【作者】 李文钊；

【导师】 田圃；

【作者基本信息】 吉林大学 ， 生物物理学， 2013， 博士

【摘要】 蛋白质是生命活动的基本功能单元，蛋白质具有柔性，存在许多不同的结构状态，蛋白质的多种构象态构成了蛋白质的构象空间，蛋白质很多生理功能的发挥是通过结构转换实现的。分子动力学方法是一种常用的研究蛋白质结构的方法，其既能体现蛋白原子微观运动也能展示其整体结构动力学和热力学的变化，是对蛋白质结构动力学研究的很好手段。蛋白质通常在水溶液的条件下，通过与其他蛋白相互作用，结构发生相应变化，最终发挥某种功能。在本文中，我们通过研究了水对蛋白质结构的影响，蛋白-蛋白相互作用中蛋白质结构的变化，并通过结构分析和构象熵的计算来对蛋白质的结构和动力学进行了分析和研究，回答了几个基础并重要的蛋白质结构问题。本论文的主要研究内容如下：1)水对蛋白质的结构形成和维持都有着重要的作用，从热力学角度讲，水是通过改变蛋白质的势能面，最终影响蛋白质结构的。所以，我们针对球蛋白质的水体系进行了分子动力学模拟，计算了蛋白质与水作用的各个能量项(蛋白质自身能量、蛋白质与水作用能量、总能量)，通过这些能量项的相关性分析和标准差计算，展示出水对蛋白能量面粗糙度的改变，最终反映出水对蛋白结构的影响。研究结果表明，水对蛋白质的结构起到了“奴役”和“润滑”的双重作用：大部分情况下水分子通过对势能面的粗糙来“奴役”蛋白质的结构，而有时候水分子也会通过“平滑”能量面来润滑蛋白质的结构。2)蛋白质通过相互作用发挥其生物功能，对于蛋白相互作用的模型，有两个经典的假说“诱导契合”假说和“构象选择”假说。我们对多组相互作用的蛋白在微秒时间尺度上进行模拟取样，对蛋白质构象进行聚类分组，对比研究了蛋白结合前后的二面角转换，构象转换，构象空间变化等，通过对比分析，我们从构象空间的角度发现，蛋白质的结合过程同时符合“构象选择”和“诱导契合”，并且对不同的蛋白相互作用，起主导作用的不同。我们从蛋白质构象空间子态的角度描述了蛋白质结构的柔性，在蛋白质的柔性区域对蛋白质的相互作用起重要作用。3）构象熵的计算可以从热力学角度对蛋白质结构进行更有效的分析。我们从蛋白质熵计算的原理出发，对构象空间进行划分，通过公式推导发现，通过对主要构象空间的熵的计算就能估算出整个蛋白构象熵的结果，我们将这个最主要的构象空间定义为“熵主导构象”，我们对两个蛋白进行分子模拟分析，进行熵的计算，结果验证了我们的公式。最后我们继续研究了具体蛋白质熵计算的应用，结果证明，微秒时间尺度的分子模拟以及多次计算取平均的方法对于研究蛋白质相互作用的熵变能获得更可信的结果，我们还对蛋白质相互作用熵进行了计算，结果表明蛋白质结合后熵变小，结合前面的对蛋白相互作用的结构分析，说明水的熵增是蛋白质相互作用和结构的变化的重要驱动力。更多还原

【Abstract】 Protein is the basic functional unit for most physiological activities. The proteinmobility which also act as flexibility is very important for their functions. Proteinsaccomplish their physiological functions with remarkably organized dynamictransitions among a hierarchical network of conformational substates. Moleculardynamics simulation is a widely used method for protein conformational substatesinvestigation. MD method could show characteristics of protein kinetics andthermodynamics, especially how protein moving on atomic level. We all know,proteins perform their functions through interaction, which mostly in the waterenvironment, accompanied with conformation transition. Therefore, it is important tofigure out how water molecules affect proteins’ conformation, how conformationchanged during protein binding. Conformation space analysis and conformationalentropy estimation should make these more clear.The main contents of this dissertation are as below:1) Water play an indispensable role in shaping dynamic behaviour of proteinsthrough molecular interactions that modify protein potential energy surface. Wesystematically analysed protein self energies and protein-water interaction energiesobtained from extensive molecular dynamics simulation trajectories of barstar. Wefound that water molecules effectively roughen potential energy surface of proteins inthe majority part of observed conformational space and smooth in the remaining part.These findings support a scenario wherein water on average slave proteinconformational dynamics but facilitate a fraction of transitions among differentconformational substates, and reconcile the controversy on the facilitating and slavingroles of water molecules in protein conformational dynamics.2）The most majority proteins perform their specially biological activity throughinteraction with others. There are two important long-stand assumptions about protein-ligand binding which are “induced-fit” and “conformational selection”. Wedemonstrate these questions from the aspects of conformational substates. Weperformed molecular dynamics simulation in a microsecond timescale for three setsof proteins and study the conformational substates change before and after proteinbinding, as well as the transform frequency of dihedral angles and conformations.We found that the two assumptions work at the same time, for most proteins“induced-fit” play a dominate role, while for some others “conformational selection”are in dominate. We found that the loop is a key area for protein interaction, we alsogive a quantitative description for the flexibility of protein structure from the aspectsof protein conformational substates.3）The estimation of entropy is a challenging problem for macromolecules suchas protein. Despite great progresses that have been made, the global samplingremains to be a challenge for computational analysis of relevant processes. Here wepropose an entropy estimation method which based on physical partition ofconfigurational space and can be readily combined with currently availablemethodologies. Tests with two globular proteins suggest that accurateconfigurational entropy estimation can be achieved simply by considering theentropically most important subspace, thus convert an exhaustive sampling probleminto a local sampling problem. We also performed some calculation for protein, andwe found entropy loss during protein-protein binding, together with the results weget from protein-protein conformation analysis, it proved that the protein-proteininteraction is a process driven by entropy.更多还原