Impact of genetic and non-genetic factors on hepatic CYP2C9 expression and activity in Hungarian subjects

CYP2C9, one of the most abundant hepatic cytochrome P450 enzymes, is involved in metabolism of 15–20% of clinically important drugs (warfarin, sulfonylureas, phenytoin, non-steroid anti-inflammatory drugs). To avoid adverse events and/or impaired drug-response, CYP2C9 pharmacogenetic testing is recommended. The impact of CYP2C9 polymorphic alleles (CYP2C9*2, CYP2C9*3) and phenoconverting non-genetic factors on CYP2C9 function and expression was investigated in liver tissues from Caucasian subjects (N = 164). The presence of CYP2C9*3 allele was associated with CYP2C9 functional impairment, and CYP2C9*2 influenced tolbutamide 4′-hydroxylase activity only in subjects with two polymorphic alleles, whereas the contribution of CYP2C8*3 was not confirmed. In addition to CYP2C9 genetic polymorphisms, non-genetic factors (co-medication with CYP2C9-specific inhibitors/inducers and non-specific factors including amoxicillin + clavulanic acid therapy or chronic alcohol consumption) contributed to the prediction of hepatic CYP2C9 activity; however, a CYP2C9 genotype–phenotype mismatch still existed in 32.6% of the subjects. Substantial variability in CYP2C9 mRNA levels, irrespective of CYP2C9 genotype, was demonstrated; however, CYP2C9 induction and non-specific non-genetic factors potentially resulting in liver injury appeared to modify CYP2C9 expression. In conclusion, complex implementation of CYP2C9 genotype and non-genetic factors for the most accurate estimation of hepatic CYP2C9 activity may improve efficiency and safety of medication with CYP2C9 substrate drugs in clinical practice.

CYP2C9*2, one of the well-characterized CYP2C9 allele, is associated with an amino acid change (Arg144 → Cys144) and decreased enzyme activity due to the 3608C > T transition in exon 3 (rs1799853). The Arg144Cys amino acid substitution has been demonstrated to affect the interaction of the enzyme with NADPH-cytochrome P450 oxidoreductase that can significantly impair the CYP2C9 catalytic function 7 . However, alterations in the CYP catalytic cycle have also been suggested to diminish metabolic activity of CYP2C9*2 8 . CYP2C9*3 allele with a nucleotide change of 42614A > C in exon 7 (rs1057910) results in an amino acid substitution of Ile359 → Leu359 in the active site of the enzyme that has been suggested to play an important role in substrate recognition or in formation of iron-oxene complex 8,9 . The Ile359Leu amino acid change leads to a significant decrease in catalytic activity due to an increase in K m values and a decrease in maximal enzyme activity (v max ) for most CYP2C9 substrates 10,11 . In addition, some single nucleotide polymorphisms (SNPs) in the promoter region of CYP2C9 have been identified, and these genetic variations were assumed to influence the constitutive expression or the pregnane X receptor mediated induction of CYP2C9 12 . As a consequence of CYP2C9 polymorphisms, substantial reduction in oral clearance of many clinically important drugs, such as S-warfarin, fluvastatin, gliclazide or glimepiride, celecoxib, phenytoin or valproate, has been reported that requires dose adjustment to avoid serious adverse reactions 4,5,11 . In patients with loss-of-function CYP2C9*2 or CYP2C9*3 alleles, reduced ability to metabolize the anticoagulant S-warfarin and therefore increased risk of over-anticoagulation are well-established 13,14 . CYP2C9 genotype-guided dosing has been demonstrated to reduce bleeding complications during initiation of warfarin therapy 13,15,16 . In vitro studies with human liver microsomes and clinical studies involving patients with epilepsy have shown that carriers of CYP2C9*2 and/or CYP2C9*3 alleles resulted in decreased valproate metabolism compared to individuals with CYP2C9*1/*1 wild-type genotypes [17][18][19] . Although the association between CYP2C9 genotype and valproate blood concentrations was clearly demonstrated in children whose metabolic pathways of valproate other than CYP2C9-mediated routes are poorly developed, the impact of CYP2C9 genetic variability on valproate clearance was not significant in adult patients 20 .
