Differences in innate immune signaling between alcoholic and non-alcoholic steatohepatitis

Abstract

The similar histopathological characteristics of alcoholic steatohepatitis (ASH) and non-alcoholic steatohepatitis (NASH), and the crucial role of the innate immune response in both conditions may lead to the assumption that ASH and NASH represent the same pathophysiological entities caused by different risk factors. In this review paper, we elaborate on the pathophysiological differences between these two entities and highlight the disease-specific involvement of signaling molecules downstream of the Toll-like receptor 4, and the differential mechanism by which the inflammasome contributes to ASH vs. NASH. Our findings emphasize that ASH and NASH have disease-specific mechanisms and therefore represent distinct biological entities. Further studies are needed to dissect the emerging differences in pathogenesis of these two conditions.

Introduction

Liver diseases represent a significant cause of morbidity and mortality worldwide, ranking as the ninth leading cause of death (1-3). Only second to viral hepatitis, alcoholic liver disease (ALD) and non-alcoholic fatty liver disease (NAFLD) represent the most prevalent liver diseases in the US and developed countries (4-6). Both entities have a broad clinical spectrum, ranging from simple steatosis to steatohepatitis with or without fibrosis, cirrhosis, and hepatocellular carcinoma. Steatosis, observed in simple ALD and in NAFLD is a benign and self-limited condition, but in 10-20% cases, the condition progresses to alcoholic steatohepatitis (ASH) or non-alcoholic steatohepatitis (NASH), which share a component of liver inflammation and injury mediated by the innate immune response (7). This is of a clinical importance because inflammation determines the long-term prognosis of patients with these diseases, whereas steatosis per se does not appear to have an adverse impact on long-term outcome (8-11).

The concept of dysregulated innate immunity as an indispensable component of ASH and NASH is supported by the findings that patients with ASH have increased antibodies against Escherichia coli in plasma (12), patients with NASH have increased serum antibodies against endotoxin (13), and that consumption of alcohol or intake of a high-fat or high-carbohydrate diet leads to an increase in gut-derived endotoxin in the portal circulation, activating resident liver macrophages to produce several pro-inflammatory cytokines (14-18). Recognition of Toll-like receptors (TLR) as the key components involved in activation of the innate immune system enabled substantial progress in understanding the mechanisms mediating ASH and NASH.

Gut-derived bacterial components are critical in the pathogenesis of ASH and NASH

Due to its unique blood supply via the portal system, the liver receives blood from the intestine, exposing hepatocytes and liver immune cells not only to nutrients but also to gut-derived microbial products, including the lipopolysaccharide (LPS, endotoxin), a component of Gram-negative bacterial wall (19). Multiple lines of evidence support the hypothesis that gut-derived endotoxin is involved in ASH and NASH. First, it has been shown that excessive intake of alcohol increases gut permeability of normally non-absorbable substances (11). Second, intestinal Gram-negative bacteria, as well as blood endotoxin, are increased in acute (12, 13) and chronic (12, 14, 15) alcohol feeding models, and in human and animal studies of NAFLD/NASH (14,20-22). The mechanisms involve bacterial overgrowth, increased intestinal permeability and translocation of endotoxin (23-26), which is increased 5- to 20-fold in the serum of patients with ASH (8, 16), 3-fold in healthy individuals on a high-fat diet (14), and 6- to 20-fold in individuals with NAFLD (21,22), compared to normal subjects. Third, intestinal sterilization with antibiotics or administration of probiotics resulted in decreased LPS levels and reduced liver inflammation, injury and fibrosis in ASH and NASH in experimental settings (25-31).

Activation of Kupffer cells has been identified as one of the key elements in the pathogenesis of ASH and NASH. Kupffer cells are the largest population of tissue macrophages, predominantly distributed in the lumen of hepatic sinusoids and exhibit endocytic activity against blood-borne materials entering the liver (10, 24). Triggering of toll-like receptor signaling drives Kupffer cells to produce inflammatory cytokines and chemokines and to initiate the inflammatory cascade (25). Indeed, the essential role of Kupffer cells as a central component of the pathomechanism of ASH or NASH has been demonstrated in studies in mice and rats that show that inactivation of Kupffer cells with gadolinium chloride or liposomal clodronate can almost fully ameliorate inflammation, steatosis and damage in ASH and NASH (24,32-34).

