Targeted restoration of the intestinal microbiota with a simple, defined bacteriotherapy resolves relapsing Clostridium difficile disease in mice

Abstract

Relapsing C. difficile disease in humans is linked to a pathological imbalance within the intestinal microbiota, termed dysbiosis, which remains poorly understood. We show that mice infected with epidemic C. difficile (genotype 027/BI) develop highly contagious, chronic intestinal disease and persistent dysbiosis characterized by a distinct, simplified microbiota containing opportunistic pathogens and altered metabolite production. Chronic C. difficile 027/BI infection was refractory to vancomycin treatment leading to relapsing disease. In contrast, treatment of C. difficile 027/BI infected mice with feces from healthy mice rapidly restored a diverse, healthy microbiota and resolved C. difficile disease and contagiousness. We used this model to identify a simple mixture of six phylogenetically diverse intestinal bacteria, including novel species, which can re-establish a health-associated microbiota and clear C. difficile 027/BI infection from mice. Thus, targeting a dysbiotic microbiota with a defined mixture of phylogenetically diverse bacteria can trigger major shifts in the microbial community structure that displaces C. difficile and, as a result, resolves disease and contagiousness. Further, we demonstrate a rational approach to harness the therapeutic potential of health-associated microbial communities to treat C. difficile disease and potentially other forms of intestinal dysbiosis.

Figure 1. Epidemic C. difficile 027/BI causes persistent infection with enhanced transmissibility compared to other virulent variants.

a) Representative fecal shedding patterns from C57BL/6 mice (n = 5 mice per group) simultaneously treated with clindamycin and exposed to human virulent C. difficile spores to mimic natural transmission. Mice were infected with C. difficile ribotype 027 (strain BI-7; n = 300), 017 (strain M68; n = 240) and 012 (strain 630; n = 50). Mice supershedding high-levels of C. difficile (>10⁸ CFU/gram fresh feces) are highly contagious (i and iii) whereas mice shedding low-levels of C. difficile (ii; <10² CFU/gram fresh feces) are non-contagious (Figure S3). Broken horizontal line indicates culture detection limit of 50 CFU/gram feces. b) i–ii) hematoxylin and eosin staining to compare cecal pathology of i) healthy, clindamycin treated mice to ii) persistent C. difficile 027/BI-7 supershedders (day 49 post-infection; C57BL/6) that display signs of hyperplasia, edema and immune cell filtrate. Scale bars represent 100 µm. iii–iv) Scanning electron micrographs of illustrating the presence of C. difficile microcolonies (iii) and biofilm-like structures (iv) on the intestinal mucosal surface of persistent supershedders. Scale bars shown in bottom right corner. c) C. difficile 027/BI-7 outcompetes C. difficile 012/R and 017/CF within susceptible host populations. Shown is the summary of two independent experiments that included 6 C. difficile infected donor mice (2 donors infected individually with either C. difficile ribotype 012, 017 or 027) housed with 14 naïve recipient mice for 30 days. The transmission rate of C. difficile 027/BI-7 infected mice is significantly higher (p<1.1e−4) than that of C. difficile 012/630 (p<0.02) or 017/M68 (p<0.22) infected mice.

Figure 2. Epidemic C. difficile 027/BI-7 induces intestinal dysbiosis in mice.

a) Temporal changes in the Shannon Diversity Indices (SDI) of the intestinal microbiota from naïve C57BL/6 mice, clindamycin treated (7 days) naïve C57BL/6 mice or clindamycin treated C57BL6 mice infected with C. difficile 027/BI-7 or 017/M68 (n = 2 mice/group). Fecal samples were collected for DNA extraction two days before clindamycin treatment/infection, 7 days post-treatment/post-infection and 49 days post-treatment/post-infection. b) Analysis of 16S rRNA gene sequences (variable regions 2–5) derived from fecal pellets of naïve mice (n = 17), C. difficile carriers (n = 5; 35–49 days post-infection), mice undergoing clindamycin treatment (n = 12), mice recovered from clindamycin treatment (n = 4; 42 days after cessation of treatment) and persisting supershedders of C. difficile 027/BI-7 (n = 15; 35–49 days post-infection). SS, supershedder; car, carrier; clin recov, mice treated with clindamycin for 7 days and then sampled 42 days later; naïve clin, naïve mice treated with clindamycin for 7 days and then sampled; 027 clin (017 clin), mice infected with C. difficile 027/BI-7 (017/M68) and treated with clindamycin for 7 days and then sampled. Community diversity patterns were determined using the Bray Curtis calculator on 336 OTUs (12,316 clones) sharing 98% identity and the Shannon Diversity Index calculated as described. Various murine genetic backgrounds were tested including, C57BL/6, C57BL/6 p40^−/−, C3H/HeN and C3H/HeJ, as indicated. c) Short chain fatty acid (SCFA) profiles of the intestinal microbiota from naïve C57BL/6 mice, clindamycin-treated C57BL/6 mice that had been allowed to recover for 49 days prior to sampling and C. difficile 027/BI-7 supershedding C57BL/6 mice (n = 5 mice/group).

