In-depth characterization of a selection of gut commensal bacteria reveals their functional capacities to metabolize dietary carbohydrates with prebiotic potential

Abstract

The microbial utilization of dietary carbohydrates is closely linked to the pivotal role of the gut microbiome in human health. Inherent to the modulation of complex microbial communities, a prebiotic implies the selective utilization of a specific substrate, relying on the metabolic capacities of targeted microbes. In this study, we investigated the metabolic capacities of 17 commensal bacteria of the human gut microbiome toward dietary carbohydrates with prebiotic potential. First, in vitro experiments allowed the classification of bacterial growth and fermentation profiles in response to various carbon sources, including agave inulin, corn fiber, polydextrose, and citrus pectin. The influence of phylogenetic affiliation appeared to statistically outweigh carbon sources in determining the degree of carbohydrate utilization. Second, we narrowed our focus on six commensal bacteria representative of the Bacteroidetes and Firmicutes phyla to perform an untargeted high-resolution liquid chromatography-mass spectrometry metabolomic analysis: Bacteroides xylanisolvens, Bacteroides thetaiotaomicron, Bacteroides intestinalis, Subdoligranulum variabile, Roseburia intestinalis, and Eubacterium rectale exhibited distinct metabolomic profiles in response to different carbon sources. The relative abundance of bacterial metabolites was significantly influenced by dietary carbohydrates, with these effects being strain-specific and/or carbohydrate-specific. Particularly, the findings indicated an elevation in short-chain fatty acids and other metabolites, including succinate, gamma-aminobutyric acid, and nicotinic acid. These metabolites were associated with putative health benefits. Finally, an RNA-Seq transcriptomic approach provided deeper insights into the underlying mechanisms of carbohydrate metabolization. Restricting our focus on four commensal bacteria, including B. xylanisolvens, B. thetaiotaomicron, S. variabile, and R. intestinalis, carbon sources did significantly modulate the level of bacterial genes related to the enzymatic machinery involved in the metabolization of dietary carbohydrates. This study provides a holistic view of the molecular strategies induced during the dynamic interplay between dietary carbohydrates with prebiotic potential and gut commensal bacteria.

Importance: This study explores at a molecular level the interactions between commensal health-relevant bacteria and dietary carbohydrates holding prebiotic potential. We showed that prebiotic breakdown involves the specific activation of gene expression related to carbohydrate metabolism. We also identified metabolites produced by each bacteria that are potentially related to our digestive health. The characterization of the functional activities of health-relevant bacteria toward prebiotic substances can yield a better application of prebiotics in clinical interventions and personalized nutrition. Overall, this study highlights the importance of identifying the impact of prebiotics at a low resolution of the gut microbiota to characterize the activities of targeted bacteria that can play a crucial role in our health.

Keywords: commensal bacteria; dietary carbohydrates; metabolomics; prebiotics; transcriptomics.

Fig 1

Functional activities of a selection of 17 commensal bacteria in the presence of carbohydrate substrates. Various carbohydrate sources were added into a low nutrient culture medium at concentrations of 0.5% (w/v for agave inulin, corn fiber, and polydextrose) or 0.1% (w/v for citrus pectin), taking into account the solubility and viscosity of dietary carbohydrates. Parallel controls included glucose at 0.5% or 0.1%, as well as a condition with no added carbohydrates. (A) Variations of OD at 600 nm reflect bacterial growth after 24 h of culture. (B) Variations in pH reveal medium acidification as a consequence of the production of fermentation end-products, such as short-chain fatty acids, after 24 h of culture. OD and pH variations correspond to the difference between OD and pH values measured after 24 h of culture, in comparison to the value measured just after the inoculation of the bacteria. Comparisons were conducted within each bacterial species between different carbon sources and the non-supplemented medium using unpaired Mann–Whitney’s non-parametric tests. The mean values annotated with * are significantly different (P-value < 0.0625) compared to the control conditions of no carbohydrates.

Fig 2

Classification of carbohydrate fermentation profiles of 17 commensal bacteria. In the dendrogram, all bacterial monocultures are grouped based on their fermentation profiles. We can observe four distinct clusters, highlighting the functional capacities of commensal bacteria to ferment various carbohydrates. The blue shading represents short-chain fatty acid (SCFA) concentrations, expressed in mM. Dotted lines denote the overall mean of total SCFA, excluding valerate, i-caproate, and caproate due to their null concentrations. The straight lines indicate the SCFA values for each bacterial culture. The hierarchical classification identified four clusters: in gray, cluster N; in orange, cluster G; in pink, cluster A; in green, and cluster G/I.

Fig 3

Metabolomic profiles of commensal bacteria according to culture conditions and carbon sources. (A) Schematic of the experimental strategy and the untargeted metabolomic data analysis. The metabolomic study was conducted on the supernatants of six commensal bacteria belonging to two different bacterial phyla, including Bacteroidetes (B. xylanisolvens, B. thetaiotaomicron, and B. intestinalis) and Firmicutes (S. variabile, R. intestinalis, and E. rectale). These bacterial species were cultivated in five low nutrient culture media (LNCM), each supplemented or not with different carbon sources. Each condition was performed in six replicates, in addition to five replicates of the initial non-inoculated LNCM. The liquid chromatography–mass spectrometry (LC-MS) metabolomic approach consisted of two types of chromatographic conditions (C18 and HILIC) and two ionization conditions (both positive and negative modes), resulting in the detection of features in the HILIC column and negative ionization mode, and features in the C18 columns and positive ionization mode. Scatter plots of the first two sparse partial least squares-discriminant analyses (sPLS-DA) components were obtained for each of the bacterial species for all the culture conditions (B) B. xylanisolvens; (C) B. thetaiotaomicron; (D) B. intestinalis; (E) S. variabile; (F) R. intestinalis; and (G) E. rectale. All ellipses were drawn assuming a multivariate t-distribution with a confidence level of 0.95.

Fig 4

Heatmaps of differentially abundant metabolites after 24-h incubation of bacterial strains (A) without any carbohydrate supplementation, (B) in the presence of 0.5% glucose, (C) 0.1% pectin, (D) 0.5% inulin, and (E) 0.5% corn fiber. The bacterial species were B. xylanisolvens, B. thetaiotaomicron, B. intestinalis, S. variabile, R. intestinalis, and E. rectale (named B. xyla, B. theta, B. int, S. var, R. int, and E. rect, respectively). Metabolites that differ between the medium inoculated with each of the bacterial species and the corresponding non-incubated medium were identified using a univariate non-parametric test (Wilcoxon test, P < 0.05). Metabolites plotted were significantly modulated in at least one condition with a fold change higher than 2. Only metabolites annotated with a high level of confidence are shown.

Fig 5

Transcriptomic profiles of four commensal bacteria in response to various carbon sources. (A) Schematic of the experimental strategy. The transcriptomic study was conducted on four commensal bacteria, belonging to two different bacterial phyla, including Bacteroidetes (B. thetaiotaomicron and B. xylanisolvens) and Firmicutes (R. intestinalis and S. variabile). These bacterial species were cultivated in five low nutrient culture media (LNCM), each supplemented or not with different carbon sources. Each condition was performed in three replicates. (B) Hierarchical clustering heatmaps of expression changes for the most overexpressed genes regrouped in CAZyme gene clusters (CGCs) that were combined across the LNCM-inulin and LNCM-corn fiber or LNCM-pectin. GH: glycoside hydrolases; GT: glycosyl transferase; PL: polysaccharide lyases; CE: carbohydrate esterases; CBM: carbohydrate-binding module; TC: transporters; TF: transcription factors.