Human intrahepatic regulatory T cells are functional, require IL-2 from effector cells for survival, and are susceptible to Fas ligand-mediated apoptosis.

UNLABELLED: Regulatory T cells (Treg ) suppress T effector cell proliferation and maintain immune homeostasis. Autoimmune liver diseases persist despite high frequencies of Treg in the liver, suggesting that the local hepatic microenvironment might affect Treg stability, survival, and function. We hypothesized that interactions between Treg and endothelial cells during recruitment and then with epithelial cells within the liver affect Treg stability, survival, and function. To model this, we explored the function of Treg after migration through human hepatic sinusoidal-endothelium (postendothelial migrated Treg [PEM Treg ]) and the effect of subsequent interactions with cholangiocytes and local proinflammatory cytokines on survival and stability of Treg . Our findings suggest that the intrahepatic microenvironment is highly enriched with proinflammatory cytokines but deficient in the Treg survival cytokine interleukin (IL)-2. Migration through endothelium into a model mimicking the inflamed liver microenvironment did not affect Treg stability; however, functional capacity was reduced. Furthermore, the addition of exogenous IL-2 enhanced PEM Treg phosphorylated STAT5 signaling compared with PEMCD8. CD4 and CD8 T cells are the main source of IL-2 in the inflamed liver. Liver-infiltrating Treg reside close to bile ducts and coculture with cholangiocytes or their supernatants induced preferential apoptosis of Treg compared with CD8 effector cells. Treg from diseased livers expressed high levels of CD95, and their apoptosis was inhibited by IL-2 or blockade of CD95.
CONCLUSION: Recruitment through endothelium does not impair Treg stability, but a proinflammatory microenvironment deficient in IL-2 leads to impaired function and increased susceptibility of Treg to epithelial cell-induced Fas-mediated apoptosis. These results provide a mechanism to explain Treg dysfunction in inflamed tissues and suggest that IL-2 supplementation, particularly if used in conjunction with Treg therapy, could restore immune homeostasis in inflammatory and autoimmune liver disease. (Hepatology 2016;64:138-150).

analysis of variance

biliary epithelial cell

enzyme‐linked immunosorbent assay

Fas ligand

interleukin

hepatic sinusoidal endothelial cell

interferon‐γ

liver‐infiltrating regulatory T cell

postendothelial migrated Treg

standard error of the mean

T helper 1

T helper 17

tumor necrosis factor α

regulatory T cell

CD4+CD25+ CD127low FOXP3+ regulatory T cells (Treg) maintain peripheral self‐tolerance by suppressing T effector cell proliferation and function.1, 2 Treg depletion or dysfunction leads to autoimmunity in mice and IPEX syndrome in humans. Treg dysfunction or an imbalance between Treg and effector cells in tissue will determine the outcome of inflammation.3, 4 Intrahepatic Treg increase in parallel with effector T cells during chronic hepatitis5 demonstrates that the presence of Treg in the liver does not prevent ongoing hepatic inflammation. Whether this is due to dysfunctional Treg or overwhelming effector responses is not known.

Understanding the fate of Treg in the inflamed liver is crucial, because Treg recruited to the liver from blood via hepatic sinusoids6 enter the hepatic microenvironment enriched with proinflammatory cytokines. The process of recruitment and/or exposure to this inflammatory microenvironment could affect Treg stability, survival, and function. Recruitment through endothelium can affect the differentiation and activation of various leukocyte subsets, and Treg can switch toward a T helper 1 (Th 1) or T helper 17 (Th 17) lineage in an inflammatory environment.7, 8 Furthermore, intrahepatic lymphocytes are removed by activation‐induced cell death triggered by receptors such as CD95 (APO‐1/Fas), leading to resolution of inflammation. Thus, differential susceptibility to apoptosis could affect the balance of Treg and T effector cells at inflammatory sites.9, 10 Treg are highly sensitive to CD95‐mediated apoptosis,11 and interleukin (IL)‐2‐dependent phosphorylation of STAT5 is crucial for their proliferation, differentiation, and survival.12 We reported previously that many intrahepatic Treg lack evidence of STAT5 phosphorylation, suggesting that the inflammatory liver microenvironment is hostile to Treg survival and function.13

In the present study, we report that recruitment through endothelium into a model of the inflamed liver microenvironment reduces the suppressive capacity of Treg and a lack of local IL‐2 enhances their susceptibility to Fas‐mediated apoptosis induced by epithelial cells. These findings support the therapeutic potential of IL‐2 therapy to restore local Treg function in inflammatory liver diseases.

