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)1,2,3,4,5,6. 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 aGVHD7,8,9,10. 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.
Similar content being viewed by others
Main
Acute graft-versus-host disease (aGVHD) is a life-threatening complication of allogenic haematopoietic cell transplantation (allo-HCT) for haematological diseases11. Patients with allo-HCT undergo many clinical interventions, such as chemotherapy, radiation and antibiotic therapy, which can affect the pathogenesis of aGVHD12. Several lines of evidence suggest that the microbiome has a considerable effect on aGVHD1,2. In the early 1970s, mice undergoing allo-HCT in germ-free conditions were found to have a lower occurrence of GVHD than conventional mice3,4. With advances in intestinal microbial analyses, disease-related alterations of the intestinal environment—such as dysbiosis—have been reported to be closely associated with aGVHD symptoms in clinical studies5,6. Under dysbiosis, some symbiotic commensal bacteria acquire pathogenic characteristics, proliferate and become directly involved in the onset and progression of the disease13. Such bacteria are known as ‘pathobionts’, and examples include adherent invasive Escherichia coli in Crohn’s disease and Clostridium ramosum in obesity and diabetes mellitus14,15. The enterococci, which are Gram-positive, facultative anaerobes, are representative pathogens detected in patients. The species Enterococcus faecalis and Enterococcus faecium are prominent causes of multidrug-resistant infections16, and are responsible for bloodstream infections in patients with allo-HCT16,17. Previous studies have also revealed the high incidence of Enterococcus domination in the intestine, which has been associated with aGVHD occurrence and higher mortality rates among patients with allo-HCT7,8,9,10. In particular, one study showed that lactose-demanding Enterococcus proliferated in the intestine of patients with allo-HCT, and E. faecalis was associated with the severity of GVHD in gnotobiotic mouse models18. In this regard, eliminating enterococci, especially E. faecalis, in the intestine of patients with allo-HCT is crucial. Faecal microbiota transplantation (FMT) can help to resolve dysbiosis. However, the positive effects of FMT on aGVHD have not been confirmed19,20,21,22. In addition, the safety of FMT has become a serious concern because of the occurrence of drug-resistant bacteria and the potential transmission of norovirus infection by FMT23,24. Therefore, performing FMT in immunocompromised patients with aGVHD can be harmful. The use of antibiotics against intestinal pathobionts also risks promoting dysbiosis by the killing of beneficial bacteria. Thus, to attenuate the increase in enterococci, developing effective and safe methods to specifically manipulate these organisms is essential for suppressing the pathogenesis of aGVHD. Bacteriophage (phage) therapy is thought to be an alternative method for eliminating pathobionts because phages infect their host bacteria specifically. We previously showed that metagenome-analysis-based information about host bacteria–phage associations helped with the detection of phage-derived antibacterial enzymes that specifically regulate pathobionts25. These methods have promising applications for eliminating Gram-positive bacteria.
Here we present a strategy for suppressing E. faecalis-associated aGVHD using a bactericidal enzyme, endolysin. We longitudinally screened the intestinal bacteriome of patients with allo-HCT and isolated Enterococcus species from faecal samples. Using shotgun sequencing data, we identified an antibacterial enzyme derived from E. faecalis-specific phages. Our results suggest that administering an E. faecalis-specific antibacterial enzyme suppresses the induction of aGVHD.
Domination of Enterococcus in allo-HCT
Sixty-four transplant cases were at first enrolled in our study, but 18 were subsequently excluded because of early withdrawal from the study or rapid worsening of their condition. The remaining 46 transplant cases included in our analysis provided at least two sequential faecal samples. One patient was recruited into our study twice because of having undergone two transplants during our study period as a result of early leukaemia relapse after the first allo-HCT. The median age of the patients was 54.5 years (range: 19–72), and 25 (54.3%) patients were male. The patients had the following underlying diseases: 26 patients (56.5%) had acute leukaemia, 12 (26.1%) had myelodysplastic syndrome or myeloproliferative neoplasms, 7 (15.2%) had malignant lymphoma and one (2.2%) had another disease. With regard to transplants, 10 (21.7%) cases were transplanted from matched related donors, 7 (15.2%) were from matched unrelated donors, 11 (23.9%) were from cord blood, 17 (37.0%) were from haploidentical donors and one (2.2%) was from a human leukocyte antigen (HLA)-mismatched unrelated donor. Graft types included bone marrow in 8 (17.4%) cases, peripheral blood stem cells in 27 (58.7%) cases and cord blood in 11 (23.9%) cases. Myeloablative, reduced intensity and nonmyeloablative conditioning therapies were performed in 24 (52.2%), 21 (45.7%) and one (2.2) cases, respectively. Fourteen (30.4%) cases received total body irradiation as part of their conditioning therapeutic regimen. In total, fluoroquinolone, sulfamethoxazole–trimethoprim, isoniazid and macrolide were prophylactically administered in 44 (95.7%), 39 (84.5%), 3 (6.5%) and 2 (4.3%) cases, respectively, before the conditioning therapies. As empiric or target therapies, cefems (cefepim or cefozopran), piperacillin–tazobactam, carbapenems (meropenem or doripenem), glycopeptides (vancomycin or teicoplanin) or lipopeptides (daptomycin) and metronidazole were administered in 21 (45.7%), 26 (56.5%), 17 (37.0%), 28 (60.9%) and 3 (6.5%) cases, respectively, during the sample collection periods (Supplementary Table 1). The detailed characteristics of each case are summarized in Supplementary Table 2.
A total of 317 faecal samples were obtained from the 46 cases. We extracted DNA from all of these samples and analysed their intestinal microbial composition by 16S ribosomal RNA (rRNA) gene sequencing (Extended Data Fig. 1a,b). Enterococcus domination, which was defined as a relative abundance of the Enterococcus genus of more than 25% in a faecal sample, on the basis of previous studies7,18, was observed in 89 samples from 30 (65.2%) cases. Analysis of clinical predictors by univariate Cox proportional hazards regression revealed that acute leukaemia was significantly associated with a higher risk of Enterococcus domination than other diseases (hazard ratio, 2.48; 95% confidence interval, 1.13 to 5.45, P = 0.024) (Supplementary Table 3). No other clinical variables were significantly associated with the cumulative incidence of Enterococcus domination. Thus, consistent with previous research7,18, Enterococcus domination was characterized in our patients with allo-HCT.
Characterization of E. faecalis
We next isolated 30 enterococci from faecal samples characterized by Enterococcus domination, and performed metagenomic analysis. We identified 11 strains of E. faecalis and 19 strains of E. faecium (Extended Data Fig. 2a). Several studies have shown that the colonization of multidrug-resistant bacteria, such as vancomycin-resistant enterococci (VRE), correlated with increased mortality among patients with allo-HCT17. We therefore performed antimicrobial susceptibility testing for the 30 isolated enterococci. Notably, all E. faecalis and E. faecium strains were sensitive to vancomycin, teicoplanin, daptomycin and linezolid, indicating that these bacterial strains were not VRE, and were consistent with the standard epidemiological features of enterococcal strains isolated in Japan26 (Extended Data Table 1). Despite the multidrug resistance of enterococci, cytolysin, which is an exocrine protein that is associated with haemolysin, contributes to the severity of enterococcal disease in humans27,28,29 and a number of animal models30,31,32,33. We detected cytolysin-associated genes, such as cylLL, cylA, cylB and cylM, in all E. faecalis strains, but not in E. faecium strains (Extended Data Fig. 2b). In addition, all of the isolated cytolysin-positive E. faecalis strains had functional haemolytic activity (Fig. 1a and Supplementary Fig. 1), indicating that highly virulent E. faecalis was dominant in patients with allo-HCT. Furthermore, although Kyoto Encyclopaedia of Genes and Genomes (KEGG) orthology (KO) terms related to biofilm formation were identified for both species, some terms, such as K01791, K07173, K05946 and K08605 were more frequently detected in E. faecalis (Fig. 1b). Because it is estimated that biofilm-forming bacteria are 10 to 1,000 times more resistant to antibiotics than are planktonic bacteria34,35, we hypothesized that cytolysin-positive virulent E. faecalis survive and proliferate by forming biofilms in the intestine of patients with allo-HCT, and thereby resist the effects of the various antibiotics administered during the treatment process.
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.
