Inhibition of Cytochrome P450 2D6: Structureâ Activity Studies Using a

450

Chem. Res. Toxicol. 2003, 16, 450-459

Inhibition of Cytochrome P450 2D6: Structure-Activity Studies Using a Series of Quinidine and Quinine Analogues J. Matthew Hutzler, Gregory S. Walker, and Larry C. Wienkers* Department of Global Drug Metabolism, Pharmacia Corporation, 301 Henrietta Street, Kalamazoo, Michigan 49007 Received December 6, 2002

Several pharmacophore models have suggested that substrates and inhibitors of cytochrome P450 2D6 (P450 2D6) possess a nitrogen with a positive charge that participates in a chargepair interaction with the aspartate 301 residue. In an effort to investigate this paradigm for P450 2D6 binding, an analogue series of the stereoisomers quinidine and quinine were synthesized and screened for binding affinity as measured by inhibition. Results revealed that bulky substituents added to the quinuclidine nitrogen (quaternary salts) did not affect the inhibitory potency of quinidine (IC50 ) 0.02 µM), suggesting minimal contribution to binding affinity of this inhibitor by the purported ionic-binding interaction. Meanwhile, substantial decreases in inhibitory potency were observed for the N-methyl, N-ethyl, and N-benzyl quininium salts, suggesting that the quaternary nitrogen of this antipode interacts with a distinct region of the P450 2D6 active site as compared to the corresponding nitrogen of quinidine. Interestingly, esterification of quinidine resulted in a substantial loss of inhibitory potency, likely due to disruption of a hydrogen-bonding interaction of the hydroxyl group. This suggests that hydrogen bonding contributes more to the tight binding of quinidine than does the charge-pair interaction of the positively charged nitrogen. Moreover, benzoyl ester formation of quinine caused the binding orientation to switch from type II to type I, with concomitant restoration of P450 2D6 inhibitory potency. Thus, it appears that both hydrogen bonding and the ionic interaction of the basic nitrogen of quinine contribute to inhibitory potency, while the hydroxyl group also apparently contributes to directing type II binding. Overall, results suggest that when analyzing a series of compounds that include stereoisomers for development of predictive pharmacophore/protein models describing P450 2D6 binding, it may be inappropriate to assume that the ionic interaction of the basic nitrogen with aspartate 301 represents the primary binding interaction.

Introduction P450 2D61 is a polymorphic member of the P450 superfamily of oxidative enzymes, being absent in 5-10% of Caucasians (1, 2). Despite representing only approximately 2-4% of total human hepatic P450 (3), P450 2D6 plays an important role in the oxidation of xenobiotics, metabolizing about 30% of drugs presently on the market (4, 5), in particular, a number of CNS and cardiovascular drugs with narrow therapeutic indices. Consequently, understanding the biochemical characteristics that enable substrates and inhibitors to bind to this polymorphic enzyme is a key topic in drug discovery and development. According to several pharmacophore models, common characteristics of substrates and inhibitors of P450 2D6 include the presence of at least one basic nitrogen 5-7 Å away from the site of oxidation, a flat hydrophobic region, and hydrogen-bonding properties (6, 7). In addition, it has been postulated that the basic nitrogen of substrates and/or inhibitors interacts with a negatively charged carboxylate group in the active site (6, 8). It has 1 Abbreviations: P450 2D6, cytochrome P450 2D6; Asp301, aspartate at amino acid position 301; Glu216, glutamate at amino acid position 216; CNS, central nervous system; IC50, concentration inhibiting 50% relative to control.

since been substantiated through site-directed mutagenesis experiments that this active site residue may in fact be an aspartate at position 301 (Asp301) (9, 10), as marked reductions in catalytic activity were observed for several molecules when this residue was substituted by neutral residues. Several groups have since included this proposed binding interaction into pharmacophore models describing the P450 2D6 active site (11, 12, and references therein). It is well-established that quinidine is one of the most potent known inhibitors of P450 2D6 and is often used as a chemical inhibitor of this P450 for drug-drug interaction screening assays (13). However, quinine, the stereoisomer of quinidine, is about 2 orders of magnitude less potent as an inhibitor of P450 2D6, illustrating the significance that stereochemistry may play in binding and inhibition of P450 2D6. While studies investigating the stereoselectivity of metabolism of certain P450 2D6 substrates have been performed (14, 15), little progress has been made in determining the important chemical characteristics of quinidine and quinine that contribute to their substantial differences in inhibition potency of P450 2D6. A thorough investigation into this phenomenon may provide valuable information for future predic-

10.1021/tx025674x CCC: $25.00 © 2003 American Chemical Society Published on Web 03/06/2003

Structure-Activity Inhibition Studies of P450 2D6

tive pharmacophore models describing inhibition of this enzyme. Consistent with previous reports by Ellis et al. (15, 16), we observed in preliminary studies that quinidine possesses a binding orientation that leads to a type I spectrum. However, its antipode, quinine, has a distinct binding orientation that results in a type II spectrum. This observation provided a sound rationale for conducting structure-activity studies using these two stereoisomers as lead molecules. To this end, a series of quinidine and quinine analogues were synthesized and screened for inhibition of P450 2D6, using dextromethorphan O-demethylation as a marker for P450 2D6 activity. Syntheses were strategically planned in an effort to disrupt inhibitory activity of quinidine/quinine by either (i) disrupting the proposed interaction of the basic nitrogen with the purported active site carboxylate residue by adding substantial bulk and simultaneously creating a permanent positively charged species; (ii) eliminating the potential hydrogen-bonding interaction of the C-9 benzylic hydroxyl group by esterification/ether formation; or (iii) adding significant steric bulk and/or lipophilicity to each functionality. Target areas for these syntheses are illustrated in Figure 1. Findings from these structure-activity studies involving inhibition of P450 2D6 are reported herein.

Experimental Procedures Chemicals. All acid chlorides (acetyl chloride, benzoyl chloride, 4-(tert-butyl)benzoyl chloride, p-anisoyl chloride, 4(chloro)benzoyl chloride, 4-(trifluoromethyl)benzoyl chloride, 4-(cyano)benzoyl chloride, and 2-naphthoyl chloride) and alkyl halides (methyl iodide, ethyl iodide, 4-(trifluoromethyl)benzyl bromide, 4-(nitro)benzyl bromide, and 2-bromomethyl naphthalene) were obtained from Sigma-Aldrich (Milwaukee, WI), as was N-benzyl quininium chloride. N-Benzyl quinidinium chloride was obtained from TCI (Portland, OR), and quininone was obtained from Maybridge Chemical Company (United Kingdom). Dextromethorphan HBr was purchased from Sigma-Aldrich, while dextrorphan and levallorphan were obtained from RBI (Natick, MA). Potassium phosphate (dibasic) was obtained from Mallinckrodt Baker (Paris, KY). DMSO (dimethyl sulfoxide) and anhydrous THF were obtained from EM Science (Gibbstown, NJ). DMSO-d6 and chloroform-d for NMR experiments were obtained from Cambridge Isotope Laboratories, Inc. (Andover, MA). Baculovirus insect cell-expressed recombinant P450 2D6 supersomes coexpressed with reductase were purchased from BD-Gentest (Woburn, MA), while purified P450 2D6 enzyme (without reductase or lipid) was purchased from Panvera (Madison, WI). All other chemicals were obtained from commercial sources and were of the highest purity available. Instrumentation and Analytical Methods. Purity of synthesis products was assessed by HPLC-UV using a Phenomenex Luna 5µm C18(2) 150 mm × 4.6 mm column connected in-line with a Perkin-Elmer Series 200 autosampler and pump module and a Perkin-Elmer 785A UV/vis detector set at 254 nm. Mobile phase A (20 mM ammonium acetate, pH 3.0) and mobile phase B (100% methanol) were delivered at 1.0 mL/min in a gradient fashion at an initial ratio of 85% A:15% B, ramped linearly to 10% A:90% B over 12 min, and then held for 2 min. The system was then returned to initial conditions over 1 min and allowed to equilibrate over 4 min. Correct masses of quinidine and quinine ester and quaternary salt products were verified by LC tandem mass spectrometry using an ion-trap instrument (Finnigan LCQ, San Jose, CA) fitted with an electrospray ion source. The spray voltage and capillary voltage were set at 5 kV and 35 V, respectively, with the capillary temperature being 200 °C. The sheath gas was set at 60 (arbitrary units), while the auxiliary gas was set at 4 (arbitrary

