The Major Genetic Defect Responsible for the Polymorphism of S-Mephenytoin Metabolism in Humans*

The metabolism of the anticonvulsant drug mepheny- toin exhibits a genetic polymorphism in humans, with the poor metabolizer trait being inherited in an autosomal recessive fashion. There are large interracial differ- ences in the frequency of the poor metabolizer phenotype, with Oriental populations having a 5-fold greater frequency compared to Caucasians. Impaired metabolism of mephenytoin and a number of other currently used drugs results from a defect in a cytochrome P450 enzyme recently identified as CYP2C19. Attempts over the past decade to define the molecular genetic basis of the polymorphism have, however, been unsuccessful. We now report that the principal defect in poor metabolizers is a single base pair (G + A) mutation in exon 5 of CYP2C19, which creates an aberrant splice site. This change alters the reading frame of the mRNA starting with amino acid 215 and produces a premature stop codon 20 amino acids downstream, which results in a truncated, non-functional protein. We further demon- strate that 7/10 Caucasian and 10/17 Japanese poor metabolizers are homozygous for this defect, indicating full-length cDNA selected which low microsomal S-mephenytoin 4’-hydroxylase ac- tivity, a high ratio of hydroxylation of the RIS enantiomers of mephenytoin and the virtual absence of CYP2C19 by immunoblotting The cDNA was in 1 x PCR buffer (67 m Tris-HC1, pH 8.8, 17 mM (NH,),SO,, 10 mM P-mercaptoethanol, 7 J ~ M EDTA, 0.2 mg/ml bovine serum albumin) containing 50 p UTP, dCTP, dGTP, and dTTP, 0.25 p PCR primers, 2.5 units of AmpliTaq DNA polymerase (Perkin-Elmer), and 1.0 mM MgC1,. The amplification was performed using a Perkin-Elmer thermocycler for 30 cycles consisting of denaturation at 94 “C for 1 min, annealing at the appropriate tempera-ture for 30 s, and extension at 72 “C for 1 min. An initial denaturation step at 94 “C for 3 min and a final extension step at 72 “C for 10 min were also performed. The PCR fragments were then subcloned into the SmaI site of pBluescript I1 SK’ (Stratagene). Plasmids were purified with Qiagen kits and sequenced with an automated sequencer, using the cycle sequencing reaction employing fluorescence-tagged dye termi-nators Applied Biosystems). were sequenced using an automated sequenator (Applied Biosystems). PCR products were purified using Microcon columns and sequenced using the same forward primer used in the PCR reaction.

be characterized as either extensive (EM)' or poor (PM) metabolizers. The latter phenotype is inherited in an autosomal recessive fashion (5, 6) with the EM phenotype comprising both the homozygous dominant and heterozygote genotypes. There are marked interracial differences in the frequency of this polymorphism. For example, the PM phenotype occurs in 2-5% of Caucasian populations but at higher frequencies (18-23%) in Oriental populations (2, 7). This polymorphism affects the metabolism of a number of other commonly used drugs, for example omeprazole (81, proguanil (9), certain barbiturates (10, 111, and citalopram (12). As a result, large interphenotypic differences occur in the disposition of these drugs, which may affect their efficacy and toxicity. The oxidation of propranolol (13), certain tricyclic antidepressants (14-161, and possibly diazepam (17) is also affected, albeit to a lesser extent.
Recent studies have shown that CYP2C19 is the enzyme responsible for the 4'-hydroxylation of S-mephenytoin in human liver and that the levels of CYP2C19 protein correlate with microsomal S-mephenytoin 4'-hydroxylase activities in human livers (18,19). However, the molecular basis of the PM phenotype is not known. The purpose of the present study was to determine the molecular genetic mechanism of the defect that is responsible for the polymorphism of S-mephenytoin metabolism in humans.
MATERIALS AND METHODS Analysis of Human Liver Microsomes-Liver microsomes were prepared by differential centrifugation from 13 human liver samples selected from organ donors that had been previously characterized in vitro (20) as varying markedly in their S-mephenytoin 4'-hydroxylase activity. For immunoblot analysis of CYP2C19, liver microsomal proteins were separated by SDS-polyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, and developed with a polyclonal antibody to CYP2C9 that also recognizes CYP2C19 using the ECL chemiluminescence kit (Amersham Corp.) as previously described (19). Results were confirmed with a specific peptide antibody to CYP2C19. Liver microsomal R-and S-mephenytoin 4'-hydroxylase activities were measured by high performance liquid chromatography analysis (21).
Amplification of CYP2C19 mRNA-Total liver RNA was isolated from liver samples using a single-step method (22) with Tri-reagent (Molecular Research Center Inc.) and reverse-transcribed as previously described (19). In initial experiments, the polymerase chain reaction (PCR) was used to amplify overlapping CYP2C19 cDNA fragments encompassing the full-length cDNA from three selected human liver samples, which had low microsomal S-mephenytoin 4'-hydroxylase activity, a high ratio of hydroxylation of the R I S enantiomers of mephenytoin (201, and the virtual absence of CYP2C19 by immunoblotting as described above. The cDNA was amplified in 1 x PCR buffer (67 m Tris-HC1, pH 8.8, 17 m M (NH,),SO,, 10 m M P-mercaptoethanol, 7 J~M EDTA, 0.2 mg/ml bovine serum albumin) containing 50 p UTP, dCTP, dGTP, and dTTP, 0.25 p PCR primers, 2.5 units of AmpliTaq DNA polymerase (Perkin-Elmer), and 1.0 m M MgC1,. The amplification was performed using a Perkin-Elmer thermocycler for 30 cycles consisting of denaturation at 94 "C for 1 min, annealing at the appropriate temperature for 30 s, and extension at 72 "C for 1 min. An initial denaturation step at 94 "C for 3 min and a final extension step at 72 "C for 10 min were also performed. The PCR fragments were then subcloned into the SmaI site of pBluescript I1 SK' (Stratagene). Plasmids were purified with Qiagen kits and sequenced with an automated sequencer, using the cycle sequencing reaction employing fluorescence-tagged dye terminators (PRISM, Applied Biosystems).
RNA from all 13 human liver donors was subsequently reverse-transcribed and amplified using the forward primer 5"ATTGAATGU-The abbreviations used are: EM, extensive metabolizer; PM, poor metabolizer; PCR, polymerase chain reaction; bp, base pair(s).