Figure 3. Fecal transplantation resolves relapsing epidemic C. difficile 027/BI-7 disease and host contagiousness.

a) C. difficile shedding patterns from mice (average shedding from 5 mice/cage) demonstrating that epidemic C. difficile infection is refractory to vancomycin treatment (van) and results in a relapsing supershedder state. Fecal transplantation suppresses high-level C. difficile 027/BI-7 shedding (brown) whereas PBS administration had no impact on C. difficile 027/BI-7 shedding levels (black). Toxins were detected in the feces of supershedders but not in the feces of carriers using the ToxA/B Quikchek (Techlab, Blackburg, VA, USA). Broken horizontal line indicates culture detection limit of 50 CFU/gram feces. b) Supershedder mice efficiently transmit C. difficile to naive mice whereas mice treated with feces and transformed to carriers become poor donors of infection to naive mice. Transmission efficiency refers to the percentage of naïve recipient mice (n = 10/group) that became infected with C. difficile 027/BI-7. c) Quantitative RT-PCR of RNA extracted from supershedder mice cecal tissue showing high-level expression of the proinflammatory genes IL-6, iNOS and Ly6G, which were suppressed to levels comparable to naive mice after fecal transplantation. Cytokine expression was normalized to Gapdh and is shown as relative values.

Figure 4. Effective bacteriotherapy re-establishes a healthy, diverse microbiota profile in epidemic C. difficile 027/BI supershedder mice.

a) Principal component analysis of the 16S rRNA gene sequences demonstrates that distinct microbiota profiles (circled) are associated with “healthy/naïve” mice, mice undergoing “clindamycin treatment” and “persisting supershedders” of C. difficile 027/BI-7. PC1 and PC2 account for 38% of the variation. Each symbol represents one microbiota (dot) or treatment (star) community. Treatment of supershedder mice with feces from healthy mice, the cultured fecal derivative or mixtures of defined, cultured bacteria are as indicated: brown - shading for healthy feces, blue - shading for fecal derivatives culture passaged once, green - shading for mixture of six suppressive bacteria (MixB) and grey - shading for Bacteroides/Lactobacillus mixture. The symbol representing the Bacteroides/Lactobacillus treatment is based on culturing counts and modified to reflect the relative abundance of each organism in the mixture. Next to the shading: pre = pre-treatment; 3 = 3 days post-treatment; 4 = 4 days post-treatment; 6 = 6 days post-treatment; 14 = 14 days post-treatment. Grey background arrows indicate the shifts in the microbiota profiles of treated mice over a 14-day period. b) Fecal shedding profiles from supershedder mice (n = 5/group) that were treated with MixA, MixB or MixC (Table S3). c) Shannon Diversity Indices of the intestinal microbiota of supershedders pre- and post-treatment (day 3, 6 and 14) with MixB and that of the corresponding input community.

Figure 5. Whole genome (maximum likelihood) phylogeny of intestinal bacteria demonstrating the phylogenetic placement of disease-resolving bacteriotherapy bacteria (MixB) and the dominant members of the supershedder microbiota.

Maximum likelihood phylogeny produced using FastTree from the concatenated protein sequence of 44 common genes (See methods). Species names marked in green indicate members of the suppressive MixB mixture, names marked in red indicate species that were commonly detected in the feces of supershedding mice, names in black are reference genomes from common intestinal bacteria that were included to provide phylogenetic context to the tree. Taxonomic designations are given at the relevant branch nodes. Adjacent pictures are transmission electron micrographs of sectioned bacterial strains that constitute MixB. Methods for sample processing and imaging have been described . Scale bars are shown below bacteria.

Figure 6. Proposed model for establishment of C. difficile-mediated dysbiosis and successful bacteriotherapy.

Intestinal homeostasis (a) is characterized by lack of pathology and a diverse, stable microbiota that produces SCFA via fermentation. Antibiotic perturbation (b–c) kills susceptible bacteria resulting in a simplified community structure (and reduced SCFA production) and a loss of colonization resistance. In the absence of opportunistic infection, the microbiota generally rebounds in diversity and SCFA production (d) to re-establish homeostasis and colonization resistance (a). However, exposure to C. difficile (e) after antibiotic perturbation (b) can lead to persistent dysbiosis (f) that is characterized by a pathogenic microbial community, reduced SCFA and pathology. Bacteriotherapy disrupts dysbiosis (g) leading to the clearance of C. difficile (h) and re-establishment of intestinal homeostasis (a).