Abstract
A panel of 15 recombinant cytochromes P450 expressed in human B-lymphoblastoid cells was used to study debrisoquine 4-hydroxylation. Both CYP2D6 and CYP1A1 carried out the reaction. The apparentKm (micromolar) andVmax (picomoles per minute per picomole of P450) for CYP2D6 were 12.1 and 18.2 and for CYP1A1 were 23.1 and 15.2, respectively. CYP1A1 debrisoquine 4-hydroxylase was inhibited by the CYP1A1 inhibitor α-naphthoflavone and the CYP1A1 substrate 7-ethoxyresorufin. Additionally and surprisingly, this reaction was also inhibited by quinidine and quinine, with respective IC50 values of 1.38 ± 0.10 and 3.31 ± 0.14 μM, compared with those for CYP2D6 debrisoquine 4-hydroxylase of 0.018 ± 0.05 and 3.75 ± 2.07 μM, respectively. Anti-CYP1A1 monoclonal antibody (mAb) 1-7-1 abolished CYP1A1 debrisoquine hydroxylase and anti-CYP2D6 mAb 50-1-3 eradicated CYP2D6 debrisoquine 4-hydroxylase. Three further CYP2D6-specific reactions were tested: dextromethorphan O-demethylation, bufuralol 1′-hydroxylation, and sparteine dehydrogenation. The CYP2D6 specificity, judged by the CYP2D6/CYP1A1 activity ratios was 18.5, 7.0, 6.0, and 1.6 for dextromethorphan, bufuralol, sparteine, and debrisoquine, respectively. Thus, debrisoquine is not a specific CYP2D6 substrate and quinidine is not a specific CYP2D6 inhibitor. These findings have significant implications for the conduct of in vitro drug metabolism inhibition studies and underscore the fallacy of “specific chemical inhibitors” of a supergene family of enzymes that have overlapping substrate specificities. The use of highly specific mAbs in such studies is mandated. It is unclear as yet whether these findings have implications for the relationship between CYP2D6 genotype and in vivo debrisoquine 4-hydroxylase activity.
Genetic polymorphism in the human cytochrome P450 CYP2D6 was first established using the antihypertensive drug debrisoquine (Mahgoub et al., 1977). Extensive metabolizer (EM) and nonmetabolizer phenotypes, together with the metabolic ratio, were all defined in this original publication. The phrase nonmetabolizer was subsequently changed to the now well known poor metabolizer (PM) (Sloan et al., 1978). In the intervening 25 years since its discovery, there has evolved an expansive scientific literature on this genetic polymorphism that includes the addition of five alternative drug substrates, sparteine (Eichelbaum et al., 1979), bufuralol (Dayer et al., 1982), dextromethorphan (Kupfer et al., 1984), metoprolol (McGourty et al., 1985), and codeine (Yue et al., 1989), for the detection of EMs and PMs. The practice of administration of debrisoquine to human subjects has all but vanished, being replaced by highly accurate phenotype forecasting using molecular genotyping methods (McElroy et al., 2000). However, there are a handful of recent studies that rely on debrisoquine administration to determine CYP2D6 phenotype (Kortunay et al., 1999; Cerqueira et al., 2000; Haffen et al., 2000). Until the development of polymerase chain reaction-based CYP2D6genotyping methodologies (Heim and Meyer, 1990), determination of CYP2D6 phenotype largely relied on the premise that the drugs used to uncover phenotype were specific substrates for CYP2D6. However, it had been observed at the outset (Mahgoub et al., 1977) that the two phenotypes EM and PM harbored a wide range of metabolic capacity, suggesting variables other than the CYP2D6 genotype might be operating to determine an individual's capacity to 4-hydroxylate a test dose of debrisoquine. Daly et al. (1991), making a population-based phenotype-genotype correlation, showed that genotypic EMs could have metabolic ratio values from 0.1 to 8, corresponding to 11 to 91% metabolism of debrisoquine; even heterozygotes alone metabolized from 11 to 77% of the administered debrisoquine. In addition to this 7-fold range of metabolism of debrisoquine within a genotypically homogeneous group, one further anomaly, observed in this and other studies, is the residual metabolism (up to 7% of the dose) of debrisoquine in the PM phenotype. Because the alleles2 that determine the PM phenotype are completely inactive, other unidentified enzymes must contribute this surplus 4-hydroxylation of debrisoquine. This mystery has remained unsolved for over a decade.
