The mammalian cytochromes P450 (P450s) constitute a family of enzymes distributed primarily in liver, kidney, and intestinal tissues that often are responsible for the metabolic clearance of therapeutic agents in humans (Wrighton and Stevens, 1992; Guengerich, 1997a). As a result, either inhibition or induction of a P450 in patients receiving polytherapy has the potential to precipitate clinically important drug interactions. Whereas P450 inhibition may lead to drug-related adverse effects that result from substantial increases in systemic concentrations of one or other of the drugs, P450 induction by one agent can decrease circulating levels of a second due to enhanced metabolic clearance and thereby compromise therapeutic effects of the latter (Lin and Lu, 1998). A relatively newer field of P450 research pertinent to drug interactions is that of heterotropic enhancement of P450 activity, in which the metabolism of a substrate is stimulated by an effector (Tang and Stearns, 2001; Hutzler and Tracy, 2002). An example of enzyme behavior of this type is found with CYP3A4 in human liver microsomes whose activity, measured by the rate of 10-hydroxylation of R-warfarin, increased severalfold in the presence of quinidine (Ngui et al., 2001). On the basis of such observations, it seems likely that enhancement of P450 activity could result in pharmacological or toxicological consequences similar to those elicited by enzyme induction. In this regard, increases in the hepatic clearance of diclofenac were evident in monkeys after coadministration of quinidine, and stimulation of aflatoxin B1 metabolism by 7,8-benzoflavone was associated with elevated mutagenic responses from Salmonella typhimurium TA98/TA100 (Buening et al., 1981; Tang et al., 1999).

In spite of these reports, there have been few attempts to extrapolate liver microsomal data to the intact liver for the purpose of projecting changes in hepatic clearance resulting from heterotropic enhancement of P450 activity. A recent effort was made to correlate the stimulation by felbamate of carbamazepine metabolism in vitro with clinical observations that the plasma concentrations of carbamazepine decreased in epileptic patients after addition of felbamate to the initial monotherapy (Egnell et al., 2003). The analysis was complex, however, due to a number of factors, including a less than 2-fold increase in magnitude of P450 activity in vitro in the presence of felbamate and a no more than 45% decline of the plasma concentrations of carbamazepine after initiation of polytherapy. Nevertheless, an approach of this type is valuable in the pharmaceutical industry, wherein the selection of drug candidates for development is dependent to some extent on metabolism data generated in vitro with human vectors. An equally important aspect in the discovery process is to establish animal models that are appropriate and readily available for evaluation of potential drug interactions that may take place in humans. In this regard, we demonstrate in the present communication that the activities of CYP3A enzymes in liver microsomes from various species are affected differently by quinidine with respect to R-warfarin metabolism, and these in vitro differences may be replicated quantitatively in intact livers with an in situ perfusion model. The structures of R-warfarin and quinidine are shown in Fig. 1.

Materials and Methods

Materials. Bovine serum albumin, 3-[(3-cholamidopropyl)dimethylammonio]propanesulfonate (CHAPS), cinchonine, 1,2-didodecanoyl-sn-glycero-3-phosphocholine, dithiothreitol, glucose, β-glucuronidase, ketoconazole, NADPH, quinidine, and troleandomycin were purchased from Sigma-Aldrich (St. Louis, MO). 3-Hydoxyquinidine, R-warfarin, 4′ -6-, 7-, 8-, and 10-hydroxywarfarin, and [phenyl-²H₅]7-hydroxywarfarin were purchased from BD Biosciences (Bedford, MA), [¹⁴C]warfarin was from Amersham Biosciences Inc. (Piscataway, NJ), and Oasis MCX extraction plates were from Waters (Milford, MA). All other chemicals were obtained from Fisher Scientific Co. (Fair Lawn, NJ).

