**2. Advantages of HepG2 transformants expressing a series of human CYP isoforms in the in vitro genotoxicity testing**

The need for metabolism, especially CYP-mediated one, for in vitro genotoxicity testing has been recognized for many years. Most target cells for genotoxicity assays lack sufficient CYP to activate many promutagens. Therefore, extracellular systems are commonly utilized to provide metabolism. The rat liver S9 fraction contains multiple CYPs and have been used with many target cell types in genotoxicity testing. However, this metabolic activation system suffers from certain limitations; (1) generation of reactive metabolites outside of the

more suitable for the genotoxic assessment of chemicals than the induced rat liver S9 fraction in the routine screening when considering human hepatic metabolism in the future. Therefore in the present thesis, we explored the usefulness of a series of 10 transformants expressing major human CYP isoforms such as CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 in HepG2 cells established previously to assess the genotoxicity of

reaction measured Transformant

*K*m (μM)

Human liver microsomes *K*m (μM)a

0.25 0.19 56

0.72 0.39 2

5.1 2.3 812 000

81 – 80 000

7.4 24 9400

45 120 25 000

8.3 16 140 000

17 40 14

88 30 120

96 89 71

Transformant *V*max

(ρmol/min/mg)

metabolites (Fig.3) (Hashizume et al., 2009; Hashizume et al., 2010).

*O*-deethylation

*O*-deethylation

hydroxylation

*O*-deethylation

hydroxylation

hydroxylation

hydroxylation

hydroxylation

hydroxylation

hydroxylation

Table 2. Characteristics of a series of 10 transformants expressing major human CYP isoforms such as CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 in HepG2 cells

**2. Advantages of HepG2 transformants expressing a series of human CYP** 

The need for metabolism, especially CYP-mediated one, for in vitro genotoxicity testing has been recognized for many years. Most target cells for genotoxicity assays lack sufficient CYP to activate many promutagens. Therefore, extracellular systems are commonly utilized to provide metabolism. The rat liver S9 fraction contains multiple CYPs and have been used with many target cell types in genotoxicity testing. However, this metabolic activation system suffers from certain limitations; (1) generation of reactive metabolites outside of the

Name of Expressed Catalytic Kinetic analysis

transformant CYP

isoform

Hepc/1A1.4 CYP1A1 7-Ethoxyresorufin

Hepc/1A2.9 CYP1A2 7-Ethoxyresorufin

Hepc/2B6.68 CYP2B6 7-Ethoxycoumarin

Hepc/2C9.1 CYP2C9 Tolbutamide 4-

Hepc/2D6.39 CYP2D6 Bufuralol 1'-

Hepc/2E1.3-8 CYP2E1 p-Nitrophenol

Hepc/3A4.2-30 CYP3A4 Testosterone 6β-

(Yoshitomi et al., 2001). a). Iwata et al., 1998.

**isoforms in the in vitro genotoxicity testing** 

Hepc/2C19.12 CYP2C19 (S)-Mephenytoin 4'-

Hepc/2A6L.14 CYP2A6 Coumarin 7-

Hepc/2C8.46 CYP2C8 Taxol 6-

target cell, (2) requirement of high exposure concentration to compensate for short exposure times and (3) differences in metabolism compared to intact tissues. To overcome these limitations, the use of genetically engineered stable cell lines expressing CYPs has studied. The liver is the tissue containing the greatest concentrations of drug-metabolizing enzymes, such as the CYP enzyme family, among many others. In human liver, about 70% of the total CYP could be accounted for by CYP1A2, 2A6, 2B6, 2C, 2D6, 2E1 and 3A proteins (Rendic and Guengerich, 1997). In the extrahepatic organs such as lungs and kidneys, CYP1A1 is present. CYPs catalyze to form toxic reactive intermediates from many chemicals. As it is well known that there are significant quantitative and qualitative differences between laboratory animals and humans in their CYP subtypes, it is necessary to use human CYP isoforms to predict the metabolism and toxicity of chemicals in humans.

Fig. 3. HepG2 transformants expressing human CYP isoforms relating drug metabolism. The pie chart shows the contribution of each CYP isoform to the human drug metabolism (Lewis, 2004). It has been concerned about the low CYP activities in HepG2 cells, so we established the HepG2 transformant system expressing a series of human CYP isoforms.

In vitro systems, particularly those derived from liver, are a commonly applied tool to gain a better understanding of the metabolism of drugs and other xenobiotics. Also in genotoxicity, a number of publications are discussed which are relevant for the use of human derived liver cell lines. One of the most promising lines is the human HepG2 cell line, originally isolated by Aden et al. in 1972 from a primary hepatoblastoma of an 11-yearold Argentine boy. This cell line retains many of the specialized functions normally lost by primary hepatocytes in culture such as secretion of the major plasma proteins. Since several publications alerted that HepG2 lacks a few drug-metabolizing enzymes such as CYP2E1 and 1A2, transfectants constitutively expressing these enzymes have been constructed. Cederbaum and coworkers developed a line, which possesses CYP2E1 activity and used it in a number of mechanistic studies (for review see Kessova and Cederbaum, 2003).

Application of a New Genotoxicity Test System with

demonstrated in the present study (Fig. 4 D).

A

B

C

D

4-hydroxy ifosfamide.

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 369

capable of metabolizing DMBA while CYP2D6 and 3A4 exhibited relatively low metabolic activity to DMBA (Shou et al., 1996). Similarly, significant MN inductions by ifosfamide were found in the transformants expressing CYP1A1, 2C9, 2C19, 2D6 and 3A4 (Fig. 4 D). Ifosfamide is mainly metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006). CYP1A1 and 2C19 are relatively minor CYP isoforms to DMBA metabolic activation, but the involvement in the genotoxicity of ifosfamide metabolite of these CYP isoforms were

Fig. 4. Metabolic activation pathway of BP (A), DMBA (B), CP (C) and ifosfamide (D). A). BP is mainly metabolized by CYP1A1 to produce the benzo(a)pyrene 7, 8-oxide. B). DMBA (7, 12- DMBA) is mainly metabolized by CYP1A1 to produce the 7, 12-DMBA-3, 4-oxide. C).

Cyclophosphamide (CP) is mainly metabolized by CYP2C9 and 3A4 to produce the 4-hydroxy cyclophosphamide. D). Ifosfamide is mainly metabolized by CYP2B6 and 3A4 to produce the

In our previous study, Yoshitomi et al. has established a series of HepG2 transformants expressing the CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 (Yoshitomi et al., 2001). Since most human drug metabolism is catalyzed by CYP1A2, 2C8, 2C9, 2C19, 2D6, 2E1 and P3A4, this HepG2 transformant system would be more suitable for the genotoxic assessment of chemicals than the induced rat liver S9 fraction in the routine screening when considering human hepatic metabolism in the future.

Therefore, we examined the advantages of HepG2 transformants expressing a series of human CYP isoforms as a better alternative for metabolic activation system in the in vitro genotoxicity testing. In section 2.1, the sensitivity of this system to detect genotoxicity requiring CYP activation was confirmed in the in vitro micronucleus (MN) tests using wellstudied model chemicals. In section 2.2, this system allowed us to investigate the genotoxicity of model chemicals for which the contributing CYP isoforms, especially those mediated by CYP1A2 or 3A4 which is known to metabolize many drugs in humans, have not yet been identified. In section 2.3, the relevance of the interaction between phase I and phase II drug-metabolizing enzymes, e.g., UGT, GST, and SULT, in the test system was demonstrated in a MN test of tamoxifen or safrole, which has been reported to be metabolized by enzymes of both phases.

#### **2.1 Basic characteristics of the HepG2 transformants on genotoxic assessment and confirmation of their sensitivity with model chemicals requiring CYP activation**

HepG2 transformants were checked for their response to known chemicals in which the CYP isoforms responsible for which genotoxicity has been reported. As model chemicals, we selected BP, DMBA, CP and ifosfamide. In BP metabolism, CYP1A1 showed clearly the highest activity among the hepatic CYP isoforms reported (Fig. 4 A). Significant formation of some metabolites was also observed with CYP1A2 and 3A4 (Bauer et al., 1995). In the DMBA metabolism (Fig. 4 B), it had been shown that CYP1A1 had clearly the highest activity among the hepatic CYP isoforms (Shimada and Fujii-Kuriyama, 2004; Shou et al., 1996) and that significant formation of some metabolites was also observed with CYP1A2, 2B6 and 2C9 (Shou et al., 1996). CP is efficiently metabolized by CYP2B6, 2C9 and 3A4 (Jing et al., 2006, Chang et al., 1993)(Fig. 4 C). Ifosfamide had been demonstrated to be efficiently metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006)(Fig. 4 D).

