從序列中預測分析抗體類澱粉沉積性

Abstract

類澱粉蛋白（amyloid）是一種不可溶的纖維性蛋白質，各種類澱粉蛋白在身體器官或組織內異常沉積會造成嚴重的類澱粉沉積症。而全身性且原發性之類澱粉沉積症，如類澱粉輕鏈沉積症（Amyloid light-chain amyloidosis），就是異常的輕鏈蛋白質纖維在不同器官中聚集導致多種器官嚴重損壞。在製造生產抗體的過程中，傳統上利用融合瘤細胞製造的鼠源抗體（murine antibody）具有高度的免疫活性、缺乏穩定性且半衰期較短，經過擬人化處理後的抗體（humanized antibody）相較於鼠源抗體能有效解決這些問題。但擬人化處理失敗的抗體容易造成抗體類澱粉沉積性。 對於發展抗體類澱粉沉積性預測系統，過去的研究多專注於辨識聚集傾向區域以及針對少數特定抗體手動調整設計專一的預測模型。然而，由於抗體種類繁多，現有的胚源依賴的方法無法廣泛應用到多樣性的抗體。為了發展廣泛的抗體類澱粉沉積預測系統，本研究經由初步分析結果，從序列蒐集數種能夠有效解釋生物現象之特徵因子，除了現有胚源依賴的方法外，並設計胚源獨立的方法針對跨胚源以及新的胚源序列預測模型來發展完整的抗體類澱粉沉積性預測系統。 分析特徵因子辨識出重要影響抗體類澱粉沉積性的因素，將有助於了解抗體類澱粉沉積性。我們發現賴胺酸在抗體序列中扮演很重要的角色。一些被辨識出有影響抗體類澱粉沉積性的重要特性，像是兩親性，疏水性，螺旋結構，等電點，可變性和核蛋白質等。此外，泛素化位置的數量多寡在類澱粉抗體序列以及非類澱粉抗體序列之間有顯著差異。此自動化且跨胚源的預測方法已建置網路伺服器提供使用者方便預測分析。

Antibody amyloidogenesis is the aggregation of soluble proteins into amyloid fibrils. Abnormal accumulation of amyloid fibrils in the organs and tissues that change the normal function of tissues may lead to amyloidosis. Amyloid light chain amyloidosis is one of the most common forms of systemic amyloidosis. The antibody-producing cells will become dysfunc-tional and produce unusual protein fibrils. Initial antibodies were simple murine antibody using hybridoma technology, and these antibodies have shortcomings of high immunogenicity and a short half-life in vivo. Humanization is usually performed to reduce the occurrence of these problems and humanized antibodies have generally replaced murine antibodies. Unfortunately antibody amyloidogenesis is one of major causes of the failures of antibody in the humanization process. For developing the antibody amyloidogenesis prediction systems, previous studies focused on predicting aggregation prone regions for general proteins and establishing individual models for each known germline. However, the antibody amyloidogenesis as a global property of an antibody have a large number of possible germlines. The existing germline-dependent methods are not able to predict sequences of novel germlines. In order to develop comprehensive antibody amyloidogenesis prediction systems, we plan to mine and analyze informative features based on the preliminary results and develop an automatic prediction server for predicting antibody amyloidogenesis of novel germlines. The further identification of informative features for each encoding scheme is of special interests that can provide better understanding of antibody amyloidogenesis. According to the feature importance analysis for amino acid composition, the amino acid lysine is ranked as the most important amino acid. Some identified informative properties are amphiphilicity, hydrophobicity, helical structure, isoelectric point, mutability and nuclear protein, etc. Additionally, the numbers of ubiquitylation sites in amyloidogenic and non-amyloidogenic antibodies are found to be significantly different. It reveals that ubiquitylation might play important roles in determining the antibody amyloidogenesis. Our method for predicting antibody amyloidogenesis is implemented as a publicly available web server.