CHARACTERIZATION OF CYTOCHROME P450 2D6.1 (CYP2D6.1), CYP2D6.2, ANDCYP2D6.17 ACTIVITIES TOWARD MODEL CYP2D6 SUBSTRATES

DEXTROMETHORPHAN, BUFURALOL, AND DEBRISOQUINE

KENDA A. MARCUCCI, ROBIN E. PEARCE, CHARLES CRESPI, DOROTHY T. STEIMEL, J. STEVEN LEEDER, AND

ANDREA GAEDIGK

Section of Developmental Pharmacology and Experimental Therapeutics, Division of Pediatric Clinical Pharmacology and Toxicology,Children’s Mercy Hospital and Clinics, Kansas City, Missouri (K.A.M., R.E.P., J.S.L., A.G.); and Gentest Corporation, Woburn, Massachussetts

(C.C., D.T.S.)

(Received December 10, 2001; accepted February 11, 2002)

This article is available online at http://dmd.aspetjournals.org

ABSTRACT:

Over 50 allelic variants of cytochrome P450 2D6 (CYP2D6) encod-ing fully functional, reduced-activity, or nonfunctional proteinshave been described. Compared with Caucasians, studies in blackpopulations demonstrate a tendency toward slower CYP2D6 ac-tivity, attributed in part to the presence of a variant allele associ-ated with reduced activity, the CYP2D6*17 allele. To investigate thekinetic characteristics of this variant protein, expression con-structs coding for CYP2D6.1, CYP2D6.2, and CYP2D6.17 geneproducts were prepared and transfected into mammalian COS-7and insect (Trichoplusia ni) cells for expression. Microsomal frac-tions containing the expressed proteins were used to determinethe kinetic parameters Km, Vmax, and intrinsic clearance (Clint) forthe model substrates dextromethorphan, bufuralol, and debriso-quine. Relative to the wild-type CYP2D6.1 protein expressed in

COS-7 cells, CYP2D6.17 exhibited a 2-fold higher Km and a 50%reduction in Vmax using dextromethorphan as the substrate. Incontrast, no appreciable change in bufuralol Km was observed withCYP2D6.17 whereas Vmax was decreased by 50%. When expressedin the baculovirus expression system, CYP2D6.17 exhibited a6-fold increase in Km but no change in Vmax with dextromethorphanas the substrate, a 2-fold higher Km and 50% reduction in Vmax withbufuralol, and a 3-fold increase in Km and no change in Vmax withdebrisoquine relative to CYP2D6.1. These data indicate thatCYP2D6.17 exhibits reduced metabolic activity toward all threecommonly used CYP2D6 substrates, although specific effects onsubstrate affinity and turnover demonstrate some substrate de-pendence.

Due to its prominent role in the biotransformation of medicationsused to treat depression, hypertension, cardiac arrhythmias, and pain,among other important conditions, interindividual variability inCYP2D6 activity within and across different ethnic groups has beenthe focus of several in vivo investigations. In Caucasian populations,CYP2D6 phenotyping studies have revealed a bimodal distribution,with a well defined antimode separating extensive and poor metabo-lizer individuals (Marez et al., 1997; Sachse et al., 1997; Gaedigk etal., 1999). By analysis of CYP2D6 genotype, a poor metabolizerphenotype can be reliably assigned based on the finding of twononfunctional or “null” alleles and can be confirmed in affectedindividuals using currently available CYP2D6 phenotyping probedrugs, such as dextromethorphan, sparteine, or debrisoquine (Marez etal., 1997; Sachse et al., 1997; Gaedigk et al., 1999). In contrast,population studies in Africans have generally observed CYP2D6activity to be unimodally distributed with a tendency toward higher

urinary metabolite ratios (lower enzyme activity) compared with thatobserved in Caucasian populations (Woolhouse et al., 1985; Ma-simirembwa et al., 1996a; Wennerholm et al., 1999), a finding that isnow attributed to a relatively high frequency (17–30%) of theCYP2D6*171 allele (Masimirembwa et al., 1996b). Furthermore,comparative phenotyping studies conducted in Ghanaian (Woolhouseet al., 1985) and Nigerian (Lennard et al., 1992) healthy volunteersprovided evidence that CYP2D6 phenotyping probes, particularlysparteine and debrisoquine, are not coordinately regulated in thesestudied populations. A lack of coordinate regulation has also beenreported between debrisoquine and metoprolol in Venda (Sommers etal., 1989), Zambian (Simooya et al., 1993), and Ghanaian (Ma-simirembwa et al., 1996a) subjects.

The issue of discordant phenotyping results with standard CYP2D6phenotyping probes was pursued in more detail by Droll et al. (1998)who conducted a comparative genotype/phenotype study in Ghanaian,Chinese, and Caucasian volunteers using sparteine, debrisoquine, anddextromethorphan as phenotyping probes. For Caucasian and Chinesesubjects, statistically significant correlations (rs values between 0.76

This study was supported by the Children’s Mercy Hospital Research VisionCore Lab Project Grant G/L 01.4878.

Address correspondence to: J. Steven Leeder, Chief, Section of Develop-mental Pharmacology and Experimental Therapeutics, Children’s Mercy Hospitaland Clinics, 2401 Gillham Road, Kansas City, MO 64108. E-mail:sleeder@cmh.edu

1 CYP2D6 allele nomenclature and nucleotide numbering system according tothe CYP Allele Nomenclature Committee (http://www.imm.ki.se/CYPalleles/). Al-leles are designated by a � (e.g., CYP2D6*1) and gene products (i.e., expressedproteins) are designated by a “.” (e.g., CYP2D6.1).

0090-9556/02/3005-595–601$7.00DRUG METABOLISM AND DISPOSITION Vol. 30, No. 5Copyright © 2002 by The American Society for Pharmacology and Experimental Therapeutics 649/980250DMD 30:595–601, 2002 Printed in U.S.A.

