|Home | About | Journals | Submit | Contact Us | Français|
Alcoholic liver disease (ALD) remains a major cause of morbidity and mortality worldwide. For example, the Veterans Administration Cooperative Studies reported that patients with cirrhosis and superimposed alcoholic hepatitis had a 4-year mortality of >60%. The poor prognosis of ALD implies that preventing disease progression would be more effective than treating end-stage liver disease. An obvious avenue of prevention would be to remove the damaging agent; however, the infamously high rate of recidivism in alcoholics makes maintaining abstinence a difficult treatment goal to prevent ALD. Indeed, although the progression of ALD is well-characterized, there is no universally accepted therapy available to halt or reverse this process in humans. With better understanding of the mechanism(s) and risk factors that mediate the initiation and progression of ALD, rational targeted therapy can be developed to treat or prevent ALD. The purpose of this review is to summarize the established and proposed mechanisms by which chronic alcohol abuse damages the liver and to highlight key signaling events known or hypothesized to mediate these effects.
Alcoholic liver disease (ALD) affects millions of patients worldwide each year. In 2004, 3.8% of all global deaths were as a result of alcohol (Rehm et al., 2009). The progression of ALD is well-characterized and is actually a spectrum of liver diseases, ranging from steatosis, to inflammation and necrosis (steatohepatitis), to fibrosis and cirrhosis, and eventually hepatocellular carcinoma (HCC) in some cases. The risk of ALD increases in a dose- and time-dependent manner with consumption of alcohol (Mann et al., 2003). However, only a minor proportion of even heavy drinkers develop the severe form of the disease, suggesting that other environmental (e.g., cigarette smoking, obesity or HCV infection) or genetic (e.g., gender or polymorphisms in key genes) factors contribute to overall risk (Day, 2000). Clinical management of ALD focuses predominantly on maintaining abstinence in the patient, and on treating sequelae associated with acute alcoholic hepatitis or cirrhosis (Diehl, 2002). Short of a successful liver transplant, the effects of decompensation (e.g., hepatorenal syndrome) usually lead to the death of the patient (Powell and Klatskin, 1968). Even if compensated cirrhosis is maintained (i.e., ‘stable cirrhotics’), the overall risk of developing hepatocellular carcinoma, which has a very low survival rate is increased ~20-fold by pre-existing cirrhosis (Gogel et al., 2000). Although the progression of ALD is well-characterized, there is no universally accepted therapy available to halt or reverse this process in humans except abstinence. With better understanding of the mechanisms (e.g., signaling pathways) that mediate the initiation and progression of ALD, rational targeted therapy can be developed to treat or prevent ALD. The purpose of this review is to summarize the established and proposed mechanisms by which chronic alcohol abuse damages the liver (see Figure 1) and to highlight key signaling events known or hypothesized to mediate these effects.
The liver is the main organ responsible for metabolizing ethanol. The major route of ethanol metabolism is the oxidation of ethanol to acetaldehyde. The dominant enzyme system involved in this process is alcohol dehydrogenases (ADH). However, cytochrome P450 systems, also called microsomal ethanol oxidizing system or ‘MEOS’ (mostly CYP2E1), and catalase also play crucial roles. ADH oxidizes ~80% of the ethanol, with MEOS mediating most of the remainder, and catalase oxidizing only a small component. Electrons from alcohol are transferred to NADP+ by ADH. By contrast, molecular oxygen (O2) is the recipient of the electrons from ethanol via the MEOS system. Lastly, hydrogen peroxide (H2O2) is reduced to water by catalase in the process of oxidizing ethanol (see Figure 2). The oxidation of ethanol to acetaldehyde is mediated by three distinct enzyme systems; aldehyde dehydrogenase (ALDH), however, is the only enzyme system that oxidizes acetaldehyde to acetate (Figure 2). Analogous to ADH, ALDH uses NAD+ as the electron acceptor for this reaction; however, ALDH is located in the mitochondria (Figure 2). Whereas the rapid removal of ethanol by the liver is clearly a protective mechanism, there are potentially toxic metabolic and biochemical processes of ethanol metabolism that could contribute to the development and progression of alcoholic liver disease, including induction of CYP2E1, production of toxic metabolites (e.g., acetaldehyde), and alteration of biochemical processes.
Whereas the relative role of the MEOS system to total alcohol metabolism is low a priori, CYP2E1 is robustly induced by alcohol and can contribute to a far greater amount of total alcohol metabolism in alcohol-dependent individuals (Lieber, 1997). This induction could be a protective effect at the organismal level, however, it might have side effects that contribute to hepatotoxicity. For example, CYP2E1 is localized in regions of the liver lobule that are damaged by alcohol. Furthermore, selective inhibitors of CYP2E1 partially inhibit liver injury caused by ethanol in animal models, supporting this hypothesis (Bardag-Gorce et al., 2000). Whereas these studies serve as proof-of-concept that CYP2E1 contributes to ALD, most CYP2E1 inhibitors, although selective for CYP2E1 among the P450s, often target other biomolecules that can confound the conclusions. More specific studies in cultured cells have historically been difficult, because cultured cells possess very low activities of CYP enzymes.
In recent years, the above-mentioned limitations of the research have been bypassed using new techniques. For example, experiments with HepG2 cells that overexpress CYP2E1 support the hypothesis that CYP2E1 is involved in hepatocyte damage owing to alcohol (Cederbaum et al., 2001). Furthermore, CYP2E1 knockout mice have also been employed (Kono et al., 1999). Interestingly, CYP2E1-deficient mice developed liver injury analogous to their wild-type counterparts (Kono et al., 1999). It has been suggested that other isoforms of CYP (e.g., CYP4A) could play a compensatory role in the onset of early alcohol-induced liver injury in the absence of CYP2E1 in the knockouts (Leclercq et al., 2001). Although it appears that CYP2E1 is not required for at least the initiation of alcohol-induced liver injury in mice, it does not preclude a role for CYP2E1 in later stages of disease progression in mice or in other species.
