Inflammatory Signals shift from adipose to liver during high fat feeding and influence the development of steatohepatitis in mice

Increased adiposity with the consequence of chronic low-grade inflammation and insulin resistance or type 2 diabetes has been linked to the development of nonalcoholic fatty liver disease (NAFLD). Currently, up to 30 percent of the general population is affected by NAFLD with 35 to 50 percent of obese adults also being diagnosed with nonalcoholic steatohepatitis (NASH). NAFLD has been described as the emerging clinical problem for the obese patient in the 21^st century [1]. The pathways that are active in promoting this disease process in the liver both in humans and in mouse models are poorly understood and are an active area of research.

There are a number of observations in the literature linking adiposity with inflammation and increased liver disease. Adipose tissue from obese people contains an increased number of CD68⁺ macrophages with a pro-inflammatory phenotype [2]. In insulin-resistant patients with fatty liver disease, there is a significant upregulation of genes involved in fatty acid partitioning and binding proteins, monocyte recruitment and inflammation [3]. Obese mice demonstrate a significant increase in plasminogen activator in the fatty liver [4]. Likewise, the absence of CCR2 protects the liver against fat accumulation in the diet-induced obese mouse [5].

In the attempt to model the human disease process in rodents, researchers have used several versions of the Western diet and have found differences in severity of disease and times of disease onset depending upon the type of fats used for feeding. Mice fed diets high in trans fats combined with high fructose in the drinking water develop very aggressive liver disease within two months whereas mice fed only 20% of calories from high fat develop liver disease in nine months [6, 7]. The genetic background of the rodent (C57BL/6 versus DBA/2) as well as cholesterol content of the diet and even the presence of endotoxin has been documented to strongly influence the development pattern of liver disease [8, 9]. Zheng et al. [10] in our institution use a rodent model which incorporates a 45% fat diet with 0.12% cholesterol to reflect approximate percentages found in the Western diet. This rodent model has all the hallmarks of obesity, insulin resistance, and liver steatosis plus it offers the further advantage of proven use for the investigation of therapeutic drugs relevant to these diseases, such as ezetimibe [10].

As a prelude to the use of the model in other drug studies, we attempted to determine the molecular pathways that were activated in this mouse model of high fat and cholesterol (HFC) feeding as the syndrome progressed towards liver steatosis and fibrosis. We used a sensitive and high throughput technology, Taqman Low Density Array (TLDA) to study message expression profiling of 92 genes representing macrophage-associated, inflammation-related and metabolism-driven genes in various tissues, including the adipose tissues and liver at 6 weeks, midway at 16 weeks and at 26 weeks post-HFC feeding. We report here that there is an initial upregulation of genes in the epididymal adipose tissue that is accompanied by a relatively quiescent liver profile at 6 weeks post-HFC followed by a dramatic shift in emphasis away from the epididymal adipose tissue to liver tissue gene activation at 16 weeks and 26 weeks. Capturing changes in gene expression profiles from different organ systems as disease progression of the liver is actively occurring will allow valuable information on molecular mechanisms leading to NAFLD and NASH to be gathered in animal models of obesity and will lead to the identification of new therapeutic targets.

To qualify our animal model as described previously by Zheng et al. [10] we have characterized the animals by tracking their body weight changes and the levels of serum mediators and cytokines throughout the time course. The percent body weight increased progressively in the HFC-fed mice over the 6 to 16 week study period and was maximal at 26 weeks post-HFC (Figure 1A). This body weight increase was accompanied by an increase in fat mass (gms) determined by MRI (Figure 1B). There was no effect of diet treatment on lean body mass. The HOMA index (Table 1) indicates that the high fat fed mice developed a significant degree of insulin resistance at the time points measured for this experiment. The epididymal fat pad measured at 6 weeks was the organ most strikingly affected when compared to the chow-fed animals. However, as the experiment progressed to 16 and 26 weeks, the epididymal fat pad weight as a percent of body weight actually decreased (Table 1). The liver weight (expressed as a percent of body weight) of the 6-week HFC-fed mice was unchanged from chow-fed controls; however, the liver weight of 16- and 26-week HFC-fed mice showed a continuous increase relative to the chow-fed mice. This increase in liver weight at 16 and 26 weeks was accompanied by an increase in the serum levels of alanine aminotransferase (ALT), indicative of progressive liver damage (Table 1).