595

and 0.93) were observed between each pair of phenotyping probes. Incontrast to previous studies (Woolhouse et al., 1985; Lennard et al.,1992), sparteine and debrisoquine metabolic ratios were significantlycorrelated (rs � 0.72, p � 0.002) in the Ghanaian volunteers (n � 21).However, the correlations between dextromethorphan and either de-brisoquine (rs � 0.52) or sparteine (rs � 0.38) did not achievestatistical significance (p � 0.05). The relatively poor correlationsamong phenotyping probes in the Ghanaian subjects were attributed tothe presence of the CYP2D6*17 allele and possibly additional novelallelic variants of CYP2D6.

The CYP2D6*17 allele carries three nonsynonymous coding regionsingle nucleotide polymorphisms conferring T107I, R296C, andS486T amino acid substitutions relative to CYP2D6*1 (Masimirem-bwa et al., 1996b). Since the latter two coding region single nucleotidepolymorphisms are also found on the CYP2D6*2 allele, the T107Isubstitution has been targeted as the most likely source of alteredsubstrate specificity/catalytic activity. Using COS-7 and yeast expres-sion systems, introduction of the T107I substitution alone did notappreciably alter enzyme activity using bufuralol as a substrate,whereas the apparent Km for codeine was increased approximately5-fold (Oscarson et al., 1997). Studies with combinations of aminoacid substitutions revealed that both the T107I and R296C substitu-tions were required to alter bufuralol hydroxylase activity (5-foldincrease in Km). In the presence of all three amino acid substitutions,apparent Km for both bufuralol and codeine was 5- to 10-fold higherthan that observed for the CYP2D6.1 protein. Data from structuralmodels and sequence alignment data indicate that Thr-107 and Arg-296 lie within distinct substrate contacting regions of the CYP2D6active site (Hasemann et al., 1995; Modi et al., 1996). Consistent withthe in vitro data of Oscarson et al. (1997), Thr-107 was found to be asubstrate contacting residue in 11 of 13 models of the CYP2D6 activesite developed using homology modeling and NMR studies of sub-strate binding (Modi et al., 1996). Nevertheless, available populationphenotyping studies and in vitro data suggest that the biotransforma-tion of CYP2D6 substrates may be differentially affected by allelicvariants present in individuals of African origin.

The purpose of this study was to resolve issues related to decreasedactivity and discordance among CYP2D6 phenotyping probes withinpopulations of black African origin by comparing the catalytic prop-erties of CYP2D6.17 with those of CYP2D6.1 and CYP2D6.2 towardsubstrates commonly used as in vivo phenotyping probes in NorthAmerica, dextromethorphan and debrisoquine, and the in vitro sub-strate, bufuralol.

Materials and Methods

Chemicals and Reagents. Bufuralol, 1�-hydroxybufuralol, 4-hydroxyde-brisoquine, and dextrorphan were purchased from Ultrafine Chemicals(Manchester, UK). Dextromethorphan, debrisoquine, glucose 6-phosphate,glucose-6-phosphate dehydrogenase, NADP, and EDTA were acquired fromSigma-Aldrich (St. Louis, MO). Hydroxybufuralol was obtained from Sigma/RBI (Natick, MA). Fetal bovine serum (FBS) was purchased from HycloneLaboratories (Logan, UT). TransfectAMINE Plus reagent was purchased fromInvitrogen (Carlsbad, CA). DMEM2 and COS-7 cells were obtained fromAmerican Type Culture Collection (Manassas, VA). Oligonucleotides werepurchased from MWG-Biotech AG (Greensboro, NC). SacI and BsrGI wereacquired from New England Biolabs (Beverly, MA). The pcDNA3.1/Hygro(�/�) vector was obtained from Invitrogen. Quantum Prep plasmid miniprepkits were purchased from Bio-Rad (Hercules, CA). Assay plates (96-well)were purchased from Nalge Nunc (Naperville, IL). NADPH-dependent cyto-chrome P450 reductase and baculovirus-expressed CYP2D6*1 were purchased

from Gentest Corp. (Woburn, MA). All other reagents were of analyticalgrade.

Construction of Expression Plasmids. A pUC19 plasmid containing theCYP2D6*1 cDNA with a 65-bp long 5� “leader sequence” optimized formammalian expression was provided by Gentest Corp. Site-directed mutagen-esis to introduce the C3 T transition at position 2850 was performed with theQuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) accordingto the manufacturer’s directions. Oligonucleotides 30-bp long with the desiredmismatches located in the center of the sequence were used without gelpurification. For restriction fragment exchange, a 645-bp fragment containingC at position 4180 was amplified using primers 5�4213 and 3�4213/4268(Gaedigk et al., 1999) from a genomic DNA sample genotyped asCYP2D6*2/*2 (Gaedigk et al., 1999) and digested with SacI and BsrGI to yielda 183-bp fragment. This fragment was exchanged with the correspondingfragment released from the intermediate 2850C � T construct. Similarly, a530-bp fragment carrying T at position 1023 was generated from a genomicDNA sample genotyped as CYP2D6*17/*17 and digested with PflMI andEagI. The resulting 73-bp fragment was exchanged with the fragment releasedby PflMI and EagI digestion of the CYP2D6*2 construct. Clones containingthe desired nucleotide changes were identified by genotyping as described indetail elsewhere (Gaedigk et al., 1999). To confirm the identity of theCYP2D6*17 cDNA, the construct was completely sequenced on a VisibleGenetics sequencing system (Visible Genetics Inc., Toronto, Canada) usingCy5 and Cy5.5 dye terminator chemistry (Amersham Biosciences, Piscataway,NJ).

For expression in COS-7 cells, cDNA and “leader sequence” were releasedfrom pUC*2 and pUC*17 with XbaI, and the resulting fragments cloned intothe XbaI site of the pcDNA3.1/Hygro (�/�) vector. Subsequently, plasmidswere purified for transfection experiments with the Quantum Prep plasmidminiprep kit, followed by an ethanol precipitation. DNA concentration andquality were evaluated by spectral photometry and agarose gel analysis.

