Estudos quimiotaxonômicos e triagem virtual de sesquiterpenos lactonizados isolados da família asteraceae com potencial atividade leishmanicida e tripanocida

Abstract

Leishmaniasis and American trypanosomiasis are the neglected diseases that affect a large percentage of the Brazil and Colombia population. Some of the difficulties for the control and elimination of these diseases are the limitation and inefficacy of the current treatments mainly in the chronic phases. Natural products are potential sources of new and selective agents for the treatment of tropical diseases caused by protozoa and other parasites. Asteraceae family has allowed the isolation of a great variety of substances that exhibit diverses biological activities, among them, sesquiterpenes lactones, which have not been studied yet deepth, but present a great potential of antiprotozoal activity. Computational methods have emerged, as an important tool in the field of medicinal chemistry, improving the effectiveness of processes to develop of new treatments, reducing costs and resources. Therefore, this study aims use, create and combine several methodologies of virtual screening of a natural products database through molecular docking, and machine learning algorithms using molecular descriptors, in order to find sesquiterpenes lactones with potential trypanocidal and leishmanicidal activity. In Chapter 2, is presented a review of the biological importance of sesquiterpenes lactones in the fight against neglected diseases including some investigations performed with in silico tools in the search of new molecules with leishmanicidal and antichagasic activity. Chapter 3 presents a modern and innovative web tool for database managing of secondary metabolites, called SISTEMATX, which was developed in the cheminformatics laboratory of the Federal University of Paraíba. The last two chapters, present computational studies aimed at finding structures with potential antichagasic and leishmanicidal multitarget activity, using an in-house databank (SISTEMATX) consisting of 1,306 sesquiterpenes lactones. Random forest classificatory models built for each one of the parasitic forms of T. cruzi and L. donovani reached a predictive power above of 75,4% of accuracy. By ligand-based virtual screening, the number of structures classified as active were: 34, 17 and 420 for amastigotes, trypomastigotes and epimastigotes of T. cruzi, as well as, 741 and 11 for amastigotes and promastigotes of L. donovani. Likewise, structure-based virtual screening was performed by molecular docking of the same set of molecules using eight target enzymes of T. cruzi and four proteins of L. donovani, including a homology model of Pteridine reductase 1 (LdPTR1). Finally, by means of a consensus analysis of the two techniques, it was sought to normalize the probability values, to verify potentially active compounds against these two neglected diseases and to establish their possible mechanism of action.