Toxicology 216 (2005) 154–167
An in vitro approach to detect metabolite toxicity due to CYP3A4-dependent bioactivation of xenobiotics Luisella Vignati ∗, Elisa Turlizzi 1, Sonia Monaci, Pietro Grossi, Ruben de Kanter, Mario Monshouwer 2
Department of Pre-Clinical Development, Nerviano Medical Sciences S.r.l., V.le Pasteur, 10, 20014, Nerviano, MI, Italy Received 22 June 2005; received in revised form 3 August 2005; accepted 3 August 2005 Available online 19 September 2005
Abstract Many adverse drug reactions are caused by the cytochrome P450 (CYP) dependent activation of drugs into reactive metabolites. In order to reduce attrition due to metabolism-mediated toxicity and to improve safety of drug candidates, we developed two in vitro cell-based assays by combining an activating system (human CYP3A4) with target cells (HepG2 cells): in the first method we incubated microsomes containing cDNA-expressed CYP3A4 together with HepG2 cells; in the second approach HepG2 cells were transiently transfected with CYP3A4. In both assay systems, CYP3A4 catalyzed metabolism was found to be comparable to the high levels reported in hepatocytes. Both assay systems were used to study ten CYP3A4 substrates known for their potential to form metabolites that exhibit higher toxicity than the parent compounds. Several endpoints of toxicity were evaluated, and the measurement of MTT reduction and intracellular ATP levels were selected to assess cell viability. Results demonstrated that both assay systems are capable to metabolize the test compounds leading to increased toxicity, compared to their respective control systems. The co-incubation with the CYP3A4 inhibitor ketoconazole confirmed that the formation of reactive metabolites was CYP3A4 dependent. To further validate the functionality of the two assay systems, they were also used as a “detoxification system” using selected compounds that can be metabolized by CYP3A4 to metabolites less toxic than their parent compounds. These results show that both assay systems can be used to screen for metabolic activation, or de-activation, which may be useful as a rapid and relatively inexpensive in vitro assay for the prediction of CYP3A4 metabolism-mediated toxicity. © 2005 Published by Elsevier Ireland Ltd.
Keywords: Metabolism-mediated toxicity; Adverse drug reactions (ADRs); Reactive metabolites; In vitro screening; CYP3A4; Cytotoxicity
1. Introduction
Abbreviations: BSO, l-buthionine S,R-sulphoximine; CYP, Adverse reactions associated with exposure of indi- cytochrome P450; DMEM, Dulbecco’s Modified Eagle’s Medium; viduals to drugs (ADRs) or xenobiotics are a com- DMSO, dimethyl-sulphoxide; GSH: reduced glutathione; MTT, 3- (4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide; PBS, mon and a significant cause of morbidity and mortality phosphate buffered saline (Lazarou et al., 1998; Meyer, 2000). From a clinical ∗ Corresponding author. Tel.: +39 0331 581335; perspective, ADRs may be classified as augmented reac- fax: +39 0331 581105. tions (type A), which are predictable from the known E-mail address: [email protected] (L. Vignati). pharmacology and often represent an exaggeration of 1 Present address: Drug profiling, Sienabiotech S.p.A., via Fiorentina, 1, 53100 Siena, Italy. the pharmacological effects of the drug that may be 2 Present address: Pharmacokinetics and Drug Metabolism, Amgen reversed by dose reduction, or as idiosyncratic reaction Inc., 1120 Veterans Blvd, South San Francisco, CA 94080, USA. (type B), which are unpredictable from the knowledge
0300-483X/$ – see front matter © 2005 Published by Elsevier Ireland Ltd. doi:10.1016/j.tox.2005.08.003 L. Vignati et al. / Toxicology 216 (2005) 154–167 155 of the basic pharmacology of the drug and show marked Over the past years, several approaches have been individual susceptibility and no simple dose dependency. used to detect metabolism-mediated toxicity in vitro. From a chemical point of view, this classification can be Because covalent binding of drugs to proteins has been expanded to reactions (type C), which are predictable associated with drug toxicity (Zhou et al., 2005), it from the chemistry of the drug or xenobiotics, and to is rather common practice within large pharmaceuti- reactions (type D), which are delayed reactions that occur cal companies to determine the extent of irreversible many years after treatment (Park et al., 1998). Pharmaco- binding to protein using radiolabeled drug and human genetic, toxicogenetic and other host-dependent factors liver microsomes (Day et al., 2005; Kitteringham et al., have been identified to be important in predisposition 1988). to type A reactions (Meyer and Gut, 2002). In contrast, Human hepatocytes represent another in vitro sys- most type B, type C and type D ADRs are mediated by tem for the evaluation of metabolism mediated toxicity toxic metabolites (Gut et al., 1995; Park et al., 1998, (Gomez-Lechon et al., 2003; Li et al., 1999), but the poor 2004). availability of human liver, the high cost and the sig- There are many examples of metabolites or reactive nificant variability among human hepatocytes prepara- intermediates of essentially non-toxic drugs or chemicals tions make this tool not applicable to an high-throughput that have been shown to exert adverse drug reactions. screening in drug discovery. More recently, cell lines that Some of these ADRs are so serious that drug treatment have been genetically modified to express a single or had to be limited and in some cases the drug had to multiple drug-metabolising enzyme(s) have been devel- be withdrawn from the market with enormous conse- oped (Bull et al., 2001; Dai and Cederbaum, 1995; Lin et quences for both patients and pharmaceutical industries al., 1999; Nakagawa et al., 1996; Philip et al., 1999; Wu (e.g. troglitazone, zomepirac and benoxaprofen) (Nassar and Cederbaum, 1996). However, most of these stably and Lopez-Anaya, 2004; Wolfgang and Johnson, 2002). transfected cell lines have very low metabolic activity, Because of this, pharmaceutical companies are making compared to hepatocytes or liver microsomes. Finally, significant efforts to predict ADRs and, currently, several the immortalized human hepatocyte cell line Fa2N-4 has tools and strategies are considered to address the for- been recently described as an in vitro tool to study drug mation of reactive metabolites and the potential conse- metabolism issues (Mills et al., 2004), but like for sta- quence for the overall safety profile on candidate drugs. bly transfected cell lines, also the use of Fa2N-4 