Potential conflict of interest: Nothing to report.
Rodrigo Liberal is supported by a doctoral grant from the Science and Technology Foundation, Science and Higher Education Ministry, Portugal; Giorgina Mieli-Vergani is supported by WellChild, Cheltenham, United Kingdom, and the Children's Liver Disease Foundation, Birmingham, United Kingdom; Maria Serena Longhi was supported by the Roger Dobson Fund, King's College Hospital Charity, United Kingdom, when this project was started and is currently supported by a Clinician Scientist Fellowship from the Medical Research Council, United Kingdom.
In autoimmune hepatitis (AIH), liver-damaging CD4 T cell responses are associated with defective CD4posCD25pos regulatory T cells (T-regs). Galectin-9 (Gal9), a β-galactosidase–binding protein expressed by T-regs, is key to their function, inhibiting T helper 1 immune responses by binding T cell immunoglobulin and mucin domain 3 (Tim-3) on CD4 effector cells. We investigated whether impaired immunoregulation in AIH results from reduced expression of Gal9 in T-regs and/or Tim-3 on CD4 effector cells. Circulating Gal9posCD4posCD25pos and Tim-3posCD4posCD25neg T cell phenotype was assessed by flow cytometry in 75 AIH patients. To evaluate whether Tim-3 expression renders CD4posCD25neg T cells amenable to T-reg control, purified CD4posCD25negTim-3pos (Tim-3pos) and CD4posCD25negTim-3neg (Tim-3neg) cells were cocultured with T-regs. To determine whether Gal9 expression is essential to function, T-regs were treated with small interfering RNA (siRNA) to repress Gal-9 translation; T-reg suppressor function was assessed by proliferation. In AIH, Tim-3pos cells within CD4posCD25neg cells and their T-betpos and RORCpos subsets were fewer and contained higher numbers of interferon-γ (IFNγ)pos and interleukin (IL)-17pos cells than healthy subjects (HS). In AIH and HS, Tim-3pos cells proliferated less vigorously and were more susceptible to T-reg control than Tim-3neg cells. In AIH, Gal9posT-regs were fewer and contained less FOXP3pos, IL-10pos, and transforming growth factor βpos and more IFNγpos and IL-17pos cells than HS. siRNA treatment of Gal-9pos T-regs drastically reduced T-reg ability to suppress CD4posCD25neg and Tim-3pos cell proliferation in AIH and HS. Tim-3pos cell percentage correlated inversely with aminotransferase and CD25negT-betpos cell values. Conclusion: Reduced levels of Tim-3 on CD4posCD25neg effector cells and of Gal9 in T-regs contribute to impaired immunoregulation in AIH by rendering effector cells less prone to T-reg control and T-regs less capable of suppressing. (HEPATOLOGY 2012)
Autoimmune hepatitis (AIH) is a progressive inflammatory liver disorder characterized by hypergammaglobulinemia, circulating autoantibodies, and histologically by a florid mononuclear cell infiltration referred to as interface hepatitis.1, 2 CD4 effector lymphocytes are the main orchestrators of liver damage in AIH, their proliferation and proinflammatory cytokine secretion (e.g., interferon-γ [IFNγ]) being correlated with the activity and severity of liver disease.3 We have provided evidence that in AIH, the extent of autoreactive CD4 T cell effector immune responses is associated with a numerical and functional impairment of CD4posCD25high regulatory T cells (T-regs),4-6 a lymphocyte subset central to immunotolerance maintenance.7, 8 Because they are defective, T-regs in AIH are unable to control proliferation and effector cytokine production (i.e., IFNγ, interleukin [IL]-17) by responder CD4 T cells stimulated in vitro with polyclonal or antigen-specific stimuli.9
