Computational structure-based identification of silymarin as a promising flavonoid inhibitor of the diphtheria toxin catalytic domain

Diphtheria toxin, a major virulence factor of Corynebacterium diphtheriae, continues to pose significant therapeutic challenges in the face of antimicrobial resistance. Flavonoids, owing to their structural diversity and bioactivity, represent attractive scaffolds for inhibitor development. This study employed advanced computational approaches to identify potent flavonoid-based inhibitors targeting the catalytic domain of diphtheria toxin. A curated flavonoid library (n = 264) was subjected to druglikeness and PAINS filtration, yielding 104 candidates for hierarchical molecular docking (HTVS → SP → XP) against diphtheria toxin (PDB ID: 1DTP). Binding free energies were validated by MM-GBSA, while induced-fit docking captured receptor conformational adaptability. Pharmacokinetic and safety attributes were evaluated using SwissADME and DeepPK platforms. Epigallocatechin gallate (EGCG) and the co-crystallized ligand served as benchmarks. Among flavonoids, silymarin recorded the lowest docking score of − 8.399 kcal/mol, indicating stronger interaction with the diphtheria toxin active site compared to EGCG (–7.698 kcal/mol), though higher than the co-crystallized ligand (–11.089 kcal/mol). MM-GBSA confirmed the thermodynamic stability of both silymarin (–64.63 kcal/mol) and EGCG (–63.43 kcal/mol). In terms of inhibition constants, the co-crystallized ligand showed the strongest binding with a $$\:{k}_{i}$$ of 7.34 nM, indicating the highest inhibitory potential. Silymarin exhibited a moderate affinity with a Ki of 694 nM, while epigallocatechin gallate had the weakest inhibition at 2210 nM. ADMET evaluation demonstrated that silymarin exhibited superior druglikeness, no PAINS alerts, and a safer toxicity profile relative to EGCG. Silymarin emerges as a promising lead flavonoid scaffold for diphtheria toxin inhibition, combining favorable binding energetics with strong pharmacokinetic and safety attributes, thereby warranting further experimental validation and optimization.