Purified cytochrome b₅ and NADPH-CYP oxidoreductase were purchased from PanVera Corp. (Madison, WI). Microsomes expressing rat CYP3A1 and 3A2, and polyclonal inhibitory antibody against CYP3A2 were purchased from BD Biosciences. Polyclonal inhibitory antibodies against rat CYP2B, 2C, and 3A were generous gifts from Dr. Stelvio Bandiera (University of British Columbia, Vancouver, BC, Canada). Microsomes expressing rabbit CYP3A6 were prepared at Merck Research Labs (West Point, PA) via baculovirus expression (T. H. Rushmore and C. Booth-Genthe, unpublished data). Monoclonal inhibitory antibody against monkey CYP3A was generated in mice via immunization with recombinant CYP3A4 (Mei et al., 1999).

Instrumentation and Analytical Methods. Liquid chromatography-tandem mass spectrometry (LC/MS/MS) was carried out on an API 3000 tandem mass spectrometer (PerkinElmerSciex Instruments, Toronto, ON, Canada) interfaced to a high-performance liquid chromatography system consisting of a Series 200 quaternary pump and a Series 200 autosampler (PerkinElmer Life and Analytical Sciences, Boston, MA). Detection of R-warfarin and its hydroxylated derivatives was performed using a Turbo IonSpray interface operated at the negative ion mode. The source temperature was set at 300°C, ionization voltage at -4.2 kV, focusing potential at -260 V, entrance potential at 10 V, and collision energy at -30 eV. The collision gas was nitrogen. Chromatography was performed on a Zorbax RX-C8 column (250 × 4.6 mm, 5 μm), and samples were delivered at a flow rate of 1.5 ml/min with 1:25 split to the mass spectrometer. The mobile phase consisted of 35% aqueous acetonitrile containing 10% methanol and 0.05% formic acid (v/v).

Detection of quinidine and 3-hydroxyquinidine was performed using a Turbo IonSpray interface operated at the positive ion mode. The source temperature was set at 300°C, ion spray voltage at 5.0 kV, focusing potential at 350 V, entrance potential at -10 V, and collision energy at 30 eV. The collision gas was nitrogen. Chromatography was performed on a Phenomenex Synergi polar-rp column (150 × 4.6 mm, 4 μm), and samples were delivered at a flow rate of 1 ml/min with a 1:5 split to the mass spectrometer. The mobile phase consisted of 40% aqueous acetonitrile containing 1 mM ammonium acetate and 0.1% trifluoroacetic acid (v/v).

Incubations with Liver Microsomes or Recombinant Enzymes. Liver microsomes were isolated from male Sprague-Dawley rats (n = 40), female New Zealand White rabbits (n = 20), and male rhesus monkeys (n = 3) by differential centrifugation (Raucy and Lasker, 1991). The microsomes were suspended in 0.1 M phosphate buffer (pH 7.4) containing 1 mM EDTA, and the final protein concentration was 1 mg/ml. R-Warfarin in methanol and quinidine in water then were added to final concentrations ranging from 0 to 750 μM and from 0 to 100 μM, respectively, whereas the concentration of methanol was 0.2% (v/v). After preincubation at 37°C for 5 min, NADPH in phosphate buffer was added to a final concentration of 1 mg/ml, and R-warfarin metabolism was initiated. Incubations were carried out for an additional 20 min, and then the reactions were quenched with 10% aqueous formic acid.

Solubilization of liver microsomes was carried out via treatment with CHAPS. Briefly, rat or monkey liver microsomes (∼40 mg/ml) were suspended in 0.1 M phosphate buffer (pH 7.4) containing 20% glycerol (v/v) and 0.1 mM EDTA, and aqueous CHAPS was added dropwise to the final concentration of 1% (w/v). The suspension was stirred at 4°C for 2 h and subsequently centrifuged at 100,000g for 30 min. The supernatant was dialyzed three times for 16, 4, and 4 h, respectively, in SnakeSkin pleated dialysis tubing at 4°C against 1000 volumes of 20 mM phosphate buffer (pH 7.4) containing 7.5% glycerol (v/v), 0.1 mM EDTA, and 0.05 mM dithiothreitol. The resulting solubilized fraction was stored at -70°C, and P450 concentrations were determined based on the reduced carbon monoxide-difference spectra (Omura and Sato, 1964). Incubations were performed by using 250 pmol/ml solubilized P450, 100 pmol/ml cytochrome b₅, 200 pmol/ml oxidoreductase, 20 μg/ml 1,2-didodecanoyl-sn-glycero-3-phosphocholine, and 0.04% CHAPS (w/v) (Zhang et al., 2004).

