Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Introduced a half-century ago as a rodenticide, warfarin has become a mainstay for prevention of thromboembolic complications in patients with atrial fibrillation (Majerus et al., 1990; Fihn, 1995). The drug, however, has a narrow therapeutic index and therefore is prone to serious drug-drug interactions when coadministered with other agents that alter warfarin clearance (Wells et al., 1994; Harder and Thurmann, 1996). Numerous such interactions have been reported, some of which were attributed to induction or inhibition of hepatic cytochrome P450 (CYP¹) enzyme(s) (Wells et al., 1994; Harder and Thurmann, 1996). In this regard, the metabolism of warfarin in humans has been shown to be catalyzed by CYP1A1/2, 2C9, and 3A4, resulting in the formation of 4'-, 6-, 7-, 8- and 10-hydroxy derivatives (Fig. 1; Rettie et al., 1992; Kaminsky and Zhang, 1997). Coadministration of CYP inhibitors such as fluconazole or miconazole was associated with elevated plasma concentrations of warfarin and prolonged prothrombin times (O'Reilly et al., 1992; Black et al., 1996). Conversely, diminished anticoagulant effects of warfarin were observed when the drug was administered together with rifampin or barbiturates and were attributed to induction of CYP enzyme(s) and a subsequent increase in warfarin metabolism (Koch-Weser and Sellers, 1971; Heimark et al., 1987). This type of drug-drug interaction, however, would become evident only after chronic dosing of enzyme inducers (Conney, 1967; O'Reilly, 1975).

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Cytochrome P450 3A4-catalyzed metabolism of warfarin and quinidine.

Modulation of warfarin therapy also was reported with quinidine; both agents may be used together for the management of atrial fibrillation (Harder and Thurmann, 1996). The outcome of this drug-drug interaction was either hypoprothrombinemic hemorrhage or a need for an increase in the anticoagulant dosage (Koch-Weser, 1968; Sylven and Anderson, 1983). While the decrease in prothrombin levels during coadministration of warfarin and quinidine had been speculated to be due to a synergistic depression of vitamin K-dependent clotting factors by the two drugs, no explanation was given for the reduced anticoagulant effect of warfarin in the presence of quinidine (Koch-Weser, 1968; Sylven and Anderson, 1983).

Quinidine is a naturally occurring cinchona alkaloid, which has been used clinically for cardioversion and is one of the most frequently prescribed antiarrhythmic drugs (Grace and Camm, 1998). The metabolism of quinidine in humans is mediated mainly by hepatic CYP3A4, resulting in 3-hydroxy and N-oxide derivatives (Fig. 1; Guengerich et al., 1986). Quinidine is known to cause drug interactions at the level of CYP3A4, although recent studies have shown that these interactions are more complex than would be expected on the basis of a simple relationship between a substrate and enzyme. For example, the drug has been suggested to inhibit CYP3A4 in a noncompetitive manner (Schellens et al., 1991; Bowles et al., 1993). Quinidine also has been demonstrated to be an effector of the enzyme, since the CYP3A4-catalyzed metabolism of diclofenac, meloxicam, and phenanthrene can be stimulated by the presence of quinidine (Ludwig et al., 1999; Ngui et al., 2000; Sai et al., 2000). In this report, we describe that the CYP3A4-mediated 4'- and 10-hydroxylation of warfarin are enhanced by quinidine in incubations with human liver microsomes as well as with human hepatocytes. Investigation of this in vitro drug-drug interaction was carried out in the context of the influence of quinidine on the K_m and V_max values of warfarin metabolism and vice versa.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. 7,8-Benzoflavone, cinchonine, NADPH, quinidine, quinine, and racemic warfarin were purchased from Sigma Chemical Co. (St. Louis, MO). S- and R-Warfarin, 4'-, 6-, 7-, 8-, and 10-hydroxywarfarin, [phenyl-²H₅]7-hydroxywarfarin, and 3-hydroxyquinidine were purchased from GENTEST Co. (Woburn, MA). 9-Epiquinidine and 9-epiquinine were from Buchler GmbH (Braunschweig, Germany). BondElut C18 extraction cartridge columns were obtained from Varian Chromatography Systems (Walnut Creek, CA), and Oasis MCX extraction plates were from Waters Co. (Milford, MA). All other chemicals were obtained from Fisher Scientific (Fair Lawn, NJ).

