Hepatic lipotoxicity and the pathogenesis of nonalcoholic steatohepatitis: The central role of nontriglyceride fatty acid metabolites


  • Brent A. Neuschwander-Tetri

    Corresponding author
    1. Division of Gastroenterology and Hepatology, Saint Louis University, St. Louis, MO
    • Division of Gastroenterology and Hepatology, Saint Louis University, 3635 Vista Avenue, St. Louis, MO 63110
    Search for more papers by this author
    • fax: 314-577-8125

  • Potential conflict of interest: Dr. Neuschwander-Tetri is a consultant for Astellas, Gilead, and Centocor.

“…the cause of steatosis, and not the fat accumulation by itself, produces cirrhosis”—Heribert Thaler, 19751

W ith nonalcoholic steatohepatitis (NASH), NASH cirrhosis, and NASH-related hepatocellular carcinoma becoming increasingly prevalent, the need for effective therapies has never been greater. Unfortunately, our understanding of what causes NASH at the molecular level remains mostly speculative, and thus our ability to design clinical trials that test rationally designed therapies is limited. More than a decade has passed since Christopher Day and Oliver James first proposed the oft-cited two-hit hypothesis of nonalcoholic fatty liver disease (NAFLD) to explain the pathogenesis of NASH.2, 3 According to this appealing hypothesis, the accumulation of lipid in the form of triglyceride is needed for the development of NASH and thus constitutes the first “hit” in this disease. The cause of injury in the setting of lipid-loaded hepatocytes was proposed to be oxidant stress leading to lipid peroxidation in the milieu of ample substrate. This second “hit” then triggers the necroinflammatory changes that we recognize histopathologically as NASH.

Although this hypothesis has intuitive appeal, emerging data now suggests that in the liver, as in other organs, triglyceride accumulation in the form of lipid droplets truly is just an “innocent bystander” in the processes leading to cellular injury and inflammation,4, 5 an explanation presciently suggested by the eminent Austrian pathologist Heribert Thaler,1 considered by others,6 and acknowledged as a viable alternative by Day and James when they proposed the two-hit hypothesis.3 Convincing evidence for a central role of oxidant stress and lipid peroxidation in causing steatohepatitis has also not been established. A decade of research has confirmed that these processes occur,7 but studies have been unable to prove that oxidant stress or lipid peroxidation are necessary for the development of steatohepatitis in humans.

As an alternative hypothesis, emerging evidence now points to metabolites of fatty acids as the real culprits in the hepatocellular injury in NASH, just as they are in other target organs of lipotoxicity (Fig. 1).8-12 Because triglyceride often accumulates as a parallel process during lipotoxic injury, it has been understandably difficult to separate the respective roles of triglyceride and fatty acid metabolites in causing injury and cell death. As with all good hypotheses, the nontriglyceride lipotoxicity hypothesis generates more questions than we can currently answer. Fortunately, it also provides new avenues for investigators to pursue in the search for effective treatments for this common disease.

Figure 1.

The lipotoxicity model of NASH. Emerging data indicate that metabolites of free fatty acids cause lipotoxic hepatocellular injury manifested as ER stress, inflammation, apoptosis, necrosis, and dysmorphic features such as ballooning and Mallory-Denk body formation. The generation of lipotoxic metabolites of fatty acids typically occurs in parallel with the accumulation of triglyceride droplets (steatosis), resulting in a phenotype recognized as NASH where steatosis and features of cellular injury are present together. A feature of this model that distinguishes it from earlier models of the pathogenesis of NASH is that the accumulation of triglyceride is not needed for the development of NASH and, in fact, it may be protective. A caveat is that steatosis does not “progress” to NASH. Metabolic abnormalities predisposing to lipotoxic injury include an increased supply or impaired disposal of free fatty acids. Insulin resistance plays a central role in these processes by allowing an excessive flow of fatty acids from adipose tissue and also impairing peripheral glucose disposal. Fatty acid disposal in the liver occurs through oxidative pathways and through the formation of triglyceride which is then stored temporarily as lipid droplets or secreted as VLDL. Supply-side salutary processes include reduction of carbohydrate precursors for de novo lipogenesis and prevention of inappropriate lipolysis by improving adipocyte insulin responsiveness. Favorable processes on the disposal side include increasing oxidative pathways (with the caveat that the intracellular antioxidant defenses must be adequate to handle the ROS produced) and increasing formation of triglyceride. Red arrows identify processes that contribute to the flux of fatty acids through hepatocytes, thus promoting lipotoxic injury; green arrows identify processes that eliminate fatty acids and thus reduce lipotoxic injury. Not shown are the mechanistic aspects of injury, repair, and fibrogenesis that ultimately determine whether lipotoxic liver injury leads to cirrhosis. ER, endoplasmic reticulum; IR, insulin resistance; P450, cytochrome P450 mixed function oxidases; ROS, reactive oxygen species; SER, smooth endoplasmic reticulum; VLCFA, very long chain fatty acids; VLDL, very low density lipoprotein.


CoA, coenzyme A; CYP, cytochrome P450; DAG, diacylglycerol; DGAT2, diacylglycerol acyltransferase 2; DNL, de novo lipogenesis; ER, endoplasmic reticulum; FABP, fatty acid binding proteins; JNK, c-Jun N-terminal kinase; LPC, lysophosphatidyl choline; NAFLD, nonalcoholic fatty liver disease; NASH, nonalcoholic steatohepatitis; PPAR, peroxisome proliferator-activated receptor; ROS, reactive oxygen species; SCD-1, stearoyl-CoA desaturase-1; SREBP-1c, sterol regulatory element binding protein 1c; VLDL, very low density lipoprotein.


