Abstract
Biochemical assays are not routinely used to assess liver damage in fish, therefore, a histopathological evaluation is usually required to determine the existence or extent of nonneoplastic liver toxicity. Many mammalian pathologists may be uncomfortable when requested to identify and interpret subtle liver changes in these unfamiliar animals. It may be reassuring to note that there are more similarities than differences between fish and mammals in terms of their macro- and microanatomy, physiological and biochemical characteristics, and pathologic responses to hepatotoxic substances. This brief overview addresses several topics pertaining to hepatotoxicity in fish, including: anatomic considerations, that is, how the anatomy of the fish liver may be predictive of its metabolic capacity, and also its microscopic appearance, following exposure to toxins; physiologic considerations, including comparisons between mammalian and fish livers regarding the uptake, elimination, toxification, or detoxification of xenobiotic compounds; morphologic responses to toxicity, in which some of the general types of findings that are most commonly observed in cases or studies of fish hepatotoxicity are highlighted; and last, responses of the fish liver to specific hepatotoxins.
Introduction
Much media attention has been given to xenobiotic-induced liver neoplasia in fish collected from natural waters, due to the logical role of fish as environmental sentinels for aquatic toxicants, and the generally heightened public concern that surrounds the subject of cancer. There is solid evidence, however, that fish should also be acknowledged as worthy models for assessing nonneoplastic hepatotoxicity, in terms of both field and laboratory-based research. Because biochemical assays are not routinely used to assess liver damage in fish (in contrast to mammals), a histopathological evaluation is usually required to determine the existence or extent of nonneoplastic liver toxicity. Many mammalian pathologists may be uncomfortable when requested to identify and interpret subtle liver changes in these unfamiliar animals. It may be reassuring to note that there are more similarities than differences between fish and mammals in terms of their macro- and microanatomy, physiological and biochemical characteristics, and pathologic responses to hepatotoxic substances. As for mammals, the recognition or interpretation of liver changes in fish can be complicated by the presence of incidental or artifactual lesions, confounding factors such as concurrent infections and nutritional disease, and the range of structural and functional diversity that exists among different species.
This brief overview is divided into several topics, including: anatomic considerations, that is, how the anatomy of the fish liver may be predictive of its metabolic capacity, and also its microscopic appearance, following exposure to toxins; physiologic considerations, including comparisons between mammalian and fish livers regarding the uptake, elimination, toxification, or detoxification of xenobiotic compounds; morphologic responses to toxicity, in which some of the general types of findings that are most commonly observed in cases or studies of fish hepatotoxicity are highlighted; and lastly, responses of the fish liver to specific hepatotoxins. More detailed information concerning hepatotoxicity in fish can be found in several excellent reviews (Gingerich, 1982; Boorman et al., 1997; Metcalfe, 1998; Hinton et al., 2001).
Anatomic Considerations
In both Osteichthyes (bony fishes) and Mammalia, the liver is a very large, discrete, encapsulated, sinusoidally perfused gland that is relatively homogeneous at the subgross level. The fish liver features the same general circulatory components as the mammalian liver; that is, blood is supplied by hepatic arterioles and portal veins, and is drained by hepatic veins. The biliary apparatus of fish is also comparable to that of mammals. Among various fishes, however, there is ample architectural divergence, as might be expected for a group of animals that contains over 23,000 species. Most species of fish have a single-lobed liver, the vasculature of which is usually divided into 2 large circulatory regions (Gingerich, 1982). A few species, such as the fathead minnow Pimephales promelas, have multiple liver lobes, in which islands of liver tissue are scattered througout the abdominal mesentery (Gingerich, 1982; Ferguson, 1989). Although subtle differences in sinusoidal structure exist, the most variable aspects of the fish liver involve the biliary system, in which there are considerable interspecies differences in the length and position of ducts (Gingerich, 1982). As examples, in fishes such as salmon, the common bile duct and cystic duct are actually intra- rather than extrahepatic, and intracellular canaliculi have been observed in some cyprinids (Gingerich, 1982; Ferguson, 1989). Similar to mammals, most fishes have a gallbladder (Hinton et al., 2001), but in certain species such as cod the gallbladder is absent (Gingerich, 1982).
