Enhanced Metabolome Coverage and Evaluation of Matrix Effects by the Use of Experimental-Condition-Matched 13C-Labeled Biological Samples in Isotope-Assisted LC-HRMS Metabolomics

Abstract

Stable isotope-assisted approaches can improve untargeted liquid chromatography-high resolution mass spectrometry (LC-HRMS) metabolomics studies. Here, we demonstrate at the example of chemically stressed wheat that metabolome-wide internal standardization by globally ¹³C-labeled metabolite extract (GLMe-IS) of experimental-condition-matched biological samples can help to improve the detection of treatment-relevant metabolites and can aid in the post-acquisition assessment of putative matrix effects in samples obtained upon different treatments. For this, native extracts of toxin- and mock-treated (control) wheat ears were standardized by the addition of uniformly ¹³C-labeled wheat ear extracts that were cultivated under similar experimental conditions (toxin-treatment and control) and measured with LC-HRMS. The results show that 996 wheat-derived metabolites were detected with the non-condition-matched ¹³C-labeled metabolite extract, while another 68 were only covered by the experimental-condition-matched GLMe-IS. Additional testing is performed with the assumption that GLMe-IS enables compensation for matrix effects. Although on average no severe matrix differences between both experimental conditions were found, individual metabolites may be affected as is demonstrated by wrong decisions with respect to the classification of significantly altered metabolites. When GLMe-IS was applied to compensate for matrix effects, 272 metabolites showed significantly altered levels between treated and control samples, 42 of which would not have been classified as such without GLMe-IS.

Keywords: GLMe-IS; abiotic stress of wheat; deoxynivalenol; internal standard; matrix effects; untargeted metabolomics.

Figure 1

(a) Experimental setup to generate native and ¹³C-labeled wheat ear sample extracts: deoxynivalenol (DON) was used for toxin treatment, while water served as the control. (b) Isotope pattern in the mass spectrum for three exemplary metabolites: the labeled form is present for all metabolites as it derives from the globally ¹³C-labeled metabolite extract (GLMe-IS) pool, while the presence of native forms may depend on the sample type (i.e., toxin and control).

Figure 2

(a) Feature plot of all detected native metabolic features and (b) qualitative distribution of native metabolites in control- and toxin-treated wheat ear extracts: toxin only (red), control only (green), or common to both sample types (grey). Only those native features that have a labeled analog in GLMe-IS could be detected by MetExtraxt II.

Figure 3

(a) Principal component analysis (PCA) scores plot considering all commonly detected ¹²C native metabolic features for control and toxin and (b) a volcano plot for a toxin and control comparison of native metabolic features.

Figure 4

(a) Correlation of abundance ratios between control and toxin constituents with and without normalization to ¹³C-metabolites of GLMe-IS (i.e., absolute and normalized abundances) and (b) a Venn diagram showing the distribution of significantly differing metabolites of the control and toxin comparison between two groups of the comparative significance test that is obtained with the application of native absolute and normalized abundance ratios.

Figure 5

(a) PCA and (b) volcano plot to demonstrate that the native matrices have similar effects on absolute abundances of labeled metabolites in both sample types.