Prediction of Binding Mode between Chemokine Receptor CCR2 and Its Known Antagonists using Ligand Supported Homology Modeling

Abstract

Chemokines play a crucial role in the trafficking of leukocytes in the body through the binding to their related receptors. Chemokine Receptor 2 (CCR2) belongs to GProtein Coupled Receptor (GPCR) family and expressed in monocytes, immature dendritic cells, activated T lymphocytes, basophils, and endothelial and vascular smooth-muscle cells. CCR2 binds several chemokines; CCL2, CCL7, CCL8, and CCL13. CCR2 and its ligand have been implicated in the pathophysiology of a number of diseases, including rheumatoid arthritis, multiple sclerosis, atherosclerosis, organ transplant rejection, and insulin resistance. Knowledge of the structural basis on CCR2-ligand interaction could help facilitate the design of novel CCR2 antagonists. We modeled and predicted the binding sites of widely known CCR2 antagonists, TAK-779 and Teijin-lead (Fig. 1), using ligand supported homology modeling method. Homology modeling predicts the three-dimensional structure of a given protein sequence based on its alignment to reference proteins of known three-dimensional structure (socalled templates). It is known as the most successful technique for predicting the three-dimensional protein structure. In conventional homology modeling method, the ligand molecules are not considered in the process of protein model building. After building the protein model, the ligand molecules are docked into the protein model. These approaches often showed the consequence that residue side chains involved in ligand binding are inappropriately modeled. The time-consuming re-modeling of the binding site or docking is required to generate sensible side-chain and ligand orientations. In the case of GPCR, the conventional homology modeling approaches for obtaining the binding site models are more challenging because the number of the available templates is very limited and the sequence identity to the templates is very low. The available templates for GPCR are the solved three-dimensional structures of bovine rhodopsin, β2-adrenergic receptor (β2AR), turkey β1adrenergic receptor, and human adenosine A2A receptor. The homology modeling from those limited number of templates may result in uncertainty not only regarding backbone positioning but also about side chain conformations. One approach to solve this is to optimize the side chains through mechanics and/or dynamics in an empty pocket; however, this method may be inaccurate due to possible disruption of the binding site. Another approach is to use dynamics and ligand data as a restraint. This method also has a possibility of poor quality. Thus, the development of modified homology modeling methods for constructing a plausible ligand binding mode is an important theme in GPCR-ligand modeling research. The incorporation of ligand information from the early stage of homology modeling has been studied as an alternative strategy to the conventional homology modeling. This approach was first developed and termed as ligand supported homology modeling by Klebe and coworkers. Recently, the improved and diversified methods have been published as ligand-steered homology modeling or homology-modeling protein-ligand interactions. The results presented reasonable binding site models. We recently reported the protocol and evaluation for the proprietary CCR2 antagonist by ligand-supported homology modeling. We here employed the protocol to predict the binding mode for known CCR2 antagonists, TAK-779 and Teijin-lead. The method in this study for structure modeling of CCR2 and its antagonists is summarized in Figure 2. In the first place of homology modeling, we retained the bound ligand, carazolol, from the crystal structure of β2AR. Then, carazolol was replaced by the various conformations of the target