Based on microscopic descriptions of histologic slides, it was once thought that fish hepatocytes were arranged in double-layered cords, which in a 3-dimensional perspective, formed hepatic plates, similar to those of mammalian livers. From more detailed morphologic studies, it is now understood that these double-layered cords actually represent a system of blind-ended, anastomosing, and branching tubules (Metcalfe, 1998; Hinton et al., 2001). The tubule lumen, which is lined by biliary epithelial cells and the apical surfaces of hepatocytes, forms the bile ductules (Metcalfe, 1998). The canaliculi that drain into these ductules are located between the lateral surfaces of hepatocytes, whereas the basal aspects of hepatocytes face the sinusoids (Hinton et al., 2001). The space of Disse, and the perforated endothelial membrane that partially defines it, have essentially the same conformation as in mammals (Gingerich, 1982). Some of the hepatocytes that are located at the branch points of tubules do not appear to have direct access to sinusoidal fluid (Hinton et al., 2001). The functional significance of this design is that only the basal and basolateral aspects of hepatocytes are directly exposed to sinusoidal perfusion (Hinton et al., 2001). Consequently, the relatively lower uptake of some chemicals by fish hepatocytes might be attributed to this architectural detail; then again, the relatively slower blood flow through fish hepatic sinusoids may at least partially compensate for the reduced hepatocyte exposure (Gingerich, 1982).
Compared to mammals, fish hepatocytes tend to be more vacuolated, corresponding to a relatively higher glycogen and/or lipid content (Gingerich, 1982; Ferguson, 1989). Such vacuolization, which tends to be uniformly distributed, is often especially apparent in the livers of captive fishes; this is presumably due to imbalances in energy intake and expenditure caused by artificial feeding and housing conditions. Figure 1 illustrates an example of nutritionally induced lipid accumulation in fish hepatocytes. Different fish species may store radically different proportions of glycogen and fat in their livers, as demonstrated in Figure 2. For example, the rainbow trout Oncorhynchus mykiss liver stores primarily glycogen (Hinton et al., 2001), whereas striped bass hepatocytes tend to contain a mixture of glycogen and fat. Sharks (Class: Chondrichthyes) often accumulate massive amounts of hepatic fat. Because sharks do not have swim bladders, it has been proposed that this high level of hepatic lipid allows them to maintain neutral buoyancy in the water column, although this theory has been challenged (Rossouw, 1987).
Constitutively present in the livers of some species of fish are small nests of pigment-containing, resident macrophages that were previously called melanomacrophage centers, but are now more commonly referred to as macrophage aggregates. Macrophage aggregates serve as repositories for products of cell membrane and erythrocyte breakdown (Fulop and McMillan, 1984; Herraez and Zapata, 1986), and appear to have an antigen-presenting function analogous to the germinal centers of mammalian lymphoid tissues (Ellis, 1980). Other purported functions of macrophage aggregates include iron storage and recycling (Agius, 1979), and the sequestration and detoxification of endogenous and exogenous substances (Mori, 1980; Agius and Roberts, 1981; Herraez and Zapata, 1986). Proliferation of macrophage aggregates has been associated with starvation and tissue catabolism (Agius, 1979; Herraez and Zapata, 1986; Agius and Roberts, 1981), nutritional imbalances (Moccia et al., 1984), infectious diseases and parasite infestations (Agius, 1979; Roberts, 1978; Vogelbein et al., 1987), toxicant-induced hemolytic anemias (Herraez and Zapata, 1986; Dawson, 1935), heat stress (Blazer et al., 1987), and sediment contamination (Benyi et al., 1989). The pigments that are present in macrophage aggregates can include lipofuscin, ceroid, hemosiderin, and melanin (Wolke, 1992).
The livers of certain fish species incorporate variable amounts of exocrine pancreatic tissue. In such livers, the pancreatic tissue is aligned along blood vessels in interstitial areas. In Japanese flounder Paralichthys olivaceus, the pancreas was observed to migrate from a point adjacent to the esophageal-intestinal junction into the liver following metamorphosis at 45 days posthatch (Kurokawa and Suzuki, 1996). The operative advantage of this arrangement, if any, has not been explored. On the negative side, inflammatory and noninflammatory diseases of the pancreas can readily extend into the adjacent hepatic parenchyma.