cells to Reactive metabolites are a common product of phase address reactive metabolites is limited due to low enzy- I oxidation reactions mediated by cytochrome P450 matic activities. (CYP)-dependent mixed function oxygenases, although In the present study, we describe the development of also examples of other phase I (e.g. flavin-mono oxyge- two in vitro models to evaluate the activation of xenobi- nases; FMOs) and phase II drug metabolizing reactions otics to toxic metabolites. In the first method, we used have been described (Zhou et al., 2005). The genera- human CYP3A4 cDNA expressed microsomes (as acti- tion of such reactive metabolites may produce adverse vating system), in combination with HepG2 cells (as reactions via different inter-related process such as for- target system). mation of free radicals, oxidation of thiols and covalent In the second model, a HepG2 cell line transiently binding with endogenous macromolecules, resulting in transfected with CYP3A4 was developed. Transient the oxidation of cellular components or inhibition of nor- transfection was preferred above stable transfection, mal cellular functions (Riley et al., 1988). Sometimes, because of the higher metabolic activity observed in and for unknown reasons, covalently modified proteins these cells. may be immunogenic and elicit an immune response HepG2 cells were selected as target cells for eval- (Knowles et al., 2000; Park et al., 2000; Uetrecht, 1999). uating toxicity because they are derived from human In contrast with adverse reactions that are usually dose liver and have been extensively used as the test system dependent and due to the pharmacology of the drug (e.g. for the prediction of toxicity, carcinogenicity and cell type A), ADRs caused by toxic metabolites are diffi- mutagenicity in humans. Moreover, HepG2 cells con- cult to predict accurately because metabolic activation of tain the co-enzymes NADPH-cytochrome P450 reduc- the drug is required to observe the toxic effect. In addi- tase and cytochrome b5, required for CYP mediated drug tion, significant species differences in drug metabolism, metabolism (Rodriguez-Antona et al., 2002; Yoshitomi chemical instability of the reactive metabolites, and the et al., 2001). presence of detoxification pathways are just some of Because of the general importance of CYP3A4 in the several factors, which make the screening for toxic drug metabolism (50% of the drugs in use today are metabolites a real challenge. metabolised by CYP3A4) (Guengerich, 2001; Nelson, 156 L. Vignati et al. / Toxicology 216 (2005) 154–167
Table 1 Test compounds and their metabolites Compound tested (Main) toxic metabolite ADR Therapeutic use Reference
Albendazole Sulfoxide Hepatotoxicity Anthielminthic Rawden et al. (2000) Carbamazepine o-Quinone, iminoquinone Skin rash, hepatic disorders Anticonvulsant Pirmohamed et al. (1992) Dapsone Nitroso-dapsone Hemolysis Antileprosy Coleman (1995) Flutamide Nitroradicals Hepatotoxicity Anticancer Berson et al. (1993) Isoniazid Isonicotinic acid Hepatotoxicity and systemic Anti-tuberculosis Walubo et al. (1998) lupus erithematosus Ochratoxin A Quinones Nephrotoxicity Food contaminant Simarro Doorten et al. (2004) Quinidine 3-Hydroxyquinidine Heart, renal and hepatic failure Antimalarial Jaeger et al. (1987) Tamoxifen ␣-Hydroxytamoxifen Cytotoxicity anti-breast cancer Antiestrogen Holownia and Braszko (2004) Troglitazone Quinone epoxide Hepatotoxicity Antidiabetic Yamamoto et al. (2002) Ziprasidone Unknown Cardiac toxicity Antipsychotic Beedham et al. (2003) Amitriptyline Detoxification by 10-hydroxylation Cardiac toxicity CNS toxicity Antidepressant Triazolam Detoxification by 1 and Hepatotoxicity reproductive Sedative 4-hydroxylation toxicity
1999; Rendic and Di Carlo, 1997) and the several exam- done was synthesized in house. All other test com- ples of the involvement of CYP3A4 in the formation of pounds were obtained from Sigma (St. Louis, MO, reactive metabolites, it was decided to focus on CYP3A4 USA). as an activating system. In both systems, the metabolic activity of CYP3A4 2.2. Cell culture was determined by incubation with testosterone and midazolam, two well known substrate markers of All cell culture media and supplements were pur- CYP3A4 metabolism. chased from Invitrogen (Carlsbad, CA, USA). The For the detection of any effect due to toxic metabo- human hepatic carcinoma cell line HepG2 was obtained lites, reproducible and sensitive endpoints are fundamen- from the European Collection of Animal Cell Cul- tal. After evaluation of the available assays it was decided tures (ECACC no. 85011430, Porton Down, UK) and that the MTT and ATP were the most suitable endpoints. cells were used between passage 25 and 28. HepG2 The CYP3A4 inhibitor ketoconazole, control micro- cells were routinely cultured in 75 cm2 vented tis- somes and a control vector plasmid were included to sue culture flasks (Corning Inc., Corning, NY, USA) make sure that toxic effects were due to CYP3A4 activ- using Dulbecco’s Modified Eagle’s Medium (DMEM) ity. containing 4500 mg/l glucose, supplemented with 10% To minimize the effect of detoxification, in particu- heat-inactivated fetal calf serum, 100 units/ml penicillin, lar the role of reduced glutathione (GSH), HepG2 cells 100 g/ml streptomycin and 2 mM l-glutamine and were treated with l-buthionine S,R-sulphoximine (BSO) ◦ maintained in a cell culture incubator at 37 Cin5% in order to deplete GSH levels. CO . To verify the applicability of both models ten com- 2 pounds that require metabolic activation to manifest their toxic effects were evaluated. In addition, two compounds 2.3. Construction of the expression vector known to be metabolised by CYP3A4 into non toxic pcDNA3.1myc/HisB-CYP3A4 metabolites were included as well. Compounds tested and their reactive metabolites are reported in Table 1 The coding region sequence of human CYP3A4 was and compound chemical structures are shown in Fig. 1. obtained from NCBI’s Nucleotide sequence database (accession no. AF182273) and cDNA was generated by RT-PCR from RNA extracted from human liver 2. Materials and methods with the use of the kit Superscript RT-PCR (Invitrogen) following the recommendations of the manufacturer. 2.1. Chemicals The forward and reverse oligonucleotide sequences were: Troglitazone was purchased from Biomol Research 5-CCGGTACCGATGGCTCTCATCCCAGAC-3 Laboratories (Plymouth Meeting, PA, USA); ziprasi- and 5-GCTCTAGAGAGGCTCCACTTACGG-3, res- L. Vignati et al. / Toxicology 216 (2005) 154–167 157
Fig. 1. Chemical structures of test compounds.