As a probe for CYP2D6 phenotype, debrisoquine 4-hydroxylation suffers from another important limitation, and that is that CYP2D6 appears not to display absolute regioselectivity with respect to the debrisoquine substrate, but does exhibit good stereoselectivity, leading to an enantiomeric excess of S-(+)-4-hydroxydebrisoquine (Fig.1, [I]) of at least 90% (Eichelbaum et al., 1988). From the earliest experiments in rat and human, debrisoquine was known to be metabolized to trace amounts of the phenolic metabolites 5-, 6-, 7-, and 8-hydroxydebrisoquine (Fig. 1, [II]) and additionally to significant quantities of ring-opened amino acid metabolites (Fig. 1, [III] and [IV]) (Allen et al., 1975). The phenolic metabolite excretion has been estimated to comprise 0 to 13.7% of the dose (cf. 5.4 to 33.4% 4-hydroxydebrisoquine in the same study) in 0- to 24-h urine in five British volunteers (Idle et al., 1979); and in 17 Ghanaian and 17 British volunteers (Woolhouse et al., 1979), the total phenolic metabolite excretion (0–8 h) was highly correlated with the 4-hydroxydebrisoquine excretion (Spearman rank correlation coefficients Rs, 0.87,P < 0.01, Ghanaian; 0.82, P < 0.01, British). In a study of Nigerian volunteers (Mbanefo et al., 1980), there was a similarly high correlation (Rs = 0.81, P < 0.01) between 4-hydroxylation and phenolic metabolites in 16 volunteers, comprising 10 EMs and six PMs; but, interestingly the phenolic metabolites in PMs (mean 3.2, range 0.8–5.2%) exceeded the 4-hydroxydebrisoquine excretion (mean 2.1, range 0.4–2.9) in this phenotype, although these differences are not statistically significant. Similarly, 1- and 3-hydroxylation of debrisoquine, resulting in the urinary excretion of the ring-opened metabolites 2-(guanidinomethyl)-phenylacetic acid (Fig. 1, [IV]) and 2-(guanidinoethyl)-benzoic acid (Fig. 1, [III]) correlated highly (Rs = 0.97 and 0.96, respectively) with 4-hydroxydebrisoquine excretion in 21 Swedish volunteers (Eiermann et al., 1998).
It might be a reasonable assumption, therefore, that debrisoquine 4-hydroxylation is a good determinant of CYP2D6 activity in vivo, and consequently the CYP2D6 phenotype. Although alternative hydroxylation pathways for debrisoquine have been described, at every available carbon atom in the molecule, these seem to be highly correlated with the principal pathway of 4-hydroxylation and can, to all intents and purposes, be ignored in the in vivo acquisition of CYP2D6 phenotype using debrisoquine. We demonstrate herein that this is a false premise because debrisoquine 4-hydroxylation is not specific to CYP2D6. During the course of a program of investigation of the “CYP2D6 humanized mouse” (Corchero et al., 2001), it emerged that CYP1A1 is also efficient in the 4-hydroxylation of debrisoquine.
Materials and Methods
Chemicals and Reagents.
Debrisoquine hemisulfate, 4-hydroxydebrisoquine, dextromethorphan, and dextrorphan were purchased from ICN (Irvine, CA). NADPH, α-naphthoflavone (α-NF), 7-ethoxyresorufin (7-ER), quinidine, and quinine were obtained from Sigma-Aldrich (St. Louis, MO). (±)-Bufuralol hydrochloride and (±)-1′-hydroxybufuralol maleate were purchased from GENTEST (Woburn, MA). Sparteine was obtained from Fluka Biochemica (Milwaukee, WI) and 2,3- and 5,6-didehydrosparteine were the gift of Dr. Ulrich Zanger (Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany). All other chemicals and high-performance liquid chromatography grade solvents were of the highest grade of purity commercially available. Microsomes prepared from human B-lymphoblastoid cell lines expressing human P450s CYP1A1, CYP1A2, CYP1B1, CYP2A6, CYP2B6, CYP2C8, CYP2C9, CYP2C18, CYP2C19, CYP2D6, CYP2E1, CYP3A4, CYP3A5, CYP4A11, and control microsomes from human B-lymphoblastoid cell lines containing the expression vector without cDNA were obtained from GENTEST. Monoclonal antibodies raised against human CYP1A1 (mAb 1-7-1) and CYP2D6 (mAb 50-1-3) enzymes were characterized previously (Friedman et al., 1985; Gelboin et al., 1997).
