Alcohol dehydrogenase–specific T-cell responses are associated with alcohol consumption in patients with alcohol-related cirrhosis§

Authors

  • Fang Lin,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    2. Liver Intensive Care Centre, Beijing 302 Hospital, Beijing, China
    Search for more papers by this author
  • Nicholas J. Taylor,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Haibin Su,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    2. Department of Transplantation, Beijing 302 Hospital, Beijing, China
    Search for more papers by this author
  • Xiaohong Huang,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Munther J. Hussain,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Robin Daniel Abeles,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Laura Blackmore,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Yunyun Zhou,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Mohammad Mashfick Ikbal,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Nigel Heaton,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    2. Liver Transplant Unit, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
  • Wayel Jassem,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    2. Liver Transplant Unit, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
    • *These authors contributed equally to this work.

  • Debbie L. Shawcross,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
    • *These authors contributed equally to this work.

  • Diego Vergani,

    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    Search for more papers by this author
    • *These authors contributed equally to this work.

  • Yun Ma

    Corresponding author
    1. Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, London, United Kingdom
    • M.D., Institute of Liver Studies, School of Medicine at King's College Hospital, King's College London, Denmark Hill, London SE5 9RS, United Kingdom===

    Search for more papers by this author
    • *These authors contributed equally to this work.

    • fax: +44 203 2993760


  • Potential conflict of interest: Nothing to report.

  • Fang Lin and Haibin Su were supported by scholarships from Beijing 302 Hospital, Robin Daniel Abeles was supported by a clinician training fellowship from the National Institute for Health Research, and Debbie L. Shawcross was supported by a 5-year clinical senior lectureship from the Higher Education Funding Council for England.

Abstract

Patients with alcohol-related liver disease (ALD) have antibodies directed to alcohol dehydrogenase (ADH), anti-ADH titers being associated with disease severity and active alcohol consumption. ADH-specific T-cell responses have not been characterized. We aimed to define anti-ADH cellular immune responses and their association with active alcohol consumption and disease severity. Using cultures of peripheral blood mononuclear cells (PBMCs) from 25 patients with alcohol-related cirrhosis (ARC; 12 were actively drinking or abstinent for <6 months, and 13 were abstinent for >6 months) and hepatic mononuclear cells (HMCs) from 14 patients with ARC who were undergoing transplantation, we investigated T-cell reactivity to 25 overlapping peptides representing the full human ADH protein (beta 1 subunit). ADH-specific peripheral T-cell responses were assessed by the quantification of T-cell proliferation and cytokine production and were correlated with the clinical course. In active alcohol consumers, proliferative T-cell responses targeted ADH31-95 and other discontinuous sequences in the ADH peptide, whereas only one sequence was targeted in abstinents. ADH peptides induced the production of interferon-γ (IFN-γ), interleukin-4 (IL-4), and IL-17. IL-4 production was lower in active drinkers versus abstinents, and IL-17 production was higher. Peptides inducing IFN-γ production outnumbered those inducing T-cell proliferation. The intensity of the predominantly T helper 1 (Th1) responses directly correlated with disease severity. Similar to PBMCs in abstinents, ADH peptides induced weak T-cell proliferation and a similar level of IL-4 production in HMCs but less vigorous Th1 and T helper 17 responses. Conclusion: This suggests that Th1 responses to ADH in ARC are induced by alcohol consumption. A Th1/T helper 2 imbalance characterizes T-cell responses in active drinkers with ARC, whereas IL-4 production prevails in abstinents. This identifies new targets for immunoregulatory therapies in ALD patients for halting detrimental effector T-cell responses, which may encourage liver fibrogenesis and progression to end-stage liver disease. (HEPATOLOGY 2013)

There has been an exponential rise in the incidence of alcohol-related liver disease (ALD) globally, and ALD is a major indication for liver transplantation in the Western world.1

There is an urgent need to define the pathogenesis of ALD in order to predict those drinkers who are at highest risk of developing liver disease and to establish novel therapeutic interventions that will treat ALD effectively. Although a close link between ALD and alcohol consumption is well established, the pathogenesis of ALD is not fully understood.1, 2 A number of contributing factors, including a genetic predisposition, environmental influence, and immune-mediated damage, have been considered. Evidence for a genetic predisposition has been sought through the definition of the polymorphisms in genes coding for enzymes metabolizing alcohol or for cytokines, including alcohol dehydrogenase 2 (ADH2), ADH3, cytochrome P450 2E1, ALDH2, interleukin-4 (IL-4), IL-6, IL-8, and IL-12,3, 4 but no direct link has been found between these genetic variations and the development of ALD.5 Similarly, the search for links between environmental influences and ALD has not been rewarding.6 Research on immunological changes, however, has provided abundant data to support the notion that immune-mediated injury plays a major role in the development, perpetuation, and outcome of ALD.7-9 Studies have also shown similarities between ALD and autoimmune liver disease. We and others have reported that several autoantibodies that are diagnostically important in autoimmune hepatitis (AIH), such as anti–liver membrane lipoprotein-specific antibodies, anti-nuclear antibodies (ANA), and anti–smooth muscle antibodies (SMAs), are detectable and associated with the severity of liver disease in patients with ALD.10-12 Cellular immunity also appears to be involved. In the late 1990s, an imbalance between T helper 1 (Th1) and T helper 2 (Th2) responses was reported, with both CD4 and CD8 T cells producing predominantly interferon-γ (IFN-γ) rather than IL-4.8 This type 1 skewing was more evident in active alcohol consumers versus those who had stopped drinking. Recently, T helper 17 (Th17) effector cells were identified in the circulation and livers of patients with ALD and especially in those with acute alcoholic hepatitis; this suggests that Th17 cells may contribute to the pathogenesis of ALD.13 Because the detection of IL-17 receptor–positive cells mainly occurred among hepatic stellate cells, it would be interesting to assess whether hepatic mononuclear cells (HMCs) produce IL-17.

