Apoptosis: The nexus of liver injury and fibrosis

Apoptosis associated with liver disease is increasingly viewed as a nexus through which many key pathways converge. Apoptotic responses incorporate soluble stimuli, inflammatory cells, resident parenchymal cells, and, it now appears, fibrogenic cells as well. Cellular apoptosis is the first cellular response to many toxic events, and accompanies viral hepatitis, alcohol-induced liver disease, nonalcoholic fatty liver disease, cholestatic liver diseases, and ischemia/reperfusion injury.1-5 Moreover, hepatocyte apoptosis is significantly increased in patients with alcoholic hepatitis and nonalcoholic steatohepatitis, and correlates with disease severity and hepatic fibrosis.1, 5 Despite their pervasive concurrence in liver damage, the relationship between apoptosis and either hepatic inflammation or fibrosis has not been fully explored, in part because of the prevailing—and overstated—dogma that cell death by apoptosis is “innocuous.” This presumption has been based primarily on analyses of organ development, where physiologic apoptosis is tightly restricted to discrete subsets of cells both spatially and temporally. In contrast, “pathologic” apoptosis in adult tissues induces large numbers of cells, is non-selective, injures large aggregates of cells, and may be sustained over decades. In contrast to physiologic apoptosis, pathologic apoptosis in liver may not only result from inflammation and fibrosis, but may in turn amplify these responses. In particular, hepatic stellate cells (HSC), the key fibrogenic cell type in liver, contribute to apoptosis and inflammation. Given these emerging and sometimes contradictory observations, a review of the relationship between apoptosis, inflammation, and fibrosis is timely. Critical insights into these complex relations may help promote rationally-based therapies for patients with chronic liver diseases. Furthermore, we anticipate that these concepts will help stimulate further investigation into mechanisms connecting cell death to inflammation and fibrosis. HSC, hepatic stellate cell; PS, phosphatidylserine; TNF, tumor necrosis factor; TRAIL, TNF-related apoptosis-inducing ligand; TGF-β1, transforming growth factor β. Apoptosis is a form of cell death characterized by organized nuclear and ultimately cellular fragmentation. During apoptosis, cells are fragmented into small membrane-bound bodies, namely apoptotic bodies, that express phosphatidylserine (PS) on the outer leaflet of their plasma membrane and are removed by phagocytosis.6-8 Apoptosis may occur by 2 fundamental pathways: (1) death receptor or extrinsic pathway; and (2) the intracellular organelle-based intrinsic pathway.9 Regulation of the apoptotic machinery in liver cells is complex, but appears to be commonly triggered through activation of death receptors.10, 11 These receptors include Fas, tumor necrosis factor (TNF)-receptor-1, and tumor necrosis factor-related-apoptosis-inducing-ligand receptors -1 and -2 (TRAIL-R1 and-R2). In addition to inducing apoptosis, these death receptors (e.g., Fas, TRAIL-R1 and R2, and TNF-receptor-1) initiate independent intracellular signaling cascades that further amplify liver injury as well as inducing cellular demise. For example, TNF-receptor-1, when ligated by TNF-α, activates nuclear factor-κB, which is a transcription factor for many proinflammatory cytokines.12, 13 The intracellular stress pathway of apoptosis can also be initiated in liver cells through several intracellular organelles. Indeed, lysosomal permeabilization, alterations in endoplasmatic reticulum calcium processing, nuclear DNA damage, and mitochondrial dysfunction all can trigger apoptosis. Mitochondrial dysfunction often plays a critical role in augmenting the apoptotic process and integrating death receptor-initiated and stress signals into a common final pathway.9, 14 Mitochondrial release of cytochrome C is a near-universal event in apoptosis and triggers a final caspase-dependent apoptosis cascade culminating in cellular fragmentation.9, 14, 15 An HSC is a resident perisinusoidal mesenchymal cell type that is a central effector of fibrosis in liver injury. In a normal liver, HSCs are quiescent, but in liver injury, they undergo a myofibroblastic transdifferentiation, or “activation”, leading to accumulation of extracellular matrix.16, 17 As the liver injury resolves, the number of activated stellate cells decreases through 1 of 2 potential pathways, either spontaneous reversion or clearance by apoptosis. To date, spontaneous reversion of myofibroblasts to quiescent cells has only been documented in culture,18, 19 but not in vivo.16 In contrast, there is now clear evidence that stellate cells undergo apoptosis during resolution of liver injury in vivo.20, 21 Thus, driving activated HSC into apoptosis may be a way to resolve fibrosis, as discussed below.21 The signals mediating HSC apoptosis have begun to emerge. Quiescent HSCs have a very low turnover, therefore, apoptosis does not appear to be