A phage-derived anti-bactericidal enzyme
Because E. faecalis, which expands in the intestine of patients with allo-HCT, is considered to be one of the pathobionts associated with the severity of aGVHD, specific elimination of this organism is desirable to suppress aGVHD8,9,18. However, biofilms exhibit strong resistance to antibiotics through mechanisms such as reduced antibiotic permeability, sequestration of antibiotics and the presence of persister cells36,37,38. Hence, it is extremely difficult to eliminate E. faecalis with existing antibiotics. Endolysins are hydrolytic enzymes produced by phages that dissolve the bacterial cell wall to release progeny phages at the end of the replication cycle39. These enzymes hydrolyse the peptidoglycan layer and kill the host bacteria40. Compared with conventional broad-spectrum antibiotics, endolysins confer advantages such as high host specificity, rapid lysis of target bacteria, a low risk of resistance, synergy with different antimicrobial agents and the ability to function effectively on biofilms and mucosal surfaces41,42,43,44,45,46. In a previous study, we successfully identified novel endolysins in Clostridioides difficile prophage sequences from metagenomic data, and confirmed their bactericidal activity against C. difficile both in vitro and in vivo25. Therefore, we next analysed shotgun sequence data relating to E. faecalis strains (Fig. 1a and Extended Data Fig. 2a,b), and extracted prophage sequences. Up to four prophage sequences classified as Siphoviridae or Podoviridae were identified from each set of sequence data (Fig. 2a). Open reading frames (ORFs) homologous to known endolysins in prophage genomes were screened (Fig. 2b). We identified nine endolysins from the metagenomic data for the E. faecalis strains shown in Fig. 1a–c, all of which were annotated as WP_002399372 (E value = 0) (Supplementary Fig. 2). In general, endolysins consist of two domains—an enzymatically active domain and a cell-wall-binding domain—that recognize specific cell-surface features. The enzymatically active domain of the detected endolysins was annotated as GH25 (glycoside hydrolase family 25)_LysA-like, which is an endo-N-acetylmuramidase (muramidase) that degrades bacterial cell walls by catalysing the hydrolysis of 1,4-β-linkages between N-acetylmuramic acid and N-acetyl-d-glucosamine residues. The cell-wall-binding domain of the detected endolysins was annotated as ZoocinA_TRD (target recognition domain). We generated a His-SUMO-tagged endolysin (Fig. 2c and Supplementary Fig. 3), and found that it had effective lytic activities in vitro (Fig. 2d and Supplementary Fig. 4). In addition, the biofilms from all isolated E. faecalis strains in this study were successfully lysed by the endolysin (Fig. 2e). Notably, the endolysin did not have effective lytic activity against E. faecium or other intestinal bacteria such as Staphylococcus epidermidis, Klebsiella pneumoniae, Klebsiella oxytoca, Citrobacter freundii and E. coli isolated from the faeces of patients with allo-HCT (Extended Data Figs. 3–5), which suggests that the endolysin is a narrow-spectrum enzyme. Furthermore, oral administration of the endolysin in gnotobiotic C57BL/6 mice that were mono-colonized with E. faecalis resulted in the effective lysis of E. faecalis-derived biofilms in the intestine (Fig. 2f and Supplementary Fig. 5). These data suggest that the endolysin detected in the E. faecalis prophage sequence from the metagenome data has the potential to lyse the biofilms formed by E. faecalis in the intestine.
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. OD600 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. OD570 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).
Suppression of aGVHD
To investigate the potential effectiveness of administering the purified endolysin against E. faecalis as an aGVHD suppression strategy, we used gnotobiotic C57BL/6 mice mono-colonized with E. faecalis (Supplementary Fig. 6a). After administration of busulfan and cyclophosphamide, bone marrow cells and splenic T cells were intravenously injected into the gnotobiotic mice to induce aGVHD (Supplementary Fig. 6a). The levels of E. faecalis in the faeces of the endolysin-treated gnotobiotic mice were significantly lower than those in the vehicle-treated gnotobiotic mice (Fig. 3a). Furthermore, the endolysin was effective against aGVHD (Fig. 3b). These findings indicated that the E. faecalis-derived endolysin regulated aGVHD by reducing the number of intestinal E. faecalis. We also analysed GVHD-induced gnotobiotic C57BL/6 mice colonized with a human faecal sample that was characterized by the domination of Enterococcus (Fig. 4a and Supplementary Fig. 6b). Levels of the cytolysin-associated gene cylLL in the faeces of the endolysin-treated humanized mice were consistently and significantly lower than those of the vehicle-treated humanized mice (Extended Data Fig. 6a). Before (pre-transplantation and day 0) and after (day 8 and day 15) transplantation, the microbial communities showed no consistent significant differences between the two groups (Fig. 4b,c). However, the endolysin-treated humanized mice showed significantly decreased levels of interferon gamma (IFNγ) in the serum (Extended Data Fig. 6b) and an increased survival rate compared with the vehicle-treated humanized mice (Fig. 4d). Together, these data suggest that the endolysin detected in the E. faecalis prophage sequences from the metagenome data has potential clinical applications for aGVHD.
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.
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.
Discussion
In this study, we analysed the intestinal bacteriome of patients with allo-HCT and revealed the domination of Enterococcus. An endolysin identified from E. faecalis prophage sequencing data was shown to lyse biofilm-forming E. faecalis. Administration of this endolysin eradicated cytolysin-positive virulent E. faecalis in the intestine and suppressed the exacerbation of aGVHD.
In our study, Enterococcus domination was observed in 65.2% of patients with allo-HCT (Extended Data Fig. 1a,b), and acute leukaemia was significantly associated with a high incidence of Enterococcus domination (Supplementary Table 3), consistent with a previous study7. Because patients with acute leukaemia are administered more cycles of highly intensified chemotherapy and broad-spectrum antibiotics before allo-HCT than are patients with other diseases, in accordance with previous studies47, gut dysbiosis and dysfunction of the intestinal barrier might have been induced. A previous report showed that bacteria living in biofilms are considerably more resistant to antibiotics than are planktonic bacteria48; therefore, biofilm-forming enterococci might have been able to remain dominant in the intestinal tracts of patients with allo-HCT.
Antimicrobial susceptibility testing confirmed that all isolated E. faecalis and E. faecium strains in this study were not VRE (Extended Data Table 1). In addition, VRE strains were not isolated through the culturing of 223 faecal samples dominated by Enterococcus on VRE-selective agar. Thus, it is unlikely that acquiring resistance to glycopeptide or oxazolidine antibiotics—which are the antibiotics most frequently administered during allo-HCT to prevent infection—contributes to the development of Enterococcus domination in the intestinal tract. However, all isolated E. faecalis strains, but not E. faecium strains, possessed cytolysin genes (Extended Data Fig. 2b), indicating that these bacteria were highly virulent. In fact, a high incidence of Enterococcus domination in the intestine has been shown to correlate with aGVHD occurrence and the higher mortality rates among patients with allo-HCT18. Therefore, inhibiting biofilm formation and eliminating cytolytic E. faecalis would be an effective clinical approach in patients with allo-HCT.
Characterization of the isolated E. faecalis strains from patients with allo-HCT revealed an increase in the levels of several KO terms associated with biofilm formation, including K01791, K05496, K07173 and K08605, compared with the E. faecium strains (Fig. 1b). The ability of E. faecalis to readily colonize the naive gut epithelium by forming biofilms49 has been previously shown in a germ-free mouse model. K01791 and K05496 are KO terms associated with the wecB and tagA genes, respectively, which play a dedicated role in polysaccharide biosynthesis50. K07173 is a KO term associated with the luxS genes, which are involved in quorum sensing and biofilm formation51,52. K08605 is a KO term associated with the gelE genes, which are essential for producing gelatinase and contribute to biofilm formation53,54. A previous study showed that the metalloprotease GelE produced by E. faecalis impaired the function of the intestinal epithelial barrier and induced chronic inflammation in IL-10−/− and TNFΔARE/Wt mice55, which suggests that, in addition to cytolysin, GelE might be a clinical target for aGVHD.
Because E. faecalis is a Gram-positive bacterium, it was predicted that endolysins, which specifically hydrolyse peptidoglycan and are involved in the formation of biofilms, would decrease the domination of E. faecalis in the intestine. Several studies have reported the identification of endolysins from meta-analyses of phage sequences25,56. In our previous study, we showed that metagenome-analysis-based information about host bacteria–phage associations is useful for detecting phage-derived antibacterial enzymes that specifically control pathobionts25. In our current study, we detect a unique endolysin from E. faecalis prophage sequencing data (Fig. 2b and Supplementary Fig. 2), and provide evidence of its lytic activity against E. faecalis strains—but not E. faecium strains—isolated from patients with allo-HCT (Fig. 2d and Extended Data Fig. 3). Notably, the identified endolysin eradicated the formation of biofilms (Fig. 2e,f) and protected against aGVHD (Figs. 3b and 4d). Furthermore, aGVHD-related mortality rates were not increased in cases associated with the cytolysin-negative E. faecalis strain JCM5803, E. faecium strain Patient019_4_1, and E. coli strain Patient025_0_122 (Extended Data Fig. 7). Future research will be required to determine whether this therapeutic approach is effective for patients with aGVHD.
When considering the possible clinical applications of the endolysin against E. faecalis, specificity of the enzyme for E. faecalis would be a desirable characteristic. Indeed, the lytic spectrum of an endolysin ranges from species-specific to targeting multiple genera. A previous study showed that enterococcal phage lysins (for example, PlyV12 and EFAP-1) have a broad-spectrum killing ability57. However, in our study, the endolysin lysed E. faecalis, but not E. faecium or other intestinal bacteria (Fig. 2d–f and Extended Data Figs. 3–5). Therefore, as well as the E. faecalis-specific endolysin LysEF-P10 previously reported58, our identified endolysin might be a narrow-spectrum enzyme, which would make it an ideal candidate to eliminate E. faecalis.
