Abstract
A rapid and sensitive radiometric assay for assessing the potential of drugs to inhibit cytochrome P450 (P450) 3A4/5 in human liver microsomes is described. In contrast to the conventional testosterone 6β-hydroxylation assay, the new method does not require high-performance liquid chromatography (HPLC) separation and mass spectrometry. The assay is based on the release of tritium as tritiated water that occurs upon CYP3A4/5-mediated 6β-hydroxylation of testosterone labeled with tritium in the 6β position. The radiolabeled product is separated from the substrate using 96-well solid-phase extraction plates. Using commercially available [1,2,6,7-3H]testosterone as substrate, we demonstrated that the reaction is NADPH-dependent, and sensitive to CYP3A4/5/5 inhibitors and a CYP3A4/5/5-specific inhibitory monoclonal antibody, but not to inhibitors of or antibodies against other P450 enzymes. The method was further improved by synthesis of testosterone specifically tritiated in the 6β position, which displayed greatly improved conversion rate with an ensuing increase in assay sensitivity. Competition experiments using tritiated and unlabeled testosterone indicated that CYP3A4/5-mediated 6β-hydroxylation exhibits positive cooperativity and a modest kinetic isotope effect. IC50 values for more than 40 structurally diverse chemical inhibitors were not significantly different from those determined in the testosterone 6β-hydroxylation assay, using HPLC-tandem mass spectrometry analysis. All the steps of the new assay, namely, incubation, product separation, and radioactivity counting, are performed in 96-well format and can be automated. This assay thus represents a high-throughput version of the classical testosterone 6β-hydroxylation assay, which is the most widely used method to assess the potential for CYP3A4/5 inhibition of new chemical entities.
The pharmacokinetic and toxicokinetic properties of pharmaceuticals depend in great part on their biotransformation by drug-metabolizing enzymes. The main drug-metabolizing system in mammals is cytochrome P450 (P450), a family of microsomal enzymes present predominantly in the liver. Multiple P450 enzymes catalyze the oxidation of chemicals of endogenous and exogenous origin, including drugs, steroids, prostanoids, eicosanoids, fatty acids, and environmental toxins (Ioannides, 1996). When a drug that is metabolized by a particular P450 enzyme is coadministered with an inhibitor of that same enzyme, changes in its pharmacokinetics can occur, which can give rise to adverse effects (Bertz and Granneman, 1997; Lin and Lu, 1998; Thummel and Wilkinson, 1998). It is therefore important to predict and prevent the occurrence of clearance changes due to metabolic inhibition. During the drug discovery process, it is routine practice in the pharmaceutical industry to assess the P450 inhibition potential of drug candidates to exclude potent inhibitors from further development (Lin and Lu, 1998; Crespi and Stresser, 2000; Riley, 2001).
CYP3A4/5 is the most abundant P450 in human liver and is involved in the metabolism of about 50% of drugs used in human therapy (Guengerich, 1999). Inhibition of CYP3A4/5 activity can give rise to clinically significant and potentially life-threatening drug interactions (Thummel and Wilkinson, 1998). Several assay methods are currently used for determining the potential of drug candidates to inhibit CYP3A4/5 activity, and each of these methods presents distinct advantages and disadvantages. The most widely used method is the testosterone 6β-hydroxylation assay, which is specific for enzymes of the CYP3A family (CYP3A4/5) (Waxman et al., 1988; Maenpaa et al., 1993; Wang et al., 1997; Yamazaki and Shimada, 1997). According to recent surveys conducted by reviewers in the Center for Drug Evaluation and Research of the United States Food and Drug Administration, the testosterone 6β-hydroxylation assay represents the most commonly used probe reaction in support of new drug applications (Yuan et al., 1999, 2002). The practical challenge posed by this assay is that it requires HPLC separation of the reaction product from the substrate, followed by UV or mass spectrometric detection. This renders the assay relatively laborious, time-consuming, and not ideally suited for screening the large number of compounds typically required in an industrial drug discovery setting.
Several alternative CYP3A4/5 assays, suitable for high-throughput screening, have been introduced in the past several years. These assays are based on the use of fluorogenic (Crespi and Stresser, 2000) or radiolabeled (Zhang and Thomas, 1996; Moody et al., 1999) substrates, eliminating the need for HPLC separation. One of the most widely used fluorogenic substrates is 7-benzyloxy-4-trifluoromethylcoumarin (Crespi and Stresser, 2000). Because this probe is not a specific substrate for CYP3A4/5, the assay cannot be run with HLMs, but requires the use of recombinant enzyme. Alternative fluorometric assays, which use CYP3A-specific substrates and can therefore be performed with HLMs, have been described (Chauret et al., 1999; Stresser et al., 2002). Even though fluorometric assays are rapid, easy to perform, and amenable to automation, they suffer from a number of limitations. First, the use of recombinant P450 instead of HLMs, which contain the full complement of P450 enzymes, may give rise to differences in inhibitory potency because test compounds may be subject to metabolism by more than one enzyme, leading to different rates of substrate depletion or formation of inhibitory metabolites. Even when this concern is eliminated by the use of a CYP3A-specific substrate and HLMs (Chauret et al., 1999; Stresser et al., 2002), a second issue is that CYP3A4/5 inhibition is substrate-dependent (Kenworthy et al., 1999; Stresser et al., 2000; Wang et al., 2000). CYP3A4/5 is a large and complex enzyme that is thought to bind substrates and inhibitors in multiple modes and binding sites (Kenworthy et al., 2001; Shou et al., 2001; Ekins et al., 2003). As a result, inhibitory interactions observed with a nonclassical CYP3A4/5 probe may not be representative of those observed with other substrates (Cohen et al., 2003). Finally, fluorescence interference is frequently observed with certain classes of compounds.