Whether impaired immunoregulation in AIH is mainly due to a primary T-reg defect or also to low responsiveness of CD4 effector cells to T-reg control is unclear. Murine studies have shown that galectin-9 (Gal9), a member of the galectin family expressed by T-regs, inhibits T helper 1 (Th1) effector immune responses after binding to the T cell immunoglobulin and mucin domain 3 (Tim-3), its receptor on CD4 effector cells.10, 11 In the context of experimental autoimmune encephalomyelitis, mice immunized with an immunodominant myelin-oligodendrocyte-glycoprotein (MOG) peptide run a more severe disease course and have a higher mortality rate than mice immunized with MOG peptide but injected with Gal9, where a loss of MOG-specific IFNγ-producing CD4 T cells is observed.12 On the other hand, the blockade of Tim-3 on Th1 cells through anti–Tim-3 antibody administration accelerates diabetes development in nonobese diabetic mice and prevents transplantation tolerance in a model of islet allograft.13, 14 In humans, down-regulation of Tim-3 has been reported in T cell clones established from the cerebrospinal fluid of patients with multiple sclerosis and found to be associated with high levels of IFNγ secretion.15, 16 In hepatitis C–infected patients, overexpression of Tim-3 characterizes CD4 and CD8 T cells and is associated with viral persistence.17, 18 Blockade of Tim-3 leads to restoration of virus-specific immune responses, particularly of hepatocyte-directed cytotoxicity,18 indicating that immunotherapeutic control of Tim-3 expression may favor viral clearance. All these investigations highlight the importance of the Gal9/Tim-3 interaction in immunoregulation, expression of Gal9 and Tim-3 reflecting the ability of T-regs to suppress and the responsiveness of effector cells to T-reg control.
The aim of the present study is to explore the role of the Gal9/Tim-3 pathway in AIH and particularly to determine whether impaired immunoregulation in this condition is the result of defective T-reg function, low responsiveness of the effectors to T-reg control, or both.
Seventy-five patients with antinuclear antibody and/or smooth muscle antibody positive AIH (44 female patients) were studied. A liver biopsy performed at the time or close to diagnosis showed histological features of interface hepatitis in all. Eighteen of them had bile duct changes characteristic of sclerosing cholangitis on retrograde cholangiography and were diagnosed as having overlap autoimmune hepatitis/sclerosing cholangitis (ASC) syndrome.19 When considered together, AIH and ASC are henceforth indicated as autoimmune liver disease (AILD). The median age at the time of the study was 13.8 years (range, 8.2-21.2 years). All patients were studied while on immunosuppressive treatment, since T-regs collected at disease presentation before immunosuppression have been shown to be inefficient at suppressing cell proliferation.4, 20 Fifty-five patients were in remission (i.e., with normal aminotransferase levels), while 20 were studied during an episode of relapse. Treatment consisted of prednisolone (2.5-5 mg daily at remission and 1-2 mg/kg/day at relapse) and azathioprine (1-2 mg/kg/day). Demographic and biochemical data are shown in Table 1. Twenty-six HS (20 female patients; median age, 29.2 years [range, 22.6-39.4 years]) served as normal controls, the disparity between patient and control age deriving from ethical constraints in obtaining blood from healthy children. To test whether age disparity may account for differences, patients were divided into two subgroups (≤14 years and >14 years). Written consent was obtained from each patient. The study was approved by the Ethical Committee of King's College Hospital, London, UK.
Table 1. Patient Demographics and Laboratory Data
No. of Patients (N = 75)
AST (Normal: <50 IU/L)
Bilirubin (Normal: <20 μmol/L)
IgG (Normal: 6.5-17 g/L)
Autoantibody Titer (Reciprocal)
Data are presented as the median (range) unless noted otherwise.
Abbreviations: ANA, anti-nuclear antibody; F, female; LKM-1, liver kidney microsomal antibody type 1; M, male; SMA, smooth muscle antibody.
P < 0.001,
P = 0.02,
P = 0.004 when comparing AST, bilirubin, and IgG levels between AILD at remission and at relapse.
Peripheral blood mononuclear cells (PBMCs) were obtained as described.5 Mononuclear cell viability as determined by trypan blue exclusion exceeded 98%.
