Streamlined system for purifying and quantifying a diverse library of compounds and the effect of compound concentration measurements on the accurate interpretation of biological assay results

Abstract

As part of an overall systems approach to generating highly accurate screening data across large numbers of compounds and biological targets, we have developed and implemented streamlined methods for purifying and quantitating compounds at various stages of the screening process, coupled with automated "traditional" storage methods (DMSO, -20 degrees C). Specifically, all of the compounds in our druglike library are purified by LC/MS/UV and are then controlled for identity and concentration in their respective DMSO stock solutions by chemiluminescent nitrogen detection (CLND)/evaporative light scattering detection (ELSD) and MS/UV. In addition, the compound-buffer solutions used in the various biological assays are quantitated by LC/UV/CLND to determine the concentration of compound actually present during screening. Our results show that LC/UV/CLND/ELSD/MS is a widely applicable method that can be used to purify, quantitate, and identify most small organic molecules from compound libraries. The LC/UV/CLND technique is a simple and sensitive method that can be easily and cost-effectively employed to rapidly determine the concentrations of even small amounts of any N-containing compound in aqueous solution. We present data to establish error limits for concentration determination that are well within the overall variability of the screening process. This study demonstrates that there is a significant difference between the predicted amount of soluble compound from stock DMSO solutions following dilution into assay buffer and the actual amount present in assay buffer solutions, even at the low concentrations employed for the assays. We also demonstrate that knowledge of the concentrations of compounds to which the biological target is exposed is critical for accurate potency determinations. Accurate potency values are in turn particularly important for drug discovery, for understanding structure-activity relationships, and for building useful empirical models of protein-ligand interactions. Our new understanding of relative solubility demonstrates that most, if not all, decisions that are made in early discovery are based upon missing or inaccurate information. Finally, we demonstrate that careful control of compound handling and concentration, coupled with accurate assay methods, allows the use of both positive and negative data in analyzing screening data sets for structure-activity relationships that determine potency and selectivity.