Any tissue trauma or excessive stress with cell injury will trigger an acute inflammatory response with activation of resident macrophages, natural killer (NK) cells and NK cells with T cell receptor (NKT) and recruitment of polymorphonuclear leukocytes (neutrophils) and mononuclear cells (monocytes, lymphocytes) into the liver. The main purpose of this innate immune response is to eliminate invading microorganisms and foreign objects, remove damaged cells and cell debris and prepare the tissue for regeneration. On the other hand, there are a number of pathophysiological situations where an excessive inflammatory response causes additional liver damage or even triggers liver failure. Under these circumstances, anti-inflammatory therapeutic interventions should limit tissue injury. However, due to the vital role of inflammation in host defense, there is an increased risk for infections and sepsis. Therefore, to be able to identify selective intervention strategies, it is necessary to understand mechanisms of inflammatory tissue damage in more detail. The current review will summarize our understanding of specific mechanisms of liver injury induced by neutrophils, which are important early responder in acute inflammation.
Abstract: Hepatic infiltration of polymorphonuclear leukocytes (neutrophils) is an early response to tissue injury, cellular stress or systemic inflammation. Neutrophil activation is vital for host-defense and the removal of cell debris but can also cause additional tissue damage or even liver failure. In order to prevent the detrimental effects of neutrophils without compromising host-defense reactions, it is important to understand the mechanisms of neutrophil hepatotoxicity. The first step in the pathophysiology is the priming and recruitment of neutrophils into the liver vasculature by inflammatory mediators, e.g. cytokines, chemokines, or complement factors. Most critical for parenchymal cell damage is the accumulation in sinusoids, which does not depend on cellular adhesion molecules. The next step is the extravasation into the parenchyma. This process requires a chemotactic signal from hepatocytes or already extravasated neutrophils and depends on cellular adhesion molecules on neutrophils (β2 or β1 integrins) and on endothelial cells (intercellular or vascular cell adhesion molecules). The third step is the direct contact with target cells (hepatocytes), which involves β2 integrins and triggers the full activation of the neutrophil with a long-lasting adherence-dependent oxidant stress and degranulation. The oxidants diffuse into hepatocytes and trigger an intracellular oxidant stress, mitochondrial dysfunction and eventually cause oncotic necrotic cell death. Neutrophil-derived proteases facilitate extravasation and are involved in the regulation of inflammatory mediator production. Based on these mechanisms, it appears that strengthening of the intracellular defense mechanisms in hepatocytes may be the most promising therapeutic approach to selectively prevent neutrophil-mediated tissue damage without compromising their host-defense function.
cellular adhesion molecules;
complement factor 5a;
intercellular adhesion molecule-1;
interleukin-8 and other CXC chemokines;
β2 integrin CD11b/CD18;
platelet activating factor;
parenchymal cell hepatocyte;
reactive oxygen species;
tumor necrosis factor- α;
vascular cell adhesion molecule-1.
Mechanisms of neutrophil-induced liver injury
Liver dysfunction and cell injury induced by neutrophils has been demonstrated in a number of experimental models including hepatic ischemia-reperfusion injury (1), endotoxin shock (2), sepsis (3), alcoholic hepatitis (4), obstructive cholestasis (5), remote organ trauma (6), and concanavalin A- or α-naphthylisothiocyanate-induced liver injury (7, 8). Neutrophil-mediated injury was also demonstrated in 2-hit models of ischemia-reperfusion injury or drug hepatotoxicity in combination with endotoxemia (9, 10). There is evidence for neutrophil activation and potential involvement in clinical situations of alcoholic hepatitis and ischemia-reperfusion injury (11, 12). On the other hand, substantial neutrophil accumulation in the liver may not always be associated with additional cell damage as shown for acetaminophen hepatotoxicity or lithocholic acid-induced bile infarcts (13–15), although this topic is controversially discussed (16, 17). Thus, although the migration of neutrophils into the liver parenchyma is required for neutrophil-induced liver injury (18), the mere presence of these phagocytes does not necessarily mean that they cause additional damage. Detailed mechanistic studies revealed factors and conditions, which contribute to neutrophil cytotoxicity in the liver.
