Abstract
CYP2C19 is a highly polymorphic enzyme that affects the metabolism of a wide range of therapeutic drugs. Almost all the identified alleles of CYP2C19 are derived from nonsynonymous single nucleotide polymorphisms (nsSNPs). The objective of this study was to functionally characterize 20 nsSNPs of CYP2C19, distributed throughout the entire coding region, most of which have not been thoroughly characterized. cDNAs of these variants were constructed and expressed in yeast cells. All variants had similar levels of apoprotein and holoprotein expression, except for CYP2C19.16 and D360N, which had significantly lower holoprotein levels than the wild-type (WT) CYP2C19 enzyme, and CYP2C19.5B, which showed only apoprotein. The activity of the CYP2C19 variants was investigated using two substrates, S-mephenytoin and omeprazole, and six different kinetic parameters were measured. CYP2C19.5B, CYP2C19.6, and CYP2C19.8 were found to be catalytically inactive. The entire dataset of the remaining 17 variants, together with the WT, was analyzed by multivariate analysis. This analysis indicated that CYP2C19.9, CYP2C19.10, CYP2C19.16, CYP2C19.18, CYP2C19.19, A161P, W212C, and D360N were substantially altered in catalytic properties in comparison with the WT, with each of these variants exhibiting either dramatically decreased catalytic activities or higher Km values. These results not only generally confirmed the function of previously reported variants but also identified additional reduced-function variants. These findings will greatly extend our understanding of CYP2C19 genetic polymorphisms in humans as well as facilitate the structure-function study of the CYP2C19 protein.
Introduction
CYP2C19 is a clinically important enzyme that plays a critical role in the metabolism of a wide variety of drugs, such as the antiulcer drugs omeprazole and lansoprazole, the anticonvulsant S-mephenytoin, the tricyclic antidepressant imipramine, the antimalarial drug proguanil, the antidiabetic drug tolbutamide, the anxiolytic drug diazepam, and the antiplatelet drug clopidogrel (Wedlund, 2000; Rosemary and Adithan, 2007; Sibbing et al., 2010). Large interindividual differences have been observed in the metabolism of these drugs in vivo, and individuals can be divided into extensive metabolizers and poor metabolizers (PMs). PMs often experience higher plasma drug concentrations and altered therapeutic responses (Kita et al., 2001; Rosemary and Adithan, 2007). In addition, there are marked interethnic differences in the distribution of the PM phenotype. For instance, only 2 to 5% of whites are PMs, compared with 13 to 23% of Asians (Wedlund, 2000). These differences can primarily be attributed to CYP2C19 genetic polymorphisms (de Morais et al., 1994a,b). To date, 38 important CYP2C19 SNPs, from which 28 haplotypes were identified, are presented on the CYP2C19 allele nomenclature Web site (http://www.cypalleles.ki.se/cyp2c19.htm). Of these, four are in the noncoding region and 34 are in the coding region, all causing amino acid changes in the CYP2C19 protein. The most predominant genetic defects responsible for the PM phenotype result from two SNPs, CYP2C19*2 (681G >A) and CYP2C19*3 (636G >A), which lead to a splicing defect and a premature stop codon, respectively, and therefore produce truncated proteins without enzyme activity (de Morais et al., 1994a,b). In studies of PMs across global populations, these two SNPs have been proposed to explain anywhere from less than 50% to more than 90% of the PM phenotype (Nakamoto et al., 2007). Other null or reduced-function alleles (*4–*10,*12,*16,*26, and *27) that may contribute to the PM phenotype have also been characterized. CYP2C19*17 (−808C>T) was ascertained to cause an increase in transcriptional activity, resulting in extensive metabolism of CYP2C19 substrates (Sim et al., 2006; Rudberg et al., 2008). These genetic variations not only affect the pharmacokinetic behavior of CYP2C19-mediated drugs but also potentially lead to differences in drug response and altered risk of adverse drug reactions. For instance, among persons treated with clopidogrel, carriers of CYP2C19 reduced-function alleles had significantly lower levels of the active metabolite of the drug, diminished platelet inhibition, and a higher rate of subsequent cardiovascular events than did noncarriers (Mega et al., 2009; Simon et al., 2009). In contrast, the CYP2C19*17 allele has been significantly associated with an enhanced response to clopidogrel and an increased risk of bleeding (Sibbing et al., 2010). Therefore, identification and functional characterization of CYP2C19 genetic polymorphisms is of great importance for better and safer drug therapy.