The role of Toll-like receptor 4 in the pathogenesis of ASH and NASH

The innate immune system recognizes conserved pathogen-associated molecular patterns, which are released during bacterial multiplication or when bacteria die or lyse (35), through pattern recognition receptors, including Toll-like receptors (TLRs) (36). For example, TLR4 recognizes LPS from Gram-negative bacteria, and is a potent activator of innate immune responses through its binding to the TLR4 complex via the co-receptors CD14 and MD-2 (37). Activation of Kupffer cells via a TLR4-dependent mechanism plays a crucial role in the pathogenesis of ASH and NASH (15,24,28,38-40). Alcoholic liver injury was prevented in C3H/HeJ mice (41), which have functional mutation in the TLR4 gene and have a defective response to bacterial endotoxin, or in mice with a genetic deficiency of TLR4 (42,43). Similarly, deficiency in TLR4 prevented development of NASH (24,40,44). Prevention of ASH or NASH-associated liver inflammation and injury in TLR4-deficient mice was associated with decreased expression of inflammatory cytokines, compared to wild-type mice.

Different pathways mediate the pathogenic effects of TLR4 signaling in ASH and NASH

TLR4 is unique among TLRs in its ability to activate two distinct pathways. One pathway is activated by the adaptors TIRAP and MyD88, which leads to activation of NF-κB and to the inducion of inflammatory cytokines. The second pathway (MyD88-independent) is activated by the adaptors TRIF and TRAM, which activates the TBK/IKKε kinase and the interferon regulatory factor 3 (IRF3) to induce Type I interferons (IFNs) as well as NF-κB activation (45,46). The two TLR4-dependent signaling pathways are induced sequentially and the TRAM-TRIF pathway is only operational from early endosomes following endocytosis of TLR4 (47).

Both MyD88-dependent and MyD88-independent pathways of TLR4 signaling were activated in mouse models of ASH or NASH, as documented by increased serum and liver inflammatory cytokines, increased nuclear binding of NF-kB to its DNA response element, and upregulation of Type I IFNs and interferon-stimulated genes in the liver (42,48,49). In addition, intraperitoneal administration of LPS to mice fed alcohol or steatogenic diet further activated both branches of the TLR4 pathway (42,44,50). Based on this data, it would be tempting to hypothesize that a similar biological scheme determines responsiveness to LPS in ASH and NASH. However, our studies do not support this notion.

Using the Lieber-DeCarli model of ASH, we observed that alcohol-fed mice deficient in MyD88 exhibited the same extent of inflammation, steatosis and injury as their wild-type controls, which contrasted with a full protection from ASH in TLR4-deficient mice (42). Further analyses showed that deficiency of MyD88 did not abrogate activation of NF-κB in the liver, and that WT or MyD88-deficient but not TLR4-deficient mice on an ethanol diet demonstrated upregulation of Type I IFNs and IFN-dependent genes in whole livers and in isolated Kupffer cells. These data suggested that TLR4, but not MyD88, leads to activation of signaling mechanisms, including the NF-κB pathway, during the development of ASH. Furthermore, this data, along with the findings of others (46) suggested a functional activation of the MyD88-independendent, IRF3-dependent pathway. We confirmed this hypothesis and observed abrogation of Type I IFN signaling along with a complete protection from alcohol-induced inflammation, steatosis and damage in alcohol-fed, IRF3-deficient mice, compared to alcohol-fed wild-type controls (48). Thus, our data demonstrated that the pathogenic effect of TLR4 signaling in ASH is mediated via the TRIF/IRF3-dependent, MyD88-independent pathway.

Similar to ASH, there is ample evidence supporting the important role of TLR4 signaling, including NF-kB activation and upregulation of inflammatory cytokines in the pathogenesis of NASH (24,40,44,51-53). In contrast to the mechanisms involved in ASH, there seems to be a crucial role of MyD88-dependent signaling in NASH. This observation is based on data demonstrating that inflammation, steatosis, liver damage and fibrosis were remarkably inhibited in MyD88-deficient mice fed choline-deficient steatogenic diet ((54) and G. Szabo, unpublished data). The role of the MyD88-dependent pathway was further supported by a significant protection from NASH that was observed in mice deficient in TLR9 which requires MyD88 for its downstream signaling (54,55). In contrast, although deficiency of IRF3 in mice abrogated induction of Type I IFNs, it did not provide any protection from NASH-associated liver inflammation, steatosis or damage (G. Szabo, manuscript in preparation).

