Machine Learning Navigated Allosteric Network to Unveil Biased Allosteric Modulation of GPCRs.

Area of Science:

• Computational chemistry and pharmacology
• Molecular modeling and simulation
• Machine learning in drug discovery

Background:

• Biased allosteric modulators (BAMs) offer potential for selective G protein-coupled receptor (GPCR) therapeutics.
• Understanding the molecular mechanisms of BAMs is challenging due to their complex nature.
• Current methods lack the ability to fully elucidate the interplay between biased and allosteric modulation.

Purpose of the Study:

• To develop and validate a novel computational strategy for investigating BAM mechanisms.
• To elucidate the molecular mechanism of a specific β-arrestin-biased modulator (SBI-553) targeting NTSR1.
• To provide a framework for analyzing biased allosteric modulation in other GPCR systems.

Main Methods:

• Proposed a machine learning-navigated allosteric network analysis (RMLNA) strategy.
• Employed molecular dynamics (MD) simulations to obtain biased conformation states.
• Utilized an interpretable deep learning model (CNN) to identify key residues.
• Performed allosteric network analysis to understand residue regulation effects.

Main Results:

• RMLNA successfully revealed the biased allosteric modulation mechanism of SBI-553 on NTSR1.
• SBI-553 was shown to stabilize a unique β-arrestin-biased state, expanding the intracellular binding site.
• Key residues in TM5, TM6, H8, and TM7 were identified as crucial for β-arrestin bias and modulation.
• Analysis highlighted the importance of communication pathways between transmembrane helices for biased signaling.

Conclusions:

• The study provides novel molecular insights into the biased allosteric modulation of SBI-553 on NTSR1.
• The developed RMLNA workflow offers a reliable and extendable approach for studying BAMs across different GPCRs.
• Findings contribute to the rational design of safer and more selective GPCR-targeted therapeutics.