Chapter 13: Toxic Responses of the Liver

The liver is the main organ where exogenous chemicals are metabolized and eventually excreted. As a consequence, liver cells are exposed to significant concentrations of these chemicals, which can result in liver dysfunction, cell injury, and even organ failure. If an industrial chemical, for example carbon tetrachloride, bromobenzene, or vinyl chloride, is identified as a hepatotoxicant, the use of the chemical may be restricted, the exposure may be minimized by mandating protective clothing and respirators, and attempts are made to replace it with a safer alternative. In the pharmaceutical industry, adverse effects on the liver are one of the most frequently cited reasons for discontinuing the development of drug candidates. In addition, hepatotoxicity recognized during the postmarketing phase is one of the main causes for withdrawing drugs from the market (Temple and Himmel, 2002). Troglitazone (Rezulin^®), a new antidiabetic drug, was removed from the market after close to 100 of the 1.9 million patients treated with the drug suffered from liver failure (Chojkier, 2005). Thus, predictable and idiosyncratic hepatotoxicities severely restrict drug discovery efforts and drug development (Lee and Senior, 2005). Furthermore, the increasing popularity of herbal medicines, which are generally plant extracts, enhances the incidence of drug-induced liver injury and liver failure (Stickel et al., 2011). Since these medicines are mixtures of sometimes hundreds of compounds, it remains a difficult task to identify the causative chemical and the mechanism of injury (Lee and Senior, 2005). Basic science and clinical aspects of drug- and chemical-induced liver injury were discussed in detail in several monographs and reviews (Zimmerman, 1999; Jaeschke et al., 2002, 2012a, b; Kaplowitz and DeLeve, 2002; Boyer et al., 2006b; Grattagliano et al., 2009; Roth and Ganey, 2010; Jones et al., 2010).

Given the unprecedented speed of drug discovery and the increasing demand and use of “natural products” as food supplements and medicine, the early identification of hepatotoxicants remains a formidable challenge for the future. The liver, with its multiple cell types and numerous functions, can respond in many different ways to acute and chronic insults. To recognize potential liver cell dysfunction and injury, it is necessary to have a general knowledge of basic liver functions, the structural organization of the liver, the processes involved in the excretory functions of the liver, and mechanisms of cell and organ injury. Each of these aspects can contribute to mechanisms of drug- and chemical-induced hepatotoxicities.

Hepatic Functions

The liver's strategic location between the intestinal tract and the rest of the body facilitates the performance of its enormous task of maintaining metabolic homeostasis of the body (Table 13-1). Venous blood from the stomach and intestine flows into the portal vein and then through the liver before entering the systemic circulation. Thus, the liver is the first organ to encounter ingested nutrients, vitamins, metals, drugs, and environmental toxicants as well as waste products of bacteria that enter portal blood. Efficient scavenging or uptake processes extract these absorbed materials from the blood for catabolism, storage, and/or excretion into bile.

All the major functions of the liver can be detrimentally altered by acute or chronic exposure to toxicants (Table 13-1). When toxicants inhibit or otherwise impede hepatic transport and synthetic processes, dysfunction can occur without appreciable cell damage. Loss of function also occurs when toxicants kill an appreciable number of cells and when chronic insult leads to replacement of cell mass by nonfunctional scar tissue. Alcohol abuse is the major cause of liver disease in most western countries (Crawford, 1999); thus ethanol provides a highly relevant example of a toxicant with multiple functional consequences (Gao and Bataller, 2011). Early stages of ethanol abuse are characterized by lipid accumulation (fatty liver) due to diminished use of lipids as fuels and impaired ability to synthesize the lipoproteins that transport lipids out of the liver. As alcohol-induced liver disease progresses, appreciable cell death occurs, the functioning mass of the liver is replaced by scar tissue, and hepatic capacity for biotransformation of certain drugs progressively declines. People with hepatic cirrhosis due to chronic alcohol abuse frequently become deficient at detoxifying both the ammonia formed by catabolism of amino acids and the bilirubin derived from breakdown of hemoglobin. Uncontrollable hemorrhage due to inadequate synthesis of clotting factors is a common fatal complication of alcoholic cirrhosis. A consequence of liver injury that merits emphasis is that loss of liver functions can lead to aberrations in other organ systems and to death (Gao and Bataller, 2011).

Structural Organization

Two concepts exist for organization of the liver into operational units, namely, the lobule and the acinus (McCuskey, 2006b). Classically, the liver was divided into hexagonal lobules oriented around terminal hepatic venules (also known as central veins). At the corners of the lobule are the portal triads (or portal tracts), containing a branch of the portal vein, a hepatic arteriole, and a bile duct (Fig. 13-1). Blood entering the portal tract via the portal vein and hepatic artery is mixed in the penetrating vessels, enters the sinusoids, and percolates along the cords of parenchymal cells (hepatocytes), eventually flows into terminal hepatic venules, and exits the liver via the hepatic vein. The lobule is divided into three regions known as centrilobular, midzonal, and periportal. The acinus is the preferred concept for a functional hepatic unit. The terminal branches of the portal vein and hepatic artery, which extend out from the portal tracts, form the base of the acinus. The acinus has three zones: zone 1 is closest to the entry of blood, zone 3 abuts the terminal hepatic vein, and zone 2 is intermediate. Despite the utility of the acinar concept, lobular terminology is still used to describe regions of pathological lesions of hepatic parenchyma. Fortunately, the three zones of the acinus roughly coincide with the three regions of the lobule (Fig. 13-1).

Acinar zonation is of considerable functional consequence regarding gradients of components both in blood and in hepatocytes (Jungermann and Kietzmann, 2000). Blood entering the acinus consists of oxygen-depleted blood from the portal vein (60%–70% of hepatic blood flow) plus oxygenated blood from the hepatic artery (30%–40%). Enroute to the terminal hepatic venule, oxygen rapidly leaves the blood to meet the high metabolic demands of the parenchymal cells. Approximate oxygen concentrations in zone 1 are 9% to 13%, compared with only 4% to 5% in zone 3 (Kietzmann and Jungermann, 1997). Therefore, hepatocytes in zone 3 are exposed to substantially lower concentrations of oxygen than hepatocytes in zone 1. In comparison to other tissues, zone 3 is hypoxic. Another well-documented acinar gradient is that of bile salts (Groothuis et al., 1982). Physiological concentrations of bile salts are efficiently extracted by zone 1 hepatocytes with little bile acids left in the blood that flows past zone 3 hepatocytes. There is difference in bile acid transporter expression between different zones (Baier et al., 2006).

Heterogeneities in protein levels of hepatocytes along the acinus generate gradients of metabolic functions. Hepatocytes in the mitochondria-rich zone 1 are predominant in fatty acid oxidation, gluconeogenesis, and ammonia detoxification to urea. Gradients of enzymes involved in the bioactivation and detoxification of xenobiotics have been observed along the acinus by immunohistochemistry (Jungermann and Katz, 1989; Gebhardt et al., 1994). Notable gradients for hepatotoxins are the higher levels of glutathione (GSH) in zone 1 and the greater amounts of cytochrome P450 proteins in zone 3, particularly the CYP2E1 isozyme inducible by ethanol (Tsutsumi et al., 1989; Niemelä et al., 2000).

Hepatic sinusoids are the channels between cords of hepatocytes where blood percolates on its way to the terminal hepatic vein. Sinusoids are larger and more irregular than normal capillaries. The three major types of cells in the sinusoids are endothelial cells, Kupffer cells, and stellate cells (Fig. 13-2) (McCuskey, 2006b). In addition, the liver contains a substantial number of lymphocytes, especially natural killer (NK) and NKT cells (Gao et al., 2009). Sinusoids are lined by thin, discontinuous endothelial cells with numerous fenestrae (or pores) that allow molecules smaller than 250 kDa to cross the interstitial space (known as the space of Disse) between the endothelium and hepatocytes. Sinusoidal endothelial cells are separated from the hepatocytes by a basement membrane-like matrix, which is not as electron-dense as a regular basement membrane (Friedman, 2000). However, this subendothelial extracellular matrix is important for the normal function of all resident liver cells (Friedman, 2000). The numerous fenestrae and the lack of basement membrane facilitate exchanges of fluids and molecules, such as albumin, between the sinusoid and hepatocytes, but hinder movement of particles larger than chylomicron remnants. Endothelial cells are important in the scavenging of lipoproteins via the apo E receptor and of denatured proteins and advanced glycation endproducts by the scavenger receptor (Enomoto et al., 2004; Elvevold et al., 2008). Hepatic endothelial cells also secrete cytokines, prostanoids, nitric oxide, and endothelins and express intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) on the cell surface (Jaeschke, 1997).

Kupffer cells are the resident macrophages of the liver and constitute approximately 80% of the fixed macrophages in the body (McCuskey, 2006b). Kupffer cells are situated within the lumen of the sinusoid. The primary function of Kupffer cells is to ingest and degrade particulate matter. Also, Kupffer cells are a major source of cytokines and eicosanoids and can act as antigen-presenting cells (APCs; Laskin, 2009). Hepatic stellate cells (HSCs; also known as Ito cells or by the more descriptive terms of fat-storing cells) are located between endothelial cells and hepatocytes (Friedman, 2008). Stellate cells are the major sites for vitamin A storage in the body (Friedman, 2008). Upon activation, these cells can synthesize and excrete collagen and other extracellular matrix proteins and express smooth muscle actin (Friedman, 2000; Lee and Friedman, 2011).

Bile Formation

Bile is a yellow fluid containing bile acids, GSH, phospholipids, cholesterol, bilirubin and other organic anions, proteins, metals, ions, and xenobiotics (Klaassen and Watkins, 1984). Formation of this fluid is a specialized function of the liver. Adequate bile formation is essential for uptake of lipid nutrients from the small intestine (Table 13-1), for protection of the small intestine from oxidative insults (Aw, 1994), and for excretion of endogenous and xenobiotic compounds. Hepatocytes begin the process of bile formation by transporting bile acids, GSH, and other osmotically active compounds including xenobiotics and their metabolites into the canalicular lumen. These molecules are the major driving force for the passive movement of water and electrolytes across the tight junctions and the hepatocyte epithelium. The canalicular lumen is a space formed by specialized regions of the plasma membrane between adjacent hepatocytes (Fig. 13-2). The canaliculi are separated from the intercellular space between hepatocytes by tight junctions, which form a barrier permeable only to water, electrolytes, and to some degree to small organic cations. Under physiological conditions, tight junctions are impermeable to organic anions allowing the high concentrations of bile acids, GSH, bilirubin diglucuronide, and other organic anions in bile. The structure of the biliary tract is analogous to the roots and trunk of a tree, where the tips of the roots equate to the canalicular lumens. Canaliculi form channels between hepatocytes that connect to a series of larger and larger channels or ducts within the liver. The large extrahepatic bile ducts merge into the common bile duct. Bile can be stored and concentrated in the gallbladder before its release into the duodenum. However, the gallbladder is not essential to life and is absent in several species, including the horse, whale, and rat.

With the identification of specific transporters, substantial progress has been made in the understanding of the molecular mechanisms of bile formation (reviewed by Jansen and Groen, 2006; Pauli-Magnus and Meier, 2006). On the basal (sinusoidal) side of the hepatocytes, there are sodium-dependent and sodium-independent uptake systems. Most conjugated bile acids (taurine and glycine conjugates) and some of the unconjugated bile acids are transported into hepatocytes by sodium/taurocholate cotransporting polypeptide (NTCP) (Fig. 13-3) (Trauner and Boyer, 2003; Hagenbuch and Dawson, 2004; Stieger et al., 2007). Sodium-independent uptake of conjugated and unconjugated bile acids is performed by members of the organic anion transporting polypeptides (OATPs) (Hagenbuch and Meier, 2004; Csanaky et al., 2011). OATP1B1 and OATP1B3 are predominantly expressed in liver and are capable of transporting conjugated and unconjugated bile acids and steroids, bromosulfophthalein, and many other organic anions. Furthermore, the OATPs are transporting numerous drugs and also some hepatotoxins, for example, phalloidin, microcystin, and amanitin (Hagenbuch and Gui, 2008; Roth et al., 2011). In addition to the uptake systems, there are ATP-dependent efflux pumps located on the basolateral membrane of hepatocytes. These carriers are members of the multidrug resistance-associated proteins (MRPs; ABCC), which are multispecific transporters for many different anions (Homolya et al., 2003) (Fig. 13-3).

