Triagem virutal para prospecção de inibidores contra a protease NS2B-NS3 de Zika vírus

• Ano de defesa: 2023
• Autor(a) principal: Beatriz Murta Rezende Moraes Ribeiro
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; Brasil; ICB - DEPARTAMENTO DE BIOQUÍMICA E IMUNOLOGIA; Programa de Pós-Graduação em Bioquímica e Imunologia; UFMG
• Palavras-chave em Português: Zika vírus; Planejamento de fármacos; Triagem virtual; Bioquímica e Imunologia; Zika virus; Preparações Famacêuticas; Triagem
• Link de acesso: http://hdl.handle.net/1843/72735

Zika disease is caused by the Zika virus and is known to cause mild symptoms but also result in severe complications such as Zika congenital syndrome in newborns and Guillain-Barré syndrome in adults. There are no approved treatments, thus further research in this direction is needed. The NS2B-NS3 protease plays a crucial role in viral replication, making it a promising target for drug development studies. This work integrated computational and experimental approaches to identify protease inhibitors. The virtual screening protocol was developed considering experimentally resolved enzyme structures, taking into account its overall structure and active site conformation. Dock6 was employed for docking, and the protocol was validated considering the reproduction of crystallographic poses (cross-docking) and its screening performance (based on evaluation of the ROC curve's AUC and enrichment factor). Cross-docking results did not succeed in any structure considered. Due to the prevalence of macrocyclic ligands, we evaluated the impact of using entry conformations more similar to the crystallographic poses on the success of cross-docking, but no significant improvement was observed. ROC curves were constructed using two test sets: decoys generated by DUD-E and inactive compounds reported in the literature taken from Abrams (2020). The curve for the first group showed low performance (AUC = 0.487), while the second exhibited excellent performance (AUC = 0.962). However, this result may be biased due to the characteristics of inactive ligands, which predominantly have charges of 0 and 1, compared to active ligands among which charges of 3 dominate. Besides, the active site has a negative electrostatic potential, and the scoring function employed does not consider the desolvation energy, favoring compounds with a positive formal charge. Virtual screening of the BraCoLi library resulted in a selection of sixteen compounds, of which 10 were made available and subjected to in vitro experiments. All showed low inhibition when evaluated at 100 µM. The ligand BR20151 (31% inhibition) underwent additional characterizations indicating an aggregating behavior for the molecule.