In the present study, of the 30 Enterococcus-dominant cases, 7 were aGVHD-positive and cytolysin-positive, 4 were aGVHD-positive and cytolysin-negative, 3 were aGVHD-negative and cytolysin-positive and 16 were aGVHD-negative and cytolysin-negative (P < 0.0147 by two-tailed Fisher’s exact test) (Supplementary Table 2). Although the sample size is small, our study is the first—to our knowledge—to show not only an experimental but also a clinical association between faecal cytolysin positivity and the development of aGVHD. Thus, the relationship between the cytolysin positivity of E. faecalis and the development of aGVHD in patients with allo-HCT should be a key consideration in the design of therapeutic strategies.
This study has several limitations. To investigate the effect of endolysin on aGVHD, but not chronic GVHD, we administered endolysin for only three weeks and observed survival rates for one month. Future studies could assess whether cytolysin-positive E. faecalis is effective against chronic GVHD. In addition, the small sample size of our current cohort study precludes a detailed investigation of cytolysin-positive E. faecalis and the severity of aGVHD in patients with allo-HCT. However, according to previous reports, E. faecalis-derived cytolysin lyses bacterial cells, as well as eukaryotic cells, in response to quorum-sensing signals59, indicating that cytolysin-positive E. faecalis, not cytolysin-negative E. faecalis, has a greater ability to proliferate as a result of the suppression of competing bacteria in the intestinal tract of patients with allo-HCT. In the present study, aGVHD-related mortality was increased in cases with cytolysin-positive E. faecalis and an Enterococcus-dominated gut microbiome, but not in cases with cytolysin-negative E. faecalis or a gut microbiome that is not dominated by Enterococcus (Fig. 3b and Extended Data Figs. 7 and 8). Future studies will be required to address whether cytolysin-positive E. faecalis is involved in the pathogenesis of aGVHD in patients with allo-HCT. Furthermore, although we did not detect VRE in all enterococci in this study, the presence of VRE has been shown to be inevitable in patients with allo-HCT16,17. Future studies are required to address whether E. faecalis-derived endolysin is effective against VRE.
In conclusion, we report a new strategy for regulating aGVHD that uses an endolysin derived from E. faecalis. As well as aGVHD, cytolytic E. faecalis in the intestine has been shown to be associated with the severity of liver disease and with mortality in patients with alcoholic hepatitis60. E. faecalis also causes other antibiotic-refractory infections, such as infective endocarditis, endophthalmitis and catheter-related urinary tract infections, by forming biofilms. Thus, our findings raise the possibility of developing new treatments for cytolytic E. faecalis-associated diseases, represented by aGVHD.
Methods
Patient cohort
An observational cohort consisting of allo-HCT recipients was recruited from patients in the Department of Haematology at Osaka Metropolitan University Hospital. Patients with haematological diseases who were undergoing allo-HCT from January 2019 to June 2020 were included. For each patient, sequential faecal samples were collected before stem cell transplantation (within 14 days before the initial date of conditioning therapy), and then weekly (at day 0 ± 3, day 7 ± 3, day 14 ± 3 and so on) until day 98 ± 3, or hospital discharge. The date of stem cell transfusion was defined as day 0. Sample collection was skipped when patients could not provide stool specimens because their general condition had worsened. This protocol was approved by the Ethics Committee of Osaka Metropolitan University and the Institute of Medical Science, The University of Tokyo, and signed informed consent was obtained from each participant (4188 and 30-92-B0320).
Antibiotics in patients with allo-HCT
Prophylactic fluoroquinolone and sulfamethoxazole–trimethoprim were started before conditioning therapy, except in patients with drug allergies or who had already received systemic antibiotics. Isoniazid or macrolide were given in addition to patients with latent tuberculosis infection or chronic sinusitis, respectively. Sulfamethoxazole–trimethoprim was stopped two days before stem cell transplantation, in consideration of its suppressive effect on the bone marrow. Our antibiotic administration strategy was based on the guidelines of the Infectious Diseases Society of America61, which specified that in the case of febrile neutropenia (defined as an axillary temperature higher than 37.5 °C and a neutrophil count lower than 0.5 × 109 per litre), fluoroquinolone was switched to a cefem or to piperacillin–tazobactam as the first-line therapy. When fever persisted, the antibiotics were further changed to other agents, such as carbapenems, as the second-line treatment. Glycopeptides or lipopeptides were also administered when a catheter-related infection, a skin or soft-tissue infection, pneumonia or haemodynamic instability was suspected. In addition, targeted antibiotics were administered according to the culture-based results or clinical signs, such as metronidazole for C. difficile enteritis.
Extraction of bacterial DNA
After collection, the patient faecal samples were immediately stored at 4 °C under anaerobic conditions, shipped to the laboratory, then divided and stored in RNA later (Invitrogen), 40% glycerol and anaerobic culture medium (2% Lab-Lemco powder (Kanto Chemical Co.), 0.1% l-cysteine (Nacalai Tesque), 0.045% KH2PO4 (Nacalai Tesque), 0.09% NaCl (Nacalai Tesque), 0.045% [NH4]2SO4 (Nacalai Tesque), 0.0045% CaCl2 (Nacalai Tesque), 0.0045% MgSO4 (Nacalai Tesque) and 40% glycerol (Nacalai Tesque) in 1 l distilled water) at a 16-fold dilution (w/v) at −80 °C until use. Bacterial DNA extraction from the faecal samples was performed as previously described25. In brief, faecal samples stored in RNA later were homogenized in 1 ml SM-plus buffer (100 mM NaCl, 50 mM Tris-HCl (pH 7.4) (Nacalai Tesque), 8 mM MgSO4·7H2O (Nacalai Tesque), 5 mM CaCl·2H2O (Nacalai Tesque) and 0.01% (w/v) gelatine in distilled water) by vortex mixing, and were then passed through a 100-µm cell strainer. SM-plus buffer was passed through a 0.22-µm syringe filter before use at each step. To extract DNA, the samples were incubated with 1 ml SM-plus buffer containing 20 mM EDTA (Nacalai Tesque), 100 µg ml−1 recombinant human lysozyme (Sigma-Aldrich) and 0.5 U ml−1 achromopeptidase (Wako Pure Chemicals) at 37 °C for 1 h. Then, supernatant samples were further incubated with a 1/400 volume of 20 mg ml−1 proteinase K (Nacalai Tesque) and a 1/20 volume of 10% sodium dodecyl sulfate (Nacalai Tesque) at 55 °C for 1 h. Next, the samples were added to an equal volume of phenol–chloroform–isoamyl alcohol (Nacalai Tesque) and mixed vigorously. After centrifugation at 16,000g for 5 min, the aqueous phase of the samples was transferred to new tubes, followed by chloroform (Nacalai Tesque) extraction. Again, the aqueous phase of the samples was transferred to new tubes and mixed with a 1/10 volume of 3 M sodium acetate (Nacalai Tesque) and an equal volume of isopropanol (Nacalai Tesque). The samples were centrifuged at 16,000g for 15 min. After discarding the supernatants, the pellets were washed with 70% ethanol (Nacalai Tesque) and centrifuged at 16,000g for 5 min. The supernatants were removed completely, and the pellets were air-dried for 5 min. The DNA was resuspended in 10 mM Tris-HCl buffer (pH 8.0) (Nacalai Tesque).
16S rRNA gene analysis
The 16S rRNA V3–V4 region was PCR-amplified using specific primers (forward primer: 5′-ACACGACGCTCTTCCGATCTCCTACGGGNGGCWGCAG-3′, reverse primer: 5′-GACGTGTGCTCTTCCGATCTGACTACHVGGGTATCTAATCC-3′, underline: overhang sequence for second-round PCR) over 20 cycles, and the PCR products were purified with Agencourt AMpure beads (Beckman Coulter) as described previously15. Then, the overhang and index sequences required for sequencing were added by a second round of PCR using NEBNext multiplex Oligos for Illumina (Dual Index Primers Set1, New England Biolabs) over eight cycles, and the products were purified with Agencourt AMpure beads. For each sample, equal amounts of each DNA amplicon library were mixed and sequenced on an MiSeq instrument (Illumina) using the MiSeq v3 Reagent kit and a 15% PhiX spike (Illumina). 16S rRNA gene analysis was performed using QIIME2 (v.2018.11; https://qiime2.org). In brief, raw sequence data were subjected to primer sequence trimming, quality filtering and paired-end read merging using the dada2 denoise-paired method (–p-trim-left-f 17 –p-trim-left-r 21–p-trunc-len-f 275 –p-trunc-len-r 215 –p-n-threads 16). Before taxonomic analysis, sequences of the 16S rRNA V3–V4 region were extracted from Greengenes 13_8 99% operational taxonomic units and our primer sequences using the q2-feature-classifier. Then, the Naive Bayes classifier was trained using the extracted Greengenes 13_8 reference sequences and Greengenes 13_8 99% operational taxonomic unit taxonomy. The taxonomic composition was visualized using a qiime taxa bar plot.
