Glycogen Synthase Kinase-3 beta (GSK-3β) is an enzyme playing a crucial role in Alzheimer’s disease by regulating key neuropathological features in the central nervous system. Allosteric inhibitors of this kinase have been validated as a promising therapeutic option due to their selective and subtle modulation, lowering the chance of producing side effects. Nonetheless, only a few GSK-3β allosteric modulators have been uncovered so far. Therefore, we developed QSAR models to discover new potential hits as well as to expand the chemical-biological diversities of compounds within this context. We collected, curated and integratrd a dataset of 88 compounds with known experimental allosteric modulation of GSK-3β (IC<sub>50</sub> activity threshold of 20 µM). QSAR models were developed in KNIME employing Morgan and RDKit descriptors along with the Random Forest algorithm. Models were developed and validated following the best practices proposed by OECD and by employing a 5-fold external cross-validation procedure to estimate the robustness of the models. Overall, the model with mixed Morgan and RDKit descriptors showed the highest correct classification rate of 0.89, sensitivity of 0.84, specificity of 0.95, and positive and negative predictive values of 0.96 and 0.82, respectively. These models were employed in a virtual screening campaign using the Molport Screening Compounds and Enamine databases (more than 10 million compounds). The top 40,000 compounds were sorted according to highest confidence/score predictions, followed by filtration of favorable ADMET properties, and finally docked using Glide to inspect preferential interactions within the allosteric pocket of the enzyme. As a result, ten potential hits were prioritized and their activity as GSK-3β allosteric modulators (currently under experimental evaluation) will be reported at the ACS meeting.

Integrated computational approaches for the discovery of putative GSK-3β allosteric modulators

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.