S 16020¹ is a cytotoxic agent derived from 9-hydroxy olivacine, structurally related to the ellipticine family (Jasztold-Howorko et al., 1994). This compound has been shown to interact with DNA by intercalation and to stimulate DNA cleavage mediated by topoisomerase II by stabilizing the covalent enzyme-DNA complex (Le Mee et al., 1998). S 16020 demonstrated a broad spectrum of antitumor activity on murine (P388 leukemia, Lewis lung carcinoma, B16 melanoma, M5076 sarcoma) and human (colon, breast, ovary, lung) tumor models (Guilbaud et al., 1996, 1997; Kraus-Berthier et al., 1997). This activity compared favorably with other structurally related compounds of the ellipticine family and was comparable to that of the widely used doxorubicin.

Several phase I dose escalation pharmacokinetic and clinical studies have been performed with S 16020, using different administration schedules, i.e., a 60-min or 3-h i.v. infusion once every 3 weeks, for 3 consecutive days every 3 weeks (Awada et al., 2002). Single-dose kinetics showed that blood or plasma concentrations increased in proportion with dose. After a single 100 mg/m² or 150 mg/m² i.v. administration of S 16020, C_inf was 1165 ± 434 ng/ml and 1669 ± 479 ng/ml, respectively. The concentrations declined biphasically, with a mean terminal half-life of about 4 h. Volume of distribution was large, and clearance appeared to be constant, over the dose range (900 ml/min) and independent of the infusion rate. Blood/plasma distribution remained unchanged with dose and dosing schedule. The pharmacokinetics of S 16020 after repeated administrations appeared to change: blood clearance increased, dependent on the dosing schedule, whereas the volume of distribution remained constant. The increase in blood clearance was dependent on the time between administrations: the closer two S 16020 dose administrations were, the higher the increase in clearance. The most pronounced was an approximately 4-fold increase in blood clearance in the 3-daily schedule, followed by the weekly schedule, where blood clearance increased 2-fold. After a 2-week washout, blood clearance of S 16020 returned to its original value. Clearance of S 16020 was mainly through metabolic elimination, and plasma protein binding was not altered by repeated administrations (unpublished data). Repeated daily exposure of Caco-2 cells to S 16020 did not affect the permeability to S 16020, suggesting that induction of P-glycoprotein or other transport proteins was not implicated (unpublished data). A likely explanation for the increase in in vivo clearance of S 16020 is an inducing effect of this drug on its own metabolism. The aim of the present study was, first, to identify the principal enzymes responsible for the metabolic clearance of S 16020 and, second, to determine whether S 16020 is a cytochrome P450 (P450) inducer that is able to induce its own metabolism using primary cultures of human hepatocytes.

Materials and Methods

Chemicals. S 16020 or 9-hydroxy-5,6-dimethyl-N-[2-(dimethylamino)ethyl)-6H-pyrido(4,3-b)-carbazole-1-carboxamide was provided by Servier Laboratories (Courbevoie, France). [³H]S 16020 (40 Ci/mmol, 96% purity) was provided by Amersham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK); rifampicin (RIF) was obtained from Sigma-Aldrich (St. Louis, MO); 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) was purchased from BCP (Lyon, France). Ketoconazole (KT), ethoxyresofurin, acetanilide, [¹⁴C]acetanilide, tolbutamide, [¹⁴C]tolbutamide, chlorzoxazone, [¹⁴C]chlorzoxazone, cyclosporin A, and [³H]cyclosporin A were from Sigma-Aldrich. S-Mephenytoin and S-[¹⁴C]mephenytoin were purchased from Amersham Biosciences UK, Ltd.

Supersomes. Microsomes prepared from insect cells (BTI-TN-5B1-4) infected with wild-type baculovirus (Autographa californica) heterologously expressing individual human cytochrome P450 enzymes (Gentest Supersomes) or FMO3 were obtained from BD Gentest (Woburn, MA). Expressed P450 enzymes included human CYP1A1, 1A2, 2A6, 2C8, 2C9, 2C19, 2D6, 2E1, or 3A4.

