Nonalcoholic fatty liver disease (NAFLD) represents a major public health problem. One experimental model which has generated a significant body of evidence regarding potential mechanisms of NAFLD pathogenesis is the ob/ob mouse. Ob/ob mice, which are leptin deficient as a result of a spontaneous mutation in the leptin gene, exhibit a number of metabolic and inflammatory features which mimic human NAFLD,1 including insulin resistance, hyperlipidaemia, hepatic steatosis, and TNF-α elevation. One of the principal applications of the ob/ob mouse has been the identification of susceptibility of the steatotic liver to inflammatory insult (exemplified by the response to lipopolysaccharide) as a key factor in the development of the nonalcoholic steatohepatitis (NASH) variant of NAFLD.2 A number of immuno-regulatory abnormalities have been identified in ob/ob mice which may contribute to their increased susceptibility to inflammatory damage. These include selective depletion from the liver (but not other organs) of NKT cells, a key population of immuno-regulatory/effecter lymphocytes which, archetypally, express phenotypic features of both “classical” T cells (CD3) and natural killer (NK) cells (NK1.1 [CD161 in humans]),3, 4 and which in their most characteristic form show specificity, through a semi-invariant surface T-cell receptor, for highly conserved glycolipid antigens presented by the MHC class I homolog CD1d. NKT cells, which are specifically enriched within the liver, have characteristic cytokine release patterns (Th-1 dominant [IFN-γ], mixed and Th-2 dominant [IL-4] depending on the mechanism of stimulation) which endow, in addition to their effector function, significant immuno-regulatory properties.5 The observation that liver NKT cells are depleted in steatosis in ob/ob mice has led to the suggestion that these cells play a key role in mediating and/or regulating inflammatory effects critical to the development of NAFLD.
Although of potential value in the study of the pathogenesis of NAFLD, conceptual problems arise with regard to the ob/ob mouse as a model for human disease due to its markedly different leptin phenotype (absent versus elevated) and the fact that leptin is itself a key immuno-modulatory cytokine.6 There are, therefore, potential mechanisms whereby leptin deficiency could modulate the immune response independent of its effects on hepatic fat accumulation (Fig. 1). In the current edition of HEPATOLOGY Li and colleagues elegantly address the latter issue.7
Li et al. used a natural obesity/steatosis model to study the effects of hepatic steatosis on hepatic innate immune system function in leptin complete animals.8 C57Bl/6 mice fed a high-fat diet showed excess weight gain and the development of hepatic steatosis. Although total hepatic mononuclear cell levels were similar in the high- and low-fat diet groups, the percentage of hepatic (but not splenic) NKT cells was significantly reduced. Within both the hepatic T-cell and NKT fractions the numbers of cells showing cytoplasmic staining for TNFα and IFNγ were, conversely, increased in the high-fat diet group (and serum IFN-γ levels were elevated), suggesting Th1 skewing of the response phenotype resulting from induced NKT cell effects. Finally, the livers of obese mice appeared to be sensitized to lipopolysaccharide injury, presumably reflecting the augmented Th1-type inflammatory cytokine response. These observations suggest that the development of hepatic steatosis per se can be associated with significant changes in liver NKT cell function. This finding would be compatible with the NKT cell changes seen in the ob/ob mice occurring as a result of hepatic steatosis that occurs in these animals, rather than the specific absence of leptin. The findings do, however, raise a number of issues which will determine the eventual value of this model for the study of human NAFLD.
The first issue is the mechanism responsible for liver NKT cell “loss,” and Th-1 skewing of the residual cells, in obese C57Bl/6 mice. Theoretically, a reduction in liver NKT cells in obese C57Bl/6 mice could result from a decreased rate of NKT cell recruitment to, or development in, the liver, an increased rate of NKT cell death or emigration from the liver, a loss of surface markers identifying the cells as NKT cells or any combination of these effects. The liver recruitment aspect of NKT cell homeostasis (an area of considerable recent progress5, 9, 10) was not addressed in the Li study. Instead, the authors argue that increased cell loss is the dominant effect, with evidence presented to suggest increased NKT cell apoptosis and increased hepatic expression of IL-12 (postulated to be a promoter of NKT cell apoptosis). There is an emerging consensus, however, that NKT cells are in fact relatively resistant to activation-induced cell death.11 Previous reports of “loss” of NKT cells following in vivo activation are instead now thought to mainly reflect downregulation of surface NK1.1.12, 13 The equation of loss of NK1.1 expression with NKT cell loss highlights a limitation of the Li study in that a relatively simple approach to the phenotyping of NKT cells was adopted (FACS-based identification of CD3+ NK1.1+ cells). An alternative methodological approach (although one which is, in turn, susceptible to postactivation TCR downregulation) is the detection of NKT cells through their distinctive TCR by the use of glycosphingolipid (α-galactosyl-ceramide; a specific NKT cell ligand) loaded CD1d tetramers.14
An alternative (and non-mutually exclusive) explanation for the Li data would be that endogeneous IL-12 released by Kupffer cells (KC) at elevated levels in the context of obesity7, 15 acts as a cofactor for the stimulation of IFN-γ release (as opposed to IL-4 release which occurs in the absence of IL-12) by physiologically activated NKT cells, with the resulting “loss” of cells occurring as a consequence of postactivation surface phenotypic shift.16 If elevation of KC released IL-12 in response to steatosis were to prove to be a factor in human fatty liver development, its well-established ability to promote breakdown of self-tolerance may explain the increasingly recognised tendency towards autoantibody formation reported in NASH patients.17, 18 Interestingly, IL-15, an important NKT cell survival factor, the deficiency of which has been implicated in NKT cell “loss” in the ob/ob mouse,15, 19 was present at unchanged levels in obese C57Bl/6 mice.
