Abordagem combinada de triagem virtual inversa e baseada em ligantes na identificação de alvos proteicos para afidicolina e novos hits contra Leishmania major

Abstract

Leishmaniasis are a group of diseases caused by protozoa of the genus Leishmania, prevalent in tropical and developing countries. The existing drugs presents serious limitations such as toxicity and parasite resistance, therefore, the search for new therapeutic agents is necessary. Molecular modeling methods are useful to facilitate the design of safer and more efficient antiparasitic drugs, especially from natural sources. There are several natural products active against Leishmania spp., including aphidicolin, a selective inhibitor of viral and human DNA polymerase-α, produced by fungi Cephalosporium aphidicola and Nigrospora sphaerica, with an unknown antiparasitic mechanism of action. Hence, this work aimed to find potential biological targets of aphidicolin and derivatives through inverse virtual screening. We used Protein Data Bank (PDB) to search Leishmania major targets. The targets selected for molecular docking on DockThor web server were those whose ligands showed molecular overlay values >0.5 relative to aphidicolin and RMSD values≤2 Å. Thus, we considered N-myristoyltransferase (NMT), methionyl t-RNA synthetase (MetRS) and map-kinase (MAPK) as possible targets of aphidicolans. Considering the poor pharmacokinetic and physicochemical properties of aphidicolin and derivatives, we used PharmaGist server to align the molecules and to build a pharmacophoric model. We evaluated the model by hierarchical cluster analysis (HCA) and Pearson's correlation in Minitab software. Due to the number of hydrophobic properties, we built a pharmacophoric model for each target and we subjected to virtual screening on the Pharmit server. After that, we filtered the hits through drug-similar property calculation in Osiris DataWarrior software, as well toxicity alerts in DEREK software and pharmacokinetics on PreADMET web server. We used SwissADME web server to predict water solubility and synthetic accessibility. In the end, we selected only 02 molecules for NMT, 09 for MetRS (02 in common with NMT) and none for MAPK. In order to evaluate the selectivity of the molecules considering the binding affinity (∆G), we anchored all of them in NTM, MetRS and MAPK on DockThor. Statistical analyzes of the 07 best values of ∆G of each complex were performed in the GraphPad Prism software by means of the ANOVA-one way and ANOVA-two way tests, using the co-crystallized ligands as positive control and the miltefosine as a negative control, because there is no known activity against the mentioned targets. The molecules showed statistically significant ∆G values only when they were anchored with NMT and MAPK, mainly the ligands MP-002-507-460, MP-002-528-375 and MP-002-911-105. However, we observed considerable selectivity for NMT. After analyzing the intermolecular interactions in Discovery Studio software, only MP-002-507-460 and MP-002-911-105 showed more interactions with the amino acid residues of the NMT binding site. The prediction of biological activity through the PASS server revealed that the two molecules, steroid analogues, exhibited a moderate probability of acting as leishmanicidal agents. Thus, the two molecules found through virtual screening are promising candidates for in vitro assays on LmNMT models to validate the theoretical results presented in this work.