Modeling Structure-Metabolism Relationships: from the Human Cytochrome P450 Consortium Initiative a New Technology to Predict P450 Inhibition, Metabolic Stability and Isoform SelectivityEmanuele Carosati1, Daniele Maiarelli2, Gabriele Cruciani1
|1Laboratory for Chemometrics and Cheminformatics, Chemistry Department, University of Perugia, Via Elce di Sotto 10, I-06123, Perugia, Italy|
2Molecular Discovery Ltd. b Molecular Discovery Ltd. 215 Marsh Road, HA5 5NE, Pinner, Middlesex, UK
|Cytochromes P450 affect the fate of drug candidates by conditioning their bioavailability and possible drug-drug interactions (DDI). They are studied in the ADME screening process, when properties such as CYP inhibition, metabolic stability, and isoform selectivity are routinely addressed with in vitro systems. The existence of reliable in silico models to predict such CYP-related properties might be helpful to cross-check experiments and predictions for early identification of either random or systematic errors, to prioritize experiments in the screening pipeline, and to forecast properties before synthesis of new molecular entities. An ever increasing number of models are present in the literature  but, unfortunately, the experimental data lacks important details such as the metabolites formed or the causes of inhibition.|
Recently, six big pharmaceutical companies joined the Human Cytochromes P450 Consortium , and a large dataset has been generated with a different philosophy, to get as much homogeneous experimental information as possible on the metabolism, the kinetics, the metabolites, and the inhibition, studying different possible mechanisms. Thus, 300 compounds have been homogeneously analyzed in terms of metabolic stability, inhibition, and selectivity using the three major P450 isoforms, namely CYP2C9, CYP2D6 and CYP3A4, that altogether are responsible for 92% of the metabolism of xenobiotics . For a significant number of compounds the experimental profile also included the identification of the metabolites when the molecule was found to be unstable, and the inspection of the mechanisms of inhibition when the molecule was found to be an inhibitor. Thus, for the first time a large set of homogeneous experimental data for human metabolism has been generated.
Such data allowed an in-depth analysis of the orientations of different atoms or groups within the catalytic site, named CYP binding modes (BMs) . BMs can be productive, unproductive, and inhibitory and correspond to different chemical substructure orientations within the catalytic site. These modes affect P450s inhibition, metabolic stability and isoform selectivity, and all of these properties can be understood as competition between BMs.
The novel in silico modeling technology works on the basis of a k-nearest neighbours (k-NN) ‘early classification’, and consequent projections onto specific submodels. Such projections work as follows: 1) Protomeric handling of the molecule (obtained with the software MoKa ) according to its acid/basic properties and the well-known affinity of the studied CYP (2C9: acid molecules; 2D6: basic; 3A4: neutral). 2) Site of metabolism (SoM) prediction with the software MetaSite . 3) Elaboration of molecular descriptors on the basis of the P450-ligand binding modes. 4) Prediction of specific properties using the derived descriptors and submodels developed in the framework on the P450 Consortium.
Here we present an overview of the data, the core of the new technology, the in silico models developed and their validation across the pharmaceutical companies which joined the P450 Consortium.
 Kirchmair J. et al. (2012) Computational Prediction of Metabolism: Sites, Products, SAR, P450 Enzyme Dynamics, and Mechanisms J Chem Inf Model, 52, 617–648.
 Human P450 Consortium page; http://www.moldiscovery.com/consortia/CYP
 Wang H. and Tompkins L.M. (2008) CYP2B6: New Insights into a Historically Overlooked Cytochrome P450 Isozyme Curr Drug Metab 9, 598-610.
 Carosati E. (2012) Modelling cytochromes P450 binding modes to predict P450 inhibition, metabolic stability and isoform selectivity. Drug Discov Today: Technol published online 17 Oct 2012. http://dx.doi.org/10.1016/j.ddtec.2012.09.007
 Milletti F. et al. (2007) New and original pKa prediction method using grid molecular interaction fields. J Chem Inf Model, 47, 2172-2181.
 Cruciani G. et al. (2012) Exposition and reactivity optimization to predict sites of metabolism in chemicals. Drug Discov Today: Technol published online 6 Dec 2012. http://dx.doi.org/10.1016/j.ddtec.2012.11.001