Integrating Spectroscopy, Molecular Docking, and Machine Learning to Decipher Mycotoxin-Protein-Binding Mechanisms.

This study investigates the molecular interactions of two mycotoxins, ochratoxin A (OTA) and citrinin (CIT), with globular proteins, such as human serum albumin (HSA) and lysozyme (LYS), using an integrated spectroscopic, computational, and machine learning (ML) approach. Fluorescence and UV-visible spectroscopy revealed that OTA binds strongly to HSA through static quenching, with binding constants of 1.12 × 105 M-1 at 298 K and 1.33 × 104 M-1 at 323 K, indicating enthalpy-driven stabilization via hydrogen bonding and van der Waals forces. In contrast, CIT displayed weaker polarity-driven interactions with both proteins. ML models, particularly random forest and Gaussian process regression, achieved near-perfect prediction of quenching behavior (R2 ≈ 0.985), surpassing traditional methods. Molecular docking corroborated spectroscopic findings, highlighting stronger OTA-HSA affinity compared to that of OTA-LYS and CIT complexes. Overall, the results demonstrate that combining spectroscopy, ML, and docking offers quantitative and mechanistic insights into mycotoxin-protein binding, with implications for toxicity assessment and detoxification strategies.