【作者】 胡建平；

【导师】 王存新；

【作者基本信息】 北京工业大学 ， 流体力学， 2003， 硕士

【摘要】 分子识别的研究具有重要的生物学和药理学意义。例如DNA的转录，激素的作用机制，免疫系统中的抗原—抗体识别，化学中的酶催化反应，以及很多药物发挥药效的机制都与之密切相关。而绝对结合自由能计算对于研究分子识别机制及其动态作用是十分重要的。自由能的正负决定了化学反应的方向，其大小决定了反应的趋势强弱。国际上有很多的实验小组投入到了自由能的计算模型的研究。Jorgensen提出的双湮灭方法被认为是一种很经典的计算绝对结合自由能的方法，并且已经得到了广泛的应用。在论文的第一部分，我们用双湮灭方法计算了谷氨酰胺结合蛋白(GlnBP)单体和底物Gln结合的绝对结合自由能。模拟得到的结果是-14.59 kcal／mol，与实验值-8.025 kcal／mol接近，并分析了双湮灭方法的优点和不足，并对该方法的改进提出了一些建议。 在过去的几十年里，分子动力学(MD)模拟已经成为研究蛋白质物理化学特性的一个重要工具。MD模拟通过给出生物大分子在原子水平上的相互作用，提供生物大分子的涨落和构象变化的详细信息。MD模拟对于研究蛋白质等生物大分子的结构与功能的关系具有重要的意义。论文的第二部分工作是选择了GlnBP单体和谷氨酰胺结合蛋白(GlnBP-Gln)复合物体系为研究对象。使用GROMOS96程序及其力场分别进行了600 ps的MD模拟。通过两个模拟所得到的轨迹，我们从体系的分子结构、自由能、铰链区柔性、口袋区相互作用及底物结合的特异性等几个方面分析了该体系的构象开合机制，得到了与实验数据是相吻合的结果。 溶剂化效应的研究是国际上异常活跃的研究热点。而静电相互作用是溶剂 北京工业大学工学硕士学位论文一 化效应中最主要的贡献，尤其当溶质分子带电荷或高度极化时更是如此。一般 来讲，广义的PB方程可以描述蛋白质溶液体系的静电效应。本课题组以前已 把用有限差分方法求解PB方程的方法与随机动力学相结合，得到FDSD模拟 程序，FDSD模拟方法在动力学模拟过程中把大量的水分子代之以连续介质。该 程序在较小的蛋白体系中是较为成功的。在论文的最后一部分，我们选择了R6 态胰岛素六聚体这个较大的蛋白为研究对象，把FDSD模拟结果与MD模拟和 通常的真空SD模拟结果作了比较。结果表明：FDSD模拟方法对于研究分子 的结构和动力学性质均比真空SD模拟有很明显的改善，与MD模拟结果接近。更多还原

【Abstract】 Molecular recognition is of central importance in biology and pharmacology. It underlies the action of hormones, the control of DNA transcription, the recognition of Antigens-Antibody in the immune system, the catalysis of chemical reactions by enzymes and the actions of many drugs. Calculations of absolute binding free energy play a very important role in the study of the mechanism and dynamics of the molecular recognition. The orientation of chemical reactions depends on the sign of binding free energy and the tendency depending on its corresponding value. Many groups have invested ingenuity and effort in the development of such models, double annihilation method developed by Jorgensen was proposed as a classical way of computing the absolute binding free energy of a complex in solution, which has been applied in a number of studies. In the first part of this thesis, The binding free energy between GlnBP monomer and its ligand Gin was computed by using the classical double annihilation method. It is found that the result (-14.59 kcal/mol) agrees well with experimental data (-8.025 kcal/mol). The advantage and disadvantage of this methods are analyzed.During the past several decades, molecular dynamics (MD) simulation of proteins has become a widely used tool to deepen our understanding of the molecules. Computer simulation techniques can be used to understand the properties of a molecular system in terms of interactions at the atomic level. The results of MD simulations are very useful to understand the relation between the macromolecularconformational changes and its biological function. In the second part of this thesis, we have chosen the protein monomer GlnBP and complex GlnBP-Gln as our study systems. The MD simulations were carried out with the GROMOS96 package and its force field for 600 ps. The MD simulation data were analyzed in terms of the overall structure, the binding free energy, the flexibility hi the region around hinges, the interactions in the binding site, pocket ligand binding specificity, and so on. The results obtained from the biochemistry experiment can be correctly identified by our MD simulations.The study on solvation effect is the focus of the international scientific research fields. The electrostatic interaction is the most important part hi the solvation effect, especially when the solute is charged or highly polarized. Commonly, the general Poisson-Boltzmannn (PB) equation can well describe the electrostatic property of a protein in solution. We use a finite difference method to solve PB equation and incorporate the solution into SD simulation, which is called as FDSD simulation procedure and developed by our group before. The solvent is considered as a continuum in FDSD simulation procedure, which gained success in simulation of smaller proteins. In the last part of this thesis, the R-state human insulin hexamer has been selected as a target. A comparison has been made among the results obtained with FDSD simulation, MD simulation and SD simulation. The results show that the FDSD simulation has obvious improvement over the SD simulation for structure and dynamics properties, and has a good agreement with that of the MD simulation.更多还原