Instrumentation. LC/MS/MS was carried out on a Perkin-Elmer Sciex API 3000 tandem mass spectrometer (Toronto, Canada) interfaced to a high performance liquid chromatography system consisting of a Perkin-Elmer Series 200 quaternary pump and a Series 200 autosampler (Norwalk, CT). LC/MS/MS experiments were performed using either a Heated Nebulizer interface or a Turbo IonSpray interface with positive ion detection. With the Heated Nebulizer interface, the source temperature was set at 500°C, corona discharge at 3.0 µA, orifice potential at 39 V, and collision energy at 40 eV. The collision gas was nitrogen. With the Turbo IonSpray interface, the source temperature was set at 150°C, ionization voltage at 5 kV, orifice potential at 50 V, and collision energy at 35 eV. The collision gas again was nitrogen.

Incubations with Human Liver Microsomes or Recombinant CYP3A4. Human liver samples from three male and two female donors were obtained from the Pennsylvania Regional Tissue Bank (Exton, PA). An agreement was made between the tissue bank and Merck & Co. for research use of the samples. Liver microsomes were isolated from individual livers by differential centrifugation (Raucy and Lasker, 1991). The activity of CYP3A4 in these microsomal preparations was estimated based on the 6-hydroxylation of testosterone (Table 1). Aliquots from each preparation then were pooled on the basis of equivalent protein concentrations to yield a representative pool of human liver microsomes.

Incubations with Human Hepatocytes. Liver samples from three human donors were obtained from the Pennsylvania Regional Tissue Bank. The death of one donor, a 26-year-old female, was caused by drug overdose; the second donor, a 47-year-old female, died from subarachnoid hemorrhage; the third donor was a 56-year-old male who died from head trauma. Hepatocytes were isolated based on a two-step perfusion procedure (Pang et al., 1997) and exhibited a viability of greater than 80% as determined by the trypan blue exclusion test. The cells were suspended in Krebs-bicarbonate buffer. Warfarin in dimethyl sulfoxide was added to the suspension to provide a final concentration of 25 µM, while quinidine in aqueous solution was added to afford final concentrations ranging from 5 to 100 µM. The final concentration of dimethyl sulfoxide was 0.1% (v/v). Controls lacked quinidine. After incubation at 37°C for 2 h, reactions were quenched with 10% trifluoroacetic acid. These experiments were performed in duplicate.

Quantification of Metabolites. Aliquots (0.1 ml) from incubations with human liver microsomes were mixed with [phenyl-²H₅]7-hydroxywarfarin and cinchonine (50 ng, internal standards) and 4 M urea (1 ml) and applied to an Oasis MCX extraction plate, which was prewashed with methanol and water. The plate then was washed with water (1 ml) and eluted with 70% aqueous acetonitrile (0.3 ml) containing 0.1% trifluoroacetic acid and 1.0 mM ammonium acetate. The eluates were analyzed by LC/MS/MS.

Kinetic Calculations. Apparent K_m and V_max values were calculated according to the Michaelis-Menten equation.

(1)

(2)

(3)

View larger version (17K): [in this window] [in a new window]   Fig. 2.   Proposed kinetic model with two substrate binding domains in the active site of CYP3A4 for interactions of a single enantiomer of warfarin, quinidine, and the enzyme.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Warfarin Metabolism in Incubations with Human Liver Microsomes. Hydroxylated derivatives of warfarin were identified by LC/MS/MS based on coincidence of high performance liquid chromatography retention times and product ion spectra with those of authentic standards. In incubations with human liver microsomes, the formation of warfarin metabolites was linear over a period of 30 min, and all studies were performed following 20 min incubations.

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Formation of 4'-hydroxywarfarin (A) and 10-hydroxywarfarin (B) in incubations of R- or S-warfarin with human liver microsomes at various quinidine concentrations. The initial concentration of R- or S-warfarin was 50 µM.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Structures of the CYP3A4 modulators examined for their effect on warfarin metabolism.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Formation of 10-hydroxywarfarin in incubations of racemic warfarin with liver microsomes from five individual donors at various quinidine concentrations. The initial concentration of warfarin was 100 µM.

Warfarin Metabolism in Incubations with Recombinant CYP3A4. Incubations of warfarin with recombinant CYP3A4 coexpressed with NADPH-CYP oxidoreductase resulted in the formation 4'- and 10-hydroxywarfarin. Once again, these biotransformation pathways were stimulated by the presence of quinidine, such that the formation of 4'- and 10-hydroxywarfarin increased approximately 4-fold (Table 3).

