Estudo in silico na busca de alcaloides potencialmente inibidores de Staphylococcus aureus

Abstract

Staphylococcus aureus is considered a species of invasive and treatmentresistant bacteria and the cause of multiple infections in humans and animals. It is the main pathogen responsible for bovine mastitis, a complex disease that is difficult to control in herds. In general, the treatment of infections caused by bacteria relies on the use of antibiotics. Therefore, the development of new antistaphylococcal drugs becomes necessary. In this context, dihydrofolate reductase (DHFR), an essential enzyme in the growth and cell proliferation of bacteria, represents a promising target in the development of new antibiotics. Therefore, the present work seeks, using in silico techniques, to select alkaloids with antistaphylococcal activity. With this objective, two databases were selected from the CHEMBL database, one with structures tested in S. aureus and the other with structures tested in SaDHFR. Descriptor selections for both banks were explored with four machine learning algorithms: linear regression (LR), pace regression (PR), support vector machine (SVM), and multilayer perceptron (MLP). The constructed models were considered acceptable, as they presented R2 and P2 values > 0.6 and Q2 > 0.5. Thus, the best models, with greater external prediction capacity, selected 30 promising alkaloids for inhibiting SaDHFR. Furthermore, in the second stage, a molecular docking study was carried out at the active site of SaDHFR to propose the binding mode of these alkaloids. Ramiflorine A, one of the three alkaloids with the highest affinities in the active site of SaDHFR, has already demonstrated in vitro inhibitory activity against S. aureus, as reported in the literature. Therefore, it is concluded that the results found in this work can contribute to the development of new candidates, which must be experimentally validated.