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 451

Figure 1. Structure of quinidine/quinine illustrating the target areas for synthesis of analogues. R1 represents ester/ether formation, and R2 represents quaternary salt formation. R groups and IC50 estimates for quinidine and quinine are shown in Tables 1 and 2, respectively. units). Analogues were separated with a Phenomenex Luna 5µm C8(2) 150 mm × 3.0 mm column connected to a Perkin-Elmer Series 200 autosampler and pump module. The mobile phase was delivered at 0.3 mL/min at an initial ratio of 75% A (20 mM ammonium acetate pH 3.0) and 25% B (100% methanol), ramped linearly to 95% B over 10 min, and held for 4 min. The system was then returned to initial conditions over 1 min and allowed to equilibrate for 9 min. 1H NMR spectra of all analogues were recorded (in DMSO-d6 or chloroform-d) using a 400 MHz Bruker Avance NMR (Billerica, MA). Quinidine and Quinine Esters. Quinidine and quinine esters were synthesized by simple alcoholysis of acid chlorides, reported to proceed via an SN2 mechanism (17). Reactions were performed by adding 1.1 M equivalents (1.35 mmol) of triethylamine (186 µL) to a stirring solution of 400 mg of quinidine (1) or quinine (16) (1.23 mmol) in approximately 30 mL of anhydrous THF, followed by addition of 1.1 M equivalents (1.35 mmol) of the respective acid chloride and allowed to react for 12 h. Reaction completeness was monitored by analysis of aliquots by HPLC-UV set at 254 nm. After the reaction was complete, THF solvent was removed by rotary evaporation and 50 mL of 100 mM KHCO3, pH 9.2, was added, followed by addition of ethyl acetate for extraction (20 mL × 3). This mixture was then placed into a separation funnel, shaken manually, with the ethyl acetate collected, dried with excess anhydrous magnesium sulfate (MgSO4), and filtered into a 500 mL round bottom flask for concentration by rotary evaporation. The product was then dissolved in 5 mL of ethyl acetate and dried under a stream of N2 gas. All products were determined to be >95% pure by HPLC-UV and evaluation of peak area percentages, while percent yields ranged from 32 to 94%. Quinidine and quinine esters were characterized by determination of the respective [M + H]+ ion, followed by tandem MS (CID of [M + H]+). All esters synthesized displayed a characteristic fragmentation pattern, of which the m/z 307 ion [C20H23N2O]+ (Figure 2A) was diagnostic of facile loss of the ester group plus water from quinidine and quinine, suggesting esterification of the hydroxyl group. This fragmentation pattern was consistent with the parent compounds quinidine and quinine. Characterization was as follows, with NMR data for quinidine and quinine included for comparative purposes: Quinidine. Calcd for C20H24N2O2 [M + H]+, 325.42; observed, 325.34. 1H NMR (400 MHz, DMSO-d6): δ 1.44 (m, 3 H), 1.68 (br, 1 H), 1.90 (m, 1 H), 2.18 (m, 1 H), 2.61 (m, 3 H), 3.00 (m, 2 H), 3.90 (s, 3 H), 5.07 (m, 2 H), 5.27 (m, 1 H), 5.64 (d, J ) 5.0 Hz, 1 H), 6.10 (m, 1 H), 7.38 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.46 (d, J ) 2.7 Hz, 1 H), 7.50 (d, J ) 4.4 Hz, 1 H), 7.92 (d, J ) 9.1 Hz, 1 H), 8.68 (d, J ) 4.4 Hz, 1 H). Compound 2 (Quinidine Methyl Ester). Yield: 430 mg, 95%. Calcd for C22H26N2O3 [M + H]+, 367.46; observed, 367.28. 1H NMR (400 MHz, CDCl ): δ 1.50 (m, 1 H), 1.58 (m, 2 H), 1.84 3 (m, 2 H), 2.16 (s, 3 H), 2.31 (m, 1 H), 2.78 (m, 1 H), 2.85 (m, 1 H), 2.96 (m, 2 H), 3.31 (m, 1 H), 3.98 (s, 3 H), 5.13 (m, 2 H), 6.04 (m, 1 H), 6.58 (br, 1 H), 7.35 (d, J ) 4.6 Hz, 1 H), 7.38 (dd, J ) 9.3, 2.7 Hz, 1 H), 7.42 (br, 1 H), 8.02 (d, J ) 9.3 Hz, 1 H), 8.75 (d, J ) 4.6 Hz, 1 H). Compound 3 (Quinidine Benzoate). Yield: 430 mg, 82%. Calcd for C27H28N2O3 [M + H]+, 429.53; observed, 429.32. 1H NMR (400 MHz, CDCl3): δ 1.61 (cm, 3 H), 1.88 (m, 1 H), 2.05

452

Chem. Res. Toxicol., Vol. 16, No. 4, 2003

Hutzler et al.

Figure 2. Spectra of product ions obtained by CID of the parent [M + H]+ ion (m/z 459) of quinidine 4-(methoxy) benzoyl ester. The m/z 307 product ion suggests that esterification occurs at the hydroxyl group of quinine and is the major fragment for all ester analogues. (m, 1 H), 2.33 (m, 1 H), 2.80 (m, 1 H), 2.91 (m, 1 H), 3.04 (m, 2 H), 3.45 (m, 1 H), 3.99 (s, 3 H), 5.12 (m, 2 H), 6.04 (m, 1 H), 6.83 (br, 1 H), 7.38 (dd, J ) 9.3, 2.7 Hz, 1 H), 7.42 (d, J ) 4.56 Hz, 1 H), 7.48 (m, 2 H), 7.53 (d, J ) 2.7 Hz, 1 H), 7.61 (m, 1 H), 8.03 (d, J ) 9.3 Hz, 1 H), 8.12 (d, J ) 8.3 Hz, 2 H), 8.73 (d, J ) 4.6 Hz, 1 H). Compound 4 (Quinidine 4-tert-Butylbenzoate). Yield: 390 mg, 65%. Calcd for C31H36N2O3 [M + H]+, 485.64; observed, 485.37. 1H NMR (400 MHz, CDCl3): δ 1.36 (s, 9 H), 1.61 (m, 3 H), 1.91 (m, 1 H), 2.07 (m, 1 H), 2.35 (m, 1 H), 2.83 (m, 1 H), 2.93 (m, 1 H), 306 (m, 2 H), 3.46 (m, 1 H), 4.01 (s, 3 H), 5.14 (m, 2 H), 6.06 (m, 1 H), 6.84 (br, 1 H), 7.39 (dd, J ) 9.3, 2.7 Hz, 1 H), 7.41 (d, J ) 4.6 Hz, 1 H), 7.50 (d, J ) 8.5 Hz, 2 H), 7.56 (br, 1 H), 8.02 (d, J ) 9.3 Hz, 1 H), 8.05 (d, J ) 8.5 Hz, 2 H), 8.71 (d, J ) 4.6 Hz, 1 H). Compound 5 (Quinidine 4-Methoxybenzoate). Yield: 50 mg, 11%. Calcd for C28H30N2O4 [M + H]+, 459.56; observed, 459.29. 1H NMR (400 MHz, CDCl3): δ 1.62 (m, 3 H), 1.90 (m, 1 H), 2.05 (m, 1 H), 2.34 (m, 1 H), 2.80 (m, 1 H), 2.91 (m, 1 H), 3.04 (m, 2 H), 3.45 (m, 1 H), 3.89 (s, 3 H), 4.01 (s, 3 H), 5.13 (m, 2 H), 6.05 (m, 1 H), 6.82 (br, 1 H), 6.96 (d, J ) 8.9 Hz, 2 H), 7.39 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.43 (d, J ) 4.6 Hz, 1 H), 7.55 (br, 1 H), 8.03 (d, J ) 9.1 Hz, 1 H), 8.08 (d, J ) 8.9 Hz, 2 H), 8.74 (d, J ) 4.6 Hz, 1 H). Compound 6 (Quinidine 4-Chlorobenzoate). Yield: 420 mg, 74%. Calcd for C27H27ClN2O4 [M + H]+, 463.97; observed, 463.33. 1H NMR (400 MHz, CDCl3): δ 1.61 (m, 3 H), 1.89 (m, 1 H), 1.98 (m, 1 H), 2.32 (m, 1 H), 2.78 (m, 1 H), 2.88 (m, 1 H), 2.99 (d, J ) 9.1 Hz, 2 H), 3.45 (m, 1 H), 4.00 (s, 3 H), 5.12 (m, 2 H), 6.03 (m, 1 H), 6.75 (d, J ) 7.1 Hz, 1 H), 7.40 (dd, J ) 9.3, 2.7 Hz, 1 H), 7.42 (d, J ) 4.6 Hz, 1 H), 7.47 (d, J ) 8.7 Hz, 2 H), 7.50 (d, J ) 2.7 Hz, 1 H), 8.04 (m, 3 H), 8.75 (d, J ) 4.6 Hz, 1 H). Compound 7 (Quinidine 4-Trifluoromethylbenzoate). Yield: 340 mg, 56%. Calcd for C28H27F3N2O3 [M + H]+, 497.53; observed, 497.31. 1H NMR (400 MHz, CDCl3): δ 1.65 (m, 3 H), 1.98 (m, 1 H), 2.03 (m, 1 H), 2.37 (m, 1 H), 2.82 (m, 1 H), 2.93 (m, 1 H), 3.04 (m, 2 H), 3.49 (m, 1 H), 4.02 (s, 3 H), 5.15 (m, 2 H), 6.04 (m, 1 H), 6.85 (br, 1 H), 7.41 (d, J ) 4.6 Hz, 1 H), 7.42 (dd, J ) 9.3, 2.5 Hz, 1 H), 7.54 (br, 1 H), 7.76 (d, J ) 8.3 Hz, 2 H), 8.05 (d, J ) 9.3 Hz, 1 H), 8.24 (d, J ) 8.3 Hz, 2 H), 8.74 (d, J ) 4.6 Hz, 1 H).

