ABSTRACT
Tuberculosis causes nearly two million deaths per year world-wide. In addition multidrug-resistant mycobacterial strains rapidly emerge so novel therapeutic approaches are needed. Recently, several promising mycobacterial target molecules were identified, which are involved in bacterial or host cell signalling e.g. the serine/threonine protein kinases, PknB and PknG, NAD kinase and the NAD synthetase. Here we describe some early efforts in the development of novel signal transduction inhibitory anti-mycobacterial drugs using a multiple target approach, with special emphasis on the kinase inhibitory field. Initially, we are using the Nested Chemical Library (NCL) technology and pharmacophore modelling. A hit-finding library, consisting of approximately 19000 small molecules with a bias for prototypic kinase inhibitors from our NCL library and commercial sources was virtually screened against these validated target molecules. Protein structures for the virtual screening were taken from the published three dimensional crystal structures of the enzymes. The hits from the virtual screening were subsequently tested in enzymatic assay systems. Potent hits were then tested for biological activity in macrophages, infected with mycobacteria. The final goal of this exercise is not only to identify potent anti-mycobacterial substances, but also a common pharmacophore for the mycobacterial target PknG in combination with PknB, NAD kinase and/or NAD synthetase. This common pharmacophore still needs to be a unique pharmacophore for the mycobacterial target proteins over human off-targets. Such a pharmacophore might then drive the optimization of a completely new profile of an antibiotic agent with activity against latent mycobacteria and resistance mycobacterial strains.
Subject(s)
Anti-Bacterial Agents/pharmacology , Anti-Bacterial Agents/therapeutic use , Mycobacterium tuberculosis/drug effects , Mycobacterium tuberculosis/metabolism , Signal Transduction/drug effects , Tuberculosis/drug therapy , Tuberculosis/metabolism , Anti-Bacterial Agents/toxicity , Drug Evaluation, Preclinical , Drug Resistance, Multiple, Bacterial , Enzyme Inhibitors/pharmacology , Enzyme Inhibitors/therapeutic use , Enzyme Inhibitors/toxicity , Humans , Macrophages/drug effects , Macrophages/enzymologyABSTRACT
Epidermal Growth Factor Receptor (EGFR) is a high priority target in anticancer drug research. Thousands of very effective EGFR inhibitors have been developed in the last decade. The known inhibitors are originated from a very diverse chemical space but--without exception--all of them act at the Adenosine TriPhosphate (ATP) binding site of the enzyme. We have collected all of the diverse inhibitor structures and the relevant biological data obtained from comparable assays and built prediction oriented Quantitative Structure-Activity Relationship (QSAR) which models the ATP binding pocket's interactive surface from the ligand side. We describe a QSAR method with automatic Variable Subset Selection (VSS) by Genetic Algorithm (GA) and goodness-of-prediction driven QSAR model building, resulting an externally validated EGFR inhibitory model built from pIC50 values of a diverse structural set of 623 EGFR inhibitors. Repeated Trainings/Evaluations (RTE) were used to obtain model fitness values and the effectiveness of VSS is amplified by using predictive ability scores of descriptors. Numerous models were generated by different methods and viable models were collected. Then, intensive RTE were applied to identify ultimate models for external validations. Finally, suitable models were validated by statistical tests. Since we use calculated molecular descriptors in the modeling, these models are suitable for virtual screening for obtaining novel potential EGFR inhibitors.