柔性多肽片段–蛋白质相互作用的全局对接方法

Global Docking Method for Flexible Peptide Segment-Protein Interactions

摘要: 多肽–蛋白质的相互作用在生物细胞中发挥着各种各样重要的作用。通常情况下，它们之间的结合信息是未知的。所以，利用计算方法预测结合位点具有重要意义。而以Rosetta为代表的常用对接软件通常具有很强的初始位置依赖性。为克服这一局限性，本研究提出了一种全局对接的方法，以受体蛋白为球状系统的中心，将多肽平均地分布在球面26个位置上；同时定义了一个区分天然结合构象和非天然结合构象的筛选参数。用上述方法预测多肽–蛋白质的结合构象，结果显示该方法能成功预测蛋白质的结合位点，且多数多肽的预测构象的Cα-RMSD在5.5 Å以下。因此，研究结果表明，所发展的方法在蛋白质多肽结合位点预测方面有很好的应用价值。

Abstract: Protein-peptide binding plays various important roles in living cells. In many cases, the peptide- binding sites of proteins are not known in prior. Then, computational prediction of the peptide- binding sites is desirable. Popular programs for protein-peptide docking usually depend strongly on the initial positions of peptides, such as Rosetta. To overcome this limitation, here we develop a global docking approach in which the peptide is initially distributed evenly on 26 surface locations of a virtual sphere around the protein, and define a selection parameter for discriminating native-like binding site from non-native sites. We used this approach to predict the native-like binding conformations of peptide-protein complexes, and in most cases the peptide-binding sites were correctly predicted, with Cα-RMSDs below 5.5 Å with respect to the crystal structures of peptides. The results of this study suggested that our approach may be very useful for the identification of peptide-binding sites of proteins.