2019 Volume 44 Issue 3 Pages 145-153
A high incidence of positive results is obtained with in vitro genotoxicity tests, which do not correlate with the in vivo negative results in many cases. To address this issue, the metabolic profile of rat liver 9000 × g supernatant fraction (S9) pretreated with phenobarbital (PB) and 5,6-benzoflavone (BNF) was characterized. Furthermore, the in vitro micronucleus tests of 10 compounds were performed with PB-BNF-induced rat S9. PB-BNF increased cytochrome P450 (CYP) activity and CYP1A1, CYP1A2, CYP2B1/2, CYP2C6, CYP3A1, and CYP3A2 expression in rat S9, whereas it decreased CYP2C11 and CYP2E1 expression. PB-BNF-induced S9 enhanced the micronucleus induction (MI) of benzo[a]pyrene (BaP), cyclophosphamide (CPA), and 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine hydrochloride (PhIP), which are metabolized by CYP1A1, CYP2C6, and CYP1A2, respectively. In contrast, coumarin and chlorpheniramine showed MI with PB-BNF-induced S9 despite the fact that they show negative results in the in vivo studies. Furthermore, diclofenac, piroxicam, lansoprazole, and caffeine showed MI regardless of the enzyme induction by PB-BNF, whereas phenacetin did not show MI. These results indicate that PB-BNF-induced rat S9 is effective in detecting the genotoxic potential of promutagens, such as BaP, CPA, and PhIP, but not of coumarin and chlorpheniramine, probably due to the differences in the in vitro and in vivo metabolic profile and its exposure levels of the drugs.
To detect the genotoxic potential of pharmaceuticals, three test systems are defined as an option in the International Conference on Harmonisation (ICH) S2(R) guideline: a test for gene mutation in bacteria, a cytogenetic test for chromosomal damage, and an in vivo micronucleus test (ICH, 2011). However, a high incidence of positive results, particularly in the in vitro cytogenetic test, has been reported for many drugs, which did not correlate with in vivo genotoxicity tests (Kirkland et al., 2005, 2007). The reason for the false positive results in the in vitro test remains unclear, but the difference in the metabolic profile of the drug between the in vitro and in vivo assay systems is a likely cause of the irrelevant outcome (Ku et al., 2007).
Many genotoxic compounds that need to be metabolized to induce genotoxicity in vivo are classified as promutagens (Ames et al., 1973; Paolini and Cantelli-Forti, 1997). As bacteria or cultured mammalian cells used in the in vitro genotoxicity tests have limited capacity for metabolizing drugs, several types of the metabolic activation systems are applied to the in vitro test systems. The rat liver 9000 × g supernatant fraction (S9) pretreated with drug-metabolizing enzyme inducers, such as Aroclor 1254 or a combination of phenobarbital (PB) and 5,6-benzoflavone (BNF), are widely used in the in vitro genotoxicity tests (Ames et al., 1973; Paolini and Cantelli-Forti, 1997), with high sensitivity for the detection of genotoxic potential of several promutagens (Ames et al., 1975; Matsuoka et al., 1979). These S9s increased the expression of cytochrome P450 (CYP) 1A and CYP2B isozymes compared with the S9 from naïve rats (Guengerich et al., 1982). Therefore, the application of these S9s in the in vitro genotoxicity test may help in detecting the genotoxic potential of promutagens metabolized by CYP1A and 2B with high sensitivity. In contrast, the expression of CYP2C or CYP2D isozymes involved in the metabolism of many drugs remained unchanged or decreased (Guengerich et al., 1982), which indicated that these S9s may not be adequate in evaluating the genotoxic potential of promutagens metabolized by CYP2C or CYP2D. Furthermore, the effect of these S9s on the expression of other isozymes, such as CYP2E and CYP4A, remains to be investigated. Although the weight of evidence approach containing metabolic condition was recommended to judge the genotoxic and carcinogenic potential of the drugs adequately (Thybaud et al., 2007), we focused on the relationship of the metabolic profile of the various compounds by PB-BNF-induced S9 and its genotoxic potential in the in vitro assay.
