HIV Articles  
NASH: a mitochondrial disease  
  ...... We believe that NASH can be considered, to some extent, as a mitochondrial disease, since mitochondrial dysfunction is involved at all successive steps leading to NASH..... insufficient mitochondrial function is skeletal muscle seems to play a major, albeit indirect, role in the genesis of hepatic steatosis..... hepatic mitochondrial dysfunction also plays a major role in the development of the main NASH lesions.....
Journal of Hepatology
June 2005
Dominique Pessayre, Bernard Fromenty
Unite INSERM 481, Faculte de Medecine Xavier-Bichat, 16 rue Henri Huchard, 75018 Paris, France
Despite our tendency to overeat, excessive fat accumulation was prevented in the past, as any excess weight soon impaired the physical fitness required to gather food, and to either fight or escape predators or foes. However, for the first time in human history, a large fraction of the population in affluent countries can now concomitantly indulge in rich food and physical idleness. The imbalance between food intake and the limited amount of fuels that can be burnt by mitochondria in inactive persons causes obesity, whose prevalence is increasing in affluent countries [1].
Obesity can lead to insulin resistance and hepatic steatosis, which triggers apoptosis, necrosis, Mallory bodies, an inflammatory cell infiltrate, and fibrosis in some patients [2]. The association of steatosis with these other liver lesions is called steatohepatitis, and the term non-alcoholic steatohepatitis (NASH) mostly refers to the steatohepatitis associated with insulin resistance.
Although the successive events triggering NASH are not fully understood, it seems that fat accumulation in myocytes triggers insulin resistance in muscle, thus causing pancreatic b-cells to release large amounts of insulin. High insulin levels increase hepatic free fatty acid (FFA) synthesis in the liver to cause steatosis, which can trigger NASH in some patients.
The purpose of this review is to discuss the evidence, mechanisms and implications of mitochondrial dysfunction at the successive steps leading to NASH, as already partially reviewed elsewhere [3-8]. Beforehand, however, it may be useful to briefly recall some salient features of mitochondria.
Mitochondria: bacterial remnants assisted and controlled by nuclear genes Like their bacterial ancestors [9], mitochondria have two membranes. The circular outer membrane surrounds the inter-membrane space, while the folded inner membrane invaginates into cristae protruding into the mitochondrial matrix. Like bacteria, mitochondria have their own circular DNA located in the matrix [10]. Although most of the ancient bacterial genes have been lost or have migrated to nuclear DNA [11], the residual human mitochondrial DNA still encodes for 13 of the polypeptides of the respiratory chain, including ATP synthase [10]. The thousand or so of other proteins present in mitochondria are encoded by nuclear DNA, synthesized within the cytoplasm and imported into the mitochondria.
Mitochondria play a major role in fuel oxidation. Although the glycolysis of glucose into pyruvate occurs in the cytoplasm, mitochondrial pyruvate dehydrogenase then converts pyruvate into acetyl-CoA, which is metabolized by the mitochondrial tricarboxylic acid cycle. The entry of long-chain FFAs into the mitochondria is critically dependent on carnitine palmitoyl transferase I (CPT-I), an outer membrane enzyme whose activity is inhibited by malonyl-CoA [12]. Malonyl-CoA is formed by acetyl-CoA carboxylase and is the first step in the synthesis of fatty acids from acetyl-CoA. Once inside mitochondria, FFAs are cut by the mitochondrial b-oxidation cycle into acetyl-CoA subunits, which are further oxidized by the tricarboxylic acid cycle.
Mitochondria play a major role in ATP formation. The oxidation of pyruvate and FFAs is associated with the conversion of NAD+ and FAD into NADH and FADH2, which transfer their electrons to the mitochondrial respiratory chain. Electrons migrate to cytochrome c oxidase (complex IV), where they combine with oxygen and protons to form water [3]. This flow of electrons is coupled with the extrusion of protons from the mitochondrial matrix into the intermembrane space. When energy is needed, protons re-enter the matrix through ATP synthase, causing the conversion of ADP into ATP.
