Immunoglobulin Y antibodies against cytolysin reduce ethanol-induced liver disease in mice

Abstract

Background:

Patients with severe alcohol-associated hepatitis have a high morbidity and mortality. Novel therapeutic approaches are urgently needed. Aims of our study were to confirm the predictive value of cytolysin-positive Enterococcus faecalis (E. faecalis) for mortality in patients with alcohol-associated hepatitis and to assess the protective effect of specific chicken immunoglobulin Y (IgY) antibodies against cytolysin in vitro and in a microbiota humanized mouse model of ethanol-induced liver disease.

Approach and Results:

We investigated a multicenter cohort of 26 subjects with alcohol-associated hepatitis and confirmed our previous findings that presence of fecal cytolysin-positive Enterococcus faecalis predicted 180-day mortality in those patients. After combining this smaller cohort with our previously published multicenter cohort, presence of fecal cytolysin has a better diagnostic area under the curve, better other accuracy measures, and a higher odds ratio to predict death in patients with alcohol-associated hepatitis than other commonly used liver disease models. In a precision medicine approach, we generated IgY antibodies against cytolysin from hyperimmunized chickens. Neutralizing IgY antibodies against cytolysin reduced cytolysin-induced cell death in primary mouse hepatocytes. Oral administration of IgY antibodies against cytolysin decreased ethanol-induced liver disease in gnotobiotic mice colonized with stool from cytolysin-positive patients with alcohol-associated hepatitis.

Conclusion:

E. faecalis cytolysin is an important mortality predictor in alcohol-associated hepatitis patients and its targeted neutralization via specific antibodies improves ethanol-induced liver disease in microbiota humanized mice.

Graphical Abstract

Introduction

Alcohol-associated liver disease is common with over 26 million people suffering from cirrhosis related to alcohol use worldwide and it has significant associated morbidity and mortality (1). Especially alcohol-associated hepatitis has an abysmal prognosis with 18.5% 28-day mortality according to a recent meta-analysis (2). The same meta-analysis found that corticosteroids nor pentoxifylline improve 180-day mortality in patients with alcohol-associated hepatitis (2). Thus, new therapeutics are desperately needed. Various liver and gut diseases are associated with intestinal dysbiosis, a reduction of beneficial microbes and proliferation of potentially pathogenic microbes (3). In particular, patients with alcohol-associated liver disease show marked microbiome alterations, including bacterial, fungal, and viral changes (4–8). Recently, we have shown that Enterococcus faecalis (E. faecalis) and its two-subunit exotoxin cytolysin are increased in the stool of patients with alcohol-associated hepatitis compared with non-alcoholic controls and patients with alcohol use disorder (9). Presence of cytolysin-positive E. faecalis predicts mortality in patients with alcohol-associated hepatitis (9). Treatment with bacteriophages specifically targeting cytolytic E. faecalis, reduced ethanol-induced liver disease in microbiota humanized mice colonized with stool from patients with alcohol-associated hepatitis (9, 10).

New emergent strategies using antigen-specific antibodies have shown positive effects in the prevention and treatment of bacterial infections (11). In particular, the use of oral chicken immunoglobulin Y (IgY) as a novel mode of immunotherapy to confer passive immunity has gained much interest as an inexpensive non-antibiotic alternative for the prophylaxis and treatment of a wide variety of infectious diseases. Chicken IgY antibodies are derived from the yolk of eggs from hyperimmunized chickens. Chickens can be immunized with antigens, pathobionts or pathogens, and serum IgY antibodies are transported into the yolk sac (12). The IgY yield from yolk remains high until at least day 81 (13), and approximately 1500 mg of IgY can be harvested each month from each laying hen (5–25 mg/egg yolk) (14). This makes IgY antibody production quick and inexpensive. IgY antibodies do not activate mammalian complement (15) and are stable between pH values from 3.5 to 11.0 (16). Orally administered IgY antibodies are active against several enteric pathogens, e.g. Rotavirus, Escherichia coli, Salmonella and Shigella, in humans (17).

In the current study, we aimed i) to confirm the predictive value of cytolysin-secreting E. faecalis for mortality in alcohol-associated hepatitis, and ii) to investigate the protective effect of specific IgY antibodies against cytolysin in vitro as well as in a microbiota humanized mouse model of ethanol-induced liver disease.

