Structure-Guided Discovery of Isoindolinone–Triazolopyrimidine Hybrids as Potential Skp1–Skp2 Inhibitors: Integrated Docking, Molecular Dynamics, and Machine learning modelling

This study presents a structure-guided computational investigation of reported isoindolinone/[1,2,4]triazolo[1,5-a]pyrimidine derivatives as potential ligands targeting the Skp1–Skp2 complex. Molecular docking, molecular dynamics (MD) simulations, MM/PBSA free-energy analysis, and machine-learning-based activity modelling were integrated to examine ligand-recognition features and binding stabilization within the Skp1–Skp2 interfacial region. Docking identified Arg356 as an important residue for initial ligand anchoring, together with additional contacts involving Phe380, His404, and Ser330. However, MD simulations revealed that ligand stabilization after structural relaxation was governed by a broader and dynamically reorganized interfacial environment, with notable contributions from residues including Ile351 and Gly391. MM/PBSA calculations showed that van der Waals interactions are the dominant favorable energetic component, whereas electrostatic contributions are partly offset by polar solvation penalties, indicating that hydrophobic packing and local shape complementarity are major determinants of binding in the evaluated sp³ series. Among the analyzed reported compounds, 4f exhibited the most favorable MM/PBSA binding free energy, while other ligands displayed differing degrees of dynamic stability during simulation. A Random Forest-based QSAR model developed in KNIME showed moderate internal agreement (R²/Q² = 0.558, RMSE = 0.103), while Y-randomization and PCA-based descriptor-space analysis supported the presence of a non-random internal SAR trend and clarified the chemical-space relationship between the reported sp³ compounds and proposed sp² analogues. On the basis of the observed conformational variability of the sp³-rich scaffolds, exploratory QSAR, HOMO–LUMO, and retrosynthetic analyses were further used to justify sp²-enriched analogues as a rational next-step design strategy. Overall, this study provides mechanistic insight into ligand recognition within the Skp1–Skp2 interface and offers a computational basis for future optimization and experimental validation of conformationally constrained inhibitors.