Metabolomics is a rapidly growing research field. It aims for quantification of all the metabolites in a biological sample such as plasma, saliva, cerebrospinal fluid or cells. Because the metabolite levels in a biological sample are the end result of the regulatory processes in cells, metabolomics is a very powerful approach for characterisation of phenotypes. Metabolomics has been used to find disease biomarkers, investigate influences of heavy metals on the metabolism and to elucidate gene function. However, analysis of the complete metabolome puts high demands on the methods used. For instance, the methods should be unbiased to accurately depict the in vivo status in the cell. Furthermore, the methods must have very high resolution and sensitivity to allow detection of all metabolites. To approach these high goals, the protocols used in metabolomics need to be thoroughly optimised.
The amount of information contained in the metabolome is immense. Consequently, the data set collected from a metabolomics study is very large. To extract the relevant information from such large sets of data, efficient methods are needed both to plan experiments and to convert the data to useful information. For this task, chemometrics is an ideal approach as it allows efficient experimental planning and multivariate data analysis. The experimental planning is sometimes referred to as statistical experimental design or design of experiments. It aims to systematically and simultaneously vary experimental factors in a structured manner. Hence, fewer experiments are generally needed to efficiently map how the system is affected by prevailing factors. The multivariate data analysis employs powerful projection and regression methods to find patterns in data, create system models and classify data. Hence, chemometrics provides a framework for efficient experimental design and an efficient approach for information retrieval.
In this thesis two thorough developments of metabolomics protocols and three metabolomics investigations, relevant to metabolic regulation in diabetes patients and insulin-producing cells, are presented. The design of experiments approach and multivariate data analysis were applied. The developed protocols were optimised and validated for the analysis of human blood plasma and adherent cell cultures, respectively, and included optimisation from the sample preparation to the analysis with gas chromatography/mass spectrometry. The first of the metabolomics studies aimed to find biomarkers reflecting metabolic regulation during an oral glucose tolerance test in humans to aid in the diagnosis of diabetes. The second study was performed on clonal β-cells and aimed to find metabolic regulation coupled to the amplifying pathway of insulin secretion. The last study aimed to identify metabolic dysregulation in clonal β-cells growing under lipotoxic and glucotoxic conditions, respectively. In all studies, metabolomics extended and deepened the understanding of metabolic regulation in cells and patients. As such, metabolomics will help to find explanations for metabolic diseases such as diabetes