Lipotoxicity is a term coined by Unger 15 years ago to describe the toxic effects of excessive free fatty acids on pancreatic beta cell survival.13 The term has never been formally defined but has been used variably to describe cellular injury and death caused by free fatty acids, their metabolites, and even triglyceride, although triglyceride is generally found to be relatively inert.8, 14 Because a wide variety of metabolites can be derived from fatty acids having varying chain lengths and varying degrees and locations of desaturation, not to mention cis versus trans configuration of double bonds, the number of permutations of fatty acid metabolites is enormous, making the identification of lipotoxic species challenging.15 Identifying important fatty acid derivatives remains an area of intense lipidomic study to understand lipotoxic diseases of the heart, pancreatic islets, and vasculature, as well as the liver.10, 12, 16-21 Shown in Fig. 2 are a few of the many putative mediators of nontriglyceride lipotoxicity. We should anticipate that as lipidomic studies advance and new metabolites are discovered, this list will appear quite naïve in retrospect.

Figure 2.

Possible fatty acid–derived mediators of lipotoxicity. Currently available data suggest that fatty acid metabolites participate in lipotoxic liver injury recognized as NASH. The reintroduction of diacylglycerol from triglyceride may play an important role in regulating the exposure of cells to these intermediaries and is highly regulated by a family of proteins.41, 67, 125 The tissue specificity and regulation of these pathways has been reviewed.66, 69, 70, 72 ACSL, acyl-CoA synthase; AGPAT, acyl-glycerolphosphate acyltransferase; ATGL, adipose triglyceride lipase; CPT, choline phosphotransferase; DAGK, diacylglycerol kinase; DGATs, diacylglycerol acyltransferases; GPAT, glycerol monophosphate acyltransferase; HSL, hormone-sensitive lipase; LPAAT, lysophosphatidic acid acyltransferase; LPAP, lysophosphatidic acid phosphatase; LysoPLD, lysophospholipase D; MAGK, monoacylglycerol kinase; MGL, monoacylglycerol lipase; MOGAT, monoacylglycerol acyltransferase; PLA, phospholipase A; PLD, phospholipase D.

Possible Molecular Mediators of Lipotoxicity


The accumulation of triglyceride as lipid droplets was once thought to be the underlying cause of insulin resistance in muscle, liver, and other tissues, but data now indicate that accumulation of lipid droplets is a parallel phenomenon and not the cause of altered insulin signaling pathways.12, 14, 22-24 Similarly, the data implicating triglyceride stored within hepatocyte lipid droplets as a cause of liver injury in human studies and animal models mostly demonstrate only an association between the two phenomena, rather than showing causality. However, triglyceride accumulation is probably not totally benign. When extreme, some have suggested that triglyceride might compromise hepatic blood flow simply by compression of the sinusoids during hepatocyte necrosis,25 although this has been debated.26 Excessive triglyceride does render donor livers susceptible to injury during ischemia and reperfusion during transplantation. The accumulation of lipid droplets has been associated with increased markers of endoplasmic reticulum stress, although the role of this cellular response in NASH is uncertain.27 Mice overexpressing diacylglycerol acyltransferase 2 (DGAT2), the final enzyme catalyzing triglyceride formation, have increased triglyceride accumulation and hepatic phosphorylation of PERK (protein kinase R–like endoplasmic reticulum kinase), a marker of endoplasmic reticulum stress. However, these mice did not exhibit increases in the proinflammatory pathways of c-Jun N-terminal kinase (JNK) phosphorylation and nuclear factor-κB activation, markers that are typically elevated in insulin resistance and steatohepatitis.27, 28 Examined in the opposite way, blocking DGAT2 expression prevented fatty acid–induced triglyceride accumulation but did not prevent lipotoxic injury caused by fatty acid exposure both in cell culture and in mice.22, 29 Overall, the experiments that manipulate DGAT2 to discern the role of triglyceride versus fatty acid–derived triglyceride precursors have been relatively consistent in demonstrating nontriglyceride lipotoxic injury. However, there has been some conflicting data, possibly because of compensatory up-regulation of oxidative disposal pathways under various experimental conditions that can reduce the burden of fatty acids and confound the interpretation of results.30-32 However, additional evidence has also dissociated lipid droplet accumulation from injury. For example, inhibiting triglyceride incorporation into nascent very low density lipoprotein (VLDL) by blocking the microsomal triglyceride transfer protein caused impaired secretion of triglyceride, leading to triglyceride accumulation, yet did not appear to cause liver injury.33

A view has emerged from recent studies that the formation of lipid droplets may actually be a protective response that prevents lipotoxicity from other fatty acid–derived species.4, 5, 10, 12, 34 Early studies of lipotoxicity demonstrated this in cardiomyocytes,35 and subsequently, a protective effect of triglyceride droplet formation has been demonstrated in cultured pancreatic islet cells36 and hepatocytes.37 However, storage of triglyceride as fat droplets is only a temporizing measure; the fatty acids in triglyceride must be released at some point, and if the cell is still unable to handle them appropriately through other metabolic pathways, the stored triglyceride could still serve as a source of lipotoxic intermediates.38

Free Fatty Acids.

