Abstract

Metabolomics is the comprehensive analytical study of the metabolome, which is composed of all of the low molecular weight (≤1500 Da) species in a biological system. Chromatographic separation of samples is implemented before detection by mass spectrometry to increase metabolome coverage. To ensure coverage of as many metabolites as possible, from hydrophobic to hydrophilic, both reversed phase liquid chromatography (RPLC) and hydrophilic interaction liquid chromatography (HILIC) are used. A variety of stationary phases are available for HILIC and can be grouped into three categories: neutral, charged, and zwitterionic. Each stationary phase varies by relative hydrophobicity, hydrogen bonding ability, and electrostatic interaction capabilities. There is currently no consensus in the literature on which of the available HILIC stationary phases provides the best results for global metabolomics applications. The first objective of this study was to compare a sulfobetaine zwitterionic ZIC-HILIC stationary phase to a charged underivatized silica HILIC stationary phase, specifically the Ascentis Si Express. The effects of salt concentration in the mobile phase and the mobile phase gradient were both investigated. The quality of peak shapes, analyte retention time, peak separation, and metabolite coverage were used to compare the results from each stationary phase. The methods were evaluated using a mixture of 37 standards covering a range of logP values (-10 to 3.73), molecular weights (59 to 776 Da), and metabolite classes. Good quality results for 7 and 14 of the metabolite standards were achieved using the silica and ZIC-HILIC columns, respectively. 14 and 2 of the standards could not be detected at all on the two phases respectively. Phospholipids, separated by HILIC based on the polarity of their head group, regardless of fatty acyl chain length or degree of saturation, can cause ion suppression. Lipid standards were analyzed to determine their retention times for both HILIC methods, aiding the interpretation of plasma analysis results. The developed methods were further compared using a complex biological sample: methanol-precipitated plasma. Metabolome coverage was greater with the silica column (3520 and 2734 compounds in positive and negative ESI respectively) compared to the ZIC-HILIC column (1612 and 1643 respectively), however peak quality and retention time reproducibility was greater with the ZIC-HILIC column. Thus, it is possible that automated data processing may overestimate the number of metabolite peaks in silica HILIC due to wider peaks and more variability in retention times. Finally, the addition of 10 mM ammonium phosphate to samples was evaluated and determined to improve the peak shape quality for standards in solvent, however no similar improvement was observed for plasma samples. The second objective of this study was to develop a dispersive solid-phase microextraction (D-SPME) protocol for global metabolomics of human plasma. Solid phase microextraction is a non-exhaustive, equilibrium-based extraction technique governed by the partitioning of analytes between the sample matrix and sorbent material. Advantages of the technique include decreased ionization suppression, decreased solvent consumption, the capability to measure free metabolite concentrations, and the large variety of sorbent materials available. Each sorbent material comes with different extraction efficiencies and selectivities. To evaluate the use of carbon nanopearls (CNPs) for D-SPME, the effect of extraction time, sorbent volume, desorption solvent, desorption solvent volume, extraction temperature and desorption temperature was evaluated in detail. Extraction time experiments indicated that short extraction times, 2 minutes in the case of standards in buffer, can be used since equilibrium appears to have been reached. The evaluation of different desorption solvents is important because the choice of desorption solvent can influence not only which compounds are detected but also the concentration of these compounds in the desorbed sample. Acetonitrile was determined to provide the greatest desorption efficiency. The extraction of metabolite standards indicated that those with greater hydrophobicity, for example diosmin and diosmetin, and those with iodine atoms in their structure, for example thyroxine and triiodothyronine, have larger distribution coefficients however further investigation of the selectivity of CNPs is required. In conclusion, both chromatographic separation and sample preparation play a role in improving metabolome coverage. HILIC remains a promising tool for the separation of the polar metabolome prior to MS although further understanding of retention mechanisms is required, and SPME is a promising tool for improving the detection of novel, possibly low abundance metabolites not detected using less selective methods such as solvent precipitation.