Neutrophil recruitment into the liver vasculature
Neutrophils can accumulate in sinusoids, postsinusoidal venules or portal venules of the liver (19, 20). Despite the expression of cellular adhesion molecules (CAMs) on sinusoidal and venular endothelial cells (21), neutrophil accumulation in sinusoids is mostly independent of CAMs (22). Neutrophils appear to be initially trapped due to mechanical reasons including the swelling of endothelial and Kupffer cells and reduced rheological properties of neutrophils (23). On the other hand, neutrophil rolling and firm adhesion in postsinusoidal venules or portal venules requires the involvement of selectins (rolling) and β2 integrin/intercellular adhesion molecule-1 (ICAM-1) or β1 integrin/vascular adhesion molecule-1 (VCAM-1) interactions (24–27). However, parenchymal cell damage is generally caused by neutrophils, which migrated out of sinusoids (5, 18, 27). The mediators that could trigger neutrophil accumulation in sinusoids include tumor necrosis factor-α (TNF-α) (28–31), interleukin-1α (IL-1α) and IL-1β (31), activated complement factors, e.g. C5a (29, 32), platelet activating factor (33) and CXC chemokines, e.g. interleukin-8 (IL-8) (34), macrophage inflammatory protein-2 (MIP-2) (31, 35), keratinocyte-derived chemokine (KC) (31, 35) and cytokine-induced neutrophil chemoattractant (CINC-1) (36, 37). The relative importance of the contribution of individual mediators to the overall effect depends on the experimental model and can vary considerably (35, 38). In addition to these classical proinflammatory mediators, there are mediators such as high mobility group box 1 protein (HMGB1), whose release is delayed (39). HMGB1 can be secreted by macrophages and is passively released by necrotic cells, where it can trigger inflammation through activation of leukocytes (40). Other products of damaged cells, i.e. lipid peroxidation products, can be chemotactic agents for neutrophils (41). However, most proinflammatory mediators do not only recruit neutrophils into liver sinusoids but also prime the cells for enhanced reactive oxygen formation (9, 32, 42, 43). In addition, neutrophil–endothelial cell interactions and the exposure to inflammatory mediators trigger the mobilization of secretory granules, which enhances the expression of Mac-1 (CD11b/CD18) on the surface of neutrophils and leads to shedding of l-selectin (44). Circulating neutrophils show higher Mac-1 and reduced L-selectin expression during endotoxemia in vivo (29, 31); hepatic neutrophils have even higher expression of Mac-1 than circulating neutrophils (45). However, neutrophils need to extravasate into the parenchyma to be fully activated and generate reactive oxygen (18, 46–48).
The migration of neutrophils into the parenchyma is a prerequisite of neutrophil-mediated injury (18). This process involves β2 integrin/ICAM-1 and β1 integrin/VCAM-1 interactions (30, 49). The firm adhesion and transmigration process triggers the exocytosis of gelatinase granules from neutrophils, which liberate matrix metalloproteinases (50). These enzymes may help in degrading extracellular matrix proteins and facilitate the migration into the parenchyma (51). On the other hand, when there is extensive endothelial cell damage, e.g. during ischemia-reperfusion, neutrophils may have direct access to the parenchyma without a CAM-dependent transmigration process (52, 53).
Independent of the status of the endothelial cell barrier, for neutrophil extravasation to occur there needs to be a signal generated from parenchymal cells or already extravasated neutrophils. CXC chemokines are potent chemotactic factors and can be generated by parenchymal cells after exposure to cytokines (54). Excessive production of CXC chemokines in parenchyma leads to neutrophil transmigration and liver injury (36). CXC chemokines also contribute to the promotion of neutrophilic hepatitis during ischemia-reperfusion injury (38, 55). However, the massive amounts of CXC chemokines generated in parenchymal cells during endotoxemia do not cause neutrophil extravasation when endotoxin is administered alone (35). In addition, neutrophil migration into the parenchyma after treatment with galactosamine/endotoxin occurs later and is independent of CXC chemokine formation (35). In this model, apoptotic cell death of parenchymal cells is the main trigger for neutrophil extravasation (56, 57). How can apoptotic cells induce a neutrophil attack? Hepatocytes undergoing apoptosis can generate CXC chemokines and thus attract neutrophils (58, 59). However, in the presence of large amounts of cytokine-induced chemokine formation during endotoxemia, this is an irrelevant effect (35). More recent studies showed the formation of endothelial cell gaps before parenchymal cells apoptosis and neutrophil transmigration in the galactosamine/endotoxin shock model (60). These gaps facilitate the direct contact of neutrophils with the underlying hepatocytes (47, 60). The recognition of surface modifications of cells undergoing apoptosis, e.g. exposure of phosphatidylserine, may trigger the migration of neutrophils. Similarly, modified cell surfaces of necrotic cells, the release of lipid peroxidation products, HMGB1 or other mediators may induce neutrophil extravasation during necrotic processes. Generally, there has to be a clear distress signal from the parenchyma to entice neutrophil migration. The most potent signals appear to be related to dead or dying cells.
