New methods for mixture analysis by liquid state NMR

Nuclear magnetic resonance (NMR) spectroscopy is the best tool we have for determining the chemical structures of unknown compounds. Hundreds of thousands of structures are identified by NMR each year, but almost all as dilute solutions of pure compounds: NMR struggles to analyse mixtures because it is very difficult, often impossible, to tell which signals in the spectrum come from which species. At the same time, mixtures lie at the heart of chemistry and biochemistry, for example in the search for drugs in plant extracts, in understanding chemical reactions - or simply in finding out what makes one wine taste better than another. One big challenge in mixture analysis is the field of metabolomics, which studies the effects of genetic makeup, disease, drugs etc. on the occurrence of metabolites in biofluids. The ability to disentangle structural information from different molecules in complex mixtures is the key to exploiting this and many other fields. Isolation, purification and concentration of individual components are tedious, expensive and time-consuming; what is needed are ways of determining the structures of individual species without separating a mixture into its individual components. By combining NMR with separation methods, spectral information can be obtained directly from mixtures. The most common example of this is LC-NMR, in which an NMR spectrometer is used as the detector for liquid chromatography (LC). A sample of a mixture flows through a column containing a fine powder, and thence into an NMR instrument. Different species stick to the column material to different extents so they emerge at different times, and the NMR spectrum of each species in turn is measured. Such methods are powerful, but they are expensive in instrumentation, in materials, and in expertise, and they struggle to deal with small amounts of material or very complex mixtures. The most potent method in current use for NMR studies of intact mixtures is diffusion-ordered spectroscopy (DOSY), which separate the signals of different species according to how rapidly they diffuse. DOSY has proven its worth in many areas of analysis, including food chemistry, biofluids, binding studies etc., but its use is limited because it only works well where the NMR signals of different species do not overlap. I propose to try to enhance significantly the power of NMR in studying mixtures by developing a new technique and by using powerful mathematical methods so far largely unexploited for NMR data. I will apply techniques currently used to measure blood flow in magnetic resonance imaging to measure directly how the rates at which different molecules flow through a chromatography column are affected by their interactions with the contents of the column. Because this method measures mean flow velocity rather than retention time, the mixture can be pumped continuously through the column in a closed cycle so, in contrast to LC-NMR, we can make repeated measurements on the same sample for as long as it takes to get a clear result with good signal-to-noise ratio. Because the motion that we measure here is coherent, as opposed to the incoherent diffusion measured in DOSY, analysis of the experimental results is straightforward and avoids the difficulties that DOSY encounters when signals overlap. The basic DOSY experiment gives data which vary in two different ways; in contrast experiments that give results which vary in sympathy as a function of three different variables - trilinear data - have special advantages when it comes to analysis. I will investigate using DOSY to allow trilinear data analysis of the NMR spectra of complex mixtures (e.g. those found in metabolomics). The prize here is to be able to extract the complete NMR spectrum of a compound of interest from a background potentially containing many hundreds of such spectra, exploiting the added resolution gained by combining diffusion and chromatographic information