Metabolism of Debrisoquine with Recombinant P450s.
The incubation medium contained P450 (20 pmol), debrisoquine (100 μM) in 0.1 M sodium phosphate buffer, pH 7.4, with a final volume of 200 μl. The mixture was preincubated for 5 min in a shaking water bath at 37°C and the reactions were initiated by the addition of 20 μl of 12 mM NADPH. Control experiments were carried out in parallel using microsomes isolated from the same cell line, containing the vector but without a cDNA insert. Reactions were stopped after 20 min by the addition of 50 μl of 1.0 M NaOH and subsequently cooling on ice for 15 min. All incubations were performed in duplicate and in the linear range with respect to microsomal protein and incubation time.
Kinetic Studies.
Kinetic experiments were performed with microsomes prepared from baculovirus-infected insect cells expressing CYP1A1 and CYP2D6. Debrisoquine 4-hydroxylase activities were determined with debrisoquine concentrations ranging from 0.5 to 500 μM. The incubations were carried out as described above. Each data point represents the average of duplicate determinations. The kinetic parameters were estimated as described below.
Chemical Inhibition Studies.
The effects were studied of coincubation with a CYP1A1 inhibitor and a substrate on the microsomal 4-hydroxylation of debrisoquine. Either 7-ER (0.1, 1.0, 10, 25, 50, and 100 μM; CYP1A1 substrate) or α-NF (0.1, 1.0, 10, and 100 μM; CYP1A1/1A2 inhibitor) was added to the CYP1A1-expressed microsomes. The mixture was preincubated for 5 min and the reaction was initiated by the addition of 50 μM debrisoquine and 1 mM NADPH. The reaction was carried out for 20 min and was terminated as described above. The effects of 7-ER and α-NF on the 4-hydroxylation of 50 μM debrisoquine were compared with the control values determined by the incubation of debrisoquine alone and expressed as percentage of the respective control value.
Immunoinhibition Studies.
Immunoinhibition of 4-hydroxydebrisoquine production was examined by preincubating of expressed CYP1A1 or CYP2D6 (25 pmol) with ascites fluid after the addition of various concentrations of the monoclonal antibodies against CYP2D6 and CYP1A1 in 100 μl of 100 mM sodium phosphate buffer, pH 7.4. The mixture was preincubated for 5 min at 37°C and the reaction was initiated by the addition of 50 μM debrisoquine and 1 mM NADPH, diluted with buffer to a final volume of 200 μl. Anti-lysozyme mAb (HyHEL) was used as a control for nonspecific binding. Reactions were incubated for 30 min and were terminated as described above.
Quantitation of Drug Metabolites by Tandem Mass Spectrometry.
Microsomal production of 4-hydroxydebrisoquine was determined using an LC/MS/MS method described previously (Corchero et al., 2001) with a minor modification of liquid-liquid extraction (Pereira et al., 2000) rather than solid-phase extraction (Scott et al., 1999). Briefly, to the total microsomal incubations were added 20 μl of internal standard solution (phenacetin, 5 μg/ml in methanol), 500 μl of propan-2-ol, 50 μl of 0.4 M sodium hydroxide, and 3 ml of methyl tert-butyl ether. The mixture was vortex-mixed for 1 min and the phases separated by centrifugation at 1000g for 10 min. The aqueous layer was frozen in dry-ice and the organic phase was transferred to a fresh borosilicate tube and evaporated to dryness under a gentle stream of air on a heating block at 30°C. The residue was reconstituted in acetonitrile/water (20:80 v/v; 100 μl), transferred to polypropylene autosampler vials, and 10 to 25 μl of sample was injected into the LC/MS/MS system.