Human Liver Lobectomy Samples. Human liver samples were obtained from three patients: FT177, a 69-year-old male with liver metastasis of a colon tumor; FT179, a 55-year-old female with an adenoma; and FT193, a 73-year-old female with liver metastasis of a colon tumor. Viral serologic analysis (hepatitis B, hepatitis C, and human immunodeficiency viruses) was negative. These liver specimens were resected for medical reasons unrelated to our research program. The tissue encompassing the tumor was dissected by the surgeon and sent for pathologic studies, while the remaining encapsulated downstream tissue was used for hepatocyte preparation. No information on the patients was available in our laboratory, apart from the reason for surgical resection. Importantly, and as accepted by the French National Ethics Committee, pathological examination of the surgical specimen was in no way hindered by the procedure used to obtain primary hepatocytes; the tissue samples used for this purpose would otherwise have been immediately discarded.

Primary Cultures of Human Hepatocytes. Primary cultures of human hepatocytes were prepared from lobectomies as described (Pichard et al., 1990, 1992; Ferrini et al., 1998). The viability of cells before plating was determined using the trypan blue exclusion test and comprised between 80 and 90%. Ten million cells in 7 ml or 4 million cells in 3 ml of culture medium were placed into 100-mm or 60-mm plastic dishes precoated with collagen (type I). The serum-free culture medium consisted of a 1:1 mixture of Ham F12 and Williams' E, supplemented as published (Isom and Georgoff, 1984). The culture medium was supplemented with 5% calf serum during the first 4 h after plating to favor the attachment of cells. Then, the medium was changed and subsequently renewed every 24 h in the absence of serum. Cultures were maintained at 37°C in a humid atmosphere of air and 5% carbon dioxide. For the treatment of cells, prototypical inducers including TCDD and RIF were diluted in dimethyl sulfoxide; S 16020 was diluted in water. Aliquots of stock solutions were then added to the culture medium to reach the final concentrations of 1 nM (TCDD), 25 μM (RIF), and 0.02, 0.2, or 2 μM (S 16020). The treatments started 12 h after plating, lasted for 96 h, and were renewed every 24 h as the culture medium was changed.

Preparation of Hepatocyte Microsomes. Microsomes were prepared from cultured hepatocytes by differential centrifugation and stored as described previously (Pichard et al., 1990, 1992). Protein concentration was determined by the bicinchoninic acid method, according to the manufacturer's protocol (Pierce Chemical, Rockford, IL). Bovine serum albumin (Pierce Chemical) was used as the standard.

Cytotoxicity Test. Cytotoxicity of S 16020 was assessed using the microculture tetrazolium assay (MTT test) on human hepatocytes. Cells were maintained in culture in 60-mm plates in the absence or presence of KT at the final concentrations of 0.2, 2, 5, 10, and 20 μM (as a positive control of cytotoxicity) and S 16020 at 0.02, 0.2, and 2 μM. Cytotoxicity was determined by the spectrophotometry method, according to the manufacturer's protocol (SigmaAldrich). Briefly, at the end of the 96-h exposure period, the medium was replaced with a phenol red-free medium containing MTT. Cells were incubated for 1 h at 37°C and Formosan crystals were formed through the activity of mitochondrial dehydrogenases. The medium was then complemented with an equal volume of the MTT solubilization solution. Cell monolayers were gently agitated on a rocking plate to solubilize the crystals of Formosan. This medium was then analyzed spectrophotometrically at 570 nm.

Metabolism of S 16020 in Supersomes. [³H]S 16020 (0.06 μM) was incubated with human Gentest Supersomes containing 50 pmol of CYP1A1, 1A2, 2A6, 2C8, 2C9, 2C19, 2D6, 2E1, or 3A4 for up to 2 h at 37°C in 0.1 M Tris buffer, pH 7.4, or with human FMO3 Supersomes containing 0.5 mg/ml of protein for 1 h and 2 h in 0.1 M Tris buffer, pH 8.4, in the presence of an NADPH-regenerating system. Proteins were precipitated with acetonitrile and supernatant was analyzed by high performance liquid chromatography (HPLC) with radiochemical detection. HPLC analysis consisted of a Prodigy ODS-2 (pore size 5 μm, 150 × 4.6 mm) column. Mobile phase was as follows: 1 ml/min for 50 min comprising mobile phase A (95% ammonium acetate, 50 mM; 5% acetonitrile; 1 ml/l formic acid) and mobile phase B (acetonitrile). Gradient from 95% A to 79% A over 35 min, 79% A to 100% B over 10 min, and 100% B for another 5 min was used.

