【作者】 马晓慧；

【导师】 王存新； 陈慰祖；

【作者基本信息】 北京工业大学 ， 流体力学， 2002， 硕士

【摘要】 在分子水平上定量地阐明蛋白质分子之间的亲合力和相互识别的机制，是正确地认识蛋白质聚集体结构和功能之间的关系并合理地设计出有用的蛋白质分子聚集体的关键。蛋白质分子间相互作用，其中最主要的是静电和疏水相互作用，使蛋白质分子能有序地结合在一起。 蛋白质的稳定性随溶液环境的变化一直是人们关心的热点问题。近年来，用计算机模拟方法对包括水化作用在内的生物分子静电效应的研究取得了一些进展，但是仍缺少相对快速且有效的预测方法。针对这个问题，我们采用简单形式电荷模型来研究蛋白质二聚体静电相互作用，建立了一种快速、有效地预测蛋白质稳定性随溶液pH值和盐离子浓度变化的方法，并将其应用于蛋白质分子结构稳定性的分析和预测。以22n胰岛素和其速效突变体为研究对象，模拟结果与实验观测相符。 在蛋白质相互作用和识别研究中，结合自由能的计算一直是一个研究热点。虽然近年来一些半经验自由能的计算方法相继出现，但是大部分模型在计算结合自由能时仍旧存在着一些不足。其中一个问题就是：对于构象熵的计算要么是简单的设置为常数，要么在计算上十分复杂，需要对每一个柔性键的转动几率进行描述。我们的目的是通过分析蛋白质复合物的界面信息来拟合计算构象熵和结合自由能的简单有效的半经验模型。本论文在此分析了20组蛋白质二聚体分子的结合界面三个有效的参数信息，即界面残基侧链可接触数（N_b）、亲水对数(N_pair)和非极性溶剂可接近表面积的变化(△ASA_apol)。结果显示N_b与结合侧链构象熵相匹配（R=0.97），而结合自由能可以通过付A、N_pair和△ASA_apol三者进行描述，拟合相关常数为0.95。此外，我们还将界面信息描 蛋白质二聚体相互作用和识别的计算机模 一 述的能量函数作为打分函数应用于蛋白质分子对接（Docking）和设计程序中。 分于对接结果表明，此打分函数能有效并简单的应用于任意蛋白质分子对接。更多还原

【Abstract】 Quantitatively describing the principle of protein-protein interactions and recognition in molecular level is the key point for understanding the relationship between structure and function of protein complexes and designing protein complexes, hi these interactions among proteins, electrostatic and hydrophobic interactions are the two most important ones, which let protein molecules bond systematically.The problem of the protein stability changed by solvent environment is in the limelight. Recently, there have been some significant advances in this direction, a fast and simple method is still not available for accurate prediction of experimental observation. To solve this problem, we used the formal charge model to study electrostatic interactions of protein complexes. And a fast and effective model for predicting the salt and pH dependent properties of protein complexes was presented here and applied to the analysis and prediction of the stability of protein structures. All simulations were performed on the native 2Zn insulin and its fast-acting mutants. It is found that the results agree well with experimental data.Binding free energy calculation is another focus of the study on protein-protein interactions and recognition. Although many methods have been developed to understand this problem for many years, most of them is still of some lacks. One problem is that the treatment of the conformational entropy is either simple to set to a constant, or still complicated to depend on the calculation of the probability of each rotamer. Our goal is to establish a simple and effective empirical approach to calculate the conformational entropy and the binding free energy by analysis ofJ:protein interface. Three variables of the binding interface information of 20 protein complexes, including the binding accessible number (Nb), hydrophilic pair (N^,) and apolar solvent-accessible surface areas (&ASAapol) are analyzed. It is found that Nbcorrelates very well (R = 0.97) with the loss of binding side-chain conformation entropy, and the binding free energy depends on Nb, Npair, and ASAapol, in which thefitting coefficient is 0.95. In addition, this model is applied to the docking program as an effective and fast target function. The results show that the score function of complexes is effective and simple for rational protein docking and its design.更多还原