Predicting Enterococcus domination
Predictive factors for Enterococcus domination, which was defined as a microbiota containing more than 25% of the genus Enterococcus, were examined using Cox proportional hazards regression. The first occurrence of Enterococcus domination in each case was defined as the end-point of interest. The following clinical variables were assessed as univariate predictors in each patient: age, sex, underlying diagnosis (acute leukaemia versus other diseases), conditioning regimen intensity (myeloablative conditioning versus reduced intensity conditioning), whether or not total body irradiation was included in the conditioning regimen, donor type (HLA-matched related donor versus HLA-matched unrelated donor, cord blood or haploidentical donor), graft type (bone marrow versus peripheral blood stem cell or cord blood) and antibiotic administration during the sample collection period (from the start of conditioning therapy to the final faecal sample collection in each case). The intensity of the conditioning regimen was defined according to a previous report from the American Society for Blood and Marrow Transplantation62. All statistical analyses were performed using EZR v.1.41 (Saitama Medical Center, Jichi Medical University), which is a graphical user interface for R63. P < 0.05 was considered significant.
Isolation of intestinal bacteria
Enterococci were isolated by culturing faecal samples with Enterococcus domination on Enterococcus selective agar medium (brain–heart infusion (BHI) broth (BD) supplemented with 20 µg ml−1 aztreonam (Nacalai Tesque), 20 µg ml−1 polymyxin B (Nacalai Tesque), 4 µg ml−1 amphotericin B (Nacalai Tesque) and 50 µg ml−1 triphenyltetrazolium chloride (Nacalai Tesque)), or vancomycin-resistant Enterococci-selective agar (BD). After culturing at 37 °C aerobically for 24–48 h, colonies were picked and further cultured in BHI broth. After growth, the bacteria were stocked at −80 °C with 80% glycerol. DNA was extracted using the PowerSoil DNA Isolation Kit (Qiagen) in accordance with the manufacturer’s protocol. The species was determined for each strain by PCR using previously designed primers: 5′-ATCAAGTACAGTTAGTCT-3′ and 5′-ACGATTCAAAGCTAACTG-3′ for E. faecalis (product size: 941 bp), and 5′-TTGAGGCAGACCAGATTGACG-3′ and 5′-TATGACAGCGACTCCGATTCC-3′ for E. faecium (658 bp)64,65. PCR amplicons were analysed using MCE-202 MultiNA (Shimadzu).
S. epidermidis, K. pneumoniae, K. oxytoca, C. freundii and E. coli were isolated by culturing faecal samples with non-Enterococcus domination on Luria–Bertani (LB) agar medium (Nacalai Tesque). After culturing at 37 °C aerobically for 24–48 h, colonies were picked and further cultured in LB broth. After growth, the bacteria were stocked at −80 °C with 80% glycerol. DNA was extracted using the PowerSoil DNA Isolation Kit (Qiagen) in accordance with the manufacturer’s protocol. The species was determined for each strain by PCR and Sanger sequencing using previously designed primers: 5′-ACACGACGCTCTTCCGATCTCCTACGGGNGGCWGCAG-3′ and 5′-GACGTGTGCTCTTCCGATCTGACTACHVGGGTATCTAATCC-3′.
Detection and quantification of cytolysin genes
Screening for cytolysin genes (cylLL, cylM, cylB and cylA) was performed by PCR using previously developed primers based on GenBank nucleotide sequences for the E. faecalis cytolysin operon (accession no. L37110): 5′-GATGGAGGGTAAGAATTATGG-3′ and 5′-GCTTCACCTCACTAAGTTTTATAG-3′ for cylLL (product size: 253 bp), 5′-AAAAGGAGTGCTTACATGGAAGAT-3′ and 5′-CATAACCCACACCACTGATTCC-3′ for cylM (2,940 bp), 5′-AAGTACACTAGTAGAACTAAGGGA-3′ and 5′-ACAGTGAACGATATAACTCGCTATT-3′ for cylB (2,020 bp) and 5′-TAGCGAGTTATATCGTTCACTGTA-3′ and 5′-CTCACCTCTTTGTATTTAAGCATG-3′ for cylA (1,282 bp)66. PCR amplicons were analysed using MCE-202 MultiNA.
The quantity of cylLL PCR amplicons was determined using MCE-202 MultiNA and cylLL copy numbers were analysed. One nanogram of faecal DNA was used as template DNA.
Haemolytic activity assays
Culture medium of each E. faecalis or E. faecium strain was streaked on Columbia agar plates supplemented with 5% horse blood (Nissui Pharmaceutical Co.). After aerobic incubation at 37 °C for 72 h, the presence of zones of clearing around the colonies was determined to indicate beta-haemolysis.
Antimicrobial susceptibility testing
A suspension of each isolated E. faecalis or E. faecium strain prepared to 1 McFarland standard with physiological saline was diluted 500 times in BHI broth, and then cultured aerobically in the solid-phase wells of an Eiken DP42 dry plate (Eiken Chemical Co.) for 24 h. The minimum inhibitory concentrations were determined according to the manufacturer’s instructions.
Whole-genome sequencing
Bacterial DNA sequencing
DNA libraries were prepared with the KAPA HyperPlus kit (KAPA Biosystems) following the manufacturer’s instructions, except that NEBNext multiplex Oligos were used for Illumina (New England Biolabs) at the adapter ligation and barcoding steps. The prepared target library size was 450–550 nucleotides. The concentration of the library was quantified with the KAPA Illumina Library Quantification kit (KAPA Biosystems) and adjusted to the mean library size measured by MCE-202 MultiNA. The libraries were pooled and sequenced on a HiSeq2500 sequencer (2 × 250 paired-end reads, HiSeq Rapid SBS Kit v2; Illumina). For each run, 10 libraries for the bacterial fractions were pooled at equimolar concentrations.
Processing of sequencing data
Sequencing reads were demultiplexed using Illumina CASAVA software and were processed with the following three steps: (1) adaptor sequence trimming; (2) nucleotide trimming and removal of duplicates; and (3) error base correction. First, adaptor sequences were removed by cutadapt software (http://cutadapt.readthedocs.io/en/stable/index.html) (v.1.18). Second, the first and last 10 nucleotides of each read were removed, then within 20 nucleotides from both ends, low-quality nucleotides with a Phred quality score of less than 20 were trimmed, and polynucleotides at the end of the sequence were also trimmed. After trimming, sequences shorter than 75 nucleotides, low-complexity sequences (DUST score greater than 7), exact duplicates, sequences containing N and singletons were filtered out. This step was performed by PRINSEQ software (http://prinseq.sourceforge.net/) (lite v.0.20.4) (-trim_right 10 -trim_left 10 -trim_qual_right 20 -trim_qual_left 20 -trim_qual_window 20 -trim_ns_right 1 -min_len 75 -lc_method dust -lc_threshold 7-ns_max_n 0 -derep 1). Third, correction of sequencing errors on the basis of the Hamming graph and Bayesian subclustering were performed using BayesHammer software (as bundled with SPAdes v.3.13.0) (spades.py –only-error-correction)67.
Metagenome assembly
The quality-filtered and error-corrected reads in each sample were assembled with MetaSPAdes v.3.13.0 with default k-mer lengths (options: --meta --only-assembler)68. To compare the abundance of contigs across the samples, the assembled contigs (with lengths ≥ 5 kb) from individual samples were pooled. CD-HIT-EST (v.4.8) was used to cluster pooled contigs at a 95% global average nucleotide identity (-c 0.95 -G 1 -n 10 -mask NX)69. From the non-redundant pooled contigs, circular contigs were identified by detecting overlaps in the 5′ and 3′ end sequences (more than 50 nucleotides overlap at 100% identity) of the contigs using Megablast (BLAST+ v.2.5)70. Each of the detected circular contigs was trimmed to remove redundant parts. Circular contigs longer than 1.5 kb and linear contigs longer than 5 kb were used for the analyses.
Bacterial taxonomic assignment
We assigned bacterial taxonomy to the contigs using PhyloPythiaS+ (v.1.4)71. The whole PhyloPythiaS+ pipeline was run (options: -n -g -o s16 mg -t -p c -r -s) using their reference database ‘NCBI201502’ with the configuration parameters: maxLeafllClades = 500 and minPercentInLeaf = 0.05 (the other parameters were set as default). To obtain the taxonomic profile in each sample, the quality-filtered and error-corrected reads were mapped to microbial taxonomy-specific marker genes with MetaPhlAn2.0 (v.2.5.0)72.
Read coverage calculation of contigs
The quality-filtered and error-corrected reads were mapped to the non-redundant pooled contigs using the bbmap (v.38.76) tool from BBtools (v.37.68) with at least 95% identity and the ambiguous mapping option (ambiguous = random). A contig was considered as ‘detected’ in a sample if more than 75% of the contig length was covered by mapped reads, as recommended in a previous study73. The abundance of a contig was calculated as the average contig coverage (number of nucleotides mapped to the contig divided by the contig length), where the abundance of a non ‘detected’ contig was set to 0, and normalized by the total number of nucleotides of the mapped reads in a sample, to obtain a total number of nucleotides equal to 109.