The high-performance liquid chromatography system consisted of a PerkinElmer Series 200 quaternary pump, a vacuum degasser, and a Series 200 autosampler with a 100-μl loop (PerkinElmer Instruments, Norwalk, CT) both interfaced to a triple-quadrupole tandem mass spectrometer (API 2000; PerkinElmer Sciex, Thornhill, Toronto, ON, Canada). 4-Hydroxydebrisoquine and the internal standard were separated on a LUNA 3-μm C18(2) column (50 × 2.0 mm; Phenomenex, Torrance, CA). The mobile phases comprised the following: solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). A linear gradient of solvent B from 5 to 95% over 4 min was applied on the column and was returned to the original condition and equilibrated for 1 min before the next injection. The samples were delivered with a flow rate of 0.20 ml/min and the run time was 5.0 min.
The mass spectrometer was operated in the turbo ion spray mode with positive ion detection. The turbo ion spray temperature was maintained at 300°C and a voltage of 4.8 kV was applied to the sprayer needle. Nitrogen was used as the turbo ion spray and nebulizing gas. The detection and quantitation of compounds were performed using MS/MS in the multiple reaction monitoring mode. Multiple reaction monitoring data acquisition was used monitoring the transitionsm/z 192→132 for 4-hydroxydebrisoquine andm/z 180→110 for the internal standard phenacetin. All raw data were processed with Sciex Analyst software, version 1.2 (PerkinElmer Sciex).
The method was linear for 4-hydroxydebrisoquine concentrations from 1 to 5000 pmol. Calibration curves were constructed in duplicate at each concentration and were computed using a 1/X2 weighting. Good linearity was achieved with correlation coefficients greater than 0.995. The lower limit of quantitation was 1.0 pmol for 4-hydroxydebrisoquine, where the coefficient of variation was less than 20%. The recoveries for 4-hydroxydebrisoquine were greater than 80%. Intraday and interday coefficients of variation were less than 10% at a concentration of 50 pmol of 4-hydroxydebrisoquine.
The LC/MS/MS assays on the metabolism of (±)-bufuralol (Needham and Brown, 2000), dextromethorphan (Bu et al., 2000), and their metabolites were performed according to published methods. Transitions monitored were m/z 258→157 (dextrorphan) andm/z 278→186 (1′-hydroxybufuralol). Sparteine metabolites formed by CYP2D6 and CYP1A1 were identified by LC/MS. The detection and quantification of the 2,3- and 5,6-didehydrosparteine metabolites were accomplished by selected ion monitoring with the protonated molecular ion m/z 233. Sample preparations and chromatography were carried out as described above for debrisoquine. Peak area measurements were used for quantitation and were compared with standards of dehydrosparteine metabolites dextrorphan and 1′-hydroxybufuralol prepared in inactive microsomes.
Determination of Apparent Km andVmax.
The formation of 4-hydroxydebrisoquine was calculated and expressed as picomoles per minute per picomole of P450. Kinetic parameters (Km andVmax) were then determined by nonlinear regression using GraphPad Prism version 3.02 (GraphPad Software, San Diego, CA). All analysis was performed on the mean values of duplicate incubations.
Results
Debrisoquine 4-Hydroxylase Activity of Recombinant Cytochromes P450.
In the control CYP2D6 and CYP1A1 microsomes, the formation of 4-hydroxydebrisoquine was linear at 37°C for incubation times up to 40 min and for microsomal protein concentrations up to 50 pmol and substrate concentrations of 50 μM. An incubation time of 20 min and a concentration of 25 pmol of microsomal protein were therefore used to ensure observation of initial velocities.
Debrisoquine 4-hydroxylase activity was determined at 100 μM debrisoquine in microsomes from 15 human B-lymphoblastoid cell lines expressing various individual P450 isozymes (Fig.2). As expected, CYP2D6 showed a high relative activity of 25 pmol min−1pmol−1 P450. All the other P450s showed low or trivial activities with the exception of CYP1A1, which, to our surprise, had a debrisoquine 4-hydroxylase activity of 21 pmol min−1 pmol−1 P450.
