One class classification for the detection of β2 adrenergic receptor agonists using single-ligand dynamic interaction...

Area of Science:

• Pharmacology and Computational Chemistry
• Structural Biology and Drug Discovery

Background:

• G protein-coupled receptors (GPCRs) are crucial drug targets involved in numerous biological processes.
• Ligand interactions with GPCRs dictate signaling outcomes, necessitating precise pharmacological profiling for drug development.
• Structure-based virtual screening (SBVS) aids ligand prioritization but relies heavily on structural data accuracy and binding mode interpretation.

Purpose of the Study:

• To develop an improved SBVS method for biased selection of ligands with specific pharmacological properties.
• To leverage conformational dynamics of protein-ligand complexes for enhanced ligand screening.
• To refine the identification of agonists for GPCR targets.

Main Methods:

• Utilized molecular dynamics (MD) simulations of a reference agonist bound to the β2 adrenergic receptor.
• Extracted interaction patterns from MD trajectories and encoded them into graphs.
• Trained a one-class machine learning classifier on these interaction graphs to identify agonist-specific binding modes.

Main Results:

• The developed method effectively filtered out irrelevant poses from retrospective virtual screening data.
• The classifier successfully discarded inactive and non-agonist ligands while identifying agonists.
• Performance was evaluated across various conditions, including agonist affinity, simulation duration, and contact types.

Conclusions:

• The proposed method enhances SBVS by incorporating conformational dynamics and machine learning.
• Consistency of the ligand binding mode during simulations is critical for successful ligand selection.
• This approach offers a more accurate way to prioritize drug candidates targeting GPCRs.