南京大学黄硕团队报道了机器学习辅助的耻垢分枝杆菌孔蛋白a（MspA）电渗阱中不同电荷蛋白质的同步结构分析。相关研究成果发表在2022年1月7日出版的国际学术期刊《美国化学会杂志》。

纳米孔正在成为单分子蛋白质传感的一种手段。然而，蛋白质表现出不同的电荷性质，使得能够同时感知不同电荷蛋白质的传感器的设计复杂化。

该文中，研究人员引入了一种不对称的电解质缓冲液和耻垢分枝杆菌孔蛋白A（MspA）纳米孔相结合，形成一个电渗流（EOF）陷阱。载脂蛋白和全肌红蛋白只有一个血红素不同，通过这种方法可以完全区分。MspA EOF陷阱可同时实现溶菌酶、载脂蛋白/全肌红蛋白和ACTR/NCBD蛋白复合物（分别为碱性、中性和酸性蛋白质）的直接鉴别。为了实现事件分类的自动化，提取了多个事件特征，建立了一个机器学习模型，准确率达到99.9%。

该方法还可用于直接从乳清蛋白粉中鉴定单分子α-乳清蛋白和β-乳球蛋白。该蛋白质传感策略有助于从混合物中直接识别蛋白质，表明其在快速、灵敏检测生物标记物或实时蛋白质结构分析中的应用前景。

附：英文原文

Title: Machine Learning Assisted Simultaneous Structural Profiling of Differently Charged Proteins in a Mycobacterium smegmatis Porin A (MspA) Electroosmotic Trap

Author: Yao Liu, Kefan Wang, Yuqin Wang, Liying Wang, Shuanghong Yan, Xiaoyu Du, Panke Zhang, Hong-Yuan Chen, Shuo Huang

Issue&Volume: January 7, 2022

Abstract: The nanopore is emerging as a means of single-molecule protein sensing. However, proteins demonstrate different charge properties, which complicates the design of a sensor that can achieve simultaneous sensing of differently charged proteins. In this work, we introduce an asymmetric electrolyte buffer combined with the Mycobacterium smegmatis porin A (MspA) nanopore to form an electroosmotic flow (EOF) trap. Apo- and holo-myoglobin, which differ in only a single heme, can be fully distinguished by this method. Direct discrimination of lysozyme, apo/holo-myoglobin, and the ACTR/NCBD protein complex, which are basic, neutral, and acidic proteins, respectively, was simultaneously achieved by the MspA EOF trap. To automate event classification, multiple event features were extracted to build a machine learning model, with which a 99.9% accuracy is achieved. The demonstrated method was also applied to identify single molecules of α-lactalbumin and β-lactoglobulin directly from whey protein powder. This protein-sensing strategy is useful in direct recognition of a protein from a mixture, suggesting its prospective use in rapid and sensitive detection of biomarkers or real-time protein structural analysis.

DOI: 10.1021/jacs.1c09259

Source: https://pubs.acs.org/doi/10.1021/jacs.1c09259