2019 Volume 44 Issue 10 Pages 701-709
Genotoxicity and carcinogenicity profiles of drugs occasionally vary across species due to species difference in drug metabolic profile. To clarify the effect of species differences in the metabolic profile on micronucleus induction, we conducted an in vitro micronucleus test for seven clastogens (benzo[a]pyrene: BaP, cyclophosphamide monohydrate: CPA coumarin, diclofenac, piroxicam, lansoprazole, and chlorpheniramine) with rat, mouse, monkey, dog, or human liver S9. BaP, CPA, coumarin, diclofenac, piroxicam, and lansoprazole induced micronucleus formation with all species of S9s, whereas chlorpheniramine did not induce micronucleus formation in any of the S9s. BaP and CPA revealed remarkable species differences in micronucleus induction, whereas coumarin, diclofenac, piroxicam, and lansoprazole did not present any differences. Interestingly, the amounts of hydroxy-BaP-epoxides and phosphamide mustard, which might be associated with micronucleus induction by BaP and CPA, respectively, were correlated with the degree of micronucleus induction among the five species. In conclusion, the species difference in micronucleus induction by BaP and CPA was attributable to the differences in the metabolic profiles of these drugs among species. Our results indicate that it is crucial to understand the effect of species differences in the metabolic profile of drug candidates on genotoxicity and carcinogenicity potential and to predict their risk in human.
Toxicity profile of the drug candidates occasionally varies across species due to the differences in its affinity or sensitivity to the target molecules, or pharmacokinetics (Gonzalez and Shah, 2008; Wang et al., 2007; Eberhart et al., 1991; Fujiwara et al., 2018). It is therefore warranted to conduct several types of toxicity studies for the drug candidates using multiple animal species in order to better understand the species difference considering its toxicity profile and its relevance in humans during drug development.
To predict the potential genotoxicity and carcinogenicity of the drug candidates, 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). In the in vivo micronucleus study, rats and mice are generally used as the standard species. Furthermore, in the in vitro cytogenetic test, the rat liver 9000 × g supernatant fraction (S9) pretreated with drug-metabolizing enzyme inducers, is widely applied as a metabolic activation system because cultured mammalian cells used in the test system have a limited capacity for metabolizing drugs (Ames et al., 1973; Paolini and Cantelli-Forti, 1997). Although rodents are the gold-standard species for carcinogenicity studies, the toxicity findings such as proliferative lesions, suggestive of potential carcinogenicity, are occasionally observed in other species including dogs and non-human primates employed in the long-term toxicity studies. Thus, to adequately assess the potential carcinogenicity of the drug candidates in humans, it is essential for investigators to understand whether the species diversity in the metabolic profile of the drug candidates affects the results of in vitro and in vivo genotoxicity studies in addition to the carcinogenicity studies.
To address this issue, we focused on the metabolic profiles of the clastogenic compounds in several species including human. We selected seven clastogenic compounds for in vitro micronucleus tests (Ishidate et al., 1988; NTP, 2017; Brambilla and Martelli, 2009) which are metabolized several types of cytochrome P450 (CYP); 2 carcinogens, benzo[a]pyrene (BaP) by CYP1A1 (McManus et al., 1990) and cyclophosphamide monohydrate (CPA) by CYP2B6, 2C9, 2C19, and 3A4/5 (Moore, 1991; Ekhart et al., 2008) and 5 marketed drugs, coumarin by CYP2A6 (Born et al., 2002; Yamano et al., 1990), diclofenac sodium and piroxicam by CYP2C9 (Hobbs and Twomey, 1981; Yan et al., 2005), lansoprazole by CYP2C19 (Pearce et al., 1996), and chlorpheniramine maleate by CYP2D6 (Yasuda et al., 2002) with rat, mouse, monkey, dog, or human liver S9. Furthermore, we identified the metabolites of BaP and CPA generated in each species, which were attributable to the positive results in the in vitro micronucleus test.
Rat pooled liver S9, which was prepared from the liver of three male Sprague-Dawley rats (7-week-old), without any treatment, was purchased from Oriental Yeast Co., Ltd. (Tokyo, Japan). Human pooled liver S9, which was prepared from the livers of 50 donors (donated for application in research), was purchased from Becton Dickinson and Company (Franklin Lakes, NJ, USA). The experiments for the preparation of mouse, monkey, and dog liver S9 were approved by the Ethics Review Committee for Animal Experimentation of Daiichi Sankyo Co., Ltd., and were conducted in compliance with the “Law Concerning the Protection and Control of Animals”, Japanese Law No. 105, October 1, 1973, revised on June 22, 2005.
