Computational Chemistry Study of Natural Alkaloids and Homemade Databank to Predict Inhibitory Potential Against Key Enzymes in Neurodegenerative Diseases

Cissampelos sympodialis Eichl is used in folk medicine for the treatment of various inflammatory diseases; several alkaloids have been isolated from this species and some of them have anti-allergic, immunomodulatory and spasmolytic activities. Treatment of rats with the total tertiary alkaloid fraction showed an antidepressant effect. One of the depression causes can be the deficiency of monoamines, which is a factor displayed in patients with Alzheimer's disease. Theoretical studies using in silico methods have aided in the process of drug discovery. From this perspective, we applied ligand-based-virtual associated with structure-based-virtual screening of alkaloids from C. sympodialis Eichl and 101 derivatives proposed by us are promising leads against some important targets (BACE, GSK-3μ and MAO-A). From the ChEMBL database, we selected a diverse set of 724, 1898 and 1934 structures, which had been tested against BACE, GSK-3μ and MAO-A, to create Random Forest (RF) models with good overall prediction rate, over 78%, for cross-validation and test set. Compounds 24 and 47 presented activity against GSK-3β and MAO-A simultaneously. The natural alkaloids roraimine and simpodialine-μ-N-oxide presented activity against BACE and liriodenine against MAO-A. The top 20 compounds with best docking performance per enzyme were selected and validated through the RF model. All 9 compounds classified as active in RF model for BACE are bisbenzylisoquinoline alkaloids and were present in the top docking scoring, demonstrating a consensus on results. Affinities of bisbenzylisoquinoline alkaloids, including two secondary metabolites (roraimine and simpodialine-μ-N-oxide), with BACE suggest that this skeleton can be used as a model to design new antagonists of this enzyme.