Antimalarial drug design: targeting the plasmodium falciparum cytochrome bc1 complex through computational modelling, chemical synthesis and biological testing

摘要

Malaria is a life-threatening disease which is responsible for roughly one million deaths annually. Previous successes in attempting to eradicate the disease have only been short lived, owing to the increased development of resistance in the parasite. There is a continued need for novel compounds which act at novel therapeutic targets, with the Plasmodium falciparum cytochrome bc1 complex (Pfbc1) representing one such target. Its inhibition halts the biochemical generation of ATP, thus resulting in parasite cell death. Work described in this thesis was concerned with utilising molecular modelling, synthesis and biological testing to develop novel antimalarial compounds, which selectively inhibit this target. The structural details of a number of compounds known to be active or inactive against Pfbc1 were used in combination with six different ligand based virtual screening techniques, and applied to the ZINC lead like library of compounds to identify potential chemotypes active against malaria. These methods included fingerprint similarity searching, principal component analysis, and naïve Bayesian classification. The hits from each of these methods were merged and formed part of a consensus analysis in which compounds identified across several methods were deemed of more interest than those which appeared less frequently. Each molecule was given a score based on its occurrence in the virtual screening methods and also its physicochemical properties. Compounds were filtered to remove those with unfavourable chemical properties, or which contained known toxicophores. 19 compounds were ultimately purchased and tested in vitro against the 3D7 strain of the malaria parasite. 5 of the compounds reported single digit µM IC50 values, with each containing novel structural chemotypes. The lead candidate contained a benzothiazole core, and reported an IC50 value against 3D7 of 4.53 ± 1.86 µM. Additional testing 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 extensively employed to rationalise the activity of Pfbc1 inhibitors such as atovaquone and HDQ. A number of quinolone containing compounds were also subject to docking, with key observations made with regard to interactions thought to be crucial to their antimalarial activity. The hits from LBVS were also the focus of docking, further supporting their potential as Pfbc1 inhibitors. QSARs were developed for a series of 4-aminoquinoline compounds which had been tested against both the NF54 and K1 strains of malaria. MLR, PLS and kNN machine learning methods were investigated, with molecular descriptors contained within valid models interpreted. Significant models were identified and shown to have strong predictive abilities for both strains. QSAR models were similarly developed for a series of thiazolide compounds with activity against hepatitis C. SVM was found to give a significant model which was able to predict the cell safety of the thiazolide derivatives. The rational design of the novel pyrroloquinolone chemotype led to the synthesis of 7 synthetic analogues to investigate its SAR, via alkylation and Winterfeldt oxidation reactions. The compounds reported 3D7 activity values between 75 nM and 1.02 µM, with molecular docking supporting their potential for Qo binding and thus Pfbc1 inhibition.