PBMCs were used either fresh or cryopreserved, the latter being stored in liquid nitrogen until the time of testing. Preliminary experiments in which the cell preparations from the same patients were tested before and after cryopreservation showed no significant difference in viability and behavior in culture when assessed for proliferation and cytokine production in the case of effectors and ability to suppress in the case of T-regs. Unfractionated cells were stained with allophycocyanin (APC)-cychrome (Cy)-7-conjugated anti-CD4, fluorescein isothiocyanate (FITC)-conjugated, or phycoerythrin (PE)-Cy-7–conjugated anti-CD25, PE-conjugated anti-CD127 (all from BD Bioscience Discovery Labware, Oxford, UK) and PE- or allophycocyanin-conjugated anti–Tim-3 (R&D Systems, Abingdon, UK) monoclonal antibodies (mAbs). Cells were incubated at 4°C in the dark for 30 minutes, washed with phosphate-buffered saline/1% fetal bovine serum, resuspended, and analyzed by flow cytometry on a Becton Dickinson fluorescent-activated cell sorter (FACSCantoII, Becton Dickinson Immunocytochemistry Systems, San Jose, CA). FACSDiva software was used for analysis. A minimum of 2 × 104 gated events was acquired for each sample.
The percentage of cells positive for FOXP3, T-bet, GATA-3, and RORC, transcription factors of T-regs, Th1, Th2 and Th17 cells, was determined by intracellular staining after cell fixation and permeabilization with Cytofix/Cytoperm (BD Bioscience) and counterstaining with FITC-conjugated anti-FOXP3 (clone PCH101), peridinin chlorophyll protein-Cy5–conjugated anti–T-bet, PE-Cy7–conjugated anti-GATA3, or PE-conjugated anti-RORC mAbs (all from eBioscience, Hatfield, UK). The percentage of Gal9-positive cells was determined after cell incubation with mouse immunoglobulin G1 (IgG1) anti-human Gal9 mAb (MBL, Nagoya, Japan) and with PE-conjugated anti-IgG1 secondary antibody (BD Bioscience).
The percentage of IFNγ, IL-17, IL-10, and transforming growth factor β (TGFβ)-producing cells was assessed before and after exposure to phorbol 12-mystrate 13-acetate (10 ng/mL)/ionomycin (500 ng/mL) (both from Sigma Aldrich, Gillingham, UK), incubation with brefeldin A (10 μg/mL; Sigma-Aldrich) for 5 hours and counterstaining with FITC- or allophycocyanin-conjugated anti–IL-17 (eBioscience), anti-IFNγ (IQ Products, Groningen, The Netherlands), anti–IL-10 (BD Bioscience) and peridinin chlorophyll protein–conjugated anti-TGFβ mAbs (R&D Systems). Flow cytometry was performed as above.
CD4posCD25pos (henceforth referred to as T-regs) and CD4posCD25neg cells were isolated from PBMCs using immunomagnetic beads (Dynal Invitrogen, Oslo, Norway) as described.4, 5
CD4posCD25neg cells were further purified according to the expression of Tim-3. In brief, CD4posCD25neg cells were incubated with PE-conjugated anti–Tim-3 mAbs for 30 minutes, then with microbeads conjugated to monoclonal anti-PE antibodies (Miltenyi Biotec, Bergisch-Gladbach, Germany) for a further 15 minutes at 4°C. CD4posCD25negTim-3neg and CD4posCD25negTim-3pos (henceforth Tim-3neg and Tim-3pos) populations were purified by negative and positive selection, respectively, using MS columns (Miltenyi Biotec) according to the manufacturer's instructions. CD4posCD25posCD127neg cells (henceforth referred to as CD127neg T-regs), or “true” T-regs,21 were purified from CD4posCD25pos cells after incubation with PE-conjugated anti-CD127 (BD Bioscience) and anti-PE microbeads (Miltenyi Biotec) (see above).
Once purified, T-regs or CD127neg T-regs were added to autologous CD4posCD25neg, Tim-3pos and Tim-3neg cells at a ratio of 1/84. Cells were cultured at 37°C and 5% CO2 for 5 days in the presence of anti-CD3/anti-CD28 T cell expander (ratio bead/cell: 1/2) (Dynal Invitrogen) and recombinant IL-2 (30 U/mL) (Chiron, Amsterdam, The Netherlands). In parallel, CD4posCD25neg, Tim-3pos, and Tim-3neg responder cells were cultured on their own under identical conditions and used as controls. All experiments were performed in duplicate. For the last 18 hours, cells were pulsed with 0.25 μCi/well 3H-thymidine and harvested using a multichannel harvester. The amount of incorporated 3H-thymidine was determined using a β-counter (Canberra Packard, Pangbourne, UK). The percentage inhibition was calculated using the following formula: (1 − cpm in the presence of T-regs or CD127neg T-regs/cpm in the absence of T-regs or CD127neg T-regs).
