Improving homology modeling from low-sequence identity templates in Rosetta: A case study in GPCRs

Area of Science:

• Structural biology
• Computational chemistry
• Pharmacology

Background:

• Protein structure determination lags behind sequence generation, hindering drug design.
• Homology modeling requires high-identity templates, limiting its application for many G-protein coupled receptors (GPCRs).
• Only 17% of druggable GPCRs have resolved structures, leaving a significant gap for structure-based drug design.

Purpose of the Study:

• To enhance homology modeling accuracy for G-protein coupled receptors (GPCRs).
• To expand the applicability of homology modeling to GPCRs with low sequence identity templates.
• To improve structure-based drug design strategies for a wider range of GPCR targets.

Main Methods:

• Developed a blended sequence- and structure-based alignment method to improve loop region modeling.
• Integrated multiple template structures into a single comparative model to maximize template utility.
• Validated the improved modeling approach on GPCRs with sequence identities as low as 20%.

Main Results:

• Achieved accurate homology modeling for GPCRs using templates with low sequence identity (down to 20%).
• The enhanced method accounts for nearly the entire druggable GPCR space.
• A comprehensive model database for non-odorant GPCRs is now publicly available at www.rosettagpcr.org.

Conclusions:

• Optimized homology modeling pipeline significantly improves accuracy for GPCR structure prediction.
• The approach enables structure-based drug design for a vast majority of previously inaccessible GPCR targets.
• Protocols and insights are provided for adapting the method to new targets, advancing pharmaceutical research.