Alcohol is metabolized in the liver by an enzymatic cascade that includes ADH.14 We have previously identified ADH as a target antigen for anti-ADH antibodies in both ALD and AIH,10, 15 and we postulate here that the humoral immune response is complemented by the cellular immune response as is the case for AIH type 2, for which the co-occurrence and interplay of antigen-specific humoral and cellular T-cell responses have been extensively documented.16, 17

The aims of the present study were (1) to define ADH-specific CD4 T-cell responses at the epitope level in alcohol-related cirrhosis (ARC), with a focus on patients who were actively drinking and patients who had stopped drinking; (2) to characterize anti-ADH–specific responses in terms of proliferation and cytokine profiles; and (3) to relate antigen-specific T-cell responses to liver disease severity.

Abbreviations

ADH, alcohol dehydrogenase; AIH, autoimmune hepatitis; ALD, alcohol-related liver disease; ALP, alkaline phosphatase; ANA, anti-nuclear antibody; APC, allophycocyanin; ARC, alcohol-related cirrhosis; AST, aspartate aminotransferase; Cy7, cyanine 7; ELISA, enzyme-linked immunosorbent assay; FITC, fluorescein isothiocyanate; GGT, gamma-glutamyl transferase; HMC, hepatic mononuclear cell; ICCS, intracellular cytokine staining; IFN, interferon; Ig, immunoglobulin; IL, interleukin; MELD, Model for End-Stage Liver Disease; mi, mild inflammatory activity; mo, moderate inflammatory activity; NASH, nonalcoholic steatohepatitis; nd, not done; neg, negative; PBMC, peripheral blood mononuclear cell; PBC, primary biliary cirrhosis; PE, phycoerythrin; PMA, phorbol 12-myristate 13-acetate; pos, positive; PSC, primary sclerosing cholangitis; SI, stimulatory index; SMA, anti–smooth muscle antibody; Th1, T helper 1; Th2, T helper 2; Th17, T helper 17.

Patients and Methods

Ethical Statement

This study was undertaken with the approval of the research ethics committee of King's College Hospital, and written informed consent was obtained from each patient in accordance with the Declaration of Helsinki (1989) of the World Medical Association.

Patients and Controls

Thirty-nine patients with ARC were investigated; peripheral blood was obtained from 25 of these patients. They were recruited at the Institute of Liver Studies at King's College Hospital. The severity of cirrhosis was defined with the Child-Pugh-Turcotte and Model for End-Stage Liver Disease (MELD) scores.18

Inclusion Criteria.

Patients required a diagnosis of ARC based on a history of excessive alcohol consumption accompanied by confirmation of the development of cirrhosis by a combination of at least two of the following modalities: clinical, biochemical, radiological, and histopathological. Group A consisted of 12 patients who either were active alcohol drinkers (n = 9) with an alcohol intake exceeding 60 g/day in males and 40 g/day in females for more than 5 years or had been abstinent for <6 months (n = 3). Group B consisted of 13 patients with periods of abstinence varying from 9 months to 20 years and included 4 patients on the waiting list for liver transplantation. The 6-month cutoff was chosen because of the significant decline in humoral immune responses to ADH 6 months into abstinence.15 HMCs were obtained from the second-passage perfusate of explanted livers from 14 patients with ARC undergoing transplantation. The severity of liver disease was determined in each case with the Child-Pugh-Turcotte and MELD scores. This allowed classification into mild disease [compensated cirrhosis (Child-Pugh A)] or severe ARC [decompensated cirrhosis (Child-Pugh B/C)]. The patients' demographic data, liver biochemistry data, and immunological profiles are shown in Table 1.

Table 1. Demographic, Biochemical, and Histological Data for Patients With ARC and Pathological Controls
 Patients With ARC (n = 39)Pathological Controls: Group D (n = 8)
Group A (n = 12)Group B (n = 13)Group C (n = 14)
  • Group A consisted of active alcohol drinkers and patients who were abstinent for <6 months, group B consisted of patients who were abstinent for >6 months, group C consisted of patients who were abstinent for >6 months and underwent liver transplantation, and group D consisted of patients with non–alcohol-related cirrhosis. All values are presented as medians and ranges unless otherwise indicated.