Mitochondrial function and biogenesis are controlled by nuclear genes [13]. Although this control is exerted in all organs, it is particularly important for muscles (Fig. 2), whose energy requirements vary markedly with physical activity. When mitochondrial function is insufficient to sustain the drain in cell ATP during prolonged exercise, ADP increases, and adenylate kinase converts 2 ADP molecules into one ATP molecule and one AMP molecule [14]. The increase in cell AMP is sensed by AMP-activated protein kinase (AMPK), which turns off ATP-consuming pathways and switches on ATP-producing catabolic pathways [15]. AMPK triggers an adaptive increase in the cell surface glucose transporter 4 thus increasing glucose uptake [14]. It also activates hexokinase, the first enzyme involved in glucose utilization [14]. AMPK phosphorylates and inactivates acetyl-CoA carboxylase, thus decreasing the synthesis of malonyl-CoA. This prevents the inhibition of CPT-I by malonyl-CoA and increases fat oxidation [14]. Finally, AMPK increases several mitochondrial enzymes, such as succinate dehydrogenase, citrate synthase and cytochrome c [14,15].
Another consequence of muscular exercise is to increase cell Ca2+, which activates Ca2+/calmodulin-dependent protein kinase (CaMK) [16]. CaMK induces the expression of peroxisome proliferator-activated receptor γ coactivator 1 (PGC-1). PGC-1 induces nuclear respiratory factors-1 and -2, resulting in an increase in the synthesis of nuclear DNA-encoded polypeptides of the respiratory chain, and the induction of mitochondrial transcription factor A [13,17]. Furthermore, PGC-1 binds to, and co-activates, the transcriptional function of nuclear respiratory factor-1 on the promoter of mitochondrial transcription factor A to increase the transcription and replication of mitochondrial DNA [17]. Finally, PGC-1 binds to, and cooperates with, PPAR-a (peroxisome proliferator-activated receptor α) in the transcriptional activation of nuclear genes encoding fatty oxidation enzymes, including CPT-I and medium-chain acyl-CoA dehydrogenase [18]. As a consequence of these concerted actions on both nuclear genes and mitochondrial DNA genes, PGC-1 increases both the number of cardiac mitochondria and also their individual capacity to oxidize fat and to generate energy through oxidative phosphorylation [19]. Interestingly, both the AMPK- and the CaMK-mediated signals seem to act coordinately, since AMPK was required for energy deprivation-induced increases in the expression of both CaMK and PGC-1[20].
Rats with low oxidative capacity develop obesity and other features of the metabolic syndrome
Rats were selectively bred for either low or high treadmill-running capacity [21]. After selection over 11 generations, low-capacity runners and high capacity runners were obtained, which differed 3.4-fold in the distance they could run before exhaustion. The rapidly exhausted rats had lower muscular PGC-1 expression and lower expression of oxidative phosphorylation polypeptides than rats able to run for long distances. The metabolically handicapped rats developed high blood glucose, insulin, and triglyceride levels, as well as visceral obesity and high blood pressure, suggesting that insufficient muscular mitochondrial function can trigger the whole spectrum of the human metabolic syndrome [21].
Mitochondrial function is insufficient in the skeletal muscle of obese and/or (pre)diabetic subjects
There is strong evidence for insufficient mitochondrial function in the skeletal muscle of obese subjects, type 2 diabetic patients and their offspring. In obese women, visceral fat was negatively correlated with mitochondrial CPT-I and citrate synthase activities in skeletal muscle [22]. The skeletal muscle mitochondria of obese patients had lower CPT, citrate synthase, and cytochrome c oxidase activities than those of lean individuals, and these activities did not improve after weight loss [23]. In the skeletal muscle of patients with type 2 diabetes, there was a 50% decrease in mitochondrial DNA copy number [24], and the skeletal muscle mitochondria of diabetic obese patients exhibited lower NADH:O2 oxidoreductase activity, lower citrate synthase activity and smaller mitochondria than those of lean volunteers [25]. In the young and lean, yet insulin-resistant offsprings of patients with type 2 diabetes, there was a 30% decrease in the rate of mitochondrial ATP production and an 80% increase in muscle lipid contents [26]. Finally, in patients with type 2 diabetes, there was a slight decrease in the transcripts for both PGC-1x and nuclear respiratory factor-1, and a coordinated decrease in the expression of genes involved in the tricarboxylic acid cycle and in oxidative phosphorylation [27,28]. These observations clearly indicate that mitochondrial function and fatty acid oxidation are deficient in these patients.
Although the reasons for this insufficient mitochondrial function are still unknown, present evidence suggests that several mechanisms could be involved, alone or in combination.