All unconjugated bile acids in hepatocytes are conjugated before being transported by the bile salt export pump (BSEP) across the canalicular membrane. Bile acid excretion is a major driving force of bile formation (bile salt-dependent bile flow). Other constituents of bile are transported by members of the multidrug resistance (MDR) P-glycoprotein family such as MDR3 (ABCC2), which transports phospholipids, and the heterodimeric transporters ABCG5/ABCG8, which transport cholesterol and plant sterols into bile (Pauli-Magnus and Meier, 2006). In addition, MRP2 (a member of the multidrug resistance-associated proteins) transports GSH, which is the main compound responsible for the bile salt-independent bile flow, as well as sulfated and glucuronidated bile acids, glutathione disulfide and glutathione conjugates, bilirubin diglucuronide, and many other conjugated drugs and chemicals (Gerk and Vore, 2002; Borst et al., 2000). Other transport systems of the canalicular membrane include the breast cancer resistance protein (BCRP; ABCG2), which can contribute to the biliary excretion of bile acids and xenobiotics.

Metals are excreted into bile by a series of processes that include (1) uptake across the sinusoidal membrane by facilitated diffusion or receptor-mediated endocytosis; (2) storage in binding proteins or lysosomes; and (3) canalicular secretion via lysosomes, a GSH-coupled event, or use of specific canalicular membrane transporter, for example MRP2 (Ballatori, 2002). Biliary excretion is important in the homeostasis of multiple metals, notably copper, manganese, cadmium, selenium, gold, silver, and arsenic (Klaassen, 1976; Gregus and Klaassen, 1986). Species differences are known for biliary excretion of several toxic metals; for example, dogs excrete arsenic into bile much more slowly than rats. Inability to export copper into bile is a central problem in Wilson's disease, a rare autosomal recessive inherited disorder characterized by a defect or the absence of a copper transporting P-type ATPase (ATP7B). This carrier is located in the trans-Golgi network and transports copper into the secretory pathway for binding to ceruloplasmin and then excretion into bile (Loudianos and Gitlin, 2000). Because biliary excretion is the only way to eliminate copper, a defect in ATP7B results in excessive copper accumulation in hepatocytes, which causes chronic hepatitis and cirrhosis (Loudianos and Gitlin, 2000).

Canalicular lumen bile is propelled forward into larger channels by dynamic, ATP-dependent contractions of the pericanalicular cytoskeleton (Watanabe et al., 1991). Bile ducts, once regarded as passive conduits, modify bile by absorption and secretion of solutes (Lira et al., 1992). Bile acids are taken up into biliary epithelial cells (cholangiocytes) by OATP1A2 (sodium-independent uptake) and by the sodium-dependent bile acid transporter ASBT (Hagenbuch and Dawson, 2004; Hagenbuch and Meier, 2004). These bile acids are then excreted on the basolateral side mainly by MRP3 and heterodimeric organic solute transporter (OSTα/OSTβ) (Ballatori et al., 2005). The bile acids excreted from cholangiocytes return to the portal circulation via the peribiliary plexus (cholehepatic shunt pathway). Biliary epithelial cells also express a variety of phase I and phase II enzymes, which may contribute to the biotransformation of chemical toxicants present in bile (Lakehal et al., 1999).

Secretion into biliary ducts is usually but not always a prelude to toxicant clearance by excretion in feces or urine. Exceptions occur when compounds such as arsenic are repeatedly delivered into the intestinal lumen via bile, efficiently absorbed from the intestinal lumen, and then redirected to the liver via portal blood, a process known as enterohepatic cycling. A few compounds, such as methyl mercury, are absorbed from the biliary tract; the extensive reabsorption of methyl mercury from the gallbladder is thought to contribute to the long biological half-life and toxicity of this toxicant (Dutczak et al., 1991). Alternatively, secretion into bile of toxicant metabolites can be a critical prelude to the development of injury in extrahepatic tissues. A clinically relevant example of bile as an important delivery route for a proximate toxicant is that of diclofenac, a widely prescribed nonsteroidal anti-inflammatory drug (NSAID) that causes small intestinal ulceration. Experiments with mutant rats lacking a functional MRP2 exporter (Fig. 13-3) have shown that these mutants secrete little of the presumptive proximate toxicant metabolite into bile and are resistant to the intestinal toxicity of diclofenac (Seitz and Boelsterli, 1998).

Toxicant-related impairments of bile formation are more likely to have detrimental consequences in populations with other conditions where biliary secretion is marginal. For example, neonates exhibit delayed development of multiple aspects of bile formation, including synthesis of bile acids and the expression of sinusoidal and canalicular transporters (Arrese et al., 1998). Neonates are more prone to develop jaundice when treated with drugs that compete with bilirubin for biliary clearance. Individuals with genetic deficiency of certain transporters are not only at risk for chronic liver injury and fibrosis, but may also be more susceptible to drugs and hepatotoxicants (Jansen and Sturm, 2003; Jansen and Groen, 2006). Patients with sepsis frequently develop cholestasis, which is mainly caused by downregulation of multiple canalicular transport systems (Geier et al., 2006). In addition, direct inhibition of BSEP, as was shown for the endothelin receptor antagonist bosentan, can lead to retention of bile acids in the liver and potentially cell injury (Fattinger et al., 2001).

Mechanisms and Types of Toxicant-Induced Liver Injury

The response of the liver to chemical exposure depends on the intensity of the insults, the cell population affected, and the duration of the chemical exposure (acute vs chronic). Milder stresses may just cause reversible cellular dysfunction, for example, temporary cholestasis after exposure to estrogens, and could cause an adaptive response (conditioning). However, acute poisoning with acetaminophen (APAP) or carbon tetrachloride triggers parenchymal cell necrosis. Exposure to ethanol induces steatosis, which may enhance the susceptibility to subsequent inflammatory insults (Table 13-2). Note that the representative hepatotoxins listed in Table 13-2 include pharmaceuticals (valproic acid, cyclosporin A, diclofenac, APAP, and tamoxifen), recreational drugs (ethanol, ecstasy), a vitamin (vitamin A), metals (Fe, Cu, and Mn), hormones (estrogens, androgens), industrial chemicals (dimethylformamide, methylene dianiline), compounds found in teas (germander) or foods (phalloidin, pyrrolidine alkaloids), and toxins produced by fungi (sporidesmin) and algae (microcystin).

Cell Death

Based on morphology, liver cells can die by two different modes, oncotic necrosis (“necrosis”) or apoptosis. Necrosis is characterized by cell swelling, leakage of cellular contents, nuclear disintegration (karyolysis), and an influx of inflammatory cells. Because necrosis is generally the result of an exposure to a toxic chemical or other traumatic conditions, for example, ischemia, large numbers of contiguous hepatocytes and nonparenchymal cells may be affected. Cell contents released during oncotic necrosis include proteins such a high-mobility group box-1 (HMGB1) and other alarmins, which are a subset of the larger class of damage-associated molecular patterns (DAMPs) (Bianchi, 2007). These molecules are recognized by cells of the innate immune system including Kupffer cells through their toll-like receptors trigger cytokine formation, which orchestrate the inflammatory response after tissue injury. Thus, an ongoing oncotic necrotic process can be identified by the release of liver-specific enzymes such as alanine (ALT) or aspartate (AST) aminotransferase into the plasma and by histology, where areas of necrosis with loss of nuclei and inflammatory infiltrates are easily detectable in H&E sections. In contrast, apoptosis is characterized by cell shrinkage, chromatin condensation, nuclear fragmentation, formation of apoptotic bodies, and, generally, a lack of inflammation. The characteristic morphological features of apoptosis are caused by the activation of caspases, which trigger the activation of enzymes such as caspase-activated DNase (CAD) responsible for internucleosomal DNA fragmentation (Nagata et al., 2003). In addition, caspases can directly cleave cellular and nuclear structural proteins (Fischer et al., 2003). Apoptosis is always a single cell event with the main purpose of removing cells no longer needed during development or eliminating aging cells during regular tissue turnover. Under these conditions, apoptotic bodies are phagocytosed by Kupffer cells or taken up by neighboring hepatocytes. In the absence of cell contents release, the remnants of apoptotic cells disappear without causing an inflammatory response. Because of effective regeneration, apoptotic cell death during normal tissue turnover or even a moderately elevated rate of apoptosis is of limited pathophysiological relevance in the liver. However, if the rate of apoptosis is substantially increased, the apoptotic process cannot be completed. In this case, cells undergo secondary necrosis with breakdown of membrane potential, cell swelling, and release of cell contents (Ogasawara et al., 1993; Bajt et al., 2000). The fundamental difference between oncotic necrosis and secondary necrosis is the fact that during secondary necrosis many apoptotic cells can still be identified based on morphology, many apoptotic characteristics such as activation of various caspases are present, and the process can be completely inhibited by potent pancaspase inhibitors (Jaeschke et al., 2004). Oncotic necrosis does not involve relevant caspase activation and is not inhibitable by caspase inhibitors.

In recent years, signaling mechanisms of apoptosis were elucidated in great detail (Fig. 13-4) (reviewed by Jaeschke, 2006a; Malhi et al., 2006; Schulze-Bergkamen et al., 2006). In the extrinsic pathway of apoptosis, ligands (eg, Fas ligand, TNF-α) bind to their respective death receptor (Fas receptor, TNF receptor type I), which triggers the trimerization of the receptor followed by recruitment of various adapter molecules and procaspases to the cytoplasmic tail of the receptor. The assembly of this death-inducing signaling complex (DISC) leads to the activation of initiator caspases (caspase-8 or -10). In hepatocytes, the active initiator caspase cleaves Bid, a member of the Bcl-2 family of proteins, and the truncated Bid (tBid) translocates together with other Bcl-2 family members such as Bax to the mitochondria. These proteins form pores in the outer membrane of the mitochondria and cause the release of intermembrane proteins such as cytochrome c and the second mitochondria-derived activator of caspases (Smac). Cytochrome c, together with apoptosis protease activating factor-1 (APAF-1), ATP, and procaspase-9, forms the apoptosome causing the activation of caspase-9, which then processes (and activates) downstream effector caspases, for example caspase-3. The effector caspases can propagate the apoptosis signal by activating CAD to initiate nuclear DNA fragmentation and by cleaving numerous cellular proteins critical to cellular function and the structural integrity of the cell and the nucleus (Fischer et al., 2003; Nagata et al., 2003). In addition to downstream substrates, caspase-3 can also process more procaspase-8 and further amplify the apoptotic signal. Although hepatocytes constitutively express Fas and TNF receptors, the death signal generated with ligation of the receptor is in most cases insufficient to trigger apoptosis. Inhibitor studies and experiments with gene-deficient mice support the hypothesis that only the amplification of the receptor-derived signal through multiple mitochondrial cycles can successfully induce apoptosis in hepatocytes (Yin et al., 1999; Bajt et al., 2000). In addition to the direct propagation of the apoptosis signal by mitochondrial cytochrome c release, the simultaneous release of Smac ensures that the cytosolic inhibitors of apoptosis proteins (IAPs) are inactivated and do not interfere with the promotion of apoptosis (Li et al., 2002). Thus, mitochondria are an important part of the extrinsic (receptor-mediated) apoptotic signal transduction pathway in liver cells after most stimuli (type II cells). However, it was recently recognized that under conditions of strong Fas receptor activation (MegaFas ligand), hepatocytes can act as type I cells where downstream caspases are activated without involvement of mitochondria (Schüngel et al., 2009).

In contrast to the extrinsic pathway, the intrinsic or mitochondrial pathway of apoptosis is initiated independent of the TNF receptor family, caspase-8 activation, and formation of the DISC. Despite the upstream differences, the postmitochondrial effects are largely similar to the extrinsic pathway. The intrinsic pathway is generally triggered by a cytotoxic stress or DNA damage, which activates the tumor suppressor p53 (Sheikh and Fornace, 2000). This protein acts as a transcription factor to promote the formation of proapoptotic Bcl-2 family members, for example, Bax. The increased Bax translocation to the mitochondria induces the release of mitochondrial intermembrane proteins including cytochrome c, Smac, endonuclease G, and apoptosis-inducing factor (AIF) (Saelens et al., 2004). An intrinsic mechanism of apoptosis has been discussed for cell death in aging livers (Zhang et al., 2002), prolonged treatment with alcohol (Ishii et al., 2003), or toxicity of benzo(a)pyrene and APAP in hepatoma cells (Boulares et al., 2002; Ko et al., 2004). For other hepatotoxic chemicals, such as carbon tetrachloride (Cai et al., 2005), galactosamine (Gomez-Lechon et al., 2002), and microcystin (Ding et al., 2000), evidence for mitochondria-dependent apoptosis has been reported in cultured hepatocytes and relevant apoptotic cell death was observed after in vivo exposure to these chemicals (Shi et al., 1998; Hooser, 2000; Gujral et al., 2003b).

