A multifaceted imbalance of T cells with regulatory function characterizes type 1 autoimmune hepatitis


  • Potential conflict of interest: Nothing to report.


Immunotolerance is maintained by regulatory T cells (Tregs), including CD4+CD25hi, CD8+CD28, γδ, and CD3+CD56+ [natural killer T (NKT)] cells. CD4+CD25hi cells are impaired in children with autoimmune hepatitis (AIH). Little is known about Tregs in adults with AIH. The aim of this study was to investigate the frequency and function of Treg subsets in adult patients with AIH during periods of active disease and remission. Forty-seven AIH patients (16 with active disease and 31 in remission) and 28 healthy controls were studied. Flow cytometry was used to evaluate surface markers and function-related intracellular molecules in γδ, CD8+CD28, NKT, and CD4+CD25hi cells. CD4+CD25hi T cell function was determined by the ability to suppress proliferation and interferon gamma (IFN-γ) production by CD4+CD25 target cells. Liver forkhead box P3–positive (FOXP3+) cells were sought by immunohistochemistry. In AIH patients, particularly during active disease, CD4+CD25hi T cells were fewer, expressed lower levels of FOXP3, and were less effective at inhibiting target cell proliferation versus healthy controls. Moreover, although the numbers of CD8+CD28 T cells were similar in AIH patients and healthy controls, NKT cells were numerically reduced, especially during active disease, and produced lower quantities of the immunoregulatory cytokine interleukin-4 versus controls. In contrast, γδ T cells in AIH patients were more numerous versus healthy controls and had an inverted Vδ1/Vδ2 ratio and higher IFN-γ and granzyme B production; the latter was correlated to biochemical indices of liver damage. There were few FOXP3+ cells within the portal tract inflammatory infiltrate. Conclusion: Our data show that the defect in immunoregulation in adult AIH is complex, and γδ T cells are likely to be effectors of liver damage. (HEPATOLOGY 2010)

Autoimmune hepatitis (AIH) is an immune-mediated liver disease characterized by high levels of aminotransferases and gamma-globulins, circulating autoantibodies, and histological evidence of interface hepatitis.1-3 Two AIH subsets are conventionally recognized according to their autoantibody profile4: type 1 AIH (AIH-1), which is characterized by positivity for anti-nuclear antibody (ANA) and/or anti–smooth muscle antibody (anti-SMA),5 and type 2 AIH, the serological hallmarks of which are anti–liver/kidney microsomal antibody type 1 and anti–liver cytosol antibody type 1.6, 7

The mechanisms underlying the breakdown of self-tolerance and leading to the development of AIH are not fully understood, although impaired immunoregulation appears to play a crucial role. Several T cell populations with regulatory properties are required to maintain immune system homeostasis; among them, the best characterized are the naturally occurring CD4+CD25+T cells, which express high levels of CD25 (CD25hi), CD45RO, and CD62L, and the function-related forkhead/winged helix transcription factor forkhead box P3 (Foxp3).8-10 CD4+CD25hi T cells suppress the proliferative and cytokine responses of effector CD4 and CD8 T cells and down-regulate the functions of macrophages, dendritic cells, natural killer cells, and B lymphocytes. Documented mechanisms of suppression include the secretion of immunosuppressive cytokines and cell-to-cell contact with antigen-presenting cells or effector T cells.11, 12 Cytotoxic T lymphocyte–associated antigen 4 (CTLA-4) has been deemed to play an important role in regulating CD4+CD25hi T cell function.13 Previous studies have shown that CD4+CD25hi T cells are numerically impaired in childhood AIH, in which they are also unable to control effector functions of CD4 and CD8 target cells.14-16

In addition to CD4+CD25hi regulatory T cell (Tregs), other T cell subsets have emerged as critical players in the maintenance of immunotolerance. After expansion in vitro, CD8-positive T lymphocytes that do not express the costimulatory molecule CD28 on their surface (CD8+CD28) are able to exert inhibitory effects and complement the CD4+CD25hi T cell function17; both act by inducing a tolerogenic phenotype on antigen-presenting cells through the inhibition of costimulatory molecules18 and by secreting soluble factors regulating activated T cell proliferation and cytotoxicity.19 Evidence that this Treg subset is involved in the induction and maintenance of immunotolerance is derived from the observation of an indirect correlation between the number of circulating CD8+ suppressor cells and the frequency of organ transplant acute rejection episodes20 and from their functional impairment in patients with active autoimmune diseases such as systemic lupus erythematosus and progressive systemic sclerosis.21