Firstly, BP treatment produced MN induction in the transformants expressing CYP1A1, CYP1A2 and CYP3A4 (Fig. 5 A). These CYP isoforms were reported to be responsible for BP activation (Bauer et al., 1995). Secondly, MN induction by DMBA in a HepG2 transformant expressing CYP1A1 was significantly higher than those in HepG2 and Hepc-Mock cells (Fig. 5 B). CYP1A1 is known to be the most active among the CYP isoforms to metabolize DMBA (Shou et al., 1996). Thirdly, CP treatment caused MN induction in the transformants expressing CYP1A2, 2B6, 2C9 and 3A4 (Fig. 5 C). CYP2B6, 2C9 and 3A4 are reported to be involved the metabolic activation of CP (Jing et al., 2006; Chang et al., 1993). Finally, in the treatment with ifosfamide, significant MN inductions were found in the transformants expressing CYP1A1, 2C9, 2C19, 2D6 and 3A4 (Fig. 5 D). Ifosfamide had been demonstrated to be efficiently metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006). These results showed HepG2 transformants system have the appropriate sensitivity to detect genotoxicity requiring CYP activation tests using well-studied model chemicals.

In addition, DMBA treatment unexpectedly produced MN induction in some transformants expressing CYP2C9, 2D6 and 3A4 (Fig. 4 B). However, it was reported that CYP2C9 was

In our previous study, Yoshitomi et al. has established a series of HepG2 transformants expressing the CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 (Yoshitomi et al., 2001). Since most human drug metabolism is catalyzed by CYP1A2, 2C8, 2C9, 2C19, 2D6, 2E1 and P3A4, this HepG2 transformant system would be more suitable for the genotoxic assessment of chemicals than the induced rat liver S9 fraction in the routine screening when

Therefore, we examined the advantages of HepG2 transformants expressing a series of human CYP isoforms as a better alternative for metabolic activation system in the in vitro genotoxicity testing. In section 2.1, the sensitivity of this system to detect genotoxicity requiring CYP activation was confirmed in the in vitro micronucleus (MN) tests using wellstudied model chemicals. In section 2.2, this system allowed us to investigate the genotoxicity of model chemicals for which the contributing CYP isoforms, especially those mediated by CYP1A2 or 3A4 which is known to metabolize many drugs in humans, have not yet been identified. In section 2.3, the relevance of the interaction between phase I and phase II drug-metabolizing enzymes, e.g., UGT, GST, and SULT, in the test system was demonstrated in a MN test of tamoxifen or safrole, which has been reported to be

**2.1 Basic characteristics of the HepG2 transformants on genotoxic assessment and confirmation of their sensitivity with model chemicals requiring CYP activation** 

metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006)(Fig. 4 D).

genotoxicity requiring CYP activation tests using well-studied model chemicals.

In addition, DMBA treatment unexpectedly produced MN induction in some transformants expressing CYP2C9, 2D6 and 3A4 (Fig. 4 B). However, it was reported that CYP2C9 was

HepG2 transformants were checked for their response to known chemicals in which the CYP isoforms responsible for which genotoxicity has been reported. As model chemicals, we selected BP, DMBA, CP and ifosfamide. In BP metabolism, CYP1A1 showed clearly the highest activity among the hepatic CYP isoforms reported (Fig. 4 A). Significant formation of some metabolites was also observed with CYP1A2 and 3A4 (Bauer et al., 1995). In the DMBA metabolism (Fig. 4 B), it had been shown that CYP1A1 had clearly the highest activity among the hepatic CYP isoforms (Shimada and Fujii-Kuriyama, 2004; Shou et al., 1996) and that significant formation of some metabolites was also observed with CYP1A2, 2B6 and 2C9 (Shou et al., 1996). CP is efficiently metabolized by CYP2B6, 2C9 and 3A4 (Jing et al., 2006, Chang et al., 1993)(Fig. 4 C). Ifosfamide had been demonstrated to be efficiently

Firstly, BP treatment produced MN induction in the transformants expressing CYP1A1, CYP1A2 and CYP3A4 (Fig. 5 A). These CYP isoforms were reported to be responsible for BP activation (Bauer et al., 1995). Secondly, MN induction by DMBA in a HepG2 transformant expressing CYP1A1 was significantly higher than those in HepG2 and Hepc-Mock cells (Fig. 5 B). CYP1A1 is known to be the most active among the CYP isoforms to metabolize DMBA (Shou et al., 1996). Thirdly, CP treatment caused MN induction in the transformants expressing CYP1A2, 2B6, 2C9 and 3A4 (Fig. 5 C). CYP2B6, 2C9 and 3A4 are reported to be involved the metabolic activation of CP (Jing et al., 2006; Chang et al., 1993). Finally, in the treatment with ifosfamide, significant MN inductions were found in the transformants expressing CYP1A1, 2C9, 2C19, 2D6 and 3A4 (Fig. 5 D). Ifosfamide had been demonstrated to be efficiently metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006). These results showed HepG2 transformants system have the appropriate sensitivity to detect

considering human hepatic metabolism in the future.

metabolized by enzymes of both phases.

capable of metabolizing DMBA while CYP2D6 and 3A4 exhibited relatively low metabolic activity to DMBA (Shou et al., 1996). Similarly, significant MN inductions by ifosfamide were found in the transformants expressing CYP1A1, 2C9, 2C19, 2D6 and 3A4 (Fig. 4 D). Ifosfamide is mainly metabolized by CYP2B6 and 3A4 (Chang et al., 1993; Jing et al., 2006). CYP1A1 and 2C19 are relatively minor CYP isoforms to DMBA metabolic activation, but the involvement in the genotoxicity of ifosfamide metabolite of these CYP isoforms were demonstrated in the present study (Fig. 4 D).

Fig. 4. Metabolic activation pathway of BP (A), DMBA (B), CP (C) and ifosfamide (D). A). BP is mainly metabolized by CYP1A1 to produce the benzo(a)pyrene 7, 8-oxide. B). DMBA (7, 12- DMBA) is mainly metabolized by CYP1A1 to produce the 7, 12-DMBA-3, 4-oxide. C). Cyclophosphamide (CP) is mainly metabolized by CYP2C9 and 3A4 to produce the 4-hydroxy cyclophosphamide. D). Ifosfamide is mainly metabolized by CYP2B6 and 3A4 to produce the 4-hydroxy ifosfamide.

Application of a New Genotoxicity Test System with

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 371

2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. Seven CYP isoforms (1A1, 1A2, 2C9, 2C19, 2D6, 2E1 and 3A4) account for 95% of this activity and with 3A4 responsible for over 65% of the metabolism of current therapeutic agents. During phase I metabolism in humans, 90% of all drugs are oxidized by CYP isoforms with different substrate selectivities. Thus, our HepG2 transformant system seems to cover most human drug metabolism mediated by human CYP isoforms but not by rat ones. Based on the results obtained in section 2.1, it was demonstrated that our HepG2 transformant system has an appropriate sensitivity for well-

For assessing the genotoxicity of chemicals with human metabolism, our HepG2 transformant system has more appropriate characteristics than other established cell lines used in toxicological testing and reviewed by Sawada and Kamataki (Sawada and Kamataki, 1998). Our HepG2 transformants are derived from hepatocyte that possesses other factors necessary for the function of CYP. Generally, the reactions catalyzed by CYP molecules require the presence of NADPH-CYP reductase and cytochrome b5 to support some CYP-mediated reactions. HepG2 has been shown to have NADPH-CYP reductase activity and cytochrome b5, although the levels are lower than those of human liver (Waxman et al., 1991; Patten et al., 1992). Therefore, our HepG2 transformant system does

studied chemicals which requires CYP activation for their genotoxicity.

not need co-expression of reductase and/or cytochrome b5 with CYP enzymes.

**2.2 An exploration using HepG2 transformant to identify the CYP isoforms** 

Given the multiplicity of CYP isoforms and the importance of other enzymes (hydrolases, transferase, etc.) in the metabolism of chemicals, there are two possible approaches to engineering cell lines. The introduction of single enzymes allows simple controlled mechanistic studies of the role of an individual enzyme in the metabolic activation of chemicals. Such system can also be viewed as the replacement of the laborious CYP purification/reconstitution analyses with a panel of engineered cells. However, to specify CYP isoform(s) involved in the activation of a certain chemical of unknown metabolism, a set of cell lines individually expressing the different CYP isoforms is needed. A series of HepG2 transformants expressing major 10 human CYP isoforms is a valuable tool, since most human hepatic drug metabolism is catalyzed by these expressing CYP isoforms.

Fig. 6. Chemical structure and significant biological features of okadaic acid. This chemical is a shellfish poisoning toxin and known as a potent phosphatase 1 and 2A. This chemical was reported to be induced micronucleus in the presence of rat liver S9 mix (Hégarat et al., 2004). In this section, the following possibility was elucidated that the set of transformants can be used for screening for the genotoxicity of newly developed pharmaceutical candidates of unknown metabolism in human in vivo. As model chemicals, we selected okadaic acid (OA)

**contributing to the genotoxicity by novel chemicals** 

and β-endosulfan.

