Poster 34: Virtual Screening to Identify Novel Antimalarial ChemotypesAlexandre S. Lawrenson1, Neil G. Berry1, Raman Sharma1, Paul M. O’Neill1, Steve A. Ward2, Giancarlo A. Biagini2, Nick Fisher2, Alison E. Shone2
|1Robert Robinson Laboratories, Department of Chemistry, University of Liverpool, Crown Street, Liverpool, L69 7ZD, UK|
2Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, U.K.
|Approximately 40% of the world’s population are exposed to the risks of malaria, resulting in roughly one million deaths annually. Previous successes in attempting to eradicate the disease were only short lived, due to increased resistance of the parasite to established drugs such as chloroquine. Computer aided-drug design is an essential part of modern drug discovery, not least in the field of antimalarial chemotherapy. Our research has been to find novel structural chemotypes which are active against the P. falciparum cytochrome bc¬1 complex (Pfbc1), which has been confirmed as an antimalarial target through the study of drugs such as atovaquone. |
Many ligand based virtual screening techniques were used to identify potential lead like structures that were active against Pfbc1, based on a number of known inhibitors previously identified and tested in-house. Several screening methods involved the use of just the structures which had previously been identified as active, and include fingerprint similarity searching, TurboSimilarity searching and substructure searching. However, several classification methods were also considered such as principle component analysis, Bayesian classification and decision tree analysis, and required both the structures of the active and inactive structures. These methods required the use 0, 1 and 2D molecular descriptors which were calculated in DRAGON, to numerically express much of the chemical/molecular information of the molecules. The virtual screening methods were used to screen the Zinc lead like library of commercially available compounds, with hits being scored based on factors such as their physicochemical properties, as well as their similarity to compounds already known to be active. Multiple ligand based virtual screening methods were employed to allow for a consensus approach, that is compounds which were identified by multiple methods were given preference over those identified by single methods, as they had the most support behind them. Filters were applied such as Lipinski’s rule and Veber’s guidelines for oral bioavailability, to remove compounds with unfavourable physicochemical properties. Additional filters were applied to remove compounds containing Rishton fragments or known toxicophores from the literature, as such structural motifs may produce unwanted biological interactions. Diversity analysis was performed to select a representative subset of compounds from the resulting hits, and included fingerprint analysis, clustering methods such as AGNES and CLARA, and simply selecting the highest scoring compounds. Following visual inspection by medicinal chemists, nineteen compounds were purchased and biological testing performed. Initial testing was against the whole cell growth inhibition assay for the 3D7 strain of the parasite, with five compounds reporting IC50 values ranging from 4.5 µM to 8 µM. Additional in vitro study showed the compounds to be inactive against bovine bc1, which is promising as strong bovine bc1 inhibition has been shown to be indicative of cardiotoxicity in humans. Molecular docking was also performed to propose docking poses of the compounds in Pfbc1. Of the resulting hits there were several promising chemotypes which have emerged. These have gone on to drive forward the next phase of the molecular design loop through synthetic optimisation, to ultimately develop more potent inhibitors of Pfbc1.