Physical inactivity
One major mechanism for insufficient mitochondrial function in the skeletal muscles of obese and/or (pre)diabetic patients could be physical inactivity, since exercise increases PGC-1 and mitochondrial biogenesis in skeletal muscle [17].
PGC-1 polymorphism
A Gly482Ser genetic polymorphism affects PGC-1 [29]. The Gly/Gly homozygous genotype was associated with lower lipid oxidation in Pima Indians [29], and an increased risk of type 2 diabetes mellitus in other studies [30,31].
Mitochondrial dysfunction could be secondarily aggravated as a consequence of fat accumulation, lipid peroxidation and other 'lipotoxic' effects, as discussed further on for hepatic mitochondria.
Comparison of healthy and lean, but elderly subjects with young participants matched for lean body fat mass and fat mass showed a 40% decrease in the oxidative and phosphorylation activities of muscle mitochondria, a 45% increase in intramuscular fat content, and a 40% decrease in insulin-stimulated peripheral glucose uptake (indicating mild insulin resistance) in elderly subjects [32]. This age-associated decline in mitochondrial function could play a role in the increased prevalence of type 2 diabetes with age [32]. Although physical inactivity may be the major reason for the insufficient function of muscle mitochondria during old age [33], mitochondrial DNA aging could also play a role, since mitochondrial DNA accumulates a large number of point mutations and also deletions during old age [34].
Insufficient mitochondrial function contributes to insulin resistance in muscle Together with excessive food intake, insufficient mitochondrial function in the muscles of obese and/or (pre)diabetic patients may cause intramyocellular fat accumulation [26], which itself seems to play a major role in the resistance of myocytes to the intracellular signaling effects of the insulin receptor (Fig. 3) [35].
In lean persons, insulin acts on the insulin receptor of myocytes to trigger the tyrosine phosphorylation of insulin receptor substrate, which activates phosphatidyl inositol 3-kinase and protein kinase B, to eventually cause the translocation of the glucose transporter from intracellular storage vesicles to the plasma membrane of myocytes [36]. Abundant expression of the glucose transporter on the plasma membrane causes rapid glucose uptake, which limits the increase in blood glucose, and therefore, limits insulin secretion by pancreatic b-cells.
In obese people, however, fat-laden myocytes are resistant to the signaling effects of insulin [36]. Acyl-CoA or other derivatives of FFAs may limit the activation of insulin receptor substrate [36]. The activation of Jun N-terminal kinase [37] and protein kinase C-o [38] may cause the serine phosphorylation and thus inactivation of insulin receptor substrate. Insufficient translocation of glucose transporters to the plasma membrane limits glucose uptake by myocytes [36].
Related effects occur in fat-engorged adipocytes, which exhibit endoplasmic reticulum stress and Jun N-terminal kinase activation [39]. Like fat-laden myocytes, fat-engorged adipocytes are insulin-resistant, with decreased translocation of glucose transporters to their plasma membrane [36].
Insulin resistance is initially compensated for by increased insulin secretion, but mitochondrial dysfunction in pancreatic b-cells may eventually blunt insulin secretion to trigger diabetes
Insulin resistance in myocytes and adipocytes tends to slightly increase blood glucose levels after meals, thus triggering a swift release of insulin by pancreatic b-cells. Mitochondria play a major role in this process (Fig. 4). Glucose equilibrates across the plasma membrane of pancreatic b-cells, and undergoes glycolysis to pyruvate in the cytosol (Fig. 4) [40]. Mitochondrial pyruvate dehydrogenase then transforms pyruvate into acetyl-CoA, which enters the mitochondrial tricarboxylic acid cycle to generate ATP. The increase in the ATP/ADP ratio closes ATP-sensitive potassium channels on the plasma membrane of the β-cell. The decreased outflow of potassium results in the depolarization of the plasma membrane, and opens voltage-sensitive calcium channels to increase cell calcium, which triggers the exocytosis of insulin-loaded vesicles [40].
In a first stage of insulin resistance, the pancreatic b-cells of obese subjects respond normally to any slight increase in blood glucose [41]. Thanks to a three to four-fold increase in basal and 24-hour insulin secretion rates, high plasma insulin levels overcome insulin resistance in muscle, and these subjects can maintain normal glucose levels [41].
