节点文献
天然产物混合物结构鉴定的MS与NMR平行动态谱和正交相关谱新型分析方法研究
【作者】 代冬梅；
【作者基本信息】 中国协和医科大学 ， 药物分析， 2008， 博士
【摘要】 本论文基于不完全分离的新型分析策略,结合化学计量学方法,主要采用MATLAB软件编写及优化数据处理程序,对系列混合物样品的LC-MS和NMR谱数据进行了整合分析,探索并建立了两个互补的新型谱学分析方法,即LC-MS与~1H NMR的平行动态谱(NMR/LC-MS PDS)和正交相关谱(NMR/LC-MS CHS)。通过从混合物谱中发掘出来自提取物中同一组分的LC-MS和NMR谱数据间的内在相关性,获取包括保留时间(Rt)、质荷比(m/z)和化学位移值(δ)等相关且互补的关键谱学数据,实现了中草药物质基础中多组分无需完全分离,高效、多容量的快速结构分析鉴定。为了建立本分析方法,首先采用结构已知的黄酮类标准品模拟不完全分离的色谱情形,设计并配制了组分含量呈连续动态变化且色谱重叠程度不同的两组人工系列混合物样品进行探索研究。在此基础上,综合运用NMR/LC-MS PDS和NMR/LC-MS CHS这两个谱学分析方法对具有抗老年痴呆和抗血栓活性的草棉植物活性提取物AB-8-2中黄酮醇类成分进行了不完全分离的同步结构鉴定研究,共分析确认了12个组分的结构,其中包括2组共流出的同分异构体。此外,初步开发了以MATLAB软件为基础的数据分析及可视化工具。本论文的研究内容主要包括以下四个部分:1.不完全分离的分析策略通过计算机模拟了几类具有代表性的不完全分离系列混合物的色谱分离情形,并以组分谱学信号强度间的相关系数为指标,分别考察了系列混合物的色谱分离度、组分相对含量,组分数目和谱学信号重叠对相关系数的影响。研究结果表明,组分浓度和组分数目对相关系数的影响不明显,但是谱学信号重叠会降低谱学信号间的相关系数,而相邻两个组分间的分离度对其相关系数具有决定性的影响。相比于完全分离的色谱分离度(R=1.5),本研究提出的不完全分离的分析策略其特点之一是对于药用植物复杂混合物体系的分析,降低了对色谱分离度的要求,其关键是在一定程度的色谱不完全分离情形下,获取组分含量呈连续动态变化的系列流份,以保证系列混合物样品中组分间的相关系数要小于同一组分的自相关系数。另外,鉴于自动化的多通道组合闪式色谱分离系统具有中等的分离度和较大上样量的特点,本研究重点采用该技术来获取所需的组分含量呈连续动态变化的系列提取物样品。2.NMR与LC-MS平行动态谱和正交相关谱分析方法的建立采用6种结构已知的黄酮类标准品,模拟不完全分离的色谱情形,设计并配制了组分含量呈连续动态变化且色谱重叠程度不同的两组人工系列混合物。然后分别进行了LC-MS和~1H NMR谱测定,并采用MATLAB软件编写的数据处理程序,对原始实验数据进行处理与分析。基于不完全分离的系列混合物样品中多个组分的浓度及其分布范围具有规律地连续动态变化的特征,将具有信号强度协同变化的多个LC-MS和~1H NMR谱平行相关及可视化,从而构成了NMR/LC-MS平行动态谱(NMR/LC-MS PDS)。在此基础上,运用化学计量学算法获取系列人工混合物组分的LC-MS和~1H NMR谱信号强度间的相关系数,并设定合适的相关系数截取值构成了NMR/LC-MS正交相关谱(NMR/LC-MS CHS)。通过发现混合物中同一组分的质荷比(m/z)和化学位移值(δ)间的相关性信息,实现了混合物组分的分子离子,碎片离子,化学位移值和偶合裂分等关键结构信息的获取。其中,通过NMR/LC-MS PDS谱的可视化处理,观察到信号强度变化范围及趋势的精细特征,从而对重叠的~1H NMR信号进行了识别和解卷积分析。此外利用不同颜色显示的XIC谱数据直观地观察到混合物样品中同分异构体的组成情况。还可以指导数据处理过程中化学位移值的分组及~1H NMR信号峰对齐处理。通过对LC-MS数据进行化学计量学的相关计算,可以预测出NMR/LC-MS CHS谱中各个组分的相关系数截取值。3.复杂活性提取物组分群的NMR与LC-MS平行动态谱和正交相关谱分析方法研究基于不完全分离的分析策略,采用具有中度分离和高载样量特点的多通道组合闪式色谱技术为主要分离手段,首先将具有抗老年痴呆和抗血栓活性的草棉植物提取物(AB-8-2)分离制备成组分含量呈连续动态变化的系列提取物样品。利用NMR/LC-MS PDS和NMR/LC-MS CHS新型分析方法,对以上系列提取物样品的LC-MS和~1H NMR谱数据进行了整合分析,并从混合物谱中获得了来自同一组分的MS和~1H NMR谱的相关性信息。在此基础上,成功地鉴定出AB-8-2混合物中相对含量在0.1～39.4%之间的12个黄酮醇类化合物的结构。除了槲皮素和山奈酚之外,包括5个以槲皮素为母核结构的糖苷类成分和3个以山奈酚结构为母核的糖苷类成分,其中还包括两组共流出的同分异构体。此外,其它相关的谱学数据,如LC-MS/MS,1D TOCSY或1D NOESY谱也可以在上述通过相关性分析已确定归属信号的引导下被结合进来,补充和丰富结构鉴定所需的谱学信息,实现准确鉴定混合物中结构相似或同分异构体结构的目的。4.数据处理程序的编写及界面开发将系列混合物样品的LC-MS谱和~1H NMR谱数据经过工作站软件的预处理后,转换成文本格式的文件。然后以MATLAB软件为平台,编写及优化了数据处理程序,进行了包括信号提取,谱峰对齐,相关性计算等处理操作,完成了以上两种谱学原始数据的处理。基于不完全分离的系列混合物样品中多个组分的浓度及其分布范围规律地连续动态变化的特征,将具有信号强度协同变化的多个LC-MS和~1HNMR谱进行平行相关及可视化,从而构成了NMR/LC-MS PDS谱。进一步通过化学计量学的方法获取系列人工混合物组分的LC-MS和~1H NMR谱信号强度间的相关系数,预测和设定合适的相关系数截取值,构成了NMR/LC-MS CHS谱。其中,将获取的LC-MS数据处理成m/z-Rt数据对作为独立的向量与NMR数据进行相关性分析,达到了将混合物中同分异构体的相关性信息按保留时间加以区分的目的。数据处理程序的重点和难点在LC-MS谱中提取离子数据的生成,及~1H NMR谱信号的峰对齐的自动化处理上。因此,在程序编写的基础上,初步设计了以MATLAB软件为平台的操作界面系统,使数据处理过程简明直观,便于掌握和操作。
【Abstract】 The NMR/LC-MS parallel dynamic spectroscopy(NMR/LC-MS PDS) and NMR/LC-MS correlation heterospectroscopy(NMR/LC-MS CHS)were established based on incompleted separation strategy for the simultaneous structure identification of natural products in crude extract. NMR/LC-MS PDS and NMR/LC-MS CHS were formed through co-analysis of multi-spectroscopic data sets from LC-MS and ~1H NMR with signal amplitude co-variation by the routines compiled and optimized in MATLAB.Based on concentration variation of