Bond-Based 2D Quadratic Fingerprints in QSAR Studies. Virtual and In Vitro Tyrosinase Inhibitory Activity Elucidation

Abstract

In this report, we show the results of QSAR (quantitative structure-activity relationship) studies of tyrosinase inhibitory activity, by using the bond-based quadratic indices as molecular descriptors (MDs) and linear discriminant analysis (LDA), togenerate discriminant functions to predict the anti-tyrosinase activity. The best two models (Eqs.8 and 14) of the total 12 QSAR models developed here show accuracies of 93.51% and 91.21%, as well as high Matthews correlation coefficients (C) of 0.86 and 0.82, respectively, in the training set. The validation external series depicts values of 90.00% and 89.44% for these best two equations 8 and 14, correspondingly. Afterwards, a second external prediction data was used, to perform a virtual screening of compounds reported in the literature as active (tyrosinase inhibitors). In a final step, a series of lignans is analyzed by using the in silico developed models, and in vitro corroboration of the activity is carried out. An aspect of great importance to remark here, is that all compounds present greater inhibition values than Kojic Acid (standard tyrosinase inhibitor: IC50 =16.67µM). The current obtained results could be used as a method to increase the speed, in the biosilico discovery of leads for the treatment of skin disorders.