Untargeted GC×GC-MS metabolic profiling of lettuce exposed to contaminants of emerging concern using wavelet transform-multivariate curve resolution followed by partial least squares-discriminant analysis

Metabolomics is systematic study of metabolic profiles, their composition, and changes in metabolome caused by genetical, environmental, nutritional, or other factors. Thus, it has important role in detecting potential biomarkers in various samples. Contaminants of emerging concern (CECs), including pharmaceuticals and personal care products, are increasingly being detected at low levels in irrigation waters and cause significant reproductive effects. Presence of CECs in irrigation waters lead to changes in plant’s metabolome in order to adapt to environmental stressors [1]. However, measurement of all metabolites is an analytical challenge due to the complexity of the sample matrices. In this regard, two-dimensional gas chromatography-mass spectrometry (GC×GC-MS) has emerged as a powerful separation technique to tackle the incomplete separation issue in complex samples matrices [2]. In the present contribution, an untargeted metabolomic study based on chemometrics was developed on control lettuce (Lactuca sativa L) samples and exposed samples to 11 CECs by irrigation. The aim of this study was identification of lettuce metabolites with significant profile alteration induced by CECs exposure. In this regard, raw GC×GC–TOFMS data from eight lettuce sample extracts (i.e., control and exposed samples) were column-wise augmented. However, due to the huge size of augmented matrix (~20 gigabytes (GB)) it was necessary to use a proper data compression approach to reduce data size without missing relevant information. For this purpose, wavelet decomposition and compression (level-2) was applied independently on every column (m/z) of the augmented matrix. Compressed augmented data matrix for eight samples was then analyzed by multivariate curve resolution-alternating least squares (MCR-ALS) using proper constraints [3]. The number of components in the data matrix was determined using singular value decomposition (SVD) which was 80 in this study. Also, orthogonal projection approach (OPA) was used to calculate initial spectral estimates to start ALS optimization. On this matter, fifty MCR-ALS components were unambiguously assigned to characteristic lettuce metabolites. Then, the peak areas of the MCR-ALS resolved elution profiles in every sample analyzed by GC×GC-TOFMS technique were arranged in a new data matrix that was then modeled by partial least squares-discriminant analysis (PLS-DA). The control and CECs exposed lettuce samples were discriminated by PLS-DA and the most relevant metabolites were estimated using Variable Importance in Projection (VIP) scores and Selectivity Ratio (SR) values. Finally, the metabolic pathways of the identified significant metabolites were found using Kyoto Encyclopedia of Genes and Genomes (KEGG) and they were interpreted from biochemistry point of view.