Metabolism of S 16020 in Human Hepatocytes. Human hepatocytes were treated for 96 h in the absence (untreated cells) or in the presence of inducers of drug-metabolizing enzymes (25 μM RIF, 1 nM TCDD) or 2 μM (780 ng/ml) S 16020. Treatments were renewed every 24 h as the culture medium was changed. After 96 h of incubation, S 16020 (2 μM) and ³H-radiolabeled S 16020 (0.025 μM, 1 μCi/ml) were added to the samples. Controls in the absence of hepatocytes were carried out to evaluate the possible binding to plastic. Aliquots of extracellular medium were collected at 0, 4, 8, and 24 h after dosing and analyzed by radiochemical-UV (254 nm) HPLC as described above.

Measurement of P450-Specific Monooxygenase Activities in Microsomes. Ethoxyresorufin O-deethylase (CYP1A1) activity (Pohl and Fouts, 1980) was determined in microsomes prepared from cultured cells. A total of 200 μg of microsomes were resuspended in 1 ml of 0.1 M potassium phosphate buffer, pH 7.4, in the presence of 5 μM ethoxyresorufin. After a 3-min incubation at 37°C, the reaction was initiated by the addition of 1 mM NADPH and monitored spectrofluorometrically (under conditions of linear kinetics). Monooxygenase activity was expressed as nanomoles of metabolite produced per minute, per milligram of protein.

Measurement of P450-Specific Monooxygenase Activities in Human Hepatocytes. Acetanilide 4-hydroxylase (CYP1A2), tolbutamide 4-hydroxylase (CYP2C9), S-mephenytoin 4-hydroxylase (CYP2C19), N-demethylase (CYP2B6), chlorzoxazone 6-hydroxylase (CYP2E1 and CYP1A1), and cyclosporin A oxidase (CYP3A4) (Pichard et al., 1990; Gerbal-Chaloin et al., 2001) were determined directly in the cultured cells (using 60-mm dishes). After a 96-h treatment with inducers, the medium was renewed in the absence of the inducer and in the presence of cold and radiolabeled substrates: 25 μM acetanilide, 26.4 μM tolbutamide, 33.6 μM S-mephenytoin, 25 μM chlorzoxazone, and 5 μM cyclosporin A. After incubation with the substrate (6-h, 8- to 24-h, 8- to 24-h, 6-h, or 4-h periods for acetanilide, tolbutamide, S-mephenytoin, chlorzoxazone, or cyclosporin A, respectively), the extracellular medium was analyzed by HPLC-radiodetection for the oxidized metabolites. These monooxygenase activities were expressed as the concentration of metabolite (micromolar) produced per 4 million cells during the incubation period.

Acetanilide 4-hydroxylase. Acetanilide was analyzed on a reverse C₁₈ ODS ultrasphere column (250 × 4.6 mm; pore size 5 μm; 100A) protected by a precolumn of the same constitution. Elution was carried out at 1 ml/min for 25 min along a linear gradient of 40:60 (v/v) to 80:20 (v/v) methyl alcohol/distilled water. The column was maintained at room temperature and the injected volume was 100 μl. The eluent was directed to a radioactive flow detector. Acetanilide eluted after 10 min and the 4-OH metabolite after 6 min.

S-Mephenytoin 4-hydroxylase and N-demethylase. Mephenytoin was analyzed on a reverse C₁₈ Hypersil column (150 × 4.6 mm; pore size 5 μm; 100A) protected by a precolumn of the same constitution. Elution was carried out at 1.3 ml/min for 15 min along a linear gradient of 30:70 (v/v) to 80:20 (v/v) methyl alcohol/distilled water. The column was maintained at room temperature and the injected volume was 100 μl. The eluent was detected as above. Mephenytoin eluted after 7 min, the 4-OH metabolite after 1.5 min, and the N-demethylated metabolite after 5.5 min.

Chlorzoxazone 6-hydroxylase. Chlorzoxazone was analyzed on a C₁₈ Zorbax Stable Bond column (150 × 4.6 mm; pore size 5 μm; 100 A) protected by a precolumn of the same constitution. Elution was carried out at 1 ml/min for 15 min along a linear gradient of 30:70 (v/v) to 40:60 (v/v) acetonitrile/distilled water/0.04% orthophosphoric acid. The column was maintained at room temperature and the injected volume was 100 μl. The eluent was detected as above. Chlorzoxazone eluted after 6 min and the 6-OH metabolite after 1.5 min.

