The Development and Application of Non- Targeted Mass Spectrometry Based Methods to Tea (Camellia Sinensis L.)

Abstract

Non-targeted analysis is an analytical approach which attempts to measure changes in the abundance of as many metabolites as possible (both known and unknown) in a single sample. Methodologies to perform this demanding task require regular improvement to keep pace with on-going technological advances in chromatography and mass spectrometry. Thus a major goal of this thesis was to develop improved, non-targeted methods using modern LChigh resolution mass spectrometry (HRMS) instrumentation and then explore their utility in investigating complex metabolite profiles of tea extracts. A reversed-phase (RP) ion-trap MS method was initially developed to detect and classify proanthocyanidins (PAs) in oolong and black tea extracts, resulting in several new PAs being identified. Two new LC-HRMS methods were then developed; the first utilising hydrophilic interaction liquid chromatography to resolve mostly polar primary metabolites such as amino acids and nucleosides; the second using RP and ultra-high pressure liquid chromatography to maximise the chromatographic resolution of mostly semi non-polar secondary metabolites such as flavonoids. Both of these methods, when combined with multivariate statistical tools, could successfully differentiate not only tea type (green, oolong and black tea), but also detect metabolites distinguishing the country of origin of the teas. These two LC-HRMS methods were applied to monitor biochemical changes occurring during oolong tea manufacturing, revealing significant changes in volatile precursors, phenolics and nucleotide/nucleoside composition at the high temperature de-greening step, along with significant changes in plant hormones and free amino acid levels during the fermentation process. These samples were also investigated using a direct analysis in real time MS (DART-MS) technique which also revealed a unique acid/base cluster of geranic acid and caffeine as a key feature of changes in volatile composition during the fermentation stages of manufacturing. This project has developed complementary LC-HRMS non-targeted methods and demonstrated their value for profiling a wide array of metabolites, and when coupled with univariate and multivariate statistical techniques, shown their benefit for detecting both known and unknown compounds for further investigation. These methods could be applied to a range of sample types, such as foods or biological fluids, to further understand the complex biochemistry.