Nonalcoholic steatohepatitis (NASH), the necroinflammatory, profibrogenic form of nonalcoholic fatty liver disease (NAFLD) that leads to cirrhosis, is inextricably related to type 2 diabetes (T2D) and metabolic syndrome.1, 2 These predicate the presence and fibrotic severity of NASH, whereas NAFLD is a risk factor for development of T2D and cardiovascular complications.1-4 The links between NASH, diabetes, and cardiovascular disease are likely to exist because they share common pathogenic factors, a key focus of which is the way the body stores fat.

Overnutrition, the first step in pathogenesis of NAFLD, is caused by excess energy intake for prevailing energy expenditure. Attention has been drawn to physical inactivity, as well as to specific nutrient excesses, such as saturated fat and fructose.1, 2, 5, 6 Surfeit energy is stored as fat, notably triacylglyceride (TG). Adipose tissue is the physiological storage site; the liver is not. Healthy subcutaneous adipose tissue (SAT) is composed mostly of small, insulin-sensitive, differentiated adipocytes that absorb circulating free fatty acids (FFAs) and lipoprotein-bound TG postprandially. They form TG (lipogenesis) and store it in lipid bodies (surrounded by a monolayer of lipase-regulating proteins) until FFAs are needed during fasting.7 They also secrete adiponectin, which by opposing hepatic lipogenesis and stimulating long chain fatty acid beta-oxidation, protects the liver from harmful effects of lipid accumulation, such as insulin resistance (IR).2, 5 In T2D and metabolic syndrome, failure of SAT to store energy efficiently leads to swollen adipocytes that are stressed and de-differentiated (Fig. 1). They continually release FFAs from TG (lipolysis)7 and recruit macrophages. Visceral adipose tissue (VAT) is inherently de-differentiated and inflamed.4 De-differentiation, coupled to recruited macrophages which release tumor necrosis factor-α, suppresses secretion and circulating levels of adiponectin.1, 2

thumbnail image

Figure 1. The respective roles of adipose tissue and liver in generating hepatic FFA accumulation is conceptualized. Healthy subcutaneous adipose tissue (SAT) takes up FFAs and TG as well as de novo synthesis, particularly driven by insulin via SREBP1. Release of FFA from TG is physiologically regulated via acyltriglyceride lipase (ATGL),7 hormone-sensitive lipase (HSL) (suppressed by insulin), and monoacylglyceride lipase (MAGL) and its coregulator, comparative gene identification-58 (CGI-58). The uncertain/ambiguous roles of PNPLA3 and PNPLA5 is indicated by question marks (?). Visceral adipose tissue (VAT), as well as inflamed, IR, and lipid-overloaded SAT continues to release FFA in an apparently unregulated, or dysregulated manner (the role of PNPLA3/PNPLA5 is unclear). FFAs are taken up by the liver in NASH (see text), particularly transported by cluster differentiation protein 36 (CD36) (reviewed in Larter et al.1), contributing to accumulation of FFA if their incorporation into TG (acylation), the proposed predominant action of PNPLA3 (or PNPLA5), is inefficient. It is not known whether, under some conditions, PNPLA3 or PNPLA5 could also act as hepatic lipases and contribute to accumulation of nonesterified fatty acids. FFAs are one possible type of lipotoxic lipid molecules that transform liver pathology from simple steatosis to steatohepatitis (NASH).

Download figure to PowerPoint

In NAFLD, T2D and metabolic syndrome, there are strong correlations between IR, VAT mass, and hepatic TG content.1-5 An early consequence of IR is hyperinsulinemia. In turn, hyperinsulinemia and hyperglycemia program hepatic synthesis of fatty acids by stimulating the transcription factors, sterol regulatory element binding protein-1 (SREBP1) and carbohydrate regulatory element binding protein (ChREBP) (Fig. 1). However, although hepatic TG levels increase up to 10-fold in NAFLD/NASH,1 tracer studies indicate that hepatic lipogenesis accounts for no more than 25% of the total; at least 60% arises from the periphery.8 TG is a storage form of lipid that it is not toxic to liver cells in vitro or in animal models.5, 6 Instead, evidence favors free fatty acids (FFAs) or other lipids (diacylglycerol, toxic phospholipids, cholesterol) as tissue damaging, proinflammatory (lipotoxic) molecules that mediate pathogenesis of NASH.5, 6 However, do these FFA originate locally or from adipose tissue?

