GPCRTooKit: A Computational Platform for Structural Comparison of GPCR Crystal Structures and Homology Model Refinement

Abstract

G-protein coupled receptors (GPCRs) constitute an important family of transmembrane receptors that regulate major life processes including sensory perception, cell proliferation and hormonal regulation. In lieu of the recent surge in GPCR crystal structures, structure based drug design methods are becoming more viable for GPCRs. However there is a huge disparity in the number of GPCR structures available and the number of GPCRs being studied in research. Thus homology models play an important role in a wide variety of GPCR research ranging from drug design to studying the functional mechanisms in these receptors. GPCR homology models do not capture the critical structural differences between the template and the GPCR being investigated.We have developed GPCRToolKit that lays the computational framework for building GPCR based modeling tools. We have developed a robust and accurate homology model refinement method called LITiCon2.0, based on optimizing the helical translation, rotational orientation, tilt and gyration angles of the seven helices. The structural differences between two GPCRs stem mostly from the rigid body degrees of freedom. The algorithm is highly parallel to enhance the computational speed. We have tested and validated this method by refining the homology models of several class A GPCRs with known crystal structures, one of them being the chemokine CXCR4 using β2-adrenergic receptor as template. We have also analyzed the statistical distribution of translations, tilts and rotations of TM helices in all available GPCR crystal structures. We found that TM3 and TM7 showed relatively less variation in rigid body orientation compared to the other helices. Overall, TM4 and TM5 showed the highest diversity in spatial orientation among the TM domains. These results along with the description and validation of the LITiCon2.0 method will be presented.