Compound 8 (Quinidine 4-Cyanobenzoate). Yield: 320 mg, 57%. Calcd for C28H27N3O3 [M + H]+, 454.54; observed, 454.26. 1H NMR (400 MHz, CDCl3): δ 1.64 (m, 3 H), 1.93 (m, 1 H), 1.99 (m, 1 H), 2.35 (m, 1 H), 2.82 (m, 1 H), 2.91 (m, 1 H), 3.00 (m, 2 H), 3.48 (m, 1 H), 4.01 (s, 3 H), 5.13 (m, 2 H), 6.02 (m, 1 H), 6.84 (br, 1 H), 7.40 (d, J ) 4.6 Hz, 1 H), 7.42 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.50 (br, 1 H), 7.80 (d, J ) 8.7 Hz, 2 H), 8.05 (d, J ) 9.1 Hz, 1 H), 8.21 (d, J ) 8.7 Hz, 2 H), 8.75 (d, J ) 4.6 Hz, 1 H). Quinine. Calcd for C20H24N2O2 [M + H]+, 325.42; observed, 325.30. 1H NMR (400 MHz, DMSO-d6): δ 1.41 (m, 1 H), 1.64 (m, 2 H), 1.74 (m, 2 H), 2.18 (m, 1 H), 2.42 (m, 2 H), 2.85 (m, 1 H), 3.06 (m, 1 H), 3.19 (m, 1 H), 3.90 (s, 3 H), 4.96 (m, 2 H), 5.23 (m, 1 H), 5.64 (d, J ) 5 Hz, 1 H), 5.88 (m, 1 H), 7.39 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.50 (m, 2 H), 7.92 (d, J ) 9.1 Hz, 1 H), 8.68 (d, J ) 4.6 Hz, 1 H). Compound 17 (Quinine Methyl Ester). Yield: 270 mg, 60%. Calcd for C22H26N2O3 [M + H]+, 367.46; observed, 367.30. 1H NMR (400 MHz, CDCl ): δ 1.58 (m, 2 H), 1.75 (m, 1 H), 1.89 3 (m, 2 H), 2.15 (s, 3 H), 2.34 (m, 1 H), 2.66 (m, 2 H), 3.08 (m, 1 H), 3.14 (m, 1 H), 3.40 (m, 1 H), 3.99 (s, 3 H), 5.03 (m, 2 H), 5.85 (m, 1 H), 6.50 (d, J ) 4.6 Hz, 1 H), 7.38 (m, 2 H), 7.47 (d, J ) 2.5 Hz, 1 H), 8.04 (d, J ) 9.3 Hz, 1 H), 8.76 (d, J ) 4.6 Hz, 1 H). Compound 18 (Quinine Benzoate). Yield: 360 mg, 68%. Calcd for C27H28N2O3 [M + H]+, 429.53; observed, 429.24. 1H NMR (400 MHz, CDCl3): δ 1.63 (m, 1 H), 1.80 (m, 2 H), 1.95 (m, 2 H), 2.35 (m, 1 H), 2.73 (m, 2 H), 3.15 (m, 1 H), 3.27 (m, 1 H), 3.53 (m, 1 H), 4.01 (s, 3 H), 5.04 (m, 2 H), 5.86 (m, 1 H), 6.81 (br, 1 H), 7.40 (dd, J ) 9.2, 2.7 Hz, 1 H), 7.44 (d, J ) 4.6 Hz, 1 H), 7.50 (m, 2 H), 7.56 (br, 1 H), 7.63 (m, 1 H), 8.04 (d, J ) 9.2 Hz, 1 H), 8.15 (m, 2 H), 8.74 (d, J ) 4.6 Hz, 1 H). Compound 19 (Quinine 4-Methoxybenzoate). Yield: 250 mg, 44%. Calcd for C28H30N2O4 [M + H]+, 459.56; observed, 459.32. 1H NMR (400 MHz, CDCl3): δ 1.63 (m, 1 H), 1.82 (m, 2 H), 1.93 (m, 2 H), 2.36 (m, 1 H), 2.74 (m, 2 H), 3.15 (m, 1 H), 3.29 (m, 1 H), 3.54 (m, 1 H), 3.89 (s, 3 H), 4.02 (s, 3 H), 5.03 (m, 2 H), 5.86 (m, 1 H), 6.81 (br, 1 H), 6.97 (d, J ) 8.9 Hz, 2 H), 7.40 (dd, J ) 9.2, 2.5 Hz, 1 H), 7.43 (d, J ) 4.6 Hz, 1 H), 7.56 (br, 1 H), 8.04 (d, J ) 9.2 Hz, 1 H), 8.07 (d, J ) 8.9 Hz, 2 H), 8.73 (d, J ) 4.6 Hz, 1 H).

