The hallmark of nonalcoholic fatty liver disease (NAFLD)/nonalcoholic steatohepatitis (NASH) is hepatic steatosis, usually macrovesicular fat. The clinical factors associated with NAFLD/NASH are multiple and ever increasing, including not only obesity/metabolic syndrome, but also drugs (e.g., Amiodarone, Tamoxifen), chemotherapy-associated steatohepatitis, and environmental toxins to name only a few.1–4 Why some patients progress from simple steatosis to more aggressive liver disease is unknown. The prevailing concept is the 2-hit theory in which there is a baseline of steatosis plus one or more second hits (e.g., cytokines, oxidative stress, mitochondrial dysfunction1, 5). A rapidly-evolving concept, which has been much more intensively investigated in other forms of organ injury, is lipotoxicity.6–10 Herein, lipids such as fatty acids are the noxious agents, and triglycerides actually serve as a “sink” or protective pathway in lipid metabolism.6, 7 There are multiple definitions for lipotoxicity; a frequently used one is adverse effects of fatty acid accumulation in nonadipose tissues (Fig. 1).
Many novel strategies are being investigated to prevent/treat NAFLD/NASH, including small molecules that alter lipid metabolism. Previous studies by Yu et al. published in HEPATOLOGY used antisense oligonucleotides to decrease the expression and activity of diacylglycerol acyltransferase (DGAT) 2 (Enzyme Commission number EC 184.108.40.206), which markedly reduced hepatic steatosis in obese (ob/ob) mice.10 Thus, there was some optimism that blocking DGAT2, which catalyzes the final step in hepatic triglyceride production, may be beneficial in NAFLD/NASH. In the study by Yamaguchi et al. in this issue of HEPATOLOGY, a more complex model of steatosis was used, employing db/db mice fed a methionine-restricted, choline-deficient (MCD) diet.11 Animal models for NAFLD/NASH are multiple, with genetic and dietary manipulations being most frequently used. The db/db mouse is an animal model of the metabolic syndrome, having obesity, hyperglycemia, insulin resistance, and hyperleptinemia, and modest increases of hepatic triglycerides. The db/db mouse also has increased intestinal permeability, portal vein endotoxemia, and increased levels of circulating inflammatory cytokines.12 Feeding the MCD diet has been shown to markedly augment steatosis, inflammation, and fibrosis in the db/db mouse.13 Feeding the MCD diet also sensitizes rodents to lipopolysaccharide, resulting in increased tumor necrosis factor production and hepatotoxicity.14 Animals fed MCD diets develop low levels of the critical methylating agent S-adenosylmethionine (SAM) and an altered SAM:S-adenosylhomocysteine ratio, which is associated with decreased activity of most methyltransferases.14 Rats remaining on the MCD diet over the long term can develop severe fibrosis/cirrhosis and even hepatocellular carcinoma. Similarly, mice deficient in the enzyme that converts methionine to SAM in the liver (MAT1A) also develop steatosis, subsequent steatohepatitis, and in the long term, may develop hepatocellular carcinoma.15 Moreover, gut microbiota (through reducing bioavailability of choline) have recently been shown to play a critical role in fatty liver development in a frequently used insulin-resistant mouse strain (129S6).16 Patients with alcoholic and nonalcoholic steatohepatitis often have alterations in hepatic methionine metabolism.1 Thus, the NASH model (db/db mouse fed the MCD diet) used by Yamaguchi and coworkers in this issue is a unique combination of a genetic and dietary model of NASH that may have important mechanistic and clinical relevance.
Feeding db/db mice the MCD diet in the Yamaguchi study caused hepatic steatosis, hepatocyte injury and cell death, oxidative stress, and fibrosis.11 The elevated hepatic triglyceride content decreased with duration of study and with increasing severity of liver injury (similar to human NASH and cryptogenic cirrhosis). Although treatment with DGAT2 antisense oligonucleotides produced an expected reduction in hepatic triglyceride content, antisense-treated animals also developed significantly greater hepatic inflammation, fibrosis, elevated alanine aminotransferase levels, terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling staining, and very prominent increases in cytochrome P450 2E1 and 4-hydroxynonenal. The MCD-fed animals (with antisense therapy) had lower body weights, lower glucose and lower insulin levels, and an improved overall insulin sensitivity profile (disassociating liver injury and insulin sensitivity). Animals receiving antisense therapy had higher adiponectin levels and lower serum free fatty acids. Importantly, hepatic free fatty acids were elevated in MCD antisense-treated animals, which is in contrast to findings of the Yu article10 in which ob/ob obese mice receiving antisense therapy had lower hepatic free fatty acid levels. Thus, the paradigm created by Yamaguchi et al. in this month's HEPATOLOGY likely represents a system of in vivo hepatic lipotoxicity, possibly in conjunction with other hepatotoxic agents such as reactive oxygen species from cytochrome P450 2E1 and injury from 4-hydroxynonenal.
Probably the most plausible inducers of lipotoxicity are the long-chain fatty acids (LCFA), either as free, nonesterified species or as activated forms, such as acyl-coenzyme A (acyl-CoA). Although LCFAs are normal, obligatory intermediates in overall fat metabolism, these lipid species become toxic when their intracellular concentrations exceed a certain level. There is no universal limit for such an increase; rather, the critical levels differ with the cell type and organ, depending upon the cell's overall capacity to metabolize free fatty acids.