  • *

    A = 5-6, B = 7-9, and C = 10-15.

  • All controls.

  • The total bilirubin level in group B was higher than the level in group A (P < 0.05) and in patients with non–alcohol-related cirrhosis (P < 0.02); the GGT level in group A was higher than the level in long-term abstinents (P < 0.02) and transplant patients (P < 0.02); and the IgA levels in groups A, B, and C tended to be higher than the level in group D (P = 0.09). There were no differences in other indices between the groups.

  • §

    3-20 μmol/L.

  • 10-50 IU/L.

  • 30-130 IU/L.

  • #

    1-55 IU/L.

  • **

    <5 mg/dL.

  • ††

    0.87-4.12 g/L.

  • ‡‡

    6.34-18.11 g/L.

  • §§

    0.53-2.23 g/L.

Age (years)53 (45-73)57 (37-64)53 (44-67)56 (25-70)
Sex: female/male (n/n)4/84/94/104/4
Child-Pugh cirrhosis*6 (5-7)6 (5-12)9 (5-11)5
MELD score8 (6-14)9 (6-20)11.8 (6-53)6 (6-14)
Total bilirubin (μmol/L),§19 (6-45)26 (12-138)27 (10-295)13 (6-37)
AST (IU/L)58 (18-151)42 (21-55)50 (20-664)39 (19-77)
ALP (IU/L)111 (72-252)112 (54-356)116 (36-364)111 (62-169)
GGT (IU/L),#314 (38-836)43 (18-754)87 (18-230)82 (14-580)
C-reactive protein (mg/dL)**<5 (<5 to 15.3)<5 (<5 to 9.9)18.7 (<5 to 95.8)<5
IgA (g/L),††3.91 (1.52-4.98)5.19 (1.8 -10.04)4.81 (1.67-12.1)2.08 (1.63-8.54)
IgG (g/L)‡‡14.41 (8.47-17.91)12.65 (9.62-20.47)13.9 (6.12-26.42)16.37 (12.59-20.98)
IgM (g/L)§§1.61 (0.8-1.88)1.72 (0.34-7.06)1.57 (0.48-4.27)0.71 (0.03-2.31)
ANA1/20, 1/160, and negative in others tested1/640 and negative in others tested1/80 and negative in others testedNegative in 7 tested
SMANegative in all testedNegative in all testedNegative in all testedNegative in 7 tested
EthnicityCaucasian12/13 Caucasian13/14 CaucasianCaucasian

Exclusion Criteria.

Patients who had received immunosuppressive therapy in the preceding 3 months were excluded. Patients with cirrhosis due to alcohol and cofactors (e.g., a virus, AIH, or hemochromatosis) were excluded.

Control Group.

Eight patients with confirmed cirrhosis due to other etiologies were recruited as pathological controls; three had primary sclerosing cholangitis (PSC), two had AIH, one had primary biliary cirrhosis (PBC), and two had nonalcoholic steatohepatitis (NASH; Table 1). Twenty-one healthy abstinent volunteers served as healthy controls.

Peptides

Twenty-five 20-mer peptides (Mimotopes, Victoria, Australia) overlapping by five amino acids and spanning the full length of the ADH beta 1 subunit (Fig. 1) were synthesized.

Figure 1.

Proliferative responses to 25 20-mer ADH peptides in ARC patients and pathological controls. The peptide numbers and their amino acid positions are shown in columns 1 and 2. The patients' numbers are shown in the top row. Proliferative responses were considered positive when a mean SI ≥ 2.5 was observed. Group A (P1-P12) consisted of nine active alcohol drinkers and three patients who were abstinent for <6 months. Group B (P13-P25) consisted of patients who were abstinent for >6 months. Group C (P26-P34) consisted of patients who were abstinent for >6 months and underwent liver transplantation because of end-stage ARC. Group D consisted of eight patients with non–alcohol-related cirrhosis. Proliferative responses to individual ADH peptides are expressed as SI values. The SI values are presented as follows: black dots for ≥2.5 to ≤3.5, vertical lines for >3.5 to ≤4.5, diagonal lines for >4.5 to ≤5.5, and white dots for >5.5. The numbers of patients reacting to peptides are indicated in the last columns; the numbers and percentages of peptides recognized by each patient's PBMCs, the positivity for anti-ADH antibodies, the mean levels of cytokines IFN-γ, IL-17, and IL-4, the immunological profiles, and the histological findings are presented in the bottom 10 rows. For ANAs, the reciprocals of titers are shown. Abbreviations: mi, mild inflammatory activity; mo, moderate inflammatory activity; nd, not done; neg, negative; pos, positive.

Antibodies

Allophycocyanin (APC) chlorophyll protein [APC–cyanine 7 (Cy7)]–conjugated anti-CD4 antibodies and phycoerythrin (PE)-Cy7 anti-CD3 antibodies were used for surface phenotype characterization (BD Bioscience, Oxford, United Kingdom). PE-conjugated anti–IL-17, anti–IL-23, and fluorescein isothiocyanate (FITC)–conjugated antibodies to human IFN-γ, IL-4, and IL-10 (eBioscience, San Diego, CA) were used for intracellular cytokine staining (ICCS). Monoclonal antibodies to human IFN-γ, IL-4, and IL-17 for enzyme-linked immunosorbent assay (ELISA) testing were purchased from R&D (Abingdon, United Kingdom).

