In vitro hepatotoxicity of Petasites hybridus extract (Ze 339) depends on the concentration, the cytochrome activity of the cell system, and the species used

Abstract Ze 339, a CO2 extract prepared from the leaves of Petasites hybridus, possesses antispasmodic and anti‐inflammatory effects and is proven to be effective in the treatment of allergic rhinitis. To study possible hepatotoxic effects of Ze 339, its main constituents and metabolites, a series of in vitro investigations were performed. Furthermore, different reconstituted fractions of extract (petasins and fatty acid fraction) were examined in three in vitro test systems using hepatocytes: Two human cell lines, with lower and higher activity of cytochrome P450 enzymes (HepG2, HepaRG) as well as a rodent cell line with high cytochrome P450 activity (H‐4‐II‐E), were used. Metabolic activity, assessed by the WST‐1 assay, was chosen as indicator of cytotoxicity. To assess potential bioactivation of Ze 339 compounds, metabolic experiments using S9 fractions from rats, dogs, and humans and isolated cytochromes (human/rat) were performed, and the formation of reactive metabolites was assessed by measuring cellular concentrations of glutathione and glutathione disulphide. Our data revealed that the cytotoxicity of Ze 339, its single constituents, and main metabolites depends on the concentration, the cytochrome activity of the cell system, and the species used.

Nine cases of severe clinical hepatotoxicity of a medicinal product containing P. hybridus root extract (Petadolor/Dolomed) for migraine prophylaxis occurred between November 2001 and February 2002 using therapeutic doses (Evers, 2009). This led to the subsequent withdrawal of the Marketing Authorization of medicinal products containing the root extract in Switzerland in 2004 (Swissmedic, 2004).
However, up to now, no evidence is available regarding the hepatotoxicity of the leaf extract Ze 339. Thus, it was the aim of this study to determine possible in vitro hepatotoxic effects of Ze 339, the petasin isomers petasin, isopetasin, and neopetasin as well as the metabolites petasol and isopetasol.
As a marker for cytotoxicity, the WST-1 assay was chosen.
To assess potential bioactivation, metabolic experiments using S9 fractions from Sprague Dawley rats, beagle dogs and humans, and isolated cytochromes (human and rat) were performed. The formation of reactive metabolites was indirectly assessed by measuring the cellular concentrations of glutathione and glutathione disulphide levels after treatment with Ze 339.
The extract Ze 339, its major active constituents petasin, isopetasin, and neopetasin, and petasol and isopetasol were tested to evaluate their contribution to possible cytotoxic effects. The latter compounds were precursors of petasin isomers, to a minor extent constituents of Ze 339 as well as metabolites. However, a rigorous species dependent formation was not studied in vivo in different species.
To investigate whether the petasin and the fatty acid fractions contribute to the cytotoxic effects, the following mixtures were pre-

| Concentrations applied
Ze 339 (containing 36.4% of total petasins) the mixtures and the reconstituted extract were applied in final concentrations between 0.02 and 100 μg/ml (containing up to 113 μM total petasins). For toxicity evaluation of the single constituents, a concentration range between 3.9 and 250 μM was used.

| Cytotoxicity
HepG2 and H-4-II-E cells were differentiated for 24 hr. HepaRG cells were directly seeded in precoated 96-well microtiter plates and were incubated in William's medium with general purpose supplements according to manufacturer's manual until adherence. All cells were treated with Ze 339 (up to 100 μg/ml) or petasin, isopetasin, neopetasin, petasol, isopetasol, petasin mixture, fatty acid mixture, and reconstituted extracts (up to 250 μM) for 4 hr.
In the three cell lines, cytotoxicity was defined as a decrease in metabolic activity of ≥20% in the WST-1 assay. At the end of the experiment, 10 μl WST-1 reagent was added to each well. Plates were then incubated for 2 hr to allow for the reduction of WST-1 reagent. Absorbance was measured using a microplate absorbance reader (Infinite M 200, Tecan Trading Ltd., Switzerland) at 450 nm, reference wavelength of 620 nm. Due to some matrix effect, the lowest concentration (0.02 μg/ml for Ze 339 and 3.9 μM for single constituents and mixtures) of each substance tested was defined as 100% metabolic activity.
Results were expressed as percentage of the control metabolic activity, and data were presented as mean values ± SEM (n = 3-6).