Materials and methods

Patient cohorts

Patient cohorts have been described (9, 18). We evaluated 26 patients with alcohol-associated hepatitis for cytolysin stool positivity in our primary cohort, as described in Table 1, and added another subset cohort of 27 patients with alcohol-associated hepatitis from a prior publication with similar clinical and biochemical profiles and available alive or dead status at 180 days (9) for further statistical analysis. We equated need for liver transplant with death for the purpose of our analysis, so calculated transplant-free survival. Alcohol-associated hepatitis patients were enrolled from the InTeam Consortium (ClinicalTrials.gov identifier number: NCT02075918) from centers in the USA, Mexico, Canada, United Kingdom, France and Spain. Inclusion criteria were active alcohol abuse (> 50 g/day for men and > 40 g/day for women) in the last 3 months, aspartate aminotransferase (AST) > alanine aminotransferase (ALT) and total bilirubin > 3 mg/dl in the past 3 months, liver biopsy and/or clinical picture consistent with alcohol-associated hepatitis. Exclusion criteria were autoimmune liver disease (antinuclear antibody (ANA) > 1:320), chronic viral hepatitis, hepatocellular carcinoma, complete portal vein thrombosis, extrahepatic terminal disease, pregnancy, and lack of signed informed consent. In all patients, the clinical picture was consistent with alcohol-associated hepatitis and in patients who underwent liver biopsy, the histology was in line with the diagnosis of alcohol-associated hepatitis. Liver biopsies were only done if clinically indicated as part of routine clinical care for diagnostic purposes of alcohol-associated hepatitis. Biospecimens were collected during their admission to the hospital. Patients were censored at the time point they were last seen alive. The model for end-stage liver disease (MELD) score, Fibrosis-4 index (FIB-4), ‘age, serum bilirubin, international normalized ratio (INR), and serum creatinine’ score (ABIC), and Maddrey’s discriminant function (Maddrey’s DF) were calculated from all alcohol-associated hepatitis patients from whom respective laboratory values were available. The protocol was approved by the Ethics Committee of Hôpital Huriez (Lille, France), Universidad Autonoma de Nuevo Leon (Monterrey, México), Hospital Universitario Vall d’Hebron (Barcelona, Spain), King’s College London (London, UK), University of Alberta (Edmonton, Canada), Yale University (New Haven, USA), University of North Carolina at Chapel Hill (Chapel Hill, USA), Weill Cornell Medical College (New York, USA), Columbia University (New York, USA), University of Wisconsin (Madison, USA), VA San Diego Healthcare System (San Diego, USA), and University of California San Diego (La Jolla, USA). Patients were enrolled after written informed consent was obtained from each patient.

Table 1.

Baseline demographic and laboratory data of the study population.

	n	Alcohol-associated Hepatitis (n=26)
Gender (male), n	26	16 (61.5%)
Age (years)	26	48.6 [41.5;53.1]
BMI (kg/m2)	24	29.4 [25.4;34.5]
AST (IU/L)	26	150 [84.0;218]
ALT (IU/L)	26	50.5 [35.0;75.0]
GGT (IU/L)	14	400 [156;2288]
AP (IU/L)	26	182 [135;241]
Bilirubin (mg/dL)	26	13.8 [6.09;22.6]
Albumin (g/dL)	26	2.55 [2.12;2.88]
INR	26	1.90 [1.60;2.31]
Creatinine (mg/dL)	26	0.76 [0.59;1.19]
Platelets (109/L)	26	138 [79.5;176]
MELD	26	23.5 [19.9;29.5]
FIB-4	26	6.34 [4.51;11.7]
FIB-4 > 3.25 (F3-F4), n	26	16 (61.5%)
Cytolysin positive, n	26	3 (11.5%)

Values are presented as median and interquartile range in brackets. The number of subjects for which data were available is indicated in the second column. ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; BMI, body mass index; FIB-4, Fibrosis-4 Index; GGT, gamma-glutamyltransferase; INR, international normalized ratio; MELD, model for end-stage liver disease.

Mice

Stool samples from cytolysin-positive patients with alcohol-associated hepatitis (Suppl. Table 4) were used for fecal transplantation in male and female C57BL/6 germ-free mice, which were bred at UC San Diego (9). Briefly, mice were gavaged with 100 μL of stool samples (1 g stool dissolved in 30 ml Luria-Bertani (LB) medium containing 15% glycerol under anaerobic conditions), starting at an age of 4–5 weeks and repeated two weeks later. Two weeks after the second gavage, mice were placed on a chronic–binge ethanol diet (NIAAA model), or control (isocaloric) diet as described (19). Mice were fed with Lieber DeCarli diet, and the caloric intake from ethanol was 0% on days 1–5 and 36% from day 6 until the end of the study period. At day 16, mice were gavaged with a single dose of ethanol (5 g/kg body weight) in the early morning and sacrificed 9 hours later. Pair-fed control mice received a diet with an isocaloric substitution of dextrose.