Cell Preparation and Proliferation Assay

Fresh or cryopreserved (viability > 90%) peripheral blood mononuclear cells (PBMCs) were isolated with a Ficoll-Hypaque density centrifugation gradient. HMCs were isolated from the second-passage perfusate of explanted livers through the portal vein. HMCs were harvested with a Ficoll-Hypaque density gradient.

PBMCs or HMCs (1 × 105/well) were cultured with an individual peptide (10 μM) in 200 μL of AIM-V medium (Invitrogen, Paisley, United Kingdom) supplemented with antibiotics and 2.5% human AB serum. Phytohemagglutinin (10 μg/mL) was used in cultures as a positive control to assess the responsiveness of patients' PBMCs or HMCs to a T-cell mitogen. The optimal ADH peptide concentration was determined in preliminary experiments with PBMCs from two ARC patients (active drinkers) cultured with 1, 10, or 50 μM ADH peptide. Proliferative responses were maximal at an individual ADH peptide concentration of 10 μM, and this concentration was used in this study.

After a 7-day culture at 37°C in a humidified 5% CO2 atmosphere, the cell cultures were pulsed with 0.25 μCi/well 3H-thymidine for 18 hours, and they were harvested onto glass fiber–lined membranes. Thymidine incorporation was measured with a β-counter (Canberra Parkard, Ltd., Pangbourne, United Kingdom) and expressed as the stimulatory index (SI), which is the ratio of the mean counts per minute from duplicate or triplicate (when the cell number was sufficient) determinations in the presence of antigenic peptides to the mean counts per minute obtained from four wells in the absence of peptides. Proliferative responses were considered positive with an SI ≥ 2.5.19

ELISA Cytokine Assays

Plasma (100 μL/well) or culture supernatants (50 μL/well) were used for the measurement of IFN-γ, IL-4, and IL-17. Supernatants were harvested on day 7 before the addition of 3H-thymidine. The concentrations of cytokines were measured by ELISA with a standard protocol.16, 20 The sensitivity of the cytokine assays was 5 to 10 pg/mL for all cytokines measured.

ICCS

The percentage of cytokine-producing T cells was determined by ICCS as previously described.21 Cells were stimulated with phorbol 12-myristate 13-acetate (PMA)/ionomycin (a leukocyte activation cocktail with GolgiPlug, BD Bioscience) for 5 hours and stained with APC-Cy7–conjugated anti-CD4 antibodies, PE-Cy7–conjugated anti-CD3 antibodies, PE-conjugated anti–IL-17 antibodies, anti–IL-23 antibodies, and FITC-conjugated antibodies to IFN-γ and IL-4. Flow cytometry was performed via the acquisition of a minimum of 1 × 104 gated events per sample (stained PBMCs and HMCs) in a BD FACSort Canto II (BD Immunocytochemistry Systems, San José, CA) with FACSDiva software (version 6.1.2).

Detection of Anti-ADH Antibodies by ELISA

Antibodies to horse liver ADH (17233, Sigma Aldrich, Poole, United Kingdom) sharing 87% homology with human ADH were detected with an in-house ELISA. In brief, each well of a 96-well, flat-bottom microplate was coated overnight with 3 ng/well ADH and incubated for 2 hours with patients' sera diluted 1/1000 in phosphate-buffered saline/0.1% bovine serum albumin. Anti-ADH antibodies were revealed by the addition of horseradish peroxidase–conjugated anti-human immunoglobulin A (IgA), IgG, and IgM for 1 hour, and this was followed by the addition of the chromogen o-phenylenediamine. The specificity of the assay was evaluated by inhibition experiments with sequentially diluted ADH and by comparison with immunoblot results, as described previously.15

Statistical Analysis

The chi-square test and the one-tailed Fisher's exact test were used to compare frequencies of peptides recognized within the different groups. Correlations between indices were determined with Pearson's correlation coefficient for log-transformed normalized data. A P value < 0.05 was considered significant.

Results

Proliferative Response to ADH Peptides in Peripheral Blood

Patients in group A, who included nine active drinkers and three short-term abstinents (<6 months), showed strong and broad cellular immune responses to ADH. Eight of the 12 patients (responders) recognized a median of 7 peptides (range = 2-12) with a median SI of 3.4 (range = 2.5-13.6). Each of peptides 3, 4, 5, 6, 8, 12, 13, 19, and 23 was recognized by 25% of the patients, and peptide 15 was recognized by 58%. Peptide 23, which was previously identified as an epitope for B cells (unpublished data), was recognized by three patients with SIs of 3, 4.8, and 5.3, respectively (Fig. 1).