For these reasons, a non-HPLC assay based on the use of a classical CYP3A4/5 substrate such as testosterone would be highly desirable. Hydroxylation of testosterone at the 6β position is accompanied by release of the corresponding hydrogen as water. When this position is labeled with tritium, CYP3A4/5-mediated hydroxylation generates radiolabeled water, which can easily be separated from the unreacted substrate. A CYP3A assay procedure based on this principle, using commercially available [1,2,6,7-3H]testosterone, has been described (Draper et al., 1998). To determine whether this method can be used for P450 inhibition screening in HLMs, we modified and optimized the assay procedure, adapted it to a high-throughput 96-well format, and studied reaction kinetics in the presence and absence of P450 inhibitors. The method was further optimized by synthesis of testosterone labeled with tritium in only the 6β position. In the present report, we describe that the new assay accurately measures the potency of CYP3A4/5 inhibitors and represents a high-throughput version of the classical testosterone 6β-hydroxylation assay.
Materials and Methods
Materials. Oasis HLB extraction plates and vacuum manifold were purchased from Waters (Milford, MA). Pooled human liver microsomes were obtained from BD Gentest (Woburn, MA). [1,2,6,7-3H]Testosterone and tritiated water were purchased from Amersham Biosciences UK, Ltd. (Little Chalfont, Buckinghamshire, UK), and [3H]lithium aluminum hydride was obtained from American Radiolabeled Chemicals (Saint Louis, MO). Testosterone-3-ethyleneacetal was purchased from Steraloids (Newport, RI). Other chemicals were purchased from Sigma-Aldrich (Milano, Italy).
Synthesis of [6β-3H]Testosterone. The synthetic scheme is described in Fig. 1.
5β,6β-Epoxy-17β-hydroxyandrostan-3-one, 3-Ethyleneacetal (2), and 5α,6α-Epoxy-17β-hydroxyandrostan-3-one, 3-Ethyleneacetal (3) (Campbell et al., 1958;Liston and Toft, 1969). To a stirring solution of 0.85 g of testosterone-3-ethyleneacetal (1) in 14 ml of chloroform was added 0.1 g of sodium acetate and 1 ml of peracetic acid (36%) with cooling in an ice-salt bath. After 2 h, the solution was washed with 1 N sodium hydroxide solution (5 ml) and water (two times, 5 ml). The organic layer was dried over anhydrous magnesium sulfate, filtered, and concentrated to dryness. The 5β,6β-epoxide (2) and the 5α,6α-epoxide (3), with overall 80% yield, were separated by silica gel column chromatography (60 g of silica gel; hexane/acetone as eluents). The fractions containing the required 5α,6α-epoxide (3) were collected and evaporated to dryness to yield 330 mg of product. LC/MS: (MH)+ 349 (100%).
[6β-3H]3,3-Ethylenedioxyandrostane-5α,17β-diol (4) (Toft et al., 1973). 5α,6α-Epoxy-17β-hydroxyandrostan-3-one, 3-ethyleneacetal (3, 3.48 mg) in dry tetrahydrofuran (2 ml) was heated under reflux with stirring under nitrogen atmosphere with [3H]lithium aluminum hydride (100 mCi) for 1 h. The reaction mixture was quenched with 1% aqueous sodium hydroxide and diluted with ether (10 ml), washed with water (two times, 5 ml), and dried over anhydrous sodium sulfate. The solvent was evaporated to dryness and used without further purification to the next reaction (total radioactivity, 20 mCi, 12.5%).
[6β-3H]Testosterone (6) (Toft et al., 1973). [6β-3H]3,3-Ethylenedioxyandrostane-5α,17β-diol (4, 20 mCi) was heated at 100°C for 15 min with a mixture of acetic acid (0.150 ml) and water (0.05 ml). The solution was cooled to room temperature and extracted with ether (two times, 10 ml). The organic layer was washed with saturated sodium bicarbonate (5 ml), water (two times, 5 ml), and dried over anhydrous sodium sulfate. The crude product (5, 10 mCi) was dissolved in methanol (5 ml) and water (2 ml), and treated with 0.1 N sodium hydroxide (0.1 ml). The solution was stirred at room temperature for 16 h and then treated with acetic acid (0.1 ml). The solution was evaporated to half-volume on a rotary evaporator. The crude product was purified by reversed-phase HPLC [Luna Phenyl Hexyl column (Phenomenex, Torrance, CA), water containing 0.1% trifluoroacetic acid/acetonitrile, 55:45, UV = 254 nm, flow rate 4 ml/min, retention time = 10.5 min]. The combined fractions were passed through Sep-Pak C-18, which was further washed with ethanol (10 ml) to yield 2.2 mCi (11%) of [6β-3H]testosterone (6). The specific activity of this tracer was 1.6 Ci/mmol as calculated by LC/MS. Tritium NMR confirmed the position of tritium to be at the 6β position of testosterone. Radiochemical purity was >98%. LC/MS 289 (MH)+, 290 (MH + 1)+, 291 (MH + 2)+. Proton NMR: (CDCl3, δ) 0.79 (3H, s 18-H3), 1.2 (3H, s, 18-H3), 5.73 (1H, s, 4-H). Tritium NMR: (CDCl3, δ) 2.1 (1 T, s, 6β-tritium).
