Abstract:
The use of computational chemistry can potentially reduce the cost of laboratory work and clinical trials, while speeding up the process of drug discovery. This can be achieved by the effective use of computational tools such as molecular modelling and in silico virtual screening. This thesis includes essential steps of the drug discovery process, including the importance of molecular modelling and virtual screening in modern drug discovery. Various approaches such as molecular modelling, similarity searching, structure-based drug design (SBDD), and ligand-based drug design (LBDD) were used successfully to find potential inhibitors for a number of enzymes and proteins. The thesis also discusses the advantages and disadvantages of different approaches. The technical aspect includes the availability of different tools and the rationale behind the method selections for these computational processes. Virtual screening is an integral part of the thesis, and all required parameters, including method validation, have been discussed to justify it. In this thesis, the usefulness of molecular modelling and virtual screening through a natural product library to find potential inhibitors for drug development are presented. Novel lead compounds that targeted biological targets such as ICL1, MCR1, and WEE1 were developed. These results were obtained by a well-designed virtual screening approach in which GOLD software and a natural product library were used for the first time in order to discover the success rate of these targets.
The work began with a virtual screening against the Isocitrate lyase I (ICL1) enzyme, a protein involved in tuberculosis. The virtual screening was carried out with 9050 natural products, and resulted in the identification of 42 potential hits. These hits were investigated and confirmed using biological assays such as NMR and thermal shift assays. We were able to discover two weak inhibitors through this virtual screening. Further development and design led to the establishment of a structure-activity relationship (SAR) with a known inhibitor itaconic acid by molecular modelling. Potential hit compounds were investigated with SAR, which was a starting point for further development of new potent inhibitors of ICL1.
The second virtual screening was carried out against Mobilised Colistin Resistance (MCR1). Out of 9050 natural products, 30 hits were identified and investigated further, using biological studies such as NMR-based assay, Minimum Inhibitory Concentration (MIC), and KD determination (binding constant). Based on those studies, two promising compounds were investigated, and collaborators synthesised new inhibitors. Synthesised molecules showed better binding affinity towards MCR1. The project is still in progress with future work appearing promising.
The third virtual screening was conducted against WEE1, a protein involved in gastric cancer. The virtual screening was carried out with 9050 natural compounds, and 33 hits were identified. Two promising compounds were identified using the MTT assay and Western Blotting technique. The inhibitions of these two compounds will be verified by kinase assay of WEE1 in the future.
Predicting biological properties by computational software is highly crucial. The reliability of this kind of tool is always questionable from the viewpoint of biologists and chemists. In the last stage of this thesis, prediction capabilities and limitations were investigated for commercially available computer software tools for five permeability descriptors, including a Blood-Brain Barrier (BBB), Caucasian Colon Adenocarcinoma (Caco-2), Madin-Darby Canine Kidney (MDCK), skin permeability, and oral bioavailability.