Statistical analysis of gene functions
The detected ORFs in each sample were annotated according to KEGG (16 September 2018) prokaryote genes and corresponding KOs using the most significant hit (E value < 1 × 10−5 and bitscore > 50). The abundance of each KO term was calculated as the sum of per base read depths of the ORFs assigned to the KO divided by the sum of the lengths of these ORFs. The read depths were counted by samtools bedcov. We calculated the median abundances of 31 KOs considered to be the single copy genes (K01889, K01890, K02528, K02600, K02863, K02864, K02867, K02871, K02874, K02876, K02881, K02886, K02890, K02892, K02895, K02906, K02931, K02933, K02946, K02948, K02950, K02952, K02956, K02961, K02965, K02967, K02982, K02988, K02992, K02994 and K02996) for each sample and performed normalization by dividing the abundances of each KO by the median abundance of this set. Samples with a median abundance of zero were discarded from the analysis. The Wilcoxon rank-sum test was performed using the R packages coin, exactRankTests and qvalue. A volcano plot was drawn using the ggplot2 (v.3.3.6) R package.
Analysis of prophages
Prophage sequences in the bacterial contigs were predicted according to known viral signatures with VirSorter (v.1.0.3)74. The bacterial contigs (≥5 kb) from individual samples were analysed by VirSorter using both RefSeqABVir (–db 1) and Viromes (–db 2). To remove virus-derived sequences from the bacterial contigs, predicted prophage sequences of VirSorter categories 4 or 5 (presence of viral hallmark genes or enrichment of viral-like genes in a prophage region) were extracted. The positions of the predicted prophage sequences on the non-redundant pooled bacterial contigs were obtained through Megablast (BLAST+ v.2.5) searches (E value < 10−100 and at least 95% identity), and prophage sequences were merged if their positions overlapped. Finally, prophage sequences longer than 3 kb were extracted and listed as candidate prophage sequences.
Viral nucleotide and protein database
To classify the prophage contigs, we prepared viral genome and protein databases. The viral RefSeq sequences (released as of 5 May 2020; 366,089 proteins) were downloaded from the NCBI FTP site (https://ftp.ncbi.nlm.nih.gov/refseq/release/viral/). Taxonomic lineage information was assigned against NCBI RefSeq nucleotide sequence data, which were searched using the term “taxid = 28883 (Caudovirales)” on the NCBI website (https://www.ncbi.nlm.nih.gov/nuccore/) with 57,556 nucleotide hits, and then all sequences were downloaded as fasta files on 24 October 2021. The ORFs for the rest of the viral contigs were predicted for classification as described below.
Viral classification
We first classified prophage contigs using viral RefSeq genomes. The prophage contig sequences were searched against the viral RefSeq data mentioned above (‘Viral nucleotide and protein database’) using blastn (BLAST+ v.2.5). The results were sorted by value (minimum E value, maximum bitscore and maximum Query Cover), and the top three results were filtered. Because some viral contig sequences contained multiple fragments that were mapped to different viral genomes, the top three significant alignments were considered to determine viral taxonomy. If the top three alignments covered more than 50% of a contig sequence, the lowest common ancestor of the top three hits was determined using blast2lca (v.0.800) (https://github.com/emepyc/Blast2lca) (modified to use accession version identifiers) with the NCBI taxonomy (downloaded on 30 November 2017), and was assigned to the contig.
Gene prediction
The ORFs on the contigs were predicted using MetaProdigal (v.2.6.3) with the metagenomics procedure (-p meta)75. To predict genes spanning the 3′ to 5′ ends of a circular contig, a temporary version of the circular contig was used in the ORF prediction, in which the first 1,500 nucleotides were duplicated and added at the end of the contig.
ORF classification
To annotate the predicted ORFs, the amino acid sequences of the ORFs were queried by GHOST-MP (v.1.3.4) against the RefSeq viral protein (released as of 5 May 2020) with an E value < 10−30 and a bitscore > 50 (ref. 76). The viral RefSeq proteins with the top three closest homologies (E value < 10−5 and bitscore > 50) were considered for each ORF. With regard to the prophage region, the ORFs were classified according to their PHROG category77. The PHROG database (v.4) was downloaded from https://phrogs.lmge.uca.fr/ as fasta files, then converted into a Blast database. The prophage ORFs were annotated using blastp, then the annotated ORFs were described using Circos78 and coloured according to their PHROG category.
Analysis of phage-derived endolysins
The ORFs that putatively encode endolysin proteins were extracted from the prophage regions found in the 11 E. faecalis strains using their ORF annotations. The viral RefSeq protein annotations containing “Endolysin”/“endolysin” were extracted using blastp (BLAST+ v.2.5) and searched against the RefSeq bacterial protein database downloaded from the NCBI ftp site (https://ftp.ncbi.nlm.nih.gov/refseq/release/bacteria/) on 17 January 2021, including non-redundant proteins. To avoid false positives, the ORFs were annotated with a significant E value < 10−30. Multiple alignments of the protein sequences in each group were performed with MUSCLE (v.3.8.31) with default settings79. Jalview (v.2.11.3.2) was used to visualize the multiple alignment results80. To find the domains of the detected proteins, we searched the accession number on the NCBI website (https://www.ncbi.nlm.nih.gov/protein/) on 25 January 2022.
Gene prediction was performed as described above and each gene is shown in Extended Data Fig. 2. His-SUMO-tagged endolysins were obtained by PCR using genomic DNA from the E. faecalis strain obtained from Patient031_14_8 as a template and were ligated into the BamHI and SalI sites of the pCold-SUMO expression vector (Creative Biogene)81. The primer sequences were as follows: forward primer, 5′-AAGGATCCATGACATTAAACGGAATTGA-3′ and reverse primer, 5′-AAGTCGACTTACCCATAGATTAATTTCT-3′. The plasmids were then transformed into BL21 (DE3) cells, which were grown in LB medium containing 100 µg ml−1 ampicillin at 37 °C until the optical density at 600 nm (OD600 nm) reached 0.4–0.6. Isopropyl β-d-thiogalactoside (IPTG) (Nacalai Tesque) was added until the final concentration reached 1 mM, and the culture was incubated at 16 °C for 18 h at 120 rpm. After centrifugation at 3,000g for 15 min, the pellet was washed with sterile deionized water. After centrifugation at 3,000g for 15 min, the precipitate was resuspended in xTractor buffer (TaKaRa). The lysate was incubated with 10 U ml−1 of DNase I (Roche) and 100 mg ml−1 of hLysozyme (Sigma-Aldrich) for 30 min, and disrupted by sonication (20 s pulse, 80 s rest, over 10 min). After centrifugation at 12,000g for 30 min, the supernatant was sterile-filtered through 0.45- and 0.2-µm filters. The target proteins were purified through Capturem His-Tagged Purification Maxiprep columns (TaKaRa). The lysate was loaded onto the column, which was equilibrated with xTractor buffer, and then centrifuged at 2,000g for 3 min at room temperature. The column was washed with wash buffer (20 mM Na3PO4 (Nacalai Tesque), 150 mM NaCl, pH 7.6), and the target protein was eluted with elution buffer (20 mM Na3PO4, 500 mM NaCl, 500 mM imidazole, pH 7.6 (Nacalai Tesque)).
All samples were desalted by HiTrap desalting columns (Cytiva) with HiTrap buffer (50 mM Na3PO4, 0.15 M NaCl, pH 7.0). The protein solution was loaded into the concentration tube (Amicon Ultra-15 10K; Merck) and centrifuged at 5,000g for 20 min at room temperature. The concentration of the target protein was measured using Protein Assay CBB Solution (Nacalai Tesque).
Western blotting
The samples collected were purified and concentrated as described above, then subjected to 5–20% SDS–PAGE. The resolved proteins were transferred to a polyvinyl difluoride membrane (Bio-Rad), which was incubated with anti-His antibody (1:2,000) (Proteintech). The membrane was washed three times with TBS-T and then incubated with goat anti-mouse IgG(H+L) (1:10,000) (Jackson Immunoresearch). Immunoreactivity was detected using SuperSignal (Thermo Fisher Scientific).
Lytic assay against planktonic E. faecalis and E. faecium
The E. faecalis strain Patient031_14_8 and E. faecium strains Patient012_7_1, Patient012_7_2, Patient016_14_1, Patient016_14_2, Patient019_14_1, Patient038_35_1, Patient038_35_6, Patient040_35_1, Patient040_35_2, Patient040_42_1, Patient040_42_3, Patient046_14_1, Patient046_14_2, Patient046_21_1, Patient046_21_2, Patient050_7_3, Patient052_21_1, Patient052_21_2 and Patient052_28_1 were grown in BHI broth aerobically and then collected by centrifugation at 3,000g for 15 min. Cell pellets were washed and resuspended in HiTrap buffer. The lytic activity of the endolysin was calculated based on reduction at an OD600 nm as measured in a TVS062CA BioPhoto recorder (ADVANTEC), with the OD600 nm measured every minute. For each sample, a 50 µg ml−1 final concentration of endolysin or a 50 µg ml−1 final concentration of vehicle including His-SUMO was added to the cell resuspension.
