Off-target prediction of SGLT2 inhibitors: an integrative bioinformatics approach to uncover structural mechanisms

Abstract

Gliflozin is known to inhibit sodium glucose transporter 2 (SGLT2) inhibitors (SGLT2i). They have been approved for the treatments of diabetes mellitus, cardiovascular diseases, and chronic kidney disease in recent years. However, the mechanisms for the multifunction remain unclear. We here propose a hypothesis of other protein targets for gliflozins. Additionally, the multistage of the protein, the specificity of the drugs, along with the structural data shortage have posed a hindrance to disclose gliflozin and its binding environment (BE). Considering the difficulties on the issue, this thesis provides an approach to uncover protein off-targets and to find the underlying pathways. We predicted pyranose was a critical substructure from > 1,500 SGLT2i. Hierarchical clustering on atom-based interaction, combined with propensity, revealed key interaction among 1,572 pyranose BEs. The other criterium presented was the concept of compound spanning space via protein pocket size. Binding sites (BSs) from proteins like SGLT2 were able to provide similar pockets, approximating the occupied space of the ligands. Finally, gliflozin binding pockets prefer (1) aromatic residues for van der Waals force, (2) aspartate or asparagine as hydrogen bond providers, and (3) protein pocket size [greater double equals]442 [A] 2. These criteria were integrated into a scoring function ST and predicted several possible proteins. This research presents a pipeline for target identification of emerging drug agents and proteins with crystallographic difficulties. The study provides an innovative computational methodology on extraction of binding features on transmembrane proteins and further pharmaceutical development.