C3H/HeNCrlCrlj mice, cynomolgus monkeys, and Narc: beagle dogs were purchased from Charles River Laboratory Japan, Inc. (Kanagawa, Japan), Hamri Co. Ltd. (Ibaraki, Japan), and Oriental Yeast Co., Ltd., respectively. Liver samples were excised under anesthetized conditions from 20 male C3H/HeNCrlCrlj mice (7-week-old), 1 male and 1 female cynomolgus monkeys (3- and 5-year old, respectively), and 1 male and 1 female beagle dogs (3- and 6-year old, respectively). They were immediately frozen with liquid nitrogen, and thereafter, the frozen liver samples were thawed and homogenized in 1.15 w/v% potassium chloride solution on ice. The homogenate was centrifuged at 9000 × g for 20 min at 4°C by a centrifuge. The supernatant fraction in each species was mixed and pooled. Protein concentrations of each S9 used in the present study were fixed at 20 mg/mL.
BaP was purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). CPA, coumarin, diclofenac sodium (diclofenac), piroxicam, lansoprazole, and chlorpheniramine maleate (chlorpheniramine) were purchased from Sigma-Aldrich Japan K.K. (Tokyo, Japan). BaP, CPA, coumarin, piroxicam, and lansoprazole were dissolved in dimethyl sulfoxide (Wako Pure Chemical Industries, Ltd.). Diclofenac and chlorpheniramine were dissolved in water for injection (Otsuka Pharmaceutical Factory, Inc., Tokushima, Japan).
Chinese hamster lung fibroblast cells (RIKEN BioResource Center, Ibaraki, Japan) were seeded in plastic chamber slides at a density of 5 × 103 cells/chamber. After 24 hr of incubation, the cells were treated with various concentrations of each test compound containing 5% liver S9 from all species for 6 hr, followed by 18 hr of recovery time in the plastic chamber slides (OECD, 2014). The concentrations tested in the micronucleus test ranged from 12.4 to 300 μg/mL for BaP, 0.348 to 50 μg/mL for CPA, 206 to 1200 μg/mL for coumarin, 76.5 to 300 μg/mL for diclofenac, 311 to 1500 μg/mL for piroxicam, 41.4 to 200 μg/mL for lansoprazole, and 120 to 700 μg/mL for chlorpheniramine. These concentrations were set based on the preliminary dose-range finding test and our previous investigation (Kishino et al., 2019a). After the treatment, the cytotoxicity of each test compound was evaluated with Cell Counting Kit-8 (Dojindo laboratories, Kumamoto, Japan) (Ishiyama et al., 1997; Tominaga et al., 1999). The relative number of viable cells (%) was calculated by subtracting mean of the absorbance at 450–650 nm in the treatment group from that in the control group. After the evaluation for cytotoxicity, the cells were treated with 0.075 M KCl for 5 min at 37°C and fixed with 6% acetic acid/methanol solution on ice. After fixation, the slides were stained with 40 μg/mL acridine orange solution and observed under 200-fold magnification. One thousand cells per chamber were examined and the number of micronucleated cells per chamber was counted and their mean values were calculated as previously reported (Kishino et al., 2019a).
Since BaP and CPA presented remarkable species differences in the micronucleus induction, metabolites of BaP and CPA attributable to the positive results were identified under the same condition as that of the in vitro micronucleus test. Briefly, the aliquots of liver S9 from each species were treated with BaP and CPA at 50 μg/mL, respectively, for 6 hr. Furthermore, the mixtures were deproteinized with three-fold volume of methanol. The metabolites generated in the mixtures were measured by an ultra-high performance liquid chromatography time-of-flight mass spectrometer (UPLC/TOFMS, ACQUITY UPLC/LCT Premier; Waters Corporation, Milford, MA, USA). The mixtures were loaded into an ACQUITY UPLC BEH C18 column (100 mm × 2.1 mm, 1.7 µm, Waters Corporation). Two mobile phases were used (distilled water/formic acid mixed at a ratio of 1000/1 [v/v] and methanol/formic acid mixed at a ratio of 1000/1 [v/v]) in a linear gradient. The full scan method was used (mass to charge ratio; m/z range from 100 to 1000). The raw data of retention time, m/z, and ion intensity of the detectable metabolites obtained from the mass spectrum were analyzed with MZmine 2 (Pluskal et al., 2010), which is an open-source software to process the liquid chromatography-mass spectrometry (LC-MS) data sets. Each metabolite of BaP and CPA was estimated based on its m/z and retention time and was further determined based on its accurate mass and constituent elements. In addition, the CPA metabolites were also identified by an LC-MS/MS system (Waters 2795 Separations Module/Quattro micro; Waters Corporation) as described previously (Kishino et al., 2019b).