Microanatomically, there are several important features of mammalian livers that fish livers lack. Perhaps most significantly, the classic, hexagonal, hepatic lobule, with distinct portal triad regions, is not readily discernable in most fish livers (Gingerich, 1982; Ferguson, 1989; Metcalfe, 1998; Hinton et al., 2001). This apparent deficiency of organization has been attributed to the almost random pattern of vascular branching that occurs during fish liver development (Gingerich, 1982). The absence of features such as a lobular pattern and obvious portal triads has very real implications for the pathologist, because a valuable tool for the morphological categorization of liver disease is not applicable to fish. Profiles of arterioles and bile ductules are relatively scarce in fish liver sections (Gingerich, 1982), despite the recognition that fish have substantially more biliary epithelial cells compared to mammals (Hinton et al., 2001), and the fish liver lacks a lymphatic drainage system (Sailendri and Muthukkaruppan, 1975). With the exception of some catfishes (Hinton and Pool, 1976) and a few other fishes, Kupffer cells are not evident in most fish livers (Gingerich, 1982; Ferguson, 1989; Hinton et al., 2001). On the other hand, fish often have more nonresident perisinusoidal macrophages compared to mammals (Hinton et al., 2001). Additionally present in the perisinusoidal space of the fish liver are the hepatic stellate cells (Ito cells, perisinusoidal cells, fat-storing cells, lipocytes) (Eastman and DeVries, 1981; Tanuma and Ohata, 1982; Yamamoto et al., 1986; Rocha et al., 1997; Hinton et al., 2001).
Physiologic Considerations
Physiologically, the livers of fish are responsible for the same basic metabolic functions as in mammals, including processing, and storage of nutrients, the synthesis of enzymes and other cofactors, bile formation and excretion, and the metabolism of xenobiotic compounds. Fish have analogous mechanisms for handling xenobiotic compounds, including both phase 1 and phase 2 biotransformation reactions, and many of the same microsomal and cytosolic enzymes as mammals (Cowey and Walton, 1989). The most studied microsomal enzyme in the fish liver is cytochrome p4501A, or CYP1A, which can be found in hepatocytes, biliary epithelial cells, and endothelial cells (Stegeman et al., 1979; Stegeman and Hahn, 1994; Hinton et al., 2001). The activity of this enzyme has been demonstrated to vary according to the type of inducer, the exposure route, and the fish species (Hinton et al., 2001). As in mammals, the goal of biotransformation is to produce metabolites that are more hydrophilic, and thus more easily excreted, than the parent compound. Fish have a relatively lower capacity than mammals to metabolize xenobiotic substances, and It has been speculated that this is may be due to the fact that fish can eliminate such substances unchanged via their gills (Parkinson, 1996). Like mammals, fish have an enterohepatic cycling mechanism (Gingerich, 1982) for the processing of substances that were not metabolized during their first pass through the liver. As in mammals, enterohepatic cycling may prolong the removal of certain compounds (Gingerich, 1982; deBethizy and Hayes, 2001).
There are a number of essential physiologic differences between mammalian and fish livers, some of which may affect the rate, pattern, and/or extent of toxicity that occurs in a fish following chemical exposure. One of the most important differences is that fish appear to have a relatively homogenous distribution of biotransforming enzymes in their livers (Gingerich, 1982; Schar et al., 1985; Metcalfe, 1998; Hinton et al., 2001) that is, there is generally no predilection for these enzymes to be located preferentially near structures such as central veins, for example. The significance of this dissimilarity will be addressed shortly. Another difference is that fish monooxygenases like CYP1A may be refractory to some classic cytochrome inducers such as phenobarbital (Ferguson, 1989), or variably responsive to other inducers such as 3-methylcholanthrene (Di Gulio et al., 1995). Depending on the type of toxicant, resistance to CYP1A induction can lead to greatly enhanced or diminished toxicity. Other differences between mammals and fish are probably less consequential. Compared to mammals, fish have a relatively lower liver to body weight ratio, they have 1/4 to 1/2 less liver perfusion, and bile formation in fish is 50-fold slower (Klassen and Plaa, 1967; Gingerich, 1984). Compared to mammals, fish have more of a tendency to preserve hepatic glycogen during periods of starvation, and there is no known mechanism for ketogenesis in fish (Segner, 1997; Hinton et al., 2001). Finally, cholestasis has not been extensively studied in fish (Hinton et al., 2001), possibly because histopathologic evidence of bile stasis is only rarely observed in fish livers (Ferguson, 1989).