The dramatically increased knowledge of the signaling mechanisms of apoptotic cell death in hepatocytes lead to the identification of many biochemical “apoptosis” parameters, most of which turned out to be not as specific for apoptosis as originally thought. Prominent examples of these tests are the DNA ladder on agarose gels and the terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assay, which demonstrate internucleosomal DNA fragmentation and DNA strandbreaks, respectively. Originally thought to specifically identify apoptotic cells, both assays are positive for most mechanisms of necrotic cell death (Grasl-Kraupp et al., 1995; Gujral et al., 2002; Jaeschke and Lemasters, 2003). As a result of the misinterpretation of many of these assays, the contribution of apoptosis to the overall pathophysiology processes and toxicological liver injuries is controversially debated (Jaeschke and Lemasters, 2003; Jaeschke et al., 2004; Malhi et al., 2006; Schulze-Bergkamen et al., 2006). However, the controversy can be avoided if the decision to label the process as apoptosis is based primarily on the morphological features of the dying cells. Because the characteristic morphology is caused by the caspase-mediated cleavage of structural proteins within the cell, relevant caspase activation, especially of downstream effector caspases such as caspase-3 or -6, is another hallmark of apoptosis. As a result, pancaspase inhibitors can effectively prevent apoptosis-induced liver injury in vivo and in isolated hepatocytes. It should be emphasized that liver cells contain enough procaspases to execute apoptosis if properly stimulated. Thus, changes in caspase gene or protein expression are not evidence for apoptosis, only a relevant increase of enzyme activity. Once the process is identified as apoptosis, additional parameters can be used to further characterize the signaling mechanism. In addition, the use of positive controls, for example, Fas ligand- or TNF-mediated hepatocellular apoptosis, can be helpful in assessing qualitative and quantitative changes of many parameters relative to a proven apoptotic process (Jaeschke et al., 2004). Another critical issue to consider is the model system that is being used. For example, both the antidiabetic drug troglitazone and the analgesic APAP clearly induce apoptosis in hepatoma cell lines (Yamamoto et al., 2001; Boulares et al., 2002). However, there is no evidence for a relevant role of apoptotic cell death in animals or patients for both drugs (Gujral et al., 2002; Chojkier, 2005; Antoine et al., 2012; McGill et al., 2012b). Thus, characterization of cell death after chemical exposure has to be primarily based on morphology and a number of additional biochemical parameters, which need to quantitatively correlate with the number of apoptotic cells. In addition, the relevance of the model system for the human pathophysiology needs to be considered.

The mechanisms of oncotic necrosis are more diverse and depend on the chemical insult to the cell (a detailed example of the mechanism of APAP-induced hepatocellular necrosis is discussed later). However, a general trend is emerging. Independent of the initial insult and signaling pathways, mitochondria are almost always involved in the pathophysiology (Jaeschke et al., 2012a; Pessayre et al., 2012). The opening of the mitochondrial membrane permeability transition (MPT) pore with collapse of the membrane potential and depletion of cellular ATP is a common final step of the mechanism of necrotic cell death (Kim et al., 2003). The loss of ATP inhibits ion pumps of the plasma membrane resulting in the loss of cellular ion homeostasis, which causes the characteristic swelling of oncotic necrosis. A special form of cell death is the more recently described programmed necrosis (necroptosis) (Vandenabeele et al., 2010). Necroptosis is generally initiated by death receptors, for example TNF receptor 1, and the formation of complex 1 with various adapter molecules including receptor-interacting protein 1 and 3 (RIP1 and -3). If caspase-8 is activated, it will cleave RIP1 and -3 and apoptosis will be initiated. However, if caspase-8 is inhibited, RIP1 and -3 activate a caspase-independent execution mechanism involving mitochondrial oxidant stress and mitochondrial dysfunction (Vandenabeele et al., 2010). Although some of these signaling mechanisms of necroptosis have been described for various cell lines, the importance for liver cell death, especially in vivo, remains unclear.

Canalicular Cholestasis

This form of liver dysfunction is defined physiologically as a decrease in the volume of bile formed or an impaired secretion of specific solutes into bile (Padda et al., 2011). Cholestasis is characterized biochemically by elevated serum levels of compounds normally concentrated in bile, particularly bile salts and bilirubin. When biliary excretion of the yellowish bilirubin pigment is impaired, this pigment accumulates in skin and eyes, producing jaundice, and spills into urine, which becomes bright yellow or dark brown. Because drug-induced jaundice reflects a more generalized liver dysfunction, it is considered a more serious warning sign in clinical trials than mild elevations of liver enzymes (Zimmerman, 1999). The histological features of cholestasis can be very subtle and difficult to detect without ultrastructural studies. Structural changes include dilation of the bile canaliculus and the presence of bile plugs in bile ducts and canaliculi. Toxicant-induced cholestasis can be transient or chronic; when substantial, it is associated with cell swelling, cell death, and inflammation. Cell injury is generally caused by the accumulation of chemicals in the liver, that is, the cholestasis-causing chemical and, as a consequence, potentially cytotoxic bile acids, bilirubin, and other bile constituents. Many different types of chemicals, including metals, hormones, and drugs, cause cholestasis (Table 13-2) (Zimmerman, 1999).

The molecular mechanisms of cholestasis are related to expression and function of transporter systems in the basolateral and canalicular membranes (reviewed by Pauli-Magnus and Meier, 2006; Padda et al., 2011) (Fig. 13-3). In principle, an increased hepatic uptake, decreased biliary excretion, and increased biliary reabsorption (cholehepatic shunting) of a drug may contribute to its accumulation in the liver. Although no case of drug toxicity has been reported in response to modifications of basolateral uptake, OATPs can contribute to the liver injury potential of toxins. The hepatotoxicity of phalloidin, microcystin, and amanitin is facilitated by the uptake through OATPs (Pauli-Magnus and Meier, 2006; Lu et al., 2008). Furthermore, there is a growing list of drugs including rifampicin, bosentan, and troglitazone, which are known to directly inhibit BSEP (Stieger et al., 2000; Fattinger et al., 2001). However, estrogen and progesterone metabolites inhibit BSEP from the canalicular side after excretion by MRP2 (Stieger et al., 2000). A substantial inhibition of bile salt excretion can lead to accumulation of these compounds in hepatocytes and may directly cause cell injury (Palmeira and Rolo, 2004). However, more recent findings indicate that most of the bile acids accumulating in the liver after obstructive cholestasis are nontoxic (Zhang et al., 2012) and instead of cell death cause proinflammatory gene expression in hepatocytes (Allen et al., 2011). Thus, liver injury after obstructive cholestasis is caused mainly by inflammatory cells (Gujral et al., 2003a). However, the increased bile acid levels can trigger compensatory mechanisms, which limit the injury potential of cholestasis (Zollner et al., 2006; Zhang et al., 2012). Bile acids are substrates for the nuclear receptor farnesoid X receptor (FXR). FXR activation stimulates the small heterodimeric partner 1 (SHP1), which downregulates NTCP and limits bile acid uptake (Denson et al., 2001). In addition, FXR activation causes the increased expression of BSEP and MDR3, which enhances the transport capacity for bile acids and phospholipids, respectively, at the canalicular membrane (Ananthanarayanan et al., 2001; Huang et al., 2003). Furthermore, the FXR-independent upregulation of the basolateral transporters MRP3 and MRP4 reduces intracellular bile acid and drug concentrations (Schuetz et al., 2001; Wagner et al., 2003; Fickert et al., 2006). Recent findings indicate that agonists of the nuclear xenobiotic receptors constitutive androstane receptor (CAR) and pregnane X receptor (PXR) can not only induce MRP3 and -4 expression, but also induce bile acid hydroxylation by Cyp3a11 and Cyp2b10 resulting in improved export and detoxification of bile acids during cholestasis (Wagner et al., 2005). In cholangiocytes, OSTα/OSTβ is upregulated at the basolateral membrane during cholestasis (Boyer et al., 2006a). This response, which is dependent on FXR, mediates the enhanced return of bile acids from bile to the plasma (Boyer et al., 2006a). Thus, the pharmacological modulation of transporter expression may counteract some of the detrimental effects of cholestasis with various etiologies (Zollner et al., 2006).

Bile Duct Damage

Another name for damage to the intrahepatic bile ducts is cholangiodestructive cholestasis (Cullen and Ruebner, 1991; Zimmerman, 1999). A useful biochemical index of bile duct damage is a sharp elevation in serum activities of enzymes localized to bile ducts, particularly alkaline phosphatase. In addition, serum levels of bile acids and bilirubin are elevated, as observed with canalicular cholestasis. Initial lesions following a single dose of cholangiodestructive chemicals include swollen biliary epithelium, debris of damaged cells within ductal lumens, and inflammatory cell infiltration of portal tracts. Chronic administration of toxicants that cause bile duct destruction can lead to biliary proliferation and fibrosis resembling primary biliary cirrhosis (PBC). A number of drugs have been implicated to cause prolonged cholestasis with features of PBC (Zimmerman, 1999). However, only in rare cases will there be permanent damage or even loss of bile ducts, a condition known as vanishing bile duct syndrome. Cases of this persisting problem have been reported in patients receiving antibiotics (Davies et al., 1994), anabolic steroids, contraceptive steroids, or the anticonvulsant carbamazepine (Zimmerman, 1999).

Sinusoidal Damage

The sinusoid is, in effect, a specialized capillary with numerous fenestrae for high permeability (Braet and Wisse, 2002). The functional integrity of the sinusoid can be compromised by dilation or blockade of its lumen or by progressive destruction of its endothelial cell wall. Dilation of the sinusoid will occur whenever efflux of hepatic blood is impeded. The rare condition of primary dilation, known as peliosis hepatis, has been associated with exposure to anabolic steroids and the drug danazol. Blockade will occur when the fenestrae enlarge to such an extent that red blood cells become caught in them or pass through with entrapment in the interstitial space of Disse. Endothelial cell gaps and injury have been shown after exposure to APAP (Ito et al., 2005), galactosamine/endotoxin (Ito et al., 2006), or an anti-Fas antibody (Ogasawara et al., 1993). These gaps can be caused by direct injury to endothelial cells by APAP (DeLeve et al., 1997) and the Fas antibody (Bajt et al., 2000) or could be just the result of detachment from the extracellular matrix (Ito et al., 2006). In general, matrix metalloproteinase inhibitors prevent the gap formation (McCuskey, 2006a). A consequence of endothelial cell injury is the loss of barrier function with extensive blood accumulation in the liver resulting in hypovolemic shock. Microcystin produces this effect within hours in rodents (Hooser et al., 1989). Microcystin dramatically deforms hepatocytes by altering cytoskeleton actin filaments, but it does not affect sinusoidal cells (Hooser et al., 1991). Thus, the deformities that microcystin produces on the cytoskeleton of hepatocytes likely produce a secondary change in the structural integrity of the sinusoid owing to the close proximity of hepatocytes and sinusoidal endothelial cells (Fig. 13-2).

Progressive destruction of the endothelial wall of the sinusoid will lead to gaps and then ruptures of its barrier integrity, with entrapment of red blood cells. These disruptions of the sinusoid are considered the early structural features of the vascular disorder known as veno-occlusive disease (DeLeve et al., 1999). Well established as a cause of veno-occlusive disease are the pyrrolizidine alkaloids (eg, monocrotaline, retrorsine, and seneciphylline) found in some plants used for herbal teas and in some seeds that contaminate food grains. Numerous episodes of human and animal poisoning by pyrrolizidine alkaloids have been reported around the world, including massive problems affecting thousands of people in Afghanistan in 1976 and 1993 (Huxtable, 1997). Veno-occlusive disease is also a serious complication in about 15% of the patients given high doses of chemotherapy (eg, cyclophosphamide) as part of bone-marrow transplantation regimens (DeLeve et al., 1999). Selective depletion of GSH within sinusoidal endothelial cells and activation of matrix metalloproteinases are critical events in the mechanism of endothelial cell injury in the pathophysiology of veno-occlusive disease (Wang et al., 2000; DeLeve et al., 2003b). Endothelial cell gap formation and injury and the resulting microcirculatory disturbances have been well established as the cause of veno-occlusive disease (DeLeve et al., 2003a).