Group B, which comprised 13 patients who had been abstinent for 9 months to 20 years (median = 4.5 years), showed weaker immune responses to ADH than group A. Seven of the 13 abstinents recognized a median of 1 peptide (range = 1-5, P = 0.017) with a median SI of 2.8 (range = 2.5-5.3, P = 0.013 versus group A).

In the ARC cohort as a whole, there was no correlation between the severity of liver disease and the number of peptides recognized or the mean SI.

Three of the eight patients in the pathological control group recognized a median of 1 peptide (range = 1-3) with a median SI of 2.9 (range = 2.5-3.46, P = 0.005 in comparison with group A). One of the two patients with NASH responded to peptide 8, which was also recognized by ARC patients. One of the two AIH patients responded to peptides 7, 12, and 21. Two of the three peptides, 7 and 21, were rarely recognized by ARC patients.

Six of the 10 healthy controls recognized a median of 2 peptides (range = 1-2) with a median SI of 2.9 (range = 2.5-3.6, P = 0.006 in comparison with group A).

HMC Proliferative Responses in ARC Patients

Proliferative responses to ADH in HMCs from nine patients who required liver transplantation for end-stage cirrhosis were assessed. HMCs from three patients recognized 1 to 2 peptides; the remaining six patients showed no proliferation in response to any of the ADH peptides (Fig. 1, group C).

Cytokine Production With Non-Antigen-Specific Stimuli

The frequency of IL-4–, IL-6–, IL-10–, IL-17–, IL-23–, and IFN-γ–producing cells after 5 hours of stimulation with PMA/ionomycin was assessed in the PBMCs of seven ARC patients (four active drinkers and three long-term abstinents) and nine healthy controls and in the HMCs of five patients with ARC. The frequencies are summarized in Table 2. Within the CD3 T-cell peripheral population, cells producing IFN-γ, IL-4, IL-17, and IL-23 were more frequent in patients versus healthy controls (P = 0.013-0.06). All cytokine-producing cells (except those producing IL-23) were less numerous in patients' HMCs versus PBMCs (P = 0.005-0.06). IL-4–, IL-6–, and IL-17–producing cells were also less numerous within the HMCs of patients versus PBMCs of healthy controls (P = 0.01-0.03). The profiles of cytokine-producing cells in PBMCs and HMCs from two representative ARC patients are shown in Fig. 2.

Table 2. Frequency of Cytokine-Producing Cells After PMA/Ionomycin Stimulation in PBMCs From Patients With ARC and Healthy Controls and in HMCs From ARC Patients
Cytokine (%)Subjects
PBMCs From Patients (n = 7)HMCs From Patients (n = 5)PBMCs From Healthy Subjects (n = 9)
  • The data are presented as means and standard errors of the mean.

  • *

    P = 0.017 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • P = 0.036 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • P = 0.06 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • §

    P = 0.015 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • P = 0.017 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • P = 0.01 (comparing the frequencies of cytokine-producing cells within PBMCs from healthy subjects and within HMCs from patients).

  • #

    P = 0.013 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • **

    P = 0.06 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • ††

    P = 0.048 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • ‡‡

    P = 0.037 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • §§

    P = 0.01 (comparing the frequencies of cytokine-producing cells within PBMCs from healthy subjects and within HMCs from patients).

  • ∥∥

    P = 0.02 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • ¶¶

    P = 0.03 (comparing the frequencies of cytokine-producing cells within PBMCs from healthy subjects and within HMCs from patients).

  • ##

    P = 0.018 (comparing the frequencies of cytokine-producing cells within PBMCs from patients and healthy subjects).

  • ***

    P = 0.01 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

  • †††

    P = 0.005 (comparing the frequencies of cytokine-producing cells within PBMCs and HMCs from patients).

CD3pos IL-42.44 ± 0.68*1.02 ± 0.320.56 ± 0.19
CD4pos IL-42.76 ± 0.811.45 ± 0.461.88 ± 0.34
CD3pos IL-60.73 ± 0.09§0.21 ± 0.050.19 ± 0.05
CD4pos IL-61.19 ± 0.60.23 ± 0.041.65 ± 0.45
CD3pos IL-10Not doneNot done0.13 ± 0.04
CD4pos IL-100.48 ± 0.11#0.18 ± 0.140.12 ± 0.10
CD3pos IL-171.61 ± 0.79**0.09 ± 0.04††0.19 ± 0.06
CD4pos IL-171.72 ± 0.720.24 ± 0.08‡‡1.21 ± 0.21§§
CD3pos IL-231.23 ± 0.4∥∥2.40 ± 0.850.19 ± 0.04¶¶
CD4pos IL-230.71 ± 0.21Not done1.34 ± 0.35
CD3pos IFN-γ14.47 ± 3.47##4.85 ± 1.60***5.95 ± 1.30
CD4pos IFN-γ7.48 ± 1.063.03 ± 1.04†††5.35 ± 1.76
Figure 2.