Morphologic Responses To Toxicity
If we look at how the above anatomic and physiologic considerations affect the fish liver from a morphologic viewpoint, we observe that the response of the fish liver to toxic exposure tends to be less severe than in mammals (Gingerich, 1984). An alternative way to state this observation is that greater concentrations of toxicant are required to cause comparable hepatic changes in fish. The relative tolerance of fish to hepatotoxins could be attributed to some of the above-mentioned factors, such as the lower perfusion rate of the fish liver, the limitation of toxic exposure to the basal hepatocyte membrane, the homogenous distribution of biotransforming enzymes, and the fact that some enzymes are not readily inducible. Another important propensity of the fish liver is that the lesion distribution following toxic exposure tends to be random, without the zonal pattern that is common to sublethal intoxications in mammals (Ferguson, 1989; Boorman et al., 1997; Hinton et al., 2001). Again, this trait has been attributed to the homogenous distribution of biotransforming enzymes in the fish liver, and the fact that the zonal architecture is not apparent.
A common morphologic response of the fish liver to toxicity is a loss of hepatic glycogen and/or lipid (Ferguson, 1989). Macroscopically, affected fish have small, dark livers; microscopically, hepatocellular vacuolization is decreased, and hepatocytes may be larger or smaller than expected, depending on the etiology. Loss of glycogen or lipid can occur as a direct effect of intoxication, or it may occur secondary to decreased body condition caused by inanition, stress, or concurrent disease. The latter is probably more typical. Paradoxically, toxic exposure can also result in accumulations of fat or glycogen in the liver. This type of dichotomy (i.e., glycogen accumulation versus depletion as a result of toxicity) was evident in experiments in which rainbow trout (Seinen et al., 1981), Japanese medaka Oryzias latipes (Wester et al., 1988), guppies Poecilia reticulata (Wester and Canton, 1987), and rats (Krajnc et al., 1984) were exposed to organotins such as bis(tri-n-butyltin)oxide (TBTO), di-n-butyltindichloride (DBTC), and tri-n-butyltin chloride (TBTC). When exposed to TBTC and TBTO, respectively, rainbow trout fry and rats both experienced a loss of hepatic glycogen that was associated with decreased body weight and inanition. In contrast, increases in hepatic glycogen were demonstrated histologically, histochemically, and biochemically in medaka and guppies that were exposed to TBTO or DBTC. In the latter 2 species, it was hypothesized that glycogen accumulation was due to decreased glycogen breakdown as a result of hepato-cellular toxicity.
In mammals, fatty liver (hepatic lipidosis, hepatic steatosis, lipoid liver disease, fatty degeneration of the liver) “. . . is the term used to describe livers that contain more visible lipid in hepatocytes than one expects to see in that organ” (Kelly, 1993). As previously discussed, many types of otherwise healthy wild and captive fishes have the capacity to store large amounts of lipid (triglyceride) in their livers. Thus, it can be difficult for the pathologist to determine the extent to which lipid-type vacuolization should be considered excessive and potentially deleterious. Unfortunately, there are no universally applicable criteria for the diagnosis of hepatic lipidosis. McLelland et al., (1995) suggest that liver lipid concentrations greater than 10–12% for gilt-head sea bream Sparus (Chrysophyrys) auratus and greater than 45–50% for European sea bass Dicentrarchus labrax are probably pathological; however, measured concentrations are not readily translated into morphologic guidelines for the histopathologist. Although hepatic lipidosis could be defined arbitrarily as the point at which hepatocellular membranes appear to visibly disintegrate, resulting in the fusion of the fat globules from neighboring hepatocytes, even this extreme manifestation may not be detrimental (or irreversible) in all fish.
