Paclitaxel 6alpha-hydroxylation Research Articles

Functional Characterization of CYP2C8.13 and CYP2C8.14: Catalytic Activities toward Paclitaxel

Cytochrome P450 2C8 (CYP2C8) plays important roles in the metabolism of various drugs, including the anti-cancer drug, paclitaxel. We recently identified two novel CYP2C8 alleles (CYP2C8*13 and CYP2C8*14; wild-type, CYP2C8*1A) with non-synonymous single nucleotide polymorphisms in a Japanese population. To precisely investigate the effect of amino acid substitutions (CYP2C8*13, Ile223Met; CYP2C8*14, Ala238Pro) on CYP2C8 function, CYP2C8 proteins of the wild-type (CYP2C8.1) and variants (CYP2C8.13 and CYP2C8.14) were heterologously expressed in yeast cells, and their paclitaxel 6alpha-hydroxylation activities were determined. The K(m), V(max) and CL(int) values for paclitaxel 6alpha-hydroxylation of CYP2C8.1 were 2.3 microM, 4.1 pmol/min./pmol CYP and 1.7 microl/min./pmol CYP, respectively. The K(m) value of CYP2C8.14 was significantly higher (2.9-fold) than that of CYP2C8.1. The V(max) value of CYP2C8.14 was comparable to that of CYP2C8.1 and the CL(int) value was reduced to 46% of CYP2C8.1. In contrast, the K(m), V(max) and CL(int) values of CYP2C8.13 were similar to those of CYP2C8.1. These results suggest that Ala238Pro substitution in CYP2C8.14 decreases the affinity toward paclitaxel of the CYP2C8 enzyme, and that the genetic polymorphism of the CYP2C8*14 allele may influence the clinical response to drugs metabolized mainly by CYP2C8.

Characterization of novel CYP2C8 haplotypes and their contribution to paclitaxel and repaglinide metabolism

Cytochrome P450 2C8 (CYP2C8) plays a major role in the metabolism of therapeutically important drugs which exhibit large interindividual differences in their pharmacokinetics. In order to evaluate any genetic influence on this variation, a CYP2C8 phenotype-genotype evaluation was carried out in Caucasians. Two novel CYP2C8 haplotypes, named B and C with frequencies of 24 and 22% in Caucasians, respectively, were identified and caused a significantly increased and reduced paclitaxel 6alpha-hydroxylation, respectively, as evident from analyses of 49 human liver samples. In healthy white subjects, CYP2C8*3 and the two novel haplotypes significantly influenced repaglinide pharmacokinetics in SLCO1B1c.521T/C heterozygous individuals: haplotype B was associated with reduced and haplotype C with increased repaglinide AUC (0-infinity). Functional studies suggested -271C>A (CYP2C8*1B) as a causative SNP in haplotype B. In conclusion, two novel common CYP2C8 haplotypes were identified and significantly associated with altered rate of CYP2C8-dependent drug metabolism in vitro and in vivo.

Effects of Acid and Lactone Forms of Eight HMG-CoA Reductase Inhibitors on CYP-Mediated Metabolism and MDR1-Mediated Transport

With the growing clinical usage of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins), the number of reports concerning serious drug-drug interaction has been increasing. Because recent studies have shown that conversion between acid and lactone forms occurs in the body, drug-drug interaction should be considered on both acid and lactone forms. Thus, we investigated the inhibitory effects of acid and lactone forms of eight statins, including one recently withdrawn, cerivastatin, and two recently developed, pitavastatin and rosuvastatin, on cytochrome P450 (CYP) 2C8, CYP2C9, and CYP3A4/5 metabolic activities and multidrug resistance protein 1 (MDR1) transporting activity. The inhibitory effects of statins on CYP metabolic activities and MDR1 transporting activity were investigated using human liver microsomes and MDR1-overexpressing LLC-GA5-COL150 cells, respectively. The acid forms had minimal inhibitory effects on all CYP activities tested, except for fluvastatin on CYP2C9-mediated tolbutamide 4-hydroxylation (IC50 = 1.7 microM) and simvastatin on CYP3A4/5-mediated paclitaxel 3-hydroxylation (12.0 microM). Lactone forms showed no or minimal inhibitory effects on CYP2C8, CYP2C9, and CYP2C19 activities, except for rosuvastatin on the CYP2C9 activity (20.5 microM), whereas they showed stronger inhibitory effects on the CYP3A4/5 activity with the rank order of atorvastatin (5.6 microM), cerivastatin (8.1 microM), fluvastatin (14.9 microM), simvastatin (15.2 microM), rosuvastatin (20.7 microM), and lovastatin (24.1 microM). Pitavastatin and pravastatin had little inhibitory effect, and a similar order was found also for testosterone 6beta-hydroxylation. MDR1-mediated transport of [3H]digoxin was inhibited only by lactone forms, and the rank order correlated with that of inhibitory effects on both CYP3A4/5 activities. Inhibitory effects on MDR1 activity, and on both CYP3A4/5 activities, could be explained by the lipophilicity; however, a significant correlation was found between the lipophilicity and inhibitory effects on CYP2C8-mediated paclitaxel 6alpha-hydroxylation. We showed the difference between the acid and lactone forms in terms of drug interaction. The lipophilicity could be one of the important factors for inhibitory effects. In the case of statins, it is important to examine the effects of both forms to understand the events found in clinical settings, including the pleiotropic effects.

