A Multistage Approach to Protein-Protein Docking

描述(申请人提供)：傅立叶相关技术是非常有效的蛋白质对接，使用表面互补性的测量作为目标函数。然而，除了近乎天然的构象外，该方法还会产生大量的假阳性(即得分较好但RMSD较大的构象)。在开发可以对对接的构象进行排序并选择接近天然构象的后处理方法方面已经取得了实质性的进展，但蛋白质对接的刚体性质仍然是一个限制。如果有很强的形状互补性，后处理就会有帮助，就像酶抑制剂复合体一样。然而，在抗体-抗原和许多其他复合物中，界面不那么紧密，其中极性相互作用和盐桥对结合更重要。对于这些复合体，傅里叶相关技术产生的命中较少，由于亲和力较低，近天然对接结构的识别变得困难，并且结果对组分蛋白质坐标的微小扰动非常敏感。这一提议的总体目标是将多级对接的力量扩展到主要由形状互补稳定的综合体之外。这将实现(1)通过基于对接构象的集群开发稳健的识别方法；(2)通过同时灵活地精炼所保留的集群，其将能够发现和提纯甚至低质量的命中；(3)通过调整最重要的表面侧链的构象，基于来自具有显式溶剂的纳秒分子动力学模拟的构象统计；以及(4)通过整合来自在前一步骤中产生的多个蛋白质结构的对接的结果，从而在一些关键侧链的构象上有所不同。初步结果表明，这些策略将显著改善相对较弱的复合体的对接结果，这些复合体经常在免疫识别、信号转导和细胞周期控制中发挥重要作用。

DESCRIPTION (provided by applicant): Fourier correlation techniques are very efficient for protein docking using measures of surface complementarity as the target function. However, in addition to near-native conformations, the method yields an enormous number of false positives (i.e., conformations with good score but large RMSD). Substantial progress has been made in developing post-processing methods that can rank the docked conformations and select the ones close to the native, but the rigid body nature of protein docking still remains a limitation. Post-processing helps if there is a strong shape complementarity as in enzyme-inhibitor complexes. However, the interface is less well-packed in antibody-antigen and many other complexes, in which polar interactions and salt bridges are more important for binding. For these complexes, Fourier correlation techniques produce fewer hits, discrimination of the near-native docked structures becomes difficult due to the lower affinity, and the results are very sensitive to small perturbations in the coordinates of the component proteins. The general goal of this proposal is to extend the power of multistage docking beyond the complexes primarily stabilized by shape complementarity. This will be achieved (1) by developing robust discrimination methods, based on the clustering of the docked conformations; (2) by simultaneous flexible refinement of the retained clusters that will be able to find and refine even low quality hits; (3) by adjusting the conformation of the most important surface side chains, based on conformational statistics from nanosecond molecular dynamics simulations with explicit solvent; and (4) by integrating results from the docking of multiple protein structures that have been generated in the previous step, and thus differ in the conformations of some key side chains. Preliminary results show that these strategies will substantially improve docking results for relatively weak complexes that frequently play important roles in immune recognition, signal transduction, and cell cycle control.