Immunoblotting of Microsomal P450 Prepared from Human Hepatocytes. P450 proteins including forms CYP1A2, 2A6, 2B6, 2C9, 2C19, 2E1, and 3A4 were quantified by immunoblotting using specific polyclonal or monoclonal antibodies, as described previously (Pichard et al., 1990, 1992; Gerbal-Chaloin et al., 2001). Routinely, 20 to 50 μg of liver microsomes prepared from cultured hepatocytes were submitted to electrophoresis on an SDS-10% polyacrylamide gel before transferring to nitrocellulose (Millipore Corporation, Bedford, MA). Membranes were incubated with specific anti-P450 antibodies: AK1, a monoclonal antibody directed against rat CYP2C6 cross-reacting with CYP2C9/18/19 (Gerbal-Chaloin et al., 2001); anti-human CYP2B6, kindly provided by P. Beaune (Institut National de la Santé et de la Recherche Médicale, Paris, France) (Gervot et al., 1999); anti-human CYP1A2 and CYP2E1 (Euromedex, Souffelweyersheim, France); and anti-baboon CYP2A and CYP3A cross-reacting with the human orthologous forms prepared in our laboratory (Dalet-Beluche et al., 1992). Blots were developed using coupling antibodies and the enhanced chemiluminescence procedure from Amersham (Amersham Biosciences UK, Ltd.). The relative amount of P450 proteins was estimated by scanning the blots using the Adobe PhotoShop LE program and quantitative analysis of the image by the NIH Image 1.6/ppc program. Authentic standards for immunoblots were microsomes from human lymphoblastoid cells transfected with the human CYP1A1, 1A2, 2A6, 2B6, 2C9, 2C19, 2E1, and 3A4 cDNA (BD Gentest).

NKNT-3 Cell Line and Transient Transfection Experiments. NKNT-3 cells obtained from large T antigen-mediated immortalization of human hepatocytes (Kobayashi et al., 2000) were kindly provided by Prof. Ira Fox (University of Nebraska Medical Center, Omaha, NE) and cultured in minimal essential medium supplemented with 5% calf serum. Plasmids p1A1-(XRE)₃-MMTV-LUC (250 ng), containing three copies of CYP1A1 xenobiotic responsive element upstream of mouse mammary tumor virus promoter and luciferase reporter gene, and pSV-β-galactosidase (25 ng, used as a transfection control vector) were transiently transfected overnight in 50,000 cells per well (24-well plate), in the presence of FuGENE-6 transfection reagent as recommended by the manufacturer (Roche Applied Science, Indianapolis, IN). Cells were then cultured in the absence (UT) or presence of 1 nM TCDD alone; 0.02, 0.2, or 2 μM S 16020 alone; or combinations of both molecules. Twenty-four hours later, cells were harvested and extracts were analyzed for luciferase and β-galactosidase activities.

Results

Identification of the Enzymes Involved in the Oxidative Metabolism of S 16020 in Humans. After i.v. administration of [¹⁴C]S 16020 in cancer patients, the drug was rapidly and extensively biotransformed to more than 15 metabolites. The recovered radioactivity was 40 to 54%, the majority of which was excreted in feces and the remainder in urine. The relative amounts of the major metabolites in urine were: S 16020-glucuronide (H26) > cyclic N-demethyl-S 16020 (S 16018) ≥ S 16020-N-oxide (S 19505) > S 16020-sulfate (S 31919), allylic N-demethyl-S 16020 (S 18717) > Y874 amide, whereas they were S 16018 > S 31919 ≥ S 18717 > S 19505 in feces (Fig. 1). The phase I metabolites including the N-demethyl and N-oxide derivatives were also observed after incubation of S 16020 with human liver microsomes. To identify the enzymes responsible for the production of these metabolites, [ethyl chain (n)-³H]S 16020 was incubated with microsomes prepared from baculovirus-infected insect cells expressing recombinant human P450 or FMO enzymes and an NADPH-regenerating system. After precipitation of proteins with acetonitrile, the supernatant was analyzed by radiochemical-UV HPLC. The results revealed a significant metabolism with CYP1A1, 1A2, and FMO3 expressing microsomes. The products generated in detectable amounts were the cyclic and allylic N-demethylated derivatives (S 16018 and S 18717) with CYP1A2 and CYP1A1, respectively, and the N-oxide derivative (S 19505) with human FMO3 (Fig. 2). In addition, very polar metabolites (H1, H2) presumably reflecting N-dealkylation of the dimethyl-ethyl amino moiety of the molecule were detected in CYP1A1 microsomes. In contrast, no metabolism was detected with microsomes expressing CYP2A6, 2C8, 2C9, 2C19, 2D6, 2E1, or 3A4.

