Tern paper

Nature | Vol 575 | 21 November 2019 | 505

Article

Bacteriophage targeting of gut bacterium attenuates alcoholic liver disease

Yi Duan1,2,29, Cristina Llorente1,2,29, Sonja Lang1, Katharina Brandl3, Huikuan Chu1, Lu Jiang1,2, Richard C. White4, Thomas H. Clarke4, Kevin Nguyen4, Manolito Torralba5, Yan Shao6, Jinyuan Liu7, Adriana Hernandez-Morales8, Lauren Lessor9, Imran R. Rahman10, Yukiko Miyamoto1, Melissa Ly11, Bei Gao1, Weizhong Sun1, Roman Kiesel1, Felix Hutmacher1, Suhan Lee1, Meritxell Ventura-Cots12, Francisco Bosques-Padilla13, Elizabeth C. Verna14, Juan G. Abraldes15, Robert S. Brown Jr16, Victor Vargas17,18, Jose Altamirano17, Juan Caballería18,19, Debbie L. Shawcross20, Samuel B. Ho1,2, Alexandre Louvet21, Michael R. Lucey22, Philippe Mathurin21, Guadalupe Garcia-Tsao23,24, Ramon Bataller12, Xin M. Tu7, Lars Eckmann1, Wilfred A. van der Donk10,25,26, Ry Young8,9, Trevor D. Lawley6, Peter Stärkel27, David Pride1,11,28, Derrick E. Fouts4 & Bernd Schnabl1,2,28*

Chronic liver disease due to alcohol-use disorder contributes markedly to the global burden of disease and mortality1–3. Alcoholic hepatitis is a severe and life-threatening form of alcohol-associated liver disease. The gut microbiota promotes ethanol- induced liver disease in mice4, but little is known about the microbial factors that are responsible for this process. Here we identify cytolysin—a two-subunit exotoxin that is secreted by Enterococcus faecalis5,6—as a cause of hepatocyte death and liver injury. Compared with non-alcoholic individuals or patients with alcohol-use disorder, patients with alcoholic hepatitis have increased faecal numbers of E. faecalis. The presence of cytolysin-positive (cytolytic) E. faecalis correlated with the severity of liver disease and with mortality in patients with alcoholic hepatitis. Using humanized mice that were colonized with bacteria from the faeces of patients with alcoholic hepatitis, we investigated the therapeutic effects of bacteriophages that target cytolytic E. faecalis. We found that these bacteriophages decrease cytolysin in the liver and abolish ethanol-induced liver disease in humanized mice. Our findings link cytolytic E. faecalis with more severe clinical outcomes and increased mortality in patients with alcoholic hepatitis. We show that bacteriophages can specifically target cytolytic E. faecalis, which provides a method for precisely editing the intestinal microbiota. A clinical trial with a larger cohort is required to validate the relevance of our findings in humans, and to test whether this therapeutic approach is effective for patients with alcoholic hepatitis.

The most severe form of alcohol-related liver disease is alcoholic hepa- titis; mortality ranges from 20% to 40% at 1–6 months, and as many as 75% of patients die within 90 days of a diagnosis of severe alcoholic hepatitis7–9. Therapy with corticosteroids is only marginally effective9. Early liver transplantation is the only curative therapy, but is offered only at select centres and to a limited group of patients10.

Alcohol-related liver disease can be transmitted via faecal micro- biota4. We investigated the microorganisms and microbial factors that are responsible for this transmissible phenotype and for progression of alcohol-related liver disease.

Cytolysin linked to increased mortality We performed 16S ribosomal RNA (rRNA) gene sequencing to deter- mine whether chronic alcohol use and alcoholic hepatitis are associated

with an altered composition of the faecal microbiota. Differences in faecal microbiota composition were noted in patients with alcohol-use disorder and alcoholic hepatitis, compared to subjects without alcohol- use disorder (controls) (Fig. 1a, Extended Data Fig. 1a, b, Supplementary Tables 1, 2). One substantial difference that we observed was an increase in the proportion of Enterococcus spp. in patients with alcoholic hepa- titis: in these patients, 5.59% of faecal bacteria were Enterococcus spp. compared with almost none in controls (0.023%; for comparison, 0.004% of all reads were Enterococcus spp. in the Human Microbiome Project) or patients with alcohol-use disorder (0.024%). Faecal sam- ples from patients with alcoholic hepatitis had about 2,700-fold more E. faecalis than samples from controls, as measured by quantitative PCR (qPCR) (Extended Data Fig. 1c), which is consistent with the 16S rRNA sequencing results. About 80% of patients with alcoholic hepatitis are positive for E. faecalis in their faeces (Extended Data Fig. 1d).

https://doi.org/10.1038/s41586-019-1742-x

Received: 9 April 2019

Accepted: 2 October 2019

Published online: 13 November 2019

A list of affiliations appears at the end of the paper.

