Ensemble-based docking using biased molecular dynamics.

Area of Science:

• Biochemistry
• Computational Chemistry
• Structural Biology

Background:

• Protein flexibility is crucial for molecular recognition and protein-ligand interactions, yet often underrepresented in drug discovery.
• Accurately modeling protein dynamics is computationally challenging for computer-aided drug design.
• Ensemble-based docking offers a strategy to address protein flexibility by docking ligands to multiple protein conformations.

Purpose of the Study:

• To present a novel approach using biased molecular dynamics (MD) simulations to generate protein ensembles for docking.
• To enhance the sampling of relevant protein conformational space by biasing simulations toward known complex structures.
• To validate the method's ability to identify correct protein-ligand binding modes.

Main Methods:

• Utilized biased MD simulations, steered by an initial protein-ligand X-ray complex, to generate diverse protein conformations.
• Employed clustering techniques to select a representative subset of protein conformations from simulation trajectories.
• Performed ensemble-based docking of ligands to the reduced set of protein conformations, including cross-docking validation.

Main Results:

• Biased MD simulations effectively sampled relevant protein conformational space while retaining key pocket-ligand information.
• Docking to the generated ensemble, particularly with cross-docking and lowest-pose selection, reliably identified correct binding modes.
• The method demonstrated validation for target proteins cyclin-dependent kinase 2 and factor Xa.

Conclusions:

• Biased MD simulations provide an efficient and effective method for generating protein ensembles for docking.
• This approach improves the modeling of protein flexibility in structure-based drug discovery.
• The validated method enhances the accuracy of predicting protein-ligand binding modes.