Incubations with recombinant P450 enzymes were performed in a manner similar to that with intact liver microsomes. The concentration of P450 was 250 pmol/ml.

In experiments involving chemical inhibitors, rabbit liver microsomes were preincubated with ketoconazole for 10 min at room temperature or preincubated with troleandomycin in the presence of NADPH for 15 min at 37°C. The inhibitors, dissolved in methanol, were added to microsomal suspensions to reach their respective final concentrations. The concentration of methanol in these suspensions was 0.2% (v/v), and the same amount of solvent was added to control incubations that lacked the inhibitors. R-Warfarin and NADPH were added sequentially and the incubations were performed at 37°C for an additional 20 min.

In immunoinhibition experiments, rat or monkey liver microsomes were preincubated with inhibitory antibodies for 15 min at room temperature. Ascites from nonimmunized animals were added to control incubations. R-Warfarin and NADPH were added thereafter, and the incubations were performed at 37°C for an additional 20 min.

In Situ Liver Perfusion. All experiments were performed according to procedures approved by the Merck Research Labs Institutional Animal Care and Use Committee.

Female New Zealand White rabbits (n = 5/group) were anesthetized with a mixture of ketamine, xylazine, and acepromazine (20, 10, and 5 mg/kg, respectively), and their pyloric vein and celiac artery were ligated after laparotomy (Chow et al., 1997). The portal vein then was cannulated, with the catheter advanced along the vein to a point close to the entrance to the liver. The cannula was connected to the outflow from a Minipulse-3 perfusion apparatus (Gilson Medical Electronics, Middleton, WI), and perfusion was initiated at a flow rate of 50 ml/min. The perfusate, exited through the abdominal inferior vena cava opened by an incision, consisted of Krebs-Henseleit bicarbonate buffer (pH 7.4) containing 3 mg/ml glucose, 1% bovine albumin (w/v), 20 μM R-warfarin (0.004 μCi/ml [¹⁴C]warfarin), and 50 μM quinidine. Control animals were treated without quinidine. A cannula then was placed through the atrium into the vena cava, with the tip of the catheter advanced to the point of entry of the hepatic vein, followed by ligation of the inferior vena cava. The outflow from the hepatic vein was collected in intervals of 20 min for 60 min.

In situ liver perfusion was similarly performed in male Sprague-Dawley rats (n = 5/group), except that the perfusate was delivered at a flow rate of 20 ml/min.

Sample Preparation and Quantitative Analysis. Samples of perfusate were incubated with β-glucuronidase (500 units/ml) for 15 h at 37°C. Aliquots (0.1 ml) of the resulting samples were mixed with [phenyl-²H₅]7-hydroxywarfarin and cinchonine (50 ng each, internal standards) and 4 M urea (1 ml) and applied to an Oasis MCX extraction plate, which was prewashed with methanol and water. The extraction plate then was washed with water (1 ml) and eluted with 70% aqueous acetonitrile (0.3 ml) containing 0.05% formic acid (v/v). Samples from liver microsomal incubations were extracted in a similar manner. The acetonitrile eluate was analyzed by LC/MS/MS (vide supra) for quantification of R-warfarin, quinidine, and their metabolites through multiple reaction monitoring of transitions m/z 307.1 → 161.0 [R-warfarin], 323.0 → 250.0 (10-hydroxywarfarin), 328.2 → 177.0 ([phenyl-²H₅]7-hydroxywarfarin), 325.2 → 307.2 (quinidine), 341.2 → 160.2 (3-hydroxyquinidine), and 295.0 → 166.0 (cinchonine). Standard curves were generated over concentration ranges of 0.016 to 81 μM for R-warfarin and quinidine and of 0.00015 to 0.774 μM for the metabolites.