Lytic assay on E. faecalis biofilm
A biofilm assay using crystal violet staining was conducted according to previous studies12,82,83. In brief, each isolated E. faecalis strain was cultured overnight in BHI medium supplemented with 20 µg ml−1 aztreonam, 20 µg ml−1 polymyxin B and 4 µg ml−1 amphotericin B at 37 °C. A sample of each overnight culture was diluted 100-fold in fresh BHI medium, inoculated into a 96-well flat-bottomed polystyrene microtiter plate (Corning) and further incubated aerobically at 37 °C for 24 h. After incubation, the 96-well plates were gently washed once with 200 µl HiTrap buffer, and then 200 µl of endolysin or vehicle including His-SUMO (50 µg ml−1 final concentration) was added. The plates were incubated overnight, then washed once with 200 µl phosphate-buffered saline (PBS) to remove planktonic cells, and dried in an inverted position. Subsequently, 200 µl of 0.5% crystal violet solution was added for staining. After 15 min of staining at room temperature, the plates were washed three times with 200 µl of PBS to remove excess dye. The plates were dried at 50 °C for 30 min, and the bound dye was dissolved by adding 200 µl of 95% ethanol (v/v) for 20 min. The absorbance of the samples was measured at 570 nm. All experiments included eight replicate experimental wells.
In vivo mouse models
Female germ-free C57BL/6 (H2kb) and female specific-pathogen-free 129SvJ/JmsSlc mice (6–8 weeks old) were purchased from SLC Japan and CLEA Japan, respectively. Mice were housed in a temperature-controlled (23 ± 2 °C) room with a dark period from 20:00 to 08:00. They were allowed free access to sterile water and standard laboratory mouse chow. Germ-free C57BL/6 mice were reared as one individual per cage. All mice were randomly housed in groups and selected for the experiments.
In the mono-colonized gnotobiotic mouse model, C57BL/6 germ-free female mice were orally inoculated with E. faecalis strain Patient031_14_8, E. faecalis strain Patient009_35_10, E. faecalis strain Patient015_56_7, E. faecium strain Patient038_35_1, E. faecium strain Patient040_35_1, E. faecalis strain JCM5803, E. faecium strain Patient019_14_1 or E. coli strain Patient025_0_122 suspensions (2.0 × 108 CFU freshly suspended in 200 µl of sterile PBS per mouse) before stem cell transplantation. The E. faecalis strain JCM5803 was obtained from the Japan Collection of Microorganisms. Three weeks after the transfer, bacterial colonization was confirmed by culturing faecal suspensions (diluted in PBS with vortex mixing, and then passed through a 100-µm cell strainer) on Enterococcus selection agar medium (BHI broth containing 20 µg ml−1 aztreonam, 20 µg ml−1 polymyxin B, 4 µg ml−1 amphotericin B and 50 µg ml−1 triphenyltetrazolium chloride) or LB agar medium.
In a humanized gnotobiotic mouse model, 200-µl suspensions of Patient031_14 faecal samples (samples obtained from Patient031 on day 14 ± 3), Patient043_42 faecal samples (samples obtained from Patient043 on day 42 ± 3), Patient032_21 faecal samples (samples obtained from Patient032 on day 21 ± 3), Patient026_7 faecal samples (samples obtained from Patient026 on day 7 ± 3) or Patient062_0 faecal samples (samples obtained from Patient062 on day 0 ± 3) in anaerobic culture medium at a 16-fold dilution (w/v) were administered before stem cell transplantation. Three weeks after the transfer, the intestinal bacterial composition of each mouse was analysed by 16S rRNA gene sequencing.
In the humanized GVHD model colonized with Patient031_14 faecal samples, Patient043_42 faecal samples or Patient032_21 faecal samples, germ-free mice were reared in two or more germ-free isolators, and each isolator contained mice from the endolysin-treated group and the vehicle-treated group. All animal experiments were performed with the approval of the Animal Care and Use Committees of Osaka Metropolitan University.
Lytic assay in the gnotobiotic mouse model
Two hundred micrograms of endolysin or vehicle, dissolved in 200 µl of HiTrap buffer, was orally administered three times a week to E. faecalis-mono-colonized gnotobiotic mice (n = 2 per group). After a week of administration, mice were euthanized and the entire intestinal tract was immediately removed and divided into the ileum and colon. Intestinal tract samples were also obtained from a germ-free mouse as control samples. Specimens of approximately 3 mm3 were prepared and then fixed in 2% formaldehyde and 2.5% glutaraldehyde in 0.1 M phosphate buffer. After extensive washing, secondary fixation in 1% osmium tetroxide solution in 0.1 M phosphate buffer was performed for 2 h, followed by further washing and chemical dehydration in a progressive ethanol series. Samples were transferred to isoamyl acetate for 30 min, and then dehydrated in a Hitachi Model HCP-2 critical point dryer (Hitachi) with liquid CO2. Samples were finally coated with 5 nm of osmium in an Neoc-Pro/PN (Meiwafosis) before imaging using a Hitachi S-4700 (Hitachi).
Two hundred micrograms of endolysin or vehicle dissolved in 200 µl of HiTrap buffer was orally administered three times a week to E. faecium-mono-colonized gnotobiotic mice (n = 4 per group). After a week of administration, the CFU was determined by culturing faecal suspensions (diluted in PBS with vortex mixing, and then passed through a 100 µm cell strainer) on Enterococcus selection agar medium (BHI broth containing 20 µg ml−1 aztreonam, 20 µg ml−1 polymyxin B, 4 µg ml−1 amphotericin B and 50 µg ml−1 triphenyltetrazolium chloride).
Gnotobiotic GVHD mouse model
A previously reported aGVHD mouse model based on chemotherapy conditioning and major-histocompatibility-complex-matched (minor-antigen-mismatched) transplantation was used in our study84.
As conditioning therapy, female germ-free C57BL/6 mice received intraperitoneal doses of busulfan (Sigma-Aldrich; 20 mg per kg per day) for five days, followed by cyclophosphamide (Sigma-Aldrich; 100 mg per kg per day) for three days. Day −2 and −1 were rest days. On day 0, recipient C57BL/6 mice were intravenously injected with 1.5 × 107 bone marrow cells and 2.0 × 106 splenic T cells from female 129SvJ/JmsSlc donor mice. Bone marrow cells were flushed from the femur and suspended in Hanks’ balanced salt solution (Nacalai Tesque). After erythrocyte lysis using Red Blood Cell Lysing Buffer Hybti-Max (Sigma-Aldrich), a single-cell suspension was prepared in PBS. A splenic T cell suspension was obtained using the Pan T Cell Isolation Kit II for mice (Milltenyi Biotec) according to the manufacturer’s instructions.
From the day of stem cell transplantation, 200 µg of endolysin or vehicle, dissolved in 200 µl HiTrap buffer, was administered orally three times a week for three weeks. Survival was monitored daily. Overall survival in each group was statistically analysed and compared using a generalized Wilcoxon test. Using the E. faecalis-mono-colonized gnotobiotic GVHD mouse model, faecal E. faecalis bacteria were quantified by culturing faecal suspensions diluted in BHI broth on Enterococcus selection agar medium and counting the colonies. All experiments were performed in duplicate. Using the humanized gnotobiotic GVHD mouse model, bacterial DNA was extracted from sequentially collected faecal samples using the PowerSoil DNA Isolation Kit (Qiagen) in accordance with the manufacturer’s protocol and 16S rRNA gene analysis was performed as described above. Faith’s phylogenetic alpha diversity estimate and principal coordinate analysis of the weighted UniFrac distance matrices were performed using QIIME2 (v.2018.11; https://qiime2.org).
The concentration of IFNγ in the serum on day 8 after the transfer was measured by ELISA using reagents from Invitrogen, according to the manufacturer’s protocol.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Data availability
All data supporting the findings of this study are provided within the manuscript and its Supplementary Information. Sequencing data generated for this study have been deposited in the NCBI Sequence Read Archive (PRJNA1095194 and PRJNA1109929). The RefSeq protein database downloaded from the NCBI ftp site on 5 May 2020 and 17 January 2021, and the NCBI RefSeq nucleotide sequence data downloaded on 24 October 2021, have been deposited in the Zenodo repository (https://doi.org/10.5281/zenodo.11196056 (ref. 85) and https://doi.org/10.5281/zenodo.11239382 (ref. 86)). Source data are provided with this paper.
Code availability
Additional scripts are available in the Zenodo repository at https://doi.org/10.5281/zenodo.11401108 (ref. 87).
References
Peled, J. U. et al. Microbiota as predictor of mortality in allogeneic hematopoietic-cell transplantation. N. Engl. J. Med. 382, 822–834 (2020).
Taur, Y. et al. The effects of intestinal tract bacterial diversity on mortality following allogeneic hematopoietic stem cell transplantation. Blood 124, 1174–1182 (2014).
Jones, J. M., Wilson, R. & Bealmear, P. M. Mortality and gross pathology of secondary disease in germfree mouse radiation chimeras. Radiat. Res. 45, 577–588 (1971).
van Bekkum, D. W., Roodenburg, J., Heidt, P. J. & van der Waaij, D. Mitigation of secondary disease of allogeneic mouse radiation chimeras by modification of the intestinal microflora. J. Natl Cancer Inst. 52, 401–404 (1974).
