Abstract

The use of model species in environmental science is confronted with several scientific limitations, such as their geographical distribution and their representativeness of the diversity of species in aquatic environments. Indeed, the knowledge related to model organisms does not allow the extrapolation and the prediction of the response(s) of the organisms present in the environment. Nevertheless, advances in new sequencing and mass spectrometry (MS) technologies allow the extension of large-scale molecular data acquisition to organisms of environmental relevance, such as the amphipod Gammarus fossarum. The overall goal of this thesis is to develop unbiased and multiomics (transcriptomics and proteomics) bioinformatics approaches to exploit the large amount of omics data available in non-model species and to overcome the limitations of functional annotation from distant species databases. To address this objective, two strategies were implemented. First, a coexpression network analysis was performed using the WGCNA method on two proteomic datasets, with the aim of identifying the molecular actors (i.e. proteins) regulating physiology and/or response to the contaminant. The first dataset combined proteomic profiles of male and female gonads at different stages of maturation and embryos during their development. The second dataset was from a previous study, and consisted of proteomic data from gammarid testes exposed to three contaminants: pyriproxyfen, cadmium and methoxyfenozide. These two studies allowed the identification of coexpressed protein modules specifically associated with different developmental and reproductive stages on the one hand, and with exposure to different contaminants on the other hand. Protein module enrichment analyses revealed and supported new hypotheses to understand the biological processes involved in gammarid physiology, as well as the mechanisms of action (MoA) of contaminants. In addition, this strategy allowed us to identify key proteins involved in physiological and contaminant-associated pathways that had not been revealed by the differential analysis used. In a second step, we apply a multiomics strategy to characterize the metabolic pathways involved in ML in G. fossarum. Indeed, we adapted the genomic annotation tool (CycADS), which allows the reconstruction of metabolic pathways, to the male and female transcriptomic data, by reducing the redundancy of transcript isoforms. Subsequently, we integrated proteogenomic data from different organs to validate MS identified enzymes with functional annotation data and to analyze ML organotropism. Metabolic pathway reconstruction identified over 70 pathways involved in ML. The MS data validated the detection of about 100 enzymes catalyzing the reactions of the identified pathways in the different organs, as well as their expression profiles that are characteristic of certain organs. Overall, this work provides a solid foundation for the use of omics data from non-model organisms in a no-priori or multiomics exploration strategy. The results identified proteins potentially involved in biological processes related to developmental and reproductive stages of G. fossarum, as well as testicular toxicity. These results also highlight the value of applying omics approaches at the organ level of sentinel species to identify and assess possible differences in contaminant MoA depending on the target organ.

Key words

Molecular pathways; Multiomics; Lipid metabolism; Gammrarus fossarum; Transcriptomic; Proteogenomic; Coexpression networks; Systems biology

Cite the thesis

Natacha Koenig. Approches bioinformatiques pour l’exploration sans a priori des voies moléculaires chez les espèces non modèles : le cas du métabolisme lipidique chez Gammarus fossarum. Bio-informatique [q-bio.QM]. Université Claude Bernard - Lyon I, 2022. Français. ⟨NNT : 2022LYO10006⟩. ⟨tel-04199684⟩

Access manuscript on HAL thèses