Materials and Methods

TISSUES AND BLOOD

Venous blood was obtained from healthy volunteers or patients with hemochromatosis who were admitted to the Liver Unit at the Queen Elizabeth Hospital, Birmingham, UK5 (Local Research Ethics Committee reference no. 04/Q2708/41) and liver tissues from patients undergoing liver transplantation for inflammatory liver diseases, including primary biliary cirrhosis, primary sclerosing cholangitis, alcoholic liver disease, and autoimmune hepatitis. Normal liver was obtained from donor liver tissue surplus to clinical requirements (Local Research Ethics Committee reference no. 06/Q2708/11, 98/CA5192).

ISOLATION OF LIVER INFILTRATING LYMPHOCYTES, BILIARY EPITHELIAL CELLS, AND PRIMARY HUMAN HEPATIC SINUSOIDAL ENDOTHELIAL CELLS

Liver‐infiltrating lymphocytes,5 biliary epithelial cells (BECs),14 and hepatic sinusoidal endothelial cells (HSECs)15 were prepared and isolated from fresh liver tissue as described previously.13

POSTENDOTHELIAL TRANSMIGRATED Treg AND CD8+ T CELL FUNCTIONAL ASSAYS

HSECs were plated and grown on six‐well collagen I precoated Transwell inserts (3 μm; Greiner Bio‐One) until confluent. Cells were then stimulated with 10 ng/mL interferon‐γ (IFN‐γ) and 10 ng/mL tumor necrosis factor α (TNF‐α) (both Peprotech) for 24 hours to mimic the inflamed environment. After washing, the lower chamber was filled with either Roswell Park Memorial Institute 1640 (RPMI‐1640) medium (Gibco) supplemented with 0.05% bovine serum albumin or inflamed liver conditioned supernatant prepared by incubating 1 g of tissue in 4 mL serum‐free RPMI 1640 for 24 hours. The supernatant was then centrifuged and filtered (0.22 μm‐pore). Freshly isolated Treg or CD8+ T cells were then added to the upper chamber of the Transwell insert in 0.05% bovine serum albumin/RPMI‐1640 and allowed to migrate across the endothelial cell monolayer into the lower chamber for 24 hours. The postendothelial transmigrated Treg (PEM Treg) or postendothelial transmigrated CD8+ T (PEM CD8) cells were collected for experiments (Treg suppression assay, biliary coculture assay, phosphorylated STAT5 assay, and Treg plasticity assay).

PEM Treg AND CD8 APOPTOSIS ASSAY

BECs were seeded into 24‐well collagen I precoated plates, grown until confluent, and stimulated with 10 ng/mL IFN‐γ and 10 ng/mL TNF‐α (Peprotech) for 24 hours. PEM Treg or PEM CD8 cells were cocultured with these BECs or with BEC supernatant for 24 hours, and apoptosis was analyzed by way of flow cytometric staining using Annexin and 7AAD (BD Pharmingen).

STATISTICAL ANALYSIS

Differences between treatments were evaluated by way of Student t test or one‐way analysis of variance (ANOVA), followed by Bonferroni multiple comparison test using GraphPad Prism version 6.0 (GraphPad Software). P < 0.05 was considered statistically significant. Data are presented as the mean ± standard error of the mean (SEM).

See the Supporting Information for details regarding immunohistochemistry, confocal microscopy, enzyme‐linked immunosorbent assay (ELISA), flow cytometry, isolation of peripheral blood Treg and CD8 and assays of Treg function, plasticity, and response to IL‐2.

