人工设计蛋白质的可折叠性的高效实验检测与优化

The De novo rational protein design occupies an exciting field in structural biology with increasing importance and huge application potentials in industrial and biomedical engineering. On the other hand, high-throughput approaches evaluating foldability of artificial designed proteins remain appealing and challenging in protein design field. Here we describe a new designing platform combining more accurate computational prediction and high-throughput approaches in designing and optimizing artificial proteins. We would utilize statistical potential to overcome the limitation of existed software while design several proteins with varies of conformations. It is an energy function derived from an analysis of known protein structures in the PDB database and previously we proved that using this potential readily improved the accuracy of artificial protein structural prediction. On the other hand, a high throughput endued split β-lactamase system which directly links the in vivo foldability of proteins to antibiotic resistance would be constructed, it's feasibility as well as parameter optimization will be studied. We would apply this system to the quantitative evaluation towards the folding of the designed proteins, and also conduct directed evolution to look for folding stabilized candidates from a high throughput library based on the wild type protein sequence. The structures of these stabilized mutants would be studied by NMR spectroscopy, as the results would serve to validate our computational prediction. Meanwhile, all experimental data (sequence, in vivo folding ability, structures, etc) would help optimize the designing process. Our long term goal is to build a generalized, theoretical and experimental approaches combined de novo protein designing platform, which would bring significant innovations to the artificial protein engineering study.

蛋白质序列设计有着广泛应用前景。如何设计热力学、动力学特性以及序列多样性与天然蛋白类似的人工序列，是并未解决的问题。目前，相关方法发展的主要障碍之一，是缺乏对序列可折叠性进行实验测定的高效方法。对设计序列是否可折叠只能逐条检验，成本高，效率低，有限的实验结果很难提示如何改进理论方法。本课题主要研究对序列可折叠性进行检测的高效实验方法：拟采用基于Split β-内酰胺酶将蛋白质折叠稳定性与宿主的抗药性关联的系统，将其应用于对理论设计序列，特别是对根据统计势能函数设计的序列的可折叠性进行系统分析；还拟利用该实验体系的高效筛选能力对设计序列的折叠能力进行实验改进和优化，从而为改进设计方法提供可靠的实验反馈和指导。

作为生命功能的主要执行者，蛋白质氨基酸序列和空间结构之间的关系是科学界悬而未决的课题。如何在广袤的序列空间中选择合适的序列，使之折叠成特定的结构，不但有助于认识这一科学问题，而且能够为按需创造人工功能蛋白打下基础。近十年来国际上蛋白质设计领域取得了一些重要进展，但有实验验证的自动设计方法只有寥寥一两种，缺乏高通量、高效率检验序列可折叠性的实验技术，是限制序列从头设计进一步发展的关键因素。.我们建立了基于β内酰胺酶活性的高效鉴定筛选普适性平台，可用于对设计蛋白折叠性的鉴定和筛选。针对特定的目标序列，已成功设计序列、鉴定其折叠性并解析其三维结构。对于未直接获得折叠性较好的目标序列，通过β内酰胺酶系统定向进化获得了优化的稳定折叠的蛋白突变体，并在此基础上将实验信息反馈于理论设计，在一定程度上指导了序列设计的优化，为理论验证和筛选提供了有效快捷的平台。基于统计能量函数构建策略建立的设计方法，直接或通过进化获得的可折叠性较好的序列，用X-射线晶体学或多维核磁共振方法进行结构解析，与设计模版结构比对表明，四个稳定折叠的人工设计蛋白的实际空间结构与设计目标高度一致，反映出此设计方法及折叠性检测优化平台的高效性。理论设计与实验反馈相辅相成，为蛋白质结构功能的设计改造提供了新方法和新工具，客观检验和拓展了我们对蛋白质序列-结构关系这一分子生物学中最基础问题的认识。本项目工作发表于2014年10月的Nature Communications。