The differential contribution of MyD88-dependent and MyD88–independent pathways of TLR4 signaling in the pathogenesis of ASH and NASH may be attributable to multiple factors. For example, the development of NASH, in contrast to ASH, involves insulin resistance and an endocrine crosstalk between adipose tissue and the liver. It has been shown that adiponectin, an anti-inflammatory adipokine secreted by adipose tissue, inhibits the TLR4/MyD88-dependent pathway in macrophages (56). A recent meta-analysis demonstrated approximately 35% decrease of serum adiponectin in patients with NAFLD, and more than 50% decrease in patients in NASH (57). In contrast, reports on the relationship of adiponectin and ASH show either increase (58-61), no change (62), or minimal decrease that poorly correlated with the extent of liver injury (63). Based on these reports demonstrating association of adiponectin levels with NAFLD/NASH vs. no correlation in ASH, we cannot exclude that downregulation of adiponectin in NAFLD/NASH may contribute to inflammatory signaling in liver macrophages with preferential induction of MyD88-dependent pathways. Therefore, signaling from the adipose tissue could potentially modulate the preference for a signaling pathway downstream of TLR4.

Another factor contributing to the differential induction of TLR4 downstream pathways in ASH and NASH may relate to the differences in dynamics between these two entities. Although both of them take years to develop in humans, animal models suggest that excessive consumption of alcohol may induce liver inflammation at an earlier timepoint than consumption of steatogenic diet. For example, it takes only one intragastric gavage of ethanol to elicit significant liver steatosis in mice ((64) and G. Szabo, unpublished observations), or less than 7 days of the ethanol-containing Lieber-DeCarli diet to initiate liver inflammation (65), but it takes at least 18-24 weeks of feeding with high fat/Western-style diet or the choline-deficient amino acid-defined diet (CDAA, (54)) to induce liver inflammation in mice (51,66). Although artificial diets such as the methionine-choline deficient (MCD) diet induce inflammation within a week, these diets represent experimental models for mechanisms involved in NASH but do not necessarily reflect liver disease development in humans ((49,66) and G. Szabo, unpublished observations). Therefore, it cannot be excluded that different pathways may be responsible for early vs. late stages of pathogenesis of ASH and NASH. This notion is supported by our findings that pharmacological blocking of IL-1 receptor, which signals through MyD88, could achieve a pronounced protective effect in mice with advanced ASH (67), and that deficiency of MyD88 or IL-1 signaling showed protective effect only in later stages of fatty liver disease in mice (51,54,66,68).

Differential activation of the inflammasome and IL-1 signaling in ASH and NASH

In the pathogenesis of ASH and NASH, activated Kupffer cells exert their pathogenic effects predominantly via inflammatory cytokines, such as TNF-α, IL-1β, IL-8 or MCP-1 (51,53,69,70). Although TLR4-dependent mechanisms are involved in upregulation of inflammatory mediators, IL-1β is specific because it is produced as inactive pro-IL-1β and requires inflammasomes for processing. Caspase-1, the effector component of the inflammasome, cleaves pro-IL-1β into the bioactive IL-1β (71), which acts in an autocrine/paracrine manner via the Type-I IL-1 receptor (IL-1R1). The activation of IL-1R1 is inhibited by its binding to the IL-1 receptor antagonist (IL-1Ra), a naturally occurring cytokine whose function is to prevent the biologic response to IL-1 (72). Studies have demonstrated a pathogenic role of IL-1 signaling in the murine model of NASH (51,54), upregulation of inflammasome components in the liver and increased serum levels of IL-1Ra in patients with NASH (66,73) and increased levels of IL-1β in patients with ASH (74). However, there was no data supporting the causal role of IL-1 signaling in ASH, and only limited data on the cellular source and mechanism of IL-1β activation in NASH.

In our studies, we observed that inflammasome and IL-1β were activated in ASH, as documented by increased expression of inflammasome components NALP3, ASC and caspase-1 in the livers of alcohol-fed mice, and by increased activity of liver caspase-1 and elevated levels of cleaved IL-1β in the liver and in the serum (67). Deficiency of inflammasome components ASC or caspase-1, significantly ameliorated alcohol-induced liver inflammation, steatosis and damage. Similar protection was observed in mice deficient in IL-1R1 which lack IL-1 signaling, and in mice treated with recombinant IL-1Ra which inhibits IL-1 signaling (67).

