Abstract
Studies were conducted to evaluate the potential mechanism-based inactivation of recombinant and human liver microsomal CYP2C8 by clinically used drugs. Several tricyclic antidepressants, calcium channel blockers, monoamine oxidase inhibitors, and various other known CYP3A4 inhibitors exhibited greater inhibition of CYP2C8 (paclitaxel 6α-hydroxylation) following preincubation, consistent with mechanism-based inactivation. Inactivation of recombinant CYP2C8 by phenelzine, amiodarone, verapamil, nortriptyline, fluoxetine, and isoniazid was of the pseudo-first order type and was characterized by respective inactivation kinetic constants (KI and kinact) of 1.2 μM and 0.243 min–1, 1.5 μM and 0.079 min–1, 17.5 μM and 0.065 min–1, 49.9 μM and 0.036 min–1, 294 μM and 0.083 min–1, and 374 μM and 0.042 min–1. Spectral scanning of recombinant CYP2C8 demonstrated the formation of metabolite-intermediate complexes with verapamil, nortriptyline, fluoxetine, and isoniazid, but not amiodarone. In contrast, inactivation by phenelzine resulted from heme destruction by free radicals. Studies with human liver microsomes (HLMs) revealed that nortriptyline, verapamil, and fluoxetine were not mechanism-based inactivators (MBIs) of CYP2C8. Simultaneous inactivation of CYP2C8 and CYP3A4 (paclitaxel 3′-phenyl-hydroxylation) was observed using amiodarone, isoniazid, and phenelzine with the efficiency of inactivation greater for the CYP3A4 pathway. With the exception of phenelzine, glutathione and superoxide dismutase failed to protect CYP2C8 (recombinant and HLMs) or CYP3A4 from inactivation by MBIs. However, the alternate CYP2C8 substrate, torsemide, prevented CYP2C8 inactivation in all cases. These data are consistent with mechanism-based inactivation of CYP2C8 by a range of commonly prescribed drugs, several of which have been implicated in clinically important drug-drug interactions.
Metabolic drug-drug interactions that arise via mechanism-based inactivation (MBI) of cytochrome P450 (P450) enzymes have generated great interest in recent years, since the reduction in metabolic clearance can be severe and long-lasting (Lin and Lu, 1998). The majority of studies performed to date have focused on CYP3A4, the principal isoform involved in the metabolic clearance of drugs in humans. Drug-drug interactions involving macrolide antibiotics (Ito et al., 2003), calcium channel blockers (Ma et al., 2000), human immunodeficiency virus protease inhibitors (Koudriakova et al., 1998), and some antidepressants (Mayhew et al., 2000) can be quantitatively better explained when the inactivation component of CYP3A4 inhibition is considered. However, it is now apparent that the function of other P450 isoforms may also be significantly impaired by drugs that act as mechanism-based inactivators (MBIs). Clinically relevant inactivation of CYP1A2 by zileuton (Lu et al., 2003), CYP2B6 by clopidogrel and ticlopidine (Richter et al., 2004), CYP2C9 by suprofen (O'Donnell et al., 2003), CYP2C19 by ticlopidine (Ha-Duong et al., 2001), and CYP2D6 by paroxetine and metoclopramide (Desta et al., 2002; Bertelsen et al., 2003) has recently been demonstrated in vitro.
In contrast to reversible inhibition, MBI requires at least one P450 catalytic cycle to produce reactive intermediate species that covalently modify the enzyme and cause impairment of function (Lin and Lu, 1998). In vitro, MBIs characteristically exhibit: 1) time-dependent inactivation, 2) saturation kinetics, 3) reduced rates of inactivation in the presence of alternate substrates, 4) a 1:1 stoichiometry for inhibitor and enzyme, 5) involvement of a catalytic step, 6) inactivation of enzyme prior to release of reactive species, and 7) quasi-irreversible or irreversible loss of function (Silverman, 1988). The maximal rate constant of inactivation (kinact) and the inactivator concentration required for half-maximal inactivation (KI) are experimentally derived in vitro kinetic constants that are useful descriptors of the potency of a drug as a MBI. The ratio of kinact/KI may additionally serve as an indicator of inactivator efficiency. In the case of P450-mediated drug-drug interactions, physiological- and non-physiological-based models have incorporated these kinetic constants to predict the degree of P450 inactivation in vivo and may provide sound estimates of the decrease in clearance when other drugs are coadministered (Mayhew et al., 2000; Ito et al., 2003).
Although the role of CYP2C8 was previously considered to be limited to the oxidation of endogenous retinoids and fatty acids, such as all-trans-retinoic acid and arachidonic acid, there is growing awareness of the importance of this enzyme in human drug metabolism. Interestingly, there appears to be a degree of overlapping substrate and inhibitor selectivity between CYP3A4 and CYP2C8. Thus, CYP2C8 contributes, in part, to the metabolism of the predominant CYP3A4 substrates carbamazepine, verapamil, and zopiclone (Ong et al., 2000). Conversely, CYP3A4 contributes to the metabolism of the predominant CYP2C8 substrates cerivastatin and paclitaxel (Ong et al., 2000). Both isoforms also metabolize repaglinide, methadone, and chloroquine (Bidstrup et al., 2003; Projean et al., 2003; Wang and DeVane, 2003). Ong et al. (2000) additionally reported that amiodarone, amitriptyline, terfenadine, quinine, midazolam, and triazolam caused >50% inhibition of CYP2C8 at concentrations corresponding to their relevant CYP3A4 Km values, suggesting similar affinities for both enzymes. Furthermore, other CYP3A4 substrates such as nifedipine, felodipine, and testosterone also appear to inhibit CYP2C8 (Harris et al., 1994).