Investigation of clinically relevant CYP2C9 polymorphisms with decreased enzyme function is important in the interpretation of altered efficacy and/or toxicity of CYP2C9 substrate drugs, particularly of those with narrow therapeutic ranges 21 . Applying CYP2C9 genotype testing in clinical practice may contribute to better understanding of phenotypic effect and thus to avoiding adverse events and/or impaired drug-response. Several clinical studies have been implemented to apply CYP2C9 genotype-based therapies; however, inter-individual variability in CYP2C9 activity can be partly attributed to genetic polymorphisms. Non-genetic factors (age, hormonal status, disease, co-medication, nutrition) can further modulate CYP2C9 expression and CYP2C9 activity resulting in phenoconversion 6,22 . Inhibition of CYP2C9 activity or transcriptional induction of CYP2C9 gene due to co-medication and other non-genetic factors can transiently switch into poor or extensive metabolizer phenotype. The CYP2C9 promoter region contains several nuclear receptor responsive elements (e.g. for pregnane X receptor, constitutive androstane receptor, glucocorticoid receptor) that are involved in transcriptional regulation of CYP2C9 expression 23,24 . Considering both the genetic and non-genetic variations may be required for an accurate estimation of CYP2C9-mediated drug metabolism. The present study attempted to investigate the impact of the two most common CYP2C9 alleles (CYP2C9*2 and CYP2C9*3) on CYP2C9 activity in liver tissue samples from adult organ donors belonging to Hungarian (Caucasian) population. The hepatic microsomal CYP2C9 activity was characterized by using tolbutamide as the CYP2C9-selective probe substrate. Since a minor role of CYP2C8 in tolbutamide 4′-hydroxylation has been suggested 25 , the impact of the clinically most relevant, loss-of-function CYP2C8*3 (rs11572080) on tolbutamide 4′-hydroxylase activity of tissue donors was also investigated. Furthermore, we aimed to identify some non-genetic factors (demographic parameters, comedication) that can potentially modify the expression or the activity of CYP2C9. Incorporating these factors in prediction of CYP2C9 activity may improve CYP2C9 genotype-based therapy.

Material and methods
Human liver microsomes and RNA samples. Human Table 1. Human livers were perfused with Euro-Collin's solution (Fresenius AG, Bad Homburg vdH, Germany) and excised. For the isolation of microsomal fraction, the tissues were homogenized in 0.1 M Tris-HCl buffer (pH 7.4) containing 1 mM EDTA and 154 mM KCl, and differential centrifugation was performed as described by van der Hoeven and Coon 26 . Microsomal protein content was determined by the method of Lowry et al. with bovine serum albumin as the standard 27 . Total RNA was also extracted from approximately 50 mg of liver tissues homogenized in 1 ml of TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. The RNA was precipitated by using ethanol and stored in 0.1% diethylpyrocarbonate contained ultra pure water at − 80 °C for further analyses.
CYP2C9 enzyme assay. The method of Miners and Birkett was followed to determine tolbutamide 4′-hydroxylation activity selective for CYP2C9 28 . The incubation mixture contained NADPH-generating system (1 mM NADP, 10 mM glucose 6-phosphate, 5 mM MgCl 2 and 2 units/ml glucose 6-phosphate dehydrogenase), human liver microsomes (1 mg/ml) and tolbutamide (1 mM). After 20-min incubation, the reaction was terminated by ice-cold methanol and the incubation mixture was centrifuged for 10 min at 10,000 × g. High-performance liquid chromatographic analysis was performed according to published method 28 . CYP2C9 enzyme assay for each donor was performed in triplicate. Analysis of CYP2C9 mRNA levels by quantitative real-time PCR. Total RNA (3 μg) was reversetranscribed into single-stranded cDNA by using Maxima First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Wilmington, DE). The real-time PCR with human cDNA was performed by using KAPA Fast Probes Mastermix (KAPA Biosystems, Cape Town, South Africa) and TaqMan probes for CYP2C9 (BioSearch Technologies, Novato, CA). The quantity of the target RNA relative to that of the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was determined. GAPDH expression is constant in all cells and independent from experimental conditions; therefore, its expression was set to 1, and CYP2C9 mRNA levels were normalized by GAPDH expression. The sequences of primers and probes used for the real-time PCR analyses of CYP2C9 and GAPDH expression were previously reported by Déri

Discussion