S-Mephenytoin and omeprazole are two commonly used probe substrates in CYP2C19 phenotyping studies. To date, the three-dimensional structure of CYP2C19 has not been resolved, and the model for binding between S-mephenytoin or omeprazole and CYP2C19 has not been established. However, by means of homology modeling or site-directed mutagenesis, previous studies (Ibeanu et al., 1996; Tsao et al., 2001) have indicated that the amino acid residues that determine substrate specificity for S-mephenytoin 4′-hydroxylation are distinct from those that confer omeprazole 5′-hydroxylation activity. This observation suggests that the same mutation could have different effects on the metabolism of S-mephenytoin and omeprazole. Therefore, to evaluate the functional consequences of CYP2C19 genetic polymorphisms, it would be beneficial to use two or more specific substrates.
Previous studies of CYP2C19 genetic variants have mainly focused on the CYP2C19*2 and CYP2C19*3 polymorphisms. To fully understand the functional consequences of CYP2C19 genetic polymorphisms, functional characterization studies should encompass additional variants. In the present study, using a valid yeast expression system established in our laboratory (Gao et al., 2010; Liu et al., 2010; Zi et al., 2010), we constructed a large enzyme panel containing 20 yeast-expressed CYP2C19 nsSNP variants and thoroughly evaluated the effects of these variants on the catalytic properties toward S-mephenytoin and omeprazole. Among the 20 nsSNPs examined in this study, 10 are functionally characterized in vitro for the first time. Whereas the other 10 have been previously evaluated in vitro, complete Michaelis-Menten kinetics have only been assessed with at least one substrate in four cases. The results obtained from this study should greatly extend our understanding of the interindividual variability in CYP2C19-mediated drug metabolism.
Materials and Methods
Materials.
The protease-deficient Saccharomyces cerevisiae strain BJ5457 integrated with human NADPH cytochrome P450 reductase cDNA was obtained from Lifegen Co. Ltd. (Xi'an, China). The galactose-inducible expression vector pYES2/CT and anti-V5 antibody were obtained from Invitrogen (Carlsbad, CA). Bradford reagent, β-NADP+, d-glucose 6-phosphate, glucose-6-phosphate dehydrogenase, and omeprazole were purchased from Sigma-Aldrich (St. Louis, MO). S-Mephenytoin, 4′-hydroxymephenytoin, and S-(+)-nirvanol were obtained from BD Gentest (Woburn, MA). 5′-Hydroxyomeprazole was purchased from Cayman Chemical (Ann Arbor, MI). All other chemicals and reagents were of the highest quality available commercially.
Construction of CYP2C19 Expression Vector.
The full-length cDNA of CYP2C19 (GenBank Accession No. NM_000769) was obtained from human liver total RNA by the reverse transcriptase-polymerase chain reaction (PCR) method using specific primers. The forward primer was 5′-TCC GGT ACC ATG GAT CCT TTT GTG GTC CTT G-3′ and the reverse primer was 5′-ACA CTC GAG GAC AGG AAT GAA GCA CAG-3′. The underlined sequences are introduced KpnI and XhoI restriction sites, respectively. PCR products were cloned into the KpnI and XhoI sites of the expression vector, pYES2/CT. The wild-type CYP2C19*1A cDNA-containing plasmid construct served as the template for introducing mutations by site-directed mutagenesis PCR. All of the CYP2C19 cDNA-containing plasmid constructs were sequenced to confirm successful mutagenesis. The 20 variants of CYP2C19 that were constructed are listed in Table 1.
Expression of CYP2C19 Enzymes and Preparation of the Yeast Microsomal Fraction.
CYP2C19 cDNA-containing plasmids were transformed into yeast cells. CYP2C19 protein expression in yeast cells and subsequent membrane preparation were performed as described previously (Gao et al., 2010; Liu et al., 2010). Microsomal proteins were quantified by the Bradford assay using bovine serum albumin as a standard, according to the manufacturer's protocol (Sigma-Aldrich).
Assay for CYP2C19 Holo- and Apoproteins.