Purification of [1,2,6,7-3H]Testosterone. To remove polar impurities, an aliquot (≅0.1 mCi) of [1,2,6,7-3H]testosterone was dried under vacuum, reconstituted in water, and loaded on a 30-mg Waters Oasis extraction cartridge (preconditioned according to the manufacturer's instructions). After washing with 5 ml of water, [1,2,6,7-3H]testosterone was eluted with 2 ml of methanol, dried again, and reconstituted in ethanol (same volume as the original aliquot). The purified probe was kept at –20°C.
Radiometric CYP3A4/5 Assay Using [1,2,6,7-3H]Testosterone. Reactions were carried out in 96-well conical microtiter plates (Corning Glassworks, Corning, NY) containing purified [1,2,6,7-3H]testosterone (0.2–0.5 μCi), unlabeled testosterone (60 μM, except otherwise noted), pooled HLMs (0.25 mg/ml, except where otherwise noted), and 0.1 M potassium phosphate buffer, pH 7.6, in a final volume of 200 μl. Inhibitors were added to the reaction mixture from stock solutions in DMSO, giving a final solvent concentration of 0.6% (v/v). Controls without inhibitors contained an equivalent amount of vehicle. After preincubation for 10 min at 37°C, reactions were started by addition of an NADPH-regenerating system containing 1 mM NADPH, 5 mM glucose 6-phosphate, 3 mM MgCl2, and 1 U/ml glucose-6-phosphate dehydrogenase. Assays were conducted for 10 min at 37°C and stopped by addition of HCl to a final concentration of 0.1 N. Plates were then centrifuged for 10 min in a microplate rotor, and supernatants were loaded on a preconditioned 30-mg Oasis HLB 96-well plate. Vacuum was applied and the flow-through was collected in the collection plate. Then, 200 μl of water was added, vacuum was applied again, and the wash was collected into the same plate. This step was repeated. Pooled flow-through and water washes were transferred into scintillation vials and counted in a β-scintillation counter. Alternatively, aliquots of this mixture were counted in 24-well scintillation plates using a TopCount scintillation counter (PerkinElmer Life and Analytical Sciences, Boston, MA). For the calculation of enzyme activity, product counts were corrected by subtraction of counts obtained in control incubations performed in the absence of NADPH. Oasis plates were regenerated by washing with 5 ml of methanol and 5 ml of water, and were reused for up to 5 times.
Radiometric CYP3A4/5 Assay Using [6β-3H]Testosterone. The assay procedure was similar to that of the [1,2,6,7-3H]testosterone hydroxylation assay, with the following modifications. The tracer was not further purified before use, total radioactivity was 100,000 to 200,000 dpm, reaction volume was 100 μl, and solvent concentration (coming from inhibitor stock solutions) was 0.3 to 0.7% DMSO, 0.5% acetonitrile (v/v). Product separation was carried out using 10-mg Oasis HLB 96-well plates. The flow-through was collected and combined with a 75-μl water wash. Radioactivity was determined using a Beckman Coulter (Fullerton, CA) β-scintillation counter or by counting 120-μl aliquots in 96-well scintillation plates using TopCount scintillation counter. IC50 values of commercial inhibitors and Merck proprietary compounds were calculated from dose-response curves consisting of 10 points per duplicate, after 1:3 dilutions starting at 100 μM, and run in at least two independent experiments.
Determination of the Kinetic Tritium Isotope Effect.TV/K, the kinetic isotope effect on the V/K ratio, was determined according to the formula (Northrop, 1982) where f is the fractional conversion of substrate to product, SA0 is the initial specific radioactivity of labeled substrate, and SAP is the specific radioactivity of product. At low values of f (<5%), such as those observed in the present experiments, this expression reduces to (Northrop, 1982) Calculation of the Apparent Rate of Formation of Unlabeled Product from Tracer Competition Experiments. If the tritium label is not retained in the hydroxylated reaction product (however, see Discussion), the tritiated water product from [6β-3H]testosterone is formed stoichiometrically with 6β-hydroxytestosterone. When assays are performed using a fixed amount of [6β-3H]testosterone and varying concentrations of unlabeled testosterone, the velocity of formation of unlabeled product, v, is given by where v* is the velocity of formation of tritiated water.
Substituting from eq. 2, we obtain: Defining v′ as the velocity of formation of unlabeled product divided by the kinetic isotope effect, i.e., we obtain v′, the apparent formation rate of unlabeled product, can thus be calculated. When plotted against the substrate concentration, S, and fitted to the Hill equation (eq. 7), V′max, S50, the substrate concentration at 50% of V′max, and n, the Hill coefficient, can be derived. where.