To assess the therapeutic efficacy of specific IgYs, egg powder containing IgY antibodies was added to the liquid diet at the start of the ethanol feeding administration (day 6). The following antibodies were used: control IgY (15 mg/day), anti-E.faecalis/Cyl IgY (against E. faecalis and cytolysin) (1, 5 and 15 mg per day) and anti-cytolysin (15 mg/day). Diets were replaced every third day. Mice were randomly assigned into groups at the beginning of the study (Figure 4a). All mouse studies were reviewed and approved by the Institutional Animal Care and Use Committee of the University of California, San Diego.

Statistical Analysis

Human data were expressed as median and interquartile range for each continuous outcome, if not stated otherwise. Area under the curve, best threshold to maximize the youden index, sensitivity, specificity, accuracy, precision, positive predictive value, and negative predictive value were calculated using the pROC library in R. Kaplan-Meier curves to visualize survival were created employing the survival library in R. Mean decrease Gini score and mean decrease accuracy were calculated with the randomForest library, and odds ratios with the questionr library in R. Continuous variables were compared using the Mann-Whitney test. Categorical variables were compared using the Pearson’s Chi-squared test. Results of the mouse studies are expressed as mean ± s.e.m. For mouse studies, the significance was evaluated using One or Two-way analysis of variance (ANOVA) with Tukey’s post-hoc test or two-stage step-up method of Benjamini, Krieger and Yekutieli. A value of p < 0.05 was considered to be statistically significant. Statistical analyses were performed using R statistical software, R version 1.3.1093, 2020 the R Foundation for Statistical Computing and GraphPad Prism v8.4.3

Additional methods can be found in Supplementary Material.

Results

Study population with alcohol-associated hepatitis

The study population from 8 international centers comprised 26 patients with alcohol-associated hepatitis, of whom 16 were male (Table 1). Median age was 48.6 years and median body mass index (BMI) was 29.4 kg/m². Liver disease severity of the cohort was characterized by a median aspartate aminotransferase (AST) of 150 IU/L, alanine aminotransferase (ALT) of 50.5 IU/L, and total bilirubin of 13.8 mg/dL. The median model for end-stage liver disease (MELD) and Fibrosis-4 index (FIB-4) were 23.5 and 6.34, respectively. Cytolysin-positive Enterococcus faecalis (E. faecalis) could be detected in the feces of 3 out of the 26 subjects with alcohol-associated hepatitis (Supplementary Figure 1).

Presence of fecal cytolysin predicts 180-day mortality in patients with alcohol-associated hepatitis

Our first aim was to confirm our prior findings and assess whether fecal E. faecalis cytolysin positivity can predict death in subjects with alcohol-associated hepatitis in a separate cohort (9). Four patients did not have study exit or mortality data available, so 22 subjects were included in the analysis. And indeed, cytolysin-positive patients with alcohol-associated hepatitis of our cohort had a significantly worse 180-day transplant-free survival than patients with alcohol-associated hepatitis, in whom cytolysin could not be detected (Figure 1). To show consistency between our studies that survival depends on cytolysin positivity in alcohol-associated hepatitis, we combined this cohort with another previously described multicenter alcohol-associated hepatitis cohort (9), and included a subset of participants from the latter who were confirmed dead or alive after 180 days of enrollment (Supplementary Table 1, see also the fifth column of Table 2). Cytolysin-negative subjects with alcohol-associated hepatitis had a significantly higher 180-day transplant-free survival than cytolysin-positive participants with alcohol-associated hepatitis in this combined population (Supplementary Figure 2).

Figure 1. Cytolysin positive patients with alcohol-associated hepatitis have a significantly worse 180-day liver transplant-free survival than cytolysin negative patients with alcohol-associated hepatitis.

Cytolysin positive n=3, cytolysin negative n=19.

Table 2.

Baseline demographic and laboratory data of alcohol-associated hepatitis cohorts confirmed dead or alive at 180 days included for mortality statistics.