506 | Nature | Vol 575 | 21 November 2019

Article

The colonization of mice with E. faecalis induces mild hepatic stea- tosis and exacerbates ethanol-induced liver disease11, by mechanisms that are unclear. Cytolysin is a bacterial exotoxin (or bacteriocin) that is produced by E. faecalis12, and which contains two post-translationally modified peptides (CylLL′′ and CylLS′′) in its bioactive form

6. The two peptides are encoded by two separate genes: cylLL and cylLS, respec- tively12. Cytolysin has lytic activity against not only Gram-positive bacte- ria, but also eukaryotic cells13. We detected cylLL and cylLS genomic DNA (cytolysin-positive) in faecal samples from 30% of patients with alco- holic hepatitis; none of the faecal samples from controls and only one sample from a patient with alcohol-use disorder was cytolysin-positive, as detected by qPCR (Fig. 1b). Importantly, 89% of cytolysin-positive patients with alcoholic hepatitis died within 180 days of admission, compared to only 3.8% of cytolysin-negative patients (P < 0.0001) (Fig. 1c). Among the cytolysin-positive patients, 72.2% (13 out of 18) died owing to liver failure (including complications related to liver

failure, such as gastrointestinal bleeding) (Supplementary Table 2). Infection was not associated with 30-day, 90-day or 180-day mortality (P = 0.403, 0.234 or 0.098) in patients with alcoholic hepatitis.

Our univariate logistic and Cox regression of laboratory and clinical parameters found an association between the detection of cytolysin- encoding genes in faeces and the international normalized ratio (INR), platelet count, the model for end-stage liver disease (MELD) score, the sodium MELD score, the age, serum bilirubin, INR and serum creatinine (ABIC) score and death (Supplementary Table 3). In the multivariate Cox analysis, detection of cytolysin-encoding genes in faeces was associated with 90-day (P = 0.004) and with 180-day mortality (P = 0.001) (Supple- mentary Table 3), even after we adjusted for the geographical origin of the patient, antibiotic treatment, platelet count, and creatinine, bilirubin and INR as components of the MELD score. We found no multicollinear- ity between the detection of faecal cytolysin-encoding genes and these cofactors (variance inflation factor < 1.6), which indicates that cytolysin is an independent predictor of mortality in patients with alcoholic hepatitis. When we performed receiver-operating-characteristic curve analysis for 90-day mortality, cytolysin had an area under the curve of 0.81, which was superior to other widely used predictors for mortality in clinical practice (Extended Data Fig. 1e). On the basis of our findings, we propose that the detection of cytolysin may be a prognostic factor for more severe liver- related outcomes and increased risk of death, and a stronger predictor of mortality than MELD, ABIC and the discriminant function score.

To determine phylogeny of E. faecalis in patients with alcoholic hepatitis, we performed targeted culturing from stool samples. Whole- genome sequencing of 93 E. faecalis isolates revealed a broad phylo- genetic diversity of cytolysin-positive E. faecalis from patients with alcoholic hepatitis (Fig. 1d), which indicates that cytolysin production is a variable trait among E. faecalis isolates and that cytolysin is car- ried in mobile genetic elements, which include both chromosomally encoded pathogenicity islands and plasmids14. Detection of any other antimicrobial resistance genes or virulence genes in E. faecalis isolates did not correlate with disease severity or mortality in patients with alcoholic hepatitis (Supplementary Table 4).

The total amount of faecal E. faecalis, or faecal E. faecalis positivity, did not correlate with disease severity or mortality in patients with alcoholic hepatitis (Supplementary Tables 5, 6). Cytolysin-positive and cytolysin-negative patients with alcoholic hepatitis had similar amounts of faecal E. faecalis (Extended Data Fig. 1f ). Although there were differences in the composition of the gut microbiota in patients with alcoholic hepatitis from different geographical regions (Extended Data Fig. 1g), the proportion of cytolysin-positive patients, total amount of faecal E. faecalis, faecal E. faecalis positivity (Extended Data Fig. 1h–j), treatment and clinical outcomes (30-day and 90-day mortality) did not differ significantly among the regions or centres (Supplemen- tary Table 7). In addition, cirrhosis was not associated with cytolysin positivity, the total amount of faecal E. faecalis or faecal E. faecalis positivity in patients with alcoholic hepatitis (Extended Data Fig. 1k–m, Supplementary Tables 4–6). These results confirm our findings that the presence of cytolysin-producing E. faecalis rather than the total amount or presence of E. faecalis per se determines the severity of alcoholic hepatitis and mortality.