Natural killer T (NKT) cells are characterized by the surface expression of both T (CD3+) and natural killer lineage markers (CD56+) and typically express a restricted T cell receptor (TCR) repertoire; they recognize glycolipid antigens in association with the major histocompatibility complex class I–like molecule CD1 and secrete high levels of regulatory cytokines, including interferon gamma (IFN-γ) and interleukin-4 (IL-4), within minutes to hours after antigen encounter.22 NKT cells increase the proliferation and enhance the surface expression of CTLA-4 on CD4+CD25hi Tregs via IL-2 production and possibly play a central role in the composite immunoregulatory network.23 Moreover, circulating NKT cells are reduced in a variety of immune-mediated conditions, such as type 1 diabetes and rheumatic and inflammatory bowel diseases.24 Moreover, NKT cells represent a large fraction of liver-resident lymphocytes.25

Unconventional T cells, rearranging the γδTCR and being double-negative for surface CD4 and CD8, though constituting a small proportion of circulating lymphocytes (1%-10%), are abundant in the liver and are involved in antitumor surveillance and immunoregulation.26 They recognize small, pathogen-derived molecules such as organophosphates and autologous proteins up-regulated by infected, transformed, or otherwise malfunctioning host cells.27 In man, two main γδ T cell subsets have been described according to the rearranged Vδ chain: Vδ1+, which is abundant among intraepithelial lymphocytes but is scarcely represented in the peripheral blood, possesses both regulatory and effector properties, and Vδ2+, which constitutes up to 80% of the whole circulating γδ T cell population, is involved in the defense against pathogens and tumors.26 γδ intraepithelial lymphocytes are directly responsible for the cytolysis of effector and antigen-presenting cells via granzyme-perforin, Fas–Fas ligand, and lymphotoxin pathways and represent a crucial population for the regulation of the immune response in the tissues.27 Although a generalized increase in the peripheral γδ T cell population characterizes patients with autoimmune disorders, including multiple sclerosis,28 Behcet's disease,29 and childhood autoimmune liver diseases,30 selective enrichment in their Vδ1 subset has been described in Takayasu arteritis31 and systemic sclerosis32; this suggests an effector involvement of γδ T cells in the pathogenesis of autoimmunity.

The aim of the present study was to explore numerical and functional characteristics of different Treg subsets in the circulation of adult patients with AIH-1 during active and quiescent stages of disease.


α-GalCer, α-galactosylceramide; [A] patients, patients with active disease; AIH, autoimmune hepatitis; AIH-1, type 1 autoimmune hepatitis; ALT, alanine aminotransferase; ANA, anti-nuclear antibody; CTLA-4, cytotoxic T lymphocyte–associated antigen 4; CY, cychrome; FITC, fluorescein isothiocyanate; FOXP3, forkhead box P3; GGT, gamma-glutamyl transpeptidase; HC, healthy control; IFN, interferon; IgG, immunoglobulin G; IL, interleukin; INR, international normalized ratio; MFI, mean fluorescence intensity; NKT, natural killer T; NS, not significant; PBMC, peripheral blood mononuclear cell; PBS, phosphate-buffered saline; PE, phycoerythrin; PerCP, peridinin chlorophyll protein; PMA, phorbol 12-myristate 13-acetate; [R] patients, patients with disease in remission; RPMI-1640, Roswell Park Memorial Institute 1640; SMA, smooth muscle antibody; TCR, T cell receptor; Treg, regulatory T cell; UNL, upper normal level.

Patients and Methods


Forty-seven consecutive patients with AIH-1 [median age = 48 years (range = 17-79 years), 79% female] were enrolled between April 2007 and April 2009; there were 16 patients with active disease ([A] patients) and 31 patients in drug-induced biochemical remission ([R] patients). Twenty-eight healthy subjects served as healthy controls [HCs; median age = 39 years (range = 23-58 years), 68% female]. Viral, metabolic, and genetic causes of liver disease were excluded by appropriate investigations, all patients being negative for anti–hepatitis C virus, hepatitis B surface antigen, Epstein-Barr virus, and cytomegalovirus serological markers of active infection.

