Identify the Novel Potential 11β-HSD1 Inhibitors Based on Deep Learning, Molecular Modelling and Activity Evaluation

Abstract

11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) has been shown to play an important role in the treatment of impaired glucose tolerance, insulin resistance, dyslipidaemia and obesity and is a promising drug target. In this study, we constructed a GRU-based recurrent neural network using 1,854,484 (processed) drug-like molecules from ChEMBL and the U.S. Patent Database, and successfully constructed a molecular generative model of 11βHSD1 inhibitors by going through the known 11β-HSD1 inhibitors to perform migration learning, and our constructed GRU model was able to accurately capture drug-like molecules evaluated by traditional machine model-related syntax, and migration learning can also easily generate potential 11β-HSD1 inhibitors. Combining Lipski and ADME/T rules to eliminate non-compliant molecules, and stepwise screening by molecular docking and molecular dynamics simulation, we finally obtained 5 potential compounds. We chose compounds 02 and 05 with the highest free energy to validate the in vitro activity experiments, and found that compound 02 possessed inhibitory activity but was not as effective as the control. In conclusion, our study provides new ideas and methods for new drug development and discovery of new 11β-HSD1 inhibitors.

Note:
Funding declaration: This study was supported by the Graduate Student Innovation Programme of Chongqing University of Technology [Grant No. gzlcx20233379].

Conflict of Interests: There are no conflicts to declare.

Keywords: 11β-hydroxysteroid dehydrogenase type 1, Gated recurrent unit, Transfer learning, Virtual screening, Molecular simulation, Activity evaluation