Cytolysin and ethanol-induced liver disease To determine whether cytolysin contributes to liver damage mediated by E. faecalis, we gavaged mice with a cytolytic E. faecalis strain (FA2- 2(pAM714)) or a non-cytolytic E. faecalis strain (FA2-2(pAM771))5; the mice were then placed on a chronic–binge ethanol diet15. Compared to mice gavaged with phosphate-buffered saline (PBS), mice fed with ethanol after they were gavaged with cytolytic E. faecalis developed more severe liver injury as indicated by a higher level of alanine amino- transferase (ALT) (Extended Data Fig. 2a) and increased hepatic stea- tosis (Extended Data Fig. 2b, c). Mice that were fed ethanol after they

−0.1

0

0.1

0.2

–0.2 –0.1 0 0.1 0.2 PC1 (29.25%)

P C

2 ( 1

7 .7

5 %

) a b

d

Cytolysin-negative Cytolysin-positive

0.15

Controls Alcohol-use disorder Alcoholic hepatitis

c

Cytolysin-positive Cytolysin-negative

0

20

40

60

80

100

S u

b je

c ts

(% )

Controls AUD Alcoholic hepatitis

0 0

25

50

75

100

Time (d)

S u

rv iv

a l(

% )

Cytolysin-negative Cytolysin-positive

180150120906030

Fig. 1 | E. faecalis cytolysin is associated with mortality in patients with alcoholic hepatitis. a, 16S rRNA sequencing of faecal samples from controls (n = 14), patients with alcohol-use disorder (n = 43) or alcoholic hepatitis (n = 75). We use principal coordinate analysis (PCoA) based on Jaccard dissimilarity matrices to show β-diversity among groups at the genus level. The composition of faecal microbiota was significantly different between each group (P < 0.01). b, Percentage of subjects with faecal samples that were positive for both cylLL and cylL S DNA sequences (cytolysin-positive), in controls (n = 25), patients with alcohol-use disorder (n = 38) or alcoholic hepatitis (n = 82), assessed by qPCR. Statistically significant differences were detected between controls and patients with alcoholic hepatitis (P < 0.01), and between patients with alcohol- use disorder and patients with alcoholic hepatitis (P < 0.001). c, Kaplan–Meier curve of survival of patients with alcoholic hepatitis whose faecal samples were cytolysin-positive (n = 25) or cytolysin-negative (n = 54) (P < 0.0001). d, Core genome single-nucleotide polymorphism (SNP) tree of E. faecalis strains isolated from patients with alcoholic hepatitis (n = 93 strains, from 24 patients), showing phylogenetic diversity of cytolysin-positive (red) E. faecalis. Genomically identical isolates from the same patient were combined, and are shown as a single dot. Scale bar represents the nucleotide substitutions per SNP site. P values are determined by permutational multivariate analysis of variance (PERMANOVA) followed by false discovery rate (FDR) procedures (a), two-sided Fisher’s exact test followed by FDR procedures (b) or two-sided log- rank (Mantel–Cox) test (c). The exact group size (n) and P values for each comparison are listed in Supplementary Table 10.

Nature | Vol 575 | 21 November 2019 | 507

were gavaged with cytolytic E. faecalis also had more liver inflamma- tion with higher expression levels of mRNAs that encode inflamma- tory cytokines and chemokines (Il1b, Cxcl1 and Cxcl2) (Extended Data Fig. 2d–f ), compared with mice given PBS. Mice that were fed ethanol after they were gavaged with non-cytolytic E. faecalis had significantly less ethanol-induced liver injury, steatosis and inflammation (Extended Data Fig. 2a–f ) and longer survival times (Extended Data Fig. 2g), as compared with mice that were fed ethanol after they were administered with cytolytic E. faecalis.

To explore the mechanism of cytolysin-associated liver damage, we measured cytolysin in the liver. CylLS was significantly increased in the liver of mice given cytolytic E. faecalis but not in the liver of mice that were not given E. faecalis or of mice gavaged with non-cyto- lytic E. faecalis after chronic ethanol administration (Extended Data Fig. 2h). E. faecalis was detectable in the liver of mice given cytolytic and

non-cytolytic E. faecalis and fed an ethanol diet, but not when mice were fed an isocaloric (control) diet (Extended Data Fig. 2i); this indicates that ethanol-induced changes in the gut barrier are necessary for the translocation of cytolytic E. faecalis from the intestine to the liver. The livers of ethanol-fed mice that were given cytolytic or non-cytolytic E. faecalis had positive E. faecalis cultures (Extended Data Fig. 2j). We observed an increased intestinal permeability in ethanol-fed mice com- pared with mice fed with an isocaloric diet, but this was independent of gavaging cytolytic or non-cytolytic E. faecalis after chronic ethanol administration (Extended Data Fig. 2k), indicating that cytolysin does not affect intestinal barrier function.