Results

FRESHLY ISOLATED HUMAN LIVER INFILTRATING Treg ARE ACTIVATED, NONEXHAUSTED, MEMORY CELLS

We compared the frequency and surface phenotype of liver‐infiltrating CD4+CD25highCD127low Treg (LITreg) in different inflammatory liver diseases with normal liver tissue (Fig. 1A,B) and with liver‐infiltrating CD8 cells (Supporting Fig. S1C). Treg isolated from normal and diseased livers differed in their expression of Treg‐associated functional surface receptors CD26, CD39, and CD69 (Fig. 1B). Very low expression of PD1 was observed on LITreg in both normal and diseased patients (9±5% versus 9.5±2%) (Fig. 1B) but high levels of CD44 were observed on LITreg (79±6% versus 87±5%) and liver‐infiltrating CD8 cells (75±9% versus 90±4%) (Supporting Fig. S1C). LITreg exhibited a memory phenotype: CD45ROhighCD45RAlow and CCR7low (85±7%) and inflamed liver tissue contained small populations of central memory CD45RAnegCCR7pos (4.7±3%) and tissue resident memory CD45RAposCCR7neg (7±3%) LITreg (Fig. 1C,D).

Figure 1

Frequency and phenotype of intrahepatic Treg in inflamed human livers. Freshly isolated LITreg from human explanted livers were phenotyped by flow cytometry. LITreg were gated as CD4+CD25highCD127low. (A) Frequency of LITreg in normal livers and different diseased livers. (B) Surface marker expressions of LITreg from normal livers, alcoholic liver disease livers, and autoimmune disease livers (see Supporting Fig. S1C for surface marker expressions of LICD8 cells from different diseased livers). LITreg (defined as CD3+CD4+CD127lowCD25high) were screened for expression of surface markers, CD26, CD39, CD44, CD69, and PD1. Data are presented as the mean ± SEM (n = 6; one‐way ANOVA followed by Bonferroni multiple comparison test). **P < 0.05. (C, D) Maturation status of LITreg. Frequencies of CD45RA− memory (CD45RA−CCR7−), CD45RO+ memory (CD45RO+CCR7−), central memory (CD45RA−CCR7+), and naïve (CD45RA+ CCR7+) LITreg populations were determined by way of flow cytometry. Representative dot plots (C) and summary data (D) of LITreg maturation status are shown. Data are presented as the mean ± SEM (n = 6).

PROINFLAMMATORY CYTOKINES ARE ELEVATED IN THE INFLAMED LIVER ENVIRONMENT, BUT THIS DOES NOT ALTER INTRAHEPATIC Treg STABILITY

We explored the intrahepatic microenvironment by measuring proinflammatory cytokines in supernatants from inflamed human liver tissue. Diseased liver supernatants contained higher concentrations of IL‐6 (8960 ± 4257 pg/mL), IL‐8 (24,033 ± 16,589 pg/mL), IL‐12 (61 ± 30 pg/mL), IFN‐γ (32 ± 7 pg/mL), and IL‐1β (363 ± 88 pg/mL) compared with normal liver supernatants (Fig. 2A).

Figure 2

Intrahepatic microenvironment is enriched with proinflammatory cytokines and PEM Treg are not plastic to other T cell lineages. (A) Cytokine profiles of supernatants collected from normal and diseased livers. Tissue from normal or diseased liver was cultured for 24 hours in RPMI‐1640 medium at 1 g/mL, and the concentrations of cytokines IL‐6, IL‐8, IL‐12, IL‐15, IFN‐γ, TNF‐α, and IL‐1β were measured using a Bio‐Plex Pro bead‐based multiplex kit. Data are presented as the mean ± SEM (n = 6; one‐way ANOVA with Bonferroni multiple comparison test). *P < 0.05. (B‐D) Treg phenotype before and after migration across stimulated hepatic sinusoidal endothelium. Treg freshly isolated from peripheral blood were transmigrated through monolayers of IFN‐γ‐ and TNF‐α‐stimulated hepatic sinusoidal endothelial cells into either RPMI‐1640 medium (control) or liver supernatant and phenotyped by way of flow cytometry. Expression of markers including CD26, CD39, CD44, PD1, CD147, CTLA‐4, CD95, CD27, and CXCR3 before and after transendothelial migration into RPMI‐1640 medium are shown (B). Data are presented as the mean ± SEM (n = 6; one‐way ANOVA followed by Bonferroni multiple comparison test). *P < 0.05. Expressions of cytokines including IL‐10, IL‐17, and IFN‐γ at 24 hours (C) and transcription factors FOXP3, ROR, and Tbet at 3 days (D) were measured before and after migration into RPMI‐1640 medium or liver supernatant (see Supporting Fig. S2 for expression of transcription factors up to 3 days of culture in inflamed liver supernatants). Data are presented as the mean ± SEM (n = 4; one‐way ANOVA). *P ≤ 0.05. (E) Expression of CD161, IL‐6 receptor, and intracellular AhR by LITreg directly ex vivo was shown. Data are presented as the mean ± SEM (n = 6; one‐way ANOVA followed by Bonferroni multiple comparison test). *P < 0.05; **P < 0.01; ***P < 0.005

