Drug-induced hepatotoxicity is a major clinical problem leading to discontinuation of drug development and withdrawal of drugs from the market. Drugs like acetaminophen (N-acetyl-p-aminophenol [APAP]) are considered safe at therapeutic doses but predictably cause various degrees of liver injury when an overdose is ingested. Because of the widespread use of this analgesic, APAP overdose is currently the most frequent cause of acute liver failure in the United States.1 In the past, mechanistic investigations focused on metabolic activation of APAP, depletion of glutathione, and covalent binding of the reactive metabolite N-acetyl-p-benzoquinone imine (NAPQI) to cellular proteins as the central cause of liver cell death.2 More recently, it was recognized that covalent binding is not sufficient to cause cell death but is an essential initiating signal for the toxicity that requires amplification within the cell.3 Intracellular events critical for cell death include mitochondrial dysfunction and formation of reactive oxygen and peroxynitrite. The mitochondrial oxidant stress triggers the mitochondrial membrane permeability transition pore opening, collapse of the mitochondrial membrane potential, adenosine triphosphate depletion, and release of intermembrane proteins, which are responsible for the characteristic nuclear DNA fragmentation of APAP-induced cell necrosis.3 Glutathione can protect liver cells by scavenging NAPQI and by detoxifying reactive oxygen and peroxynitrite. This mechanism is the basis for the clinical use of N-acetylcysteine, a glutathione precursor, as antidote against APAP toxicity.4
In addition to the intracellular signaling mechanisms of toxicity, cells of innate immunity have been considered as potential contributors to APAP-induced liver injury during the last decade. Kupffer cells, the resident macrophages of the liver, were among the first nonparenchymal cell types evaluated. It was reported that inactivation of Kupffer cells with gadolinium chloride completely eliminated APAP-induced peroxynitrite formation and hepatotoxicity in mice.5 However, because the most active Kupffer cells are located in the periportal areas, it is difficult to envision how these cells are able to selectively generate a centrilobular injury characteristic of APAP overdose. In contrast, gp91phox-deficient mice, whose phagocytes are unable to generate reactive oxygen through reduced nicotinamide adenine dinucleotide phosphate oxidase, showed the same oxidant stress, peroxynitrite formation, and hepatotoxicity as wild-type animals.6 These data do not support a relevant role of Kupffer cell–derived oxidant stress in the injury process. More importantly, true elimination of Kupffer cells with liposomal chlodronate actually increased APAP-induced liver injury.7 Although the reason for the enhanced toxicity is not completely understood, Kupffer cell depletion resulted in a dramatically reduced gene expression of interleukin-10 (IL-10), IL-6, and other mediators.7 IL-10 functions to limit inducible nitric oxide synthase expression and peroxynitrite-induced liver injury after APAP overdose, and IL-6 promotes regeneration. These and potentially other Kupffer cell–derived mediators may contribute to the protective effect.7 Together, these data support the hypothesis that the activation of Kupffer cells during APAP hepatotoxicity is overall beneficial because the anti-inflammatory effects outweigh potential protoxicity effects.
Another cell type of the innate immune system is the neutrophil, which is highly mobile and can effectively attack and kill any potential cellular target. An involvement of neutrophils in the pathophysiology has been documented in a number of acute liver disease processes including ischemia-reperfusion injury, obstructive cholestasis, and endotoxemia and sepsis.8 A potential role of neutrophils in the pathogenesis of APAP-induced liver injury was investigated.9, 10 Neutrophils initially accumulate in sinusoids, i.e., outside the area of injury, with a time course that closely follows the development of cell necrosis.9, 10 However, an antibody against β2 integrins (CD18) did not protect against APAP hepatotoxicity.9 Moreover, the fact that an antibody against high mobility group box-1 protein, which is released from necrotic cells, attenuated hepatic neutrophil accumulation but did not affect liver injury suggests that indeed dead or dying cells release mediators that activate neutrophils and recruit them into the liver vasculature.11 In addition, a substantial number of different interventions directed against neutrophils, including various inhibitors of adhesion molecules or NADPH oxidase or the respective gene knock-out mice, did not reduce APAP hepatotoxicity.6, 9, 10 Only one experimental approach yielded different results. Two groups independently showed that a neutropenia-inducing antibody administered 24 hours before APAP substantially attenuated liver injury.12, 13 However, this effect could not be reproduced when the antibody was injected 2 hours after APAP administration but before injury development and neutrophil accumulation.10 Why would this antibody protect only as a pretreatment? The answer is connected to the fate of circulating neutrophils after exposure to the antibody. The antibody-tagged neutrophils accumulate in capillaries of the liver and potentially other organs. They are functionally inactivated and do not pose any direct threat to the liver.14 Eventually, these cells or their remnants will be phagocytosed by Kupffer cells, which are being activated by this process.14 As a consequence, mediators produced by these activated macrophages trigger a stress response in hepatocytes, which up-regulate a number of genes including heme oxygenase-1 and in particular metallothionein.15 Induction of each of these stress genes has been shown to be protective against APAP hepatotoxicity. Thus, given the extensive evidence against neutrophils, the most likely explanation for the protective effect of pretreatment with a neutropenia-inducing antibody is the up-regulation of protective genes within hepatocytes rather than elimination of neutrophils.15