Transfection of COS-7 Cells and Preparation of Microsomes. COS-7cells were plated onto 100-mm dishes (typically 10 dishes per construct) inDMEM supplemented with 10% FBS, 200 U/ml penicillin, and 200 �g/mlstreptomycin. With the cells �50% confluent, the media was changed toDMEM without penicillin/streptomycin 12 to 24 h prior to transfection.Plasmids were transfected into the COS cells using TransfectAMINE Plus(Invitrogen) according to the manufacturer’s guidelines. Optimal cell viabilityand transfection efficiency were achieved with 6 �g of DNA/dish, 15 �l ofTransfectAMINE Plus reagent/dish, and 20 �l of lipid/dish. Cells transfectedwith vector alone served as a control. Twenty-four hours after transfection,medium was replaced with DMEM containing FBS supplemented with peni-cillin/streptomycin. Forty-eight hours after transfection, cells were washedonce and harvested by scraping in homogenization buffer containing 50 mMTris-HCl, 150 mM KCl, and 2 mM EDTA. After sonication, microsomes wereprepared by differential centrifugation at 9,000g for 20 min and centrifugationof the resulting supernatant at 100,000g for 60 min. The microsomal pellet wasresuspended in 250 mM sucrose and stored at �70°C. The CYP2D6*1,CYP2D6*2, and CYP2D6*17 constructs were transfected on three separateoccasions into COS-7 cells passaged in succession.

Quantitation of Expressed CYP2D6 Variant Proteins. Total proteincontent of microsomes prepared from COS-7 cells was measured using theBCA protein assay kit (Pierce, Rockford, IL). Microsomal CYP2D6 apoproteincontent was determined by immunoblot analysis. Microsomal protein (5 �g/lane) was resolved on 4 to 12% Novex polyacrylamide gel electrophoresisgradient gels (Invitrogen) in MOPS buffer and transferred to Hybond-Cnitrocellulose (Amersham Biosciences) using a semidry transfer apparatus(Invitrogen). CYP2D6 protein was probed with an anti-CYP2D6 antibodyraised against a multivalent antigenic peptide (MAP) antigen prepared from theCYP2D6-derived peptide NH2-DPAQPPRDLTEAFLA-COOH coupled to thepolylysine core using carbodiimide chemistry. MAP antigen (2 mg) wasdissolved in 1 ml of phosphate-buffered saline and emulsified with an equalvolume of Freund’s Complete Adjuvant. After preimmune serum was ob-tained, 2-kg male, pathogen-free New Zealand White rabbits (Charles RiverCanada, Montreal, Quebec, Canada) were immunized subcutaneously withemulsified MAP antigen (1 mg per rabbit) distributed over multiple sites.Injections with MAP antigen (1 mg in Freund’s Incomplete Adjuvant) dividedinto six to eight subcutaneous sites were repeated 4, 8, and 16 weeks after the

2 Abbreviations used are: DMEM, Dulbecco’s modified Eagle’s medium; FBS,fetal bovine serum; MAP, multivalent antigenic peptide; OR, oxidoreductase; bp,base pair; MOPS, 4-morpholinepropanesulfonic acid.

596 MARCUCCI ET AL.

initial immunization. Rabbits were exsanguinated under pentobarbital anesthe-sia 21 days after the final immunization, and serum recovered and heat-inactivated (56°C for 30 min) prior to storage at �20°C. Immunoblotting withthe anti-CYP2D6 antibody was performed under optimized conditions fordilution (1:100,000) and incubation time (1 h, 22°C). A serial dilution ofbaculovirus-expressed CYP2D6.1 was included on each blot to serve as a“standard curve” for the expressed CYP2D6. The alkaline phosphatase anti-rabbit secondary antibody (Promega, Madison, WI) was diluted 1:10,000, andbands were visualized by incubation with 5-bromo-4-chloro-3-indolyl phos-phate/nitro blue tetrazolium (Kirkegaard and Perry Laboratories, Gaithersburg,MD). Blots were scanned with a Kodak 440CF Image Station and subjected todensitometric analysis using Kodak Digital Science 1D software (EastmanKodak, Rochester, NY). Microsomal CYP2D6 apoprotein content was extrap-olated from linear regression of the CYP2D6.1 standards and expressed aspicomoles per milligram of microsomal protein.

Baculovirus Expression, Microsome Preparation, and Quantitation ofExpressed CYP2D6. CYP2D6*2 and CYP2D6*17 cDNAs were modified bythe addition of EcoRI linkers and were coexpressed with human NADPH-dependent cytochrome P450 oxidoreductase (OR) in a baculovirus expressionsystem (Sata et al., 2000). Virus bearing active CYP protein was identified byassaying virus-infected Sf9 cell lysates using the fluorometric substrate 3-[2-(N,N-diethyl-N-methylammonium)ethyl]-7-methoxy-4-methylcoumarin at afinal concentration of 25 �M (Chauret et al., 2001), and the presence of ORwas confirmed by measurement of cytochrome c reduction. Virus containingboth of these activities was amplified, and microsomes were prepared fromTrichoplusia ni cells infected with the appropriate virus (Sata et al., 2000).

In Vitro Incubation Conditions. For proteins expressed in COS-7 cells,dextromethorphan O-demethylation and bufuralol 1�-hydroxylation assayswere carried out using 0.5 pmol of enzyme (apoprotein) per incubation.Reactions were performed in round bottom, 96-well plates in a total assayvolume of 100 �l containing 50 mM potassium phosphate buffer (pH 7.4), 3mM MgCl2, 1 mM EDTA, 5 mM glucose 6-phosphate, 1 U/ml glucose-6-phosphate dehydrogenase, and 100 mM NADP and supplemented with 6.0 �gof OR. Dextromethorphan and bufuralol concentrations ranged between 0 and100 �M and 0 and 200 �M, respectively. Reactions were conducted for 60 minat 37 � 1°C with gentle shaking. Appropriate metabolite standards wereincluded in each assay. Reactions were terminated by the addition of 100 �l ofmethanol, and samples were centrifuged at 10,000 rpm for 5 min to pelletprotein prior to high-performance liquid chromatography analysis.