Similar to glycogen overload, hepatic lipidosis may be nutritionally induced, as described in a captive African stone-fish Synanceja verrucosa (Penrith et al., 1994). In such cases, lipid accumulation may be the result of overfeeding an excessively energy-rich diet. Alternatively, it may be associated with lipid peroxidation, caused by the feeding of diets high in polyunsaturated fats and/or by the suppression of vitamin E (Ferguson, 1989). Spisni et al., (1998) suggested that captive marine teleosts may be particularly predisposed to hepatic lipidosis, due a reduced capacity for hepatocyte peroxisome proliferation that is a distinctive characteristic of marine (versus freshwater) fishes, coupled with the feeding of artificial diets that often contain high proportions of C18:1 and other mono-unsaturated fatty acids. Lipid peroxidation in fish may also be toxicant-induced. For example, lipid peroxidation occurred in channel catfish that were exposed to sediment contaminated with polyaromatic hydrocarbons, polychlorinated biphenyls (PCBs), and metals (Di Gulio et al., 1993) and in brown bullhead and channel catfish exposed to tert-butyl hydroperoxide (Ploch et al., 1999). There are a variety of other conditions in which hepatic lipid accumulation has been associated with exposure to toxicants, and examples include channel catfish Ictalurus punctatus fed Fusarium moniliforme toxins (Lumlertdacha et al., 1995), Rivulus marmoratus exposed to diethylnitrosamine (DEN) (Thiyagarajah and Grizzle, 1985), and juvenile grey mullet Liza ramada exposed to atrazine (Biagianti-Risbourg and Bastide, 1995).
Lipid or glycogen vacuolization can cause an increase in the size of hepatocytes; however, Hinton et al., (1992) identified 3 additional potential causes of hepatocellular enlargement: organelle proliferation (hypertrophy); the failure of sublethally-injured hepatocytes to mitotically divide (megalocytosis); and vacuolar swelling of the endoplasmic reticulum cisternae (hydropic degeneration). Hepatocyte hypertrophy was observed in rainbow trout that were simultaneously exposed to endosulfan and disulfoton (Arnold et al., 1995). Ultrastructural changes in this study included increased numbers of organelles such as myelinated bodies, mitochondria, glycogenosomes, peroxisomes, and lysosomes, and changes in rough endoplasmic reticulum (increased amount, vesiculation, and dilatation of the RER). Physiologic hypertrophy of hepatocytes can be seen in reproductively active female fish, or in male fish that have been exposed to exogenous estrogenic compounds, due to the ability of estrogens to boost hepatocyte metabolism for vitellogenin (yolk lipoprotein) production (Wester et al., 2003). Whether due to physiologic or toxicologic causes, hepatocyte hypertrophy is often accompanied by basophilia (Figure 3), due to a loss in glycogenic vacuolization and increased mRNA content (Wester et al., 2003). Megalocytosis, in which there is both cytoplasmic and nuclear enlargement, has been observed in cases of algal toxicity in farmed Atlantic salmon (Figure 4) (Kent et al., 1988; Andersen et al., 1993; Stephen et al., 1993). Other types of nuclear changes, such as enlarged, binucleate, or bizarre nuclei, are seen occasionally, most often in the repair stages of toxicosis (Ferguson, 1989).
Another finding that can be exacerbated by toxins is an increase in the size and/or number of macrophage aggregates in the liver and other tissues. Often, macrophage aggregate proliferation is more obvious in organs other than the liver, such as the kidney, spleen, and gonads. As previously discussed, macrophage aggregates may also become more prominent secondary to nontoxic causes, such as aging or infectious disease, so these etiologies must be considered where appropriate. Diagnostically, macrophage aggregates are morphologically distinct from granulomatous inflammation (Figure 5), and because these 2 findings may have different causes, they should be diagnosed separately whenever possible. The distinction between macrophage aggregates and inflammation may become hazy in livers with moderate to severe inflammation, in which inflammatory foci (granulomas) may arise in, or merge with, macrophage aggregates (Boorman et al., 1997).
A finding that is commonly seen in both control and toxin-exposed fish, such as the medaka in Figure 6, is cystic degeneration of the liver. This condition has been termed either “spongiosis hepatis” or “hepatic cysts,” depending on the size of the cysts, their contents, and whether they are multilocular or unilocular (Boorman et al., 1997). Despite published criteria, there appears to be significant overlap between spongiosis hepatis and hepatic cysts as viewed in tissue sections, which suggests that they are probably two morphologic subtypes of the same degenerative process; therefore, it may be more appropriate to combine these terms into the single diagnosis of cystic degeneration. A potential relationship between piscine cystic degeneration and hepatic stellate cells has been suggested but not fully established (Couch, 1991). In a study in which medaka were exposed to DEN, cystic degeneration was exacerbated among fish of the high-dose group (Boorman et al., 1997). In sheepshead minnows Cyprinodon variegatus, cystic degeneration was associated with toxic exposure and a possible progression to spindle cell neoplasms (Couch, 1991). The fact that neoplasms often occur concomitant with cystic degeneration does not necessarily indicate causality, and the theory that cystic degeneration represents a preneo-plastic lesion in rats or fish has been challenged (Karbe and Kerlin, 2002).
