Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

CYP2D6 is responsible for the DEB 4-hydroxylase polymorphism, which has important clinical consequences, including pronounced interindividual variation in the disposition of many important drugs such as -adrenergic blocking agents, antiarrhythmics, antidepressants, and analgesics. Since it was first discovered, CYP2D6 has been the most studied human genetic polymorphism in drug metabolism with more than 75 identified alleles within most human populations and racial groups (http://www.imm.ki.se/CYPalleles/cyp2d6.htm). Approximately 7 to 10% of the Caucasian population inherit mutant CYP2D6 alleles as an autosomal recessive trait (Mahgoub et al., 1977). This polymorphism stratifies the population depending on the copy number of wild-type alleles: poor (PM, zero), intermediate (one), extensive (EM, two), and ultra-rapid metabolizers (multiple copies) (Gonzalez, 1996; Wolf and Smith, 1999). The CYP2D gene locus has been isolated, sequenced, and mapped to human chromosome 22q13.1, and contains the only active CYP2D6 gene downstream of two inactive pseudogenes CYP2D7P and CYP2D8P (Gonzalez et al., 1988; Kimura et al., 1989).

Clinical studies are fundamental to the identification of human polymorphisms, to the establishment of pharmacokinetic profiles, and to drug-drug interaction effects. However, to determine how a drug is metabolized, what toxic effects it may produce, or how pathophysiological conditions affect drug metabolism, animal models or in vitro systems must be developed.

Because of marked differences between humans and experimental animals, the results from animal studies can be misleading and need to be interpreted very cautiously. For example, the CYP2D family in humans has a single active member, CYP2D6, whereas rats and mice have at least five genes (Gonzalez and Nebert, 1990; Nelson et al., 1996). DEB is hydroxylated to 4-hydroxydebrisoquine (4-OH-DEB) by humans and by Sprague-Dawley rats. However, Dark Agouti rats have been found to posses a low capacity to metabolize DEB (Al-Dabbagh et al., 1981). Similarly, no significant formation of 4-hydroxy DEB was detected by liver microsomes from three strains of mice and by purified cyp2d9-11 (Masubuchi et al., 1997).

The transcriptional factor hepatic nuclear factor 4 (HNF4) is known to play a major role in liver organogenesis, maintenance, and gene transcription (Sladek, 1994). Several lines of evidence suggest that HNF4 controls constitutive levels of expression of the CYP2D6 gene. Indeed, it was shown that cotransfection of the minimal CYP2D6 promoter-CAT construct (392 bp) with a mammalian HNF4 expression vector produced a 30-fold induction of CAT activity in COS-7 cells (Cairns et al., 1996). Moreover, using adenovirus-mediated antisense targeting that selectively diminishes HNF4 content in human hepatocytes, CYP2D6 gene expression, among other P450s, is down-regulated by the depletion of HNF4 (Jover et al., 2001).

In addition, the R154X mutation in the human HNF4 gene provokes maturity onset diabetes of the young, a genetically and clinically heterogeneous subtype of non-insulin-dependent diabetes mellitus. Persons carrying this mutation in HNF4 are predicted to have reduced levels of this transcription factor in the tissues where it is expressed (Yamagata et al., 1996; Lindner et al., 1997). Because this pathology occurs concomitantly with other diseases like obesity, hypertension, coronary insufficiency, and heart failure among others, multiple drug regimes are common within this population (Scheen and Lefebvre, 1995). Because CYP2D6 metabolizes a large number of clinically prescribed drugs, it would be relevant to determine whether the lack of HNF4 could modulate CYP2D6 metabolic activity in vivo. However, because of the embryonic lethality of a standard gene HNF4 knock-out approach, preclinical evaluation of this hypothesis using a intact animal model is not possible. Fortunately, the generation of a conditional liver-specific knock-out of the HNF4 gene mouse line has overcome this issue (Hayhurst et al., 2001).

