Endotoxemia and gut barrier dysfunction in alcoholic liver disease

Abstract

Agrowing body of evidence indicates that endotoxemia is closely associated with alcoholic liver disease (ALD). Endotoxins stimulate different cells in the liver releasing cytokines, chemokines, and reactive oxygen species (ROS) by toll-like receptor-4 (TLR-4)–mediated mechanisms. Intestinal microflora is the source of circulating endotoxins, and the gut barrier dysfunction leading to elevated intestinal permeability is considered the main cause of endotoxemia in ALD. The mechanism of ethanol-induced gut barrier disruption is an active area of investigation. Evidence indicates that intestinal microflora, the metabolism of ethanol, and acetaldehyde-induced cell signaling are involved in ethanol-induced intestinal barrier dysfunction. Recent advances in alcoholic endotoxemia, the mechanism of epithelial barrier disruption, and the factors that prevent alcoholic endotoxemia are discussed in this article. The current understanding of these issues is illustrated in Fig. 1. Diagrammatic representation of ethanol-induced intestinal permeability and endotoxemia. Ethanol in the colonic lumen is metabolized to acetaldehyde by microflora and mucosa. Acetaldehyde disrupts epithelial barrier function and allows the diffusion of bacterial LPS into the portal circulation. In the liver, LPS activates Kupffer cells via an LBP/CD14/TLR-4–dependent mechanism. Abbreviations: ADH, alcohol dehydrogenase; LBP, lipopolysaccharide binding protein; LPS, lipopolysaccharide; NFκB, nuclear factor kappa B; ROS, reactive oxygen species; TLR-4, toll-like receptor 4. Endotoxins are lipopolysaccharides (LPSs) derived from the cell walls of gram-negative bacteria. Dead bacteria and the LPS shed from the cell walls of viable organisms contribute to the circulating endotoxins. Normally, Kupffer cells in the liver detoxify endotoxins by phagocytosis. When the flux of endotoxins overwhelms the phagocytotic capacity of Kupffer cells, the endotoxins spill into the systemic circulation. Endotoxemia is conventionally believed to exist when the plasma endotoxin level rises higher than 2.5 endotoxin units per milliliter. Endotoxemia in ALD was first recognized by the detection of antibodies against Escherichia coli in the plasma of patients with ALD.1 Numerous subsequent studies have demonstrated that the plasma endotoxin level in patients with ALD is several times greater than that in healthy subjects2-7 or in patients with nonalcoholic cirrhosis.4 Endotoxemia was recorded in alcoholics who showed only minimal symptoms of ALD7; therefore, endotoxemia appears to occur early during the pathogenesis of this disease.7 The plasma endotoxin level in normal subjects ranges from 0.3 to 10.4 pg/mL,2-8 whereas the endotoxin level in ALD patients ranges from 8.5 to 206 pg/mL.2-8 Despite the wide variability in endotoxin values, the plasma endotoxin levels in patients with ALD have always been found to be 5- to 20-fold greater than those in normal subjects. Although one study has indicated the existence of a positive correlation between the degree of endotoxemia and the severity of disease,3 other studies have failed to demonstrate such a correlation.7 Endotoxemia in ALD has been confirmed in experimental models of alcoholic liver injury. Plasma endotoxin levels are elevated by acute9, 10 or chronic9, 11-13 ethanol administration, and the levels correlate well with the development of liver injury. The elevation of the plasma endotoxin level with acute ethanol administration indicates that endotoxemia occurs during the early stage of liver damage. Although there exists no single animal model that depicts different stages of ALD, there is a striking agreement between plasma endotoxin levels in patients with ALD and ethanol-fed animals. There is no evidence of species-dependent differences in alcoholic endotoxemia; however, a gender-dependent difference has been observed.12, 14 ADH, alcohol dehydrogenase; AJ, adherens junction; ALD, alcoholic liver disease; EGF, epidermal growth factor; LBP, lipopolysaccharide binding protein; LPS, lipopolysaccharide; MLCK, myosin light chain kinase; NFκB, nuclear factor kappa B; PKC, protein kinase C; PLCγ, phospholipase C-γ; PTPase, protein tyrosine phosphatase; PTP1B, protein tyrosine phosphatase-1B; ROS, reactive oxygen species; TJ, tight junction; TLR, toll-like receptor; ZO-1, zonula occludens-1. The attenuation of alcohol-induced endotoxemia and liver damage by