pectively. The amplified product was cloned into 2.4. GSH measurement and depletion the KpnI-XbaI site of pcDNA3.1myc/HisB vector (Invitrogen). The plasmid obtained was purified using To assess the intracellular content of glutathione Qiagen Endofree Maxiprep (Qiagen, Chatsworth, CA, (GSH), HepG2 cells were plated in 25 cm2 vented tis- USA) according to the manufacturer’s instructions. sue culture flasks (Corning) and incubated with DMEM 158 L. Vignati et al. / Toxicology 216 (2005) 154–167 alone or DMEM with l-buthionine S,R-sulphoximine at or pcDNA3.1myc/HisB-control vector and with a final concentration of 50 M for 24 and 48 h and at 0.2 g/well of pCMV*SPORT -Gal plasmid (Invit- 100 M for 24, 48 and 72 h. GSH content was mea- rogen) using Lipofectamine Plus reagent (Invitrogen) sured using the ApoAlert® Glutathione Detection Kit according to the recommendations of the manufac- (Clontech, Palo Alto, CA, USA) as described by the turer and left for further 24 h in the cell incubator. manufacturer. The amount of GSH in experimental sam- The efficiency of the transfection was determined by ples was calculated from a standard curve prepared with measuring the expression of -galactosidase according GSH and expressed as nmol/106 cells. Cell number was to the instructions of -Gal assay Kit (Invitrogen). determined by trypan blue exclusion method. Cells transfected with pcDNA3.1myc/HisB-CYP3A4 will be named, for shortness, HepG2-3A4 in the next 2.5. SupersomesTM-HepG2 cells co-culture assay sections of the paper and control cells transfected with pcDNA3.1myc/HisB-control vector will be named Human CYP3A4 SupersomesTM and Insect Cell Con- HepG2-control. trol SupersomesTM were obtained from BD-Gentest Test compounds were added as described for the (Woburn, MA, USA). At cell confluency of 80–85%, SupersomesTM-HepG2 co-culture assay. After adding HepG2 cells were washed twice with phosphate buffered the chemicals, hepG2-3A4 and HepG2-control cells saline (PBS) and detached with 0.25% trypsin-EDTA, were incubated for 24 h before analyzing the cytotox- resuspended in DMEM containing 10% fetal calf serum, icity of the cell cultures. and seeded in 96-well cell-culture treated plates (Corn- ing) at a density of 1 × 105 cells/well. Cells were allowed 2.7. Preparation of microsomes from transiently to attach for 4 h before treatment with 100 M BSO for transfected HepG2 cells 48 h. Solutions of test compounds, at different concentra- Microsomes were prepared from HepG2-3A4 and tions, were prepared in DMSO and diluted in DMEM HepG2-control cells after scraping the cells at 80–85% containing 100 M BSO, but without serum. The final confluency from four 75 cm2 tissue culture flasks and concentration of DMSO was kept constant in each sam- homogenizing in ice-cold 0.1 M phosphate buffer (pH ple at 0.1% (v/v) for all the test compounds except for car- 7.4) containing 0.15 M KCl and 1 mM EDTA with the bamazepine (0.5% v/v DMSO). Compound preparations use of a Potter-Elvehjem homogenator and centrifuga- were added to the cells in the presence of 20 pmol/ml tion at 9000 × g for 20 min. The supernatant obtained human CYP3A4 or Insect Cell Control SupersomesTM was centrifuged at 100,000 × g for 90 min; the pellet was and 1 mM of NADPH as a co-factor. The final volume then washed in phosphate buffer without EDTA and cen- of each well was 200 l. In every experiment cells incu- trifuged again at 100,000 × g for 90 min. Microsomes bated without compounds (DMEM alone) and cells incu- were frozen in liquid nitrogen and stored at −80 ◦C until bated with 10% cell lysis buffer (Triton® × 100, Sigma) use. were included as controls for cell viability tests. DMSO at 0.1% v/v (or 0.5% v/v) in the absence of test com- 2.8. Determination of catalytic activity of human pounds was also included as control. Each test compound CYP3A4 concentration (in five-fold) was incubated for 1 h at 37 ◦C under gentle agitation, followed by an additional 24 h in Transiently transfected HepG2 cells and microsomes ◦ a cell culture incubator at 37 C and 5% CO2. At the end prepared from these cells were both incubated with of the incubation period cells were analyzed for cytotox- 250 M testosterone and 10 M midazolam for 4 h (for icity. transfected cells) or 10 min (for microsomes incubated in phosphate buffer containing 1 mM NADPH) and 2.6. Cells transient transfection and compounds metabolites from medium samples were extracted with treatment acetonitrile (1:1 v/v). The supernatants of centrifuged samples were ana- HepG2 cells were plated onto 96-well cell-culture lyzed by LC–MS/MS, using a Turbo ion Spray treated plates (Corning) at a density of 1 × 105 cells/well Source and a Triple Quadrupole API 4000 instru- and allowed to recover for 4 h before treatment with ment (Perkin-Elmer, Woodbridge, Canada). A 4.6 (inner 100 M BSO. diameter) × 12.5 mm C8 column (Zorbax, Agilent Tech- After 24 h cells were transiently transfected nologies) was applied. A mobile phase containing with 0.2 g/well of pcDNA3.1myc/HisB-CYP3A4 10 mM ammonium formate (pH 4.0) and acetonitrile, L. Vignati et al. / Toxicology 216 (2005) 154–167 159 was used; for midazolam, acetonitrile was increased fied by the manufacturer (ATPLite, Packard Bioscience, from 5 to 95% within 0.4 min and then back to 5% in Meriden, CT, USA). After an incubation period of 24 h 1.4 min. The flow rate was 1.5 ml/min for the first 0.2 min with test compounds, HepG2 cells were lysed and total to equilibrate the column quickly and after 0.2 min after ATP content per well was determined using a TopCount injection the flow rate was reduced to 0.6 ml/min. For plate reader (Packard Instrument Company, Meriden, testosterone, acetonitrile in the