Geoffrey C. Farrell, MD, FRACP
I was honoured to be invited to deliver the Georges Brohée Lecture for the 2013 WCOG in this wonderful city of Shanghai, and thank the organisers and the Belgian Society of Gastroenterology for bestowing this privilege on me.
Nonalcoholic fatty liver disease (NAFLD) is very common in Asia, Europe, North and South America and Australasia. Studies from Shanghai seven years ago confirmed that fatty liver disease in adults over the age of 30 was 20-25% in men, 15-30% in women and associated with metabolic factors such as obesity, high blood pressure, and diabetes rather than alcohol. In Hong Kong, 27% of the population have fatty liver by proton magnetic resonance spectroscopy (PMRS), and the prevalence of fatty liver depended on the number of metabolic criteria, such as central obesity, dyslipidemia, high blood sugar and high blood pressure; with three or more risk factors, 60-80% of individuals have fatty liver. The clinical outcomes of NAFLD include increased standardized mortality, diabetes, cardiovascular events, obesity-related cancers (breast, prostate, colorectal, uterine, cervix, adenocarcinoma of oesophagus, pancreatic, renal, etc.) followed by cirrhosis and primary liver cancer, especially hepatocellular carcinoma (HCC).
NAFLD is associated with a range of liver pathology from simple steatosis through steatohepatitis (NASH), a condition of liver cell injury with inflammation that leads to fibrosis and cirrhosis. NASH is even more strongly associated with metabolic risk factors than other cases of NAFLD; 98% have insulin resistance, 85% fully developed metabolic syndrome, and at least 60% established diabetes or diabetes on oral glucose tolerance testing. Family history of diabetes is also strongly associated with NASH and adverse liver outcomes of NAFLD.
An earlier concept was that the metabolic factors favoured development of steatosis (the first hit), but a second injurious or proinflammatory process (second hit) was needed for NASH pathogenesis. This concept has been outmoded for at least ten years because it fails to explain the strong relationships of more severe metabolic disorder to NASH (versus “not NASH” NAFLD). It also ignores the roles of the lipids themselves. Such molecules as saturated free fatty acids (satFFA), free cholesterol (FC) and diacylglycerol (DAG) activate a series of searing, serine/threonine kinases which block physiological tyrosine phosphorylation of insulin receptor sub-strate and other intermediates in insulin receptor signalling. In this way, lipid accumulation in nonadipose tissues causes insulin resistance (essential for NASH), but also cell injury, cell death, inflammation and a tissue wound healing response of regeneration and fibrosis. These phenomena are grouped together as lipotoxicity, regarded as the mechanism of pancreatic beta cell injury in type 2 diabetes and muscle injury in metabolic syndrome.
To be a candidate for causing NASH, a lipotoxic lipid should be higher in NASH than in non-NASH NAFLD livers, found in models of metabolic syndrome-related to NASH (versus nutritional depletion or genetically manipulated models), its accumulation should be explained by either genetic or metabolic pathogenesis, and direct exposure of liver cells to the lipid molecule should cause apoptosis and necrosis with release of proinflammatory molecules. The latter aspect could be regarded as the toxic equivalent of Koch’s postulates for infectious organisms.
Triglycerides do not cause NASH; experimentally they protect fatty livers against injury. SatFFA cause JNK dependent lipo-apopotosis in vitro, but do not selectively accumulate in NASH livers versus simple steatosis. There is variable but not very strong evidence that DAG or various phospholipids accumulate in NAFLD (ceramide was not increased in two studies), but the strongest evidence for a lipotoxic molecue in NASH is cholesterol, more specifically the highly reactive free cholesterol (FC) fraction.
Three human lipidomic studies show selective accumulation of FC in NASH versus non-NASH, NAFLD. Dietary studies in which high fat diet with added cholesterol causes NASH in rodents appears strongly related to liver cholesterol content, as are some gene-manipulated mouse models (LDLR-/- and APOE knock-in mice. ABCB4 mutant opossums develop NASH associated with liver cholesterol accumulation. Mari and colleagues showed that cholesterol-loaded livers are sensitive to injury caused by death-inducing cytokines, such as Fas and TNF.
