Chemometric strategy for untargeted lipidomics: biomarker detection and identification in stressed human placental cells

Abstract

A lipidomic study was developed in a human placental choriocarcinoma cell line (JEG-3) exposed to tributyltin (TBT) and to a mixture of perfluorinated chemicals (PFCs). The method was based on the application of multivariate curve resolution-alternating least squares (MCR-ALS) to data sets obtained by ultra-high performance liquid chromatography coupled to time-of-flight mass spectrometry (UHPLC-TOF-MS) using an untargeted approach. Lipids from exposed JEG-3 cells were solid-liquid extracted and analyzed by UHPLC-TOF-MS in full scan mode, together with control samples. Raw UHPLC-TOF-MS data of the different cell samples were subdivided into 20 distinct chromatographic windows and each window was further organized in a column-wise augmented data matrix, where data from every sample was in an individual data matrix. Then, the 20 new augmented data matrices were modeled by MCR-ALS. A total number of 86 components were resolved and a statistical comparative study of their elution profiles showed distinct responses for the lipids of exposed versus control cells, evidencing a lipidome disruption attributed to the presence of the xenobiotics. Results from one-way ANOVA followed by a multiple comparisons test and from discriminant partial least squares (PLS-DA) analysis were compared as usual strategies for the determination of potential biomarkers. Identification of 24 out of the 33 proposed biomarkers contributed to the better understanding of the effects of PFCs and TBT in the lipidome of human placental cells. Overall, this study proposes an innovative untargeted LC-MS MCR-ALS approach valid for -omic sciences such as lipidomics.

Keywords: Biomarkers; Liquid chromatography–mass spectrometry; Multivariate curve resolution-alternating least squares; Obesogens; Partial least squares-discriminant analysis; Untargeted lipidomics.