Disruption of the Cytoskeleton

Phalloidin and microcystin disrupt the integrity of hepatocyte cytoskeleton by affecting proteins that are vital to its dynamic nature. The detrimental effects of these two potent hepatotoxicants are independent of their biotransformation and are exclusive for hepatocytes, because there is no appreciable uptake of either toxin into other types of cells. Tight binding of phalloidin to actin filaments prevents the disassembly phase of the normally dynamic rearrangement of the actin filament constituent of the cytoskeleton. Phalloidin uptake into hepatocytes leads to striking alterations in the actin-rich web of cytoskeleton adjacent to the canalicular membrane; the actin web becomes accentuated and the canalicular lumen dilates (Phillips et al., 1986). Experiments using time-lapse video microscopy have documented dose-dependent declines in the contraction of canalicular lumens between isolated hepatocyte couplets after incubation with a range of phalloidin concentrations (Watanabe and Phillips, 1986).

Microcystin uptake into hepatocytes leads to hyperphosphorylation of cytoskeletal proteins secondary to this toxicant's covalent binding to the catalytic subunit of serine/threonine protein phosphatases (Runnegar et al., 1995b). Reversible phosphorylations of cytoskeletal structural and motor proteins are critical to the dynamic integrity of the cytoskeleton. Extensive hyperphosphorylation produced by large amounts of microcystin leads to marked deformation of hepatocytes due to a unique collapse of the microtubular actin scaffold into a spiny central aggregate (Hooser et al., 1991). Lower doses of microcystin, insufficient to produce the gross structural deformations, diminish uptake and secretory functions of hepatocytes in association with preferential hyperphosphorylation of the cytoplasmic motor protein dynein (Runnegar et al., 1999). Dynein is a mechanicochemical protein that drives vesicles along microtubules using energy from ATP hydrolysis; central to the hydrolysis of the dynein-bound ATP is a cycle of kinase phosphorylation and phosphatase dephosphorylation. Thus, hyperphosphorylation of dynein freezes this motor pump. Chronic exposure to low levels of microcystin has raised new concerns about the health effects of this water contaminant. Specifically, low levels of microcystin promote liver tumors and kill hepatocytes in the zone 3 region, where microcystin accumulates (Solter et al., 1998).

Information about the binding of phalloidin and microcystin to specific target molecules is valuable for two reasons. First, the linkages of specific binding to loss of target protein functions provide compelling evidence that such a binding constitutes a defined molecular mechanism of injury. Second, the demonstrations of high-affinity binding to a target molecule without confounding effects on other processes or tissues have translated into applications of these toxins as tools for cell biology research. For example, phalloidin complexed with a fluorochrome (eg, rhodamine phalloidin or Texas Red phalloidin) is used to visualize the actin polymer component of the cytoskeleton in all types of permeabilized cells. The collapse of actin filaments into spiny aggregates after microcystin treatment was visualized by fluorescence microscopy of cells stained with rhodamine phalloidin (Hooser et al., 1991). Low levels of microcystin are being used to discriminate the roles of dynein from other cytoskeletal motor proteins (Runnegar et al., 1999).

Fatty Liver

Fatty liver (steatosis) is defined biochemically as an appreciable increase in the hepatic lipid (mainly triglyceride) content, which is <5 wt% in the normal human liver. Histologically, in standard paraffin-embedded and solvent-extracted sections, hepatocytes containing excess fat appear to have multiple round, empty vacuoles that displace the nucleus to the periphery of the cell. Use of frozen sections and special stains is needed to document the contents of the vesicles as fat. Based on the size of the fat droplets, one can distinguish between macrovesicular and microvesicular steatosis.

Currently, the most common cause of hepatic steatosis is insulin resistance due to central obesity and sedentary lifestyle. However, acute exposure to many hepatotoxins, for example, carbon tetrachloride and drugs can induce steatosis (Zimmerman, 1999). Compounds that produce prominent steatosis associated with lethality include the antiepileptic drug valproic acid (Scheffner et al., 1988) and the antiviral drug fialuridine (Honkoop et al., 1997). Ethanol is by far the most relevant drug or chemical leading to steatosis in humans and in experimental animals. Often, drug-induced steatosis is reversible and does not lead to death of hepatocytes. The metabolic inhibitors ethionine, puromycin, and cycloheximide cause fat accumulation without causing cell death. Although steatosis alone may be benign, it can develop into steatohepatitis (alcoholic or nonalcoholic), which is associated with significant liver injury (Farrell, 2002; Pessayre et al., 2002; Stravitz and Sanyal, 2003; Saito et al., 2007; Neuschwander-Tetri, 2010). Steatohepatitis can progress to fibrosis and even hepatocellular carcinoma (Farrell and Larter, 2006). Livers with steatosis can be more susceptible to additional insults such as hepatotoxins (Donthamsetty et al., 2007) or hepatic ischemia (Selzner and Clavien, 2001).

Free fatty acids (FFAs) can be newly synthesized in hepatocytes (mainly from carbohydrate-derived acetyl-coenzyme A). FFAs released from adipose tissue can be taken up into hepatocytes, or they are generated in the liver from hydrolysis of absorbed fat (chylomicrons). Once in the cytosol, FFAs can be imported into mitochondria for degradation (β-oxidation), or esterified into triglycerides for incorporation into very low density lipoproteins (VLDL), which transports the FFAs to the peripheral adipose tissue. FFA uptake into mitochondria depends on the activity of the mitochondrial enzyme carnitine palmitoyl transferase 1, which can be downregulated by malonyl-coenzyme A, the first intermediate of FFA synthesis. Thus, FFA synthesis, consumption, and storage are in a state of equilibrium with no relevant accumulation of triglycerides in the liver (Pessayre et al., 2002). However, if there is chronic excess food consumption with obesity and insulin resistance, excess uptake of FFAs derived from adipose tissue and food into hepatocytes leads to an overload of FFAs, which cannot be degraded and are therefore esterified to triglycerides. One part of the excess triglycerides is incorporated into VLDL, and the other part is stored in the liver gradually leading to steatosis (Pessayre et al., 2002). Drug-induced steatosis is mainly caused by compounds such as 4,4′-diethylaminoethoxyhexestrol, amiodarone, tamoxifen, perhexiline, amineptine, doxycycline, tetracycline, tianeptine, and pirprofen, which accumulate in mitochondria and inhibit β-oxidation and mitochondrial respiration (Berson et al., 1998; Larosche et al., 2007, Pessayre et al., 2012). This effect does not only lead to steatosis, but also to increased reactive oxygen formation and lipid peroxidation. In addition, amineptine, amiodarone, tetracycline, pirprofen, and tianeptine can inhibit directly microsomal triglyceride transfer protein, which lipidates apolipoprotein B to form triglyceride-rich VLDL particles (Letteron et al., 2003). Drugs with this dual effect on β-oxidation and VLDL secretion are generally most steatogenic.

The previously preferred hypothesis of nonalcoholic steatohepatitis (NASH) considered triglyceride accumulation in hepatocytes as the first hit and any additional stress (oxidant stress, lipid peroxidation) as a second hit leading to the progression from steatosis to steatohepatitis (Day and James, 1998a). However, more recent data have clearly demonstrated that triglyceride accumulation does neither cause insulin resistance nor cell injury (Neuschwander-Teri, 2010). A new hypothesis postulates that nonalcoholic fatty liver disease (NAFLD) is mainly caused by lipotoxicity of nontriglyceride fatty acid metabolites (Neuschwander-Tetri, 2010). Although the specific fatty acids or their metabolites causing NAFLD in patients have not been identified, the emerging evidence suggests that the excessive burden of fatty acids in the liver from either inappropriate lipolysis in adipose tissue or synthesis in the liver may cause liver injury (Neuschwander-Tetri, 2010). Mechanisms of lipotoxicity elucidated in cell culture experiments include endoplasmic reticulum stress, activation of the mitochondrial cell death pathway, and lysosomal dysfunction (Ibrahim et al., 2011). Thus, it is likely that the typical histological phenotype of NAFLD may be caused by a variety of different lipotoxicity mechanisms (Neuschwander-Tetri, 2010).

Fibrosis and Cirrhosis

Hepatic fibrosis (scaring) occurs in response to chronic liver injury and is characterized by the accumulation of excessive amounts of fibrous tissue, specifically fibril-forming collagens type I and III, and a decrease in normal plasma membrane collagen type IV (reviewed by Bataller and Brenner, 2005; Rockey and Friedman, 2006; Gutierrez-Ruiz and Gomez-Quiroz, 2007). Fibrosis can develop around central veins and portal tracts or within the space of Disse. The excessive extracellular matrix protein deposition and the loss of sinusoidal endothelial cell fenestrae and of hepatocyte microvilli limit exchange of nutrients and waste material between hepatocytes and sinusoidal blood. With continuing collagen deposition, the architecture of the liver is disrupted by interconnecting fibrous scars. When the fibrous scars subdivide the remaining liver mass into nodules of regenerating hepatocytes, fibrosis has progressed to cirrhosis and the liver has limited residual capacity to perform its essential functions. The primary cause of hepatic fibrosis/cirrhosis in humans worldwide is viral hepatitis. However, biliary obstruction and, in particular, alcoholic and NASH are of growing importance for the development of hepatic fibrosis (Bataller and Brenner, 2005). In addition, fibrosis can be induced by chronic exposure to drugs and chemicals including ethanol and by heavy metal overload (Gutierrez-Ruiz and Gomez-Quiroz, 2007). Repeated treatment with carbon tetrachloride, thioacetamide, dimethylnitrosamine, aflatoxin, or other chemicals has been associated with hepatic fibrosis in experimental animals and humans (Zimmerman, 1999).

Central to the development of fibrosis is the activation of HSC (Fig. 13-2), which are the main cell type producing extracellular matrix proteins (Bataller and Brenner, 2005; Gressner and Weiskirchen, 2006; Rockey and Friedman, 2006). Products formed during liver cell injury initiate HSC activation. Activating signals can be reactive oxygen species and lipid peroxidation products generated in injured hepatocytes. In addition, Kupffer cells can release reactive oxygen and proinflammatory cytokines during the phagocytosis of cell debris or apoptotic bodies, thereby recruiting more inflammatory cells and enhancing the injury and oxidant stress (Tsukamoto, 2002). Damaged sinusoidal endothelial cells contribute to the activation of HSC by generating a splice variant of cellular fibronectin and by release of the urokinase-type plasminogen activator, which processes latent transforming growth factor-β1 (TGF-β1) (Friedman, 2000). Furthermore, accumulating platelets at the site of injury can produce TGF-β1 and growth factors such as platelet-derived growth factor (PDGF) (Rockey and Friedman, 2006). Together these stimuli cause the activation of HSC, which undergo phenotypic changes including proliferation, fibrogenesis and matrix remodeling, chemotaxis and proinflammatory mediator formation, and contractility (Rockey and Friedman, 2006). The proliferation of HSC is induced by the formation of mitogens such as PDGF. In addition, it also involves the upregulation of the PDGF receptor, which further enhances the responsiveness of HSC to this mitogen (Pinzani, 2002). Another hallmark of HSC activation is enhanced contractility due to the increased expression of α-smooth muscle actin. Increased expression of endothelin-1 receptors on HSC together with the general imbalance between vasodilator (nitric oxide, carbon monoxide) and vasoconstrictor (ET-1) formation contributes to the development of portal hypertension during fibrosis (Rockey, 2003). The increased accumulation of HSC at sites of injury is caused by migration and proliferation of HSC. PDGF, monocyte chemotactic protein 1 (MCP-1), and other chemokines have been shown to be HSC chemoattractants (Marra, 2002). One of the central events in HSC activation is the excessive formation of extracellular matrix proteins induced mainly by TGF-β1. The effects of TGF-β1, which is generated to a large degree by HSCs, are amplified by the increased expression of TGF-β receptors on HSC (Gressner and Weiskirchen, 2006). However, during fibrogenesis there is not only an overall increase in extracellular matrix deposition, but also a change from the basement membrane-like matrix dominated by nonfibril-forming collagens (types IV, VI, and XIV) to a basement membrane-type matrix involving fibril-forming collagen types I and III. The effect is caused by the differential expression and release of matrix metalloproteinases (MMPs) and their inhibitors (TIMPs) from HSC and Kupffer cells. TIMP1 and TIMP2 are upregulated and MMP1 (collagenase I) is downregulated during fibrogenesis leading to the reduced degradation of fibril-forming collagens, for example, type I. At the same time, MMP2 and -9 (collagenase IV) are activated causing the accelerated degradation of nonfibril-forming collagens (Arthur, 2000). Previously, it was assumed that fibrotic changes, especially the state of cirrhosis, were irreversible. However, more recent insight into the pathophysiology indicated the possibility for reversal of fibrosis. Stimulation of apoptosis in activated HSC and the expression of different MMPs and TIMPs can reduce matrix deposition and enhance degradation resulting in a gradual reversal of fibrosis (Arthur, 2000; Bataller and Brenner, 2005; Rockey and Friedman, 2006). This area is of considerable interest for pharmaceutical intervention.

