Antibiotic-induced microbiome perturbations are associated with significant alterations to colonic mucosal immunity in rhesus macaques

Abstract

The diverse bacterial communities that colonize the gastrointestinal tract play an essential role in maintaining immune homeostasis through the production of critical metabolites such as short-chain fatty acids (SCFAs) and this can be disrupted by antibiotic use. However, few studies have addressed the effects of specific antibiotics longitudinally on the microbiome and immunity. We evaluated the effects of four specific antibiotics: enrofloxacin, cephalexin, paromomycin, and clindamycin, in healthy female rhesus macaques. All antibiotics disrupted the microbiome, including reduced abundances of fermentative bacteria and increased abundances of potentially pathogenic bacteria, including Enterobacteriaceae in the stool, and decreased Helicobacteraceae in the colon. This was associated with decreased SCFAs, indicating altered bacterial metabolism. Importantly, antibiotic use also substantially altered local immune responses, including increased neutrophils and Th17 cells in the colon. Furthermore, we observed increased soluble CD14 in plasma, indicating microbial translocation. These data provide a longitudinal evaluation of antibiotic-induced changes to the composition and function of colonic bacterial communities associated with specific alterations in mucosal and systemic immunity.

Figure 1. Study Schedule.

Animals (n=3 per group) were treated with enrofloxacin, cephalexin or paromomycin for nine days, or clindamycin for six days. Two sets of samples were collected prior to the treatment. During the treatments, non-invasive samples were collected three times and mucosal samples collected once. Animals were tracked for 63 days after initiation of the antibiotic treatments.

Figure 2. Bacterial community composition of the colonic mucosa throughout the antibiotic treatments.

(A) Mucosal bacterial communities of all animals showed a high abundance of Helicobacteraceae prior to the antibiotic treatment. The antibiotics disrupted the mucosal bacterial communities, but clindamycin showed no impact on the predominant Helicobacteraceae. (B) Stool-associated bacterial communities of all of the animals were highly diverse, predominated by bacteria from the phyla Firmicutes and Bacteroides. Each antibiotic disrupted these communities, but the effects on community composition varied between the different antibiotics. (C) Random Forest analysis demonstrated that Enterobacteriaceae were the most strongly discriminating bacterial group in the stool bacterial communities at time points when antibiotics were being administered. (D) Other discriminating groups included Verrucomicrobiaceae, Helicobacteraceae, and Streptococcaeae, among others. The most abundant genera of Enterobacteriaceae included Escherichia, Proteus, and Enterobacter with a small proportion of Klebsiella. (E) The relative abundance of Enterobacteriaceae in the stool bacterial communities increased rapidly upon initiation of the antibiotic treatments (Day 0–7), persisted in some animals until Day 14, and then returned to pre-antibiotic levels by Day 49. (F) Akkermansia, detected in the stool bacterial communities, increased in relative abundance near the end of the antibiotic treatment and then persisted through the end of the study. (G) Ruminococcaceae displayed decreased relative abundance on Days 7 and 14, and increased back to pre-antibiotic levels on Day 28. (E-G) Green points indicate samples taken prior to the antibiotic treatment, red points indicate samples during the antibiotic treatment, and blue points indicate samples taken after cessation of the antibiotic treatment. Statistics were performed across all antibiotic treatment groups using a Wilcoxon signed-rank test and corrected for multiple comparisons using a post-hoc Bonferroni correction.

Figure 3. Antibiotics alter diversity of mucosal and stool bacterial communities.

(A) Species diversity, as measured by the Inverse Simpson Index, was altered by antibiotic treatment across all animals in the four antibiotic treatment groups. The colonic mucosa showed increased species diversity during (Day 3) and shortly after (Day 14) the antibiotic treatments. (B) Stool communities showed reduced diversity during (Days 0, 3, 7) and shortly after (Day 14) the antibiotic treatment. (C-D) Principal coordinates analysis confirmed that the antibiotics disrupted the community composition in the mucosa (C) and stool (D) during the antibiotic treatment. Green points indicate samples taken prior to the antibiotic treatment, red points indicate samples during the antibiotic treatment, and blue points indicate samples taken after cessation of the antibiotic treatment. Statistics were performed across all groups using a Wilcoxon signed-rank test and and corrected for multiple comparisons using a post-hoc Bonferroni correction.

