Stable isotope fingerprinting can directly link intestinal microorganisms with their carbon source and captures diet-induced substrate switching in vivo

Abstract

Diet has strong impacts on the composition and function of the gut microbiota with implications for host health. Therefore, it is critical to identify the dietary components that support growth of specific microorganisms in vivo. We used protein-based stable isotope fingerprinting (Protein-SIF) to link microbial species in gut microbiota to their carbon sources by measuring each microbe's natural ¹³C content (δ¹³C) and matching it to the ¹³C content of available substrates. We fed gnotobiotic mice, inoculated with a 13 member microbiota, diets in which the ¹³C content of all components was known. We varied the source of protein, fiber or fat to observe ¹³C signature changes in microbial consumers of these substrates. We observed significant changes in the δ¹³C values and abundances of specific microbiota species, as well as host proteins, in response to changes in ¹³C signature or type of protein, fiber, and fat sources. Using this approach we were able to show that upon switching dietary source of protein, fiber, or fat (1) some microbial species continued to obtain their carbon from the same dietary component (e.g., protein); (2) some species switched their main substrate type (e.g., from protein to carbohydrates); and (3) some species might derive their carbon through foraging on host compounds. Our results demonstrate that Protein-SIF can be used to identify the dietary-derived substrates assimilated into proteins by microbes in the intestinal tract; this approach holds promise for the analysis of microbiome substrate usage in humans without the need of substrate labeling.

Significance: The gut microbiota plays a critical role in the health of animals including humans, influencing metabolism, the immune system, and even behavior. Diet is one of the most significant factors in determining the function and composition of the gut microbiota, but our understanding of how specific dietary components directly impact individual microbes remains limited. We present the application of an approach that measures the carbon isotope "fingerprint" of proteins in biological samples. This fingerprint is similar to the fingerprint of the substrate used to make the proteins. We describe how we used this approach in mice to determine which dietary components specific intestinal microbes use as carbon sources to make their proteins. This approach can directly identify components of an animal's diet that are consumed by gut microbes.

Keywords: Gut microbiota; metaproteomics; microbiome; protein-SIF.

Figure 1:. Overview of experimental design and procedure.

A. Timeline and design of the experiment. The diet in the blue weeks (week 3 in Exp1 and week 4 in Exp2) was the same exact diet. B. Natural carbon isotopic signatures (δ¹³C values) of the macronutrients used in the diets. Signatures of dietary components were averaged between two replicates measured by EA-IRMS (measurement uncertainties of ± 0.42 ‰ or less; ± 0.13 ‰ averaged uncertainty). C. Overview of the metaproteomics protein-SIF approach.

Figure 2:. Changes in isotopic signatures of gut microbes in response to changes in the isotopic signature or source of dietary protein.

(A) Overview of experiment 1 as described in Materials & Methods (Figure 1). (B, D, F, H, J) Protein-SIF δ¹³C values for mouse, A. muciniphila, B. thetaiotaomicron, M. formatexigens and B. uniformis. Displayed δ¹³C values for dietary protein sources, corn oil, corn starch, sucrose, and corn fiber were measured by isotope ratio mass spectrometry (IRMS). Significance is denoted by * and determined by T-tests corrected by BH (p < 0.05; n=5 unless otherwise stated). (C, E, G, I) Relative proteinaceous biomass of A. muciniphila, B. thetaiotaomicron, M. formatexigens and B. uniformis determined according to the method described by (32). Letters that do not overlap denote significantly different groups as determined by Tukey HSD (p < 0.05).

Figure 3:. Changes in isotopic signatures of gut microbes in response to changes in the isotopic signature or source of dietary fiber or fat

A. Overview of Experiment 2 as described in Materials & Methods (Figure 1). (B, D, F, H, J, L, N) Protein-SIF δ¹³C values for mouse, A. muciniphila, B. thetaiotaomicron, M. formatexigens, B. uniformis, B. caccae, and E. coli, respectively. Displayed δ¹³C values for dietary fiber and fat sources, corn starch, sucrose, and egg white protein were measured by isotope ratio mass spectrometry (IRMS). Significance is denoted by * and determined by T-tests corrected by BH (p < 0.05; n=6 unless otherwise stated). (C, E, G, I, K, M) Relative proteinaceous biomass of A. muciniphila, B. thetaiotaomicron, M. formatexigens and B. uniformis, B. caccae, and E. coli. Letters that do not overlap denote significantly different groups as determined by Tukey HSD (p < 0.05).

Figure 4:. Hierarchical clustering of M. formatexigens proteins that significantly differed between casein, egg, white and soy protein.

A. Table listing 17 proteins that were more abundant in the soy cluster. B. Hierarchical clustering of the z-score values of the 60 proteins that changed significantly in abundance between the three dietary protein sources (ANOVA, q < 0.05, n=5). C. Abundant proteins that potentially explain the increase in the δ¹³C value in the soy diet.(Figure 2H). Glutamine synthetase is highlighted in red; proteins from a Sugar ABC transporter gene neighborhood are highlighted in blue. Bars represent average% orgNSAF abundance and error bars represent the standard deviation.

Figure 5:. Hierarchical clustering of B. thetaiotaomicron proteins that significantly differed between cellulose, inulin, and corn fiber.

Hierarchical clustering of the z-score values of the 147 proteins whose abundances changed significantly between the three dietary fiber sources (ANOVA, q < 0.05, n=5). The table represents the 41 proteins abundant in the inulin cluster. Proteins from PUL22 are highlighted in red.

Figure 6:. B. thetaiotaomicron proteins belonging to a PUL that significantly differed in abundance between the diets with different fiber sources.

Heatmap ordered by diet (cellulose, inulin, corn fiber) representing the z-scored abundances of 50 proteins from PULs ordered by the PULs. If the substrate of the PUL is known, it is described. All PULs numbers are from the literature-derived numbering in PULDB