CYP2C9 is in the focus of pharmacogenetic studies for genotype-based drug therapy, because it is one of the most abundant hepatic CYP enzymes and catalyses the metabolism of many clinically important drugs, particularly of those with narrow therapeutic concentration range 4,5,35 . The association between the clearance of the anticoagulant S-warfarin and CYP2C9 polymorphism was the most intensively investigated, and to avoid bleeding complications, genotype-guided dosing was recommended [13][14][15]42 . Although no significant difference in in vitro losartan metabolism was found between CYP2C9*1/*2 and CYP2C9*1/*1 genotypes, a substantial decrease was detected in hepatic microsomes from subjects with CYP2C9*1/*3, CYP2C9*2/*2, and CYP2C9*3/*3 genotypes comparing to those with the homozygous wild-type genotype 43 . Similarly, no or negligible reduction of flurbiprofen metabolism was observed in healthy volunteers carrying CYP2C9*1/*2, whereas the metabolic rates of flurbiprofen were markedly decreased in subjects with CYP2C9*1/*3 or with two polymorphic alleles (CYP2C9*2/*2, CYP2C9*2/*3, CYP2C9*3/*3) comparing to those carrying two wild-type alleles 44 . Genetic variability of CYP2C9, however, can partly explain the substantial inter-individual variations in pharmacokinetics of CYP2C9 substrates. Due to phenoconversion, non-genetic factors can significantly modify CYP2C9 expression and/or enzyme activity predicted from the genotype 36,45 . The aim of the present work was to elucidate the impact of CYP2C9 polymorphisms common in Caucasian populations on the activity and expression of CYP2C9 in liver tissues from 164 Hungarian organ donors. CYP2C9 enzyme activity was characterized by tolbutamide 4′-hydroxylation in microsomal fractions isolated from the liver tissues. Since a minor role of CYP2C8 enzyme in the hydroxylation of tolbutamide was suggested 25 , the contribution of the most common CYP2C8*3 allele to the reduction of tolbutamide 4′-hydroxylation was also investigated. The allele frequencies for CYP2C9*2, CYP2C9*3 and CYP2C8*3 in the 164 tissue donors were similar to the allele frequency data previously reported in Caucasian populations [4][5][6]33 . The significant linkage between the CYP2C9*2 and CYP2C8*3 alleles (D′ = 0.87) demonstrated in the liver tissue donors was concordant with previous findings 32, 46,47 . Since several CYP2C9 substrates (montelukast, fluoxetine, ibuprofen, rosiglitazone, tolbutamide, zopiclone) are also metabolized to some extent by CYP2C8 48 , reduced activity of CYP2C8*3 has been assumed to contribute to the low metabolic rates of CYP2C9 substrates in CYP2C9*2 carriers. A substantial decrease in ibuprofen clearance and as a consequence, increased risk of gastrointestinal bleeding have been reported in heterozygous CYP2C9*1/*2 subjects when CYP2C9*2 was present in combination with CYP2C8*3 46,49,50 . In vitro studies on diclofenac metabolism identified CYP2C9 as the major catalyst with minor contribution of CYP2C8 51 ; however, the association between diclofenac clearance and genetic polymorphisms of CYP2C9 and CYP2C8 appear to be controversial 52,53 . The impact of CYP2C9 genetic polymorphisms on tolbutamide clearance has clearly been demonstrated 54 ; however, no data available for the CYP2C8 polymorphic alleles. Although both CYP2C9*2 and CYP2C9*3 alleles are associated with decreased enzyme activity 55,56 , in heterozygous subjects carrying one wild-type and one polymorphic alleles (CYP2C9*1/*2 or CYP2C9*1/*3), the role of the CYP2C9*2 in modification of tolbutamide 4′-hydroxylase activity was found to be negligible in the liver tissues, whereas CYP2C9*3 was associated with significant reduction of CYP2C9 activity comparing to those with CYP2C9*1/*1 genotype. Furthermore, tolbutamide 4′-hydroxylase activity of the liver tissue donors with two loss-of-function alleles (CYP2C9*2/*2, CYP2C9*2/*3) was lower than in those with CYP2C9*1/*1. Our data are consistent with those of Kirchheiner et al. 54 and Jetter et al. 57 , who demonstrated tolbutamide clearance in CYP2C9*1/*2 carriers to be indistinguishable from the subjects with two wild-type alleles, whereas the oral plasma clearance was significantly reduced in CYP2C9*1/*3 heterozygotes and in subjects with two polymorphic alleles comparing with those carrying CYP2C9*1/*1. In the present work, we have demonstrated for the first time that the presence of the loss-of-function CYP2C8*3 had no impact on tolbutamide 4′-hydroxylation in CYP2C9*1/*2 carriers. Even more, CYP2C8*3 in two of the CYP2C9*1/*1 carriers did not result in significant reduction of tolbutamide 4′-hydroxylase activity, but both subjects were high intermediate metabolizers.