To date, there have been no investigations of potential mechanism-based inactivation of CYP2C8. Given the overlapping substrate and inhibitor selectivities of CYP3A4 and CYP2C8 and the emerging role of CYP2C8 in the metabolism of drugs, this work aimed to evaluate a range of CYP3A4 inhibitors and other drugs as mechanism-based inactivators of CYP2C8, characterize the kinetics of CYP2C8 inactivation, and determine the mechanisms of CYP2C8 inactivation. Drugs investigated as potential MBIs of CYP2C8 included a series of macrolide antibiotics and calcium channel blockers, 17α-ethynylestradiol, amiodarone, fluoxetine, and tamoxifen (all known MBIs of CYP3A4) and several monoamine oxidase inhibitors (MAOIs) and tricyclic antidepressants (TCAs), classes that have been implicated as MBIs in previous studies (Muakkassah et al., 1981; Bensoussan et al., 1995; Wen et al., 2002).
Materials and Methods
Chemicals
Testosterone, 6β-hydroxytestosterone, paclitaxel, troleandomycin, erythromycin, roxithromycin, amitriptyline, nortriptyline, desipramine, norclomipramine, diltiazem, verapamil, nicardipine, felodipine, mibefradil, selegiline, tranylcypromine, isoniazid, phenelzine, fluoxetine, 17α-ethynylestradiol, diethylstilbestrol, 4-methylumbelliferone, trimethoprim, glucose 6-phosphate, glucose-6-phosphate dehydrogenase, β-NADP, mannitol, reduced glutathione, and superoxide dismutase were purchased from Sigma-Aldrich (Sydney, Australia). 6α-Hydroxypaclitaxel was purchased from BD Gentest (Woburn, MA). Other chemicals were kindly donated by the following sources: azithromycin, Pfizer Australia (Sydney, Australia); clarithromycin, Abbott Laboratories (Chicago, IL); torsemide, Roche Diagnostics (Mannheim, Germany); desmethylnortriptyline, F. Hoffmann-La Roche (Basel, Switzerland); amiodarone and desethylamiodarone, Sanofi UK Ltd (Manchester, England); indinavir, Merck Sharp and Dohme Pty Ltd (Sydney, Australia); tamoxifen, Imperial Chemical Industries PLC (Manchester, England); and ketoconazole, Janssen-Cilag Pty (Sydney, Australia). Restriction enzymes were purchased from New England Biolabs Inc. (Hitchin, Hertfordshire, UK) and Escherichia coli DH5α cells from Invitrogen (Melbourne, Australia). All other chemicals and reagents were of analytical reagent grade.
Construction of CYP2C8, CYP3A4, and rOxR Expression Plasmids
N terminus modifications known to promote high levels of bacterial expression of human P450 were incorporated into the wild-type CYP2C8 (accession no. Y00498) and CYP3A4 (accession no. AF182273) cDNAs (Boye et al., 2004). The N-terminal membrane anchor of CYP2C8 was replaced with modified sequence derived from CYP17A as previously described (Richardson et al., 1995). In addition, the bovine 17α-hydroxylase leader sequence (MALLLAVFL) was immediately followed by the CYP2C consensus sequence (GLSCLLLLS). Generation of the 17α-hydroxylase leader sequence in CYP3A4 cDNA utilized polymerase chain reaction-directed mutagenesis to delete codons 3 to 10 using the following primers: sense, 5′ TACATATGGCTCTGTTATTAGCAGTTTTTCTGGTGCTCCTCTATCTATATGG 3′; antisense, 5′ AATGTCGACTCAGGCTCCACTTACGGTGCC 3′. To facilitate directional ligation into the pCW plasmid, NdeI and SalI restriction sites (bold text) were incorporated into the sense and antisense CYP3A4 oligonucleotides, respectively. The 1486-base pair 17α-CYP2C8 and 1496-base pair 17α-CYP3A4 polymerase chain reaction products were digested with NdeI/HindIII and NdeI/SalI prior to ligation into the pCW ori(+) plasmid.
The cDNA coding for rat cytochrome P450 oxidoreductase (rOxR) consisted of the OmpA signal sequence (MKKTAIAIAVALAGFATVAQA) fused upstream of the full-length native rat OxR sequence. The rOxR expression construct was generated using the bacterial plasmid pACYC184, as documented previously (Boye et al., 2004).
Heterologous Coexpression of pCW-17αCYP2C8 and pCW-17αCYP3A4 with pACYC OmpA-rOxR
pCW-17αCYP2C8 and pCW-17αCYP3A4 constructs were cotransformed into DH5α E. coli cells with pACYC OmpA-rOxR. Ampicillin/chloramphenicol-selected colonies were screened for the correct plasmid by restriction enzyme analysis. Plasmid DNA was purified with the QIAprep spin miniprep kit (QIAGEN Operon, Alameda, CA) and confirmed on both strands by sequencing (ABI Prism 3100). Cells were cultured and membrane fractions separated as described (Boye et al., 2004). The total protein concentration of membrane fractions was determined (Lowry et al., 1951) and holoenzyme quantified (Omura and Sato, 1964). The rate of reduction of cytochrome c was used as a measure of rOxR activity in membrane fractions (Yasukochi and Masters, 1976).
