A protective role for cyclooxygenase-2 in drug-induced liver injury in mice.

Despite the utility of cyclooxygenase (COX) inhibition as an antiinflammatory strategy, prostaglandin (PG) products of COX-1 and -2 provide important regulatory functions in some pathophysiological states. Scattered reports suggest that COX inhibition may also promote adverse drug events. Here we demonstrate a protective role for endogenous COX-derived products in a murine model of acetaminophen (APAP)-induced acute liver injury. A single hepatotoxic dose caused the selective induction of COX-2 mRNA and increased PGD2 and PGE2 levels within the livers of COX(+/+) male mice suggesting a role for COX-2 in this model of liver injury. APAP-induced hepatotoxicity and lethality were markedly greater in COX-2(-/-) and (-/+) mice in which normal PG responsiveness is altered. The significantly increased toxicity linked to COX-2 deficiency could be mimicked using the selective COX-2 inhibitory drug, celecoxib, in COX(+/+) mice and was not due to alterations in drug-protein adduct formation, a surrogate for bioactivation and toxicity. Microarray analyses indicated that increased injury associated with COX-2 deficiency coincided, most notably, with a profoundly impaired induction of heat shock proteins in COX-2(-/+) mice suggesting that PGs may act as critical endogenous stress signals following drug insult. These findings suggest that COX-2-derived mediators serve an important hepato-protective function and that COX inhibition may contribute to the risk of drug-induced liver injury, possibly through both nonimmunological and immunological pathways.

Halothane-induced liver injury in outbred guinea pigs: role of trifluoroacetylated protein adducts in animal susceptibility.

Halothane causes a mild form of liver injury in guinea pigs that appears to model the hepatotoxicity seen in approximately 20% of patients treated with this drug. In previous studies, it was concluded that the increased susceptibility of some outbred guinea pigs to halothane-induced liver injury is not caused by their inherent ability to metabolize halothane to form toxic levels of trifluoroacetylated protein adducts in the liver. In this study, we reevaluated the role of trifluoroacetylated protein adducts in halothane-induced liver injury in guinea pigs. Male outbred Hartley guinea pigs were treated with halothane intraperitoneally. On the basis of serum alanine aminotransferase levels and liver histology, treated animals were designated as being susceptible, mildly susceptible, or resistant to halothane. Immunoblot studies with the use of anti-trifluoroacetylated antibodies showed that susceptible guinea pigs for the most part had higher levels of trifluoroacetylated protein adducts in the liver 48 h after treatment with halothane than did less susceptible animals. In support of this finding, the level of trifluoroacetylated protein adducts detected immunochemically in the sera of treated guinea pigs correlated with sera levels of alanine aminotransferase activity. In addition, the levels of cytochrome P450 2A-related protein but not those of other cytochrome P450 isoforms, measured by immunoblot analysis with isoform-specific antibodies, correlated with the amount of trifluoroacetylated protein adducts detected in the livers of guinea pigs 8 h after halothane administration. The results of this study indicate that the susceptibility of outbred guinea pigs to halothane-induced liver injury is related to an enhanced ability to metabolize halothane in the liver to form relatively high levels of trifluoroacetylated protein adducts. They also suggest that cytochrome P450 2A-related protein might have a major role in catalyzing the formation of trifluoroacetylated protein adducts in the liver of susceptible guinea pigs. Similar mechanisms may be important in humans.

Expression profiling of acetaminophen liver toxicity in mice using microarray technology.

Drug-induced hepatotoxicity causes significant morbidity and mortality and is a major concern in drug development. This is due, in large part, to insufficient knowledge of the mechanism(s) of drug-induced liver injury. In order to address this problem, we have evaluated the modulation of gene expression within the livers of mice treated with a hepatotoxic dose of acetaminophen (APAP) using high-density oligonucleotide microarrays capable of determining the expression profile of >11,000 genes and expressed sequence tags (ESTs). Significant alterations in gene expression, both positive and negative, were noted within the livers of APAP-treated mice. APAP-induced toxicity affected numerous aspects of liver physiology causing, for instance, >twofold increased expression of genes that encode for growth arrest and cell cycle regulatory proteins, stress-induced proteins, the transcription factor LRG-21, suppressor of cytokine signaling (SOCS)-2-protein, and plasminogen activator inhibitor-1 (PAI-1). A number of these and other genes and ESTs were detectable within the liver only after APAP treatment suggesting their potential importance in propagating or preventing further toxicity. These data provide new directions for mechanistic studies that may lead to a better understanding of the molecular basis of drug-induced liver injury and, ultimately, to a more rational design of safer drugs.