Fig. 5. Micronucleus induction of BP (A), DMBA (B), CP (C) and ifosfamide (D) by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. A). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 200 ng/ml (light gray bars) or 400 ng/ml (medium gray bars) BP. B). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 78 ng/ml (light gray bars) or 156 ng/ml (medium bars) DMBA. C). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% saline (heavy gray bars), 1 mg/ml (light gray bars) or 2 mg/ml (medium gray bars) CP. D). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% saline (heavy gray bars), 500 μg/ml (light gray bars) or 1000 μg/ml (medium gray bars) ifosfamide. Values were normalized with the mean DMSO- or saline-treated control value of 3 experiments for each transformant. Each bar represents the mean ± S.D. Data were tested using Student's t-test when the variance was homogeneous or Aspin & Welch t test when the variance was heterogeneous (\*P<0.05, compared with Hepc-Mock).

Mock

Based on the results in section 2.1, it was showed that genotoxic metabolites could be produced by not only the most active CYP isoform but also by other less active CYPs and that this transformant system could detect the genotoxic potential of chemicals requiring CYP activation not tested routinely in the early stage of drug development.

One of major advantages of our system is the variety of human CYP isoforms. When considering replacement of the rat induced liver S9 fraction, increasing the number of the principal CYP isoforms would be desirable in order to cover the diverse CYP activities. As mentioned in the general introduction, our HepG2 system includes CYP1A1, 1A2, 2A6, 2B6,

Mock

A B

0.0

8.0

C D

0.0

HepG2 Hepc-Mock

4.0

**Fold induction to each control group**

HepG2 Hepc-Mock 1A1 1A2 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4

CYP1A1 CYP1A2 CYP2A6 CYP2B6 CYP2C8 CYP2C9 CYP2C19 CYP2D6 CYP2E1 CYP3A4

 

4.0

**Fold induction to each control group**

8.0

\*

\*

Fig. 5. Micronucleus induction of BP (A), DMBA (B), CP (C) and ifosfamide (D) by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. A). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 200 ng/ml (light gray bars) or 400 ng/ml (medium gray bars) BP. B). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 78 ng/ml (light gray bars) or 156 ng/ml (medium bars) DMBA. C). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% saline (heavy gray bars), 1 mg/ml (light gray bars) or 2 mg/ml (medium gray bars) CP. D). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% saline (heavy gray bars), 500 μg/ml (light gray bars) or 1000 μg/ml (medium gray bars) ifosfamide. Values were normalized with the mean DMSO- or saline-treated control value of 3 experiments for each transformant. Each bar represents the mean ± S.D. Data were tested using Student's t-test when the variance was homogeneous or Aspin & Welch t

test when the variance was heterogeneous (\*P<0.05, compared with Hepc-Mock).

CYP activation not tested routinely in the early stage of drug development.

Based on the results in section 2.1, it was showed that genotoxic metabolites could be produced by not only the most active CYP isoform but also by other less active CYPs and that this transformant system could detect the genotoxic potential of chemicals requiring

One of major advantages of our system is the variety of human CYP isoforms. When considering replacement of the rat induced liver S9 fraction, increasing the number of the principal CYP isoforms would be desirable in order to cover the diverse CYP activities. As mentioned in the general introduction, our HepG2 system includes CYP1A1, 1A2, 2A6, 2B6,

0

0.0

HepG2 Hepc-Mock

2.5

5.0

**Fold induction to saline-treated control**

7.5

10.0

HepG2 Hepc-Mock

\* \* \*

1A1 1A2 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4

\*

1A1 1A2 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4

\* \*

\*

5

10

**Fold induction to DMSO-treated control**

15

2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. Seven CYP isoforms (1A1, 1A2, 2C9, 2C19, 2D6, 2E1 and 3A4) account for 95% of this activity and with 3A4 responsible for over 65% of the metabolism of current therapeutic agents. During phase I metabolism in humans, 90% of all drugs are oxidized by CYP isoforms with different substrate selectivities. Thus, our HepG2 transformant system seems to cover most human drug metabolism mediated by human CYP isoforms but not by rat ones. Based on the results obtained in section 2.1, it was demonstrated that our HepG2 transformant system has an appropriate sensitivity for wellstudied chemicals which requires CYP activation for their genotoxicity.

For assessing the genotoxicity of chemicals with human metabolism, our HepG2 transformant system has more appropriate characteristics than other established cell lines used in toxicological testing and reviewed by Sawada and Kamataki (Sawada and Kamataki, 1998). Our HepG2 transformants are derived from hepatocyte that possesses other factors necessary for the function of CYP. Generally, the reactions catalyzed by CYP molecules require the presence of NADPH-CYP reductase and cytochrome b5 to support some CYP-mediated reactions. HepG2 has been shown to have NADPH-CYP reductase activity and cytochrome b5, although the levels are lower than those of human liver (Waxman et al., 1991; Patten et al., 1992). Therefore, our HepG2 transformant system does not need co-expression of reductase and/or cytochrome b5 with CYP enzymes.

#### **2.2 An exploration using HepG2 transformant to identify the CYP isoforms contributing to the genotoxicity by novel chemicals**

Given the multiplicity of CYP isoforms and the importance of other enzymes (hydrolases, transferase, etc.) in the metabolism of chemicals, there are two possible approaches to engineering cell lines. The introduction of single enzymes allows simple controlled mechanistic studies of the role of an individual enzyme in the metabolic activation of chemicals. Such system can also be viewed as the replacement of the laborious CYP purification/reconstitution analyses with a panel of engineered cells. However, to specify CYP isoform(s) involved in the activation of a certain chemical of unknown metabolism, a set of cell lines individually expressing the different CYP isoforms is needed. A series of HepG2 transformants expressing major 10 human CYP isoforms is a valuable tool, since most human hepatic drug metabolism is catalyzed by these expressing CYP isoforms.

Fig. 6. Chemical structure and significant biological features of okadaic acid. This chemical is a shellfish poisoning toxin and known as a potent phosphatase 1 and 2A. This chemical was reported to be induced micronucleus in the presence of rat liver S9 mix (Hégarat et al., 2004).

In this section, the following possibility was elucidated that the set of transformants can be used for screening for the genotoxicity of newly developed pharmaceutical candidates of unknown metabolism in human in vivo. As model chemicals, we selected okadaic acid (OA) and β-endosulfan.

Application of a New Genotoxicity Test System with

lactone [WHO, 1999].

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 373

changed with a fresh one containing 0.5% DMSO, 5 and 7.5 ng/ml OA. After 48 h, the cells were collected and sampled. Values were normalized with the mean DMSO-treated control value without siRNA of three experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with non-targeting siRNA). D). Effects of external metabolic activation system for CYP1A2 on micronucleus induction by okadaic acid in Hepc. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO, 5 or 7.5 ng/ml OA in the absence (heavy gray bars) or presence of Insect Cell Control SupersomesTM (light gray bars) or Human CYP1A2 SupersomesTM (medium gray bars). Values were normalized with the mean DMSO-treated control value without microsomes of 3 experiments in each

transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using

OA significantly increased the fold induction of MN in the transformant expressing CYP1A2 compared with that obtained in Hepc-Mock (Fig. 7 A). Furthermore, inhibitory effects of a specific inhibitor of CYP1A2 and siRNA to CYP1A2 on MN induction by OA were shown (Figs.7 B and 7 C, respectively). Moreover, co-treatment with OA and microsomes expressing CYP1A2 showed MN induction in Hepc-Mock cells (Fig. 7 D). These results indicated that MN induction by OA could be associated with the presence of CYP1A2

Fig. 8. Proposed metabolic pathway of β-endosulfan. In mammalian systems, β-endosulfan is metabolized to endosulfan sulfate, which is the most persistent metabolite, and also to endosulfan diol, which is further metabolized to endosulfan ether, hydroxyether, and

β-Endosulfan, shown in Fig. 8, is also reported to induce MN in HepG2 cells, suggesting that CYP activation might be involved in the MN induction (Lu et al., 2000); however the contributing CYP isoform to induce MN has not yet been investigated, to the best of our knowledge. Therefore we examined whether a series of HepG2 transformants could identify the CYP isoform contributing to the MN induction by β-endosulfan as a model chemical. β-Endosulfan significantly increased the fold induction of MN in the transformant expressing CYP3A4 compared with that obtained with the transformant Mepc-Mock (Fig. 8 A). Furthermore, inhibitory effects of a specific inhibitor of CYP3A4 and of siRNA to CYP3A4 on MN induction by β-endosulfan were shown (Figs. 8 B and 8 C, respectively). The activity of CYP3A4 in the transformant using the luminogenic substrate demonstrated that these inhibitory conditions decreased the activity to approximately 10% compared to control level. These results indicated that MN induction by β-endosulfan could be

activity, suggesting that CYP1A2 is involved in the genotoxic activation of OA.

Student's t-test (\*P<0.05, compared with the control microsomes).

Hégarat et al. found that OA enhanced formation of MN in the presence of a metabolic activation system (Hégarat et al., 2004), although the CYP isoforms involved in the MN induction were not reported. Thus we selected OA as a model chemical to evaluate the ability of our system to investigate which CYP isoform is involved in producing unknown genotoxic metabolites.