Structure-Activity Inhibition Studies of P450 2D6 Compound 20 (Quinine 4-Chlorobenzoate). Yield: 180 mg, 32%. Calcd for C27H27ClN2O4 [M + H]+, 463.97; observed, 463.32. 1H NMR (400 MHz, CDCl3): δ 1.65 (m, 1 H), 1.79 (m, 2 H), 1.95 (m, 2 H), 2.37 (m, 1 H), 2.75 (m, 2 H), 3.15 (m, 1 H), 3.25 (m, 1 H), 3.53 (m, 1 H), 4.03 (s, 3 H), 5.06 (m, 2 H), 5.85 (m, 1 H), 6.80 (br, 1 H), 7.41 (m, 2 H), 7.48 (d, J ) 8.5 Hz, 2 H), 7.55 (br, 1 H), 8.04 (m, 3 H), 8.74 (d, J ) 4.6 Hz, 1 H). Compound 21 (Quinine 4-Trifluoromethylbenzoate). Yield: 320 mg, 52%. Calcd for C28H27F3N2O3 [M + H]+, 497.53; observed, 497.33. 1H NMR (400 MHz, CDCl3): δ 1.63 (m, 1 H), 1.73 (m, 1 H), 1.80 (m, 1 H) 1.97 (m, 2 H), 2.36 (m, 1 H), 2.73 (m, 2 H), 3.13 (m, 1 H), 3.22 (m, 1 H), 3.55 (m, 1 H), 4.02 (s, 3 H), 5.06 (m, 2 H), 5.87 (m, 1 H), 6.80 (br, 1 H), 7.42 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.44 (d, J ) 4.6 Hz, 1 H), 7.54 (br, 1 H), 7.77 (d, J ) 8.3 Hz, 2 H), 8.05 (d, J ) 8.3 Hz, 1 H), 8.22 (d, J ) 7.9 Hz, 2 H), 8.75 (d, J ) 4.6 Hz, 1 H). Compound 22 (Quinine Naphthoate). Yield: 370 mg, 63%. Calcd for C31H30N2O3 [M + H]+, 479.59; observed, 479.33. 1H NMR (400 MHz, CDCl3): δ 1.65 (m, 1 H), 1.87 (m, 2 H), 1.97 (m, 2 H), 2.37 (m, 1 H), 2.77 (m, 2 H), 3.16 (m, 1 H), 3.31 (m, 1 H), 3.59 (m, 1 H), 4.04 (s, 3 H), 5.06 (m, 2 H), 5.88 (m, 1 H), 6.89 (br, 1 H), 7.41 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.51 (d, J ) 4.6 Hz, 1 H), 7.61 (m, 3 H), 7.93 (m, 2 H), 8.00 (d, J ) 7.9 Hz, 1 H), 8.06 (d, J ) 9.1 Hz, 1 H), 8.13 (dd, J ) 8.5, 1.6 Hz, 1 H), 8.68 (d, J ) 1.6 Hz, 1 H), 8.76 (d, J ) 4.6 Hz, 1 H). Quinidine and Quinine Quaternary Ammonium Salts. Quaternary ammonium salts were synthesized by alkylation of the quinuclidine tertiary nitrogen (Menshutkin reaction) (18). This was performed by adding 1.1 M equivalents (1.35 mmol) of each alkyl halide to a stirring solution of 400 mg of quinidine (1) or quinine (16) (1.23 mmol) in approximately 30 mL of anhydrous THF and allowed to react for 12 h. Salt formation was often observed by precipitation during the reaction and was centrifuged and washed (15 mL × 3) using either ethyl acetate or hexane to extract remaining reactants. The solvent was then decanted off, and the precipitate was dried under N2 gas. For those products that remained dissolved in THF despite salt formation (due to overall lipophilicity), hexane was added to precipitate the salt product, followed by washing with hexane (15 mL × 3) and drying under N2 gas. Synthesis of N-methyl methoxy quinidinium salt (compound 9) was accomplished by dissolving 400 mg of quinidine (1) (1.23 mmol) in approximately 30 mL of anhydrous THF in a 250 mL round bottom flask. This solution was cooled in an ice-bath prior to addition of 60 mg (2.50 mmol) of sodium hydride (60% dispersion). This solution was then warmed to room temperature, and 230 µL (3.70 mmol) of methyl iodide (CH3I) was added and allowed to react ∼12 h. The reaction was then quenched by addition of 1 mL of MeOH, evaporated to dryness by rotary evaporation, and dissolved in acetone. The acetone solution was washed with hexane (15 mL × 2) and with H2O (15 mL × 1), redissolved in acetone, and dried under N2 gas. All salts were determined to be >95% pure by HPLC-UV, with percent yields ranging from 5 to 95%. Quaternary salt formation was first diagnosed by formation of the appropriate [M]+ molecular ion, followed by tandem MS (CID of [M]+). Most synthesized salts displayed a characteristic fragmentation pattern with an m/z 160 ion representing loss of the quinoline ring [C10H9NO + H]+, consistent with alkylation of the quinuclidine nitrogen and not the quinoline ring nitrogen, while all salts displayed an m/z ion consistent with loss of H2O (-18). Compounds 13 and 16 both showed an m/z fragment of 202 [C13H15NO + H]+, representing the quinoline ring system plus the benzylic carbon and two carbons from the quinuclidine ring system, supporting alkylation of the quinuclidine ring nitrogen as well. Tandem MS of compounds 15 and 27 suggested facile loss of the naphthyl group [C11H9]+ (m/z 141) from the quinuclidine nitrogen, which resulted in a fragmentation pattern consistent with the ester analogues fragmentation patterns. Characterization was as follows. Compound 9 (N-Methyl Methoxy Quinidinium Iodide). Yield: 410 mg, 94%. Calcd for C22H29N2O2 [M]+, 353.48; observed, 353.37. 1H NMR (400 MHz, CDCl3): δ 1.14 (m, 1 H),

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 453 1.91 (m, 2 H), 2.08 (m, 1 H), 2.12 (m, 1 H), 2.41 (m, 1 H), 2.96 (m, 1 H), 3.45 (s, 3 H), 3.84 (s, 3 H), 4.11 (s, 3 H), 4.22 (m, 4 H), 5.30 (m, 2 H), 5.90 (m, 2 H), 7.34 (br, 1 H), 7.44 (dd, J ) 9.3, 2.5 Hz, 1 H), 7.66 (br, 1 H), 8.10 (d, J ) 9.3 Hz, 1 H), 8.83 (d, J ) 4.6 Hz, 1 H). Compound 10 (N-Methyl Quinidinium Iodide). Yield: 280 mg, 67%. Calcd for C21H27N2O2 [M]+, 339.46; observed, 339.38. 1H NMR (400 MHz, DMSO-d6): δ 1.02 (m, 1 H), 1.87 (m, 3 H), 2.27 (m, 1 H), δ 2.78 (m, 1 H), 3.31 (s, 3 H), 3.57 (m, 3 H), 3.76 (m, 1 H), 4.03 (s, 3 H), 4.20 (m, 1 H), 5.25 (m, 2 H), 6.03 (m, 1 H), 6.20 (br, 1 H), 6.66 (d, J ) 3.3 Hz, 1 H), 7.26 (d, J ) 2.7 Hz, 1 H), 7.48 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.71 (d, J ) 4.6 Hz, 1 H), 8.00 (d, J ) 9.1 Hz, 1 H), 8.80 (d, J ) 4.6 Hz, 1 H). Compound 11 (N-Ethyl Quinidinium Iodide). Yield: 200 mg, 92%. Calcd for C22H29N2O2 [M]+, 353.48; observed, 353.41. 1H NMR (400 MHz, DMSO-d ): δ 0.97 (m, 1 H), 1.52 (t, J ) 7.0 6 Hz, 3 H), 1.83 (m, 3 H), 2.27 (m, 1 H), 2.75 (m, 1 H), 3.38 (m, 1 H), 3.67 (m, 6 H), 4.00 (s, 3 H), 4.08 (m, 1 H), 5.25 (m, 2 H), 6.00 (m, 1 H), 6.17 (br, 1 H), 6.65 (d, J ) 3.1 Hz, 1 H), 7.23 (d, J ) 2.7 Hz, 1 H), 7.49 (dd, J ) 9.3, 2.7 Hz, 1 H), 7.72 (d, J ) 4.6 Hz, 1 H), 8.00 (d, J ) 9.3 Hz, 1 H). Compound 12 (N-Benzyl Quinidinium Chloride). See Chemicals section. Compound 13 (N-4-Trifluoromethyl Benzyl Quinidinium Bromide). Yield: 286 mg, 48%. Calcd for C28H30F3N2O2 [M]+, 483.55; observed, 483.42. 1H NMR (400 MHz, DMSO-d6): δ 1.11 (m, 1 H), 1.77 (m, 2 H), 1.90 (m, 1 H), 2.40 (m, 1 H), 2.66 (m, 1 H), 2.95 (m, 1 H), 3.50 (m, 1 H), 3.85 (m, 1 H), 3.97 (m, 1 H), 4.05 (s, 3 H), 4.24 (m, 1 H), 4.83 (d, J ) 12.6 Hz, 1 H), 5.06 (d, J ) 12.6 Hz, 1 H), 5.23 (m, 2 H), 6.02 (m, 1 H), 6.50 (br, 1 H), 6.82 (d, J ) 3.3 Hz, 1 H), 7.43 (d, J ) 2.7 Hz, 1 H), 7.51 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.76 (d, J ) 4.6 Hz, 1 H), 7.96 (s, 4 H), 8.02 (d, J ) 9.1 Hz, 1 H), 8.82 (d, J ) 4.6 Hz, 1 H). Compound 14 (N-4-Nitrobenzyl Quinidinium Bromide). Yield: 500 mg, 88%. Calcd for C27H30N3O4 [M]+, 460.55; observed, 460.37. 1H NMR (400 MHz, DMSO-d6): δ 1.13 (m, 1 H), 1.77 (m, 2 H), 1.90 (br, 1 H), 2.40 (1, J ) m Hz, 1 H), 2.63 (m, 1 H), 2.97 (m, 1 H), 3.50 (m, 1 H), 3.86 (m, 1 H), 4.01 (m, 1 H), 4.07 (s, 3 H), 4.26 (m, 1 H), 4.91 (d, J ) 12.4 Hz, 1 H), 5.15 (d, J ) 12.4 Hz, 1 H), 5.23 (m, 2 H), 6.03 (m, 1 H), 6.50 (br, 1 H), 6.84 (d, J ) 3.7 Hz, 1 H), 7.43 (d, J ) 2.5 Hz, 1 H), 7.52 (dd, J ) 9.1, 2.5 Hz, 1 H), 7.78 (d, J ) 4.6 Hz, 1 H), 8.03 (d, J ) 9.1 Hz, 1 H), 8.05 (d, J ) 8.7 Hz, 2 H), 8.42 (d, J ) 8.71 Hz, 2 H), 8.82 (d, J ) 4.6 Hz, 1 H). Compound 15 (N-Naphthyl Quinidinium Bromide). Yield: 540 mg, 94%. Calcd for C31H33N2O2 [M]+, 465.62; observed, 465.39. 1H NMR (400 MHz, DMSO-d6): δ 1.09 (m, 1 H), 1.73 (m, 2 H), 1.87 (br, 1 H), 2.38 (m, 1 H), 2.60 (m, 1 H), 2.98 (m, 1 H), 3.59 (m, 1 H), 3.88 (m, 1 H), 3.99 (m, 1 H), 4.05 (s, 3 H), 4.25 (m, 1 H), 4.90 (d, J ) 12.6 Hz, 1 H), 5.11 (d, J ) 12.6 Hz, 1 H), 5.21 (m, 2 H), 6.02 (m, 1 H), 6.56 (br, 1 H), 6.84 (d, J ) 3.2 Hz, 1 H), 7.44 (d, J ) 2.4 Hz, 1 H), 7.50 (dd, J ) 9.1, 2.4 Hz, 1 H), 7.64 (m, 2 H), 7.77 (m, 2 H), 8.01 (m, 2 H), 8.07 (m, 2 H), 8.30 (s, 1 H), 8.81 (d, J ) 4.4 Hz, 1 H). Compound 23 (N-Methyl Quininium Iodide). Yield: 400 mg, 96%. Calcd for C21H27N2O2 [M]+, 339.46; observed, 339.41. 1H NMR (400 MHz, DMSO-d ): δ 1.34 (m, 1 H), 1.92 (m, 1 H), 6 2.04 (m, 1 H), 2.14 (m, 2 H), 2.80 (m, 1 H), 3.39 (s, 3 H), 3.42 (m, 1 H), 3.63 (m, 2 H), 3.69 (m, 1 H), 4.00 (s, 3 H), 4.07 (m, 1 H), 5.01 (m, 1 H), 5.13 (m, 1 H), 5.75 (m, 1 H), 6.22 (d, J ) 3.5 Hz, 1 H), 6.52 (d, J ) 3.5 Hz, 1 H), 7.20 (d, J ) 2.7 Hz, 1 H), 7.48 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.70 (d, J ) 4.6 Hz, 1 H), 8.00 (d, J ) 9.1 Hz, 1 H), 8.79 (d, J ) 4.6 Hz, 1 H). Compound 24 (N-Ethyl Quininium Iodide). Yield: 23 mg, 5%. Calcd for C22H29N2O2 [M]+, 353.48; observed, 353.43. 1H NMR (400 MHz, DMSO-d6): δ 1.34 (m, 1 H), 1.45 (t, J ) 7.1 Hz, 3 H), 1.91 (m, 1 H), 2.01 (m, 1 H), 2.14 (m, 2 H), 2.79 (m, 1 H), 3.38 (m, 1 H), 3.46 (m, 1 H), 3.68 (m, 1 H), 3.77 (m, 3 H), 3.94 (m, 1 H), 3.99 (s, 3 H), 5.00 (m, 1 H), 5.13 (m, 1 H), 5.76 (m, 1 H), 6.62 (d, J ) 3.3 Hz, 1 H), 6.53 (d, J ) 3.3 Hz, 1 H), 7.26 (d, J ) 2.6 Hz, 1 H), 7.48 (dd, J ) 9.1, 2.7 Hz, 1 H), 7.72 (d,