Shono, Y. et al. Increased GVHD-related mortality with broad-spectrum antibiotic use after allogeneic hematopoietic stem cell transplantation in human patients and mice. Sci. Transl. Med. 8, 339ra371 (2016).
Gavriilaki, M., Sakellari, I., Anagnostopoulos, A. & Gavriilaki, E. The impact of antibiotic-mediated modification of the intestinal microbiome on outcomes of allogeneic hematopoietic cell transplantation: systematic review and meta-analysis. Biol. Blood Marrow Transplant. 26, 1738–1746 (2020).
Taur, Y. et al. Intestinal domination and the risk of bacteremia in patients undergoing allogeneic hematopoietic stem cell transplantation. Clin. Infect. Dis. 55, 905–914 (2012).
Ilett, E. E. et al. Associations of the gut microbiome and clinical factors with acute GVHD in allogeneic HSCT recipients. Blood Adv. 4, 5797–5809 (2020).
Kusakabe, S. et al. Enterococcus: a predictor of ravaged microbiota and poor prognosis after allogeneic hematopoietic stem cell transplantation. Biol. Blood Marrow Transplant. 26, 1028–1033 (2020).
Holler, E. et al. Metagenomic analysis of the stool microbiome in patients receiving allogeneic stem cell transplantation: loss of diversity is associated with use of systemic antibiotics and more pronounced in gastrointestinal graft-versus-host disease. Biol. Blood Marrow Transplant. 20, 640–645 (2014).
Jagasia, M. et al. Risk factors for acute GVHD and survival after hematopoietic cell transplantation. Blood 119, 296–307 (2012).
Ciernikova, S. et al. Targeting the gut microbiome: an emerging trend in hematopoietic stem cell transplantation. Blood Rev. 48, 100790 (2021).
Kamada, N. et al. Regulated virulence controls the ability of a pathogen to compete with the gut microbiota. Science 336, 1325–1329 (2012).
Barrios-Villa, E. et al. Comparative genomics of a subset of adherent/invasive Escherichia coli strains isolated from individuals without inflammatory bowel disease. Genomics 112, 1813–1820 (2020).
Fujimoto, K. et al. Antigen-specific mucosal immunity regulates development of intestinal bacteria-mediated diseases. Gastroenterology 157, 1530–1543 (2019).
Dubin, K. & Pamer, E. G. Enterococci and their interactions with the intestinal microbiome. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.bad-0014-2016 (2014).
Ford, C. D. et al. Vancomycin-resistant Enterococcus colonization and bacteremia and hematopoietic stem cell transplantation outcomes. Biol. Blood Marrow Transplant. 23, 340–346 (2017).
Stein-Thoeringer, C. K. et al. Lactose drives Enterococcus expansion to promote graft-versus-host disease. Science 366, 1143–1149 (2019).
Kakihana, K. et al. Fecal microbiota transplantation for patients with steroid-resistant acute graft-versus-host disease of the gut. Blood 128, 2083–2088 (2016).
Kaito, S. et al. Fecal microbiota transplantation with frozen capsules for a patient with refractory acute gut graft-versus-host disease. Blood Adv. 2, 3097–3101 (2018).
Goeser, F. et al. Fecal microbiota transfer for refractory intestinal graft-versus-host disease—experience from two German tertiary centers. Eur. J. Haematol. 107, 229–245 (2021).
Zhao, Y. et al. Safety and efficacy of fecal microbiota transplantation for grade IV steroid refractory GI-GvHD patients: interim results from FMT2017002 trial. Front. Immunol. 12, 678476 (2021).
DeFilipp, Z. et al. Drug-resistant E. coli bacteremia transmitted by fecal microbiota transplant. N. Engl. J. Med. 381, 2043–2050 (2019).
Bilinski, J. et al. Eosinophilic gastroenteritis and graft-versus-host disease induced by transmission of Norovirus with fecal microbiota transplant. Transpl. Infect. Dis. 23, e13386 (2021).
Fujimoto, K. et al. Metagenome data on intestinal phage–bacteria associations aids the development of phage therapy against pathobionts. Cell Host Microbe 28, 380–389 (2020).
Fujiya, Y. et al. Transmission dynamics of a linear vanA-plasmid during a nosocomial multiclonal outbreak of vancomycin-resistant enterococci in a non-endemic area, Japan. Sci. Rep. 11, 14780 (2021).
Huycke, M. M. & Gilmore, M. S. Frequency of aggregation substance and cytolysin genes among enterococcal endocarditis isolates. Plasmid 34, 152–156 (1995).
Huycke, M. M., Spiegel, C. A. & Gilmore, M. S. Bacteremia caused by hemolytic, high-level gentamicin-resistant Enterococcus faecalis. Antimicrob. Agents Chemother. 35, 1626–1634 (1991).
Ike, Y., Hashimoto, H. & Clewell, D. B. High incidence of hemolysin production by Enterococcus (Streptococcus) faecalis strains associated with human parenteral infections. J. Clin. Microbiol. 25, 1524–1528 (1987).
Dupont, H., Montravers, P., Mohler, J. & Carbon, C. Disparate findings on the role of virulence factors of Enterococcus faecalis in mouse and rat models of peritonitis. Infect. Immun. 66, 2570–2575 (1998).
Gentry-Weeks, C. R., Karkhoff-Schweizer, R., Pikis, A., Estay, M. & Keith, J. M. Survival of Enterococcus faecalis in mouse peritoneal macrophages. Infect. Immun. 67, 2160–2165 (1999).
Ike, Y., Hashimoto, H. & Clewell, D. B. Hemolysin of Streptococcus faecalis subspecies zymogenes contributes to virulence in mice. Infect. Immun. 45, 528–530 (1984).
Jett, B. D., Jensen, H. G., Nordquist, R. E. & Gilmore, M. S. Contribution of the pAD1-encoded cytolysin to the severity of experimental Enterococcus faecalis endophthalmitis. Infect. Immun. 60, 2445–2452 (1992).
Hoyle, B. D. & Costerton, J. W. Bacterial resistance to antibiotics: the role of biofilms. Prog. Drug Res. 37, 91–105 (1991).
Potera, C. Forging a link between biofilms and disease. Science 283, 1837–1839 (1999).
Lewis, K. Multidrug tolerance of biofilms and persister cells. Curr. Top. Microbiol. Immunol. 322, 107–131 (2008).
Mah, T. F. & O’Toole, G. A. Mechanisms of biofilm resistance to antimicrobial agents. Trends Microbiol. 9, 34–39 (2001).
Mah, T. F. et al. A genetic basis for Pseudomonas aeruginosa biofilm antibiotic resistance. Nature 426, 306–310 (2003).
Briers, Y. et al. Engineered endolysin-based “Artilysins” to combat multidrug-resistant gram-negative pathogens. mBio 5, e01379–01314 (2014).
Vollmer, W. & Bertsche, U. Murein (peptidoglycan) structure, architecture and biosynthesis in Escherichia coli. Biochim. Biophys. Acta 1778, 1714–1734 (2008).
Nelson, D., Loomis, L. & Fischetti, V. A. Prevention and elimination of upper respiratory colonization of mice by group A streptococci by using a bacteriophage lytic enzyme. Proc. Natl Acad. Sci. USA 98, 4107–4112 (2001).
Schmelcher, M., Donovan, D. M. & Loessner, M. J. Bacteriophage endolysins as novel antimicrobials. Future Microbiol. 7, 1147–1171 (2012).
Lopez, R., Garcia, E., Garcia, P. & Garcia, J. L. The pneumococcal cell wall degrading enzymes: a modular design to create new lysins? Microb. Drug Resist. 3, 199–211 (1997).
Briers, Y. et al. The high-affinity peptidoglycan binding domain of Pseudomonas phage endolysin KZ144. Biochem. Biophys. Res. Commun. 383, 187–191 (2009).
Domenech, M., Garcia, E. & Moscoso, M. In vitro destruction of Streptococcus pneumoniae biofilms with bacterial and phage peptidoglycan hydrolases. Antimicrob. Agents Chemother. 55, 4144–4148 (2011).
Schuch, R., Nelson, D. & Fischetti, V. A. A bacteriolytic agent that detects and kills Bacillus anthracis. Nature 418, 884–889 (2002).
Rashidi, A. et al. Dysbiosis patterns during re-induction/salvage versus induction chemotherapy for acute leukemia. Sci. Rep. 9, 6083 (2019).
Stewart, P. S. Antimicrobial tolerance in biofilms. Microbiol. Spectr. https://doi.org/10.1128/microbiolspec.MB-0010-2014 (2015).
Barnes, A. M. T. et al. Enterococcus faecalis readily colonizes the entire gastrointestinal tract and forms biofilms in a germ-free mouse model. Virulence 8, 282–296 (2017).
Kattke, M. D. et al. Structure and mechanism of TagA, a novel membrane-associated glycosyltransferase that produces wall teichoic acids in pathogenic bacteria. PLoS Pathog. 15, e1007723 (2019).