The liver contains a resident lymphocyte population, which consists mainly of natural killer (NK) and natural killer T (NKT) cells.16 Hepatic NK/NKT cells are known as an important defense against tumor cells, viruses, and other pathogens.16 In recent years evidence was provided that these lymphocytes may also play a role in several models of acute liver injury, most notably in APAP hepatotoxicity.17 Liu et al. demonstrated that depletion of NK and NKT cells significantly attenuated APAP-induced liver injury and improved survival.17 The effect of NK and NKT cells was related to the secretion of interferon-γ (IFN-γ), which modulated inflammatory chemokine formation, enhanced recruitment of leukocytes, and increased expression of Fas ligand on innate immune cells.17 Although there is neither evidence for a role of neutrophils9, 10 nor for Fas-mediated apoptosis18 in APAP hepatotoxicity, the report by Liu et al.17 was considered an important advance in our understanding of the pathophysiology even if the detailed mechanism was not yet elucidated. However, Masson et al. were initially unable to reproduce these findings.19 Upon further investigation, which included the consultation of the authors of the previous communication, Masson et al. realized that the involvement of NK and NKT cells in APAP hepatotoxicity was linked to the presence of the solvent dimethyl sulfoxide (DMSO) used to aid the dissolution of APAP.19 In the absence of DMSO, NK and NKT cells did not produce IFN-γ after the administration of various doses of APAP in male or female mice, and depletion of these lymphocytes had no protective effect.19 This novel observation prompted the authors to investigate the effect of DMSO on NK and NKT cell number and activation status in vivo. They could clearly demonstrate that low amounts of DMSO can promote recruitment of additional NKT cells into the liver and causes the activation of both NK and NKT cells as measured by increased IFN-γ and granzyme B expression.19 In addition, the authors demonstrated that the effect of DMSO was indeed responsible for the participation of NK and NKT cells in the pathophysiology of APAP hepatotoxicity. In other words, the use of the solvent DMSO modulated the mechanism of APAP-induced liver injury in vivo!
The findings reported by Masson et al. have several relevant pathophysiological implications. First, the results add another important biological effect of DMSO to its already known characteristics of being an antioxidant and free radical scavenger as well as being a potent inhibitor of cytochrome P450. In fact, a recent study pointed out that previously reported beneficial effects of a caspase inhibitor in the APAP model were caused exclusively by the reduced metabolic activation of APAP by DMSO, which was used to dissolve the inhibitor.20 However, in contrast to this artifact, which could have been avoided by using proper solvent controls, the effect of DMSO on NK and NKT cells as described by Masson et al. is unexpected and was previously not detected because all experimental groups were treated with APAP in DMSO. Thus, given the widespread use of solvents to facilitate the dissolution of drugs and chemicals for experimental use, more awareness of these potential pitfalls on the part of investigators is clearly necessary to avoid future misinterpretations of experimental data and disease mechanisms caused by use of these solvents. A second critical finding of these data is the fact that activation of NK and NKT cells through an independent stimulus can trigger the pathophysiological involvement of these innate immune cells in APAP hepatotoxicity. Because bacterial and viral pathogens and many other agents can activate NK and NKT cells,16 these mediators may have the potential to modulate an individual's susceptibility to liver injury induced by APAP overdose or other drugs. In addition, activation of innate immune cells by pathogens or other stimuli may contribute to certain idiosyncratic drug toxicities.
In summary, the manuscript by Masson et al. is the latest in a series of articles which question the recent “discoveries” of detrimental effects of Kupffer cells, neutrophils, and NK/NKT cells in the pathophysiology of APAP-induced liver injury. The limited role of innate immune cells in APAP hepatotoxicity under normal conditions is consistent with the fact that isolated mouse hepatocytes can be killed by APAP with a mechanism similar to the one observed in vivo. Nevertheless, these articles indicate that innate immune cells are either activated during APAP-induced liver injury or that activation through independent stimuli may have the potential to affect the mechanism of toxicity. This suggests that under specific circumstances innate immune cells may aggravate the toxicity of APAP in certain individuals. Although this insight may enhance our understanding of individual susceptibility to drug-induced liver disease, the majority of experimental evidence points toward targeting intracellular signaling mechanisms of cell injury in hepatocytes as the most effective way to limit liver damage after APAP overdose. This is the reason why the currently used antidote N-acetylcysteine is unlikely to be replaced anytime soon.