Similar to ASH, the methionine-choline deficient diet (MCD)-based mouse model of NASH demonstrated activation of Caspase-1 in the liver and increased levels of cleaved IL-1β in the liver and in the serum after 6 weeks of treatment (66,68). Using the high-fat diet model of NASH, we observed that Caspase-1 and IL-1β became activated at a later timepoint of 9 months along with increased inflammation, but not at 4 weeks when liver pathology was dominated by steatosis only (66). This finding contrasted with our ASH data which demonstrated that inflammasome activation occurs very early in the course of alcohol treatment (67). Furthermore, deficiency of caspase-1 significantly ameliorated only liver inflammation induced by the MCD diet, but did not alleviate liver damage (68). Some protection from MCD diet-induced liver damage was observed in mice lacking IL-1α or IL-1β; however, this protection was observed only after 18 weeks of the experiment and not at earlier time points (51). In the context of early activation of the inflammasome in ASH and a significant protection from all components of alcoholic liver disease observed in mice deficient in inflammasome components or IL-1 signaling, the available data suggested differential role of inflammasomes in the pathogenesis of ASH and NASH.

In search for mechanisms that would explain this discrepancy, we investigated the cellular source of activated inflammasome and IL-1β. Both in ASH and NASH, the baseline levels of caspase-1 protein or pro-Casp-1, Asc, Nlrp3 and pro-IL-1b mRNA were substantially higher in liver immune cells, compared to hepatocytes (66,67). Specific for ASH, analysis of liver immune cells or primary hepatocytes isolated from alcohol-fed mice showed that alcohol increased the active fragment of caspase-1 and IL-1β only in liver immune cells but not in primary hepatocytes. As this data suggested that liver immune cells were the predominant cell type that activates caspase-1 and IL-1β in ASH, we generated caspase-1-chimeric mice using a combination of clodronate-mediated Kupffer cell depletion, irradiation and bone-marrow transplantation. Using this model, we confirmed our hypothesis that caspase-1 expressed in Kupffer cells was involved in alcohol-induced liver inflammation, steatosis and injury, and we did not find any evidence for a pathogenic role for caspase-1 in liver parenchymal cells in the development of ASH (67).

In addition to Kupffer cell-specific inflammasome activation in ASH (67), we observed that activation of the inflammasome occurred also in isolated hepatocytes in NASH (66). Specifically, primary hepatocytes isolated from the MCD-fed mice had increased expression of NALP3, ASC, caspase-1 and pro-IL-1b mRNA (49,66). Taking into account that fatty livers had elevated expression of inflammasome components and that this process occurred in hepatocytes which accumulate lipids, we tested whether fatty acids exert any effects on inflammasome in hepatocytes. We observed that in vitro treatment of primary mouse hepatocytes with palmitoic acid, a saturated fatty acid, resulted in upregulation of the inflammasome component Nalp3, priming of caspase-1 for subsequent activation by LPS and induction of IL-1β secretion. Using the pan-caspase inhibitor ZVAD, we demonstrated that these events were caspase-dependent, and we also showed that they were caused by saturated fatty acids whereas non-saturated fatty acids had no effect. We further showed that hepatocytes exposed to palmitoic acid produced inflammasome-mediated danger signals, which in turn activated liver macrophages in a caspase-dependent manner (66).

Taken together, our findings have outlined several differences in inflammasome/IL-1 signaling between ASH and NASH. First, activation of inflammasome seems to be an early event in ASH and late event in NASH. Second, deficiency in inflammasome components or absence of IL-1 signaling significantly ameliorates inflammation, steatosis and liver damage in ASH, whereas only protection from liver steatosis is consistently observed in NASH ((51,54,68) and G.S., unpublished data). Third, whereas inflammasome activation in ASH is specific to bone marrow-derived Kupffer cells, hepatocytes are involved in inflammasome activation in NASH. We hypothesize that this difference may be due to the predominance of cytotoxic free fatty acids and increased hepatocyte lipoapoptosis in NASH, compared to ASH in which the majority of fatty acids in hepatocytes is in esterified, less toxic form (75).

Conclusion

In spite of comparable histopathological characteristics of ASH and NASH, their similar pattern of progression to advanced liver disease, and the crucial role of innate immune signaling in both conditions, it is unlikely that the same immunopathogenic mechanisms contribute to ASH and NASH. Further studies are needed to dissect the emerging differences in pathogenesis of these two conditions.