antibiotics indicates that endotoxin plays a crucial role in alcoholic liver damage.15 An emerging body of evidence indicates that LPS and ethanol synergistically affect the liver cells. LPS by itself fails to mimic ethanol-induced steatosis or hepatitis; however, ethanol and LPS together effectively induce liver damage. Ethanol feeding sensitizes the liver to LPS-induced cellular injury in experimental animals16-18 and exacerbates LPS-induced cytokine release in the liver.19 The mechanism of synergism between ethanol and LPS may involve multiple factors, such as down-regulation of interleukin-10–mediated protection,18 nicotinamide adenine dinucleotide phosphate oxidase–dependent production of ROS,16 and adrenergic stimulation.17 Cellular targets of LPS in the liver include Kupffer cells,20 sinusoidal endothelial cells,20 stellate cells,21 neutrophils, and hepatocytes.22 Inactivation of Kupffer cells prevents ethanol-induced liver injury in rats.23 LPS stimulates sinusoidal endothelial cells to release cytokines and chemokines.20 In stellate cells, LPS pretreatment enhances ethanol-induced collagen secretion.21 Lipopolysaccharide binding protein (LBP) presents LPS to CD14, a 55-kDa glycoprotein.24 CD14 specifically binds to LPS and interacts with TLR-4.25 Chronic ethanol feeding enhances the expression of LBP11 and CD14,26 and ethanol-induced liver injury is absent in LBP,27 CD14,26 and TLR-428 knockout mice. TLR-4–mediated stimulation of different liver cells leads to the secretion of a variety of proinflammatory factors such as cytokines,29 chemokines,30 and ROS.31 Intestinal microflora is the source of circulating endotoxins. Normally, endotoxin absorption is impeded by the mucosal barrier function. Three major factors may contribute to alcoholic endotoxemia: (1) delayed endotoxin clearance from the circulation, (2) ethanol-induced bacterial overgrowth, and (3) a dysfunctional gut barrier leading to elevated endotoxin absorption. Evidence indicates that clearance of LPS from the circulation is delayed by ethanol feeding.32 Ethanol impairs the phagocytic function of Kupffer cells and attenuates the endotoxin uptake by these cells.33 The endotoxins that escape Kupffer cells spill into the circulation and contribute to endotoxemia. The numbers of both aerobic and anaerobic bacteria were found to be high in the jejunal aspirates of alcoholics in comparison with those in normal subjects.34 Ethanol has been shown to delay gastrointestinal motility,35 and delayed gastrointestinal transit is known to increase bacterial growth in the lumen. Therefore, it is possible that delayed gastrointestinal motility is responsible for bacterial overgrowth in alcoholics. Although delayed clearance and bacterial overgrowth are potential contributors, enhanced intestinal permeability to endotoxins appears to be the primary cause of alcoholic endotoxemia. Alcohol ingestion has been shown to increase intestinal permeability to macromolecules in both normal subjects and alcoholics,13, 34, 36 and this suggests that the disruption of intestinal epithelial barrier function plays an important role in causing endotoxemia in alcoholics. It is not clear whether an ethanol-induced increase in permeability to macromolecules is caused by gastroduodenal barrier disruption or intestinal barrier disruption. This is an important issue as the intestinal microflora is confined to the colon and the distal ileum. One study has suggested that acute ethanol may increase the gastroduodenal permeability without altering the intestinal permeability, whereas chronic ethanol may elevate the intestinal permeability without altering the gastroduodenal permeability.37 Another important question is whether ethanol-induced intestinal permeability is a transient or persistent response. Studies by Bjarnason et al.38 demonstrated that elevated intestinal permeability to ethylene diamine tetraacetic acid existed in alcoholics after 3 days of abstention from alcohol, whereas it persisted even after 2 weeks of sobriety in alcoholics with liver cirrhosis. Therefore, it appears that the effect of ethanol on intestinal barrier function is a transient effect in normal subjects and in alcoholics without cirrhosis, whereas the ethanol effect is likely to persist much longer in alcoholics with the liver disease. As intestinal epithelial cells are renewed every 48 to 72 hours, it is likely that chronic ethanol feeding causes a delay in cell migration and renewal. Increased intestinal permeability