De novo structure prediction of helical transmembrane proteins using RosettaTMH and pseudocontact shift NMR data

膜蛋白占所有细胞蛋白质的近30%，占所有已知药物靶点的50%以上。它们在包括信号转导、代谢物运输和神经传递在内的大量生理过程中发挥关键作用，但也参与癌症、糖尿病和肥胖症等病理情况的发展。尽管它们具有巨大的医学相关性，但膜蛋白占蛋白质数据库中所有蛋白质结构的不到2%。尽管结构生物学方法如X射线晶体学和核磁共振光谱学（NMR）取得了重大进展，但膜蛋白结构测定仍然是一个艰巨的挑战，通常只能获得有限的实验数据。计算方法即使在缺乏实验数据的情况下也可以提供结构模型。通过这种方式，可以生成可溶性蛋白质的结构模型;然而，用于膜蛋白质结构预测的计算方法还处于起步阶段。作为比较建模模板的同源蛋白质的结构缺失，因此从氨基酸序列开始的从头折叠是必要的。这些方法的成功，例如Rosetta-Membrane，迄今为止仅限于小蛋白质（<150个氨基酸），但可以通过纳入有限数量的实验结构信息来大幅改善。拟议研究项目的目的是开发一种新的，创新算法的从头结构预测的α-螺旋膜蛋白使用一组有限的伪接触位移（PCS）NMR数据。PCS是由顺磁性镧系元素离子引起的，并产生长距离和取向信息，这些信息可以指导结构预测算法并验证生成的结构模型。该项目旨在开发一个计算框架，用于将PCS集成到RosettaTMH中。该算法将预测的跨膜螺旋（TMH）组装成膜蛋白拓扑结构。随后的肽片段插入完成环区域并对蛋白质构象空间进行详尽的采样。将在包含不同大小（2-15 TMH）和折叠的膜蛋白的基准组上评价所开发的方法（称为RosettaTMH+PCS）的性能。此外，RosettaTMH+PCS将通过测定膜整合蛋白CNIH 1（chornicon蛋白家族成员）的结构进行实验性检测。因此，该项目旨在开发一种实验方案，用于膜蛋白的镧系元素标记和使用基于TROSY的NMR实验收集PCS数据的时间效率，这将加快结构预测过程。所提出的方法被认为是膜蛋白结构预测的一般路线，预计将产生更准确的结构模型，特别是对于大的膜蛋白，比以前的算法。

Membrane proteins constitute nearly 30% of all cellular proteins and more than 50% of all known drug targets. They play a key role in a plethora of physiological processes including signal transduction, transport of metabolites and neurotransmission, but are also involved in the development of pathological situations as cancer, diabetes and obesity. Despite of their enormous medical relevance, membrane proteins make up less than 2% of all protein structures present in the Protein Data Bank. Despite significant advances in methods of structural biology as X-ray crystallography and nuclear magnetic resonance spectroscopy (NMR), membrane protein structure determination remains a formidable challenge often yielding limited experimental data. Computational methods can provide structural models even in the absence of experimental data. In this way, structural models of soluble proteins can be generated; however, computational methods for the structure prediction of membrane proteins are in their infancy. Structures of homolog proteins as template for a comparative modeling are missing, for which reason a de novo folding, starting from the amino acid sequence, is necessary. The success of these methods, e.g. Rosetta-Membrane, is so far limited to small proteins (<150 amino acids), but can be drastically improved by the incorporation of a limited amount of experimental structural information.The aim of the proposed research project is the development of a novel, innovative algorithm for the de novo structure prediction of alpha-helical membrane proteins using a limited set of pseudocontact shift (PCS) NMR data. PCSs are caused by paramagnetic lanthanide ions and give rise to long range distance and orientation information that can guide the structure prediction algorithm and verify the generated structural models. The project aims to develop a computational framework for the integration of PCSs into RosettaTMH. This algorithm will assemble predicted transmembrane helices (TMHs) into membrane protein topologies. Subsequent peptide fragment insertion completes loop regions and exhaustively samples the protein conformational space. The performance of the developed method, termed RosettaTMH+PCS, will be evaluated on a benchmark set comprising membrane proteins of different sizes (2-15 TMHs) and folds. Furthermore, RosettaTMH+PCS will be tested experimentally through the structure determination of the integral membrane protein CNIH1, a member of the chornicon protein family. Thereby, the project aims to develop an experimental protocol for the lanthanide tagging of membrane proteins and the time efficient collection of PCS data using TROSY-based NMR experiments, which will speed up the structure prediction process. The proposed method is conceived as a general route for membrane protein structure prediction and is expected to produce more accurate structural models, especially for large membrane proteins, than previous algorithms.