V′max = Vmax/(TV/K), i.e., the apparent maximal rate of product formation. Quantification of 6β-Hydroxytestosterone and 2β-Hydroxytestosterone. Aliquots of the assay reaction mixture and of metabolite standard curves were analyzed by HPLC-MS/MS using an Agilent HP1100 liquid chromatograph (Agilent Technologies, Palo Alto, CA) equipped with a CTC Analytics (Zwingen, Switzerland) PAL autosampler. Chromatography was performed on an XTERRA MS C18 column (4.6 mm × 5 cm; 5 μm; Waters) at a flow rate of 1 ml/min, using a mobile phase consisting of a mixture of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) (linear gradient for 4 min from 5 to 100% B; 2 min at 100% B). The eluate was diverted to waste for the first minute and then to a Sciex API-2000 (Applied Biosystems/MDS Sciex, Foster City, CA) triple quadrupole mass spectrometer with a Turbo IonSpray ionization source operated in the positive ion mode. Metabolites and internal standard were detected and identified using the transition m/z 305.0→269.1. Metabolite concentrations were determined by weighted linear least-squares regression analysis, using Analyst Quantitation Wizard software version 1.4 (Applied Biosystems, Foster City, CA).
Curve Fitting. Curve fitting of enzyme kinetics data to the Hill equation or to a four-parameter logistic inhibition model (Rodbard and Frazier, 1975) was performed by nonlinear regression analysis using Xlfit 4.0 (IDBS, Guildford, UK).
Results
Separation of Radiolabeled Testosterone and Tritiated Water Using 96-Well Solid-Phase Extraction Plates. When a solution of assay buffer containing purified [1,2,6,7-3H]testosterone (from 104 to 107 dpm) and stopping solution was applied to 96-well extraction plates containing 10 or 30 mg of Oasis resin, greater than 99.9% of the radioactivity was retained on the plate. The tracer could be recovered by elution with methanol (data not shown). Only 0.045 ± 0.006% (average ± S.E.M., n = 7) of radioactivity eluted in the combined flow-through and aqueous wash. For [6β-3H] testosterone, 0.13 ± 0.1% (average ± S.E.M., n = 3) of radioactivity eluted in the aqueous fraction. In contrast, tritiated water (from 102 to 105 dpm) was not retained on the extraction plates under the same conditions. In both 10-mg and 30-mg plates, recovery of tritiated water eluted in the combined flow-through and the 100-μl or 400-μl aqueous wash, respectively, was quantitative (94 ± 6%, average ± S.D., n = 6). Recoveries of radiolabeled testosterone and water were not affected by the presence of unlabeled testosterone at concentrations up to 600 μM (data not shown). For chronological reasons, experiments with [1,2,6,7-3H]testosterone and [6β-3H] testosterone were performed using 30-mg and 10-mg Oasis plates, respectively.
Formation of Tritiated Water from [1,2,6,7-3H]Testosterone and [6β-3H]Testosterone in HLMs. When [1,2,6,7-3H]testosterone was incubated with HLM in the presence of an NADPH-regenerating system, tritiated water was formed in a time-dependent manner (Fig. 2A). At HLM concentrations of 0.05, 0.1, and 0.25 mg/ml, the reaction was linear for up to 20 min, with conversion rates of 0.006%/min, 0.011%/min, and 0.024%/min, respectively (correlation coefficients, r2, of 0.990, 0.963, and 0.963, respectively). Product formation at 10 min increased linearly with the concentration of microsomes up to a protein concentration of 0.5 mg/ml. Formation of tritiated water was dependent on NADPH (see below). Similar results were obtained using [6β-3H]testosterone as substrate (Fig. 2B). The main differences between the two substrates were the signal to noise ratios and conversion rates. We define signal to noise ratio as the ratio between product counts obtained in the presence versus absence of NADPH. The specific conversion rate is the percentage of total radiolabeled substrate converted into tritiated water per unit of time and per milligram of microsomal protein. As summarized in Table 1, signal to noise ratios were 5 to 6 with [1,2,6,7-3H]testosterone and 19 to 25 with [6β-3H]testosterone. Specific conversion rates were 0.4 to 0.6%/min/mg for [1,2,6,7-3H]testosterone and 16- to 19-fold higher for [6β-3H]testosterone. Since the reaction volumes were not identical for the two tracers, it is more appropriate to compare their clearance, rather than conversion rates. At substrate concentrations of 10 and 60 μM, clearance for [1,2,6,7-3H]testosterone was 0.74 and 1.1 μl/min/mg, and that for [6β-3H]testosterone was 5.9 and 10.5 μl/min/mg, respectively.
Effect of P450 Inhibitors and Anti-P450 Antibodies. To confirm that product formation is mediated by CYP3A4/5, reactions were performed in the presence or absence of a series of selective chemical inhibitors (Bourrie et al., 1996; Eagling et al., 1998) or monoclonal antibodies (Mei et al., 1999; Shou et al., 2000). Chemical inhibitors chosen were furafylline (1A2 inhibitor), coumarin (2A6), sulfaphenazole (2C9), quinidine (2D6), diethyldithiocarbamate (2E1), and ketoconazole (3A4/5). Monoclonal antibodies used were inhibitors of CYP2A6, CYP2C9, CYP2C19, CYP2D6, and CYP3A4/5. As shown in Fig. 3, none of these agents significantly affected formation of tritiated water in HLMs, with the exception of ketoconazole and the anti-CYP3A4/5 monoclonal antibody, which inhibited the reaction by over 90%. Taken together, these results indicate that formation of tritiated water from [1,2,6,7-3H]testosterone in HLMs is mediated by CYP3A4/5.