Kinetics of Debrisoquine 4-Hydroxylase with Recombinant Cytochromes P450.
Kinetic analyses were then performed with recombinant CYP1A1 and CYP2D6 (Fig. 3). Kinetic parameters are presented in Table 1. TheKm of CYP2D6 and CYP1A1 for debrisoquine 4-hydroxylation was 12.1 and 23.1 μM, respectively, and the corresponding Vmax values were 18.2 and 15.2 pmol min−1pmol−1 P450. Accordingly, the calculated intrinsic clearance of debrisoquine for 4-hydroxylation by CYP2D6 was only about twice that of CYP1A1 (1.5 versus 0.7; Table 1).
Inhibition of Debrisoquine 4-Hydroxylase by α-Naphthoflavone, 7-Ethoxyresorufin, Anti-CYP2D6, and Anti-CYP1A1 Monoclonal Antibodies.
The effects of chemical inhibitors and anti-P450 antibodies on debrisoquine 4-hydroxylation were investigated using microsomes from human B-lymphoblastoid cell lines expressing CYP1A1 and CYP2D6. α-NF (Murray and Reidy, 1990; Rodrigues and Prough, 1991), an inhibitor of CYP1A1/1A2 at 100 μM, inhibited the 4-hydroxylation of 50 μM debrisoquine by 90% in microsomes expressing CYP1A1 (Fig.4A). 7-ER (Pelkonen et al., 1986;Rodrigues and Prough, 1991), a putative substrate for CYP1A1, inhibited the formation of 4-hydroxydebrisoquine in a concentration-dependent manner in microsomes expressing CYP1A1 (Fig. 4B). The monoclonal antibody against CYP1A1 (mAb 1-7-1) also inhibited the formation of 4-hydroxydebrisoquine by more than 95% (Fig.5). As expected, the anti-CYP2D6 monoclonal antibody (mAb 50-1-3) inhibited debrisoquine 4-hydroxylation, at 50 μM debrisoquine, in microsomes expressing CYP2D6, by more than 98% (Fig. 5).
Inhibition of CYP2D6- and CYP1A1-Mediated 4-Hydroxylation of Debrisoquine by Quinine and Quinidine.
Quinidine has long been used as the cardinal specific and high-affinity inhibitor of CYP2D6.von Bahr et al. (1985) first showed that quinidine was an unusually potent inhibitor (Ki = 0.27 μM) of debrisoquine hydroxylase in vitro, using desmethylimipramine 2-hydroxylation as the probe; the isomeric quinine had aKi of 12 μM. Complete inhibition of an oxidative reaction by quinidine, with aKi in the submicromolar range, has generally been taken as evidence that the reaction was mediated by CYP2D6 (Speirs et al., 1986; Otton et al., 1988; Branch et al., 2000;Dalen et al., 2000). Addition of either 0.01 to 10 μM quinidine or 0.1 to 100 μM quinine strongly inhibited the 4-hydroxylation of debrisoquine by CYP2D6, as anticipated, but also, unexpectedly, debrisoquine 4-hydroxylation by CYP1A1 (Fig.6). The IC50 values (μM) determined for CYP2D6 for quinidine and quinine were 0.018 ± 0.05 and 3.75 ± 2.07, respectively, and for CYP1A1 with quinidine and quinine were 1.38 ± 0.10 and 3.31 ± 0.14, respectively.
Metabolism of Dextromethorphan, Bufuralol, and Sparteine by CYP2D6 and CYP1A1.
Recombinant P450s were used to determine whether other CYP2D6 substrates such as 10 μM dextromethorphan, 10 μM bufuralol, and 10 μM sparteine were metabolized by CYP1A1. As expected, CYP2D6-expressing cells exhibited the high-level activity for these four substrates (Fig. 7). In contrast, cells expressing the recombinant CYP1A1 exhibited the highest levels of activity for debrisoquine, compared with dextromethorphan, bufuralol, or sparteine. Activity ratios (CYP2D6 activity/CYP1A1 activity), which give an idea of the specificity toward CYP2D6, relative to CYP1A1, of the various substrates, were 1.6 for debrisoquine, 18.5 for dextromethorphan, 7.0 for bufuralol, and 6.0 for sparteine, suggesting that dextromethorphan is the one substrate with the greatest specificity for CYP2D6 relative to CYP1A1.