The potential mechanisms by which CYP2E1 contributes to ALD are still being explored; however, there are some plausible hypotheses. For example, CYP2E1 has been shown to contribute to oxidative stress caused by alcohol. This enzyme is relatively loosely coupled with cytochrome reductase; it can therefore leak electrons to oxygen to form O2•−, or catalyze lipid peroxidation (Ekstrom and Ingelman-Sundberg, 1989). Furthermore, CYP2E1 has also been shown to bioactivate several hepatotoxic agents (e.g., acetaminophen). Therefore, the induction of this enzyme by chronic abuse of ethanol can increase the risk of liver damage by other agents. Furthermore, CYP2E1 can be induced in the resident macrophages, the Kupffer cells and macrophages overexpressing CYP2E1 have a more robust response to stimulation in culture (Cao et al., 2005), which could contribute to the ‘priming’ effect of alcohol on these cells (see below).
Whereas acetaldehyde is oxidized to acetate by ALDH (see Figure 2), the kinetics of this reaction are relatively slow and therefore allow detectable increases in acetaldehyde in humans consuming alcohol. Acetaldehyde is toxic and several of the systemic effects of alcohol abuse (e.g., flushing, headaches and nausea) are mediated, at least in part, by direct or indirect effects of elevated acetaldehyde levels. Furthermore, it is proposed that acetaldehyde might also play a causal role in ALD (Lieber, 1988). Acetaldehyde can form adducts with reactive residues on proteins or small molecules (e.g., cysteines). These chemical modifications can alter and/or interfere with normal biologic processes, such as signal transduction, and/or be directly toxic to the cell. Modified biologic molecules could also stimulate the host immune response and cause an autoimmune-like disease. Antibodies against such oxidatively modified proteins have been reported in both humans and animal models of ALD (Klassen et al., 1995; Niemela, 2001). Furthermore, acetaldehyde promotes enhanced GSH utilization/turnover and significantly depletes GSH stores (Lieber, 1997), thus contributing to oxidative stress (see below).
The concentrations of ethanol in the systemic blood can reach high levels; for example a blood alcohol concentration (BAC) of 0.08% equates to an alcohol concentration of approximately 20 mm in the blood. The hepatic concentrations of ethanol are much higher than systemic owing to the first-pass effect of the liver on ethanol. High alcohol concentrations, coupled with the remarkable rate of metabolism by the liver, results in biochemical stress for liver cells. Indeed, whereas acetate from ethanol oxidation can enter the citric acid cycle after conversion to acetyl-CoA, the various metabolic and biochemical alterations caused by ethanol exposure result in a negative energy balance (see Lieber, 1997, for review). It is suggested that these biochemical changes caused by alcohol metabolism, at least in part, mediate and/or exacerbate ALD.
Steatosis, one of the earliest hepatic changes caused by alcohol, was originally thought to be a pathologically inert histological change. More recent work, however, showed that steatosis could play a crucial role not only in the initiation, but also in the progression of ALD (Yang et al., 1997; Day and James, 1998). Fatty livers are more sensitive to hepatotoxicity caused by agents such as endotoxin (Eastin et al., 1997; Yang et al., 1997). Furthermore, the degree of fat accumulation is predictive of the severity of later stages of ALD (i.e., fibrosis and cirrhosis) (Sorensen et al., 1984; Teli et al., 1995). As mentioned above, the oxidation of ethanol to acetaldehyde by ADH and subsequent oxidation to acetate by ALDH utilizes NAD+ as an electron acceptor (see Figure 2), causing a pronounced shift in the NADH:NAD+ ratio to a more reduced state. This increase in the reduced state of pyridine nucleotides is also proposed to be involved in the accumulation of lipids during alcohol ingestion. Specifically, the shift in the NADH:NAD+ ratio increases the rate of fatty acid synthesis and esterification, while simultaneously decreasing mitochondrial β-oxidation of free fatty acids. This change in the redox state can also impair normal carbohydrate metabolism, resulting in multiple effects, including a decreased supply of ATP to the cell (Lieber, 2000). Furthermore, it has been demonstrated that this shift in the pyridine nucleotide redox state can activate the sirtuin family of NAD+-dependent histone deacteylases (You et al., 2008), which can alter gene expression profiles and thereby indirectly affect the metabolic state and signal transduction of the liver. For example, SIRT1 has been shown to activate AMP-activated protein kinase (AMPK, see below) via deacetylation/activation of serine threonine liver kinase B1 (LKB1; Ruderman et al., 2010). However, a decrease in energy state or activation of AMPK also leads to activation of SIRT1, possibly by increasing NAD+ or the NADH:NAD+ ratio (Canto and Auwerx, 2009).
As mentioned above, the oxidation of ethanol by the MEOS consumes oxygen (see Figure 2). Furthermore, alcohol causes an acute hypermetabolic state in the liver, resulting in a doubled oxygen consumption rate (Yuki and Thurman, 1980). This increase in the oxygen consumption rate also increases the intralobular oxygen gradient in the liver (Ji et al., 1982) with subsequent pericentral hypoxia (Arteel et al., 1996, 1997). Hypoxia aggravates the metabolic stress on the liver by further increasing the pyridine nucleotide redox state. A decrease in cellular oxygen tension will also exacerbate the impaired mitochondrial electron flow caused by alcohol by reducing the delivery of O2 to the mitochondria in the steep intracellular oxygen gradient (De Groot et al., 1985). After alcohol levels decrease, the subsequent reoxygenation could increase pro-oxidant production via hypoxia/reoxygenation. This effect together with the impairment of free radical defenses caused by hypoxia (Shan et al., 1989) can add to the observed oxidative stress in the liver after ethanol exposure (Shan et al., 1989).
Reactive oxygen and nitrogen species (ROS and RNS, respectively) are products of normal cellular metabolism and have beneficial effects. For example, ROS have been shown to mediate a variety of cellular signaling pathways. However, owing to the potential of these molecules also to damage normal tissue, the balance between pro-oxidants and antioxidants is crucial for the survival and function of aerobic organisms. If the balance is tipped to favor overproduction of these species, oxidative stress can occur. Oxidative stress has been proposed to be crucially involved in ALD (Shaw et al., 1981; Arteel, 2003). In support of this hypothesis, numerous antioxidants have been shown to protect against the damaging effects of ethanol in vitro and in vivo models of ALD (Kono et al., 2000a, 2001a,b). Although important advances have been made in understanding the role of pro-oxidants in experimental alcohol-induced liver injury, this work has yet to translate into an accepted antioxidant therapy for ALD.