In subjects with homozygous wild-type genotype (CYP2C9*1/*1), hepatic CYP2C9 mRNA expression has been reported to correlate with CYP2C9 activity 29 . However, the polymorphic CYP2C9*2 and CYP2C9*3 alleles seemed to have no impact on CYP2C9 expression. No significant differences in CYP2C9 mRNA levels were found between the subjects with CYP2C9*1/*1 genotype and those carrying one or two polymorphic CYP2C9 alleles. The C > T transition at 3608 position in CYP2C9*2 allele and the nucleotide change of 42614A > C in CYP2C9*3 are well-described to result in amino acid changes that significantly impair CYP2C9 activity; however, these SNPs seem to have no impact on CYP2C9 mRNA expression 4 . Our findings confirmed that the CYP2C9*2 and CYP2C9*3 polymorphic alleles did not influence the hepatic CYP2C9 mRNA expression. It also means that in those carrying polymorphic CYP2C9 alleles, CYP2C9 expression does not inform about hepatic CYP2C9 activity, www.nature.com/scientificreports/ whereas CYP2C9 mRNA levels correlate with CYP2C9 activity merely in those subjects with CYP2C9*1/*1 genotype, in line with previous findings 29 .
Because of the risk of serious adverse reactions, Clinical Pharmacogenetic Implementation Consortium has recommended CYP2C9 genotype-based activity scoring to clinicians when prescribing CYP2C9 substrate drugs, such as non-steroidal anti-inflammatory agents, anticoagulant warfarin or anticonvulsant phenytoin 35,58,59 ; however, internal and environmental non-genetic factors have also been suggested to consider. Due to phenoconversion, the concomitant treatments with CYP2C9 inhibitors and inducers have been proposed to improve CYP2C9 phenotype prediction. As a consequence of CYP2C9 inhibitor therapy, an individual with CYP2C9*1/*1 genotype can transiently become low intermediate or poor metabolizer, while the poor-metabolizer status of subjects with two polymorphic alleles has been assumed to be not affected by inhibitors 34 . In liver tissue donors carrying CYP2C9*1/*1 genotype, medication with CYP2C9 inhibitors (amlodipine, tamoxifen) was associated with low tolbutamide 4′-hydroxylation activity; furthermore, amoxicillin + clavulanic acid therapy and chronic alcohol consumption resulted in a decrease in CYP2C9 activity and mRNA levels. The racemic mixture of the anti-hypertensive amlodipine is used for therapeutic purposes; however, vasodilation is ascribed only to its S-enantiomer. Amlodipine has been reported to inhibit CYP2C9 activity in a stereoselective manner, and R-enantiomer was proved to be more potent CYP2C9 inhibitor than S-amlodipine 41 . The risk of drug interactions with CYP2C9 substrates has also been predicted during co-administration of the selective estrogen receptor modulator tamoxifen 60 . Tamoxifen and its anti-estrogenic hydroxylated metabolites (4-hydroxy-tamoxifen, endoxifen and norendoxifen) potently inhibited the activities of CYP2C enzymes. These findings were confirmed by the reduced tolbutamide 4′-hydroxylation activities in hepatic microsomes of tissue donors treated with amlodipine or tamoxifen. The pathomechanism of chronic alcohol consumption induced liver disease has long been studied, and there is a large body of evidence indicating that impaired drug metabolism is related to severe liver disease [61][62][63] . Amoxicillin, the widely used antibiotic is often applied in combination with clavulanic acid. Hepatotoxic effect of this combination is generally mild; however, amoxicillin + clavulanic acid therapy rarely leads to drug-induced liver injury or severe acute liver failure, for which liver transplantation is the only lifesaving intervention 64,65 . Although both chronic alcohol consumption and amoxicillin + clavulanic acid therapy have been reported to exert liver injury, information about their CYP2C9 inhibitory potential is hardly available. In contrast to amlodipine and tamoxifen that have the capability to inhibit CYP2C9 function, chronic alcohol consumption and amoxicillin + clavulanic acid therapy evoking liver injury were likely to have non-specific impact on CYP enzyme function rather than direct CYP2C9 inhibitory properties 66,67 .