Induction of P450 Proteins and Related Specific Monooxygenase Activities by S 16020 in Primary Human Hepatocytes. Since phase I studies in cancer patients revealed the possibility that S 16020 could induce its own metabolism, we decided to evaluate the inducing effect of this molecule on P450 proteins and related monooxygenase activities using primary human hepatocytes as a model.

First, we determined whether S 16020 is toxic to the cells. For this purpose, hepatocytes were treated for 96 h with increasing concentrations of S 16020 between 0.02 and 2 μM (that is, under the standard conditions used in this work) or with 20 μM KT in parallel as an authentic cytotoxic molecule (Maurel, 1996), and the cells were analyzed using the MTT test. Whereas KT (at 20 μM) appeared to be toxic to the cells as expected, S 16020 was not, even at the highest concentration used (2 μM) (data not shown). Consistent with this, examination of cells by phase contrast microscopy, under the same conditions, revealed no toxicity either.

Next, hepatocytes were cultured in the absence of any treatment (UT) or in the presence of S 16020 (under standard conditions; i.e., 0.02–2 μM for 96 h) or, in parallel, 1 nM TCDD or 25 μM RIF, two prototypical inducers of CYP1A and CYP2A/2B/2C/3A, respectively. At the end of the period of treatment, cells were analyzed for the accumulation of P450 proteins by immunoblotting and for related specific monooxygenase activities evaluated directly in the culture dishes, except for ethoxyresorufin O-deethylase, which was measured in microsomes extracted from these cultures. The results on the immunoblot analysis and on monooxygenase activities are summarized in Fig. 3, Table 1, and Table 2, respectively. CYP1A and CYP2A/2B/2C/3A were induced in response to TCDD and RIF, respectively, as expected. The data clearly show that S 16020 (at the greatest concentration tested, i.e., 2 μM) is a potent inducer of the CYP1A family in the three different cultures analyzed in this work. The levels of CYP1A2 protein determined by immunoblot (Fig. 3) were on average approximately 3 times greater in S 16020-treated cells than in untreated or RIF-treated cells and similar to those found in TCDD-treated cells. This trend was further confirmed with the observation of a concomitant increase in the rate of acetanilide hydroxylation and ethoxyresorufin O-deethylation, two activities related to CYP1A2 and 1A1, respectively. Indeed, both activities appeared to be strongly induced by S 16020 and TCDD (approximately 50 times on average). The greater induction ratio characterizing these activities in S 16020- or TCDD-treated cells as compared with the induction ratio characterizing CYP1A2 levels is likely due to the contribution of CYP1A1 enzyme to both activities. Another argument in favor of a strong induction of CYP1A1 came from the analysis of the rate of chlorzoxazone 6-hydroxylation. This reaction had been first suggested as a marker of human CYP2E1 (Peter et al., 1990). However, it has been subsequently reported that CYP1A1 is also able to catalyze this reaction very efficiently (Carriere et al., 1993). Hence, when this activity was evaluated in our cultures, it was found to be increased approximately 6- to 8-fold in response to TCDD and S 16020, respectively, whereas in the meantime, the levels of CYP2E1 protein were not affected (Tables 1 and 2). This observation most likely reflects the induction of CYP1A1. In summary, these results suggest that S 16020 is a potent aryl hydrocarbon-like inducer in primary human hepatocytes.