To circumvent all of these methodological problems, a humanized mouse line expressing CYP2D6 would offer a unique approach to answering fundamental questions about the specific role of CYP2D6 in drug metabolism and drug interactions. Such experiments would be performed in the context of the entire animal and would overcome many limitations inherent in in vitro experiments. To this end, the complete wild-type allele of the human CYP2D6 gene, including its regulatory sequence, was microinjected into a fertilized FVB/N mouse egg to produce a humanized transgenic mouse line. In addition, to explore the in vivo regulation of CYP2D6 expression by HNF4, a mouse line carrying both the CYP2D6 gene and the conditional HNF4 mutation has been generated. Using DEB as a substrate, only the humanized CYP2D6 mice are able extensively to convert DEB to its 4-hydroxy metabolite and to display a pharmacokinetic profile similar to humans.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Animals. Adult males from all the genotypes described in this work (25-30 g, 2-4 months) were maintained under conditions of controlled temperature (23 ± 1°C) and lighting (lights on 6:00 AM-6:00 PM), with food and water provided ad libitum. All animal experiments were conducted under National Institutes of Health guidelines for the use and care of laboratory animals, and approved by National Institutes of Health Animal Care and Use Committee.

Generation of the Humanized Mouse. The CYP2D6 gene (GenBank accession number M33388) previously isolated and sequenced (Kimura et al., 1989) was microinjected into a fertilized FVB/N mouse egg to produce a transgenic mouse line. Incorporation of the CYP2D6 DNA within the mouse genome was determined by both PCR and Southern blot analysis. The transgenic founder was mated to a nontransgenic FVB/N (wild-type), and animals from this cross were subsequently crossed to each other to produce homozygous mice. Mice homozygous for the transgene were confirmed by crossing them with wild-type mice and testing the progeny for transgene transmission. Heterozygous animals were generated by crossing homozygous and wild-type mice. Wild-type and homozygous littermates were bred and maintained by brother-sister mating.

Generation of Liver-Specific Deletion of HNF4 in the CYP2D6 Transgenic Line. HNF4 conditional knock-out mice had been generated previously (Hayhurst et al., 2001). The breeding scheme to produce liver-specific deletion of HNF4 in the CYP2D6 transgenic line was designed using two different mice genotypes: a) HNF4 fl/wt, AlbCre +/ mice generated by crossing HNF4 fl/fl homozygous to albumin-Cre heterozygous (Alb-Cre) kindly provided by Dr. Derek LeRoith (National Institute of Diabetes and Digestive and Kidney Diseases, Bethesda, MD) (Yakar et al., 1999); and b) HNF4 fl/fl, CYP2D6 +/ mice: generated by crossing HNF4 fl/fl homozygous mice to CYP2D6 homozygous animals and back-crossing the resultant offspring to homozygous HNF4 fl/fl mice.

HNF4

HNF4

PCR Genotyping Procedures. Genomic DNA was isolated from tails as described previously (Laird et al., 1991). For CYP2D6 PCR analysis, approximately 50 ng of tail DNA was amplified in 25 µl of reaction mixture containing 2.5 mM MgCl₂, 0.2 mM deoxynucleoside triphosphates (dNTPs), 1.25 U of AmpliTaq (PerkinElmer, Foster City, CA), and 20 pmol of CYP2D6 gene-specific primers CYP2D6F 5'-AGAAGGGGAAGCAGGTTTG-3' and CYP2D6R 5'-CGGCACTCAGGACTA ACTCATC-3', and microsomal epoxide hydrolase (mEH) gene-specific primers MEHF 5'-AAGTGAGTTTGCATGGCGCAGC-3' and MEHR 5'-CCCTTTAGCCCCTTCCCTCTG-3'. Cycling conditions were 94°C for 5 min, and then 33 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, followed by a 5-min extension at 72°C. mEH primers served as a positive control for amplification, yielding a fragment of 341 bp in all samples (Miyata et al., 1999). An additional band of 241 bp was amplified exclusively in CYP2D6 humanized animals. For HNF4 fl/fl mice and AlbCre transgenic PCR procedures were described previously (Hayhurst et al., 2001).