Quinidine Metabolism in Incubations with Human Liver Microsomes. The formation of 3-hydroxyquinidine and quinidine N-oxide was linear over a period of 20-min incubation with either the individual or pooled human liver microsomes. The K_m and V_max values were estimated at 24 µM and 150 pmol/min/mg of protein for the 3-hydroxylation and 89 µM and 76 pmol/min/mg of protein for the N-oxidation in incubations with the pooled liver microsomes. Therefore, the N-oxidation reaction represented a minor pathway for quinidine metabolism as compared with 3-hydroxylation. This assessment is consistent with previously published data (Guengerich et al., 1986).

Kinetic Model for the Interaction of Warfarin and Quinidine with CYP3A4. The proposed kinetic scheme for the interaction of R-warfarin and quinidine with CYP3A4 was based on assumptions of a rapid equilibrium between the enzyme and substrates (Fig. 2). Experimental data were fit to the equations derived from the model, and resulting velocities (z-axis) were plotted against both substrate (x-axis) and effector (y-axis) concentrations (Fig. 6). The dissociation constants were estimated at ~250 µM for the equilibrium between warfarin and CYP3A4 and 18 to 25 µM for the equilibrium between quinidine and the enzyme (Table 5). These values decreased slightly upon binding of a second substrate or an effector (ES₁  S₂ES₁  S₂E; Figs. 2 and 6). The formation of 10-hydroxywarfarin increased with increasing effector (quinidine) concentrations (Fig. 6A). The maximal velocity for the formation of 10-hydroxywarfarin from the warfarin-CYP3A4-quinidine complex was 6-fold higher than that from the warfarin-CYP3A4 complex (Table 7). The velocity for the formation of 3-hydroxyquinidine from the quinidine-CYP3A4-warfarin complex was similar to that from the quinidine-CYP3A4 complex; in other words, the metabolism of quinidine was negligibly influenced by the presence of R-warfarin (Fig. 6B and Table 7).

View larger version (53K): [in this window] [in a new window]   Fig. 6.   A, CYP3A4-mediated 10-hydroxylation of R-warfarin in the presence of quinidine, and B, CYP3A4-mediated 3-hydroxylation of quinidine in the presence of warfarin. The surfaces represent the theoretical fits to eqs. 1 and 2 (see Experimental Procedures), respectively.

Stimulation of Warfarin Metabolism in Human Hepatocyte Suspensions. Five monohydroxylated derivatives of warfarin, namely 4'-, 6-, 7-, 8-, and 10-hydroxywarfarin, were identified by LC/MS/MS in incubations of racemic warfarin with human hepatocytes. The formation of 4'- and 10-hydroxywarfarin increased when incubations were performed in the presence of quinidine, although intersubject difference in the increases were evident (Table 1). The 6-, 7-, and 8-hydroxylation pathways of warfarin metabolism in human hepatocyte suspensions were not affected by the presence of quinidine (data not shown).

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Among CYP enzymes, CYP3A4 is the most abundant in human liver and is involved in the metabolism of numerous therapeutic agents (Wrighton and Stevens, 1992). While activity of the enzyme can be modulated by either inhibition or induction, CYP3A4 may exhibit stimulation (positive cooperativity) in the presence of certain xenobiotic compounds (Guengerich, 1997; Szklarz and Halpert, 1998). For instance, the catalytic activity of CYP3A4 was reported to be enhanced by quinidine in the metabolism of diclofenac, meloxicam, and phenanthrene (Ludwig et al., 1999; Ngui et al., 2000; Sai et al., 2000). The work presented here provides another example of stimulation of CYP3A4, wherein conversion of warfarin to its monohydroxylated derivatives increased significantly in incubations with human liver microsomes in the presence of quinidine.

Therapeutically, warfarin is used as the racemic mixture of R- and S-enantiomers (Harder and Thurmann, 1996). Both enantiomers are subject to 4'- and 10-hydroxylation, although 10-hydroxywarfarin derives preferentially from R-warfarin, while 4'-hydroxywarfarin results mainly from metabolism of S-warfarin. The 10-hydroxylation of R-warfarin is predominant over the corresponding 4'-hydroxylation pathway. These differences are reflected in the respective intrinsic clearance values for the 4'- and 10-hydroxylation of warfarin enantiomers. It is of interest to note that the metabolism of R-warfarin was inhibited by S-warfarin and vice versa, although the nature of inhibition is unknown at this time. A similar phenomenon has been reported for the 7-hydroxylation of S-warfarin, which is inhibited by R-warfarin (Kunze et al., 1991; Yamazaki and Shimada, 1997).