454


J ) 4.6 Hz, 1 H), 8.00 (d, J ) 9.1 Hz, 1 H), 8.80 (d, J ) 4.6 Hz, 1 H). Compound 25 (N-Benzyl Quininium Chloride). See Chemicals section. Compound 26 (N-4-Trifluoromethyl Benzyl Quininium Bromide). Yield: 170 mg, 29%. Calcd for C28H30F3N2O2 [M]+, 483.55; observed, 483.44. 1H NMR (400 MHz, DMSO-d6): δ 1.45 (m, 1 H), 1.81 (m, 1 H), 2.01 (m, 1 H), 2.14 (m, 1 H), 2.25 (m, 1 H), 2.67 (m, 1 H), 3.24 (m, 1 H), 3.30 (m, 1 H), 3.68 (m, 1 H), 3.85 (m, 1 H), 4.00 (s, 3 H), 4.21 (m, 1 H), 4.79 (d, J ) 12.3 Hz, 1 H), 5.00 (m, 1 H), 5.09 (m, 1 H), 5.47 (d, J ) 12.3 Hz, 1 H), 5.74 (m, 1 H), 6.56 (d, J ) 3.9 Hz, 1 H), 6.71 (d, J ) 3.9 Hz, 1 H), 7.38 (d, J ) 2.5 Hz, 1 H), 7.51 (dd, J ) 9.1, 2.5 Hz, 1 H), 7.76 (d, J ) 4.5 Hz, 1 H), 7.92 (d, J ) 8.5 Hz, 2 H), 7.95 (d, J ) 8.5 Hz, 2 H), 8.03 (d, J ) 9.1 Hz, 1 H), 8.82 (d, J ) 4.5 Hz, 1 H). Compound 27 (N-Naphthyl Quininium Bromide). Yield: 200 mg, 35%. Calcd for C31H33N2O2 [M]+, 465.62; observed, 465.37. 1H NMR (400 MHz, DMSO-d6): δ 1.44 (m, 1 H), 1.78 (m, 1 H), 1.98 (m, 1 H), 2.14 (m, 1 H), 2.24 (m, 1 H), 2.62 (m, 1 H), 3.30 (m, 2 H), 3.40 (m, 1 H), 3.68 (m, 1 H), 3.88 (m, 1 H), 4.00 (s, 3 H), 4.23 (m, 1 H), 4.83 (d, J ) 12.2 Hz, 1 H), 4.99 (m, 1 H), 5.08 (m, 1 H), 5.49 (d, J ) 12.2 Hz, 1 H), 5.73 (m, 1 H), 6.60 (d, J ) 3.7 Hz, 1 H), 6.72 (d, J ) 3.7 Hz, 1 H), 7.42 (d, J ) 2.6 Hz, 1 H), 7.50 (dd, J ) 9.1, 2.6 Hz, 1 H), 7.63 (m, 2 H), 7.73 (dd, J ) 8.6, 1.9 Hz, 1 H), 7.75 (d, J ) 4.5 Hz, 1 H), 8.01 (m, 2 H), 8.08 (d, J ) 8.6 Hz, 1 H), 8.23 (d, J ) 1.9 Hz, 1 H), 8.81 (d, J ) 4.45 Hz, 1 H). Stability of Analogues. Chemical and metabolic stability of all quinidine and quinine analogues was assessed. For chemical stability, each analogue was placed into 100 mM potassium phosphate buffer, pH 7.4, at a concentration of 10 µM and incubated at 37 °C for 4 h, an appropriate time considering an incubation time of only 10 min (see below). Stability was assessed by analysis of peaks using the HPLCUV system described above. Metabolic stability was assessed by incubating each analogue (2 µM) in 100 mM potassium phosphate buffer, pH 7.4, with 10 pmol expressed P450 2D6 and 1 mM NADPH for 45 min, in a final volume of 200 µL. Reactions were performed in a 96 well plate, quenched with 100 µL of acetonitrile, and centrifuged, and 10 µL was injected onto the LC-MS (LCQ) system described above. Hydroxylation (M + 16) and demethylation (M - 14) metabolic pathways were monitored. Incubation Conditions for IC50 Determinations. Incubations were performed in a 96 well plate, with a final volume of 170 µL. Briefly, dextromethorphan was incubated with recombinant P450 2D6 supersomes (2 pmol) at a concentration approximately equal to the Km (1.7 µM) in 100 mM potassium phosphate, pH 7.4, buffer. Quinidine, quinine, and analogues were dissolved in either DMSO or acetone and were added at concentrations ranging from 0.0001 to 300 µM (in triplicate) such that incubations contained 1% organic (v/v). After 3 min of preincubation, reactions were initiated by addition of 1 mM NADPH, allowed to continue for 10 min in a 37 °C incubator, and quenched with 75 µL of acetonitrile containing levallorphan (0.125 µg/mL) as internal standard. The plate was then centrifuged at 1374g for 5 min, and the dextromethorphan Odemethylation product (dextrorphan) was measured by LC-MS analysis. Dextrorphan Analysis. Analysis of dextrorphan formation was performed by LC-MS on a PerkinElmer (Norwalk, CT) Sciex API 150 single quadrupole mass spectrometer connected to Perkin-Elmer (Norwalk, CT) Series 200 micropumps and autosampler. Ions generated by electrospray ionization were detected using selected-ion monitoring (SIM) in positive-ion mode. Dextrorphan formation was detected by analysis of [M + H]+ ion at m/z 258, with m/z 284 for levallorphan as internal standard. Analytical separation was accomplished using a Zorbax SB-C18 2.1 mm × 15 cm column with mobile phase A (0.1% formic acid) and B (100% acetonitrile) delivered at 0.3 mL/min at an initial ratio of 80% A:20% B with a linear gradient to 90% B over the first 2 min, held for another 1.5 min, and then back to initial