Free fatty acids are highly noxious to biological systems. As a protective mechanism, circulating free fatty acids are bound with high affinity to albumin, and intracellular fatty acids are bound to fatty acid binding proteins (FABPs) within cells that have mechanisms for fatty acid uptake (e.g., hepatocytes, adipocytes, and enterocytes). Measurement of fatty acid concentrations in tissues does not provide information about the net flux through the system, but in patients with NASH, the flux of fatty acids through the liver is increased.39-41 Because fatty acids are tightly bound to FABPs, the levels are mostly a function of FABP concentrations and not reflective of how much is flowing through the system. Studies have confirmed that the intracellular fatty acid concentration is fairly constant in the liver,17 even with increased levels in the blood as seen with NASH,42 suggesting that elevations of fatty acid concentrations in hepatocytes are unlikely to be associated with lipotoxicity, whereas the increased delivery of fatty acids to enzymes and organelles that process them is markedly increased.14 On the other hand, overexpression of FABP in cardiac cells promotes lipotoxicity,43 and deletion of liver FABP prevented steatosis in the liver and liver cell lines,44, 45 observations that confirm the role of this delivery system in facilitating lipotoxicity. FABPs serve as the intracellular conduit for delivery of fatty acids to enzymes that use them as substrates, and the bigger the conduit (i.e., higher FABP concentrations) the more can be delivered. A recent study demonstrated decreased immunostaining of FABP in the livers of patients with NASH compared to those with fatty liver but no steatohepatitis, the significance of which remains uncertain.46

Because fatty acids can be shuttled off to so many metabolic pathways, it has been difficult to prove whether free fatty acids or their metabolites are responsible for the cellular injury seen in lipotoxicity.38 Saturated fatty acids can serve as activating ligands for toll-like receptor-4, leading to a cascade of events precipitating apoptosis.47 One study demonstrated that inhibition or loss of function of toll-like receptor-4 can prevent steatohepatitis in mouse models.48 Fatty acids are also capable of directly destabilizing lysosomal membranes, leading to inappropriate release of cathepsin B and causing activation of apoptotic pathways.49, 50 As ligands for peroxisome proliferator-activated receptor alpha (PPARα), PPARγ, and G protein–coupled receptors, fatty acids also mediate diverse metabolic effects that could indirectly play protective and causative roles in lipotoxicity.11 Activation of PPARα and PPARγ serves as an adaptive and protective response against lipotoxicity by promoting the disposal of fatty acids through oxidative and storage pathways, respectively.

Despite data suggesting that fatty acids could directly cause lipotoxic injury, other studies have shown that the formation of acyl-coenzyme A (acyl-CoA) is a requisite step in the development of fatty acid–induced lipotoxicity. Experiments in Chang liver cells and pancreatic β-cells have shown that blocking acyl-CoA synthesis with triacsin C prevented lipotoxicity caused by saturated fatty acids.51, 52 Therefore, other metabolites of fatty acids need to be considered as more likely causes of lipotoxic liver injury.


Ceramides are fatty acid metabolites that were initially thought to play a major role in lipotoxic cellular injury53 and hepatic levels are increased in fatty liver disease.16, 54 Ceramide lipotoxicity has been shown in pancreatic islet cells,55 but several studies demonstrated that blocking ceramide synthesis in other cell types did not diminish fatty acid–induced injury. For example, blocking the final step of ceramide synthesis with the fumonisin B1 did not prevent palmitate lipotoxicity in liver-derived cell lines,51, 56 and ceramides did not seem to be involved in palmitate-induced lipotoxic cell death in Chinese hamster ovary cells despite their levels being elevated.57 Whether these results will be relevant to human liver disease remains to be shown, but studies are now pointing to other lipid species as more likely candidates.


Diacylglycerol (DAG) species are well-known activating ligands of most protein kinase C isoforms, a group of kinases essential for diverse cellular signaling pathways. Thus, the synthesis of DAG species is carefully regulated to coordinate the functions of DAGs in signal transduction and as substrates for the formation of triglyceride and phospholipids.58 Increases in DAG species have been suspected as a major contributor to hepatic lipotoxicity;34, 59 however, lipidomic analysis of liver biopsy specimens from patients with NAFLD with or without steatohepatitis found similar DAG levels, although both had higher levels than was found in normal liver.17 Thus, the increased DAG levels may reflect only increased trafficking of fatty acids through the liver, but may not be causally related to the cellular injury of steatohepatitis. Further studies of the role of DAGs in promoting steatohepatitis are needed to further clarify their role in lipotoxic liver injury.

Lysophosphatidyl Choline.

Phosphatidyl choline, or lecithin, is a major component of cell membrane bilayers and lipid droplet envelop monolayers, and it is a necessary component of VLDL required for its formation and secretion. Removal of one fatty acid from the sn-2 position (the middle position on the glycerol backbone) generates lysophosphatidyl choline (LPC), a biologically active molecule that has been implicated in direct cellular injury as well as monocyte activation.60 One recent study found that LPC could be an important mediator of hepatic lipotoxicity in NASH.51 In this study, LPC levels in liver biopsies from a small group of patients with NAFLD were found to be increased and levels were elevated in proportion to disease severity. A direct role of LPC in lipotoxic liver injury was then demonstrated using human liver cell lines, mouse hepatocytes in primary culture, and in mice. Mechanistically, the LPC-induced lipotoxic injury appeared to depend on an unidentified G protein–coupled receptor with induction of apoptosis through mitochondrial membrane depolarization. LPC species can be generated through the action of phospholipase A2 on lecithin, and inhibitors of phospholipase A2 were shown to prevent palmitate-induced lipotoxicity in cell culture. Although this important study needs validation, it provides provocative evidence for the identity of one potential mediator of fatty acid–induced hepatic lipotoxicity.