The 10-hydroxylation of S- and R-warfarin is known to be catalyzed by CYP3A4 (Rettie et al., 1992). However, it was believed that the 4'-hydroxylation of warfarin involves CYP2C, based on studies with recombinant enzymes (Kaminsky and Zhang, 1997), while a separate study indicates that the 4'-hydroxylation of S-warfarin is catalyzed by both recombinant CYP3A4 and 2C9, with the rate of metabolite formation being 6-fold higher in incubations with CYP3A4 (Rettie et al., 1992). The 4'-hydroxylation of R-warfarin was shown to be catalyzed by CYP3A4 (Rettie et al., 1992). In the present study, the formation of 4'- and 10-hydroxywarfarin, from either R- or S-warfarin, in incubations with human liver microsomes was inhibited 90% or more by a monoclonal antibody against CYP3A4. These data are consistent with earlier findings and appear to suggest that CYP3A4 plays a dominant role in the 4'- and 10-hydroxylation of warfarin.

The 4'- and 10-hydroxylation of R- or S-warfarin in incubations with human liver microsomes were enhanced by the presence of quinidine. Thus, under the conditions used, the formation of 4'-hydroxywarfarin from S-warfarin increased approximately 3-fold, while the formation of 10-hydroxywarfarin from R-warfarin increased 5-fold. The enhancement of warfarin metabolism most likely is due to interactions of quinidine with CYP3A4 because the stimulatory effect of quinidine was diminished when human liver microsomes were pretreated with an inhibitory directed antibody against CYP3A4. This interpretation also is supported by the fact that the formation of 4'- and 10-hydroxywarfarin in incubations with recombinant CYP3A4 increased in the presence of quinidine.

Stimulation of the CYP3A4-catalyzed metabolism of warfarin also was observed with diastereoisomers of quinidine, namely quinine and the threo epimers, indicating that the stereochemistry of the effectors is not critical in this type of interaction. On the other hand, limited data generated in this study with five different preparations of human liver microsomes indicate that there may be variability among individuals with respect to cooperativity of CYP3A4. For example, stimulation of the 4'-hydroxylation of S-warfarin was not observed with one liver microsomal preparation. The magnitude of increases in the 10-hydroxylation of R-warfarin appeared to be similar within the five liver microsomal preparations, despite large variations in the CYP3A4 activities in these livers.

The quinidine-mediated stimulation of warfarin 4'- and 10-hydroxylation in incubations with pooled human liver microsomes was characterized by increases in V_max values with minimal changes in K_m values. Conversely, the V_max for the 3-hydroxylation of quinidine was not influenced by the presence of warfarin. These phenomena are similar to those observed in studies of interactions between diclofenac and quinidine, wherein the formation of 5-hydroxydiclofenac in incubations with recombinant CYP3A4 was stimulated by quinidine, but quinidine metabolism was not affected by diclofenac (Ngui et al., 2000). On the basis of these results, it was proposed that the CYP3A4 active site contained two distinct binding domains, one for diclofenac and a second for quinidine (Ngui et al., 2000). However, interactions involving warfarin, quinidine, and CYP3A4 appear to be more complex, taking into account the results with the two enantiomers of warfarin. Thus, the 10-hydroxylation of R-warfarin and 4'-hydroxylation of S-warfarin exhibited different K_m values and were subject to inhibition by the respective antipodes. The EC₅₀ of quinidine was different for stimulation of the 4'- versus the 10-hydroxylation of warfarin. These findings may imply the existence of two or even three binding domains in the active site of CYP3A4, one for each enantiomer of warfarin and the third for quinidine. It also is conceivable that there may be an allosteric binding site for quinidine that may, or may not, be identical to that for quinidine metabolism.