We have been studying mice with a defect in appetite regulation, Alms1 mutant (foz/foz) mice that undergo postnatal loss of hypothalamic appetite-sensing cilia. Foz/foz mice gain weight rapidly after weaning, develop leptin and insulin resistance and, when fed an atherogenic diet, develop diabetes, hypercholesterolemia, high blood pressure, low serum adiponectin and NASH with fibrosis. We have studied mechanisms of lipotoxicity in these mice. FC accumulates at 12 weeks in association with NASH pathology. The mechanisms for FC accumulation include up-regulation of low density lipoprotein receptor (LDLR) (and CD36) on hepatocytes, suppression of cholesterol biotransformation (to form bile acids) through CYP7A, suppression of cholesterol export into bile (via ABCG5/8) and suppression of bile acid transport into bile (via ABCB4 and Bsep). There is also increased expression of cholesterol ester hydrolase (CEH), which could partly explain why free cholesterol accumulates rather than the nontoxic, long chain fatty acid esters (CE). Min and colleagues used our findings as a road map to study pathways of cholesterol turnover in human livers of patients with NASH. As in mice, SREBP2 was increased, but they did not find resultant upregulation of LDLR. Instead they found evidence of increased cholesterol synthesis based on indirect data (HMG-CoA reductase-phosphoprotein; peripheral metabolites), whereas in our mouse studies we measured enzyme activity directly and found the suppression expected physiologically from cholesterol accumulation. On the other hand, Min et al confirmed up-regulation of CEH, decreased CYP7A, decreased cholesterol secretion into bile and in blood, and decreased pathways of bile acid secretion. Thus, while some details differ between humans and mice, the net effect is a profound accumulation of free cholesterol in liver cells in NASH, but not in less severe forms of fatty liver disease. We also showed by experiments in primary hepatocytes that the high insulin concentrations that circulate in obese, diabetic mice directly upregulate SRABP2, LDLR and suppressed the bile salt export pump. Thus, we think that hyperinsulinemia resulting from insulin resistance is the primary cause of hepatic cholesterol dysregulation in NASH.
Our colleagues in Seattle (George Ioannou, Chris Savard) have demonstrated that livers of patients with NASH, in their atherogenic diet-fed mice and in our foz/foz mice with high FC accumulation, all exhibit abundant cholesterol crystals. In recent studies we have loaded primary hepatocytes with FC by incubation with LDL and shown that cholesterol accumulates in the plasma membrane, in the endplasmic reticulum, and in mitochondria. There is a dose-dependent relationship between hepatocyte FC content, LDH leakage, apoptosis and necrosis.
In our hepatocyte system, we observed JNK1 activation with nuclear accumulation of phospho-c-Jun, while JNK1-/- hepatocytes are refractory to FC-induced cell death. Further, potent JNK1 inhibitors, such as CC-401 and CC-903 (Celegene) completely protect hepatocytes from FC-induced cell death. We have adduced direct evidence that FC causes mitochondrial generation of ROS, with oxidative stress and ATP depletion as early as six hours after free cholesterol loading. Transmission electron microscopy confirms mitochondrial injury with swelling and fragmentation of cristae. There is direct evidence of membrane pore transition (MPT) with cytochrome C leakage out of the mitochondria into the cystosol. Inhibitors of MPT (cyclosporine A) or of resultant active caspase 3 protect cholesterol-loaded hepatocytes against apoptosis and necrosis. We found no evidence of ER stress or protection of by ER chaperone against cell death.
Incubation of hepatocytes with LDL caused dose dependent release of HMGB1. When we added conditioned culture medium (hepatocytes undergoing FC lipotoxicity versus control cells) to resting Kupffer cells, activation (p65 nuclear migration; morphology on SEM, inflammasome activation with release of IL1β) was clearly evident. Anti-HMGB1 antiserum completely blocked such KC activation. Further, TLR4 knock-out Kupffer cells were refractory to activation by FC lipotoxicity conditioned medium.
Together with studies from Shanghai in TLR4-/- mice, the present results lead us to propose that free cholesterol, either alone or in concert with satFFA, activate JNK1 in hepatocytes which descends on mitochondria to generate ROS with further activation of JNK1, and eventually MPT with cytochrome C release and activation of the aptosome. The concomitant fall in membrane potential leads to a decline of cellular ATP levels with consequences for the necrotic cell death pathway. Either or both redox stress and necrosis cause release of HMGB1, which can feed back via TLR4 on hepatocytes to further activate JNK and accentuate this type of injury. In addition, HMGB1, and likely other danger-associated molecular patterns interact with TLR4 on Kupffer cells to activate NF-kB and the inflammasome with resultant further release of chemokines and cytokines that an integral part of inflammation in NASH.
We have published data recently showing that lowering hepatic FC with combination of atorvastatin and ezetimibe reduces liver injury, reverses NASH pathology and, most importantly, reduces liver fibrosis in our obese, diabetic mouse model. At the population and clinical level, attempts to combat insulin resistance by improving physical activity (as well as weight control) should prevent hepatic FC accumulation. However, lipid lowering therapy may be required once this state has been established. In addition to cholesterol-lowering therapy, novel agents such as JNK1 inhibitors and agents that protect the mitochondria or block TLR4 activation could be introduced as novel therapies for established NASH.