Southern Blot Analysis and Determination of Transgene Copy Number. Tail genomic DNA (15 µg/lane) was digested with BamHI and subjected to electrophoresis on a 0.5% agarose gel containing 0.5× Tris/borate/EDTA. The DNA was hydrolyzed in 0.2 M HCl and transferred on to a Gene Screen Plus nylon membrane (DuPont, Wilmington, DE) by capillary blotting in 0.4 M NaOH. Blots hybridized with random-primer ³²P-labeled CYP2D6 cDNA probe (Gonzalez et al., 1988) at 42°C overnight, washed twice in 2× SSC (1× SSC is 150 mM NaCl plus 15 mM sodium citrate) and 0.5% SDS at 65°C for 15 min, twice in 0.1 × SSC and 0.5% SDS for 5 min, and exposed to a PhosphorImager screen (Molecular Dynamics, Sunnyvale, CA) for 2 to 4 h.

Northern Blot Analysis. Total RNA was isolated from the livers of all the HNF4 fl/fl, AlbCre, and CYP2D6 humanized mice used in this study by the acidic guanidine isocyanate/phenol/chloroform extraction method using the Ultraspec RNA isolation system (Biotecx Laboratories, Houston, TX). Ten micrograms of total RNA per sample was subjected to electrophoresis on a 1% agarose gel containing 220 mM formaldehyde in 20 mM MOPS, 8 mM sodium acetate, and 1 mM EDTA buffer, pH 7.0, and transferred to Gene Screen Plus membranes by capillary blotting in 20× SSC. Blots were hybridized with a random-primer ³² P-labeled probe specific for HNF4 exon 4 and 5 probe (Hayhurst et al., 2001) at 42°C overnight, washed twice in 2× SSC and 0.5% SDS at 65°C for 15 min, twice in 0.1× SSC and 0.5% SDS for 5 min, and exposed to a PhosphorImager screen for 2 to 4 h.

Western Blot Analysis. Tissues were homogenized in ice-cold buffer (1.15% KCl, 50 mM Tris-HCl, and 1 mM EDTA, pH 7.4) and microsomes were prepared by differential centrifugation as described previously (Sinal et al., 1999). Microsomal protein concentration was determined using a bicinchoninic acid protein kit (Pierce, Rockford, IL) using bovine serum albumin as a standard. SDS-polyacrylamide gel electrophoresis and Western blot analysis of microsomal proteins (40 µg) were performed with a 4% stack and 10% separating gel and transferred to nitrocellulose (Schleicher & Schuell, Keene, NH) using standard methods. The primary mouse monoclonal anti-CYP2D6 antibody (Gentest Corp., Woburn, MA) was diluted 3,000-fold in 0.5% nonfat dry milk, 1× phosphate-buffered saline. The primary monoclonal antibody against rat CYP2E1 (kindly provided by Dr. Harry V. Gelboin) was diluted 500-fold in 3% milk and 1× phosphate-buffered saline. Bound antibody was detected with a horseradish peroxidase-conjugated secondary antibody anti-mouse IgG, that was diluted 2,000- and 10,000-fold for CYP2D6 and CYP2E1 blotting, respectively. P450 proteins were visualized by use of the enhanced chemiluminescence reagent (Amersham Pharmacia Biotech, Piscataway, NJ).

Drug Administration and Blood and Urine Samples. DEB hemisulfate (ICN, Irvine, CA) was dissolved in sterile water and administered orally by gavage at 2.5 mg/kg. For pharmacokinetic evaluations, blood samples were collected from suborbital veins 0, 0.5, 1, 2, 4, 6, 8, 12, and 24 h after DEB administration (2.5 mg/kg). Each time point was analyzed with three to four animals. Serum was separated by centrifugation at 1000g, 4°C, for 10 min. For the urine excretion analysis, three to four animals per group were dosed with DEB (2.5 mg/kg) and placed in metabolic chambers (Jencons, Leighton Buzzard, UK). Total urinary excretion from individual mice was collected for 24 h. Serum aliquots (0.2-0.5 ml) or urine volumes (0.8-1.5 ml) were stored at 80°C until analyzed. At the end of experiments, mice were euthanized by carbon dioxide.