Other Potential Mediators.

A wide variety of other biologically active metabolites of free fatty acids, both known and unknown, could play a role in the lipotoxicity caused by excessive exposure to free fatty acids. For example, lysophosphatidic acid species are biologically active molecules that act through a family of specific G protein–coupled receptors, and phosphatidic acid species promote membrane bending or folding, and they also interact with a large number of intracellular signaling molecules.

Factors Predisposing to Production of Lipotoxic Intermediates and NASH

When homeostatic mechanisms regulating fat metabolism are running smoothly, a process termed “liporegulation”,53 production of lipotoxic metabolites from free fatty acids is minimized. However, metabolic or genetic stresses that alter this homeostasis by increasing the flux of fatty acids through the liver, either through increased supply or impaired disposal of precursor fatty acids, promote lipotoxicity. Up-regulation of the enzymatic pathways that generate lipotoxic intermediates from fatty acids or inhibition of pathways that dispose of them could also be speculated to promote lipotoxicity. Because the understanding of lipotoxicity is just emerging and the major mediators at the molecular level remain unknown, there are currently no data that inform us of the role of such hypothetical pathways in lipotoxicity. Stresses on liporegulation that promote lipotoxic liver injury have been characterized primarily in cell culture and animal models, but corroborating evidence in human disease is limited. The findings of seminal studies are briefly reviewed below.

The major mechanisms of hepatic fatty acid exposure and associated triglyceride accumulation and disposal have been known for more than 4 decades.61 The sources of hepatic free fatty acids are primarily adipocyte triglyceride lipolysis with release of free fatty acids into the blood, or hepatocyte de novo lipogenesis (DNL), the formation of new fatty acids from excess carbohydrates and amino acids as substrates (Fig. 1).62 Adipose tissue lipolysis is the major source of fatty acids delivered to the liver, whereas DNL normally contributes only about 5%; in patients with NASH, DNL may contribute up to 26% of the fatty acids.40 Relatively minor contributors to the hepatic free fatty acid pool include uptake of short chain fatty acids delivered directly from the gut and lysosomal breakdown of lipoprotein remnants and autophagosomes.63 Except for uptake of the limited residual triglyceride in lipoprotein remnants, circulating triglycerides are not a direct source of fat for the liver. This is important when designing treatments for NASH, because efforts to reduce serum triglyceride levels pharmacologically might not have a significant direct benefit on the liver.31, 33

Excessive or Inappropriate Peripheral Lipolysis.

The role of the supply side of fatty acid delivery to the liver, either through excessive peripheral lipolysis or excessive synthesis, in the development of NASH has been well described.62, 64 Enzymatic hydrolysis of triglyceride stored in adipose tissue with the release of fatty acids into the circulation is regulated by the interplay of hormonal, neurologic, and pharmacologic stimuli and is much more complex than simple regulation of the enzyme hormone-sensitive lipase, as was once thought (Table 1).65 The process is mediated by several neutral lipases, lipid droplet proteins, and regulators of adipocyte cyclic adenosine monophosphate levels.66-71 Insulin is a major inhibitor of the lipolytic release of fatty acids from adipose tissue triglyceride and, as discussed further below, insulin resistance at the level of adipose tissue is largely responsible for inappropriate lipolysis that leads to lipotoxic diseases.

Table 1. Regulators of Adipose Tissue Lipolysis that Impact the Liver
Effect on LipolysisMechanismReference
  1. ACTH, adrenocorticotropic hormone; cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; EP3, prostaglandin E receptor-3; GPCR, G protein–coupled receptor; HSL, hormone sensitive lipase; PGE, prostaglandin E; PLA2, phospholipase A2; Plin, perilipin; PPAR, peroxisomal proliferator activated receptor.

Decrease lipolysis  
 Insulin1. Akt/protein kinase B (PKB) activation of phosphodiesterase E3, reduction of cAMP67
2. Activation of protein phosphatase 1 with dephosphorylation of hormone-sensitive lipase
 Adiponutrin (product of the gene PNPLA3)Promotes triglyceride synthesis and storage183, 184
 PPARγPromotes expression of lipid droplet proteins, stabilizes droplets185
 PGE2Through the EP3 GPCR186
 Adipocyte PLA2Release of arachidonic acid for PGE2 synthesis73, 186
 AdenosineAdenosine receptor–mediated inhibition of adenylate cyclase187
 Phosphodiesterase E3See insulin above 
 Nicotinic acidSpecific GPCR-mediated inhibition of adenylate cyclase188
Increase lipolysis  
 CatecholaminesIncreased cAMP via adrenergic receptors (alpha in humans, beta in rodents)66
 GlucagonIncreased adenylyl cyclase activity189
 ACTHIncreased cAMP via GPCR 
 Methylxanthines (e.g., caffeine)Increased cAMP via phosphodiesterase inhibition and increased catecholamines67, 190
 Natriuretic peptides (atrial and brain natriuretic peptides [ANP, BNP])Increased cGMP, activation of PKG; induced by exercise191
 cAMPActivation of protein kinase A192
 Protein kinase AActivation of hormone-sensitive lipase69
Phosphorylation of Plin1193
 Adipocyte triglyceride lipase (ATGL), desnutrin (product of the gene PNPLA2)Lipases that have a role in fatty acid release from triglyceride droplets; initiates lipolysis by removal of the first fatty acid66
 HSLMajor triglyceride lipase; necessary for release of fatty acids from DAGs69
 CGI-58 (ABHD5)Needed for ATGL activity194
 PerilipinsActivates ATGL66
 TNFαJNK activation, multiple other effects195
 c-Jun N-terminal kinase (JNK)Inhibition if insulin signaling by serine phosphorylation of insulin receptor substrate-1140

