Electrospray ionization and heterogeneous matrix effects in liquid chromatography/mass spectrometry based meta-metabolomics: A biomarker or a suppressed ion?

Rationale: Correct biomarker determination in metabolomics is crucial for unbiased conclusions and reliable applications. However, this determination is subject to several drifts, e.g. matrix effects and ion suppression in Liquid Chromatography/Mass Spectrometry (LC/MS)-based approaches. This phenomenon provokes critical issues for biomarker determination, particularly during comparative studies dealing with samples exhibiting heterogeneous complexities. Methods: Occurrence of the issue was coincidentally noticed when studying the environmental impact of a complex bioinsecticide: Bacillus thuringiensis israelensis. The studied samples comprised insecticide-spiked sediments and untreated control sediments. QuEChERS extractions followed by LC/ESI-Q/ToF analyses were performed on sediments after 15 days of incubation. Meta-metabolomes containing pesticide xenometabolites and sediment endometabolites were analyzed in depth using XCMS-based computational data preprocessing. Multivariate statistical analyses (PCA, OPLS-DA) and raw data crosschecks were performed to search for environmental biomarkers. Results: Multivariate analyses and raw data crosschecks led to the selection of nine metabolites as biomarker candidates. However, when exploring the mass spectra, co-elutions were noticed between seven of these metabolites and multi-charged macromolecules originating from the pesticide. Provoked false positives were thus suspected due to a potential ion suppression exclusively occurring in the spiked samples. A dilution-based approach was then applied. It confirmed five metabolites as suppressed ions. Conclusions: Ion suppression should be considered as a critical issue for biomarker determination when comparing heterogeneous metabolic profiles. Raw chromatograms and mass spectra crosschecks are mandatory to reveal potential ion suppressions in such cases. Dilution is a suitable approach to filter reliable biomarker candidates before their identification and absolute quantification.