Several lines of evidence implicate an inadequate adipose response to lipid storage as important in NASH (see reviews1-6). In patients, the distribution of bodily fat is central (visceral), serum adiponectin levels correlate inversely with steatosis severity/steatohepatitis transition, and therapeutic response to pioglitazone depends on reversal of “adipose-IR”.9 Experimentally, Alms1 mutant (foz/foz) mice fed an atherogenic diet develop IR, diabetes, hypoadiponectinemia, and NASH, but only after adipose stores fail to expand further (adipose restriction).10 In ob/ob mice, development of severe steatosis, diabetes, and dyslipidemia with fall in serum adiponectin is averted by the insertion of an adiponectin transgene, which improved insulin sensitivity and reduced steatosis as TG was “redistributed” back to SAT.11 However, the strongest evidence that an impaired adipose response to overnutrition contributes to NASH pathogenesis has come from the identification of human genetic polymorphisms.

Genes implicated in NAFLD include those affecting bodily lipid distribution, lipoprotein metabolism (e.g., apolipoprotein C312), and adiponectin levels.1,2 Greatest interest has come from genome-wide association studies that identified a single-nucleotide polymorphism, rs738409 (G allele), encoding isoleucine substitution for methionine (I148M) in a gene designated as patatin-like phospholipase A3 (PNPLA3),13, 14 also termed adiponutrin-3. The polymorphism is more common in those of Southern European ancestry.15 It is not associated with higher frequency of obesity or insulin resistance, but among the overweight it correlates closely with central obesity (waist circumference) and hepatic steatosis (mass resonance spectrometry).13 In Dallas, TX, rs738409G accounts for virtually all the ethnic differences in NAFLD frequency, from ∼40% in Hispanics, through ∼30% in Europeans, to ∼20% for African Americans.13 The PNPLA3 polymorphism also correlates with raised serum alanine aminotransferase,15–17 indicating predilection to liver injury in subjects with NAFLD, and it has now been linked to higher rates of NASH,18 and fibrosis with NAFLD and alcoholic liver disease.18, 19

One might anticipate that knowing how PNPLA3 mutation is related to hepatic lipid distribution and liver injury would give profound insights into the pathogenesis of NASH. Unfortunately, information about the location (adipose or liver) and regulated roles of PNPLA3 in TG synthesis and lipolysis remains fragmentary and ambiguous.7, 15, 20 Although predominantly expressed in adipose, it is also present in liver, more so in humans than mice.20 PNPLA3 was discovered in the search for more complete understanding of TG turnover. Earlier attention had focused on hormone-suppressible lipase which catalyzes hydrolysis of diacylglycerol, the second step in TG lipolysis, and mono-acylglyceride lipase, which with its coregulator, comparative gene identification-58, catalyzes the third step.7 The first step is catalyzed by acyltriglyceride lipase (ATGL) (adiponutrin 2).7 The adiponutrins seem to play cooperative roles in both lipolysis and its opposite process of transacylation during TG synthesis.7, 15, 20 PNPLA3 expression is suppressed by fasting and induced by a carbohydrate-rich diet; it may therefore be involved with TG synthesis and storage during times of energy excess. Its strong regulation by insulin (via SREBP1) accords with that function.15, 20 In the early stages of NAFLD pathogenesis, when partial IR activates SREBP1,1 PNPLA3, acting as a transacylation pathway in lipogenesis, could play a role in expanding adipose TG stores, but it is unclear whether this differs between SAT and VAT, or whether defective PNPLA3 would liberate more FFA to be taken up by the liver (Fig. 1). Conversely, if the main function of PNPLA3 is to regulate lipolysis, its inactivity would favor TG accumulation, which is desirable in adipose, but potentially increases TG storage in liver. To date, experimental approaches show that I148M substitution abolishes TG hydrolysis (lipolysis), despite not altering the subcellular distribution of PNPLA3 between membranes and lipid droplets.20 The result is that mutant (I148M) but not wild-type (WT) PNPLA3 increases hepatocellular TG content in vitro and in vivo.20