Hutzler et al. conditions for reequilibration over the next 4.5 min. Data were analyzed using Sciex Analyst version 1.2 software, and IC50 values were estimated by plotting percent activity remaining vs log10 of the inhibitor concentration and using nonlinear regression analysis in Graph Pad PRISM 3.0 graphing software (San Diego, CA). Spectral Binding. Binding spectra were recorded at room temperature on a Hitachi (Danbury, CT) U3300 dual-beam spectrophotometer, and data were analyzed using UV Solutions 1.2 software. Purified P450 2D6 (without reductase) was suspended in 100 mM potassium phosphate, pH 7.4, buffer to 1 µM and evenly divided (0.9 mL) between sample and reference 10 mm semimicroblocked cuvettes. After a baseline recording, P450 2D6 was titrated with methanolic solutions of quinidine, quinine, or analogue via 1 µL aliquots to the sample cuvette, with 1 µL of methanol added to the reference cuvette such that the added volume did not exceed 2% of the sample volume. Spectra were recorded between 350 and 500 nm after each substrate aliquot addition. For type I binding, the peak absorbance was ∼390 nm and the trough was ∼420 nm, while for type II spectra, the trough was ∼410 nm and the peak was ∼435 nm. The difference in absorbance between the peak and the trough for type I and type II spectra was then plotted vs titrant concentration and fit to the following equation for estimation of Ks values using Graph Pad PRISM 3.0 graphing software (San Diego, CA).

∆A ) BmaxS/Ks + S

(1)

Lipophilicity. ClogP values were calculated using the Hansch/Leo method within BioByte ClogP software (Claremont, CA).

Results and Discussion Binding interactions of many substrates and inhibitors of P450 2D6 with the active site have been extensively studied and modeled in an effort to understand the important structural features and biochemical interactions of this polymorphic enzyme. As suggested from a number of these studies, the underlying feature that is common for most substrates and inhibitors of P450 2D6 appears to be the presence of at least one basic nitrogen 5-7 Å away from the site of metabolism (8, 19). In addition, it has been reported that this nitrogen may form an ion pair with a negatively charged carboxylate group within the active site (6). On the basis of site-directed mutagenesis studies, it has been suggested that this ionic interaction takes place with an aspartate residue at position 301 within the I helix (9, 10). This information has since been combined with protein structural data to generate small molecule models in an effort to include steric restrictions and orientational preferences within the P450 2D6 active site (11 and references therein). Initial findings from our study suggest that quinidine and quinine possess distinct preferred binding orientations within the active site, explaining their drastically different inhibition potency. As shown in Figure 3A, quinidine displays a type I binding spectrum upon binding with P450 2D6, while quinine binding results in a type II spectrum (Figure 3B). A type I binding spectrum indicates that quinidine binding causes a low to high spin state transition resulting from displacement of water as the sixth ligand to the heme iron (20). Meanwhile, the type II binding spectrum of quinine suggests that the sp2 nitrogen in the quinoline ring, likely to be uncharged at pH 7.4 (pKa ) 5.3), is complexed with the heme iron (20), which leads to a more definitive binding orientation, clearly distinct from that of quinidine. Thus, the differences in binding orientation may explain the inhibition


Figure 3. Spectral binding for (A) quinidine, showing a type I binding orientation, and (B) quinine, showing a type II binding orientation, illustrating the effects of different stereochemistry of these inhibitors on overall binding orientation within the P450 2D6 active site.

potential differences displayed by these stereoisomers (quinidine IC50 ) 0.02 µM; quinine IC50 ) 4.9 µM), consistent with previously reported data (16). Interestingly, both of these compounds possess a quinuclidine ring with a basic tertiary nitrogen, questioning the validity of some assumptions made to date regarding the binding interaction of basic nitrogen-containing compounds and the aspartate 301 residue of P450 2D6. Quinidine and Quinine Quaternary Salts. The intention of this portion of the study was to investigate the importance of the interaction of the basic nitrogen


for both quinidine and quinine, by alkylating the quinuclidine ring tertiary nitrogen with bulky substituents, while at the same time creating a permanent positively charged species. All quinidine and quinine salt analogues were found to be chemically stable over a 4 h incubation period in potassium phosphate buffer, pH 7.4, at 37 °C. In addition, they were metabolically stable over a 45 min incubation with expressed P450 2D6 (data not shown), consistent with previously reported data for the parent compounds quinidine and quinine (21, 22). This observation provided a level of confidence that inhibitor consumption, which has been shown previously to result in inaccurate inhibition parameter estimates (23), would not be a confounding factor. Determined IC50 values for quinidine and its synthesized salt analogues are shown in Table 1 (compounds 10-15). The observation that addition of functional groups as large as a naphthyl ring to the quinuclidine basic nitrogen had no effect on inhibition potency of quinidine was unanticipated. This suggested that perhaps the interaction of the positively charged nitrogen with an active site carboxylate is far more accommodating than expected. Moreover, this specific interaction may not be an important point of contact allowing potent inhibition by quinidine. It would be reasonable to assume that either less inhibition would be observed due to steric interference of the additional N-substituted functional groups or perhaps an increase due to the permanent positive charge on the nitrogen, as opposed to a pH-dependent protonation state. The latter of these hypotheses is born out of recent evidence that protonation states of P450 2D6 substrates may affect binding to this enzyme (24). Interestingly, this same group has recently found that spirosulfonamide and a number of steroids, compounds lacking a basic nitrogen, have high affinity for P450 2D6, suggesting that the aspartate 301 interaction is not necessary for catalytic activity by P450 2D6 (25). This may also be true of some inhibitors of P450 2D6, which is consistent with our results generated from the quinidine quaternary salt analogues (compounds 10-15).

Table 1. IC50 Estimates for Quinidine (1), Synthesized Ester (2-8), and Quaternary Salt (9-15) Analoguesa

a

Each IC50 was estimated by evaluation of 13 concentrations (each in triplicate) of inhibitor.

456


Hutzler et al.

Table 2. IC50 Estimates for Quinine (16), Synthesized Ester (17-22), and Quaternary Salt (23-27) Analoguesa

a

Each IC50 was estimated by evaluation of 13 concentrations (each in triplicate) of inhibitor.

Conversely, alkylation of the quinine tertiary quinuclidine nitrogen resulted in substantial changes to inhibitory potency (Table 2). Compounds 23-25 all had substantially higher IC50 values as compared to quinine (compound 16). This observation is in agreement with Strobl et al., as this group reported that N-benzylquininium chloride (compound 25) had a Ki of 25 µM, much higher than the Ki of 4.6 µM observed for quinine (7). In this study by Strobl et al., analogues of quinidine and quinine were examined in an effort to generate a pharmacophore for inhibition of P450 2D6 (7). However, the number of quinine salt analogues analyzed in that study was minimal and binding orientation within the active site was not considered. Meanwhile, in the current study, inhibition potency was restored once a trifluoromethyl benzyl group (compound 26) or naphthyl group (compound 27) was added to the tertiary nitrogen, possibly a result of added lipophilicity. While results of the current study cannot definitively distinguish which of the basic nitrogens of quinidine and quinine interacts with the Asp301 residue, it appears that they likely do not have the same point of contact within the active site. In light of these observations, it may be inappropriate to overlay these compounds when constructing a pharmacophore model meant to predict binding of these stereoisomers. Interestingly, it has been suggested that in addition to Asp301, the glutamate residue at position 216 (Glu216) within the F helix may also play a role in binding of certain P450 2D6 substrates/inhibitors. Modeling studies performed by Venhorst et al. suggested that diMMAMC, an analogue of 7-methoxy-4-(aminomethyl)coumarin, interacts with Glu216, and not the Asp301 residue (26). Similar studies by de Groot et al. have also implicated Glu216 as playing a role in binding of certain substrates of P450 2D6 (11). Most recently, another group has demonstrated through site-directed mutagenesis studies that removal of the negative charge at position 216 by substitution with neutral residues substantially increased the Km values for bufuralol and dextromethorphan (27), further implicating this residue as important in binding certain compounds. Although appropriate experiments would have to be performed to investigate this possibility, the potential contribution of the Glu216

Figure 4. IC50 plots for quinidine, quininone (a commercially available compound with a carbonyl in the place of the hydroxyl group), compound 9, and quinine, illustrating the effects of disrupting the hydrogen-bonding character of the hydroxyl group on inhibition of P450 2D6 by quinidine.

