Better QSAR: Amphiphilicity in transport and multiple-mode receptor binding

This study deals with two important aspects of quantitative structure-activity relationships (QSAR) in drug action: trans-bilayer transport and drug-receptor interactions. Accordingly, it was formally divided into two subprojects.; The first subproject aims at improvement in the transport description by characterization of drug amphiphilicity. Surrogate systems are used to characterize trans-bilayer transport because the process has a complicated kinetics in cellular systems due to the presence of multiple bilayers, protein binding, and metabolism. Current surrogates exhibit various shortcomings, such as (1) the lack of structure (1-octanol/water) that makes it unsuitable for amphiphilic compounds; (2) insufficient stability (liposomes) that precludes large-scale screening; and (3) smaller thickness and restricted flexibility of fatty acyl chains (immobilized artificial membranes-IAM), affecting partitioning of larger lipophilic molecules. A novel system overcoming these disadvantages was developed. It consists of nonporous octadecylated silica particles with an adsorbed monolayer of phospholipids. The system was used to characterize the amphiphilicity of 40 compounds.; The second subproject is oriented towards a more realistic description of drug-receptor interactions. Without structural information, Comparative Molecular Field Analysis (CoMFA) and other 3D-QSAR methods infer details of the binding site using experimental association constants K and structures of a series of ligands. The methods are based on the assumption that the ligands bind in one binding mode (conformation and/or orientation in the binding site) that was repeatedly disproved experimentally. We modified CoMFA to accommodate multi-mode ligand binding using a treatment of multiple-mode equilibria that describe K as the sum of the partial association constants of individual modes. Multi-mode CoMFA analysis was implemented into the most frequently used commercial software for CoMFA (Sybyl(TM), Tripos, Inc.) using its scripting language. In addition to the standard output (a receptor map and a prediction equation), our procedure determines contributions of individual binding modes to overall binding and effectively selects the appropriate binding modes. The user can start the multi-mode CoMFA procedure with a dozen or so plausible binding modes of each ligand instead of the anxious selection of one mode that is currently available in one-mode CoMFA. This feature of our conceptual approach opens the door to automatization of 3D-QSAR.