Intestinal microbiota containing Barnesiella species cures vancomycin-resistant Enterococcus faecium colonization

Abstract

Bacteria causing infections in hospitalized patients are increasingly antibiotic resistant. Classical infection control practices are only partially effective at preventing spread of antibiotic-resistant bacteria within hospitals. Because the density of intestinal colonization by the highly antibiotic-resistant bacterium vancomycin-resistant Enterococcus (VRE) can exceed 10(9) organisms per gram of feces, even optimally implemented hygiene protocols often fail. Decreasing the density of intestinal colonization, therefore, represents an important approach to limit VRE transmission. We demonstrate that reintroduction of a diverse intestinal microbiota to densely VRE-colonized mice eliminates VRE from the intestinal tract. While oxygen-tolerant members of the microbiota are ineffective at eliminating VRE, administration of obligate anaerobic commensal bacteria to mice results in a billionfold reduction in the density of intestinal VRE colonization. 16S rRNA gene sequence analysis of intestinal bacterial populations isolated from mice that cleared VRE following microbiota reconstitution revealed that recolonization with a microbiota that contains Barnesiella correlates with VRE elimination. Characterization of the fecal microbiota of patients undergoing allogeneic hematopoietic stem cell transplantation demonstrated that intestinal colonization with Barnesiella confers resistance to intestinal domination and bloodstream infection with VRE. Our studies indicate that obligate anaerobic bacteria belonging to the Barnesiella genus enable clearance of intestinal VRE colonization and may provide novel approaches to prevent the spread of highly antibiotic-resistant bacteria.

Fig 1

Fecal transplantation clears VRE colonization in antibiotic-treated mice. Mice were infected with 10⁸ VRE CFU after 1 week of ampicillin treatment. One day after infection, ampicillin treatment was stopped. Mice received PBS or a suspension of fecal pellets from untreated mice by oral gavage for three consecutive days, starting 1 day after antibiotic cessation. (A) Numbers of VRE CFU in the fecal pellets of infected mice were quantified every other day for two consecutive weeks (n = 6). Black dots represent PBS-treated mice and red dots represent fecal transplant-treated mice. (B) Composition of the microbiotas of four PBS and four fecal transplant (FE) mice was analyzed 15 days following infection and compared with that of the microbiotas of three untreated mice (NT). Hierarchical clustering was used to cluster samples by their microbiota composition at the genus level. Each column represents one mouse. Each row represents one genus. The most predominant phyla (left) and genera (right) are indicated.

Fig 2

Commensal anaerobic bacteria suppress VRE colonization in antibiotic-treated mice. Mice were infected with 10⁸ VRE CFU after 1 week of ampicillin treatment. One day after infection, ampicillin treatment was stopped. Mice were orally gavaged for three consecutive days, starting 1 day after antibiotic cessation, with PBS, a suspension of fecal pellets from untreated mice (feces), or an aerobic (aero) or anaerobic (anaero) culture of fecal microbiota from untreated mice. Numbers of VRE CFU in the fecal pellets of infected mice were analyzed 5 weeks after infection (n = 8 to 10). Limit of detection, 10 CFU/10 mg. ***, significantly different (P < 0.001) from the PBS group; ns, not significant.

Fig 3

Microbiota reconstitution and VRE clearance vary after fecal transplantation. Mice were infected with 10⁸ VRE CFUs after a week of ampicillin treatment. One day after infection, ampicillin treatment was stopped. Mice were orally gavaged for three consecutive days, starting 1 day after antibiotic cessation, with PBS, a suspension of fecal pellets from untreated mice (feces), or an aerobic or anaerobic culture of fecal microbiota from untreated mice. (A) Genus level composition. (B) VRE CFU numbers. (C) Shannon diversity index of murine fecal samples obtained 5 weeks after infection. Each column represents one mouse. For the microbiota composition, the most abundant bacterial taxa are indicated with different colors.

Fig 4

Colonization with the Barnesiella genus correlates with VRE elimination. (A) Genus level composition of the fecal microbiota (red heat map) and VRE CFU levels (blue heat map) for the different mice described in Fig. 2 5 weeks after VRE infection. Only the microbial taxa representing at least 1% of the microbiota are shown. Each heat map row represents one mouse, and mice are sorted by VRE colonization levels in descending order. (B) Spearman correlation analysis of the different bacterial genera present in the fecal microbiota of mice with the VRE CFU levels 5 weeks after infection. Spearman correlation coefficient values range from +1 (maximum positive correlation value) to −1 (maximum inverse correlation value). Each point represents one genus. To analyze the statistical significance of a given coefficient Spearman value, P values were computed using the asymptotic t approximation. Spearman coefficient values with a P value of <0.05 are indicated.

Fig 5

Commensal anaerobic bacteria prevent VRE intestinal colonization. Mice were treated with ampicillin and, after antibiotic treatment was stopped, were reconstituted with either PBS or an aerobic or anaerobic culture of the fecal microbiota. (A) Two weeks after reconstitution, the fecal microbiota composition of the mice was determined and mice were infected orally with 10⁸ VRE CFU. (B) The level of VRE colonization was determined 1 week after infection. Each column represents one mouse.

Fig 6

Barnesiella diminishes the risk of VRE domination in allo-HSCT patients. (A) Composition of the intestinal microbiota of allo-HSCT patients before and during VRE domination. (B) Shannon diversity index of samples from patients during VRE domination and prior to VRE domination and from patients that never developed VRE domination. (C) Relative abundance (parts per unit; total microbiota = 1) of the Barnesiella genus in samples from patients that did not develop VRE domination compared to samples from patients that developed VRE domination, taken before domination occurred. The y axis scale is divided into two sections (from 0 to 0.005 and from 0.01 to 0.11 parts per unit). Blue dashed lines indicate the means in each group of samples.