We then examined the stability of intrahepatic Treg in the hepatic microenvironment by modeling recruitment and intrahepatic conditions in vitro. Blood Treg that had transmigrated across TNF‐α and IFN‐γ‐stimulated HSECs had a phenotype similar to the LITreg phenotype, allowing us to use them to model the inflamed liver in vitro (Fig. 2B). PEM Treg were migrated across stimulated HSECs into either control media or supernatants from inflamed liver tissue cultures (Supporting Fig. S5) and secretion of IL‐10, IL‐17, and IFN‐γ by Treg measured 24 hours postmigration (Fig. 2C) and expression of FOXP3, ROR, and Tbet analyzed up to 3 days (Fig. 2D, Supporting Fig. S2). A subset of LITreg expressed CD161 (20%) or IL‐6 receptor (27%). We did not detect AhR expression (Fig. 2E).

PEM Treg REMAIN FUNCTIONALLY SUPPRESSIVE

After migration across endothelium into inflamed tissue supernatants, Treg maintained their ability to suppress the proliferation of T effector cells (Fig. 3A, Supporting Fig. S5A). Freshly isolated Treg were functionally suppressive (Treg/T effector cell ratio = 1:8) before any cell contact (Fig. 3Ai) and after contact with endothelium (Treg/T effector cell ratio = 1:4) but before transmigration (Fig. 3Aii). After transmigration across TNF‐α and IFN‐γ‐stimulated HSECs into control media (Fig. 3Aiii), Treg had a reduced suppressive capacity at a 1:2 ratio. This was reduced further after migration into inflamed liver supernatant with suppression only seen at a 1:1 cell ratio (Fig. 3Aiv).

Figure 3

Intrahepatic Treg remain functional and express Treg functional markers. (A) Treg representative of intrahepatic Treg phenotype were generated by transmigration of freshly isolated blood Treg across TNF‐α‐ and IFN‐γ‐stimulated human sinusoidal endothelium into RPMI‐1640 control medium or inflamed liver supernatant for 24 hours. The suppressive function of Treg was analyzed by testing their ability to suppress the proliferation of CellTrace Violet prestained CD4 T effector cells using Miltenyi Treg suppression inspector. Treg populations compared included: (i) Treg directly isolated, without contact with any cells or supernatant; (ii) Treg in contact with endothelium but not transmigrated; (iii) PEM Treg migrated into control RPMI media; and (iv) PEM Treg migrated into liver supernatant. (B) Comparison of the expressions of Treg functional markers by freshly isolated human LITreg and CD8 T cells. Data are presented as the mean ± SEM (n = 5; one‐way ANOVA followed by Bonferroni multiple comparison test). *P < 0.05; **P < 0.01; ***P < 0.005. (C) Representative flow cytometry overlays for each functional marker on LITreg (CD3+CD4+CD25highCD127low) (top panel) and LICD8 (CD3+CD8+) (bottom panel).