Total functional CYP2C19 protein levels (holoprotein) were measured by the reduced CO difference spectrum method (Omura and Sato, 1964) using a UV-visible spectrophotometer (Shimadzu, Tokyo, Japan). Total CYP2C19 protein levels of holo and apo forms in yeast cell microsomes were determined by semiquantitative immunoblot analysis. In brief, microsomal fractions (10 μg of microsomal protein) were separated by 10% SDS-polyacrylamide gel electrophoresis. Mouse anti-human V5 antibody (diluted at 1:15,000) and peroxidase-conjugated goat anti-mouse IgG (diluted at 1:300,000) were used as the primary and secondary antibodies, respectively. Immunoreactive proteins were visualized by a SuperSignal West Femto Trial Kit (Thermo Fisher Scientific Inc., Rockford, IL), and the band densities were determined with BIO-1D software (SIM International Group Co., Newark, NJ). The relative protein level of each detected microsomal sample was expressed as the ratio of its band density to that of CYP2C19.1A (i.e., the relative protein level of CYP2C19.1A was taken as 100%). All the assays were performed in triplicate with independent microsomal preparations.
S-Mephenytoin 4′-Hydroxylation Activity Assay.
For the enzymatic activity assay of recombinant CYP2C19s, the reaction mixture was composed of recombinant CYP2C19 protein, substrate, an NADPH-regenerating system (consisting of 1.3 mM NADP+, 3.3 mM glucose 6-phosphate, and 0.4 U/ml glucose-6-phosphate dehydrogenase), 3.3 mM MgCl2, and 50 mM potassium phosphate buffer (pH7.4). For the S-mephenytoin 4′-hydroxylation activity assay, all of the incubations were performed under the linearity range with respect to microsome protein concentration (0.125–1.0 mg/ml) and incubation time (0–60 min). To determine the kinetic constants for S-mephenytoin 4′-hydroxylation by CYP2C19s, 2-fold serial diluted S-mephenytoin (6.25–200 μM) was incubated with 0.5 mg/ml recombinant proteins in a final volume of 200 μl of reaction mixture. S-Mephenytoin was dissolved in acetonitrile. The final concentration of acetonitrile in the incubation mixture was 1% or less. The reaction was initiated by addition of the NADPH-regenerating system and terminated after 40 min by addition of 100 μl of iced acetonitrile containing 8 μM S-nirvanol as an internal standard. The reaction mixture was extracted twice with dichloromethane. After centrifugation at 12,000g for 10 min, the organic phase was evaporated to dryness under a gentle stream of nitrogen at 35°C. The residue was dissolved in 100 μl of acetonitrile-water (40:60, v/v) and analyzed using the Agilent 1200 Series HPLC system (Agilent Technologies, Santa Clara, CA) with an HC-C18 column (5 μm, 4.6 × 250 mm; Agilent Technologies) at 30°C. Elution was monitored at 204 nm using a mobile phase of 28% acetonitrile at a flow rate of 0.8 ml/min. Standard curve samples were prepared in the same manner as incubation samples.
Omeprazole 5′-Hydroxylation Activity Assay.
Omeprazole 5′-hydroxylation assays were performed using the same incubation system described above in a total volume of 100 μl. The linearity range for omeprazole 5′-hydroxylation activity of CYP2C19s was between 0.125 and 1.0 mg/ml protein and at least up to a 30-min incubation time at 37°C. For the kinetic analysis, 2, 5, 10, 20, 50, and 100 μM omeprazole was incubated with 0.5 mg/ml recombinant proteins in a final volume of a 100-μl reaction mixture. Omeprazole was dissolved in methanol. The final concentration of methanol in the incubation mixture was 1% or less. Each reaction was terminated by the addition of 50 μl of iced acetonitrile after a 30-min incubation time. The incubation mixture was centrifuged at 12,000g for 15 min, and then an aliquot of the supernatant was analyzed by the Agilent 1200 Series HPLC System with a HC-C18 column (5 μm, 4.6 × 150 mm, Agilent Technologies) at 30°C. Elution was monitored at 302 nm using a mobile phase consisting of 0.1% formic acid with 20% acetonitrile at a flow rate of 0.8 ml/min. Standard curve samples were prepared in the same manner as incubation samples.
Data Analysis and Statistics.