Fig. 7. Micronucleus induction of OA. A). Micronucleus induction of OA by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 5 ng/ml (light gray bars) or 10 ng/ml (medium gray bars) OA. After 48 h, the cells were collected and sampled. One thousand interphase cells per each treatment were scored and the incidence of cells with micronuclei was calculated. Values were normalized with the mean DMSO-treated control value of 3 experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with Hepc-Mock). B). Effects of furafylline, a CYP1A2 specific inhibitor for micronucleus induction by various chemicals in the transformant expressing CYP1A2. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO, 5 or 7.5 ng/ml OA, 400 ng/ml BP and 25 ng/ml MMC in the absence (heavy gray bars) or presence of 5 μM (lihgt gray bars) or 50 μM (medium gray bars) furafylline. Values were normalized with the mean DMSO-treated control value without an inhibition of three experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with no inhibition). C). Effects of siRNA to CYP1A2 on micronucleus induction by OA in the transformant expressing CYP1A2. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h in the presence of 50 nM siRNA for non-targeting (heavy gray bars) or CYP1A2 (light gray bars). Medium was

Hégarat et al. found that OA enhanced formation of MN in the presence of a metabolic activation system (Hégarat et al., 2004), although the CYP isoforms involved in the MN induction were not reported. Thus we selected OA as a model chemical to evaluate the ability of our system to investigate which CYP isoform is involved in producing unknown

A B

0

0.0

Fig. 7. Micronucleus induction of OA. A). Micronucleus induction of OA by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 5 ng/ml (light gray bars) or 10 ng/ml (medium gray bars) OA. After 48 h, the cells were collected and sampled. One thousand interphase cells per each treatment were scored and the incidence of cells with micronuclei was calculated. Values were normalized with the mean DMSO-treated control value of 3 experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test

(\*P<0.05, compared with Hepc-Mock). B). Effects of furafylline, a CYP1A2 specific inhibitor for micronucleus induction by various chemicals in the transformant expressing CYP1A2. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO, 5 or 7.5 ng/ml OA, 400 ng/ml BP and 25 ng/ml MMC in the absence (heavy gray bars) or presence of 5 μM (lihgt gray bars) or 50 μM (medium gray bars) furafylline. Values were normalized with the mean DMSO-treated control value without an inhibition of three experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with no inhibition). C). Effects of siRNA to CYP1A2 on micronucleus induction by OA in the transformant expressing CYP1A2. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h in the presence of 50 nM siRNA for non-targeting (heavy gray bars) or CYP1A2 (light gray bars). Medium was

2.5

5.0

**Fold induction to DMSO-treated control without microsomes** 

7.5

C D

Saline OA 5 ng/ml OA 7.5 ng/ml BP 400 ng/ml MMC 25 ng/ml

DMSO OA 5 ng/ml OA 7.5 ng/ml

\* \*

\*

\*

\*

5

10

**Fold induction to DMSO-treated control without a inhibitor** 

15

20

genotoxic metabolites.

\*

0

0.0

2.5

5.0

**Fold induction to DMSO-treated control in the non-targeting** 

**siRNA group** 

7.5

10.0

HepG2 Hepc-Mock

1A1 1A2 2A6 2B6 2C8 2C9 2C19 2D6 2E1 3A4

\*

DMSO OA 5 ng/ml OA 7.5 ng/ml

5

10

15

**Fold induction to DMSO-treated control**

20

25

changed with a fresh one containing 0.5% DMSO, 5 and 7.5 ng/ml OA. After 48 h, the cells were collected and sampled. Values were normalized with the mean DMSO-treated control value without siRNA of three experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with non-targeting siRNA). D). Effects of external metabolic activation system for CYP1A2 on micronucleus induction by okadaic acid in Hepc. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO, 5 or 7.5 ng/ml OA in the absence (heavy gray bars) or presence of Insect Cell Control SupersomesTM (light gray bars) or Human CYP1A2 SupersomesTM (medium gray bars). Values were normalized with the mean DMSO-treated control value without microsomes of 3 experiments in each transformant. Each bar represents the mean ± S.D. of 3 experiments. Data were tested using Student's t-test (\*P<0.05, compared with the control microsomes).

OA significantly increased the fold induction of MN in the transformant expressing CYP1A2 compared with that obtained in Hepc-Mock (Fig. 7 A). Furthermore, inhibitory effects of a specific inhibitor of CYP1A2 and siRNA to CYP1A2 on MN induction by OA were shown (Figs.7 B and 7 C, respectively). Moreover, co-treatment with OA and microsomes expressing CYP1A2 showed MN induction in Hepc-Mock cells (Fig. 7 D). These results indicated that MN induction by OA could be associated with the presence of CYP1A2 activity, suggesting that CYP1A2 is involved in the genotoxic activation of OA.

Fig. 8. Proposed metabolic pathway of β-endosulfan. In mammalian systems, β-endosulfan is metabolized to endosulfan sulfate, which is the most persistent metabolite, and also to endosulfan diol, which is further metabolized to endosulfan ether, hydroxyether, and lactone [WHO, 1999].

β-Endosulfan, shown in Fig. 8, is also reported to induce MN in HepG2 cells, suggesting that CYP activation might be involved in the MN induction (Lu et al., 2000); however the contributing CYP isoform to induce MN has not yet been investigated, to the best of our knowledge. Therefore we examined whether a series of HepG2 transformants could identify the CYP isoform contributing to the MN induction by β-endosulfan as a model chemical.

β-Endosulfan significantly increased the fold induction of MN in the transformant expressing CYP3A4 compared with that obtained with the transformant Mepc-Mock (Fig. 8 A). Furthermore, inhibitory effects of a specific inhibitor of CYP3A4 and of siRNA to CYP3A4 on MN induction by β-endosulfan were shown (Figs. 8 B and 8 C, respectively). The activity of CYP3A4 in the transformant using the luminogenic substrate demonstrated that these inhibitory conditions decreased the activity to approximately 10% compared to control level. These results indicated that MN induction by β-endosulfan could be

Application of a New Genotoxicity Test System with

treated with β-endosulfan.

activation system.

CYP enzymes.

metabolic activation system.

**metabolizing enzymes** 

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 375

Lee et al. reported that β-endosulfan is metabolized by CYP3A4 based on the results in the study with CYP isoform-selective inhibitor in human liver microsomes and with the incubation study of cDNA-expressed enzymes (Lee et al., 2006). They have also reported that human liver microsome incubation of β-endosulfan in the presence of NADPH resulted in the formation of endosulfan sulfate (Lee et al., 2006). Based on these reports, we examined the genotoxicity of endosulfan sulfate in the Hepc-Mock cells in order to investigate whether this sulfate is the metabolite that induces MN in the β-endosulfan–treated transformant expressing CYP3A4. As shown in Fig. Fd, endosulfan sulfate induced MN with statistical significance at 12.5 μg/ml. This result demonstrated that endosulfan sulfate was the genotoxic metabolite and that this metabolite was formed by CYP3A4 in the transformant

Based on the results obtained in the OA and β-endosulfan treatments, it was clearly demonstrated that the HepG2 transformant system was able to identify the CYP isoform relating to the genotoxicity of chemical metabolite(s) and was useful to elucidate the genotoxicity of a new chemical or a drug candidate in the presence of the metabolic

More effort as for CYP induction is necessary, but the results obtained in the present study demonstrated the availability of these transformants expressing human CYP to elucidate the genotoxic potential of the chemicals that require metabolic activation to create risk to humans. In order to validate these transformants, an additional study is in progress with more chemicals that have been well studied in the metabolic activation or inactivation by

These results clearly demonstrated that the HepG2 transformant system was able to identify the CYP isoform related to the genotoxicity of chemical metabolite(s) and was useful to elucidate the genotoxicity of a new chemical or a drug candidate in the presence of the

**2.3 Genotoxic assessment of chemicals metabolized by phase I and phase II drug-**

of carboxylic acids; [Glatt et al., 1998; Sallustio et al., 2006]).

As mentioned in the Introduction section, to detect chemicals which require bioactivation to electrophiles to exhibit a genotoxic and carcinogenic response, the standard in vitro genotoxicity testing also include incubation of the test chemicals with liver microsomal or S9 fractions, as activation systems so that chemically stable xenobiotics can be converted to reactive electrophiles (Malling, 1971; Ames et al., 1973; Levin et al., 1984). However, it is possible that the Aroclor induced rat liver S9 system is not the most appropriate metabolite generation system for detecting drugs that may pose a carcinogenic hazard to humans. This is because the system is set up to favor CYP-mediated metabolism and the other enzymes present in the system that could be responsible for detoxication of reactive intermediates are not supplemented with the appropriate cofactors (e.g., UGT, GST, methyl transferases, etc), thus potentially providing an unrealistic metabolic profile. In a recent work by Obach and Dobo, it was revealed that many human in vivo metabolites arise via conjugation reactions with the limited 16 drugs (Obach and Dobo, 2008). This can be important not only for reducing potential false positives (e.g., reactive electrophiles that would be rapidly quenched by conjugation in vivo before being able to cause mutation) but can also be important for false negatives because some conjugation reactions can yield metabolites that are more reactive than their substrate (e.g., sulfation of aliphatic alcohols or glucuronidation

Endosulfan sulfate

associated with the presence of CYP3A4 activity and suggested that CYP3A4 is involved in producing the genotoxic metabolites of β-endosulfan.

