An enterococcal phage-derived enzyme suppresses graft-versus-host disease

Abstract

Changes in the gut microbiome have pivotal roles in the pathogenesis of acute graft-versus-host disease (aGVHD) after allogenic haematopoietic cell transplantation (allo-HCT)^1-6. However, effective methods for safely resolving gut dysbiosis have not yet been established. An expansion of the pathogen Enterococcus faecalis in the intestine, associated with dysbiosis, has been shown to be a risk factor for aGVHD^7-10. Here we analyse the intestinal microbiome of patients with allo-HCT, and find that E. faecalis escapes elimination and proliferates in the intestine by forming biofilms, rather than by acquiring drug-resistance genes. We isolated cytolysin-positive highly pathogenic E. faecalis from faecal samples and identified an anti-E. faecalis enzyme derived from E. faecalis-specific bacteriophages by analysing bacterial whole-genome sequencing data. The antibacterial enzyme had lytic activity against the biofilm of E. faecalis in vitro and in vivo. Furthermore, in aGVHD-induced gnotobiotic mice that were colonized with E. faecalis or with patient faecal samples characterized by the domination of Enterococcus, levels of intestinal cytolysin-positive E. faecalis were decreased and survival was significantly increased in the group that was treated with the E. faecalis-specific enzyme, compared with controls. Thus, administration of a phage-derived antibacterial enzyme that is specific to biofilm-forming pathogenic E. faecalis-which is difficult to eliminate with existing antibiotics-might provide an approach to protect against aGVHD.

Fig. 1. Characterization of enterococcal strains in patients with allo-HCT.

a, Detection of functional haemolytic activity. Representative image from two independent experiments. b, Volcano plot comparing the KO terms of E. faecalis and E. faecium strains. Circles coloured blue and red indicate significantly different pathways (more than twofold, q < 0.05, according to the Wilcoxon rank-sum test (two-sided)) between E. faecalis and E. faecium strains. Source Data

Fig. 2. Lysis of biofilms by a phage-derived antibacterial enzyme derived from an E. faecalis-specific prophage.

a, Number of detected prophage sequences and their viral taxa in each E. faecalis strain. b, Representative bacterial contig showing the whole sequence of a viral contig classified as a putative member of the Podoviridae containing endolysin in an E. faecalis strain obtained from Patient031_14_8. c, Expression and purification of endolysin by western blotting. Representative blot obtained from two independent experiments. d, Bacteriolytic capacity of endolysins against an E. faecalis strain obtained from Patient031_14_8. Representative experiment and image from two independent experiments. OD_600 nm, optical density at 600 nm. e, Mature biofilm assay with crystal violet staining 24 h after incubation with endolysin or vehicle, including His-SUMO (n = 8 in each group). *P = 0.0001554. Significance was determined using the Wilcoxon rank-sum test (two-sided). The line inside the box represents the median. The whiskers represent the range of points up to 1.5 times the interquartile range. Data are representative of two independent experiments. OD_570 nm, optical density at 570 nm f, Oral administration of endolysin or vehicle in germ-free mice (n = 2) or gnotobiotic mice that were mono-colonized with E. faecalis (n = 2). Representative scanning electron microscope images of the small and large intestines. Scale bars, 200 µm (low-magnification images); 5 µm (high-magnification images). Source Data

Fig. 3. Analysis of E. faecalis phage-derived endolysin in E. faecalis-mono-colonized gnotobiotic GVHD mice.

a, Colony-forming units (CFU) (left) and percentage CFU (right) of E. faecalis in faecal samples before and after transplantation. Each dot and line represent the data from one faecal sample. Significance was determined using the Wilcoxon rank-sum test (two-sided). Patient031_14_8: n = 8 for endolysin-treated mice, n = 6 for vehicle-treated mice. Patient009_35_10: n = 8 for endolysin-treated mice, n = 8 for vehicle-treated mice. Patient015_56_7: n = 8 for endolysin-treated mice, n = 7 for vehicle-treated mice. b, Kaplan–Meier survival plots of endolysin-treated mice and vehicle-treated mice. Significance was determined using the log-rank test (two-sided). The numbers of endolysin-treated mice and vehicle-treated mice for each of the patient samples are as in a. Source Data

Fig. 4. Analysis of E. faecalis phage-derived endolysin in humanized gnotobiotic GVHD mice.