Kinetic parameters such as Km and Vmax were estimated by analyzing Michaelis-Menten plots using Prism software (version 4.0; GraphPad Software, Inc., San Diego, CA). Intrinsic clearance (Clint) values were determined as the ratio of Vmax/Km. All values are expressed as the mean ± S.D. of three separate experiments derived from independent microsomal preparations.
Statistical comparisons were performed to assess the differences between wild-type and variant enzymes. To evaluate the differences in one independent variable, one-way analysis of variance with Dunnett's test was used, with P < 0.05 being considered statistically significant. The overall trend governing the differences between the polymorphic variants and the WT in the entire set of kinetic parameters was globally analyzed by multivariate analysis with principal component analysis (PCA) (Urban et al., 2009). All the enzyme kinetic data (the Km, Vmax, and Clint for the two substrates) were preprocessed by dividing the values of the initial data set by the row-column double variance: Xnorm = (Xobs/Varcol/Varrow). This type of double normalization is appropriate when different enzyme preparations are being compared and when different types of substrates and detection methods are used. The PCA was performed with the SPSS statistical program (SPSS Inc., Chicago, IL).
Computational Prediction of the Functional Effect of CYP2C19 nsSNPs.
In addition to the in vitro assessment, the PANTHER and PolyPhen software programs were used to predict the effects of selected CYP2C19 nsSNPs on protein function. PANTHER is based on an alignment of evolutionarily related proteins. The variability of the particular amino acid positions at which the variants occur in evolutionarily related proteins (output as a substitution position-specific evolutionary conservation score, subPSEC) was calculated to estimate the probability that a given coding variant will cause a deleterious functional change in a PANTHER hidden Markov model (Thomas et al., 2003). PolyPhen takes account of sequence, phylogenetic, and structural information to estimate the impact of an amino acid replacement on the three-dimensional structure and function of the protein (Ramensky et al., 2002).
A three-dimensional homology model of CYP2C19 was constructed using the automatic modeling mode on SWISS-MODEL (http://swissmodel.expasy.org). The CYP2C9 crystal structure (Protein Data Bank code: 1OG2) was used as the template for CYP2C19 homology modeling because of the high sequence identity (∼91%) (Wada et al., 2008). Iron protoporphyrin IX was added to the CYP2C19 protein using the X-ray structure data of CYP2C9. PyMOL (v1.3r1; DeLano Scientific LLC, South San Francisco, CA; http://pymol.org/) was used for molecular visualization and measurement of hydrogen bonds.
Results
Expression of Wild-Type and Variant CYP2C19s in Yeast Cells.
Immunoblot analysis indicated that the immunoreactive protein expression levels were similar for CYP2C19.1A and the 20 variant CYP2C19s. The CO difference spectra of all recombinant CYP2C19s were also determined to quantify the CYP2C19 holoprotein content. The holoprotein level of CYP2C19.1A was 56.16 ± 10.14 pmol/mg microsomal protein. Most variants showed comparable or slightly lower holoprotein levels than CYP2C19.1A, except for CYP2C19.5B, CYP2C19.16, and D360N. CYP2C19.5B showed a total absence of the 450-nm peak in microsomal protein. For the CYP2C19.16 and D360N constructs, a significantly reduced 450-nm peak and a large 420-nm peak were detected (Fig. 1). The percentage expression levels of CYP2C19 holo- and apoprotein, for each variant compared with CYP2C19.1A, are shown in Fig. 2.
Enzymatic Properties for S-Mephenytoin 4′-Hydroxylation by Wild-Type and Variant CYP2C19s.
The S-mephenytoin 4′-hydroxylation activities of the recombinant enzymes were determined at two concentrations, one near the Km of CYP2C19 (25 μM) and one near the Vmax of the enzyme (200 μM). For CYP2C19.5B and CYP2C19.6, no S-mephenytoin 4′hydroxylation activity was detected even at the highest microsomal concentration of 2 mg/ml and at the highest substrate concentration. CYP2C19.8 only showed negligible activity (approximately 4% of CYP2C19.1A) at 200 μM. Therefore, these three variants were excluded from subsequent kinetic analysis. CYP2C19.9, CYP2C19.10, CYP2C19.13, CYP2C19.16, CYP2C19.18, CYP2C19.19, E92D, M74T, A161P, E122A, W212C, and D360N exhibited significantly decreased activities compared with CYP2C19.1A at 25 μM; the same trend was seen at 200 μM with the exception of CYP2C19.13, CYP2C19.19, and E122A, which showed activities comparable to those of CYP2C19.1A (Table 2).