Determination of Reversible Protein Binding. Reversible binding of R-warfarin to proteins in perfusate and in liver microsomal incubations was determined by equilibrium dialysis. Briefly, an aliquot (0.5 ml) of perfusate or microsomal suspensions was treated with R-warfarin to final concentrations ranging from 5 to 750 μM (0.004 μCi/ml [¹⁴C]warfarin) and with quinidine to concentrations of 10 to 100 μM. The samples then were subjected to dialysis against 0.5 ml of isotonic phosphate buffer for 6 h in chambers (Bel-Art, Pequannock, NJ) separated by a membrane that had a molecular weight cutoff at 12.4 kDa (Sigma-Aldrich). The fraction of unbound drug was determined by comparing the differences between radioactivity in the perfusate (or microsomal suspensions) and that in the buffer solution.

Reversible binding of quinidine to proteins was evaluated in a similar manner, except that quantification of the drug was based on LC/MS/MS analysis (vide supra).

Calculation of Intrinsic Hepatic Clearance. An enzyme model consisting of two binding sites was adopted to describe the kinetics of R-warfarin metabolism in rabbit liver microsomal incubations (Fig. 2; Ngui et al., 2001). In the presence of quinidine, the rate of formation of 10-hydroxywarafrin was defined by eqs. 1 and 2. where ν_W is the initial velocity; V_max^W is the maximum velocity; k_W and k_Q are the rate constants for metabolism of R-warfarin and quinidine, respectively; [S_W] and [S_Q] are unbound R-warfarin and quinidine concentrations, respectively; K_W and K_Q are the dissociation constants for enzyme-warfarin and enzyme-quinidine complexes, respectively; and β is the factor by which V_max^W value increases (or decreases) when quinidine is in contact with the enzyme (Fig. 2).

In the absence of quinidine ([S_Q] = 0), the rate of 10-hydroxylation of R-warfarin was calculated according to eq. 3:

These equations were solved by using a SCIENTIST software package from MicroMath Inc. (Salt Lake City, UT).

In cases where R-warfarin metabolism took place at substrate concentrations well below the value of the dissociation constant ([S_W] << K_W), the intrinsic hepatic clearance of R-warfarin by 10-hydroxylation (CL_int^W) was estimated based on eqs. 4 and 5. In the presence of quinidine, In the absence of quinidine,

After in situ liver perfusion, CL_int^W was calculated according to eq. 6. where υ_W is the rate of formation of 10-hydroxywarfarin, and C_ssW is the concentration of unbound R-warfarin at the steady state. Estimation of the C_ssW was based on a logarithmic average of unbound drug concentrations before and after the perfused organ, as shown in eq. 7. where C_inW and C_outW represent unbound R-warfarin concentrations in the input and output perfusates, respectively (Xu et al., 1993). The unbound quinidine concentration (C_ssQ) at steady state was similarly determined after liver perfusion.

Statistical comparison was based on one-tailed t test at a significance level of P < 0.05.

Results

R-Warfarin Metabolism in Liver Microsomal Incubations.R-Warfarin metabolism in incubations with liver microsomal preparations from rats, rabbits, and monkeys resulted in the formation of 4′-, 6-, 7-, 8-, and 10-hydroxy derivatives, based on LC/MS/MS analysis with the aid of synthetic standards. All incubations were performed for 20 min because preliminary studies demonstrated that the rates of R-warfarin metabolism remained constant for 30 min after initiation of the reactions.