Human Liver Microsomes
The Flinders Medical Centre Ethics Review Committee approved the use of human liver tissue for in vitro drug metabolism studies obtained from the liver “bank” of the Department of Clinical Pharmacology. Microsomes from six livers (H6, H7, H10, H12, H13, and H40) were prepared by differential centrifugation. Liver portions in phosphate buffer (0.1 M, pH 7.4) containing 1.15% w/v potassium chloride were homogenized sequentially with a Janke and Kunkle Ultra Tarex (24,000 rpm) and a Potter-Elvehjem homogenizer (mechanical drive at 1480 rpm). The homogenate was centrifuged at 700g for 10 min and then at 10,000g for another 10 min. The supernatant fraction was aspirated and centrifuged at 105,000g for 60 min at 4°C. The resulting pellet was resuspended in phosphate buffer (0.1 M, pH 7.4) containing 1.15% w/v potassium chloride and centrifuged at 105,000g for 60 min at 4°C. The microsomal pellet was suspended in phosphate buffer (0.1 M, pH 7.4) containing 20% glycerol and stored at –80°C until use. Microsomal protein concentrations were determined by the method of Lowry et al. (1951) using bovine serum albumin as standard. The kinetics of testosterone and paclitaxel metabolism were determined using the microsomes from the six livers. Inhibition studies were conducted using pooled HLMs with equal amounts of microsomal protein from each liver.
Inhibition of CYP3A4 and CYP2C8 in the MBI Screen
Coincubation. Inhibitors were incubated with HLMs or recombinant enzyme and NADPH-regenerating system in the presence of probe substrates for 15 min (testosterone assay) or 10 min (paclitaxel assay) at 37°C. Coincubation mixtures contained either 0.1 mg/ml HLMs, recombinant CYP3A4 (10 pmol/ml) or CYP2C8 (10 pmol/ml), NADPH-regenerating system (1 mM NADP, 10 mM glucose 6-phosphate, 2 IU/ml glucose-6-phosphate dehydrogenase, 5 mM MgCl2), inhibitor, probe substrate, and phosphate buffer (0.1 M, pH 7.4) in a total volume of 500 μl (testosterone assay) or 200 μl (paclitaxel assay). Probe substrates were included at concentrations close to reaction Km values, 40 μM testosterone using HLMs and recombinant CYP3A4 or 10 μM paclitaxel using HLMs and 1.5 μM paclitaxel using recombinant CYP2C8 (see Results). Total solvent concentration (acetonitrile for the testosterone assay and methanol for the paclitaxel assay) was either 1 or 2%. The testosterone assay was terminated by the addition of 5 μl of 70% perchloric acid, and 0.05 μg of 4-methylumbelliferone was added as the internal standard. The paclitaxel assay was terminated by the addition of 200 μl of ice-cold acetonitrile containing 0.1 μg of diethylstilbestrol as the internal standard. Mixtures were vortex mixed, chilled on ice, and then centrifuged (4000g) before analysis of the supernatant fraction by HPLC. Rates of formation of all metabolites (6β-hydroxytestosterone, 6α-hydroxypaclitaxel, and 3′-phenyl-hydroxypaclitaxel) were linear with respect to protein concentration and incubation time.
Preincubation. Inhibitors were preincubated with HLMs or recombinant enzyme and NADPH-regenerating system in the absence of the probe substrates for 30 min at 37°C. A preincubation time of 20 min was employed for CYP3A4 studies with the macrolides due to their marked effect on activity. Preincubation mixtures contained either HLMs (0.1 mg/ml), recombinant CYP3A4 (10 pmol/ml) or CYP2C8 (10 pmol/ml), NADPH-regenerating system, inhibitor, and phosphate buffer (0.1 M, pH 7.4) in a total volume of 500 μl (testosterone assay) or 200 μl (paclitaxel assay). Following preincubation, probe substrates were added at the concentrations specified for the coincubations, and the reactions were allowed to proceed. Total solvent concentration (acetonitrile for the testosterone assay and methanol for the paclitaxel assay) was either 1 or 2%. Reactions were terminated and the samples were prepared for HPLC analysis as described above.
Kinetics of Paclitaxel Metabolism Inactivation
A two-step incubation method was utilized to characterize the time- and concentration-dependent inhibition of paclitaxel metabolism by selected inactivators in vitro. Inactivation assays (200 μl) contained recombinant CYP2C8 (100 pmol/ml) or HLMs (1 mg/ml), NADPH-regenerating system, and selected inactivators (at least five different concentrations) in phosphate buffer (0.1 M, pH 7.4). At selected preincubation times, a 20-μl aliquot was removed and diluted 10-fold to the activity assay containing either 30 μM (recombinant CYP2C8) or 100 μM (HLMs) paclitaxel and NADPH-regenerating system. Reactions were terminated after 10 min, and the concentration of the 6α- and 3′-phenyl-hydroxylated metabolites of paclitaxel were determined by HPLC.
Effect of Trapping Agents and Torsemide on Inactivation of Paclitaxel Metabolism
The inactivation of paclitaxel metabolism by selected drugs was investigated in the presence of the nucleophile trapping agent glutathione (2 mM), the scavengers of reactive oxygen species superoxide dismutase (1000 U/ml) and mannitol (1 mM), and the alternate CYP2C8 substrate, torsemide (500 μM). These components were included individually with drugs in an inactivation assay (10 min) with either recombinant CYP2C8 (100 pmol/ml) or HLMs (1 mg/ml) prior to the determination of the remaining paclitaxel 6α- and 3′-phenyl-hydroxylation activity. Control activities were determined in the absence of inactivating drugs.