When fatty acids are released from adipose tissue, they can be re-esterified in the liver to return as VLDL triglyceride to the adipocytes; this seemingly futile cycling has been proposed to play a role in thermogenesis because it consumes adenosine triphosphate.72 Although the prevailing view has been that fatty acids liberated by adipocyte lipases are simply released and transported to other tissues, newer data now also suggest that oxidative pathways within adipocytes might also be an important fate, playing a role in energy wasting and thermogenesis,73 although this continues to be debated.70 Bile acids can also interact directly with adipose tissue to regulate thermogenesis by adipose tissue. Acting through the receptor TGR5 to increase cyclic adenosine monophosphate levels, bile acids activate thyroid hormone locally to promote energy wasting and thermogenesis.74, 75

The impact of peripheral lipolysis on the liver can be modulated by muscle fatty acid uptake. It is well recognized that unbridled adipocyte lipolysis, such as occurs during prolonged fasting, can send more fatty acids to the liver than the liver can handle.76, 77 Older studies demonstrated the adverse impact of prolonged fasting on accelerating liver injury,78, 79 but newer studies have also shown that the effect of such peripheral lipolysis on the liver can be attenuated by compensatory oxidative metabolism by muscle. Muscle-specific deletion of lipoprotein lipase spares muscle from the effects of excessive fatty acid uptake, but at the expense of deleterious effects of fatty acids on other organs.80 Also, most mouse strains will develop fasting-induced hepatic steatosis, but mice with impaired mitochondrial biogenesis are particularly prone to fasting-induced liver steatosis,81 and the SJL/J mouse strain that has a significant compensatory increase in muscle fat oxidation tolerates fasting without developing hepatic steatosis.82 Manipulating muscle fatty acid oxidation or recognizing genetic defects in humans that might even predispose to sedentary behaviors and NAFLD are certainly areas that need exploration in clinical trials, especially with respect to exercise capacity and the common assumption that exercise would benefit all patients equally.

Excessive De Novo Lipogenesis.

The de novo synthesis of fatty acids is a major fate of carbohydrate presented to the liver from excess dietary intake and impaired peripheral glucose disposal. Dietary habits that include excessive carbohydrate consumption, especially in the form of high fructose corn syrup,83, 84 may be common causes of overwhelming the supply side of the liver's fatty acid burden through DNL. Fructose enters the glycolytic pathway to fatty acid synthesis downstream from phosphofructokinase, the rate-limiting enzyme for this pathway, and thus fructose provides an unregulated source of acetyl-CoA as the substrate for DNL.84, 85 Studies in mice have confirmed the importance of dietary sugars in precipitating steatohepatitis,86 and a recent cross-sectional study in humans has now linked fructose consumption with more advanced NASH.87

To efficiently incorporate newly synthesized fatty acids into triglyceride, a fraction of synthesized fatty acids must undergo desaturation to monounsaturated fatty acids because the middle position of the glycerol backbone is optimally esterified with an unsaturated fatty acid. The enzyme stearoyl-CoA desaturase-1 (SCD-1) effects this desaturation and is essential to the efficient formation of triglyceride in the absence of adequate dietary unsaturated fatty acids. Recent studies have shown that inhibitors of SCD-1 diminish triglyceride accumulation, but at the expense of increased lipotoxicity.23 Similarly, forcing excessive DNL with a diet high in sucrose but deficient in unsaturated fats caused severe liver injury but no triglyceride accumulation in mice lacking SCD-1.88 Such studies add to the evidence that triglyceride formation is a protective response rather than a cause of injury. Previous mouse studies had shown that inhibiting SCD-1 is often accompanied by increases in compensatory oxidative pathways,89, 90 indicating that as long as alternative pathways of fatty acid disposal are available, lipotoxic injury can be averted.91, 92

De novo lipogenesis can also be amplified by increases in sterol regulatory element binding protein 1c (SREBP-1c), the transcription factor largely responsible for inducing the expression of enzymes responsible for DNL. Overexpression of SREBP-1c in animal models has been identified as a contributor to lipotoxic liver injury.93, 94 SREBP-1c is physiologically down-regulated by bile acids acting through the nuclear receptor farnesoid X receptor, a potential rationale for pursuing farnesoid X receptor ligands as a means of thwarting inappropriate lipogenesis and preventing NASH.95, 96

Lysosomal Catabolism as a Source of Fatty Acids.