Kinetic Isotope Effect. CYP3A4/5-mediated loss of tritium from [1,2,6,7-3H]testosterone may occur as a consequence of hydroxylation at either the 6β or 2β positions (Waxman et al., 1988; Yamazaki and Shimada, 1997). It was recently reported that 1β-hydroxytestosterone is another minor hydroxylation product formed by CYP3A4/5 (Krauser et al., 2004). The activity of testosterone 2β-hydroxylase in the batch of HLMs used in the present experiments was ≅11% of that of the 6β-hydroxylase (data not shown). The 1β-hydroxylase activity was not determined. The formation of both 6β- and 2β-hydroxylated metabolites was mediated by CYP3A4/5, since it was completely inhibited by 10 μM ketoconazole over the entire substrate concentration range (data not shown). The relative proportions of label in the 2β and 6β positions of [1,2,6,7-3H]testosterone are not known, and it cannot, therefore, be determined from which position the tritium loss occurred.
Following CYP3A4/5-mediated hydroxylation of [6β-3H]testosterone, tritium can be lost only from the 6β position. To determine whether tritium substitution gives rise to a kinetic isotope effect, the rate of formation of tritiated water was compared with the rate of formation of 6β-hydroxytestosterone in the same reaction mixture. Considering that the hydrogen lost upon hydroxylation of testosterone is formed stoichiometrically with 6β-hydroxytestosterone, the specific radioactivity of the tritiated water product can be calculated by dividing product counts by the amount of 6β-hydroxytestosterone formed (Table 2). Together with the conversion rate of tritiated tracer and its known initial specific radioactivity, this information can be used to calculate the kinetic isotope effect on V/K. As shown in Table 2, TV/K is 2.3, which is equal to the substrate/product specific radioactivity ratio.
Competition between Radiolabeled and Unlabeled Testosterone. The effect of unlabeled testosterone on the formation of tritiated water in HLM is depicted in Fig. 4A. The curve displays a “low dose hook;” i.e., product formation rate increased with increasing concentration of unlabeled substrate, reached a peak at a testosterone concentration of 30 to 40 μM, and then decreased. Since [6β-3H]testosterone is used as an isotopic tracer, the formation rate of tritiated water (v*) is representative of that of unlabeled product, namely, water derived from 6β-hydroxylation of testosterone (which is formed stoichiometrically with 6β-hydroxytestosterone). The apparent formation rate of unlabeled product, v′, is defined as v* divided by the specific radioactivity of the tracer. The dependence of v′ on substrate concentration (S) can be used to obtain information about the dependence on substrate concentration of the unlabeled product, even if the latter is not measured directly (see Materials and Methods). As depicted in Fig. 4B, the curve of v′ versus S could be fitted to the Hill equation, with S50 = 98 ± 18 μM, n = 1.4 ± 0.2, and V′max = 1.3 ± 0.1 nmol/min/mg (average ± S.E.M., n = 3). The Hill coefficient was greater than 1, suggesting positive cooperativity. Indeed, at low substrate concentrations, a sigmoidal relationship was observed between v′ and S, as shown in the inset of Fig. 4B.
The kinetics of 6β-hydroxytestosterone formation is depicted in Fig. 4C. The reaction had a S50 of 65 ± 17 μM, Vmax of 3.4 ± 0.4 nmol/min/mg protein, and Hill coefficient of 1.5 ± 0.2 (average ± S.E.M., n = 5). Note that the ratio between V′max and Vmax is 2.6, which is close to the kinetic isotope effect on V/K, as expected from the definition of V′max (see Materials and Methods). The formation of tritiated water and of 6β-hydroxytestosterone was completely inhibited by ketoconazole, confirming that both products are formed via CYP3A4/5-mediated metabolism.
Similarly, substrate competition between a fixed concentration of [1,2,6,7-3H]testosterone and increasing concentrations of unlabeled testosterone displayed a low-dose hook, and tritiated water formation from this substrate was completely inhibited by ketoconazole (Fig. 4D). Since neither the specific radioactivity of the tracer nor the proportion of product coming from the 2β and 6β positions are known, the formation rates of the corresponding unlabeled products cannot be determined from isotopic dilution studies.
Kinetics of Inhibition by CYP3A4/5 Inhibitors. The effect of known inhibitors of CYP3A4/5 on the rate of formation of tritiated water from [1,2,6,7-3H]testosterone and [6β-3H]testosterone is shown in Fig. 5. IC50 values are summarized in Table 3. With [1,2,6,7-3H]testosterone, inhibition experiments were carried out in the presence of two different concentrations of unlabeled testosterone, 10 and 60 μM. No significant differences (<3.5-fold) in IC50 values were observed at the two different substrate concentrations. IC50 values determined with the two different tracers were very similar, with differences of less than 2-fold. To confirm that IC50 values obtained with the radiometric assay reflect those of the conventional assay, the effect of the inhibitors on tritiated water and 6β-hydroxytestosterone formation was determined in the same reaction mixture. As shown in Table 3, almost identical IC50 values were obtained.
