NADPH oxidase–derived free radicals are key oxidants in alcohol-induced liver disease

Introduction

Alcohol is the third leading cause of preventable mortality in the US and contributes to about 100,000 deaths annually. The economic burden of alcoholism on the US economy, in great part due to health care expenditures, rose 42% to $148 billion from 1985 to 1992 (1, 2). Despite the fact that the population in general is well aware of the adverse and often fatal consequences of alcohol consumption, it is estimated that more than 7% of individuals 18 and over have problems with drinking. Alcoholic liver disease is one of the major drinking-related health problems and a primary cause of liver disease among Caucasians.

Histopathological features of alcoholic liver disease include fat accumulation and hepatitis, followed by fibrosis and cirrhosis. A major breakthrough in understanding the molecular mechanisms of these pathological changes was the development of an enteral animal model of alcoholic liver disease in the rat by Tsukamoto et al. (3). In the rat liver, when ethanol-containing diet is delivered continuously, pathological changes reflective of human alcoholic liver disease occur (i.e., fat accumulation, inflammation, and fibrosis). One hypothesis to account for the mechanism of alcohol-induced liver injury is that CYP2E1, induced predominantly in hepatocytes by ethanol, increases production of free radicals. Indeed, ethanol causes formation of α-hydroxyethyl radical in the Tsukamoto-French rat model (4); however, CYP2E1 is most likely not involved in this process, as no differences in liver injury and free radical formation were observed when wild-type and CYP2E1 knockout mice were compared (5, 6).

On the other hand, evidence has been presented in support of the hypothesis that Kupffer cells, the resident hepatic macrophages, play a key role in alcohol-induced liver damage (7, 8). Specifically, it is known that alcohol increases gut permeability for Gram-negative bacterial endotoxin. Endotoxin is a potent activator of Kupffer cells, which release toxic cytokines and inflammatory mediators (e.g., TNF-α) as well as reactive oxygen species. Indeed, Kupffer cells may be the source of oxidants (9). NADPH oxidase is a major oxidant-generating enzyme in activated macrophages (10). Here, therefore, we have adapted the enteral alcohol feeding model to the mouse (11) so that knockout technology can be used to test the hypothesis that oxidants from NADPH oxidase are involved in early alcohol-induced liver injury. This hypothesis was evaluated by comparing wild-type and NADPH oxidase–deficient (p47^phox knockout [–/–]) mice (12) treated continuously with ethanol.

Methods

Results and Discussion

Alcohol-containing or control diets were initiated 1 week after surgery to allow for complete recovery, and all mice were healthy during the enteral feeding period. No differences in behavior between wild-type and p47^phox–/– mice fed ethanol were observed. Mice gained weight steadily during 4 weeks of continuous enteral feeding of liquid diets with or without ethanol, indicating adequate nutrition (see Table 1). In both rats and mice, urine alcohol levels fluctuated in a cyclic pattern even though ethanol was infused continuously (3, 11). Recently, it was reported that the cyclic oscillation of the urinary alcohol level in rats fed ethanol involves fluctuation of body temperature and thyroid hormone release (25). Since similar patterns were observed here in p47^phox–/– mice, it is concluded that NADPH oxidase is not involved. Average urine alcohol concentrations over 4 weeks were 248 ± 17 mg/dl (mean ± SEM) in wild-type and 255 ± 22 mg/dl in p47^phox–/– mice (see Table 1). Serum ALT levels were about 30 U/l in wild-type mice fed a high-fat control diet for 4 weeks; however, values were increased significantly, about threefold, by enteral ethanol (see Table 1). In contrast, increases were blocked completely in p47^phox–/– mice fed ethanol.

After 4 weeks of the high-fat control diet, there were no pathological changes in the livers of wild-type mice (Figure 1a) or unexpected changes in p47^phox–/– mice fed a high-fat diet (Figure 1b) (12). Wild-type mice fed enteral ethanol developed severe steatosis in pericentral to midzonal regions, as well as inflammation and necrosis (Figure 1c), with a total pathology score of 5.4 ± 0.4 (see Table 1). A macro- or microvesicular pattern of fat accumulation was observed mainly in pericentral to midzonal regions, except for 1 to 3 layers of hepatocytes around central veins. In contrast, these pathological changes were prevented nearly completely in p47^phox–/– mice fed enteral ethanol (Figure 1d) (total pathology score, 0.6 ± 0.1).

