Acetaminophen (AAP) is an effective and safe pain-relieving drug when therapeutic doses are taken. However, an overdose of AAP causes centrilobular necrosis, which in severe cases can lead to liver failure, in both experimental animals and humans 12. It is well established that formation of a reactive metabolite, presumably N-acetyl-p-benzoquinone imine (NAPQI), by microsomal P450 isoenzymes, is essential for the development of AAP-induced liver toxicity 2. NAPQI is readily conjugated with glutathione and excreted from hepatocytes 3. However, excessive NAPQI formation results in covalent binding to sulfhydryl groups of proteins 24.

Although protein binding is a critical early event in AAP hepatotoxicity, this mechanism alone cannot explain the severe cell injury. Therefore, several amplifying mechanisms have been postulated. AAP treatment leads to Kupffer cell activation 5 and recruitment of neutrophils into the liver 6. Furthermore, AAP metabolism causes mitochondrial dysfunction 789, which results in mitochondrial oxidant stress 10 and peroxynitrite formation 11. Recently, we could show that selective scavenging of peroxynitrite with glutathione (GSH) effectively protects parenchymal cells in the liver against AAP-induced cell injury despite continued mitochondrial oxidant stress 12. This suggests that peroxynitrite plays a critical role in the mechanism of AAP-induced hepatocellular toxicity.

Microvascular disturbances and injury may also be relevant for the progression of AAP-induced liver injury. Walker et al. described sinusoidal endothelial cell (SEC) injury with trapping of red blood cells in the space of Disse (hemorrhage) during AAP-induced liver injury in mice 1314. Recently, Ito et al. demonstrated SEC swelling and impaired endothelial scavenger function, which preceded parenchymal cell injury 15. Subsequent accumulation of erythrocytes in the space of Disse and reduced sinusoidal blood flow indicate substantial microvascular dysfunction 15. These microvascular changes may cause ischemic injury, platelet aggregation and thrombosis, neutrophil accumulation and inflammatory injury 16. Severe hemorrhage can cause hypovolemic shock 17. However, the mechanism and pathophysiological relevance of these events for AAP-induced liver injury remain unclear. Activated Kupffer cells can generate reactive oxygen and nitric oxide 18 and may cause vascular injury 19. Kupffer cells have been implicated in the mechanism of hepatocellular injury and peroxynitrite formation after AAP overdose 2021. Therefore, the objective of this investigation was to test the hypothesis that Kupffer cells may cause SEC injury through formation of vascular peroxynitrite formation.

A dose of 300 mg/kg AAP had no effect on liver tissue at 1 h but caused severe centrilobular necrosis with hemorrhage at 6 h (Figure 1). Red blood cells were trapped in the space of Disse due to sinusoidal cell injury. The large number of red blood cells in the liver was responsible for the increased hepatic hemoglobin content at that time (Figure 2). The time course of hemoglobin accumulation indicated that the sinusoidal cell injury occurred between 2 and 4 h after AAP administration (Figure 2). Thus, the damage to the vascular lining cells developed parallel to the parenchymal cell injury 1122. Immunohistochemical staining for nitrotyrosine, an indicator for peroxynitrite formation, demonstrated selective staining of vascular lining cells at 1 h after AAP (Figure 1). However, at 6 h, staining of centrilobular hepatocytes was evident. To test if Kupffer cells may be responsible for the oxidant stress and injury, animals were pretreated with GdCl₃ to inactivate Kupffer cells. GdCl₃ treatment had no significant effect on liver injury as indicated by high plasma ALT values (Figure 3). Animals treated with GdCl₃ still had severe hemorrhage and centrilobular necrosis (Figure 4). On the other hand, pretreatment with allopurinol, which previously was shown to protect against AAP-induced parenchymal cell injury by preventing mitochondrial injury 11, completely eliminated the increase in plasma ALT values (Figure 3) and evidence of necrosis (Figure 4). Furthermore, allopurinol treatment prevented hemorrhage, which suggests that it also prevented sinusoidal endothelial cell injury. Similarly, nitrotyrosine staining as indicator for AAP-induced peroxynitrite formation was neither attenuated in vascular endothelial cells at 1 h nor in parenchymal cells at 6 h by GdCl₃ treatment (Figure 5). On the other hand, allopurinol treatment prevented endothelial cell and parenchymal cell nitrotyrosine staining (Figure 5).