Administration of cytolytic or non-cytolytic E. faecalis to mice did not significantly change the composition of the intestinal microbiota, as shown by 16S rRNA gene sequencing (Extended Data Fig. 2l). Cytolytic E. faecalis did not affect intestinal absorption or hepatic metabolism

a b

c

e f g

h

d

Control

Ethanol

Alcoholic hepatitis cytolysin-negative

no. 3

Alcoholic hepatitis cytolysin-positive

no. 2

Alcoholic hepatitis cytolysin-positive

no. 1

Alcoholic hepatitis cytolysin-negative

no. 4

S e ru

m le

ve ls

o f

A L T

( U

l– 1 )

0

100

200

300

400 *

Control Ethanol

***

** ***

H e p

a ti c t

ri g

ly c e ri d

e c o

n te

n t

(m g

p e r

g li

ve r)

0

10

20

30

40

50

Control Ethanol

**** ****

* ***

Alcoholic hepatitis cytolysin-positive no. 1 Alcoholic hepatitis cytolysin-positive no. 2 Alcoholic hepatitis cytolysin-negative no. 3 Alcoholic hepatitis cytolysin-negative no. 4

Il1b

F o

ld c h

a n

g e

0

2

4

6

8

Control Ethanol

* *

** **

Cxcl1

F o

ld c h a n g

e

0

2

4

6

8

Control Ethanol

** **

** **

Col1a1 F

o ld

c h a n g

e

0

5

10

15

20

Control Ethanol

*** **

* *

Liver cylLS

M ic

e (%

)

0

20

40

60

80

100

ND ND ND

Control Ethanol

ND ND ND

**** ****

**** ****

LDH

C yt

o to

xi c it y

(% )

0

10

20

30

40

0 0

0 200

0 400

200 200

400 400

0 0

0 200

0 400

200 200

400 400

+– Ethanol

CylLL″ (nM) CylLS″ (nM)

* ** **

*

Fig. 2 | Transplantation of faeces from cytolysin-positive patients with alcoholic hepatitis exacerbates ethanol-induced liver disease in gnotobiotic mice. a–g, C57BL/6 germ-free mice were colonized with faeces from two cytolysin-positive and two cytolysin-negative patients with alcoholic hepatitis, and subjected to the chronic–binge feeding model. a, Serum levels of ALT. b, Hepatic triglyceride content. c, Representative sections of liver stained with haematoxylin and eosin (H & E). d–f, Hepatic levels of mRNAs that encode Il1b, Cxcl1 and Col1a1. g, Proportions of mice that were positive for cytolysin in the liver, measured by qPCR for cylL S . h, Lactate dehydrogenase (LDH) assay to measure cytotoxicity of hepatocytes isolated from mice that were fed an oral isocaloric control diet (five groups, left) or chronic–binge ethanol diet (five

groups, right), and incubated with vehicle, CylLS′′, CylLL′′ or both of the cytolysin subunits at the indicated concentrations without (–) or with (+) ethanol (25 mM) for 3 h. The survival of hepatocytes was determined in three independent experiments. Scale bar, 100 μm. Results are expressed as mean ± s.e.m. (a, b, d–f, h). P values are determined by one-way analysis of variance (ANOVA) with Tukey’s post hoc test (a, b, d–f ), two-sided Fisher’s exact test followed by FDR procedures (g) or two-way ANOVA with Tukey’s post hoc test (h). All results were generated from at least three independent replicates. The exact group size (n) and P values for each comparison are listed in Supplementary Table 10. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

508 | Nature | Vol 575 | 21 November 2019

Article

of ethanol, as indicated by serum levels of ethanol and hepatic levels of Adh1 and Cyp2e1 mRNAs (which encode the two primary enzymes that metabolize ethanol in the liver) (Extended Data Fig. 2m, n). These results indicate that E. faecalis that produce cytolysin promote ethanol- induced liver disease in mice.

To extend our findings to humans, we colonized germ-free mice with faeces from cytolysin-positive and cytolysin-negative patients with alcoholic hepatitis (Supplementary Table 8). Consistent with our findings from mice colonized with cytolytic E. faecalis, gnotobi- otic C57BL/6 mice colonized with faeces from two cytolysin-positive patients developed more severe ethanol-induced liver injury, steatosis, inflammation and fibrosis than mice given faeces from two cytolysin- negative patients (Fig. 2a–f, Extended Data Fig. 3a–d). Transplantation of faeces from cytolysin-positive patients reduced the survival time of the mice (Extended Data Fig. 3e) and increased translocation of