Tumors

Chemically induced neoplasia can involve tumors that are derived from hepatocytes, bile duct progenitor cells, the ductular “bipolar” progenitor cells, and the periductular stem cells (Sell, 2002). The rare, highly malignant angiosarcomas are derived from sinusoidal lining cells. Hepatocellular cancer has been linked to chronic abuse of androgens, alcohol, and a high prevalence of aflatoxin-contaminated diets. In addition, viral hepatitis, metabolic diseases such as hemochromatosis and α₁-antitrypsin deficiency, and NASH are major risk factors for hepatocellular carcinoma (Zimmerman, 1999; McKillop et al., 2006; Wands and Moradpour, 2006). The synergistic effect of coexposure to aflatoxin and hepatitis virus B is well recognized (Henry et al., 2002). The prevalence of hepatitis B and C viruses and environmental factors make hepatocellular carcinoma one of the most common malignant tumors worldwide (Bosch et al., 2005). Angiosarcomas have been associated with occupational exposure to vinyl chloride and arsenic (Zimmerman, 1999). Exposure to Thorotrast (radioactive thorium dioxide used as contrast medium for radiology) has been linked to tumors derived from hepatocytes, sinusoidal cells, and bile duct cells (cholangiocarcinoma) (Zimmerman, 1999). The compound accumulates in Kupffer cells and emits radioactivity throughout its extended half-life. One study of Danish patients exposed to Thorotrast found that the risk for bile duct and gallbladder cancers was increased 14-fold and that for liver cancers more than 100-fold (Andersson and Storm, 1992). Furan is the only chemical known to cause cholangiocarcinomas experimentally in rats (Hickling et al., 2010).

The molecular pathogenesis of hepatocellular carcinoma is complex and poorly understood. The malignant transformation of hepatocytes occurs as a result of increased cell turnover due to chronic liver injury, persistent inflammation, regeneration, and cirrhosis (Wands and Moradpour, 2006). Direct DNA binding of carcinogens or their reactive metabolites (eg, aflatoxin metabolites) or indirect DNA modifications by reactive oxygen species generated during inflammation and cell injury can lead to genetic alterations in hepatocytes resulting in impaired DNA repair, the activation of cellular oncogenes, and inactivation of tumor suppressor genes. An overall imbalance between stimulation of proliferation and inhibition of apoptosis in the liver leads to the survival and expansion of these preneoplastic cells (Fabregat et al., 2007). This concept is supported by the observation that 30% of hepatocellular carcinomas show mutations in the tumor suppressor gene p53; the mutation rate is up to 70% in areas with high aflatoxin exposure (Wands and Moradpour, 2006). The functional inactivation of p53 by mutations prevents the induction of apoptosis. Because most chemotherapeutic agents require p53 to induce apoptosis in these cancer cells, hepatocellular carcinomas are mostly resistant to conventional chemotherapy (Bruix et al., 2006). However, p53 mutations alone are not sufficient to initiate carcinogenesis. It was shown that telomere dysfunction and chromosomal instability in combination with p53 mutations are critical for the progression from neoplasms to malignant carcinomas (Farazi et al., 2006). Because telomere dysfunction (shortening) limits the capacity of cancer cells to proliferate, the activity of the telomere-synthesizing enzyme telomerase is activated in advanced hepatocellular carcinomas (Satyanarayana et al., 2004). Stabilization and repair of telomeres promote the expansion of the tumor. Additional tumor cell survival mechanisms include the disruption of TGF-β apoptosis signaling and activation of phosphatidylinositol-3-kinase/AKT survival pathways (Thorgeirsson et al., 1998). Furthermore, NF-κB activation during inflammation is responsible for induction of prosurvival genes such as Bcl-X_L and XIAP, the downregulation of the Fas receptor on hepatocytes, and the reduced expression of the proapoptotic Bax gene (Fabregat et al., 2007). These effects are combined with the overexpression and dysregulated signaling of promitogenic and antiapoptotic growth factors such as insulin-like growth factor (IGF), hepatocyte growth factor (HGF), wingless (Wnt), and transforming growth factor-α (TGF-α)/epidermal growth factor (EGF) (Breuhahn et al., 2006). Many of these pathways may offer novel therapeutic targets to prevent or eliminate hepatocellular carcinoma (Aravalli et al., 2008).

Critical Factors in Toxicant-Induced Liver Injury

Why is the liver the target site for many chemicals of diverse structure? Why do many hepatotoxicants preferentially damage one type of liver cell? Our understanding of these fundamental questions is incomplete. Influences of several factors are of obvious importance (Table 13-3). Location and specialized processes for uptake and biliary secretion produce higher exposure levels in the liver than in other tissues of the body, and strikingly high levels within certain types of liver cells. Then, the abundant capacity for bioactivation reactions influences the rate of exposure to proximate toxicants. Subsequent events in the pathogenesis appear to be critically influenced by responses of sinusoidal cells and the immune system. Discussion of the evidence for the contributions of these factors to the hepatotoxicity of representative compounds requires commentary about mechanistic events.

In vitro systems using tissue slices, the isolated perfused liver, primary isolated cultured liver cells, and cell fractions allow observations at various levels of complexity without the confounding influences of other systems. Some hepatoma cell lines with maintained drug metabolizing enzyme expression can be useful to study mechanisms of drug hepatotoxicity (Guguen-Guillouzo et al., 2010; McGill et al., 2011). Models using cocultures or chemicals that inactivate a given cell type can document the contributions and interactions between cell types. A limitation of these in vitro systems is the fact that the artificial cell culture conditions will modify the basal gene expression profile (Boess et al., 2003) and may influence or even dominate the response of cells to a chemical. In particular, the generally high oxygen concentrations used during cell culture (room air, 21% oxygen) compared to physiologically relevant oxygen levels (4%–10% oxygen) can affect the mechanism of drug-induced cell injury (Yan et al., 2010). Whole animal models are essential for assessment of the progression of injury and responses to chronic insult and the confirmation of in vitro results. Use of chemicals that induce, inhibit, deplete, or inactivate gene products can define roles of specific processes, although potential influences of nonspecific actions can confound interpretations. Application of molecular biology techniques for gene transfection or repression attenuates some of these interpretive problems. Gene knockout animals are extremely useful models to study complex aspects of hepatotoxicity. However, it is critical that the relevance of in vivo systems for the human pathophysiology is being established. In addition, compensatory responses due to the loss of a specific gene always need to be considered when using gene knockout animals (Ni et al., 2012b).

Uptake and Concentration

Hepatic “first pass” uptake of ingested chemicals is facilitated by the location of the liver downstream of the portal blood flow from the gastrointestinal tract. Lipophilic compounds, particularly drugs and environmental pollutants, readily diffuse into hepatocytes because the fenestrated epithelium of the sinusoid enables close contact between circulating molecules and hepatocytes. Thus, the membrane-rich liver concentrates lipophilic compounds. Other toxins are rapidly extracted from blood because they are substrates for transporters located on the sinusoidal membrane of hepatocytes (Hagenbuch and Gui, 2008; Roth et al., 2011).

Phalloidin and microcystin are illustrative examples of hepatotoxins that target the liver as a consequence of extensive uptake into hepatocytes by sinusoidal transporters (Frimmer, 1987; Runnegar et al., 1995a). Ingestion of the mushroom Amanita phalloides is a common cause of severe, acute hepatotoxicity in continental Europe and North America. Microcystin has produced numerous outbreaks of hepatotoxicity in sheep and cattle that drank pond water containing the blue-green alga Microcystis aeruginosa. An episode of microcystin contamination of the water source used by a hemodialysis center in Brazil led to acute liver injury in 81% of the 124 exposed patients and the subsequent death of 50 of these patients (Jochimsen et al., 1998). Microcystin contamination was verified by analysis of samples from the water-holding tank at the dialysis center and from the livers of patients who died. This episode indicates the vulnerability of the liver to toxicants regardless of the route of administration. Because of its dual blood supply from both the portal vein and the hepatic artery, the liver is presented with appreciable amounts of all toxicants in the systemic circulation.

An early clue to preferential uptake as a factor in phalloidin's target-organ specificity was the observation that bile duct ligation, which elevates systemic bile acid levels, protects rats against phalloidin-induced hepatotoxicity in association with an 85% decrease in hepatic uptake of phalloidin (Walli et al., 1981). Subsequent studies found that cotreatment with substrates (eg, cyclosporin A, rifampicin) known to prevent the in vivo hepatotoxicity of phalloidin or microcystin would also inhibit their uptake into hepatocytes by sinusoidal transporters for bile acids and other organic anions (Ziegler and Frimmer, 1984; Runnegar et al., 1995a). Recently, conclusive evidence utilizing Oatp1b2-null mice has demonstrated that the OATP transporter family is responsible for the hapatic uptake and toxicity of phalloidin (Lu et al., 2008).

Accumulation within liver cells by processes that facilitate uptake and storage is a determining factor in the hepatotoxicity of vitamin A and several metals. Vitamin A hepatotoxicity initially affects stellate cells, which actively extract and store this vitamin. Early responses to high-dose vitamin A therapy are stellate cell engorgement, activation, increase in number, and protrusion into the sinusoid (Geubel et al., 1991). Cadmium hepatotoxicity becomes manifest when the cells exceed their capacity to sequester cadmium as a complex with the metal-binding protein, metallothionein (MT). This protective role for MT was definitively documented by observations with MT transgenic and knockout mice. Overexpression of MT in the transgenic mice rendered them more resistant than wild-type animals to the hepatotoxicity and lethality of cadmium poisoning (Liu et al., 1995). In contrast, MT gene knockout mice were dramatically more susceptible to cadmium heptatotoxicity (Liu et al., 1996).

Iron poisoning produces severe liver damage. Hepatocytes contribute to the homeostasis of iron by extracting this essential metal from the sinusoid by a receptor-mediated process and maintaining a reserve of iron within the storage protein ferritin. Acute Fe toxicity is most commonly observed in young children who accidentally ingest iron tablets (Chang and Rangan, 2011). The cytotoxicity of free iron is attributed to its function as an electron donor for the Fenton reaction, where hydrogen peroxide is reductively cleaved to the highly reactive hydroxyl radical, an initiator of lipid peroxidation. Accumulation of excess iron beyond the capacity for its safe storage in ferritin is initially evident in the zone 1 hepatocytes, which are closest to the blood entering the sinusoid. Thus, the zone 1 pattern of hepatocyte damage after iron poisoning is attributable to location for (1) the preferential uptake of iron and (2) the higher oxygen concentrations that facilitate the injurious process of lipid peroxidation (Table 13-3). Chronic hepatic accumulation of excess iron in cases of hemochromatosis is associated with a spectrum of hepatic disease including a greater than 200-fold increased risk for liver cancer (Ramm and Ruddell, 2005).

Bioactivation and Detoxification

One of the vital functions of the liver is to eliminate exogenous chemicals and endogenous intermediates. Therefore, hepatocytes contain high levels of phase I enzymes, which have the capacity to generate reactive electrophilic metabolites. Hepatocytes also have a wide variety of phase II enzymes, which enhance the hydrophilicity by adding polar groups to lipophilic compounds and target these conjugates to certain carriers in the canalicular or plasma membrane for excretion. Generally, phase II reactions yield stable, nonreactive metabolites. Although electrophiles may be effectively conjugated and excreted, if the intermediate is highly reactive, some of these compounds can react with proteins and other target molecules before an interaction with a phase II enzyme is possible. In contrast, if the amount of the reactive metabolite exceeds the capacity of the hepatocyte to detoxify it, covalent binding to cellular macromolecules will occur and potentially result in cell injury (Park et al., 2011). Thus, the balance between phase I reactions, which generate the electrophile, and conjugating phase II reactions determines whether a reactive intermediate is safely detoxified or may cause cell dysfunction or injury. Because the expression of phase I and II enzymes and of the hepatic transporters can be influenced by genetics (eg, polymorphism of drug-metabolizing enzymes) and lifestyle (eg, diet, consumption of other drugs and alcohol), the susceptibility to potential hepatotoxicants can vary markedly between individuals. Several prominent and important examples are discussed.