Polymorphism in Mephenytoin Hydroxylation
CATCAGGATTG-3' and the reverse primer 5"GTAAGTCAGCTG-CAGTGATTA-3' and the strategy shown in Fig. 1 to detect aberrant splicing of exon 5. PCR conditions were similar to those described above. PCR products were analyzed on 3% agarose gels and stained with ethidium bromide. Selected PCR products were sequenced directly, after purification using Microcon columns (Amicon).
Phenotyping Procedures-The in vivo phenotype of most of the Swiss subjects was based on their "hydroxylation index" values (2), where a value above 5.6 identifies a PM. The phenotype of American subjects, Japanese subjects, and one Swiss subject was based on the urinary SIR ratio as described previously (3), with a poor metabolizer being defined as having a ratio > 0.9. One American subject was of a rare intermediate phenotype characterized by the extent of 4'-hydroxylation being greater than in PMs, but with the rate of the metabolite's formation being slower than in EMS (23).
Genotyping Procedure-DNA was isolated (24) from human blood of selected Caucasian and Japanese subjects who had been previously phenotyped as described above. These populations contained an intended overrepresentation of PMs. PCR conditions were similar to those described previously, except that reactions used 200 ng of genomic DNA, 3 mM MgCI,, and an initial denaturation a t 94 "C for 5 min. The forward primer was 5'-AATTACAACCAGAGC'I"GGC-3' and the reverse primer 5'-TATCACT"TCCATAAAAGCAAG-3'. PCR products were restricted with SmaI in the PCR buffer, without purification. Digested PCR products were analyzed on 4% agarose gels stained with ethidium bromide. PCR products of genomic DNA from three individuals who were homozygous-extensive, heterozygous-extensive, and homozygouspoor metabolizers (based on their SmaI restriction digests and their in vivo phenotypes) were sequenced using an automated sequenator (Applied Biosystems). PCR products were purified using Microcon columns and sequenced using the same forward primer used in the PCR reaction.

