Patrik Rydberg

SMARTCyp: A 2D Method for Prediction of Cytochrome P450-Mediated Drug Metabolism

SMARTCyp is an in silico method that predicts the sites of cytochrome P450-mediated metabolism of druglike molecules. The method is foremost a reactivity model, and as such, it shows a preference for predicting sites that are metabolized by the cytochrome P450 3A4 isoform. SMARTCyp predicts the site of metabolism directly from the 2D structure of a molecule, without requiring calculation of electronic properties or generation of 3D structures. This is a major advantage, because it makes SMARTCyp very fast. Other advantages are that experimental data are not a prerequisite to create the model, and it can easily be integrated with other methods to create models for other cytochrome P450 isoforms. Benchmarking tests on a database of 394 3A4 substrates show that SMARTCyp successfully identifies at least one metabolic site in the top two ranked positions 76% of the time. SMARTCyp is available for download at http://www.farma.ku.dk/p450.

The SMARTCyp cytochrome P450 metabolism prediction server

SUMMARY The SMARTCyp server is the first web application for site of metabolism prediction of cytochrome P450-mediated drug metabolism. AVAILABILITY The SMARTCyp server is freely available for use on the web at www.farma.ku.dk/smartcyp where the SMARTCyp Java program and source code is also available for download. CONTACT smartcyp@farma.ku.dk; lo@farma.ku.dk SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.

RS-Predictor models augmented with SMARTCyp reactivities: Robust metabolic regioselectivity predictions for nine CYP isozymes

RS-Predictor is a tool for creating pathway-independent, isozyme-specific, site of metabolism (SOM) prediction models using any set of known cytochrome P450 (CYP) substrates and metabolites. Until now, the RS-Predictor method was only trained and validated on CYP 3A4 data, but in the present study, we report on the versatility the RS-Predictor modeling paradigm by creating and testing regioselectivity models for substrates of the nine most important CYP isozymes. Through curation of source literature, we have assembled 680 substrates distributed among CYPs 1A2, 2A6, 2B6, 2C19, 2C8, 2C9, 2D6, 2E1, and 3A4, the largest publicly accessible collection of P450 ligands and metabolites released to date. A comprehensive investigation into the importance of different descriptor classes for identifying the regioselectivity mediated by each isozyme is made through the generation of multiple independent RS-Predictor models for each set of isozyme substrates. Two of these models include a density functional theory (DFT) reactivity descriptor derived from SMARTCyp. Optimal combinations of RS-Predictor and SMARTCyp are shown to have stronger performance than either method alone, while also exceeding the accuracy of the commercial regioselectivity prediction methods distributed by Optibrium and Schrödinger, correctly identifying a large proportion of the metabolites in each substrate set within the top two rank-positions: 1A2 (83.0%), 2A6 (85.7%), 2B6 (82.1%), 2C19 (86.2%), 2C8 (83.8%), 2C9 (84.5%), 2D6 (85.9%), 2E1 (82.8%), 3A4 (82.3%), and merged (86.0%). Comprehensive datamining of each substrate set and careful statistical analyses of the predictions made by the different models revealed new insights into molecular features that control metabolic regioselectivity and enable accurate prospective prediction of likely SOMs.

Sulfoxide, sulfur, and nitrogen oxidation and dealkylation by cytochrome P450

The oxidation and dealkylation of dimethylsulfoxide (DMSO), dimethylsulfide (DMS), and trimethylamine (TMA) by cytochrome P450 has been studied with density functional theory calculations. The results show that the oxidation reactions always occur on the doublet spin surface, whereas dealkylations can take place for both the doublet and quartet spin states. Moreover, DMS is more reactive than DMSO, and S-oxidation is more favorable than S-dealkylation, whereas N-dealkylation is more favorable than N-oxidation. This is in perfect agreement with experimental results, showing that density functional activation energies are reliable and comparable for widely different reactions with cytochrome P450.

General transition-state force field for cytochrome p450 Hydroxylation

We have developed force-field parameters for the hydrogen-abstraction transition state of aliphatic hydroxylation by cytochrome P450 using the Q2MM approach. The parametrization is based on quantum chemical (B3LYP) transition-state structures and Hessian matrices for 24 diverse substrate models (14 in the training set and 10 in the test set). The force field is intended to be applicable to any druglike molecule by the use of the general Amber force field (GAFF) for the substrates. The parameters reproduce the geometries within 0.1 Å and 1.2° for bond lengths and angles, respectively, with no significant differences between the training and test sets. The Hessian matrix is also well reproduced with a correlation coefficient of 0.99. The parametrization is performed by the ideal iterative approach of Norrby and Liljefors, which we have implemented for the Amber software.