Effect of Ultrafiltration on Inactivation of Paclitaxel Metabolism
Ultrafiltration studies were performed to determine whether the catalytic function of recombinant CYP2C8 could be restored following preincubation with putative MBIs. Inactivation assays (250 pmol/ml recombinant CYP2C8) were chilled on ice following a 10-min preincubation, transferred to Ultrafree-MC filters (30,000 nominal molecular weight limit regenerated cellulose membrane; Millipore, Yonezawa, Japan), and centrifuged at 5000g for 15 min. Samples were washed with 200 μl of phosphate buffer (0.1 M, pH 7.4), centrifuged at 5000g for 30 min, and then resuspended with 200 μl of phosphate buffer (0.1 M, pH 7.4). A 20-μl aliquot was removed to determine the rate of 6α-hydroxypaclitaxel formation. Control measurements were performed in the absence of inactivating drugs.
Spectral Difference Scanning
Difference spectra (500–380 nm) between reference and sample cuvettes were recorded using a Cary 300 double-beam UV-visible spectrophotometer (Varian Inc., Melbourne, Australia). Incubations (500 μl in 0.1 M phosphate buffer, pH 7.4) containing recombinant CYP2C8 (1000 or 2000 pmol/mol for isoniazid), with or without NADPH-regenerating system, were divided into two 250-μl cuvettes. Drugs were added to the sample cuvette (200 μM nortriptyline, 100 μM verapamil, 200 μM fluoxetine, 10 μM amiodarone, 500 μM isoniazid, or 10 μM phenelzine), and an equivalent volume of solvent was added to the reference cuvette. Both were placed in a shaking water bath at 37°C for 0 to 30 min before difference spectra were recorded. Potassium ferricyanide (200 μM) was then added to both cuvettes and the difference spectrum re-recorded.
Cytochrome P450 Reduced CO-Difference Spectroscopy
Incubations (500 μl in 0.1 M phosphate buffer, pH 7.4) containing recombinant CYP2C8 (1000 pmol/ml), NADPH-regenerating system, and inactivating drugs were performed for 10 min at 37°C and then terminated with 50 μl of ice-cold phosphate buffer (0.1 M pH 7.4) containing 10% glycerol and 1% Emulgen 913. Following the addition of sodium dithionite, the mixture was divided between two 250-μl cuvettes. CO gas was gently bubbled through the sample cuvette for 1 min. The spectrum of the reduced CO complex was recorded between 500 and 400 nm. Before termination, a 20-μl aliquot was removed to determine the 6α-hydroxypaclitaxel concentration. Control samples were prepared in the absence of inactivating drugs.
HPLC Assays
Testosterone 6β-Hydroxylation (CYP3A4). 6β-Hydroxytestosterone was quantified by reversed-phase HPLC using an Agilent 1100 series HPLC system (Agilent Technologies, Sydney, Australia) fitted with a Waters (Waters, Milford, MA) Nova-Pak C18 column (1.5 cm × 3.9-mm i.d., 4-μm particle size). The mobile phase, which consisted of water (A) and acetonitrile (B), was delivered at a flow rate of 1 ml/min according to the gradient: initial conditions, 80% A/20% B changed to 70% A/30% B over 7 min, then 30% A/70% B over 1 min, which was held for 10 s before returning to the initial conditions. Retention times of 4-methylumbelliferone (assay internal standard), 6β-hydroxytestosterone, and testosterone measured using UV detection at 241 nm were 2.5, 6.2, and 9.8 min, respectively. Concentrations of 6β-hydroxytestosterone in incubations were determined by comparison of the peak area ratios with those of a standard curve.
Paclitaxel 6α-Hydroxylation (CYP2C8) and 3′-Phenyl-hydroxylation (CYP3A4). 6α-Hydroxypaclitaxel and 3′-phenyl-hydroxypaclitaxel were quantified by reversed-phase HPLC as described (Kerdpin et al., 2004). Under these conditions, respective retention times for 3′-phenyl-hydroxypaclitaxel, 6α-hydroxypaclitaxel, diethylstilbestrol, and paclitaxel were 8.2, 10.6, 12.1, and 14.4 min. 3′-Phenyl-hydroxypaclitaxel and 6α-hydroxypaclitaxel were quantified by using paclitaxel as the standard (Ong et al., 2000).
Data Analysis
All results represent the mean of duplicate determinations. Kinetic constants for paclitaxel 6α- and 3′-phenyl-hydroxylation and testosterone 6β-hydroxylation were derived (EnzFitter; Biosoft, Cambridge, UK) by fitting either to the Michaelis-Menten equation, where v is the rate of metabolite formation, Vmax is the maximum velocity, Km is the Michaelis constant (substrate concentration at half-maximal velocity), and [S] is the substrate concentration, or the Hill equation, which describes sigmoidal kinetics, where S50 is the substrate concentration resulting in half-maximal velocity and n is the Hill coefficient.
The preincubation effect in the initial screen for MBI was determined by subtracting the percentage of control activity remaining following preincubation from the percentage of control activity remaining following coincubation at the same inhibitor concentration.