prominent feature of this cell type until it is “activated” in liver injury. The relative apoptotic activity of stellate cells reflects a balance between apoptotic stimulation and survival signals. Apoptotic stimuli include the death receptors, Fas, and TRAIL-R2,20, 22 all of which may provoke cell death as they do in activated T-cells.23 The TRAIL-R2 receptor is upregulated in activated HSCs, which are susceptible to TRAIL-mediated apoptosis in vitro22; studies in TRAIL-deficient animals are anticipated to validate this concept in vivo. Additionally a Fas/Fas-ligand-dependent mechanism of HSC apoptosis has been reported.20 Interestingly, transformed HSC also express peripheral benzodiazepine receptors, which render the cells sensitive to peripheral benzodiazepine receptor-ligand-induced apoptosis.24 Stellate cells also express the nerve growth factor receptor p75, which mediates nerve growth factor-induced apoptosis of cultured stellate cells.25 Stimuli promoting survival of stellate cells are less well characterized, but, in principle, these factors favor the sustained deposition of extracellular matrix by preventing loss of matrix-producing cells. The tissue inhibitor of metalloproteinase, a molecule that inhibits the activity of matrix-degrading proteases, has the independent effect of preventing stellate cell apoptosis.26 Through its dual activities of sustaining stellate cells and preventing matrix clearance, tissue inhibitor of metalloproteinase-1 has emerged as an important target to antagonize in attempting to reduce fibrosis.27 In addition, the accumulation of extracellular matrix, in particular collagen I, may represent another important survival signal. This concept has been underscored by a recent study in which mice with a mutation in collagen-1 (col-1aIr/r) that renders them resistant to degradation by interstitial collagenase, have delayed recovery from liver fibrosis, persistence of activated HSC, and diminished hepatocyte regeneration.28 These newly appreciated but complex relationships between HSC apoptosis and fibrosis are critical to understanding the pathogenesis of cirrhosis. A direct link between accumulation of inflammatory mediators and apoptosis has also emerged. Apoptosis may induce inflammation by several mechanisms. First, dysregulated apoptosis in pathologic conditions can disrupt hepatocyte integrity. For example, after experimental induction of apoptosis with Fas-agonists, mice develop fulminant hepatic failure with massive necrosis and inflammation.29 When the magnitude of apoptosis overcomes the capacity to clear cellular debris, apoptotic bodies undergo spontaneous disruption and release their contents, inciting tissue damage that elicits an inflammatory response.30 Because the liver cannot simply shed apoptotic cells into a lumen like many other tissues (e.g., colon, breast, prostate, etc.), apoptotic bodies may accumulate because their clearance is compromised. Second, death receptor-mediated apoptosis may contribute to liver inflammation, possibly by initiating deleterious signaling cascades.31-34 For example, Fas agonists induce chemokine (e.g., macrophage inflammatory protein-2, CXC ligand-1) expression that promotes neutrophil infiltration into the liver, and stimulates hepatic inflammation.35 In support of this concept, Jaeschke et al. demonstrated that hepatocyte apoptosis is a potent stimulus to neutrophil extravasation and enhancement of endotoxin-induced liver injury.32, 33, 36 In human and in experimental alcoholic hepatitis, apoptotic hepatocytes colocalize with neutrophils, which correlates strongly with the severity of tissue damage.36, 37 Consistent with these data, inhibition of hepatocyte apoptosis blocks neutrophil transmigration into the liver during injury.31-34 Finally, the disposition of apoptotic bodies may also link apoptosis to inflammation in the liver. For example, engulfment of neutrophil apoptotic bodies by macrophages and/or Kupffer cells can induce expression of death ligands, especially Fas,38, 39 thereby accelerating apoptosis. Like the proverbial “weasel in the chicken coop,” once the Kupffer cell has tasted its prey (i.e., the apoptotic cells) it will kill again (i.e., express death ligands).40 Fas ligand is also strongly pro-inflammatory.41 Another consequence of hepatic inflammation is the generation of soluble mediators and oxidative stress that also promote the activation of HSC into a myofibroblast-like phenotype.42 Rather than being passive targets of inflammatory stimuli, however, HSCs also mediate the inflammatory response via expression of cytokines and adhesion molecules, including monocyte chemotactic protein-1, macrophage colony stimulating factor, among others.42-45 Moreover, HSC activation leads to up-regulation of toll-like receptor-4, cluster of differentiation (CD)-14 and expression of myeloid differentiation protein (MD)-2, which also mediate inflammation.46 Activation of toll-like receptor-4 by lipopolysaccharide is also associated with an up-regulation of proinflammatory