cytolytic E. faecalis to the liver after ethanol administration (Fig. 2g). The overall composition of the intestinal microbiota was not differ- ent between mice fed the control diet and colonized with faeces from cytolysin-positive or cytolysin-negative donors with alcoholic hepatitis, as shown by 16S rRNA gene sequencing. Mice transplanted with faeces from one of the cytolysin-positive patients with alcoholic hepatitis (patient no. 2) showed a microbiota that was significantly different from that of the other mouse groups after ethanol administration (Extended Data Fig. 3f ). Non-cytolytic E. faecalis was not detected in stool samples from donors with cytolytic E. faecalis (Extended Data Fig. 3g). We did not observe differences in intestinal absorp- tion or hepatic metabolism of ethanol between mice colonized with faeces from cytolysin-positive versus cytolysin-negative patients (Extended Data Fig. 3h, i). Together, these results provide further evi- dence that cytolysin promotes ethanol-induced liver disease.

b c

d

f g h

e

Control

Ethanol

Cytolysin-positive E. faecalis phages

C. crescentus phages

Cytolysin-positive E. faecalis phages

C. crescentus phages

Patient no. 5 Patient no. 2

Ef5.1 Ef5.2 Ef5.3 Ef5.4

Ef2.1 Ef2.2 Ef2.3

a

0

100

200

300

400 ***

Control

Patient no. 5

Patient no. 2

Ethanol

Patient no. 5

Patient no. 2

0

10

20

30

40

50

Control

Patient no. 5

Patient no. 2

Ethanol

Patient no. 5

Patient no. 2

*** C. crescentus phages

Cytolysin-positive E. faecalis phages

C. crescentus phages

Cytolysin-positive E. faecalis phages

Il1b

F o

ld c h a n g

e

0 2 4 6 8

10 12

Control

Patient     no. 5

Patient     no. 2

Ethanol

Patient     no. 5

Patient     no. 2

****

Cxcl1

F o

ld c h a n g

e

0 2 4 6 8

10 12

Control

Patient     no. 5

Patient     no. 2

Ethanol

Patient     no. 5

Patient     no. 2

****

Col1a1

F o

ld c h a n g

e

0 3 6 9

12 15 18

Control

Patient     no. 5

Patient     no. 2

Ethanol

Patient     no. 5

Patient     no. 2

** Liver cylLS

M ic

e (%

)

0

20

40

60

80

100

ND ND ND ND

Control

Patient no. 5

Patient no. 2

Ethanol

Patient no. 5

Patient no. 2

****

S e ru

m le

ve ls

o f

A L T

( U

l– 1 )

H e p

a ti c t

ri g

ly c e ri d

e c o

n te

n t

(m g

p e r

g li

ve r)

Fig. 3 | Phage therapy against cytolytic E. faecalis abolishes ethanol-induced liver disease in gnotobiotic mice. a, Transmission electron microscopy revealed that the phages we isolated were either siphophages (Ef5.1, Ef5.2, Ef5.3, Ef5.4 and Ef2.2) or myophages (Ef2.1 and Ef2.3). Scale bar, 50 nm. b–h, C57BL/6 germ-free mice were colonized with faeces from two cytolysin- positive patients with alcoholic hepatitis (faeces from one of these patients were also used in Fig. 2) and subjected to the chronic–binge feeding model, gavaged with control phages against C. crescentus (1010 plaque-forming units (PFUs)) or a cocktail of three or four different phages that target cytolytic E. faecalis (1010 PFUs), 1 day before an ethanol binge. b, Serum levels of ALT.

c, Hepatic triglyceride content. d, Representative H & E-stained liver sections. Scale bar, 100 μm. e–g, Hepatic levels of mRNAs that encode Il1b, Cxcl1 and Col1a1. h, Proportions of mice that were positive for cytolysin in the liver, measured by qPCR for cylL S . Results are expressed as mean ± s.e.m. (b, c, e–g). P values are determined by two-way ANOVA with Tukey’s post hoc test (b, c, e–g) or two-sided Fisher’s exact test followed by FDR procedures (h). All results are generated from at least three independent replicates. The exact group size (n) and P values for each comparison are listed in Supplementary Table 10. *P < 0.05, **P < 0.01, *** P < 0.001.

Nature | Vol 575 | 21 November 2019 | 509

To determine the mechanism by which cytolysin increases liver dis- ease, we isolated hepatocytes from mice fed ethanol or control diets, and stimulated them with pure bioactive cytolysin peptides (CylLL′′ and CylLS′′)

6. Incubation of the primary mouse hepatocytes with the two cytolysin subunits caused a dose-dependent increase in cell death compared to hepatocytes that were incubated with vehicle or with one subunit only (Fig. 2h). When we isolated hepatocytes from ethanol-fed mice and then incubated these hepatocytes with ethanol, we did not observe increased levels of cytolysin-induced cell death compared to hepatocytes isolated from mice on the control diet, which indicates that cytolysin-induced hepatocyte cell death was independent of ethanol. The cytotoxic effects of cytolysin are possibly mediated by pore forma- tion, resulting in cell lysis14.