Metabolism of Repaglinide by CYP2C8 and CYP3A4 in vitro: Effect of Fibrates and Rifampicin

Repaglinide is an antidiabetic drug metabolised by cytochrome P450 (CYP) 2C8 and CYP3A4 enzymes. To clarify the mechanisms of observed repaglinide drug interactions, we determined the contribution of the two enzymes to repaglinide metabolism at different substrate concentrations, and examined the effect of fibrates and rifampicin on CYP2C8, CYP3A4 and repaglinide metabolism in vitro. We studied repaglinide metabolism using pooled human liver microsomes, recombinant CYP2C8 and recombinant CYP3A4 enzymes. The effect of quercetin and itraconazole on repaglinide metabolism, and of gemfibrozil, bezafibrate, fenofibrate and rifampicin on CYP2C8 (paclitaxel 6alpha-hydroxylation) and CYP3A4 (midazolam 1-hydroxylation) activities and repaglinide metabolism were studied using human liver microsomes. At therapeutic repaglinide concentrations (<0.4 microM), CYP2C8 and CYP3A4 metabolised repaglinide at similar rates. Quercetin (25 microM) and itraconazole (3 microM) inhibited the metabolism of 0.2 microM repaglinide by 58% and 71%, and that of 2 microM repaglinide by 56% and 59%, respectively. The three fibrates inhibited CYP2C8 (Ki: bezafibrate 9.7 microM, gemfibrozil 30.4 microM and fenofibrate 92.6 microM) and repaglinide metabolism (IC50: bezafibrate 37.7 microM, gemfibrozil 111 microM and fenofibrate 164 microM), but had no effect on CYP3A4. Rifampicin inhibited CYP2C8 (Ki 30.2 microM), CYP3A4 (Ki 18.5 microM) and repaglinide metabolism (IC50 13.7 microM). In conclusion, both CYP2C8 and CYP3A4 are important in the metabolism of therapeutic concentrations of repaglinide in vitro, but their predicted contributions in vivo are highly dependent on the scaling factor used. Gemfibrozil is only a moderate inhibitor of CYP2C8 and does not inhibit CYP3A4; inhibition of CYP-enzymes by parent gemfibrozil alone does not explain its interaction with repaglinide in vivo. Rifampicin competitively inhibits both CYP2C8 and CYP3A4, which can counteract its inducing effect in humans.

FUNCTIONAL CHARACTERIZATION OF FIVE NOVEL CYP2C8 VARIANTS, G171S, R186X, R186G, K247R, AND K383N, FOUND IN A JAPANESE POPULATION

Cytochrome P450 2C8 is one of the primary enzymes responsible for the metabolism of a wide range of drugs such as paclitaxel, cerivastatin, and amiodarone. We have sequenced the CYP2C8 gene from 201 Japanese subjects and found five novel nonsynonymous single nucleotide polymorphisms (SNPs): 511G>A (G171S), 556C>T (R186X; X represents the translational stop codon), 556C>G (R186G), 740A>G (K247R), and 1149G>T (K383N), with the allele frequency of 0.0025. The CYP2C8 variants were heterologously expressed in COS-1 cells and functionally characterized in terms of expression level, paclitaxel 6alpha-hydroxylase activity, and intracellular localization. The prematurely terminated R186X variant was undetectable by Western blotting and inactive toward paclitaxel 6alpha-hydroxylation. The G171S, K247R, and K383N variants exhibited properties similar to those of the wild-type CYP2C8. Paclitaxel 6alpha-hydroxylase activity of the R186G transfectant was only 10 to 20% that of wild-type CYP2C8. Furthermore, the R186G variant displayed a lower level of protein expression in comparison to the wild type, which was restored by the addition of a proteasome inhibitor (MG-132; Z-Leu-Leu-Leu-aldehyde). The reduced CO-difference spectral analysis using recombinant proteins from an insect cell/baculovirus system revealed that the R186G variant has a minor peak at 420 nm in addition to the characteristic Soret peak at 450 nm, suggesting the existence of improperly folded protein. These results indicate that the novel CYP2C8 SNPs, 556C>T (R186X) and 556C>G (R186G), could influence the metabolism of CYP2C8 substrates such as paclitaxel and cerivastatin.