Fig. 9. Micronucleus induction of β-endosulfan by CYP3A4-mediated activation. a). A micronucleus induction of β-endosulfan by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 6.25 μg/ml (light gray bars) or 12.5 μg/ml (medium gray bars) β-endosulfan. b). Effects of ketoconazole, a CYP3A4 specific inhibitor for micronucleus induction by various chemicals in the transformant expressing CYP3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% saline, 100 ng/ml mitomycin C, 1000 μg/ml cyclophosphamide, 0.5% DMSO, 6.25 or 12.5 μg/ml β-endosulfan in the absence (heavy gray bars) or presence (light gray bars) of 1 μM ketoconazole. c). Effects of siRNA to CYP3A4 on micronucleus induction by various chemicals in the transformant expressing CYP3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 8 h in absence (heavy gray bars) or the presence of 50 nM siRNA for Negative control (light gray bars) or CYP3A4 (medium gray bars). d). Micronucleus induction of endosulfan sulfate in Hepc-Mock, the transformant expressing an empty vector only. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 8 μg/ml, (light gray bars), 10 μg/ml (medium gray bars) or 12.5 μg/ml (solid bars) endosulfan sulfate. Statistical analysis was done in the same procedure as Fig. E, except for the Student's t- test (#P<0.05, compared with the DMSOtreated control group) in Fig. F d).

associated with the presence of CYP3A4 activity and suggested that CYP3A4 is involved in

A B

Fig. 9. Micronucleus induction of β-endosulfan by CYP3A4-mediated activation. a). A micronucleus induction of β-endosulfan by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 6.25 μg/ml (light gray bars) or 12.5 μg/ml (medium gray bars) β-endosulfan. b). Effects of ketoconazole, a CYP3A4 specific inhibitor for micronucleus induction by various chemicals in the transformant expressing CYP3A4. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% saline, 100 ng/ml mitomycin C, 1000 μg/ml cyclophosphamide, 0.5% DMSO, 6.25 or 12.5 μg/ml β-endosulfan in the absence (heavy gray bars) or presence (light gray bars) of 1 μM ketoconazole. c). Effects of siRNA to CYP3A4 on micronucleus induction by various

chemicals in the transformant expressing CYP3A4. The cells (1×105 cells) were seeded onto a

induction of endosulfan sulfate in Hepc-Mock, the transformant expressing an empty vector only. The cells (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 8 μg/ml, (light gray bars), 10 μg/ml (medium gray bars) or 12.5 μg/ml (solid bars) endosulfan sulfate. Statistical analysis was done in the same procedure as Fig. E, except for the Student's t- test (#P<0.05, compared with the DMSO-

24-well plate for 8 h in absence (heavy gray bars) or the presence of 50 nM siRNA for Negative control (light gray bars) or CYP3A4 (medium gray bars). d). Micronucleus

DMSO Endosulfan (6.25) Endosulfan (12.5)

C D

0.0

0.0

1.0

2.0

**Fold induction to the control group in the non-inhibition** 

**condition** 

3.0

Saline Mitomycin C (0.025)

Cyclophosphamide (2000)

Endosulfan sulfate

DMSO Endosulfan (6.25)

#

Endosulfan (12.5)

5.0

**Fold induction to the control group in the non-inhibition** 

**condition** 

10.0

producing the genotoxic metabolites of β-endosulfan.

CYP1A1 CYP1A2 CYP2A6 CYP2B6 CYP2C8 CYP2C9 CYP2C19 CYP2D6 CYP2E1 CYP3A4

0.0

0.0

Saline Mitomycin C (0.025)

Cyclophosphamide (2000)

treated control group) in Fig. F d).

5.0

10.0

**Fold induction to the control group in the non-inhibition** 

**condition** 

HepG2 Hepc-Mock

5.0

**Fold induction to each control group**

10.0

Lee et al. reported that β-endosulfan is metabolized by CYP3A4 based on the results in the study with CYP isoform-selective inhibitor in human liver microsomes and with the incubation study of cDNA-expressed enzymes (Lee et al., 2006). They have also reported that human liver microsome incubation of β-endosulfan in the presence of NADPH resulted in the formation of endosulfan sulfate (Lee et al., 2006). Based on these reports, we examined the genotoxicity of endosulfan sulfate in the Hepc-Mock cells in order to investigate whether this sulfate is the metabolite that induces MN in the β-endosulfan–treated transformant expressing CYP3A4. As shown in Fig. Fd, endosulfan sulfate induced MN with statistical significance at 12.5 μg/ml. This result demonstrated that endosulfan sulfate was the genotoxic metabolite and that this metabolite was formed by CYP3A4 in the transformant treated with β-endosulfan.

Based on the results obtained in the OA and β-endosulfan treatments, it was clearly demonstrated that the HepG2 transformant system was able to identify the CYP isoform relating to the genotoxicity of chemical metabolite(s) and was useful to elucidate the genotoxicity of a new chemical or a drug candidate in the presence of the metabolic activation system.

More effort as for CYP induction is necessary, but the results obtained in the present study demonstrated the availability of these transformants expressing human CYP to elucidate the genotoxic potential of the chemicals that require metabolic activation to create risk to humans. In order to validate these transformants, an additional study is in progress with more chemicals that have been well studied in the metabolic activation or inactivation by CYP enzymes.

These results clearly demonstrated that the HepG2 transformant system was able to identify the CYP isoform related to the genotoxicity of chemical metabolite(s) and was useful to elucidate the genotoxicity of a new chemical or a drug candidate in the presence of the metabolic activation system.

#### **2.3 Genotoxic assessment of chemicals metabolized by phase I and phase II drugmetabolizing enzymes**

As mentioned in the Introduction section, to detect chemicals which require bioactivation to electrophiles to exhibit a genotoxic and carcinogenic response, the standard in vitro genotoxicity testing also include incubation of the test chemicals with liver microsomal or S9 fractions, as activation systems so that chemically stable xenobiotics can be converted to reactive electrophiles (Malling, 1971; Ames et al., 1973; Levin et al., 1984). However, it is possible that the Aroclor induced rat liver S9 system is not the most appropriate metabolite generation system for detecting drugs that may pose a carcinogenic hazard to humans. This is because the system is set up to favor CYP-mediated metabolism and the other enzymes present in the system that could be responsible for detoxication of reactive intermediates are not supplemented with the appropriate cofactors (e.g., UGT, GST, methyl transferases, etc), thus potentially providing an unrealistic metabolic profile. In a recent work by Obach and Dobo, it was revealed that many human in vivo metabolites arise via conjugation reactions with the limited 16 drugs (Obach and Dobo, 2008). This can be important not only for reducing potential false positives (e.g., reactive electrophiles that would be rapidly quenched by conjugation in vivo before being able to cause mutation) but can also be important for false negatives because some conjugation reactions can yield metabolites that are more reactive than their substrate (e.g., sulfation of aliphatic alcohols or glucuronidation of carboxylic acids; [Glatt et al., 1998; Sallustio et al., 2006]).

Application of a New Genotoxicity Test System with

exerting its genotoxicity in the metabolic pathway.

more relevant data than current genotoxicity tests.

Fig. 12. Metabolic activation and inactivation pathways of safrole.

inactive substance.

GST (Rietjens et al., 2005).

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 377

tested the effect of UGT inhibitor on the MN induction by tamoxifen. A small but significant increase in MN by tamoxifen was observed in the presence of the UGT inhibitor, diclofenac, in the transformants expressing CYP3A4 which contribute to the metabolic activation of tamoxifen to α-hydroxytamoxifen (Fig. 11 B). This result indicated that the CYP3A4 metabolite, α-hydroxy-tamoxifen, was further metabolized by UGT to a genotoxically

Another example is safrole. Safrole is reported to be hydroxylated predominantly by CYP2A6, 2C9, 2D6 or 2E1 and further metabolized by SULT to 1'-sulfooxysafrole (Andrew and Brian, 2007; Rietjens et al., 2005)(Fig. 12). This intermediate forms the electrophilic carbocation of safrole, suggesting the production of DNA adduct (Rietjens et al., 2005). On the other hand, the safrole-2', 3'-oxide formed from the parent safrole by epoxide hydrolases or 1'-hydroxysafrole-2', 3'-oxide from 1'-hydroxysaforle are reported to be detoxified by

As shown in Fig. 13 A, safrole did not significantly induce MN at any concentration tested in any transformant. At much higher concentrations of each chemical, the frequencies of the micronuclei were decreased (data not shown), suggesting that the tested concentrations were appropriate to evaluate MN induction. Then to investigate the involvement of the detoxification pathway by GST in the metabolism of safrole, we tested the effect of GST inhibitor on the MN induction by safrole. Significant increases were seen in the presence of the GST inhibitor, ethacrynic acid, in the transformants expressing CYP2D6 responsible for the genotoxic activation of safrole to 1'-hydroxysafrole (Fig. 13 B). This result suggested that CYP2D6-mediated metabolite, 1'-hydroxysafrole, was further metabolized by GST not

The results for tamoxifen and safrole clearly demonstrated that interaction between the phase I and II drug-metabolizing enzymes was crucial to assess the genotoxicity of chemicals in the presence of a metabolic activation system. The interplay between the phase I and II enzymes is lacking in the NADPH-supplemented rat liver S9 system due to an absence of co-factor necessary for several phase II enzymes such as UGT or GST. Furthermore, the reactive intermediates have to be formed in the target cell because some conjugates have poor membrane permeability. These results raise the possibility that the induced rat liver S9 system may generate mutagenic metabolites of no relevance, or worse even may not generate a mutagenic metabolite that would be generated by human enzymes. Therefore, a set of HepG2 transformants is a superior test system for mimicking the metabolism occurring in the human liver and the use of this system can potentially provide

Fig. 10. Metabolic activation and inactivation pathways of tamoxifen.