The Michaelis-Menten kinetics for S-mephenytoin 4′-hydroxylation by the catalytically active enzymes were determined (Supplemental Fig. 1a). The kinetic parameters are summarized in Table 3. CYP2C19.1A had a Km value of 30.72 ± 2.93 μM, which falls within the range observed with human liver microsomes (Uttamsingh et al., 2005) and other recombinant CYP2C19s in previous studies (Blaisdell et al., 2002; Hanioka et al., 2007). The apparent affinities of CYP2C19.9, CYP2C19.10, CYP2C19.18, CYP2C19.19, A161P, D360N, and W212C for S-mephenytoin were lowered, with significantly higher Km values (ranging from 1.57- to 2.98-fold) than that of CYP2C19.1A, whereas the apparent affinity of E92D was increased, with a 1.72-fold lower Km value than that of CYP2C19.1A. The Vmax values of CYP2C19.10, CYP2C19.16, A161P, D360N, and W212C were significantly decreased, whereas those of CYP2C19.15 and CYP2C19.19 were significantly increased. CYP2C19.9, CYP2C19.10, CYP2C19.16, A161P, D360N, and W212C demonstrated a significant decrease in their Clint values, which ranged from 10 to 39% of that of CYP2C19.1A. CYP2C19.18 and CYP2C19.19 also showed a moderate decrease in Clint (approximately 53 and 60% of that of CYP2C19.1A, respectively).
Enzymatic Properties for Omeprazole 5′-Hydroxylation by Wild-Type and Variant CYP2C19s.
Similar to the S-mephenytoin 4′-hydroxylation assays, the omeprazole 5′-hydroxylation activities of the CYP2C19s were initially determined at two concentrations, one near the Km of CYP2C19 (5 μM) and one near the Vmax of the enzyme (100 μM). CYP2C19.5B and CYP2C19.6 showed no omeprazole 5′hydroxylation activity, even at the highest microsomal concentration of 2 mg/ml and at the highest substrate concentration. The catalytic activity of CYP2C19.8 (approximately 12% of that of CYP2C19.1A) was only detected at 100 μM omeprazole. Therefore, these three variants were excluded from the kinetic analysis. Of the remaining variants, only CYP2C19.16, A161P, and W212C exhibited significantly decreased activities compared with those of CYP2C19.1A at both substrate concentrations (Table 2).
The Michaelis-Menten kinetics for omeprazole 5′-hydroxylation by WT and variant CYP2C19 enzymes were determined (Supplemental Fig. 1b). As shown in Table 3, CYP2C19.1A had a Km value toward omeprazole of 5.42 ± 1.43 μM, which was also comparable with what was observed in human liver microsomes (Shu et al., 2000) and other recombinant CYP2C19s in previous studies (Hanioka et al., 2008) (Lee et al., 2009). The Km values of CYP2C19.9, CYP2C19.10, CYP2C19.19, E92D, E122A, W212C, and D360N were significantly higher (ranging from 1.44- to 2.68-fold) than that of CYP2C19.1A, which indicated that the apparent affinity of these variants for omeprazole was decreased. In contrast, CYP2C19.1A, CYP2C19.16, A161P, D360N, and W212C showed significantly decreased Vmax values. The Clint values of CYP2C19.9, CYP2C19.10, CYP2C19.16, CYP2C19.19, A161P, D360N, E122A, and W212C were significantly reduced, to 12 to 44% of those for CYP2C19.1A. E92D and F168L also showed a moderately decreased Clint (approximately 53 and 63% of CYP2C19.1A, respectively).
Principal Component Analysis of the Differences between Wild-Type and Variant CYP2C19s.
