Mass spectrometry-based metabolomics and chemometric analysis reveal host tissue-level classification of Aspergillus flavus isolates

Abstract Mass spectrometry-based metabolomics has emerged as a powerful approach for investigating fungal metabolic diversity and identifying biomarkers related to pathogenicity and variability among isolates. Here, we characterized and distinguished Aspergillus flavus strains isolated from the leaves and roots of soybean plants via ultrahigh-performance liquid chromatography-high-resolution mass spectrometry (UHPLC-HRMS) integrated with chemometric analyses. A total of 42 metabolites, including peptides, organic acids, alkaloids, trichothecenes, nitro compounds, terpenes, phospholipids, and phenolic compounds with antimicrobial, phytotoxic, and plant growth-promoting activities, were annotated in the methanolic extracts of A. flavus isolates. Hierarchical clustering analysis (HCA) showed that UHPLC-HRMS was able to distinguish A. flavus isolates from soybean leaves and roots in both negative and positive ion modes. The partial least squares-discriminant analysis (PLS-DA) method revealed the classification of A. flavus methanolic extract samples in terms of host plant tissue by the formation of two more distant clusters, with adjustment quality and satisfactory prediction, generating values of R 2 = 0.96 and Q 2 = 0.99 for negative ion mode and R 2 = 0.93 and Q 2 = 0.98 for positive ion mode in cross-validation. Citric acid and 13-hydroxy-5'- O -methylmeledonal were biomarkers of A. flavus isolates from soybean leaves, and cyclopiazonic acid and galactitol were biomarkers of A. flavus isolates from roots. This study reports a novel application of UHPLC‒HRMS and chemometrics in the classification of plant pathogenic fungi at the host tissue level.