Cytokine-producing T cells for two representative patients with ARC. ICCS of IL-4, IL-17, IL-23, and IFN-γ with PBMCs from an active drinker with ARC and with HMCs from a patient with end-stage ARC was performed. Quadrants were set up with the appropriate isotype controls. Cytokine-producing cells are shown in the upper right quadrants. In both PBMCs and HMCs, IFN-γ–producing cells predominated. There were more IL-4– and IL-17–producing CD4 T cells in PBMCs versus HMCs.

Cytokine Production With an Antigen-Specific Stimulus

PBMCs.

Cytokines produced by PBMCs cultured in the presence of ADH peptide was investigated in 18 patients (nine in group A and nine in group B) and in three healthy subjects. Cytokines produced by HMCs from nine patients with ARC were also tested.

IFN-γ.

IFN-γ was quantitatively the predominant cytokine released in the presence of 19 peptides that included all peptides but one (peptide 6) inducing a proliferative response (Fig. 3A,B). In the undivided ARC patient group, the levels of IFN-γ correlated with the serum bilirubin level (r = 0.53, P < 0.025) but not with other liver biochemistry indices. After the patients were divided into groups A and B, the same correlations persisted in group A, with the levels of IFN-γ correlating with serum bilirubin levels (r = 0.68, P < 0.025) and also with liver disease severity as defined by the Child-Pugh score (r = 0.61, P < 0.05).

Figure 3.

Proliferative responses and mean levels of cytokines in the culture supernatant of PBMCs from 18 patients with ARC after 7 days of stimulation with individual ADH peptides. Levels of IFN-γ, IL-4, and IL-17 were measured by ELISA. The y axes represent the mean SI values in panel A and the levels of cytokines in panels B to D; the x axes represent the 25 ADH peptides tested. The dotted lines indicate the baseline cytokine production (mean values) in the culture supernatant in the absence of a peptide: 167 pg/mL for IFN-γ, 158 pg/mL for IL-4, and 32 pg/mL for IL-17. In the presence of ADH peptides, PBMCs from patients with ARC produced all three types of cytokines. Open bars indicate peptides that induced the production of cytokines only; solid bars indicate peptides that also induced proliferative responses.

There was no difference in the mean IFN-γ levels in culture supernatants from drinking and nondrinking patients (406.7 ± 46.9 versus 351.48 ± 28.3 pg/mL, P = not significant; Fig. 4A).

Figure 4.

Mean levels of cytokines in the culture supernatant of PBMCs from patients [nine in group A (solid bars) and nine in group B (open bars)] with ARC after 7 days of stimulation with individual ADH peptides. Levels of IFN-γ, IL-4, and IL-17 were measured by ELISA. The y axes represent the levels of cytokines, and the x axes represent the 25 ADH peptides tested. (A) In the presence of ADH peptides, PBMCs from active drinkers (group A) produced IFN-γ when they were exposed to 19 peptides; PBMCs from patients who were abstinent for >6 months (group B) produced IFN-γ against 16 peptides, with the levels of IFN-γ being similar in the two groups. (B) PBMCs from patients in group A produced IL-17 against more ADH peptides than PBMCs from patients in group B (12 versus 8), with the IL-17 levels being significantly higher in group A versus group B. (C) Patients in group A responded to fewer ADH peptides than patients in group B (5 versus 11), with the levels of IL-4 being significantly lower in group A versus group B.

IL-17.

In the undivided ARC patients, 10 ADH peptides induced the secretion of IL-17 (Fig. 3C). Five (peptides 3, 8, 12, 15, and 19) corresponded to those inducing proliferation, and all peptides inducing the secretion of IL-17 also induced the secretion of IFN-γ. The levels of IL-17 in group A patients were significantly higher than the levels in group B patients (128.9 ± 17.6 versus 57.8 ± 6.1 pg/mL, P < 0.03; Fig. 4B). There was no correlation between the level of IL-17 and the Child-Pugh or MELD scores.

IL-4.

In all ARC patients, 10 peptides induced IL-4 secretion (Fig. 3D), with 4 corresponding to those inducing proliferation and 5 corresponding to IFN-γ– and IL-17-inducing peptides. The IL-4 levels correlated with aspartate aminotransferase (AST; r = 0.49, P < 0.025) among the undivided patients, but no correlation with other indices of liver damage was seen. The levels of IL-4 in group B patients were higher than those in group A patients (219.7 ± 21.6 versus 101.5 ± 17.4 pg/mL, P < 0.01; Fig. 4C).

After ADH peptide stimulation, the mean levels for all cytokines in the PBMCs of three healthy controls were below the baselines.

HMCs.

HMCs from nine patients produced all three cytokines after ADH peptide stimulation (Fig. 5A-C). Sixteen peptides induced IFN-γ, and all but 3 (peptides 11, 16, and 25) corresponded with those stimulating IFN-γ production in PMBCs; 10 peptides induced IL-4 production, and 6 of them corresponded to those stimulating IL-4 production in PBMCs. Seven peptides induced the production of IL-17, and five of them corresponded to those stimulating IL-17 production in PBMCs.

Figure 5.