P450 inhibitors tested in these experiments had to be dissolved in organic solvents. Since inhibitory effects of solvents on CYP3A4/5 activity have been reported (Chauret et al., 1998; Hickman et al., 1998), we assessed the effect of two different concentrations of DMSO on the release of tritiated water from [6β-3H]testosterone and on IC50 values of selected inhibitors. DMSO inhibited product formation, with about 70% and 50% of control activity remaining at solvent concentrations of 0.3% and 0.7% (v/v), respectively. However, DMSO had no effect on the IC50 values of ethynylestradiol, ketoconazole, miconazole, bromocriptine, nicardipine, nifedipine, quinidine, and verapamil (data not shown). Inhibition assays can therefore be performed using DMSO concentrations between 0.3 and 0.7%, as long as care is taken to include solvent controls, and activity data are expressed relative to these controls.
Comparison with the Conventional Testosterone 6β-Hydroxylation Assay for a Large Number of Compounds. IC50 values for 39 structurally diverse NCEs from different Merck programs were determined in the radiometric assay using [6β-3H]testosterone and compared with data that had been previously generated using the conventional testosterone 6β-hydroxylation assay. The results of this comparison are depicted in Table 4. Three compounds had IC50 values >100 μM in both assays. Six compounds had IC50 >100 μM in one assay and IC50 >45 μM in the other assay. One compound had IC50 <0.4 μM in one assay and 0.2 μM in the other assay. The IC50 values of the remaining 29 compounds differed less than 4.2-fold between the two assays, and in 25 of these (86%), they differed less than 3-fold. For the compounds with measurable IC50 values in both assays, linear regression analysis of the IC50 in the radiometric assay versus IC50 in the conventional assay resulted in a line with a slope of 1.09 and a correlation coefficient, r2, of 0.562 (Fig. 6A). The point that deviated most from the trend line corresponded to a weak inhibitor (compound 28) with a less than 3-fold difference in IC50 values between the two assays. When this compound was excluded from the analysis, the r2 increased to 0.757 (Fig. 6B).
Discussion
The release of tritium that accompanies hydroxylation of a substrate has been used to measure the activity of cytochrome P450 enzymes (Daly, 1970; Hayakawa and Udenfriend, 1973; Thompson and Siiteri, 1974; Reed and Ohno, 1976; Tomaszewski et al., 1976; Osawa and Coon, 1987). Draper et al. (1998) described a CYP3A assay procedure based on this principle, in which tritiated water generated from CYP3A-mediated metabolism of [1,2,6,7-3H]testosterone was separated from the unreacted substrate by charcoal extraction. Several features of the procedure described by these authors render it less than ideal for high-throughput screening, and we therefore modified and adapted the method for this purpose. The present procedure, which is amenable to automation, allows the assays to be run in 96-well format throughout the incubation and extraction steps and collection of the reaction product for scintillation counting in multiwell plates. The following modifications were introduced. First, the assay volume was reduced from 500 μlto200 μl and the microsome concentration from 0.4 mg/ml to 0.25 mg/ml, with a final amount of microsomes of 50 μg instead of 200 μg. Second, incubations were performed in 96-well plates. Finally, tritiated water was separated from [1,2,6,7-3H]testosterone using 96-well solid-phase extraction plates containing Oasis reverse phase resin.
Previous studies with [1,2,6,7-3H]testosterone showed that the release of tritium-labeled water from [1,2,6,7-3H]testosterone is catalyzed by recombinant CYP3A4/5 at a 6-fold higher rate than by recombinant P450s 1A2, 2A6, 2B6, 2C9, 2C19, 2D6, and 2E1 (Draper et al., 1998). NADPH-dependent metabolism of unlabeled testosterone in HLM occurred mainly (>75%) by 6β-hydroxylation, and the rate of release of tritiated water from [1,2,6,7-3H]testosterone correlated with that of testosterone 6β-hydroxylation in HLM from 12 different donor livers. On the basis of these findings, Draper et al. (1998) proposed that the NADPH-dependent release of tritium from [1,2,6,7-3H]testosterone in HLM could be used as a probe reaction for CYP3A4/5 activity (Draper et al., 1998). Convincing validation of this claim would come from the demonstration that the reaction can be completely inhibited by specific CYP3A4/5 inhibitors, but not by inhibitors of other P450 enzymes. The only inhibitor used in the study by Draper et al. (1998) was the competitive substrate erythromycin, which displayed a relatively high IC50 of 130 μM. Other authors have reported that erythromycin and testosterone only weakly inhibit the metabolism of each other (Wang et al., 1997). Erythromycin is therefore not an ideal probe to demonstrate CYP3A4/5 specificity. To clarify this point, we used monoclonal antibodies that specifically neutralize the activity of CYP3A4/5 or other P450 enzymes (Mei et al., 1999; Shou et al., 2000) (CYP2A6, 2C9, 2C19, and 2D6), as well as specific chemical inhibitors of CYP3A4/5, 1A2, 2A6, 2C9, 2D6, and 2E1. Among all these agents, only ketoconazole and an anti-CYP3A4/5 monoclonal antibody potently inhibited release of tritiated water from [1,2,6,7-3H]testosterone in HLM, demonstrating that this reaction is mediated almost exclusively (>95%) by CYP3A4/5. Further proof for this notion comes from the results of a more detailed analysis of the inhibitory potencies of a series of CYP3A4/5 inhibitors. Ketoconazole, nifedipine, bromocriptine, miconazole, nicardipine, and the proprietary compound A inhibited release of tritiated water from [1,2,6,7-3H]testosterone with IC50 values that were not significantly different from those for inhibition of testosterone 6β-hydroxylation, measured by LC-MS/MS in the same reaction mixture. Taken together, these results demonstrate that the present method faithfully measures the activity of microsomal CYP3A4/5-mediated hydroxylation of testosterone and that it can be used to analyze the inhibitory potencies of investigational drugs.