To investigate whether the susceptibility of the responder cells to T-reg control is related to the release of effector cytokines, CD4posCD25neg, Tim-3pos, and Tim-3neg cells were exposed to anti-IFNγ and anti–IL-17 neutralizing antibodies (10 μg/mL) (R&D Systems) for 12 hours before T-regs were added and during the 5-day coculture period. To assess whether T-reg control over responder cell proliferation relates to IL-10 secretion, anti-human IL-10–neutralizing antibodies (R&D Systems) were added at 10 μg/mL to T-regs cocultured with the effectors.
Gal9 Small Interfering RNA
To evaluate whether the expression of Gal9 by T-regs is relevant to their ability to suppress, T-regs and CD127neg T-regs were treated with a set of three Stealth RNAis to block the expression of Gal9. Cells were resuspended in Opti-MEM medium (Invitrogen Life Technologies) at 2-3 × 106/mL and transfected using Lipofectamine RNAiMAX reagent (Invitrogen Life Technologies) according to the manufacturer's instructions. Gal9-specific Stealth RNAis were used at a final concentration of 3 nM. A glyceraldehyde 3-phosphate dehydrogenase Stealth RNAi and a negative control Stealth RNAi (Invitrogen Life Technologies) served as positive and negative controls for transfection, respectively. Following overnight incubation at 37°C and 5% CO2, aliquots of 2.5 × 105 cells were collected to extract RNA and assess Gal9 expression by real-time polymerase chain reaction as described.22 Following small interfering RNA (siRNA) treatment, cells were cocultured with CD4posCD25neg, Tim-3pos, and Tim-3neg responder cells, whose proliferation was assessed as detailed above. The effect of Gal9 gene knockdown on T-reg suppressor ability was also evaluated after sorting CD25pos cells into CD25high and CD25low. CD25pos cells were sorted using a FACSAria (Becton Dickinson Immunocytochemistry Systems). The purity of both CD25low and CD25high cells exceeded 98%.
The normality of variable distribution was assessed by the Kolmogorov-Smirnov goodness of fit test; once the hypothesis of normality was accepted (P > 0.05), comparisons were performed by paired and unpaired Student t test as appropriate. Correlation analysis was determined by Pearson's correlation coefficient. Results are expressed as the mean ± SEM. P < 0.05 was considered significant; P ≥ 0.05 and P ≤ 0.15 were considered to indicate a trend toward significance.
Enumeration and Characterization of CD4posCD25negTim-3pos Cells.
The percentage of Tim-3pos lymphocytes within CD4posCD25neg cells was lower in patients (4.2 ± 0.4) than in HS (7.9 ± 0.8; P < 0.001) whether they were studied at relapse or during remission (Fig. 1A,B). Tim3pos cells were lower at relapse than during remission (Fig. 1B). Within AILD patients, no differences were observed between AIH and ASC patients and between the two age subgroups.
The percentage of Tim-3pos cells within the T-betpos and the RORCpos effector subsets was lower in AILD (T-betpos: 5.1 ± 0.7; RORCpos: 3.9 ± 0.3) than in health (T-betpos: 13.6±1.7, P < 0.001; RORCpos: 8.2 ± 1.1, P = 0.02).
Compared with HS (Table 2), the Tim-3pos subset in AILD contained more RORCpos, IFNγpos, and IL-17pos cells and similar numbers of T-betpos, GATA-3pos, FOXP3pos, IL-10pos, and TGFβpos cells; the Tim-3neg subset in AILD contained more T-betpos, RORCpos, IFNγpos, and IL-17pos cells, fewer GATA-3pos cells, and similar proportions of FOXP3pos, IL-10pos, and TGFβpos lymphocytes. Compared with the Tim-3neg cell population, Tim-3pos lymphocytes in AILD contained more T-betpos, FOXP3pos, IFNγpos, IL-17pos, IL-10pos, and TGFβpos cells, fewer GATA-3pos cells, and similar numbers of RORCpos cells; in HS, Tim-3pos lymphocytes contained more T-betpos, IFNγpos, IL-17pos, IL-10pos, and TGFβpos cells, fewer GATA-3pos cells, and similar numbers of RORCpos and FOXP3pos cells (Table 2).