To address this issue, we characterized the metabolic profile of the liver S9 from rats pretreated with PB and BNF (PB-BNF-induced rat S9) and compared it with that from untreated rats (untreated rat S9). Next, we conducted an in vitro micronucleus test with PB-BNF-induced rat S9 for nine clastogenic and one non-clastogenic compounds metabolized by several types of CYP isozymes: clastogenic compounds include benzo[a]pyrene (BaP) (Ishidate et al., 1988; McManus et al., 1990), 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine hydrochloride (PhIP) (Knasmüller et al., 1999; McManus et al., 1990), caffeine (Ishidate et al., 1988; Tassaneeyakul et al., 1993), coumarin (Born et al., 2002; NTP, 2017), cyclophosphamide (CPA) (Ishidate et al., 1988; Yu et al., 1999), diclofenac (Brambilla and Martelli, 2009; Yan et al., 2005), piroxicam (Brambilla and Martelli, 2009; Hobbs and Twomey, 1981), lansoprazole (Brambilla and Martelli, 2009; Pearce et al., 1996), and chlorpheniramine (Brambilla and Martelli, 2009; Yasuda et al., 2002) and non clastogenic compound is phenacetin (NTP, 2017; Tassaneeyakul et al., 1993). The results indicated that the PB-BNF-induced rat S9 was effective in detecting the genotoxic potential of promutagens, such as BaP, CPA, and PhIP, but not of coumarin and chlorpheniramine, probably due to the differences in the in vitro and in vivo metabolic profile of the drugs. Our results emphasize that it is essential to select an optimal metabolic condition based on the metabolic profile of the drug candidate to adequately assess its genotoxic potential in the in vitro micronucleus test system.
Benzo[a]pyrene (BaP), 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine hydrochloride (PhIP) and caffeine were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Phenacetin, coumarin, cyclophosphamide monohydrate (CPA), diclofenac sodium, piroxicam, lansoprazole, and chlorpheniramine maleate were purchased from Sigma-Aldrich Japan K.K. (Tokyo, Japan). BaP, coumarin, CPA, piroxicam, lansoprazole, and phenacetin were dissolved in dimethyl sulfoxide (DMSO, Wako Pure Chemical Industries, Ltd.). PhIP, caffeine, diclofenac, and chlorpheniramine were dissolved in water for injection (Otsuka Pharmaceutical Factory, Inc., Tokushima, Japan).
Two types of rat liver S9s were purchased from Oriental Yeast Co., Ltd. (Tokyo, Japan). One was prepared from the liver of 7-week-old male Sprague–Dawley (SD) rats which were pretreated with a single administration of 30 mg/kg PB, followed by a 3-day repeated administration of 60 mg/kg PB, and a single administration of 80 mg/kg of BNF at the last day of PB treatment (PB-BNF-induced rat S9). The livers were excised at two days after the last PB treatment under isoflurane anesthesia. The other was prepared from the liver of untreated 7-week-old male SD rats (untreated rat S9).