Discussion
We have demonstrated that microsomes from human lymphoblastoid cells expressing CYP1A1 are efficient in the 4-hydroxylation of debrisoquine, with an intrinsic clearance of debrisoquine ∼50% that of similar microsomes expressing CYP2D6, the so-called “debrisoquine hydroxylase” (Kahn et al., 1982) (EC 1.14.99; OMIM entry 124030 http://www.ncbi. nlm.nih.gov/htbin-post/Omim/dispmim?124030). TheVmax for CYP1A1 was similar to that for CYP2D6 (15.2 and 18.2 pmol min−1pmol−1 P450, respectively), but the CYP1A1 debrisoquine 4-hydroxylase Km was almost double that of the CYP2D6 debrisoquine 4-hydroxylase (23.1 and 12.1 μM, respectively). CYP1A1 debrisoquine 4-hydroxylase activity was almost entirely inhibited by NF and ER (Fig. 4) with IC50 values of 20 and 7 μM, respectively. Furthermore, CYP1A1 debrisoquine hydroxylase was abolished by the specific anti-CYP1A1 monoclonal antibody mAb 1-7-1 (Fujino et al., 1984). Collectively, these findings demonstrate that CYP1A1, whose constitutive catalytic activity is generally associated with water-insoluble nonpolar substrates such as polycyclic aromatic hydrocarbons and various substituted coumarins (Penman et al., 1994), metabolizes the highly basic and water-soluble substrate debrisoquine, with a kinetic profile that approaches that of CYP2D6. The extent of both regioselectivity and stereoselectivity displayed by CYP1A1 with respect to debrisoquine hydroxylation is not known at present.
In addition, we have demonstrated that the apparently specific in vitro and in vivo inhibitor of CYP2D6, quinidine (Kobayashi et al., 1989;Branch et al., 2000), also strongly inhibits the 4-hydroxylation of debrisoquine by CYP1A1 with an IC50 of 1.38 μM (cf. 0.018 μM for CYP2D6). Its diastereomer quinine, known also to be an in vitro CYP2D6 inhibitor (Kobayashi et al., 1989) with little activity in vivo (Ayesh et al., 1991), also inhibited CYP1A1 debrisoquine 4-hydroxylase with an IC50 of 3.31 μM (cf. 3.75 μM for CYP2D6). Although the quinidine IC50 is some 80-fold lower for CYP2D6 than for CYP1A1, the quinine IC50 values are comparable for the two isozymes. Recently, Ching et al. (2001) reported that quinidine and quinine inhibited 7-ethoxyresorufinO-deethylation mediated by CYP1A1 expressed in yeast, with respective “[I]0.5 ” values of 1.0 and 1.1 μM, which compares well with the data reported herein for CYP1A1 debrisoquine 4-hydroxylase.
The finding that CYP1A1 4-hydroxylates debrisoquine is a surprising one. Apart from the minimal homology between CYP1A1 and CYP2D6, there are many related CYP2D isozymes with little or no debrisoquine 4-hydroxylase activity, such as mouse isoforms cyp2d9–11, cyp2d22, and cyp2d26 (Masubuchi et al., 1997), and the rat cyp2d1 (Wan et al., 1997;Schulz-Utermoehl et al., 1999). CYP2D6 is an enzyme that typically metabolizes relatively small nitrogenous compounds (Wolf and Smith, 1999; Miller et al., 2001) that are usually water-soluble. It has been proposed that the substrate specificity and enzyme activity are determined by the binding of a basic nitrogen atom in the substrate to the carboxyl group in the CYP2D6 Asp-301 residue (Ellis et al., 1995;Hanna et al., 2001). In contrast, what little is known about substrate binding to the CYP1A1 active site suggests a role for the Val-382 residue, at least with regard to the O-dealkylation of alkoxyresorufins2. In stark contrast to CYP2D6, typical CYP1A1 substrates are not basic, are lipophilic, and are often large planar molecules, such as the polycyclic aromatic hydrocarbons. Only the most flimsy of clues suggest a relationship between these two disparate P450 isozymes. First, mexiletine, a substrate of CYP2D6 (Vandamme et al., 1993; Senda et al., 2001) is a preferential inhibitor of CYP1A isozymes (Konishi et al., 1999). Second, CYP1A1 is efficient in steroid 16α-hydroxylation (Schwarz et al., 2000), a reaction mediated in the mouse by cyp2d9. Consideration of the metabolism of the 7-alkoxycoumarins shows the nonoverlapping substrate preferences of CYP1A1 and CYP2D6; CYP1A1 metabolizes the neutral lipid-soluble 7-ethoxy-coumarin and 7-ethoxy-4-trifluoromethyl-coumarin (Penman et al., 1994), whereas CYP2D6 metabolizes a series of 4-aminomethyl-7-alkoxycoumarins (Nakamura et al., 2001) that are both basic and more water-soluble. The major finding in this article would not, therefore, have been predicted from first principles.