	n	Total (n=37)	New Cohort (n=10)	Prior Cohort (n=27)	P value
Gender (male), n	37	23 (62.2%)	6 (60.0%)	17 (63.0%)	1.000
Age (years)	37	50.0 [41.0;56.0]	51.1 [41.5;55.6]	50.0 [42.2;57.3]	0.837
BMI (kg/m2)	34	27.6 [24.1;31.7]	28.5 [26.1;36.7]	27.2 [23.1;31.3]	0.335
AST (IU/L)	37	136 [98.0;217]	190 [92.5;241]	128 [101;176]	0.505
ALT (IU/L)	37	48.0 [29.0;72.0]	60.5 [28.2;74.0]	48.0 [29.5;71.5]	0.905
GGT (IU/L)	20	130 [87.0;293]	143 [137;348]	120 [86.0;230]	0.432
AP (IU/L)	36	164 [124;232]	186 [165;233]	135 [117;225]	0.258
Bilirubin (mg/dL)	37	17.3 [9.95;20.5]	14.8 [7.02;19.7]	17.7 [10.9;21.0]	0.356
Albumin (g/dL)	34	2.40 [2.02;2.88]	2.45 [2.10;2.90]	2.35 [1.98;2.82]	0.623
INR	37	2.00 [1.70;2.50]	2.05 [1.82;2.45]	2.00 [1.65;2.48]	0.837
Creatinine (mg/dL)	37	0.99 [0.62;1.37]	0.99 [0.64;1.51]	0.99 [0.62;1.26]	0.771
Platelets (109/L)	35	115 [68.0;154]	83.5 [71.5;170]	123 [67.0;146]	0.913
MELD	37	25.6 [22.5;32.5]	27.4 [21.3;32.0]	25.3 [22.7;30.5]	0.824
FIB-4	35	9.47 [6.29;16.5]	9.48 [6.82;17.5]	9.47 [6.06;15.6]	0.884
FIB-4 > 3.25 (F3-F4), n	35	34 (97.1%)	9 (90.0%)	25 (100%)	0.286
Cytolysin positive, n	37	22 (59.5%)	3 (30.0%)	19 (70.4%)	0.056

Values are presented as median and interquartile range in brackets. The number of subjects for which data were available is indicated in the second column. Continuous variables were compared using the Mann-Whitney test. Categorical variables were compared using the Pearson’s Chi-squared test. Level for statistical significance is P<0.05. ALT, alanine aminotransferase; AP, alkaline phosphatase; AST, aspartate aminotransferase; BMI, body mass index; FIB-4, Fibrosis-4 Index; GGT, gamma-glutamyltransferase; INR, international normalized ratio; MELD, model for end-stage liver disease.

Presence of fecal cytolysin has better predictive mortality features than established liver disease markers in alcohol-associated hepatitis

To achieve a higher power for additional statistical calculations, we combined the alcohol-associated hepatitis cohort characterized in Table 1 with the aforementioned cohort from another previously described multicenter alcohol-associated hepatitis cohort (9), focusing on a subset of subjects who were confirmed dead or alive after 180 days of enrollment (Table 2). We compared the predictive value of cytolysin positivity for 180-day mortality with commonly employed liver disease scores in our combined alcohol-associated hepatitis cohort. We found that cytolysin had a markedly higher area under the curve (AUC) than FIB-4, the ‘age, serum bilirubin, international normalized ratio (INR), and serum creatinine’ score (ABIC), Maddrey’s discriminant function (Maddrey’s DF), and MELD with an AUC of 0.82 versus 0.69, 0.77, 0.68, and 0.80, respectively (Table 3). Similarly, the sensitivity, specificity, accuracy, precision, positive, and negative predictive value (PPV and NPV) for death within 180 days were markedly higher for presence of fecal cytolysin compared with FIB-4, ABIC, Maddrey’s DF, and MELD. The best threshold that maximized the youden index was determined for each marker (Table 3). The odds ratios (ORs) for death within 180 days for values exceeding the best threshold for the respective marker were calculated, and all ORs for cytolysin positivity and high liver disease markers were significant per simple logistic regression: cytolysin OR=24.00 (95% confidence interval 3.54–488.40; p=0.006), high FIB-4 OR=11.50 (1.91–99.13; p=0.012), high ABIC OR=8.75 (1.70–68.02; p=0.016), and high MELD OR=16.89 (2.55–338.98; p=0.013) (Figure 2a). The sensitivity for Maddrey’s DF was 1.00 at the best threshold, implying the absence of false negative values. Thus, the OR for Maddrey’s DF could not be determined. After multiple logistic regression using all markers, only cytolysin positivity retained a significant OR for mortality with 15.03 (95% confidence interval 1.34–473.34; p=0.046) vs high FIB-4 OR=7.39 (0.35–234.20; p=0.175), high ABIC OR=1.67 (0.05–27.91; p=0.720), and high MELD OR=6.37 (0.48–165.55; p=0.172) (Figure 2b).

Table 3.

Predictive accuracy for mortality of cytolysin in comparison with established liver disease markers.