The 10-hydroxylation of R-warfarin in rat liver microsomal incubations was inhibited by polyclonal antibodies against rat CYP3A but was not affected by those against rat CYP2B or 2C (Table 1). Similarly, the formation of 10-hydoxywarfarin in monkey liver microsomal incubations was blocked by a monoclonal inhibitory antibody against monkey CYP3A (Table 1). Although inhibitory antibodies against rabbit CYP3A were not available to us at the time these experiments were performed, the 10-hydroxylation of R-warfarin in rabbit liver microsomal incubations was nearly completely inhibited by ketoconazole, and to a lesser extent, by troleandomycin (Table 1). Both ketoconazole and troleandomycin are selective inhibitors of human CYP3A and also were shown to inhibit rabbit CYP3A6 when carbamazepine and N-desethylamiodarone were evaluated as substrates (Mesdjian et al., 1999; Kozlik et al., 2001). These data collectively suggest that CYP3A enzymes are responsible for the 10-hydroxylation of R-warfarin in rats, rabbits, and monkeys. The same reaction is known to be catalyzed by CYP3A4 in humans (Kaminsky and Zhang, 1997).

Effect of Quinidine on R-Warfarin Metabolism in Liver Microsomal Incubations.R-Warfarin metabolism was enhanced by quinidine in incubations with liver microsomes from female rabbits, and the degree of enhancement was dependent on effector concentrations, with an approximately 25-fold increase in the formation of 10-hydroxywarfarin at 100 μM quinidine (Fig. 3). Experiments with liver microsomes from male rabbits led to a similar conclusion (data not shown). The rate of 10-hydroxylation of R-warfarin also increased in incubations with recombinant rabbit CYP3A6 in the presence of quinidine, although the magnitude of augmentation was one-eighth that with rabbit liver microsomes (Fig. 3). This finding may not be surprising in light of a recent report that enhancement by quinidine of CYP3A4 activity was evident with human liver microsomes but not with a recombinant enzyme system, the difference being attributed to different compositions of the host membranes surrounding the P450 protein (Zhang et al., 2004). Nevertheless, CYP3A6 responded positively to stimulation by quinidine, suggesting that elevated activities of this enzyme are responsible for the observed enhancement of R-warfarin metabolism in rabbit liver microsomal incubations.

R-Warfarin metabolism also was enhanced by quinidine in human liver microsomal incubations; a 4-fold increase in the formation of 10-hydroxywarfarin was noted at 100 μM effector (Fig. 3). These data are similar to the result reported previously (Ngui et al., 2001). In contrast, the 10-hydroxylation of R-warfarin was little affected (<2-fold increase) or even inhibited (∼50%) by quinidine when incubations were performed with monkey and rat liver microsomes, respectively (Fig. 3). Although recombinant monkey CYP3A enzymes were not available for evaluation, R-warfarin metabolism was inhibited by quinidine in incubations with recombinant rat CYP3A1 or 3A2 (Fig. 3).

Rat and monkey liver microsomes were subsequently solubilized by CHAPS and further evaluated for interactions between R-warfarin and quinidine. Enhancement by quinidine of R-warfarin metabolism was evident with solubilized P450 from monkey liver microsomes; the formation of 10-hydroxywarfarin increased >5-fold in incubations containing 100 μM quinidine (Fig. 4). A slight increase in the 10-hydroxylation of R-warfarin (∼1.7-fold) was associated with solubilized P450 from rat liver microsomes in the presence of quinidine (Fig. 4).

With respect to other pathways of R-warfarin metabolism, the rates of formation of 4′-, 6-, 7-, and 8-hydroxywarfarin were little affected by 100 μM quinidine in rabbit and human liver microsomal incubations, but they were inhibited approximately 20 and 50% in monkey and rat liver microsomal incubations, respectively (data not shown).

Estimation of Intrinsic Hepatic Clearance with Liver Microsomes. Using equilibrium dialysis, reversible binding of R-warfarin to rabbit and rat liver microsomal proteins was determined to be 4%, whereas that of quinidine was 20%, over the concentration ranges 5 to 750 μM and 5 to 100 μM, respectively. These values are similar to those reported previously for binding of the two drugs to liver microsomal proteins from rats, dogs, monkeys, and humans (Obach, 1997, 1999). In the current study, the fraction of unbound R-warfarin was not affected by the presence of quinidine, and vice versa, in incubations with rat or rabbit liver microsomes.