LITreg expressed cell surface receptors associated with suppressive function (Fig. 3B,C) including CTLA‐4 and CD147,16 whereas factors associated with cytolytic activity, granzyme‐B, and perforin were expressed at higher levels on LICD8 T cells compared with LITreg (53% versus 8% and 27% versus 8%). There were no significant differences in the expression of LAG‐3 (21% versus 8%), CD137 (13% versus 4%), and TIM‐3 (8% versus 4%) between LITreg and LICD8 (Fig. 3B,C).

LIVER RESIDENT Treg AND CD8 RESIDE AROUND Fas LIGAND (FASL)‐POSITIVE BILE DUCTS

FOXP3+ LITreg were detected throughout the hepatic parenchyma, fibrous septa, and sinusoids as well as close to bile ducts in portal tracts (Fig. 4A‐C) in inflamed liver tissue. CD95 ligand (FASL/CD178) was detected on CK19+ bile ducts in all inflammatory liver disease but not on normal liver tissue (Fig. 4D),14 with no difference in expression between different liver diseases.

Figure 4

Human diseased liver bile ducts expressing FASL and LITreg are present across liver lobules, with some residing in the peribiliary region. Immunohistochemistry (A‐D) and confocal fluorescence staining (Biii) of paraffin‐embedded human liver sections for FOXP3, CD8, CK19 (biliary epithelial cell marker), and FASL are shown. (A) intrahepatic FOXP3+ Treg are present across the parenchyma (i), septa (ii), and portal tract (iii). (B) LICD8+ T cells reside close to bile ducts in the portal tracts in diseased human liver sections (Bi) alcoholic liver disease (ALD), (Bii) primary biliary cirrhosis (PBC), (Biii) Localization of CD8 cells on bile ducts in AIH liver tissue (red = CD8 [phycoerythrin]; green = bile ducts/CK19 [fluorescein isothiocyanate]; blue = nuclear stain [4′,6‐diamidino‐2‐phenylindole]). (C) Localization of FOXP3+ cells around the bile duct in diseased livers (Ci) autoimmune hepatitis (AIH), (Cii) hepatitis C virus (HCV), (Ciii) non A non B seronegative hepatitis. (D) Expression of FASL (right panel) by human CK19‐expressing bile ducts (left panel) in normal and diseased livers.

CD95 EXPRESSION IS INCREASED ON LIVER INFILTRATING Treg AND MEDIATES CELL APOPTOSIS IN RESPONSE TO BEC FASL

Treg from diseased livers were more susceptible to apoptosis (Fig. 5A). LITreg expressed CD154, OX40, CD40, CD95, and CD27 and high levels of CD95 in all diseased livers (Fig. 5B). Liver‐infiltrating CD8 cells were expressed at lower levels (Fig. 5C) Freshly isolated Treg and CD8 cells were transmigrated across TNF‐α and IFN‐γ‐stimulated hepatic sinusoidal endothelium into either control media, stimulated BEC supernatant, or contact with BECs (Fig. 6, Supporting Fig. S5B). Treg underwent apoptosis in this model, whereas very few CD8 T cells did. Treg apoptosis in response to contact with BECs or exposure to BEC supernatant was prevented by blocking CD95 ligand or by the addition of IL‐2 (Fig. 6). Apoptotic immune cells were present around the bile duct, and we observed FOXP3 Treg and caspase‐3 dual‐positive apoptotic cells in peribiliary regions in tissue slides (Fig. 6C).

Figure 5

Apoptotic potential and expression of the TNF receptor CD95 is increased on LITreg in diseased livers. (A) The apoptotic potential of LITreg in both normal and diseased livers was assessed by way of flow cytometry with Annexin 5 staining directly ex vivo. (B, C) Expression of TNF superfamily (TNFSF) and TNF receptor superfamily (TNFRSF) markers by LITreg (CD3+CD4+CD127lowCD25high) (B) and liver‐infiltrating CD8 (CD3+CD8+) (C) in normal livers (white bar = normal liver) and diseased livers (black bar = diseased liver). Data are presented as the mean ± SEM (n = 6; one‐way ANOVA followed by Bonferroni multiple comparison test). *P < 0.05.

