Assessment of insect GABA A receptor homology models performance for virtual screening and their validation by molecular dynamics simulations

MARCOS A. VILLARREAL; IVÁN FELSZTYNA; DANIEL A. GARCÍA; MARÍA SALOMÉ GASTALDI

Congreso; Reunión Anual SAB 2020, Sociedad Argentina de Biofísica; 2020

Insect nervous system is the main target for the most used insecticides, as is the case of GABAA receptor (GABAA-R) non-competitive antagonists (NCAs). In insects, the homopentamer formed by the Rdl subunit (Rdlhomo5) is the most representative type of GABAA-R. Rdlhomo5 presents several binding sites for NCAs. The NCA-IA site is located inside the channel pore and is targeted by polychlorocycloalkanes, cyclodienes and phenylpyrazoles. Also, a NCA-II site is located in the transmembrane domains interfaces and is targeted by fluralaner. Since there are no experimental 3D structures of any insect Rdlhomo5, in a previous work we performed homology modelling of Aedes aegypti Rdlhomo5 using templates in different pharmacological states. We aim to validate structural models suitable for performing virtual screening of new NCAs. We have now evaluated ten different Rdlhomo5 homology models by a retrospective virtual screening. A set of active ligands for the NCA-I site and a set of property-matched decoys were subjected to molecular docking. Virtual screening performance in these models was evaluated by the calculation of ROC curves and ligand enrichment factors. The model with the best performance was embedded in a POPC bilayer and subjected to ̴200 ns molecular dynamics simulations, both in the apo state and in complex with fipronil, a NCA-I site binder insecticide. The protein showed small RMSD values in both states, indicating the structure stability of the homology model. Also, the key interactions between fipronil and the Rdlhomo5 remained stable through the entire MD simulation time. We are currently evaluating the Rdlhomo5 models in regards to their performance for molecular docking in the NCA-II site. Our results show that several models based on different templates must be evaluated to obtain a trustworthy structure for the application of structure-based virtual screening in insect GABAA-R homology models.