residue may explain some observations from this part of the study. Quinidine and Quinine Esters. Esterification of quinidine was conducted to test the hypothesis that the C-9 benzylic hydroxyl group of quinidine may contribute to tight binding by a hydrogen-bonding interaction, consistent with suggestions by other groups (13, 28) but never verified experimentally. Similar to the salt analogues, all synthesized esters were stable for at least 4 h when incubated in potassium phosphate buffer, pH 7.4, at 37 °C and were metabolically stable. As expected, analogues lacking the potential for hydrogen bonding had higher IC50 estimates. Figure 4 illustrates the shift of the IC50 inhibition curve due to disruption of the hydrogenbonding interaction of quinidine, as formation of an ether (methylation of the hydroxyl group) with minimal steric bulk (compound 9) changed the potency of P450 2D6 inhibition by quinidine to that of a quinine-like compound. According to Lewis and Dickins, hydrogen bonding represents a major factor governing binding affinity of substrates for P450 enzymes (-2.0 kcal/mol), supporting the notion that loss of this interaction would greatly reduce the affinity of P450 2D6 for quinidine (29). In addition, when the hydroxyl group was replaced by a carbonyl group (quininone), inhibition potency was reduced as well (Figure 4). These observations, coupled with the results of the quinidine quaternary salts,


Figure 5. Correlation of the Hammett substituent parameter (σ) for field effects and IC50 estimates for a series of quinidine benzoyl esters, illustrating the importance of an electrostatic interaction within the active site of P450 2D6 for inhibition.

suggest that the hydrogen-bonding interaction of quinidine is of equal if not more importance than the ionic interaction of the basic nitrogen in contributing to the binding and inhibition potency of this compound. For quinine, it has been suggested that the stereochemistry at C-9 (R to quinoline ring) positions the hydroxyl group in such a way that it is sterically less favorable for a hydrogen bond to take place (28), explaining the lower inhibitory potential. However, our findings suggest that this hydroxyl group may indeed be involved in a hydrogen-bonding interaction, despite the reduced inhibition potential. Similar to quinidine, upon esterification of quinine with an acetyl group (compound 17), inhibitory potency was drastically decreased (Table 2), arguing in favor of a role for a hydrogen-bonding interaction, albeit in a different vicinity of the active site relative to the hydrogen bond of quinidine. Interestingly, upon esterification with bulkier benzoyl groups (compounds


18-21), inhibition potency was restored (Table 2). In addition, we observed that quinine benzoyl ester formation resulted in quinine switching from type II to a type I binding orientation, which suggests that not only does the hydroxyl group contribute to the inhibition potency of quinine, but it also apparently contributes in directing the overall type II binding orientation within the active site of P450 2D6 (see Spectral Binding section). One of the more interesting observations from this study involves the quinidine benzoyl ester series (compounds 3-8). It was discovered that upon addition of functional groups with distinctive electronic properties on the para position of the benzoyl ester, there existed a rank order of IC50 values from 26.6 µM down to 0.4 µM (Table 2). From these data, a high coefficient of determination (r2 ) 0.96) was found between the Hammett substituent parameter (σ) for field effects of these functional groups (30) and the IC50 (Figure 5). This suggests that the para functional groups may influence binding interactions of another group on the quinidine molecule through space or solvent (field effect), depending on the geometry of the molecule in the active site. The reason for this observed phenomenon is unclear at this time but may be explained by the negatively charged electrostatic fields of differing strength around the para substituents, which may result in an electrostatic interaction with a positively charged region of the active site. Interestingly, correlations involving Hammett σ constants have been observed with other metabolic enzymes as well. Van der Aar et al. reported that for certain glutathione-S-transferase (GST) isoforms, conjugation of 2-substituted 1-chloro-4-nitrobenzenes depended on σ values of the ortho substituents (31). Overall, our observations are consistent with that of Strobl et al., whose group suggested that

Figure 6. Spectral binding for compound 13 (A) and compound 8 (B), illustrating maintainence of type I binding for quinidine salt and ester analogues, respectively. For quinine analogues, compound 27 (C) displayed a type II binding spectra like quinine, while benzoyl ester formation (compound 18) (D) caused a switch in preferred binding orientation to type I binding.

458


compounds with high inhibitory potency for P450 2D6 possess functional groups with negative molecular electrostatic potential (7). Lipophilicity. Previous studies have shown that a good correlation exists between overall lipophilicity and inhibition of P450 2D6 (32, 33). Furthermore, Venhorst et al. found a correlation of 0.98 for log IC50 vs calculated lipophilicities of a series of 4-substituted 7-methoxy coumarins (26). In contrast to these observations, determined ClogP values for all quinidine analogues, as well as the quinine ester analogues, did not correlate well with estimated P450 2D6 IC50 values. However, there appeared to be a trend between lipophilicity of the quinine quaternary salts and IC50 (Table 2), which suggests the presence of at least one hydrophobic pocket in the P450 2D6 active site. Spectral Binding. In preliminary studies, it was observed that quinidine gives a type I binding spectrum, while quinine binds in a type II manner (Figure 3A,B, respectively). Analogues selected for spectral binding studies were chosen as representatives within each analogue series for quinidine and quinine due to their low IC50 estimates, which suggested tight binding and thus a good binding spectrum. Upon screening a number of quinidine analogues for binding, we observed that neither N-substitution of the basic nitrogen nor esterification of quinidine affected overall binding orientation, as both compounds 13 (Figure 6A) and 8 (Figure 6B), selected as representatives from within the quaternary salt and ester series, respectively, showed type I binding. However, while N-alkylation (compound 27) did not affect the type II binding orientation of quinine (Figure 6C), esterification (compound 18) of quinine switched the binding orientation from type II to that of a type I orientation (Figure 6D). It is possible, however, that the added steric bulk of the benzoyl esters caused the orientation conversion. In an effort to examine this possibility, attempts were made to acquire a binding spectrum for compound 17 to determine if this analogue with low steric bulk was also converted from type II to type I binding. Unfortunately, the low binding affinity of this compound toward P450 2D6 (IC50 ) 29.1 µM) precluded obtaining this information. Nonetheless, it is evident from this observation that when overlaying and comparing structural analogues for pharmacophore modeling, it is imperative to consider how additional functional groups may direct binding orientation within the active site, as this may compromise the predictive ability of these models. In addition, a good correlation (r2 ) 0.97) was observed between spectral binding constant (Ks) and IC50 estimates for several type I compounds (Figure 7). The ability to obtain binding spectra for a number of the tight binding analogues ensured that these compounds were accessing the active site and interacting in some type of competitive manner, providing confidence in our IC50 values as a measure of inhibition. Concluding Remarks. In summary, structure-activity studies with an analogue series of the stereoisomers quinidine and quinine suggest that hydrogen bonding of the hydroxyl group, and not the basic nitrogen interaction with an active site acid residue, controls the inhibitory potency of quinidine. Substantial increases in IC50 were observed after disruption of this hydrogen-bonding interaction, while no effect was observed after addition of bulky functional groups to the quinuclidine tertiary nitrogen. Meanwhile, the hydroxyl group of quinine also

Hutzler et al.

Figure 7. Correlation of spectral binding constant (Ks) with IC50 estimates for several analogues displaying a type I binding spectra. Numbers in parentheses indicate the compound numbers designated in Tables 1 and 2.

appears to be involved in a hydrogen-bonding interaction, which at least partly contributes to the type II binding orientation of quinine. Last, formation of quaternary salts of quinine resulted in substantial decreases in inhibitory potency, while added lipophilic substituents eventually restored inhibitory potential. Overall, results from this study suggest that a simple pharmacophore model describing commonalities between substrates/inhibitors of P450 2D6 may be inappropriate, as it appears that certain compounds containing a basic nitrogen may possess distinct binding orientations, compromising the predictive ability of such models.

Acknowledgment. We gratefully acknowledge Nico P. E. Vermeulen for helpful discussion and Mark P. Grillo and Frank J. Powers for assistance with syntheses of analogues.

