Une méthode permettant de caractériser l’ensemble du métabolome secondaire de moisissures a été appliquée à la caractérisation des métabolomes de Penicillium nordicum, Penicillium verrucosum et Fusarium graminearum. Le substrat représentant l’unique source de carbone et d’azote des moisissures, chacun des champignons ont été mis en culture sur trois types de grains de blé: (i) grains naturels, (ii) grains marqués à 97% de 13C, et (iii) grains marqués à 53% 13C et 97% de 15N. Les extraits ont été analysés par HRMS. Les métabolites secondaires ont été spécifiquement détectés et leurs formules brutes ont été caractérisées. La caractérisation de nouveaux métabolites secondaires a ensuite été assistées par la génération de réseaux moléculaires de similarités MS/MS. L’étude de P. verrucosum et P. nordicum a permis de détecter 98 et 92 métabolites secondaires respectivement. Parmi eux, 80% étaient inconnus. La génération de réseaux moléculaires a permis de mettre en évidence un groupe de 25 composés se fragmentant de manière similaire. Seize de ces composés ont été identifiés comme étant des dérivés de fungisporines, des métabolites suspectés d’intervenir dans la croissance aérienne des champignons. Des analyses structurales ont permis de caractériser de nouveaux composés potentiellement impliqués dans l’infestation des denrées alimentaires. Le marquage du métabolome de F. graminearum par des isotopes stables a permis de mettre en évidence la production de 37 métabolites secondaires dont 29 inconnus lorsque le champignon se développe in vitro. Des analyses par MSn ont permis d’élucider les structures des fusaristatines C et D. / Characterization of fungal secondary metabolomes became a great challenge in the last decades due to both the emergence of fungal threats, and the industrial interest of many natural products; In view of this, we recently developed an analytical strategy for fungal secondary metabolome characterization (Cano P. et al. Anal. Chem. (2013) 85:8412) based on untargeted MS metabolomics applied to labeled samples. This strategy has been here validated by application to the analysis of the complex secondary metabolomes of Penicillium verrucosum and Penicillium nordicum. HRMS acquisitions performed on specific isotopically labelled samples, MS/MS experiments and in-silico emerging tools such as molecular networks, allowed to characterize 181 metabolites, including 80% of new compounds, and the structural determination of seven potential new mycotoxins. Penicillium verrucosum (NRRL 5571) and Penicillium nordicum (NRRL 6062) were grown on wheat grains (Triticum aestivum) presenting different isotopic enrichments: (i) naturally enriched grains, (ii) 97% 13C, and (iii) 53% 13C / 97% 15N. Extracts of each culture were analyzed by HPLC coupled to a LTQ-Orbitrap mass spectrometer equipped with electrospray ionization, operating in the positive or the negative mode. Metabolites were then specifically detected according to the specific isotopic pattern of their respective isotopic enrichments. Known secondary metabolites were annotated using the Antibase database, then identified by comparison with standard compounds when available. Unknown secondary metabolites were annotated using molecular networks of MS/MS similarities (Watrous J. et al.; PNAS (2012) 109 E1743). Wheat grains representing the only source of carbon and nitrogen for fungal growth, the produced fungal secondary metabolites were either unlabeled (naturally enriched cultures), singly labeled (13C cultures) or doubly labeled (13C/15N cultures). This feature allowed discrimination of fungal metabolites against non-fungal compounds which remained unlabeled in the three substrates. Fungal origin was further confirmed by analysis of a control 12C wheat extract (without fungus). Furthermore, the comparison of m/z ratios of a same metabolite detected in the three different cultures, led to the unambiguous determination of the number of carbon and nitrogen atoms and therefore to the unambiguous characterization of its chemical formula. This approach previously developed and validated on a well characterized fungus, has been here successfully applied to the characterization of the complex and unknown secondary metabolomes of P. verrucosum and P. nordicum. Analyses of the two studied fungal strains allowed the detection of 181 secondary metabolites. Interestingly, only 20% of them are suspected to match known metabolites according to databases, meaning that 80% of this metabolome is unknown. To enhance unknown identification efficiency, a molecular network of MS/MS similarities has been generated from our data. A group of 24 metabolites with highly similar MS/MS spectra was highlighted on P. nordicum and P. verrucosum. Fifteen of them were identified as cyclic tetrapeptides from the fungisporin family. Tandem mass spectrometry experiments were performed to characterize the structure of these secondary metabolites. To the best of our knowledge, this is the first time these molecules are pointed out on these Penicillium species. More interestingly, seven of the other metabolites display some similarities with fungisporins, but have never been detected on fungal metabolomes. Furthermore, although the two studied strains are genetically close, these new metabolites seem to be strain specific.