The study of ROS/RNS in vivo is complicated. Owing to their rapid reactions with biomolecules, they generally cannot be measured directly. Therefore, to assess these reactions the products of the reaction of pro-oxidants with endogenous (e.g., lipid peroxides) or exogenous (e.g., spin traps) targets are measured. This indirect detection makes it difficult to identify the parent oxidant. However, there are species with relatively clear data supporting their involvement in experimental ALD. These species are usually derived from superoxide (O2•−) or nitric oxide (NO•). There has been major debate over the identity of the major source(s) of oxidants during alcohol exposure in ALD. Based on experimental data, key pro-oxidant enzymes or enzyme systems have been consistently proposed to play a role in alcohol-induced liver injury localized in hepatocytes, inflammatory cells (e.g., Kupffer cells and neutrophils) and other nonparenchymal cells (e.g., stellate cells).
Superoxide (O2•−) is believed to play a central role in alcohol-induced liver injury. Not only is O2•− readily produced by numerous processes in vivo, but many other oxidants found in vivo are also derived from O2•−. Indeed, exogenous SOD has protective/antioxidant effects in both in vitro and in situ models of alcohol exposure. Genetic overexpression of either cytosolic Cu/Zn-SOD or mitochondrial Mn-SOD in liver cells has been shown to prevent alcohol-induced liver injury in rats fed enteral alcohol (Wheeler et al., 2001a,b). The finding that both cytosolic and mitochondrial SOD isoforms were protective against alcohol-induced liver injury suggests at least two distinct pools of O2•− production are involved.
There is clear evidence that pro-oxidant formation is dependent on the production of O2•− during experimental ALD. However, the presence of this radical alone is not sufficient to explain the oxidative damage caused by alcohol. Indeed, O2•− is not a strong oxidant and does not oxidize many biologic molecules. A well-known radical that can be detected after alcohol exposure is the α-hydroxyethyl free radical. The formation of this radical depends on O2•− production (Wheeler et al., 2001a,b), however, O2•− is too weak an oxidant to directly react with ethanol to form this product (Knecht et al., 1993). Alternatively, O2•− can react via catalytic pathways in the cell to form more potent oxidants. For example, the reduction of O2•− by SOD forms H2O2 and H2O2 plus transition metals can lead to formation of hydroxyl radicals [OH•; the Fenton reaction (Fridovich, 1995)]. Superoxide could also react with NO• to form peroxynitrite (ONOO−), another strong oxidizing and nitrating species (Beckman et al., 1990; Beckman, 1996). Both of these products (OH• and ONOO−) are potent oxidants and capable of reacting with ethanol and form the α-hydroxyethyl radical. Therefore, although O2•− is not a potent pro-oxidant per se, it appears to be a key initiator of oxidative stress during alcohol exposure.
Most RNS found in vivo are derived from NO•. Whether NO• production is protective or damaging in ALD is not clear (Hon et al., 2002). For example, hepatic vasoconstriction caused by acute ethanol or by cirrhosis is, at least in part, owing to low production of NO• (Oshita et al., 1994). Nitric oxide is also antiapoptotic in hepatocytes and is required for normal hepatic regeneration (Rai et al., 1998; Kim et al., 2000). Nitric oxide can also terminate lipid peroxidation chain reactions by attacking lipid peroxyl radicals (Rubbo et al., 2000). Studies in support of the hypothesis that NO• plays a protective role in ALD showed that the NOS-inhibitor N(G)-nitro-l-arginine methyl ester (l-NAME) exacerbated experimental ALD, and that arginine supplementation reversed liver injury caused by ethanol (Nanji et al., 1995, 2001). However, NO• also plays a potentially damaging role in ALD by producing strong reactive species (e.g., ONOO−). These NO•-derived RNS can cause nitration reactions (e.g., 3-nitrotyrosine formation) and nitrosation reactions (e.g., nitrosothiol formation), as well as oxidation reactions during alcohol exposure. Reactive intermediates formed during ONOO− degradation can also cause one-electron oxidation reactions with ethanol, leading to the formation of hydroxyethyl radicals (Gatti et al., 1998). Thus, NO• might play a dual role in ALD, mediating both protective effects and tissue damage by overproduction of RNS. These events in vivo are critically dependent on the cell type, NOS isoform and stage of disease.
The most obvious pathologic changes to the liver during alcohol exposure occur in the hepatocytes. Moreover, the accumulation of indices of oxidative stress (e.g., lipid peroxides) are predominantly a hepatocellular event during alcohol administration. Therefore, hepatocellular oxidant production probably plays a key role in alcoholic liver injury. The proposed major sources of pro-oxidants in hepatocytes are the ethanol-inducible CYP2E1, mitochondria and NAD(P)H oxidase (NOX). In chronic liver disease, activation of NADPH oxidase isoforms leads to increased ROS generation and oxidative stress, resulting in inflammation and fibrogenesis. Activation of NADPH oxidase by hepatotoxic agents (e.g., bile salts) activate death pathways such as CD95-induced signaling to induce cell apoptosis (Reinehr et al., 2005b). Moreover, NOX isoforms found in hepatocytes participate in a variety of signal transduction cascades (Lambeth et al., 2000). Elevated pro-oxidant production not only increases the net amount of pro-oxidants in the hepatocyte, but also directly damages mitochondrial proteins and DNA, which can aggravate mitochondrial aging and activate apoptotic pathways (see Cunningham and Bailey, 2001, for review). In support of this hypothesis, overexpression of Mn-SOD in rats protects against alcohol-induced liver injury (Wheeler et al., 2001b). Furthermore, alcohol depletes mitochondrial GSH (Fernandez-Checa et al., 1997), resulting in increased hepatocellular apoptosis (Colell et al., 1998). Therefore, hepatocellular pro-oxidant production is a crucial step in the progression of ALD.
