Recent evidence suggests that macrophages and/or other nonparenchymal cells may release important mediators contributing to the hepatic necrosis induced by high doses of acetaminophen (APAP). The nature and causative role of these mediators has remained elusive, however. To investigate the role of the proinflammatory cytokine, tumor necrosis factor (TNF) in the initiation and early propagation of APAP-induced liver injury, we have used mice deficient in both TNF and the closely related lymphotoxin-␣ (LT-␣). Male TNF/LT-␣ knockout mice and C57BL/6 wild-type mice were treated with a hepatotoxic dose of APAP (400 mg/kg, intraperitoneally), and the development of liver injury was monitored over 8 hours. Both genotypes exhibited similar basal activities of hepatic cytochrome P450 2E1 and 1A2. After APAP administration, both the rate of glutathione consumption and the extent of subsequent selective protein binding did not differ significantly in the knockout and wild-type mice. The TNF/LT-␣-deficient mice developed severe centrilobular necrosis and exhibited highly increased levels of serum alanine aminotransferase and aspartate aminotransferase, the extent of which was not significantly different from that in wild-type mice. In C57BL/6 mice exposed to APAP, no increases in hepatic transcripts of TNF or LT-␣ were found by reverse transcription-polymerase chain reaction, nor was immunoreactive serum TNF detected by enzyme-linked immunosorbent assay over 8 hours posttreatment. These data indicate that, in the absence of the genes encoding for TNF and LT-␣, APAP bioactivation was not altered and mice still developed severe hepatic necrosis. Thus, TNF is unlikely to be a key mediator in the early pathogenesis of APAP-induced hepatotoxicity. (HEPATOLOGY 1998;27:1021-1029.) Despite considerable research, the mechanisms responsible for acetaminophen (APAP) hepatotoxicity in humans and experimental animals are not yet fully understood (for reviews see Nelson, 1 Vermeulen et al., 2 Hinson et al., 3 and Cohen et al. 4
). Metabolic activation 5 of APAP and subsequent binding of its electrophilic metabolite N-acetyl-p-benzoquinone imine (NAPQI) to glutathione (GSH) and, after GSH depletion, to nucleophilic protein targets 6 seems to be necessary but is not sufficient for toxicity. 4,7 Several other mechanistic components of parenchymal cell injury have been proposed, including the generation of reactive oxygen species, 8-11 mitochondrial dysfunction [12][13][14] followed by disruption of calcium homeostasis and energy supply, or induction of apoptosis. 15 Apart from these events occurring intracellularly in parenchymal cells after APAP exposure, an important role has been attributed to nonparenchymal cells. [16][17][18] In particular, macrophage recruitment and activation has been implicated in APAP-induced liver injury. 17,19 According to this concept, hepatocytes initially injured by APAP would release factors that chemoattract and subsequently activate Kupffer cells and other mononuclear phagocytes to the centrilobular regions. These activated macrophages are then thought to contribute to hepatocellular damage by releasing cytotoxic mediators. The nature of these mediators has not been elucidated, but recent evidence suggests that tumor necrosis factor (TNF) may be a possible candidate, at least in the progressive phases of liver injury. 18,[20][21] However, the role of TNF in the early pathogenesis of APAP-induced hepatocellular damage is unknown. In particular, it remains unclear whether TNF release is a primary event triggered by APAP bioactivation or a secondary mechanism that follows inflammation and necrosis. Specifically, no clear information is available on the severity of APAP-induced liver injury when TNF synthesis and release are inhibited.
With the advent of gene-targeting techniques it has become possible to selectively knock out genes and to study their roles in drug toxicity. Here, to specifically explore the role of TNF, we have compared the early hepatic changes in mice deficient for TNF with those in normal C57BL/6 wild-type mice. We have chosen a double-knockout model, in which mice were made deficient for both TNF and lymphotoxin-␣ (LT-␣), 22 because these two closely related cytokines may cross-compete for binding at the TNF receptors on target cells. 23 The results indicate that, although the genes for