Quantification of DEB and 4-OH-DEB by Liquid Chromatography/Tandem Mass Spectrometry. Serum and urine concentrations of the DEB and 4-OH-DEB were determined using a previously described LC/MS/MS method with a minor modification using liquid-liquid extraction (Scott et al., 1999) and solid phase extraction (Pereira et al., 2000). Briefly, 100 µl of serum or urine (supplemented with 30 mg of sodium chloride) was spiked with 20 µl of internal standard (phenacetin, 30 µM in methanol), 500 µl of isopropanol, 50 µl of aqueous sodium hydroxide (400 mM), and 3 ml of methyl-t-butyl ether were added. The mixture was vortex-mixed for 1 min and the phases were separated by 10 min of centrifugation at 1000g, 4°C. 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 at 30°C using a heating block (Pierce, Rockford, IL). The residue was reconstituted in 100 µl of acetonitrile-water (20:80, v/v) and transferred to polypropylene autosampler vials and 10 to 25 µl of sample was injected into the LC/MS/MS system.

Pharmacokinetic Analysis. Pharmacokinetics parameters for DEB and its metabolite 4-OH-DEB were estimated from the serum concentration-time data by a noncompartmental approach with the software package WinNonlin Standard version 1.5 (Scientific Consulting Inc., Cary, NC). The peak concentration in serum (C_max) and the time to reach serum concentration (T_max) were obtained from the original data. The area under the serum concentration versus time curves from 0 to 24 h (AUC_{0-24 h}) was determined with a combination of linear and logarithmic trapezoidal methods. The elimination rate constant () was determined by use of log-linear regression of the terminal portion of the concentration versus time curve using at least four data points. The elimination half-life (t_1/2) was calculated from 0.693/. The apparent oral clearance was calculated from dose/AUC_{0-24 h}. From urine concentrations, the percentage of the total dose excreted was determined.

Statistics. Statistical analyses were performed with SigmaStat software (Jandel Corporation, San Rafael, CA) using one-way or two-way analysis of variance followed by the Student-Newman-Keul's test when comparing three or four groups, respectively. Differences were considered significant if the probability that they were due to chance was less than 5%.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Generation of a CYP2D6 Humanized Mouse Line. The CYP2D6 gene was isolated and sequenced in an earlier study (Kimura et al., 1989). The sequence indicated that it was a wild-type allele. The complete CYP2D6 gene (Fig. 1A), including its promoter and regulatory elements, was microinjected into a fertilized FVB/N mouse egg to produce a transgenic mouse line. Successful incorporation of the CYP2D6 DNA into the germ line was assessed by Southern blot analysis. Genomic DNA from wild-type and CYP2D6 humanized mice was digested with BamHI, and probed with the CYP2D6 cDNA. Hybridization signal was found only in the lanes with transgenic DNA but not in the lanes with wild-type DNA. The size of the bands corresponds exactly with the predicted sizes calculated from the sequence of the CYP2D6 gene (Fig. 1B). Furthermore, the results of Southern blot analysis were corroborated by PCR. Genomic DNA from wild-type and CYP2D6 humanized mice was amplified with two different sets of specific primers: CYP2D6, and mEH (Miyata et al., 1999). Figure 1C shows the amplification of mEH PCR product (341 bp) in both wild-type and CYP2D6 humanized animals, which served as a positive control for amplification. CYP2D6 PCR product (241 bp) is only present in CYP2D6 humanized mice. The transgene was present at 5 ± 1 copies per haploid genome.

View larger version (25K): [in this window] [in a new window]   Fig. 1.   Generation and analysis of the CYP2D6 humanized mouse. A, schematic diagram of the wild-type CYP2D6 gene used for microinjection (GenBank accession number M33388). Restriction sites for EcoRI (E) and BamHI (B) are depicted. Black boxes represent CYP2D6 exons. The bar represents 1 kb. B, Southern blot genotyping of wild-type and CYP2D6 humanized mice. Tail DNA (15 µg) was digested with BamHI and probed with CYP2D6 cDNA. Hybridization signals were present only in CYP2D6 humanized mice, and their sizes were as expected from the CYP2D6 sequence. C, PCR genotyping of wild-type and CYP2D6 humanized mice. Tail DNA was amplified with mEH (internal PCR control) and CYP2D6 gene-specific primers. The PCR products (341 bp for mEH, 241 bp for CYP2D6) were separated on a 1.5% agarose gel. D, Western blot analysis of CYP2D6 protein expression in wild-type and CYP2D6 humanized mice. Liver (L), intestine (I), and kidney (K) microsomal proteins (40 µg) were separated by SDS-polyacrylamide gel electrophoresis and transferred to a nitrocellulose membrane. A CYP2D6-specific monoclonal antibody was used to assess CYP2D6 protein expression. The antibody only reacted against CYP2D6 expressed protein but did not recognize any of the mouse CYP2D proteins. Human liver microsomes (HLM) were used as a control.