From 9 of 16 [A] patients, blood was obtained before treatment; the remaining 7 [A] patients were studied at relapse during immunosuppression tapering (3 were on 4 mg of methylprednisolone daily and 4 were on 2-4 mg of methylprednisolone and 50 mg of azathioprine daily).

The 31 [R] patients had normal alanine aminotransferase (ALT) and gamma-globulin levels for a median of 40 months (range = 8-120 months); 21 were on 2 to 4 mg of methylprednisolone daily, 3 were on 50 mg of azathioprine daily, and 7 were on 2 to 4 mg of methylprednisolone and 50 to 100 mg of azathioprine daily. The median duration of immunosuppression was 42 months (range = 12-237 months).

Liver biopsy showed histological features of interface hepatitis in all 38 patients at diagnosis.

In the group of [A] patients, ANAs were present in 12 patients, SMAs were present in 6, soluble liver antigen was present in 1, and anti-mitochondrial antibodies were present in 1 (ANAs and SMAs co-occurred in 3 individuals). At diagnosis, all 31 [R] patients tested positive for autoantibodies (ANAs and SMAs were detected in 19 and 21 subjects, respectively, and co-occurred in 15), whereas at the time of study, 7 (28%) had lost all autoreactivity (ANAs and/or SMAs were still present in 22 subjects and co-occurred in 4). At diagnosis, all patients fulfilled the diagnostic criteria of the International Autoimmune Hepatitis Group.5

Clinical and laboratory features are summarized in Table 1. All 16 [A] patients had high aminotransferase, gamma-glutamyl transpeptidase (GGT), and bilirubin levels, increased international normalized ratios (INRs), high gamma-globulin and immunoglobulin G (IgG) levels, and seropositivity for autoantibodies. All [R] patients had normal biochemical tests, but autoantibodies were still detectable in half.

Table 1. Clinical and Laboratory Features of AIH Patients at the Time of the Study
 AIH [A] Patients (n = 16)AIH [R] Patients (n = 31)P
  • Unless otherwise indicated, data are expressed as means and standard deviations.

  • Abbreviation: NS, not significant; UNL, upper normal level.

  • *

    Mann-Whitney test.

  • Normal total bilirubin range = 0.2-1.1 mg/dL.

  • Normal INR range = 0.8-1.2.

  • §

    Normal gamma-globulin range = 0.6-1.6 g/L.

  • Normal IgG range = 700-1900 mg/dL.

Age (years), median (range)44 (17-69)48 (17-79)NS
Female sex, %6984NS
ALT × UNL (U/L)11.39 ± 8.490.60 ± 0.27< 0.001*
GGT × UNL (U/L)2.56 ± 1.510.59 ± 0.510.009*
Total bilirubin (mg/dL), median (range)1.8 (0.6-4.0)0.6 (0.3-0.9)0.04*
INR1.38 ± 0.441.01 ± 0.02NS
Gamma-globulin (g/L)§2.77 ± 1.211.25 ± 0.550.006*
IgG (mg/dL)2638 ± 10091119 ± 1300.008*
ANA, n of positive patients (%)12 (75%)13 (42%)NS
Anti-SMA, n of positive patients (%)6 (38%)15 (48%)NS

Autoantibody Testing.

Sera were tested for non–organ-specific autoantibodies by indirect immunofluorescence on cryostatic sections of rat liver, kidney, and stomach specimens at an initial serum dilution of 1:40.33 Anti–soluble liver antigen was detected by enzyme-linked immunosorbent assay according to the manufacturer's instructions (Euroimmun, Lubeck, Germany).

Cell Separation and Purification.