We measured a variety of serum cytokines and mediators from these animals at the observation points. We found that there was a large degree of variability in these animals despite pre-selection for diet-induced obesity (DIO). We routinely kept the animals on a HFC diet for 3 weeks prior to entrance into the experimental cohorts to ensure that all animals chosen had at least a 30% increase in body weight when compared to chow-fed mice. Of the adipokines measured, serum leptin levels continually increased over time (Table 1). Adiponectin decreased only at 16 weeks of HFC feeding. Of the chemokines tested, MCP-1 (CCL2) was elevated throughout the observation periods in the HFC-fed mice; KC levels although higher than those of the chow-fed mice were not significantly elevated until 26 weeks post-HFC. Of the pro-inflammatory cytokines measured, IL-6 showed a modest increase at 16 and 26 weeks post-HFC. We did not obtain appreciable increases in circulating levels of GM-CSF, TNF-α, IL-12p70 and IL-1β in these HFC-mice. Serum amyloid A (SAA) levels were variable at 6 and 16 weeks post-HFC but were significantly elevated in the HFC-fed mice at 26 weeks post-HFC. IL-10 levels were increased in the serum of the HFC-fed mice at 16 weeks but were highly variable. At 26 weeks, IL-10 levels were more consistently elevated over the chow-fed controls. These measurements over the course of HFC feeding demonstrated that there was an inflammatory milieu in these mice.

Our work on gene activation in the HFC-fed mouse model is an attempt to accurately predict the sites of drug intervention and to possibly discover new targets in the mechanisms leading up to the induction of overt NASH disease. This paper describes a kinetic study for the direct temporal and spatial comparison of gene activation in a variety of organs as compared to age-matched chow-fed controls. First, a time dependent, diet-induced development of liver histopathology reminiscent of NASH, including steatosis, fibrosis and inflammatory infiltrates, was observed in the animal model used in our study of gene expression. Analysis of temporal inflammatory gene expression in multiple organs was carried out by TLDA. The epididymal adipose, mesenteric adipose and liver tissues show definitive time-dependent gene activation. In particular, those genes representing inflammatory mediators change dramatically over time. Gene activation to a various degree was also observed in other organ systems such as the blood cells, spleen, lymph nodes, gastrocnemius muscle, or pancreas but these gene patterns do not appear in any discernible pattern that would impact the induction of liver disease.

We report here that gene activation occurs early in the epididymal adipose with the peak activation at 6 weeks post-HFC. This activity in this particular adipose tissue is followed by a definitive increase in inflammatory signals in the liver at 16 weeks and 26 weeks post-HFC feeding. On the contrary, activated genes were downregulated in the epididymal fat pad at these later time points. This switch in activation profile from epididymal adipose tissue to liver as determined by quantitative RT-PCR was corroborated by the finding that there were significant increases in CD11b⁺ or CD11c⁺ macrophages in the liver at 16 weeks and 26 weeks post-HFC as well as by an increased accumulation of fatty droplets in the liver. These "activated macrophages" were not seen in the 6 week HFC livers when evaluated versus age-matched chow-fed control mice nor were there as many fatty droplets at this early time point. This increase in cellularity in the liver also appears in human disease. Genes involved in monocyte/macrophage recruitment are over-expressed in the livers of insulin-resistant human patients [3] and it is well-established that macrophages will accumulate both in adipose and liver under the influence of inflammatory signals [15, 16]. Furthermore, the reduction in gene activation observed in the epididymal adipose tissues at the later time point was accompanied by a decreased number of macrophages in this fat pad at 26 weeks post-HFC. These data suggest that a trigger for induction of inflammation was first set off in the adipose tissue and then sent out to other organ systems as fat feeding continued over time. This adipose-initiated signal sets up a process which results in overt liver disease by 26 weeks.

