Comparative analysis of amino acid auxotrophies and peptidase profiles in non-dysbiotic and dysbiotic small intestinal microbiomes

Abstract

Small Intestinal Bacterial Overgrowth (SIBO) is linked to various diseases and has been associated with altered serum amino acid levels. However, the direct role of the gut microbiome in these changes remains unconfirmed. This study employs genome-scale metabolic modeling to predict amino acid auxotrophy and peptidase gene profiles in the small intestinal microbiomes of SIBO and non-SIBO subjects. Auxotrophy and peptidase gene profiles were further examined in the large intestinal microbiome under non-dysbiotic conditions to assess their similarity to the microbial SIBO profile. Our analysis revealed that the abundance of auxotrophic bacteria is higher in the microbiota of the small intestine than in the large intestine in non-dysbiotic controls. In patients with SIBO, the abundance of auxotrophies in the small intestine decreased compared to non-SIBO subjects. Peptidase gene profiles in non-dysbiotic individuals were distinct between small and large intestinal microbiomes, with fewer extracellular peptidase genes in small intestine microbiomes. In SIBO, extracellular peptidase genes increased compared to non-SIBO subjects. Further, there were more significant associations between the abundance of auxotrophies and peptidase genes in microbiomes of the small intestine compared to the large intestine. In conclusion, the auxotrophy and peptidase gene profiles of the small and large intestinal microbiomes were distinct. In SIBO, the small intestinal microbiome shifts towards a metabolic state resembling that of the large intestine, particularly in its increased abundance of extracellular peptidase genes. This highlights the potential of genome-scale metabolic modeling in identifying metabolic disruptions associated with SIBO, which could inform the development of targeted interventions.

Keywords: Amino acids; Auxotrophies; Peptidases; SIBO; Small intestinal microbiome.

Graphical abstract

Fig. 1

Flowchart of the study. Genome-scale metabolic models were reconstructed from genomes of the human reference gut microbiome (HRGM) catalog with gapseq . Auxotrophies were predicted with metabolic modeling using flux-balance analysis. The peptidase profiles were determined by scanning the peptidase sequences of the HRGM catalog for peptidases from the MEROPS database . The distribution of peptidases and auxotrophies was predicted in the SIBO and Reimagine study cohorts after mapping on HRGM genomes. The farthest distance refers to the distance that could be reached during the sampling. Free available icons were taken from www.flaticon.com (creators: photo3idea_studio, surang, Icon home Eucalyp, Kiranshastry, Becris, Nadiinko).

Fig. 2

Comparison of auxotrophy profiles in small and large intestinal microbiomes predicted by genome-scale metabolic modeling. (a) Comparison of individual microbial amino acid auxotrophy abundances in the SI (Small intestine)-duodenum, SI-jejunum, SI-FD (farthest distance reached in the small intestine), and large intestine (LI). Grey boxes report p values and Kendall’s τ (tau) from Kendall rank sum correlation tests between the relative abundance of auxotrophies (y-axis) and the intestinal regions (color-coded), which is treated as an ordered factor. Reported p values are FDR-adjusted. (b) Abundance-weighted average of auxotrophies per genotype in the duodenal, jejunal, farthest distance, and colonic microbiomes. Horizontal bars denote pairwise tests for statistical differences using Wilcoxon rank sum tests. Labels above bars indicate significance levels: ns – not significant (p value > 0.05), * * (p < 0.01), * ** * (p < 0.0001).

Fig. 3

Comparison of peptidase gene profiles in small and large intestinal microbiomes. (a) Heatmap of individual peptidase abundances (range-scaled) in the microbiomes of the small intestine(SI)-duodenum, SI-jejunum, SI-farthest distance (SI-FD), and large intestine (LI) microbiomes. (b) Abundance-weighted average of peptidases. Horizontal bars denote pairwise tests for statistical differences using Wilcoxon rank sum tests. Labels above bars indicate significance levels: ns – not significant (p value > 0.05), * (p < 0.05), * ** * (p < 0.0001). (c) PCoA plot of peptidase profiles of the microbiomes of the four different intestinal regions (PERMANOVA based on Bray-Curtis Distances of pairwise peptidase profiles, R²=0.397, p = 0.001).

Fig. 4

Associations between predicted peptidases and auxotrophies in duodenal (SI-Duodenum) and large intestinal (LI-stool) microbiomes. Spearman rank sum correlations were calculated between the relative abundance of the respective amino acid auxotrophy (x-axis) and the relative abundance of specific peptidases (y-axis). p values were FDR-adjusted.

Fig. 5

Genome-scale metabolic modeling predicted auxotrophy profiles in the duodenal microbiomes of non-SIBO and SIBO samples. (a) Individual amino acid auxotrophy abundances in the duodenal microbiome of SIBO and non-SIBO patients. Asterisks denote the statistical significance of differences in the relative abundance of the respective auxotrophy between the SIBO and the non-SIBO group (Wilcoxon rank sum test, * p < 0.05, ** p < 0.01, *** p < 0.001). p values are FDR-adjusted. (b) Abundance-weighted average of the number of auxotrophies per gut microbial genotype in SIBO and non-SIBO (Wilcoxon rank sum test, p = 0.036).

Fig. 6

Microbial peptidase gene profiles in duodenal microbiomes of SIBO and non-SIBO samples. (a) Heatmap of individual peptidase abundances (range-scaled) in small intestinal microbiomes from SIBO and non-SIBO samples. (b) Abundance-weighted average of peptidases in the microbiomes of SIBO and non-SIBO samples (Wilcoxon rank sum test, p value < 0.0001). (c) PCoA plot of peptidase profiles of the microbiomes of SIBO and non-SIBO samples (PERMANOVA based on Bray-Curtis Distances of pairwise peptidase profiles, R²=0.17, p = 0.001).

Fig. 7

Dissimilarities of amino acid auxotrophy (a) and peptidase gene profiles (b) of small intestine microbiomes to profiles of large intestine microbiomes. Small intestine microbiomes in SIBO patients show slightly higher similarity to large intestine microbiomes in auxotrophy and peptidase gene profiles than non-SIBO individuals. The y-axis displays the median Bray-Curtis-Dissimilarities for each small intestine microbiome sample from the patients with SIBO to all large intestine microbiomes from non-dysbiotic individuals. The SIBO and non-SIBO groups were compared by Wilcoxon rank sum tests.