Fig. 11. Micronucleus induction of tamoxifen by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 (A), and effects of a phase II inhibitro on micronucleus indcution by tamoxifen (B). A). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 1 μg/ml (light gray bars) or 2 μg/ml (medium gray bars) tamoxifen. No statistically significant increase was observed when compared with both Hepc-Mock and HepG2. B). The transformant cells expressing CYP3A4 (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 2 μg/ml (light gray bars) or 3 μg/ml (medium gray bars) tamoxifen in the absence or presence of 50 μM diclofenac, a UGT inhibitor.

In order to evaluate the relevance of the interaction between phase I and phase II drugmetabolizing enzymes in the test system, the transformants were treated with tamoxifen as shown in Fig. 10. Tamoxifen is reported to be metabolized by CYP3A4 to αhydroxytamoxifen and further metabolized by SULT to α-hydroxytamoxifen sulfate ester as the putative reactive intermediate (Brown, 2009; White, 2003; Zhao et al., 2009). This intermediate reacts with the exocyclic amino group of guanines (the major reaction) and adenines (a minor reaction) in DNA (Osborne et al., 1996). UGT plays a detoxification role through the glucuronidation of α-hydroxytamoxifen (Brown, 2009; White, 2003; Zhao et al., 2009).

As shown in Fig. 11 A, tamoxifen did not significantly induce MN at any concentration tested in any transformant. At much higher concentrations of each chemical, the frequencies of the micronuclei were decreased (data not shown), suggesting that the tested concentrations were appropriate to evaluate MN induction. Then to investigate the involvement of the detoxification pathway by UGT in the metabolism of tamoxifen, we

0.0

No inhibitor 50 μM Diclofenac

1.0

**Fold induction to the control group in the non-inhibition** 

Fig. 11. Micronucleus induction of tamoxifen by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8,

indcution by tamoxifen (B). A). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 1 μg/ml (light gray bars) or 2 μg/ml (medium gray bars) tamoxifen. No statistically significant increase was observed when compared with both Hepc-Mock and HepG2. B). The transformant cells expressing CYP3A4 (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 2 μg/ml (light gray bars) or 3 μg/ml (medium gray bars) tamoxifen in the

In order to evaluate the relevance of the interaction between phase I and phase II drugmetabolizing enzymes in the test system, the transformants were treated with tamoxifen as shown in Fig. 10. Tamoxifen is reported to be metabolized by CYP3A4 to αhydroxytamoxifen and further metabolized by SULT to α-hydroxytamoxifen sulfate ester as the putative reactive intermediate (Brown, 2009; White, 2003; Zhao et al., 2009). This intermediate reacts with the exocyclic amino group of guanines (the major reaction) and adenines (a minor reaction) in DNA (Osborne et al., 1996). UGT plays a detoxification role through the glucuronidation of α-hydroxytamoxifen (Brown, 2009; White, 2003; Zhao et al.,

As shown in Fig. 11 A, tamoxifen did not significantly induce MN at any concentration tested in any transformant. At much higher concentrations of each chemical, the frequencies of the micronuclei were decreased (data not shown), suggesting that the tested concentrations were appropriate to evaluate MN induction. Then to investigate the involvement of the detoxification pathway by UGT in the metabolism of tamoxifen, we

2C9, 2C19, 2D6, 2E1 and 3A4 (A), and effects of a phase II inhibitro on micronucleus

**condition** 

2.0

Fig. 10. Metabolic activation and inactivation pathways of tamoxifen.

A B

CYP1A1 CYP1A2 CYP2A6 CYP2B6 CYP2C8 CYP2C9 CYP2C19 CYP2D6 CYP2E1 CYP3A4

absence or presence of 50 μM diclofenac, a UGT inhibitor.

0.0

2009).

HepG2 Hepc-Mock

1.0

**Fold induction to each control group**

2.0

tested the effect of UGT inhibitor on the MN induction by tamoxifen. A small but significant increase in MN by tamoxifen was observed in the presence of the UGT inhibitor, diclofenac, in the transformants expressing CYP3A4 which contribute to the metabolic activation of tamoxifen to α-hydroxytamoxifen (Fig. 11 B). This result indicated that the CYP3A4 metabolite, α-hydroxy-tamoxifen, was further metabolized by UGT to a genotoxically inactive substance.

Another example is safrole. Safrole is reported to be hydroxylated predominantly by CYP2A6, 2C9, 2D6 or 2E1 and further metabolized by SULT to 1'-sulfooxysafrole (Andrew and Brian, 2007; Rietjens et al., 2005)(Fig. 12). This intermediate forms the electrophilic carbocation of safrole, suggesting the production of DNA adduct (Rietjens et al., 2005). On the other hand, the safrole-2', 3'-oxide formed from the parent safrole by epoxide hydrolases or 1'-hydroxysafrole-2', 3'-oxide from 1'-hydroxysaforle are reported to be detoxified by GST (Rietjens et al., 2005).

As shown in Fig. 13 A, safrole did not significantly induce MN at any concentration tested in any transformant. At much higher concentrations of each chemical, the frequencies of the micronuclei were decreased (data not shown), suggesting that the tested concentrations were appropriate to evaluate MN induction. Then to investigate the involvement of the detoxification pathway by GST in the metabolism of safrole, we tested the effect of GST inhibitor on the MN induction by safrole. Significant increases were seen in the presence of the GST inhibitor, ethacrynic acid, in the transformants expressing CYP2D6 responsible for the genotoxic activation of safrole to 1'-hydroxysafrole (Fig. 13 B). This result suggested that CYP2D6-mediated metabolite, 1'-hydroxysafrole, was further metabolized by GST not exerting its genotoxicity in the metabolic pathway.

Fig. 12. Metabolic activation and inactivation pathways of safrole.

The results for tamoxifen and safrole clearly demonstrated that interaction between the phase I and II drug-metabolizing enzymes was crucial to assess the genotoxicity of chemicals in the presence of a metabolic activation system. The interplay between the phase I and II enzymes is lacking in the NADPH-supplemented rat liver S9 system due to an absence of co-factor necessary for several phase II enzymes such as UGT or GST. Furthermore, the reactive intermediates have to be formed in the target cell because some conjugates have poor membrane permeability. These results raise the possibility that the induced rat liver S9 system may generate mutagenic metabolites of no relevance, or worse even may not generate a mutagenic metabolite that would be generated by human enzymes. Therefore, a set of HepG2 transformants is a superior test system for mimicking the metabolism occurring in the human liver and the use of this system can potentially provide more relevant data than current genotoxicity tests.

Application of a New Genotoxicity Test System with

should not be relied on when testing new compounds.

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 379

The induced rat liver S9 fraction has another limitation. To be detected as mutagens, some genotoxic metabolites have to be formed within the target cell by enzymes that are not represented in standard in vitro test systems. SULT-dependent activations are not uncommon. Using genetically modified target cells, activation by SULTs has been demonstrated for more than 100 chemicals, including various carcinogens (such as tamoxifen, cyproterone acetate, safrole, nitrofen and some nitrotoluenes) that are missed in conventional test systems (Glatt 2000, Glatt 2005, Glatt and Meinl, 2005). Depending on the compound, varying SULT forms were required for the activation. Like N-sulfooxy-2-acetylaminofluorene, several other sulfo conjugates [e.g. furfuryl sulfate and 1-(L-sulfooxyethyl) pyrene] had to be formed within the target for a positive test result. Other reactive sulfo conjugates undergo spontaneous substitution reactions with components of the culture medium, such as chloride anions, leading to the formation of secondary, membrane-penetrating active species (Glatt et al., 1990). Moreover, cDNA-mediated expression of organic anion transporters in target cells enhanced the genotoxic effects of some reactive sulfuric acid esters externally added (Bakhiya et al., 2006). Such uptake mechanisms might play a role in the organotropism of reactive species, but