DEB Pharmacokinetics in CYP2D6 Humanized Mice. Nonlinear pharmacokinetics in the elimination of CYP2D6 substrates has been reported (Brøsen and Gram, 1988; Brøsen, 1990). To determine a dose within the linear range of DEB elimination, a dose-effect study (up to 30 mg/kg) on DEB AUC was performed. We observed nonlinear pharmacokinetics of DEB elimination at doses of 10 mg/kg or higher. Therefore, the characterization of CYP2D6 humanized mouse pharmacokinetics was performed using a dose of 2.5 mg/kg, which is within the linear range of DEB elimination (data not shown). The average DEB serum concentration versus time curves in wild-type, CYP2D6 heterozygous and CYP2D6 homozygous humanized mice were determined (Fig. 2A). After a single oral dose of DEB (2.5 mg/kg), both CYP2D6 heterozygous and homozygous mice had DEB serum levels significantly lower than in wild type. Consistently, 4-OH-DEB levels are highest in CYP2D6 homozygous, intermediate in CYP2D6 heterozygous and extremely low in the wild-type (Fig. 2B).

View larger version (29K): [in this window] [in a new window]   Fig. 2.   Time course of serum concentrations of DEB (A) and 4-OH-DEB (B) from wild-type, CYP2D6 humanized heterozygous, and CYP2D6 humanized homozygous mice after one single oral administration of DEB (2.5 mg/kg). Venus blood was obtained 0, 0.5, 1, 2, 4, 6, 8, 12, and 24 h after DEB administration. Values represent the mean and the vertical lines (±S.E.M.) of DEB and 4-OH-DEB from three to four mice.

Urinary Metabolic Balance in the CYP2D6 Humanized Mice. CYP2D6 integration in the mouse genome did not affect any of the physiological parameters concerning renal function (data not shown). Twenty four hours after a single oral dose of DEB, CYP2D6 humanized mice excreted significantly higher amounts of 4-OH-DEB (28.9 ± 12.5% of dose) and lower amounts of DEB (14.6 ± 6.4%) than the wild-type mice (6.2 ± 3.1% and 61.0 ± 9.0%, respectively). Total recoveries of DEB + 4-OH-DEB were 67.2 ± 10.7% and 43.5 ± 18.9% for the wild-type and CYP2D6 humanized mice, respectively (Fig. 3). This latter finding perhaps indicates that the human CYP2D6 gene provokes metabolism of DEB to additional metabolites not detected in this study. The metabolic ratio (Mahgoub et al., 1977) for the wild-type mice fell from 9.8 to 0.5 after insertion of the human transgene.

View larger version (33K): [in this window] [in a new window]   Fig. 3.   Percentage of a single oral dose of DEB (2.5 mg/kg) excreted in urine over 24 h from wild-type, CYP2D6 humanized heterozygous and CYP2D6 humanized homozygous mice. Columns represent the mean and S.E.M of DEB and 4-OH-DEB from three to four mice. *, values of 4-OH-DEB levels from CYP2D6 humanized mice that are significantly different (p < 0.05) form wild-type mice. **, values of DEB levels from CYP2D6 humanized mice that are significantly different (p < 0.05) form wild-type mice.

Regulation of CYP2D6 Activity by Conditional Knock-Out of the HNF4 Allele. To determine the mechanism of regulation of the CYP2D6 gene, the transgenic mouse was bred with a HNF4 conditional null mouse line. The HNF4 conditional null mouse allows the study of the role of this hepatic factor in an intact model, and circumvents the embryonic lethality of a standard HNF4 null mouse (Hayhurst et al., 2001). A mouse line carrying the CYP2D6 gene and the conditional HNF4 mutation was generated (see Materials and Methods). To this end, HNF4 fl/+; AlbCre +/ mice, previously produced, were crossed with HNF4 fl/fl; CYP2D6 +/ obtained in the present work. The F1 generation of this cross yielded HNF4 fl/fl; AlbCre +/; CYP2D6 +/ and littermate control mice, HNF4 fl/fl; AlbCre /; CYP2D6 +/, HNF4 fl/fl; AlbCre +/; CYP2D6 / and HNF4 fl/fl; AlbCre /; CYP2D6 /. Genotypes of all mice were assessed by PCR of tail DNA (data not shown).