Epidemic of liver disease caused by hydrochlorofluorocarbons used as ozone-sparing substitutes of chlorofluorocarbons.

BACKGROUND: Hydrochlorofluorocarbons (HCFCs) are used increasingly in industry as substitutes for ozone-depleting chlorofluorocarbons (CFCs). Limited studies in animals indicate potential hepatotoxicity of some of these compounds. We investigated an epidemic of liver disease in nine industrial workers who had had repeated accidental exposure to a mixture of 1,1-dichloro-2,2,2-trifluoroethane (HCFC 123) and 1-chloro-1,2,2,2-tetrafluoroethane (HCFC 124). All nine exposed workers were affected to various degrees. Both compounds are metabolised in the same way as 1-bromo-1-chloro-2,2,2-trifluoroethane (halothane) to form reactive trifluoroacetyl halide intermediates, which have been implicated in the hepatotoxicity of halothane. We aimed to test whether HCFCs 123 and 124 can result in serious liver disease. METHODS: For one severely affected worker liver biopsy and immunohistochemical stainings for the presence of trifluoroacetyl protein adducts were done. The serum of six affected workers and five controls was tested for autoantibodies that react with human liver cytochrome-P450 2E1 (P450 2E1) and P58 protein disulphide isomerase isoform (P58). FINDINGS: The liver biopsy sample showed hepatocellular necrosis which was prominent in perivenular zone three and extended focally from portal tracts to portal tracts and centrilobular areas (bridging necrosis). Trifluoroacetyl-adducted proteins were detected in surviving hepatocytes. Autoantibodies against P450 2E1 or P58, previously associated with halothane hepatitis, were detected in the serum of five affected workers. INTERPRETATION: Repeated exposure of human beings to HCFCs 123 and 124 can result in serious liver injury in a large proportion of the exposed population. Although the exact mechanism of hepatotoxicity of these agents is not known, the results suggest that trifluoroacetyl-altered liver proteins are involved. In view of the potentially widespread use of these compounds, there is an urgent need to develop safer alternatives.

UDP-glucose:glycoprotein glucosyltransferase associates with endoplasmic reticulum chaperones and its activity is decreased in vivo by the inhalation anesthetic halothane.

Halothane causes an idiosyncratic hepatitis that is thought to result, in part, from immune reactions against one or more lumenal endoplasmic reticulum (ER) proteins that have been covalently modified by the trifluoroacetyl chloride metabolite of halothane. In this study, we have identified a 170 kDa protein target of halothane in the liver of rats. The 170 kDa protein was first detected when proteins in lysates of hepatocytes from halothane-treated rats were immunoprecipitated with antisera against several resident ER proteins. This 170 kDa protein was found to be associated with other protein targets of halothane, including protein disulfide isomerase, a protein disulfide isomerase isoform, a 59 kDa carboxylesterase, and 78 kDa glucose-regulated protein. Immunoblotting with antiserum directed against the trifluoroacetylated hapten indicated that the 170 kDa protein was trifluoroacetylated. Based upon its subcellular localization, molecular mass, N-terminal amino acid sequence, and antigenicity, the trifluoroacetylated 170 kDa protein was identified as UDP-glucose:glycoprotein glucosyltransferase (UGGT), a lumenal ER protein that is thought to have a role in the folding of N-linked glycoproteins. Moreover, treatment of rats with halothane caused a 44% decrease in the activity of liver microsomal UGGT, and at least 36% of the change in the activity of the enzyme could be due to a decrease in the level of the protein. The results suggest that the function of UGGT in folding of N-linked glycoproteins may be affected by other resident ER proteins or xenobiotics such as halothane.