Neutrophil-induced parenchymal cell killing
Once extravasated, the neutrophil will adhere to the target, i.e. parenchymal cells. This adherence is mediated by members of the β2 integrin family on neutrophils, i.e., LFA-1 (CD11b/CD18) and Mac-1 (CD11b/CD18), and its counterreceptor on hepatocytes, ICAM-1 (61). Interestingly, only LFA-1 interacts with ICAM-1. The counterreceptor on hepatocytes used by Mac-1 is unclear (61). Nevertheless, the adherence through Mac-1 triggers a long-lasting adherence-dependent oxidant stress and degranulation (release of specific, azurophilic and gelatinase granules) (50, 62). Upon activation, neutrophils generate superoxide through NADPH oxidase, which is a multicomponent enzyme system that assembles at the cell membrane (63). Superoxide dismutates spontaneously to oxygen and hydrogen peroxide, which is then used by neutrophil-derived myeloperoxidase (MPO) to generate hypochlorous acid (HOCl) (63). HOCl is a potent oxidant and chlorinating agent. It also can react with amino groups and form toxic chloramines (64). On the other hand, azurophilic, specific and gelatinase granules contain a large number of proteolytic enzymes, e.g. elastase, cathepsins, and proteinase-3, and bactericidal proteins, e.g. presenilin 1, defensins, bactericidal/permeability-increasing protein, matrix metalloproteinases, and lysozyme (50).
There is a longstanding controversy regarding the mechanisms of neutrophil-induced cell injury in the liver. Most in vitro coculture experiments using activated neutrophils and cultured hepatocytes showed that only protease inhibitors but not antioxidant enzymes were able to prevent the neutrophil-mediated cell injury after 15–20 h (65, 66). However, there are a number of concerns with this approach. Neutrophils do not attack and kill normal hepatocytes in vivo (67). Hepatocytes are exposed to the same inflammatory mediators as neutrophils, upregulate ICAM-1 and generate chemokines (53–55, 68). Thus, hepatocytes are not a passive target of the neutrophil attack. Neutrophil can kill hepatocytes in vivo within 1 h. This correlates with the appearance of an intracellular oxidant stress and the formation of hypochlorite-mediated chlorotyrosine protein adducts (48, 67, 69). Furthermore, mice deficient in glutathione peroxidase-1 showed enhanced susceptibility against the neutrophil cytotoxicity (69). Thus, under in vivo conditions, neutrophils appear to kill mainly through reactive oxygen species, which diffuse into the adherent hepatocytes and cause an intracellular oxidant stress (Fig. 1). Although there is some evidence of lipid peroxidation caused by this neutrophil-derived oxidant stress (67), on a quantitative basis this amount of lipid peroxidation is an order of magnitude below what it would take to cause direct cell killing (70, 71). In contrast, the neutrophil-mediated oxidant stress may induce mitochondrial dysfunction in hepatocytes, which can trigger an additional mitochondrial oxidant stress (72). This can eventually trigger mitochondrial membrane permeability transition pore opening and the collapse of the mitochondrial membrane potential resulting in oncotic necrosis (72).
Although the neutrophil-derived oxidant stress on its own is able to kill hepatocytes, proteases are clearly involved in cell killing during a neutrophilic hepatitis (23). Treatment with urinary trypsin inhibitor (UTI) or the elastase inhibitor ONO-5046 reduced reperfusion injury after warm ischemia or cold storage (73–75). However, the protective effect of protease inhibitors appears to be not limited to preventing a direct protease-mediated cell injury as demonstrated in vitro (65, 66). Neutrophil elastase can induce CINC-1 formation in Kupffer cells, which is inhibited by UTI (76). Both UTI and ONO-5046 reduced CINC-1 formation after ischemia-reperfusion in vivo and attenuated neutrophil accumulation and activation in the postischemic liver (76–78). In a clinical study, protease inhibitors improved postischemic liver injury and reduced proinflammatory cytokine release after tumor resections (79). These experimental and clinical data are consistent with the more recently recognized role of neutrophil proteases in regulating the processing of cytokines, growth factors and their receptors (80). The anti-inflammatory mechanism of protease inhibitors may actually be the main effect in vivo rather than the general inhibition of protein degradation.