Uptake of lipoprotein remnants such as chylomicron and VLDL remnants with intracellular release of fatty acids from the residual triglyceride by lysosomal lipases is a relatively minor contributor to the fatty acid flux through the liver. Recently recognized and possibly quite significant by comparison is the role of autophagy in releasing fatty acids from triglyceride within hepatocyte lipid droplets. Termed lipophagy, this process could play a major role in the turnover of hepatocyte lipid droplets.63, 97 Autophagosome lipolysis is also mediated by lysosomal lipases, and inhibition of these lipases as a way of reducing the generation of fatty acids has been investigated. However, understanding the effects of inhibiting intracellular lipases is complicated by the difficulties in knowing if the effects of drugs or genetic manipulations are affecting the liver, adipose tissue, or both. Inhibiting lysosomal lipases with 18 β-glycyrrhetinic acid prevented NAFLD in lard-fed rats and in HepG2 cells, the latter finding suggesting that inhibition at the level of the liver could play a role in this response.98

Impaired Disposal: Oxidative Pathways.

An increased supply of fatty acids might not be a problem if disposal pathways are robust. One means of disposal is to oxidize fat to carbon dioxide and water through three different metabolic pathways. As an important caveat, endogenous antioxidant mechanisms must be able to handle reactive oxygen species (ROS) such as superoxide and hydrogen peroxide that are inevitably generated in small amounts as electrons are transferred from carbon bonds to oxygen as fatty acids are oxidized. Not coincidentally, the liver is well-endowed with antioxidants. Glutathione is present in high concentrations in hepatocytes and hepatocyte mitochondria; glutathione eliminates hydrogen peroxide and lipid peroxides via glutathione peroxidase and is a substrate for the abundant glutathione S-transferases to eliminate reactive molecules generated during oxidative stress. Substantial depletion of hepatic glutathione has not been reliably shown to occur in NASH,99, 100 suggesting that under most circumstances, this cytoprotective mechanism is capable of handling the load of ROS known to occur during increased oxidative metabolism of fatty acids.7, 101-104 However, mitochondria cannot make glutathione, and studies have shown that this sensitive pool can be depleted, resulting in apoptotic cell death and steatohepatitis.105, 106

After meals, fatty acids delivered to the liver are not needed as a source of energy and are shuttled away from mitochondrial β-oxidation to the formation of triglyceride. By comparison, in the fasting state, fatty acids are used as a source of energy through mitochondrial β-oxidation to generate adenosine triphosphate and ketone bodies. Partitioning fatty acids to mitochondrial oxidative pathways versus triglyceride formation is tightly controlled. In pancreatic β-cells, high glucose levels shut off β-oxidation and predispose to fatty acid induced lipotoxicity.52 In the liver, β-oxidation is physiologically decreased after feeding but if peripheral lipolysis in insulin-resistant adipose tissue continues to send fatty acids to the liver, then this normal metabolic switch puts the disposal pathways at a disadvantage and may predispose to lipotoxicity. Deleting acetyl-CoA carboxylase 2, the gatekeeper to β-oxidation, in mice prevented diet-induced NAFLD, suggesting a central role of mitochondrial β-oxidation in preventing lipotoxic liver injury.107

Mitochondrial dysfunction, whether genetic or acquired, is often associated with steatohepatitis.64, 108-111 Whether the steatohepatitis that develops under these circumstances is due solely to altered metabolism in the liver or is the consequence of mitochondrial dysfunction elsewhere such as muscle, adipose tissue, or other peripheral tissues has not been established. Rats selected for poor aerobic capacity have been shown to exhibit impaired muscle mitochondrial function and are prone to NAFLD.112, 113 The possibility that genetic or acquired mitochondrial defects predispose to both exercise intolerance and NAFLD in humans requires further investigation.

Peroxisomal β-oxidation is a second route of oxidative fatty acid β-oxidation that could be adaptive when mitochondria are dysfunctional.113 The role of peroxisomal fatty acid oxidation in NASH was difficult to understand in the context of the prior paradigm, because the associated production of ROS was given a central role in causing NASH. Yet, loss of the peroxisomal pathway in rodents seemed to cause NASH, and its stimulation was found to be protective.114-118 Unlike mitochondrial β-oxidation in which oxygen is fully reduced to water, peroxisomal β-oxidation enzymes transfer only two electrons to oxygen to generate hydrogen peroxide. Peroxisomes are well endowed with catalase to handle this, however, and ROS that escape catalase are likely eliminated by cytosolic antioxidants such as glutathione peroxidase. Pharmacologically up-regulating peroxisomal pathways as a treatment for human NASH has met with mixed results. Bezafibrate, a PPARα inducer available in Japan, has been shown to be beneficial in limited human studies.119, 120 However, studies of other fibrates have not been fruitful in clinical trials, perhaps because of differences in the drugs or the significantly greater role of PPARα in rodents compared to humans in fatty acid metabolism121, 122 and the down-regulation of PPARα shown in human NASH.123

The third oxidative pathway, via smooth ER cytochrome P450 (CYP) enzymes, oxidizes a diverse array of lipophilic compounds including free fatty acids. This system is capable of handling unusual fatty acids such as branched chain molecules and fatty acids with odd numbers of carbons or unusual placement of double bonds. Omega oxidation of fatty acids by CYP4A isoforms leads to dicarboxylic acids that are further metabolized by peroxisomes.124 Up-regulation of various CYP isoforms has been documented in NASH. Although initially thought to be a source of pathogenic ROS generation, further experimental work has not found this to be the case.

