Lipidome variation of industrial Saccharomyces cerevisiae strains analyzed by LC-QTOF/MS-based untargeted lipidomics.

Although various yeast strains used in the food industry have been characterized by multilayer analysis, knowledge of the variation of lipid profiles involved in fermentation characteristics and stress tolerance remains in its infancy. In this study, untargeted lipidomics was applied to 10 yeast strains, including laboratory, baker's, wine, and sake yeasts, which exhibit distinct fermentation phenotypes, to obtain a comprehensive overview of the yeast lipidome. The relative standard deviation (RSD) in the abundance of the 352 identified lipid molecular species was investigated to reveal the specific and common lipids. Lipids containing very long-chain fatty acids and hydroxy long-chain fatty acids showed relatively large RSD, whereas lipids containing acyl chains, which are commonly found in yeast, such as C16-C18, showed less RSD among the 10 strains. Furthermore, principal component analysis of lipid profiles showed similar trends among industrial yeast strains. As lipids are involved in yeast phenotypes, including stress tolerance and fermentation characteristics, correlation analysis was performed with lipid abundance and phenotypes. The results revealed that molecular species with a high RSD in abundance among the 10 strains were correlated with specific stress tolerance and fermentation phenotypes.

Data Processing of Product Ion Spectra: Quality Improvement by Averaging Multiple Similar Spectra of Small Molecules.

In metabolomics studies using high-resolution mass spectrometry (MS), a set of product ion spectra is comprehensively acquired from observed ions using the data-dependent acquisition (DDA) mode of various tandem MS. However, especially for low-intensity signals, it is sometimes difficult to distinguish artifact signals from true fragment ions derived from a precursor ion. Inadequate precision in the measured m/z value is also one of the bottlenecks to narrowing down the candidate compositional formula. In this study, we report that averaging multiple product ion spectra can improve m/z precision as well as the reliability of fragment ions that are observed in such spectra. A graph-based method was applied to cluster a set of similar spectra from multiple DDA data files resulting in creating an averaged product-ion spectrum. The error levels for the m/z values declined following the central limit theorem, which allowed us to reduce the number of candidate compositional formulas. The improved reliability and precision of the averaged spectra will contribute to a more efficient annotation of product ion spectral data.