As in mammals, foci of cellular alteration can be found in the livers of otherwise normal fish, but it has also been shown that these proliferative lesions may be induced by exposure to a number of carcinogenic or estrogenic compounds (Boorman et al., 1997; Baumann and Okihiro, 2000). The most commonly used nomenclature for altered foci in fish mimics the system used for rodents, that is, eosinophilic, basophilic, clear cell, and vacuolated varieties can be discerned (Figure 7), and similar diagnostic criteria are employed (Boorman et al., 1997). It has been postulated, and to some degree demonstrated, that basophilic foci may be precursors of primary hepatocellular neoplasms (Hendricks et al., 1984, 1995; Hinton et al., 1988). In medaka, at least, it may be difficult to differentiate eosinophilic foci from a little-studied lesion that has been termed “hepatocyte hyalinization” (Boorman et al., 1997). Hepatocyte hyalinization is characterized by the focal to diffuse presence of variably enlarged hepatocytes that contain discrete or pancytoplasmic inclusions of refractile, eosinophilic material (Figure 8). Hyalinized hepatocytes can be found within some primary hepatic neoplasms, and Boorman et al. (1997) speculated that this condition represents a degenerative change.
Most of the previously discussed responses to hepatotoxins could also be observed in unexposed control or wild fish as incidental findings, especially in older animals. On the other hand, a common, clearly pathologic, response of the fish liver to toxins is hepatocyte necrosis. The most characteristic reaction to toxicity is an apoptotic type of single cell death (Ferguson, 1989; Boorman et al., 1997), however, coagulative, liquefactive, or massive types of necrosis can also be seen in more severe intoxications. Coagulative necrosis often implies regional vascular impairment (Metcalfe, 1998), and a random to diffuse pattern of coagulative necrosis was observed in medaka exposed to DEN, for example (Boorman et al., 1997). As previously mentioned, a random pattern of necrosis would be more typical than a zonal pattern for the fish liver, although a perivenous distribution of necrosis is occasionally seen (Casillas, et al., 1983). Necrosis of the biliary epithelium can also occur, as was induced in rainbow trout exposed to the bile duct toxin alpha-naphthylisothiocyanate (Metcalfe, 1998). Unlike mammals, in which significant damage to the liver parenchyma often results in hepatic fibrosis during the reparative phase (Kelly, 1993), cirrhosis is rarely a sequel to hepatocellular necrosis in the fish liver (Ferguson, 1989).
Although hepatic parenchymal fibrosis is rare, it is not unusual to observe a scirrhous reaction in the fish liver that is centered around bile ducts. As in mammals, cholangiofibrosis in fish can have inflammatory or toxic causes, or it can be idiosyncratic. Piscine cholangiofibrosis may be accompanied by bile duct dilatation, increased numbers of bile duct profiles (bile duct hyperplasia), increased numbers of bile ductule and/or oval cells, and nonneoplastic proliferation of the biliary epithelium (Figure 9). As an example, bile duct hyperplasia appeared to be a regenerative response following DEN exposure in medaka (Boorman et al., 1997). Based on a study of lampreys Petromyzon marinus L. (Class: Spidomorphi) undergoing biliary atresia during metamorphosis, Yamamoto et al., (1986) suggested that lipocytes (hepatic stellate cells) may be the cell type responsible for periductal fibrosis (in at least some cases). Differentiating cholangiofibrosis from cholangioma can occasionally be challenging; ostensibly, cholangiomas are characterized by compression of, or expansion into, the hepatic parenchyma (Boorman et al., 1997).
Responses of the Fish Liver to Specific Hepatotoxins
Numerous hepatotoxic substances representing various classes of organic compounds have been studied in fish, and a comprehensive list of these has been published (Metcalfe, 1998). For comparative purposes, it is interesting to look briefly at how fish respond to some classic hepatotoxins such as carbon tetrachloride, acetaminophen, and allyl formate.
