Computational Studies on Pharmaceutical Targets in Human Diseases

Bacterial multidrug resistance (i.e. the ability of some bacterial species to survive in presence of various drugs) has become a primary challenge at a global level. Due to various factors, such as the overuse of antibiotics in human activities like health care and farming or inadequate diagnostic, many bacteria have indeed evolved acquiring novel and highly efficient resistance mechanisms. Some species, in particular, have become resistant to almost all in-use drugs. Among the several mechanisms of resistance, efflux pumps of the RND superfamily (resistance-nodulation-cell division) play a major role. These complexes span the cell wall and are able to expel a wide range of noxious compounds, including antibiotics of many different classes. In order to reinvigorate the action of these drugs, a viable route is to hinder their transport out of the cell through co-administration of efflux pumps inhibitors (EPIs). At present several EPIs have been identified, but none of them is usable in clinical therapies due to adverse effects. Moreover, several questions are still open regarding the mode of action of known EPIs as well as the functioning mechanism of RND efflux pumps. Further research in this field is thus needed. In order to characterize the mode of action of several EPIs of this pump, we applied computational techniques such as molecular docking and molecular dynamics (MD) simulations. Specifically, we focused on the EPIs: (i) amitriptyline and chlorpromazine, repurposed drugs which were proven to act as inhibitors against AcrB; (ii) PAβN, a known inhibitor of the pump whose mode of action is not fully understood. This thesis focuses on the inhibition of the AcrB efflux pump, the best known representative of the RND superfamily. High-resolution structural data are indeed available for this protein (specifically, for its Escherichia coli orthologue). Moreover, a fluoroquinolone resistant variant of this pump has been detected in clinical environments. With regard to amitriptyline and chlorpromazine, our in silico investigations revealed that both compounds tend to occupy a known binding pocket of AcrB. Their binding mode presents considerable similarities with that of several substrates and other EPIs of the pump, indicating that amitriptyline and chlorpromazine may inhibit the AcrB pump through competitive binding. In the case of PAβN, MD simulations were compared with experimental data from hydrogen-deuterium exchange mass spectrometry. From these analyses, it emerged that PAβN can significantly restrain the conformational dynamics of AcrB and its fluoroquinolone resistant variant. This EPI, therefore, may act by preventing conformational changes that are functional for AcrB. Importantly, our MD simulations revealed that PAβN and the antibiotic ciprofloxacin can simultaneously occupy the same binding pocket, suggesting that the EPI does not act by competitive binding. Further computational analyses were conducted on structural models of Salmonella Typhimurium AcrB. Experimental structural data on this wt protein are indeed missing, while the structure of its fluoroquinolone resistant variant has recently been solved through cryo-electron microscopy (cryo-EM). In order to assess the structural differences between the two proteins, we derived their structural models through homology modelling and MD simulations (modeling of the fluoroquinolone resistant variant was integrated with cryo-EM data). Structural analyses were then performed, with focus on the binding pockets of the protein. Considerable differences were detected regarding the volume as well as the hydration properties of the pockets. Although not strictly related to EPI development, this information may be valuable for the design of novel drugs and/or inhibitors of AcrB from Salmonella.