Reducing signal interference in complex NMR spectra : characterization and quantification of metabolites in mixtures

Abstract

Nuclear magnetic resonance (NMR) spectroscopy is a versatile analytical technique that can be used for identification, characterization, and quantification of various compounds. However, many sample types, including most biological samples, are mixtures of numerous different compounds with large differences in concentration and physico-chemical properties. When such samples are studied with NMR spectroscopy, they typically give rise to complicated NMR spectra that are difficult to analyse due to large dynamic range and extensive signal overlap. Methods are therefore needed that reduce signal overlap and other interferences in NMR spectra of complex mixtures so that individual compounds can be characterized and quantified. The work in this thesis revolves around two such methods. One common form of signal interference is caused by intense, unwanted signals that overlap with, and sometimes obscure, signals of interest. Here, an NMR experiment was developed that enables selective suppression of the unwanted signals so that other signals can be studied. After evaluation and optimization of the experiment, it was used in the first complete NMR spectral assignment of the minor furanose forms of glucose. Broad signals from lipids or macromolecules is another type of interference. Such signals are frequently encountered in NMR spectra of metabolomics samples, where they prevent accurate quantification of certain metabolites. In this thesis, an automated workflow was devised and optimized that first eliminates the interference from broad signals and then calculates absolute metabolite concentrations. The entire workflow was performed in less than one second per spectrum.