Apparent kinetic constants for the inactivation of paclitaxel metabolism were determined using the observed inactivation rate constant (kobs) at each MBI concentration. Since P450 activity declines in the presence of NADPH-regenerating system alone (see Results), control values at each preincubation sampling time represent the maximum possible catalytic function at that time (i.e., 100% activity in the absence of MBIs). The values of kobs at each MBI concentration were determined from the slope when the logarithm of the remaining activity as percentage of control was plotted against the preincubation time. Inactivation data were fitted to the following equation using nonlinear least-squares regression. where kinact, I, and KI are the maximum rate of inactivation, concentration of inactivator, and the inactivator concentration required for half-maximal inactivation.
Results
Expression of Recombinant CYP2C8 and CYP3A4. Levels of expression of CYP2C8/rOxR (328 nmol P450/liter culture and 313 nmol rOxR/liter culture) and CYP3A4/rOxR (128 nmol P450/liter culture and 117 nmol rOxR/liter culture) were comparable with those described previously (Boye et al., 2004).
Kinetics of Testosterone and Paclitaxel Metabolism. Available evidence indicates that the conversion of paclitaxel to 6α-hydroxypaclitaxel and 3′-phenyl-hydroxypaclitaxel by HLMs is catalyzed exclusively by CYP2C8 and CYP3A4, respectively (Cresteil et al., 2002; Foti and Fisher, 2003). The identity of 3′-phenyl-hydroxypaclitaxel formed by incubations with HLMs was confirmed using the CYP3A inhibitor troleandomycin. Preincubation with 10 μM troleandomycin totally inhibited paclitaxel 3′-phenyl-hydroxylation without affecting the 6α-hydroxylation pathway, consistent with previous observations (Harris et al., 1994). Furthermore, 3′-phenyl-hydroxypaclitaxel was formed in incubations with recombinant CYP3A4, but not by recombinant CYP2C8. The identity of 6α-hydroxypaclitaxel and 6β-hydroxytestosterone was confirmed by reference to authentic standards. The kinetics of paclitaxel 6α-hydroxylation by HLMs and recombinant CYP2C8 was described by the Michaelis-Menten equation, whereas testosterone 6β-hydroxylation and paclitaxel 3′-phenyl-hydroxylation formation by HLMs were sigmoidal and described by the Hill equation (Fig. 1). Derived kinetic parameters are given in Table 1.
Screening for Preincubation Effects with CYP3A4 and CYP2C8. MBIs exhibit greater inhibition following a preincubation step, and this permits screening of drugs to identify and/or confirm a time-dependent component to CYP2C8 and CYP3A4 inhibition. Table 2 shows the percent inhibition difference observed between co- and preincubation conditions for the potential inhibitors investigated here.
With the exception of desmethylnortriptyline, desethylamiodarone, and indinavir using either recombinant CYP3A4 or HLMs as the enzyme source, and tamoxifen and selegiline using HLMs as the enzyme source, all drugs inhibited testosterone 6β-hydroxylation to a greater extent following preincubation compared with coincubation (i.e., inhibition difference was positive) (Table 2). The macrolide antibiotics (troleandomycin, erythromycin, clarithromycin, roxithromycin, and azithromycin) did not inhibit paclitaxel 6α-hydroxylation at concentrations up to 100 μM. The secondary amine tricyclic antidepressants, nortriptyline, desipramine, and norclomipramine, exhibited positive inhibition differences with paclitaxel 6α-hydroxylation by recombinant CYP2C8. This same preincubation effect was not observed with the corresponding tertiary and primary amine tricyclic antidepressants (amitriptyline and desmethylnortriptyline) and was minor (<10%) with HLMs. The calcium channel blockers, diltiazem, verapamil, nicardipine, felodipine, and mibefradil, and the selective serotonin reuptake inhibitor fluoxetine also showed positive inhibition differences in experiments with recombinant CYP2C8. However, this effect was either reduced or not observed in parallel experiments conducted with HLMs. Consistency between enzyme sources with respect to the observation of a preincubation effect was noted for 17α-ethynylestradiol, amiodarone, desethylamiodarone, isoniazid, and phenelzine, although the magnitude was generally greater using recombinant enzyme. There was no difference between the co- and preincubation inhibition by selegiline, but tranylcypromine showed a negative inhibition difference (i.e., coincubation inhibition > preincubation inhibition). A preincubation effect on paclitaxel 6α-hydroxylation was observed for tamoxifen using HLMs, but not with recombinant CYP2C8.
Inactivation Kinetics of Paclitaxel 6α-Hydroxylation Catalyzed by Recombinant CYP2C8. Based on the initial screening results, representative drugs from different therapeutic classes were selected for inactivation studies. Drugs were selected on the basis of inhibitory potency, the magnitude of preincubation effect, the apparent disparity between results obtained using recombinant CYP2C8 and HLMs, and the possible mechanistic basis of MBI. Recombinant CYP2C8 activity declined to approximately 35% of original activity when incubated for 30 min in the presence of NADPH-regenerating system alone (between day coefficient of variation <5%). Figure 2 shows the time- and concentration-dependent inhibition of recombinant CYP2C8 by nortripyline, verapamil, fluoxetine, amiodarone, isoniazid, and phenelzine, all of which exhibited pseudo-first order inactivation kinetics over the preincubation times and concentrations studied. Inactivation was saturable in all cases. Kinetic constants for the inactivation of recombinant CYP2C8 by these drugs are presented in Table 3. The rank order of inactivation efficiency was phenelzine > amiodarone > verapamil > nortriptyline > fluoxetine > isoniazid.
