In silico structure-based design of superior DYRK1A enzyme inhibitors.

Abstract

DYRK1A kinase is a multifunctional enzyme involved in cellular homeostasis and signaling regulation, and its dysregulation is associated with various diseases, including neurological disorders. In this study, a 3D-QSAR model was developed using the experimentally determined inhibitory values of 30 potent compounds. The model exhibited strong statistical performance (Q² = 0.504, R² = 0.996, R²test = 0.659), which was further validated by a high R²m value (0.936), a significant F-value (≈569.14), a high bootstrap coefficient (R²bs = 0.995), Williams plot analysis, and Y-randomization testing. 3D-QSAR contour analysis and the enzyme surface electrostatic potential map confirmed favorable compound alignment in the active site. Guided by the model's contour maps, 138,651 novel inhibitors were designed via appropriate substitutions at key positions. After comprehensive screening, including pIC₅₀ prediction via the 3D-QSAR model, drug-likeness evaluation, and Lipinski's Rule of Five, 756 compounds were selected for molecular docking. The novel compounds exhibited higher binding affinity and stronger interactions with Lys188 and Asp307 compared to the dataset compounds. The top five compounds with binding energies between -16.706 and -17.193 kcal/mol were selected for further analysis with 400 ns molecular dynamics simulations. RMSD values below 0.5 nm, low RMSF, stable hydrogen bonds, and consistent SASA profiles indicated the structural stability of all complexes. PBSA and GBSA binding free energy calculations yielded values of -14.980 to -37.380 and -30.320 to -45.640 kcal/mol, respectively, indicating strong ligand-receptor interactions. Comprehensive QTAIM and IGMH analyses provided further insight into the nature and strength of intermolecular interactions.