Other conjugating enzymes (some UGTs, GSTs and NATs) have also been expressed in target cells. The activation of promutagens by UGTs in such models has not yet been reported (and not been studied). However, co-expression of human UGT1A1 provided protection against the mutagenicity and cytotoxicity of PhIP in CHO-derived cells engineered for expression of CYP1A2 (Malfatti et al., 2005). Human GST T1, expressed in Salmonella typhimurium, strongly enhanced the mutagenicity of various dihalogenated alkanes as well as diepoxybutane (Thier et al., 1996). The activation of some of these agents could also be demonstrated using external GSH-conjugating systems (Rannug et al., 1978), but the extent of the uptake and its dependence on the structures of the reactive GSH conjugates are largely unexplored. Heterologous expression of GSTs in mammalian cells conferred resistance against various alkylating agents; in some cases, this protection was enhanced by, or was even strictly dependent on, the co-expression of an export pump (MRP-1 or MRP-2) (Smitherman et al., 2004). The expression of endogenous acetyltransferases in Salmonella may be a reason for the high mutagenic activity observed in the Ames test with many amino- and nitro-arenes, whose final activation step is often an O-acetylation. Salmonella strains are available in which O-acetyltransferase has been replaced by a mammalian NAT (Glatt and Meinl 2005, Grant et al., 1992), which differ in substrate specificity. Thus, various aromatic hydroxamic acids are activated to mutagens by human NATs, but not by OAT. Such differences may often lead to misleading results when the standard bacterial strains are used. Unlike typical phase II metabolites, acetyl conjugates are uncharged. Nevertheless the site of their formation can strongly affect the outcome of mutagenicity experiments. Thus, PhIP shows much higher mutagenicity in S. typhimurium TA98 compared to an O-acetyltransferase-deficient variant of this strain; however, purified O-acetyltransferase in the presence of its cofactor acetyl-CoA had drastically reduced its bacterial mutagenicity (although it strongly enhanced the covalent binding to naked DNA) (Saito et al., 1985). Various standard mammalian target cells, including most sublines of V79 cells, do not express any endogenous NAT. Heterologous expression of human NATs in these cells strongly enhanced the genotoxic effects of many amino- and nitro-arenes (Glatt, 2005; Glatt, 2006). For example, induction of gene mutations by 3-nitrobenzanthrone required 100 times lower substrate concentrations in NAT2-expressing compared to control V79 cells. The isomer, 2-nitrobenzanthrone, was mutagenic in cells engineered for

Drug metabolism is generally regarded as proceeding via 2 stages, phase I and phase II. The induced rat liver S9 fraction as an exogenous metabolic activation system is supplemented with only NADPH for CYP-mediated metabolism. The appropriate cofactors for phase II drug-metabolizing enzymes (e.g. UGT, GST, SULT and NAT) are absent. This means these phase II enzymes are not active in the rat liver S9 fraction and that this leads not only for potential false positives (e.g., reactive electrophiles that would be rapidly quenched by conjugation in vivo before being able to cause mutation) but also for false negatives because some conjugation reactions can yield metabolites that are more reactive than their substrate (Dashwood, 2002; Ku et al., 2007). In other words, the use of an S9 system with NADPH may represent an incomplete picture of the metabolism that can occur in vivo. In particular, it is well-studied that SULTs are able to sequester some proximate mutagens through the

Fig. 13. Micronucleus induction of safrole by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 (a), and effects of a phase II inhibitro on micronucleus indcution by safrole (b). a). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 83.3 μg/ml (light gray bars) or 125 μg/ml (medium gray bars) safrole. No statistically significant increase was observed when compared with both Hepc-Mock and HepG2. b). The transformant cells expressing CYP2D6 (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 41.7 μg/ml (light gray bars), 62.5 μg/ml (medium gray bars) or 83.3 μg/ml (solid bars) safrole in the absence or presence of 15 μM ethacrynic acid, a GST inhibitor.

transfer of a sulfuryl group. However are these are also not active in the standard testing system because the necessary cofactor, 3'-phosphoadenosine-5'-phosphosulfate, is not added (Glatt, 2000; Glatt, 2005). From the study by Obach and Dobo using 16 drugs commonly used, not all metabolites observed as significant in humans in vivo are generated in the system using rat or human S9 fractions (Obach and Dobo, 2008).. Furthermore, they reported that a metabolite observed in humans in vivo was only seen in the rat S9 system and not the human S9 system in a few cases, (Obach and Dobo, 2008). Many human in vivo metabolites arise via conjugation reactions, which will not be observed in the in vitro S9 system as presently supplemented in standard in vitro genotoxicity tests. In addition, with regard to similarity to in vivo metabolite profiles, the results of the in vitro testing presented in the literature by Obach and Dobo clearly demonstrate a limitation of both systems, in that both human and rat S9 predominantly generate metabolites that are the result of one to two metabolic reactions (>90%).

Drug metabolism is generally regarded as proceeding via 2 stages, phase I and phase II. The induced rat liver S9 fraction as an exogenous metabolic activation system is supplemented with only NADPH for CYP-mediated metabolism. The appropriate cofactors for phase II drug-metabolizing enzymes (e.g. UGT, GST, SULT and NAT) are absent. This means these phase II enzymes are not active in the rat liver S9 fraction and that this leads not only for potential false positives (e.g., reactive electrophiles that would be rapidly quenched by conjugation in vivo before being able to cause mutation) but also for false negatives because some conjugation reactions can yield metabolites that are more reactive than their substrate (Dashwood, 2002; Ku et al., 2007). In other words, the use of an S9 system with NADPH may represent an incomplete picture of the metabolism that can occur in vivo. In particular, it is well-studied that SULTs are able to sequester some proximate mutagens through the

A B Fig. 13. Micronucleus induction of safrole by expression of CYP1A1, 1A2, 2A6, 2B6, 2C8, 2C9, 2C19, 2D6, 2E1 and 3A4 (a), and effects of a phase II inhibitro on micronucleus indcution by safrole (b). a). The cells (1×105 cells) were seeded onto a 24-well plate for 24 h, and then treated with 0.5% DMSO (heavy gray bars), 83.3 μg/ml (light gray bars) or 125 μg/ml (medium gray bars) safrole. No statistically significant increase was observed when compared with both Hepc-Mock and HepG2. b). The transformant cells expressing CYP2D6 (1×105 cells) were seeded onto a 24-well plate for 24 h and then treated with 0.5% DMSO (heavy gray bars), 41.7 μg/ml (light gray bars), 62.5 μg/ml (medium gray bars) or 83.3 μg/ml (solid bars) safrole in the absence or presence of 15 μM ethacrynic acid, a GST

transfer of a sulfuryl group. However are these are also not active in the standard testing system because the necessary cofactor, 3'-phosphoadenosine-5'-phosphosulfate, is not added (Glatt, 2000; Glatt, 2005). From the study by Obach and Dobo using 16 drugs commonly used, not all metabolites observed as significant in humans in vivo are generated in the system using rat or human S9 fractions (Obach and Dobo, 2008).. Furthermore, they reported that a metabolite observed in humans in vivo was only seen in the rat S9 system and not the human S9 system in a few cases, (Obach and Dobo, 2008). Many human in vivo metabolites arise via conjugation reactions, which will not be observed in the in vitro S9 system as presently supplemented in standard in vitro genotoxicity tests. In addition, with regard to similarity to in vivo metabolite profiles, the results of the in vitro testing presented in the literature by Obach and Dobo clearly demonstrate a limitation of both systems, in that both human and rat S9 predominantly generate metabolites that are the result of one to two

inhibitor.

metabolic reactions (>90%).

The induced rat liver S9 fraction has another limitation. To be detected as mutagens, some genotoxic metabolites have to be formed within the target cell by enzymes that are not represented in standard in vitro test systems. SULT-dependent activations are not uncommon. Using genetically modified target cells, activation by SULTs has been demonstrated for more than 100 chemicals, including various carcinogens (such as tamoxifen, cyproterone acetate, safrole, nitrofen and some nitrotoluenes) that are missed in conventional test systems (Glatt 2000, Glatt 2005, Glatt and Meinl, 2005). Depending on the compound, varying SULT forms were required for the activation. Like N-sulfooxy-2-acetylaminofluorene, several other sulfo conjugates [e.g. furfuryl sulfate and 1-(L-sulfooxyethyl) pyrene] had to be formed within the target for a positive test result. Other reactive sulfo conjugates undergo spontaneous substitution reactions with components of the culture medium, such as chloride anions, leading to the formation of secondary, membrane-penetrating active species (Glatt et al., 1990). Moreover, cDNA-mediated expression of organic anion transporters in target cells enhanced the genotoxic effects of some reactive sulfuric acid esters externally added (Bakhiya et al., 2006). Such uptake mechanisms might play a role in the organotropism of reactive species, but should not be relied on when testing new compounds.

