Inhibiting NHEJ: In-Silico Approach to Stratifying DNA-PK Inhibitors, Predicting Radio-Halogenation Potential of DNA-PK Inhibitors, and Assessing AlphaFold2 Prediction Accuracy

Background: The current study integrates AlphaFold2-based structural modeling of DNA-PKcs with an experimentally resolved DNA-PKcs structure, large-scale molecular docking of over 1500 known DNA-PK inhibitors, aiming to identify clinically relevant candidates for next-generation pharmacological delivery methods, evaluate receptor model suitability, and predict how potential radio-halogenation may influence binding behavior and translational radiotherapeutic potential of current clinically relevant candidates. AlphaFold2-based prediction of the DNA-PKcs structure was the starting point, followed by comparison of the AI-derived model with an experimentally resolved structure to evaluate the model’s accuracy and docking suitability. A large database of 1369 known DNA-PK inhibitors was then screened by molecular docking, after which six clinically relevant compounds were analyzed for molecular dynamics. Finally, a focused docking study was conducted on 67 possible radio-halogenated derivatives derived from these clinically relevant scaffolds to investigate the effect of potential iodination, bromination, or astatination on binding behavior. These findings indicate that large-scale docking can be used to comparatively rank DNA-PK-targeting compounds and explore the impact of theoretical radiohalogenation on predicted binding behavior. However, the results also underline the limitations of current receptor models and molecular simulation workflows, supporting the need for experimental validation before translational conclusions are drawn.