eluens was increased CT, USA). from 5 to 50% during the first 6 min using a flow of 0.5 ml/min and then back to 5% in 2 min. After injection 2.11. Statistical analysis of 20 l of the sample ion spectra were acquired in posi- tive mode using Q1 → Q3 of m/z 342 → 324, 324 → 234, All data are given as means ± standard deviation 305 → 269 for 1-OH midazolam, 4-OH midazolam and (S.D.) and are average values from five values per exper- 6-OH testosterone, respectively, and the quantification iment; experiments were repeated at least twice. Sta- was performed by comparing the peak areas with authen- tistical evaluation among groups was carried out using tic standards of each metabolite. Data were expressed in two-tailed Student’s t-test, and p < 0.05 was considered pmol/min/mg protein following determination of cellu- significant. lar or microsomal protein measured using the BioRad Protein assay Kit (BioRad, Munchen,¨ Germany). 3. Results
2.9. Inhibition of CYP3A4 activity 3.1. Cell viability
Test compounds incubated with the SupersomesTM- Increasing concentrations of cell lysis buffer (10% HepG2 cells system and with transiently transfected cells Triton® X100), a non-selective cell damaging agent, were also co-incubated in the presence of 1 M keto- were used to evaluate the applicability of MTT and ATP conazole to further characterize the role of CYP3A4 for as cytotoxicity endpoints. Cell viability, expressed as the observed toxic effects. percentage of non-treated cells (control: 100% viabil- ity) in HepG2 cells after 24 h incubation with increasing 2.10. Cell viability tests concentrations of cell lysis buffer is shown in Fig. 2A. For both endpoints a dose-dependent decrease in cell 2.10.1. MTT viability was observed. MTT (3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl In order to establish the maximal tolerable dose tetrazolium bromide), a yellow tetrazolium salt, can be of DMSO, HepG2 cells were incubated with increas- reduced mainly by the mitochondria of metabolically ing concentrations of DMSO. HepG2 cells viability active cells to form a blue formazan dye precipitate that appeared to be higher than 80% when DMSO concentra- can be extracted using organic solvents. tion was lower than 0.5% (v/v), as shown in Fig. 2B. For After treatment with test compounds, culture medium this reason, DMSO concentration was kept lower than to was replaced with 100 l/well of a 0.5 mg/ml solution of 0.5% (v/v) during all the experiments. MTT in PBS and cells were incubated for 2 h at 37 ◦C. At the end of this period MTT solution was removed and 3.2. Enzyme activity 50 l of isopropanol was added to each well. The plates were shaken to solubilize the blue formazan produced To be sure that CYP3A4 transfection of HepG2 cells and absorbance was measured at a wavelength of 570 nm resulted in an active CYP3A4 enzyme, testosterone with background subtraction at 690 nm in a plate reader. and midazolam hydroxylase activities were measured. Control refers to incubations in the absence of test com- For this, transiently transfected HepG2 cells were incu- pounds, in the presence of vehicle only (DMSO 0.1%) bated for 4 h with testosterone (250 M) and midazolam and was considered as 100% viability value. (10 M) as markers substrates for CYP3A4 activity. CYP3A4 mediated enzymatic activities for testosterone 2.10.2. ATP and midazolam, expressed as pmol of product/min/mg The ATP assay is based on the detection of a chemi- of cellular protein are reported in Table 2. luminescent signal proportional to cellular ATP levels To further characterize the enzymatic activities, caused by a luciferin–luciferase reaction. The amount also microsomes obtained from transiently transfected of ATP is directly proportional to the number of viable HepG2 cells were incubated with 250 M testos- cells in culture. The assay was performed as speci- terone and 10 M midazolam for 10 min. Results 160 L. Vignati et al. / Toxicology 216 (2005) 154–167
3.3. GSH content
To minimize the effect of detoxification due to GSH trapping of reactive metabolites, HepG2 cells were treated first with BSO to deplete GSH levels. After a 48 h treatment with 50 and 100 M of BSO, GSH lev- els were reduced by 74 and 98%, respectively (Table 4). Despite a 20% decrease in cell viability due to BSO at 100 M (data not shown), it was decided that for the pur- pose of this study, a 48 h incubation of HepG2 cells with 100 M of BSO was the most appropriate.
3.4. Incubation of CYP3A4 substrates with SupersomesTM-HepG2 cells system
Ten CYP3A4 substrates, known to generate toxic metabolites, were incubated at different concentra- tions for 24 h in the presence of human CYP3A4 SupersomesTM or Insect Cell Control SupersomesTM following the procedure as described in Section 2. The effect of each test compound on cell viability is presented in Fig. 3. Because the ATP and MTT data were comparable, only MTT data is presented and differences between ATP and MTT are highlighted in Fig. 3. For all test compounds except carbamazepine, the concentra- Fig. 2. HepG2 cell viability is expressed as percent of viability tion dependent decrease in cell viability was more pro- in respect to medium only. MTT, 3-(4,5-dimethyl-thiazol-2-yl)-2,5- diphenyl tetrazolium bromide. nounced in the presence of CYP3A4. For carbamazepine the presence of CYP3A4 did not have an additional effect demonstrated that enzymatic activities were comparable on loss of cell viability. For isoniazid, troglitazone and with testosterone hydroxylase and midazolam hydrox- tamoxifen at high concentrations, the effect of CYP3A4 TM ylase activities obtained with CYP3A4 Supersomes was not visible anymore due to significant toxicity of (Table 3). parent compound.