	AUC	Threshold	Sens	Spec	Acc	Prec	PPV	NPV
Cytolysin	0.82	detectable	0.75	0.89	0.78	0.95	0.95	0.53
FIB-4	0.69	6.53	0.82	0.71	0.80	0.92	0.92	0.50
ABIC	0.77	8.28	0.71	0.78	0.73	0.91	0.91	0.47
Maddrey’s DF	0.68	29.17	1.00	0.40	0.88	0.86	0.86	1.00
MELD	0.80	25.13	0.68	0.89	0.73	0.95	0.95	0.47

The best threshold was determined to maximize the youden index (= sensitivity + specificity) for each marker. Cytolysin n=37, FIB-4 n=35, ABIC n=37, Maddrey’s DF n=24, and MELD score n=37. ABIC, Age, serum bilirubin, INR, and serum creatinine score; acc, accuracy; AUC, area under the curve; FIB-4, Fibrosis-4 Index; INR, international normalized ratio; Maddrey’s DF, Maddrey’s discriminant function; MELD, model for end-stage liver disease; NPV, negative predictive value; PPV, positive predictive value; pre, precision; sens, sensitivity; spec, specificity.

Figure 2. Cytolysin positivity has a higher predictive value for mortality in patients with alcohol-associated hepatitis than established liver disease composite scores and various liver biopsy findings.

(a-b) Odds ratios for death within 180 days per (a) simple logistic regression and (b) multiple logistic regression were calculated for cytolysin positivity and various liver disease markers for values that exceeded the best thresholds per Table 3 (95% confidence interval indicated by black brackets for each marker). Cytolysin n=37, FIB-4 n=35, ABIC n=37, and MELD score n=37. (c) Mean decrease Gini score and (d) mean decrease accuracy by random forest analysis were quantitated for presence of cytolysin and multiple liver biopsy findings to determine their respective feature importance for death within 180 days (n=15). Significance is indicated by *P<0.05, and **P<0.01. ABIC, Age, serum bilirubin, INR, and serum creatinine score; FIB-4, Fibrosis-4 Index; INR, international normalized ratio; MELD, model for end-stage liver disease; PMN, polymorphonuclear leukocytes.

Cytolysin positivity has a higher predictive value for mortality in patients with alcohol-associated hepatitis than liver histology features

Given several reports about the predictive accuracy of liver biopsy findings for mortality in subjects with alcohol-associated hepatitis, including degree of fibrosis, polymorphonuclear leukocytes (PMN) infiltration, presence and degree of bilirubinostasis, and presence of giant mitochondria (20, 21), we aimed to compare the predictive value for mortality for fecal cytolysin with these histologic markers (see Supplementary Table 2 for demographic and laboratory data of the combined cohort with liver biopsies). We performed random forest analysis and found that cytolysin positivity has a higher feature importance for death within 180 days with a higher mean decrease Gini score (Figure 2c) and higher mean decrease accuracy (Figure 2d) than either of the aforementioned liver histology components.

IgY antibodies against cytolysin reduce the cytotoxic effects of cytolysin-positive E. faecalis

The described results demonstrate the high significance of cytolysin for outcome in human subjects with alcohol-associated hepatitis. We have previously demonstrated that targeting cytolysin-positive E. faecalis with bacteriophages reduces ethanol-induced liver disease in a microbiota humanized mouse model (9). Our second aim was to further develop this precision medicine approach by using specific antibodies targeting cytolysin to neutralize its toxic effects. Neutralizing polyclonal antibodies specifically target the pathogens in the body without affecting the microbiome (11) and is thus an effective alternative to antibiotics. We therefore immunized chickens to generate IgY antibodies. IgY from hyperimmunized chicken egg yolks was extracted and specific activity of IgY antibody against E. faecalis and cytolysin antigens were measured by indirect ELISA. ELISA results indicated positive binding of IgY to the specific antigens when compared with the control IgY with a high titer of 1:32,000 (Supplementary Figure 3a–b). The specific antibody titer to cytolysin alone is > 1:80,000 (Supplementary Figure 3c).

To neutralize the effects of cytolysin produced by E. faecalis, we used freeze-dried egg powder containing specific chicken IgY antibodies raised against cytolysin alone (anti-Cyl IgY) and E. faecalis, E. faecium and cytolysin (anti-Tri IgY). Control IgY served as experimental control. To determine the effect of IgY antibodies on native cytolysin produced and secreted from E. faecalis, we used a hemolysis assay with red blood cells. Culture supernatant from cytolysin-positive E. faecalis, cytolysin-negative E. faecalis (E. faecalis^Δcyl), Gram-positive Staphylococcus aureus (S. aureus) and Gram-negative Escherichia coli (E. coli) was incubated with red blood cells. Supernatant from cytolysin-positive E. faecalis caused a significant increase in hemolysis, while supernatant from E. coli and S. aureus did not (Figure 3a). Supernatant from cytolysin-negative E. faecalis caused significantly less hemolysis as compared with cytolysin-positive E. faecalis. Adding control IgY did not affect hemolysis induced by cytolysin-positive E. faecalis (Figure 3a). Anti-Cyl IgY [10 ng/mL, 100 μg/mL and 200 μg/mL] and anti-Tri IgY [100 μg/mL and 200 μg/mL] reduced hemolysis induced by cytolysin-positive E. faecalis as compared with hemolysis induced by cytolysin-positive E. faecalis alone (Supplementary Figure 4 and Supplementary Table 3). Red blood cell death was reduced to the level of hemolysis induced by cytolysin-negative E. faecalis alone (Figure 3a). These results indicate that our specific IgY antibodies effectively neutralize endogenous cytolysin produced and secreted by E. faecalis.