Mean levels of cytokines in the culture supernatant of HMCs from nine patients with end-stage ARC after 7 days of stimulation with ADH peptides. The y axes represent the mean cytokine levels, and the x axes represent the 25 ADH peptides tested. Sixteen peptides induced the production of IFN-γ, with all but three overlapping with those inducing IFN-γ production in PBMCs. Seven peptides induced IL-17 production, with five overlapping with those inducing IL-17 production in PBMCs (peptides 3, 7, 8, 13, and 19). Ten peptides induced IL-4, with 6 overlapping with those inducing IL-4 production in PBMCs (peptides 2, 7, 8, 9, 13, and 19).

The mean level of IFN-γ was lower in HMCs versus PBMCs in both groups A and B [271.6 ± 25.9 versus 406.7 ± 46.9 (P < 0.01) and 351.48 ± 28.3 pg/mL (P < 0.05)]. The mean level of IL-4 in HMCs was higher than that in the PBMC culture supernatant for group A (214.9 ± 19.8 versus 101.5 ± 17.4 pg/mL, P = 0.003) but similar to that for group B (219.7 ± 21.6 pg/mL, P = not significant). The mean level of IL-17 in HMCs was significantly lower than that for group A (33.5 ± 5.7 versus 128.9 ± 17.6 pg/mL, P < 0.001) and tended to be lower than that for group B (57.8 ± 6.1 pg/mL, P = 0.06). The levels of IL-17 correlated with three biochemical indices of liver damage—serum bilirubin levels (r = 0.61, P < 0.05), alkaline phosphatase (ALP) levels (r = 0.60, P < 0.05), and AST levels (r = 0.54, P < 0.1)—and with the MELD score (r = 0.73, P < 0.025). The levels of IL-4 inversely correlated with the serum bilirubin levels (r = −0.61, P < 0.05), AST levels (r = −0.72, P < 0.025), Child-Pugh score (r = −0.77, P < 0.01), and MELD score (r = −0.71, P < 0.025).

Plasma.

IFN-γ and IL-4 were detectable in patients' plasma. The levels of IFN-γ in group A (median = 15 pg/mL, range = 0-667 pg/mL) tended to be higher than the levels in group B (median = 0 pg/mL, range = 0-245 pg/mL, P = 0.12). There was no correlation between IFN-γ levels in plasma and in culture supernatants of PBMCs exposed to ADH peptides. IL-4 was detected in only one patient in group A (89 pg/mL). IL-17 was detected in none of the plasma tested.

Correlation Between Humoral and Cellular Immune Responses to ADH

Anti-horse ADH antibodies were detected in 8 of 10 patients (80%) in group A and in 6 of 11 patients (55%) in group B (P = 0.2). Six of the eight anti-ADH–positive patients (75%) in group A and four of the six patients in group B (67%, P = not significant) had cellular immune responses to ADH. The prevalence of anti-ADH tended to be higher in patients with elevated AST levels (7/8 or 88%) versus those with normal AST levels (7/13 or 54%, P = 0.1). The titers of anti-ADH antibodies (optical density [OD]) correlated with ALP levels (R = 0.42, P < 0.05). No correlation was observed between anti-ADH antibody titers and the measurement of cellular immune responses to ADH.

Discussion

This study shows that actively drinking patients with ARC mount cellular immune responses against the ethanol-oxidizing enzyme ADH, and these responses correlate with disease severity.

T-cell proliferative responses target the broad ADH31-95 region and five additional discontinuous sequences on ADH. The breadth and intensity of these proliferative responses are associated with active alcohol consumption in ARC patients. Conversely, the T cells from patients on sustained abstinence [confirmed by low gamma-glutamyl transferase (GGT) levels] do not proliferate in response to ADH. In ARC patients, alcohol intake favors immune recognition of ADH; although the mechanism for this can only be conjectured, excessive production of acetaldehyde, the oxidative metabolite of ADH and an efficient promoter of adduct formation, may account for ADH neo-antigenicity.

In actively drinking ARC patients, 19 ADH sequences were able to stimulate IFN-γ production, the key cytokine of Th1 effector cells. This soluble mediator is able to inflict liver cell damage.22 The peptides able to stimulate IFN-γ production far outnumber those inducing proliferation and include ADH301-320, a sequence involved in substrate binding and subunit interaction: its targeting may interfere with ADH enzymatic function.

Testing anti-ADH responses on mononuclear cells from the liver gave results of interest: none of the ADH sequences was able to induce proliferative responses in HMCs, whereas 16 peptides that induced IFN-γ production in peripheral cells were also able to do so in mononuclear cells from the liver. This indicates first that the ADH-specific Th1 immune responses observed in the peripheral blood are mirrored in the liver, and second, these IFN-γ–restricted immune responses are likely to have a damaging effect because the production of IFN-γ directly results in liver damage.22 This possibility is supported by our observation that the severity of liver disease correlates with the levels of IFN-γ in culture. The fact that the HMC population secretes IFN-γ but proliferates weakly when it is exposed to ADH peptides suggests that it is mainly composed of memory T cells, which are poor proliferators but good IFN-γ producers when they are antigenically stimulated.23 Memory T cells may also explain the ability of long-term abstinents to mount IFN-γ anti-ADH immune responses. The data thus far presented suggest that the vigorous IFN-γ T-cell immune responses seen in the periphery and livers of ARC patients are pathogenic, and this suggests the IFN-γ–inducing ADH sequences as potential targets for therapeutic interventions. ADH peptides with inhibitory properties may be designed and used to restrain damaging anti-ADH immune responses.24 Alternatively, the inhibition of IFN-γ production by cytokine modulation could be considered. Kawaratani et al.25 recently showed in a rat model of ALD that Y-40138, a cytokine modulator, effectively suppressed the production of IFN-γ and enhanced IL-10; this resulted in decreased inflammation, fibrosis, and oxidative stress in the liver.