TLDA results of the mesenteric adipose tissue from HFC-fed mice indicate that inflammatory mediators and cell activation signals are also induced in this fat depot. This adipose tissue, however, does not appear to shift as dramatically in gene activation profile as that of the epididymal fat pad. This continual generation of cytokine and mediator gene signals found within the mesenteric fat amplifies the shift toward enhanced gene activation of the liver in these HFC-fed mice. It has been reported that the omental fat pad in obese humans can serve as a steady-state generator of inflammatory mediators which will then impact the development of disease [17].

Within the panel of genes included in our analysis, which ranges from metabolic to cellular markers to inflammatory mediators, IL-1β is identified as one of the most relevant inflammatory mediator as the disease induction process shifts from the inflamed epididymal adipose tissue to the liver. This enhanced IL-1 signal is relevant to current research findings in human metabolic disease. In obese patients, mRNA levels of IL-1 receptor antagonist (IL-1Ra) and IL-1 receptor type 1(IL-1R1) are markedly upregulated in white adipose tissue. In addition, IL-1β gene expression was selectively increased in the visceral fat (but not subcutaneous fat) from obese subjects [18]. In patients with type 2 diabetes, the use of Anakinra (IL-1Ra) improved glycemia and beta-cell secretory function and reduced markers of systemic inflammation (CRP, IL-6) [19]. Recent positive clinical results from a small trial with high affinity anti-human IL-1β (XOMA 052) also support targeting IL-1β mediated inflammatory damage to pancreatic beta cells as a potential therapeutic approach for type 2 diabetes [20]. The use of TLDA technology applied to this mouse model can reveal further activated or dysregulated gene signals that will be relevant to human disease.

Excessive pro-inflammatory stimulation enhances the development of steatosis and NASH. In patients with acute and chronic liver diseases, production of IL-1 alters as the liver disease shifts in intensity from acute to chronic cirrhosis [21]. At present, it appears that a certain mix of cytokines plus a genetic predisposition to hepatic immune defects may be needed to facilitate progression of NASH to cirrhosis [22]. Our work demonstrates that there is very strong and relevant cross-talk between organ systems of the adipose tissue and the liver in our HFC mouse model which is also dependent upon the genotype of the mouse [23]. Use of the TLDA technology to track gene changes over a time line will help to unravel some of this immune conversation and to identify whether new therapeutics can make these gene signals quiescent.

In the present study, we utilized the mouse model with a diet containing 45% fat and 0.12% cholesterol (as found in the Western diet) to investigate the molecular signals that trigger insulin resistance, hyperinsulinemia and progression to liver steatosis and fibrosis in the diet-induced obesity (DIO) mouse model. We used a sensitive and high throughput technology, Taqman Low Density Array (TLDA), to study gene expression profiling of 92 important genes representing macrophage-associated, inflammation-related and metabolism-driven genes in multiple tissues at 6 weeks, 16 weeks and 26 weeks post high fat and cholesterol (HFC) feeding. We demonstrate from our analysis of the gene activation profiles that macrophage infiltration accompanied by severe inflammation and metabolic changes occurs in both adipose and liver tissues with a temporal shift in the levels of these signals depending upon the duration of HFC feeding. This gene activation initiates early at 6 weeks post-HFC feeding in epididymal adipose tissues. Activation signals then switch to the liver at 16 and 26 weeks post-HFC feeding. These findings of time dependent development of steatosis in the liver are supported by immunohistochemistry. Taken together, the evidences of gene expression profile, elevated serum alanine aminotransferase, increased liver to body weight ratio, and histological data support a progression driven by diet-induced inflammation towards nonalcoholic fatty liver disease and even nonalcoholic steatohepatitis in these HFC-fed mice within the time frame of 26 weeks.