Structure and function of hepatic cytochromes P450 : implications for drug development

Abstract

Prediction of human drug metabolism in vivo by use of in vitro systems is of immense importance in drug development since it allows avoidance of drugs coming out on the market that are subjected to interindividual variability in metabolism or causes drug-drug interactions, both of which in turn can lead to adverse drug reactions and result in non-optimal treatment. In the present investigations the importance of P450 haplotype on substrate specificity was studied as well as a comparison of different in vitro systems for optimal prediction of in vivo drug metabolism. Regions important for membrane interactions of one form of P450 were elucidated in order to engineer a soluble form of the enzyme of use for crystallization studies.

A new CYP2D6 allele (CYP2D6*17) common in Black African populations carrying three mutations Thrl07IIe, Arg296Cys and Ser486Thr was investigated by recombinant expression and the enzyme characterized using CYP2D6 specific substrates. It was found that both the Thrl07IIe, Arg296Cys substitutions were required in order to yield an enzyme with altered substrate specificity in a manner seen in vivo, emphasizing the necessity to consider haplotypes rather than single SNPs for functional analyses of many polymorphic genes.

Various in vitro systems like homology models of P450 enzymes, recombinant enzymes, liver microsomes, hepatocyte primary cultures and liver slices were compared with respect to predictability of in vivo metabolism of 3 different drugs in the EUROCYP project where I was responsible for the recombinant enzyme systems using yeast. The expression systems were optimized for nine different human CYPs. It was found that the recombinant system under-predicted the overall clearance of almokalant, mainly being a substrate for UDP-glucuronosyltransferases. For carbamazepine, only CYP3A4 among the 9 different forms of P450 examined was active. The addition of cytochrome b5 lead to a overestimated clearance putting emphasis on the composition of cofactors when dealing with CYP3A4. For selegiline we found that, in contrast to previous studies, CYP2B6 and CYP2C19 were the most active enzymes, but some contribution from CYP1A2 and CYP3A4 was also predicted. Scaling up our recombinant data to the human situation yielded values slightly higher but not significantly different from the in vivo clearance. Hence, the recombinant system correctly predicted selegiline to be a high clearance drug. It was concluded that the recombinant system was of good value for determination of the enzyme specificity but not for in vivo clearance of drug candidates.

The role of the hydrophobic NH2-terminal of rat CYP2E1 for mitochondrial targeting and membrane interactions was studied by expressing NH2-terminally truncated variants of this enzyme in Saccharomyces cerevisiae. Six variants were expressed along with the wt CYP2E1, four were truncated at position 29, 64, 82 or 95 (delta95-2E1), respectively and one variant had Leu to Asp amino acid exchanges at positions 90 and 91 (delta82mut-2E1). With the exception of delta95-2E1 and delta82mut-2E1, all variants were active in NADPH supported chlorzoxazone 6-hydroxylation, but when the artificial electron donor cumene hydroperoxide was used, also these were active. A region between as 82 and 95 was found to be critical for mitochondrial import of CYP2E1. It was found that delta82mut-2E1 was soluble and catalytically active in phosphate buffer and could be used for crystallization studies. It is concluded that the B-helix of CYP2E1 and possibly other CYPs is important for mitochondrial targeting as well as for membrane association where interactions of ionic nature occur.