Another crucial component in the development of alcohol-induced liver injury is inflammation, involving both resident (e.g., Kupffer cells) and recruited (e.g., neutrophils and lymphocytes) inflammatory cells. Unlike parenchymal cells where ROS production is a ‘side effect’ of biochemical processes caused by electron leakage, inflammatory cells are actively generating high concentrations of ROS/RNS into their surrounding environment. Although the production of these species is key for host defense, they can also cause damage to normal tissue if inappropriately upregulated.
It is known that inappropriate activation of Kupffer cells plays a key role in the initiation of alcoholic liver injury (see Thurman, 1998, for review). The production of pro-oxidants is stimulated in activated Kupffer cells. The proposed major sources of pro-oxidants in these cells are NOX and the inducible form of NOS (iNOS or NOS2). Work from Thurman's group showed that the NAD(P)H oxidase inhibitor, diphenyliodonium (DPI), blocks alcohol-induced liver injury in rats exposed to enteral alcohol administration (Kono et al., 2001b). However, DPI is a flavoprotein inhibitor rather than a specific NAD(P)H oxidase inhibitor and could therefore have non-specific effects. Moreover, mice deficient in NAD(P)H oxidase (p47phox knockout mice), were also protected against experimental ALD (Kono et al., 2000b). Together, these results support the hypothesis that O2•− production from this enzyme plays a key role in the initiation of oxidative stress and experimental ALD.
Previous work has also shown that iNOS knockout mice are protected against oxidative stress caused by alcohol and nitration of tyrosine residues was prevented as well (McKim et al., 2003). Similar results were demonstrated using the highly specific iNOS inhibitor, 1400W, in wild-type mice. The finding that injury and oxidative stress are prevented in both iNOS knockout mice (McKim et al., 2003), and in NAD(P)H oxidase-deficient mice (Kono et al., 2000b), demonstrates that the damaging oxidant in experimental ALD is dependent on the production of O2•− and NO•. This in turn gives weight to the idea that ONOO− might be a key player in the onset of alcoholic liver damage. These results also support the hypothesis that NO• production might play dual roles in ALD, with iNOS activity in macrophages a critical early step in the initiation and the progression of liver injury by increasing RNS production. Disadvantages of ‘specific’ NOS inhibitors in vivo are significant cross-reactions with other NOS isoforms (Boer et al., 2000), particularly when these isoforms serve different functions in the disease state, as in ALD (Wiest and Groszmann, 2002).
Recent studies have shown that oxidative stress could play a role not only in infiltrating inflammatory cells, Kupffer cells or hepatocytes, but also in the transformation of stellate cells into myofibroblasts, the critical matrix-depositioning cell in the fibrotic liver (Poli, 2000). It is unclear at this time how much this cell type contributes to oxidative stress in the liver, or whether pro-oxidant production in this cell acts mainly as an autocrine/paracrine signaling event. Because activation of this cell type is crucial in the progression to severe ALD, more research in this area is required.
Oxidative stress can be mediated by an increase in ROS/RNS production, by a decrease in the antioxidant defense, or an increase in the reactivity of ROS/RNS (Sies, 1986). Alcohol causes modifications to the cell that might also favor oxidative stress via these mechanisms. For example, alcoholics have lower antioxidant levels owing to the nutritional deficiencies (Lieber, 2003). Hypoxia caused by alcohol exposure can impair antioxidant defenses (Jones, 1985; Tribble and Jones, 1990). Free iron is mobilized by alcohol (Shaw et al., 1988), which can also lead to an increase in transition-metal catalysis to potent oxidants (e.g., the Fenton reaction). Another example is that alcohol exposure inhibits the 26S proteosome in hepatocytes (Bardag-Gorce et al., 2000), which is responsible for degrading proteins damaged by ROS/RNS. The inhibition of this complex results in an accumulation of proteins damaged by ROS/RNS (Donohue, 2002). Finally, there is a mass of proteins and systems involved in the ‘antioxidant network’. This family does not directly block pro-oxidants, but serves as an additional reductant and maintains the catalytic activity of antioxidant proteins or small molecules. These reactions are energy-dependent to maintain such cycles. Biochemical stress (see above) caused by alcohol exposure can therefore indirectly impair cellular antioxidant defenses.
Chemical modification/damage to biologic molecules, which can alter and/or interfere with normal processes within the cell, is a major mechanism by which ROS/RNS can cause cellular injury. It is also known that pro-oxidants can mediate and/or amplify their signal by modifying signaling cascades within the cell. Many reviews have focused on the role of signaling cascades in damage owing to oxidative stress (Kamata and Hirata, 1999; Allen and Tresini, 2000; Forman and Torres, 2001; Droge, 2002). Signaling cascades that are oxidant sensitive include small molecules [e.g., intracellular Ca2+ (Ermak and Davies, 2002)], stress-activated protein kinases [e.g., SAPK, JNK, ERK1/2, and p38 (Suzuki et al., 1997)], transcription factors [e.g., AP-1, HIF-1 and NF-κB (D'Angio and Finkelstein, 2000)], and modulators of apoptosis signaling [e.g., caspases, Bad and Bcl-2 (Hoek and Pastorino, 2002)]. Ethanol has been shown to alter the signal of many of these pathways in vitro and/or in experimental ALD (see Mandrekar and Szabo, 2009; Szabo and Bala, 2010, for review); however, whether or not these effects are mediated by oxidative stress per se is often unclear.
The concept of priming and sensitization is important in alcoholic liver damage. As noted, the natural history of ALD is characterized by chronic inflammation in the liver, involving an increase in proinflammatory cytokines and a decrease in anti-inflammatory cytokines. Activation of the inflammatory response is, at least in part, owing to increased levels of activators of this response, such as lipopolysaccharides (LPS; Spencer et al., 1983; Bigatello et al., 1987). However, elevated levels of LPS found in alcoholics and in experimental ALD are relatively low compared with those found in experimental endotoxemia and human sepsis. Furthermore, liver injury caused by alcohol cannot be mimicked by chronic low-dose LPS in the absence of ethanol (Deaciuc et al., 1999). However, inflammatory cells seem to be primed to activation by chronic alcohol administration. For example, peripheral blood monocytes from patients with alcoholic hepatitis spontaneously produce proinflammatory mediators (e.g., TNFα). In response to LPS, these monocytes produce more proinflammatory mediators than their control counterparts (McClain and Cohen, 1989).
