Abstract:
Throughout the last decades, substantial technological innovations in mass spectrometry (MS) fueled a continuous progress in analytical chemistry and enabled the accelerated exploration of various “omics” branches. Lipidomics, a subset of metabolomics, has recently attracted increasing attention, since lipids, with their diverse physiological functions as structural components of membranes, energy depots and signaling molecules, have been recognized as significant factors in disease onset and progression, e.g. for central health concerns like cardiovascular disease and cancer. Besides the improved instrument hardware, also analytical strategies and bioinformatics have advanced to yield reliable qualification and quantification, which aid in the ongoing decryption of the lipidome and its pathways and networks.
To obtain sufficient sensitivity, selectivity and analyte coverage, hybrid quadrupole time-of-flight (QTOF) mass spectrometers are often utilized, as their rapid acquisition rates allow to combine the recording of high resolution mass spectra with the hyphenation to ultra-high performance liquid chromatography (UHPLC). Along with the introduction of sequential window acquisition of all theoretical fragment ion mass spectra (SWATH), a powerful tool for true comprehensive analysis was made available.
In this thesis, the potential of a UHPLC-QTOF platform was exploited to achieve a maximum yield of extractable information from biological samples in the context of lipidomic workflows. Besides the comparison of different lipid extraction strategies for HeLa cells, also the development of a specialized sample preparation protocol for plasma steroids was conducted. Together with SWATH acquisition and a thorough optimization of MS parameters, absolute quantification of low picomolar levels of testosterone and estradiol was attained. An obstacle for the quantification of endogenous compounds is typically the lack of a true blank matrix, which was compensated by surrogate calibration via 13C3-labeled target analyte analogues. The established method was validated according to international guidelines and the accuracy and precision were additionally verified by the analysis of external quality control (QC) samples. Later, over 300 clinical samples were analyzed and the obtained results were utilized to monitor and interpret the influence of estradiol treatment on food intake in healthy men. The observed lowered protein consumption was shown to be independent from alterations in macronutrient ingestion induced by insulin administration. Furthermore, the merged targeted/untargeted study design enabled simultaneous screening of additional steroids and revealed a significant reduction in epitestosterone, dihydrotestosterone and hydroxyprogesterone levels after estradiol intake.
Moreover, in the framework of an exclusively untargeted lipidomic study, different normalization strategies, which are usually required to control for unwanted variation, were assessed. After compiling a QC-based workflow to effectively compare the performance of normalization methods, novel guidelines for selecting the best suitable strategy, while maintaining data integrity, were proposed. In addition, a script-based statistical and bioinformatical tool was provided as an open source solution to facilitate the implementation of these guidelines into the existing data processing workflows. Eventually, a contribution towards the necessary harmonization of data handling approaches in untargeted lipidomics was supplied for the scientific community.
In spite of the increasing efforts in untargeted analysis, the majority of studies is still only able to report results as relative foldchanges. Without absolute quantification, though, comparability to other studies or databases is limited and follow-up experiments have to be conducted to estimate reference or abnormal levels of potential biomarkers. To overcome this issue, a study with deuterated, lipid class-specific standards as class-wide surrogate calibrants was designed. Matrix-matched calibrants and QCs were incorporated into the analytical sequence of an untargeted plasma study and precision and accuracy for representative surrogate lipids was validated. Lipid species were separated with a reversed-phase UHPLC method and data was acquired using SWATH. Due to different ionization efficiencies and instrument responses between lipid species, which depend on carbon chain length, degree of saturation, matrix effects and solvent composition during elution, response factors had to be considered for class-wide extrapolation of absolute concentrations. It could be demonstrated that surrogate calibration resulted in more accurate quantification of lipid levels than one-point calibration. With post-acquisition re-calibration, which is describing the experimental determination of response factors after analysis and data processing for lipids of interest, a workflow, that is capable to estimate lipid levels in untargeted assays, was suggested.