MetScore: Site of Metabolism Prediction Beyond Cytochrome P450 Enzymes.

The metabolism of xenobiotics by humans and other organisms is a complex process involving numerous enzymes that catalyze phase I (functionalization) and phase II (conjugation) reactions. Herein we introduce MetScore, a machine learning model that can predict both phase I and phase II reaction sites of drugs in a single prediction run. We developed cheminformatics workflows to filter and process reactions to obtain suitable phase I and phase II data sets for model training. Employing a recently developed molecular representation based on quantum chemical partial charges, we constructed random forest machine learning models for phase I and phase II reactions. After combining these models with our previous cytochrome P450 model and calibrating the combination against Bayer in-house data, we obtained the MetScore model that shows good performance, with Matthews correlation coefficients of 0.61 and 0.76 for diverse phase I and phase II reaction types, respectively. We validated its potential applicability to lead optimization campaigns for a new and independent data set compiled from recent publications. The results of this study demonstrate the usefulness of quantum-chemistry-derived molecular representations for reactivity prediction.

Site of Metabolism Prediction Based on ab†initio Derived Atom Representations.

Machine learning models for site of metabolism (SoM) prediction offer the ability to identify metabolic soft spots in low-molecular-weight drug molecules at low computational cost and enable data-based reactivity prediction. SoM prediction is an atom classification problem. Successful construction of machine learning models requires atom representations that capture the reactivity-determining features of a potential reaction site. We have developed a descriptor scheme that characterizes an atom's steric and electronic environment and its relative location in the molecular structure. The partial charge distributions were obtained from fast quantum mechanical calculations. We successfully trained machine learning classifiers on curated cytochrome P450 metabolism data. The models based on the new atom descriptors showed sustained accuracy for retrospective analyses of metabolism optimization campaigns and lead optimization projects from Bayer Pharmaceuticals. The results obtained demonstrate the practicality of quantum-chemistry-supported machine learning models for hit-to-lead optimization.

Activation barriers of oxygen transformation at the active site of [FeFe] hydrogenases.

Oxygen activation at the active sites of [FeFe] hydrogenases has been proposed to be the initial step of irreversible oxygen-induced inhibition of these enzymes. On the basis of a first theoretical study into the thermodynamics of O2 activation [Inorg. Chem. 2009, 48, 7127] we here investigate the kinetics of possible reaction paths at the distal iron atom of the active site by means of density functional theory. A sequence of steps is proposed to either form a reactive oxygen species (ROS) or fully reduce O2 to water. In this reaction cascade, two branching points are identified where water formation directly competes with harmful oxygen activation reactions. The latter are water formation by O-O bond cleavage of a hydrogen peroxide-bound intermediate competing with H2O2 dissociation and CO2 formation by a putative iron-oxo species competing with protonation of the iron-oxo species to form a hydroxyo ligand. Furthermore, we show that proton transfer to activated oxygen is fast and that proton supply to the active site is vital to prevent ROS dissociation. If sufficiently many reduction equivalents are available, oxygen activation reactions are accelerated, and oxygen reduction to water becomes possible.

Theoretical 57Fe Mössbauer spectroscopy for structure elucidation of [Fe] hydrogenase active site intermediates.

[Fe] hydrogenase is a hydrogen activating enzyme that features a monoiron active site, which can be well characterized by Mössbauer spectroscopy. Mössbauer spectra have been measured of the CO and CN(-) inhibited species as well as under turnover conditions [Shima, S. et al., J. Am. Chem. Soc., 2005, 127, 10430]. This study presents calculated Mössbauer parameters for various active-site models of [Fe] hydrogenase to provide structural information about the species observed in experiment. Because theoretical Mössbauer spectroscopy requires the parametrization of observables from first-principles calculations (i.e., electric-field gradients and contact densities) to the experimental observables (i.e., quadrupole splittings and isomer shifts), nonrelativistic and relativistic density functional theory methods are parametrized against a reference set of Fe complexes specifically selected for the application to the Fe center in [Fe] hydrogenase. With this methodology, the measured parameters for the CO and CN(-) inhibited complexes can be reproduced. Evidence for the protonation states of the hydroxyl group in close proximity to the active site and for the thiolate ligand, which could participate in proton transfer, is obtained. The unknown resting state measured in the presence of the substrate and under pure H2 atmosphere is identified to be a water-coordinated complex. Consistent with previous assignments based on infrared and X-ray absorption near-edge spectroscopy, all measured Mössbauer data can be reproduced with the active site's iron atom being in oxidation state +2.

Electric-field effects on the [FeFe]-hydrogenase active site.

The effect of a homogeneous electric field--as exerted by the protein environment and by an electrode potential--on the reactivity of the active site of [FeFe] hydrogenases is unravelled by density functional theory calculations.

Kinetic modeling of hydrogen conversion at [Fe] hydrogenase active-site models.

By means of density functional theory, we investigate the catalytic cycle of active-site model complexes of [Fe] hydrogenase and study how ligand substitutions in the first coordination sphere of the reactive Fe center affect the free-energy surface of the whole reaction pathway. Interestingly, dispersion interactions between the active site and the hydride acceptor MPT render the hydride transfer step less endergonic and lower its barrier. Substitution of CO by CN(-), which resembles [FeFe] hydrogenase-like coordination, inverts the elementary steps H(-) transfer and H2 cleavage. A simplified kinetic model reveals the specifics of the interplay between active-site composition and catalysis. Apparently, the catalytic efficiency of [Fe] hydrogenase can be attributed to a flat energy profile throughout the catalytic cycle. Intermediates that are too stable, as they occur, e.g., when one CO ligand is substituted by CN(-), significantly slow down the turnover rate of the enzyme. The catalytic activity of the wild-type form of the active-site model could, however, be enhanced by a PH3 ligand substitution of the CO ligand.