The proposed kinetic scheme for interactions of R-warfarin and quinidine with rabbit CYP3A was based on the assumption of a rapid equilibrium between the enzyme, substrate, and effector (Fig. 2). Thus, the rate of formation of 10-hydroxywarfarin in the absence of quinidine is proportional to the product of the maximal velocity and unbound substrate concentration but inversely related to the sum of the dissociation constant and unbound substrate concentration. This expression is similar to that derived from the classic Michaelis-Menten model. In the presence of quinidine, the rate of R-warfarin metabolism becomes a function of variables involving both the substrate (R-warfarin) and the effector (quinidine), including dissociation constants and unbound drug concentrations. The velocity equations then were solved via numerical regression of the rate of formation of 10-hydroxywarfarin at various concentrations of unbound R-warfarin and quinidine. The dissociation constant therefore was estimated to be 125 to 155 μM for the equilibrium between R-warfarin and the P450, regardless the presence or absence of quinidine. The maximal velocity, defined as a product of the “basal” rate constant and the “total” enzyme concentration, also seemed to be independent of the presence of quinidine, with values ranging from 40 to 60 pmol/min/mg protein (Table 2). These results may represent “self-consistency” in the kinetic model because they are derived from the best fit of two separate equations with different sets of data (i.e., [S_Q] = 0 and [S_Q] ≠ 0). The rate of formation of 10-hydroxywarfarin, however, was severalfold faster from the warfarin-P450-quinidine complex than it was from the warfarin-P450 complex, due primarily to an increase in the rate constant (Table 2).

Defined as the rate of metabolism at a given substrate concentration, the intrinsic hepatic clearance of R-warfarin by 10-hydroxylation (CL_int^W) is expressed as a ratio of the maximal velocity to the dissociation constant in cases where unbound substrate concentrations are markedly lower than the dissociation constant, and the value was estimated to be 6 μl/min/g liver on the basis of 25 mg microsomal protein/g rabbit liver (Tanaka et al., 1999). The expression of CL_int^W in the presence of quinidine is complex, in that not only does it depend on parameters of the substrate but also on those involving the effector. At 21 μM unbound quinidine (vide infra), the CL_int^W increased approximately 20-fold compared with that in the absence of the effector (Table 2).

Determination of Hepatic Clearance via Liver Perfusion. In situ perfusion of rat or rabbit livers was performed via infusion of R-warfarin or R-warfarin plus quinidine through the portal vein for 60 min. Steady-state drug concentrations were achieved at 30 min in the outlet from the hepatic vein. Reversible binding of R-warfarin and quinidine to proteins in the perfusates (inlet and outlet) was approximately 94 and 50%, respectively, whereas the fraction of unbound R-warfarin was not affected by coadministration of quinidine. Steady-state unbound drug concentrations therefore were estimated based on a logarithmic average of unbound drug concentrations in portal and hepatic veins (Table 3).

Hepatic intrinsic clearance essentially reflects the rate of metabolism “normalized” to a given concentration of the unbound drug at steady state during liver perfusion. For R-warfarin metabolism in perfused rat or rabbit livers, 4′-, 6-, 7-, 8-, and 10-hydroxy derivatives and their respective glucuronic acid conjugates were detected. Further quantification of these hydroxy derivatives in the outlet from the hepatic vein was carried out after treatment of the samples with β-glucuronidase, assuming that oxidative metabolism occurs before glucuronidation. By this approach, the mean value of CL_int^W in rabbits was determined to be 7 μl/min/g liver, whereas it increased 22-fold after coadministration of quinidine (Table 3). Although the results exhibited large individual variation, the enhancement by quinidine of R-warfarin metabolism in rabbits was statistically significant. In contrast, the CL_int^W in rats decreased in the presence of quinidine (Table 3). The other pathways of R-warfarin metabolism remained unchanged in rabbits but were inhibited 60 to 80% in rats after cotreatment with quinidine, estimated based on the “total” concentrations of 4′-, 6-, 7-, and 8-hydroxywarfarin (hydroxy derivatives plus their respective glucuronides) in the perfusates collected from the hepatic vein (data not shown).