Peripheral blood mononuclear cells (PBMCs) were prepared from 20 mL of peripheral blood with preservative-free heparin (10 U/mL), diluted 1:1 with Roswell Park Memorial Institute 1640 (RPMI-1640) medium (Invitrogen Life Technologies, Paisley, United Kingdom), and separated with Ficoll-Hypaque (Amersham Pharmacia Biotech, Ltd., Little Chalfont, United Kingdom). PBMCs were collected and washed twice with RPMI-1640. Viability, determined by trypan blue exclusion, always exceeded 98%. Cryopreserved PBMCs, stored in liquid nitrogen for 1 to 12 months before analysis, were also used. Preliminary experiments showed no significant differences in viability and surface marker staining between freshly prepared and cryopreserved cells.

CD4+CD25+ T cells were isolated from PBMCs by CD4-negative selection with antibodies to CD14, CD56, CD19, CD8, CD235a, and CD45RA and depletion beads (Dynal Invitrogen, Oslo, Norway) coated with fragment crystallizable–specific human immunoglobulin G4 antibody; this was followed by CD25-positive selection with immunomagnetic beads coated with anti-human CD25 antibodies (Dynal Invitrogen). Purified CD4+CD25+ T cells localized in the CD4+CD25hi cell gated area, as previously described.16

Flow Cytometry.

Three-color flow cytometry analysis was performed on fresh and frozen PBMCs. Unfractionated cells were stained with fluorescein isothiocyanate (FITC)–conjugated anti-CD4, anti-Vδ1, or anti-Vδ2 monoclonal antibodies, phycoerythrin (PE)-conjugated anti-CD25, anti-CD28, or anti-γδTCR monoclonal antibodies, or cychrome (CY)-conjugated anti-CD8, anti-CD3, or anti-CD56 monoclonal antibodies in the following combinations: FITC-CD4/PE-CD25, PE-CY7-CD4/FITC-CD25/PE-CD45RO, PE-CY7-CD4/FITC-CD25/PE-CD62L, FITC-CD8/PE-CD28, peridinin chlorophyll protein (PerCP)–CD3/PE-CD56, CY-CD3/PE-γδTCR, CY-CD3/FITC-Vδ1, and CY-CD3/FITC-Vδ2 [the monoclonal antibodies were obtained from BD Pharmingen (Oxford, United Kingdom), except for anti-Vδ1 and anti-Vδ2, which were obtained from Pierce (Rockford, IL)]. Cells were incubated at 4°C for 35 minutes, washed with phosphate-buffered saline (PBS)/1% fetal bovine serum, and stored at 4°C until the analysis. At least 50,000 cells were used per experiment.

Flow cytometry was performed on a Becton Dickinson fluorescence activated cell sorter (FACSCanto II, Becton Dickinson Immunocytochemistry Systems, San José, CA); CellQuest software and FACSDiva software were used for the analysis. On average, 20,000 lymphocyte-gated events were acquired.

Intracellular Staining.

Purified CD4+CD25hi T cells from 15 patients (7 [A] patients and 8 [R] patients) and 9 controls were stained with an FITC-conjugated anti-CD4 monoclonal antibody, permeabilized, fixed with Cytoperm/Cytofix, and stained with PE-conjugated anti-FOXP3 (eBioscience, Inc., San Diego, CA) or anti–CTLA-4 monoclonal antibodies (BD Pharmingen).

Unfractionated cells from 24 patients (12 [A] patients and 12 [R] patients) and 16 controls were exposed to phorbol 12-myristate 13-acetate (PMA; 10 ng/mL)/ionomycin (500 ng/mL) to stimulate the production of granzyme B and IFN-γ, and they were incubated for 5 hours at 37°C in 5% CO2; after washing, the following surface/intracellular staining combinations were used: PE-conjugated anti-γδTCR/FITC-conjugated anti–granzyme B monoclonal antibody (BD Pharmingen) and FITC-conjugated anti-Vδ1 or anti-Vδ2/PE-conjugated anti–IFN-γ monoclonal antibody (Pierce). Unfractionated cells from 13 patients (7 [A] patients and 6 [R] patients) and 6 controls were seeded at 2 × 105/well in a 96-well plate, cultured for 5 days at 37°C in 5% CO2 in RPMI-1640 [supplemented with 2 mM L-glutamine, 25 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, 100 U/mL penicillin, 0.1 mg/mL streptomycin, 2.5 μg/mL amphotericin B, 10% inactivated fetal bovine serum, and rIL-2 (Chiron) at a concentration of 30 U/mL], and used in two different settings. In the first setting, cells were exposed for the last 5 hours of culture to PMA (10 ng/mL)/ionomycin (500 ng/mL) to stimulate the production of IL-10.34 After washing, the following surface/intracellular staining combination was used: FITC-conjugated anti-CD8, PE-conjugated anti-CD28, and Alexa-Fluor 647-conjugated anti–IL-10 (e-Bioscience). In the second setting, α-galactosylceramide (α-GalCer) was added to cultures (2 μg/mL) on day 1 to stimulate NKT cell expansion. Cells were then exposed for the last 5 hours to PMA (10 ng/mL)/ionomycin (500 ng/mL) to stimulate the production of IL-4 and IFN-γ; after washing, the following surface/intracellular staining combination was used: PerCP-conjugated anti-CD3, PE-conjugated anti-CD56, PE-Cy7–conjugated IFN-γ (e-Bioscience), and FITC-conjugated anti–IL-4 (e-Bioscience).