Alternative mechanisms of neutrophil-induced cytotoxicity
Other cytotoxic mediators, such as bactericidal proteins, phospholipases, and fatty acids, released by neutrophils may contribute synergistically to the damaging potential of oxidants and proteases (81). This aspect of the mechanism of neutrophil-induced cell injury has not been investigated in the liver. More recently, there is increasing support for the involvement of the FasL/Fas system in the pathophysiology of inflammatory liver injury (16, 82, 83). Neutrophils express FasL (16, 82) and could kill Fas receptor expressing hepatocytes through apoptosis. However, there is no evidence that neutrophils cause apoptotic liver injury (83); in contrast, a neutrophil attack on hepatocytes undergoing apoptosis will result in oncotic necrotic cell death (56, 67). On the other hand, stimulation of the Fas receptor can cause upregulation of inducible nitric oxide synthase and accelerate glutathione depletion, which may enhance the susceptibility of hepatocytes to cytotoxic mediators (84). In addition, Fas receptor activation can trigger CXC chemokines formation (58). Thus, there are multiple possibilities for the involvement of the FasL/Fas system in the mechanism of neutrophil-mediated liver injury.
Neutrophil-mediated injury in steatotic livers
Excessive fat accumulation in the liver increases the risk for more severe liver injury after ischemia-reperfusion and endotoxemia (85–87). Part of the increased injury is related to aggravated impairment of the microcirculation (88). The role of inflammation and in particular neutrophils in this increased injury is controversial. There is evidence for increased neutrophil recruitment and activation after ischemia reperfusion in models of alcohol- and nonalcohol-induced steatosis (89, 90). Furthermore, alcohol-induced steatosis sensitizes the liver for endotoxin-induced neutrophil recruitment and injury (91). On the other hand, neutrophil accumulation in the postischemic liver was not affected in obese Zucker rats (92). In our experience, the increase in inflammatory gene expression, hepatic recruitment and transmigration of neutrophils was actually attenuated in ob/ob mice (93). In contrast to liver tissue of lean mice, the steatotic livers of ob/ob mice did not stain for hypochlorite-modified proteins, a specific indicator for a neutrophil-induced oxidant stress (93). These findings indicate that the contribution of the inflammatory response to the postischemic injury can vary considerably depending on the experimental model of steatosis.
Neutrophil-mediated parenchymal cell damage in the liver is initiated by the priming and subsequent accumulation of neutrophils in the hepatic vasculature, in particular hepatic sinusoids (Fig. 2). This initial step is mediated by proinflammatory mediators such as cytokines, CXC chemokines, complement factors and others. If the primed neutrophils receive a chemotactic signal from the parenchyma, the cells will extravasate. This process is dependent on integrins on neutrophils and cellular adhesion molecules (ICAM-1, VCAM-1) on endothelial cells. During the extravasation process and the subsequent adhesion to the target cell (hepatocyte), the neutrophil will be fully activated. The adhesion via Mac-1 (CD11b/CD18) triggers superoxide formation by NADPH oxidase and degranulation with the release of MPO and proteases. Whereas the proteases appear to be mainly involved in the further promotion of inflammatory cytokine and chemokine formation, hydrogen peroxide and MPO-derived hypochlorite induce an intracellular oxidant stress in hepatocytes and eventually cause oncotic necrosis.
Perspectives and future developments
Therapeutic strategies aimed at preventing inflammatory and in particular neutrophil-mediated liver injury have to take into consideration the vital host-defense function of these leukocytes. Interventions, which primarily target the neutrophil's ability to migrate and to generate cytotoxic mediators, e.g., blocking of CAMs or inhibition of NADPH oxidase, effectively prevent neutrophil-induced liver injury but increase the risk for infections and sepsis. Based on our improved understanding of neutrophil-mediated cell killing in the liver and the documentation that reactive oxygen species and not necessarily proteases are the central cytotoxic mediators, the most promising therapeutic approach may be to strengthen the intracellular defense mechanisms against oxidant stress. In fact, recent insight into the mechanisms of preconditioning induced by ischemia, heat shock or chemicals such as atrial natriuretic peptide indicate that a major component of the beneficial effects is the enhanced antioxidant defense induced by these interventions (94, 95). This therapeutic approach accomplishes the goals of protecting the target cells against inflammatory tissue injury without paralyzing host defense functions of leukocytes.
Work in the authors' laboratory was supported in part by National Institutes of Health grants AA12916 and DK070195.