Table 2 CYP3A4 activity in transiently transfected HepG2 cells Testosterone Midazolam
6-Hydroxylation 1-Hydroxylation 4-Hydroxylation
HepG2-3A4 51.2 1.96 0.38 HepG2-control Data are expressed in pmol/min/mg cellular protein. Table 3 CYP3A4 activity in microsomes obtained from transiently transfected HepG2 cells and in SupersomesTM Testosterone Midazolam 6-Hydroxylation 1-Hydroxylation 4-Hydroxylation HepG2-3A4 microsomes 16922 1094 109 HepG2-control microsomes Data are expressed in pmol/min/mg microsomal protein. L. Vignati et al. / Toxicology 216 (2005) 154–167 161 Table 4 Effect of BSO on the depletion of GSH in HepG2 cells 0 M BSO (control) 50 M BSO 100 M BSO nmol/106cells nmol/106cells % reduction nmol/106cells % reduction 24 h 26.7 ± 0.11 9.7 ± 0.06 63.7 3.8 ± 0.02 85.8 48 h 25.9 ± 0.08 6.8 ± 0.12 73.7 0.6 ± 0.07 97.7 72 h 26.9 ± 0.10 n.d. n.d. 1.2 ± 0.09 95.5 Values represent mean of three independent determinations ± S.D. Percent of reduction was calculated vs. control at the correspondent time point. n.d., not determined. GSH, reduced glutathione; BSO, l-buthionine S,R-sulphoximine; S.D., standard deviation. To further confirm, the involvement of CYP3A4 in the a major cause of adverse drug reactions observed in formation of toxic metabolites, all test compounds were the clinic (Gad, 2003). As a result of this, both regu- co-incubated with 1 M of ketoconazole, a well known latory authorities as well as pharmaceutical industries, inhibitor of CYP3A4 activity. For all test compounds are currently paying significant attention to address incubated in the presence of CYP3A4, the decrease in metabolism-mediated toxicity. cell viability was prevented by ketoconazole. Over the past years, several methods have been pro- posed to predict the potential of chemicals to form toxic 3.5. Incubation of CYP3A4 substrates with metabolites, but until now there is no predictive tool transiently transfected HepG2 cells available that is widely applicable in the early phase of drug discovery. HepG2 transiently transfected with pcDNA3.1myc/ Many reactive metabolites are unstable and therefore HisB-CYP3A4 (HepG2-3A4) or pcDNA3.1myc/HisB- direct detection is very difficult. However, many of these control (HepG2-control) plasmids were incubated with unstable and highly reactive metabolites are able to bind the same ten compounds as described above. Results covalently to proteins. Several methods are available to obtained with transfected cells were in agreement detect or quantify covalent binding of drugs and their TM with data obtained from the Supersomes -HepG2 metabolites to macromolecules, including radiochemi- cells system, with the exception of carbamazepine cal and immunological methods. The covalent binding (Fig. 4). For all compounds, including carbamazepine, and formation of drug-protein adducts are generally con- the concentration-dependent decrease in cell viability sidered to be related to drug toxicity, and selective protein was greater in the CYP3A4 transfected HepG2 cells covalent binding by drug metabolites may lead to selec- than in control. Ketoconazole confirmed the involvement tive organ toxicity. However, the mechanisms involved of CYP3A4, and co-incubation of test compounds with in the protein adduct-induced toxicity are still largely ketoconazole decreased the CYP3A4 mediated toxicity. undefined (Evans et al., 2004; Nassar and Lopez-Anaya, 2004; Zhou et al., 2005). 3.6. CYP3A4 detoxification activity Another way to investigate the relevance of reactive and/or toxic metabolites is by looking at the toxicological To confirm the validity of the two described systems, consequences of these metabolites. In the early nineties, compounds known to be more toxic than their respective Riley et al. (1990) described how cytotoxicity could be metabolites were also incubated. Fig. 5 shows the toxic assessed by co-incubation of drug and liver microsomes effect of amitriptyline and triazolam in SupersomesTM- with human mononuclear leucocytes, which served as HepG2 cells system and transiently transfected HepG2 target cells. However, these studies required a high cells. MTT and ATP showed again comparable results, amount of liver microsomes and a two compartments and therefore only the MTT is presented. For both sys- system, comprising two adjacent teflon chambers sepa- tems, the presence of CYP3A4 reduced the toxic effect rated by a semi-permeable membrane. More recently, in of both compounds. Co-incubation of amitriptyline and vitro systems based on cell-lines expressing CYP iso- triazolam with 1 M ketoconazole confirmed the role of forms have been described as a useful tool to detect CYP3A4. cytotoxicity related to the production of reactive or cyto- 4. Discussion toxic metabolites. In these models the metabolic activity of transfected cells is in general very low and there is The biotransformation of xenobiotics to reactive a significant risk that the amount of toxic metabolite intermediates or toxic metabolites is considered to be formed is too low to cause any measurable toxic effect 162 L. Vignati et al. / Toxicology 216 (2005) 154–167 Fig. 3. MTT test in HepG2 cells after 24 h exposure to different concentrations of test compounds with human CYP3A4 SupersomesTM or Insect Cell Control SupersomesTM. Cell viability is expressed as percentage of solvent only treated cells: 0.1% DMSO except for carbamazepine: 0.5% DMSO. Each bar represents the mean ± S.D. of five separate determinations of a representative experiment. Data were analyzed using Student’s t-test (*p < 0.05; **p < 0.01 vs. corresponding Insect Cell Conrol SupersomesTM for both MTT and ATP assay). (§) ATP significantly different. (#) ATP non significantly different. KCZ, 1 M ketoconazole; MTT, 3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide; DMSO, dimethyl sulfoxide; S.D., standard deviation. L. Vignati et al. / Toxicology 216 (2005) 154–167 163 Fig. 4. MTT test in HepG2-3A4 and HepG2-control cells after 24 h exposure to different concentrations of test compounds. Cell viability is expressed as percentage of solvent only treated cells: 0.1% DMSO except for carbamazepine: 0.5% DMSO. Each bar represents the mean ± S.D. of five separate determinations of a representative experiment. Data were analyzed using Student’s t-test (*p < 0.05; **p < 0.01 vs. corresponding HepG2-control for both MTT and ATP assay). (§) ATP significantly different. (#) ATP non significantly different. KCZ, 1 M ketoconazole; MTT, 3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide; DMSO, dimethyl sulfoxide; S.D., standard deviation. 