The dislocation between CYP2D6 and its apparent specific inhibitor quinidine had a slightly greater expectation. Porcine CYP2D25, the microsomal vitamin D3 25-hydroxylase, shares only a 77% homology with human CYP2D6, but nevertheless is able to hydroxylate the CYP2D6 substrate tolterodine (Hosseinpour and Wikvall, 2000). This reaction is not inhibited by quinidine. Moreover, the deamination of amphetamine and benzphetamine, and theN-dealkylation of benzphetamine are all mediated in rabbit liver by CYP2C3, and these reactions are potently inhibited by quinidine (Shiiyama et al., 1997). The very low IC50 values for the inhibition of both CYP1A1 and CYP2D6 debrisoquine 4-hydroxylase activities, reported herein, diverge from the affinities of both quinidine and quinine for the enzyme that actually metabolizes them both, CYP3A4. Quinidine undergoes both stereoselective (3S)-3-hydroxylation andN-oxidation by CYP3A4, withKm values of 74.2 and 76.1 μM, respectively (Nielsen et al., 1999). Similarly, the mean apparentKm for quinine 3-hydroxylation by CYP3A isozymes in 10 humans livers was 83 μM (Zhang et al., 1997). The higher affinity by over 2 orders of magnitude of quinine and quinidine for CYP1A1 and quinine for CYP2D6, and over 4 orders of magnitude of quinidine for CYP2D6 remains a riddle yet to be solved.
One is bound to ask why the 4-hydroxylation of debrisoquine by CYP1A1 has gone unreported for so long. The answer may lie in the similarity in the kinetics of the 4-hydroxylation reactions carried out by the two isozymes and the fact that both are inhibited by quinidine, albeit with different potencies. Thus, a meager contamination of CYP2D6 4-hydroxylation by CYP1A1 hydroxylation should go undetected, unless highly specific tools were used, based on structure and not activity, such as mAbs. Previous studies reported, for example, that debrisoquine 4-hydroxylation by human liver microsomes was competitively inhibited by sparteine (Nakano and Inaba, 1984) and sparteine dehydrogenation was competitively inhibited by debrisoquine (Otton et al., 1982). However, this can be completely rationalized by the fact that both drugs are substrates for CYP2D6 and CYP1A1 (Fig. 7). Initial studies on the nature of debrisoquine 4-hydroxylase reported that polyclonal antibodies raised against the purified protein failed to completely inhibit the reaction in human hepatic microsomes (Wolff et al., 1985). This, too, may be rationalized by the occurrence of two distinct enzyme activities. Furthermore, attempts to phenotype human liver samples in vitro using debrisoquine 4-hydroxylase “met with limited success” (Boobis et al., 1985). In relation to these aforementioned findings, there are reports that the metabolism of archetypal CYP2D6 substrates [(+)- and (−)-bufuralol and dextromethorphan, but, interestingly, not debrisoquine] in human liver microsomes is biphasic (Kronbach et al., 1987). Moreover, the apparent CYP2D6-specific reactions of dextromethorphan O-demethylation and (±)-bufuralol 1′-hydroxylation are both also mediated by non-CYP2D6 P450 isozymes, CYP2C9 and CYP2C19 for dextromethorphan (von Moltke et al., 1998), and CYP2C19 for bufuralol (Mankowski, 1999). Interestingly, in neither of these studies was CYP1A1 considered as a candidate. All of this notwithstanding, the paucity of expression of CYP1A1 in human liver would render difficult the detection of an in vivo contribution from this isozyme to debrisoquine 4-hydroxylation.
