Design, Synthesis, and Characterization of 7-Methoxy-4-(aminomethyl

Chem. Res. Toxicol. 1999, 12, 555-559

555

Communications Design, Synthesis, and Characterization of 7-Methoxy-4-(aminomethyl)coumarin as a Novel and Selective Cytochrome P450 2D6 Substrate Suitable for High-Throughput Screening Rob C. A. Onderwater, Jennifer Venhorst, Jan N. M. Commandeur, and Nico P. E. Vermeulen* Leiden/Amsterdam Center for Drug Research (LACDR), Division of Molecular Toxicology, Department of Pharmacochemistry, Vrije Universiteit, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands Received November 13, 1998

In this study, a selective substrate for cytochrome P450 2D6 was designed using a small molecule model developed by M. J. De Groot et al. [(1997) Chem. Res. Toxicol. 10, 41-48]. The substrate, 7-methoxy-4-(aminomethyl)coumarin (MAMC), and its putative O-demethylated metabolite 7-hydroxy-4-(aminomethyl)coumarin (HAMC) were synthesized, and their respective fluorescence properties were characterized. The selectivity of MAMC for P450 2D6 was characterized using microsomes containing single human P450 isoenzymes and human liver microsomes. Formation of the metabolic product HAMC was easily assessed in real time with fluorescence spectroscopy, since MAMC and HAMC excitation and emission wavelengths differed significantly. HPLC analysis confirmed that HAMC was the single metabolic product of MAMC and that HAMC formation accounts for the total increase in fluorescence. It was found that, in microsomes from yeast or lymphoblastoid cells selectively expressing P450 isoenzymes, MAMC was selective for P450 2D6 at a concentration of 25 µM with only P450 1A2 contributing significantly to the formation of HAMC. P450s 2A6, 2B6, 2C8, 2C9, 2C19, 2E1, 3A4, and 3A5 were shown not to metabolize MAMC at a concentration of 25 µM. Km and vmax values of MAMC for P450 2D6 were found to be 26.2 ( 2.8 µM and 2.9 ( 0.07 min-1, respectively. For P450 1A2, MAMC was found to have a Km value of 29.7 ( 6.2 µM and a vmax of 0.57 ( 0.07 min-1. Formation of HAMC in human liver microsomes could be completely inhibited by quinidine, at a concentration of 0.5 µM selective for P450 2D6, and furafylline, at a concentration of 30 µM selective for P450 1A2. In conclusion, O-demethylation of 7-methoxy4-(aminomethyl)coumarin is a rapid and easily determined parameter for P450 2D6 activity and, due to the fluorescent properties of the metabolite formed, may be a valuable new tool for high-throughput screening purposes.

Introduction Cytochrome P450 2D6 is a member of the P450 superfamily, which is absent in 5-9% of the Caucasian population as a result of recessive inheritance of gene mutations (1, 2). First discovered as the debrisoquine/ sparteine polymorphism, this absence results in deficiencies in the metabolism of numerous drugs. Deficiencies in metabolism can lead to a number of problems, such as undesirably high plasma levels of the drug and concomitant side effects or lack of activation of a prodrug (3). It is, therefore, important to discover in an early stage of drug development whether potential drug candidates might be substrates of P450 2D6. * To whom correspondence should be addressed. Telephone: +31204447590. Fax: +31204447610. E-mail: [email protected].

Presently, an increasing need exists in drug discovery and development for highly selective P450 substrates, which give rise to easily detectable products, preferably suited for high-throughput screening. Until recently, only HPLC and GC methods have been described for P450 2D6 model substrates such as bufuralol, dextromethorphan, debrisoquine, sparteine, and metoprolol. These assays are all time-consuming and therefore not suitable for high-throughput screening purposes. Recently, a fluorescence microplate reader-based method for highthroughput screening was described by Crespi et al. (4). This method is based on single expressed P450 isoenzymes with high catalytic activity. However, the substrate used in this assay, i.e., 7-ethoxy-3-cyanocoumarin, was not selective for P450 2D6 and had a low rate of turnover.