Identification of cytochrome P450 isoforms involved in the metabolism of loperamide in human liver microsomes.

The purpose of the present study was to elucidate the cytochrome P450 (P450) isoform(s) involved in the metabolism of loperamide (LOP) to N-demethylated LOP (DLOP) in human liver microsomes. Three established approaches were used to identify the P450 isoforms responsible for LOP N-demethylation using human liver microsomes and cDNA-expressed P450 isoforms: (1) correlation of LOP N-demethylation activity with marker P450 activities in a panel of human liver microsomes, (2) inhibition of enzyme activity by P450-selective inhibitors, and (3) measurement of DLOP formation by cDNA-expressed P450 isoforms. The relative contribution of P450 isoforms involved in LOP N-demethylation in human liver microsomes were estimated by applying relative activity factor (RAF) values. The formation rate of DLOP showed biphasic kinetics, suggesting the involvement of multiple P450 isoforms. Apparent Km and Vmax values were 21.1 microM and 122.3 pmol/min per milligram of protein for the high-affinity component and 83.9 microM and 412.0 pmol/min per milligram of protein for the low-affinity component, respectively. Of the cDNA-expressed P450 s tested, CYP2B6, CYP2C8, CYP2D6, and CYP3A4 catalyzed LOP N-demethylation. LOP N-demethylation was significantly inhibited when coincubated with quercetin (a CYP2C8 inhibitor) and ketoconazole (a CYP3A4 inhibitor) by 40 and 90%, respectively, but other chemical inhibitors tested showed weak or no significant inhibition. DLOP formation was highly correlated with CYP3A4-catalyzed midazolam 1-hydroxylation (rs=0.829; P<0.01), CYP2B6-catalzyed 7-ethoxy-4-trifluoromethylcoumarin O-deethylation (rs=0.691; P<0.05), and CYP2C8-catalyzed paclitaxel 6alpha-hydroxylation (rs=0.797; P<0.05). CYP2B6, CYP2C8, CYP2D6, and CYP3A4 catalyze LOP N-demethylation in human liver microsomes, and among them, CYP2C8 and CYP3A4 may play a crucial role in LOP metabolism at the therapeutic concentrations of LOP. Coadministration of these P450 inhibitors may cause drug interactions with LOP. However, the clinical significance of potential interaction of LOP metabolism by CYP2C8 and CYP3A4 inhibitors should be studied further.

Mechanism-based inactivation of human cytochrome P4502C8 by drugs in vitro.

Studies were conducted to evaluate the potential mechanism-based inactivation of recombinant and human liver microsomal CYP2C8 by clinically used drugs. Several tricyclic antidepressants, calcium channel blockers, monoamine oxidase inhibitors, and various other known CYP3A4 inhibitors exhibited greater inhibition of CYP2C8 (paclitaxel 6alpha-hydroxylation) following preincubation, consistent with mechanism-based inactivation. Inactivation of recombinant CYP2C8 by phenelzine, amiodarone, verapamil, nortriptyline, fluoxetine, and isoniazid was of the pseudo-first order type and was characterized by respective inactivation kinetic constants (KI and kinact) of 1.2 microM and 0.243 min(-1), 1.5 microM and 0.079 min(-1), 17.5 microM and 0.065 min(-1), 49.9 microM and 0.036 min(-1), 294 microM and 0.083 min(-1), and 374 microM and 0.042 min(-1). Spectral scanning of recombinant CYP2C8 demonstrated the formation of metabolite-intermediate complexes with verapamil, nortriptyline, fluoxetine, and isoniazid, but not amiodarone. In contrast, inactivation by phenelzine resulted from heme destruction by free radicals. Studies with human liver microsomes (HLMs) revealed that nortriptyline, verapamil, and fluoxetine were not mechanism-based inactivators (MBIs) of CYP2C8. Simultaneous inactivation of CYP2C8 and CYP3A4 (paclitaxel 3'-phenyl-hydroxylation) was observed using amiodarone, isoniazid, and phenelzine with the efficiency of inactivation greater for the CYP3A4 pathway. With the exception of phenelzine, glutathione and superoxide dismutase failed to protect CYP2C8 (recombinant and HLMs) or CYP3A4 from inactivation by MBIs. However, the alternate CYP2C8 substrate, torsemide, prevented CYP2C8 inactivation in all cases. These data are consistent with mechanism-based inactivation of CYP2C8 by a range of commonly prescribed drugs, several of which have been implicated in clinically important drug-drug interactions.

