Metabolomic profiling of a modified alcohol liquid diet model for liver injury in the mouse uncovers new markers of disease

Abstract

Metabolomic evaluation of urine and liver was conducted to assess the biochemical changes that occur as a result of alcohol-induced liver injury. Male C57BL/6J mice were fed an isocaloric control- or alcohol-containing liquid diet with 35% of calories from corn oil, 18% protein and 47% carbohydrate/alcohol for up to 36 days ad libitum. Alcohol treatment was initiated at 7 g/kg/day and gradually reached a final dose of 21 g/kg/day. Urine samples were collected at 22, 30 and 36 days and, in additional treatment groups, liver and serum samples were harvested at 28 days. Steatohepatitis was induced in the alcohol-fed group since a 5-fold increase in serum alanine aminotransferase activity, a 6-fold increase in liver injury score (necrosis, inflammation and steatosis) and an increase in lipid peroxidation in liver were observed. Liver and urine samples were analyzed by nuclear magnetic resonance spectroscopy and electrospray infusion/Fourier transform ion cyclotron resonance-mass spectrometry. In livers of alcohol-treated mice the following changes were noted. Hypoxia and glycolysis were activated as evidenced by elevated levels of alanine and lactate. Tyrosine, which is required for l-DOPA and dopamine as well as thyroid hormones, was elevated possibly reflecting alterations of basal metabolism by alcohol. A 4-fold increase in the prostacyclin inhibitor 7,10,13,16-docosatetraenoic acid, a molecule important for regulation of platelet formation and blood clotting, may explain why chronic drinking causes serious bleeding problems. Metabolomic analysis of the urine revealed that alcohol treatment leads to decreased excretion of taurine, a metabolite of glutathione, and an increase in lactate, n-acetylglutamine and n-acetylglycine. Changes in the latter two metabolites suggest an inhibition of the kidney enzyme aminoacylase I and may be useful as markers for alcohol consumption.

Introduction

In North America, liver disease due to alcohol consumption is an important cause of death in adults, although its pathogenesis remains largely unknown. The disease process is characterized by early steatosis, inflammation, and necrosis (often called alcoholic steatohepatitis). In some individuals who abuse alcoholic beverages, liver disease progresses to fibrosis and cirrhosis, which significantly impairs liver function, and can ultimately lead to the development of hepatocellular carcinoma (Purohit et al., 2005, Purohit and Brenner, 2006).

Much has been learned about the molecular mechanisms of liver damage by alcohol from experiments in model organisms, especially rats and knockout mice (Thurman, 1998). The Lieber–DeCarli liquid diets contain carbohydrates (11% of calories), protein (18%), fat (35%), and alcohol or maltose dextrin (36%) (Lieber and DeCarli, 1982). The diet is administered to animals ad libitum and, together with its low-fat and high-protein modifications, is readily available commercially which makes this model easy to use. While animals on Lieber–DeCarli diets have circulating blood alcohol levels around 200 mg/dl after as early as one week of feeding (Sibley and Jerrells, 2000) and liver steatosis develops, the injury has not been reported to progress much farther in mice. Thus, the Lieber–DeCarli model provides only a limited resource for studies of alcohol-related liver disease.

The continuous intragastric enteral alcohol feeding protocol in the rat, known as Tsukamoto–French model (Tsukamoto et al., 1984), was a major development in alcohol-induced liver injury research. Although this method requires animal surgery and is very labor intensive, it produces liver injury that is consistent with the human disease starting with steatosis, progressing to steatohepatitis and achieving fibrosis after 3 months of exposure in the rat (Tsukamoto et al., 1990). This force-feeding (through an intragastric cannula) protocol was also adapted to the mouse which allowed mechanistic studies in genetically engineered animals (Kono et al., 1999, Yin et al., 1999). Interestingly, alcohol must be delivered in a high-fat corn oil-based diet in order to facilitate the progression of the liver injury in rodents (Nanji et al., 1989). Recent studies have also implicated fish oil as a source of potentiation of injury in female as compared to males rats using a liquid diet which is formulated similarly to the enteral model (Donohue et al., 2007).

The advantages of the enteral model over liquid diet models include continuous delivery of alcohol 24 h a day which simulates binge drinking, robustness of the liver injury phenotype, and a characteristic cycling in urine alcohol concentrations due to diurnal variation in basal metabolism and hormones. The criticisms of the Tsukamoto–French model have been that it is labor intensive, does not follow a normal nocturnal feeding pattern for rodents, and that the diet is not nutritionally adequate for induction of cytochrome P450 to the same extent as a standard Lieber–DeCarli liquid diet (Ronis et al., 1991), although it has been reported recently that P450s are well induced in mice and rats after enteral alcohol treatment (Isayama et al., 2003, Bradford et al., 2005). Because the “success” of each model is assessed largely by its effectiveness to produce liver damage as the key endpoint, the development of an easy-to-use model that produces significant liver injury is desired.

Recently, it was shown that by feeding rats a high-fat corn oil-based modified Lieber–DeCarli diet, together with daily administration of a large dose of alcohol by gavage, produced liver injury similar to that shown by the Tsukamoto–French model after 8 weeks of treatment (Enomoto et al., 1999). Similar experiments were not successful in the mouse (N. Enomoto, personal communication), possibly due to the stress of daily gavage. Thus, this study was conducted to determine if by using corn oil as the source of fat in the standard Lieber–DeCarli liquid diet it is possible to achieve a liver injury phenotype more similar to human disease. We show that this protocol results in the development of steatohepatitis comparable to that in the Tsukamoto–French model. Furthermore, we utilized metabolomic analysis of the urine and liver extracts to assess the biochemical and molecular changes of the disease progression toward alcohol-induced liver injury.