For baculovirus-expressed enzymes, assay conditions for dextromethorphanO-demethylation and bufuralol 1�-hydroxylase activities were identical tothose described for the COS-7-expressed proteins with the exception that theamount of added CYP2D6 variant was based on spectral determination ratherthan apoprotein content, exogenous reductase was not added, and the incuba-tion time was 30 min. Assays for debrisoquine 4-hydroxylation activity used 2pmol of enzyme per incubation and were incubated for 1 h using debrisoquineconcentrations ranging from 0 to 200 �M.

Analytical Procedures. All analytic procedures were conducted with aHewlett Packard HP1100 high-performance liquid chromatography system(Hewlett Packard Instruments, Palo Alto, CA) equipped with a degasser,binary or quaternary pump, autosampler, column heater, and programmable1100 series variable fluorescence detector. All data were collected and inte-grated with Hewlett Packard Chemstation V A.0401 software.

The O-demethylation of dextromethorphan to dextrorphan was analyzedusing a method modified from that described by Abdel-Rahman et al. (1999).Aliquots (25–75 �l) of assay supernatants were injected onto a PhenomenexPhenyl guard column (4 mm 3 mm i.d., 5-�m particle size; Phenomenex,Torrance, CA) followed by separation on a Novapak Phenyl column (3.9mm 15 cm, 4-�m particle size, 40°C; Waters, Milford, MA) that had beenequilibrated with a mobile phase consisting of a 75:25 mixture of buffer (20mM potassium phosphate and 20 mM hexane sulfonic acid, pH 4.0) andacetonitrile at a flow rate of 1.2 ml/min followed by fluorescence detectionwith excitation and emission wavelengths of 235 and 310 nm, respectively.Under these conditions, dextromethorphan and dextrorphan had retention timesof 4.1 and 15.7 min, respectively. A five-point standard curve was used toquantify the amount of dextrorphan present in sample injections. The limit ofquantitation for the assay was 0.75 pmol. The analytical method demonstratedlinearity between 1.125 and 112.5 pmol injected, the range of standards

evaluated (r2 � 0.997). Intraday and interday assay variability for dextrorphanstandards over this range of standards varied from 0.6 to 9.3% and 2.2 to 10%,respectively.

Analytical methods to determine bufuralol 1�-hydroxylation and debriso-quine 4-hydroxylation activities were modified from the method describedabove for quantitation of dextromethorphan and dextrorphan. A NovapakPhenyl column preceded by a Phenomenex Phenyl guard column was used forboth methods. For bufuralol 1�-hydroxylation, the mobile phase consisted of a65:35 mixture of buffer (20 mM potassium phosphate and 20 mM hexanesulfonic acid, pH 4.0) and acetonitrile at a flow rate of 1.0 ml/min and acolumn temperature of 40°C with fluorescence detection at excitation andemission wavelengths of 252 and 302 nm, respectively. Under these condi-tions, the retention times of 1�-hydoxybufuralol and bufuralol were 2.1 and 6.5min, respectively. The analytical method demonstrated linearity over the rangeof standards evaluated, 0.25 to 5.0 pmol injected (r2 � 0.999), and the lowerlimit of quantitation for the assay was 0.1 pmol. Intraday and interday assayvariability for the range of standards was 0.5 to 4.8% and 0.9 to 4.8%,respectively. For debrisoquine and 4-hydroxydebrisoquine, the acetonitrilecontent of the mobile phase was reduced to 20% at a flow rate of 1.0 ml/minand a column temperature of 40°C with fluorescence detection of columneluants at excitation and emission wavelengths of 208 and 290 nm, respec-tively. 4-Hydroxydebrisoquine eluted at 2.9 min under these conditions,whereas the parent compound eluted at 6.8 min. The lower limit of quantitationwas 3.75 pmol, and the method demonstrated linearity over a 100-fold rangebetween 3.75 and 375 pmol injected (r2 � 0.99). Intraday and interday assayvariability for 4-hydroxydebrisoquine varied from 0.2 to 21% and 2.8 to 18%,respectively.

Data Analysis. Kinetic parameters for metabolite formation were estimatedfrom the line of best-fit using least-squares linear regression analysis ofLineweaver-Burk plots (reciprocal substrate concentration versus reciprocalvelocity). Data presented for dextromethorphan and bufuralol biotransforma-tion by CYP2D6 variants expressed in COS-7 cells represent the mean �S.E.M. of three separate transfection experiments. For baculovirus-expressedproteins, experiments with dextromethorphan and bufuralol were conducted intriplicate whereas experiments with debrisoquine were conducted in duplicate.For each expression system, enzyme activity at every substrate concentrationstudied was determined in duplicate. Statistical comparisons were made usingDunnett’s procedure for comparing multiple treatments to a control.

Results

A representative immunoblot of microsomal fractions preparedfrom COS-7 cells expressing CYP2D6.1, CYP2D6.2, andCYP2D6.17 proteins is presented in Fig. 1. The lane labeled “control”contains microsomes prepared from cells transfected with vectorwithout insert and confirm that CYP2D6 is not constitutively ex-pressed in COS-7 cells. The absolute amount of expressed CYP2D6

FIG. 1. Detection by immunoblot analysis of CYP2D6 protein expressed inmicrosomes isolated from COS-7 cells.

Representative results of expressed CYP2D6.1, CYP2D6.2, and CYP2D6.17,with known quantities of baculovirus-expressed CYP2D6.1 (Gentest Corp.) rangingfrom 0.1 to 0.5 pmol/lane serving as standards for quantitation.

