Antibiotic-induced shifts in the mouse gut microbiome and metabolome increase susceptibility to Clostridium difficile infection

Abstract

Antibiotics can have significant and long-lasting effects on the gastrointestinal tract microbiota, reducing colonization resistance against pathogens including Clostridium difficile. Here we show that antibiotic treatment induces substantial changes in the gut microbial community and in the metabolome of mice susceptible to C. difficile infection. Levels of secondary bile acids, glucose, free fatty acids and dipeptides decrease, whereas those of primary bile acids and sugar alcohols increase, reflecting the modified metabolic activity of the altered gut microbiome. In vitro and ex vivo analyses demonstrate that C. difficile can exploit specific metabolites that become more abundant in the mouse gut after antibiotics, including the primary bile acid taurocholate for germination, and carbon sources such as mannitol, fructose, sorbitol, raffinose and stachyose for growth. Our results indicate that antibiotic-mediated alteration of the gut microbiome converts the global metabolic profile to one that favours C. difficile germination and growth.

Figure 1. Susceptible and resistant states of C. difficile infection

Each circle represents the state of the intestinal environment for each of the 4 treatment groups. C57BL/6 WT mice (represented at baseline by state R1) were treated with cefoperazone for 10 days. The structural and functional status of the gut was determined two days (S1) or 6 weeks (R3) after discontinuation of the antibiotic. An additional group of mice (R2) was held without any antibiotic treatment to serve as age-matched controls to the R3 animals. Each treatment group was challenged with C. difficile spores to assess if they were resistant or susceptible to colonization and disease. There were 3 states that were fully resistant to colonization and disease (R1, R2, R3) and one susceptible state (S1).

Figure 2. Untargeted metabolomics of the gut metabolome

(a) A heatmap of the metabolites grouped by KEGG pathway found in the gut metabolome of CDI susceptible and resistant mice from all four-treatment groups (n=8, A–H). Mice susceptible to CDI (S1) showed significant changes in their intestinal metabolome compared to the three groups of mice with intestinal environments that were resistant to C. difficile colonization (R1–R3). (b) A heatmap of the relative amounts of bile acids in the gut metabolome of CDI susceptible and resistant mice from all four-treatment groups. The heatmap scale ranges from −6 to +8 on a log₂ scale. (c) A heatmap of carbohydrates in the gut metabolome of CDI susceptible and resistant mice from all four-treatment groups. The heatmap scale ranges from −5 to +10 on a log₂ scale.

Figure 3. Targeted metabolomics of gut metabolome

(a) Bile acids were analyzed by LC-MS from the cecal content of non-antibiotic treated (R1) and cefoperazone treated mice (S1) (n=4 for each group). Significant changes in the concentrations of taurocholate, cholate and deoxycholate resulted from antibiotic treatment (Mann-Whitney non-parametric t-test). Error bars represent the mean ± SEM. (b) Concentrations of sugar alcohols (mannitol and sorbitol) measured by LC-MS from cecal content of untreated (R1) and cefoperazone-treated mice (S1) (n=4 for each group). Antibiotic treatment significantly changes the levels of these sugar alcohols (Mann-Whitney non-parametric t-test). Error bars represent the mean ± SEM. (c) Short chain fatty acid levels (acetate, propionate, butyrate) were analyzed by GC-MS from cecal content of non-antibiotic treated (R1, n=3), 2 days (S1, n=4) and 6 weeks after cefoperazone treated mice (R3, n=3). Changes in SCFAs were significant between groups R1 and S1 (non-parametric Kruskal-Wallis one-way analysis of variance test followed by Bonferroni-Dunn Multiple Comparison Test). Error bars represent the mean ± SEM.

Figure 4. C. difficile in vitro and ex vivo germination and growth studies

(a) In vitro assays were performed to assess the ability of taurocholate (black bars) and deoxycholate (white bars) to trigger C. difficile spore germination. Spores were incubated in BHIS with each bile salt at the indicated concentrations. Spores were incubated with 0.1% bile salt for 30 minutes or with 0.01% bile salt for 6 hours. Data presented represent mean ± SD of triplicate experiments and were significant (Students t-test). (b) In vitro growth of C. difficile was done in a defined minimal media which included essential amino acids and vitamins required by C. difficile for growth. Carbohydrates that were found to be increased after antibiotic treatment during the CDI susceptible state (S1) were supplemented in the media. C. difficile can utilize many of the carbohydrates in the murine gut after antibiotic treatment. Only carbohydrates that supported growth of C. difficile are shown here (Supplementary Table 1). Error bars represent the mean ± SEM of triplicate experiments. (c) Ex-vivo germination and outgrowth of C. difficile was done measured in cecal contents from untreated and from cefoperazone-treated mice two days after antibiotics. C. difficile VPI 10463 spores inoculated into the antibiotic-treated cecal contents (S1, n=6) were able to germinate and outgrow over a 6 hour period where as spores in the non-antibiotic treated cecal content (R1, n=3) did not. Significance between groups was done by Mann-Whitney non-parametric t-test. Error bars represent the mean ± SEM.

Figure 5. Correlation analysis of the microbiome and metabolome

(a) Spearman’s correlation analysis was done with all 84 OTUs in the microbiome color-coded by phylum and grouped based on unsupervised clustering. All 480 metabolites in the metabolome were similarly clustered and then color-coded by KEGG super pathway. There were four distinct clusters of OTUs that were seen, O1–O4, and two distinct clusters of metabolites, M1–M2. The heatmap scale ranges from +1.0 to −1.0. (b) Relative levels of metabolites from all treatment groups (R1, R2, S1 and R3) depicted in a heatmap but keeping with the same order of metabolites as in (a). The heatmap scale for relative levels of metabolites ranges from +10 to −10 on a log₂ scale. (c) Relative abundance of OTUs from all treatment groups (R1, R2, S1 and R3) depicted in a heatmap but keeping with the same order of OTUs as in (a). The heatmap scale ranges from 0 to 100% relative abundance.