Liquid chromatography–quadrupole time of flight tandem mass spectrometry–based targeted metabolomic study for varietal discrimination of grapes according to plant sterols content

• Varietal discrimination of grapes according to phytosterols content.
• LC-QTOF-MS system is a useful tool to global metabolite profiling.
• Data analysis is a key step in “omics” research.
• Supervised pattern recognition method (PLS-DA) for class prediction analysis.
• MS/MS analysis provides high level of confidence for metabolite identification.

Grapevine and derived products are rich in a wide range of compounds and its quality mainly depends on its metabolites, as a result of viticulture practices. Plant sterols, also called phytosterols (PS), are secondary metabolites regarded as bioactive substance present in grape berries and other plant-based food. The present study deals with a metabolomic approach focusing on phytosterols family in six varieties of Rioja grapes (Cabernet Sauvignon, Tempranillo, Graciano, Garnacha, White Garnacha and Viura), in order to find significant differences among them. Liquid chromatography- mass spectrometry with a quadrupole-time of flight mass analyzer (LC-QTOF) was used to find as many metabolites as possible in the different grape berry fractions, and using statistics to help finding significant clustering of the metabolic profile of pulp, peel and seeds in relation to the variety. The best chromatographic and detection conditions were achieved by gas phase ionization via atmospheric pressure chemical ionization (APCI) in positive mode. Furthermore, analysis with electrospray (ESI) is also needed for phytosterol derivatives confirmation.Putative compounds of interest in the analyzed samples were found by an automated compound extraction algorithm (Molecular Feature Extraction, MFE) and an initial differential expression from the data was created with the aid of commercial software. Once the data were collected, the results were filtered, aligned and normalized, and evaluating applying one-way analysis of variance (ANOVA) with a 95% significance level. For sample class prediction, partial least square-discriminant analysis (PLS-DA) is used as a supervised pattern recognition method and excellent separation among the grape varieties is shown. An overall accuracy of 93.3% (pulp samples), 100.0% (peel) or 96.7% (seeds) in discriminating between grape varieties was achieved when comparing the different fractions. In general, 7 PS derivatives were identified with ID scores higher than 84%.