164 L. Vignati et al. / Toxicology 216 (2005) 154–167 Fig. 5. MTT test in HepG2 cells after co-incubation for 24 h of amitriptyline and midazolam with human CYP3A4 or Insect Cell Control SupersomesTM (panel A) and with transiently transfected HepG2 cells (panel B). Cell viability is expressed as percent of control response (DMSO 0.1% v/v). The co-incubation of 1 M ketoconazole was considered to confirm CYP3A4-dependent de-activation of compounds. Each bar represents the mean ± S.D. of five separate determinations of a representative experiment. Data were analyzed using Student’s t-test (*p < 0.05; **p < 0.01 vs. corresponding CYP3A4 expressing system for both MTT and ATP assay). (#) ATP non significantly different. KCZ, 1 M ketoconazole; MTT, 3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide; DMSO, dimethyl sulfoxide; S.D., standard deviation. (Valentine et al., 1996). Improved metabolic activities lines and were comparable to the activities reported for have been obtained by expressing CYP3A4 in conjunc- example with V79 and NIH-3T3 cells co-transfected tion with human NADPH-cytochrome P450 reductase with NADPH-cytochrome P450 reductase (Schneider and a few successful examples of metabolism mediated et al., 1996; Simarro Doorten et al., 2004) and in toxicity screening have been described by using these human hepatocytes (Kostrubsky et al., 1999; Roymans cell-lines (Schneider et al., 1996; Simarro Doorten et et al., 2005; Venkataramanan et al., 2000; Wen et al., al., 2004; Valentine et al., 1996; Yoshitomi et al., 2001). 2002). The enzymatic activities observed in CYP3A4 In this study, two in vitro cell-based bioactivation SupersomesTM were in agreement with the activities screening assays are described. In one system the bioac- assured by the supplier. It is obvious that metabolite for- tivation of xenobiotics was mediated by CYP3A4 cDNA mation rate in both systems will be different and it is expressed microsomes (SupersomesTM), whereas in the expected that the rate of metabolism in CYP3A4 tran- other system CYP3A4 transiently transfected HepG2 siently transfected HepG2 will be linear over a much cells were used for bioactivation. For both in vitro sys- longer period as compared to CYP3A4 SupersomesTM. tems the cytotoxicity was evaluated in HepG2 cells as Although other well known viability assays, such as target cells. neutral red uptake and LDH leakage, were considered For the development of these in vitro systems, the as well, MTT reduction and the measurement of ATP following five major requirements were considered; high levels were chosen because of their good sensibility, as enzymatic activities (1), a simple viability/cytotoxicity shown in Fig. 2A and B, and their applicability. Neutral endpoint (2), good sensitivity (3), cytotoxicity mediated red uptake and LDH leakage are in general considered as by a single CYP isoform (4), and lastly, amenable to high late stage markers of toxicity and are very often preceded throughput screening (5). by an effect on MTT and/or ATP. Therefore, a longer The metabolic activity of CYP3A4 transiently trans- incubation period would be required to observe an effect fected HepG2 cells was evaluated by testosterone and on neutral red uptake and/or LDH leakage. midazolam hydroxylation. Enzymatic activities were To address the sensitivity requirement of the in vitro significant higher as compared to stably transfected cell assays, it was decided to treat HepG2 cells with BSO L. Vignati et al. / Toxicology 216 (2005) 154–167 165 to deplete GSH levels. GSH is known to have a protec- cells has a high level of flexibility. The CYP3A4 cDNA tive effect on metabolism mediated toxicity by trapping expressed microsomes can be easily replaced by other reactive metabolites and as a consequence, high levels CYP isoform or other (non CYP) drug metabolizing of GSH would reduce the sensitivity of these cell based enzymes, like flavin-mono oxygenases or glucuronosyl- systems (Dai and Cederbaum, 1995; Zhang et al., 2001). transferases. Also the HepG2 cells can be substituted by The advantage of a metabolic activation system based another cell line or even primary cell cultures. on only CYP3A4 allows to relate the observed toxic Although, the transfection model is less flexible, the effect directly to this human CYP isoform. In this study, fact that this assay do not require a stable cell line, allows the CYP3A4 involvement was further confirmed by modifying the assay by simply transfecting (transiently) using ketoconazole as an inhibitor of CYP3A4 medi- with a plasmid of a different CYP isoform. ated metabolism. To know which enzyme is responsible The applicability of these models is not limited to the of the formation of a toxic metabolite may be particular identification of reactive/toxic metabolites and currently useful in the case of drug-drug interactions: compounds the possibility to use these cell based systems to identify that induce metabolizing enzyme activity may further pharmacological active metabolites are explored in our enhance the toxicity of co-administered drugs if these department as well. are activated by the same enzyme. Finally, the model systems described here are rapid, The results obtained using the two screening assays relatively inexpensive and amenable for high-throughput were comparable for all the compounds examined, screening. They can be applied during the early phase of except for carbamazepine. Carbamazepine showed only drug discovery and require neither a large amount of a dose-dependent increase in cytotoxicity after incuba- compound, nor a radiolabeled compound. However, it is tion with CYP3A4 