Bacteriophage treatment in liver disease To further demonstrate the potential causative role of cytolytic E. faecalis for the development of ethanol-induced steatohepatitis,

we investigated the effects of treatment with bacteriophages (here- after, phages). Phages are ubiquitous in bacteria-rich environments, including the gut16. E. faecalis phages that are highly strain-specific can be isolated17, which potentially makes the direct editing of gut microbiota feasible. It has previously been shown that Atp4aSl/Sl mice, which lack gastric acid, have overgrowth of intestinal enterococci, which is associated with increased susceptibility to alcohol-induced steatohepatitis11. The gavaging of wild-type mice with an E. faecalis strain isolated from Atp4aSl/Sl mice led to increased ethanol-induced steatohepatitis11. We found that this same E. faecalis strain expressed cytolysin. We then isolated four distinct phages from sewage water. These phages lyse the cytolytic E. faecalis strain isolated from Atp4aSl/Sl mice. All four phages were podophages of the virulent Picovirinae group (Extended Data Fig. 4). Atp4aSl/Sl mice and their wild-type littermates were then placed on the chronic–binge ethanol diet and gavaged with the lytic phage cocktail. Phages directed against Caulobacter crescen- tus, a bacterium that is present in freshwater lakes and streams18 but that does not colonize humans or rodents19,20, were used as controls.

b c

d

f g h

e

Control

Ethanol

Cytolysin-negative E. faecalis phages

C. crescentus phages

Cytolysin-negative E. faecalis phages

C. crescentus phages

Patient no. 6 Patient no. 7

Ef 7.4

Ef6.1 Ef6.2 Ef6.3 Ef 6.4

Ef7.1 Ef7.2 Ef7.3

a

0

100

200

300

400

Control

Patient no. 6

Patient no. 7

Ethanol

Patient no. 6

Patient no. 7

0

10

20

30

40

50

Control

Patient no. 6

Patient no. 7

Ethanol

Patient no. 6

Patient no. 7

C. crescentus phages

Cytolysin-negative E. faecalis phages

C. crescentus phages

Cytolysin-negative E. faecalis phages

Il1b

F o

ld c h

a n

g e

0

1

2

3

4

5

Control

Patient     no. 6

Patient     no. 7

Ethanol

Patient     no. 6

Patient     no. 7

Cxcl1

F o

ld c h

a n

g e

0

1

2

3

4

5

Control

Patient     no. 6

Patient     no. 7

Ethanol

Patient     no. 6

Patient     no. 7

Col1a1

F o

ld c h

a n

g e

0

2

4

6

Control

Patient     no. 6

Patient     no. 7

Ethanol

Patient     no. 6

Patient     no. 7

F a e c a l E

n te

ro c o

c c u

s (C

F U

g –1

)

105 106 107 108 109

1010 **

Control

Patient no. 6

Patient no. 7

Ethanol

Patient no. 6

Patient no. 7

S e ru

m le

ve ls

o f

A L T

( U

l– 1 )

H e p

a ti c t

ri g

ly c e ri d

e c o

n te

n t

(m g

p e r

g li

ve r)

Fig. 4 | Phages that target non-cytolytic E. faecalis do not reduce ethanol- induced liver disease in gnotobiotic mice. a, Transmission electron microscopy revealed that the phages we isolated were either podophages (Ef6.2, Ef6.3, Ef 7.2, Ef 7.3 and Ef 7.4) or siphophages (Ef6.1, Ef6.4 and Ef 7.1). Scale bar, 50 nm. b–h, C57BL/6 germ-free mice were colonized with faeces from two cytolysin-negative patients with alcoholic hepatitis and subjected to the chronic–binge feeding model, gavaged with control phages against C. crescentus (1010 PFUs) or a cocktail of four different phages that target non-

cytolytic E. faecalis (1010 PFUs), 1 day before an ethanol binge. b, Serum levels of ALT. c, Hepatic triglyceride content. d, Representative H & E-stained liver sections. Scale bar, 100 μm. e–g, Hepatic levels of mRNAs that encode Il1b, Cxcl1 and Col1a1. h, Faecal colony-forming units (CFUs) of Enterococcus. Results are expressed as mean ± s.e.m. (b, c, e–h). P values are determined by two-way ANOVA with Tukey’s post hoc test (b, c, e–h). All results were generated from at least three independent replicates. The exact group size (n) and P values for each comparison are listed in Supplementary Table 10. *P < 0.05.

