统计相关NMR和LC-MS数据对复杂混合物结构解析和定量分析的研究

Identification and quantification of each component from the complex mixture sample is the core problem in many research fields, such as metabolomics (metabolite analysis), analysis of natural products extracts and medicine, etc. The nuclear magnetic resonance (NMR) spectroscopy and HPLC/UPLC hyphenated mass spectroscopy (LC-MS) are the most common and powerful tools widely used for structure and quantitative analysis. With statistically correlated NMR spectra and LC-MS data, a statistical paradigm with high information densities can be generated, which is of great help and significance to solve the problem mentioned above. In this project, model mixture consist of several typical metabolite compounds (nicotinamide，L-leucine，creatinine，betaine ，3-hydroxybutyric acid，citric acid and hippurate acid) is prepared to set up a method to achieve structure elucidation and quantification each component from the complex mixture sample. This method, after being established and optimized, is used for the analysis of practical samples (metabolite of caenorhabditis elegans). Other than using conventional 1H 1D spectrum in which signals are easily overlapped, 13C detected 1D experiments such as 1D 13C spectrum and DEPT 90 and 135 spectra are used taking advantage of the wide dispersion of the 13C chemical shifts, together with 2D experiment such as TOCSY and 2D 1H -13C correlation spectra (1H -13C HSQC and HMBC in this project). The multiple NMR experiment (1D and 2D experiments above) can make sure the data accuracy and high information densities for Identification and quantification of certain component from the complex mixture sample. Through the establishment and application of the method, a more useful and accurate tool can be provided for individual compound identification and quantification in many research fields.

复杂混合物体系中单个化合物的结构指认和定量分析是诸如代谢组学、天然产物分析、药物分析等领域的核心难题。核磁共振技术（NMR）和液相色谱质谱联用技术（LC-MS）是结构解析和定量研究最常用也最有力的两种检测手段，而通过统计相关的方法将二者数据结合起来可以获得高内涵的结构与定量信息，有助于解决上述核心难题。本项目拟以模式混合物为对象，建立通过NMR和LC-MS数据的相关性提取混合物中每个化合物结构和定量信息的方法，最终将该方法应用于实际复杂混合物样品（线虫代谢物）的结构与定量研究；其中NMR数据来自于多种实验，包括：一维碳谱，dept90，dept135和高分辨二维TOCSY、HSQC、HMBC，区别于以往方法中使用的一维氢谱带来的NMR数据误差。通过此方法的建立和应用，为代谢组学等研究领域提供更准确更有效的结构解析和定量分析研究手段。

项目开展三年以来，我们已完成该项目计划的主要内容，达到预期研究目标。.在本项目中，我们针对目前代谢组学普遍遇到的问题——如何快速、准确、可靠地鉴别代谢物结构，提出一种新策略，即利用核磁共振技术(NMR)和液相色谱质谱联用技术(LC-MS)的互补优势，基于统计相关原理对相同样品的LC-MS和多种NMR代谢组学数据进行融合分析，帮助代谢物化学结构的快速鉴别，提升化学结构鉴别的可靠性。.我们对常见代谢物溶解度及磷酸盐缓冲液对其一定浓度范围内的水溶液的缓冲能力进行考察，筛选出6个常见代谢物包括亮氨酸、谷氨酰胺、牛磺酸、肌酐、甜菜碱和烟酰胺，构建一种模式混合物体系。制备10个模式混合物样品，同时采集LC-MS(正、负离子模式)和多种NMR(1D 13C、DEPT 90、DEPT 135、1H-13C HSQC、1H-1H TOCSY)数据。采用开源R包XCMS完成LC-MS数据的前处理，主要包括峰识别、峰匹配和对齐、峰补齐，并利用CAMEAR包对LC-MS特征(或变量)进行同位素、加合物注释；利用代谢组工具包(由Arthur S. Edison教授友好提供的matlab语言软件包)完成NMR数据的批量处理。基于同一代谢物的LC-MS信号和NMR信号随浓度变化而变化，采用实验室内部matlab脚本实现统计相关分析和去卷积聚类，并重构各组分的伪MS和伪NMR图谱，尝试性解析出各组分的化学结构。随后将各组分重构图谱与其相应的实验图谱重叠比较，结果显示高度的一致性，表明LC-MS和NMR统计相关分析结果准确可靠。.本项目实现了一整套LC-MS、1D NMR和2D NMR代谢组数据处理分析流程，利用模式混合物证实LC-MS和NMR数据统计相关分析的可靠性，这对于帮助真实代谢组研究中代谢物的鉴别具有重要意义。