We had originally observed, in a study of the CYP2D6 humanized mouse, that wild-type mice orally administered 2.5 mg kg−1 debrisoquine excreted 61.0 ± 9.0% dose as unchanged debrisoquine in 0- to 24-h urine and 6.2 ± 3.1% as 4-hydroxydebrisoquine (Corchero et al., 2001). Because the mouse cyp2d isozymes do not 4-hydroxylate debrisoquine (Masubuchi et al., 1997), it is possible that the residual 4-hydroxylation in the wild-type mouse comes from cyp1a1 because cyp1a1 expression (Oladapo and Forkert, 1995) and activity (Delker et al., 1996) can be detected in mouse liver.
Does this same phenomenon occur in humans and to what extent might it confound both in vivo CYP2D6 phenotyping and therefore associations between phenotype and clinical observations, such as disease occurrence or adverse drug reactions? CYP1A1 has been described as being absent from human liver (Murray et al., 1993). Other reports cite a 20-fold intersubject variation in CYP1A1 hepatic expression (Schweikl et al., 1993). CYP1A1 has been found in human small intestine (Vang et al., 1999; Zhang et al., 1999), occasionally with levels of ethoxyresorufinO-deethylase activity that exceed the liver (Paine et al., 1999). Native expression of CYP1A1 has been detected in human lung (Mace et al., 1998; Wei et al., 2001), in bronchial epithelial cells (Willey et al., 1996, 1997), and in human placenta (Hakkola et al., 1996). It is possible, therefore, that although first-pass metabolism of debrisoquine and other “CYP2D6-specific substrates” is mediated principally by hepatic CYP2D6 with some contribution from enteric CYP1A1, the systemic metabolism of the drug may be subject to significant 4-hydroxylation by CYP1A1, both affecting the urinary metabolic ratio and contributing significantly to debrisoquine 4-hydroxylation in CYP2D6 genotypic PMs. These speculations, however, await experimental confirmation.
There is a significant body of literature in which metabolic transformations have been assigned to CYP2D6 based upon correlations with “known” CYP2D6 substrates and inhibition of the reaction with quinidine. Our findings demonstrate that this practice is unsound and should be abandoned in favor of the use of a panel of recombinant cytochromes P450 and the use of inhibitory monoclonal antibodies that, unlike chemical inhibitors, show much less cross-reactivity with other isozymes.
We propose that CYP1A1 debrisoquine 4-hydroxylase and its inhibition by quinidine are potentially more important for the interpretation of in vitro metabolic studies than for the relationship between CYP2D6 genotype and extent of debrisoquine 4-hydroxylation in vivo. This additional variable in debrisoquine metabolism, however, is a potential confounder in the association studies between debrisoquine metabolic phenotype and disease, for example, lung cancer, and might explain, in part, the weak phenotypic association (Rostami-Hodjegan et al., 1998) in the absence of a genotypic association (Laforest et al., 2000).
Footnotes
-
↵1 Current address: Zlatá 34, 36005 Karlovy Vary, Czech Republic (on leave of absence).
-
↵2 CYP2D6*3A, *4A, *4B, *4C, *4D, *4K, *5, *6A, *6B, *6C, * 7, *8, *11, *12, * 13, *14, *16, *18, *19, and *20 (http://www.imm.ki.se/CYPalleles/cyp2d6.htm).
-
J.R.I. received a travel grant from U.S. Smokeless Tobacco Company.
- Abbreviations:
- P450
- cytochrome P450
- EM
- extensive metabolizer
- PM
- poor metabolizer
- NF
- α-naphthoflavone
- 7-ER
- 7-ethoxyresorufin
- mAb
- monoclonal antibody: LC/MS/MS, liquid chromatography/tandem mass spectrometry
- Received January 25, 2002.
- Accepted March 1, 2002.
- U.S. Government