Table 2. Percentage of T-betpos, GATA-3pos, RORCpos, FOXP3pos, IFNγpos, IL-17pos, IL-10pos, and TGFβpos Cells Within CD4posCD25neg, Tim-3pos, and Tim-3neg Cells
Data are presented as the mean ± SEM and refer to 20 AILD patients and 12 HS.
Abbreviation: NS, not significant.
Within the Tim-3pos subset between AILD and HS.
Within the Tim-3neg subset between AILD and HS.
Between the Tim-3pos and the Tim-3neg subset in AILD.
Between the Tim-3pos and the Tim-3neg subset in HS.
91.2 ± 0.8
93.1 ± 0.5
10.1 ± 0.9
6.1 ± 0.9
0.71 ± 0.23
0.78 ± 0.24
2 ± 0.5
2.8 ± 0.4
7.2 ± 0.9
2.9 ± 0.5
5.9 ± 0.4
3.5 ± 0.5
2.84 ± 0.85
2.74 ± 0.49
0.5 ± 0.2
1.6 ± 0.8
11.2 ± 0.8
4.2 ± 0.6
6.09 ± 0.7
2.4 ± 0.3
7.9 ± 1.2
4.3 ± 0.8
3.1 ± 0.7
1.5 ± 0.2
7.1 ± 1
9.9 ± 1.2
5.1 ± 0.7
6.7 ± 1
6.8 ± 0.4
9.8 ± 1.9
3.9 ± 1.2
4.7 ± 0.4
Enumeration and Characterization of CD4posCD25posGal9pos Cells.
The percentage of Gal9pos cells within CD4posCD25pos cells was lower in patients (30.6 ± 3.1) than in HS (49.4 ± 3; P < 0.001), this difference being evident also when the CD25low (23.9 ± 2.9 versus 44.7 ± 3.2; P < 0.001), CD25med (29.5 ± 3.1 versus 49.9 ± 2.8; P < 0.001), and CD25high (35.2 ± 2.1 versus 65.6 ± 3.8; P < 0.001) fractions were analyzed separately (Fig. 2). CD127neg T-regs contained the highest percentage of Gal9pos cells. This percentage was lower in AILD (40.9 ± 4.1) than in HS (72.2 ± 4.8; P < 0.001). Within AILD, there was no difference between patients with AIH and those with ASC and between the two age subgroups.
Compared with HS (Table 3), the Gal9pos T-reg subset in AILD contained more IFNγpos and IL-17pos cells, fewer FOXP3pos, IL-10pos, and TGFβpos cells and similar numbers of T-betpos, GATA-3pos, and RORCpos cells; the Gal9neg subset in AILD contained more RORCpos, IFNγpos and IL-17pos cells, fewer FOXP3pos cells, and similar proportions of T-betpos, GATA3pos, IL-10pos, and TGFβpos lymphocytes. Compared with the Gal9neg population, Gal9pos T-regs in AILD contained more FOXP3pos and IL-10pos cells, fewer RORCpos and IL-17pos cells, and similar numbers of T-betpos, GATA-3pos, IFNγpos, and TGFβpos cells; in HS, Gal9pos T-regs contained more FOXP3pos, IL-10pos, and TGFβpos cells, fewer RORCpos cells, and similar numbers of T-betpos, GATA-3pos, IFNγpos, and IL-17pos cells.
Table 3. Percentage of T-betpos, GATA-3pos, RORCpos, FOXP3pos, IFNγpos, IL-17pos, IL-10pos, and TGFβpos Cells Within CD4posCD25pos, Gal9pos, and Gal9neg Cells
Data are presented as the mean ± SEM and refer to 23 AILD patients and 17 HS.
Abbreviation: NS, not significant.