Human cytochrome P450 2E1 is a major autoantigen associated with halothane hepatitis.

Autoantibodies against specific human cytochrome P450s have been found in the sera of patients suffering from a variety of diseases, including those caused by drugs. In the cases of tienilic acid- and dihydralazine-induced hepatitis, patients have serum autoantibodies directed against cytochromes P450 2C9 and P450 1A2, respectively. In the present study, we have found that 25 of 56 (45%) patients diagnosed with halothane hepatitis have autoantibodies that react with human cytochrome P450 2E1 that was purified from a baculovirus expression system. The autoantibodies inhibited the activity of cytochrome P450 2E1 and appeared to be directed against mainly conformational epitopes. In addition, because cytochrome P450 2E1 became trifluoroacetylated when it oxidatively metabolized halothane, it is possible that the covalently altered form of cytochrome P450 2E1 may be able to bypass the immunologic tolerance that normally exists against cytochrome P450 2E1. A similar mechanism may explain the formation of autoantibodies that have been found against other cellular targets of the reactive trifluoroacetyl chloride metabolite of halothane.

Anti-cytochrome P450 autoantibodies in drug-induced disease.

Drugs may induce hepatitis through immune mechanisms. In this review we have used the examples of 2 drugs to elucidate the first steps leading to the triggering of such disease, namely tienilic acid (TA) and dihydralazine (DH). These drugs are transformed into reactive metabolite(s) by cytochrome P450 (2C9 for TA and 1A2 for DH) (step 1). The reactive metabolites produced are very short-lived and bind directly to the enzymes which generated them (step 2). A neoantigen is thus formed which triggers an immune response (step 3), characterized by the presence of autoantibodies in the patient's serum (step 4). The autoantibodies are directed against the cytochrome P450 which generated the metabolite(s). Although the process by which TA and DH induce-hepatitis has been elucidated, further studies are necessary to generalize this mechanism. In addition, an animal model will also be useful to fully understand the immune mechanism of this type of disease.

cDNA cloning and baculovirus expression of the human liver endoplasmic reticulum P58: characterization as a protein disulfide isomerase isoform, but not as a protease or a carnitine acyltransferase.

The function of a 58-kDa liver microsomal protein (P58) is controversial. To help clarify the physiological function of this protein, particularly in humans, a full-length human liver cDNA clone was isolated, sequenced, and expressed in milligram quantities with the use of a baculovirus expression system. The deduced amino acid sequence of the mature protein contained two thioredoxin-like active site motifs (CGHC) and in its C-terminus a nuclear localization motif (KPKKKKK), and an ER-retention/retrieval motif (QEDL). The mature form of human P58 shared 95% amino acid sequence identity with the deduced amino acid sequences of a bovine liver cDNA, 93% with a murine B lymphocyte cDNA, and 91% with a rat basophilic leukemia cell cDNA. In contrast to reports on the activities of nonhuman forms of P58, the purified expressed human P58 showed no carnitine acyltransferase or protease activities. However, it did have protein disulfide isomerase activity, indicating that the physiological activity of human liver P58 may be attributed, at least in part, to this activity.

Interactions of dihydralazine with cytochromes P4501A: a possible explanation for the appearance of anti-cytochrome P4501A2 autoantibodies.

The antihypertensive drug dihydralazine may, on rare occasions, cause immunoallergic hepatitis characterized by anti-cytochrome P450 (P450)1A2 autoantibodies. To understand the first steps leading to this immune reaction, we studied the covalent binding fo dihydralazine metabolites to microsomes from rat and human livers. Upon incubation with NADPH and microsomes, dihydralazine formed metabolites that reacted with heme (as evidenced by destruction of heme, formation of 445-nm light-absorbing complexes, and covalent binding of heme to P450 apoprotein) and covalently bound to microsomal proteins. Formation of these metabolites was shown (by NADPH dependence, induction by beta-naphthoflavone, and immunoinhibition by anti-P4501A antibodies) to be mediated by P4501A. Finally, these metabolites appeared to bind to P4501A2, which produced them. These results support the following scheme for the first steps of this autoimmune reaction: P4501A2 metabolizes dihydralazine into reactive metabolites that then bind to it, forming a neoantigen that triggers an immune response characterized by autoantibodies against P4501A2.