constituents in series of incompletely separated fractions,the intrinsic correlation between retention time,m/z and chemical shift data of the same constituent from mixture spectra can be discovered.The two methods were primarily designed using two series of artificial mixtures,made of known flavonoids to simulate the fractions from incompleted chromatographic separation,in which constituents had different concentration variation profiles and chromatographic overlap degree.Furthermore,NMR/LC-MS PDS and NMR/LC-MS CHS were applied successfully to identification of 12 constituents in active herbal extract(AB-8-2) involving flavonol glycosides,including two sets of coeluted isomers.Besides,a data processing and visualization tool was developed based on MATLAB software.1.Ineompleted separation analysis strategySeveral series of incompleted separation mixtures were simulated with computer, then the four factors of chromatographic resolution,relative concentration,number of constituents and signal overlap were studied with the correlation coefficient between constituents as index.The results indicated that relative concentration and number of constituents had a minor influence,signal overlap reduced the correlation coefficients, and the chromatographic resolution had a notable influence.However,compared with the resolution of 1.5,one of the strategy had a lower request on chromatographic resolution, the key was to acquire the series of fractions with concentration variation of constituents, to ensure that the correlation coefficient between constituents were smaller than that of the self-correlations.Besides,based on the moderate resolution and larger sample loading of the automatic flash column chromatography,it was used to prepare the series of fractions with different concentration variation requested in this new analytical strategy.2.The establishment of NMR/LC-MS PDS and NMR/LC-MS CHSTwo series of artificial mixtures with concentration variation of constituents and different chromatographic overlaps were made of six known flavonoids,to simulate the fractions from incompleted chromatographic separation.The LC-MS and ~1H NMR spectra of the series of artificial fractions were acquired,and then processed with routines compiled and optimized in MATLAB.NMR/LC-MS PDS spectra was formed through co-analysis of parallel visualized multi-spectroscopic data sets from LC-MS and ~1H NMR with signal amplitude co-variation,based on regular concentration variation and sample distribution of constituents in series of incompletely separated fractions. Furthermore,the correlation coefficients between the signal intensities of LC-MS and NMR spectra from all the mixture samples were calculated with chemometrics algorithms, and were visualized with a reasonable cutoff to form NMR/LC-MS CHS.On the basis of integrated analysis,the correlation relationship between ions(m/z) and chemical shifts(δ) deriving from the same molecule can be extracted simultaneously from the mixture spectra.,thus the structural information including retention time,molecular ion,fragment ion,chemical shift and couple pattern of a molecule can be obtained.With the visualization of NMR/LC-MS PDS spectrum,the