HNF4

HNF4

HNF4

HNF4

HNF4

View larger version (54K): [in this window] [in a new window]   Fig. 4.   Regulation of CYP2D6 by conditional knock-out of HNF4. A, Northern blot analysis of mouse liver HNF4 expression. Hepatic mRNA (10 µg) from HNF4; AlbCre +/; CYP2D6 +/ and littermate control mice, HNF4 fl/fl; AlbCre /; CYP2D6 +/, HNF4 fl/fl; AlbCre +/; CYP2D6 / and HNF4 fl/fl; AlbCre /; CYP2D6 / was separated on a 1% agarose gel, transferred to a nylon membrane, and hybridized with an HNF4 exon 4 and 5 cDNA probe. HNF4 was completely deleted after Cre-mediated recombination. -actin was used as a loading control. B, Western blot analysis of CYP2D6 protein expression in liver microsomal protein (40 µg) from the same animal groups of A. Deletion of HNF4 resulted in a more than 50% of CYP2D6 protein expression. The blots were re-probed with an antibody to CYP2E1 to verify loading and the presence of microsomal proteins in all the samples.

HNF4

HNF4,

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Although DEB has been extensively and successfully used as a clinical probe for the CYP2D6 polymorphism, the knowledge about its disposition in humans is limited. Most phenotyping studies have been carried out using urine excretion analysis; and during the last 20 years, only three studies of DEB pharmacokinetics have been reported (Silas et al., 1978; Sloan et al., 1983; Dalen et al., 1999). Human disposition of DEB depends on the number of functional CYP2D6 alleles present in each subject. Using individuals that carried 0, 1, 2, 3 to 4, and 13 functional alleles, it was shown that there is a gene dose effect on DEB and 4-OH DEB pharmacokinetic profile. When the number of CYP2D6 active copies increases, 4-hydroxylase activity also increases, and the plasma levels of DEB decrease, so the drug is eliminated faster (Dalen et al., 1999).

In vitro, DEB 4-hydroxylase activity in wild-type mice is almost absent (Masubuchi et al., 1997). The data presented in this study provides an in vivo confirmation for this previous finding and show that, once the CYP2D6 gene is integrated, constitutively expressed, and transmitted into the germ line, DEB 4-hydroxylase activity is clearly present. After a single oral dose of DEB, the half-life, clearance, AUC values for wild-type, heterozygous and homozygous CYP2D6 humanized mice reflect the gene dose effect found in the human population. All these data show that the humanized CYP2D6 mouse model mimics the pharmacokinetic and metabolic properties of human CYP2D6-mediated metabolism.

Clinical studies on CYP2D6-mediated metabolism are based on phenotype and genotype tests. Although phenotyping analysis reflects the real metabolic status of the patient, and genotyping analysis offers pure genetically based prediction of individual metabolic profile, there are a number of nongenetic factors (age, sex, smoking and drinking habits, and pathological status) that can affect drug metabolism and interfere with an experimentally determined phenotype or genotype (Wolf and Smith, 1999). CYP2D6 humanized mice have a tremendous potential in phenotype-genotype correlations, because these animals share the same genetic background except for the presence or absence of the human CYP2D6 gene. Because the environmental conditions through the present work have been identical for all the animals used, the phenotypic differences determined in this study are caused by the presence/absence of the CYP2D6 gene.

