As a consequence of the worldwide epidemic of diabesity, the prevalence of nonalcoholic fatty liver disease (NAFLD) is continuously rising.1 According to current concepts, the subset of patients with nonalcoholic steatohepatitis (NASH), histologically characterized by hepatocyte ballooning, inflammation, fibrosis, in addition to steatosis,2 are at significant risk for adverse hepatic outcomes due to cirrhosis and hepatocellular carcinoma.3 Interestingly, patients with advanced NASH and related cirrhosis may no longer display pronounced hepatic lipid accumulation, a constellation known as “burnt-out NASH.” This may have contributed to underestimation of this etiology since such patients may have frequently been classified as “cryptogenic” in the past. While many possible explanations such as diversion of insulin and nutrients from the liver due to portal hypertension and the general catabolic state in liver cirrhosis have been brought forward,4 the molecular mechanisms explaining this paradox have remained poorly understood. In this issue, Van der Poorten et al.5 have now addressed this important question in a series of metabolically and histologically well-characterized NASH patients and made the key observation that raising serum adiponectin levels were inversely correlated with hepatic fat content in advanced NASH while increases of certain bile acid (BA) species may contribute to this phenomenon.
Adiponectin is a 247 amino acid protein secreted by adipocytes belonging to the collagen superfamily and forming low and high molecular weight complexes in blood, the latter being the most biologically active with a key role in glucose and lipid metabolism.6 In liver, adiponectin signals by way of two distinct receptors, adipoRI and adipoRII,7 which have distinct biological roles: RI controls AMP-activated protein kinase (AMPK) activation responsible for the regulation of de novo lipogenesis and fatty acid (FA) oxidation, while RII controls peroxisome proliferator activated receptor alpha (PPARα) and thereby counteracts inflammation and oxidative stress (Fig. 1).8 In addition, using high-throughput RNA sequencing in mouse, adiponectin was shown to regulate glycolysis, cholesterol, triglycerides content, and FA synthesis by way of hepatocyte nuclear factor 4 alpha (HNF4α) (Fig. 1).9 Therefore, studies on the role of adiponectin in advanced NASH patients have great potential to reveal new molecular mechanisms responsible for progression from NASH to cirrhosis.
The current study by Van der Poorten et al.5 established that adiponectin levels were paradoxically rising while hepatic fat declined. Notably, high serum adiponectin was the strongest predictor of burnt-out NASH. At first glance this finding may be surprising, since reduced adiponectin levels have been implicated in pathogenesis of NAFLD.10 In order to explore whether serum adiponectin levels were functionally relevant for hepatic lipid metabolism, the authors demonstrated a correlation between hepatic adiponectin staining and AMPK and acetyl-CoA carboxylase 1 and 2 (ACC1/ACC2) phosphorylation. Notably, phosphorylation of ACC2, the main ACC isoform in human liver, inhibits this enzyme and reduces its product, malonylCoA, a CPT1α inhibitor, therefore indirectly promoting FA oxidation.11 Previous studies in patients with NASH revealed that adipoRI expression was stable, while adipoRII was reduced concomitantly with reduced hepatic adiponectin expression.12 Given the specific roles of adipoRI and RII, their relative amounts in advanced NASH may be important to estimate the net response. Since adipoRI remains stable and regulates AMPK, it is also remarkable that FA synthesis and its main regulator SREBP1c are inhibited in burned-out NASH patients.13 It is thus possible that adiponectin activation of adipoRI leads to AMPK activation which subsequently inhibits SREBP1c expression by phosphorylation of a serine residue near the cleavage site of SREBP1c, thereby repressing endogenous FA synthesis and forcing lipid droplet catabolism in the liver (Fig. 1). Conversely, low adipoRII expression may concomitantly promote oxidative stress and inflammation (Fig. 1). Unfortunately, potential changes of hepatic adipoRI or RII receptors expressions were not addressed in the current study.
A key question concerns potential mechanisms which might cause elevated adiponectin levels in advanced NASH. Adiponectin levels are known to be elevated in experimental models and patients with liver cirrhosis,14 with the highest levels being observed in cholestasis,15 suggesting a potential link to biliary constituents such as BAs. Previous studies established that adiponectin levels correlated with fibrotic markers such as transient elastography, hyaluronate, and serum BA.16 Serum BA levels are increased in NASH patients and correlate with disease progression.17 Adiponectin is secreted into the bile, which represents an important way of elimination, as reflected by increased levels in cholestatic patients and bile duct-ligated mice.15 It is therefore plausible that parallel increases of serum adiponectin and BA levels might simply reflect progressive liver dysfunction leading to burnt-out NASH.