510 | Nature | Vol 575 | 21 November 2019

Article Compared to Atp4aSl/Sl mice gavaged with control phages or vehicle, Atp4aSl/Sl mice gavaged with phages that target cytolytic E. faecalis had less severe liver injury, steatosis and inflammation after chronic ethanol feeding (Extended Data Fig. 5a–f ). Administration of E. faecalis phages significantly reduced levels of cytolysin in the liver (Extended Data Fig. 5g) as well as faecal amounts of Enterococcus (Extended Data Fig. 5h). Phage administration did not affect the overall composition of the faecal microbiome, intestinal absorption or hepatic metabolism of ethanol (Extended Data Fig. 5i–k).

To develop a therapeutic approach to precisely edit the intestinal microbiota, we cultured cytolytic E. faecalis strains from the faecal samples of patients with alcoholic hepatitis. We then isolated lytic phages from sewage water against these cytolytic E. faecalis strains; these phages had siphophage or myophage morphology (Fig. 3a, Extended Data Fig. 6). Gnotobiotic mice were colonized with faeces from two cytolysin-positive patients with alcoholic hepatitis (Sup- plementary Table 8) and given three or four different—but patient- specific—lytic phages against cytolytic E. faecalis. The phages against cytolytic E. faecalis abolished ethanol-induced liver injury and steatosis, as shown by lower levels of ALT, lower percentages of hepatic cells positive for terminal deoxynucleotide transferase-mediated dUTP nick-end labelling, and lower levels of hepatic triglycerides and oil red O-staining (Fig. 3b–d, Extended Data Fig. 7a, b), as well as by decreased hepatic levels of Il1b, Cxcl1, Cxcl2, Col1a1 and Acta2 mRNAs, and reduced hepatic levels of cylLS, as compared with mice given control phages (against C. crescentus) (Fig. 3e–h, Extended Data Fig. 7c, d). Treatment with phages against cytolytic E. faecalis also reduced faecal amounts of Enterococcus (Extended Data Fig. 7e) without affecting the overall composition of the gut microbiota (Extended Data Fig. 7f ). Intestinal absorption of ethanol and hepatic metabolism were similar in all groups (Extended Data Fig. 7g, h).

To demonstrate that the effect of phage treatment occurs via the targeting of cytolysin-positive E. faecalis, rather than a reduction in cytolysin-negative E. faecalis, we colonized gnotobiotic mice with faeces from cytolysin-negative patients with alcoholic hepatitis (Sup- plementary Table 8). Phages against non-cytolytic E. faecalis from patients were isolated from sewage water; they had siphophage or podophage morphology (Fig. 4a, Extended Data Fig. 8). These phages did not reduce features of ethanol-induced liver disease compared with control phages (Fig. 4b–g, Extended Data Fig. 9a–h), despite the reduc- tion of faecal Enterococcus (Fig. 4h). Our findings indicate that treat- ment with lytic phages can selectively attenuate the ethanol-induced liver disease caused by cytolysin-positive E. faecalis in humanized mice.

Discussion Phage-based therapies have predominantly been studied in patients with bacterial infections in the gastrointestinal tract21–23, urinary tract24,25 and other organ systems26–28. The results of these studies— although mixed in terms of efficacy—strongly suggest that phage treat- ment offers a safe alternative to antibiotics26,27. However, safety studies are required for complex populations (such as patients with alcoholic hepatitis), because phages can induce a strong immune reaction29. Further work is required to determine whether phages that target cyto- lytic E. faecalis might be used to treat patients with alcoholic hepatitis, a life-threatening disease that at present has no effective treatment. Eradication of this specific bacterial strain might produce better out- comes than current treatments, and environmental sources can be used to easily isolate phages that target cytolysin-positive E. faecalis. Here we provide an example of the efficacy of approaches based on phages in mice for a disease that is not considered a classic infectious disease. Our data also suggest that cytolysin may be used as a predictive biomarker of severe alcoholic hepatitis; an independent, prospective

cohort is therefore needed to validate cytolysin as a biomarker, and to extend the phage findings in mice to human patients.

Online content Any methods, additional references, Nature Research reporting sum- maries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author con- tributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-019-1742-x.

1. Lozano, R. et al. Global and regional mortality from 235 causes of death for 20 age groups in 1990 and 2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380, 2095–2128 (2012).

2. Lee, B. P., Vittinghoff, E., Dodge, J. L., Cullaro, G. & Terrault, N. A. National trends and long- term outcomes of liver transplant for alcohol-associated liver disease in the United States. JAMA Intern. Med. 179, 340–348 (2019).

3. Rehm, J. et al. Burden of disease associated with alcohol use disorders in the United States. Alcohol. Clin. Exp. Res. 38, 1068–1077 (2014).

