Simulated isotope exchange patterns for protein structure determination

细胞功能是由一个巨大的蛋白质相互作用网络（PPIs）协调的，在人类中，这个网络被认为涉及蛋白质之间大约500,000个二元接触。几乎每个细胞过程都需要某种形式的PPI，理论上可以通过分子治疗来调节，为药物开发开辟了新的可能性。蛋白质对接已经出现，以满足在分子水平上理解大量不同PPIs的挑战。对接是用来预测从其组成单体的高分辨率结构的蛋白质组装的结构。虽然最近在蛋白质对接方面取得了许多进展，但从通常产生的潜在状态的高背景中提取天然组装体仍然是一个重大挑战。需要新的方法来改进蛋白质对接输出的排序。在数据驱动的对接中，蛋白质结构的生物物理/生化方面被用作约束来指导从蛋白质对接模拟中选择天然构象。实验输出通常对每个对接姿势进行模拟，从而根据模拟轮廓与实验所得轮廓的一致程度对其进行排名。有许多不同类型的实验约束被用于指导蛋白质对接，包括那些报道蛋白质大小、形状和相互作用强度的约束。还有一系列方法可以直接洞察蛋白质-蛋白质界面，并为数据驱动的对接提供了巨大的潜力。然而，利用这些所谓的足迹方法提供的独特见解已被证明具有挑战性。氢氘交换质谱（HDX-MS）是研究蛋白质构象的一种强大的足迹和生物物理方法。该技术报告了在分解为5 - 20个氨基酸的肽单元的D2O溶剂中，蛋白质的主酰胺质子与氘交换时质量的变化。对于二元PPI， HDX-MS测量通常报告为结合和未结合蛋白质之间的差异图。这些数据为相互作用的蛋白质区域提供了直接的见解，因为在蛋白质界面上的同位素吸收在组装中由于溶剂排斥而较慢。HDX-MS快速，敏感，需要最少的样品制备，易于自动化，具有彻底改变蛋白质对接的潜力。然而，该技术目前仅限于提供蛋白质界面的定性信息，不能用于批判性地评估不同的对接姿势或对模拟输出进行排序。在本研究中，我将开发直接从蛋白质对接输出中模拟HDX-MS数据的方法，以实现高通量和高效的整个对接模拟排序。我已经提供了重要的初步结果，清楚地显示了这种方法从蛋白质对接输出中提取天然结构的潜力。由于HDX-MS最近的商业化，这种方法是完全新颖的，也及时地增加了对HDX-MS的兴趣。它将在生物科学和生物制药研究的广泛基础上得到应用，并将为HDX-MS和分子对接的从业者提供一种新的工作方式。需要资金来开发方法和优化代码，使它们更适合蛋白质界面。该方法将被编写为预编译的可执行文件，并根据要求提供给研究人员，允许他们从任何蛋白质对接输出中模拟HDX-MS数据。该技术还将用于表征葡萄球菌抑制蛋白（Stl）和phi-11 dUTPase之间的四级组织，以揭示这一重要生物系统的新亮点。

Cell function is orchestrated by a large network of protein-protein interactions (PPIs) which in humans is thought to involve around 500,000 binary contacts between proteins. Almost every cellular process requires some form of PPI which could in theory be modulated a molecular therapeutic opening up new possibilities for drug development. Protein docking has emerged to meet the challenge of understanding the vast number of different PPIs at the molecular level. Docking is used to predict the structure of a protein assembly from the high resolution structures of its constituent monomers. While there have been many recent advances in protein docking extracting native assemblies from the high background of potential states that are typically generated remains a significant challenge. New approaches that allow improved ranking of protein docking outputs are required.In data-driven docking biophysical/biochemical aspects of protein structure are used as restraints to guide the selection of native conformations from protein docking simulations. Experimental outputs are typically simulated for each docking pose allowing them to be ranked according to the degree to which their simulated profiles agree to those obtained by experiment. There are many different types of experimental restraints that have been used to guide protein docking including those that report on protein size, shape and interaction strength. There are also a range of methods that provide direct insight into protein-protein interfaces and which offer enormous potential for data-driven docking. Nevertheless, leveraging the unique insights provided by these so-called footprinting methods has proven challenging.Hydrogen deuterium exchange mass spectrometry (HDX-MS) is a powerful footprinting and biophysical approach to study protein conformations. The technique reports on the change in mass accompanying the exchange of a protein's backbone amide protons for deuterium in D2O solvent resolved in peptide units of between 5 - 20 amino acids. For a binary PPI HDX-MS measurements are typically reported as difference plots between the bound and unbound proteins. These data provide direct insight into interacting protein regions as the isotope uptake at protein interfaces is slower in the assemblies due to solvent exclusion. HDX-MS is fast, sensitive, requires minimal sample preparation and is easy to automate and has the potential to revolutionise protein docking. However, the technique is currently limited to providing qualitative information on protein interfaces that cannot be used to critically evaluate different docking poses or to rank the simulated outputs.In this research I will develop methods to simulate HDX-MS data directly from protein docking outputs to allow high throughput and efficient ranking of whole docking simulations. I have provided significant primary results that clearly show the potential of this method to extract native structures from protein docking outputs. This approach is completely novel and also timely given the increased interest in HDX-MS due to its recent commercialisation. It will have applications across a broad base of bio-science and bio-pharmaceutical research and will represent a new way of working for practitioners of HDX-MS and molecular docking. Funding is required for development of the approach and optimisation of the codes so they are more suited for protein interfaces. The method will be written as a precompiled executable and be provided to researchers on request allowing them to simulate HDX-MS data from any protein docking outputs. The technique will also be applied to characterise the quaternary organisation between the Staphylococcal repressor protein (Stl) and the phi-11 dUTPase to shed new light on this important biological system.