蛋白激酶與受質磷酸化網路之建構

Abstract

透過蛋白激酶 (protein kinase) 所催化的蛋白質磷酸化 (protein phosphorylation) 機制是一種常見可逆的蛋白質轉譯後修飾作用，扮演著訊息傳遞路徑上的重要角色。Manning 等作者在2002 年發現了518 個人類蛋白激酶基因，也提供了一個蛋白質磷酸化網路研究的切入點。隨著高通量的質譜儀蛋白質體學技術，實驗驗證的蛋白質磷酸化資料也快速的增加，但是，只有20%的磷酸化位置有註解是被哪個蛋白激酶催化的。為了完整地探討蛋白激脢如何調控細胞內的機制，需要詳盡且精確的方法來辨識受質 (substrate) 上面的磷酸化位置是被哪個特定蛋白激酶所催化。因此我們發展了一個叫作RegPhos 的方法，整合了電腦模型與蛋白質相關性(包含蛋白質交互作用、功能相關性以及細胞內位置)來辨識某個磷酸化位置被哪個蛋白激酶催化。為了評估RegPhos方法的效能，四個已知的蛋白激酶 (CDK、PKC、PIKK 和INSR) 的磷酸化資料被用來測試是否能正確的預測作用的蛋白激酶，RegPhos 跟單純用電腦模型的方法比起來，可以 改善5 到10%的準確度。這些完整且準確被分析預測蛋白激酶與受質的交互作用可以被用來建構細胞內從細胞膜上的受體蛋白激酶 (receptor kinase) 到細胞核內的轉錄因子 (transcription factor) 的磷酸化網路，並且用實驗表現證據(如:基因微陣列資料) 來檢視蛋白激酶跟受質是否有統計上顯著的相似表現行為。

Protein phosphorylation, catalyzed by protein kinases, is a ubiquitous reversible post-translational modification (PTM) and plays a crucial role in signaling pathway. Manning et al. have identified 518 human kinase genes, the so-called “kinome", that provides a starting point for comprehensive analysis of protein phosphorylation networks. With the high-throughput mass spectrometry (MS) proteomics, the number of in vivo phosphorylation sites is increasing rapidly. However, only 20% of the experimentally verified phosphorylation sites have the annotation of catalytic kinases. To understand how protein kinases regulate their substrates in intracellular processes, it is necessary to link these sites to specific kinases. Therefore, we propose an approach that incorporates machine learning method with protein associations (protein-protein interactions, functional associations, and subcellular localization) for identifying the catalytic kinase for each experimental phosphorylated site. Four well-annotated kinase families, such as CDK, PKC, PIKK, and INSR, are used to test the ability to correctly predict which kinases are responsible for catalyzing them. The presented approach can improve 5 - 10% predictive accuracy more than purely using machine learning method. The identified kinase-substrate interactions are used to construct the intracellular phosphorylation network starting from receptor kinases to transcription factors. Moreover, the experimental expression evidence such as time-series microarray gene expression profiles is adopted to validate the syn-expression of kinase and substrate with statistical significance.