Scenarios involving multiple ligands binding in the active site of CYP3A4, or substrate bindings plus a distinct allosteric binding site, have been hypothesized to describe the cooperative properties of CYP3A4. These proposals are based on studies involving site-directed mutagenesis (Harlow and Halpert, 1998; Domanski et al., 2000), enzyme kinetics (Shou et al., 1994; Ueng et al., 1997; Korzekwa et al., 1998), computational docking modeling (Szklarz and Halpert, 1997; Ekins et al., 1999), and ligand binding determinations (Hosea et al., 2000). In most of these studies, 7,8-benzoflavone was described as an effector of CYP3A4. However, in the present studies, this flavonoid was inhibitory toward the metabolism of warfarin. The distinction between 7,8-benzoflavone and quinidine in terms of CYP stimulation may reside in differences in effector-mediated changes of enzyme conformations. In this regard, based on measurements of the rate of reformation of the ferroheme-carbon monoxide complex, it was proposed that CYP3A4 may exist in two or more subpopulations or conformations (Koley et al., 1995, 1997). Hence, the stimulation of warfarin metabolism by quinidine could be explained in the context that the subpopulation (conformation) of CYP3A4 responsible for hydroxylation of warfarin increased upon binding of quinidine to the enzyme. This perturbation of the equilibrium between different CYP3A4 conformations in favor of warfarin metabolism would be reflected by an enhancement of substrate (warfarin) turnover without impact on binding affinity, consistent with the observed increases in V_max but minimal changes in the K_m values. Conversely, the lack of effect of warfarin on the 3-hydroxylation of quinidine could be interpreted as indicating that binding of warfarin to the enzyme does not perturb the equilibrium among CYP conformations.

The interactions of warfarin, quinidine, and CYP3A4 also were considered in the context of a kinetic model containing two binding domains in the active site of CYP3A4. In this model, only the R-antipode of warfarin was considered, and relationships among binding sites were simplified by assuming that these sites were independent from each other. The rate of formation of 10-hydroxywarfarin was therefore described to increase 6-fold upon binding of quinidine to the enzyme, whereas the 3-hydroxylation of quinidine was affected minimally by the presence of warfarin. A kinetic model appears to have the advantage of demonstrating the cooperative properties of CYP3A4 with mathematic equations.

The stimulation of the 4'- and 10-hydroxylation of warfarin by quinidine in incubations with human liver microsomes is unlikely to be an in vitro artifact because a similar enhancement of warfarin metabolism was observed in human hepatocyte suspensions. In this regard, it may be expected that this type of drug-drug interaction will occur in vivo, the effect of which would be to alter the pharmacokinetics of a therapeutic agent. Data generated in animal species supporting this hypothesis include increases in zoxazolamine metabolism in rats treated concurrently with flavone and elevation of the hepatic clearance of diclofenac in rhesus monkeys following coadministration of quinidine (Lasker et al., 1982; Tang et al., 1999). In a controlled human trial, the clearance of warfarin was found to be increased slightly, but to a statistically significant extent, following coadministration of 3-hydroxy-10,11-dihydroquinidine (Trenk et al., 1993). In the present study, 3-hydroxyquinidine, a close analog of 3-hydroxy-10,11-hydroquinidine, was observed to stimulate the metabolism of warfarin.

While warfarin is used therapeutically as a racemic mixture, S-warfarin has been shown to be 2- to 5-fold more potent as an anticoagulant than its R counterpart (Harder and Thurmann, 1996). The clearance of warfarin in humans is due mainly to hepatic metabolism involving multiple CYP enzymes, although the duration of the anticoagulant effect is determined primarily by the rate of 7-hydroxylation of S-warfarin catalyzed by CYP2C9 (Rettie et al., 1992). Evidence also is available to indicate that the metabolism of S-warfarin is inhibited by the presence of R-warfarin and vice versa. Warfarin and quinidine often are used together for the treatment of atrial fibrillation, and drug-drug interactions between these two agents have been reported, the outcome of which may require an increase in the warfarin dose (Sylven and Anderson, 1983). While it is tempting to speculate in light of current findings that the diminished effect of warfarin under these circumstances is due to stimulation of its metabolism by quinidine, it may be argued that the 4'-hydroxylation of warfarin is a minor pathway for the clearance of S-isomer, and increases in the 10-hydroxylation of R-warfarin represent at best a secondary effect in altering the disposition of S-warfarin by removing the inhibitory R-enantiomer.

In summary, the present investigation has demonstrated that the 4'- and 10-hydroxylation of warfarin in incubations with human liver microsomes or hepatocytes are enhanced by the presence of quinidine. While this drug-drug interaction is mediated by CYP3A4, its clinical significance remains unclear in light of the complexities associated with intersubject variability, the racemic nature of warfarin, and pharmacological side effects elicited by quinidine. Nevertheless, the findings of this in vitro study reinforce emerging views on the existence of multiple binding domains in CYP3A4 that underlie the complex kinetics and drug-drug interaction characteristics of this important enzyme.