Identification of Selective α-Glucosidase Inhibitors via Virtual Screening with Machine Learning

Area of Science:

• Biochemistry
• Pharmacology
• Computational Chemistry

Background:

• Existing alpha-glucosidase inhibitors have limitations.
• There is a critical need for novel, selective inhibitors.

Purpose of the Study:

• To identify selective alpha-glucosidase inhibitors using machine learning-integrated virtual screening.
• To validate the potency and selectivity of identified compounds.

Main Methods:

• Machine learning-integrated virtual screening of 100 K+ compounds.
• Inhibition kinetics and fluorescence quenching assays.
• Molecular docking and in vitro/in vivo starch digestion assays.

Main Results:

• A series of selective alpha-glucosidase inhibitors were identified.
• These inhibitors demonstrated superior selectivity and potency over acarbose.
• Mechanistic studies revealed competitive inhibition via interactions with ASP526.

Conclusions:

• Machine learning-driven virtual screening is effective for discovering potent alpha-glucosidase inhibitors.
• The identified compounds possess a promising inhibitory profile for therapeutic applications.
• Key interactions within the enzyme's active site contribute to enhanced inhibition.