| Metabolism of isolated constituents
The metabolism of petasin, isopetasin, and neopetasin was investigated using three different S9 fractions originating from three different species (isolated from rat, dog, and human liver homogenates containing cytochrome P450 isoforms) and isolated cytochromes of two different species (human and rat; rat analogues of the respective human cytochromes were used). For the expression of the different cytochromes, a baculovirus expression system was used according to supplier's manual. The S9 fractions or respectively single P450 cytochromes were used to mimic the liver metabolism by in vitro incubation.
The assay was carried out in triplicate in 1.5 ml glass vials with a final concentration of 20 μM of petasin, isopetasin, and neopetasin.
The incubation was performed in a 37 C warm water bath. S9 fractions were thawed slowly on ice. Four different approaches were pursued by testing: (a) petasin, neopetasin, and isopetasin before (0 min) and after 60 min incubations with S9 fraction; (b) petasin, neopetasin, and isopetasin without S9 fraction after 60 min incubation; (c) vehicle as negative controls; and (d) cofactors of S9 fraction after 60-min incubation.
A mixture of NADPH regeneration systems A (20-fold) and B (100-fold), phosphate buffer, and H 2 O was added, preincubated at 37 C, and shaken gently for 5 min. Afterwards, the recommended amount between 10 and 40 pmol/ml of enzyme or enzyme mixture was added according to manufacturer's instruction and incubated for further 60 min at 37 C and gently shaken. Afterwards, the reactions were stopped by addition of acetonitrile (400 μl). For the control sample, acetonitrile was added at 0 min incubation. The samples were centrifuged for 3 min at 10,000 × g. The supernatant was transferred to high-performance liquid chromatography vials and analyzed by ultra-performance liquid chromatography mass spectrometry (UPLC/ MS). The UPLC analysis was performed with an UPLC Acquity H Class from Waters (Waters Corporation, Switzerland) including a quaternary high-pressure gradient pump, an automatic sample injector, and a column thermostat. Chromatographic separation was achieved on an Acquity BEH C18, 1.7 μl, 50 × 2.1 mm column (Waters, Switzerland).
The mobile phase consisted of 0.1% formic acid (A) and acetonitrile (B). The initial gradient condition was 10% (B) followed by a step to 90% (B) until 6.0 min. The column temperature was adjusted to 40 C.
The flow rate was 0.5 ml/min and the injection volume 5 μl. The mass spectrometry was performed in positive electrospray mode using an ACQUITY QDa mass detector (Waters Corporation, Switzerland). The single ion recording was set to 317.2 Da.

| Effects on GSH and GSSG
The kinetics of glutathione (GSH) depletion and glutathione disulfide (GSSG) formation after incubation with Ze 339 was determined in HepaRG cells. Cells were seeded with a density of 20 × 10 4 cells in 100 μl medium per well in a 96-well assay plate. Afterwards, Ze 339 (concentrations up to 100 μg/ml) and the solvent (0.5% DMSO) were added. Then, cells were incubated for further 4 hr at 37 C with 5% CO 2 atmosphere. The amounts of GSH and GSSG were determined using GSH/GSSG-Glo™ assay, a luminescence-based assay for detection and quantification of total glutathione (GSH + GSSG), GSSG and GSH/GSSG ratios in cultured cells. The assay was performed according to manufacturer's instruction (Tecan Infinite M200, Switzerland).

| Statistics
Data were analyzed by descriptive analysis and graphically displayed by Origin 2017 Software (Originlab Corp., USA). Concentration-response analysis was done using the sigmoidal Hill equation:

F I G U R E 2
Mean metabolic activity (WST-1 test) of Ze 339 and isolated constituents (petasin, neopetasin, isopetasin, petasol, and isopetasol) after administration in different cell lines.

(a) Effects of Ze 339 in HepaRG, HepG2, and H-4-II-E cells (n = 6); (b) effects of isolated constituents in HepG2 cells (n = 3); (c) effects of isolated constituents in HepaRG cells (n = 3); and (d) effects of isolated constituents in H-4-II-E cells (n = 3) [Colour figure can be viewed at wileyonlinelibrary.com]
comparable, mild cytotoxicity, in the highest concentration a decrease in metabolic activity down to 60-80% of the control (Figure 3a).
In HepaRG cells, neither the fatty acid mixture alone nor the petasin mixture inhibited metabolic activity. The combination of these mixtures, the reconstituted extract, however, showed a moderate reduction in metabolic activity (30-50%) similar to Ze 339 in HepaRG cells (Figure 3b).
In H-4-II-E cells, the toxic effects of all treatments seemed to be intensified: The petasin mixture, the reconstituted extract as well as Ze 339 showed marked reduction in metabolic activity (below 30%).
Even the fatty acid mixture indicated a mild 26% reduction in metabolic activity (Figure 3c).

| Metabolism of Ze 339 and single constituents
After incubation of S9 fractions from different species with the three isolated constituents petasin, isopetasin, and neopetasin, a significant (p < .01 or p < .001) species-dependent metabolism after 60 min incubation was observed with the following rank order: rat>dog>human ( Figure 4a). Incubation with either solvent or only cofactors of S9 fraction as negative controls did not show relevant metabolism of petasin, isopetasin, and neopetasin (with solvent: <1%, <1%, and <1% and with cofactor: 10%, <1%, and <1%, respectively, data not shown).
Positive controls for each of the isolated cytochromes showed an extent of metabolism between 31.2% (CYP1A2) and 100% (CYP2C9) for human cytochromes and between 23.1% (CYP2D2) and 100% (CYP2B1) for rat cytochromes (data not shown). In HepaRG cells, the effect of cytochrome inhibition on the toxic effects of Ze 339 was investigated ( Figure 5a); addition of quinidine (CYP2D6 inhibitor) and montelukast (CYP2C8 inhibitor) attenuated (p < .001 and p = .041, respectively); most of the cytotoxic effect of Ze 339 demonstrated for the highest concentration studied (100 μg/ml). No significant effect on Ze 339 toxicity was observed for the other CYP inhibitors (see Figure S1).
The results of the latter experiment indicate that several cytochromes contribute at least partly to the toxic effect, presumably by formation of reactive metabolites (biotransformation). The biggest contribution was shown for CYP2D6 ( Figure 5a).