A large Th1/Th2 imbalance is present in active ARC drinkers, whose levels of IL-4 are remarkably low; this finding echoes an observation in a murine model of ALD in which ethanol feeding inhibited the production of the key Th2 cytokine IL-4. Notably, the cytokine profile of the immune responses in long-term abstinents, including those undergoing liver transplantation, differed from that of active drinkers. Upon exposure to ADH peptides, a strong IL-4 immune response was detectable in the long-term abstinents with ARC in association with an IFN-γ response; this IL-4 response was much lower in active drinkers. Although a switch from a proinflammatory Th1 cytokine response to a sustained Th2 profile has been reported to facilitate hepatic fibrosis26, 27 and our abstinent patients relevantly were all at a late stage of their alcoholic liver disease (4 were on the transplantation list waiting list, and 15 had undergone transplantation), we observed an inverse correlation between IL-4 levels secreted by HMCs and the severity of liver damage. This indicates that IL-4 does have a key role in counterbalancing the damaging Th1 immune response through the inhibition of IFN-γ–producing T cells and autoantigen-specific cytotoxic CD8 T cells at the site of inflammation.28 However, levels of IL-4 in PBMCs also correlated with AST in contrast to the correlation seen in HMCs, and this indicates that the role of IL-4 in the pathogenesis of ALD is complex.

Recently, Lemmers et al.13 reported that the IL-17 pathway is involved in the pathogenesis of acute alcoholic hepatitis, a condition characterized by a high frequency of IL-17 receptor–positive hepatic stellate cells, which are deemed responsible for the intrahepatic recruitment of neutrophils. Our results showed that Th17-positive T cells and an anti-ADH Th17 immune response were detectable in PBMCs and HMCs of patients with ARC (both active drinkers and, to a lesser extent, abstinents). Although Th17 immune responses were much less vigorous than those exerted by Th1 cells, importantly, a close association between Th17 and liver damage was observed in the liver. The Th17 pathway thus appears to have a prominent role in ARC and clearly warrants further investigation within the ALD spectrum. This may allow us to define the function of T-cell subsets that can suppress Th1 and Th17 immune responses, such as CD4posCD25posFOXP3pos regulatory T cells, which are scarce in the livers of ARC patients (data not shown).

We found no direct association between humoral and cellular anti-ADH immune responses, probably because of the different forms of antigens used in the investigation of the two arms of the immune system: a set of overlapping peptides reproducing the human protein for testing cellular immunity and a full-length equine protein for the assessment of humoral immunity because of the unavailability of the human enzyme. It should be noted that the presence of cellular immune responses to ADH is unrelated to positivity for anti-ADH antibodies and is common in both antibody-positive and antibody-negative patients. It is not surprising that anti-ADH cellular immunity is frequently detected in ARC patients. This could be the result of ADH release from the cytoplasm of damaged hepatocytes acting as a stimulus to trigger autoimmune responses. Although the hepatocyte damage is likely to be multifactorial, autoimmunity almost certainly enhances hepatocyte damage to form a vicious circle.

Anti-ADH immune responses warrant future studies in acute alcoholic hepatitis, a condition characterized by florid inflammation leading to severe hepatic injury with a mortality rate of 30% at 28 days.29 A possible immune pathogenesis for this condition is suggested by an improvement in mortality with prednisolone.30, 31 Our findings showing that ADH-specific Th1 and Th17 responses are closely correlated with the Child-Pugh and MELD scores and indices of liver damage in ARC indicate the potential value of testing effector anti-ADH responses (both Th1 and Th17) in acute alcoholic hepatitis also. This may also assist in the selection of patients suitable for glucocorticoid treatment.

In conclusion, we have found that Th1 cellular immunity to ADH may represent a link between alcohol consumption and liver damage in patients with ARC, and Th17 immunity persists in abstinents. These immune responses may also provide a biomarker for identifying patients drinking in excess, who are at high risk for progressing to cirrhosis. Our data also provide a starting point for the development of more specific and effective therapeutic interventions for the treatment of this complex condition. The challenge will now be to validate this biomarker in patients who drink alcohol in excess but have yet to develop clinical evidence of ALD so that we can investigate whether it can differentiate those who will develop cirrhosis from those who will not.

Acknowledgements

The authors thank the staff at the Biomedical Research Centre (King's College London) for providing help with flow cytometry.

Ancillary