The P450 contents in each S9 were determined from the spectral difference in absorbance at 490 nm and 450 nm and the protein concentrations in each S9 were determined by measurement of the absorbance at 750 nm as previously reported (Omura and Sato, 1964). As markers of total CYP activity, the 7-methoxycoumarin O-demethylase (MCD), 7-ethoxycoumarin O-deethylase (ECD), and 7-propoxycoumarin O-depropylase (PCD) activities in each S9 were measured as previously reported (Matsubara et al., 1983) using a spectrophotofluorometer (F-2000, Hitachi Ltd., Tokyo, Japan). For western blot analysis, the protein concentration of each S9 was adjusted to 2 mg protein/mL for CYP1A1, CYP1A2, CYP2B1/2, CYP2C6, CYP2C11, CYP2D1, CYP2E1, CYP3A1, CYP3A2, and CYP4A and NADPH P450 reductase. These samples were diluted two-fold to 1 mg protein/mL with Tris-SDS-Mercapto Ethanol Sample Loading Buffer (Cosmo Bio Co., Ltd., Tokyo, Japan) and heated at 95°C for 5 min. Then, 10 μL (for CYP1A1, CYP1A2, CYP2B1/2, CYP2E1, CYP3A1, CYP3A2 and CYP4A) or 3 μL (for CYP2C6, CYP2C11, and CYP2D1 and NADPH P450 reductase) of each dilution was applied onto 7.5% SDS-polyacrylamide gel (Funakoshi Co., Ltd., Tokyo, Japan). These proteins were separated by SDS-polyacrylamide gel electrophoresis and blotted onto Immobilon-P Transfer Membranes (Millipore Corporation, Bedford, MA, USA). The membranes were blocked with ECL blocking agent (GE Healthcare Japan Corporation, Tokyo, Japan) and treated with primary antibodies (anti-rat CYPs sheep antibody, Merck Japan, Tokyo, Japan; mouse antibody, Abcam K.K. Tokyo, Japan or Fitzgerald Industries International Inc., Concord, MA, USA; rabbit antibody, GE Healthcare Japan Corporation), biotin-labeled secondary antibodies (biotinylated anti-sheep, mouse, or rabbit antibody, GE Healthcare Japan Corporation), streptavidin-horseradish peroxidase conjugate (GE Healthcare Japan Corporation), and ECL western blotting detection reagent (GE Healthcare Japan Corporation) and exposed to instant film (Fujifilm Corporation, Tokyo, Japan). For identifying the contribution of PB or BNF to expression levels for each CYP isozyme in PB-BNF-induced rat S9, the western blot analysis was also conducted with rat microsomal fractions pretreated with PB at 100 mg/kg/day for 7 days or 3-MC at 20 mg/kg/day for 7 days. 3-MC instead of BNF was used because it increases protein expression of CYP1A1 and CYP1A2 by the same mechanisms as BNF (Thomas et al., 1983).
Chinese hamster lung fibroblast cell line (CHL) obtained from RIKEN BioResource Center (Ibaraki, Japan) was maintained in Eagle’s medium (Nissui Pharmaceutical Co., Ltd., Tokyo, Japan) containing 10% calf serum (HyClone Laboratories, Inc., South Logan, UT, USA), 2 mM L-glutamine (Wako Pure Chemical Industries, Ltd.), and 25 mM NaHCO3 (Otsuka Pharmaceutical Factory, Inc.). Cells were cultured in a humidified atmosphere with 5% CO2 at 37°C.
The cells were seeded in 6-well plates at a density of 1 × 105 cells/well. After 24 hr of incubation, the cells were treated with nutrient growth medium containing various concentrations of the test compounds. The final concentration of vehicles was 10% for water for injection or 1% for DMSO. The metabolic activation system (S9 mix) was prepared with each S9 and additives: 4 mM HEPES (Dojindo Laboratories, Kumamoto, Japan), 5 mM MgCl2 (Wako Pure Chemical Industries, Ltd.), 33 mM KCl (Wako Pure Chemical Industries, Ltd.), 5 mM G6P·Na2 (Sigma-Aldrich Japan K.K.), and 4 mM NADP·Na (Sigma-Aldrich Japan K.K.) according to the method reported by Matsuoka et al. (1979). The concentration of S9 in each culture medium was fixed as 5%. The treatment period and recovery period were set at 24 hr and 18 hr for continuous treatment without PB-BNF-induced rat S9 mix and 6 hr and 36 hr for short treatment with or without PB-BNF-induced rat S9 mix, respectively. In this study, precipitation was not considered for selection of the dose because the objective of the present study was to clarify the effect of inducer on micronucleus induction.