An intriguing and apparently unaccountable observation is the major decline in the recovery of DEB plus 4-OH-DEB in the transgenic mice compared with the wild-type mice (Table 2). The wild-type mice had a total recovery of 67.2 ± 10.7% dose, which fell to 40.9 ± 6.6% and 43.5 ± 18.9% in the heterozygous and homozygous humanized mice, respectively (unpaired t test, t = 2.90, df = 7, P = 0.02). Apparently, the fall in excretion of DEB due to CYP2D6 metabolism is not compensated by a concomitant rise in excreted 4-OH-DEB; some 25% of the dose is unaccounted for, a similar proportion of the dose as is excreted as 4-OH-DEB. The answer lies in alternative metabolic products produced by CYP2D6. Early studies (Allen et al., 1976; Idle et al., 1979) showed that DEB, in addition to 4-hydroxylation, was also metabolized to 5-, 6-, 7- and 8-hydroxy-DEB (phenolic metabolites) in humans, together with two ring-opened products, which are thought to arise from 1- and 3-hydroxylation of the drug and also are mediated by CYP2D6 (Eiermann et al., 1998). Therefore, DEB is metabolized at positions 1, 3, 4, 5, 6, 7, and 8 (i.e., at every C-H bond in the ring system). Quantitatively, the phenols represented 7.1 ± 5.8% dose eliminated in the 0- to 24-h human urine, with 7-hydroxylation predominating (Idle et al., 1979). In the female Wistar rat (Al-Dabbagh et al., 1981), 6-hydroxy-DEB (14.3 ± 1.5% dose) and an unidentified phenolic metabolite (11.3 ± 2.3% dose, calculated as a phenol; probably a dihydroxy metabolite) accounted for significant proportions of the 0- to 24-h excreted dose, compared with DEB (3 ± 0%) and 4-OH-DEB (23.7 ± 2.5%). The hydroxylations in positions 1 and 3 may contribute significantly to the excretion profile. Swedish investigators (Dalen et al., 1999) found that, as the number of functional CYP2D6 alleles increased in their human study population from 0 to 13, the total recovery of DEB and 4-OH-DEB decreased, a situation analogous to that reported here in the mouse. The early studies (Allen et al., 1976; Idle et al., 1979) had observed up to 15% dose excreted in 24 h as the ring-opened metabolites, and the Swedish study (Eiermann et al., 1998) suggests that the 1- and 3-hydroxylations may account for even more of the dose than this. However, until this point, it has not been possible to gain a true insight in vivo into relative quantitative importance of 4-hydroxylation and non-4-hydroxylation by CYP2D6. There were indications from older literature that the non-4-hydroxylation pathways may be major routes of DEB metabolism in man, rat, and dog (Allen et al., 1976; Idle et al., 1979; Al-Dabbagh et al., 1981). Using the CYP2D6 humanized mouse, it is possible to visualize that the non-4-hydroxylation of DEB by CYP2D6 may comprise about 50% of the total metabolism of this drug.

View this table: [in this window] [in a new window]   TABLE 3 DEB urine metabolic ratios from the animal groups in Figure 4A Animals were dosed with DEB (2.5 mg/kg) and placed in urine metabolic chambers for 24 h. Values from HNF4-deleted, CYP2D6 humanized mice that are significantly different (p < 0.05) from HNF4-nondeleted, CYP2D6 humanized mice (see Table 2 for key).

In addition to the pharmacokinetic study of CYP2D6 substrates, of which DEB is an archetype, the CYP2D6 humanized mouse offers us an opportunity to understand better the genetic foundation of the CYP2D6 polymorphism. The EM trait was recognized at the outset (Mahgoub et al., 1977) as the dominant phenotype, but few occasions have presented themselves for the quantification of the degree of dominance of the EM over the PM trait. The first determinations were made in family pedigrees that permitted identification of obligate heterozygotes (e.g., EM children of PM probands). The mean MR values for homozygous EMs (calculated), heterozygous EMs (observed), and PMs (observed) were 0.4, 1.9, and 33.9, respectively. These data permitted calculation of the degree of dominance of EM over PM to be approximately 30% (Evans et al., 1980). These findings were thought to explain the major phenotypic overlap between the homozygous and heterozygous EM genotypes and demonstrated just why molecular genetic methods would be required to determine the genotype reliably. However, the expression of the human transgene in the mouse background offers yet further insights. The MR values in the mouse for 0, 1, and 2 human CYP2D6 genes are 15.2 ± 7.0 (n = 6), 0.7 ± 0.2 (n = 3), and 0.5 ± 0 (n = 3). Using log-transformed data according to the method of Evans et al. (1980), these results yield a degree of dominance of EM over PM, for the human gene in the mouse, of 78% (95% confidence interval, 60 to 100%) (Fig. 5). This is significantly greater than the human estimate of 30% and may reflect the fact that transcription and translation of the CYP2D6 transgene proceeds very efficiently. Perhaps the mouse hepatic background provides a more transcriptionally active environment than the endogenous human one.