10.1021/tx980248q CCC: $18.00 © 1999 American Chemical Society Published on Web 06/18/1999

556 Chem. Res. Toxicol., Vol. 12, No. 7, 1999

Several small molecule and homology models have been developed for P450 2D6 (reviewed in ref 5). One derived by Koymans et al. (6) and extended and refined by De Groot et al. (7, 8) suggests that substrates of P450 2D6 attach with a basic nitrogen moiety to the carboxylate group of aspartate 301, that oxidation occurs at 5 or 7 Å from this basic nitrogen moiety, and that the substrates contain a flat usually aromatic region coplanar to the oxidation site. In this study, this small molecule model of P450 2D6 substrates was used to design a new and selective P450 2D6 substrate which would give rise to a fluorescent metabolite and be suitable for high-throughput screening. This computer-predicted substrate was synthesized and characterized for substrate selectivity in microsomes of 10 different heterologously expressed human P450s and human liver microsomes.

Materials and Methods Materials. 7-Methoxy-4-(bromomethyl)coumarin, resorcinol, 4-chloroacetoacetate ethyl ester, and quinidine were obtained from Aldrich (Zwijndrecht, The Netherlands). Furafylline was obtained from RBI (Natick, MA). NADPH was obtained from Applichem (Darmstadt, Germany). Microsomes of lymphoblastoid cells overexpressing P450 1A2, 2A6, 2B6, 2E1, 2D6, and 3A4 were obtained from GENTEST (Woburn, MA). Microsomes containing P450 2C8, 2C9, 2C18, and 3A5 and human liver microsomes were a kind gift from P. Beaune (INSERM U75, Paris, France). Molecular Modeling of 7-Methoxy-4-(aminomethyl)coumarin (MAMC).1 Generation of the initial geometry of the structure as well as a full conformational analysis was carried out with the molecular mechanics modeling package MacroModel (5.0) (9). The resulting conformations were subsequently minimized with the BatchMin facility (10) using the AMBER force field. Further optimization of the lowest-energy conformation resulting from the conformational search was performed with the Amsterdam Density Functional (ADF) program version 2.3 using the double-z basis set (11). The fitting procedure was essentially as described previously (8) and was carried out with the molecular modeling program Chem-X 1998 (12). The energy difference between the fitted and lowest-energy conformation was determined by means of a single-point calculation with ADF. Synthesis of MAMC. 7-Methoxy-4-(bromomethyl)coumarin (500 mg) was added to 50 mL of acetone and 2 mL of 25% ammonium hydoxide. The resulting colorless solution was stirred at room temperature for 60 min. The progress of the reaction was followed by silica TLC with acetone as the mobile phase. The product with an Rf of 0.6 was detected by UV (256 or 366 nm) and by ninhydrin reactivity. After completion of the reaction, the yellow reaction mixture was acidified to pH 6 with 6 N hydrochloric acid and the acetone was evaporated. The product was extracted at pH 10 with ethyl acetate and taken up in 2-propanol. To this solution 10 drops of concentrated hydrochloric acid was added, and the product crystallized pentane as a hydrochloric acid salt (beige plates). The identity of the product (350 mg, >95% yield) was established by 1H NMR and with GC/MS after derivatization with acetic acid anhydride: 1H NMR (DMSO-d6) δ 3.85 (s, 3H, CH3O), 4.40 (s, 2H, CH2), 6.40 (s, 1H, dCH), 7.00 (m, 2H, Ar), 7.70 (d, 1H, Ar); GC/ MS (relative intensity) m/z 247 ([M+•], 53%), 205 ([M+ - Od CdCH2], 100%), 190 ([M+ - HNCOCH2], 22%), 176 (51%), 162 (44%), 161 (53%). MAMC exhibited a fluorescence excitation maximum at 326 nm and a fluorescence emission maximum at 396 nm (Figure 1). 1Abbreviations: HAMC, 7-hydroxy-4-(aminomethyl)coumarin; MAMC, 7-methoxy-4-(aminomethyl)coumarin.