At the end of the treatment and recovery periods, the cells were treated with 0.01% trypsin and collected. The number of cells was counted with a cell viability analyzer (ViCELL XR, Beckman Coulter Inc., Brea, CA, USA). Two wells per concentration were evaluated for each test article, and their mean values were calculated. The relative cell counts (RCC) was calculated by the following formula:
RCC (%) = mean of cell number in treatment group/mean of cell number in control group
The cells were seeded in plastic chamber slides at a density of 5 × 103 cells/chamber. After 24 hr of incubation, the cells were treated with the same procedure as in subjected to the cell growth assay. At the end of the treatment and recovery periods, the cells were treated with pre-warmed 0.075 M KCl (Wako Pure Chemical Industries, Ltd.) for 5 min at 37°C and fixed with ice-cooling fixative [solution of 6% acetic acid /methanol (Wako Pure Chemical Industries, Ltd.)]. Then, the slide chambers were sufficiently air-dried. Two chambers per concentration were prepared for each test compound. The slides were stained with 40 μg/mL acridine orange (Wako Pure Chemical Industries, Ltd.) solution and observed under a microscope (200× magnification). The micronucleus induction was evaluated at the dose levels without severe cytotoxicity, which could be counted the sufficient number of cells. Namely, one thousand cells per chamber (2000 cells per concentration) were examined; the number of micronucleated cells per chamber was counted and their mean values were calculated. The number of micronucleated cells included both mono- and poly-micronucleated cells was selected as an endpoint in this study to clarify the relationship between the metabolic profile of compounds and genotoxic potential.
Based on the results of the cell growth assay and in vitro micronucleus test of the dose range-finding test, 3 to 7 dose levels included the dose level showing 50% relative survival or less were selected in the main test. In the main test, the short treatment with PB-BNF-induced rat S9 or untreated rat S9 was conducted because all compounds examined in the dose range-finding study did not induce micronucleus induction in the continuous treatment. At the end of the treatment and recovery periods, cell growth assay and in vitro micronucleus test were conducted as described above.
The P450 contents and CYP activities were expressed as mean of duplicates in single assay. The RCC and the number of micronucleated cells were expressed as mean values in the two independent dose-range finding and main studies with two parallels. For the in vitro micronucleus tests, the incidence of micronucleated cells were statistically analyzed by two-tailed Fisher’s exact test. The dose-dependency was also evaluated with Cochran-Armitage trend test. EXSUS (ver. 8.1.0, CAC Croit Corporation, Tokyo, Japan) was used for these statistical analyses. A P value less than 5% was considered statistically significant. When the number of micronucleated cells showed statistically significant increase with dose-dependency, the result was judged as positive.
In PB-BNF-induced rat S9, P450 content and MCD, ECD, and PCD activities were 1.26 nmol/mg protein and 0.68, 2.49, and 2.54 nmol/min/mg protein, respectively. They were about 4-5 times higher than those in untreated rat S9 (0.24 nmol/mg protein and 0.15, 0.45, and 0.62 nmol/min/mg protein, respectively) (Fig. 1). Western blot analysis revealed that protein expression of CYP1A1, CYP1A2, CYP2B1/2, CYP2C6, CYP3A1, and CYP3A2 and NADPH P450 reductase increased in PB-BNF-induced rat S9 compared with that in untreated rat S9 (Fig. 2). Conversely, the protein expression of CYP2C11 and CYP2E1 decreased in PB-BNF-induced rat S9. The expression of CYP2D1 and CYP4A was unchanged between them (Fig. 2). The western blot analysis of microsomal fractions showed that PB increased CYP2B1/2, CYP2C6, CYP3A1, and CYP3A2 and NADPH P450 reductase levels and decreased CYP1A2, CYP2C11, and CYP2E1 levels compared with those in untreated microsomal fractions. In addition, PB did not change the expression of CYP1A1, CYP2D1, and CYP4A (Fig. 2). 3-MC increased CYP1A1 and CYP1A2 levels and decreased CYP2B1/2 level but did not change CYP2C6, CYP2C11, CYP2D1, CYP2E1, CYP3A1, CYP3A2, and CYP4A and NADPH P450 reductase levels (Fig. 2).