Inactivation of Paclitaxel Metabolism by HLMs. The differential metabolism of paclitaxel by CYP2C8 (6α-hydroxylation) and CYP3A4 (3′-phenyl-hydroxylation) was used to assess the possibility of simultaneous MBI in HLMs. CYP2C8 and CYP3A4 activity declined to approximately 55% and 75% of original activity, respectively, when HLMs were incubated for 30 min in the presence of NADPH-regenerating system alone (between day coefficient of variation <10%). Figure 3 shows the observed inactivation profiles for the time- and concentration-dependent inhibition of paclitaxel metabolism by amiodarone, isoniazid, and phenelzine. The corresponding inactivation kinetic constants are given in Table 3. Inactivation of human liver microsomal 6α- and 3′-phenyl-hydroxylation was saturable and exhibited pseudo-first order inactivation kinetics over the preincubation times and concentrations studied; however, the determination of 3′-phenyl-hydroxylation inactivation kinetics by phenelzine was not possible because time-dependent inhibition was only observed at concentrations greater than 10 μM. In contrast to results with recombinant CYP2C8, inhibition of HLM paclitaxel 6α-hydroxylation by nortriptyline, verapamil, and fluoxetine was not time-dependent, although verapamil and fluoxetine did inhibit paclitaxel 3′-phenyl-hydroxylation in a manner consistent with MBI (data not shown). kinact/KI ratios for phenelzine, amiodarone, and isoniazid inactivation of paclitaxel 6α-hydroxylation were lower in comparison with recombinant CYP2C8, suggesting reduced efficiency of inactivation. However, the rank order of inactivation efficiency was the same: phenelzine > amiodarone > isoniazid.
Effect of Glutathione, Reactive Oxygen Scavengers, and Torsemide on CYP2C8 Inactivation. The addition of glutathione, superoxide dismutase, and mannitol to incubations did not prevent or slow inactivation of recombinant CYP2C8 by nortriptyline, verapamil, fluoxetine, amiodarone, and isoniazid (Table 4); however, the rate of phenelzine-mediated inactivation was reduced in the presence of glutathione and superoxide dismutase, although not by mannitol. Similarly, with the exception of phenelzine, glutathione, superoxide dismutase, and mannitol did not prevent or slow CYP2C8 and CYP3A4 inactivation by drugs that were shown to inactivate paclitaxel 6α- and 3′-phenyl-hydroxylation by HLMs (data not shown). Addition of the alternate CYP2C8 substrate torsemide (500 μM) to incubations prevented or significantly reduced the inactivation of paclitaxel 6α-hydroxylation in all cases (Table 4).
Effect of Ultrafiltration on Recombinant CYP2C8 Inactivation. Recombinant CYP2C8 activity was not restored by ultrafiltration following preincubation with nortriptyline, verapamil, fluoxetine, and amiodarone (Table 4). The apparent inactivation caused by phenelzine was partially reversed although full catalytic function was restored in the case of isoniazid (Table 4) and following preincubation with trimethoprim, which is a known reversible CYP2C8 selective inhibitor (data not shown).
Spectral Difference Scanning with Recombinant CYP2C8. Further studies were conducted to investigate whether the inactivation of recombinant CYP2C8 by nortriptyline, verapamil, fluoxetine, amiodarone, isoniazid, and phenelzine occurred via the formation of metabolite-intermediate complexation (MIC). Figure 4 shows the spectral difference scans for these six drugs; for clarity only, the scans for 10-min incubations are provided. When incubated with recombinant CYP2C8 in the presence of NADPH-regenerating system, nortriptyline, verapamil, and fluoxetine showed time-dependent increases in absorbance maxima in the Soret region (456–449 nm) that were sensitive to ferricyanide, consistent with MIC formation by alkylamine drugs. Incubations with isoniazid and NADPH-regenerating system resulted in an increase in absorbance between 510 and 420 nm. These spectral changes were sensitive to ferricyanide and transitionary, reaching a maximum by 10 min, but then rapidly decreasing. This is consistent with MIC formation by isoniazid observed using rat liver microsomes (Muakkassah et al., 1981). Amiodarone, which contains an amine function, did not form a MIC with recombinant CYP2C8. Rather, there was a NADPH-dependent, but ferricyanide insensitive, increase in absorbance with a maximum that shifted from around 435 nm to 428 nm with increasing incubation time. In contrast, phenelzine bound to CYP2C8 in the absence of NADPH-regenerating system to give a typical type II difference-binding spectrum with an absorbance maximum at 428 nm. When phenelzine was incubated with NADPH-regenerating system, the absorbance was ferricyanide insensitive, and the maximum increased and shifted from 442 to 438 nm as incubation time increased.
Reduced CO-Difference Spectra with Recombinant CYP2C8. Reduced CO-difference spectra were recorded following incubations of recombinant CYP2C8 with nortriptyline, verapamil, fluoxetine, amiodarone, isoniazid, and phenelzine. Consistent with MIC formation, the decrease in P450 content corresponded to the reduction in paclitaxel 6α-hydroxylation activity following incubations with nortriptyline, verapamil, and fluoxetine (Fig. 5). For isoniazid, the decrease in P450 content was significantly greater than the reduction in CYP2C8 activity, presumably due to the transient nature of the hydrazide MIC. Amiodarone did not decrease P450 content, and the majority of the inactivation caused by phenelzine was explained by destruction of the heme component of CYP2C8.
