GPCR跨膜螺旋的结构拓扑建模及其预测方法

Modeling the Structural Topology and Predicting the Three-Dimensional Structure for Transmembrane Helixes of GPCR

7个α跨膜螺旋组成的螺旋束是G蛋白偶联受体的最主要拓扑特征,其三维结构的预测精度直接影响完整受体的三维结构预测、配体对接及功能分析的准确性.近期许多研究小组提出了各种方法,同时也遇到了一个共同的问题:采样时难以在7个跨膜螺旋结构的保守性与局部多样性之间获得平衡,其实质是未将两者统一到一个系统模型中.文中针对跨膜螺旋的空间结构特点,建立了兼顾保守性与多样性的结构拓扑模型,并利用该模型形成了4阶段的结构优化方法,试图获得采样广度与深度的平衡.同时,引入基于结构拓扑的能量项与约束,起到了优化评判标准和剪裁采样空间的作用,有效地预测了跨膜螺旋的三维结构.使用文中方法展开了3组验证实验,用8个已解构的目标分别与GPCRDOCK2010的参赛结果、知名结构预测工具Swiss和MODELLER进行了比较.与Swiss的比较中,文中方法有5个目标获得了更优的三维螺旋结构;与单模板、多模板的MODELLER的比较中,文中方法分别在6个目标与7个目标上取得了优势.

G protein-coupled receptors are a class of proteins characterized by α-helix bundle composed of seven transmembrane helixes. The prediction accuracy of bundle＇s three-dimensional structure directly affects the accuracy of the entire receptor, the accuracy of ligand complex, and structure-based functional analysis. After establishing a reasonable structural topology mapping the structural characters of transmembrane helixes, this paper develops a four-stage optimization method, introduces a new energy item for evaluating the bundle structure and constraints for tailoring sampling space to predict the three-dimensional structure of transmembrane helixes. Three in silio experiments are performed to evaluate the ability of the method. After comparing with structures submitted by GPCRDOCK2010 participants, the well-known structure prediction tools Swiss and MODELLER against eight resolved GPCR targets, respectively, the results show the method proposed in this paper obtains better accuracy on five targets than Swiss and gets lower TM RMSD on six targets and seven targets than single-and multi-template MODELLER respectively.