597IN VITRO CHARACTERIZATION OF CYP2D6.1, .2, AND .17

apoprotein (in picomoles) from each transfection experiment is pre-sented in Table 1. Although factors such as cell confluency at the timeof transfection, concentration of the plasmid preparation, and cellviability at the time of harvest may contribute to some degree ofvariability between transfections, the data demonstrated that for agiven transfection experiment expression of each CYP2D6 allelicproduct was comparable, suggesting that differences in the stability ofthe three expressed CYP2D6 variants were minimal.

In contrast to expression in the baculovirus system, insufficientamounts of microsomal protein were available from the COS-7 cellexpression experiments to determine P450 holoenzyme levels basedon spectral analysis. Therefore, the rate of product formation byCOS-7 cell microsomes was expressed relative to immunoreactiveapoprotein. Experimentally determined values for the kinetic param-eters Km and Vmax describing dextromethorphan and bufuralol bio-transformation by CYP2D6.1, CYP2D6.2, and CYP2D6.17 expressedin COS cells are summarized in Table 2. Compared with the wild-typeprotein, the R296C and S486T amino acid substitutions present inCYP2D6.2 had no apparent effect on Km and a marginal 30% reduc-tion in Vmax with dextromethorphan as a substrate and a modest 30%increase in Km and no apparent effect on Vmax with bufuralol as thesubstrate. The net result of the R296C and S486T substitutions was anapproximately 30% reduction in the intrinsic clearance (CLint; Vmax/Km) for each substrate that did not achieve statistical significance(p � 0.06).

Relative to CYP2D6.2, further addition of the T107I substitution inCYP2D6.17 resulted in an approximately 2-fold increase in Km and a33% decrease in Vmax for dextromethorphan, but no effect on Km, anda 45% decrease in Vmax with bufuralol as the substrate. The CLint ofdextromethorphan and bufuralol by CYP2D6.17 was significantly lessthan that observed with CYP2D6.1 being 25% (p � 0.01) and 37%(p � 0.02) of the CYP2D6.1 activity, respectively. The CLint of eachsubstrate was also significantly lower (p � 0.05) with CYP2D6.17compared with CYP2D6.2 (Table 2).

In general, when microsomes prepared from baculovirus-trans-fected T. ni cells were used, Km estimates were 2- to 3-fold lower andVmax estimates were 10- to 20-fold higher than those determined usingmicrosomes from COS-7 cells (Fig. 2 and Table 3). Consistentchanges in the kinetic parameters characterizing dextromethorphan,bufuralol, and debrisoquine biotransformation by CYP2D6.2 relativeto CYP2D6.1 were observed for all three substrates, with a 40 to 50%

TABLE 1

Inter- and intraexperiment variability in the expression of CYP2D6.1, CYP2D6.2,and CYP2D6.17 apoprotein in COS-7 cells

Protein

Total Recovery of Expressed CYP2D6 Apoprotein perTransfection

Transfection 1 Transfection 2 Transfection 3

pmol

CYP2D6.1 17.2 30.8 29.6CYP2D6.2 15.6 24.0 31.6CYP2D6.17 17.6 25.6 25.2

TABLE 2

Mean � S.E.M. kinetic parameters from CYP2D6.1, CYP2D6.2, and CYP2D6.17 expressed in COS-7 cells

ProteinDextromethorphan Bufuralol

Km Vmax Vmax/Km Vmax/Km Km Vmax Vmax/Km Vmax/Km

�M pmol/pmol/min �l/min/nmol % CYP2D6.1 �M pmol/pmol/min �l/min/nmol % CYP2D6.1

CYP2D6.1 5.4 � 1.3 0.68 � 0.06 126 � 20 25 � 3 1.0 � 0.13 40 � 4CYP2D6.2 5.4 � 0.3 0.49 � 0.00 90 � 3 71 32 � 3 0.94 � 0.05 29 � 4 72CYP2D6.17 9.9 � 2.9 0.32 � 0.02 32 � 16 25 33 � 9 0.51 � 0.05 15 � 5 37

FIG. 2. Kinetics of dextromethorphan O-demethylation, bufuralol 1�-hydroxylation, and debrisoquine 4-hydroxylation in microsomes prepared from

insect cells expressing CYP2D6.1 (squares), CYP2D6.2 (circles), andCYP2D6.17 (triangles).

Representative data from a single experiment are presented.

598 MARCUCCI ET AL.

increase in Km value combined with an approximately 50% increasein Vmax such that CLint values were the same or slightly (10–15%)greater for CYP2D6.2 (Table 3). The reduction in CLint associatedwith CYP2D6.17 also appeared to be similar for all three substrates,ranging from 18% of CYP2D6.1 activity for dextromethorphan (p �0.01) to 22% of CYP2D6.1 activity for bufuralol and debrisoquine(p � 0.05). However, there seemed to be subtle substrate-dependentdifferences in the effects on Km and Vmax, with a tendency for changesin dextromethorphan and debrisoquine Km to be greater than changesin Vmax, whereas equivalent (2-fold) increases in each parameter wereobserved when bufuralol was the substrate.

Discussion

In theory, phenotyping individuals for specific drug biotransforma-tion pathways prior to initiation of drug therapy has considerablepotential for improving pharmacotherapy, especially for drugs withnarrow therapeutic indices. In principle, the phenotype (usually aurinary metabolite ratio) is a surrogate measure of enzyme activityfrom which drug clearance by the eliminating organ can be inferred.The implicit intent of this process is to tailor initial dose selectionaccording to the individual clearance estimate of the patient such thattarget therapeutic drug concentrations can be achieved as quickly aspossible while minimizing the risk of concentration-dependent toxic-ity. CYP2D6 pharmacogenetics figure prominently in this process dueto the number of clinically useful compounds (tricyclic antidepres-sants, serotonin selective reuptake inhibitors, and cardiovascularagents) that are dependent upon this polymorphically expressed en-zyme for their elimination from the body. In particular, relationshipsbetween CYP2D6 genotype and/or phenotype and drug clearance (andas a result, plasma concentrations achieved) have been established forseveral agents with relatively narrow therapeutic ranges such asnortriptyline (Nordin et al., 1985; Yue et al., 1999), desipramine(Spina et al., 1997), paroxetine (Sindrup et al., 1992; Yoon et al.,2000), and venlafaxine (Lessard et al., 1999), although CYP2D6phenotype appears to predict toxicity better than it predicts clinicalefficacy (Spina et al., 1997; Lessard et al., 1999).

