Cosolvent Simulations with Fragment-Bound Proteins Identify Hot Spots to Direct Lead Growth

Area of Science:

• Computational chemistry
• Drug discovery and development
• Structural biology

Background:

• Fragment-based drug design involves sequential addition of chemical groups to weak-binding fragments to create potent lead molecules.
• Static protein structures from crystallography cannot fully capture dynamic binding site adaptations, hindering lead optimization.
• A dynamic approach is crucial for understanding and predicting binding site behavior during lead development.

Purpose of the Study:

• To introduce a dynamic computational framework for identifying binding hot spots to guide lead molecule growth.
• To explore the utility of mixed-solvent molecular dynamics (MixMD) for mapping potential lead space in protein targets.
• To investigate binding cooperativity between protein fragments and potential chemical additions.

Main Methods:

• Utilized mixed-solvent molecular dynamics (MixMD) simulations on 20 protein-fragment systems and their apo forms.
• Analyzed simulation data to identify energetic "hot spots" indicative of favorable binding locations.
• Assessed the influence of the bound fragment on the binding of simulated solvent probes.

Main Results:

• MixMD successfully mapped binding hot spots in 17 out of 20 tested systems, guiding potential chemical modifications.
• Identified specific directions and types of chemical modifications for improving binding affinity.
• Observed cases where mapping was appropriately absent due to lead growth away from the protein or minimal binding opportunities.

Conclusions:

• MixMD provides a generalizable framework for extracting molecular features to select chemical groups for lead molecule expansion.
• The method is computationally tractable due to short simulation timescales, making it suitable for the pharmaceutical industry.
• This study represents the first application of cosolvent MD with bound inhibitors for hot spot identification.