Figure 3. Anti-Tri and anti-cytolysin IgY antibodies reduce the cytotoxic effect of E. faecalis cytolysin.

(a) Red blood cell (RBC) culture assay using anti-Tri IgY (against E. faecalis, E. faecium and cytolysin) and anti-cytolysin IgY with a final concentration of 200 μg/mL. Red blood cells were incubated with culture supernatant from cytolysin positive and negative E. faecalis, E. coli and S. aureus. 7–8 independent RBC cultures were determined in all conditions (Supplemental Figure 4). Determination of cell death in primary mouse hepatocytes (3–6 independent isolations) (b-c) and human primary hepatocytes (4 independent isolations) (d) with lactate dehydrogenase (LDH) assay to measure cytotoxicity of cytolysin protein at indicated concentrations and over time. (e) LDH assay was used to evaluate the efficacy of IgY antibodies against recombinant cytolysin protein. Survival of hepatocytes was determined in 3 independent experiments of hepatocytes isolated from C57BL/6 mice. Results expressed as mean ± s.e.m. P values are determined by One-way ANOVA corrected for Two-stage step-up method of Benjamini, Krieger and Yekutieli. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.

Previous results from our group demonstrated that the combination of the recombinant large (CylL_L) and small (CylL_s) cytolysin subunits induces cell death in primary mouse hepatocytes (9). In this study, to better understand the role of the cytolytic E. faecalis we used an in vitro assay to measure hepatocyte cytotoxicity with a recombinant fusion protein of the two cytolysin subunits (Supplementary Figure 5). Treatment of primary mouse hepatocytes with cytolysin caused a dose-dependent increase in cell death with a maximum cytotoxicity using 800 nM after 3 hours, compared with hepatocytes that were incubated with vehicle (Figure 3b–c). To confirm our findings in human cells, primary human hepatocytes were treated with 800 nM cytolysin for 3 hours. Cytolysin significantly increase the cytotoxicity in human primary hepatocytes (Figure 3d).

To determine the effect of anti-Tri and anti-Cyl IgY antibodies for cytolysin-induced cell death, we pre-treated primary mouse hepatocytes with cytolysin (800 nM) for 15 minutes, and then we added anti-Tri IgY and anti-Cyl IgY (200 μg/mL) to the culture medium. A significant reduction of cell death became manifest 3 hours after the incubation with anti-Tri and anti-Cyl specific IgYs as compared to the no treatment or control IgY treated groups (Figure 3e). These results show that IgY against cytolysin reduces cytolysin-induced cell death in cultured primary mouse hepatocytes.

IgY antibodies against cytolysin attenuate ethanol-induced liver disease in gnotobiotic mice

We have previously shown that gnotobiotic mice colonized with stool from cytolysin-positive patients with alcohol-associated hepatitis develop more ethanol-induced liver disease as compared with mice colonized with stool from cytolysin-negative patients with alcohol-associated hepatitis (9). In addition, mice gavaged with cytolysin-positive E. faecalis show more severe ethanol-induced liver disease than mice gavaged with an isogenic E. faecalis strain lacking cytolysin (9). Our aim was to determine whether IgY antibodies against cytolysin reduce ethanol-induced liver disease in gnotobiotic mice. In preliminary dose findings studies, we used freeze-dried egg powder containing specific chicken IgY antibodies against E. faecalis and cytolysin (anti-Ef/Cyl IgY) and control IgY as experimental control. We colonized germ-free mice with stool from a cytolysin-positive patient with alcohol-associated hepatitis (Supplementary Table 4, Patient 1), subjected them to the chronic-binge ethanol feeding model and supplemented their liquid diet with vehicle, control IgY (15 mg/day) or three different doses of anti-Ef/Cyl IgY (1, 5 and 15 mg per day) (Figure 4a). Body weight, liver weight and food intake were not significantly different in ethanol-fed groups (Supplementary Figure 6a–c). Drug consumption was similar to the expected amount (Supplementary Figure 6d). Treatment with anti-Ef/Cyl did not affect systemic levels of ethanol, or hepatic expression of ethanol metabolizing genes Adh1 and Cyp2e1 (Supplementary Figure 6e–g). Mice receiving 5 mg and 15 mg anti-Ef/Cyl showed significantly less liver injury as indicated by a decreased serum ALT level as compared with mice receiving control IgY following chronic ethanol feeding (Supplementary Figure 6h). No significant differences were seen in hepatic triglycerides and hepatic cholesterol (Supplementary Figure 6i–j), although there was a trend toward lower hepatic steatosis in the anti-Ef/Cyl treated mice. Histological analysis confirmed the attenuation of lipid accumulation in the liver after anti-Ef/Cyl treatment (Supplementary Figure 6k). To explain the reduction of liver damage after the anti-Ef/Cyl treatment, we measured the translocation of cytolytic E. faecalis to the liver after ethanol administration. CylL_s was significantly reduced in the liver of mice treated with anti-Ef/Cyl administration, as compared with mice given control-IgY (Supplementary Figure 6l).