The in vitro micronucleus test with liver S9 of each species was conducted in duplicate assays, wherein the mean of incidence of micronucleated cells was statistically analyzed by the two-tailed Fisher’s exact test. The dose-dependency was also evaluated with the Cochran–Armitage trend test (Matsushima et al., 1999). As for metabolite analysis, the ion intensities of the metabolites were expressed as the mean ± standard deviation (SD) in triplicate assays. They were compared using the Turkey’s multiple comparison 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 as statistically significant. The relationship between the amount of metabolites and the number of micronucleated cells was evaluated to determine the correlation coefficient by regression analysis with Microsoft Excel 2013 add-ins (Microsoft Corporation, Redmond WA, USA).
Among the seven compounds used in the micronucleus test, BaP, CPA, coumarin, diclofenac, piroxicam, and lansoprazole statistically increased the number of the micronucleated cells. Among these clastogens, BaP and CPA showed notable species differences in the micronucleus induction (Fig. 1). BaP dose-dependently induced the micronucleus when the cells were treated with rat, mouse, monkey, or human liver S9, and the monkey liver S9 was particularly susceptible to BaP. In contrast, BaP with dog liver S9 did not induce micronucleus. Furthermore, CPA dose-dependently increased the number of micronucleated cells when the cells were treated with rat, mouse, money, and dog liver S9s, but not human liver S9. In contrast, BaP and CPA did not reveal any cytotoxicity up to the maximum concentration tested (Fig. 1). Coumarin, diclofenac, piroxicam, and lansoprazole also increased the number of micronucleated cells with all the S9s, whereas chlorpheniramine did not induce the micronucleus formation with any species of S9s, even at the concentrations in which the relative number of viable cells was approximately 50% (Figs. 2 and 3).
In vitro micronucleus induction and cytotoxicity by benzo[a]pyrene and cyclophosphamide with rat, mouse, monkey, dog, and human liver S9. A; in vitro micronucleus induction, B; cytotoxicity, Open circle; rat liver S9, Open triangle; mouse liver S9, Open square; monkey liver S9, Closed circle; dog liver S9, Closed triangle; human liver S9. *P < 0.01: Significantly different from the control by two-tailed Fishers’ exact test. BaP: benzo[a]pyrene, CPA: cyclophosphamide.
In vitro micronucleus induction by coumarin, diclofenac, piroxicam, and lansoprazole with rat, mouse, monkey, dog, and human liver S9. Open circle; rat liver S9, Open triangle; mouse liver S9, Open square; monkey liver S9, Closed circle; dog liver S9, Closed triangle; human liver S9. *P < 0.01: Significantly different from the control by two-tailed Fishers’ exact test.
Cytotoxicity by coumarin, diclofenac, piroxicam, and lansoprazole with rat, mouse, monkey, dog, and human liver S9. Open circle; rat liver S9, Open triangle; mouse liver S9, Open square; monkey liver S9, Closed circle; dog liver S9, Closed triangle; human liver S9.
In the medium containing BaP with liver S9 from each species, six metabolites such as BaP-epoxide, 9-hydroxy-BaP (9-OH-BaP), BaP-7,8-dihydrodiol, and hydroxy-BaP-epoxide (OH-BaP-epoxide) 1–3 were determined as major metabolites based on their accurate mass and constitute elements (Fig. 4). The BaP-epoxide, 9-OH-BaP, BaP-7,8-dihydrodiol, and OH-BaP-epoxides were most abundant in the BaP medium with monkey liver S9, whereas they were extremely scarce in the BaP medium with human liver S9. A close correlation was observed between the number of micronucleated cells and the amount of OH-BaP-epoxides generated in the BaP medium with liver S9s from all the species (Fig. 5).