Free radical adducts were barely detectable in bile from wild-type mice (Figure 2a) and p47^phox–/– mice (Figure 2b) using the spin trapping technique and ESR. In contrast, 4 weeks of enteral ethanol caused formation of radical adducts in wild-type mice (Figure 2c). Spectral simulation revealed the presence of two spectra with hyperfine coupling constants: (I) a^N = 15.8 ± 0.2 G and a_β^Η = 2.5 ± 0.2 G; and (II) a^N = 15.6 ± 0.2 G and a_β^Η = 3.2 ± 0.2 G. Spectrum I, which comprises 80% of the spectral intensity, is most likely derived from an approximately 50/50 mixture of α-hydroxyethyl and lipid-derived free radical adducts (4, 26). ESR studies of such systems using ¹³C-ethanol have demonstrated the similarity of the hyperfine coupling constants and the presence of both radical adducts following chronic ethanol administration using Tsukamoto-French protocol (4). Spectrum II, with its unique β hydrogen hyperfine coupling of 3.2 G, is most likely the POBN/CO₂•^– free radical adduct formed by the abstraction of a hydrogen atom from endogenous formate to give CO₂•^–, which then reacts with POBN (27). Importantly, all three above-mentioned free radical adducts are formed from attacks of oxidants on the respective parent molecule.

On the other hand, radical adduct formation was not increased in p47^phox–/– mice fed ethanol (Figure 2d; P < 0.05). Endotoxin-activated Kupffer cells and infiltrating neutrophils are two possible sources of oxidant production. However, several sets of evidence point to the NADPH oxidase in Kupffer cells as a likely source of free radicals due to alcohol treatment. First, it was shown that inactivation of Kupffer cells with GdCl₃ blocks formation of the α-hydroxyethyl radical in the chronic enteral ethanol feeding model in rats (4). Furthermore, the number of infiltrating neutrophils does not increase until 2–4 weeks of enteral alcohol feeding, and liver injury is prevented early in p47^phox–/– mice when few neutrophils are present (data not shown). Furthermore, it was shown that acute ethanol treatment activates NADPH oxidase in rat Kupffer cells (28). Indeed, ethanol activates LPS-stimulated superoxide production in Kupffer cells determined by SOD-inhibitable reduction of cytochrome c (29). In Kupffer cells isolated by standard procedures (30) from wild-type mice, superoxide was increased from 1.5 ± 0.2 nmol/10⁶ cells per 15 minutes to 3.2 ± 0.3 by acute ethanol treatment (5 g/kg, intragastrically for 21 hours). However, LPS did not increase superoxide production in cells from p47^phox–/– mice (control: 0.3 ± 0.2 nmol/10⁶ cells per 30 minutes; LPS [10 μg/ml]: 0.3 ± 0.1). Furthermore, the radical scavenger allopurinol blocks alcohol-induced liver injury in the rat (31).

Oxidants can function not only as toxic substances, but also as second messengers (32, 33). For instance, the transcription factor NF-κB is redox-sensitive and plays an important role in signaling (34). Since both alcohol and the peroxisome proliferator WY-14,643 activate NF-κB nearly exclusively in Kupffer cells in the rat (35), it was hypothesized that NF-κB is involved in alcohol-induced liver injury by stimulating TNF-α production in Kupffer cells. In wild-type mice fed a high-fat control diet for 4 weeks, NF-κB activity was minimal; however, activity was increased significantly, about twofold, by enteral ethanol in wild-type mice (Figure 3, a and b). In contrast, this increase was blocked completely in p47^phox–/– mice. Thus, the key transcription factor NF-κB is not activated by alcohol in NADPH oxidase–deficient mice.

Binding sites for NF-κB have been identified within the regulatory elements of genes for proinflammatory cytokines such as TNF-α (36). Recent evidence from this laboratory supports the hypothesis that TNF-α plays an essential role in early alcohol-induced liver injury. Specifically, anti–TNF-α serum reduces inflammatory cell infiltration and necrosis in the rat enteral alcohol model (37). Furthermore, alcohol-induced liver injury is present in wild-type mice fed enteral ethanol chronically, but not in TNF receptor-1 knockout mice (11). In this study, expression of TNF-α and IL-6 mRNA was increased in wild-type but not in p47^phox–/– mice fed ethanol-containing diet (Figure 3c). TNF-α stimulates sinusoidal endothelial cells to synthesize adhesion molecules (e.g., ICAM-1 and others) that are involved in inflammation (38–40). Further, liver injury in wild-type mice fed enteral ethanol was prevented in ICAM-1 knockout mice (Kono et al., unpublished observations). TNF-α is also implicated in hepatic steatosis caused by ethanol (41). In this study, hepatic steatosis observed in wild-type mice was prevented nearly completely in p47^phox–/– mice (Figure 1), suggesting that TNF-α production enhanced by oxidants is involved in fat mobilization and/or metabolism during alcohol feeding.

Collectively, the data presented here are consistent with the following sequence of events: ethanol, endotoxin, Kupffer cell CD14, NADPH oxidase, oxidants, NF-κB, TNF, cytotoxicity. Taken together, these results strongly support the hypothesis that oxidants from NADPH oxidase, most likely of Kupffer cell origin, play a key role in the pathogenesis of early alcohol-induced liver injury.