Increases in P450 content and alkoxycoumarin dealkylase activity in PB-BNF-induced rat S9. a, P450 contents; b, alkoxycoumarin O-dealkylase (ACD) activities; Open bars, untreated rat S9; Gray bars; combination of phenobarbital and 5,6-benzoflavone-induced rat S9; MCD: methoxycoumarin-O-demethylase activity; ECD: ethoxycoumarin-O-deethylase activity; PCD: propoxycoumarin-O-depropylase activity.
Western blot analysis of cytochrome P450 isozymes and NADPH P450 reductase in rat liver fractions. 1, untreated rat S9; 2, PB-BNF-induced rat S9; 3, untreated rat liver microsomal fraction; 4, rat liver microsomal fraction pretreated with phenobarbital; 5, rat liver microsomal fraction pretreated with 3-methylcholanthrene; CYP: cytochrome P450.
As the results of the dose range-finding tests, among 10 compounds, 8 compounds (BaP, PhIP, coumarin, chlorpheniramine, CPA, diclofenac, piroxicam, and lansoprazole) significantly increased the number of micronucleated cells only on short treatment with PB-BNF-induced rat S9, but did not increase on short and continuous treatment without PB-BNF-induced rat S9. Caffeine increased them on short treatment with and without PB-BNF-induced rat S9, and phenacetin did not increase them under any treatment condition. In the main test, the reproducibility for results of the micronucleus induction by each test compound with PB-BNF-induced rat S9 was confirmed: BaP, PhIP, coumarin, and chlorpheniramine statistically significant increases in the number of micronucleated cells with PB-BNF-induced rat S9, while they had little or no impact on the induction of micronucleated cells with untreated rat S9 (Fig. 3). For coumarin, the micronucleus induction was observed at the dose levels in which the RCC was less than 50% in the main test, although the induction was noted at the dose levels in which the RCC was more than 50% in the dose-range finding study (data not shown). In addition, the micronucleus induction due to PhIP, coumarin, and chlorpheniramine was statistically dose-dependent (P < 0.001). Furthermore, the cytotoxicity of these drugs with PB-BNF-induced rat S9 was higher than those with untreated rat S9. CPA showed statistically significant induction of micronucleated cells with both PB-BNF-induced rat S9 and untreated rat S9, although the induction of micronucleated cells and cytotoxicity by CPA were higher in PB-BNF-induced rat S9 (Fig. 3). Diclofenac, piroxicam, lansoprazole, and caffeine induced dose-dependent increases in the number of micronucleated cells regardless of the enzyme induction by PB-BNF (P < 0.001). The degree of the increase in micronucleated cells and cytotoxicity by these drugs were comparable between both S9s (Fig. 3). Phenacetin did not show micronucleus induction with either type of S9 and cytotoxicity was comparable between both S9s (Fig. 3). The dose levels of all compounds inducing cytotoxicity and micronucleus induction were comparable between dose range-finding and main tests.
In vitro micronucleus test of 10 compounds with a PB-BNF-induced rat S9 and untreated rat S9. Open circles and squares indicate the RCC (%) of cultured cells with untreated rat S9 and PB-BNF-induced rat S9, respectively. Closed circles and squares indicate the micronucleus induction with untreated and PB-BNF-induced rat S9, respectively. In PhIP, coumarin, CPA, diclofenac and lansoprazole, the dose levels between PB-BNF-induced S9 and untreated S9 conditions or the dose levels evaluated for RCC and micronucleus induction were different based on their cytotoxicity. *P < 0.01: Significantly different from the control by two-tailed Fishers’ exact test. BaP: benzo[a]pyrene, CPA: cyclophosphamide; PhIP: 2-amino-1-methyl-6-phenylimidazo[4,5-b] pyridine hydrochlorid, RCC (%): relative cell count.