Hepatocytes also appear to be sensitized to inflammatory stimuli by alcohol administration. For example, although TNFα is proproliferative in hepatocytes isolated from naive animals, it is proapoptotic in cells isolated from ethanol-treated animals (Colell et al., 1998), via cellular ‘death domain’ pathways (Liu et al., 2002). Other cell types such as stellate cells also appear to be primed/sensitized owing to alcohol administration (Reeves et al., 2000; Kim et al., 2001). The concept of priming and sensitization also means that there could be a series of sequential events in the progression of liver damage owing to alcohol exposure. Specifically, the priming of inflammatory cells by ethanol leads to a more robust cell-killing response that is increased in sensitized hepatocytes, explaining why blocking the activation of Kupffer cells (Adachi et al., 1994) or employing knockouts with an impaired Kupffer cell response (Kono et al., 2000b) prevent hepatocyte damage.
It is known that ALD is associated with an increase in apoptotic cell death. Hepatocyte death via apoptosis is associated with alcohol consumption in animals and humans and the number of apoptotic cells detected in the liver correlates with the development of ethanol-induced liver injury (Deaciuc et al., 1999). Studies have shown that hepatocyte apoptosis owing to ethanol can be linked to pathways involving oxidative stress mechanisms via the induction enzymes such as CYP2E1, the effects of cytokines (e.g., TNFα), or the involvement of death receptors (e.g., Fas/CD95; see McVicker et al., 2007 for review). The extrinsic death receptor pathway via TNFα and Fas involves the downstream activation of ‘initiator caspases’ (e.g., caspase 8), which then activate ‘executioner caspases’ (e.g., caspase 3). Intertwined in these pathways are proteins that control the intrinsic or extrinsic routes of apoptosis (e.g., BCL-2 and Bax). Under normal conditions, the highly regulated apoptotic system is counterbalanced by cytoprotective signals that maintain tissue homeostasis. Alcohol consumption however, might increase expression of proapoptotic death factors and result in subsequent apoptotic cell death (see McVicker et al., 2007, for review). The expression of membrane-bound Fas, soluble Fas and Fas ligand have been increased with pathological conditions associated with alcoholic hepatitis (Taieb et al., 1998). This increased expression of Fas ligand in hepatocytes owing to ethanol has been also shown to induce apoptosis of adjacent cells by interacting with their Fas receptors (Taieb et al., 1998). Furthermore, it has been demonstrated that oxidative stress contributes to Fas receptor activation (Reinehr et al., 2005a). Whether this is true for the enhancement of ethanol-induced Fas receptor activation has yet to be determined.
As mentioned above, the enhanced production of TNFα is considered a key factor in the priming effect of ethanol on macrophage activation. This effect has largely been studied in the context of stimulating macrophages with LPS. Ethanol causes changes in several components along the signaling cascade from the LPS receptor partners (CD14/TLR4) to expression of TNFα protein, all leading to enhanced production (see Figure 3). For example, the expression of CD14 on the surface of Kupffer cells is upregulated by oxidant-dependent activation of the transcription factor AP-1 by alcohol (see Figure 3; Wheeler and Thurman, 2003). Two transcription factors that are critical in TNFα expression are NF-κB and EGR-1. Studies have shown that ethanol increases NF-κB activation via both ROS-dependent (Hill et al., 1999) and ROS-independent pathways (Roman et al., 1999). Ethanol also increases EGR-1 transcriptional activity via enhancing ERK1/2 signaling (see Nagy, 2004a, for review). Therefore oxidative stress is hypothesized to be key factor in the priming effect of ethanol. Furthermore, alcohol itself enhances LPS-stimulated ROS production by macrophages via MyD88-independent TLR4 signaling (Hritz et al., 2008), which implies that alcohol induces a feed-forward mechanism to further enhance the above-mentioned pathways and subsequently the expression of TNFα (see Figure 3). It is important to note that there are great similarities between obesity related fatty liver and alcoholic steatohepatitis. These include increased gut permeability with subsequent endotoxemia and TLR4 activation, with both priming and sensitization. However, the TLR4 signaling pathway for hepatotoxicity diverge, with alcohol being MyD88 independent and obesity being MyD88 dependent.
In addition to alcohol-induced oxidative stress, other alcohol-mediated mechanisms also probably contribute to the enhanced production of TNFα protein by macrophages after alcohol exposure (see Figure 3). For example, alcohol activates phosphodiesterase 4B (PDE4B), which is a cAMP-specific PDE isozyme (Gobejishvili et al., 2008). The subsequent decrease in cAMP is proposed to enhance TNFα production via disinhibition of NF-κB activity (see Figure 3; Gobejishvili et al., 2006). Ethanol-induced TNFα expression is not only regulated at the level of transcription but also by increased mRNA stability. Increased LPS stimulation of p38 mitogen-activated protein kinase contributes to this stabilization of TNFα mRNA in macrophages (see Figure 3). Moreover, studies have shown that at least one mRNA-binding protein, HuR, is also involved in stabilization of TNFα mRNA stability after chronic exposure to ethanol (see Nagy, 2004b, for review). The net effect is an increase in TNFα protein release in alcohol-exposed macrophages after stimulation (see Figure 3).
As noted, the sensitization effect of alcohol is described as an increase in sensitivity of parenchymal cells to cytotoxic killing. TNFα-induced signaling via TNFR1 appears to play a key role. In normal hepatocytes, TNFα is a mitogenic stimulus. However, alcohol is proposed to cause changes in intracellular signaling that switch this mitogenic response to a cytoxic response. Although this pathway is not completely elucidated, the sustained activation of JNK (which is an apoptotic signal) coupled with the inhibition of NF-κB activation in parenchymal cells are proposed to mediate the sensitization effect of ethanol (see Han et al., 2009, for review).
