Designing interactions mediated by disordered protein regions

蛋白质是细胞的主要成分，它们几乎参与了细胞中发生的所有过程。它们的关键作用是调节细胞功能和传递信号的相互作用。在过去的三十年中，在分子内和分子间蛋白质相互作用的结构和功能表征方面取得了巨大进展。此外，已经开发出计算和实验蛋白质工程工具，使我们能够设计和选择具有所需相互作用特征的新蛋白质。这一进展对折叠成明确三级结构的“经典”蛋白质尤其有影响。然而，许多蛋白质含有大的内在无序区（IDRs）。对于这些通常更动态的蛋白质元件，在结构表征和功能再造方面的进展要温和得多。鉴于idr在变构和位阻活性调控以及调控和信号网络中枢纽蛋白的动态组装中发挥的重要作用，很明显，我们需要开发新的工具，使我们能够更好地理解idr介导的相互作用，并设计具有所需相互作用特性的idr。在本研究计划中，我们建议采用计算和实验方法相结合的方法，以便为具有所需相互作用特性的idr的成功工程建立新的管道。在我们的第一个目标中，我们将开发基于深度学习的工具，用于识别通过瞬时分子内相互作用作为蛋白质活性开关的idr。利用这些新工具和基于大规模分子动力学（MD）模拟的突变筛选，我们将设计具有具有特定开关行为的idr的转录因子。在第二个目标中，我们将使用基于流式细胞术的筛选方法结合深度学习来确定哪些序列特征赋予idr相分离能力，分别是定位应力颗粒的能力。我们将利用所学到的知识来设计具有可调相分离行为的idr，即期望的分子间相互作用特性。这里提出的研究计划旨在为最近发展和非常通用的idr的开发开辟新的途径，这将对我们设计蛋白质内部和蛋白质之间相互作用的能力产生重大影响，从而引导特定的细胞过程在学术界和工业界的应用。

Proteins are the principle constituents of cells, and they are involved in nearly all processes that take place in cells. Key to their contributions are interactions that regulate cellular functions and transmit signals. In the last three decades, tremendous progress has been made in the structural and functional characterization of intra and intermolecular protein interactions. Moreover, computational and experimental protein engineering tools have been developed that enable us to design and select for new proteins with desired interaction features. This progress has been particularly impactful for "classical" proteins that fold into well-defined tertiary structures. However, many proteins contain large intrinsically disordered regions (IDRs). For these generally more dynamic protein elements, progress in structural characterization and functional reengineering has been much more modest. Given the important role IDRs play in allosteric and steric activity regulation as well as dynamic assembly of pivot proteins in regulatory and signaling networks, it is clear that we need to develop new tools that allow us better understand IDR-mediated interactions as well as engineer IDRs with desired interaction properties. In this research program, we propose to employ a combination of computational and experimental approaches in order to set up new pipelines for the successful engineering of IDRs with desired interaction properties. In our first aim, we will develop deep learning-based tools for the identification of IDRs that, via transient intramolecular interactions, act as switches of protein activity. Taking advantage of these new tools and large-scale, molecular dynamics (MD) simulation-based mutation screens, we will then design transcription factors that have IDRs with specific switch behaviour. In the second aim, we will use a flow cytometry-based screening method in combination with deep learning to identify which sequence features confer phase-separation capabilities to IDRs, respectively, the ability to locate to stress granules. We will exploit what has been learned to design IDRs with tunable phase separation behaviors, i.e. desired intermolecular interaction properties. The research program proposed here aims to open up new avenues for the exploitation of the recently evolved and very versatile IDRs, which will have significant impact on our ability to design interactions within and between proteins and thus steer specific cellular processes for applications in academia and industry.