CYP2D6 phenotyping studies indicate that Chinese (Bertilsson etal., 1992) and individuals of black African origin (Woolhouse et al.,1985; Masimirembwa et al., 1996a; Wennerholm et al., 1999) tend tohave lower CYP2D6 activity on a population basis, relative to Cau-casians. Black African populations are notable for the presence of theCYP2D6*17 allele, which occurs at a frequency as high as 34% instudied populations (Bradford et al., 1998), and for an apparent lackof coregulatory control between debrisoquine and sparteine pheno-types (Woolhouse et al., 1985; Lennard et al., 1992) and debrisoquineand metoprolol phenotypes (Sommers et al., 1989; Simooya et al.,1993; Masimirembwa et al., 1996a). The CYP2D6*17 allele, whichhas been observed to have reduced activity in vitro (Oscarson et al.,1997), is therefore likely to be at least partially responsible for thelower CYP2D6 activity observed in black African populations (Ma-simirembwa et al., 1996b; Griese et al., 1999; Wennerholm et al.,1999). However, its contribution to the observed discrepancies inphenotype assignment noted above has not been addressed. Therefore,the specific goal of this investigation was to determine whether theCYP2D6.17 protein had differential effects on the biotransformationof model substrates, dextromethorphan and debrisoquine, commonlyused as in vivo phenotyping probes in North America.

In essence, the results of these studies are consistent with theCYP2D6*17 allele contributing toward the observation of lowerCYP2D6 activity in populations of black African origin relative towhite American or European populations. The intrinsic clearance(estimated as Vmax/Km) of all three substrates (dextromethorphan,

TA

BL

E3

Mea

n�

S.E

.M.

kine

tic

para

met

ers

from

bacu

lovi

rus-

expr

esse

dC

YP

2D6.

1,C

YP

2D6.

2an

dC

YP

2D6.

17

Prot

ein

Dex

trom

etho

rpha

nB

ufur

alol

Deb

riso

quin

e

Km

Vm

axV

max

/Km

Vm

ax/K

mK

mV

max

Vm

axK

mV

max

/Km

Km

Vm

axV

max

/Km

Vm

ax/K

m

�M

pmol

/pm

ol/m

in�

l/m

in/n

mol

%C

YP

2D6.

1�

Mpm

ol/p

mol

/min

�l/

min

/nm

ol%

CY

P2D

6.1

�M

pmol

/pm

ol/m

in�

l/m

in/n

mol

%C

YP

2D6.

1

CY

P2D

6.1

1.0

�0.

32.

4�

0.4

2400

�27

38.

1�

0.9

10.2

�0.

712

59�

287

13.4

�3.

32.

8�

0.1

209

�43

CY

P2D

6.2

1.4

�0.

33.

7�

0.4

2600

�45

710

913

.0�

1.7

19.3

�0.

514

85�

186

117

21.2

�0.

14.

5�

0.0

212

�0.

010

1C

YP2

D6.

176.

1�

0.6

2.7

�0.

445

0�

3818

16.5

�0.

44.

5�

0.0

273

�6

2246

.3�

2.5

2.2

�2.

547

�1.

022

599IN VITRO CHARACTERIZATION OF CYP2D6.1, .2, AND .17

debrisoquine and bufuralol) catalyzed by CYP2D6.17 was reduced toapproximately 15 to 20% of that observed with the CYP2D6.1 andCYP2D6.2 proteins. Furthermore, since CYP2D6.17 affected the in-trinsic clearance of all three substrates to a similar extent, it wouldappear that the CYP2D6*17 allele alone cannot account for the poorcorrelation between dextromethorphan and debrisoquine metabolicratios (rs � 0.52, p � 0.05) observed by Droll et al. (1998) inGhanaian volunteers. Since sparteine and metoprolol were not specif-ically investigated in this study, it is not possible to address the effectsof CYP2D6.17 on their metabolism nor the contribution of CYPD26.17to the poor correlations between these two phenotyping probes anddebrisoquine in vivo (Woolhouse et al., 1985; Sommers et al., 1989;Lennard et al., 1992; Simooya et al., 1993; Masimirembwa et al., 1996a;Droll et al., 1998).

A novel variant allele with reduced functional activity,CYP2D6*29, was recently discovered at a frequency of 20% in ablack Tanzanian population (Wennerholm et al., 2001) compared withonly one allele observed in 672 European subjects (Marez et al.,1997). This allele codes for a protein equivalent to CYP2D6.2 withtwo additional amino acid changes, V136I and V338M. CYP2D6.29expressed in COS-1 cells and supplemented with exogenous OR had26% of the catalytic activity of CYP2D6.1 with bufuralol as a sub-strate whereas debrisoquine biotransformation was less affected (63%of CYP2D6.1 activity) (Wennerholm et al., 2001). Thus, CYP2D6.29rather than CYP2D6.17 may account for the poor correlation betweenphenotyping probes observed in individuals of black African origin.