Communications

Figure 1. Fluorescence spectra of 7-methoxy-4-(aminomethyl)coumarin (MAMC) and 7-hydroxy-4-(aminomethyl)coumarin (HAMC) at a concentration of 100 µM in 0.1 M potassium phosphate buffer at pH 7.4 and 37 °C: (A) excitation spectrum of MAMC (at an emission wavelength of 396 nm), (B) excitation spectrum of HAMC (at an emission wavelength of 470 nm), (C) emission spectrum of MAMC (at an excitation wavelength of 326 nm), and (D) emission spectrum of HAMC (at an excitation wavelength of 370 nm). Synthesis of 7-Hydroxy-4-(aminomethyl)coumarin (HAMC). At 0 °C, 50 mL of sulfuric acid (98%) was added to 11.1 g of resorcinol and 18.5 g of 4-chloroacetoacetate ethyl ester while the mixture was being stirred for 45 min. The mixture was left stirring for 18 h at 20 °C and subsequently was poured over ice. The progress of the reaction was followed by silica TLC with 9/1 acetone/25% ammonium hydroxide as the mobile phase. The product (Rf ) 0.5) was detected by UV (256 or 366 nm). The product 7-hydroxy-4-(chloromethyl)coumarin was recrystallized from methanol, and its identity was established by 1H NMR: 1H NMR (DMSO-d6) δ 4.90 (s, 2H, CH2), 6.40 (s, 1H, d CH), 6.70 (m, 2H, Ar), 7.75 (d, 1H, Ar). 7-Hydroxy-4-(chloromethyl)coumarin (1 g) was added under anaerobic conditions to 50 mL of 25% ammonium hydroxide. The resultant yellow solution was stirred under nitrogen for 60 min at 50 °C. The progress of the reaction was followed by silica TLC with 9/1 acetone/25% ammonium hydroxide as the mobile phase. The product (Rf ) 0.3) was detected by UV (256 or 366 nm) and by ninhydrin reactivity. After completion of the reaction, the mixture was acidified with 6 N hydrochloric acid and filtered. The filtrate was gently neutralized with 2 N sodium hydroxide, and the product was filtered off as a beige solid. After the product (460 mg) had been dried, its identity was established by 1H NMR: 1H NMR (DMSO-d6) δ 3.95 (s, 2H, CH2), 6.35 (s, 1H, dCH), 6.85 (m, 2H, Ar), 7.55 (d, 1H, Ar). HAMC exhibited a fluorescence excitation maximum at 370 nm and a fluorescence emission maximum at 470 nm (Figure 1). Microsomal MAMC Incubations. Incubations were performed at 37 °C in a Shimadzu RF-5000 spectrofluorometer with the excitation wavelength set at 405 nm (5 nm slit) and emission set at 480 nm (10 nm slit). These wavelength settings deviated slightly from the excitation and emission maxima of the metabolic product in order to eliminate background fluorescence from the parent substrate and NADPH. NADPH (10 µL of a 1 mM solution) and 10 µL of the microsomal fraction were added to 970 µL of 0.1 M potassium phosphate buffer (pH 7.4) containing 0.4 mM EDTA. After equilibration at 37 °C, 10 µL of 2.5 or 25 mM MAMC in DMSO was added and the real-time increase in fluorescence was recorded. After following the reaction for several minutes, 10 µL of 10 µM HAMC in buffer was added and the reaction rate was quantified from the resulting increase in fluorescence. The reaction rate remained constant for at least 20 min. In the experiments with the inhibitors quinidine and furafylline, 10 µL of solutions with concentrations of 50 µM (in water) and 3 mM (in DMSO) was added, respectively. Furafylline was added 5 min prior to addition of MAMC.