to macromolecules has also been observed in animal models of alcoholic liver damage.39 Gastrointestinal permeability to macromolecular markers and LPS has been elevated by both acute and chronic ethanol administration in experimental animals.9, 35, 36, 40, 41 It is likely that barrier disruption occurs much before liver injury, and this has been confirmed by a recent study.42 An ethanol-induced increase in gastrointestinal permeability has been associated with endotoxemia and liver damage.9, 36, 39 Thus, experimental models of alcoholic liver damage are beneficial in understanding the pathogenesis of ALD. One of the important questions that have been addressed for a decade is whether ethanol directly disrupts intestinal barrier function. All of the in vitro studies have indicated that intestinal barrier disruption requires an ethanol concentration greater than 1%. Increases in paracellular permeability in rat jejunal everted sacs43 and Caco-2 cell monolayers44, 45 required exposure to at least 2% ethanol. Ethanol concentrations of 1.2% and 5% were used to disrupt the barrier function in intestinal epithelial cells (IEC-6)46 and Madin-Darby canine kidney47 cell monolayers, respectively. Such a high concentration of ethanol is not expected to exist in the distal intestine of alcoholics. An early study by Halsted et al.48 demonstrated that 5% ethanol could be measured in the jejunal contents after alcohol consumption, whereas only 0.2% to 0.25% ethanol was detectable in the ileal and colonic lumen. Therefore, the results obtained in studies using ethanol concentrations greater than 0.25% may be relevant to barrier dysfunction in gastroduodenal and jejunal mucosa. A significant body of evidence indicates that acetaldehyde plays a crucial role in the disruption of intestinal epithelial barrier function.41 An acetaldehyde level as high as 0.4 mM has been measured in the saliva of alcoholics.49 The concentration of acetaldehyde in the rat colonic lumen after ethanol administration was found to be millimolar.50 Acetaldehyde at a concentration of 0.1 to 0.6 mM disrupts the barrier function in Caco-2 cell monolayers.13, 45 A similar acetaldehyde-induced barrier disruption has been demonstrated in human colonic mucosa.51 Furthermore, in vitro incubation of rat colonic strips mounted to Ussing chambers showed that ethanol up to 18 mM does not affect the barrier function.41 However, acetaldehyde (50-160 μM) dose-dependently increased the paracellular permeability, and this demonstrated that the metabolism of ethanol into acetaldehyde in the colonic lumen is required for barrier disruption. The cellular mechanism involved in acetaldehyde-induced barrier dysfunction involves the disruption of epithelial tight junctions (TJs) and adherens junctions (AJs).13, 52-57 In addition to being the source of circulating endotoxins, intestinal microflora plays an important role in ethanol metabolism. Although intestinal epithelial cells express alcohol dehydrogenase, bacterial alcohol dehydrogenase seems to play a predominant role in the generation and accumulation of acetaldehyde in the intestinal lumen.58 The capacity of microflora and intestinal mucosa to further metabolize acetaldehyde into acetate by aldehyde dehydrogenase is low, and this results in acetaldehyde accumulation in the colonic lumen. A recent study41 demonstrated that antibiotics partially reduce colonic acetaldehyde. Therefore, intestinal microflora seems to play an important role in the production of acetaldehyde in the intestinal lumen. Ethanol-induced gastrointestinal permeability was attenuated by the pretreatment of rats with antibiotics for 12 days, and this was also associated with the attenuation of ethanol-induced endotoxemia. TJs are the specialized intercellular junctional complexes that form a barrier to the diffusion of macromolecules across the epithelial monolayers.59 TJs are assembled by the organization of a variety of specific proteins such as occludin, claudins, and zonula occludens. AJs are other intercellular junctional complexes that lie beneath the TJs. AJs do not form a physical barrier to macromolecular diffusion. However, AJs indirectly regulate the integrity of TJs and therefore control the barrier function of the epithelium. AJs are organized by the interaction between E-cadherin and β-catenin, by the interaction between β-catenin and α-catenin, and by the binding of α-catenin to the actin cytoskeleton. Both TJs and AJs are regulated by intracellular