Fast prediction of cytochrome P450 mediated drug metabolism

Cytochrome P450 mediated metabolism of drugs is one of the major determinants of their kinetic profile, and prediction of this metabolism is therefore highly relevant during the drug discovery and development process. A new rule‐based method, based on results from density functional theory calculations, for predicting activation energies for aliphatic and aromatic oxidations by cytochromes P450 is developed and compared with several other methods. Although the applicability of the method is currently limited to a subset of P450 reactions, these reactions describe more than 90 % of the metabolites. The rules employed are relatively few and general, and when combined with solvent‐accessible surface area calculations to account for steric accessibility, the method gives a major P450 metabolite as first‐ranked position for 75 % of the substrates, and ranked in the top three for 90 % of substrates for a set of 20 substrates. In combination with docking, it can predict isoform‐specific metabolism, and we apply this on CYP1A2 with very good results on 81 substrates, for which we find a major metabolite ranked in the top three for 90 % of the substrates (100 % in the training set and 87 % in the larger test set).

Prediction of activation energies for aromatic oxidation by cytochrome P450.

We have estimated the activation energy for aromatic oxidation by compound I in cytochrome P450 for a diverse set of 17 substrates using state-of-the-art density functional theory (B3LYP) with large basis sets. The activation energies vary from 60 to 87 kJ/mol. We then test if these results can be reproduced by computationally less demanding methods. The best methods (a B3LYP calculation of the activation energy of a methoxy-radical model or a partial least-squares model of the semiempirical AM1 bond dissociation energies and spin densities of the tetrahedral intermediate for both a hydroxyl-cation and a hydroxyl-radical model) give correlations with r(2) of 0.8 and mean absolute deviations of 3 kJ/mol. Finally, we apply these simpler methods on several sets of reactions for which experimental data are available and show that we can predict the reactive sites by combining calculations of the activation energies with the solvent-accessible surface area of each site.

The Accuracy of Geometries for Iron Porphyrin Complexes from Density Functional Theory

Iron porphyrin complexes are cofactors in many important proteins such as cytochromes P450, hemoglobin, heme peroxidases, etc. Many computational studies on these systems have been done over the past decade. In this study, the performance of some of the most commonly used density functional theory functionals is evaluated with regard to how they reproduce experimental structures. Seven different functionals (BP86, PBE, PBE0, TPSS, TPSSH, B3LYP, and B97-D) are used to study eight different iron porphyrin complexes. The results show that the TPSSH, PBE0, and TPSS functionals give the best results (absolute bond distance deviations of 0.015-0.016 A), but the geometries are well-reproduced by all functionals except B3LYP. We also test four different basis sets of double-zeta quality, and we find that a combination of double-zeta basis set of Schafer et al. on the iron atom and the 6-31G* basis set on the other atoms performs best. Finally, we remove the porphyrin side chains and increase the basis set size systematically to see if this affects the results. We show that basis sets larger than double-zeta quality are not necessary to get accurate geometries, and nonaromatic side chains do not affect the geometries.

Ligand-Based Site of Metabolism Prediction for Cytochrome P450 2D6

A ligand-based method based on the SMARTCyp approach that predicts the sites of cytochrome P450 2D6-mediated metabolism of druglike molecules has been developed. The method uses only two descriptors besides the reactivity from SMARTCyp: the distance to a protonated nitrogen atom and the distance to the end of the molecule. Hence, the site of metabolism is predicted directly from the 2D structure of a molecule, without requiring calculation of electronic properties or generation of 3D structures. Testing on an independent test set gives an area under the curve value of 0.94, and a site of metabolism is found among the top two ranked atoms for 91% of the compounds.

WhichCyp: prediction of cytochromes P450 inhibition

SUMMARY In this work we present WhichCyp, a tool for prediction of which cytochromes P450 isoforms (among 1A2, 2C9, 2C19, 2D6 and 3A4) a given molecule is likely to inhibit. The models are built from experimental high-throughput data using support vector machines and molecular signatures. AVAILABILITY The WhichCyp server is freely available for use on the web at http://drug.ku.dk/whichcyp, where the WhichCyp Java program and source code is also available for download.