Communications

Chem. Res. Toxicol., Vol. 12, No. 7, 1999 557

Figure 2. Metabolism of MAMC to its fluorescent metabolite HAMC by P450 2D6. Incubations were performed in the presence of microsomes from yeast or lymphoblastoid cells selectively expressing P450 isoenzymes (10 pmol of P450/incubation) and human liver microsomes. HPLC Analysis. MAMC and its O-demethylated metabolite HAMC were separated on a 4.6 mm × 25 cm Phenomenex Spherex 5 C18 column with 25% methanol/1% triethylamine (pH 3) as the mobile phase. The flow rate was set at 0.4 mL/min. Compounds were detected using a Shimadzu RF 530 fluorescence detector with excitation set at 370 nm and emission set at 470 nm. Under these conditions, HAMC and MAMC were found to have retention times of 8.4 and 13.5 min, respectively.

Results and Discussion This study was aimed at the development of a selective P450 2D6 substrate which would give rise to a fluorescent metabolite. Because of its fluorescent properties, the coumarine skeleton was chosen as a starting point. 7-Hydroxycoumarine has fluorescent properties which strongly differ from those of 7-alkoxycoumarines; thus, the site of oxidation of the potential substrate should preferably be the R-carbon of an alkoxycoumarine. Since all selective, high-turnover P450 2D6 substrates known to date contain a basic nitrogen moiety 5-7 Å from their site of oxidation, an amino-containing substituent was added to the coumarin skeleton, to fulfill this requirement. The small molecule model derived by Koymans et al. (6) and extended and refined by De Groot et al. (7, 8) was used to predict where such a moiety would have to be placed to achieve oxidation at the R-carbon of an alkoxycoumarin. A methylamine moiety at the 4-position of the coumarine ring was proposed to direct the site of oxidation to the R-carbon of alkoxycoumarins (see Figure 2 for the chemical structures of MAMC and its fluorescent metabolic HAMC). As can be seen in Figure 3, MAMC fits well in the refined substrate model according to the criteria previously defined. The difference between the energy level of the lowest-energy conformation of MAMC and the fitted conformation was 7.94 kcal/mol. The fit distances between the CR and Cβ atoms of the aspartic acid residue coupled to MAMC and the two template molecules were 0.130 and 0.163 Å for dextromethorphan and 0.224 and 0.430 Å for debrisoquine, respectively, which satisfy the fit criteria of the model of 0.5 Å. In addition, MAMC contains a flat aromatic region coplanar to the oxidation site. Incubations were performed at 37 °C in a Shimadzu RF-5000 spectrofluorometer with a crystal cuvette, and the increase in fluorescence from incubations was measured in real time, analogous to the well-known assay of ethoxyresorufin O-deethylation as a marker for P450 1A1/2 activity. MAMC proved to be a relatively selective P450 2D6 substrate. In a series of microsomes containing 10 different human P450s, only P450 2D6 and to a lesser

Figure 3. Overlay of debrisoquine (green), dextromethorphan (blue), and 7-methoxy-4-(aminomethyl)coumarin (MAMC, red) in the small molecule model for P450 2D6 derived by Koymans et al. (6) and extended and refined by De Groot et al. (7, 8). The red arrow points to the site of oxidation. The green arrow points to the interaction site between the basic nitrogen moiety of the substrate and aspartate 301 of the enzyme.