Principal component analysis, a procedure that transforms a high number of possibly correlated variables into a reduced number of uncorrelated variables denoted by principal components, was used to project the present data from a six-dimensional space (i.e., the three kinetic parameters for each of the two substrates) into a two-dimensional space for ease of visualization (Fig. 3.). After two-variance normalization, the first analysis was by a Scree plot analysis (data not shown). This analysis indicated that the two first principal components retained 90% of the initial variance; therefore, these two principal components explain almost all the significant variations in the data. The analysis seems to indicate that the trend of variation that governs our dataset stems from the difference between those enzymes with WT-like activity (CYP2C19.1A, CYP2C19.1B, CYP2C19.11, CYP2C19.13, CYP2C19.14, CYP2C19.15, E92D, M74T, E122A, and F168L) and non-WT-like enzymes (CYP2C19.9, CYP2C19.10, CYP2C19.16, CYP2C19.18, CYP2C19.19, A161P, W212C, and D360N). The non-WT-like enzymes can be further divided into two subgroups. One subgroup contains CYP2C19.9, CYP2C19.18, CYP2C19.19, and D360N, characterized by having higher Km values and/or moderately decreased catalytic activities (with Clint values from 33 to 60% of those of CYP2C19.1A for S-mephenytoin and from 36 to 88% of those of CYP2C19.1A for omeprazole); the other subgroup consists of CYP2C19.10, CYP2C19.16, A161P, and W212C, which all demonstrate extremely decreased catalytic activities (with Clint values from 10 to 16% of those of CYP2C19.1A for S-mephenytoin and from 12 to 34% of those of CYP2C19.1A for omeprazole).
Prediction of the Functional Effects of CYP2C19 nsSNPs by PANTHER and PolyPhen.
According to the PANTHER and PolyPhen programs, 12 of the 20 CYP2C19 nsSNPs, namely L17P, S51G, M74T, W120R, R132Q, R144H, A161P, W212C, P227L, R410C, R433W, and R442C, were predicted to have deleterious effects on protein function, giving either a significantly low subPSEC score (less than −3, with a corresponding Pdeleterious score greater than 0.5), a high PSIC score, or both (Supplemental Table 1). In contrast, the other eight nsSNP mutations were predicted to have benign effects on protein function.
Discussion
In the present study, using a valid yeast expression system established in our laboratory (Gao et al., 2010; Liu et al., 2010; Zi et al., 2010), we constructed a large enzyme panel containing 20 yeast-expressed human CYP2C19 nsSNP variants and examined the functional consequences of CYP2C19 genetic polymorphisms. Of the 20 variants, the effects on CYP2C19 activity of 10 variants (CYP2C19.1B, CYP2C19.5, CYP2C19.6, CYP2C19.8, CYP2C19.9, CYP2C19.10, CYP2C19.11, CYP2C19.13, CYP2C19.18, and CYP2C19.19) have been experimentally characterized previously (Ibeanu et al., 1998a,b, 1999; Blaisdell et al., 2002). However, the full kinetic profiles, determined using two probe substrates, have only been reported for four variants (CYP2C19.1B, CYP2C19.10, CYP2C19.18, and CYP2C19.19). For the other 10 variants (CYP2C19.13, CYP2C19.14, CYP2C19.16, M74T, E92D, E122A, A161P, F168L, W212C, and D360N), in vitro functional data have not been provided before. In this study, the full kinetic profiles of the 20 variants were extensively evaluated, using two specific substrates, S-mephenytoin and omeprazole.
We confirmed the expression of cytochrome P450 holo- and apoprotein of recombinant CYP2C19 enzymes by reduced CO difference spectral and immunoblot analyses. The immunoreactive protein levels of WT and variant proteins were similar, indicating that these single amino acid substitutions hardly affected the translation level of CYP2C19. Most of the variants also showed holoprotein levels comparable to those of the wild type, with the exception of CYP2C19.5B, CYP2C19.16, and D360N, which showed nonexistent or significantly lowered holoprotein levels. These three residues (Arg433, Arg442, and Asp360, respectively) are located in or near the heme binding region (Supplemental Fig. 2), suggesting that these amino acid changes could alter the stability and/or folding efficiency of the CYP2C19 holoprotein, thus resulting in a large proportion of the functional protein being denatured (Johansson et al., 1994; Ibeanu et al., 1998a).