View larger version (7K): [in this window] [in a new window]   Fig. 5.   Estimation of the mean degree of dominance of the EM [transgenic] over the PM [endogenous] phenotype in mice. Horizontal double-headed arrows represent 95% confidence intervals. The vertical dotted line shows the position of the mid-point between the two homozygous genotypes (+/+) and (/), the position of zero dominance. When the mean heterozygote (+/) value falls to the left of this dotted line, then (+/+) is dominant over (/), the excursion to the left of this mid-point revealing the degree of dominance. For all calculations, only log-transformed data were used, because logMR is normally distributed, unlike its arithmetic values that are generally skewed to the right (Evans et al., 1980).

HNF4 is known to regulate the basal levels of expression of a several P450s. Promoter activity, mobility shift, and antisense targeting assays have shown that HNF4 regulates, among others, the expression of human CYP2D6 (Cairns et al., 1996; Jover et al., 2001) but not rat CYP2D5 (Lee et al., 1994) or human CYP2E1 (Liu and Gonzalez, 1995). The results from the present study provide, for the first time, in vivo confirmation of these earlier findings.

The most intriguing finding with the HNF4a conditional knock-out experiments is not that in the wild-type and CYP2D6 humanized mice the metabolic ratio (MR) rises from 18.2 ± 6.2 to 32.0 ± 2.9 (t = 3.93, df = 5, P = 0.011) and from 2.6 ± 0.2 to 3.6 ± 0.6 (t = 3.31, df = 6, P = 0.016), respectively, but that the percentage dose recovered in urine as DEB plus 4-OH-DEB fell from 54.5 ± 9.1% to 26.6 ± 2.0% (t = 6.17, df = 5, P = 0.002) and from 25.5 ± 4.9% to 13.0 ± 0.6% (t = 5.06, df = 6, P = 0.002), respectively. This situation is somewhat similar to that described above for the decrease in urinary recovery associated with insertion of the CYP2D6 transgene into the mouse. However, it is inconceivable that the conditional knock-out of HNF4a in the mouse liver, either in wild-type or humanized mice, could result in switching to alternative and undisclosed pathways of metabolism in the mouse. There is no evidence that DEB is metabolized by any P450s other than the CYP2D family. Conditional HNF4a knock-out mice harbor several biochemical abnormalities, together with hepatomegaly (Hayhurst et al., 2001). The single most dramatic biochemical change in these mice was an 18-fold elevation in serum bile acids from 16.7 ± 7 µM in wild-type mice to 283 ± 61 µM (t = 8.70, df = 6, P = 0.0001) in the HNF4 conditional null mice (conditionally hepatic HNF4a-null mice). Elevated bile acids, as is seen in biliary obstruction, for example, are known to impair renal function (Shen et al., 1990; Bomzon et al., 1997) and thus reduce urine output. The reduced urinary recovery of DEB and 4-OH-DEB, relative to controls, reported here in both the heterozygous CYP2D6 humanized mice with conditional hepatic HNF4a knock-out (fl/fl, CRE+/, 2D6+/) and wild-type mice with the same HNF4a knock-out (fl/fl, CRE+/, 2D6/), is almost certainly secondary to the massively elevated serum bile acid concentration and concomitantly reduced renal function. In human populations with reduced urine volumes caused by dehydration, for example in Saudis and Egyptians (Islam et al., 1980), the 0- to 8-h urinary recovery of DEB plus 4-OH-DEB fell from 41% in British subjects to 15 to 16% in the Arab populations.

In conclusion, we report here the first humanized P450 model that predicts human drug metabolism and reflects the polymorphic status of the gene in the human population. Moreover, we show in vivo regulation of CYP2D6 by HNF4a. This preclinical model may be of considerable interest for the identification of in vivo drug interactions and for the preclinical evaluation of novel CYP2D6 substrates or inhibitors. The CYP2D6 humanized mouse will also permit investigation into the physiological significance of CYP2D6 and its polymorphism.