Comparing pharmacophore models derived from crystal structures and from molecular dynamics simulations.

Area of Science:

• Computational chemistry
• Drug discovery
• Structural biology

Background:

• Structure-based pharmacophore models are crucial in computer-aided drug discovery for hit identification and lead optimization.
• The accuracy of these models relies heavily on the precise representation of ligand-protein interactions.
• Concerns exist regarding the fidelity of crystal structures, including non-physiological contacts and solvent effects, which can impact model reliability.

Purpose of the Study:

• To investigate the impact of molecular dynamics (MD) simulations on refining protein-ligand complex structures for pharmacophore modeling.
• To compare pharmacophore models generated from initial Protein Data Bank (PDB) structures versus those derived from the final structures of MD simulations.
• To assess whether MD-refined models improve the discrimination between active and decoy compounds.

Main Methods:

• Generating pharmacophore models using initial protein-ligand complex structures from the PDB.
• Generating pharmacophore models using the final structures obtained from molecular dynamics (MD) simulations of protein-ligand systems.
• Comparing the resulting pharmacophore models in terms of feature number, feature type, and their ability to distinguish active ligands from decoys.

Main Results:

• Pharmacophore models derived from initial PDB structures and final MD simulation structures exhibit differences in feature composition and type.
• In several cases, pharmacophore models built using the final structures from MD simulations demonstrated a superior capacity to differentiate between active and decoy ligand sets.
• The refinement of protein-ligand complex structures through MD simulations can lead to more effective pharmacophore models.

Conclusions:

• Molecular dynamics simulations offer a valuable strategy for refining protein-ligand complex structures, thereby enhancing the quality of structure-based pharmacophore models.
• Utilizing MD-refined structures in pharmacophore modeling can lead to improved performance in virtual screening and hit-to-lead optimization.
• This study highlights the importance of considering dynamic structural aspects in computational drug discovery workflows.