Figure 4. Anti-cytolysin and anti-E. faecalis/cytolysin IgY attenuate ethanol-induced liver disease in gnotobiotic mice colonized with cytolysin-positive feces.

C57BL/6 germ-free mice were colonized with feces from cytolysin-positive patients with alcohol-associated hepatitis and subjected to the chronic-binge ethanol feeding model (n=11 per group). Egg powder containing the following IgY antibodies: control-IgY (15 mg/day), anti-E.faecalis/Cyl IgY (15 mg/day) and anti-Cyl (15 mg/day) were added to the liquid diet. Control mice were fed an isocaloric diet (n=5 per group). (a) Experimental design. (b) Serum level of alanine aminotransferase (ALT). (c) Hepatic triglyceride content. (d) Hepatic cholesterol content. (e) Representative Oil Red O-stained liver sections. Hepatic CCl2 (f), Mpo (g), Lys6g (h) and Col1a1 (i) mRNA. (j) Serum LPS measured by ELISA. (k) Proportion of mice that were positive for cytolysin in the liver, measure by qPCR for CylL_S. Results expressed as mean ± s.e.m. P values are determined by Two-way ANOVA with Tukey’s post-hoc test or Two-stage step-up method of Benjamini, Krieger and Yekutieli. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Bar size = 100 μm. 

Based on these preliminary results, our subsequent experiments were performed with a dose of 15 mg anti-IgY antibodies per day. Furthermore, we extended our study using a specific anti-cytolysin (anti-Cyl) neutralizing IgY and IgY antibodies raised against E. faecalis and cytolysin (anti-Ef/Cyl) in microbiota-humanized mice subjected to a chronic–binge ethanol diet. Gnotobiotic mice were colonized with stool from three different cytolysin-positive patients with alcohol-associated hepatitis (Supplementary Table 4). No significant differences were found in body weight, liver weight, food intake and drug consumption (Supplementary Figure 7a–d). Mice treated with anti-Ef/Cyl and anti-Cyl IgY in the liquid diet showed less severe ethanol-induced liver injury, as indicated by lower level of ALT, and lower lipid accumulation, as shown by reduced hepatic triglycerides and cholesterol, and Oil Red O stained liver sections (Figure 4b–e). Mice treated with anti-Cyl IgY also had reduced hepatic expression of inflammation-related genes such as of chemokine C-C motif ligand 2 (Ccl2), myeloperoxidase (Mpo), a marker of neutrophil azurophilic granules, lymphocyte antigen 6 complex locus G6D (Lys6g), a marker for monocytes, granulocytes and neutrophils, and the fibrosis-related gene collagen type I alpha 1 (Col1a1) (Figure 4f–i). Serum LPS was lower after anti-cytolysin IgY as compared with control-IgY antibody treatment (Figure 4j). CylL_s in the liver was significantly decreased in the liver of mice given anti-Ef/Cyl and anti-Cyl IgY, but not in the liver of mice that were treated with control-IgY (Figure 4k). Treatment with neutralizing antibodies did not affect absorption or hepatic metabolism of ethanol, determined by serum levels of ethanol and hepatic gene expression levels of enzymes that metabolize ethanol in the liver, Adh1 and Cyp2e1 (Supplementary Figure 7e–g). These results indicate that treatment with neutralizing IgY antibodies against cytolysin selectively attenuates ethanol-induced liver disease in microbiota humanized mice colonized with stool from cytolysin-positive E. faecalis patients with alcohol-associated hepatitis.