Discussion
The present in vitro studies provide the first data demonstrating MBI of CYP2C8. The overlapping substrate and inhibitor selectivities between CYP3A4 and CYP2C8 were used initially as the basis to screen potential MBIs of CYP2C8. Since MBIs reduce the availability of functional P450 during preincubation, greater inhibition following preincubation is expected (Silverman, 1988). Using testosterone 6β-hydroxylation, preincubation effects were observed for previously described CYP3A4 MBIs, including the macrolide antibiotics (Ito et al., 2003), several calcium channel blockers (Ma et al., 2000), isoniazid (Wen et al., 2002), amiodarone (Ohyama et al., 2000b), fluoxetine (Mayhew et al., 2000), and 17α-ethynylestradiol (Lin et al., 2002). This was not the case for desethylamiodarone, indinavir, or the potent reversible inhibitor, ketoconazole. Small preincubation effects (<10%) were observed for the secondary amine TCAs, nortriptyline, desipramine, and norclomipramine, but the significance of these remains to be fully explored. Consistent with a previous report, tamoxifen showed a preincubation effect with recombinant CYP3A4 which was not apparent in corresponding studies with HLMs (Zhao et al., 2002). Greater inhibition of CYP3A4 following preincubation was also observed for several MAOIs. Phenelzine was previously shown to cause loss of the heme content of rat microsomal P450 by free radical formation (Muakkassah and Yang, 1981), whereas selegiline and tranylcypromine were studied as reversible inhibitors of human P450 without consideration of MBI (Taavitsainen et al., 2000, 2001).
Many of the confirmed CYP3A4 MBIs also demonstrated preincubation effects when corresponding screening studies were performed with CYP2C8 (Table 2). Notable exceptions were the macrolide antibiotics, which did not inhibit CYP2C8 (Harris et al., 1994) and tranylcypromine. Interestingly, desethylamiodarone showed a preincubation effect for CYP2C8, but not with CYP3A4. Although consistency between results with recombinant CYP2C8 and HLMs was not always observed, several drugs including isoniazid, phenelzine, amiodarone, desethylamiodarone, and 17α-ethynylestradiol were identified by these initial studies as potentially significant MBIs of CYP2C8.
A weakness of the MBI screening approach is that it is not possible to differentiate between the preincubation effects caused by MBI and the preincubation effects that may result from the generation of a potent reversible inhibitory metabolite(s). In the absence of a dilution step and using a probe substrate concentration near Km, an observation of time-dependent inhibition is an indicator of possible MBI rather than verification. Despite this, the approach was validated by correctly confirming or rejecting a drug as a CYP3A4 MBI based on previous reports as described above.
Nortriptyline, verapamil, fluoxetine, and isoniazid may be classified as MBIs of CYP2C8 using recombinant enzyme based on the following observations: 1) time- and concentration-dependent inhibition (Fig. 2), 2) saturable inactivation, 3) inactivation proceeded via a catalytic step(s) as indicated by the requirement of NADPH, 4) inactivation rate was reduced by the alternate substrate torsemide (Table 4), 5) inactivation rate was not reduced in the presence of glutathione and scavengers of reactive oxygen species, suggesting that the production of reactive intermediates remained within the active site prior to inactivation (Table 4), and 6) loss of activity was quasi-irreversible consistent with MIC formation (Fig. 4; Table 4). Many alkylamine drugs are associated with MIC formation, and it was therefore not surprising that the inactivation of CYP2C8 by nortriptyline, verapamil, and fluoxetine, proceeded via this mechanism (Bensoussan et al., 1995). The unusual spectral changes observed for isoniazid were also consistent with the formation of a MIC, albeit from the metabolism of the hydrazide moiety (Muakkassah et al., 1981). Interestingly, disruption of this MIC was possible by ultrafiltration, a characteristic not observed for nortriptyline, verapamil, and fluoxetine (Table 4). The transient nature of the hydrazide MIC and the reversibility of the decrease in P450, observed by the partial restoration of CYP2C8 activity during activity assays (Fig. 5), is distinctly different from the more stable MIC formed from alkylamine drugs.
Amiodarone was also expected to form a stable MIC with CYP2C8 given the significant role of CYP2C8 in amiodarone deethylation (Ohyama et al., 2000a) and given that MIC formation by amiodarone in rodents was consistent with tertiary amine metabolism to nitrosoalkane reactive intermediates (Larrey et al., 1986). However, the spectral studies did not detect MIC formation or heme loss (Figs. 4 and 5) despite irreversible loss of catalytic function (Table 4). Along with the tertiary amine function, the furan ring in amiodarone is a structural moiety that has previously been associated with MBI; however, metabolism of the furan ring of furafylline was recently excluded as a basis for CYP1A2 inactivation (Racha et al., 1998). Furthermore, other furan-containing drugs, such as methoxsalen, are potent suicide inhibitors associated with loss of spectrally observable heme (Tinel et al., 1987), a feature not observed here. Thus, the exact mechanism by which amiodarone inactivates CYP2C8 remains unclear, although covalent binding to the CYP2C8 apoprotein may explain the NADPH-dependent spectral differences observed.
