Nuclear magnetic resonance and liquid chromatography–mass spectrometry combined with an incompleted separation strategy for identifying the natural products in crude extract

Abstract

NMR and LC–MS combined with an incompleted separation strategy were proposed to the simultaneous structure identification of natural products in crude extracts, and a novel method termed as NMR/LC–MS parallel dynamic spectroscopy (NMR/LC–MS PDS) was developed to discover the intrinsic correlation between retention time (Rt), mass/charge (m/z) and chemical shift (δ) data of the same constituent from mixture spectra by the co-analysis of parallelly visualized multispectroscopic datasets from LC–MS and ¹H NMR. The extracted ion chromatogram (XIC) and ¹H NMR signals deriving from the same individual constituent were correlated through fraction ranges and intensity changing profiles in NMR/LC–MS PDS spectrum due to the signal amplitude co-variation resulted from the concentration variation of constituents in a series of incompletely separated fractions. NMR/LC–MS PDS was applied to identify 12 constituents in an active herbal extract including flavonol glycosides, which was separated into a series of fractions by flash column chromatography. The complementary spectral information of the same individual constituent in the crude extract was discovered simultaneously from mixture spectra. Especially, two groups of co-eluted isomers were identified successfully. The results demonstrated that NMR/LC–MS PDS combined with the incompleted separation strategy achieved the similar function of on-line LC–NMR–MS analysis in off-line mode and had the potential for simplifying and accelerating the analytical routes for structure identification of constituents in herbs or their active extracts.

Introduction

Analysis of individual components in a complex mixture is important in many research fields such as in the discovery of new potential drugs from herbal medicines. Because of the complexity of the crude extracts, various on-line hyphenated techniques have been developed for the analysis of the complex mixtures. These techniques include liquid chromatography (LC) tandem mass spectrometry (MS/MS) [1], [2], [3], [4], [5], [6], [7], LC–nuclear magnetic resonance (NMR) spectrometry [8], [9], [10], [11], [12] and LC–NMR–MS [13], [14], [15], [16], [17]. They facilitate the structure determination of unknown constituents in crude extracts. Among the hyphenated techniques, LC–NMR–MS was the most powerful method, as it can provide a straightforward and real-time MS and NMR information of an individual component. Despite the promising performance of LC–NMR or LC–NMR–MS, some technical drawbacks still exist in direct hyphenated methods involving the use of NMR [18], [19], [20], such as lower detection sensitivity, high cost and narrow range of deuterated solvents as LC eluents and the compatibility of the chromatographic peak volume with that of the NMR flow cell. Recently, many improvements on the hyphenated NMR techniques were reported, including the use of miniaturized flow cells, higher magnetic field strengths, and cryogenically cooled NMR probes and preamplifiers [20], [21]. In addition, LC solid-phase extraction (SPE) NMR [22], [23], [24], [25], [26] has been developed by using a SPE interface between LC and NMR to enable multiple SPE trapping of an analyte of interest eluted from LC. The combination of capillary LC (capLC) and microcoil NMR (capNMR) has also been investigated (e.g., LC-SPE-capNMR [25] and capLC-capNMR [19]) to work with small amounts of samples made possible by the increased sensitivity of solenoidal NMR microcoils. These hyphenated techniques shared the common character of “separation prior to analysis”. Without the complete chromatographic separation of individual components in a mixture, the co-eluting analytes will produce broad and unsymmetrical peaks compromising the quality of the NMR spectra [19].

To address the problem of structure identification of individual components in a complex mixture, new analytical methods based on chemometrics have been applied to deconvolute the mixture MS or/and NMR spectra. For examples, band-target entropy minimization (BTEM) [27], [28], [29], molecular factor analysis (MFA) [30], [31], mean field independent component analysis (MF-ICA) [32], [33], DemixC [33], [34], and hard modeling methods [35]. In metabonomics, statistical total correlation spectroscopy (STOCSY) [36], [37], [38], statistical diffusion-ordered spectroscopy (S-DOSY) [39], and heteronuclear statistical total correlation spectroscopy (HET STOCSY) [40], [41] were developed for the analysis of complex ¹H NMR spectra to facilitate the discovery of potential biomarkers. The statistical heterospectroscopy (SHY) [42] was established as a new statistical paradigm for the co-analysis of multiple ¹H NMR and LC–MS datasets and was also applied to discover potential biomarkers in metabonomics.

In the past few years, we reported a series of methods for rapid structure identification of constituents in crude extracts from herbal medicines and traditional Chinese medicines (TCM) by MS/MS and LC–MS/MS [1], [2], [3], [43], [44], [45], [46], [47]. However, NMR data were sometimes needed to provide complementary information to validate the chemical structure determined by MS/MS.

In the present study, we focus on simplifying the separation and analysis procedure for complex mixtures with a goal of simultaneous and unambiguous structure identification of individual constituents in an active herbal extract. A novel method termed as NMR/LC–MS parallel dynamic spectroscopy (NMR/LC–MS PDS) has been developed and is shown in Scheme 1. This method is based on the off-line analysis of a series of incompletely separated chromatographic fractions with different concentration changing profiles of the constituents. Furthermore, NMR/LC–MS PDS was applied successfully for simultaneous identification of the major constituents in an active extract obtained from Gossypium herbaceam L., named AB-8-2, which was separated into a series of fractions with different concentrations by flash column chromatography [48], [49], [50] for its large sample loading and moderate resolution. Through the co-analysis of visualized MS and NMR data with signal amplitude co-variation in the NMR/LC–MS PDS spectra, we can discover the intrinsic correlation between retention time (Rt), mass/charge (m/z) and chemical shift (δ) data of the same individual constituent in the LC fractions. As a consequence, the complementary spectral information was obtained from mixture spectra for unambiguous structure identification of individual constituents in crude extracts.