Activation of the AhR by S 16020. To more definitely confirm the inducing properties of S 16020, we decided to evaluate directly its effect on the transcriptional activity of the AhR by transfection assay. For this purpose, NKNT-3 cells were transfected with a plasmid harboring three copies of the CYP1A1 xenobiotic responsive element (XRE; TGCGTG, a motif specifically recognized by the AhR-AhR nuclear translocator heterodimer in CYP1A and other AhR target genes) (Hankinson, 1995), upstream of the mouse mammary tumor virus promoter, and luciferase as the reporter gene. After transfection, cells were treated for 16 h with 1 nM TCDD alone, 0.02 to 2 μM S 16020 alone, and combinations of both molecules, and the activity of the reporter gene was measured. A low concentration of TCDD (1 nM) was used to detect any possible antagonist effect of S 16020. The results are reported in Fig. 4. TCDD stimulated the gene reporter activity as expected. Interestingly, and in agreement with the data presented above, S 16020 was a potent activator of the AhR at the greatest concentration tested, i.e., 2 μM. Moreover, no antagonist effect was observed when this compound was used in combination with TCDD; in contrast, the inducing effects of both molecules were additive.

Metabolism of S 16020 in Primay Human Hepatocytes. Although the results presented above are in favor of the hypothesis that S 16020 induces its own metabolism in humans, we decided to verify this point directly in our cultures. For this purpose, hepatocytes were first cultured in the absence or presence of TCDD, RIF, or S 16020 under standard conditions for 96 h. Next, cells were incubated in the presence of 2 μM ³H-radiolabeled S 16020 for an additional 4, 8, or 24 h. At these times, extracellular medium was collected and analyzed for S 16020 and metabolites by radiochemical-UV HPLC. Control experiments were carried out under the same conditions but in the absence of cells to evaluate the possible binding of radiolabeled molecules to plastic dishes and glass. The results are presented in Fig. 5 and Table 3. In untreated human hepatocytes, S 16020 was extensively metabolized so that less than 2% of the parent drug remained in the extracellular medium after 24 h of incubation. Six major metabolites (each one accounting for ≥5% of the initial radioactivity) were detected, including the S 16020 glucuronide (H26) and sulfate (S 31919) conjugates, the N-oxide (S 19505), two derivatives of oxidative N-dealkylation (H1 and H2), and an unidentified compound, H8. A similar, if not identical, metabolic profile was observed in cells pretreated with RIF. This is consistent with the fact that neither CYP2B6, 2C9, nor 3A4 enzymes, the levels of which are strongly increased in these cells, are involved significantly in S 16020 biotransformation. The absence of significant amounts of the N-demethyl derivatives of S 16020 (S 16018 and S 18717) is likely due to the low level (if any) of CYP1A2 and CYP1A1 proteins in untreated or RIF-treated cells. However, in cells pretreated with either S 16020 or TCDD, the drug biotransformation was much faster than in UT or RIF-treated cells, so that only approximately 2% of the parent drug remained in the extracellular medium after 4 h of incubation. This was accompanied by a significant reduction in the formation of the glucuronide (M26) and sulfate (S 31919) conjugates of S 16020, presumably due to the rapid metabolism of S 16020 in these cultures. Only one of the CYP1A-dependent N-demethyl derivatives (S 16018) was detected at a high level (12% after 4 h of incubation) in these cells. However, its level decreased rapidly with time, suggesting the formation of secondary derivatives. The reason for the absence of the other N-demethyl derivative (S 18717), found to be CYP1A-dependent as well, is not clear. One possibility is that this metabolite is highly unstable in hepatocytes (in contrast to microsomes) and rapidly converted to secondary derivatives. In fact, this metabolite is also found at a very low level in urine in humans. Conversely, at least 14 metabolites were observed, including those mentioned above (note the significant increase in the formation of the phase I metabolites H1 and S 16018) and several unidentified derivatives, H6, H7, H13, H18, H20, H30, and H33, the formation of which generally increased with time. In addition, the decrease with time of N-oxide accumulation (S 19505) in these cells was faster as compared with that observed in untreated or RIF-treated cells. These results clearly confirm that S 16020 is a potent inducer of its own metabolism in human hepatocytes.

Discussion

The data presented here show that the metabolism of S 16020 in humans is dependent on phase I and II enzymes including CYP1As and FMO3, on the one hand, and glucuronosyl- and sulfotransferases, on the other hand, respectively. In addition, S 16020 appears to be a potent activator of the Ah receptor and an inducer of the CYP1A family, so that this molecule is expected to be an inducer of its own metabolism. This is in good agreement with the previous observation that the blood clearance of this drug increases in cancer patients after repeated dosing.