Figure 4. O-Demethylation of MAMC by microsomes of heterologously expressed human P450s at 25 and 250 µM (n ) 3). P450 1A2, 2A6, 2B6, 2E1, 2D6, and 3A4 were obtained from GENTEST. P450 2C8, 2C9, 2C18, and 3A5 were a kind gift from P. Beaune. nd means not detectable.

extent 1A2 exhibited activity at a substrate concentration of 25 µM (Figure 4). The increase in fluorescence in incubations of P450 2D6-containing microsomes with 250 µM MAMC and 0.1 mM NADPH was found to be linear for at least 20 min. MAMC was found to have a Km value of 26.2 ( 2.8 µM and a vmax of 2.9 ( 0.07 min-1 for P450 2D6. MAMC was found to have a Km value of 29.7 ( 6.2 µM and a vmax of 0.57 ( 0.07 min-1 for P450 1A2 (Table 1). Activity in microsomes containing P450 2D6 could be completely inhibited by addition of 0.5 µM quinidine, a

558 Chem. Res. Toxicol., Vol. 12, No. 7, 1999 Table 1. Enzyme Kinetic Parameters of MAMC Metabolism by Relevant Human Cytochrome P450 Isoformsa

a

P450 isoform

Km (µM)

vmax (min-1)

1A2 2D6

29.7 ( 6.2 26.2 ( 2.8

0.57 ( 0.07 2.9 ( 0.07

n ) 3.

Figure 5. Overlay of representative HPLC traces of the O-demethylation of 100 µM MAMC by P450 2D6 (100 pmol/ mL): (A) HAMC peak of a 0 min incubation, (B) HAMC peak of a 30 min incubation, and (C) HAMC peak of a 30 min incubation in the presence of 10 µM quinidine. The positions of the parent compound MAMC and the inhibitor quinidine are also indicated.

Communications

O-demethylation activity of human liver microsomes exhibit an excellent correlation to their respective vmax/ Km values. At a concentration of 250 µM, other P450 isoforms also metabolize MAMC (Figure 4); therefore, we recommend that MAMC be used in human liver microsomal preparations at a concentration near its Km value for P450 2D6. HPLC analysis of incubations of 100 µM MAMC (4Km) incubated for 30 min with P450 2D6-containing microsomes and stopped with 0.3 M trichloroacetic acid clearly showed the formation of HAMC as the single metabolite. The formation rate of HAMC (3.5 ( 0.51 min-1) was linear over 30 min and could be completely inhibited by addition of quinidine (Figure 5). HPLC analysis of the rate of formation of HAMC indicated that the entire real-time increase in fluorescence could be ascribed to the formation of HAMC. Recently, a high-throughput screening method was developed for several human P450s (4). In this assay, 7-ethoxy-3-cyanocoumarin was used as a substrate for P450 2D6. However, although useful, 7-ethoxy-3-cyanocoumarin does not appear to be a selective, high-turnover substrate for P450 2D6 as can be derived from the substrate model for 2D6 (8). MAMC might easily be incorporated in this high-throughput method by replacing 7-ethoxy-3-cyanocoumarin. In conclusion, MAMC is a highly selective P450 2D6 substrate. Due to its strong fluorescence, 7-hydroxy-4(aminomethyl)coumarin (HAMC), the O-demethylated metabolite of MAMC, can be quantified directly in the incubation mixture with negligible interference from substrate or NADPH fluorescence. In human liver microsomes, only P450 2D6 and 1A2 contribute to the O-demethylation of MAMC. After preincubation with the selective P450 1A2 inhibitor furafylline, all MAMC O-demethylation activity in human liver microsomes can be ascribed to P450 2D6. Thus, the O-demethylation of MAMC may provide a rapid and easy tool for investigating drug-drug interactions in vitro.