fine characters of changing ranges and profiles of signal intensity can be observed,which will facilitate the reorganization and deconvolution analysis of overlapped ~1H NMR signals;with the color coded XIC data,the isomers existed in mixture can be observed intuitionally;it also can guide the grouping and peak alignment processing of ~1H NMR data.With the correlation calculation of LC-MS data,the cutoff value of correlation coefficient of each constituent in NMR/LC-MS CHS spectrum can be forecasted.3.Appling NMR/LC-MS PDS and NMR/LC-MS CHS on the study of complex active herbal extractBased on incompleted separation analysis strategy,the active herbal extract AB-8-2 with remarkable activity of resisting mental retardation and antithrombotic was separated into series of fractions with concentration variation of constituents by the automatic flash column chromatography,which had the character of moderate resolution and larger sample loading.Based on the integrated analysis of the LC-MS and 1H NMR data of the obtained series of fractions by NMR/LC-MS PDS and NMR/LC-MS CHS,the correlation relationship between MS and ~1H NMR data deriving from the same molecule can be extracted from the mixture spectra.Furthermore,12 flavonol compounds with relative concentration ranged from 0.1%to 39.4%in active herbal extract were identified successfully.Besides quercetin and kaempferol,there were five quercetin-based glycosides and three kaempferol-based glycosides,as well two sets of co-eluted isomers.Moreover,other relevant spectral data, including LC-MS/MS,1D TOCSY or 1D NOESY can be integrated under the guide of the assigned signals,which will provide the complemented and comprehensive data for the unambiguous structure identification of the constituents with similar structure or isomers existed in mixture.4.Data processing routines and user interfaceWith the preprocessing of the LC-MS and ~1H NMR data by the workstation software, the raw data were converted into text format file.Then the data were processed by the routines compiled and optimized in MATLAB,including signal extraction,peak alignment and correlation calculation.NMR/LC-MS PDS spectra was formed through co-analysis of parallel visualized multi-spectroscopic data sets from LC-MS and ~1H NMR with signal amplitude co-variation,based on regular concentration variation and sample distribution of constituents in series of incompletely separated fractions; NMR/LC-MS CHS with a reasonable cutoff was formed by visualizing the correlation coefficients between the signal intensities of LC-MS and NMR spectra from all the mixture samples calculated with chemometrics algorithms.Moreover,the LC-MS data were converted into series of independent vectors,which were used in the correlation calculation with the NMR data,such a data processing mode can facilitate to distinguish the correlation information of isomers by retention time.The emphases and difficulties in routines compiling were existed in the data extraction of LC-MS data and the automation of peak alignment of ~1H NMR signals.Based on the routines,the user interface was developed,which will facilitate the data processing courses and was easy to master and operate.