After being stained, cells were washed once with PBS/1% fetal bovine serum, resuspended, and stored at 4°C until the analysis. At least 50,000 cells were analyzed in each experiment. The flow cytometry analysis was carried out as previously discussed.


Paraffin-embedded liver sections, available from seven patients, were stained with anti-FOXP3 monoclonal antibody. Samples were deparaffinized with xylene and then ethanol. After rehydration, sections were immersed in a trishydroxymethylaminomethane/ethylene diamine tetraacetic acid buffer (pH 9), microwaved for 25 minutes, cooled for 15 to 30 minutes, and placed in 1× PBS for 5 minutes. After an endogenous peroxidase block and a treatment with a protein block solution, sections were washed with 1× PBS for 5 minutes, stained for 1 hour with anti-FOXP3 (diluted 1:100; clone number ab22510, Abcam, Cambridge, United Kingdom), washed, and incubated for another 30 minutes with anti-mouse IgG polymer horseradish peroxidase–labeled antibody (Novolink polymer detection system, Novocastra, Newcastle upon Tyne, United Kingdom). The bound antibody was revealed by the addition of a diaminobenzidine solution. Specimens were counterstained with Carazzi's hematoxylin solution.

Proliferation Assay.

The suppressive function of CD4+CD25hi cells from 15 patients (5 [A] patients and 10 [R] patients) and 10 controls was evaluated in a proliferation assay. After purification, CD4+CD25hi T cells were added at a ratio of 1:8 (selected as optimal on the basis of preliminary experiments in which ratios of 1:16, 1:8, and 1:4 were compared) to autologous CD4+CD25 cells seeded at 5 × 105/well in a 96-well plate. Control cultures with CD4+CD25 T cells instead of Tregs and CD4+CD25 cells cultured on their own were performed under identical conditions. Experiments were performed in duplicate. Cells were cultured for 5 days with a T cell expander capable of preserving the original T cell function (CD3/CD28 Dynabeads, Dynal Biotech). rIL-2 (Chiron) was added at 30 U/mL on day 1. On the last day of culture, cell proliferation was tested by the addition of 0.25 μCi of [3H]thymidine/well, and this was followed by harvesting 18 hours later. Incorporated [3H]thymidine was assessed with a β-counter (Canberra Packard, Ltd., Pangbourne, United Kingdom). The inhibition percentage was calculated as follows:

equation image

Statistical Analysis.

The normality of the variable distribution was assessed by the Kolmogorov-Smirnov goodness-of-fit test; once the hypothesis of normality was accepted (P > 0.05), a comparison was performed with the Student t test. If the values were not normally distributed, the analysis was performed with the Mann-Whitney test. Categorical variables were compared with Fisher's exact test. Correlations were assessed with Pearson's or Spearman's correlation coefficient. A P value < 0.05 was considered significant.


Frequency of Treg Populations (Table 2).

The percentage of CD4+CD25hi T cells in patients with AIH was lower than that in HCs; this difference reached statistical significance not only when undivided AIH patients were considered but also when the subgroups of [A] patients and [R] patients were analyzed separately ([A] patients versus HCs, P = 0.05; [R] patients versus HCs, P = 0.02).