Chronic alcohol feeding has been shown to sensitize to the hepatotoxicity induced by gut-derived endotoxin and TNFα in animal models (Szabo and Bala, 2010), and patients with alcoholic hepatitis (AH) have increased basal and endotoxin stimulated monocyte TNFα production (McClain and Cohen, 1989). Moreover, mice that were given anti-TNFα antibodies or mice deficient in TNFR1 were protected against the development of experimental ALD (Iimuro et al., 1997; Yin et al., 1999). However, recent human studies have shown no therapeutic efficacy for anti-TNFα antibodies or TNFα soluble receptors in AH (Naveau et al., 2004; Boetticher et al., 2008). Thus, it appears that completely blocking TNFα is not a viable therapeutic option in ALD, possibly because of the necessary role for a basal level of TNFα in liver regeneration (Fausto, 2006). Another strategy is to attenuate TNFα production/activity with agents such as pentoxyfilline, a broad phosphodiesterase inhibitor. As noted previously, in animal models of ALD highly specific PDE4 inhibition is very effective at attenuating LPS-stimulated TNFα production. Importantly, human trials support a role for pentoxyfilline and phosphodiesterase inhibition in the treatment of alcoholic hepatitis, with the most recent randomized trial showing superiority to corticosteroids (De et al., 2009).
As mentioned above, alcohol-induced steatosis has historically been thought to be the direct result of alcohol metabolism. However, whereas alcohol metabolism is indeed likely to contribute to alcohol-induced steatosis, this process alone does not fully explain the phenomenon. For example, experimental alcohol-induced steatosis can be blunted by using pharmacologic agents but alcohol metabolism per se is not affected (Kono et al., 2001a,b). Similarly, many knockout strains [e.g., NADPH oxidase, iNOS and CD14 (Kono et al., 2000b; Yin et al., 2001a; McKim et al., 2003)] are protected against experimental alcohol-induced steatosis without changes in alcohol metabolism. Therefore, these results support the hypothesis that alcohol metabolism is not the only cause of alcoholic fatty liver.
A possible alternative mechanism by which alcohol can cause steatosis is via the production of cytokines such as TNFα. Indeed, many of the treatments/knockouts discussed above all block the increase in cytokine production caused by alcohol. In support of the role of cytokines in alcohol-induced steatosis, TNFR1 knockout mice were also almost completely protected from alcohol-induced fatty liver (Yin et al., 1999, 2001b). Furthermore, TNFα and other cytokines have been shown to influence lipid metabolism in both liver and the periphery (see Pessayre et al., 2002, for review). TNFα increases free fatty acid release from adipocytes in the periphery (Hardardottir et al., 1992), augments lipogenesis in hepatocytes (Feingold and Grunfeld, 1987), and blocks β-oxidation of fatty acids (Nachiappan et al., 1994). Alcohol-induced cytokines might also impair transport and secretion of triglycerides as VLDL (Navasa et al., 1998). The net effect of alcohol exposure is that cytokines could increase the supply of fatty acids to the liver while simultaneously impairing the ability of the hepatocytes to metabolize and secrete them. However, the specific mechanism(s) by which cytokines might mediate these effects have not been completely elucidated.
It has been hypothesized that cytokines could alter hepatic fat metabolism by increasing the expression and DNA binding activity of the transcription factor, sterol regulatory element binding protein 1 (SREBP-1; You and Crabb, 2004). SREBP-1 activation could directly cause steatosis by upregulating the expression of lipogenic enzymes (e.g., fatty acid synthetase) and indirectly by impairing β-oxidation via increasing the production of malonyl-CoA by acetyl CoA carboxylase (ACC). Malonyl CoA is an allosteric inhibitor of carnitine palmitoyl-transferase-1 (CPT-1), which is a crucial player in the β-oxidation of fatty acids (see Rawson, 2003, for review). In the cell, the effects of SREBP-1 on lipid metabolism appear to be almost exactly opposed to the effects of AMPK (Zhou et al., 2001). It has therefore been proposed that inducers of AMPK might protect against alcohol-induced steatosis by counteracting the effects of SREBP-1.
There are still crucial gaps in our understanding of ALD but the major concepts discussed above are well-established. More recent findings expand our understanding of ALD; these include the potential roles of S-adenosylmethionine (SAM), zinc and fibrin metabolism in the initiation and progression of ALD.
SAM is a precursor for syntheses of polyamines, choline, and is the major methylating agent for a host of molecules via specific methyltransferases (see Figure 4; Avila et al., 2005). Furthermore, SAM is a precursor for glutathione, a major endogenous antioxidant that prevents cellular injury by scavenging free radicals. Under normal conditions, most of the SAM produced is used in transmethylation reactions in which SAM is converted to SAH by transferring the methyl group to diverse biological acceptors (Finkelstein, 1990). SAH hydrolase then catalyzes the conversion of SAH to homocysteine and adenosine in a reversible reaction (Finkelstein, 1990). The conversion of homocysteine to methio-nine is an essential reaction to conserve methionine, detoxify homocysteine, and generate SAM (see Figure 4).
It has been known for over 50 years that alcohol consumption leads to deficiencies of components involved in the transulfuration/methylation pathways (Best et al., 1949). However, the knowledge of the contribution of methyl donor deficiency in human alcoholic liver disease has recently increased. Specifically, alcohol consumption causes a hepatic deficiency in SAM, and this appears to be key to the development of animal model and human ALD (Lu et al., 2002). Indeed, hepatic-specific methionine adenosyl transferase (MAT) is highly sensitive to oxidative stress, and it is probable that decreased hepatic MAT activity in ALD is owing in part to oxidation of the active site. Furthermore, homocysteine accumulates after alcohol exposure in both humans and rats (Hultberg et al., 1993; Cravo and Camilo, 2000; Barak et al., 2001). The mRNA expression of the two enzymes responsible for remethylation of homocysteine, methylfolate-homocysteine methyltransferase (methionine synthase; MS) and betaine-homocysteine methyltransferase (BHMT) is also decreased (Avila et al., 2000). Levels of precursors needed for remethylation of homocysteine (e.g., folate) might be depleted in alcoholics (Cravo et al., 1996). Taken together, the increase in homocysteine observed in alcoholics can at least be explained in part by impaired ability of the liver to remethylate this compound (see Figure 4). It has been shown that patients with compensated alcoholic cirrhosis who were randomized to receive SAM (1200 mg/day orally) for 2 years had decreased liver mortality/liver transplantation (16% vs. 30%) compared with the placebo-treated group (Mato et al., 2002). Betaine supplementation was also shown to prevent alcohol-induced liver damage in mice (Ji and Kaplowitz, 2003). These studies suggest that altered methionine metabolism owing to alcohol intake contributes to ALD.