RESULTS AND DISCUSSION
We initially amplified and sequenced overlapping CYP2C19 cDNA fragments from liver samples of selected human organ donors, which had low microsomal S-mephenytoin 4"hydroxylase activity, a high ratio of hydroxylation of the R IS enantiomers of mephenytoin (20), and the virtual absence of CYP2C19 by immunoblotting. One aberrant CYP2C19 cDNA fragment was identified with a 40-bp deletion at the beginning of exon 5 (from bp 643 to bp 682), which included the deletion of a SmaI restriction site. This alteration results in the deletion of amino acids 215-227 and shifts the reading frame beginning a t amino acid 215, producing a premature stop codon 20 amino acids downstream. The resultant truncated 234-amino acid protein would lack the heme-binding region and, therefore, would be catalytically inactive.
RNA was then reverse-transcribed from a total of 13 liver donors, which were selected based on a wide range of S-mephenytoin hydroxylase activities as determined in vitro, and the PCR strategy shown in The genomic sequence of CYP2C19 is not currently known; hence, primers for the intron 4lexon 5 junction were developed empirically. This involved the use of multiple primers for intron 4 based on the sequence of this region in CYP2C9 (25), which is a closely related gene that shows 95% similarity to CYP2C19 in the upstream region and several exons: and a specific reverse primer for exon 5 of CYP2C19. One primer pair amplified a DNA fragment with the correct predicted size in both EMS and PMs; however, only the fragment from EMS could be digested with SmaI. Sequencing of this fragment yielded sequence information from which a specific forward primer was generated and used in subsequent PCR reactions to genotype individuals.
DNA from 28 unrelated Swiss and American Caucasian subjects whose phenotype had been established in vivo was amplified using specific primers and restricted with SmaI using the strategy outlined in Fig. 2 A . Fig   Oriental populations have a much greater frequency of the mephenytoin PM phenotype compared to Caucasians (2, 3, 7). Accordingly, we analyzed DNA from 29 unrelated Japanese subjects (Fig. 2 0 ) . Eight of the 12 EMS were homozygous and 4 were heterozygous for CYP2C19,,,,. CYP2C19,, accounted for a similar percentage (74%) of the alleles (25 of 34) in Japanese PMs as found in Caucasian PMs. Ten of 17 PMs were homozygous for the mutant allele, and 5 were heterozygous. Thus, the major mutation responsible for the PM phenotype in Japanese is identical to that found in Caucasians. However, the defect was not present in all PMs regardless of whether they were Caucasian or Japanese. It is therefore likely that additional mutations exist, which result in the PM phenotype in both populations. In a similar manner, a point mutation a t a splice site consensus sequence is the single most common mutation in CYP206, accounting for >75% of mutant alleles in PMs of debrisoquine (26,27). However, several minor mutant alleles have also been identified that contribute to this phenotype.
We also genotyped a Japanese family that had been studied previously with respect to the inheritance of the PM trait ( 5 ) (Fig. 2B). There was complete concordance between the CYP2C19 genotype and the in vivo phenotype consistent with the previously reported Mendelian autosomal recessive mode of inheritance (5, 6).
DNA from individuals representative of the two CYP2C19 Mephenytoin Hydroxylation

Aberrant (XlSm)
~---~---~---" "~" -~ " " . I"fro"d"""-" " " " " " " " _" genotypes was amplified as described above and then directly sequenced (Fig. 3). The sequence information verified that only CYP2C19 was amplified in the genotyping test. Surprisingly, the sequence of intron 4 of the defective gene was identical to that of the normal gene. The only difference was a G + A substitution in the coding sequence of exon 5 of CYP2C19, corresponding to position 681 in the cDNA. This change produces a cryptic splice site in the exon, which shows a similar degree of homology with the mammalian 3"splice site consensus sequence (28) as the normal 3"splice site. There are also potential branch points near this cryptic splice site. The apparently complete selection of the cryptic site is somewhat surprising, and the reasons for this are not clear. Comparison of the genotypes of liver samples 13 (extensive), 21 (intermediate), and 35 (poor) with their cDNA analysis patterns ( Fig. 1) indicated complete agreement between the genotype and splicing pattern. Interestingly, cDNA from CYP2C8 and CYP2C18 have potential 3"splice sites at the same position in exon 5 (291, yet the full-length CYP2C8 protein is present in human liver microsomes (19), indicating that this gene is spliced correctly. Moreover, we have amplified the exon 4exon 5 junction of the cDNA for CW2C18 and CYP2C8 and have seen no evidence for abnormal splicing (data not shown). However, homology to the mammalian consensus sequence for the introdexon junctions and branch points is not the sole determinant of splice site selection. For example, higher order RNA structure can also facilitate cleavage preferentially at a particular site (30) and could explain the preferential selection of the cryptic CYP2C19 splice site.

f t a a t . a a f f a t t g f t t t~t~t t~~t , t g~, * t~* t t t t~~~~~t~t~~t t g~t t . t t t~~~~~A A C C C A l A
In conclusion, the present study identifies the primary genetic defect (CYP2C19,) that is responsible for the poor metabolism of mephenytoin and which also affects metabolism of several other widely used drugs (8)(9)(10)(11)(12)(13)(14)(15)(16)(17). CYP2C19, accounts for 75% of the defective alleles in both Caucasian and Japanese PMs. The defect consists of a single base pair substitution in exon 5 of CYP2C19, which produces a cryptic 3"splice site, resulting in an aberrantly spliced mRNA and the absence of the CYP2C19 protein in livers of PMs. We have developed a simple PCR-based genetic test for the defective CYP2C19, allele, which will be useful in clinical studies investigating the importance of this genetic defect in drug metabolism in humans.