Determinação de energia livre de ligação por métodos in silico para ligantes da enzima InhA (EC 1.3.1.9) de Mycobacterium tuberculosis

Abstract

Tuberculosis is an infectious disease responsible for about 1.3 million deaths annually worldwide. Despite the appearance of multi-resistant strains of Mycobacterium tuberculosis (Mtb), since the 80´s there is a gap in the development of new antimicrobials. With the advent of bioinformatics and computational molecular biophysics, became possible to test, from a established molecular target, numerous molecules, especially associated with the prediction of binding free energy. In the current dissertation, were selected 14 compounds with recognized activity against the enzyme 2-trans-enoil-ACP redutase (InhA, EC 1.3.1.9) of Mtb. These molecules were divided into three groups. Set 1: 5 compounds with distant values of binding free energy. Set 2: 9 compounds with close binding free energy values and similar molecular structures (derived from Genz 10850). Set 3: 14 ligands, corresponding to the sum of the set´s 1 and 2. Sampling obtained from molecular dynamics and 2 ns of simulations, in explicit solvent, allowed to estimate the free energy of bind associated with the methods MM/GBSA, MM/PBSA, (QM)MM/GBSA, LigScore, DrugScore, AutoDock and SQM. The ranking of the compounds were based in the correlation (R2) between the predicted and experimental values. Results showed similar values of R2 in all tested methods. More accurate methods, such as SQM and (QM)/MM/GBSA, not obtained better correlations in comparison with simplified methods, as LigScore and DrugScore. In general, Set 1 obtained a moderate correlation (R2 of 0.00-0.80). Set´s 2 and 3, showed weak correlation (R2 < 0.40). Despite the satisfactory results of Set 1, the tested methods presented limitations in the ranking of compounds with close values of experimental binding free energy and similar molecular structures, as Set 2.