Within the Gal9pos subset between AILD and HS.
Within the Gal9neg subset between AILD and HS.
Between the Gal9pos and the Gal9neg subset in AILD.
Between the Gal9pos and the Gal9neg subset in HS.
9.5 ± 1.3
9.4 ± 4.4
9.7 ± 1.5
8.1 ± 1.6
7.3 ± 1.2
7.4 ± 3.8
5.6 ± 1.6
5.6 ± 1
1.1 ± 0.6
0.9 ± 0.4
8.6 ± 0.9
4.4 ± 0.8
14.4 ± 2
42.8 ± 3.1
2.6 ± 0.5
16.4 ± 1.5
4.4 ± 0.6
2.1 ± 0.3
5.7 ± 0.7
2.2 ± 0.4
4.1 ± 0.6
1.75 ± 0.2
7.9 ± 1.2
2.3 ± 0.7
5.1 ± 0.6
9.1 ± 0.5
2.3 ± 0.2
3.3 ± 0.4
6.3 ± 0.7
8 ± 0.4
5.3 ± 0.5
5.6 ± 1.1
Proliferation of CD4posCD25neg, Tim-3pos, and Tim-3neg Cells and Responsiveness to T-reg Control.
The proliferation of unfractionated CD25neg cells was lower than that of Tim-3neg cells (AILD, P = 0.02; HS, P = 0.001) and higher than that of Tim-3pos cells (AILD, P = 0.03; HS, P = 0.04) (Fig. 3). Addition of T-regs reduced cell proliferation by 26% (P not significant) in AILD and 53% (P = 0.007) in HS when unfractionated CD25neg cells were used as responders; by 23% (P not significant) and 25% (P not significant) when Tim-3neg cells were the responders; and by 47% (P = 0.03) and 62% (P = 0.001) when the responder cells were Tim-3pos (Fig. 4).
Addition of CD127neg T-regs reduced proliferation of CD25neg cells by 46% in AILD (P < 0.001) and 60% in HS (P < 0.001) and proliferation of Tim-3pos cells by 56% in AILD (P = 0.006) and 69% in HS (P < 0.001).
As the percentage of IFNγ-producing and, to a lesser extent, IL-17–producing cells was higher in Tim-3pos than in Tim-3neg cells, we investigated the effect of IFNγ and IL-17 neutralization on the ability of CD25neg, Tim-3neg, and Tim-3pos cells to be regulated by T-regs. Addition of anti-IFNγ–neutralizing antibodies did not change CD25neg and Tim-3neg cell responsiveness to T-reg control (Fig. 4A,B). In contrast, treatment with anti-IFNγ abrogated Tim-3pos responsiveness to T-reg control (Fig. 4C). Exposure to anti-IFNγ–neutralizing antibodies, while leaving unchanged the percentage of Tim-3pos lymphocytes within undivided CD4 cells in both AILD patients and HS, tended to decrease that of Tim-3pos lymphocytes within CD25neg cells in HS after 12-hour (8.5 ± 0.4 versus 6.2 ± 1.2; P = 0.12) and 5-day (15.5 ± 1.6 versus 12.2 ± 0.6; P = 0.1) culture. Similarly, while no difference in Tim-3 expression on a cell-per-basis (expressed as mean fluorescence intensity) was noted in CD4 undivided cells in both AILD patients and HS, within CD25neg cells from HS, Tim-3 mean fluorescence intensity tended to decrease after 12-hour (798.5 ± 35.3 versus 678 ± 62.2; P = 0.14) and 5-day (978.2 ± 62.6 versus 808.2 ± 86.3; P = 0.14) culture. In HS, anti-IFNγ–induced decrease in Tim-3pos CD25neg cells and in Tim-3 mean fluorescence intensity was abrogated by T-reg addition.
Treatment with anti–IL-17–neutralizing antibodies did not change the responsiveness of CD25neg, Tim-3neg, and Tim-3pos responder cells to T-reg control in both patients and HS. Addition of anti-IFNγ and anti–IL-17–neutralizing antibodies together left unchanged the responsiveness of CD25neg and Tim-3neg cells, while reduced that of Tim-3pos cells (Fig. 4A-C).