Acetaminophen

One of the most widely used analgesics, acetaminophen (APAP) is a safe drug when used at therapeutically recommended doses. However, an overdose can cause severe liver injury and even liver failure in experimental animals and in humans (Lee, 2004). About half of all overdose cases are caused by suicide attempts, but an increasing number of cases are reported with unintentional overdosing (Larson et al., 2005). Although the toxicity is a rare event compared to the millions of patients taking the drug daily, APAP-mediated liver injury represents a significant clinical problem. During the last 10 years, APAP-induced hepatotoxicity became the most frequent cause of drug-induced liver failure in the United States and in the United Kingdom (Lee, 2004; Larson et al., 2005).

Because >90% of a therapeutic dose of APAP is conjugated with sulfate or glucuronide, the limited formation of a reactive metabolite, that is, N-acetyl-p-benzoquinone imine (NAPQI), poses no risk for liver injury. In fact, long-term studies with APAP in osteoarthritis patients did not reveal any evidence of liver dysfunction or cell injury even in patients consuming the maximal recommended daily dose of APAP for 12 months (Kuffner et al., 2006; Temple et al., 2006). In contrast, after an overdose, the formation of large amounts of NAPQI leads first to depletion of cellular GSH stores and subsequently causes covalent binding of NAPQI to intracellular proteins (Jollow et al., 1973; Mitchell et al., 1973) (Fig. 13-5). The generally higher levels of P450 enzymes combined with the lower GSH content in centrilobular hepatocytes are the main reasons for the predominant centrilobular necrosis observed after APAP poisoning. Consistent with the critical role of protein binding for cell injury are the findings that APAP protein adducts are located predominantly in centrilobular hepatocytes undergoing necrosis (Roberts et al., 1991) and that no APAP hepatotoxicity is observed without protein binding (Nelson, 1990). Because protein binding can be prevented by conjugation of NAPQI with GSH, any manipulation that reduces hepatic GSH levels, for example, fasting or protein malnutrition, potentially enhances the toxicity of APAP. In contrast, interventions such as the supply of cysteine, the rate-limiting amino acid for GSH synthesis, promote the detoxification of NAPQI and limit cell injury (Mitchell et al., 1973). Based on this fundamental insight into the mechanism of APAP hepatotoxicity, N-acetylcysteine was introduced in the clinic as intervention therapy (Smilkstein et al., 1988). This highly successful approach, which saved the lives of many patients who took an APAP overdose, is still the most effective treatment available (Lee, 2004). More recent evidence indicates that N-acetyl cysteine treatment not only promotes cytosolic GSH synthesis to detoxify NAPQI, but also replenishes the depleted mitochondrial GSH content, which scavenges reactive oxygen and peroxynitrite. In addition, excess N-acetyl cysteine is degraded and supports the mitochondrial energy metabolism (Saito et al., 2010b).

A significant factor in APAP hepatotoxicity can be the consumption of alcoholic beverages. In addition to potential malnutrition in alcoholics, ethanol is a potent inducer of CYP2E1, which is the main enzyme responsible for the metabolic activation of APAP in humans (Gonzalez, 2007). Whereas the simultaneous exposure of ethanol and APAP competitively inhibits NAPQI formation and therefore prevents APAP-induced toxicity (Sato and Lieber, 1981), the increased expression of CYP2E1 can enhance APAP toxicity after ethanol metabolism (Gonzalez, 2007). In addition, the presence of higher-chain alcohols, for example, isopentanol, in alcoholic beverages can induce additional P450 isoenzymes such as CYP3A, which can significantly enhance APAP hepatotoxicity (Sinclair et al., 2000; Guo et al., 2004). Despite the clear experimental evidence that alcohol consumption can increase the susceptibility to APAP (Sato et al., 1981) and the clinical observation of severe APAP hepatotoxicity in alcoholics, it remains controversial whether alcohol can actually induce hepatotoxicity at therapeutic doses of APAP as suggested by some case reports (Zimmerman and Maddrey, 1995). However, an extensive review of the literature involving APAP consumption in alcoholics suggests no relevant risk for APAP hepatotoxicity at therapeutic levels in this patient population (Dart et al., 2000). In addition, a randomized, double-blind, placebo-controlled trial with multiple therapeutic doses of APAP showed no evidence of liver dysfunction or cell injury in alcoholics (Kuffner et al., 2001). Thus, alcohol consumption does not increase the risk for liver injury after therapeutic doses of APAP. This finding may apply to the potential interaction with other drugs and dietary chemicals. Nevertheless, consistent with experimental data and clinical experience, inducers of CYPs aggravate liver injury after a hepatotoxic dose of APAP.

Although the focus of early mechanistic investigations was on the role of covalent binding in APAP-induced hepatotoxicity, it became apparent during the last decade that protein adduct formation is an important biomarker for APAP overdose (Davern et al., 2006), but protein binding alone was not sufficient to explain cell injury (Fig. 13-5). In fact, low levels of protein adducts are even observed after therapeutic doses of APAP in mice (McGill et al., 2012a) and humans (Heard et al., 2011). Because no APAP-induced cell injury is observed without covalent binding of NAPQI to cellular proteins, in particular mitochondrial proteins, it is considered a critical initiating event of the toxicity that requires amplification (Jaeschke et al., 2003, 2012a). Mitochondrial protein binding causes inhibition of mitochondrial respiration, a selective mitochondrial oxidant stress, mitochondrial peroxynitrite formation, and declining ATP levels in the liver (Jaeschke and Bajt, 2006). The early mitochondrial translocation of Bax and Bid, members of the Bcl-2 family of proteins, triggers the release of mitochondrial intermembrane proteins including endonuclease G and AIF (Jaeschke and Bajt, 2006). These endonucleases, which translocate to the nucleus after APAP exposure, cause the initial nuclear DNA fragmentation after mitochondrial Bax pore formation (Jaeschke et al., 2012a). However, the continued exposure of GSH-depleted mitochondria to peroxynitrite results in nitration of mitochondrial proteins and mitochondrial DNA modifications (Cover et al., 2005). The continued oxidant stress will eventually trigger the MPT pore opening with breakdown of the membrane potential, mitochondrial swelling, and rupture of the outer membrane (Kon et al., 2004; Ramachandran et al., 2011). These events lead to the loss of mitochondrial ATP synthesis capacity, more extensive nuclear DNA fragmentation due to the amplified release of intermembrane proteins after the MPT, and eventually oncotic necrotic cell death (Gujral et al., 2002). Because of the central role of the mitochondrial oxidant stress in APAP hepatotoxicity, its regulation could be an important therapeutic target. Once initiated by early protein binding, the mitochondrial oxidant stress triggers activation (phosphorylation) of c-jun-N-terminal kinase (JNK), which then translocates to the mitochondria and amplifies the mitochondrial oxidant stress (Han et al., 2010; Jaeschke et al., 2012a). JNK is phosphorylated by apoptosis signal regulating kinase-1 (ASK-1), which is liberated from the ASK-1-thioredoxin-1 complex in the cytosol after oxidation of thioredoxin-1 by the mitochondrial oxidant stress (Han et al., 2010; Jaeschke et al., 2012a) (Fig. 13-5).

In addition to these intracellular signaling mechanisms leading to cell death, additional events may expand the area of necrosis. The release of calpains, which are Ca²⁺-activated proteases, during necrosis can promote further cell injury in neighboring cells (Mehendale and Limaye, 2005). Likewise, the release of DNase-1 enhances nuclear DNA fragmentation in adjacent cells and aggravates the injury after APAP overdose. Also, the release of intracellular proteins such as the nuclear protein HMGB-1 from necrotic cells can stimulate macrophages to produce proinflammatory cytokines. This way, the necrotic cell death during APAP hepatotoxicity can promote an innate immune response with recruitment of neutrophils and other inflammatory leukocytes, which may clear cell debris and prepare for regeneration of the lost tissue (Jaeschke et al., 2012b) but, under certain conditions, may cause additional injury (see the “Inflammation” section).

Although many details of the mechanism still remain to be elucidated, the newly gained insight into signaling events in response to APAP overdose suggests two fundamentally new developments. First, necrotic cell death is in most cases not caused by a single catastrophic event but can be the result of a cellular stress, which is initiated by metabolic activation and triggers sophisticated signaling mechanisms culminating in cell death (Fig. 13-5). Second, the multitude of events following the initial stress offers many opportunities for therapeutic interventions at later time points. Because these events are not occurring in all cells to the same degree and at the same time, delayed interventions may not completely prevent cell damage but limit the area of necrosis enough to prevent liver failure. Delayed treatment with GSH to accelerate the recovery of mitochondrial GSH levels effectively scavenged peroxynitrite, reduced the area of necrosis, and promoted regeneration resulting in improved survival after APAP overdose (Bajt et al., 2003). Overexpression of calpastatin, an inhibitor of calpains, attenuated APAP-induced liver injury and enhanced survival (Limaye et al., 2006). Delayed treatment with a JNK inhibitor attenuated the mitochondrial oxidant stress and prevented the MPT and tissue injury (Hanawa et al., 2008; Saito et al., 2010a). Further support for the central role of mitochondrial dysfunction in APAP toxicity in hepatocytes comes from the recent observation that removal of damaged mitochondria by autophagy (mitophagy) limits APAP-induced liver injury in vivo and in cultured cells (Ni et al., 2012a). Together these findings underscore the concept that the later stages of APAP-induced liver injury can be potentially affected at the level of intracellular signaling in hepatocytes and during the propagation of the injury to neighboring cells.

Ethanol

Morbidity and mortality associated with the consumption of alcohol is mainly caused by the toxic effects of ethanol on the liver (Stewart and Day, 2006). This targeted toxicity is due to the fact that >90% of a dose of ethanol is metabolized in the liver. Three principal pathways of ethanol metabolism are known (Fig. 13-6). Alcohol dehydrogenase (ADH) oxidizes ethanol to acetaldehyde with a Km of 1 mM; the electrons are transferred to NAD⁺, which leads to the production of NADH. Acetaldehyde is further oxidized to acetate in a NAD-dependent reaction by acetaldehyde dehydrogenase (ALDH). This pathway is mainly regulated by the mitochondrial capacity to utilize NADH and regenerate NAD⁺ (Stewart and Day, 2006). The formation of excess reducing equivalents and acetate stimulates fatty acid synthesis and is a major factor in the development of alcohol-induced steatosis. Both ADH and ALDH exhibit genetic polymorphisms and ethnic variations, which play a role in the development of alcoholism and liver damage (Agarwal, 2001; Day, 2006). A toxicologically relevant polymorphism involves the mitochondrial ALDH2, where the ALDH2*2 form shows little or no catalytic activity. The increased levels of acetaldehyde present in individuals that carry this polymorphism is thought to cause the “flushing” syndrome after ethanol exposure. The inactive form of ALDH is found in 50% of Asians but is absent in Caucasians. This may be the reason for the overall reduced incidence of alcoholism in Asia compared to Europe and North America (Chen et al., 1999). However, heterozygotes of ALDH2*2 were found to develop more severe liver injury in response to lower alcohol consumption, suggesting a higher susceptibility to alcoholic liver disease (Enomoto et al., 1991). These findings underscore the importance of acetaldehyde in the pathophysiology.

The second major pathway involves the alcohol-inducible enzyme CYP2E1, which oxidizes ethanol to acetaldehyde (Fig. 13-6). The enzyme is located predominantly in hepatocytes of the centrilobular region and requires oxygen and NADPH. Because the Km of CYP2E1 for ethanol is >10 mM, this reaction is most relevant for high doses of ethanol and, due to the enzyme's inducibility, for chronic alcoholism (Stewart and Day, 2006). The third pathway involves catalase in peroxisomes. In this reaction, ethanol functions as an electron donor for the reduction of hydrogen peroxide to water. Thus, the capacity of this pathway is limited due to the low levels of hydrogen peroxide. It is estimated that <2% of an ethanol dose is metabolized through this pathway (Stewart and Day, 2006).