Mechanism-based inhibition of human liver microsomal cytochrome P450 1A2 by zileuton, a 5-lipoxygenase inhibitor.

Zileuton, a 5-lipoxygenase inhibitor, was evaluated as an inhibitor of cytochrome P450 activity in human liver microsomes. In the absence of preincubation, the racemate was found to be a weak inhibitor (IC50 > 100 microM) of phenacetin O-deethylation (POD) (CYP1A2), paclitaxel 6alpha-hydroxylation (CYP2C8), diclofenac 4'-hydroxylation (CYP2C9), (S)-mephenytoin 4'-hydroxylation (CYP2C19), bufuralol 1'-hydroxylation (CYP2D6), testosterone 6beta-hydroxylation (CYP3A4), chlorzoxazone 6-hydroxylation (CYP2E1), and bupropion hydroxylation (CYP2B6). When preincubated with NADPH-fortified human liver microsomes in the absence of substrate, zileuton (racemate) was shown to inhibit POD. The effect was NADPH-, time-, and concentration-dependent, and was characterized by a kinact (maximal rate of enzyme inactivation) and apparent KI(inhibitor concentration that supports half the maximal rate of inactivation) of 0.035 min(-1) and 117 microM, respectively (kinact/KIratio of 0.0003 min-1 microM(-1)). Preincubation-dependent inhibition of POD activity was also observed with the individual (S)-(-)- and (R)-(+)-enantiomers of zileuton [(S)-(-)-zileuton; kinact, 0.037 min(-1), KI, 98.2 microM, kinact/KIratio, 0.0004 min(-1) microM(-1); (R)-(+)-zileuton; kinact, 0.012 min(-1), KI, 66.6 microM, kinact/KIratio, 0.0002 min(-1) microM(-1)]. In addition, the inhibition of CYP1A2 was not reversed in the presence of reduced glutathione, catalase, and superoxide dismutase and was refractory to dialysis. Therefore, zileuton was characterized as a mechanism-based inhibitor of human liver microsomal CYP1A2. Mechanism-based inhibition of CYP1A2 may explain why zileuton decreases the oral clearance of antipyrine, propranolol, (R)-warfarin, and theophylline, at doses that have a minimal effect on the pharmacokinetics of (S)-warfarin, phenytoin, and terfenadine.

Cytochrome P450 2C8 and CYP3A4/5 are involved in chloroquine metabolism in human liver microsomes.

Chloroquine has been used for many decades in the prophylaxis and treatment of malaria. It is metabolized in humans through the N-dealkylation pathway, to desethylchloroquine (DCQ) and bisdesethylchloroquine (BDCQ), by cytochrome P450 (CYP). However, until recently, no data are available on the metabolic pathway of chloroquine. Therefore, the metabolic pathway of chloroquine was evaluated using human liver microsomes and cDNA-expressed CYPs. Chloroquine is mainly metabolized to DCQ, and its Eadie-Hofstee plots were biphasic, indicating the involvement of multiple enzymes, with apparent Km and Vmax values of 0.21 mM and 1.02 nmol/min/mg protein 3.43 mM and 10.47 nmol/min/mg protein for high and low affinity components, respectively. Of the cDNA-expressing CYPs examined, CYP1A2, 2C8, 2C19, 2D6 and 3A4/5 exhibited significant DCQ formation. A study using chemical inhibitors showed only quercetin (a CYP2C8 inhibitor) and ketoconazole (a CYP3A4/5 inhibitor) inhibited the DCQ formation. In addition, the DCQ formation significantly correlated with the CYP3A4/5-catalyzed midazolam 1-hydroxylation (r = 0.868) and CYP2C8-catalyzed paclitaxel 6alpha-hydroxylation (r = 0.900). In conclusion, the results of the present study demonstrated that CYP2C8 and CYP3A4/5 are the major enzymes responsible for the chloroquine N-deethylation to DCQ in human liver microsomes.

