In silico testing of flavonoids as potential inhibitors of protease and helicase domains of dengue and Zika viruses

Background. Dengue and Zika are two major vector-borne diseases. Dengue causes up to 25,000 deaths and nearly a 100 million cases worldwide per year, while the incidence of Zika has increased in recent years. Although Zika has been associated to fetal microcephaly and Guillain-Barre syndrome both it and dengue have common clinical symptoms such as severe headache, retroocular pain, muscle and join pain, nausea, vomiting, and rash. Currently, vaccines have been designed and antivirals have been identified for these diseases but there still need for more options for treatment. Our group previously obtained some fractions from medicinal plants that blocked dengue virus (DENV) infection in vitro. In the present work, we explored the possible targets by molecular docking a group of molecules contained in the plant fractions against DENY and Zika virus (ZIKV) NS3-helicase (NS3-hel) and NS3-protease (NS3-pro) structures. Finally, the best ligands were evaluated by molecular dynamic simulations. Methods. To establish if these molecules could act as wide spectrum inhibitors, we used structures from four DENV serotypes and from ZIKV. ADFR 1.2 rcl software was used for docking analysis; subsequently molecular dynamics analysis was carried out using AMBER20. Results. Docking suggested that 3,5-dicaffeoylquinic acid (DCA01), quercetin 3-rutinoside (QNR05) and quercetin 3,7-diglucoside (QND10) can tightly bind to both NS3-hel and NS3-pro. However, after a molecular dynamics analysis, tight binding was not maintained for NS3-hel. In contrast, NS3-pro from two dengue serotypes, DENV3 and DENV4, retained both QNR05 and QND10 which converged near the catalytic site. After the molecular dynamics analysis, both ligands presented a stable trajectory over time, in contrast to DCA01. These findings allowed us to work on the design of a molecule called MOD10, using the QND10 skeleton to improve the interaction in the active site of the NS3-pro domain, which was verified through molecular dynamics simulation, turning out to be better than QNR05 and QND10, both in interaction and in the trajectory. Discussion. Our results suggests that NS3-hel RNA empty binding site is not a good target for drug design as the binding site located through docking is too big. However, our results indicate that QNR05 and QND10 could block NS3-pro activity in DENV and ZIKV. In the interaction with these molecules, the sub-pocket-2 remained unoccupied in NS3-pro, leaving opportunity for improvement and drug design using the quercetin scaffold. The analysis of the NS3-pro in complex with MOD10 show a molecule that exerts contact with sub-pockets S1, S1', S2 and S3, increasing its affinity and apparent stability on NS3-pro.