A machine learning model for the prediction and verification of the protective intestinal microbiota-derived L-valine in sepsis

ABSTRACT:

The intestinal microflora and metabolites produced by these microbes serve as important regulators of the development of sepsis. Accordingly, this study was designed to systematically explore the relationships between the regulation of septicemia and both the intestinal flora and fecal metabolites by examining the functional roles of metabolites in the protection against sepsis-associated intestinal damage. To that end, fecal and peripheral blood mononuclear cell (PBMC) samples were collected from sepsis patients and healthy controls. A series of longitudinal multi-omics analyses were then used to assess the links between the intestinal flora or associated metabolites and PBMCs in sepsis patients, while animal model studies were further used to probe the protective effects of intestinal flora-derived metabolites on intestinal damage and immunity in the context of sepsis. These analyses revealed that intestinal dysbiosis was a common finding in sepsis patients, which commonly exhibited higher levels of deleterious bacteria and/or reductions in beneficial bacteria. A machine learning approach was used to identify samples from sepsis patients, revealing that at the genus level, sepsis samples could be distinguished by the presence of Bifidobacterium, Bacteroides, Porphyromonas, Prevotell, Enterococcus, Anaerococcus and Veillonella species. Metabolomics analyses indicated that there were significant differences in the levels of intestinal flora-derived metabolites including L-serine, L-valine and L-tyrosine when comparing samples from the sepsis and control groups, while corresponding transcriptomic analyses of PBMC samples using an ImmunecellAI analytical approach revealed a significant sepsis-related increase in the abundance of T cells and Th17 cells. Single-cell sequencing data from sepsis-associated PBMCs was also downloaded from the GEO database, confirming the observation that Th17 cell levels and those of other immune cells rose significantly in the context of septicemia. Animal model experiments revealed that intestinal microbiota-derived L-valine was able to alleviate inflammation and protest against sepsis-induced intestinal damage by inhibiting Th17 cell activation. Overall, these results thus highlight the successful application of machine learning to distinguish between sepsis and control samples based on the composition of the intestinal flora while demonstrating the potential therapeutic benefits of L-valine as an inhibitor of Th17 cell activity that may offer value as a means of alleviating or preventing intestinal damage in treated individuals. 

INSTRUMENT(S): Liquid Chromatography MS - negative - reverse phase, Liquid Chromatography MS - positive - reverse phase

SUBMITTER: Zetian Wang 

PROVIDER: MTBLS7866 | MetaboLights | 2023-07-28

REPOSITORIES: MetaboLights