Over the years, however, the insulin secretory response of pancreatic b-cells after meals may become delayed and blunted in some patients [41]. Not enough insulin is released to compensate for insulin resistance, and glucose intolerance and diabetes can develop [41].
Because mitochondrial ATP formation plays a major role in insulin secretion, it was tempting to look for a mitochondrial explanation to the decline in insulin secretion by pancreatic b-cells.
Overexpression of uncoupling protein 2 and impaired insulin secretion Compared to control islets, the pancreatic islets isolated from type 2 diabetic patients exhibited an increased protein expression of uncoupling protein 2, a decreased mitochondrial membrane potential, decreased ATP levels, and reduced insulin secretion in response to glucose [42]. As discussed further on, obesity and insulin resistance are associated with increased plasma FFAs levels and with an increased cellular formation of reactive oxygen species (ROS). Both FFAs and ROS induce uncoupling protein 2 in pancreatic b-cells [43-45]. Uncoupling protein 2 overexpressed in the mitochondrial inner membrane of pancreatic b-cells allows the re-entry of protons from the intermembrane space into the mitochondrial matrix [43-45]. This re-entry decreases the mitochondrial membrane potential and ATP generation, and decreases glucose-stimulated insulin secretion (Fig. 4) [43-45].
Uncoupling protein 2 polymorphism
Interestingly a -866G/A polymorphism modulates the activity of the promoter of human uncoupling protein 2 [46]. The A allele has higher promoter activity than the G allele [46], and glucose-stimulated insulin secretion was lower in humans with the homozygous A/A genotype [47]. Although uncoupling protein 2 blunts ATP formation and insulin secretion in pancreatic b-cells, on the other hand, its expression in many other tissues may waste energy as heat, unleash the flow of electrons in the respiratory and increase respiration and mitochondrial fuel consumption. This may explain why A/A UCP2 homozygotes exhibited a lower prevalence of obesity than G/G homozygotes [48].
Mitochondrial DNA mutations
In addition to increased uncoupling protein 2 expression, the accumulation of somatic point mutations and deletions in mitochondrial DNA during old age would be expected to also blunt insulin release by pancreatic b-cells. This could explain the increased prevalence of diabetes with age. That mitochondrial DNA mutations can indeed impair insulin secretion is shown by a relatively rare form of diabetes, which is due primarily to inborn mitochondrial DNA mutations. A large kindred was reported with maternally transmitted type 2 diabetes and deafness, which was due to a heteroplasmic mutation in the mitochondrial tRNA for leucine [49]. In another kindred, mitochondrial diabetes and deafness were due to a maternally transmitted large mitochondrial DNA deletion [50].
Finally, as frank diabetes develops, a further vicious cycle may ensue, because the combination of high glucose and high FFAs synergistically triggers β-cell apoptosis [51], to finally decrease pancreatic β-cell mass in humans with type 2 diabetes [52].
High glucose and/or insulin levels induce hepatic lipogenesis to cause hepatic steatosis
Normally, insulin inhibits the hormone-sensitive lipase of adipocytes to block adipose tissue lipolysis [53]. In contrast, the fat-engorged, insulin-resistant adipocytes of obese persons keep releasing FFAs in the circulation despite high insulin levels. The increased release of FFAs from adipose tissue increases plasma FFAs [54]. The high plasma FFA levels may either increase hepatic FFA uptake, or may at least maintain a normal rate of hepatic FFA uptake, despite increased hepatic FFA levels.
Concomitantly, insulin resistance in myocytes increases plasma levels of glucose and/or insulin. Both glucose and insulin increase the hepatic synthesis of FFAs (Fig. 5) [55]. Glucose activates carbohydrate response element binding protein, with two consequences. Firstly, carbohydrate response element binding protein stimulates liver-type pyruvate kinase, thus increasing the glycolysis of glucose into pyruvate, which forms the acetyl-CoA and then the malonyl-CoA required for FFA synthesis [56]. Secondly, carbohydrate response element binding protein stimulates the transcription of all lipogenic genes, to increase hepatic FFA and triacylglycerol synthesis [57].