All of the oxidative pathways, including mitochondrial, peroxisomal, and CYP4A-mediated pathways, are strongly regulated by the PPARα nuclear receptor. PPARα thus serves as a sensor for a broad array of lipophilic species and, as a heterodimer with the retinoid X receptor, induces this multipronged approach to fatty acid oxidation. In contrast, the nuclear receptor PPARγ is also a sensor for lipophilic species but promotes the formation and export or storage of fatty acids as triglyceride.

Impaired Disposal: Triglyceride Formation and Secretion.

The accumulation of triglyceride as lipid droplets in NASH is orchestrated by a complex array of facilitating proteins125 and factors that impair this process, predisposing to lipotoxic injury and defects in insulin signaling in cell culture23, 28, 126 and in animal studies.22, 54, 127, 128 Corroborating human data has been difficult to obtain. Impaired secretion of VLDL, the basis of abetalipoproteinemia and hypobetalipoproteinemia, is not consistently associated with liver disease, suggesting that additional defects may be necessary for liver disease to develop in this disease.33, 129 This was also suggested to be the case in the primate Suncus murinus, which manifests abetalipoproteinemia and fatty liver without steatohepatitis when fasted.130

Insulin Resistance and Lipotoxicity

Multiple studies have implicated insulin resistance in the pathogenesis of NASH.6, 64, 131-135 As our understanding of insulin resistance has expanded, questions have arisen regarding which insulin-resistant tissues are most important for the development of NASH. In our hepatocentric way of thinking, a common assumption has been that because NASH is a disease of the liver, insulin resistance in the liver must be pathogenic. However, liver-specific deletion of insulin receptors fails to cause NAFLD or lipotoxic liver injury in animals.136 Because insulin normally up-regulates lipogenic pathways in the liver, one could even postulate that hepatic insulin resistance could be an adaptive protective response that blunts excessive de novo lipogenesis from abundant carbohydrates.

Human data demonstrate that insulin resistance at the level of adipose tissue is the critical location for the development of lipotoxic diseases65 including NASH.10, 64, 135, 137 Studies have further shown that adipose tissue insulin resistance may be a major target of thiazolidinedione therapy.137-139 Activation of JNK in adipose tissue may be a major mediator of adipose tissue insulin resistance, and ligand-induced activation of the PPARγ nuclear receptor or overexpressing PPARγ in adipocytes have been shown to reduce inappropriate lipolysis and prevent the associated NAFLD in animal studies.139, 140

Because insulin resistance is closely tied to NASH, estimating insulin sensitivity may be an important adjunct in evaluating potential therapies for NASH. Hepatic insulin sensitivity can be estimated using the product of glucose × insulin, either expressed as the HOMA (Homeostasis Model Assessment) score or the inverse-log transformed QUICKI (Quantitative Insulin Sensitivity Check Index).141 Perhaps more informative with respect to NASH and the role of adipose insulin resistance would be an estimate of adipose insulin sensitivity. Similar to hepatic insulin sensitivity, adipose insulin sensitivity can be estimated using the product of insulin × free fatty acid levels to calculate the “adipo-IR” index,137 a newly described measure that might be worth considering in future studies of NASH and in the management of patients with NASH.

Triglyceride accumulation in the liver was once thought to impair insulin signaling and be a cause of insulin resistance in the liver and muscle.8, 142, 143 However, newer data have convincingly shown that triglyceride does not impair insulin signaling just as it does not cause lipotoxic injury; instead, it is simply a reflection of the increased supply of fatty acids or their impaired disposal leading to both lipotoxic cell stress and defects in the insulin signaling that occur in parallel.28, 54, 126, 144-149

Mechanisms of Lipotoxic Cellular Injury and Death

A detailed overview of the current understanding of the mechanisms of lipotoxic hepatocellular injury and the resulting response to injury that leads to the histological phenotype of NASH has been provided elsewhere.8, 14 Oxidative stress certainly occurs, but its role in cell injury and death has not been established.150, 151 Whether it plays a pathogenic role in lipotoxic injury continues to be investigated152; however, the absence of consistent therapeutic efficacy data for antioxidants would argue against ROS as a major contributor to liver injury in all patients with NASH.153, 154 Moreover, a certain degree of oxidant stress may be important in maintaining normal insulin signaling, and some studies have shown that excessive antioxidants can actually contribute to insulin resistance.155-157 Apoptotic cell death appears to be central to the pathogenesis of lipotoxic injury in the liver14, 158 as it is in other tissues12 and activation of caspases,50 and JNK may be major effectors of this process.10, 27, 159, 160 Exciting progress is being made in understanding the response of the liver to lipotoxic injury in animal models of NASH with the establishment of the hedgehog signaling pathway as a mechanism leading from injury to fibrosis.161, 162

The lipotoxicity model of NASH likely represents an oversimplification because it does not take into account a number of other known contributors to the phenotype of liver injury recognized as NASH. For example, impaired mitochondrial adenosine triphosphate production,163, 164 depletion of mitochondrial glutathione,105, 106 excessive accumulation of cholesterol in cell and mitochondrial membranes,105, 128 hypoxia associated with impaired blood flow or obstructive sleep apnea,165-167 gut-derived endotoxin and ethanol,168, 169 dysregulated adipokine production such as adiponectin deficiency or resistance,170, 171 and dysregulated innate immunity172, 173 are likely to be important contributors to varying degrees in different patients, but are not directly accounted for in this model. This underscores the likelihood that what is currently recognized as NASH on a liver biopsy actually reflects a mechanistically diverse collection of abnormalities, yet with a common histologic phenotype.

