Discovery of Novel GABAAR Allosteric Modulators Through Reinforcement Learning

Area of Science:

• Computational chemistry
• Pharmacology
• Neuroscience

Background:

• In vitro screening limitations necessitate advanced computational methods for drug discovery.
• Transmembrane receptors, like GABAAR, present significant screening challenges.

Purpose of the Study:

• To apply reinforcement learning (RL) guided virtual screening for identifying novel gamma-aminobutyric acid A receptor (GABAAR) modulating peptides.
• To demonstrate the efficacy of RL in discovering modulators for difficult-to-screen targets.

Main Methods:

• Structure-based virtual screening utilizing a GABAAR homology model.
• Synthesis and patch-clamp analysis of screened, novel molecules.
• Application of reinforcement learning to guide the virtual screening process.

Main Results:

• 13 novel peptide molecules were synthesized.
• 11 synthesized peptides demonstrated positive allosteric modulation of the GABAAR.
• Two peptides achieved 50% activation at low micromolar concentrations.

Conclusions:

• Reinforcement learning guided virtual screening is a viable strategy for discovering novel modulators of challenging transmembrane receptors.
• This approach accelerates the identification of potential therapeutics for difficult-to-screen targets like GABAAR.