Discovery of Novel NMDA Receptor Inhibitors via Deep Learning

Abstract

This study presents a deep learning-based approach for discovering novel inhibitors of GluN1/GluN3A NMDA receptors. NMDA receptors play critical roles in neurological functions, and dysregulation is implicated in various neurological diseases. The GluN1/GluN3A subtype is an emerging therapeutic target, but selective inhibitors remain scarce.

We applied AI-based virtual screening methods to systematically explore the chemical space and identify novel lead compounds against the GluN1/GluN3A NMDA receptor.

Method

The discovery pipeline combines multiple computational approaches:

• Deep Learning-Based Virtual Screening: Leveraging trained molecular representation models to screen large compound libraries
• Molecular Docking: Structure-based assessment of binding poses and affinities
• Lead Optimization: Iterative optimization of initial hits using computational tools
• Experimental Validation: Electrophysiology assays to validate predicted inhibitors

Key Results

• Novel Inhibitors: Discovered multiple novel GluN1/GluN3A NMDA receptor inhibitors
• AI-Driven Efficiency: Deep learning methods significantly accelerated the hit identification process
• Validated Leads: Identified compounds showed inhibitory activity in electrophysiology experiments
• Therapeutic Potential: The discovered compounds provide new starting points for neurological drug development

BibTeX

@article{wang2025discovery,
  title={Discovery of novel GluN1/GluN3A NMDA receptor inhibitors using a deep learning-based method},
  author={Wang, Shi-hang and Zeng, Yue and Yang, Hao and Tian, Si-yuan and Zhou, Yong-qi and Wang, Lin and Chen, Xue-qin and Wang, Hai-ying and Gao, Zhao-bing and Bai, Fang},
  journal={Acta Pharmacologica Sinica},
  year={2025},
  doi={10.1038/s41401-025-01571-1}
}