There are two main natural sources, both dietary, of increased intracellular concentration of LCFA and/or their CoA derivatives: high-fat or high-carbohydrate diets. Nondietary causes, such as drugs that impair fatty acid oxidation, have also been identified, and they also lead to increased intracellular LCFA and or acyl-CoA levels. Removal of LCFA and/or their CoA derivatives may reduce or reverse their lipotoxic effects. Among the natural ways to decrease the intracellular levels of the LCFA/acyl-CoAs is their esterification to form less toxic or nontoxic compounds, such as triacylglycerols.6 Moreover, inhibiting the pathway of esterification to triacylglycerols may lead to high intracellular levels of LCFA and/or their CoA derivatives, therefore perpetuating their lipotoxic effects. This has been well documented in other tissues or organs. For example, oleic acid and palmitate metabolism in Chinese hamster ovary cells have been extensively studied. Although oleic acid supplementation led to well-tolerated triglyceride accumulation, excess palmitic acid was poorly incorporated into triglyceride and caused apoptosis.6 Unsaturated fatty acids rescued palmitate-induced apoptosis by channeling palmitate into triglyceride pools and away from pathways leading to apoptosis. Moreover, when triglyceride synthesis was inhibited, oleate induced lipotoxicity. The NASH model used by Yamaguchi with DGAT2 antisense intervention likely represents an in vivo model of hepatic lipotoxicity, with inhibited hepatic triglyceride formation, disruption of other pathways of lipid metabolism due to the MCD diet, and hepatic fatty acid accumulation.11
The critical concept highlighted by Yamaguchi et al., is that increased intracellular levels of LCFA and/or their CoA derivatives can be toxic to cells.11 Several mechanisms underlying the LCFA-induced lipotoxicity are mentioned in the Discussion section. Two additional mechanisms include sphingolipid formation in a sequence of reactions that starts with serine:palmitoyl-CoA transferase (EC 220.127.116.11); and protein palmitoylation (N-palmitoyl-CoA transferase; EC 18.104.22.168).17, 18 Sphingolipid formation may lead to increased ceramide generation, which may in part explain the elevated hepatocyte apoptosis seen in the study.17 Protein palmitoylation, although there are very few studies in liver steatosis, may also be of interest because it is necessary for anchoring proteins to various cell membranes.18 A final important concept possibly associated with lipotoxicity is fatty acid-induced inflammation (“lipoinflammation”). It is now clear that certain fatty acids can activate Toll-like receptor 4 (TLR4), cause nuclear factor κB activation, and result in production of proinflammatory cytokines such as tumor necrosis factor.19 It is the lipid A moiety of endotoxin that activates TLR4, and certain common fatty acids also do this. Interestingly, the fatty acids that activate TLR4, such as palmitate, are ones that classically cause lipotoxicity, while fatty acids such as oleate do not activate TLR4. This TLR4 activation has not only been demonstrated in vitro, but in animals in vivo following lipid infusion and feeding high fat diets.
An important issue in most animal and human studies of NASH is an almost exclusive focus on the liver. Whereas the hepatologist deals with hepatic steatosis, the cardiologist studies lipid accumulation in atherosclerotic plaques and the heart, and the endocrinologist examines adipose tissue. Obesity and the metabolic syndrome involve altered energy metabolism and lipid accumulation in many tissues/organs that critically impact one another. Unfortunately, potential therapeutic interventions for the metabolic syndrome may be beneficial for one organ system but possibly deleterious to another. For example, the nutritional supplement, conjugated linoleic acid (CLA), reduces adiposity and is sold in health food stores as a weight loss agent.20 However, its unwanted consequence is hepatic steatosis. CLAs given by gavage to mice caused marked inflammation of adipose tissue and markedly reduced adiponectin levels in less than 1 week. CLAs modulate critical genes in opposite directions in adipose tissue versus liver; the scavenger receptor CD36 (which takes up both fatty acids and endotoxin similar to TLR4) is up-regulated in liver but decreased in adipose tissue.20 On the other hand, the insulin sensitizer rosiglitazone decreases hepatic steatosis but increases body weight and adipose tissue. Investigators from Dr. Belury's laboratory have recently combined these agents in mice fed a high-fat diet to maximize the benefits of each product while minimizing negative effects.20 We predict that future studies of NASH will increasingly evaluate not only the liver but other tissues/organs, such as muscle, adipose tissue, in order to evaluate critical cross-talk between tissues. We may see more combination therapy (similar to the Belury laboratory approach) in the future.20 We must develop a better understanding of lipids (both extracellular and intracellular) and their mechanisms and role in hepatic inflammation, injury, and fibrosis. We also must evaluate effects of lipids on individual cell types in the liver, such as stellate cells. In this study, blocking DGAT2 accelerated fibrosis. Stellate cells exhibit both DGAT1 and DGAT2 activity, and how these enzymes modulate lipid/vitamin A metabolism and hepatic fibrosis are important areas for future investigation. Finally, we likely will find even more pathways leading to NASH, and may have to reconsider how we view hepatic triglyceride (both as a marker in noninvasive imaging and as a “good” fat versus a “bad” fat).