Clinical Implications

The hypothesis that NASH is caused by nontriglyceride lipotoxicity has several important clinically relevant caveats. First, if lipotoxicity occurs independent of triglyceride accumulation, then there are probably rare patients who have lipotoxic liver injury but who do not have readily detectable triglyceride accumulation. This hypothetical condition could arise if hepatic triglyceride synthesis is impaired and the flux of fatty acids to the liver is not counterbalanced by oxidative disposal. Such nonsteatotic lipotoxic liver injury might present with unexplained liver enzyme elevations and a biopsy showing inflammation, Mallory-Denk bodies, and possibly fibrosis, but no significant steatosis. Whether ballooning would be present in such patients is uncertain. Although a common view is that ballooned cells are hydropic, electron microscopy studies suggest that what is seen as ballooning on routine histological evaluation might actually represent foamy accumulation of very small triglyceride droplets that are lost from biopsy specimens during routine formalin preservation for paraffin embedding.174 A recent study showing liver enzyme elevations but only trivial changes in liver fat content after eating large amounts of fast food may be an example of such nonsteatotic lipotoxicity.175

A second caveat is that steatosis does not mechanistically “progress” to NASH. Although longitudinal studies confirm this conclusion, the paradigm of steatosis progressing to steatohepatitis has become firmly entrenched in our way of thinking about the disease, possibly because of its intuitive appeal. There are, of course, occasional cases of fibrosis developing in the setting of steatosis alone, or steatohepatitis being found on a follow-up biopsy when only steatosis was evident on an earlier biopsy. These relatively infrequent cases could be explained by the presence of low-level lipotoxic injury not initially detected by our relatively insensitive techniques of identifying ballooned cells or Mallory-Denk bodies or alternatively, lipotoxic injury occurring in some patients due to episodic environmental factors such as fluctuations in trans-fat consumption. The observation in mice that a high trans-fat diet induced severe steatohepatitis176 raises the possibility that human NASH could be more dependent on this dietary component than is currently recognized. This hypothesis is supported by correlative data in humans177 and in the study of students with liver enzyme elevations caused by fast food consumption.175, 178

Even though triglyceride accumulation does not appear to cause liver injury, measuring liver fat content can be a sensitive barometer of excessive fatty acid trafficking from peripheral adipose tissue and thus might serve as a sensitive indicator of peripheral insulin resistance.179 It has been tempting to assess the efficacy of treatments that improve peripheral liporegulation by their effects on hepatic steatosis. Caution is warranted when taking this approach, however, because the severity of steatosis did not correlate with the presence of steatohepatitis in one study,180 nor did changes in steatosis correlate with improvement in indices of injury in a clinical trial.181 Furthermore, if an intervention were to specifically impair the formation of triglyceride, this could diminish fat accumulation in the liver, yet predispose to lipotoxic injury. Thus, an awareness of putative mechanisms of therapeutic agents is needed before using improvement in hepatic steatosis as a sole surrogate marker for diminished flux of fatty acids through the liver and improvement of the associated lipotoxicity.

Finally, the field has wrestled with issues of developing an appropriately descriptive nomenclature for the various liver abnormalities seen in patients with excess flux of fatty acids through the liver and lipotoxic liver injury.182 The term “nonalcoholic steatohepatitis” is widely used, but is problematic because identifying a disease as not being another disease—in this case, alcoholic steatohepatitis—is unsatisfactory.182 And now, if it is incorrect to use the “steato” prefix because steatosis is not essential to the pathogenesis of injury, then the need for a better name is only more pressing. Nontriglyceride lipotoxic liver injury is not terribly poetic in full form or acronym, but is certainly descriptive and differentiates it from the idea that triglyceride can cause lipotoxic injury. However, it is still a “non” name, so perhaps simply lipotoxic liver injury should suffice, with the caveat that triglyceride is generally inert and not a contributor to lipotoxicity. These suggestions and others might be considered by investigators in the field in order to develop a new nomenclature, not only for lipotoxic liver injury but also for steatosis without significant inflammation or injury (i.e., non-NASH NAFLD, a name of three “non's”).


A significant body of evidence now forces us to rethink the causes of NASH. Once thought to be a disease caused by triglyceride accumulation in hepatocytes with subsequent oxidant stress and lipid peroxidation causing inflammation and fibrosis, new data from animal studies and a limited number of human studies now provide convincing evidence that triglyceride accumulation does not cause insulin resistance or cellular injury in the liver.

The lipotoxic liver injury hypothesis for the pathogenesis of NASH suggests that we need to focus our therapeutic efforts on reducing the burden of fatty acids going to the liver or being synthesized in the liver. This can be accomplished by improving insulin sensitivity at the level of adipose tissue to prevent inappropriate peripheral lipolysis and by preventing unnecessary de novo lipogenesis in the liver. Excess carbohydrates are the major substrates for de novo lipogenesis, and thus, reducing carbohydrate consumption through dietary changes and increasing muscle glucose uptake through exercise remain important cornerstones of treatment and prevention of lipotoxic liver injury, a disease hitherto called NASH.


I am indebted to Elizabeth Brunt for her thoughtful comments regarding this review. My apologies to the many investigators in the field of lipotoxicity whose contributions to the field were inadvertently not included.