Discussion

In addition to being a drug of choice for prophylactic treatment of thromboembolic complications in patients with atrial fibrillation, warfarin has often been used as a pharmacological tool to probe various forms of P450 and their active sites. This is due to the fact that warfarin metabolism catalyzed by different P450 enzymes leads to regioselective hydroxylation of the drug (Fasco et al., 1978; Kaminsky et al., 1980). For example, CYP3A1 and 3A4 have been suggested to be the primary P450s involved in the 10-hydroxylation of R-warfarin in rats and humans, respectively (Namkung et al., 1988; Kaminsky and Zhang, 1997). Consistent with this assessment, it was found in the current study with rat liver microsomes that the formation of 10-hydroxywarfarin was inhibited by antibodies against CYP3A but not by those specific for CYP2B and 2C. A similar conclusion was derived from immuno- and chemical inhibition experiments that showed that CYP3A is responsible for the 10-hydroxylation of R-warfarin in rabbit and monkey liver microsomal preparations. Whereas multiple forms of CYP3A exist in rats and monkeys, CYP3A6 currently is the only member in the family identified in rabbits (Guengerich, 1997b).

Stimulation by quinidine of the 10-hydroxylation of R-warfarin was demonstrated initially with human liver microsomes and was attributed to increases in P450 activity instead of an in vitro artifact on the basis that the drug interaction was replicated with recombinant CYP3A4 and also with human hepatocyte systems (Ngui et al., 2001). This drug interaction was exploited in the current study to probe whether heterotropic enhancement of P450 activity is conserved across various species and would affect hepatic clearance of a drug in vivo. Subsequent experiments were performed with liver microsomes from rats, rabbits, and monkeys, because CYP3A enzymes were demonstrated to be responsible for the 10-hydroxylation of R-warfarin in these species. With rabbit liver microsomes, increases in the formation of 10-hydroxywarfarin were readily detected in the presence of quinidine. Enhancement by quinidine of R-warfarin metabolism also was apparent with recombinant CYP3A6, suggesting a similar behavior of this enzyme in comparison with its counterpart in humans with respect to their responses to heterotropic stimulation.

Whereas the mechanisms underlying heterotropic stimulation of P450 activity remain to be clarified, multiple binding sites have been proposed to exist on P450 proteins (Korzekwa et al., 1998; Hosea et al., 2000). One hypothesis involves the presence of two independent binding sites proximal to the ferroheme-oxygen complex, binding at either of which would permit a substrate access to the reactive oxygen species without interference from a second substrate and vice versa. An alternative scenario implicates one substrate binding site plus a distal allosteric binding site. According to these models, increases in P450 activity could result either from changes of P450 protein conformation or perturbation of the equilibrium between various reactive oxygen species in favor of a particular metabolic pathway upon effector binding (Koley et al., 1997; Vaz et al., 1998; Tang and Stearns, 2001). These concepts have been incorporated into a kinetic scheme in which a P450 enzyme consists of two independent binding sites, one for its substrate and the other for an effector that may also be a substrate (Korzekwa et al., 1998; Ngui et al., 2001; Shou et al., 2001). Formation of the products may result from three “active” forms of P450, namely, the substrate-enzyme, effector-enzyme, and substrate-enzyme-effector complexes, although the nature of these species remains to be defined (Fig. 2). This model was adopted in the current study for evaluation of interactions between R-warfarin and quinidine, with emphases placed on R-warfarin metabolism in vitro with rabbit liver microsomes. Stimulation by quinidine of the 10-hydroxylation of R-warfarin was therefore ascribed primarily to increases in the rate of metabolism with little contribution from changes in binding affinity of the substrate to the enzyme. Whereas the model may not represent the “true mechanism” by which P450 activity increases, it has provided a reasonably “good fit” to the existing experimental data and thus affords a mathematical basis for projecting the rate of formation of 10-hydroxywarfarin at various effector concentrations (vide infra).