Consistent with previous findings (Ibeanu et al., 1998a,b, 1999), CYP2C19.5, CYP2C19.6, and CYP2C19.8 showed abolished or negligible catalytic activities toward S-mephenytoin and omeprazole, indicating that these are inactive variants. To observe the overall trend governing the differences between the remaining 17 catalytically active variants and CYP2C19.1A in their kinetic behavior, principle component analysis was performed on the entire set of kinetic data. This analysis showed that CYP2C19.1B, CYP2C19.11, CYP2C19.13, CYP2C19.14, CYP2C19.15, E92D, M74T, E122A, and F168L are similar to CYP2C19.1A, indicating that these mutations may have little effect on CYP2C19 catalytic properties. This result was generally consistent with findings from a previous study (Blaisdell et al., 2002) and with prediction analysis by PANTHER and PolyPhen (Supplemental Table 1), which predicted that almost all these amino acid substitutions would be benign.
The PCA also revealed that CYP2C19.9, CYP2C19.10, CYP2C19.16, CYP2C19.18, CYP2C19.19, A161P, W212C, and D360N significantly differed from the WT in catalytic properties. Of these, CYP2C19.10, W212C, CYP2C19.16, and A161P are the most distinguishable variants, because of their dramatically decreased intrinsic clearance (each less than approximately 30% of CYP2C19.1A) toward both S-mephenytoin and omeprazole. These experimental data also correlated well with the results computationally predicted by PANTHER and PolyPhen, which indicated that these substitutions may cause impairment of protein function (Supplemental Table 1).
Previous studies (Blaisdell et al., 2002; Lee et al., 2009) found that CYP2C19.10 (P227L) largely decreased the apparent affinity and catalytic efficiency of the enzyme for S-mephenytoin and omeprazole; this decrease was reproduced in the current study. Pro227, a residue that is highly conserved in the CYP2 family, is located within the F-G loop (Supplemental Fig. 2), which forms part of the substrate access channel in CYP2C19 and is important for substrate specificity and activity (Ibeanu et al., 1996; Tsao et al., 2001; Wada et al., 2008). This proline forms a tight turn between helices G′ and G and hence its replacement by leucine, which is a highly stabilizing residue within an α-helix (Fersht, 1999), would be expected to destabilize the F-G loop conformation, resulting in reduced activity toward substrate. Similar to Pro227, Trp212 is also located in the F-G loop and is adjacent to Gln214, which is the active residue that forms a hydrogen bond to omeprazole (Lewis, 2002). The effect observed in W212C suggests that substitution of the aromatic residue by Cys may affect the substrate access channel, thus significantly impairing protein function.
CYP2C19.16 (R442C) exhibited reduced catalytic activities in the present study, and it also showed a lower in vivo capacity for CYP2C19-mediated 4′-hydroxylation of mephobarbital (Morita et al., 2004). Arg442 is located in the L helix on the proximal surface opposite the substrate binding site across the heme group (Williams et al., 2003; Wester et al., 2004) (Supplemental Fig. 2). This residue seems to be involved in electron transfer, and thus its replacement by a cysteine would decrease the binding to NADPH-P450 reductase, leading to reduced metabolic activity (Wada et al., 2008).
A161P exhibited significantly reduced intrinsic clearance (only 12–16% of that of CYP2C19.1A). Ala161 is located in the junction area between the D helix and E helix (Supplemental Fig. 2), which appears to not be at any putative substrate recognition site (Gotoh, 1992) or binding site on CYP2C19 (Lewis, 2002; Oda et al., 2004). However, alanine is the best helix-forming residue, whereas proline is a well known helix breaker and is commonly found in turns, so the introduction of Pro161 may refold the D-E loop and change the accessibility to the active site, thus affecting the metabolic activity of the enzyme.
Several other variants, namely CYP2C19.9, CYP2C19.18, CYP2C19.19, and D360N, also showed kinetic behaviors that differed from that of CYP2C19.1A. Consistent with previous findings (Blaisdell et al., 2002), CYP2C19.9 exhibited decreased activity toward S-mephenytoin and also had the same effect on omeprazole metabolism in vitro. The substrate affinity of CYP2C19.9 for the two substrates was lowered, with significantly higher Km values compared with that for CYP2C19.1A. CYP2C19.18 significantly differed from the WT by having a 2.44-fold higher Km for S-mephenytoin, which is in contrast with a previous report that did not show a significant Km difference with CYP2C19.1B (Hanioka et al., 2007). However, similar to another report (Hanioka et al., 2008), CYP2C19.18 did not show any influence on omeprazole 5′-hydroxylation. The differences in the degree of alteration found in the present study and the previous study may be a result of the different experimental conditions. CYP2C19.19 differed from the WT by demonstrating significantly higher Km values toward the two substrates, which was highly consistent with the published data (Hanioka et al., 2007, 2008) and confirmed that CYP2C19.19 decreases the substrate affinity of the CYP2C19 enzyme.