Given the strong links between a functionally inactive variant of PNPLA3 and NASH, and that pathways of TG formation and lipolysis are highly conserved across species, creation of a Pnpla3 gene-deleted mouse should be useful to study NASH pathogenesis. In this issue of HEPATOLOGY, Chen and colleagues report such a line, produced by gene targeting.21 Hepatic and adipose Pnpla3 expression was abrogated. Loss of Pnpla3 had no effect on body weight, adipose mass or development, insulin sensitivity, or glucose tolerance. Thus, they challenged these animals with three dietary regimes associated with steatosis, or steatohepatitis in the case of methionine and choline deficiency,22 and cross-bred them with ob/ob mice. None of these challenges seemed to worsen the NAFLD disease phenotype in Pnpla3/ versus WT mice. These resoundingly negative results add further mystery to the function of Pnpla3, and seem to challenge its role in NASH pathogenesis.

Among possible explanations that come to mind, the first is that Pnpla3 might not be relevant to liver and adipose TG storage and/or lipolysis in mice, a species difference from humans. The second is that adiponutrins are relevant, but Pnpla5 can substitute for Pnpla3 gene deletion. In the present work, a high-sucrose diet increased hepatic Pnpla3 and Pnpla5 messenger RNA markedly and to a similar extent in WT mice. It did not alter liver or adipose ATGL messenger RNA in Pnpla3/ mice, but there was a disproportionate rise in adipose, not liver,8 Pnpla5 messenger RNA. In vitro experiments failed to show enhanced catecholamine-stimulated adipose lipolysis in Pnpla3 knockouts, but this may not simulate the role of Pnpla3 or Pnpla5 for basal lipolysis in animals with obesity and IR, or exclude a role for transacylation in protection against NASH. It therefore remains possible that redundancy in this metabolic pathway is why Pnpla3/ mice failed to recapitulate the NASH phenotype. Pnpla3.Pnpla5 double knockout and tissue-specific gene deletion experiments will be of interest. It is also possible that gene deletion may not be equivalent to a “dominant-negative” effect of gene mutation; the variant protein remained normally distributed between membranes and lipid droplets,20 and might still interact physically with other regulators of lipogenesis and lipolysis to displace alternative pathways that could be activated in response to gene deletion. More basic studies into the regulation of TG turnover in both adipose and liver are required before data from one knockout line can be fully interpreted.

The other key consideration is that the experimental models used in this work, despite their popularity, may have failed to recapitulate the essential preconditions for NASH pathogenesis: overeating, dietary factors, under-activity, visceral adiposity, and IR.1, 22 PNPLA3 polymorphisms cause neither obesity nor IR. Rather, they come into play to increase hepatic lipid partitioning and liver injury/fibrosis among individuals who are overweight and have IR.15-17 Dietary animal models have major limitations for understanding NASH.22 Most develop only modest weight gain, and, as in this study, relatively minor increases in hepatic TG content to cause simple steatosis, not NASH. The methionine- and choline-deficient diet does cause steatohepatitis, but is associated with weight loss, increased insulin sensitivity, and serum adiponectin, the opposite metabolic changes that cause NASH.22 In ob/ob.Pnpla3 double knockout mice, effects of leptin deficiency on inflammatory recruitment and hepatic fibrosis (not assessed by Chen et al.) may have negated the effects of altered hepatic lipid turnover, or the experimental duration may have been too short. Studies employing metabolic models of NASH (hyper-alimentation, foz/foz, apolipoprotein E−/− mice),22 extending over 24-36 weeks, are required before dispelling a role for Pnpla3 inactivation in NASH.

Meanwhile, the link between PNPLA3 polymorphisms and predisposition to fatty liver during overnutrition is too strong to disregard,13-17 as are the burgeoning data that support its links to severity of NASH and alcoholic cirrhosis.16-19 Identifying how this one genetic factor interacts with others1, 2, 12 and with environmental factors to confer a high risk of NASH will be a key step toward understanding the causation of metabolic forms of fatty liver disease.


  1. Top of page
  2. Acknowledgements
  3. References

The author gratefully acknowledges the assistance of Derrick van Rooyen and Betty Rooney in preparation of this manuscript.


  1. Top of page
  2. Acknowledgements
  3. References