Insulin also potently stimulates hepatic lipogenesis [55]. Insulin activates phosphatidyl inositol 3-kinase and protein kinase C-λ, which increases the transcription of sterol regulatory element-binding protein-1c [58], which increases the transcriptional activation of all lipogenic enzymes [55,58]. Although other effects of insulin can be partially blunted by insulin resistance in the liver, the hepatic lipogenic effects of insulin are not prevented [55]. In insulin resistant states, Foxo1 may retain some activity despite high insulin levels, thus allowing persistent hepatic gluconeogesesis [59,60]. In contrast, Foxa2 may be completely inactivated by insulin, even in insulin resistance states, thus permitting hepatic fat accumulation [59,60].
While both glucose and insulin increase the hepatic synthesis of FFA in diabetic obese patients, insulin alone suffices to increase hepatic lipogenesis in patients with insulin resistance but normal blood glucose levels [61].
Although the liver is laden with fat in patients with NASH, it does not enlarge indefinitely. A new steady is achieved, whereby the increased hepatic synthesis of FFAs, and possibly their increased hepatic uptake, are compensated for by an increased removal of lipids from the liver.
One pathway, which removes fat from the liver, is very low density lipoprotein secretion [62]. In the lumen of the endoplasmic reticulum and Golgi apparatus, microsomal triglyceride transfer protein lipidates apolipoprotein B into triglyceride-rich very low density lipoprotein particles, which are transported to the plasma membrane by vesicles to be released into the plasma [62]. Microsomal triglyceride transfer protein mRNA and activity, as well as the in vivo hepatic secretion of triglyceride-rich lipoproteins, are increased in genetically ob/ob mice [63]. Similarly, plasma very low density lipoproteins and triglycerides can be increased in insulin-resistant patients [64,65]. All these data suggest that the hepatic secretion of triglycerides can be increased in insulin-resistant obese subjects.
However, divergent effects are observed concerning hepatic apolipoprotein B secretion. While the hepatic secretion of apolipoprotein B was slightly increased in obese patients without NASH, it was decreased in obese patients with NASH [66]. One possible explanation could be that hyperinsulinemia may be even higher in patients with NASH and might thus exceed hepatic insulin resistance. Hyperinsulinemia tends to increase the hepatic degradation of apolipoprotein B, thus limiting its hepatic secretion [67]. Interestingly, insulin also blocks apolipoprotein B secretion by the intestine [68]. Although patients with NASH exhibited a higher postprandial increase in total triglyceride and very low density lipoprotein triglycerides than control subjects, apolipoprotein B levels increased much less in NASH patients than controls [69].
Another pathway which can limit excess fat accumulation in the liver is an increased FFA oxidation. Indeed, hepatic FFA oxidation is increased in genetically obese mice [70,71] and in patients with NASH [72-74]. Although this increase may seem paradoxical given the mitochondrial dysfunction to be described further on, it is noteworthy that mitochondrial FFA oxidation is not altered until respiration is severely impaired [75]. Furthermore, several adaptive changes may allow hepatic mitochondria to oxidize more fat in patients with NASH. A first mechanism could be the increased hepatic FFA concentrations, which may force the entry of FFAs into the mitochondria. A second mechanism may involve changes in CPT-I expression and its sensitivity to malonyl-CoA inhibition. CPT-I is up-regulated in rodent models of type 1 or type 2 diabetes, and its affinity for its physiological inhibitor, malonyl-CoA, is decreased [70,76]. Loss of CPT-I inhibition by malonyl-CoA could explain why b-oxidation can increase in type 2 diabetic states, despite high insulin and malonyl-CoA levels, which, normally would block the entry of long-chain fatty acids into mitochondria. Finally, another mechanism increasing fatty acid oxidation may be the proliferation and enlargement of hepatic peroxisomes in patients with a fatty liver [77,78]. An important upstream mediator of these diverse changes may be PPAR-a. PPAR-a is activated by long-chain FFAs, and, furthermore, its hepatic mRNA is upregulated in murine models of obesity [79]. PPAR-a activation increases the expression of enzymes involved in peroxisomal and mitochondrial β-oxidation, including CPT-I [79,80]. PPAR-a activation also increases uncoupling protein 2 mRNA [81]. This uncoupling protein could decrease the mitochondrial membrane potential, unleash the flow of electrons in the respiratory chain, increase mitochondrial respiration, and allow better re-oxidation of NADH into the NAD+ required for fatty acid oxidation. However, although both the uncoupling protein 2 mRNA and an immunoreactive uncoupling protein 2-like protein were increased in the fatty liver of genetically obese mice [82], it has been argued that the uncoupling protein 2 protein itself may not be expressed in the liver [83]. Thus, it remains unclear whether uncoupling protein 2 itself, or some other uncoupling protein, may be part of the adaptive changes that develop in fatty livers.