transiently transfected HepG2 cells; important to mention that these methods do not replace in the co-incubated SupersomesTM-HepG2 model car- the available later-stage predictive tools, but are comple- bamazepine did not demonstrate a CYP3A4-dependent mentary. Because, in the end, it will be a combination of toxicity. Carbamazepine is metabolized by CYP3A4 to several assays, which will help us to minimize the risk of form a 2,3-epoxide, an iminoquinone species and a 10,11 a late-stage attrition due to clinical relevant metabolism epoxide that is further metabolized to quinone species mediated toxicity. that are known to be unstable and highly reactive inter- mediates (Miao and Metcalfe, 2003; Zhou et al., 2005). Acknowledgements In the CYP3A4 transiently transfected cells, the forma- tion of this highly reactive intermediate is intracellular The authors wish to thank Ms. Anna Moscone for and this might explain why carbamazepine toxicity was LC–MS analyses, Dr. Roberta Cunati for her support and only observed in CYP3A4 transfected cells. encouragement and Dr. Italo Poggesi for critical reading To ensure the detoxification capacity of the two sys- of the manuscript. tems, the cytotoxicity of amitriptyline and triazolam was evaluated. These two compounds are metabolized References by CYP3A4 into non toxic metabolites, resulting in a reduced toxicity of amitriptyline and triazolam in the Beedham, C., Miceli, J.J., Obach, R.S., 2003. Ziprasidone metabolism, presence of CYP3A4. aldehyde oxidase, and clinical implications. J. Clin. Psychophar- In conclusion, this study demonstrates the applica- macol. 23, 229–232. Berson, A., Wolf, C., Chachaty, C., Fisch, C., Fau, D., Eugene, D., bility of two in vitro cell based systems as a predictive Loeper, J., Gauthier, J.C., Beaune, P., Pompon, D., et al., 1993. screening tool for CYP3A4 mediated toxicity of xeno- Metabolic activation of the nitroaromatic antiandrogen flutamide biotics. by rat and human cytochromes P-450, including forms belonging The combination of the two systems permits to distin- to the 3A and 1A subfamilies. J. Pharmacol. Exp. Ther. 265, 366– guish if a reactive or cytotoxic metabolite is sufficiently 372. Bull, S., Langezaal, I., Clothier, R., Coecke, S., 2001. A Genetically long-lived to escape and diffuse from the site of for- engineered cell-based system for detecting metabolism-mediated mation, penetrate the cell membrane, gain access to its toxicity. Altern. Lab. Anim. 29, 703–716. cellular target and lead in this way to cytotoxicity at some Coleman, M.D., 1995. Dapsone toxicity: some current perspectives. distance away, or, alternatively, if it is so reactive that its Gen. Pharmacol. 26, 1461–1467. cytotoxic action may be observed only at the site of pro- Dai, Y., Cederbaum, A.I., 1995. Cytotoxicity of acetaminophen in human cytochrome P4502E1-transfected HepG2 cells. J. Pharma- duction (as observed with carbamazepine). col. Exp. Ther. 273, 1497–1505. It is obvious that the system based on a co-incubation Day, S.H., Mao, A., White, R., Schulz-Utermoehl, T., Miller, R., of CYP3A4 cDNA expressed microsomes and HepG2 Beconi, M.G., 2005. A semi-automated method for measuring the 166 L. Vignati et al. / Toxicology 216 (2005) 154–167 potential for protein covalent binding in drug discovery. J. Phar- Nassar, A.E., Lopez-Anaya, A., 2004. Strategies for dealing with reac- macol. Toxicol. Methods 52, 278–285. tive intermediates in drug discovery and development. Curr. Opin. Evans, D.C., Watt, A.P., Nicoll-Griffith, D.A., Baillie, T.A., 2004. Drug Discov. Dev. 7, 126–136. Drug-protein adducts: an industry perspective on minimizing the Nelson, D.R., 1999. Cytochrome P450 and the individuality of species. potential for drug bioactivation in drug discovery and development. Arch. Biochem. Biophys. 369, 1–10. Chem. Res. Toxicol. 17, 3–16. Park, B.K., Kitteringham, N.R., Maggs, J.L., Pirmohamed, M., Gad, S.C., 2003. Active drug metabolites in drug development. Curr. Williams, D.P., 2004. The role of metabolic activation in drug- Opin. Pharmacol. 3, 98–100. induced hepatotoxicity. Annu. Rev. Pharmacol. Toxicol.. Gomez-Lechon, M.J., Donato, M.T., Castell, J.V., Jover, R., 2003. Park, B.K., Kitteringham, N.R., Powell, H., Pirmohamed, M., 2000. Human hepatocytes as a tool for studying toxicity and drug Advances in molecular toxicology-towards understanding idiosyn- metabolism. Curr. Drug Metab. 4, 292–312. cratic drug toxicity. Toxicology 153, 39–60. Guengerich, F.P., 2001. Common and uncommon cytochrome P450 Park, B.K., Pirmohamed, M., Kitteringham, N.R., 1998. Role of drug reactions related to metabolism and chemical toxicity. Chem. Res. disposition in drug hypersensitivity: a chemical, molecular, and Toxicol. 14, 611–650. clinical perspective. Chem. Res. Toxicol. 11, 969–988. Gut, J., Christen, U., Frey, N., Koch, V., Stoffler, D., 1995. Molec- Philip, P.A., Ali-Sadat, S., Doehmer, J., Kocarek, T., Akhtar, A., Lu, ular mimicry in halothane hepatitis: biochemical and structural H., Chan, K.K., 1999. Use of V79 cells with stably transfected characterization of lipoylated autoantigens. Toxicology 97, 199– cytochrome P450 cDNAs in studying the metabolism and effects 224. of cytotoxic drugs. Cancer Chemother. Pharmacol. 43, 59–67. Holownia, A., Braszko, J.J., 2004. Tamoxifen cytotoxicity in hepato- Pirmohamed, M., Kitteringham, N.R., Breckenridge, A.M., Park, B.K., blastoma cells stably transfected with human CYP3A4. Biochem. 1992. The effect of enzyme induction on the cytochrome P450- Pharmacol. 67, 1057–1064. mediated bioactivation of carbamazepine by mouse liver micro- Jaeger, A., Sauder, P., Kopferschmitt, J., Flesch, F., 1987. Clinical fea- somes. Biochem. Pharmacol. 44, 2307–2314. tures and management of poisoning due to antimalarial drugs. Med. Rawden, H.C., Kokwaro, G.O., Ward, S.A., Edwards, G., 2000. Rel- Toxicol. Adverse Drug Exp. 2, 242–273. ative contribution of cytochromes P-450 and flavin-containing Kitteringham, N.R., Lambert, C., Maggs, J.L., Colbert, J., Park, B.K., monoxygenases to the metabolism of albendazole by human liver 1988. A comparative study of the formation of chemically reactive microsomes. Br. J. Clin. Pharmacol. 