Abstract |
Cytochrome P450s enzyme belongs to the superfamily of heme-containing proteins,
responsible for metabolizing more than 90% of clinical drugs. One of the most significant
enzymes in this family, Cytochrome P450 2D6 (CYP2D6), metabolizes ~25% of the
clinically used drugs including crucial and commonly administered drugs such as
antidepressants, chemotherapeutics, beta-blockers and opioids. Variations in CYP2D6, a
highly polymorphic loci in the genome, could alter its activity influencing the efficacy and
toxicity of numerous drugs. More than 100 haplotypes (star alleles) of the drug metabolizing
enzyme CYP2D6 have been reported in the Pharmacogene Variation Consortium (PharmVar,
www.pharmvar.org), resulting in wide intraindividual variability in drug metabolism activity
and changes of the drug plasma concentration. The complete connecting link between the
genetic variants and the metabolizer phenotype is still an open and challenging question. Our
main objective was to investigate the key factors that determine the metabolizer phenotype by
exploiting and appropriately employing molecular dynamics (MD) methods. MD is an
elaborate computational method that enables the prediction of the time evolution of atomic
positions within interacting systems of molecules. To this end, we have probed the dynamics
of numerous CYP2D6 variants, as enzyme models with normal and no function, at all-atom
resolution. We concluded that changes in residue b-factors and Dynamical Cross-correlation
analysis could be used as markers in the discrimination of the two classes of metabolizing
activity. Molecular docking analysis between CYP2D6 variants and BACE1 inhibitor
confirmed our observations and highlighted the role of helix I and of K-K’ loop and their
relative movement in the activity of the enzyme. Classical MD runs on the CYP2D6 *1
(wild-type) were used for identifying the important residues for the protein conformational
space using Markov State Modeling. Based on these residues and using the data from the
tICA/MSM analysis, a dataset for each variant has been produced which was then used to
build a prediction model for the metabolizer phenotype. This is the first time such a tool has
been developed. Results of this work are of great importance for areas like Personalized
Medicine, Adverse Drug Reaction (ADR) prediction and drug discovery.
|