Identification of novel DYRK1A inhibitors as treatment options for alzheimer's disease through comprehensive in silico approaches.

This study aims to identify potential DYRK1A inhibitors from a curated database and utilize a QSAR model to predict the bioactivity of drug compounds in inhibiting the enzyme involved in tau protein oligomerization, a key process in AD pathology. 192 compounds were sourced from the SuperNatural 3.0 database and docked against DYRK1A using Maestro 12.5. The top five lead compounds and the reference drug Abemaciclib underwent ADMET profiling via the AI Drug Lab Server and a 200 nanosecond molecular dynamics simulation using Desmond. A machine learning-based Quantitative Structure-Activity Relationship (QSAR) analysis was then performed to predict their biological activity based on pIC50 values. The top five compounds, identified as 45,934,388, CNP0344929, CNP0360040, CNP0309850, and CNP0426983, demonstrated binding affinities of -13.337, -12.746, -11.712, -11.656, and - 11.416 kcal/mol, respectively, outperforming Abemaciclib (-6.528 kcal/mol). None of the compounds violated Lipinski's Rule of Five, and all exhibited favorable ADMET profiles, including optimal blood-brain barrier penetration and structural stability. The QSAR model successfully predicted the pIC50 values of the hit compounds (6.16, 5.758, 5.752, 6.003, 5.982), comparable to Abemaciclib (6.32). These findings highlight five promising DYRK1A inhibitors with potential therapeutic applications for AD.