Dynamics of Substrate-Protease Interactions

该提案的主要目的是实现膜内蛋白酶底物的更好的机制理解，并通过理论预测扩展已知底物的谱。为此，我们将研究底物跨膜结构域的结构和动力学要求，并将它们与使用硅建模和生物信息学的序列联系起来。因此，我们希望揭示将蛋白质特性与可切割性联系起来的代码。在目标1中，我们将使用分子动力学模拟来表征已知γ分泌酶底物跨膜结构域的局部和全局动力学。我们将通过允许识别关键动力学基元的灵活性概况来表征特定地点的动力学。此外，我们将研究γ -分泌酶底物是否具有大规模骨干动力学的共同模式，以及影响裂解的突变是否会干扰这些全局运动。伽马分泌酶底物的骨干动力学将与菱形底物PINK1和SPPL蛋白酶的底物进行比较。要回答的关键问题将是灵活性谱是否区分酶结合位点、切割位点和铰链，以及该联盟将识别的底物跨膜结构域的结构动力学与非底物跨膜结构域的动力学相比如何。目标2是基于序列的结构柔性区域预测。我们将开发一种机器学习方法，根据两种类型的数据从序列中预测结构灵活性：从已知的跨膜蛋白3D结构中获得的晶体学b因子和由分子动力学模拟生成的灵活性概况。在目标3中，将通过序列分析和机器学习来预测膜内蛋白酶的新底物，使用该联盟确定的扩展底物序列集以及其他研究人员确定的越来越多的底物，特别是γ -分泌酶。除了单纯的序列基序，我们将利用与TM区域有关的广泛的结构特征，包括灵活性概况。此外，我们将利用各种类型的基因组背景，如底物与其同源蛋白酶的共表达以及分子相互作用网络的拓扑结构，来发现额外的候选底物，这些底物不一定包含可识别的切割位点基序。

The main objective of the proposal is to achieve a better mechanistic understanding of intramembrane protease substrates and to extend the spectrum of known substrates by theoretical predictions. To this end we will investigate the structural and dynamical requirements of a substrate transmembrane domain and relate them to sequence using in silico modeling and bioinformatics. We thus hope to uncover the code that links protein properties to cleavability. In Goal 1 we will use molecular dynamics simulations in order to characterize the local and global dynamics of the transmembrane domain of known gamma-secretase substrates. We will characterize site-specific dynamics by flexibility profiles which allow the identification of key dynamical motifs. Further, we will investigate whether gamma-secretase substrates share a common pattern of large-scale backbone dynamics and whether or not mutations affecting cleavage interfere with these global motions. The backbone dynamics of gamma-secretase substrates will be compared to that of the rhomboid substrate PINK1 and the substrates of SPPL proteases. The crucial questions to answer will be whether flexibility profiles discriminate between enzyme binding sites, cleavage sites, and hinges and how the structural dynamics of substrate transmembrane domains compares to the dynamics of non-substrate transmembrane domains to be identified by this consortium.Goal 2 is the sequence-based prediction of structurally flexibible regions. We will develop a machine learning approach to predict structural flexibility from sequence based on two types of data: crystallographic B-factors derived from known 3D structures of transmembrane proteins and flexibility profiles generated by molecular dynamics simulations. In Goal 3 novel substrates of intramembrane proteases will be predicted by sequence analysis and machine learning using the expanded set of substrate sequences to be determined by this consortium as well as the growing number of substrates determined by other researchers, in particular for gamma-secretase. Beyond mere sequence motifs we will exploit a broad spectrum of structural features pertaining to TM regions, including the flexibility profiles. Furthermore, we will exploit various types of genomic context, such as co-expression of substrates with their cognate proteases as well as the topology of the molecular interaction network, to uncover additional candidate substrates that do not necessarily contain recognizable cleavage site motifs.