Development of NMR tools and experiments for the structure elucidation of carbohydrates

The market for carbohydrates used in pharmaceuticals and personal care products has been steadily increasing over the past few decades. Notable examples of such products - which contain carbohydrates as active ingredients - include: anti-coagulants, anti-thrombosis drugs, anti-metastasis drugs, eczema cream, medication for hay fever relief, and skin care products. Nevertheless, as most carbohydrates are structurally complex, and/or exist naturally as a mixture, their structural analysis can be extremely difficult. Whilst Nuclear Magnetic Resonance (NMR) spectroscopy can be used to determine carbohydrate structures, the process is challenging, error prone, and time consuming. This thesis presents novel NMR experiments, as well as semi-automated algorithms for analysis of NMR spectra of carbohydrates, all of which contribute to alleviating the difficulties encountered in carbohydrate studies. To demonstrate both the power and limitations of NMR spectroscopy, the steps taken to elucidate the structure of a complex polysaccharide from marine algae (supplied by GlycoMar) were explored. This structure was solved by manual analysis of homo- and hetero-correlated 800 MHz NMR spectra. Moreover, in collaboration with the Dorfmueller group, University of Dundee, NMR spectroscopy was used to determine the structures of the acceptors and products involved in the second and third steps of the Biosynthesis pathway of Group A Carbohydrate (GAC) – which comprises between 40 to 60% of the bacterial cell wall of Group A Streptococcus. Using selected NMR experiments in conjunction with algorithms written in Python, the extraction and assignment of 1H and 13C chemical shifts, as well as linkage determination of oligo- and polysaccharides, has been semi-automated. The experiments involved in these processes include hetero-correlated 2D techniques such as 1H,13C HSQC, HSQC-TOCSY, HSQC-NOESY, and HMBC. Whilst further testing and optimisation is required, these algorithms provide a good base for further development. \newline Additionally, a common issue in the NMR analysis of carbohydrates is a severe overlap of 2D cross peaks, resulting from a very limited range of chemical shifts of complex carbohydrates; this is especially the case in 1H-1H TOCSY (or 1H-1H NOESY) spectra. A novel homonuclear 2D 1H-1H TOCSY NMR experiment was developed which uses a perfect echo pulse sequence and sophisticated phase cycling to increase signal separation in the F1 dimension up to fourfold; as a result, this significantly improves the spectral resolution in this dimension. The accompanying drop in sensitivity (up to fourfold) is often not critical on spectrometers using cryoprobes. Processing the data using linear prediction in MNova software further improves the achievable resolution. Increasing resolution in F1 is necessary should the automated assignment be attempted on homonuclear spectra that are inherently more crowded than their heteronuclear counterparts. The conformation of carbohydrates underpins their biological functions. Amongst NMR parameters - which can be interpreted in terms of conformation - are scalar and residual dipolar couplings; in particular, the latter are notoriously difficult to measure. A novel "pure-shift" NMR method, named SHARPER-J (Sensitive, Homogeneous And Resolved PEaks in Real time - J-coupling), was developed in order to allow for the accurate measurement of scalar and residual dipolar 1H-1H couplings of carbohydrates. This method removes all but a mutual splitting between the two involved spins, and uses upscaling of splittings - which is essential for measuring small unresolved couplings - whilst decreasing the spectral linewidth close to the values dictated solely by spin-spin relaxation, thus removing the contribution of the magnetic field inhomogeneity