Although a new equilibrium is finally achieved, whereby the increased fat input is now compensated by increased output pathways, this new equilibrium is achieved at the expense of expanded pools of hepatic FFAs and triglycerides, thus causing hepatic steatosis. The outcome of this steatosis differs in different subjects. In some patients, hepatic steatosis is, and remains, the only liver lesion. Other patients, however, silently develop apoptosis, necrosis, an inflammatory infiltrate and fibrosis [2]. Similar liver lesions can evolve in patients or animals with chronic hepatic steatosis due to diverse other causes, and with or without insulin resistance, suggesting that the prolonged presence of fat can be harmful to the liver [7]. Although the reasons for the deleterious effects of fat ('lipotoxicity') are still incompletely understood, they could involve lipid peroxidation, increased TNF-a (tumor necrosis factor-α) expression and mitochondrial dysfunction [7].
Hepatic mitochondrial dysfunction contributes to the genesis of NASH lesions Patients with NASH have an impaired in vivo ability to re-synthesize ATP after a fructose challenge, which transiently depletes hepatic ATP [84]. Their hepatic mitochondria exhibit ultrastructural lesions, with the presence of para-cristalline inclusions in megamitochondria [72,85]. Patients with NASH have decreased hepatic mitochondrial DNA levels [86], decreased protein expression of several mitochondrial DNA-encoded polypeptides [86,87], and lower activity of complexes I, III, IV and V (ATP synthase), which are partly encoded by mitochondrial DNA, and also complex II [87], which is only encoded by nuclear DNA.
Although the mechanisms for hepatic mitochondrial dysfunction in NASH are still unknown, possible mechanisms may involve lipid peroxidation, TNF-a and ROS.
Lipid peroxidation
Even in the basal (fat-free) state, hepatocytes produce ROS at different sites, including the mitochondrial respiratory chain and microsomal and mitochondrial cytochrome P-450 2E1. This basal ROS formation is further increased in fatty livers. First, mitochondrial ROS formation may be increased, as discussed further on. Second, patients with NASH have increased hepatic levels of cytochrome P450 2E1 [88]. Third, the expression of endotoxin receptors on Kupffer cells is increased in animals with either obesity- or alcohol-induced hepatic steatosis [89,90]. Increased sensitivity of Kupffer cells to bacterial endotoxin might activate NAD(P)H oxidase and increase ROS formation by these cells. This abundant ROS formation oxidizes the unsaturated lipids of fat deposits to cause lipid peroxidation [7]. Indeed, eleven different treatments causing acute or chronic steatosis increased hepatic thiobarbituric acid-reacting lipid peroxidation products in mice, and also increased ethane exhalation, an in vivo index of lipid peroxidation [91]. After a single dose of tetracycline or ethanol, there was a parallel time course in the rise and fall of hepatic triglycerides, and the rise and fall of ethane exhalation, suggesting a cause-and-effect relationship between the presence of oxidizable fat in the liver and lipid peroxidation [91]. Extensive lipid peroxidation also occurs in animals with hepatic steatosis due to a choline/methionine-deficient diet [92], in genetically obese leptin-deficient ob/ob mice (personal unpublished results), and in patients with NASH [72].
Lipid peroxidation products directly attack and inactivate respiratory chain components, including cytochrome c oxidase, the terminal oxidase of the respiratory chain (Fig. 6) [93,94].
TNF-a is released by fat-engorged adipocytes [95] and fat-laden hepatocytes [96,97], and might also be released by endotoxin-stimulated Kupffer cells, as a consequence of the increased expression of endotoxin receptors on these cells. As discussed further on, TNF-a acts on its hepatocyte receptor to trigger permeability of the mitochondrial membranes. This releases cytochrome c from the mitochondrial intermembrane space into the cytosol, thus partially blocking the flow of electrons from complex III to complex IV (Fig. 6).