Draper et al. (1998) previously reported that water release from [1,2,6,7-3H]testosterone and 6β-hydroxylation exhibited similar Km and Vmax values, based on the assumption that the reaction of the tritiated tracer proceeds without any isotope effect, and that formation of 1 Eq of (unlabeled) 6β-hydroxytestosterone corresponds to formation of 1 Eq of tritiated water. However, as discussed below, the assumption that hydroxylation of the tritiated tracer is not subject to a kinetic isotope effect is not valid, and it is not possible, therefore, to determine molar reaction velocities using commercially available [1,2,6,7-3H]testosterone. To calculate the velocity of product formation from the tritiated substrate, it would be necessary to know the specific activity of radiolabel present in the 6β position. For commercial [1,2,6,7-3H]testosterone, this information is not available. According to the information supplied by the vendor, ∼20% of the label is in the 1α position, 19% in 2α, 20% in 6α, 25% in 7β, 10% in (1β plus 7β), and less than 6% in positions 2β and 6β, but the exact proportion of label present in the 6β position is not known.
Aside from the impossibility to determine kinetic reaction parameters, the use of commercial [1,2,6,7-3H]testosterone as a CYP3A probe presents some additional drawbacks. Because of the low amount of label in the 6β position, it is necessary to use quite high amounts of radioactive substrate (0.5–1 × 106 dpm/assay) to obtain adequate product counts. This obviously has a significant impact on the cost of the assay and renders it difficult to miniaturize the product-counting procedure, for example, using 96-well scintillation counters. Furthermore, it cannot be formally excluded that tritium loss from positions other than the 6β position contributes to product formation. The availability of testosterone specifically labeled in the 6β position represents a significant advantage with regard to these issues. When using this novel tracer, the assay volume could be further reduced to 100 μl and the amount of microsomes to 25 μg, and reasonable product counts (2000–4000 dpm) could be obtained with as little as 100,000 dpm/assay of substrate probe. The signal to noise ratio was significantly improved to 25-fold. The 10-fold higher clearance rate of [6β-3H]testosterone relative to that of [1,2,6,7-3H]testosterone is slightly less than expected from the relative proportions of label in the 6β position (100% versus ≤6%). However, as pointed out above, the exact proportion of label is not known for the commercial tracer, and it cannot be excluded that it is slightly higher than the estimate provided by the supplier. Formation of tritiated water from [6β-3H]testosterone was NADPH-dependent and inhibited by a series of CYP3A/5 inhibitors with IC50 values that were not significantly different from those obtained with [1,2,6,7-3H]testosterone. This improved assay thus represents a high-throughput radiometric version of the classical testosterone 6β-hydroxylation assay and should be suitable for rapid screening of the inhibitory potential of investigational drugs, as discussed further below.
Using [6β-3H]testosterone, for which the specific radioactivity is known, as substrate probe allows determination of the kinetic isotope effect, TV/K. The large disparity between the concentrations of tritiated and unlabeled substrate makes it impossible to measure their separate reaction velocities. When trace concentrations of a tritiated substrate are used to measure reaction velocities in the presence of competing unlabeled substrate, it is possible to measure the kinetic isotope effect on the V/K ratio (but not on the individual values of Vmax and Km) (Northrop, 1982). By measuring formation of tritiated water and 6β-hydroxytestosterone in the same reaction mixture, the specific activity of the tritiated water product can be calculated and compared with the initial specific radioactivity of the substrate, and these values can be used to derive TV/K. The kinetic isotope effect on TV/K was 2.3-fold. It was previously reported that 6β-hydroxylation of deuterated testosterone proceeds in the absence of a kinetic isotope effect (Bjorkhem, 1972). This is most likely due to the difference between C–D and C–T bond energies. It cannot be excluded that the modest difference between the formation rates of tritiated water and 6β-hydroxytestosterone involves abstraction of the 6α-hydrogen with partial retention of the isotopic label in the hydroxylated product, as reported for P450cam-mediated oxidation of camphor, where either the exo or the endo hydrogen is abstracted, but rebound occurs only on the exo side (Gelb et al., 1982). Independently of the mechanism, the inhibitory potencies of competitive CYP3A inhibitors are not affected, as demonstrated by the observation that IC50 values for inhibition of tritiated water formation were not significantly different from those for 6β-hydroxytestosterone formation and were not modified by 6-fold isotopic dilution of the substrate probe.