Metabolic profile of benzo[a]pyrene in the assay system with rat, mouse, monkey, dog, and human liver S9. **P < 0.01: Significantly different from the rat liver S9; †P < 0.05, ††P < 0.01: Significantly different from the mouse liver S9; ‡P < 0.05, ‡‡P < 0.01: Significantly different from the monkey liver S9; §§P < 0.01: Significantly different from the dog liver S9 by Tukey’s multiple comparison test. BaP: benzo[a]pyrene, OH-BaP: hydroxy-benzo[a]pyrene.
Relationship between the ion intensity of hydroxy-benzo[a]pyrene-epoxides and micronucleus induction by benzo[a]pyrene in rat, mouse, monkey, dog, and human S9. Open circle; OH-BaP-epoxide 1, Open triangle; OH-BaP-epoxide 2, Open square; OH-BaP-epoxide 3 OH-BaP: hydroxy-benzo[a]pyrene.
As the major metabolites of CPA, 4-hydroxyl-CPA (4-OH-CPA), carboxyphosphamide, 4-keto-CPA, and phosphamide mustard were detected in the CPA medium with all the liver S9s (Fig. 6). The 4-OH-CPA and phosphamide mustard were most abundant in the CPA medium with dog liver S9. The CPA content was specifically decreased in the CPA medium with monkey liver S9 compared to that in the control medium (Fig. 6), reflecting the production of tremendous carboxyhphosphamide and 4-keto-CPA.
Metabolic profile of cyclophosphamide in the assay system with rat, mouse, monkey, dog, and human liver S9. *P < 0.05, **P < 0.01: Significantly different from the rat liver S9; †P < 0.05, ††P < 0.01: Significantly different from the mouse liver S9; ‡P < 0.05, ‡‡P < 0.01: Significantly different from the monkey liver S9; §§P < 0.01: Significantly different from the dog liver S9 by Tukey’s multiple comparison test. CPA: cyclophosphamide, 4-OH-CPA: 4-hydroxyl-cyclophosphamide.
In the present study, we evaluated the species differences in the micronucleus induction of seven clastogens using liver S9 obtained from five animal species including humans. Among them, six clastogens (BaP, CPA, coumarin, diclofenac, piroxicam, and lansoprazole) induced the micronucleus formation with all the S9s. Interestingly, BaP and CPA revealed the remarkable species diversity in the micronucleus induction, whereas coumarin, diclofenac, piroxicam, and lansoprazole did not present any species differences. We had previously reported that micronucleus induction by CPA was attenuated at higher concentrations using rat liver S9 which might be related to its cytotoxicity (Kishino et al., 2019a). However, in the present study, similar attenuation of micronucleus induction by CPA at higher concentrations was observed using rat and dog liver S9s without any cytotoxicity. The reason for the discrepancy remains unclear. BaP is metabolized into BaP-epoxide mediated by cytochrome P450 (CYP) 1A1, followed by conversion to BaP-7,8-dihydrodiol or 9-OH-BaP, and further metabolized to anti-BaP-7,8-dihydrodiol-9,10-epoxide or hydroxy-BaP-4,5-epoxide. These epoxides react with DNA to form the DNA adducts (Hodek et al., 2013; Baird et al., 2005). In addition, CYP1A2 oxidizes BaP to BaP-epoxide, thus forming DNA adduct in the rat liver (Hodek et al., 2013). In metabolite analysis of BaP, OH-BaP-epoxides were formed in all the S9s, and were the most abundant in monkey liver S9 and most scarce in human liver S9. Furthermore, a close correlation was observed between the amount of OH-BaP-epoxides and micronucleus induction, suggesting that OH-BaP-epoxides were major determinants of micronucleus induction in all the species. In the present study, anti-BaP-7,8-dihydrodiol-9,10-epoxide, which leads to the formation of the aforementioned DNA adduct, was not detected. In addition, the portion of hydroxylation or epoxidation of OH-BaP-epoxides was not determined. In contrast, Hodek et al. (2013) suggested that BaP-DNA adducts other than anti-BaP-7,8-dihydrodiol-9,10-epoxide or 9-hydroxy-BaP-4,5-epoxide were associated with BaP-induced carcinogenicity; however, the impacts of OH-BaP-epoxides on micronucleus formation in the present study remain to be explored.