4. Llopis, M. et al. Intestinal microbiota contributes to individual susceptibility to alcoholic liver disease. Gut 65, 830–839 (2016).

5. Ike, Y., Clewell, D. B., Segarra, R. A. & Gilmore, M. S. Genetic analysis of the pAD1 hemolysin/bacteriocin determinant in Enterococcus faecalis: Tn917 insertional mutagenesis and cloning. J. Bacteriol. 172, 155–163 (1990).

6. Tang, W. & van der Donk, W. A. The sequence of the enterococcal cytolysin imparts unusual lanthionine stereochemistry. Nat. Chem. Biol. 9, 157–159 (2013).

7. Maddrey, W. C. et al. Corticosteroid therapy of alcoholic hepatitis. Gastroenterology 75, 193–199 (1978).

8. Dominguez, M. et al. A new scoring system for prognostic stratification of patients with alcoholic hepatitis. Am. J. Gastroenterol. 103, 2747–2756 (2008).

9. Thursz, M. R. et al. Prednisolone or pentoxifylline for alcoholic hepatitis. N. Engl. J. Med. 372, 1619–1628 (2015).

10. Mathurin, P. & Lucey, M. R. Management of alcoholic hepatitis. J. Hepatol. 56, S39–S45 (2012).

11. Llorente, C. et al. Gastric acid suppression promotes alcoholic liver disease by inducing overgrowth of intestinal Enterococcus. Nat. Commun. 8, 837 (2017).

12. Gilmore, M. S. et al. Genetic structure of the Enterococcus faecalis plasmid pAD1- encoded cytolytic toxin system and its relationship to lantibiotic determinants. J. Bacteriol. 176, 7335–7344 (1994).

13. Cox, C. R., Coburn, P. S. & Gilmore, M. S. Enterococcal cytolysin: a novel two component peptide system that serves as a bacterial defense against eukaryotic and prokaryotic cells. Curr. Protein Pept. Sci. 6, 77–84 (2005).

14. Van Tyne, D., Martin, M. J. & Gilmore, M. S. Structure, function, and biology of the Enterococcus faecalis cytolysin. Toxins 5, 895–911 (2013).

15. Bertola, A., Mathews, S., Ki, S. H., Wang, H. & Gao, B. Mouse model of chronic and binge ethanol feeding (the NIAAA model). Nat. Protoc. 8, 627–637 (2013).

16. Ogilvie, L. A. & Jones, B. V. The human gut virome: a multifaceted majority. Front. Microbiol. 6, 918 (2015).

17. Chatterjee, A. et al. Bacteriophage resistance alters antibiotic-mediated intestinal expansion of enterococci. Infect. Immun. 87, e00085-19 (2019).

18. Poindexter, J. S. Biological properties and classification of the Caulobacter group. Bacteriol. Rev. 28, 231–295 (1964).

19. Shin, J. et al. Analysis of the mouse gut microbiome using full-length 16S rRNA amplicon sequencing. Sci. Rep. 6, 29681 (2016).

20. Human Microbiome Project Consortium. Structure, function and diversity of the healthy human microbiome. Nature 486, 207–214 (2012).

21. Marcuk, L. M. et al. Clinical studies of the use of bacteriophage in the treatment of cholera. Bull. World Health Organ. 45, 77–83 (1971).

22. Sarker, S. A. et al. Oral phage therapy of acute bacterial diarrhea with two coliphage preparations: a randomized trial in children from Bangladesh. EBioMedicine 4, 124–137 (2016).

23. Dalmasso, M., Hill, C. & Ross, R. P. Exploiting gut bacteriophages for human health. Trends Microbiol. 22, 399–405 (2014).

24. Ujmajuridze, A. et al. Adapted bacteriophages for treating urinary tract infections. Front. Microbiol. 9, 1832 (2018).

25. Khawaldeh, A. et al. Bacteriophage therapy for refractory Pseudomonas aeruginosa urinary tract infection. J. Med. Microbiol. 60, 1697–1700 (2011).

26. Schooley, R. T. et al. Development and use of personalized bacteriophage-based therapeutic cocktails to treat a patient with a disseminated resistant Acinetobacter baumannii infection. Antimicrob. Agents Chemother. 61, e00954-17 (2017).

27. Dedrick, R. M. et al. Engineered bacteriophages for treatment of a patient with a disseminated drug-resistant Mycobacterium abscessus. Nat. Med. 25, 730–733 (2019).

28. Fish, R. et al. Compassionate use of bacteriophage therapy for foot ulcer treatment as an effective step for moving toward clinical trials. Methods Mol. Biol. 1693, 159–170 (2018).

29. Górski, A. et al. Phages and immunomodulation. Future Microbiol. 12, 905–914 (2017).

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© The Author(s), under exclusive licence to Springer Nature Limited 2019