Results and Discussion

In our previous studies, it was reported that three-dimensional cultures of a human FLC-4 cell line synthesized and secreted liver-specific proteins (albumin, α-fetoprotein, etc.). These cells also have a high capacity to metabolize drugs and were used to produce reactive metabolites of carbamazepine (Kato et al., 2014). In the hepatocyte supernatant, carbamazepine was decreased (0.1–0.048 mM), and carbamazepine-10,11-epoxide (1 μM) and 2-hydroxyiminostilbene (0.26 μM) were detected in the hepatocyte supernatant with 7-day incubation of carbamazepine (Supplemental Fig. 1). In addition, there was a metabolite with an m/z 253, consistent with 2- and/or 3-hydroxycarbamazepine. Given that the epoxide is chemically reactive and 2-hydroxyiminostilbene is a product of the reactive iminoquinone, undoubtedly the metabolism through these pathways was more extensive than indicated by the observed metabolites. We found that THP-1 cells were not activated by carbamazepine or carbamazepine-10,11-epoxide, but they were activated by 2-hydroxyiminostilbene, leading to the release of IL-1β and production of active caspase-1 (Fig. 2A). However, the culture supernatant from FLC-4 hepatocytes with carbamazepine did lead to the activation of inflammasomes in THP-1 cells (Fig. 2B). IL-1β production was inhibited by the addition of the cytochrome P450 inhibitor (ABT). In addition, 2-hydroxyiminostilbene and the supernatant of FLC-4 cells increased caspase-1 activity in differentiated THP-1 cells (Fig. 2, C and D).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

THP-1 cells incubated with 2-hydroxyiminostilbene or the supernatant of FLC-4 cells with carbamazepine led to release of IL-1β and an increase in caspase-1 activity. (A) IL-1β secretion by differentiated THP-1 cells, which were treated with carbamazepine, carbamazepine-10,11-epoxide, or 2-hydroxyiminostilbene for 24 hours. (B) IL-1β produced by incubation of differentiated THP-1 cells, which were treated with the supernatant of FLC-4 cells, with carbamazepine for 7 days, with or without a cytochrome P450 inhibitor (ABT) or a caspase-1 inhibitor [Ac-YVAD-CHO (YVAD)]. Caspase-1 activity of differentiated THP-1 cells, which were treated with 2-hydroxyiminostilbene (C) or the supernatant from hepatocytes with carbamazepine (D), with or without YVAD, for 24 hours. Statistical comparisons were performed using the Tukey multiple comparison tests. A P value <0.05 was considered to be statistically significant (**P < 0.01, n = 3–5). RLU, relative light unit.

We also identified DAMPs in the culture supernatant of hepatocytes. Higuchi et al. (2012) reported that the mRNA expression of S100A8 and S100A9 significantly increased in mice with carbamazepine-induced liver injury. However, in the Higuchi et al. study, a very high dose of carbamazepine was used; this induced acute toxicity, which is presumably not immune-mediated. In our study, although S100A8 and S100A9 proteins were not detected, HSP60 was significantly increased in the culture supernatant of hepatocytes treated with carbamazepine (Figs. 2 and 3). The cytochrome P450 inhibitor, ABT, prevented this increase in HSP60. Furthermore, the supernatant from hepatocytes cultured with carbamazepine and treated with an anti-HSP60 antibody did not increase IL-1β production in THP-1 cells, which suggests that HSP60 activates the inflammasome. It has been reported that HSP60 mediates IL-1β production by regulating the NLRP3 inflammasome pathway (Swaroop et al., 2018). Therefore, the release of HSP60 from hepatocytes caused by reactive metabolites of carbamazepine may be responsible for the activation of inflammasomes in differentiated THP-1 cells.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

HSP60 was released from hepatocytes as DAMPs. (A) Western blot analysis of proteins in culture medium [high mobility group box 1 (HMGB1), HSP32, HSP40, HSP60, HSP90, S100A8, and S100A9] that were released into the supernatant from incubation of hepatocytes with carbamazepine incubated for 7 days with carbamazepine (100 μM), and their quantitative analysis of HSP40 (B), HSP60 (C), and HSP90 (D). (E) IL-1β secreted by differentiated THP-1 cells, which were treated with carbamazepine, with or without anti-HSP60 antibody (HSP60Ab) for 24 hours. HSP60 rabbit polyclonal antibody was added to the supernatant from the FLC-4 cells, which had been cultured with carbamazepine for 7 days. Statistical comparisons were performed using the Tukey multiple comparison test. A P value <0.05 was considered to be statistically significant (**P < 0.01, n = 3).

Additionally, signal 2, which is produced by costimulatory molecules on APCs, is required to induce an immune response (danger hypothesis) (Matzinger, 1994). The results of this study are consistent with the working hypothesis that the mechanism of carbamazepine-induced hypersensitivity reactions involves oxidation of carbamazepine in the liver to the iminoquinone of 2-hydroxyiminostilbene. This can covalently bind to proteins in the liver, causing the release of DAMPS. It can also be reduced to 2-hydroxyiminostilbene and reach the blood. In other tissues, the 2-hydroxyiminostilbene can be reoxidized to the iminoquinone by peroxidases and activate inflammasomes there. Reactive metabolites may activate the inflammasome indirectly by inducing the release of DAMPs, or may directly activate inflammasomes. The activated inflammasome results in caspase-1–mediated conversion of pro-IL-1β into its bioactive mature form, IL-1β. IL-1β is a strong proinflammatory molecule capable of promoting an immune response (Guo et al., 2015). Therefore, the activation of inflammasomes may be an important bridge between reactive metabolite formation and the induction of an immune response. In some patients with the right HLA/T cell receptor combination, this activation can lead to a hypersensitivity reaction.

An independent piece of information also argues against involvement of the 10,11-epoxide in the mechanism of carbamazepine-induced hypersensitivity reactions. Oxcarbazepine also causes hypersensitivity reactions similar to those of carbamazepine. The mechanism must be similar because oxcarbazepine-induced toxic epidermal necrolysis is associated with the same HLA genotype as toxic epidermal necrolysis caused by carbamazepine (Chen et al., 2017). That implies that the drug-modified protein binds with a similar orientation. Oxcarbazepine has the potential to form an iminoquinone similar to that of carbamazepine, but it cannot form a 10,11-epoxide. However, we cannot rule out the possibility that the epoxide or other metabolites contribute to the release of DAMPs from hepatocytes.

In conclusion, our results support the hypothesis that idiosyncratic or hypersensitivity reactions associated with carbamazepine involve oxidation of carbamazepine into reactive metabolites in hepatocytes, the most likely being the iminoquinone. This results in the production of DAMPs that activate inflammasomes, thereby resulting in an immune response. This method may provide a method to study the mechanism of idiosyncratic drug reactions and even predict which drug candidates are likely to cause such adverse reactions.