signal transduction The disruption of TJs and AJs by ethanol and acetaldehyde has been demonstrated in Caco-2 cell monolayers by immunofluorescence microscopy.44, 52-57 Acetaldehyde induces a redistribution of occludin and zonula occludens-1 (ZO-1) from the intercellular junctions52, 54-56 and dissociates these proteins from the actin cytoskeleton.57 Acetaldehyde also causes a redistribution of E-cadherin and β-catenin from the intercellular junctions, and this indicates the disruption of AJs.52, 55 A careful time course analysis of the redistribution of E-cadherin and β-catenin indicated that acetaldehyde disrupts AJs as early as 10 minutes, whereas the distribution of occludin and ZO-1 is unaffected. This observation suggests that acetaldehyde alters the integrity of AJs first, and this in turn may trigger the disruption of TJs.55 Acetaldehyde-induced disruption of TJs and AJs has been further validated in human colonic mucosa.51 Acetaldehyde induced the disruption of TJs and AJs in human colonic mucosal biopsies, as indicated by the redistribution of TJ and AJ proteins from the intercellular junctions and the dissociation of these proteins from the actin cytoskeleton. Signaling elements such as protein kinases60-64 and protein phosphatases63 regulate the integrity of TJs in different epithelia. Acetaldehyde-induced disruption of TJs and AJs has been associated with a rapid increase in the tyrosine phosphorylation of ZO-1, E-cadherin, and β-catenin, and the paracellular permeability is attenuated by tyrosine kinase inhibitors.52 Acetaldehyde inhibits protein tyrosine phosphatase (PTPase) activity, and this is a likely cause of increased tyrosine phosphorylation. Acetaldehyde inhibits PTPase activity in cell-free fractions, and this indicates that it directly interacts with the PTPases. Acetaldehyde not only inhibits the activity of PTPase but also disrupts its interaction with E-cadherin.55 The current knowledge supports a model in which inhibition and dissociation of protein tyrosine phosphatase-1B (PTP1B) from E-cadherin lead to tyrosine phosphorylation of E-cadherin and β-catenin and a loss of interaction between these two proteins (Fig. 2). Mass spectrometric analysis has demonstrated that acetaldehyde increases phosphorylation of β-catenin on Y331, Y333, Y654, and Y670. In vitro protein-protein interaction studies using recombinant proteins have demonstrated that tyrosine phosphorylation of β-catenin reduces its interaction with E-cadherin, but tyrosine phosphorylation of E-cadherin has no effect on this interaction (Fig. 2). On the other hand, tyrosine phosphorylation of E-cadherin results in the loss of its interaction with PTP1B. Therefore, acetaldehyde disrupts the interaction between E-cadherin and β-catenin by a phosphorylation-dependent mechanism. Diagrammatic model showing acetaldehyde-induced loss of interaction between PTP1B, E-cadherin, and β-catenin. Cadherin-based cell-cell adhesion is mediated by the interaction of E-cadherin with β-catenin. PTP1B binds to the intracellular domain of E-cadherin and dephosphorylates β-catenin on tyrosine residues. (1) The treatment of cell monolayers with acetaldehyde induces the inhibition and dissociation of PTP1B from E-cadherin, and (2) this results in increased tyrosine phosphorylation of β-catenin and E-cadherin. (3) Tyrosine phosphorylation of β-catenin results in the loss of its interaction with E-cadherin, and (4) the loss of the interaction between E-cadherin and β-catenin leads to a loss of the homophilic interaction between extracellular domains of E-cadherin and disruption of adherens junctions. (5) The disruption of adherens junctions leads to the disruption of tight junctions. (6) Ethanol at high doses alters the cytoskeletal structure by a nitric oxide–dependent and MLCK-dependent mechanism, (7) which in turn disrupts tight junctions. Abbreviations: MLCK, myosin light chain kinase; PTP1B, protein tyrosine phosphatase-1B. Disruption of AJs is the likely signal that leads to disruption of TJs. Although the precise mechanism involved in AJ-mediated TJ disruption is unknown, recent studies have demonstrated that acetaldehyde induces tyrosine phosphorylation of ZO-1 and threonine dephosphorylation of occludin. The functions of occludin dephosphorylation and ZO-1 phosphorylation are unclear. However, such changes in the phosphorylation status of occludin and ZO-1 have been consistently observed during the disruption of TJs and the barrier function by other