Mitochondrial ROS
A last factor which damages hepatic mitochondria in NASH is increased mitochondrial ROS formation. Previous studies have shown that mitochondrial ROS formation can be increased by augmenting the delivery of electrons to the respiratory chain during fatty acid oxidation [98] or glucose oxidation [99], or by partially blocking the flow of electrons within this chain, for example in cytochrome c-depleted mitochondria [100]. In patients with NASH, the increased mitochondrial b-oxidation rate [72-74] augments the formation of NADH and FADH2, and thus the delivery of electrons to the respiratory chain. Concomitantly, the decreased activity of respiratory chain complexes [87] partially blocks the flow of electrons within the respiratory chain. The imbalance between an increased input and an impaired outflow of electrons may cause the accumulation of electrons in complex I and complex III of the respiratory chain [98], to increase the formation of the superoxide anion radical (Fig. 6). Indeed, an increased mitochondrial ROS formation has been demonstrated in genetically obese mice [101] or rats fed a choline-deficient diet [102], an animal model of steatohepatitis.
Once mitochondrial ROS formation starts to increase in NASH, it could trigger several vicious cycles. ROS directly damage respiratory chain polypeptides, thus further blocking electron flow in the respiratory chain. ROS oxidize the unsaturated lipids of cytoplasmic hepatic fat deposits and the cardiolipin of mitochondria, to liberate reactive lipid peroxidation products, which inactivate cytochrome c oxidase [93,94]. Both ROS and lipid peroxidation products attack mitochondrial DNA. Oxidative mitochondrial DNA lesions, such as oxidized DNA bases [103], mitochondrial DNA depletion [86], and possibly acquired mitochondrial DNA mutations, may all impair the synthesis of respiratory chain polypeptides. All these effects may further block the flow of electrons in the respiratory chain, to further increase mitochondrial ROS formation. Furthermore, the resulting decrease in the mitochondrial membrane potential could prevent the potential-driven import of DNA repair enzyme into the mitochondria. Decreased import of a mismatch-repairing enzyme has been observed in animal models of steatohepatitis [103], which could possibly favor the occurrence of point mutations. Finally, ROS may deplete some antioxidants to further aggravate ROS-induced damage. Low vitamin E levels are found in some obese children with NASH [104], and supplementation with vitamin E can decrease transaminases in obese children [105].
By increasing mitochondrial ROS formation, hepatic mitochondrial dysfunction may contribute to the genesis of the hepatic lesions of NASH. Indeed, ROS increase the expression of several cytokines, including TNF-b (transforming growth factor-β), interleukin-8, TNF-a and Fas ligand, and ROS also trigger lipid peroxidation, which releases the biologically active aldehydes, malondialdehyde and 4-hydroxynonenal [3]. Both cytokines and lipid peroxidation products may act together to trigger the diverse lesions of NASH [7].
Inflammatory infiltrate and mallory bodies
TNF-b, 4-hydroxynonenal and interleukin-8 are chemoattractants for human neutrophils, which may account, in part, for the neutrophil infiltrate [3]. TNF-b also induces tissue transglutaminase [106], which cross-links cytokeratins [107], and could participate in the generation of Mallory bodies, which are formed of aggregated cytoskeletal proteins, in particular polymerized cytokeratins [3].
Cell death
Normally, hepatocytes express Fas (a membrane receptor), but not Fas ligand, preventing them from killing their neighbors [108]. However, the expression of Fas is increased in the liver of patients with NASH [109] or in fat-laden mouse hepatocytes [110]. Furthermore, several other conditions increasing ROS formation, such as drugs or alcohol abuse, cause Fas ligand expression by hepatocytes, so that Fas ligand on one hepatocyte can interact with Fas on another hepatocyte, to cause fratricidal apoptosis [108]. ROS also increase the synthesis of TNF-a, and patients with NASH have both high hepatic TNF-a mRNA levels and high TNF-a receptor 1 expression [96].
The interaction of Fas ligand with Fas, or TNF-a with TNF-a receptor 1, activates procaspase-8 into caspase-8, which cuts Bid (BH3 interacting domain death agonist) [108]. Truncated Bid can enter the outer mitochondrial membrane to make this membrane leaky, and it also induces a conformational change in Bax (Bcl-2-associated x protein), which translocate to mitochondria and associates with its analogue Bak, to form channels in the outer mitochondrial membrane [111]. Increased permeability of the outer mitochondrial membrane releases cytochrome c from the intermembrane space of mitochondria, thus partially blocking the flow of electrons into the respiratory chain and increasing mitochondrial ROS formation [112].
