Combined computational and bioinformatic approach to uncover the pre-covalent protein interactions of auranofin and its chlorido derivative Au(PEt₃)Cl

Auranofin (AF) is a clinically approved gold(I) metallodrug with recognized anti-inflammatory and anticancer properties, whose mechanism of action relies on the covalent binding at key selenoproteins and thiols causing their irreversible deactivation. While the final covalent binding event is well-documented, the initial non-covalent recognition phase that precedes it, and which likely governs the drug's selectivity, remains poorly characterized by experimental methods. To address this gap, we employed density functional theory (DFT) calculations to systematically investigate the weak, pre-covalent interactions between auranofin (AF) or its chlorido derivative, Au(PEt₃)Cl (AFCl), with model protein residues. Our results reveal distinct non-covalent interactions preferences for each drug: AF shows a stronger affinity for charged amino acid residues, while AFCl exhibits a marked preference for aromatic and some charged residues. We demonstrate that these initial non-covalent interactions induce a significant redistribution of electron density. This effect alters the local electronic properties of the gold center and its bond to the labile ligand, effectively priming the drug for subsequent covalent attack. We then utilized the computationally derived geometric assets to perform a comprehensive motif search within the Protein Data Bank (PDB) database, which identified ten protein targets with significant therapeutic relevance. This bioinformatic analysis provided a general picture of how these gold compounds navigate their biological environment and led to the identification of targets. This pre-covalent interaction with protein is not a random anchoring process but a crucial preparatory step for the targeted attachment of gold-based drugs.