Apart from their detergent properties in lipid digestion, BAs have more recently been recognized to possess additional hormonal actions that control a range of metabolic and immune functions throughout the body by way of the farnesoid X receptor (FXR) and the G-protein coupled BA receptor (GPBAR1/TGR5).18 As such, BA-activating FXR and TGR5 regulate cholesterol, triglyceride and glucose metabolism, as well as energy expenditure.19 Therefore, another interesting observation from the current study comes from its insights in BA composition: although no changes of primary BA cholic (CA) and chenodeoxycholic (CDCA) as well as ursodeoxycholic acid (UDCA) levels were observed, another secondary BA deoxycholic (DCA) correlated with adiponectin levels. DCA is formed after bacterial 7 alpha dehydroxylation of CA in the colon and is a potent natural TGR5 agonist20, 21 and a ligand activating FXR.22-24 Using mouse 3T3 cells as adipocytes, the authors could further demonstrate that both TGR5 and FXR activation was able to induce adiponectin expression. However, it must be kept in mind that the presence of FXR in adipose tissue may be questionable25 and that in addition to adipocytes also inflammatory cells could significantly contribute to TGR5 expression within fat.21 Since TGR5 and FXR have different affinities for DCA, the absolute serum concentrations would have been of interest in order to estimate which receptor may be most likely involved. Moreover, the possibility to activate TGR5 and FXR in vivo by specific ligands and using specific knockout mice should help to decipher in the near future which receptor plays a key role in regulation of adiponectin expression. Whether such agonists regulate adiponectin in humans could be addressed in ongoing and future clinical trials with FXR and TGR5 activators. On the other hand, it is important to consider that high adiponectin levels are associated with increased cardiovascular mortality despite improved inflammatory, atherogenic, and insulin-sensitizing effects,26 which might result from adiponectin resistance.27
It is known that DCA represses endogenous BA synthesis by way of Cyp7a1, but without suppressing cholesterol synthesis, in contrast to CDCA.28 Increased hepatic cholesterol synthesis promotes progression of NAFLD.29 Therefore, a DCA increase in patients with advanced burnt-out NASH might also contribute to deterioration of the liver condition by failing to repress the endogenous cholesterol synthesis (Fig. 1). In addition, a high fat diet is known to increase DCA levels in mouse, which in turn increases the intestinal permeability and thus propagates inflammation.30 It is therefore possible that a fat-enriched diet in human favors ectopic fat storage in the liver and DCA formation, which then keeps endogenous cholesterol synthesis at a high level and promotes intestinal leakage by enhancing bacterial translocation promoting inflammation and fatty liver development.
The identification of DCA as an important BA in NASH patients in the current study could also indicate a potential role of the gut flora. In the gut, FXR maintains epithelial barrier integrity by induction of multiple genes involved in intestinal mucosal defense against inflammation and microbes, which, together with direct antibacterial detergent actions, help to control the gut microbiota.18 Conversely, the gut microbiota is also able to alter BA metabolism and signaling by way of deconjugation and dehydroxylation, resulting in formation of secondary more hydrophobic BA with altered ligand binding properties for TGR5 and FXR. Patients with fatty liver display abnormal small intestine bacteria overgrowth and leaky intestinal tight junctions that could increase bacterial translocation and promote formation of secondary BA.31 Since adiponectin is also an immune-modulatory protein promoting macrophage polarization towards an antiinflammatory phenotype,32 it is tempting to speculate that adiponectin rise in advanced NASH patients might be part of a homeostatic response not primarily targeting lipid metabolism, but rather attempting to restore an adequate immune control in order to avoid or limit further liver injury, a mechanism that may be hampered by low adipoRII expression.
Taken together, this important work expands our knowledge on how fat might disappear from the liver in advanced NASH patients by uncovering a previously unknown link between adiponectin and BA metabolism. The findings of this study expand our understanding of the emerging hormonal actions of BA in fine-tuning hepatic glucose and lipid metabolism through activation of FXR and TGR5.18 The availability of recombinant adiponectin, or novel BA-based therapies targeting hormones such as adiponectin, might constitute an interesting therapeutic option in the earlier stages of NASH limiting overshooting inflammation.33