Target repurposing applied to the search of new chemotherapies against echinococcosis

ALBERCA, L.N.; TALEVI, A. ; FRANCHINI, G. ; CÓRSICO, B.

Buenos Aires

Congreso; DRUG DISCOVERY FOR NEGLECTED DISEASES INTERNATIONAL CONGRESS 2018; 2018

/ Instituto de la Química y Metabolismo del Fármaco FFyB UBA

Human echinococcosis is a parasitic disease caused by tapeworms of the genus Echinococcus, responsible for two forms of the disease in humans: cystic echinococcosis (hydatidosis) and alveolar echinococcosis [1].Fatty acid binding proteins (FABPs) are low molecular weight proteins that bind long chain fatty acids as well as other hydrophobic ligands. Among their functions, we may mention promoting solubilization, trafficking and metabolism of intracellular fatty acids [2]. It has been demonstrated that compounds that modify human FABP function may provide tissue-specific control of lipid signaling pathways, inflammatory responses and metabolic regulation [3].Two Echinococcus granulosus FABPs have been characterized (EgFABP1 and EgFABP2) [4-5], displaying similar structures with some human FABPs [6].Application of computer-aided target repurposing strategies [7] by the development of computational models capable of recognizing new human FABP4 inhibitors, subsequent application of virtual screening and assay of the emerging hits against Echinococcus granulosus.After compilation of known binders of human aFABP from literature, we have inferred 1000 ligand-based classificatory models capable of predicting if an untested compound could be a potential aFABP inhibitor. The models weregenerated using the semicorrelation approach [8]. Validation of the models was performed by retrospective virtual screening, using the DUD-E Library as a source of decoys [9-11]. To improve the predictive ability of the individual models a meta-classifier was obtained from combination of five individual models. The ensemble was used to screen DrugBank 5 database.The best individual model achieved a remarkable AUROC of 0.95 in the retrospective virtual screening. The 5-model ensemble using the MIN operator achieved a significantly higher AUROC (0.99, p