenesis independently from other key lipogenic transcription factors, but has also been demonstrated to induce differentiation of adipocytes, hence affecting adipogenesis. (4) Decreased XBP1 function leads to increased JNK phosphorylation and downstream signaling via a mechanism that involves overactivation of IRE1 in the context of ER stress. JNK in turn affects insulin resistance by phosphorylation of IRS-1 on Ser307 and may contribute to inflammatory processes. (5) Similarly, the key proinflammatory transcription factor NF-κB is activated by IRE1. (6) In addition, ER stress has been demonstrated to inhibit leptin signaling. (7) Only recently has it been demonstrated that TLR2 and TLR4 activate IRE1 and its downstream target XBP1.114

Importantly, two reports have recently opened a completely new aspect for XBP1 demonstrating that certain subunits of the insulin signaling pathway (phosphatidyl inositol 3-kinase [PI3K], namely p85α and p85β) interact and increase the nuclear translocation of XBP1s.91, 92 XBP1 has evolved as a critical molecule that is able to regulate all aspects of NAFLD, namely lipid synthesis/accumulation, leptin resistance, adipogenesis, inflammation, and insulin signaling/resistance.93, 94

Autophagy has recently evolved as an additional pathway affecting hepatic lipid metabolism. Autophagy, a phylogenetically conserved response to cellular starvation, regulates lipid metabolism because inhibition of autophagy in cultured hepatocytes and murine livers increases triglyceride storage.95 Autophagy seems to critically interact with ER stress because impaired hepatic autophagy results in elevated ER stress and defective insulin signaling.96 Therefore, ER stress and autophagy appear as closely intertwined pathways in inflammatory diseases.

Genetics and NAFLD

Genetic factors might be attractive candidates to explain why a certain percentage of patients with fatty liver develop inflammation. NAFLD is a heritable disorder, suggesting there are genetic components that predispose to these traits. Polymorphisms in patatin-like phospholipase 3 (PNPLA3), encoding a protein of unknown function with homology to lipid acyl hydrolases, has been strongly associated with increased hepatic fat content in NAFLD.97 Several other genetic variants have been identified, although with less convincing evidence, except for apolipoprotein C3.98PNPLA3 findings have been confirmed by several groups99-101 and recent studies have demonstrated that homozygosity for the PNPLA3 148M polymorphism is associated with severity of disease (degree of inflammation, liver fibrosis).102-104PNPLA3-deficient mice develop neither fatty liver, elevated aminotransferases, nor insulin resistance.105 How PNPLA3 mutations confer this high risk for NASH therefore remains unclear. We have to keep in mind that NAFLD is part of a syndrome strongly overlapping with obesity and insulin resistance and therefore it seems likely that common genetic aspects for all those diseases exist. Whereas genetic factors overall may play a minor role in the current epidemic of obesity, certain genetic factors might well offer explanations for a more progressive disease course in NAFLD.

Multiple Hits: Testable Hypotheses

NAFLD is a complex disease with no simple answers. Presented data, however, suggest that extrahepatic tissues could play an important role in the evolution of liver inflammation. Before advancing toward therapeutic human studies, e.g., interfering with our microbiota, more information on the natural history of this disease is needed. Human studies investigating the microbiota/microbiome should be initiated to define whether there exists a “NASH-associated” (core) microbiome.106 To support our hypothesis, tissue-specific knockout animal models (adipose-specific, epithelial-specific, and macrophage-specific knockout mice) with special emphasis on mediators directing innate immune processes are needed. Interbreeding these mice will enable experiments to prove that “a defect” at both levels could induce a more inflammatory and progressive disease phenotype. Obesity and related disorders including NAFLD are the consequence of our current lifestyle and therefore “inflammatory” diets such as those rich in trans fatty acids and/or fructose, diets that activate the AhR, or others have to be investigated in various animal models to better define the “major triggers” in our diet.

Based on our hypothesis, various potential treatment targets may evolve and treatment approaches beyond focusing on insulin resistance might be important. There might also be a need for combination therapies targeting various pathways in the disease process. A good example is vitamin E which as been recently demonstrated to show certain efficacy in the treatment of NASH.107 Interestingly, treatment with vitamin E did not affect insulin resistance, suggesting that improvement in NASH may take place independent of interference with insulin resistance. This is of interest because at least certain animal models suggest that presence of insulin resistance might accelerate steatohepatitis and degree of fibrosis.108 Because vitamin E suppresses proinflammatory cytokines and induces adiponectin, regulation of such key mediators in the disease process might be of considerable importance.109 Interference with ER stress might be another treatment option in the future. Chemical chaperones such as ursodeoxycholic acid (UCDA) reduced ER stress and improved metabolic functions in a mouse model of diabetes.110 A beneficial effect for UDCA in the treatment of NASH has not been observed so far in various clinical trials,111, 112 An explanation for these discrepancies could be the fact that in vitro effects were observed using much higher UDCA concentrations than ever achievable in humans.110 Studies using more powerful ER chaperones are eagerly awaited.


Simple hepatic steatosis, which is a benign condition and nonprogressive in the majority of patients, and NASH may reflect different disease entities. Inflammation may precede steatosis in NASH. In contrast, only a small group of patients with simple steatosis might advance toward inflammation and fibrosis. In case of simple steatosis, various protective mechanisms including liver trigylcerides and hyperleptinemia might be operative, thereby protecting the liver from toxic lipid insults. A number of very diverse parallel processes might contribute to the development of liver inflammation. Our model suggests that inflammatory mediators derived from various tissues but especially from the gut and adipose tissue could play a central role in the cascade of inflammation, fibrosis, and finally tumor development. Within the adipose and liver tissue, increased lipid storage, lipogenesis, and (adipo)cytokine synthesis may act as stress signals for the ER. XBP1 might reflect an ideal pathway linking many components observed in this disease. Because a high-fat diet is needed in almost all experimental models to induce pathology, it is evident that dietary factors and nutrient sensing are cornerstones of these diseases.113 Whereas genetic factors overall may play a minor role in the current epidemic of obesity, certain genetic factors might well offer explanations for a more progressive disease course in NAFLD.97 Many of the events discussed here might often take place rather in parallel than consecutively, therefore not allowing the exact dissection of the evolution of steatosis and inflammation. Our concept of “multiple parallel hits” might reflect more precisely current knowledge of this metabolic disease and could explain why this disease might also occur in rather lean subjects.


We gratefully acknowledge Dr. Arthur Kaser, Medical University Innsbruck, Department of Gastroenterology and Hepatology, for very helpful discussions and critical reading of the manuscript.