Untargeted detection of the carbonyl metabolome by chemical derivatization and liquid chromatography-tandem mass spectrometry in precursor ion scan mode: Elucidation of COVID-19 severity biomarkers.

ABSTRACT

Metabolomics (both targeted and untargeted) has become the gold standard in biomarker discovery. Whereas targeted approaches only provide information for the selected markers, thus hampering the determination of out-of-the-box markers, the common bottleneck of untargeted metabolomics is the identification of detected biomarkers. In this study, we developed a strategy based on derivatization and LC-MS/MS detection in a precursor ion scan for the untargeted determination of a specific part of the metabolome (carbonyl-containing metabolites). The usefulness of this guided metabolomics approach has been demonstrated by elucidating carbonyl-containing biomarkers of COVID-19 severity. First, the LC-MS/MS behavior of 63 model compounds after O-benzylhydroxylamine derivatization was studied. A precursor ion scan of m/z 91 was selected as a suitable approach for the untargeted detection of carbonyl-containing metabolites. The method was able to detect ≈300 potential carbonyl-containing molecules in plasma, including mono-/di-/tricarbonylic compounds with satisfactory intra-day and inter-day repeatability and RSDs commonly <15%. Additionally, the semiquantitative nature of the precursor ion scan method was confirmed by comparison with a fully validated targeted method. The application of the guided metabolomics method to COVID-19 plasma samples revealed the presence of four potential COVID-19 severity biomarkers. Based on their LC-MS/MS behavior, these biomarkers were elucidated as 2-hydroxybutyrate, 2,3-dihydroxybutyrate, 2-oxobutyrate and 2-hydroxy-3-methylbutyrate. Their structures were confirmed by comparison with reference materials. The alterations of these biomarkers with COVID-19 severity were confirmed by a target analysis of a larger set of samples. Our results confirm that guided metabolomics is an alternative approach for the untargeted detection of selected families of metabolites; this approach can accelerate their elucidation and provide new perspectives for the establishment of health/disease biomarkers.