CYP2C8 and CYP3A4 are the principal enzymes involved in the human in vitro biotransformation of the insulin secretagogue repaglinide.

To identify the principal human cytochrome P450 (CYP) enzyme(s) responsible for the human in vitro biotransformation of repaglinide. Previous experiments have identified CYP3A4 as being mainly responsible for the in vitro metabolism of repaglinide, but the results of clinical investigations have suggested that more than one enzyme may be involved in repaglinide biotransformation. [14C]-Repaglinide was incubated with recombinant CYP and with human liver microsomes (HLM) from individual donors in the presence of inhibitory antibodies specific for individual CYP enzymes. Metabolites, measured by high-performance liquid chromatography (HPLC) with on-line radiochemical detection, were identified by liquid chromatography-mass spectrophotometry (LC-MS) and LC-MS coupled on-line to a nuclear magnetic resonance spectrometer (LC-MS-NMR). CYP3A4 and CYP2C8 were found to be responsible for the conversion of repaglinide into its two primary metabolites, M4 (resulting from hydroxylation on the piperidine ring system) and M1 (an aromatic amine). Specific inhibitory monoclonal antibodies against CYP3A4 and CYP2C8 significantly inhibited (> 71%) formation of M4 and M1 in HLM. In a panel of HLM from 12 individual donors formation of M4 and M1 varied from approximately 160-880 pmol min-1 mg-1 protein and from 100-1110 pmol min-1 mg-1 protein, respectively. The major metabolite generated by CYP2C8 was found to be M4. The rate of formation of this metabolite in HLM correlated significantly with paclitaxel 6alpha-hydroxylation (rs = 0.80; P = 0.0029). Two other minor metabolites were also detected. One of them was M1 and the other was repaglinide hydroxylated on the isopropyl moiety (M0-OH). The rate of formation of M4 in CYP2C8 Supersomes was 2.5 pmol min-1 pmol-1 CYP enzyme and only about 0.1 pmol min-1 pmol-1 CYP enzyme in CYP3A4 Supersomes. The major metabolite generated by CYP3A4 was M1. The rate of formation of this metabolite in HLM correlated significantly with testosterone 6beta-hydroxylation (rs = 0.90; P = 0.0002). Three other metabolites were identified, namely, M0-OH, M2 (a dicarboxylic acid formed by oxidative opening of the piperidine ring) and M5. The rate of M1 formation in CYP3A4 Supersomes was 1.6 pmol min-1 pmol-1 CYP enzyme but in CYP2C8 Supersomes it was only approximately 0.4 pmol min-1 pmol-1 CYP enzyme. The results confirm an important role for both CYP3A4 and CYP2C8 in the human in vitro biotransformation of repaglinide. This dual CYP biotransformation may have consequences for the clinical pharmacokinetics and drug-drug interactions involving repaglinide if one CYP pathway has sufficient capacity to compensate if the other is inhibited.

Trimethoprim and sulfamethoxazole are selective inhibitors of CYP2C8 and CYP2C9, respectively.

To evaluate the inhibitory effects of trimethoprim and sulfamethoxazole on cytochrome P450 (P450) isoforms, selective marker reactions for CYP1A2, CYP2A6, CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4 were examined in human liver microsomes and recombinant CYP2C8 and CYP2C9. The in vivo drug interactions of trimethoprim and sulfamethoxazole were predicted in vitro using [I]/([I] + K(i)) values. With concentrations ranging from 5 to 100 microM, trimethoprim exhibited a selective inhibitory effect on CYP2C8-mediated paclitaxel 6alpha-hydroxylation in human liver microsomes and recombinant CYP2C8, with apparent IC(50) (K(i)) values of 54 microM (32 microM) and 75 microM, respectively. With concentrations ranging from 50 to 500 microM, sulfamethoxazole was a selective inhibitor of CYP2C9-mediated tolbutamide hydroxylation in human liver microsomes and recombinant CYP2C9, with apparent IC(50) (K(i)) values of 544 microM (271 microM) and 456 microM, respectively. With concentrations higher than 100 microM trimethoprim and 500 microM sulfamethoxazole, both drugs lost their selectivity for the P450 isoforms. Based on estimated total hepatic concentrations (or free plasma concentrations) of the drugs and the scaling model, one would expect in vivo in humans 80% (26%) and 13% (24%) inhibition of the metabolic clearance of CYP2C8 and CYP2C9 substrates by trimethoprim and sulfamethoxazole, respectively. In conclusion, trimethoprim and sulfamethoxazole can be used as selective inhibitors of CYP2C8 and CYP2C9 in in vitro studies. In humans, trimethoprim and sulfamethoxazole may inhibit the activities of CYP2C8 and CYP2C9, respectively.