cytokines (interleukin-8, monocyte chemotactic protein-1) via nuclear factor-κB.46 Activated stellate cells also express CD40, a receptor whose ligand is present on immune effector cells, thereby providing a direct link between inflammatory and fibrogenic cells.47 Activation of CD40 leads to nuclear factor-κB mediated expression of chemokines, which amplify inflammation.47 Recent microarray data suggest that as HSCs age in culture with repeated passaging, they undergo a phenotypic switch toward a more inflammatory but less fibrogenic state,48 an observation that has not yet been validated in vivo. Clearance of apoptotic debris through phagocytosis may directly stimulate fibrogenesis.49-52 HSCs are ideally positioned to engulf apoptotic bodies through their intimate proximity to hepatocytes, a key source of apoptotic particles during liver injury. Indeed stellate cells may possess many more phagocytic functions than previously appreciated. For example, HSCs not only can phagocytose apoptotic bodies, but they also express nicotinamide adenine dinucleotide phosphate oxidase, an enzyme normally associated with phagocytic functions.53 Kupffer cells can also migrate into the hepatic cords and phagocytose apoptotic bodies and produce fibrogenic stimuli, as described below. Although “professional” phagocytes, such as macrophages, have been assigned the major role in clearance of apoptotic bodies, epithelial cells and even fibroblasts may also serve this function.54, 55 Phagocytes recognize apoptotic cells through an array of surface receptors, among them CD14, CD36, complement receptors, scavenger receptors, a PS-specific receptor, membrane tyrosine kinase c-mer receptor and through milk fat globule epidermal growth factor 8, a glycoprotein, which facilitates rapid clearance.56 These cell membrane receptors elicit a signaling response when activated that may modulate inflammatory responses.56, 57 For example, mice lacking the intracellular domain of c-mer, a signaling kinase, display increased TNF-α production and mortality following lipopolysaccharide administration in vivo.58 Apoptotic cells exhibit numerous “eat-me” signals to provoke engulfment, which includes alterations in the composition or exposure of cell membrane constituents, including PS. In particular, the PS receptor appears to play a dominant role in the phagocytosis of apoptotic bodies,6, 59 supported by the observation that apoptotic cells that do not express PS are poorly phagocytosed.49, 60 Thus, ligation of PS receptor on phagocytes delivers a “tickle” signal, which stimulates the phagocytosis of apoptotic cells.57 The engulfment of apoptotic bodies is not simply a degradative or “clean-up” process to remove cellular corpses. Rather, engulfment, like most other receptor ligand interactions, initiates intracellular signaling cascades in the engulfing cell with discrete biologic responses including cytokine generation.49, 61 In particular, transforming growth factor (TGF)-β is generated when cells phagocytose apoptotic bodies. Although TGF-β is a potent inhibitor of T-cell function, it is also a strong fibrogenic signal in liver.16, 62 Expression of TGF-β and cytokines is mediated by mitogen activated protein kinase and phosphoinositol-3-kinase pathways.50, 63, 64 While inhibiting these pathways does not interfere with phagocytosis of apoptotic bodies, it prevents TGF-β expression.50 Thus, corpse clearance can be dissociated from the expression of TGF-β, suggesting that inhibition of kinase activity in stellate cells might be antifibrotic provided that activity of such an inhibitor could be limited to this cell type. Hepatic stellate cells express the core machinery to phagocytose apoptotic bodies, in particular the PS receptor. Not only can HSCs engulf apoptotic bodies, but their signaling response to phagocytosis is conserved. Apoptotic body engulfment in HSCs stimulates TGF-β expression and induces collagen I, a major constituent of the cirrhotic scar.50 Both quiescent and activated HSC will phagocytose apoptotic bodies in vitro. More importantly, phagocytosis of apoptotic bodies by quiescent HSC facilitates the phenotypic transformation to myofibroblasts, as manifested by induction of the classic activation marker, α-smooth muscle actin.65 However, in vivo studies determining the disposition of apoptotic bodies are currently lacking in part, because elimination of endogenous apoptotic hepatocytes is extremely efficient and difficult to identify and quantitate. Such studies are in progress. The relationship between apoptosis and fibrosis is also bidirectional, wherein fibrosis may in turn stimulate apoptosis. An altered extracellular matrix in an injured liver could induce pro-apoptotic gene expression in parenchymal cells. For example, fibrosis accompanying tissue injury may lead to up-regulation of Fas/ Fas L.66-68 A recent study suggests that activating the Fas system in mice enhances pulmonary fibrosis, whereas suppressing apoptosis attenuates fibrosis in this model.68 Thus, many of the emerging