Untargeted metabolomics analysis with direct ESI MS for discrimination of Botrytis cinerea infection in grapes

Infection of grapes (Vitis vinifera) by Botrytis cinerea (grey mould) occurs frequently in vineyards exposed to wet and humid conditions and leads to detrimental impacts on yield and quality. Growth of B. cinerea in grapes causes the oxidisation of phenolic compounds resulting in loss of colour and formation of a suite of off-flavours and odours in wine made from infected fruit.

In this study, metabolites were extracted from sample homogenates using acetonitrile. Our data set comprised a total of 140 healthy and infected bunches of grapes representing different vintages, cultivars, regions, and maturity stages. Sample extracts were randomly analysed by direct injection into a Q trap mass spectrometer, including regular quality assurance samples, with data collected from 50-2000 m/z for 1 min. Molecular feature abundances were summed from 0.1 to 0.4 min and normalised prior to PCA for quality assurance. Samples were randomly assigned to a calibration and independent test data set, with feature reduction, a two-class model PLS-DA, cross-validation and permutation testing performed with the calibration data set. Prediction of sample class in the independent test samples demonstrated an overall predictive error of less than 5%. Feature importance was assessed using a combined VIP and selectivity ratio plot which demonstrated a high level of correlation with standard volcano plots. Annotation of important molecular features was undertaken using a highresolution Orbitrap MS detector, and LC/MS -QToF of selected samples from healthy and infected extracts. The rapid sample preparation, analysis and data workflow presented could be applied for analytical demands requiring discrimination of complex samples including environmental or biological
specimens.