蛋白質結構-動力學-功能之關係(I)

Abstract

由於結構基因體學的進展，使得蛋白質數據庫中的蛋白質結構快速增加。蛋白質 的結構對於瞭解此蛋白質的生物功能是非常重要的，但許多重要的生物學過程與蛋白 質動態有關，無法僅由蛋白質結構來推推論。例如：在C 型肝炎病毒 NS3 解旋酵素（NS3 helicase）的ATPase 活性與解旋酵素活性在動態上息息相關。上皮細胞生長因子接受器 （EGFR）是最早被發現與人類癌細有關的的接受器，當其被活化時結構產生巨大的改 變。換句話說，蛋白質動態與蛋白質結構對於蛋白質功能佔有同等重要的地位。然而 目前的方法分子動態學方法（Molecular Dynamics），計算極為耗時，對於大蛋白質的 計算更是難上加難。 先前本實驗室所發展的蛋白質固定點模型可有效計算原子波動，計算值與實驗值 極為一致。蛋白質固定點模型根據原子與蛋白質中最小波動點的空間位置計算出原子 波動與各原子運動的相關性。與目前分子動力學方法相比，固定點模型計算速度遠比 分子動力學方法要快到幾個級數的差別；此外在動力學特性的預測上，固定點模型的 預測結果也較分子動力學方法來得正確。而且，蛋白質固定點模型可應用於極大蛋白 質上，但此方法的限制是無法使用在蛋白質複合體與多域蛋白質上。在這個計畫中， 我們將會改進蛋白質固定點模型，初步結果令人滿意。 在實際的應用上，我們將研究EGFR 的蛋白質動態，EGFR 與癌症的發生有極強的 關連性。近來，EGFR 基因突變從non-small-cell 的肺癌病人中找到。然而儘管EGFR 與ATP 複合體及其抑制物的三級結構已被解開，但結構仍無法澄清其作用之機制。我 們初步的研究顯示，這些突變的胺基酸有動態的關連性。EGFR 提供一個絕佳的例子 讓我們進行蛋白質動態與功能的研究。我們相信，我們的研究將有助於瞭解藥物對抑 制EGFR 的作用機制。 最終，我們將建立一個完整的蛋白質動態資料庫。雖然蛋白質動力學能提供很多 蛋白質結構與功能之間的相關資訊，但生物學家目前並無如此的資訊。我們將計算PDB 中所有蛋白質結構的動態資料，為生物學家提供一個完整的蛋白質動力學資料庫。我 們相信，我們的結果將會刺激蛋白質結構、動力學及蛋白質功能的相關研究。

Protein structures are deposited in Protein Data Bank in an amazing speed due to the progress of structure genomics initiatives. Though knowing the shapes of proteins is crucial for understanding their biological functions, many important biological processes cannot be inferred from sheer structures due to large-scale protein dynamics that frequently occurrs in biological reactions. For example, the ATPase activity and helicase activity of hepatitis C virus NS3 helicase are dynamically coupled to each other, while the activation of EGFR, the fist cell-surface receptor to be linked directly to human cancer, has been shown to be closely related to large scale domain rearrangements. In other word, protein dynamics is in fact as important as protein structure in understanding protein function. At present, current methods such as Molecular Dynamics (MD) are very computationally expensive and, consequently, are impractical for large biological systems. Here we will develop novel approaches to compute protein dynamical properties, based on the protein fixed-point (PFP) model previously proposed by us. This model compute both atomic fluctuations and motion correlation using the positional vectors of atoms issuing from the fixed point, which is the point of the least fluctuation in protein. The PFP model is faster than current methods by about orders of magnitude and its prediction of the average dynamical properties is much more accurate. Another major advantage of the protein fixed-point model is that it can be applied to very large proteins. However, the PFP model cannot be applied to protein complexes or multidomain proteins. Here, we will propose novel ideas to improve the PFP model. Our preliminary results are very encouraging. As a practical application of our approach, we will study the protein dynamics of epidermal growth factor receptor (EGFR), the first cell-surface receptor to be linked directly to human cancer. Recently. Mutations of the EGFR gene have been identified in specimens from patients with non-small-cell lung cancer. But the mechanism of the drug resistance is still unclear despite the availability of the 3D X-ray structure complex with ATP or the EGFR inhibitors. Our preliminary study shows that these mutated residues are dynamically coupled. This presents an excellent chance for us to study the protein dynamics in EGFR. We believe that our results will help shed light on the mechanism of the drug resistance of EGFR. Finally, we will build establish a comprehensive Protein Dynamics Database. Though protein dynamics provides rich information about the relationship between protein structure and function, the general biologists cannot access such information easily. We will compute the dynamical properties for all structures in Protein Data Bank and provide a comprehensive database of protein dynamics for general biologists. We believe that our results will stimulate interest in investigating protein structure-dynamics-function relationship.