Figure 6

Apoptosis of PEM Treg but not CD8 is FASL dependent and can be rescued by blocking FASL or by supplementing with IL‐2. Treg and CD8+ T cells were transmigrated across TNF‐α‐ and IFN‐γ‐stimulated human sinusoidal endothelium to obtain postendothelial migrated cells, which were then cocultured with BECs or BEC supernatant in the presence or absence of recombinant IL‐2 or anti‐FASL antibody. Apoptosis was analyzed using flow cytometry for Annexin V. (A) Representative dot plots of apoptosis for Treg and CD8 before and after endothelial migration. (B, C) Apoptosis of PEM Treg (B) and PEM CD8 cells (C) with either BECs or BEC supernatant in the presence or absence of IL‐2 or anti‐FASL (one‐way ANOVA followed by Bonferroni multiple comparison tests). **P ≤ 0.05. (D) Terminal deoxynucleotidyl transferase‐mediated deoxyuridine triphosphate nick‐end labeling revealed that apoptotic immune cells were present around bile ducts. (E) Dual staining of FOXP3 (brown) and caspase‐3 (blue) in different diseased livers.

We detected high expression of CD27 on LITreg and CD70 the ligand for CD27 on liver‐infiltrating dendritic cells. However, recombinant CD70 had no effect on Treg or CD8 T cell proliferation (Supporting Fig. S4).

EFFECTOR T CELLS ARE THE MAIN SOURCE OF INTRAHEPATIC IL‐2 AND Treg RESPOND BY PHOSPHORYLATION OF STAT5

IL‐2 was nearly undetectable in supernatants prepared from normal and chronic liver disease tissues (Fig. 7A). Secretion of IL‐2 by human hepatocytes, HSECs, stromal cells, and biliary epithelial cells was minimal (Fig. 7B). IL‐2 expression by liver‐infiltrating activated CD4/CD8 cells was observed by way of flow cytometry (Fig. 7C) and secreted IL‐2 was detected in the supernatant of liver‐infiltrating CD4 and CD8 lymphocytes after stimulation with anti‐CD3/CD28 beads, suggesting that effector lymphocytes are the main source of IL‐2 in inflamed livers (Fig. 7D).

Figure 7

IL‐2 is deficient in the liver microenvironment, and activated T lymphocytes are the source of IL‐2. (A) IL‐2 concentrations in normal and inflamed liver tissue supernatants are shown. Chronic diseased and normal liver tissues were diced and cultured for 24 hours in RPMI‐1640 medium at 1 g/mL, and the concentration of IL‐2 was measured by way of ELISA. (B) Minimal IL‐2 synthesis by liver‐resident cells under normal and inflamed conditions. Cultures of primary BECs, HSECs, hepatocytes, stellate cells, and fibroblasts from human livers were treated with the inflammatory cytokines TNF‐α (10 ng/mL) and IFN‐γ (100 ng/mL) or left untreated, and the concentration of IL‐2 in 24‐hour supernatants was determined by way of Luminex. Data are presented as the mean ± SEM (n = 3). (C, D) Intrahepatic CD4+ and CD8+ T cells express and secrete IL‐2 upon T cell receptor activation. CD4+ and CD8+ T cells were isolated from normal and diseased human liver tissues by way of cell sorting and were cultured for 12 hours at 100 cells/mL with or without anti‐CD3/anti‐CD28 stimulation 4 cells/bead. (C) IL‐2 expression was examined by way of flow cytometry. (Ci) Representative flow cytometry dot plots for IL‐2 production by stimulated (CD3/CD28 beads) or unstimulated CD4 and CD8 T cells in the presence or absence of additional PMA and ionomycin stimulation. (Cii, Ciii) Summary frequencies for IL‐2 expression without (Cii) and with (Ciii) additional PMA and ionomycin stimulation. (D) IL‐2 secretion was examined by way of ELISA. PMA, phorbol 12‐myristate 13‐acetate.

A high level of STAT5 phosphorylation was induced in 80%‐90% of Treg, both non‐transmigrated and PEM Treg (Fig. 8, Supporting Fig. S4), and was not affected by transmigration into the inflamed liver microenvironment (Fig. 8). Significantly fewer CD8 cells (27%‐36%) responded compared with Treg (Supporting Fig. S4), and the level of phosphorylated STAT5 generated was also significantly greater in Treg than in CD8 T cells (Fig. 8). Thus, IL‐2‐CD25 signaling and STAT5 phosphorylation remained intact in Treg after transmigration into the inflamed hepatic microenvironment.