Depletion of SAM can change the methylation status of DNA, RNA, biogenic amines, phospholipids, histones, and other proteins, which probably contributes to liver injury. Altered DNA methylation might also be a post-translational modification that can affect transcription of key genes involved in the inflammatory response to alcohol. SAM also decreases TNFα expression in animal models of liver injury and in peripheral blood monocytes or macrophage cell lines in vitro (Song et al., 2005). This effect is most probably mediated by blunting LPS-induced NF-κB transcriptional activation of TNFα expression (Veal et al., 2004). Furthermore, SAM increases IL-10 concentrations in animal models or macrophage cell lines (Hevia et al., 2004; Song et al., 2005). Therefore, SAM deficiency can contribute to the priming effect of alcohol on inflammatory cells, with both an increase in proinflammatory TNFα and a decrease in anti-inflammatory IL-10. Other members of the transmethylation pathway (e.g., SAH, homocysteine) could also contribute to the sensitizing effect of ethanol via altering signal transduction responses (Song et al., 2007). Indeed, Song and co-workers showed that elevated SAH sensitized hepatocytes in a dose-dependent fashion to TNF hepatotoxicity. Moreover, a mitochondrial SAM transporter has been identified, and cytosolic SAM is exchanged for mitochondrial SAH. Inhibiting the SAM transporter or increasing cytosolic SAH have both been shown to sensitize to LPS/TNF hepatotoxicity. Thus, altered methionine/SAM metabolism plays a crucial role in both sensitization and priming to hepatotoxicity.
Zinc is an essential trace element, and zinc deficiency is a common occurrence in ALD (Kang and Zhou, 2005). Selected complications of zinc deficiency of relevance to ALD include skin lesions, anorexia, depressed wound healing, hypogonadism, altered immune function, impaired night vision and depressed mental function with possible encephalopathy (Kang and Zhou, 2005). The mechanisms for altered zinc metabolism in ALD are multifactorial and include impaired intake, impaired absorption and increased urinary zinc excretion. Stress/inflammation caused by a variety of factors including LPS/TNFα also cause an internal redistribution of zinc, with loss of zinc from tissues (deficiency) and targeting to other tissues/organs such as the liver. Moreover, zinc deficiency can occur through an oxidative stress setting in which thiol oxidation of zinc-finger transcription factors causes zinc loss from these proteins leading to loss of DNA-binding activity (Zhou et al., 2007). Zinc participates in cellular function through hundreds of zinc proteins including zinc metalloenzymes and crucial zinc transcription factors (Kang and Zhou, 2005). Zinc deficiency has been associated with multiple forms of clinical and experimental liver injury, and it sensitizes to experimental LPS-induced hepatotoxicity (Kang and Zhou, 2005; Shea-Budgell et al., 2006). In ALD, alcohol intake and oxidative stress can cause disruption of tight junctions in the intestine, which leads to translocation of bacterial products, such as endotoxin (Zhong et al., 2010). Endotoxin causes TLR4 activation and TNFα production with subsequent oxidative stress and liver injury. Endotoxin and TNFα also play a crucial role in liver fibrosis as well as liver injury. Disruption of tight-junction proteins occurs not only in the intestine but also in the lung and probably at the blood-brain barrier related to oxidative stress and thiol oxidation, thus potentially predisposing to lung injury and hepatic encephalopathy (Joshi et al., 2009). Zinc treatment in experimental animals with ALD has been shown to attenuate the increased gut permeability, endotoxemia, decrease TNFα production, oxidative stress, and decrease liver injury (Kang and Zhou, 2005). Thus, zinc supplementation targets most postulated mechanisms for the development of ALD. It is well documented that zinc supplementation corrects the manifestations of zinc deficiency in human ALD, and a human pilot trial suggests it might stabilize or cause regression of hepatic fibrosis (Takahashi et al., 2007).
The liver is a critical organ in the regulation of blood coagulation. Hepatocytes synthesize and release numerous coagulation factors into the blood, including fibrinogen and prothrombin, as well as anticoagulant factors such as protein C (Mammen, 1992; Monroe and Hoffman, 2009). Under normal physiological conditions, thrombin, the main effector protease of the coagulation cascade, catalyzes the conversion of fibrinogen to fibrin, which then mediates clot formation (see Figure 5). Homeostasis of the coagulation cascade is crucial for normal organ/organism function; too little activity can lead to edema and clotting dysfunction, and too much activity can lead to hypercoagulation and hemostasis owing to increased fibrin extracellular matrix (ECM) accumulation (Northup et al., 2008). In addition to fibrin deposition by coagulation, the level of fibrin ECM is also regulated by degradation of the existing matrix (fibrinolysis; Kolev and Machovich, 2003). Specifically, inhibition of fibrinolysis can cause ECM to accumulate, even in the absence of enhanced deposition by the thrombin cascade (see Figure 5). A major inhibitor of fibrinolysis is plasminogen activator inhibitor 1 (PAI-1), via blocking the activation of plasmin by plasminogen activators (uPA and tPA; Arteel, 2008).
Thrombin is an acute phase protein that is rapidly activated under conditions of stress in the body. The activation of this protease initiates the coagulation cascade that ultimately causes fibrin to accumulate in the extracellular matrices. Thrombin also has a host of direct actions on the cells (Coughlin, 2000). Most of the cellular effects of thrombin are mediated via proteinase-activated receptor 1 (PAR1). PAR1 receptors are present on hepatic stellate cells which, when activated, deposit fibrous matrix (Fibbi et al., 1999). Increased thrombin generation could thus directly promote hepatic fibrogenesis. Indeed, thrombin inhibition decreased experimental liver fibrosis caused by CCl4 in rats (Duplantier et al., 2004). Furthermore, PAR1 antagonists also ameliorate experimental hepatic fibrosis (Fiorucci et al., 2004). In addition to direct activation by PAR1, thrombin might indirectly activate hepatic stellate cells via stimulating cytokine secretion by macrophages and platelets, which also express PAR1 (Marra et al., 1995). Thrombin signaling could also affect endothelial cells and angiogenesis (Zania et al., 2008), which in principle could also contribute to the pathologic changes in liver caused by alcohol. However, the potential effect of thrombin under these conditions is not well studied.
