Estudo in silico de inibidores da glicogênio sintase quinase 3 beta (GSK-3beta)

Abstract

Alzheimer's disease (AD) is the leading cause of dementia worldwide and is characterized by the accumulation of β-amyloid (Aβ) plaques and the formation of intracellular neurofibrillary tangles. The enzyme GSK-3β plays a central role in these processes, making it a promising target for the development of new therapies. Given the limitations of currently available treatments, computer-aided drug design (CADD) methodologies were employed, including Quantitative Structure–Activity Relationship (QSAR) modeling, 2D similarity-based virtual screening, and structure-based drug design approaches. QSAR models were developed from a dataset of 60 pyrazolone derivatives described as GSK-3β inhibitors. The models were generated using two different approaches: (i) based on SMILES representations (1D/2D descriptors, via AlvaDesc/OCHEM) and (ii) based on DFT-optimized structures (electronic, topological, and geometric descriptors, via Dragon 6.0). These approaches resulted in two main models: D1, containing 12 descriptors (R² = 0.908; Q²LOO = 0.855), and O1, containing 10 descriptors (R² = 0.874; Q²LOO = 0.808). Both models underwent internal and external validation, demonstrating robustness, statistical significance, and good predictive performance. 2D similarity-based virtual screening (performed using SwissSimilarity) yielded 9,343 compounds, which were subsequently reduced to 16 hits based on the Tanimoto coefficient. In parallel, molecular docking screening (performed with PyRx) led to the selection of 28 compounds, which were subjected to consensus docking (GOLD); among these, three were selected for molecular dynamics simulations. The QSAR models were applied to the compounds obtained from 2D similarity screening to predict their biological activity. Pharmacokinetic evaluation (performed using the BOILED-Egg model) of the 19 compounds selected in conjunction with the other analyses, the results reinforced the selection of two promising candidates, compound J and compound 7178171, both showing potential to permeate the blood–brain barrier. These results highlight the potential of in silico approaches for identifying promising GSK-3β inhibitors, contributing to the development of new therapeutic strategies for AD.