Untargeted LC-MS/MS metabolomics reveals farming practice- and cultivar-driven modulations of pea (Pisum sativum L.) seed metabolome across multiple biosynthetic pathways

This work presents a comprehensive biochemical landscape of pea seeds captured by untargeted LC-MS-based metabolomics of ten pea cultivars grown at three Danish field sites following different agricultural practices. More than 1200 metabolite features were detected in methanolic extracts of pea seed flours. Of these, nearly 300 features were identified using mass spectral libraries and advanced computational tools. Approximately 40 metabolites were found to be associated with location effect, independent of cultivar type. Organically grown pea samples showed lower levels of the main pea triterpene glycoside soyasaponin I and higher levels of nitrogen-abundant amino acids, indicating increased nitrogen availability in soil. More than 100 metabolites were associated with the location-independent cultivar effect. Akooma and Greenway cultivars showed the most distinct metabolome with greater levels of polyunsaturated fatty acids and lipid oxidation products known to give ‘beany’ off-flavors. The commonly cultivated pea variety, Ingrid, was devoid of compounds derived from the phenylpropanoid pathway including hydroxycinnamic acid amides such as caffeoyl, feruloyl, and coumaroyl aspartates that were present in all other cultivars. Three chloroauxin metabolites, reported here for the first time, were identified through molecular networking within GNPS platform and propagation of annotation from a computationally predicted indole-3-acetic acid catabolite. Overall, the results indicate biochemical adaptation of pea plants to location or agricultural practices as reflected in their seed metabolome.