In vitro inhibitory effects of troglitazone and its metabolites on drug oxidation activities of human cytochrome P450 enzymes: comparison with pioglitazone and rosiglitazone

1. Investigated were the effects of a new oral antidiabetic drug, troglitazone, and its three metabolites and antidiabetic drug candidates pioglitazone and rosiglitazone on xenobiotic oxidations catalyzed by nine recombinant human cytochrome P450 (P450 or CYP) enzymes and by human liver microsomes. 2. Troglitazone (5 muM) significantly inhibited CYP2C8-dependent paclitaxel 6alpha- hydroxylation and CYP2C9-dependent S-warfarin 7-hydroxylation. On the other hand, pioglitazone and rosiglitazone (50 muM) only slightly inhibited these xenobiotic oxidation activities catalyzed by CYP2C enzymes. 3. The inhibitory potential of troglitazone (50% inhibition concentration, IC50) was ~5 muM for drug oxidations catalyzed by CYP2C9 and CYP2C8 and C20 muM for activities catalyzed by CYP2C19 and CYP3A4 respectively. For the three metabolites of troglitazone tested, a quinone-type metabolite (M3) was the most potent inhibitor for CYP2C enzymes, followed by a sulphate conjugate (M1); effects of a glucuronide (M2) were very weak. The inhibitory effects of the parent drug were more potent than those of metabolites. Troglitazone and M3 inhibited P450 activities mainly through a competitive manner with Ki=0.2-1.7 muM and 1.4-8.8 muM respectively. 4. In three human liver microsomes, troglitazone and its metabolites also inhibited paclitaxel 6alpha-hydroxylation, S-warfarin 7-hydroxylation, S-mephenytoin 4'-hydroxylation, and testosterone 6beta-hydroxylation with similar IC50, as observed for the recombinant P450 enzyme systems. 5. These results suggest that xenobiotic oxidations by P450 enzymes are more substantially affected by troglitazone and its metabolites than pioglitazone or rosiglitazone, and that drug interactions may be of much importance to understand the basis for the pharmacological and toxicological actions of this new oral antidiabetic drug.

Characterization of the cytochrome P450 enzymes involved in the in vitro metabolism of rosiglitazone.

To identify the human cytochrome P450 enzyme(s) involved in the in vitro metabolism of rosiglitazone, a potential oral antidiabetic agent for the treatment of type 2 diabetes-mellitus. Method The specific P450 enzymes involved in the metabolism of rosiglitazone were determined by a combination of three approaches; multiple regression analysis of the rates of metabolism of rosiglitazone in human liver microsomes against selective P450 substrates, the effect of selective chemical inhibitors on rosiglitazone metabolism and the capability of expressed P450 enzymes to mediate the major metabolic routes of rosiglitazone metabolism. Result The major products of metabolism following incubation of rosiglitazone with human liver microsomes were para-hydroxy and N-desmethyl rosiglitazone. The rate of formation varied over 38-fold in the 47 human livers investigated and correlated with paclitaxel 6alpha-hydroxylation (P<0.001). Formation of these metabolites was inhibited significantly (>50%) by 13-cis retinoic acid, a CYP2C8 inhibitor, but not by furafylline, quinidine or ketoconazole. In addition, both metabolites were produced by microsomes derived from a cell line transfected with human CYP2C8 cDNA. There was some evidence for CYP2C9 playing a minor role in the metabolism of rosiglitazone. Sulphaphenazole caused limited inhibition (<30%) of both pathways in human liver microsomes and microsomes from cells transfected with CYP2C9 cDNA were able to mediate the metabolism of rosiglitazone, in particular the N-demethylation pathway, albeit at a much slower rate than CYP2C8. Rosiglitazone caused moderate inhibition of paclitaxel 6alpha-hydroxylase activity (CYP2C8; IC50=18 microm ), weak inhibition of tolbutamide hydroxylase activity (CYP2C9; IC50=50 microm ), but caused no marked inhibition of the other cytochrome P450 activities investigated (CYP1A2, 2A6, 2C9, 2C19, 2D6, 2E1, 3A and 4A). Conclusion CYP2C8 is primarily responsible for the hydroxylation and N-demethylation of rosiglitazone in human liver; with minor contributions from CYP2C9.