Table 2. Frequencies of Treg Populations
 Undivided AIH PatientsHCsAIH [A] PatientsAIH [R] Patients
  • Values are expressed as percentages of unfractionated PBMCs (means and standard deviations). Comparisons are made between undivided AIH patients and HCs and between AIH [A] patients and AIH [R] patients. Mann-Whitney and t tests were used as appropriate.

  • *

    P < 0.005.

  • P < 0.05.

CD4+CD25hi1.19 ± 0.20*3.56 ± 1.14*0.70 ± 0.181.38 ± 0.19
CD8+CD2824.57 ± 2.8821.71 ± 1.8422.91 ± 6.1425.54 ± 2.92
CD3+CD56+ (NKT)8.26 ± 1.09*16.19 ± 1.83*6.16 ± 1.179.32 ± 1.50
CD3+γδTCR+5.03 ± 0.693.85 ± 0.896.81 ± 1.694.15 ± 0.74
CD3+Vδ1+4.38 ± 1.133.32 ± 0.465.70 ± 2.203.16 ± 1.12
CD3+Vδ2+3.64 ± 0.645.75 ± 1.142.97 ± 1.074.26 ± 0.73
Vδ1+/Vδ2+ ratio1.49 ± 0.250.84 ± 0.152.19 ± 0.430.91 ± 0.18

CD45RO and CD62L expression did not differ between patients and controls; it was present in about 70% of CD4+CD25hi T cells in both groups.

CD8+CD28 T cell numbers were similar in patients and controls.

The number of CD3+CD56+ (NKT) cells mirrored the pattern of CD4+CD25hi cells: they were significantly lower in undivided AIH patients versus controls and lower in [A] patients versus [R] patients (AIH [A] patients versus HCs, P = 0.001; AIH [R] patients versus HCs, P = 0.005).

The numbers of γδTCR-expressing cells were comparable between AIH and controls, but the physiological ratio of circulating Vδ1 and Vδ2 cells, preserved in [R] patients, was inverted in [A] patients (AIH [A] patients versus HCs, P = 0.001).

Functional Characterization.

The expression of FOXP3 and CTLA-4 was evaluated in magnetically purified CD4+CD25+ lymphocytes and recorded both as the percentage of positive cells and as the mean fluorescence intensity (MFI; Table 3). FOXP3 expression was significantly reduced in AIH patients versus controls (P = 0.038 for the percentage and P = 0.012 for the MFI), regardless of disease activity, whereas CTLA-4 was similarly expressed in AIH patients and controls.

Table 3. Intracellular Staining
 All AIH PatientsHCs
  • % refers to the percentage of positive cells (means and standard deviations). Granzyme B production was evaluated in γδTCR-positive cells. IL-4 and INF-γ were evaluated on an NKT cell subset after α-GalCer stimulation. Comparisons are made between undivided AIH patients and HCs. Mann-Whitney and t tests were used as appropriate.

  • *

    P < 0.05.

  • P = 0.001.

Purified CD4+CD25+ T lymphocytes  
  %92.00 ± 1.73*96.35 ± 0.47*
  MFI28.85 ± 2.21*39.41 ± 2.87*
  %96.75 ± 1.7898.94 ± 0.39
  MFI85.53 ± 5.5395.75 ± 7.40
γδTCR and NKT-gated PBMCs  
 Granzyme B  
  %51.20 ± 4.31*37.52 ± 4.19*
  MFI32.75 ± 4.7813.78 ± 2.13
 IL-4, %16.10 ± 2.40*43.18 ± 11.49*
 INF-γ, %54.29 ± 8.2951.85 ± 10.19

CD8+CD28 lymphocytes from both HCs and AIH patients did not produce significant amounts of IL-10 in our ex vivo culture setting (data not shown).

Granzyme B expression by γδTCR-positive cells was higher in AIH patients versus controls in terms of both the percentage of positive cells and the fluorescence intensity (Table 3), the latter mirroring disease activity and being significantly higher in [A] patients versus [R] patients (44.05 ± 7.83 and 22.39 ± 4.04, P = 0.027; Fig. 1). The granzyme B MFI correlated with the ALT levels (r2 = 0.16, P = 0.047). Granzyme B expression, considered as both the percentage of positive cells and the MFI, correlated with the bilirubin levels (Fig. 2).