Other conjugating enzymes (some UGTs, GSTs and NATs) have also been expressed in target cells. The activation of promutagens by UGTs in such models has not yet been reported (and not been studied). However, co-expression of human UGT1A1 provided protection against the mutagenicity and cytotoxicity of PhIP in CHO-derived cells engineered for expression of CYP1A2 (Malfatti et al., 2005). Human GST T1, expressed in Salmonella typhimurium, strongly enhanced the mutagenicity of various dihalogenated alkanes as well as diepoxybutane (Thier et al., 1996). The activation of some of these agents could also be demonstrated using external GSH-conjugating systems (Rannug et al., 1978), but the extent of the uptake and its dependence on the structures of the reactive GSH conjugates are largely unexplored. Heterologous expression of GSTs in mammalian cells conferred resistance against various alkylating agents; in some cases, this protection was enhanced by, or was even strictly dependent on, the co-expression of an export pump (MRP-1 or MRP-2) (Smitherman et al., 2004). The expression of endogenous acetyltransferases in Salmonella may be a reason for the high mutagenic activity observed in the Ames test with many amino- and nitro-arenes, whose final activation step is often an O-acetylation. Salmonella strains are available in which O-acetyltransferase has been replaced by a mammalian NAT (Glatt and Meinl 2005, Grant et al., 1992), which differ in substrate specificity. Thus, various aromatic hydroxamic acids are activated to mutagens by human NATs, but not by OAT. Such differences may often lead to misleading results when the standard bacterial strains are used. Unlike typical phase II metabolites, acetyl conjugates are uncharged. Nevertheless the site of their formation can strongly affect the outcome of mutagenicity experiments. Thus, PhIP shows much higher mutagenicity in S. typhimurium TA98 compared to an O-acetyltransferase-deficient variant of this strain; however, purified O-acetyltransferase in the presence of its cofactor acetyl-CoA had drastically reduced its bacterial mutagenicity (although it strongly enhanced the covalent binding to naked DNA) (Saito et al., 1985). Various standard mammalian target cells, including most sublines of V79 cells, do not express any endogenous NAT. Heterologous expression of human NATs in these cells strongly enhanced the genotoxic effects of many amino- and nitro-arenes (Glatt, 2005; Glatt, 2006). For example, induction of gene mutations by 3-nitrobenzanthrone required 100 times lower substrate concentrations in NAT2-expressing compared to control V79 cells. The isomer, 2-nitrobenzanthrone, was mutagenic in cells engineered for

Application of a New Genotoxicity Test System with

chemicals.

**4. Summary and conclusion** 

CYP isoform but also by other less active CYPs.

activation system.

Human Hepatocyte Cell Lines to Improve the Risk Assessment in the Drug Development 381

expression of genotoxicity by a chemical. Despite the proof that most chemical carcinogens undergo metabolic conversion into DNA-reactive intermediates, some compounds do not bind to DNA and are not mutagenic, yet they are carcinogenic in animal models and possibly also in humans. These non-genotoxic mechanisms such as induction of inflammation, immunosuppression, formation of reactive oxygen species, activation of receptors such as arylhydrocarbon receptor or estrogen receptor, and epigenetic silencing. Therefore ,another approach based on the non-genotoxic mechanism is necessary to predict the carcinogenic action from a certain chemical. Together, these genotoxic and nongenotoxic mechanisms can alter signal-transduction pathways that finally result in hypermutability, genomic instability, loss of proliferation control, and resistance to apoptosis — some of the characteristic features of cancer cells. In this regard, we need to learn much more about the role and interplay of susceptibility and resistance function targeted by human carcinogens or involved in modulating human responses to carcinogenic

Many carcinogens are known to be procarcinogens and require metabolic activation to exert their genotoxicity through the formation of reactive intermediates. Therefore, for hazard identification on the genotoxic potential of drug candidate and its metabolites, S9 fraction prepared from the livers of rats pretreated with phenobarbital and 5,6-benzoflavone or with Aroclor 1254 to induce drug-metabolizing enzyme activity must be used in the in vitro genotoxicity testing. However, it is frequently questioned as to whether such an in vitro metabolite generation system is the most relevant for human risk, or whether the assay would be better served by using a human-derived in vitro system. In the present study, we examined the advantages of HepG2 transformants expressing a series of human CYP isoforms as a better

In section 2.1, the sensitivity of this system to detect genotoxicity requiring CYP activation was confirmed in the in vitro micronucleus tests using well-studied model chemicals. These results showed HepG2 transformants system have the appropriate sensitivity to detect genotoxicity requiring CYP activation tests using well-studied model chemicals. In addition, based on results obtained in the DMBA and ifosfamide treatments, HepG2 transformant system showed that genotoxic metabolites would be produced by not only the most active

In chapter 2.2, this system allowed us to investigate the genotoxicity of model chemicals for which the contributing CYP isoforms, especially those mediated by CYP1A2 or 3A4 which is known to metabolize many drugs in humans, have not yet been identified. Based on the results obtained in the okadaic acid and β-endosulfan treatments, it was clearly demonstrated that the HepG2 transformant system was able to identify the CYP isoform relating to the genotoxicity of chemical metabolite(s) and was useful to elucidate the genotoxicity of a new chemical or a drug candidate in the presence of the metabolic

In chapter 2.3, the relevance of the interaction between phase I and phase II drugmetabolizing enzymes, e.g., UGT, GST, and SULT, in the test system was demonstrated in a MN test of tamoxifen or safrole, which has been reported to be metabolized by enzymes of both phases. Based on the results for tamoxifen and safrole, it was clearly demonstrated that the interaction between the phase I and phase II drug-metabolizing enzymes was crucial to

alternative for metabolic activation system in the in vitro genotoxicity testing.

expression of human SULT1A1, but not in control cells. 2-Amino-3-methylimidazo[4,5-f] quinoline induced gene mutations in V79 cells co-expressing human NAT2 or NAT1 together with human CYP1A2, even at a concentration of 0.01 and 1 μM, respectively, but was inactive (even at 30 μM) in cells expressing only CYP1A2 (Glatt, 2005; Glatt, 2006).

Genotoxicity is a branch of the field of toxicology that assesses the effects of chemicals on DNA or genetic processes of living cells. Such effects can be accessed directly by measuring the interaction of chemicals with DNA or more indirectly through the production of gene mutation or chromosome alterations. The observations of these consequences in the genotoxicity tests suggest the carcinogenic concern of a chemical. Thus it is important to improve the genotoxicity test system to evaluate accurately based on the in vivo situation in human as much as possible. In the present research, I tried to imitate human metabolism by using human hepatocyte cell line expressing human CYP enzymes. My results indicated that metabolism focused only on CYP was not sufficient to evaluate the genotoxicity of the chemicals such as tamoxifen and safrole. A comprehensive metabolic pathway not only by phase I drug-metabolizing enzymes but also by phase II enzymes would be needed for the accurate assessment of genotoxicity. Moreover, other cellular defense systems (i.e., antioxidant system, GSH or ascorbic acid, and DNA repair system) are involved in the expression of genotoxicity by a chemical. Despite the proof that most chemical carcinogens undergo metabolic conversion into DNA-reactive intermediates, some compounds do not bind to DNA and are not mutagenic, yet they are carcinogenic in animal models and possibly also in humans. These non-genotoxic mechanisms such as induction of inflammation, immunosuppression, formation of reactive oxygen species, activation of receptors such as arylhydrocarbon receptor or estrogen receptor, and epigenetic silencing. Therefore ,another approach based on the non-genotoxic mechanism is necessary to predict the carcinogenic action from a certain chemical. Together, these genotoxic and nongenotoxic mechanisms can alter signal-transduction pathways that finally result in hypermutability, genomic instability, loss of proliferation control, and resistance to apoptosis — some of the characteristic features of cancer cells. In this regard, we need to learn much more about the role and interplay of susceptibility and resistance function targeted by human carcinogens or involved in modulating human responses to carcinogenic chemicals.

#### **3. Future considerations**

Genotoxicity is a branch of the field of toxicology that assesses the effects of chemicals on DNA or genetic processes of living cells. Such effects can be accessed directly by measuring the interaction of chemicals with DNA or more indirectly through the production of gene mutation or chromosome alterations. The observations of these consequences in the genotoxicity tests suggest the carcinogenic concern of a chemical. Thus it is important to improve the genotoxicity test system to evaluate accurately based on the in vivo situation in human as much as possible. In the present research, I tried to imitate human metabolism by using human hepatocyte cell line expressing human CYP enzymes. My results indicated that metabolism focused only on CYP was not sufficient to evaluate the genotoxicity of the chemicals such as tamoxifen and safrole. A comprehensive metabolic pathway not only by phase I drug-metabolizing enzymes but also by phase II enzymes would be needed for the accurate assessment of genotoxicity. Moreover, other cellular defense systems (i.e., antioxidant system, GSH or ascorbic acid, and DNA repair system) are involved in the expression of genotoxicity by a chemical. Despite the proof that most chemical carcinogens undergo metabolic conversion into DNA-reactive intermediates, some compounds do not bind to DNA and are not mutagenic, yet they are carcinogenic in animal models and possibly also in humans. These non-genotoxic mechanisms such as induction of inflammation, immunosuppression, formation of reactive oxygen species, activation of receptors such as arylhydrocarbon receptor or estrogen receptor, and epigenetic silencing. Therefore ,another approach based on the non-genotoxic mechanism is necessary to predict the carcinogenic action from a certain chemical. Together, these genotoxic and nongenotoxic mechanisms can alter signal-transduction pathways that finally result in hypermutability, genomic instability, loss of proliferation control, and resistance to apoptosis — some of the characteristic features of cancer cells. In this regard, we need to learn much more about the role and interplay of susceptibility and resistance function targeted by human carcinogens or involved in modulating human responses to carcinogenic chemicals.