Under normal conditions, PAI-1 is only expressed in adipocytes and endothelial cells; however, PAI-1 can be upregulated to high levels in multiple cell types under conditions of injury and/or inflammation, most probably via the MAP kinases pathways (Kruithof, 1988; Fearns and Loskutoff, 1997). PAI-1 is the major inhibitor of both tissue-type plasminogen activator (tPA) and urokinase-type plasminogen activator (uPA). It therefore plays a key regulatory role in fibrinolysis by inhibiting the activation of plasminogen. Specifically, high PAI-1 levels can contribute to fibrin ECM accumulation by blocking degradation (see Figure 5).
A common problem in patients with cirrhosis is hyperfibrinolysis (high PA/PAI-1 ratio). It has been shown that these patients have more severe consequences and poor outcome of the disease (Violi et al., 1992; Hu et al., 2001). In contrast to end-stage liver disease, PAI-1 is known to be upregulated with alcohol consumption (Marques-Vidal et al., 1995; Mukamal et al., 2001), and its level is an index of severity during disease development (Tran-Thang et al., 1989). However, whether increased PAI-1 expression is a cause or an effect in the development of alcoholic liver disease remains unclear.
It was recently shown that acute ethanol rapidly and robustly induced PAI-1 expression in the mouse liver, and that steatosis under these conditions was prevented by genetic (PAI-1−/− mice) or pharmacologic inhibition of PAI-1 expression (Bergheim et al., 2006b). The mechanisms by which PAI-1 blunts steatosis were most probable via increasing HGF receptor (cMET) activation, leading to enhanced VLDL synthesis and export from the hepatocyte. Steatosis owing to chronic enteral alcohol exposure was also blunted by preventing the induction of PAI-1 expression (Bergheim et al., 2006b). Furthermore, PAI-1 has been shown to contribute to hepatic inflammation caused by alcohol (Bergheim et al., 2006b; Beier et al., 2009), as well as experimental hepatic fibrosis (Bergheim et al., 2006a). The latter effect is most probable via inhibiting ECM degradation during fibrogenesis; the former effect is still being elucidated, but is probably mediated via a fibrin matrix and/or fibrin signaling (see below).
Hepatic injury in models of alcohol- or drug-induced liver disease involve dysregulation of the coagulation cascade/fibrinolysis, favoring the formation of fibrin clots in the hepatic sinusoids (Beier et al., 2008, 2009). This accumulation might be mediated by de novo fibrin deposition via the coagulation cascade, or also by a decrease in fibrinolysis (e.g., by PAI-1, see above). The fibrin clots disrupt the flow of blood within the hepatic parenchyma (i.e., hemostasis), causing microregional hypoxia and subsequent hepatocellular death (Pearson et al., 1996; Ganey et al., 2004). For example, previous studies have shown that ethanol enhances LPS-induced liver damage via mechanisms involving excess fibrin accumulation. Exaggerated fibrin accumulation induced by ethanol was not associated with an enhancement of LPS-induced coagulation, per se, as thrombin-antithrombin levels were similar in the LPS and ethanol/LPS groups.
One potential mechanism by which fibrinogen could alter signal transduction cascades is via enhanced hemostasis. Specifically, fibrin clots disrupt the flow of blood within the hepatic parenchyma (i.e., hemostasis), the subsequent micro-regional hypoxia and hepatocellular death might directly (e.g., HIF1α) and indirectly cause changes in cell signaling (Pearson et al., 1996; Ganey et al., 2004). Fibrin(ogen) ECM not only serves as physical structure, but also binds/interacts with several biomolecules that can directly or indirectly alter responses. One family of receptors for which fibrin(ogen) is a ligand are the integrins. Integrins are receptors that mediate attachment between a cell and the tissues surrounding it, which might be other cells or ECM. Integrins transfer information from the ECM to the cell, allowing rapid and flexible responses to changes in the environment. Integrins play a myriad of roles within the body, including proliferation/angiogenesis, as well as inflammation and apoptosis (Hodivala-Dilke et al., 2003; Zhou et al., 2009). Fibrin(ogen) is a known ligand for several integrins, including integrin αIIbβ3, and integrin αMβ2 and integrin αvβ3. These integrins are found on several non-parenchymal cells in the liver. Therefore, fibrin(ogen) ECM has the potential to alter intracellular signaling in liver via a myriad of mechanisms. Modulators of integrin function are in human clinical trials for nonhepatic diseases, and might be ultimately used in hepatic problems such as ALD.
As summarized in this review, there are multiple signaling pathways that are proposed in the mechanisms by which alcohol causes liver disease. These proposed mechanisms are not mutually exclusive and probably work in tandem to cause initiation and progression of ALD (see Figure 1). Whereas there is no universally-accepted therapy to treat ALD or halt its progression, there are many potential therapies being tested that target at least one of these potential mechanisms (see Bergheim et al., 2005, for review). Likewise, owing to the multiple mechanisms by which alcohol can damage the liver, it is unlikely that one therapy will be sufficient to treat ALD. Effective therapy might probably be a mix of lifestyle modification, dietary care and pharmacologic intervention.
Another avenue in which improvements need to be made is in modeling the disease. Specifically, most research to date focuses on early stages of ALD and often involves concomitant administration of the potential therapy with ethanol, or in the case of transgenic/knockout mice, using strains in which the expression of the protein of interest is a life-long event. Such research has given us useful information and is the foundation on where we now stand. However, the relevance of this research to the clinical situation in which alcoholics are seeking therapy to reverse existing ALD is unclear. Further translational research and research involving multiple modulators/targets are required.
Supported, in part, by grants from National Institute on Alcohol Abuse and Alcoholism (R37 AA010762, P01 AA017103, R01 AA015970, RC2 AA019388, R01 AA018869). J.I.B. was supported by a postdoctoral (T32) fellowship from National Institute of Environmental Health Science (ES011564).