| Determination of glutathione depletion and glutathione disulphide formation
The investigations of the effect of Ze 339 on the cellular redox status revealed a concentration-dependent decrease in GSH concentration from 19.46 μM at 0.02 μg/ml down to 13.54 μM at 100 μg/ml Ze 339 (Figure 5b). Correspondingly, the GSSG formation was concentration-dependently increased after treatment with Ze 339.

| DISCUSSION
The cytotoxicity of Ze 339 and its single constituents petasin, isopetasin, neopetasin, isopetasol, and petasol was shown to be dependent on the concentration used, the cytochrome activity, and species. Comparison of two human cell lines (HepG2 and HepaRG) with different activities of cytochrome P450 enzymes revealed that Ze 339 and its single constituents petasin, isopetasin, neopetasin, isopetasol, and petasol exerted higher toxicity in cells having a higher cytochrome activity. Furthermore, Ze 399 and petasin, isopetasin, and neopetasin were more toxic in rat cells (H-4-II-E), which are known to have a high expression of cytochrome P450 (Fujimura et al., 2012;Westerink et al., 2008)  expressed cytochromes. Besides daily dose (≥100 mg) and lipophilicity (LogP≥3), bioactivation was identified as a significant risk factor for the development of drug-induced hepatotoxicity (Chen, Borlak, & Tong, 2013;Yu et al., 2014). The importance of these risk factors was also recently confirmed by us in larger dataset (Hammann, Schöning, & Drewe, 2019). This may explain the observed species-specific effects in preclinical in vivo studies, with rats being the most sensitive species, followed by dogs, and the good tolerability profile in clinical studies and post-marketing observations. Experiments with isolated cytochromes showed that the metabolic pathway of petasin, isopetasin, and neopetasin involves different cytochromes, further explaining the species-specific sensitivity to the toxic effects of Ze 339. Inhibition of different cytochromes in HepaRG showed an attenuation of the toxic effects, which was pronounced for CYP2D6 and to a minor extend CYP2C8. One limitation of the study is the use of HepaRG cells, which is an immortalized hepatic cell line that retains many characteristics of primary human hepatocytes (Antherieu et al., 2010). Exposure of HepaRG cells to potential inducers resulted in the induction of most of the major drug metabolizing CYP isoenzymes (Aninat et al., 2006;Kanebratt & Andersson, 2008). Therefore, the European Medicines Agency (EMA, 2012) classified HepaRG cells as supportive model to cultured (fresh or cryopreserved) human hepatocytes. However, HepaRG cells are generated from a donor carrying a polymorphism for CYP2D6 (Guillouzo et al., 2007), which leads to a lower expression of this enzyme in Hep-aRG cells compared with, for example, primary hepatocytes (Kanebratt & Andersson, 2008;Zanelli, Caradonna, Hallifax, Turlizzi, & Houston, 2012), and therefore, CYP2D6 activity might be underestimated. On the other hand, because CYP2D6 inhibition by quinidine resulted in attenuated toxicity of Ze 339 in HepaRG cells, this still emphasizes the involvement of this cytochrome enzyme in the development of the hepatotoxicity.
The hypothesis of metabolic bioactivation of Ze 339 is further supported by the observation that Ze 339 administration concentration-dependently decreases cellular GSH concentration and GSSG formation, which may indicate the response to oxidative stress.
Comparison of the toxicity of different fraction mixtures and combinations thereof showed that the petasin mixture (36.4% of the extract) alone was less toxic than Ze 339 itself. The fatty acid mixture alone (~30% of the extract) did not exert any toxic effects. However, the combination of the petasin and the fatty acid mixture (reconstituted extract) had comparable toxic effects as Ze 339 itself. One possible explanation for this synergistic effect is that the fatty acid mixture may increase the solubility of the highly lipophilic petasin isoforms (XlogP of about 4.5, https://pubchem.ncbi.nlm.nih.gov/) in the hydrophilic extracellular medium, thus, enhancing the cellular uptake of the fatty acid/petasin complex.
The clinically recommended daily dose of Ze 339 for the treatment of allergic rhinitis is two to three tablets containing up to 80-120 mg of the extract (corresponding to 16-24 mg petasins).
In a clinical study in 21 healthy volunteers (Vogt & Will-Shahab, 1999), two tablets were administered resulting in mean petasin plasma concentrations of 25.5 ng/ml. We could demonstrate in our present experiments that administration of the petasin mixture at a concentration above 25 μg/ml resulted in only a subtoxic reduction of metabolic activity in HepaRG cells by 12.5 % (Figure 3b). Thus, using clinically recommended doses of Ze 339 in patients, the resulting peak plasma concentration is 1,000-fold lower, indicating that toxic effects of Ze 339 occur only after high, supra-pharmacologic doses.
Since the Marketing Authorization of a medicinal product con-