Figure 8

STAT5 phosphorylation is functionally intact in response to supplemented IL‐2 in postendothelial migrated Treg in the liver microenvironment. STAT5 phosphorylation in pre‐endothelial migrated and PEM Treg and CD8 cells in response to IL‐2 stimulation is shown. Peripheral blood mononuclear cells were transmigrated overnight across 24‐hour TNF‐α‐ and IFN‐γ‐stimulated human sinusoidal endothelium into RPMI‐1640 medium or inflamed liver supernatant. Postendothelial migrated cells and non‐transmigrated cells were collected into separate fractions and were stimulated for 10 minutes with 0 or 100 IU/mL IL‐2. (A, B) Phosphorylated STAT5 (pY694) expression by (A) CD4+CD25+CD127− Treg and (B) CD8 cells in the fractions was examined by way of flow cytometry in the absence of IL‐2 stimulation (untreated) and after 10 minutes of treatment with 100 IU/mL IL‐2. (C) Comparison of the percentages of Treg and CD8 cells expressing phosphorylated STAT5 after treatment with IL‐2. (D) Comparison of the median fluorescence intensity (MFI) of STAT5 phosphorylation in Treg and CD8 cells after treatment with IL‐2. Data are presented as the mean ± SEM (n = 4; paired t test). *P < 0.05, **P < 0.01, ***P < 0.0001. Statistical tests compare cells from the same fraction (either untreated and IL‐2 treated Treg or CD8 cells (A and B) or IL‐2 treated Treg and CD8 cells (C and D)).

Discussion

Functionally active Treg are required to suppress effector cells and maintain immune homeostasis,13 and the balance of Treg and T effector cells will contribute to the outcome of liver inflammation.17 Therapeutic modulation of tissue resident Treg has potential for treating autoimmune liver diseases.17, 18 Treg enter the liver from blood via hepatic sinusoids5 before being localized within the liver in response to signals from the hepatic stromal compartment, hepatocytes, and cholangiocytes.19 Once Treg have migrated across the hepatic sinusoid, they are exposed to cytokines including IL‐1β, IL‐6, IL‐8, IL‐12, and IFN‐γ in the inflamed hepatic microenvironment (Fig. 2). This proinflammatory milieu has the potential to influence the stability and function of intrahepatic Treg.20 We observed that PEM Treg that had migrated through activated endothelium into liver tissue supernatants showed no changes in FOXP3, RORc, or Tbet expression at 3 days, suggesting that Treg differentiation is not affected by the liver milieu. In particular, the lack of ligand‐activated transcription factor aryl hydrocarbon receptor and low levels of CD161, which are associated with Th 17 polarity, suggest few cells differentiate into Th 17 cells within the inflamed environment.21, 22, 23

The frequency of LITreg is increased across all liver diseases,24 and these cells are not exhausted, with very low expression of PD‐1 and high levels of CD69 and maintained functional properties25, 26 (Fig. 3Aiv). LITreg exposed to a liver microenvironment maintain high levels of CD39, an ectonucleotidase used by Treg to generate immunosuppressive adenosine from extracellular nucleotides.27 Low expression of CD39 has been reported in autoimmune liver disease, and our data suggest that the hepatic microenvironment cannot explain this finding.28 Similarly, LITreg expressed high levels of CD44, which has been associated with FOXP3 expression and suppressive function.29 After transendothelial migration into inflamed liver supernatant, Treg maintained IL‐10 secretion and CD147 expression, a marker of activated and suppressive Treg.16 In addition, we also observed an enrichment of CTLA‐4 expression in LITreg. CTLA‐4 interacts with and removes CD80/CD86 from dendritic cells by way of transendocytosis, resulting in impaired costimulation via CD28 and is thus critical for Treg function.30 Overall, LITreg are equipped with key surface markers and intracellular cytokines required to execute their suppressive capacity.31, 32 These findings are encouraging for proposed studies of adoptive Treg therapy in autoimmune and inflammatory liver diseases because they suggest that therapeutic cells will remain functional and will not differentiate into effector cells in an inflammatory environment.17