He, Z. et al. Use of the quorum sensing inhibitor furanone C-30 to interfere with biofilm formation by Streptococcus mutans and its luxS mutant strain. Int. J. Antimicrob. Agents 40, 30–35 (2012).
Ali, L. et al. Molecular mechanism of quorum-sensing in Enterococcus faecalis: its role in virulence and therapeutic approaches. Int. J. Mol. Sci. 18, 960 (2017).
Kristich, C. J., Li, Y. H., Cvitkovitch, D. G. & Dunny, G. M. Esp-independent biofilm formation by Enterococcus faecalis. J. Bacteriol. 186, 154–163 (2004).
Thurlow, L. R. et al. Gelatinase contributes to the pathogenesis of endocarditis caused by Enterococcus faecalis. Infect. Immun. 78, 4936–4943 (2010).
Steck, N. et al. Enterococcus faecalis metalloprotease compromises epithelial barrier and contributes to intestinal inflammation. Gastroenterology 141, 959–971 (2011).
Fernandez-Ruiz, I., Coutinho, F. H. & Rodriguez-Valera, F. Thousands of novel endolysins discovered in uncultured phage genomes. Front. Microbiol. 9, 1033 (2018).
Duerkop, B. A., Palmer, K. L. & Horsburgh, M. J. in Enterococci: From Commensals to Leading Causes of Drug Resistant Infection (eds Gilmore, M. S. et al.) Ch. 8 (Massachusetts Eye and Ear Infirmary, 2014).
Cheng, M. et al. Endolysin LysEF-P10 shows potential as an alternative treatment strategy for multidrug-resistant Enterococcus faecalis infections. Sci. Rep. 7, 10164 (2017).
Van Tyne, D., Martin, M. J. & Gilmore, M. S. Structure, function, and biology of the Enterococcus faecalis cytolysin. Toxins 5, 895–911 (2013).
Duan, Y. et al. Bacteriophage targeting of gut bacterium attenuates alcoholic liver disease. Nature 575, 505–511 (2019).
Freifeld, A. G. et al. Clinical practice guideline for the use of antimicrobial agents in neutropenic patients with cancer: 2010 update by the infectious diseases society of america. Clin. Infect. Dis. 52, e56–e93 (2011).
Bacigalupo, A. et al. Defining the intensity of conditioning regimens: working definitions. Biol. Blood Marrow Transplant. 15, 1628–1633 (2009).
Kanda, Y. Investigation of the freely available easy-to-use software ‘EZR’ for medical statistics. Bone Marrow Transplant. 48, 452–458 (2013).
Dutka-Malen, S., Evers, S. & Courvalin, P. Detection of glycopeptide resistance genotypes and identification to the species level of clinically relevant enterococci by PCR. J. Clin. Microbiol. 33, 1434 (1995).
Cheng, S. et al. A PCR assay for identification of Enterococcus faecium. J. Clin. Microbiol. 35, 1248–1250 (1997).
Semedo, T. et al. Comparative study using type strains and clinical and food isolates to examine hemolytic activity and occurrence of the cyl operon in enterococci. J. Clin. Microbiol. 41, 2569–2576 (2003).
Nikolenko, S. I., Korobeynikov, A. I. & Alekseyev, M. A. BayesHammer: Bayesian clustering for error correction in single-cell sequencing. BMC Genomics 14, S7 (2013).
Nurk, S., Meleshko, D., Korobeynikov, A. & Pevzner, P. A. metaSPAdes: a new versatile metagenomic assembler. Genome Res. 27, 824–834 (2017).
Li, W. & Godzik, A. Cd-hit: a fast program for clustering and comparing large sets of protein or nucleotide sequences. Bioinformatics 22, 1658–1659 (2006).
Camacho, C. et al. BLAST+: architecture and applications. BMC Bioinformatics 10, 421 (2009).
Gregor, I., Droge, J., Schirmer, M., Quince, C. & McHardy, A. C. PhyloPythiaS+: a self-training method for the rapid reconstruction of low-ranking taxonomic bins from metagenomes. PeerJ 4, e1603 (2016).
Truong, D. T. et al. MetaPhlAn2 for enhanced metagenomic taxonomic profiling. Nat. Methods 12, 902–903 (2015).
Roux, S., Emerson, J. B., Eloe-Fadrosh, E. A. & Sullivan, M. B. Benchmarking viromics: an in silico evaluation of metagenome-enabled estimates of viral community composition and diversity. PeerJ 5, e3817 (2017).
Roux, S., Hallam, S. J., Woyke, T. & Sullivan, M. B. Viral dark matter and virus-host interactions resolved from publicly available microbial genomes. eLife 4, e08490 (2015).
Hyatt, D., LoCascio, P. F., Hauser, L. J. & Uberbacher, E. C. Gene and translation initiation site prediction in metagenomic sequences. Bioinformatics 28, 2223–2230 (2012).
Kakuta, M., Suzuki, S., Izawa, K., Ishida, T. & Akiyama, Y. A massively parallel sequence similarity search for metagenomic sequencing data. Int. J. Mol. Sci. 18, 2124 (2017).
Terzian, P. et al. PHROG: families of prokaryotic virus proteins clustered using remote homology. NAR Genom. Bioinform. 3, lqab067 (2021).
Krzywinski, M. et al. Circos: an information aesthetic for comparative genomics. Genome Res. 19, 1639–1645 (2009).
Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32, 1792–1797 (2004).
Waterhouse, A. M., Procter, J. B., Martin, D. M., Clamp, M. & Barton, G. J. Jalview Version 2—a multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191 (2009).
Guo, H. et al. Self-assembly of virus-like particles of rabbit hemorrhagic disease virus capsid protein expressed in Escherichia coli and their immunogenicity in rabbits. Antiviral Res. 131, 85–91 (2016).
Yang, Y., Li, W., Hou, B. & Zhang, C. Quorum sensing LuxS/autoinducer-2 inhibits Enterococcus faecalis biofilm formation ability. J. Appl. Oral Sci. 26, e20170566 (2018).
Shao, C. et al. LuxS-dependent AI-2 regulates versatile functions in Enterococcus faecalis V583. J. Proteome Res. 11, 4465–4475 (2012).
Riesner, K., Kalupa, M., Shi, Y., Elezkurtaj, S. & Penack, O. A preclinical acute GVHD mouse model based on chemotherapy conditioning and MHC-matched transplantation. Bone Marrow Transplant. 51, 410–417 (2016).
Sato, N. RefSeq protein sequences of bacteria and viruses. Zenodo https://doi.org/10.5281/zenodo.11196056 (2024).
Sato, N. Nucleotide sequences of Caudovirales obtained from RefSeq. Zenodo https://doi.org/10.5281/zenodo.11239382 (2024).
Kimura, Y. et al. Shotgun metagenomics processing pipeline. Zenodo https://zenodo.org/records/11401108 (2024).
Acknowledgements
We thank B. Baigalmaa, K. Shishido, F. Furuta, Y. Koike, M. Fujimori and S. Hatakeyama for technical assistance; M. Maeda, K. Ogawa and K. Suetsugu for administrative assistance; S. Kikuchi for patient enrolment; and the Research Support Platform of the Graduate School of Medicine, Osaka Metropolitan University for technical support. This study was supported by the Japan Agency for Medical Research and Development (AMED) (to K.F., JP21ae0121048; and to S.U., JP21fk0108619 and JP21ae0121040), the Japanese Society for the Promotion of Science (JSPS) (to K.F., 22K16329; to S.U., 21K19495 and 22H00477), the Takeda Science Foundation (to S.U.), the Canon Foundation (to S.U.) and the Center of Innovation Program from the Japan Science and Technology Agency (JST) (to S.U. and S.I., JPMJCE1304). The super-computing resource was provided by the Human Genome Center, Institute of Medical Science, University of Tokyo (http://sc.hgc.jp/shirokane.html). We thank Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript.
Author information
Authors and Affiliations
Contributions
K.F., T.H., S.I. and S.U. conceived and designed the study. K.F. and T.H. performed most of the experiments. K.F., T.H., M.Y. and N.S. performed computational analysis of metagenomic datasets. M.Y. and S.I. developed the data analysis pipeline. T.H., M.S., D.M., C.Y., M.W., Y.H., Y.K., Y.Y., T.Y. and S.S. collected the human faecal samples. T.H., A.H. and M.N. created protocols for patient enrolment and sample collection at Osaka Metropolitan University Hospital. N.S., M.U. and K.Y. helped with the data analysis. H.N., Y.F., Y.A. and M.H. provided scientific insights and critical review of the manuscript. K.F., T.H., S.I. and S.U. wrote and edited the manuscript. S.U. directed the research.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Peer review
Peer review information
Nature thanks Jeremy Barr, Marcel van den Brink and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Extended data figures and tables
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.
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 cylLL, 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.
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.
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).
Extended Data Fig. 6 Analysis of the E. faecalis cylLL 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.
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).
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).
Supplementary information
Source data
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Fujimoto, K., Hayashi, T., Yamamoto, M. et al. An enterococcal phage-derived enzyme suppresses graft-versus-host disease. Nature (2024). https://doi.org/10.1038/s41586-024-07667-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41586-024-07667-8
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