a, Microbial composition of the gut, on the basis of the relative abundance of operational taxonomic units at the genus level, for the donor faeces and the faecal samples from humanized mice before and after transplantation. Patient031: n = 10 for endolysin-treated mice, n = 9 for vehicle-treated mice. Patient043: n = 16 for endolysin-treated mice, n = 13 for vehicle-treated mice. Patient032: n = 8 for endolysin-treated mice, n = 8 for vehicle-treated mice. The mean of the relative abundance in each group is shown. b, Bacterial alpha diversities of faecal microbial communities using Pielou’s evenness index. Significance was determined using the Kruskal–Wallis test (two-sided). The line inside the box represents the median. The whiskers represent the range of points up to 1.5 times the interquartile range. There were some deaths in the day-15 Patient043 group and this group was therefore excluded. c, Principal coordinate analysis of the weighted UniFrac distance matrices for the faecal microbial communities. Significance was determined using a pairwise PERMANOVA. NS, not significant. Red, endolysin-treated mice; blue, vehicle-treated mice. There were some deaths in the day-15 Patient043 group and this group was therefore excluded. d, Kaplan–Meier survival plots of endolysin-treated mice and vehicle-treated mice. Significance was determined using the log-rank test (two-sided). Patient031: n = 10 for endolysin-treated mice, n = 9 for vehicle-treated mice. Patient043: n = 16 for endolysin-treated mice, n = 13 for vehicle-treated mice. Patient032: n = 8 for endolysin-treated mice, n = 8 for vehicle-treated mice. Source Data

Extended Data Fig. 1. Gut bacterial composition in patients with allo-HCT.

a,b, Gut bacterial composition at the genus level based on the relative abundance of operational taxonomic units for faecal samples from patients with allo-HCT. a, Cases with Enterococcus domination (defined as >25% of the genus Enterococcus) at any time point. b, Cases without Enterococcus domination. Source Data

Extended Data Fig. 2. Cytolysin-associated genes in E. faecalis and E. faecium strains.

a, Detection of E. faecalis and E. faecium strains isolated from faecal samples of allo-HCT patients by PCR. A representative image obtained from two independent experiments is shown. b, Detection of cylL_L, cylA, cylB, and cylM in each strain by PCR. A representative image obtained from two independent experiments is shown.

Extended Data Fig. 3. Analysis of the bacteriolytic capacity of endolysins against E. faecium.

Bacteriolytic capacity of endolysins against E. faecium isolated from the faeces of patients with allo-HCT. A representative experiment from two independent experiments is shown. Source Data

Extended Data Fig. 4. Analysis of the bacteriolytic capacity of endolysins against intestinal bacteria.

Bacteriolytic capacity of endolysins against S. epidermidis, K. pneumoniae, K. oxytoca, C. freundii and E. coli isolated from the faeces of patients with allo-HCT. A representative experiment from two independent experiments is shown. Source Data

Extended Data Fig. 5. Analysis of E. faecium CFU in colonized gnotobiotic mice.

CFU of E. faecium strain Patient038_35_1 and E. faecium strain Patient040_35_1 in faecal samples. Each dot represents the data from one faecal sample. n = 4 for endolysin-treated mice mono-colonized with E. faecium strain Patient038_35_1; n = 4 for vehicle-treated mice mono-colonized with E. faecium strain Patient038_35_1; n = 4 for endolysin-treated mice mono-colonized with E. faecium strain Patient040_35_1; and n = 4 for vehicle-treated mice mono-colonized with E. faecium strain Patient040_35_1. Significance was determined using the Wilcoxon rank-sum test (two-sided). Source Data

Extended Data Fig. 6. Analysis of the E. faecalis cylL_L gene copy number in faecal samples before and after transplantation, and IFNγ levels in the serum of E. faecalis-colonized gnotobiotic GVHD mice.

a, Each dot and line represents the data from one faecal sample. Significance was determined using the Wilcoxon rank-sum test (two-sided). Patient031: n = 10 for endolysin-treated mice, n = 9 for vehicle-treated mice. Patient043: n = 16 for endolysin-treated mice, n = 13 for vehicle-treated mice. Patient032: n = 8 for endolysin-treated mice, n = 8 for vehicle-treated mice. b, IFNγ production in serum. Each dot represents the data from one serum sample. Significance was determined using the Wilcoxon rank-sum test (two-sided). The numbers of endolysin-treated mice and vehicle-treated mice for each of the patient samples are as in a. Source Data

Extended Data Fig. 7. Analysis of survival rates in germ-free mice and in cytolysin-negative E. faecalis-, E. faecium- and E. coli-mono-colonized gnotobiotic GVHD mice.

Kaplan–Meier survival plots of germ-free mice (n = 7) and mice mono-colonized with E. faecalis strain JCM5803 (n = 8), E. faecium strain Patient019_14_1 (n = 8), and E. coli strain Patient025_0_122 (n = 7). Source Data

Extended Data Fig. 8. Analysis of the gut microbial composition and survival rates of humanized GVHD mice.

a, Gut microbial composition based on the relative abundance of operational taxonomic units at the genus level for the donor faeces and the faecal samples from humanized mice before and after transplantation. n = 8 samples for Patient026, and n = 8 samples for Patient062. b, Kaplan–Meier survival plots of mice colonized with faeces from Patient026 (n = 8) and mice colonized with faeces from Patient062 (n = 8). Source Data