Prediction of interactions involving intrinsically disordered proteins

固有无序(ID)蛋白在蛋白质相互作用中起着核心作用，因为它们的ID片段经常介导大量的相互作用。我们以前已经证明，由ID蛋白片段介导的相互作用的特征是将它们与其他蛋白质相互作用区分开来。我们假设这些特征可以用来(I)确定球状蛋白上优先与ID蛋白结合的相互作用部位，以及(Ii)确定ID蛋白片段在结合时采用的结构。基于这一假设，我们提出了以下目标：*目标1：开发一种识别与ID蛋白结合的球形蛋白质表面区域的预测器。*目标2：开发一种确定其中一个成员是ID蛋白的二元复合体结构的方法。*目标3：将这些方法应用于自抑制蛋白质。*目标1：使用区分ID蛋白结合表面与其他蛋白质表面的特定特征(包括柔韧性、静电势或残基守恒)，我们将训练分类器来区分这两种类型的表面。分类器的性能将通过十次交叉验证以及在训练中未使用的一组测试结构上进行测试。*目标2：我们将扩展最近公布的对接协议，使其可以用于确定其中一个成员是ID蛋白的复合体的结构。该协议使用多个步骤的低分辨率和高分辨率对接以及Rosetta势来对生成的结构进行评分。所开发的方案将以一组包含ID蛋白的复合体的已知高分辨率结构为基准。*目标3：在这里，我们将测试在目标1和2中开发的方法是否可以用于确定蛋白质的结构，其中ID片段在顺式结构中与相同多肽链的结构域相互作用，即作为自动抑制开关。我们将首先以一组结构已知的自体抑制蛋白为基准，然后用它来确定我们预测为自体抑制的结核分枝杆菌ABC转运蛋白Rv1747的自体抑制状态的结构。*我们相信，在这个研究计划中开发的工具将为未来开发用于设计分别抑制ID蛋白介导的模拟相互作用的治疗性多肽和蛋白质的计算方法奠定坚实的基础。

Intrinsically disordered (ID) proteins play a central role in protein interactomes because their ID segments often mediate large numbers of interactions. We have shown previously that interactions mediated by ID protein segments are characterized by features that distinguish them from other protein interactions. We hypothesize that these features can be used to (i) identify interaction sites on globular proteins that preferentially bind ID proteins and (ii) determine the structures that ID protein segments adopt upon binding. Based on this hypothesis, we propose the following aims: ***  ***Aim 1: Development of a predictor that identifies surface regions on globular proteins that bind ID proteins.***Aim 2: Development of a method to determine the structure of binary complexes in which one member is an ID protein.***Aim 3: Application of these methods on autoinhibited proteins.*******Aim1: Using specific features that distinguish ID protein-binding surfaces from other protein surfaces (including flexibility, electrostatic potential or residue conservation), we will train a classifier to distinguish the two types of surfaces. The performance of the classifier will be tested by ten-fold cross-validation as well as on a test set of structures not used in the training. *******Aim 2: We will extend a recently published docking protocol such that it can be used for the determination of the structure of complexes in which one member is an ID protein. The protocol uses multiple steps of low- and high-resolution docking and the Rosetta potential for the scoring of generated structures. The developed protocol will be benchmarked on a set of known high-resolution structures of complexes that contain ID proteins.*******Aim 3: Here, we will test whether the methods developed in Aim 1 and 2 can be used to determine the structure of proteins in which an ID segments interacts in cis with a domain of the same polypeptide chain, i.e., acts as an autoinhibitory switch. We will first benchmark the approach on a set of autoinhibited proteins for which the structure is known, and then use it to determine the structure of the autoinhibited state of the ABC transporter Rv1747 from Mycobacterium tuberculosis that we predict to be autoinhibited.******We are confident that the tools developed in this research program will constitute a solid base for the future development of computational methods used in the design of therapeutic peptides and proteins that inhibit, respectively, mimic interactions mediated by ID proteins.