The function of the D360N variant was first revealed in this study. Its kinetic behavior was distinct from that of CYP2C19.1A, with significantly higher Km values and lower Vmax and Clint values toward both substrates. The same mutation was also identified in CYP2C9 in an African-American patient with major hemorrhage while receiving warfarin therapy (Goldstein et al., 2009). Another mutation at 360, namely D360E (CYP2C9*5), was found to greatly decrease the affinity and catalytic activity of the CYP2C9 enzyme (Dickmann et al., 2001). D360 lies within substrate recognition site 5 (residues 359–369), close to the residues (L362 and L366) that form the substrate-binding pockets of CYP2C9 and CYP2C19 (Lewis, 2002; Williams et al., 2003; Oda et al., 2004; Wester et al., 2004). Using the CYP2C19 homology model, we found that Asp360 forms hydrogen bonding interactions with Gln356 and Thr392 (Fig. 4a). Once Asp360 is replaced with Asn, the hydrogen-bonding interaction with Thr392 is lost, and the destabilizing effect on the protein secondary structure possibly alters the substrate binding or catalytic behavior of the enzyme (Fig. 4b).
Although most of the variants showed similar effects on S-mephenytoin and omeprazole metabolism, a substrate-dependent effect was observed with some variants. For example, E92D demonstrated a 1.73-fold lower Km value toward S-mephenytoin, but a 1.61-fold higher Km value toward omeprazole compared that for with CYP2C19.1A. E122A showed normal catalytic activity toward S-mephenytoin compared with that of CYP2C19.1A but significantly decreased omeprazole 5′-hydroxylation activity, with a higher Km and lower Clint. Because these residues appear to lie far away from the binding or active cavity of CYP2C19 for omeprazole (Ibeanu et al., 1996; Lewis, 2002) and S-mephenytoin (Tsao et al., 2001; Oda et al., 2004), further studies are required to identify the mechanism underlying the substrate-dependent effect of these variants.
In conclusion, in this report we described the thorough characterization of a large number of CYP2C19 polymorphic variants and obtained functional data that could be readily compared across variants. These data not only generally confirmed the function of previously reported alleles but also identified additional reduced-function variants, such as A161P, W212C, and D360N, as well as variants whose activities were similar to those of the WT. This work extends our understanding of the functional consequences of CYP2C19 genetic polymorphisms and contributes to the increased accuracy of CYP2C19 phenotype prediction in humans. To date, the crystal structure of CYP2C19 has not been resolved. The 20 nsSNPs described in this article extend over the entire cDNA region of CYP2C19. The functional data we obtained may thus contribute to the structure-function relationship study of the CYP2C19 protein and hence aid in elucidation of the CYP2C19 structure.
Authorship Contributions
Participated in research design: Hu. Wang, Zhu, and Chen.
Conducted experiments: Hu. Wang, An, Ha. Wang, Gao, and Bian.
Performed data analysis: Hu. Wang and Chen.
Wrote or contributed to the writing of the manuscript: Hu. Wang, D. Liu, Zhu, and Chen.
Acknowledgments
We thank Dr. Qing Zhou for assistance in homology modeling. We also thank Dr. Melanie Webster for critical reading of the manuscript.
Footnotes
This work was supported in part by the National High-Tech Research and Development Program of China (863 Program) [Grant 2006AA020705, 2009AA022710]; the Program for Changjiang Scholars and Innovative Research Team in University [Grant IRT0648]; and the Scientific and Technological Innovation Project of Shaanxi Province (13115 Project) [Grant 2007ZDKG-76, 2007ZDKG-70].
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.110.037549.
↵ The online version of this article (available at http://dmd.aspetjournals.org) contains supplemental material.
-
ABBREVIATIONS:
- PM
- poor metabolizer
- SNP
- single nucleotide polymorphism
- ns
- nonsynonymous
- PCR
- polymerase chain reaction
- WT
- wild type
- PCA
- principal component analysis.
- Received December 5, 2010.
- Accepted February 16, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics