Isotopologue pattern based data mining for selenium species from HILIC–ESI–Orbitrap–MS-derived spectra

Abstract

AbstractAutomated and specific picking of selenium-containing molecular entities has not been an obvious option for software tools associated with electrospray high-resolution mass spectrometry (MS). In our study, a comprehensive pattern matching approach based on intra-isotopologue distance and isotopologue ratio data was critically evaluated in terms of reproducibility and selenium isotope selection on three samples, including selenized Torula yeast and the selenium hyperaccumulator plant Cardamine violifolia. Hydrophilic interaction liquid chromatography was applied to provide a one-step separation for water soluble metabolites to put an end to the need for either orthogonal setups or poor retention on reversed phase chromatography. Assistance from inductively coupled plasma–MS was taken only for chromatographic verification purposes, and the involvement of absolute mass defect (MD) data in selenometabolite-specific screening was assessed by multivariate statistical tools. High focus was placed on screening efficiency and on the validation of discovered selenized molecules to avoid reporting of artefacts. From the >1000 molecular entries detected, selenium-containing molecules were picked up with a recovery rate of >88% and a false positive rate of <10%. Isotop(ologu)e pairs of 78Se–80Se and 80Se–82Se proved to be the most performant in the detection. On the basis of accurate mass information and hypothetical deamination processes, elemental composition could be proposed for 72 species out of the 75 selenium species encountered without taking into account selenocompound databases. Absolute MD data were used to significantly differentiate a potentially sample-specific subgroup of false positive molecular entities from non-selenized and selenized entities.