Figure 1.

Intracellular granzyme B staining of CD3+/γδTCR+ cells. Mann-Whitney and t tests were used as appropriate. % refers to the percentage of positive cells. 1P < 0.05; 2P < 0.005.

Figure 2.

Correlation between granzyme B expression by CD3+/γδTCR+ cells and bilirubin levels: (▴) percentage of CD3+/γδTCR+ cells producing granzyme B, (▪) granzyme B MFI, (—) correlation between the percentage of CD3+/γδTCR+ cells producing granzyme B and bilirubin levels (P = 0.05), and (---) correlation between the granzyme B MFI and bilirubin levels (P = 0.002).

The number of Vδ1-positive cells producing IFN-γ after stimulation with PMA and ionomycin was higher in AIH patients versus HCs (3.69% ± 0.66% versus 1.76% ± 0.36%, P = 0.02), with no difference between [A] patients and [R] patients. IFN-γ MFI levels and production by the Vδ2 subset were comparable in all groups. No correlations were found between IFN-γ production and laboratory indices.

The stimulation of PBMCs with α-GalCer resulted in a higher expansion of CD3+CD56+ cells with respect to the baseline in patients (567% ± 153%) versus HCs (190% ± 25%), the poststimulation NKT cell frequency being similar in the two groups (25.0% ± 6.2% in HCs versus 19.8% ± 11.2% in AIH patients, P = 0.51).

Although no difference in the frequency of IFN-γ–producing NKT cells was noted between the two groups, the frequency of IL-4–producing NKT cells was lower in AIH patients versus HCs (Table 3), this decrease being particularly evident in AIH [A] patients (15.0% ± 2.5%, P = 0.035; Table 3).


FOXP3+ cells were detected in the portal tracts of five of seven liver biopsy samples from tested AIH patients (all were histologically active, and two had normal aminotransferase levels; Fig. 3, Supporting Fig. 1, and Supporting Table 1). FOXP3+ cells represented a small proportion of the portal tract inflammatory infiltrate, their presence and number being unrelated to the liver disease stage.

Figure 3.

Immunohistochemical staining of FOXP3+ cells in the liver biopsy samples of two AIH patients at a low magnification (upper panels) and a high magnification (lower panels). Both had florid portal tract inflammatory infiltrates, but FOXP3+ cells (stained in brown) were seen only in patient 1.

Proliferation Assays.

After the addition of CD4+ CD25hi T lymphocytes, the mean CD4+CD25 T cell count per minute decreased by 57% in HCs (from 27,150 ± 7172 to 12,948 ± 4697 cpm, P = 0.001) and by 34% in AIH patients (from 22,114 ± 3167 to 16,424 ± 3170 cpm, P = 0.02), with the inhibition percentage lower than that in HCs (P = 0.009). No significant difference was noted in the suppression ability of Tregs between AIH [A] patients and AIH [R] patients, the inhibition of CD4+CD25 T cell proliferation being 27% in the former and 37% in the latter. Control experiments in which CD4+CD25 T lymphocytes were used instead of Tregs had no detectable effect on the proliferation of CD4+ CD25 T cells in AIH patients or HCs.


Liver damage in AIH is orchestrated by CD4+ T lymphocytes that recognize autoantigenic liver cell epitopes.35 If not effectively controlled by immunoregulation, these autoreactive T cells perpetuate self-aggression against the liver and lead to chronic hepatitis and cirrhosis.3

Compelling evidence obtained from animal models indicates that CD4+CD25hi lymphocytes prevent or cure autoimmune disorders by restoring immunotolerance to autoantigens.8, 9 Numerical and functional CD4+CD25hi cell impairment has been reported in a number of organ-specific autoimmune diseases, including diabetes,36 multiple sclerosis,37 rheumatoid arthritis,38 and primary biliary cirrhosis.39

Impairment of suppressor cell function in AIH was described in the 1980s and was found to be corrected after in vitro exposure to therapeutic doses of steroids.40, 41 In those early studies, however, a detailed phenotypic and functional characterization of the defective suppressor cells could not be provided. In children with AIH, CD4+CD25hi Tregs are impaired in both number and function versus normal controls, and these defects, being more evident at diagnosis than during drug-induced remission, parallel the clinical expression of the disease.14-16

Akin to the results in the juvenile form of AIH, our data indicate that CD4+CD25hi T cells are numerically reduced, express lower levels of FOXP3, and have less effective inhibitory activity in adults with AIH-1 compared to HCs. The numerical decrease in Tregs is more marked during active disease but also persists during immunosuppression-induced remission.