Discussion

In this study, we confirm in an independent multicenter cohort that alcohol-associated hepatitis patients who test positive for cytolysin-secreting E. faecalis have significantly worse 180-day survival than those who test negative for cytolysin-secreting E. faecalis. Further, we show that testing for the presence of fecal cytolysin has a markedly better diagnostic area under the curve, better other accuracy measures, and a higher odds ratio to predict death in alcohol-associated hepatitis patients than commonly employed liver disease scores including MELD, FIB-4, or ABIC. It also is a more important feature to predict 180-day survival than liver biopsy findings. We further show how the toxic and deleterious effects of cytolysin can be neutralized in vitro via specifically designed antibodies targeting cytolysin. We finally demonstrate how these antibodies can be used to alleviate ethanol-induced liver disease in microbiota-humanized mice in a precision medicine approach.

We have previously shown that proton pump inhibitors promote progression of alcohol-associated liver disease, non-alcoholic fatty liver disease, and non-alcoholic steatohepatitis in mice by increasing numbers of intestinal Enterococcus spp., which translocate and lead to hepatic inflammation and hepatocyte death (22). We have also demonstrated that patients suffering from alcohol use disorder are more likely to test positive for E. faecalis and its exotoxin cytolysin in their stool than control subjects, and more alcohol-associated hepatitis patients test positive for E. faecalis and its cytolysin in their stool than patients suffering from alcohol use disorder (9). In that study, we have shown that E. faecalis cytolysin associates with mortality in patients with alcohol-associated hepatitis, which we have confirmed in the current study. Here, we have also detailed further the superior sensitivity, specificity, accuracy, precision, positive and negative predictive values of cytolysin compared with established liver disease composite scores.

Various reports have indicated the predictive value of liver biopsy findings for mortality in patients with alcohol-associated hepatitis, including degree of fibrosis, PMN infiltration, presence and degree of bilirubinostasis, and presence of giant mitochondria (20, 21). We have compared those histologic components with fecal cytolysin and found that cytolysin positivity has a higher feature importance to predict death within 180 days per random forest analysis than any of the liver histology markers, further underlining the prognostic value of cytolysin.

Of note, the excellent predictive value of cytolysin does not appear to extend to all liver diseases, as E. faecalis nor cytolysin are related to disease severity in subjects with non-alcoholic steatohepatitis (23).

Using a precision and microbiome-centered therapeutic approach, we custom-designed chicken IgY antibodies targeting cytolysin. In vitro, anti-cytolysin IgY antibodies had high binding specificity and inhibited the cytotoxic effects of cytolysin in cultured primary mouse hepatocytes and mitigated ethanol-induced liver disease in microbiota humanized mice. Interestingly, IgY against E. faecalis did not add additional protective effects as compared with anti-cytolysin antibodies alone. It also indicates that E. faecalis exerts its disease exacerbating effect via cytolysin. Cytolysin likely causes pore formation in the cell membrane resulting in cell death (24). Interestingly, anti-cytolysin IgY antibody treatment lowered serum LPS and likely stabilized the gut barrier. We did not find differences in serum LPS when C57BL/6 mice were gavaged with PBS, cytolyin positive E. faecalis or an isogenic E. faecalis strain lacking cytolysin in our prior study (9). Oral administration of IgY antibodies provided improved disease outcome in oral and gastrointestinal infections (25), treatment of fungal infections (26), and recently for the diagnosis and treatment of viral infections, such as SARS-CoV-2 (27). IgY antibody therapy offers a precise and relatively inexpensive intervention without affecting the composition of the microbiota. A randomized clinical trial is warranted to evaluate whether the promising results of our designed anti-cytolysin antibodies can be translated to patients with alcohol-associated hepatitis, for whom so far no pharmacological approaches exist with significant long-term benefits (2).

Abbreviations:

Adh1

alcohol dehydrogenase 1

ALT

alanine amino-transferase

ANOVA

analysis of variance

AP

alkaline phosphatase

AST

aspartate aminotransferase

BMI

body mass index

Cxcl2

chemokine (C-X-C motif ligand 2) ligand

Cyp2e1

cytochrome P450 family 2 subfamily E polypeptide 1

E. faecalis

Enterococcus faecalis

E. coli

Escherichia coli

FIB-4

Fibrosis-4 Index

GGT

gamma-glutamyltransferase

IgY

Immunoglobulin Y

INR

international normalized ratio

Maddrey’s DF

Maddrey’s discriminant function

MELD

model for end-stage liver disease

NPV

negative predictive value

PPV

positive predictive value

pre

precision

sens

sensitivity

spec

specificity

qPCR

quantitative PCR

S. aureus

Staphylococcus aureus