References well-known and selective inhibitor of P450 2D6 (13). Activity in microsomes containing P450 1A2 could be completely inhibited by preincubation for 5 min with 30 µM furafylline, a well-known and selective inhibitor of P450 1A2 (13). In human liver microsomes (n ) 3; obtained from P. Beaune and designated as A1, A8, and A11), MAMC O-demethylation activity (25 µM MAMC) was inhibited by 33.9 ( 8.9% by 0.5 µM quinidine and by 53.4 ( 3.9% by preincubation for 5 min with 30 µM furafylline. Residual activity after preincubation with 30 µM furafylline could be completely inhibited by addition of 0.5 µM quinidine. These observations indicate that in human liver microsomes, MAMC is exclusively metabolized by P450 2D6 and 1A2. Although P450 1A2 has a much lower vmax value for MAMC O-demethylation than 2D6, due to the higher level of 1A2 (12.7 ( 6.2% of the total P450) compared to 2D6 (1.5 ( 1.3% of the total P450) in human liver (14), the contribution of 1A2 to the overall MAMC O-demethylation in human liver microsomes is significant. In fact, when their respective microsomal levels are taken into account, the observed contributions of P450 1A2 and 2D6 to the overall MAMC

(1) Mahgoub, A., Idle, J. R., Dring, L. G., Lancaster, R., and Smith, R. L. (1977) Polymorphic hydroxylation of Debrisoquine in man. Lancet II (8038), 584-586. (2) Eichelbaum, M., and Gross, A. S. (1990) The genetic polymorphism of debrisoquine/sparteine metabolism: Clinical Aspects. Pharmacol. Ther. 46, 377-394. (3) Tucker, G. T. (1994) Clinical implications of genetic polymorphism in drug metabolism. J. Pharm. Pharmacol. 46, 417-424. (4) Crespi, C. L., Miller, V. P., and Penman, B. W. (1997) Microtiter plate assays for inhibition of human, drug-metabolizing cytochromes P450. Anal. Biochem. 248, 188-190. (5) De Groot, M. J., and Vermeulen, N. P. E. (1997) Modeling the active sites of cytochrome P450s and glutathione S-transferases, two of the most important biotransformation enzymes. Drug Metab. Rev. 29, 747-799. (6) Koymans, L. M. H., Vermeulen, N. P. E., Van Acker, S. A. B. E., Te Koppele, J. M., Heykants, J. J. P., Lavrijsen, K., Meuldermans, W., and Donne´-Op den Kelder, G. M. (1992) A predictive model for substrates of cytochrome P450-debrisoquine (2D6). Chem. Res. Toxicol. 5, 211-219. (7) De Groot, M. J., Bijloo, G. J., Hansen, K. T., and Vermeulen, N. P. E. (1995) Computer prediction and experimental validation of cytochrome P4502D6-dependent oxidation of GBR 12909. Drug Metab. Dispos. 23, 667-669. (8) De Groot, M. J., Bijloo, G. J., Martens, B. J., Van Acker, F. A. A., and Vermeulen, N. P. E. (1997) A refined substrate model for human cytochrome P450 2D6. Chem. Res Toxicol. 10, 41-48. (9) Department of Chemistry, Columbia University (1995) Macromodel, version 5.0, Columbia University, New York.

Communications (10) Department of Chemistry, Columbia University (1995) BatchMin, version 4.0, Columbia University, New York. (11) Department of Theoretical Chemistry, Vrije Universiteit Amsterdam (1997) ADF, version 2.3, Vrije Universiteit Amsterdam, Amsterdam. (12) Chemical Design Ltd. (1998) ChemX, version January 1998, Chemical Design, Ltd. (13) Hickman, D., Wang, J.-P., Wang, Y., and Unadkat, J. D. (1998) Evaluation of the selectivity of in vitro probes and suitability of

Chem. Res. Toxicol., Vol. 12, No. 7, 1999 559 organic solvents for the measurement of human cytochrome P450 monooxygenase activities. Drug Metab. Dispos. 26, 207-215. (14) Shimada, T., Yamazaki, H., Mimura, M., Inui, Y., and Guengerich, F. P. (1994) Interindividual variations in human liver cytochrome P-450 enzymes involved in the oxidation of drugs, carcinogens and toxic chemicals: studies with liver microsomes of 30 Japanese and 30 Caucasians. J. Pharmacol. Exp. Ther. 270, 414-423.

TX980248Q