Improving flexible protein-ligand docking using implicit ligand sampling

Mengang Xu, Purdue University

Abstract

Accurate prediction of protein-ligand interactions and the associated binding affinity is a major task in computer aided drug design. It is crucial for the selection of potential lead and drug candidates. Docking is an efficient molecular modeling method to predict the nativelike ligand binding mode and the corresponding binding affinity. To incorporate protein flexibility and dynamics into docking, a computationally efficient method that samples protein conformations when binding to structurally diverse ligands is highly demanded. We have developed a novel methodology, named 'Ligand model' concept (Limoc) to explore the receptor flexibility during docking simulation. In the ligand-model concept, short molecular-dynamics simulations are performed with a virtual ligand, represented by a collection of functional groups that binds to the protein and dynamically changes its shape and properties during the simulation. The ligand model essentially represents a large ensemble of different chemical species binding to the same target protein. We have demonstrated that Limoc yields significant improvements in predicting native-like binding poses and quantifying binding affinities compared to static docking and ensemble docking simulations into protein structures generated from long and computationally demanding MD simulation of apo receptor structure. We utilized Limoc to generate ensembles of holo-like protein structures in combination with the relaxed complex scheme (RCS) to perform ligand virtual screening. We developed different schemes to reduce the size of the ensemble of protein structures to increase efficiency and enrichment quality. Utilizing experimental knowledge about a small set of actives for a target protein allows the reduction of the ensemble size to a minimum of three protein structures increasing enrichment quality and efficiency simultaneously. An important factor contributing to the failure of docking to accurately predict binding affinities is the lack of inclusion of entropic information of the protein-ligand complex. We developed an efficient method to estimate configurational entropy of the bound ligand and interacting residues in the binding site. As part of the docking process, short Monte Carlo simulations of the ligand in a pre-generated ensemble of protein structures are combined with covariance matrix analysis to estimate the entropy of the bound ligand. A mixed scoring scheme dividing interacting residues into explicit and implicit residues allows for an efficient estimation of protein-ligand interactions. The entropy estimations of our method correlate quite well with the entropies derived from extensive MD simulations for each protein-ligand complex.

Degree

Ph.D.

Advisors

Lill, Purdue University.

Subject Area

Biochemistry|Pharmacy sciences|Biophysics

COinS