論隱藏在蛋白質結構的動態資訊

Abstract

蛋白質的三級結構與其功能有緊密的相關。蛋白質結構不僅僅只是一個固定的結構而是一個動態的結構。研究顯示蛋白質的功能受到影響來自於遠離功能處一個胺基酸突變，其原因來自其突變改變了蛋白質的動態穩定。因此研究蛋白質的動態結構可以幫助我們了解蛋白質的功能與機制。分子動態模擬是基於原子與原子間各項的化學作用力，一般可用來分析蛋白質的動態及其功能的關係。但是，蛋白質結構動態模擬非常的消耗電腦的計算能量以及需要長時間的模擬。因此我們發展出了數個利用蛋白質三級結構推演出其內部原子的動態資訊而不需要任何化學作用力的機制。我們的方法非常快速可以得到蛋白質的動態資訊且的到的動態資訊而且算出的動態資訊跟實驗與模擬的結果一致。這篇論文第一想闡明蛋白質的結構隱含了其蛋白質動態資訊，更進一步的指出不需要胺基酸的資訊。第二、想藉由研究蛋白質酵素內的執行催化的胺基酸來研究結構資訊與動態資訊是否可以找出的執行催化的胺基酸的特徵。第三、藉這些執行催化的胺基酸在蛋白質序列演化上是高度保留的，我們發現結構資訊，動態資訊及演化資訊有高度相關性。並且，僅由蛋白質結構資訊就可以精確的推出其動態資訊與演化資訊，其精確程度可以達到單一胺基酸的程度，相較於之前的研究只能探討平均的趨勢有極大的優勢。最後我們展示出蛋白質與蛋白質之間的交互作用的交界面可以利用結構資訊與演化資訊相互搭配來找出。

Protein structure is closely related to its function. However, protein structure is not static, due to its constant thermal fluctuations. A mutation in a residue far away from the active site will greatly affect a protein’s function. Therefore, knowledge of protein dynamics will help shed light on protein function. Molecular Dynamics simulation, which is based on molecular mechanics, is a powerful tool to compute protein’s trajectories, from which useful dynamical properties can be derived for the analysis of the relationship between protein structure and its function. However, molecular simulation is computationally expensive, and becomes impractical to simulate large proteins. Here, we have developed geometry-based approaches to derive atomic thermal fluctuations from a single protein structure without using mechanical model. Our approaches are fast and comparable with molecular simulation to reproduce thermal fluctuation from a single structure. This thesis will start with our recent work to extract atomic fluctuations and motional correlation directly from protein structures through several commonly models. The results from sophisticated to simple models all suggest that dynamics are determined by structures. Then, catalytic residues are usually evolutionary conserved; thus evolutionary information from homologous sequences is essential in prediction of catalytic residues. In our studies, dynamics information from structures can predict better catalytic residues than evolutionary information. Therefore, we illustrate that the evolutionary information are hidden in a single structure. Finally, we will show that the coevolution between proteins in complex might be identified by the relationship between structure and conservation.