Although fulfilling several in vitro MBI criteria, the inactivation of recombinant and human liver microsomal CYP2C8 by phenelzine was not, strictly speaking, mechanism-based. Inactivation rate was reduced in the presence of the nucleophile glutathione and the reactive oxygen species scavenger superoxide dismutase, indicating that reactive intermediates escape the active site prior to inactivation (Table 4). Consistent with these observations, heme destruction via free radical formation during oxidation of phenelzine has been proposed previously (Muakkassah and Yang, 1981). In addition, a potent reversible metabolite(s) appears to contribute toward time-dependent inhibition because activity was partially restored by ultrafiltration (Table 4). Such free radicals and metabolite(s) are likely to cause inhibition and inactivation of other P450 enzymes. Hence, phenelzine should be considered a metabolically activated inactivator (Silverman, 1988).
The differential metabolism of paclitaxel by HLMs was used to study the kinetics of simultaneous CYP2C8 and CYP3A4 inactivation by amiodarone, isoniazid, and phenelzine (Fig. 3). Although Foti and Fisher (2003) recently reported the 3′-phenyl-hydroxylation pathway as a metabolic route specific to CYP3A4 and not CYP3A5, the 3′-phenyl-hydroxylation inactivation kinetic constants obtained here for verapamil, fluoxetine, and amiodarone agree well with those reported using the nonselective CYP3A substrates midazolam and testosterone (Mayhew et al., 2000; Ohyama et al., 2000b; Wang et al., 2004). The efficiency of CYP3A4 inactivation was greater compared with CYP2C8 for all three drugs tested, although interenzyme comparisons were not possible for phenelzine. The kinact/KI ratio is used to evaluate the potential clinical impact of MBI (Lu et al., 2003). In comparison with other clinically significant MBIs, kinact/KI ratios for the effects of amiodarone and isoniazid on human liver microsomal paclitaxel 6α-hydroxylation are low. It is therefore unlikely that these drugs would significantly reduce the in vivo clearance of other CYP2C8 substrates. However, inactivation of CYP2C8 by phenelzine was comparable with CYP3A4 inactivation by erythromycin, a drug commonly associated with CYP3A4-mediated drug-drug interactions (Ito et al., 2003).
In contrast to results with recombinant CYP2C8, nortriptyline, verapamil, and fluoxetine were not MBIs of the human liver microsomal isoform. Furthermore, the efficiency of inactivation was reduced for amiodarone, isoniazid, and phenelzine, as reflected by lower kinact/KI ratios. Similar discrepancies between recombinant and human liver microsomal MBI kinetic data have been noted previously for 5-fluoro-2-[4-[(2-phenyl-1H-imidazol-5-yl)methyl]-1-piperazinyl]pyrimidine (SCH 66712), tamoxifen, and verapamil (Palamanda et al., 2001; Zhao et al., 2002; Wang et al., 2004). This may arise from the nonspecific binding of certain drugs to HLMs, which would effectively lower the “available” concentration present in incubations and confound the determination of kinetic constants. In comparing the results for HLMs and recombinant CYP2C8, the higher apparent Km for paclitaxel 6α-hydroxylation, the smaller preincubation effects observed during the MBI screen, and the greater KI values for amiodarone and phenelzine all provide supporting evidence for this hypothesis. Despite this, the exact differences (if any) between the binding of drugs to human liver microsomes and E. coli membrane fractions are unknown. Alternatively, Palamanda et al. (2001) suggest that sufficient catalytic turnover to the reactive intermediate occurs with the isolated recombinant enzyme, whereas these reactions are of lesser significance due to competing metabolic pathways in HLMs that lower the MBI concentration. However, both explanations fail to explain the results for nortriptyline, verapamil, and fluoxetine. Time-dependent inhibition of CYP2C8 in HLMs was not observed despite a significant component of reversible inhibition at concentrations that resulted in MBI of recombinant CYP2C8 (data not shown). Hence, concentration was not limiting in the presence of other enzymes or considering the likely nonspecific binding of these drugs in HLMs. Thus, the exact reason(s) why some drugs behave as in vitro MBIs using recombinant systems without the corresponding effect on the human liver microsomal enzyme remains an intriguing question.
In conclusion, we have demonstrated that CYP2C8 is susceptible to MBI by drugs that are common CYP3A4 inhibitors, several of which have been implicated in clinically important drug-drug interactions. Data are consistent with MBI of recombinant CYP2C8 by nortriptyline, verapamil, fluoxetine, amiodarone, and isoniazid, whereas phenelzine acts as a metabolically activated inactivator. Studies with HLMs revealed that nortriptyline, verapamil, and fluoxetine were not MBIs of CYP2C8, although simultaneous inactivation of CYP2C8 and CYP3A4 was described using amiodarone, isoniazid, and phenelzine. Inactivation of human P450 by commonly prescribed drugs may be more widespread than currently believed.
Footnotes
-
This work was supported by a grant from the National Health and Medical Research Council of Australia. T.M.P. is the recipient of an Australian postgraduate award.
-
Part of this work will be presented at the 8th World Congress of Clinical Pharmacology and Therapeutics, Brisbane, Australia, August 1–6, 2004 and at the 7th International ISSX (International Society for the Study of Xenobiotics) Meeting, Vancouver, Canada, August 29–September 3, 2004.
-
doi:10.1124/jpet.104.071803.
-
ABBREVIATIONS: MBI, mechanism-based inactivation; P450, cytochrome(s) P450; MBI, mechanism-based inactivator; MAOI, monoamine oxidase inhibitor; TCA, tricyclic antidepressant; HLM, human liver microsome; MIC, metabolite-intermediate complexation; HPLC, high-performance liquid chromatography.
- Received May 23, 2004.
- Accepted August 6, 2004.
- The American Society for Pharmacology and Experimental Therapeutics