relationships between fibrosis and apoptosis may be generalizable to other tissues. The growing panoply of interactions between apoptotic, inflammatory, and fibrotic responses has therapeutic implications. For example, phosphatidylcholine, a polyunsaturated phospholipid extracted from soy beans, has been explored for its antifibrotic potential in alcoholic liver injury.69-71 A related compound, dilinoleoylphosphatidylcholine, prevents TGF-β−mediated collagen accumulation in cultured rat HSCs through an unknown mechanism.64 Intriguingly, phosphatidylcholine blocks PS receptor mediated uptake of apoptotic bodies, which is associated with reduced generation of TGF-β.57, 60 It is tempting to speculate that phosphatidylcholine and related compounds may exert antifibrotic activity in liver by blocking engulfment of apoptotic bodies, but direct evidence is still lacking. Similarly, the phosphotidylinositol-3-kinase inhibitor, LY294002, and the p38 mitogen-activated kinase inhibitor, SB203580, both down-regulate collagen messenger RNA levels in primary rodent stellate cells.63, 64, 72 Inhibiting phosphotidylinositol-3-kinase or p38 mitogen-activated kinase in HSCs also reduces collagen I and TGF-β1 messenger RNAs, following their engulfment of apoptotic bodies.50 Targeted kinase inhibition therefore might facilitate removal of apoptotic corpses without eliciting downstream signaling that generates fibrogenic mediators. From another perspective, inhibiting apoptosis may also prevent liver inflammation and fibrosis. For example, small molecule caspase inhibitors currently are being developed for clinical use, which could reduce hepatocyte apoptosis thereby attenuating inflammation, reducing HSC activation, and decreasing fibrosis. Another strategy might be to block apoptotic signaling through expression of small interfering RNAs. For example, Fas-specific small interfering RNA attenuates hepatic fibrosis following repeated concavalin A administration in mice,73 a model of immune-mediated liver damage. An alternative antifibrotic strategy could selectively target activated HSCs with apoptotic stimuli. Proof-of-principle has been accomplished with the use of gliotoxin, a fungal metabolite, that reduces hepatic fibrosis through accelerated apoptosis of HSCs, but does not alter hepatocyte apoptosis in vivo following carbon-tetrachloride administration.74 Similarly, ligands for TRAIL-R2 induce apoptosis in activated HSCs, which are not toxic to a normal liver because this receptor is not expressed on either quiescent stellate cells or hepatocytes.22 These early results merit further study. We propose the following model linking hepatocyte apoptosis to inflammation and HSC activation based on current data (Fig. 1). Pro-apoptotic stimuli induce hepatocyte apoptosis, which, combined with death receptor signaling, promote hepatic inflammatory responses through the release of extracellular signals, including lipid mediators.51 Engulfment of the hepatocyte apoptotic bodies by HSC and Kupffer cells enhances their expression of pro-fibrogenic genes and death ligands. Persistent activation of these cells promotes further hepatocyte apoptosis, which culminates in hepatic inflammation, with generation of CXC chemokines and sustained stellate cell activation. At the same time, anti-apoptotic signals including tissue inhibitor of metalloproteinase-1 and collagen I that accumulate during fibrogenesis render activated stellate cells relatively resistant to apoptotic stimuli, further perpetuating the accumulation of matrix. As candidate compounds are developed to manipulate apoptotic responses in liver disease, a key challenge will be to limit apoptosis to parenchymal cells while accelerating “therapeutic” apoptosis of hepatic stellate cells. Such targeted approaches will require a comprehensive knowledge of cell-specific apoptotic signaling pathways and mediators. Schematic representation of the proposed model linking hepatocyte apoptosis to inflammation and fibrosis. Proapoptotic stimulus induces hepatocyte apoptosis. These apoptotic cells release lipid signals for their uptake by Kupffer cells and hepatic stellate cells (HSC). Engulfment of the apoptotic bodies by HSC and Kupffer cells enhances their expression of pro-fibrogenic genes and of death ligands (e.g., FasL). Persistent activation of these cells promotes further hepatocyte apoptosis, which culminates in hepatic inflammation, with generation of CXC chemokines (Interleukin-8, macrophage inflammatory protein-2, growth-related oncogene -α;-β;-γ, and CXC ligand-1) and further HSC activation in a feed-forward-loop process. Anti-apoptotic signals including tissue inhibitor of metalloproteinase-1 and collagen I that accumulate during fibrogenesis render activated stellate cells relatively resistant to apoptotic stimuli, further perpetuating the accumulation of scar matrix. The authors acknowledge Erin Bungum for her excellent secretarial assistance and Dr. Maria Eugenia Guicciardi for reviewing the manuscript and providing critical input.