Using Multiple Screening Strategies to Biologically and Chemically Characterize Natural Products
MetadataShow full item record
Natural products play an important role in the discovery and development of therapeutics and biological probes. However, in recent decades therapeutic screening efforts have moved away from using natural product libraries, instead opting for large synthetic molecule library. This move has corresponded with a move from phenotypic screens to target based screening approaches. The perceived incompatibility of natural product libraries with target based screening efforts is often cited for these shifts. Herein we discuss the benefits of phenotypic screens and advances in bioinformatic approaches to improve natural product discovery using phenotypic screens. We describe the development and implementation of a screen using a natural product fraction library of ~9000 fractions to screen 26 non-small cell lung cancer cell lines for selective cytotoxic compounds. A screen of this magnitude is unprecedented in academia, therefore we developed a process to rapidly identify and characterize natural products of interest. Natural product fractions with selective toxicity were filtered to ~1000 high priority natural product fractions. Using LC-MS analysis and bioinformatic approaches, Elastic Net (EN) and Functional Signatures of Ontology (FuSiOn),we further prioritized these 1000 natural product fractions based on sensitivity and mechanism of action predictions, and chemical complexity. The implementation of this prioritization process has resulted in an effective discovery pipeline. Ikarugamycin, a selective cytotoxin and endocytosis inhibitor, has been characterized for cytotoxicity and as a chemical tool for inhibiting endocytosis. Piericidin A, a known complex I inhibitor, displays extreme selective toxicity to a subset of cancer cell lines independent of its complex I inhibition, however we can predict this sensitivity using a common genetic biomarker. Other natural products have also been identified, although their biological characterization is ongoing. Using FuSiOn we effectivity identified a minor metabolite (N6,N6-dimethyladenosine) responsible for AKT inhibition. FuSiOn was implemented in conjunction with our non-small cell lung cancer cell line screen to characterize the mechanisms of action of natural product fractions. This correlation led to the rapid identification of bafilomycin among our prioritized 1000 natural product fractions. The process we outline herein effectively uses bioinformatics (EN and FuSiOn) in concert with primary screening data to select for those natural product fractions of greatest biologic and chemical significance.