The mechanisms of alcohol-induced liver disease are complex and still incompletely understood. Steatosis is a common feature of chronic alcohol consumption. It is caused by the excessive supply of acetate and NADH, which promotes fatty acid synthesis. In addition, ethanol and acetaldehyde inhibit the DNA binding of peroxisome proliferator-activated receptor-α (PPAR-α), which regulates constitutive and inducible expression of mitochondrial and peroxisomal fatty acid metabolizing enzymes (Aoyama et al., 1998). In addition to the enhanced synthesis and reduced consumption of fatty acids, ethanol exposure inhibits the transfer of triglycerides from liver to adipose tissue. Acetaldehyde inhibits the microsomal triglyceride transfer protein, which incorporates triglycerides into VLDL (Lin et al., 1997), and disrupts the export mechanism of VLDL by interfering with microtubular function (Kannarkat et al., 2006). These effects of ethanol and its metabolites can be compounded in the presence of a high-fat diet. Although steatosis alone does generally not develop into more severe liver disease, it has been hypothesized that it plays a critical role in the advancement of the disease process (Day and James, 1998a, b). Steatosis is considered the “first hit,” which requires a “second hit” to progress to severe alcoholic liver disease (Day and James, 1998a, b). However, more recent data support the concept of lipotoxicity as a critical determinant of disease progression (Neuschwander-Tetri, 2010). CYP2E1 is a relevant source of reactive oxygen formation during ethanol metabolism (Dey and Cederbaum, 2006). This intracellular oxidant stress in hepatocytes can ultimately induce mitochondrial dysfunction and cell death of hepatocytes, but also activate stellate cells and promote fibrosis (Dey and Cederbaum, 2006). In addition to the intracellular events, alcohol exposure causes an inflammatory response, which contributes to the oxidant stress (Arteel, 2003; Hines and Wheeler, 2004). Gut-derived endotoxin and other bacteria-derived products can activate Kupffer cells through toll-like receptor activation to produce reactive oxygen species and cytokines such as TNF-α (Seki and Schnabl, 2012). The formation of these mediators can be further amplified by feedback loops, which enhance cytokine and chemokine formation through priming of the redox-sensitive transcription factor NF-κB in Kupffer cells (Arteel, 2003). In addition, TNF-α can induce the inducible nitric oxide synthase (iNOS, NOS2) leading to the formation of peroxynitrite, a potent oxidant and nitrating species (Arteel, 2003). TNF-α can also directly promote cell death by acting on hepatocytes, which are primed by ethanol-induced depletion of mitochondrial GSH (Colell et al., 1998). Inhibition of the proteasome pathway, a well-recognized feature of chronic alcohol exposure, can enhance chemokine formation in hepatocytes and promote inflammatory liver injury (McClain et al., 2005). Additional proinflammatory mediators and immune responses can be triggered by protein adducts of acetaldehyde and malondialdehyde (Freeman et al., 2005) and the release of other DAMPs (Miller et al., 2011). Interestingly, ethanol inhibits hepatic NK cells. As NK cells can kill HSC, the major cell type promoting hepatic fibrosis, ethanol can indirectly support fibrogenesis by preventing the elimination of activated stellate cells (Miller et al., 2011). Another defense mechanism that is activated during alcoholic liver disease is autophagy, which can remove damaged cell organelles and modified proteins (Ding et al., 2011) and thereby reduce the activation of the innate immune response. Thus, alcoholic liver disease is a complex interplay between the activation of pro-cell death mechanisms (reactive metabolite formation, oxidant stress, protein adducts, and stimulation of proinflammatory and profibrotic innate immune responses) and activation of defense mechanisms (antioxidants, autophagy, and NK cell activation) (Gao and Bataller, 2011).

Allyl Alcohol

An industrial chemical used in the production of resins, plastics, and fire retardants, allyl alcohol is also used as a model hepatotoxin due to its preferential periportal (zone 1) hepatotoxicity. The alcohol is metabolized by ADH to acrolein, a highly reactive aldehyde, which is then further oxidized by ALDH to acrylic acid. The fact that the toxicity depends on depletion of hepatic GSH levels is prevented by inhibitors of ADH but enhanced by inhibitors of ALDH suggests that acrolein formation is the critical event in liver injury (Jaeschke et al., 1987; Rikans, 1987). Age and gender differences in allyl alcohol hepatotoxicity can be explained by variations in the balance between ADH and ALDH expression (Rikans and Moore, 1987). The preferential occurrence of allyl alcohol injury in zone 1 hepatocytes (Table 13-3) is caused by the predominant uptake of allyl alcohol in the periportal region and the oxygen dependence of the toxicity (Badr et al., 1986). Although protein binding of the reactive metabolite acrolein and subsequent adduct formation appears to be the main cause of liver cell death (Kaminskas et al., 2004), lipid peroxidation can become a relevant mechanism of cell injury under conditions of a compromised antioxidant status (Jaeschke et al., 1987). Lipid peroxidation is caused by a reductive stress where the excessive NADH formation leads to mobilization of redox-active iron from storage proteins (Jaeschke et al., 1992).

Carbon Tetrachloride

Cytochrome P450-dependent conversion of CCl₄ to •CCl₃ and then to CCl₃OO• is the classic example of xenobiotic bioactivation to a free radical that initiates lipid peroxidation by abstracting a hydrogen atom from the polyunsaturated fatty acid of a phospholipid (Recknagel et al., 1989; Weber et al., 2003). The metabolic activation of CCl₄ involves primarily CYP2E1 in vivo as indicated by the absence of toxicity in CYP2E1 knockout mice (Wong et al., 1998). CCl₄-induced lipid peroxidation increases the permeability of the plasma membrane to Ca²⁺, leading to severe disturbances of the calcium homeostasis and necrotic cell death (Weber et al., 2003). Recent research indicates that CCl₄ also induces significant mitochondrial damage, which is dependent on lipid peroxidation events and on CYP2E1 activity (Knockaert et al., 2012). In addition, the •CCl₃ radical can directly bind to tissue macromolecules and some of the lipid peroxidation products are reactive aldehydes, for example, 4-hydroxynonenal, which can form adducts with proteins (Weber et al., 2003). In addition to the intracellular events, Kupffer cell activation can contribute to liver injury (elSisi et al., 1993). Kupffer cells may enhance the injury by oxidant stress (elSisi et al., 1993) or TNF-α generation, which may lead to apoptosis (Shi et al., 1998). In support of these different components of the mechanism of CCl₄-mediated cell and organ damage, beneficial effects were shown with inhibition of CYPs, preservation of Ca²⁺ homeostasis, antioxidants, and anoxia (Weber et al., 2003). In contrast, treatments with chemicals that induce CYP2E1, for example, ethanol or acetone, enhance the injury. This was confirmed in humans. A case report showed the higher vulnerability of workers with a history of alcohol abuse to CCl₄ vapors compared to similarly exposed moderately drinking coworkers (Manno et al., 1996). Although the use of CCl₄ is restricted and human exposure is limited, it is still a popular model hepatotoxin to study mechanisms of cell injury and fibrosis.

Regeneration

The liver has a high capacity to restore lost tissue and function by regeneration. Loss of hepatocytes due to hepatectomy or cell injury triggers proliferation of all mature liver cells. This process is capable of restoring the original liver mass (Michalopoulos and DeFrances, 1997; Fausto, 2000). However, if hepatocyte replication is blocked, hepatic stem cells or oval cells may proliferate to replace the lost parenchyma (Michalopoulos and DeFrances, 1997; Fausto, 2000). This effect is caused by the reduced sensitivity of oval cells to the proliferation-inhibiting cytokine transforming growth factor-β (TGF-β) (Nguyen et al., 2007). Hepatocytes are normally quiescent, that is, they are in G₀ phase of the cell cycle. In order to proliferate, they need to enter the cell cycle. The process is initiated by cytokines (TNF-α, IL-6), which prime hepatocytes to respond to essential growth factors such as HGF and TGF-α (Fausto, 2000). In contrast to Fausto's priming/progression hypothesis, Michalopoulos (2007) argues that hepatocytes do not fit the two stage model but there are primary direct mitogens for hepatocytes (eg, HGF) and secondary proregenerative substances (TNF-α), which enhance the effect of the primary mitogen (Michalopoulos, 2007). Both cytokines and growth factors are involved in the activation of transcription factors and ultimately expression of cell cycle-regulating proteins, that is, cyclins, the activators of cyclin-dependent kinases (CDKs), and p18, p21, and p27, inhibitors of CDKs (Trautwein, 2006). The coordinated expression of individual cyclins and inhibitors of CDKs guides the cell through the different phases of the cell cycle including DNA synthesis (S phase) and mitosis (M phase). For details on the intracellular signal mechanisms of hepatocyte regeneration, the reader is referred to excellent reviews on this subject (Trautwein, 2006; Michalopoulos, 2007).

In recent years, work from Mehendale and coworkers demonstrated extensively that regeneration is not just a response to cell death but is a process that actively determines the final injury after exposure to hepatotoxic chemicals such as thioacetamide, APAP, chloroform, bromobenzene, trichloroethylene, CCl₄, galactosamine, and allyl alcohol (Mehendale, 2005; Anand et al., 2005). Inhibition of mitosis with colchicine prevented tissue repair and aggravated liver injury after thioacetamide (Mangipudy et al., 1996) and other chemicals (Mehendale, 2005). In contrast, stimulation of repair by exposure to a moderate dose of a hepatotoxicant strongly attenuates tissue damage of a subsequent high dose of the same chemical (autoprotection) or a different hepatotoxin (heteroprotection) (Mehendale, 2005). Tissue repair follows a dose–response up to a threshold where the injury is getting too severe and cell proliferation is inhibited (Mangipudy et al., 1995). In addition to the dose of the hepatotoxicant, other factors such as age, nutritional status, and disease state may influence tissue repair (Mehendale, 2005). Of particular interest is the potential increased susceptibility of diabetic animals to hepatotoxicants. Streptozotocin-induced diabetes reduced liver injury after APAP overdose in rats (Price and Jollow, 1982) and in mice (Shankar et al., 2003). The mechanism of protection included the faster clearance of APAP due to enhanced sulfation and glucuronidation and stimulated tissue repair (Price and Jollow, 1982; Shankar et al., 2003). However, it remains to be evaluated if the reduced susceptibility is caused by the chemical streptozotocin rather than diabetes. In a genetic model of diabetes and obesity, APAP hepatotoxicity is actually enhanced (Kon et al., 2010). This may have been caused by the increased oxidant stress in these steatotic livers (Kon et al., 2010).

Inflammation

The activation of resident macrophages (Kupffer cells), NK and NKT cells, and the migration of activated neutrophils, lymphocytes, and monocytes into regions of damaged liver is a well-recognized feature of the hepatotoxicity produced by many chemicals. The main reason for an inflammatory response is to remove dead and damaged cells. However, under certain circumstances, these inflammatory cells can aggravate the existing injury by release of directly cytotoxic mediators or by formation of pro- and anti-inflammatory mediators (Fig. 13-7).

Kupffer cells and neutrophils are potent phagocytes, which have a vital function in host defense and removal of cell debris. Formation of reactive oxygen species by NADPH oxidase is a critical tool for these cells. Upon activation, Kupffer cells generate mainly hydrogen peroxide, which can diffuse into neighboring liver cells and create an intracellular oxidant stress leading to cellular stress and injury (Bilzer et al., 1999). Kupffer cells can be activated by bacterial products, opsonized particles, and activated complement factors to cause oxidant stress and cell injury (Bilzer et al., 2006). A detrimental role of Kupffer cells in the pathogenesis of toxicant-induced liver injury has been suggested for a number of chemicals including ethanol, APAP, CCl₄, and 1,2-dichlorobenzene (Laskin, 2009) although the detrimental role Kupffer cells in APAP hepatotoxicity has been questioned (Ju et al., 2002). Despite the capacity to directly cause cell damage, a prominent function of Kupffer cells is to generate inflammatory mediators (Decker, 1990). Recent evidence suggests that not only bacterial products but also intracellular proteins, for example, HMGB-1, which are released during necrotic cell death, can bind to toll-like receptors on Kupffer cells and trigger cytokine and chemokine formation (Schwabe et al., 2006). These mediators may aggravate injury by recruiting cytotoxic neutrophils into the liver (Bajt et al., 2001), directly cause apoptotic cell death in susceptible hepatocytes (Nagai et al., 2002), or promote cytotoxic mechanisms such as induction of iNOS during APAP hepatotoxicity (Bourdi et al., 2002). However, Kupffer cells can also generate anti-inflammatory mediators such as prostaglandin E₂ and interleukin-10 (Decker, 1990), which downregulate formation of proinflammatory cytokines and attenuate toxin-induced liver injury (Bourdi et al., 2002; Ju et al., 2002). Thus, Kupffer cells can promote or inhibit an injury process and assist in removal of cell debris and apoptotic bodies. In addition, newly recruited mononuclear cells (macrophages) can function in a similar way as Kupffer cells in liver. Although there is a capacity for additional damage by these cells, in general, the recruitment of macrophages into the damaged liver and even the formation of proinflammatory mediators are important signals for inducing regeneration and repair of the damaged tissue (Holt et al., 2008; Laskin, 2009; Adams et al., 2010).

