Solid-phase microextraction as sample preparation method for metabolomics

Abstract

The main objective of the emerging field of metabolomics is the analysis of all small molecule metabolites present in a particular living system in order to provide better understanding of dynamic processes occurring in living systems. This type of studies is of interest in various fields including systems biology, medicine and drug discovery. The main requirements for sample preparation methods used in global metabolomic studies are lack of selectivity, incorporation of a metabolism quenching step and good reproducibility. The efficiency of metabolism quenching and stability of analytes in selected biofluid or tissue dictate how accurately the analytical results represent true metabolome composition at the time of sampling. However, complete quenching of metabolism is not easily accomplished, so sample preparation can significantly affect metabolome's composition and the quality of acquired metabolomics data. In this research, the feasibility of the use of solid-phase microextraction (SPME) in direct extraction mode for global metabolomic studies of biological fluids based on liquid chromatography-mass spectrometry (LC-MS) was investigated for the first time. Initial research presented in this thesis focused on resolving several outstanding issues regarding the use of SPME for the analysis of biological fluids. SPME was not simultaneously capable to provide high-sample throughput and high degree of automation when coupled to LC-MS. This was successfully addressed through the development and evaluation of a new robotic station based on a 96-well plate format and an array of 96 SPME fibres. The parallel format of extraction and desorption allowed increased sample throughput of >1000 samples/day which represents the highest throughput of any SPME technique to date. This exceeds sample throughput requirements for a typical metabolomics study whereby ~100 samples/day are processed. SPME can also be used for direct in vivo sampling of flowing blood of an animal without the need to isolate a defined sample volume. This format of SPME is particularly attractive for metabolomic studies as it decreases the overall number of steps and also eliminates the need for metabolism quenching step because only small molecular weight species are extracted by the device, whereas large biological macromolecules such as proteins are not extracted by the coating. In current work, in vivo SPME sampling was successfully applied for sampling of mice for the first time. The proposed sampling procedure was fully validated against traditional terminal and serial sampling approaches for a pharmacokinetic study of carbamazepine and its metabolite. Excellent agreement of pharmacokinetic parameters such as systemic clearance, steady-state volume of distribution and terminal half-life was found for all three methods, with no statistically significant differences (p>0.05). The performance of new prototype commercial SPME devices based on hypodermic needle was also evaluated within the context of the study. The availability of such single-use devices with excellent inter-fibre reproducibility (<10% RSD) presents an important step forward in order to gain wider acceptance of in vivo SPME sampling. Finally, existing SPME coatings were not suitable for the simultaneous direct extraction of both hydrophilic and hydrophobic species, which is one of the requirements for a successful global metabolomics study. To address this issue, a systematic study of 40 types of commercially available sorbents was carried out using a metabolite standard test mixture spanning a wide molecular weight (80-777 Da) and polarity range (log P range of -5 to 7.4). The best performance for balanced extraction of species of varying polarity was achieved by (i) mixed-mode coating containing octadecyl or octyl group and benzenesulfonic acid ion exchange group, (ii) polar-enhanced polystyrene-divinylbenzene polymeric coatings and (iii) phenylboronic acid coatings. The second aspect of the research focused on the evaluation of SPME for a global metabolomics study of human plasma using two complementary LC-MS methods developed on benchtop Orbitrap MS system: reverse-phase method using pentafluorophenyl LC stationary phase and HILIC method using underivatized silica stationary phase. The parameters influencing overall method sensitivity such as voltages, mass ranges and ion inject times into C-trap were optimized to ensure best instrument performance for global metabolomic studies. Orbitrap system provided a powerful platform for metabolomics because of its high resolution and mass accuracy, thus helping to distinguish between metabolites with same nominal mass. The acquisition speed of the instrument at the highest resolution setting was insufficient for use with ultrahigh performance liquid chromatography (UHPLC), so all methods were developed using conventional LC. However, overall metabolite coverage achieved in current study compared well or even exceeded metabolite coverage reported in literature on different LC-MS or UHPLC-MS platforms including time-of-flight, quadrupole time-of-flight and hybrid Orbitrap instruments. The performance of SPME was fully compared versus traditional methods for global metabolomics (plasma protein precipitation and ultrafiltration). The main findings of this systematic study show that SPME provides improved coverage of hydrophobic metabolites versus ultrafiltration and reduces ionization suppression effects observed with both plasma protein precipitation and ultrafiltration methods. Using SPME, <5% and <20% of peaks showed significant matrix effects in reverse phase and HILIC methods, respectively and the observed effects were mostly correlated to elution within retention time window of anticoagulant for the majority of metabolites showing this effect. This improves overall quality of collected metabolomics data and can also improve metabolite coverage. For example, the highest number of metabolite features (3320 features) was observed using SPME in combination with negative ESI reverse-phase LC method, while in positive ESI mode plasma protein precipitation with methanol/ethanol mixture provided the most comprehensive metabolite coverage (3245 features versus 1821 features observed for SPME). Method precision of SPME method was excellent as evaluated using median RSD (11-18% RSD) of all metabolites detected. A proof-of-concept in vivo SPME study was also performed on mice to study the effects of carbamazepine administration and shows that SPME can be used as successful sample preparation method for global metabolomic studies in combination with unsupervised statistical data analysis techniques. This study highlights important advantages of in vivo sampling approaches including the ability to capture short-lived and/or unstable metabolites, to achieve truer representation of the metabolome at the time of sampling than achievable by blood withdrawal methods and the ability to use smaller animal cohorts while obtaining highly-relevant data sets. The experimental results provide new and useful insight into the effects of different sample preparation methods on the collected metabolomics data, and establish both in vitro and in vivo SPME as a new tool for global LC-MS metabolomics analysis for the first time.