Considered the “archetype of toxic metabolism” (Kelly, 1993), the primary mechanism of carbon tetrachloride (CCl4) intoxication involves necrogenic membranous lipoperoxidation following a microsomal transformation that produces a highly reactive trichloromethyl radical (Parkinson, 1996). CCl4 is generally less toxic to fish than mammals (Hinton et al., 2001). One explanation for this disparity is that the higher oxygen tension of the fish liver allows glutathione to function more efficiently as an antioxidant (Hinton et al., 2001). Another theory is that the toxicity of CCl4 in fish is decreased because fish livers lack Kupffer cells, the presence of which has been demonstrated to enhance the toxicity of carbon tetrachloride and other substances, such as cadmium chloride (CdCl2), in mammals (Sauer et al., 1997; Hinton et al., 2001). Lesions produced by exposure to CCl4 in English sole included coagulative necrosis of subcapsular and centrally located hepatocytes, sinusoidal congestion, and fatty change, and biochemical alterations included elevations in serum alanine aminotransferase and aspartate aminotransferase (Casillas et al., 1983). Among fishes in general, CCl4 exposure does not appear to produce any consistent pattern of hepatocellular necrosis (Droy and Hinton, 1988), and it has been postulated that the subcapsular necrosis observed in some cases may actually be an experimental artifact attributable to intraperitoneal injection (Hinton et al., 2001). The conspicuous pattern of hepatic lipidosis observed in cases of sublethal CCL4 intoxication in rodents is not a prominent feature in fish (Gingerich, 1982; Plaa and Charbonneau, 2001).
The consequences of 2 other well-known hepatotoxins are fundamentally similar to CCL4. Acetaminophen, which characteristically produces centrilobular necrosis in the rodent liver (Mitchell et al., 1973), is also less toxic to fish than it is to mammals, and toxic exposure in fish results in rare focal hepatocyte necrosis without any zonal pattern (Blair et al., 1990). In mammals, cell damage caused by acetaminophen follows microsomal enzyme-induced bioactivation of this compound to an electrophilic intermediate, N -acetylbenzoquinoneimine, which may then overwhelm and deplete its detoxifying agent, glutathione (Moslen, 1996; Plaa and Charbonneau, 2001). The reduced hepatotoxic effects of acetaminophen in fish may stem from a relative inability to convert it to this reactive intermediate (Thomas and Wofford, 1984; Hinton et al., 2001). Allyl formate is a classic cause of periportal necrosis in rodents (Kelly, 1993). Its toxicity is dependent on the enzyme alcohol dehydrogenase, which converts allyl formate to the reactive metabolite acrolein, an alkylating aldehyde that causes membrane damage (Kelly, 1993; Moslen, 1996). In mammalian livers, alcohol dehydrogenase is preferentially located in hepatocytes adjacent to portal triads (Moslen, 1996), however, no such localization for this enzyme is evident in fish (Schar et al., 1985). Allyl formate causes severe necrosis with hemorrhage in trout due to endothelial cell toxicity (Droy et al., 1989; Metcalfe, 1998). At high doses there is no zonal distribution pattern, but a perivenous pattern is evident at low doses (Droy et al., 1989).
It is unlikely that any of the above compounds will be encountered by fish in other than experimental settings. A group of naturally occurring hepatotoxins that have been extensively studied in fish are the microcystins. Produced by cyanobacteria (blue-green algae) such as Microcystis aeruginosa, microcystins have been implicated as a potential cause of fish kills (Kotak et al., 1996). The best characterized agent of this group is microcystin-LR, the form of toxin that contains
Summary
The livers of fishes have essentially the same basic anatomic components and metabolic machinery as mammalian livers. Due to certain anatomic and physiologic considerations, hepatic toxicity in fish tends to be less severe than in mammals, and fish do not generally display a zonal response pattern to hepatic intoxication. Morphologic features of liver toxicity are often exacerbations of findings that may be observed in normal or control fish. Finally, the ability of fish to survive extensive liver necrosis suggests they may be valuable animal models for studying sublethal hepatic toxicity and liver regeneration.
Footnotes
Acknowledgments
The authors are indebted to the following persons and institutions who contributed photographs and/or reference materials to this project: Dr. John W. Fournie; Dr. Daniel R. Diet-rich; Ms. Lourdes Mena; The National Toxicology Program, The National Institutes of Health; The Registry of Tumors in Lower Animals, The National Cancer Institute, National Institutes of Health, and The United States Army.