Figure 4. Concentrations of SCFA decrease during antibiotic treatment.

Across all four antibiotic treatment groups, stool concentrations of the SCFA acetate, propionate, butyrate, isobutyrate, valerate, and isovaerate were significantly reduced during the antibiotic treatment (indicated by gray bar) and rapidly returned to pre-antibiotic levels following cessation of the antibiotic treatments. Enrofloxacin – red circles, cephalexin – blue squares, paromomycin – green triangles, clindamycin – black upside-down triangles. The table below shows P values that were determined using a paired-t test with Bonferroni Correction for for multiple comparisons (NS: not significant). Statistics were performed across all antibiotic treatment groups.

Figure 5. Neutrophils infiltrate the colonic mucosa during antibiotic treatment.

(A) In all treatment groups, flow cytometry analysis demonstrated a significant increase of neutrophils in the colonic mucosa during the antibiotic treatment (indicated by the gray bar), which persisted to Day 14 and then returned to pre-antibiotic levels by Day 28–63. (B) Representative flow cytometry plots demonstrating accumulation of neutrophils on Day 3 of the antibiotic treatment. Neutrophils were specifically identified as CD45+CD3-CD20-CD14+CD11b+ leukocytes. CD11bHiCD14Hi cells were specifically gated out in order to exclude monocytes from the analysis. (C/D) We observed a significant decrease in Caspase-3+ neutrophils at days 14 and 28, which returned to pre-antibiotic levels by Day 63. (E-F) RNA-seq analysis demonstrated increased expression of cxcl8, the gene encoding IL-8 (E), as well as the IL-8 production pathway (F) with strong temporal alignment with the neutrophil infiltration observed through the flow cytometry analysis. Interestingly, the enrofloxacin group showed reduced expression of cxcl8 and the IL-8 production pathway on Day 3, but increased expression on Day 14. Enrofloxacin – red circles, cephalexin – blue squares, paromomycin – green triangles, clindamycin – black upside-down triangles. P values were determined using a paired-t test with Bonferroni Correction for for multiple comparisons. Statistics were performed across all antibiotic treatment groups.

Figure 6. IL-17 producing CD4+ T cells increase in the colonic mucosa following the antibiotic treatment.

(A) Flow cytometry analysis demonstrated an increase of IL-17 producing CD4+ T cells (defined as CD45+CD3+CD4+IL17+) in the colonic mucosa after the antibiotic treatment (indicated by the gray bar) at Day 14 and a significant increase at Day 28, which returned to pre-antibiotic values by Day 63. (B) Representative flow cytometry plots demonstrating increased frequency of IL-17-producing CD4+ T cells on Day 28. (C) We observed an increase in the frequency of IL-22-producing CD4+ T cells. (D) mRNA-seq analysis similarly demonstrated increased expression of genes in the IL-17 pathway on Day 14–28 after the antibiotic treatment. Enrofloxacin – red circles, cephalexin – blue squares, paromomycin – green triangles, clindamycin – black upside-down triangles. P values were determined using a paired-t test with Bonferroni Correction for for multiple comparisons. Statistics were performed across all antibiotic treatment groups.

Figure 7. Plasma concentrations of sCD14 increased during antibiotic treatment.

Across all treatment groups, we observed an increase in plasma concentrations of sCD14 during the antibiotic treatment (indicated by the gray bar), which reached significance at Day 7. Plasma sCD14 levels returned to pre-antibiotic values following cessation of antibiotics. These data support an inflammatory response during the antibiotic treatment. Enrofloxacin – red circles, cephalexin – blue squares, paromomycin – green triangles, clindamycin – black upside-down triangles. P values were determined using a paired-t test with Bonferroni Correction for for multiple comparisons. Statistics were performed across all antibiotic treatment groups.