CYP2C9 inducers have also been reported to modify CYP2C9 metabolic activity; therefore, concomitant treatment with CYP2C9 inducer drugs is recommended to take into account during phenotype prediction 34 . As a consequence of inducer therapy such as rifampicin, dexamethasone, carbamazepine or phenobarbitone, higher CYP2C9 activity is expected in patients with one or two wild-type CYP2C9 alleles than predicted from the genotype, whereas CYP2C9 function is assumed to be unchanged in those subjects with two polymorphic alleles 34 . The promoter region of the CYP2C9 gene contains a number of nuclear receptor binding sites through which the gene can be transcriptionally induced. The synthetic glucocorticoid dexamethasone or the corticosteroid methylprednisolone are known to activate glucocorticoid receptor and pregnane X receptor, and to increase CYP2C9 transcription, whereas midazolam has been reported to act as a pregnane X receptor activator and to induce the expression of CYP2C9 3,38-40,68,69 . The liver tissue donors known to receive dexamethasone, methylprednisolone or midazolam displayed high CYP2C9 mRNA expression and activity. In those subjects carrying one or two wild-type alleles, the high tolbutamide 4′-hydroxylation activity and CYP2C9 mRNA level were associated with CYP2C9 inducer drug therapy, resulting in extensive or high intermediate metabolizer phenotypes. However, in those with two polymorphic alleles (CYP2C9*2/*2 or CYP2C9*2/*3), the effect of CYP2C9 inducer therapy on increasing CYP2C9 expression and function was observed, which did not confirm the phenoconversion prediction by Mostafa et al. 34 . Since both CYP2C9*2 and CYP2C9*3 display some residual activity, it is reasonable to assume that the presence of CYP2C9 inducers ameliorated the poor function of CYP2C9 predicted from genotype. In vitro studies with various CYP2C9 allelic variants demonstrated that the inducibility of CYP2C9*2 and CYP2C9*3 by rifampicin was similar to that of CYP2C9*1 12 . As a result of rifampin treatment (pregnane X receptor activator), an increase in tolbutamide clearance (generally twofold) was reported in healthy volunteers with various CYP2C9 genotypes, and the CYP2C9 inducibility by rifampin was observed in all genotype groups, even in CYP2C9*2/*2, CYP2C9*3/*3 or CYP2C9*2/*3 carriers 70 .
Some limitations of the present study should be considered. First, CYP2C9*3/*3 genotype was not detected in the present population, although the prevalence of this genotype was reported to be 0-5.7% in Caucasian populations 5,32,[71][72][73] . Definite conclusion nevertheless could be drawn from CYP2C9*1/*3 genotype regarding the impact of CYP2C9*3 allele on the enzyme activity and the expression of CYP2C9. Second, we did not assess CYP2C9 alleles other than CYP2C9*2 and CYP2C9*3; however, the prevalence of other clinically relevant CYP2C9 alleles in Caucasian populations are extremely low 74 . Third, one may assume that the medical history of some subjects was incompletely reported, and some relevant non-genetic factors that can decrease or increase CYP2C9 activity were not included.
In conclusion, the impact of CYP2C9 polymorphic alleles (CYP2C9*2, CYP2C9*3) and non-genetic factors on CYP2C9 function and mRNA expression was demonstrated in human liver tissues. The role of the CYP2C9*3 allele in functional impairment was clearly confirmed, whereas the influence of CYP2C9*2 allele on hepatic tolbutamide 4′-hydroxylation activity was evident in those subjects carrying two polymorphic alleles. Furthermore, the contribution of CYP2C8*3 to tolbutamide 4′-hydroxylase activity was not confirmed. Although CYP2C9 genotype was found to be a major factor in CYP2C9 function, non-genetic factors such as co-medication with CYP2C9 inhibitors and inducers as well as non-specific factors including amoxicillin + clavulanic acid therapy and chronic alcohol consumption significantly altered the CYP2C9 phenotype predicted from genotype. In more than two thirds of the liver tissue donors, the combined effect of the CYP2C9 genotype and non-genetic www.nature.com/scientificreports/ factors was found to correspond to CYP2C9 function. It should be emphasized that non-genetic factors affected tolbutamide 4′-hydroxylation activity of CYP2C9 both in CYP2C9*1/*1 carriers and in those subjects with one or two polymorphic alleles. In contrast, substantial variability in hepatic CYP2C9 mRNA levels, irrespective of the CYP2C9 genotype, was demonstrated; however, CYP2C9 induction and non-specific non-genetic factors potentially resulting in liver injury appeared to contribute to CYP2C9 expression. As such these results supported the complex implementation of CYP2C9 genotype and non-genetic factors for the most accurate estimation of hepatic CYP2C9 enzyme activity that can improve efficiency and safety of medication with CYP2C9 substrate drugs in clinical practice. However, a CYP2C9 genotype-phenotype mismatch still existed in 32.6% of the subjects. www.nature.com/scientificreports/