Systematic classification of phosphorylation sites for an integrative analysis of kinase signalling

细胞中的一系列生化事件被称为信号通路，在所有生物体的正常生理功能调节中起着重要作用。信号通路调节的过程包括细菌向食物源移动、酵母出芽、植物对病原体的反应以及哺乳动物的糖代谢。因此，监测信号通路的能力对于理解所有生物的生物化学是非常重要的。这些途径的活性是由一组被称为激酶的酶驱动的，这些酶将一种被称为磷酸盐的化学基团附着在其他蛋白质上。人类体内有超过500种不同的蛋白激酶它们彼此之间以及与其他蛋白质之间的关系非常复杂。蛋白激酶的活性可以通过分析附着在其他蛋白质上的磷酸盐来检测。基于质谱（MS）技术的现代方法现在可以检测到数千个这样的磷酸化事件。这种技术被称为磷酸蛋白质组学，这种方法提供的信息有可能揭示一套关于细胞中激酶是如何调节的以及它们在疾病中是如何改变的巨大的新知识。为了最大限度地从磷酸化蛋白质组学数据中获得信息，我们最近开发了一种名为激酶底物富集分析（KSEA）的计算方法，该方法将MS鉴定的磷酸化位点与上游激酶联系起来。然后，KSEA算法计算数据集中属于给定激酶的底物的富集。我们发现KSEA给出的值可以用来测量底物已知的所有激酶的活性。然而，在MS检测到的磷酸化位点中，只有大约10%被上游激酶所注释。因此，在磷蛋白质组学实验中获得的数据中，只有一小部分对理解细胞生物化学具有实际意义。为了解决这个问题，在这个应用程序中，我们的目标是组装一个带有它们所属的信号通路注释的磷酸化位点数据库。我们的假设是，当与KSEA一起使用时，这个磷酸化位点数据库将能够以前所未有的深度测量信号传导，从而显著推进我们对生物系统基本特性的理解。我们将首先关注在人类细胞中运作的信号通路，但同样的方法可以用于促进对其他生物信号传导的理解。信号通路的数据库将通过对蛋白质磷酸化事件进行分类来建立，这些磷酸化事件是由靶向激酶的药物增加还是减少的，以及由已知的激活蛋白激酶的药物调节的模式。这些实验现在是可能的，因为大量的激酶抑制剂最近被开发出来，用作治疗癌症和炎症等疾病的药物，而且因为定量磷酸化蛋白质组学技术的最新发展。我们期望用至少100种激酶抑制剂治疗细胞，针对至少50种不同的激酶。通过在不同细胞系中系统地进行这种分类，我们将确定不同激酶之间的关系，并将区分几种细胞类型的核心信号事件与细胞类型特异性的信号事件。磷酸化位点的系统分类还将识别信号标记，这些标记可用于测量这些事件如何在因疾病或因对治疗不敏感而改变其特征的细胞中被重塑。我们还假设，基于激酶抑制剂调节模式的磷酸化位点分类将有助于构建模型来预测可以改变给定表型的最佳激酶抑制剂。

A series of biochemical events in cells known as signalling pathways play important roles in the regulation of normal physiological functions in all organisms. Examples of processes regulated by signalling pathways include the movement of bacteria towards a food source, budding of yeast, the response of plants to pathogens, and sugar metabolism in mammals. Therefore, the ability to monitor signalling pathways is important for understanding the biochemistry of essentially all living beings. The activity of these pathways is driven by a group of enzymes known as kinases which attach a type of chemical group, known as phosphate, to other proteins. There are more than 500 different protein kinases in humans and their relationship with each other and with other proteins is very complex. The activity of protein kinases can be detected in cells by analysing phosphates attached to other proteins. Modern methods based on a technique named mass spectrometry (MS) can now detect several thousands of such phosphorylation events. This technique is known as phosphoproteomics and the information provided by this method has the potential to reveal an immense new set of knowledge on how kinases are regulated in cells and how these are altered in disease. To maximize the information that can be derived from phosphoproteomics data, we recently developed a computational approach named Kinase Substrate Enrichment Analysis (KSEA), which links the phosphorylation sites identified by MS to the kinases acting upstream. KSEA algorithms then calculate the enrichment of substrates belonging to given kinases in the dataset. We found that values given by KSEA can be used to measure the activities of all kinases for which substrates are known. However, only about 10% of phosphorylation sites detectable by MS are annotated with the kinases acting upstream. Therefore, only a small fraction of the data obtained in a phosphoproteomic experiment are actually informative for understanding cell biochemistry.To address this issue, in this application we aim to assemble a database of phosphorylation sites annotated with the signalling pathway they belong to. Our hypothesis is that, when used together with KSEA, this database of phosphorylation sites will have the ability to measure signalling with unprecedented depth, thus significantly advancing our understanding of the fundamental properties of biological systems. We will initially focus on signalling pathways operating in human cells but the same approaches could be used to advance the understanding of signalling in other organisms. The database of signalling pathways will be built by classifying phosphorylation events on proteins based on whether these are increased or decreased by drugs that target kinases and by their patterns of modulation by agents known to activate protein kinases. These experiments are now possible because a large array of kinase inhibitors have recently been developed, to be used as drugs to treat diseases such as cancer and inflammation, and because of the recent development of techniques for quantitative phosophoproteomics. We expect to treat cells with at least 100 kinase inhibitors, targeting a minimum of 50 different kinases.By performing this classification systematically and in different cells lines, we will identify relationships between different kinases and will discriminate signalling events that are core for several cell types from those that are cell type specific. Systematic classification of phosphorylation sites will also identify markers of signalling that can be used to measure how these events are remodelled in cells that have changed their characteristics due to disease or because they have become insensitive to therapy. We also hypothesise that a classification of phosphorylation sites based on their patterns of modulation by kinase inhibitors will be useful in constructing models to predict the best kinase inhibitors that can modify a given phenotype.