49, 313–322. drug metabolites by human liver microsomes. Br. J. Clin. Pharma- Rendic, S., Di Carlo, F.J., 1997. Human cytochrome P450 enzymes: a col. 26, 13–21. status report summarizing their reactions, substrates, inducers, and Knowles, S.R., Uetrecht, J., Shear, N.H., 2000. Idiosyncratic drug reac- inhibitors. Drug Metab. Rev. 29, 413–580. tions: the reactive metabolite syndromes. Lancet 356, 1587–1591. Riley, R.J., Maggs, J.L., Lambert, C., Kitteringham, N.R., Park, B.K., Kostrubsky, V.E., Ramachandran, V., Venkataramanan, R., Dorko, K., 1988. An in vitro study of the microsomal metabolism and cellular Esplen, J.E., Zhang, S., Sinclair, J.F., Wrighton, S.A., Strom, S.C., toxicity of phenytoin, sorbinil and mianserin. Br. J. Clin. Pharma- 1999. The use of human hepatocyte cultures to study the induction col. 26, 577–588. of cytochrome P-450. Drug Metab. Dispos. 27, 887–894. Riley, R.J., Roberts, P., Coleman, M.D., Kitteringham, N.R., Park, Lazarou, J., Pomeranz, B.H., Corey, P.N., 1998. Incidence of adverse B.K., 1990. Bioactivation of dapsone to a cytotoxic metabolite: drug reactions in hospitalized patients: a meta-analysis of prospec- in vitro use of a novel two compartment system which contains tive studies. JAMA 279, 1200–1205. human tissues. Br. J. Clin. Pharmacol. 30, 417–426. Li, A.P., Lu, C., Brent, J.A., Pham, C., Fackett, A., Ruegg, C.E., Silber, Rodriguez-Antona, C., Donato, M.T., Boobis, A., Edwards, R.J., Watts, P.M., 1999. Cryopreserved human hepatocytes: characterization P.S., Castell, J.V., Gomez-Lechon, M.J., 2002. Cytochrome P450 of drug-metabolizing enzyme activities and applications in higher expression in human hepatocytes and hepatoma cell lines: molec- throughput screening assays for hepatotoxicity, metabolic stabil- ular mechanisms that determine lower expression in cultured cells. ity, and drug-drug interaction potential. Chem. Biol. Interact. 121, Xenobiotica 32, 505–520. 17–35. Roymans, D., Annaert, P., Van Houdt, J., Weygers, A., Noukens, Lin, H.L., Parsels, L.A., Maybaum, J., Hollenberg, P.F., 1999. N- J., Sensenhauser, C., Silva, J., Van Looveren, C., Hendrickx, J., Nitrosodimethylamine-mediated cytotoxicity in a cell line express- Mannens, G., Meuldermans, W., 2005. Expression and induction ing P450 2E1: evidence for apoptotic cell death. Toxicol. Appl. potential of cytochromes P450 in human cryopreserved hepato- Pharmacol. 157, 117–124. cytes. Drug Metab. Dispos. 33, 1004–1016. Meyer, U.A., 2000. Pharmacogenetics and adverse drug reactions. Schneider, A., Schmalix, W.A., Siruguri, V., de Groene, E.M., Hor- Lancet 356, 1667–1671. bach, G.J., Kleingeist, B., Lang, D., Bocker, R., Belloc, C., Meyer, U.A., Gut, J., 2002. Genomics and the prediction of xenobiotic Beaune, P., Greim, H., Doehmer, J., 1996. Stable expression of toxicity. Toxicology 181–182, 463–466. human cytochrome P450 3A4 in conjunction with human NADPH- Miao, X.S., Metcalfe, C.D., 2003. Determination of carba- cytochrome P450 oxidoreductase in V79 Chinese hamster cells. mazepine and its metabolites in aqueous samples using liquid Arch. Biochem. Biophys. 332, 295–304. chromatography-electrospray tandem mass spectrometry. Anal. Simarro Doorten, A.Y., Bull, S., van der Doelen, M.A., Fink- Chem. 75, 3731–3738. Gremmels, J., 2004. Metabolism-mediated cytotoxicity of ochra- Mills, J.B., Rose, K.A., Sadagopan, N., Sahi, J., de Morais, S.M., 2004. toxin A. Toxicol. In Vitro 18, 271–277. Induction of drug metabolism enzymes and MDR1 using a novel Uetrecht, J.P., 1999. New concepts in immunology relevant to idiosyn- human hepatocyte cell line. J. Pharmacol. Exp. Ther. 309, 303–309. cratic drug reactions: the “danger hypothesis” and innate immune Nakagawa, T., Sawada, M., Gonzalez, F.J., Yokoi, T., Kamataki, T., system. Chem. Res. Toxicol. 12, 387–395. 1996. Stable expression of human CYP2E1 in Chinese hamster Valentine, C.R., Valentine, J.L., Seng, J., Leakey, J., Casciano, D., cells: high sensitivity to N,N-dimethylnitrosamine in cytotoxicity 1996. The use of transgenic cell lines for evaluating toxic metabo- testing. Mutat. Res. 360, 181–186. lites of carbamazepine. Cell Biol. Toxicol. 12, 155–165. L. Vignati et al. / Toxicology 216 (2005) 154–167 167 Venkataramanan, R., Ramachandran, V., Komoroski, B.J., Zhang, S., Yamamoto, Y., Yamazaki, H., Ikeda, T., Watanabe, T., Iwabuchi, H., Schiff, P.L., Strom, S.C., 2000. Milk thistle, a herbal supple- Nakajima, M., Yokoi, T., 2002. Formation of a novel quinone epox- ment, decreases the activity of CYP3A4 and uridine diphosphoglu- ide metabolite of troglitazone with cytotoxicity to HepG2 cells. curonosyl transferase in human hepatocyte cultures. Drug Metab. Drug Metab. Dispos. 30, 155–160. Dispos. 28, 1270–1273. Yoshitomi, S., Ikemoto, K., Takahashi, J., Miki, H., Namba, M., Walubo, A., Smith, P., Folb, P.I., 1998. The role of oxygen free radi- Asahi, S., 2001. Establishment of the transformants expressing cals in isoniazid-induced hepatotoxicity. Methods Find. Exp. Clin. human cytochrome P450 subtypes in HepG2, and their applica- Pharmacol. 20, 649–655. tions on drug metabolism and toxicology. Toxicol. In Vitro 15, 245– Wen, Y.H., Sahi, J., Urda, E., Kulkarni, S., Rose, K., Zheng, X., Sin- 256. clair, J.F., Cai, H., Strom, S.C., Kostrubsky, V.E., 2002. Effects of Zhang, K., Chew, M., Yang, E.B., Wong, K.P., Mack, P., 2001. Mod- bergamottin on human and monkey drug-metabolizing enzymes in ulation of cisplatin cytotoxicity and cisplatin-induced DNA cross- primary cultured hepatocytes. Drug Metab. Dispos. 30, 977–984. links in HepG2 cells by regulation of glutathione-related mecha- Wolfgang, G.H., Johnson, D.E., 2002. Web resources for drug toxicity. nisms. Mol. Pharmacol. 59, 837–843. Toxicology 173, 67–74. Zhou, S., Chan, E., Duan, W., Huang, M., Chen, Y.Z., 2005. Drug Wu, D., Cederbaum, A.I., 1996. Ethanol cytotoxicity to a transfected bioactivation, covalent binding to target proteins and toxicity rele- HepG2 cell line expressing human cytochrome P4502E1. J. Biol. vance. Drug Metab. Rev. 37, 41–213. Chem. 271, 23914–23919.