Systematic identification of phosphorylation-mediated protein interaction switches

Proteomics techniques can identify thousands of phosphorylation sites in a single experiment, the majority of which are new and lack precise information about function or molecular mechanism. Here we present a fast method to predict potential phosphorylation switches by mapping phosphorylation sites to protein-protein interactions of known structure and analysing the properties of the protein interface. We predict 1024 sites that could potentially enable or disable particular interactions. We tested a selection of these switches and showed that phosphomimetic mutations indeed affect interactions. We estimate that there are likely thousands of phosphorylation mediated switches yet to be uncovered, even among existing phosphorylation datasets. The results suggest that phosphorylation sites on globular, as distinct from disordered, parts of the proteome frequently function as switches, which might be one of the ancient roles for kinase phosphorylation.

蛋白质组学（Proteomics）技术可在单次实验中识别数千个磷酸化位点，其中绝大多数为全新位点，且缺乏关于其功能或分子机制的精准信息。本研究提出一种快速方法，通过将磷酸化位点映射至已知结构的蛋白质-蛋白质相互作用（protein-protein interactions, PPI）界面，并分析该蛋白界面的特性，以预测潜在的磷酸化开关（phosphorylation switches）。本研究共预测出1024个可潜在调控特定蛋白质相互作用的位点。我们对筛选出的部分磷酸化开关开展了实验验证，结果显示磷酸化模拟突变确实会对蛋白质相互作用产生影响。即便在现有的磷酸化数据集之中，我们估算仍有数千个由磷酸化介导的开关有待发掘。研究结果表明，相较于蛋白质组中无序区域，球状结构区域上的磷酸化位点常发挥开关功能，这或许是激酶（kinase）磷酸化的古老功能之一。