mediators. Several factors that ameliorate ethanol/acetaldehyde-induced barrier dysfunction have been identified. Understanding the mechanisms involved in the actions of these protective factors may be beneficial for the development of therapeutic strategies for ALD. Epidermal growth factor (EGF), secreted in saliva and other gastrointestinal secretions,65 is known to protect the gastrointestinal mucosa from various insults.66, 67 A recent study has demonstrated that EGF prevents acetaldehyde-induced TJ disruption56 by preventing the reorganization of the actin cytoskeleton and dissociation of TJ and AJ proteins from the actin cytoskeleton. EGF-mediated prevention of acetaldehyde effects on TJs involves the activation of phospholipase C-γ (PLCγ).57 Knockdown of PLCγ1 attenuates EGF-mediated protection of TJs from acetaldehyde. EGF induces membrane translocation of protein kinase C β I (PKCβI) and PKCϵ by a PLCγ-dependent mechanism.57 Prevention of PKCβI and PKCϵ translocation by interfering peptides also attenuates EGF-mediated protection of TJs. This study indicates that PLCγ-mediated activation of PKCβI and PKCϵ may be involved in stabilization of the actin cytoskeleton. Other mechanisms may be involved in barrier dysfunction by a high dose of ethanol. Ethanol activates myosin light chain kinase44 and induces nitric oxide generation,42 which alters the actin and microtubule cytoskeletal structures. Therefore, more than one mechanism may synergistically attenuate barrier function. Glutamine is an essential nutrient for intestinal epithelial cell growth and differentiation.66 Pretreatment of Caco-2 cell monolayers with L-glutamine attenuated acetaldehyde-induced permeability to inulin and LPS.56 This protective effect of glutamine was associated with the prevention of acetaldehyde-induced redistribution of occludin, ZO-1, E-cadherin, and β-catenin from the intercellular junctions. Interestingly enough, the protective effect of L-glutamine was mediated by the transactivation of EGF receptor. Glutamine rapidly activated EGF receptor, and a selective inhibitor of EGF receptor kinase attenuated the glutamine effect on TJs. Other studies have shown that feeding oat bran or zinc attenuates ethanol-induced liver injury. Zinc supplementation attenuates an ethanol-induced increase in plasma alanine aminotransferase and liver pathology,40 and this has been associated with a reduction of ethanol-induced intestinal permeability and endotoxemia. Similarly, oat bran supplementation prevents ethanol-induced liver injury.36 Once again, oat bran supplementation attenuates ethanol-induced intestinal permeability and endotoxemia. Another potential factor that may ameliorate alcoholic liver injury is probiotics. Probiotics attenuate ethanol-induced liver injury in rats.68, 69 Probiotics are well known to reduce the growth of harmful bacteria,70 so it is likely that probiotics prevent intestinal barrier disruption and endotoxemia in alcoholics. Additionally, Lactobacillus rhamnosus GG has been shown to effectively metabolize acetaldehyde into acetate.71 Therefore, by reducing the acetaldehyde-producing commensal bacteria and by metabolizing acetaldehyde, probiotics may alleviate the colonic acetaldehyde level. However, further studies are necessary to confirm this possibility. In summary, the evidence is clear that alcohol consumption leads to increased intestinal permeability and endotoxemia, which are involved in the pathogenesis of ALD. Intestinal microflora and the generation of acetaldehyde in the colonic lumen play crucial roles in alcoholic intestinal permeability and endotoxemia. Evidence indicates that intestinal microflora not only is the source of circulating endotoxins but also plays a role in the generation and accumulation of acetaldehyde in the colonic lumen and has a subsequent influence on epithelial barrier dysfunction. The major mechanism of acetaldehyde-induced barrier dysfunction involves disruption of AJs and TJs by a phosphorylation-dependent mechanism. Gastrointestinal mucosal protective factors such as EGF, glutamine, zinc, oat bran, and probiotics prevent ethanol/acetaldehyde-induced intestinal permeability, endotoxemia, and liver damage. It is therefore important to delineate the mechanisms involved in ethanol-induced intestinal epithelial barrier disruption, endotoxemia, and endotoxin-mediated liver cell injury to understand the pathogenesis of ALD and to design preventive and therapeutic strategies for the treatment of ALD.