Predicting intrinsic clearance for drugs and drug candidates metabolized by aldehyde oxidase.

Metabolism by aldehyde oxidase (AO) has been responsible for a number of drug failures in clinical trials. The main reason is the clearance values for drugs metabolized by AO are underestimated by allometric scaling from preclinical species. Furthermore, in vitro human data also underestimates clearance. We have developed the first in silico models to predict both in vitro and in vivo human intrinsic clearance for 8 drugs with just two chemical descriptors. These models explain a large amount of the variance in the data using two computational estimates of the electronic and steric features of the reaction. The in vivo computational models for human metabolism are better than in vitro preclinical animal testing at predicting human intrinsic clearance. Thus, it appears that AO is amenable to computational prediction of rates, which may be used to guide drug discovery, and predict pharmacokinetics for clinical trials.

Impact of pH on plasma protein binding in equilibrium dialysis.

Many pharmacokinetic analyses require unbound plasma concentrations, including prediction of clearance, volume of distribution, drug-drug interactions, brain uptake analysis, etc. It is most often more convenient to measure the total drug concentration in plasma rather than the unbound drug concentration. To arrive at unbound plasma concentrations, separate in vitro determinations of the plasma protein binding of a drug are usually carried out in serum or in plasma, and the plasma pharmacokinetic results are then mathematically adjusted by this fraction unbound ( f u,p). Plasma protein binding or the drug fraction unbound in plasma ( f u,p) is known to be affected by protein, drug, free fatty acid concentrations, lipoprotein partitioning, temperature, pH, and the presence or absence of other drugs/displacing agents within plasma samples. Errors in f u,p determination caused by lack of adequate pH control in newer assay formats for plasma protein binding (e.g., 96-well equilibrium thin walled polypropylene dialysis plates) will have significant drug-specific impact on these pharmacokinetic calculations. Using a diverse set of 55 drugs and a 96-well equilibrium dialysis plate format, the effect of variable pH during equilibrium dialysis experiments on measured values of f u,p was examined. Equilibrium dialysis of human plasma against Dulbecco's phosphate buffered saline at 37 degrees C under an air or 10% CO 2 atmosphere for 22 h resulted in a final pH of approximately 8.7 and 7.4, respectively. The ratio of f u,p at pH 7.4 (10% CO 2) vs pH 8.7 (air) was >or=2.0 for 40% of the 55 compounds tested. Only one of the 55 compounds tested had a ratio <0.9. Select compounds were further examined in rat and dog plasma. In addition, physicochemical properties were calculated for all compounds using ACD/Labs software or Merck in-house software and compared to plasma protein binding results. Changes in plasma protein binding due to pH increases which occurred during the equilibrium dialysis experiment were not species specific but were drug-specific, though nonpolar, cationic compounds had a higher likely hood of displaying pH-dependent binding. These studies underscore the importance of effectively controlling pH in plasma protein binding studies.

Use of density functional calculations to predict the regioselectivity of drugs and molecules metabolized by aldehyde oxidase.

Aldehyde oxidase is a molybdenum hydroxylase that catalyzes the oxidation of aldehydes and nitrogen-containing heterocycles. The enzyme plays a dual role in the metabolism of physiologically important endogenous compounds and the biotransformation of xenobiotics. Using density functional theory methods, geometry optimization of tetrahedral intermediates of drugs and druglike compounds was examined to predict the likely metabolites of aldehyde oxidase. The calculations suggest that the lowest energy tetrahedral intermediate resulting from the initial substrate corresponds to the observed metabolite >or=90% of the time. Additional calculations were performed on a series of heterocyclic compounds where the products resulting from metabolism by xanthine oxidase and aldehyde oxidase differ in many instances. Again, the lowest energy tetrahedral intermediate corresponded to the observed product of aldehyde oxidase metabolism >or=90% for the compounds examined, while the observed products of xanthine oxidase were not well predicted.

Empirical regioselectivity models for human cytochromes P450 3A4, 2D6, and 2C9.

Cytochromes P450 3A4, 2D6, and 2C9 metabolize a large fraction of drugs. Knowing where these enzymes will preferentially oxidize a molecule, the regioselectivity, allows medicinal chemists to plan how best to block its metabolism. We present QSAR-based regioselectivity models for these enzymes calibrated against compiled literature data of drugs and drug-like compounds. These models are purely empirical and use only the structures of the substrates, in contrast to those models that simulate a specific mechanism like hydrogen radical abstraction, and/or use explicit models of active sites. Our most predictive models use three substructure descriptors and two physical property descriptors. Descriptor importances from the random forest QSAR method show that other factors than the immediate chemical environment and the accessibility of the hydrogen affect regioselectivity in all three isoforms. The cross-validated predictions of the models are compared to predictions from our earlier mechanistic model (Singh et al. J. Med. Chem. 2003, 46, 1330-1336) and predictions from MetaSite (Cruciani et al. J. Med. Chem. 2005, 48, 6970-6979).

Inhibition of recombinant cytochrome P450 isoforms 2D6 and 2C9 by diverse drug-like molecules.

The affinities of a diverse set of 500 drug-like molecules to cytochrome P450 isoforms 2C9 and 2D6 were measured using recombinant expressed enzyme. The dose-response curve of each compound was fitted with a series of equations representing typical or various types of atypical kinetics. Atypical kinetics was identified where the Akaike Information Criterion, plus other criteria, suggested the kinetics was more complex than expected for a Michaelis-Menten model. Approximately 20% of the compounds were excluded due to poor solubility, and approximately 15% were excluded due to fluorescence interference. Of the remaining compounds, roughly half were observed to bind with an affinity of 200 microM or lower for each of the two isoforms. Atypical kinetics was observed in 18% of the compounds that bind to cytochrome 2C9, but less than 2% for 2D6. The resulting collection of competitive inhibitors and inactive compounds were analyzed for trends in binding affinity. For CYP2D6, a clear relationship between polar surface area and charge was observed, with the most potent inhibitors having a formal positive charge and a low percent polar surface area. For CYP2C9, no clear trend between activity and physicochemical properties could be seen for the group as a whole; however, certain classes of compounds have altered frequencies of activity and atypical kinetics.