To explore whether T-reg control over responder cell proliferation is influenced by IL-10 secretion, anti–IL-10–neutralizing antibodies were added to T-regs cocultured with effectors obtained from four AILD patients and four HS. Following treatment, T-reg inhibition of CD25neg cell proliferation did not change in AILD but was decreased from 51% to 34% (P = 0.07) in HS. Anti–IL-10–neutralizing antibody treatment resulted in no change in the ability of T-regs to suppress Tim-3neg cells, but decreased T-reg inhibition over Tim-3pos cell proliferation from 42% to 36% (P = 0.06) in AILD and from 56% to 48% (P = 0.04) in HS.
Gal9 Gene Knockdown.
Treatment of T-regs with Gal9-RNAi led to a decrease in the expression of Gal9 gene by 86% in AILD and by 88% in HS.
After addition of Gal9-siRNA–treated T-regs, inhibition of cell proliferation was reduced by 43.5% in AILD (P < 0.001) and by 67% in HS (P < 0.001) when CD25neg cells were used as responders, by 43% (P < 0.001) and 48.7% (P = 0.038) when Tim-3neg cells were the responders, and by 70.4% and 73.5% when the responders were Tim-3pos lymphocytes.
After treatment with Gal9-RNAi, CD127neg T-reg suppressor function was reduced by 54.4% (P < 0.001) in AILD and 66.7% (P < 0.001) in HS when responders were CD25neg cells; by 36.4% (P = NS) in AILD and 51.62% (P = 0.06) in HS when responders were Tim-3neg cells; and by 75% (P = 0.001) in AILD and 71% (P < 0.001) in HS when responders were Tim-3pos cells.
In two HS where higher numbers of cells were available, the effect of Gal9 gene knockdown on T-reg suppression was also evaluated after sorting T-regs according to their CD25 expression. CD25high T-reg suppressor function was markedly reduced by Gal9 siRNA treatment: inhibition of CD25neg cell proliferation was 57% following addition of untreated CD25high T-regs and 14% after addition of Gal9 siRNA-treated CD25high T-regs. No difference in the ability of CD25low T-regs to suppress was noted before and after Gal9 siRNA treatment (25% versus 27%).
The percentage of Tim-3poscells was inversely correlated with the levels of aspartate aminotransferase (AST) (r = −0.47; P = 0.002). The percentage of CD25negT-betpos cells was positively correlated with AST levels (r = 0.69; P < 0.01) and negatively with that of Tim-3pos cells (r = −0.91; P < 0.001).
The percentage of Gal9pos T-regs was inversely correlated with the titer of IgG (r = −0.39, P < 0.02), smooth muscle antibody (r = −0.32, P = 0.05), and, to a lesser extent, antinuclear antibody (r = −0.27, P < 0.1).
This study shows that in AILD numerical and functional T-reg impairment is accompanied by low susceptibility of responder cells to the control exerted by T-regs, indicating that there is a defect both at the regulatory and effector cell levels. Thus, the present results, besides confirming our previous observations on reduced number and suppressor function of T-regs in AILD, provide evidence that low expression of Gal9 could be one of the mechanisms responsible for T-reg impairment, as highlighted by gene knockdown experiments where T-regs treated with a set of Gal9-specific Stealth RNAis were less effective at controlling the proliferation of responder cells, especially in the case of CD25high T-regs.