From a technical perspective, estimates of Km for dextromethor-phan and bufuralol tended to be 2- to 3-fold lower with baculovirus-expressed enzymes compared with the same proteins expressed inCOS-7 cells. In addition, Vmax values were 5- to 10-fold higher fordextromethorphan and approximately 10-fold higher for bufuralolusing the baculovirus-expressed enzymes. A number of factors maycontribute to these observed differences including the higher level ofprotein expression in T. ni cells compared with COS-7 cells and therelative efficiency of coexpressed reductase coupling with the ex-pressed CYP2D6 variants compared with exogenous supplementa-tion, as was the case with COS-7-expressed enzyme. In addition, theCYP2D6.2 variant has been demonstrated to be more labile thanCYP2D6.1 when expressed in insect cells, and the amount of holo-protein is reduced to a greater extent than apoprotein (Zanger et al.,2001). In our studies, incubations with baculovirus-expressed enzymewere conducted using equivalent amounts of each variant form basedon spectral determination whereas those with COS cell-expressedenzyme were based on immunochemically determined apoproteincontent. Thus, it is possible that CYP2D6.17 may demonstrate labilitysimilar to that observed by others with CYP2D6.2 and that the extentof lability may differ between the two expression systems. Regardless,the rank order of dextromethorphan Km estimates with CYP2D6.1,CYP2D6.2, and CYP2D6.17 proteins determined in either the COS-7or baculovirus systems in the present study compares favorably withthe values reported for morphine formation from codeine using thesame enzymes expressed in yeast (Oscarson et al., 1997). In contrast,bufuralol Vmax was essentially identical for all three CYP2D6 variantsexpressed in yeast (Oscarson et al., 1997) whereas a 50% reduction invalue was observed for CYP2D6.17 in both the COS-7 and baculo-virus expression systems. Nevertheless, both the results of Oscarson etal. (1997) and those presented here confirm that the intrinsic clearanceof all substrates tested is reduced with CYP2D6.17 compared withCYP2D6.1. Therefore, since both the mammalian COS-7 and thebaculovirus expression systems were able to detect the reduced cata-lytic activity associated with CYP2D6.17, we conclude that theCOS-7 expression system is useful for rapid, initial characterization of

the functional consequences of newly identified allelic variants, butsystems yielding higher levels of expression are preferred for moreaccurate estimation of kinetic parameters.

In conclusion, this in vitro study confirms in vivo reports that theCYP2D6*17 allele confers reduced catalytic activity. However, theissue of discrepancies in phenotype assignment when using differentmodel CYP2D6 probes remains unresolved. Population phenotypingstudies imply that metabolism and thus clearance of CYP2D6 sub-strates may be reduced in African Americans relative to Caucasians,and this appears to be the case for nortriptyline. However, clearanceof metoprolol appears to be similar (Johnson and Burlew, 1996) andthe clearance of propranolol (Sowinski et al., 1996), by both CYP1A2and CYP2D6 (Johnson et al., 2000), is increased in African Ameri-cans compared with Caucasians. Thus, clearer insights into the con-sequences of existing and yet undiscovered population-specific alleleson a broad range of individual substrates are necessary to optimizepharmacotherapy of CYP2D6 substrates in individuals of black Afri-can origin.

Acknowledgments. We gratefully acknowledge Thomas Walkerfor valuable assistance with bufuralol and debrisoquine assay devel-opment; we also acknowledge Mark Marcucci, Ph.D. for kindlyconducting the statistical analyses.

References

Abdel-Rahman SM, Marcucci K, Boge T, Gotschall RR, Kearns GL, and Leeder JS (1999)Potent inhibition of cytochrome P-450 2D6-mediated dextromethorphan O-demethylation byterbinafine. Drug Metab Dispos 27:770–775.

Bertilsson L, Lou Y-Q, Du Y-L, Liu Y, Kuang T-Y, Liao X-M, Wang K-Y, Reviriego J, IseliusL, and Sjoqvist F (1992) Pronounced differences between native Chinese and Swedishpopulations in the polymorphic hydroxylations of debrisoquin and S-mephenytoin. ClinPharmacol Ther 51:388–397.

Bradford LD, Gaedigk A, and Leeder JS (1998) High frequency of CYP2D6 poor and “inter-mediate” metabolizers in black populations: a review and preliminary data. PsychopharmacolBull 34:797–804.

Chauret ND, B., Lackman RL, Bateman K, Nicoll-Griffith D, Stresser DM, Ackermann JM,Turner SD, Miller VP, and Crespi CL (2001) The use of 3-[2-(N, N-diethyl-N-methylammonium)ethyl]-7-methoxy-4-methylcoumarin (AMMC) as a specific CYP2D6probe in human liver microsomes. Drug Metab Dispos 29:1196–1200.

Droll K, Bruce-Mensah K, Otton SV, Gaedigk A, Sellers EM, and Tyndale RF (1998) Com-parison of three CYP2D6 probe substrates and genotype in Ghanaians, Chinese and Cauca-sians. Pharmacogenetics 8:325–333.

Gaedigk A, Gotschall RR, Forbes NS, Simon SD, Kearns GL, and Leeder JS (1999) Optimizationof cytochrome P450 2D6 (CYP2D6) phenotype assignment using a genotyping algorithmbased on allele frequency data. Pharmacogenetics 9:669–682.

Griese E-U, Asante-Poku S, Ofori-Adjei D, Mikus G, and Eichelbaum M (1999) Analysis of theCYP2D6 gene mutations and their consequences for enzyme function in a West Africanpopulation. Pharmacogenetics 9:715–723.

Hasemann CA, Kurumbail RG, Boddupalli SS, Peterson JA, and Deisenhofer J (1995) Structureand function of cytochromes P450: a comparative analysis of three crystal structures. Structure3:41–62.

Johnson JA and Burlew BS (1996) Metoprolol metabolism via CYP2D6 in ethnic populations.Drug Metab Dispos 24:350–355.

Johnson JA, Herring VL, Wolfe MS, and Relling MV (2000) CYP1A2 and CYP2D6 4-hydroxy-late propranolol and both reactions exhibit racial differences. J Pharmacol Exp Ther 294:1099–1105.

Lennard MS, Iyun AO, Jackson PR, Tucker GT, and Woods HF (1992) Evidence for adissociation in the control of sparteine, debrisoquine and metoprolol metabolism in Nigerians.Pharmacogenetics 2:89–92.