CPA is metabolized into 2-dechloroethylcyclophosphamide and chloroacetaldehyde by CYP3A4/5, and into 4-OH-CPA by CYP2B6, CYP2C9, CYP2C19, and CYP3A4/5 in different pathways (Moore, 1991; Ekhart et al., 2008). Phosphamide mustard directly interacts with DNA and forms cross-linked structures (Moore, 1991), leading to gene mutation with a high frequency (Anderson et al., 1995). CPA is also reported to be metabolized into the same metabolites in rats and mice via the same pathway as in humans (Chen et al., 2004; Ramirez, et al., 2019). In the present study, the micronucleus induction and formation of genotoxic-metabolites due to CPA were largest with the dog liver S9. In dogs, CPA is reported to be metabolized into 4-OH-CPA by CYP2B11, which is an ortholog of CYP2B6 in human or CYP2B1 in rats (Heikkinen et al., 2012; Ramirez et al., 2019). In addition, the 4-hydroxylation activity of CYP2B11 in dogs is suggested to be higher than that of CYP2B6 in humans and CYP2B1 in rats (Heikkinen et al., 2012). High content of phosphamide mustard and 4-OH-CPA and greater micronucleus induction observed in the present study are therefore, at least in part, due to the difference in CYP isozyme activity of each species. In contrast, the formation of genotoxic-metabolites and micronucleus induction caused by CPA were smallest with the human liver S9 among the five species. It is well-documented that a large inter-individual variability exists in the activity of the human liver S9 fraction, resulting in distinct results of the in vitro genotoxicity tests (Hakura et al., 2003). We previously reported that a wide inter-individual variability of human liver microsomal fractions was correlated with micronucleus induction due to CPA (Kishino et al., 2019b). Collectively, the enzyme activities of human liver S9 employed in the present study might be lower than those of the aforementioned reports (Kishino et al., 2019b; Hakura et al., 2003).
It has been reported that species diversity exists in the amount of the metabolites generated and its formation rate of coumarin (Lake, 1999; Rietjens et al., 2008), diclofenac (Tang et al., 1999; John, 1979), and piroxicam (Klopas et al., 1998; Hobbs and Twomey, 1981); however, no apparent differences in micronucleus induction for these three clastogens were noted among the five species in the present study, suggesting that there was no difference in the formation of the metabolites associated with micronucleus induction. For lansoprazole, our result indicated that there were no apparent species differences in the formation of genotoxic metabolite among the species. Although coumarin, diclofenac, piroxicam, and lansoprazole induced micronucleus in the present study, consistent with our previous in vitro genotoxicity assay with Chinese hamster lung fibroblast cells (Kishino et al., 2019a), this assay platform is known to produce false positive results of several clastogens with high incidence (Kirkland et al., 2007; Kishino et al., 2019a). Therefore, it is important to adequately understand the metabolism of drug candidate and use the optimum assay platform to avoid an irrelevant outcome in the in vitro micronucleus test. In fact, diclofenac and lansoprazole were negative in the Ames test (Brambilla and Martelli, 2009). In addition, diclofenac and piroxicam did not induce sister-chromatid exchange in human lymphocytes (Kullich and Klein, 1986). Furthermore, coumarin, diclofenac, piroxicam, and lansoprazole did not show genotoxicity in the in vivo studies (Api, 2001; Brambilla and Martelli, 2009), and therefore, the genotoxic risk of these marketed drugs in humans is considered to be low. Chlorpheniramine was reported to induce micronucleus formation when the in vitro micronucleus test was conducted with rat liver S9 pretreated with phenobarbital (PB) and 5,6-benzoflavone (BNF), but not non-treated rat liver S9 (Kishino et al., 2019a). In addition, this positive result was due to the differences in the amount of genotoxic metabolites, which were produced by several CYP isozymes induced by PB and BNF (Kishino et al., 2019a). However, chlorpheniramine did not induce micronucleus and carcinogenicity in rodents (NTP, 2017), indicating that the positive result with rat liver S9 pretreated with PB and BNF was not relevant to humans.
In conclusion, the present study demonstrated that the species diversity in the micronucleus induction caused by BaP and CPA was attributable to the differences in the metabolic profile of these drugs among the species. In cases of BaP and CPA, liver S9 from non-rodent species can detect the potential genotoxicity with high sensitivity and could be a useful tool for detecting the genotoxic and carcinogenic risk. In contrast, it is crucial to understand the species differences in the metabolic profile and its relevance in humans because false positive results might be obtained in the in vitro genotoxicity study or in vivo long-term toxicity studies using sensitive species, which was not human relevant.
The authors thank Kyoko Watanabe for technical assistance with the metabolite analysis.
The authors declare that there is no conflict of interest.