Neutralization experiments showing reduced T-reg suppression following treatment with anti–IL-10–neutralizing antibodies indicate that Gal9pos T-regs partly act through IL-10 production/secretion. This finding suggests that Gal9pos T-regs, in addition to regulating effectors via induction of their apoptosis,23 can deliver suppression via IL-10 secretion. That Gal9pos T-regs exert a control over disease activity was supported by the observation of an inverse correlation between Gal9pos T-regs and IgG levels and titers of autoantibodies, which are the serological markers of the disease. These results echo a murine study wherein the context of autoimmune encephalomyelitis Gal9 was shown to have a direct effect on disease severity, as mice immunized with a MOG immunodominant peptide and injected in vivo with Gal9 had a less severe disease course and a lower mortality rate than mice that were not injected.12
In AILD, decreased percentage of Gal9pos T-regs is mirrored by a down-modulation of Tim-3 on the surface of CD4posCD25neg cells. Whether down-regulation of Tim-3 on effectors is the result of low Gal9pos T-reg number or a defective Th1 cell maturation/differentiation process is unclear. Our data demonstrating low percentage of Tim-3 on T-betpos cells support the second hypothesis and suggest that in AILD, Th1 cells arrest at a T-bet positive status having acquired effector properties, without differentiating into the T-bet/Tim-3 double positive status, which characterizes terminally differentiated effector cells that are susceptible to T-reg control. This is in line with the report of Anderson et al.,24 who documented that T-bet−/− mice have defective Tim-3 expression on CD4 T cells and that T-bet binds directly to the Tim-3 promoter, regulating its expression. Future studies in AILD should also measure soluble Tim-3, as its role is controversial: while administration of soluble Tim-3-Ig fusion protein augments Th1 immune responses in SJL/J mice immunized with a myelin proteolipid protein peptide,14 the overexpression of soluble Tim-3 by in vivo delivery of plasmid DNA in C57BL/6 mice inoculated with melanoma cells results in the inhibition of Th1 cytokines and impaired T cell antitumor response.25
That Tim-3 renders effectors responsive to T-reg control was clearly shown in the current study by a series of experiments in which the Tim-3neg and the Tim-3pos cell fractions were tested for their ability to be regulated by T-regs. The Tim-3pos fraction was the most amenable to T-reg control, especially when CD127neg T-regs were used as suppressors; responsiveness of Tim-3pos cells, however, was more evident in health than in AILD, indicating that in this condition, immunoregulation impairment includes low responder cell susceptibility to T-reg suppression in addition to a defective T-reg function. It is therefore plausible that, as supported by our data, in AILD T-betpos/Tim-3neg cells prevail over the T-betpos/Tim-3pos cells, accounting for the high number of poorly controllable effectors that contribute to the unfolding of the liver damage. That Tim-3 down-modulation is associated with disease activity was confirmed by the strong negative correlation between percentage of Tim-3pos effectors and AST levels.
Further phenotypic and functional analyses performed in the current study indicate that regulation of Tim-3pos cells by T-regs depends on IFNγ production, as blockade of IFNγ results in reduced Tim-3 expression of CD4posCD25neg cells and lower susceptibility of Tim-3pos cells to T-reg control. These data suggest that a proinflammatory signature is necessary for the effectors in order to be “seen” and consequently regulated by T-regs via the Gal9/Tim-3 pathway.
The higher percentage of IL-17–producing and RORCpos cells within the Tim-3pos fraction shows that Tim-3 is expressed also on other types of effectors, i.e. Th17 cells, a subset involved in the autoimmune liver damage.1 Tim-3 expression by Th17 cells is likely to contribute to their ability to be controlled by T-regs, as reported by Chen et al.26 in mice and Hastings et al.27 in humans. Relevant to our findings, in the collagen-induced arthritis model, treatment with Gal9 blocks Th17 cell induction and IL-17 messenger RNA expression even when CD4-naïve cells are exposed to TGFβ and IL-6, cytokines central to Th17 development in mice.28
The information obtained in the present study is crucial for devising T-reg–based immunotherapy for the possible treatment/cure of AILD. In view of the reported direct role of Gal9 in T-reg induction and differentiation in a murine setting,28 further studies should explore whether in the context of AILD T-reg suppression could be restored by transfecting T-regs with human Gal9 complementary DNA or, alternatively, by culturing them in the presence of Gal9. In this context, it is of interest that culture of PBMCs in the presence of Gal9 results in the expansion of CD4posCD25posFOXP3posCD127low T-regs both in hepatitis C–infected patients and healthy subjects.29
In conclusion, our data show that defective immunoregulation is due not only to T-reg inability to suppress, but also to a low susceptibility of effector cells to T-reg control due to decreased expression of Tim-3. Because Tim-3 expression is reduced when the disease is active, adoptive transfer of autologous T-regs may be more efficacious once inflammation is dampened by immunosuppression to render effectors more amenable to T-reg inhibition.