ROS could then act on the same or other mitochondria to open an inner membrane pore, the mitochondrial permeability transition pore. Due to high matrix osmotic pressure, the opening of this pore causes an influx of water into the mitochondrial matrix. The inner mitochondrial membrane has several folds and can accommodate this increased volume, without bursting. In contrast, the unfolded outer mitochondrial membrane breaks when the mitochondrion swells [113]. Due to increased permeability and rupture of the outer membrane, cytochrome c and other pro-apoptotic factors leave the mitochondrial intermembrane space to activate caspase-9 in the cytosol. Caspase-9 then activates caspase-3, which re-amplifies the mitochondrial loop of apoptosis by further cutting caspase-8, to eventually trigger a major activation of effector caspases and apoptosis [113]. Indeed, apoptosis is an important mechanism of cell death in patients with NASH [109].
Hepatic fibrosis results from the activation of Kupffer cells and stellate cells. After ingesting apoptotic bodies, Kupffer cells and hepatic stellate cells release TNF-b, which activates stellate cells into collagen-producing myofribroblastic cells [114,115]. Lipid peroxidation products further activate fibrogenesis in two ways. They enhance the production of TNF-b by macrophages [116], and they also directly enhance collagen production by activated stellate cells [117]. Finally, leptin, which is secreted by fat-engorged adipocytes, is increased in obese patients [118]. Leptin has a permissive effect on hepatic fibrogenesis [119,120]. First, leptin acts on Kupffer cells, hepatic sinusoidal endothelial cells and stellate cells to release more TNF-b [120-122]. Second, leptin also directly increases collagen formation by stellate cells [122].
Hepatocellular carcinoma
The marked apoptosis rate in patients with NASH [109] requires a compensatory increase in the cell proliferation rate of progenitor cells to maintain liver mass (Fig. 7). Concomitantly, both ROS and lipid peroxidation products damage DNA. The combination of DNA damage and increased cell proliferation causes gene mutations. As these mutations accumulate over the years, and as there is a constant apoptotic pressure, cells that resist apoptosis and/or escape the control of the cell cycle may be selected, to finally allow the development of a malignant cell clone.
We believe that NASH can be considered, to some extent, as a mitochondrial disease, since mitochondrial dysfunction is involved at all successive steps leading to NASH.
Firstly, lack of exercise (and possibly a PGC-1 or other genetic polymorphisms) may decrease mitochondrial biogenesis and fat oxidation in skeletal muscle, and this insufficient mitochondrial function could be further impaired during old age, as a consequence of the accumulation of somatic mitochondrial DNA mutations. Together with excessive food intake, the insufficient oxidation of fat in skeletal muscle causes intramyocellular fat accumulation, which seems to trigger insulin resistance. Initially, increased insulin secretion by pancreatic b-cells compensates for insulin resistance, and these subjects maintain normal blood glucose levels. However, the high plasma insulin levels increase sterol regulatory element-binding protein 1c expression in the liver, to induce lipogenesis and cause hepatic steatosis. Thus, insufficient mitochondrial function is skeletal muscle seems to play a major, albeit indirect, role in the genesis of hepatic steatosis.
Secondly, in fat-engorged hepatocytes, several vicious cycles involving TNF-a, ROS and chemically reactive lipid peroxidation products alter mitochondrial respiratory chain polypeptides and mitochondrial DNA to partially block the flow of electrons in the respiratory chain and increase mitochondrial ROS formation. The resulting oxidative stress releases both biologically active lipid peroxidation products and cytokines, which act together to trigger the diverse hepatic lesions of NASH. In particular, the permeabilization of mitochondrial membranes is critically involved in cytokine-induced apoptosis, which in turn may contribute to fibrosis, and paradoxically may also contribute to cancer, through the stimulation of compensatory cell proliferation and the selection of apoptosis-resistant clones. Thus hepatic mitochondrial dysfunction also plays a major role in the development of the main NASH lesions.
Finally, in pancreatic b-cells, increased uncoupling protein 2 expression and possibly age-related mitochondrial DNA mutations can eventually blunt glucose-mediated ATP formation and insulin secretion, to finally cause diabetes. Thus, mitochondrial dysfunction in b-cells seems to play a major role in the development of diabetes. The resulting increase in blood glucose then induces carbohydrate response element binding protein in the liver, to further increase hepatic lipogenesis and hepatic steatosis.
  icon paper stack View Older Articles   Back to Top