The similarity between the extent of induction of CYP1A2 and of CYP1A-related monooxygenase activities and the metabolic profile of S 16020 in either S 16020- or TCDD-treated cells strongly suggests that this drug is an aryl hydrocarbon-like inducer. This was fully confirmed by transfection experiments in NKNT-3 cells. Indeed, treatment of these cells by S 16020 or TCDD strongly increased the activity of a reporter gene placed under the control of the CYP1A1-specific XRE, thus confirming the activation of the AhR by these compounds. The inducing effect of S 16020 in hepatocytes (CYP1A2 protein and CYP1A activities) and in NKNT-3 cell (transfection experiments) was only significant at the greatest concentration tested (2 μM), suggesting either that this molecule is a moderate-affinity ligand but a strong activator of AhR or that rapid metabolism is responsible for the absence of effect at low concentrations. The finding that S 16020 is a substrate and inducer of CYP1As was expected from its chemical planar structure (Gasiewicz et al., 1996). In fact, this compound is structurally related to ellipticine, which has been characterized for a long time as an aryl hydrocarbon-like inducer (Lesca et al., 1976; Phillipson et al., 1982), although it becomes an antagonist of AhR at high concentrations (Fernandez et al., 1988; Roy et al., 1988; Gasiewicz et al., 1996). In addition, ellipticine is known to be an inhibitor and presumably a substrate of CYP1As (Lesca et al., 1978, 1979, 1980, 1981; Delaforge et al., 1980; Cresteil et al., 1982). The current data show that S 16020 is a substrate of CYP1As. In addition, it is interesting that S 16020 is not an antagonist of the AhR, in contrast to ellipticine.

CYP1A1 protein is generally hardly detectable by immunoblot in liver and hepatocyte microsomes, so that only the CYP1A2 blot is shown in Fig. 2. However, we have provided indirect evidence of CYP1A1 induction as follows: 1) S 16020 is a potent activator of AhR, which is well known to control both CYP1A1 and 1A2 expression; 2) CYP1A2 is indeed induced, and there is no case reported in which CYP1A2 but not 1A1 is induced by AhR activators; 3) the rate of chlorzoxazone 6-hydroxylation (a CYP2E1- and 1A1-specific substrate) is strongly induced without corresponding induction of CYP2E1; 4) the rate of ethoxyresorufin O-deethylation, highly specific for CYP1A1, is strongly induced; and, finally, 5) in all experiments, S 16020 behaved as TCDD, the prototypical inducer of CYP1A1 and 1A2.

The metabolism of S 16020 is apparently very complex, with the formation of more than 30 metabolites, most of which have not yet been identified. Among these, however, the N-oxide (FMO3), the glucuronide, and the sulfate conjugates account for more than 60% of the initial amount of drug in untreated hepatocytes after 4 h of incubation. The absence of N-demethyl derivatives (CYP1As) in untreated cells is consistent with the finding that CYP1A proteins are not expressed in uninduced hepatocytes. The very strong increase in production of other metabolites in cells treated with S 16020 and TCDD suggests that CYP1A and maybe other AhR-dependent systems contribute actively to the metabolism of this drug. This is consistent with the results obtained with microsomes prepared from baculovirus-infected insect cells expressing recombinant human P450s.

This work offers another example of the usefulness of preclinical screening of drugs with primary human hepatocytes. We and others have demonstrated that, provided these cells are cultured under appropriate conditions, they retain a differentiated phenotype including the expression of the battery of phase I and phase II functional enzymes (Clement et al., 1984; Kremers et al., 1994; Li et al., 1995, 1997; Maurel, 1996; Pichard-Garcia et al., 2002). Here we show that there is a good agreement between the in vitro data obtained in primary human hepatocytes and the in vivo data in patients. In particular, the major metabolites of S 16020 found in urine and feces after repeated dosing in humans are observed in hepatocyte extracellular medium after treatment of cells with either TCDD or S 16020, although the overall conditions are very different. Indeed, it is worth mentioning that not only phase I metabolism systems (P450s) are expressed in these cultures, but phase II conjugation systems as well, including both UDP glucuronyl- and sulfotransferases. Thus, although primary human hepatocytes are difficult to use for routine measurements, they remain the gold standard model to predict the metabolism and drug-mediated enzyme induction in the human liver.