Improving the performance of the PLB index for ligand-binding site prediction using dihedral angles and the solvent-accessible surface area.

Chen Cao,Shutan Xu

doi:10.1038/srep33232

Abstract

Protein ligand-binding site prediction is highly important for protein function determination and structure-based drug design. Over the past twenty years, dozens of computational methods have been developed to address this problem. Soga et al. identified ligand cavities based on the preferences of amino acids for the ligand-binding site (RA) and proposed the propensity for ligand binding (PLB) index to rank the cavities on the protein surface. However, we found that residues exhibit different RAs in response to changes in solvent exposure. Furthermore, previous studies have suggested that some dihedral angles of amino acids in specific regions of the Ramachandran plot are preferred at the functional sites of proteins. Based on these discoveries, the amino acid solvent-accessible surface area and dihedral angles were combined with the RA and PLB to obtain two new indexes, multi-factor RA (MF-RA) and multi-factor PLB (MF-PLB). MF-PLB, PLB and other methods were tested using two benchmark databases and two particular ligand-binding sites. The results show that MF-PLB can improve the success rate of PLB for both ligand-bound and ligand-unbound structures, particularly for top choice prediction.

Highlights

Proteins perform their biological functions by interacting with other molecules, such as DNA, antigens, drugs or even other proteins
We evaluated the ligand-binding site prediction performances of the propensity for ligand binding (PLB), multi-factor PLB (MF-PLB), Ligsite-csc and other methods
These researchers created the PLB index based on the amino acid preferences for the ligand-binding site

Summary

Introduction

Proteins perform their biological functions by interacting with other molecules (ligands), such as DNA, antigens, drugs or even other proteins. Protein-DNA binding sites show the most obvious amino acid preferences because positively charged residues, such as arginine and lysine, likely face the negatively charged phosphate backbone of DNA1. In this manuscript, we focus on the detection of small molecule ligand-binding sites because it is a prerequisite for protein-ligand docking and the first step of structure-based drug discovery[2,3]. To identify a potential ligand-binding site, Soga et al developed an index known as the propensity for ligand binding (PLB), which is calculated by summing up the RA (preference factor for an amino acid) of all residues involved in the site[22]. Our previous study revealed that the dihedral angles of the residues in specific regions of the Ramachandran plot reveal preferences for ligand-binding sites[24]

Methods

Results

Discussion

Conclusion