Immunotoxicology and expression of human cytochrome P450 in microorganisms.

Drug-induced hepatitis can be caused by an abnormal immunological response. In the case of tienilic acid- and dihydralazine-induced hepatitis, we postulated a scheme in which a P450 produced a reactive metabolite (step 1); this reactive metabolite bound to the P450 producing it (step 2) leading to a neoantigen triggering the immune response (step 3); the autoantibodies produced during the immune response recognized the P450 producing the reactive metabolite (step 4). The use of microorganisms (yeast or bacteria) expressing cloned human P450 helped in proving some steps of this postulated scheme, particularly steps 1 and 4.

Anti-liver microsomes autoantibodies and dihydralazine-induced hepatitis: specificity of autoantibodies and inductive capacity of the drug.

Anti-liver microsomes (anti-LM) autoantibodies in patients with dihydralazine-induced hepatitis were found to react specifically with cytochrome P4501A2 (P4501A2) but not with P4501A1 expressed in yeast and bacteria. These results were confirmed by immunoinhibition of methoxyresorufin-O-demethylase activity (supported by the P4501A subfamily); anti-LM antibodies more strongly inhibited this activity in yeast expressing P4501A2 than in yeast expressing P4501A1. Anti-LM were shown to be specific to the disease; in three cases, these autoantibodies were present at high titers during disease, whereas the titers decreased upon recovery and became undetectable a few months after recovery. Thus, there exists a time-dependent relationship between the disease and the autoantibodies, which does not prove that the autoantibodies are causative of the hepatitis; they might only be a marker. The inductive capacity of dihydralazine toward P450 was also studied. In rats treated in vivo and in human hepatocytes treated in vitro with dihydralazine, a 2-fold increase in P4501A2- and P4501A-supported monooxygenase activities was found. The levels of the other P450 isoforms tested were unchanged during treatment, both in vivo in rats and in vitro in cultures of human hepatocytes. In human hepatocytes, dihydralazine produced a dose-dependent increase in the level of P4501A up to 0.1 mM; induction of P4501A was less strong at 0.2 mM and disappeared at 0.5 mM. The same treatment did not change the level of P4503A4, taken as control. The strong heterogeneity in the expression of P4501A enzymes in human liver and the capacity of these enzymes for induction by dihydralazine and by other compounds might be predisposing factors in this autoimmune disease.

Anti-liver endoplasmic reticulum autoantibodies are directed against human cytochrome P-450IA2. A specific marker of dihydralazine-induced hepatitis.

Sera from patients with dihydralazine-induced hepatitis were shown to contain anti-liver microsomal autoantibodies (anti-LM) by indirect immunofluorescence. These anti-LM antibodies were different from anti-liver/kidney microsomes (anti-LKM) 1 or 2 autoantibodies which have been previously described. Sera recognized a single 53,000 = Mr polypeptide in human liver microsomes as judged by immunoblotting, and the target antigen was identified as cytochrome P-450IA2 (P-450IA2) by (a) comparison of immunoblotting patterns with anti-human P-450IA2 and anti-rat P-450IA2 and with five anti-LM sera, and (b) specific immunoinhibition of microsomal ethoxyresorufin and phenacetin O-deethylation activities (both P-450IA2 supported reactions) by anti-LM antibodies. Finally, purified human P-450IA2 was recognized by these anti-LM sera. The anti-LM antibodies are specific for the disease because none of the other antisera tested behaved in the same manner as anti-LM, even those from patients treated with dihydralazine and without hepatic disease. A possible role of P-450IA2 in the metabolism of dihydralazine was suggested by competitive inhibition of ethoxyresorufin-O-deethylase observed in microsomal incubations. Thus, a new example is presented in which a cytochrome P-450 may be a target for autoantibodies in drug-induced hepatitis.