A Multi-Omics Approach to Disclose Metabolic Pathways Impacting Intestinal Permeability in Obese Patients Undergoing Very Low Calorie Ketogenic Diet

Abstract

A very low calorie ketogenic diet (VLCKD) impacts host metabolism in people marked by an excess of visceral adiposity, and it affects the microbiota composition in terms of taxa presence and relative abundances. As a matter of fact, there is little available literature dealing with microbiota differences in obese patients marked by altered intestinal permeability. With the aim of inspecting consortium members and their related metabolic pathways, we inspected the microbial community profile, together with the set of volatile organic compounds (VOCs) from untargeted fecal and urine metabolomics, in a cohort made of obese patients, stratified based on both normal and altered intestinal permeability, before and after VLCKD administration. Based on the taxa relative abundances, we predicted microbiota-derived metabolic pathways whose variations were explained in light of our cohort symptom picture. A totally different number of statistically significant pathways marked samples with altered permeability, reflecting an important shift in microbiota taxa. A combined analysis of taxa, metabolic pathways, and metabolomic compounds delineates a set of markers that is useful in describing obesity dysfunctions and comorbidities.

Keywords: 16S microbiota profile; VOCs; calorie restrictive diet; dietary intervention; metabolomics; obesity.

Figure 1

PLS-DA plot. A normalized matrix composed of fecal and urine metabolites, plus 16S taxa, was used as input for the PLS regression analysis. (a) Component 1 versus component 2 PLS-DA score plot based on 16S metataxonomic and metabolomics (fecal and urinary VOCs) variables. (b,d) Relative to other component combinations and, precisely, component 1 vs. 3 and component 2 vs. 3, respectively. (c) Top fifteen “Variable Importance in Projection” (VIP) scores, including both VOCs and microbiome taxa with a value greater than 2.0.

Figure 2

Extended error bar plots of statistically significant bacterial genera that differed as a consequence of VLCKD dietary treatment in post-altered versus pre-altered samples (A) and post-versus pre-normal samples (B). Both the plots report the effect size and associated confidence interval for each significative feature (corrected p < 0.05) obtained in STAMP software by applying corrected Welch test (BH) statistics based on the normalized matrix from QIIME2. Difference in genus mean proportions (95% of confidence intervals) resulted from a Welch pairwise test (corrected by applying a Benjamini–Hochberg multiple test). Post-treatment groups are colored light blue (post-altered) or orange (post normal), and green (pre-altered) or violet (pre-norm). All the reported genera were statistically significant after correction (corrected p < 0.05). Because of the direction of the comparison, the differences in mean proportion for pre-altered samples appeared as negative values.

Figure 3

Log2 fold change vs. significance plot. Statistically significant Picrust2 pathways obtained by comparing post-altered and pre-altered sample groups visualized as a volcano plot. Fold change is indicative of increasing (red) and decreasing (violet) abundance in post-altered predicted pathways.

Figure 4

Pearson correlation between statistically significant predicted pathways and fecal/urinary VOCs, as derived from the comparison between pre- and post-VLCKD-administered subjects with altered intestinal permeability. Network plot of significant linear correlations reported as linear connection (p-value < 0.05 and r > 0.7). Blue lines indicate positive correlations among variables (nodes). Red font indicates fecal and urinary VOCs, whereas black font is relative to metabolic pathways.