Comprehensive MS/MS profiling by UHPLC-ESI-QTOF-MS/MS using SWATH data-independent acquisition for the study of platelet lipidomes in coronary artery disease

Highlights: • Comprehensive MS and MS/MS non-targeted lipidomic profiling by UHPLC-ESI-QTOF-MS. • Workflow for data-independent acquisition of lipidomics data with SWATH. • SWATH parameter optimization. • Parallel semi-targeted analysis of oxidized phospholipids. • Clinical application to patient platelets. -- Abstract: A non-targeted lipidomics workflow based on C8 core-shell particle ultra high-performance liquid chromatography (UHPLC) hyphenated to ESI-QTOF-MS in data-independent acquisition (DIA) mode with sequential window acquisition of all theoretical fragment ion spectra (SWATH) was developed and applied to differential platelet lipidomics profiling of cardiovascular disease patients (stable angina pectoris (n = 10), ST-elevated myocardial infarction (n = 13)) against healthy controls (n = 10). DIA with SWATH generates comprehensive MS and MS/MS data throughout the entire chromatograms and all study samples. Hence, chromatograms can be extracted based on precursors or fragments which provided some benefits in terms of assay specificity in some cases. SWATH acquisition offers flexible experimental design with variable Q1 isolation windows. Liquid chromatography as well as SWATH settings were optimized to cover the lipidome of human platelets. The flexibility of the SWATH experiment design was utilized to implement target SWATH windows with narrow 5 Da Q1 precursor ion selection width (multiple reaction monitoring (MRM)-like SWATH windows) for the detection of low abundant oxidized phospholipids. Data processing was performed with MS-DIAL, and its feasibilities and caveats are discussed by illustrative examples. Thereby, identification of lipids is still a bottleneck in non-targeted lipidomics workflow. MS-DIAL, however, offers automatic identification via spectral matching using an in silico library. In total 1971 molecular features were detected cross the samples of which 611 were identified (total score >70%). The quality of the acquired data was validated with embedded quality control samples (n = 11). 80.3% of all features detected in the QC samples showed a coefficient of variation of below 30%. Multivariate statistics were used to visualize differences in the lipidome of distinct sample groups at a false discovery rate of 5%.