In a murine model of AIH,42 the numerical defect in Tregs observed in the circulation was attributed to hepatic sequestration because massive portal tract infiltration by CD8 T cells was accompanied by an equally abundant presence of CD4+CD25+FOXP3+ T cells, which were deemed by the authors to be recruited in the tissue to counterbalance the CD8-mediated damaging immunoresponse. Ebinuma et al.43 reported that in liver specimens from AIH patients, FOXP3+ cells were confined to portal tracts with marked cellular infiltration. In the present study, FOXP3+ cells were detected in most liver biopsy samples from AIH patients, but they represented a small component of the florid portal tract inflammatory infiltrate; in some biopsy samples, they were absent despite severe interface hepatitis. Further immunohistochemical studies of a larger number of patients are necessary to clarify the quantitative and spatial relationship between liver infiltrating effector and regulatory cells at different stages of disease activity. Isolation of tissue Tregs from explanted livers at the time of transplantation will also help in clarifying their functional properties.

A novel finding of this study is related to the behavior of NKT cells, which mirrors that of CD4+CD25hi T cells. The number of NKT cells is particularly low during active disease and is only partially restored after drug-induced remission. In addition, NKT cells from AIH patients produce lower amounts of the regulatory cytokine IL-4 than those from HCs, especially during the active phase of the disease but also during drug-induced remission, and this indicates a role for defective NKT cell numbers and function in permitting liver autoaggression. The similar behavior of CD4+CD25hi T cells and NKT cells may be explained by the fact that they share essential signaling pathways that could account for concerted responses, such as secretion by activated NKT cells of IL-2, a cytokine essential for CD4+CD25hi T cell function in both mice and humans.23 The observation that, despite biochemical remission of the disease, these two regulatory cell populations remain numerically and functionally impaired may provide an explanation for the propensity of AIH to relapse during the withdrawal of immunosuppressive treatment, even in patients with resolution of the inflammatory infiltrate on histology.44, 45 In contrast to what has been observed for CD4+CD25hi T cells and NKT cells, our data suggest that CD8+CD28 Tregs are unlikely to play a major role in AIH because no difference in their number was observed between patients and controls.

In sharp contrast to the behavior of CD4+CD25hi T cells and NKT cells, the number of γδ T cells is elevated during the active phases of the disease. Importantly, we have demonstrated increased production of their effector molecule, granzyme B, whose level of expression correlates with biochemical indices of liver damage such as ALT and bilirubin levels; this suggests the direct involvement of this cell population in hepatic injury. This notion is further supported by the observation that in [A] patients, there is an inversion of the physiological Vδ1/Vδ2 ratio in favor of Vδ1 cells, which have an enhanced ability to produce the proinflammatory cytokine IFN-γ. This finding is similar to what we have observed in patients with hepatitis C virus–related chronic liver disease, in which an inverted Vδ1/Vδ2 ratio is associated with active hepatitis with markedly elevated aminotransferase levels.46 Disruption of the physiological γδ T cell balance has also been linked to the inflammatory process in other autoimmune diseases.31, 32

In conclusion, our data show that a profound impairment of T cell regulation characterizes the adult form of AIH and is not confined to classical CD4+CD25hi T cells, in that it also involves NKT cells. Moreover, we have demonstrated that γδ T cells, normally able to perform both regulatory and effector functions, in AIH are skewed toward the latter and are likely to be involved in the pathogenesis of liver damage. Further studies are needed to dissect the complex interplay between regulatory and effector cell functions in the circulation and in the livers of patients with AIH. This knowledge is essential for the establishment of cell-based immunotherapies aimed at restraining the inflammatory autoimmune attack while reconstituting tolerance to liver autoantigens.


The authors are grateful to Dr. Alberto Quaglia for his assistance with the photography.