We detected LITreg and LICD8 cells throughout the parenchyma and fibrous septa as well as around bile ducts in patients with liver disease. Biliary epithelial cells can secrete chemokines that localize lymphocytes to portal tracts.5, 13, 33, 34 We recently reported that VCAM‐1 on bile ducts supports effector T cells survival by binding to α4β1,35 but apoptosis pathways affecting different lymphocyte subsets in the inflamed liver are poorly understood. BECs express FASL, which can induce T cell death by activating Fas. Thus, expression or secretion of FASL by BECs could contribute to intrahepatic T cell apoptosis.36

Circulating Treg are highly susceptible to CD95‐FASL‐dependent apoptosis but not to TCR‐mediated cell death in contrast to effector T cells.11 Treg in the tumor undergo CD95‐dependent cell death as a consequence of FASL expression by tumor cells.37 We have reported previously that BECs express FASL in inflammatory liver diseases,14 but we have not explored the role of Fas in the differential fate of bile duct‐associated Treg and CD8 cells. In the present study, we demonstrate that PEM Treg are more susceptible to CD95‐FASL‐mediated apoptosis either in contact with BECs or in response to BEC supernatant compared with PEM CD8 cells. Treg could be rescued by blocking FASL on BECs or in the inflamed biliary supernatant. If apoptosis of Treg affects only those in contact with bile ducts, and not those elsewhere in the liver, this may not have a substantial effect on the overall frequency of Treg in the diseased liver. In addition, the total frequency of Treg in the inflamed liver will depend on other factors, including the balance between recruitment, retention, proliferation, and exit of cells from the liver.

IL‐2 is a crucial cytokine for Treg survival that activates STAT5 phosphorylation in response to activation of the IL‐2 receptor. IL‐2 can selectively expand Treg populations in vitro and in vivo.38, 39 We detected minimal IL‐2 in both the normal and inflamed liver microenvironment, which might account in part for reduced Treg functional capacity in the inflammatory tissue. The intrahepatic source of IL‐2 is activated in CD4 and CD8 cells, which may be rapidly consumed by both effector and Treg for their survival and proliferation.40, 41 Exogenous IL‐2 restored levels of STAT5 phosphorylation and protected Treg from Fas‐mediated apoptosis but had no effect on CD8 T cells. Thus, the lack of IL‐2 in the inflamed liver microenvironment may result in unopposed Fas‐mediated Treg apoptosis as well as contributing to defective intrahepatic LITreg function. Because the requirement for IL‐2 is much higher for CD8 cells, they may depend on different survival mechanisms. This is in agreement with a previous study on the effect of IL‐2 on peripheral blood lymphocyte subsets.42 This enhanced phosphorylation of STAT5 in Treg compared with CD8 cells in response to IL‐2 occurred in the presence and absence of cytokine‐enriched inflamed hepatic supernatant, suggesting that the differential effect of IL‐2 on Treg compared with effector cells is maintained in an inflammatory environment. This finding provides support for the therapeutic manipulation of LITreg by treating these patients with IL‐2 along with Treg therapy in the future to maintain functional and surviving LITreg.

In conclusion, we show for the first time that LITreg are phenotypically stable within the inflamed hepatic environment and exhibit minimal plasticity. They remain functionally suppressive but are susceptible to apoptosis via the Fas pathway mediated by FASL on bile ducts, and this can be inhibited by exogenous IL‐2. Therefore, IL‐2 therapy may successfully restore intrahepatic Treg numbers and function in liver disease.

Author names in bold denote shared co‐first authorship.

Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.28517/suppinfo.

Supporting Information Figure S1

Click here for additional data file.

Supporting Information Figure S2

Click here for additional data file.

Supporting Information Figure S3

Click here for additional data file.

Supporting Information Figure S4

Click here for additional data file.

Supporting Information Figure S5

Click here for additional data file.

Supporting Information

Click here for additional data file.