Lessard E, Yessine MA, Hamelin BA, O’Hara G, LeBlanc J, and Turgeon J (1999) Influence ofCYP2D6 activity on the disposition and cardiovascular toxicity of the antidepressant agentvenlafaxine in humans. Pharmacogenetics 9:435–443.

Marez D, Legrand M, Sabbagh N, Lo Guidice J-M, Spire C, Lafitte J-J, Meyer UA, and BrolyF (1997) Polymorphism of the cytochrome P450 CYP2D6 gene in a European population:characterization of 48 mutations and 53 alleles, their frequencies and evolution. Pharmaco-genetics 7:193–202.

Masimirembwa C, Hasler J, Bertilsson L, Johansson I, Ekberg O, and Ingelman-Sundberg M(1996a) Phenotype and genotype analysis of debrisoquine hydroxylase (CYP2D6) in a blackZimbabwean population. Reduced enzyme activity and evaluation of metabolic correlation ofCYP2D6 probe drugs. Eur J Clin Pharmacol 51:117–122.

Masimirembwa C, Persson I, Bertilsson L, Hasler J, and Ingelman-Sundberg M (1996b) A novelmutant variant of the CYP2D6 gene (CYP2D6*17) common in a black African population:association with diminished debrisoquine hydroxylase activity. Br J Clin Pharmacol 42:713–719.

Modi S, Paine MJ, Sutcliffe MJ, Lian L-Y, Primrose WU, Wolf CR, and Roberts GCK (1996)A model for human cytochrome P450 2D6 based on homology modeling and NMR studies ofsubstrate binding. Biochemistry 35:4540–4550.

Nordin C, Siwers B, Benitez J, and Bertilsson L (1985) Plasma concentrations of nortriptyline

600 MARCUCCI ET AL.

and its 10-hydroxy metabolite in depressed patients-relationship to the debrisoquine hydroxy-lation metabolic ratio. Br J Clin Pharmacol 19:832–835.

Oscarson M, Hidestrand M, Johansson I, and Ingelman-Sundberg M (1997) A combination ofmutations in the CYP2D6*17(CYP2D6Z) allele causes alterations in enzyme function. MolPharmacol 32:1034–1040.

Sachse C, Brockmoller J, Bauer S, and Roots I (1997) Cytochrome P450 2D6 variants in aCaucasian population: allele frequencies and phenotypic consequences. Am J Hum Genet60:284–295.

Sata F, Sapone A, Elizondo G, Stocker P, Miller VP, Zheng W, Raunio H, Crespi CL, andGonzalez FJ (2000) CYP3A4 allelic variants with amino acid substitutions in exon 7 and 12:evidence for an allelic variant with altered catalytic activity. Clin Pharmacol Ther 67:48–56.

Simooya OO, Njunju E, Rostami Hodjegan A, Lennard MS, and Tucker GT (1993) Debrisoquineand metoprolol oxidation in Zambians: a population study. Pharmacogenetics 3:205–208.

Sindrup SH, Brøsen K, Gram LF, Hallas J, Skjelbo E, Allen A, Allen GD, Cooper SM, MellowsG, Tasker TCG, and Zussman BD (1992) The relationship between paroxetine and thesparteine oxidation polymorphism. Clin Pharmacol Ther 51:278–287.

Sommers DK, Moncrieff J, and Avenant J (1989) Non-correlations between debrisoquine andmetoprolol polymorphisms in the Venda. Human Toxicol 8:365–368.

Sowinski KM, Lima JJ, Burlew BS, Massie JD, and Johnson JA (1996) Racial differences inpropranolol enantiomer kinetics following simultaneous IV and oral administration. Br J ClinPharmacol 42:339–346.

Spina E, Gitto C, Avenoso A, Campo GM, Caputi AP, and Perucca E (1997) Relationship

between plasma desipramine levels, CYP2D6 phenotype and clinical response to desipramine:a prospective study. Eur J Clin Pharmacol 51:395–398.

Wennerholm A, Johansson I, Hidestrand M, Bertilsson L, Gustaffson LL, and Ingelman-Sundberg M (2001) Characterization of the CYP2D6*29 allele commonly present in a blackTanzanian population causing reduced catalytic activity. Pharmacogenetics 11:417–427.

Wennerholm A, Johansson I, Massele AY, Jande M, Alm C, Aden-Abdi Y, Dahl M-L,Ingelman-Sundberg M, Bertilsson L, and Gustaffson LL (1999) Decreased capacity fordebrisoquine metabolism among black Tanzanians: analyses of the CYP2D6 genotype andphenotype. Pharmacogenetics 9:707–714.

Woolhouse NM, Eichelbaum M, Oates NS, Idle JR, and Smith RL (1985) Dissociation ofco-regulatory control of debrisoquin/phenformin and sparteine oxidation in Ghanaians. ClinPharmacol Ther 37:512–521.

Yoon Y-R, Chan I-J, Shon J-H, Kim K-A, Cha Y-N, Jang I-J, Park C-W, Shin S-G, FlockhartDA, and Shin J-G (2000) Relationship of paroxetine disposition to metoprolol metabolite ratioand CYP2D610 genotype of Korean subjects. Clin Pharmacol Ther 67:567–576.

Yue Q-Y, Zhong Z-H, Tybring G, Dalen P, Dahl M-L, Bertilsson L, and Sjoqvist F (1999)Pharmacokinetics of nortriptyline and its 10-hydroxy metabolite in Chinese subjects ofdifferent CYP2D6 genotypes. Clin Pharmacol Ther 64:384–390.

Zanger UM, Fischer J, Raimundo S, Stuven T, Evert BO, Schqab M, and Eichelbaum M (2001)Comprehensive analysis of the genetic factors determining expression and function of hepaticCYP2D6 Pharmacogenetics 11:573–585.

601IN VITRO CHARACTERIZATION OF CYP2D6.1, .2, AND .17