In spite of being catalyzed by rat and monkey CYP3A, the 10-hydroxylation of R-warfarin was not enhanced by quinidine in vitro with liver microsomes from these two species. A similar example of the phenomenon was reported previously, wherein a stimulatory effect of 7,8-benzoflavone on benzo-[a]pyrene metabolism was apparent with rabbit and human liver microsomes but not with the organelles from guinea pigs and rats (Huang et al., 1981). This type of selective enhancement of P450 activity by an effector is likely due to variations in the structures of enzyme proteins from different species, such that minor changes in amino acid sequences would alter complex interactions between a P450 and the effector. Alternatively, it may be hypothesized that certain membrane-bound P450 enzymes are restricted by their hosts so that they cannot respond to stimuli. In this regard, stimulation by quinidine of R-warfarin metabolism was evident with solubilized P450 from monkey liver microsomes treated with a detergent. A similar treatment of rat liver microsomes led to a slightly increase in the rate of 10-hydroxylation of R-warfarin in the presence of quinidine. These data provide a tentative explanation for the lack of response by CYP3A in intact rat and monkey liver microsomes to stimulation by quinidine.

The kinetic properties of a P450 are often described by the Michaelis-Menten model, with which a hyperbolic relationship is generated for metabolite formation versus substrate concentrations. It has been suggested, however, that the model may not be appropriate for those enzymes that are subject to homotropic or heterotropic stimulation; a “forced” fit of experimental data to a hyperbolic curve would result in miscalculation of kinetic parameters and consequently of intrinsic clearance (Houston and Kenworthy, 2000; Tracy, 2003). In the current study, an attempt was made with rabbit liver microsomes to quantify the rate of 10-hydroxylation of R-warfarin by a kinetic scheme that incorporated influences from quinidine (vide supra). In the absence of quinidine, the mathematic form of the scheme would collapse to a simpler version that resembled the Michaelis-Menten expression. Accordingly, the kinetic parameters were generated with free fractions of the substrate and effector, based on the principle that only unbound drugs are available for interactions with enzymes. The resulting maximal velocity and dissociation constants then were used in combination with quinidine concentrations for subsequent estimation of the hepatic intrinsic clearance of R-warfarin via the 10-hydroxylation pathway, namely, CL_int^W. In addition to mechanistic considerations, this approach is preferred from a technical stand point, in that CL_int^W is readily assessed at any given effector concentration. For example, the values were approximately 6 and 128 μl/min/g liver, respectively, in the absence and presence of 21 μM unbound quinidine. These estimates essentially “predict” the results from studies involving perfused rabbit livers, with which CL_int^W was determined to increase from 7 to 156 μl/min/g liver after increasing the hepatic exposure of unbound quinidine from 0 to 21 μM. The significance of this exercise resides in 1) its demonstration of heterotropic enhancement of P450 activity taking place in intact livers, and 2) the predictability with liver microsomes of changes in hepatic clearance resulting from stimulation of P450-mediated metabolism. Although the origin of the striking species differences remain to be defined, the utility of liver microsomal data also was reflected by replication with perfused rat livers of the inhibitory effect of quinidine on R-warfarin metabolism, wherein CL_int^W decreased to 26% of the control value at 22 μM unbound effector.

In summary, stimulation by quinidine of the 10-hydroxylation of R-warfarin is conserved in rabbits but not in rats, in which R-warfarin metabolism is inhibited by quinidine. This represents a rare case wherein drug interactions in two species occur in opposite directions, in spite of the fact that the metabolism in both species is catalyzed by enzymes in the CYP3A family. A good correlation seems to exist between liver microsomes and intact livers with respect to evaluation of the impact of drug interactions on the rate of metabolism, regardless of stimulation or inhibition. In particular, the hepatic clearance of R-warfarin by 10-hydroxylation was projected quantitatively in rabbits from liver microsomal data based on a kinetic scheme incorporating two substrate binding sites. These findings highlight the variability of drug interactions in different species, hence the importance of selecting appropriate animal models for evaluation of potential drug interactions in humans, and provide further evidence that data generated with liver microsomes may be of predictive value in the drug discovery process.