References (1) Mahgoub, A., Idle, J. R., Dring, L. G., Lancaster, R., and Smith, R. L. (1977) Polymorphic oxidation of debrisoquine in man. Lancet 11, 584-586. (2) Eichelbaum, M., Spannbrucker, N., Steineke, B., and Dengler, H. J. (1979) Defective N-oxidation of sparteine in man: a new pharmacogenetic effect. Eur. J. Clin. Pharmacol. 16, 183-187. (3) 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. (4) Guengerich, F. P. (1995) Human Cytochrome P450 Enzymes. In Cytochrome P450: Structure, Mechanism, and Biochemistry (Ortiz de Montellano, P. R., Ed.) pp 473-535, Plenum Press, New York. (5) Smith, D. A., Abel, S. M., Hyland, R., and Jones, B. C. (1998) Human cytochrome P450s: selectivity and measurement in vivo. Xenobiotica 28, 1095-1128. (6) Koymans, L., Vermeulen, N. P., van Acker, S. A., te Koppele, J. M., Heykants, J. J., Lavrijsen, K., Meuldermans, W., and DonneOp den Kelder, G. M. (1992) A predictive model for substrates of cytochrome P450-debrisoquine (2D6). Chem. Res. Toxicol. 5, 211219. (7) Strobl, G. R., von Kruedener, S., Stockigt, J., Guengerich, F. P., and Wolff, T. (1993) Development of a pharmacophore for inhibition of human liver cytochrome P-450 2D6: molecular modeling and inhibition studies. J. Med. Chem. 36, 1136-1145. (8) Wolff, T., Distlerath, L. M., Worthington, M. T., Groopman, J. D., Hammons, G. J., Kadlubar, F. F., Prough, R. A., Martin, M. V., and Guengerich, F. P. (1985) Substrate specificity of human liver cytochrome P-450 debrisoquine 4-hydroxylase probed using immunochemical inhibition and chemical modeling. Cancer Res. 45, 2116-21122. (9) Ellis, S. W., Hayhurst, G. P., Smith, G., Lightfoot, T., Wong, M. M., Simula, A. P., Ackland, M. J., Sternberg, M. J., Lennard, M. S., Tucker, G. T., and et al. (1995) Evidence that aspartic acid 301 is a critical substrate-contact residue in the active site of cytochrome P450 2D6. J. Biol. Chem. 270, 29055-29058.

Structure-Activity Inhibition Studies of P450 2D6 (10) Mackman, R., Tschirret-Guth, R. A., Smith, G., Hayhurst, G. P., Ellis, S. W., Lennard, M. S., Tucker, G. T., Wolf, C. R., and Ortiz de Montellano, P. R. (1996) Active-site topologies of human CYP2D6 and its aspartate-301 f glutamate, asparagine, and glycine mutants. Arch. Biochem. Biophys. 331, 134-140. (11) de Groot, M. J., Ackland, M. J., Horne, V. A., Alex, A. A., and Jones, B. C. (1999) Novel approach to predicting P450-mediated drug metabolism: development of a combined protein and pharmacophore model for CYP2D6. J. Med. Chem. 42, 1515-1524. (12) Ekins, S., Bravi, G., Binkley, S., Gillespie, J. S., Ring, B. J., Wikel, J. H., and Wrighton, S. A. (1999) Three and four dimensionalquantitative structure activity relationship (3D/4D-QSAR) analyses of CYP2D6 inhibitors. Pharmacogenetics 9, 477-489. (13) Otton, S. V., Crewe, H. K., Lennard, M. S., Tucker, G. T., and Woods, H. F. (1988) Use of quinidine inhibition to define the role of the sparteine/debrisoquine cytochrome P450 in metoprolol oxidation by human liver microsomes. J. Pharmacol. Exp. Ther. 247, 242-247. (14) Bichara, N., Ching, M. S., Blake, C. L., Ghabrial, H., and Smallwood, R. A. (1996) Propranolol hydroxylation and Ndesisopropylation by cytochrome P4502D6: studies using the yeast-expressed enzyme and NADPH/O2 and cumene hydroperoxide-supported reactions. Drug Metab. Dispos. 24, 112-118. (15) Ellis, S. W., Rowland, K., Ackland, M. J., Rekka, E., Simula, A. P., Lennard, M. S., Wolf, C. R., and Tucker, G. T. (1996) Influence of amino acid residue 374 of cytochrome P-450 2D6 (CYP2D6) on the regio- and enantioselective metabolism of metoprolol. Biochem. J. 316, 647-654. (16) Ellis, S. W., Hayhurst, G. P., Lightfoot, T., Smith, G., Harlow, J., Rowland-Yeo, K., Larsson, C., Mahling, J., Lim, C. K., Wolf, C. R., Blackburn, M. G., Lennard, M. S., and Tucker, G. T. (2000) Evidence that serine 304 is not a key ligand-binding residue in the active site of cytochrome P450 2D6. Biochem. J. 345, 565571. (17) Bentley, T. W., Llewellyn, G., and McAlister, J. A. (1996) S(N)2 mechanism for alcoholysis, aminolysis, and hydrolysis of acetyl chloride. J. Org. Chem. 61, 7927-7932. (18) Persson, J., Berg, U., and Matsson, O. (1995) Steric effects in SN2 reactions. Primary carbon kinetic isotope effects in Menshutkin reactions. J. Org. Chem. 60, 5037-5040. (19) Islam, S. A., Wolf, C. R., Lennard, M. S., and Sternberg, J. E. (1991) A three-dimensional molecular template for substrates of human cytochrome P450 involved in debrisoquine 4-hydroxylation. Carcinogenesis 12, 2211-2219. (20) Schenkman, J. B., Remmer, H., and Estabrook, R. W. (1967) Spectral studies of drug interaction with hepatic microsomal cytochrome. Mol. Pharmacol. 3, 113-123. (21) Otton, S. V., Brinn, R. U., and Gram, L. F. (1988) In vitro evidence against the oxidation of quinidine by the sparteine/debrisoquine monooxygenase of human liver. Drug Metab. Dispos. 16, 15-17. (22) Guengerich, F. P., Muller-Enoch, D., and Blair, I. A. (1986) Oxidation of quinidine by human liver cytochrome P-450. Mol. Pharmacol. 30, 287-295.

Chem. Res. Toxicol., Vol. 16, No. 4, 2003 459 (23) Gibbs, M. A., Kunze, K. L., Howald, W. N., and Thummel, K. E. (1999) Effect of inhibitor depletion on inhibitory potency: tight binding inhibition of CYP3A by clotrimazole. Drug Metab. Dispos. 27, 596-599. (24) Miller, G. P., Hanna, I. H., Nishimura, Y., and Guengerich, F. P. (2001) Oxidation of phenethylamine derivatives by cytochrome P450 2D6: the issue of substrate protonation in binding and catalysis. Biochemistry 40, 14215-14223. (25) Guengerich, F. P., Miller, G. P., Hanna, I. H., Martin, M. V., Leger, S., Black, C., Chauret, N., Silva, J. M., Trimble, L. A., Yergey, J. A., and Nicoll-Griffith, D. A. (2002) Diversity in the oxidation of substrates by cytochrome P450 2D6: Lack of an obligatory role of aspartate 301-substrate electrostatic bonding. Biochemistry 41, 11025-11034. (26) Venhorst, J., Onderwater, R. C., Meerman, J. H., Commandeur, J. N., and Vermeulen, N. P. (2000) Influence of N-substitution of 7-methoxy-4-(aminomethyl)-coumarin on cytochrome P450 metabolism and selectivity. Drug Metab. Dispos. 28, 1524-1532. (27) Paine, M. J., McLaughlin, L. A., Flanagan, J. U., Kemp, C. A., Sutcliffe, M. J., Roberts, G. C., and Wolf, C. R. (2002) Residues glutamate-216 and aspartate-301 are key determinants of substrate specificity and product regioselectivity in cytochrome P450 2D6. J. Biol. Chem. In press. (28) Lewis, D. F., Eddershaw, P. J., Goldfarb, P. S., and Tarbit, M. H. (1997) Molecular modelling of cytochrome P4502D6 (CYP2D6) based on an alignment with CYP102: Structural studies on specific CYP2D6 substrate metabolism. Xenobiotica 27, 319-339. (29) Lewis, D. F. V., and Dickins, M. (2002) Substrate SARs in human P450s. Drug Discovery Today 7, 918-925. (30) Smith, M. B., and March, J. (2001) Effects of Structure on Reactivity. March’s Advanced Organic Chemistry: Reactions, Mechanisms, and Structure, pp 363-380, John Wiley & Sons, New York. (31) Van der Aar, E. M., Bouwman, T., Commandeur, J. N. M., and Vermeulen, N. P. E. (1996) Structure-activity relationships for chemical and glutathione S-transferase-catalysed glutathione conjugation reactions of a series of 2-substituted 1-chloro-4nitrobenzenes. Biochem. J. 320, 531-540. (32) Ferrari, S., Leeman, T., and Dayer, P. (1991) The role of lipophilicity in the inhibition of polymorphic cytochrome P4502D6 oxidation by beta-blocking agents in vitro. Life Sci. 48, 22592265. (33) Ching, M. S., Blake, C. L., Ghabrial, H., Ellis, S. W., Lennard, M. S., Tucker, G. T., and Smallwood, R. A. (1995) Potent inhibition of yeast-expressed CYP2D6 by dihydroquinidine, quinidine, and its metabolites. Biochem. Pharmacol. 50, 833-837.

TX025674X

Inhibition of Cytochrome P450 2D6: Structureâ Activity Studies Using a

Recommend Documents

Inhibition of Cytochrome P450 2D6: Structureâ Activity Studies Using a