When reaction velocity is measured in the presence of a fixed amount of radiolabeled substrate and increasing concentrations of unlabeled substrate, product counts are expected to decrease when the concentration of unlabeled substrate becomes sufficiently high to compete with binding of the tracer to the active site of the enzyme. As depicted in Fig. 4A, at low concentrations of unlabeled testosterone, the formation of tritiated water is actually enhanced as unlabeled substrate increases, and starts to decrease only at high concentrations (>40 μM) of unlabeled substrate. This effect, called low dose hook, is characteristic for positively cooperative ligand displacement interactions (De Lean and Rodbard, 1979). The reason for the increased formation of tritiated product is that at low substrate concentrations, the reaction velocity of a positively cooperative enzyme increases more than dose proportionally with increasing substrate concentration. When the apparent formation rate of unlabeled product was calculated from the isotopic dilution experiment and plotted against the substrate concentration, a sigmoidal relationship was observed, confirming that the reaction displays positive cooperativity. Data could be fitted to the Hill equation, with a Hill coefficient of 1.4 and a S50 of 98 μM. Measurement of 6β-hydroxytestosterone formation under the same conditions demonstrates that reaction kinetics of unlabeled testosterone indeed displays positive cooperativity, with a Hill coefficient of 1.5 and a S50 of 65 μM. The differences between the values of S50 and the Hill coefficient (n) obtained in the radiometric versus LC-MS/MS assays were not statistically significant (p > 0.1). A Hill coefficient of 1.3 has previously been reported for CYP3A4/5-mediated 6β-hydroxylation of testosterone (Ueng et al., 1997). The good agreement between kinetic parameters obtained from the tracer competition (isotopic dilution) data with those determined by direct measurement of unlabeled product reinforces the notion that formation of tritiated water from [6β-3H]testosterone in HLM is mediated by the same mechanism as formation of 6β-hydroxytestosterone. A low dose hook was also observed using [1,2,6,7-3H]testosterone as substrate. It should be noted that hyperbolic (nonsigmoidal) kinetics have been reported for this tracer (Draper et al., 1998). The most likely explanation for this apparent discrepancy is that the low dose hook was observed in the present studies only at substrate concentrations between 0.4 and 20 μM, whereas Draper et al. (1998) studied reaction kinetics at substrate concentrations greater than 14 μM.
To validate the use of the new assay as a screening method for CYP3A4/5 inhibition, we determined IC50 values for a large number of structurally diverse investigational compounds from different Merck programs and compared the results with those obtained in the conventional testosterone 6β-hydroxylation assay (with product quantification by LC-MS/MS). The results of this analysis indicate that IC50 values obtained with the new radiometric assay are very similar to those of the conventional assay. IC50 values differed less than 4-fold in every case, and for 86% of the compounds, less than 3-fold. Most importantly, not a single compound of the 39 tested would have been misclassified as either a strong or a weak inhibitor based on the results of the radiometric assay.
Several authors have recommended that CYP3A4/5 inhibition should be assessed using more than one substrate (Stresser et al., 2000; Yuan et al., 2002; Cohen et al., 2003). Even though testosterone is not necessarily a better predictor of drug interactions than other substrates, it is one of the most commonly used probes for CYP3A4/5 inhibition studies. A detailed comparison for a large number of compounds of IC50 values obtained with a fluorogenic probe versus testosterone revealed that the correlation between these assays was relatively poor (Cohen et al., 2003). Results obtained in our laboratory support the same conclusion (data not shown). The authors of this study recommended that screening with fluorogenic probes should be followed up by studies with conventional substrates, such as testosterone. The present assay, which combines the advantages of speed, high throughput, and the use of a conventional substrate, should prove to be a valuable tool for rapidly determining the potential of compounds to inhibit CYP3A4/5 in a drug discovery setting. Using appropriately radiolabeled substrates, the assay principle should be adaptable to other P450 enzymes. In the accompanying paper (Di Marco et al., 2005), we describe the synthesis of [4′-3H]diclofenac and the development of a high-throughput CYP2C9 assay using this substrate.
Acknowledgments
We thank Regina Wang for providing IC50 values for NCEs determined by LC-MS/MS assay, Dennis Dean and David Melillo for contributions to the design of [6β-3H]testosterone synthesis, George Doss and Allen Jones for tracer characterization, Magang Shou for providing monoclonal antibodies, and Tom Baillie and the members of the CYP task force of the Merck Drug Metabolism Council for helpful suggestions.
Footnotes
-
This work was supported in part by a grant from the Ministero dell'Istruzione, dell'Università e della Ricerca.
-
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
-
doi:10.1124/dmd.104.002873.
-
ABBREVIATIONS: P450, cytochrome P450; HLM, human liver microsome; HPLC, high-performance liquid chromatography; LC/MS, liquid chromatography-mass spectrometry; DMSO, dimethyl sulfoxide; NCE, new chemical entity; MS/MS, tandem mass spectrometry; IC50, concentration of drug required to inhibit activity by 50%; S50, concentration of substrate at which 50% of maximal activity is observed.
- Received November 4, 2004.
- Accepted December 10, 2004.
- The American Society for Pharmacology and Experimental Therapeutics