Neutrophils are activated and accumulate in the liver vasculature in response to extensive cell injury or bacterial infection (Fig. 13-7). The main purpose of hepatic neutrophil recruitment is to remove bacteria and cell debris, at least in part through interactions with the resident macrophages (Gregory and Wing, 2002). Neutrophils generate the aggressive oxidant and chlorinating species hypochlorous acid through NADPH oxidase and myeloperoxidase (El-Benna et al., 2005). In addition, neutrophils can release a large number of proteolytic enzymes and bacteriocidal proteins (Wiedow and Meyer-Hoffert, 2005). The capability of neutrophils to migrate out of the vasculature, adhere to and generate potent cytotoxins in close proximity to its target makes this leukocyte an effective killer of invading microorganisms and a remover of dead or dying cells. However, if the cytotoxicity is directed against still viable liver cells, this can cause additional tissue injury or even liver failure (Jaeschke, 2006b). Recent insight into the pathomechanisms revealed that neutrophil-induced liver cell injury is a multistep process (Jaeschke and Hasegawa, 2006; Ramaiah and Jaeschke, 2007) (Fig. 13-7). It requires exposure to inflammatory mediators, which upregulate adhesion molecules such as Mac-1 (CD11b/CD18) on the surface, prime the neutrophils for reactive oxygen formation, and cause their accumulation in vascular beds of the liver. If a chemotactic signal is received from the parenchyma, neutrophils will extravasate and adhere to the target. In contrast to other vascular beds, in the liver this process can take place in both sinusoids (capillaries) and venules (portal and/or postsinusoidal venules). However, extravasation from sinusoids is most critical for parenchymal cell injury (Chosay et al., 1997). At this time, the neutrophil becomes fully activated, that is, initiates a prolonged adherence-dependent oxidant stress and releases proteolytic enzymes (Jaeschke and Hasegawa, 2006). Cell killing is predominantly caused by hypochlorous acid diffusing into the target cell and causing an intracellular oxidant stress (Jaeschke, 2006b). Although proteases can also be directly involved in the injury process, the main function of neutrophil-derived proteases appears to be the promotion of the inflammatory process by generation of inflammatory mediators and facilitation of neutrophil migration (Jaeschke and Hasegawa, 2006). It has previously been assumed that the killing of “innocent bystanders” mainly caused the aggravation of liver injury by neutrophils during the attack on dying hepatocytes. More recent findings suggest that neutrophils only attack distressed or damaged, but not healthy cells (Gujral et al., 2004). Thus, the aggravation of liver injury by neutrophils is mainly caused by the killing of distressed cells, which would actually survive the original insult (Jaeschke, 2006b). Neutrophils have been shown to be involved in the injury process during hepatic ischemia-reperfusion, alcoholic hepatitis, alpha-naphthylisothiocyanate hepatotoxicity, obstructive cholestasis, and halothane-induced liver injury (Ramaiah and Jaeschke, 2007). The involvement of neutrophils in APAP hepatotoxicity is controversial but most of the experimental evidence in animals and humans support the hypothesis that neutrophils do not actively contribute to the injury process (Jaeschke et al., 2012b). Although many chemicals cause liver injury without neutrophil participation or do not cause injury at moderate doses, initiation of an inflammatory response with endotoxin triggers a neutrophil-induced injury or aggravates the existing injury after ethanol, allyl alcohol, aflatoxin B₁, monocrotaline, ranitidine, diclofenac, and trovafloxacin (Ganey and Roth, 2001; Shaw et al., 2010). Thus, a detrimental effect of neutrophils only occurs when activated neutrophils are recruited to the site of injury and if a relevant number of distressed cells, which are killed by neutrophils, would survive without the neutrophil attack (Jaeschke, 2006b).

Identification of the NALP3 inflammasome and its role in inflammation and autoimmunity (Agostini et al., 2004) has again led to resurgence in studies of inflammation as a critical factor in drug hepatotoxicity (Jaeschke et al., 2012b). The inflammasome forms in response to NALP1/2/3, forming an oligomer with the linker protein ASC and caspase 1 to cleave pro-IL-1β or pro-IL-18, thereby forming the active cytokine (Agostini et al., 2004). Activation of the NALP3 inflammasome has been associated with APAP-induced liver injury (Imaeda et al., 2009). However, detailed studies into NALP3 inflammasome-dependent caspase-1 activation and the role of IL-1β in the pathophysiology confirmed caspase-1 activation and formation of active IL-1β, but could not demonstrate a relevant impact of these inflammatory mediators in the injury mechanism (Williams et al., 2010; 2011). A risk of using interventions that target the immune response in drug hepatotoxicity are off-target effects modulating drug metabolism and disposition, which can impact the injury process independent of the assumed immune mechanism (Jaeschke et al., 2012b).

Immune Responses

In addition to the activation of an inflammatory response, immune-mediated reactions may also lead to severe liver injury (Ju, 2005; Adams et al., 2010). Drugs and chemicals that have been suggested to cause immune-mediated injury mechanisms in the liver include halothane, tienilic acid, and dihydralazine (Ju, 2005; Uetrecht, 2007). A delay in onset of the injury or the requirement for repeated exposure to the drug and the formation of antibodies against drug-modified hepatic proteins are characteristic features of immune reactions (Ju, 2005; Zhang et al., 2011). However, the mechanisms of these immune-mediated liver injuries are not well understood. The hapten hypothesis assumes that a reactive metabolite covalently binds to cellular proteins and the drug-modified protein is taken up by APCs, cleaved to peptide fragments, which are then presented within the major histocompatibility complex (MHC) to T cells (Ju, 2005; Uetrecht, 2007). In support of the hapten hypothesis, antibodies against drug-modified proteins were detected in the serum of patients with halothane hepatitis (Vergani et al., 1980; Satoh et al., 1989) or with liver injury caused by ethanol, tienilic acid, and dihydralazine (Bourdi et al., 1994; Lecoeur et al., 1996; Tuma, 2002). However, the hapten hypothesis does not explain why other drugs (eg, APAP), which also form reactive metabolites and drug-modified proteins, do not trigger an immune response. This suggests that additional activating factors may be necessary to induce immune-mediated liver injury. The danger hypothesis (Fig. 13-8) postulates that damaged cells release danger signals, which induce the upregulation of B7 on APCs and, the interaction of B7 with CD28 on T cells generates a costimulatory signal (Uetrecht, 2007). A cytotoxic immune response occurs only when the T-cell receptor stimulation with the antigen is accompanied by an independent costimulation of the T cell. In the absence of this costimulatory signal, the antigens derived from drug-modified proteins induce immune tolerance (Ju, 2005; Uetrecht, 2007). Both liver sinusoidal endothelial cells and Kupffer cells can function as APCs in the liver and can be inducers of tolerance to hapten-induced immunological responses (Knolle et al., 1999; Ju et al., 2003). More recently another mechanism of tolerance has been proposed. Hepatotoxic doses of APAP caused a loss of lymphocytes from spleen, thymus, and draining hepatic lymph node and immunesuppression (Masson et al., 2007). These mechanisms could be the reason that tolerance appears to be the default reaction to drug-induced protein modifications in most people. However, impairment of these mechanisms in a limited number of patients can make them susceptible to the immune-mediated liver disease (Ju, 2005; Uetrecht, 2007; Zhang et al., 2011).

Idiosyncratic Liver Injury

Idiosyncratic drug hepatoxicity is a rare but potentially serious adverse event, which is not clearly dose-dependent, is at this point unpredictable, and affects only very few of the patients exposed to a drug or other chemicals. However, idiosyncratic toxicity is a leading cause for failure of drugs in clinical testing and it is the most frequent reason for posting warnings, restricting use, or even withdrawal of the drug from the market (Li, 2002; Kaplowitz, 2005) (Table 13-4). In addition, idiosyncratic hepatotoxicity is observed after consumption of herbal remedies and food supplements (Stickel et al., 2011). There are no known mechanisms of cell injury specific for idiosyncratic hepatotoxins. A number of drugs including halothane (anesthetic), nitrofurantoin (antibiotic), and phenytoin (anticonvulsant) are thought to cause injury mainly by immune (allergic) mechanisms as described in the previous paragraph (Kaplowitz, 2005; Uetrecht, 2007). Other drugs, for example, isoniazid (antituberculosis), disulfiram (alcoholism), valproic acid (anticonvulsant), or troglitazone (antidiabetic), are considered nonimmune (nonallergic) idiosyncratic hepatotoxins (Kaplowitz, 2005); however, some immune mechanisms might be involved (Metushi et al., 2011). Diclofenac (analgesic) can elicit allergic and nonallergic mechanisms of toxicity (Boelsterli, 2003). Because only very few patients (1 in 10,000 or less) treated with these drugs actually experience significant hepatotoxicity, the prevailing opinion at the present time is that an enhanced individual susceptibility with the failure to adapt to a mild adverse drug reaction is a key factor in the pathogenesis (Watkins, 2005). Thus, only small subsets of patients who show elevated plasma ALT levels during drug treatment actually develop severe tissue damage or liver failure (Watkins, 2005). In support of this concept, it was reported that 10% to 20% of patients treated with isoniazid show increased levels of plasma ALT levels as an indicator of hepatocellular injury (Mitchell et al., 1975). However, only a small subgroup of these patients develops severe hepatotoxicity (Mitchell et al., 1975). This raises the possibility that one or several gene defects, which prevent effective adaption to drug-induced cellular stress, may be involved in idiosyncratic reactions (Watkins, 2005). Recent findings appear to support this hypothesis. The antidiabetic drug troglitazone (Rezulin^®) was withdrawn from the market due to idiosyncratic hepatotoxicity. In preclinical studies, troglitazone did not cause any relevant liver toxicity and despite extensive investigations since withdrawal of the drug, the mechanism of toxicity remains unclear (Chojkier, 2005). Several studies suggest that very high concentrations of troglitazone can induce mitochondrial dysfunction in vitro (Haskins et al., 2001; Tirmenstein et al., 2002). Because the conditions applied in these studies are not relevant for human exposure, the proposed mechanism cannot explain the idiosyncratic toxicity in humans. However, the data suggest that troglitazone can cause a subclinical mitochondrial stress, which could sensitize hepatocytes to troglitazone. In fact, mice partially deficient in mitochondrial manganese superoxide dismutase (Mn-SOD; SOD2) showed mitochondrial dysfunction and mild liver injury after treatment with 30 mg/kg troglitazone for 28 days (Ong et al., 2007). Although the animals did not develop severe hepatic injury as observed in humans, the injury in mice occurred also after a lag time, which is consistent with the hypothesis that a certain threshold of mitochondrial stress has to be reached to cause cell injury (Ong et al., 2007). Overall, these findings support the concept that a clinically silent genetic deficiency in individuals can trigger the hepatotoxicity of a drug, which by itself may only cause a mild and clinically silent cellular stress. This recent insight indicates the need for a paradigm shift for preclinical toxicity studies (Jaeschke, 2007). The assumption in traditional toxicity studies is that an adverse effect of a drug can be detected by progressively increasing the dose. The experience with troglitazone suggests that this is not always the case. It may be necessary to include experiments with genetically deficient animals in these studies if there is any evidence for clinically silent adverse effects of these drugs. In addition to the genetic makeup, which may render individuals more susceptible to stress induced by the metabolism of drugs or chemicals, a second “hit” such as a systemic inflammatory response can also contribute to the unmasking of the toxicity at least in experimental models (Ganey and Roth, 2001; Shaw et al., 2010). A major argument against the systemic inflammatory hypothesis is the fact that neutrophils are prominent players in animal models (Shaw et al., 2010) but not in humans (Zhang et al., 2011). Because idiosyncratic hepatotoxicity is a rare event for most drugs, it is likely that a combination of gene defects and adverse events need to be present simultaneously in an individual to trigger the severe liver injury. A detailed genomic analysis of patients with idiosyncratic responses to drug exposure may give additional insight what gene expression profile renders a patient susceptible (Watkins, 2005).

Continued progress in the understanding of drug- and chemical hepatotoxicity will depend on the use of relevant in vivo and in vitro models including human hepatocytes and analysis of human liver tissue. Traditional mechanistic investigations in combination with genomic and proteomic approaches have the greatest potential to yield important new insight into pathomechanisms. Progress in the understanding of the liver's response to known hepatotoxins and other adverse conditions will not only aid in the development of therapies to limit and reverse acute and chronic liver injury, but also improve the predictability of the potential hepatotoxicity of new drugs and other chemicals (Guengerich and MacDonald, 2007; O'Connell and Watkins, 2010).