蛇毒 C 型凝集素衍生之拮抗劑與血小板醣蛋白陸交互作用的 結構探討

Abstract

心肌梗塞和中風是已開發國家當中最常見的兩種死因，這兩種疾病的起因皆與血管受傷時血小板上醣蛋白(GPVI)的反應有關，凝血機制透過GPVI 與其受體膠原蛋白、蛇毒等產生凝血反應，以聚集許多GPVI 來啟動凝血機制。因此我們可以透過阻止GPVI 的凝聚反應，來開發治療人類心血管疾病的抗血栓劑。 本篇論文中，利用分子接合與分子動態模擬來探勘許多蛇毒片段對於GPVI 接合的性質，我們推論出具活性蛋白質片段的長度以及接合位置的條件。蛋白質片段應介於6 到8 個胺基酸之間，如果蛋白質片段的長度過長會導致接合位置位於GPVI的外側，降低蛋白質片段的活性。為了提高蛋白質片段的活性，我們利用了丙胺酸掃描的數值模型來分析先前生物實驗結果當中活性最高的蛋白質片段，並且藉由丙胺酸掃描的結果和分子動態模擬來尋找出具有更高活性的胺基酸組合，希望能藉此提高蛋白質片段對GPVI 的抑制活性。 在全新蛋白質設計問題上，我們提出一個新的長片段胜肽設計演算法應用於蛋白質藥物設計。此類問題，傳統上利用基因演算法針對受體的表面來設計新的胜肽，基因演算法並不適用於長片段的胜肽設計，因為其演算的速度過慢，並且無法篩選全部的組合，容易收斂於局部的受體表面。我們提出的新方法結合了胺基酸位置適合度矩陣來加快篩選並用基因演算法進行最佳化，我們根據前人研究的E3 泛素連接酶(E3 ubiqutin ligase)來進行驗證，重現研究中提出的胜肽，以及胜肽上各個點突變之間的關係。此驗證證明了方法的可用性，將可應用於設計新的抗血栓胜肽。

Myocardial infarction and stroke, the two most common causes of death in the developed world, were related to the platelet response mediated through human glycoprotein VI to vessel injuries. Natural ligands of Human GPVI, for example, collagen and several snake venoms, activated the signal transduction by first linking several GPVI. Therefore, designing inhibitors which prevent GPVI aggregation can be used as anti-thrombotic agents to treat human cardiovascular disease. In this study, molecular docking and molecular dynamics simulation were performed to explore the binding characteristics of several snake venom fragments to GPVI. We found an optimal the peptide length of six to eight amino acids and the relative position for good affinity. If the peptide length is larger than six, the fragment binding position would shift to the outer region of GPVI and result in a lower inhibition activity. To further explore the activity of protein fragment, the computational alanine scanning was performed to the most potent hexapeptide obtained from previous biological experiments. According to these results, we searched new sequence combinations that may have higher inhibited activities to GPVI. In protein de novo design part, a new methodology was proposed and can be applied on long peptide drug design. Traditionally, novel peptides without template VII peptide can be generated by genetic algorithm according to the interaction surface of the target. However, the genetic algorithm was not applicable for long length peptide design due to the slow calculation resulted in incomplete exploration of peptide assembly. In addition, genetic algorithm is easy to converge in local minimum at the interaction surface. Therefore, we proposed a new method combined with amino acid positional fitness to pre-filter all the possible combinations before applying genetic algorithm for optimization. The method was validated by successful repeating results from a set of an E3 ubiquitin ligase complex with a series of potent peptide obtained from previous studies. The method not only re-generated the peptide and their activity trend but also confirmed the point mutation relationship in the peptide. The validation proves the appliance of this new method in peptide drug design and can be implemented to the design strategy of novel anti-thrombotic peptides.