The present study investigated the impact of PB-BNF-induced S9 on the results of the in vitro micronucleus tests for several test compounds that are metabolized by several rat or human CYPs. The major human enzymes involved in phase 1 metabolism of drugs are CYP3A4/5/7, CYP2D6, CYP2Cs, CYP1A2, CYP2E1, CYP2A6, and CYP2B6 (Guengerich, 2003). In addition, some promutagens were reported to be metabolized by CYP1A1/2 (McManus et al., 1990). Therefore, in the present study, the rat isozymes, CYP3A1/2, CYP2D1, CYP2C6, CYP2C11, CYP2C12, CYP1A2, CYP2E1, and CYP2B1/2, corresponded to those of human, and CYP1A1 were selected and their expression levels were evaluated. Consequently, alkoxycoumarin O-dealkylase activity known as markers for CYP activity and protein expression of CYP1A1, CYP1A2, CYP2B1/2, CYP2C6, CYP3A1, CYP3A2, and NADPH P450 reductase, were increased in the PB-BNF-induced S9, but the protein expression of CYP2D1 and CYP4A or CYP2C11 and CYP2E1 were unchanged or decreased, respectively. Consistent with the present results, Guengerich et al. (1982) reported that b-naphthoflavone increased CYP1A1 and CYP1A2 expression levels and decreased CYP2C6, CYP2C11 and CYP3A expression levels. In addition, phenobarbital increased CYP2B1/2, CYP2C6, and CYP3A expression levels, and decreased CYP2C11 expression level (Guengerich et al., 1982). Taking our results and the above reports into consideration, BNF was attributable to increase in CYP1A1 and CYP1A2 expression levels, whereas PB was attributable to increase in CYP2B1/2, CYP2C6, CYP3A1, CYP3A2, and decrease in CYP2C11 expression levels in our experimental condition with PB-BNF-induced S9.
In the in vitro micronucleus test, PB-BNF-induced S9 promoted the micronucleus induction by BaP, PhIP, coumarin, and chlorpheniramine and enhanced it by CPA compared with untreated rat S9. No dose-dependent micronucleus induction due to BaP with PB-BNF induced rat S9 may be attributable to precipitation of the drug at 100 μg/mL because the number of micronucleated cells and the cytotoxicity of BaP were decreased at 100 μg/mL or more, although the observation of precipitation was not conducted. Among these drugs, BaP, CPA, and PhIP are metabolized into its genotoxic metabolites by CYP1A1, CYP2C6, and CYP1A2, respectively (Clarke and Waxman, 1989; McManus et al., 1990; Yu et al., 1999). In this study, the expression of these CYP isozymes prominently increased in the PB-BNF-induced S9. The results of the in vivo micronucleus and carcinogenicity studies of BaP (NTP, 2017), CPA (NTP, 2017), and PhIP (Okada et al., 2013; Shirai et al., 1997) were positive, which suggested that our assay condition with PB-BNF-induced S9 adequately predicted carcinogenicity with high sensitivity. Coumarin is metabolized into a cytotoxic intermediate, o-hydroxyphenylacetaldehyde (o-HPA), by CYP2E and CYP1A in rats (Born et al., 2002). In addition, rat microsomal CYP2B can metabolize coumarin into o-HPA (Peters et al., 1991). Therefore, under our assay condition, it is possible to consider that CYP1A1/2 and CYP2B1/2 overexpressed by PB and BNF produce excessive o-HPA, resulting in the positive result in the in vitro micronucleus test. This was supported by the strong cytotoxic action of coumarin on the CHL cells with PB-BNF-induced S9. Conversely, coumarin did not induce micronuclei in the bone marrow or peripheral blood of mice (Api, 2001; NTP, 2017) or unscheduled DNA synthesis in rat hepatocytes (Edwards et al., 2000). Since o-HPA was detoxified immediately to o-hydroxyphenylacetic acid (o-HPAA) in rats (Vassallo et al., 2004), the difference in the abundance of o-HPA under in vitro and in vivo conditions may be responsible for the discrepancy. Chlorpheniramine does not induce micronucleated cells and carcinogenicity in rodents (NTP, 2017). The reason for the discrepancy in the results between in vivo studies and our in vitro study remains to be clarified, but the difference in CYP isozymes involved in chlorpheniramine metabolism may have resulted in the positive result in the in vitro micronucleus assay. N-Demethylation of chlorpheniramine, which is stereoselectively mediated by CYP2C11, is a major pathway of its metabolism in rats (Nomura et al., 1997). However, the expression of CYP2C11 decreased in the PB-BNF-induced S9. In contrast, CYP1A and CYP2B also contribute to chlorpheniramine metabolism in the rat liver microsomal fraction. In this study, the expression of CYP1A and CYP2B increased in PB-BNF-induced S9, suggesting that the metabolites of chlorpheniramine mediated by CYP1A and CYP2B may be responsible for the positive results in the in vitro micronucleus test. However, the involvement of the other metabolites for chlorpheniramine in the positive results in the in vitro micronucleus test remains unclear, and therefore further investigation should be necessary.
