Type 2 Diabetes Mellitus has reached epidemic levels globally, and several studies have confirmed a link between gut microbial dysbiosis and aberrant glucose homeostasis among people with diabetes. While the assumption is that abnormal metabolomic signatures would often accompany microbial dysbiosis, the connection remains largely unknown. In this study, we investigated how diet changed the gut bacteriome, mycobiome and metabolome in people with and without type 2 Diabetes.1 Differential abundance testing determined that the metabolites Propionate, U8, and 2-Hydroxybutyrate were significantly lower, and 3-Hydroxyphenyl acetate was higher in the high fiber diet compared to low fiber diet in the healthy control group. Next, using multi-omics factor analysis (MOFA2), we attempted to uncover sources of variability that drive each of the different groups (bacterial, fungal, and metabolite) on all samples combined (control and DM II). Performing variance decomposition, ten latent factors were identified, and then each latent factor was tested for significant correlations with age, BMI, diet, and gender. Latent Factor1 was the most significantly correlated. Remarkably, the model revealed that the mycobiome explained most of the variance in the DM II group (12.5%) whereas bacteria explained most of the variance in the control group (64.2% vs. 10.4% in the DM II group). The latent Factor1 was significantly correlated with dietary intake (q < 0.01). Further analyses of the impact of bacterial and fungal genera on Factor1 determined that the nine bacterial genera (Phocaeicola, Ligilactobacillus, Mesosutterella, Acidaminococcus, Dorea A, CAG-317, Caecibacter, Prevotella and Gemmiger) and one fungal genus (Malassezia furfur) were found to have high factor weights (absolute weight > 0.6). Alternatively, a linear regression model was fitted per disease group for each genus to visualize the relationship between the factor values and feature abundances, showing Xylose with positive weights and Propionate, U8, and 2-Hydroxybutyrate with negative weights. This data provides new information on the microbially derived changes that influence metabolic phenotypes in response to different diets and disease conditions in humans.
© 2023. Springer Nature Limited.