The possibility that NKT cell activation is responsible, through activation induced cell death and/or postactivation phenotypic change, for “reduction” in hepatic NKT-cells in obese C57Bl/6 mice, and through cytokine release, for liver damage, raises the important question of the mechanism of that activation. Most previous work on NKT cell activation has used nonphysiological ligands (anti-CD3 and anti-TCR). Although the recent identification of α-galactosyl-ceramide has highlighted the potential importance of glycolipids as natural ligands for NKT, it is unlikely, given its marine sponge origin, that this agent is a physiological ligand in mice or humans. At present the identity of the invivo physiological ligand for NKT cells, the extent to which TCR-mediated as opposed to cytokine-driven mechanisms (such as via IL-12) are required for activation, and the extent to which different activation pathways result in different cytokine response phenotypes, remain areas of speculation. One potentially highly intriguing link between hepatic steatosis and NKT cell activation has emerged with the observation that microsomal triglyceride transfer protein, deficiency of which in mice is associated with hepatic steatosis, and functional polymorphisms in the encoding gene of which have shown significant associations with NASH in humans,20 plays a key role in the acquisition of glycolipid antigens by CD1d.21 One approach to dissecting out the mechanisms of NKT cell activation and loss in obese C57Bl/6 mice would be to ultilize NKT cell adoptive transfer and tracking methodologies in recombinant NKT cell–deficient mice in combination with NKT cell activation and appropriate cytokine blocking.
The second issue raised is that of the natural history of the relationship between steatosis development and NKT cell changes. It would be of particular interest, using this model, to explore the extent to which changes are reversible following dietary modification. This would have real potential significance for our understanding of the extent to which dietary approaches are likely to be effective in the management of NAFLD.
The third and perhaps key issue is the relevance, if any, of the findings made in the obese C57Bl/6 mouse to human NAFLD. Two issues are paramount here. The first is the extent to which the obese C57Bl/6 mouse mimics metabolically human NAFLD, and the second is the extent to which liver NKT cells in humans and mice share phenotypic and functional characteristics. With regard to the first of these issues, although Li et al. did not specifically address the issue of leptin levels in obese mice it has previously been demonstrated that a similar high diet protocol induces high leptin levels, hyperglycaemia and hyperinsulinaemia in the resulting obese C57Bl/6 mice, thereby replicating key metabolic changes seen in human NAFLD.8 With regard to the second issue, it is important to note that there are potentially significant differences between hepatic NKT cells in humans and mice. Human thymic NKT cell levels are typically significantly lower than those seen in mice and can vary markedly between individuals (reflecting the contrast between the outbred human population and the genetically homogenous inbred populations used almost exclusively in murine studies22). Even more marked inter-species differences are seen with regard to the liver, with NKT cells forming a significantly lower proportion of the liver lymphocyte pool in humans (c1%) than in mice. Moreover, human liver NKT cells appear to have distinct activation requirements and properties.23 These identified differences reflect, in part, practical problems faced in phenotyping human liver NKT cells (particularly the diverse “non-classical” (non-CD161 and non-invariant TCR/CD1d restricted) populations24 and the practical and ethical issues faced in obtaining sufficient quantities of normal and steatotic human liver to do the appropriate isolation and FACS studies (in situ studies give rise to their own specific practical problems). These practical problems notwithstanding, the areas of NKT cell numbers and physiology in human NAFLD are ones where study is badly needed.
In conclusion, the steatosis model outlined by Li and colleagues is of potential importance for two reasons. The first is that the demonstration of NKT cell modulation in response to a physiologically relevant dietary insult opens new and interesting avenues for the study of the physiology of hepatic NKT cells. The second is that it opens another window to our understanding of the pathogenesis of NAFLD by potentially delineating a first “hit” for NASH development.25 This will not only give rise to a model in which to study potential second “hits,” but also a setting in which to study therapeutic approaches (both physiological and, potentially, pharmacological) to reversal of the first “hit.”