Diclofenac, piroxicam, lansoprazole, and caffeine induced micronucleated cells to same extent regardless of the enzyme induction by PB-BNF in the in vitro micronucleus test. These results suggested that these compounds were metabolized to genotoxins through the metabolic pathways unaffected by the PB-BNF-induced S9. These genotoxins may be further metabolized or detoxified by pathways other than the PB-BNF-induced S9 in vivo, as evidenced by the negative results obtained in the rat in vivo cytogenetic tests for diclofenac (Brambilla and Martelli, 2009), piroxicam (Gris et al., 2008), lansoprazole (TOXNET, 2017), and caffeine (Aeschbacher et al., 1986).
Diclofenac is known to be metabolized into reactive metabolites by some CYP isozymes and rapidly detoxified as glutathione conjugates (Tang et al., 1999). In addition, lansoprazole was reported to be eliminated immediately from the circulation (Youssef et al., 2003). Therefore, the different kinetics of genotoxic metabolites by detoxification due to GSH or elimination rate of the parent compound in the in vivo condition is one of the likely causes of the discrepancy, however, further investigation is necessary to clarify its mechanism including the metabolism in vivo. Phenacetin did not induce the micronucleated cells in the present in vitro micronucleus test, which is consistent with a previous finding (De Flora et al., 1985). However, phenacetin induces micronucleated cells in mice (Higashikuni et al., 1992) and carcinogenicity in rats (Johansson, 1981). Therefore, the present in vitro test condition with rat PB-BNF-induced S9 was considered to be inadequate for evaluating the genotoxic potential of phenacetin, because the genotoxicity of phenacetin could be detected with Aroclor 1254-induced hamster liver S9 more efficiently than with PB-BNF-induced rat liver S9 (De Flora et al., 1985).
As described above, the PB-BNF-induced rat S9 was effective in detecting the genotoxic potential of promutagens such as CPA, BaP, and PhIP. However, it produced false positive results in the in vitro micronucleus tests for coumarin and chlorpheniramine, probably due to the differences in the in vitro and in vivo metabolic profile of the drug and its exposure levels (Yamamura et al., 2018). In addition, the assessment of the hydrophilic metabolites or the metabolites generated by non-CYP enzymes such as sulfotransferases and acetyltransferases is difficult in the in vitro micronucleus test using S9 systems (Kirkland et al., 2007; Ku et al., 2007). Therefore, it is important to adequately understand the drug metabolism of a drug candidate to avoid an irrelevant outcome in the in vitro micronucleus test. Our approach to clarify the metabolic profile including identification of metabolites and CYP isozymes attributable to the positive results in the in vitro micronucleus study could improve the predictability of in vivo genotoxicity study and human risk assessment. Our findings emphasize the need to select optimal metabolic conditions based on the in vivo metabolic profile of the drug candidates to appropriately evaluate their genotoxic potential in the in vitro test system. In case the metabolic profile of test compound is unclear, several assay systems including human liver S9 (Hakura et al., 2003), HepG2 systems expressed human CYP isozymes (Hashizume et al., 2009), HepaRG cell line (Jossé et al., 2012), and 3-dimensional in vitro assay system (Shah et al., 2018) could be valuable for detecting genotoxic potential of drugs.
The authors thank Dr. Toshiyuki Watanabe for useful comments and suggestions. The authors also thank Donald J. Hinman for English proofreading.
The authors declare that there is no conflict of interest.