The objective of this investigation was to test the hypothesis that Kupffer cell-derived reactive oxygen and peroxynitrite could be responsible for vascular and parenchymal cell injury. Previous studies suggested that Kupffer cells are activated after AAP overdose 5 and are relevant contributors to the overall liver injury in rats 20. More recently, a similar conclusion was reached using the mouse model 21. However, the reported baseline ALT activities for GdCl₃-pretreated animals did not correlate with histology, which still showed severe centrilobular necrosis 21. Our data indicate that Kupffer cells play at most a minor role in the pathophysiology. Gadolinium chloride (GdCl₃), which functionally inactivates Kupffer cells to produce less reactive oxygen 23, neither reduced the early vascular nor the later parenchymal cell staining for nitrotyrosine. Moreover, GdCl₃ treatment had no significant effect on vascular cell injury (hemorrhage) or hepatocellular necrosis. These findings are consistent with recent preliminary data showing no attenuation of nitrotyrosine staining or injury in phox-deficient mice, which have no functional NADPH oxidase, the main superoxide producing enzyme in Kupffer cells 24. In addition to this direct evidence against the involvement of Kupffer cells in the murine model of AAP-induced liver injury, there are other observations that argue against this hypothesis. In general, the most active Kupffer cells are located in the periportal areas 2526. Activation of these cells results in a predominantly periportal to midzonal injury 27. In contrast, AAP causes a strict centrilobular necrosis and hemorrhage (Figure 1) with the earliest and most severe injury affecting the cells closest to the central vein 1122. Thus, overall our results are consistent with a number of observations, which do not support a role of Kupffer cells in vascular peroxynitrite formation and injury.

Another potential source of vascular oxidant stress could be infiltrating neutrophils 2829. These phagocytes are recruited into the liver in response to the injury 6, i.e., several hours after the occurrence of vascular nitrotyrosine staining 11. No evidence for a systemic activation of neutrophils was found at any time after AAP treatment 6. In addition, antibodies against CD18, the common subunit of beta₂ integrins, which functionally inactivate hepatic neutrophils 3031, had no effect on AAP-induced hepatotoxicity 6. These findings suggest that neutrophils are not a relevant source of reactive oxygen species in the vasculature after AAP overdose.

In parenchymal cells, AAP induces mitochondrial swelling 32 and dysfunction 78, oxidant stress 10, cytochrome c release 9, peroxynitrite formation 11 and a reduction in cellular ATP levels 10. Preventing mitochondrial dysfunction with allopurinol treatment eliminated the oxidant stress, peroxynitrite formation and cell injury 1011. On the other hand, if peroxynitrite was scavenged by GSH, injury was attenuated despite continued mitochondrial dysfunction 12. The findings suggest that peroxynitrite is a critical mediator of AAP-induced liver injury. Our present data show that allopurinol treatment prevented the vascular nitrotyrosine staining and hemorrhage. It was previously shown that AAP caused severe depletion of GSH and injury in cultured sinusoidal endothelial cells 33. These observations document the capacity of sinusoidal endothelial cells to metabolically activate AAP. Since NAPQI, the reactive metabolite of AAP, is responsible for the mitochondrial oxidant stress and peroxynitrite formation in parenchymal cells 11, and the fact that allopurinol prevented vascular nitrotyrosine staining and injury suggests that AAP may have caused a mitochondrial oxidant stress and peroxynitrite formation in endothelial cells. Thus, endothelial cell damage and hemorrhage occurred parallel to the parenchymal cell injury through similar mechanisms.

What is the pathophysiological relevance of the vascular injury in the liver? Without interventions, the massive hemorrhage can lead to hypovolemic shock and death, as shown in other models of sinusoidal endothelial cell injury 17. However, the early vascular injury between 2 and 4 h had no effect on hepatic ATP levels 10. These results suggest that the initial hemorrhage does not lead to significant tissue ischemia.

Nevertheless, injury and prolonged dysfunction of sinusoidal endothelial cells, even without severe hemorrhage, can be expected to have a negative impact on liver function. Sinusoidal endothelial cells have not only a barrier function in the liver vasculature but play an important role in clearing a large number of macromolecules and colloids from the circulation. Collagen-, mannose-, Fc gamma-, and hyaluranon scavenger receptors are vital for the turnover of extracellular matrix proteins and the removal of immune complexes 34. Reduced uptake of formaldehyde-treated serum albumin as early as 2 h after AAP administration demonstrated dysfunction of the hyaluranon scavenger receptor 15.

Our data argue against Kupffer cells as relevant source of vascular oxidant stress during AAP-induced sinusoidal endothelial cell injury. Our data suggest that, similar to the mechanism in parenchymal cells, mitochondrial oxidant stress and peroxynitrite formation may be critical for sinusoidal endothelial cell injury.