In drug-induced, immune-mediated hepatitis, interleukin-33 reduces hepatitis and improves survival independently and as a consequence of FoxP3+ T-cell activity

Abstract

Immune-mediated, drug-induced hepatitis is a rare complication of halogenated volatile anesthetic administration. IL-4-regulated Th2-polarized reactions initiate this type and other types of hepatitis, while the mechanisms that regulate the severity remain elusive. IL-33 is an innate, IL-4-inducing, Th2-polarizing cytokine that has been detected in patients with liver failure and has been associated with upregulated ST2+Foxp3+CD4+CD25+ T cells; however, roles for IL-33 in drug-induced hepatitis are unclear. We investigated IL-33 in an anesthetic, immune-mediated hepatitis modeled in BALB/c, IL-33−/− and ST2−/− mice, as well as in patients with anesthetic hepatitis. The hepatic IL-33 and ST2 levels were elevated in BALB/c mice (p < 0.05) with hepatitis, and anti-IL-33 diminished hepatitis (p < 0.05) without reducing IL-33 levels. The complete absence of IL-33 reduced IL-10 (p < 0.05) and ST2+Foxp3+CD4+CD25+ T cells (p < 0.05), as well as reduced the overall survival (p < 0.05), suggesting suppressive roles for IL-33 in anesthetic, immune-mediated hepatitis. All of the mice demonstrated similar levels of CD4+ T-cell proliferation following direct T-cell receptor stimulation, but we detected splenic IL-33 and ST2-negative Foxp3+CD4+CD25+ T cells in ST2−/− mice that developed less hepatitis than BALB/c mice (p < 0.05), suggesting that ST2-negative Foxp3+CD4+CD25+ T cells reduced hepatitis. In patients, serum IL-33 and IPEX levels were correlated in controls (r² = 0.5, p < 0.05), similar to the levels in mice, but not in anesthetic hepatitis patients (r² = 0.01), who had elevated IL-33 (p < 0.001) and decreased IPEX (p < 0.01). Our results suggest that, in anesthetic, immune-mediated hepatitis, IL-33 does not regulate the CD4+ T-cell proliferation that initiates hepatitis, but IL-33, likely independent of ST2, reduces hepatitis via upregulation of Foxp3+CD4+CD25+ T cells. Further studies are needed to translate the role of IL-33 to human liver disease.

Graphical abstract

Our results show that, in anesthetic hepatitis, IL-33, likely independent of ST2, reduces hepatitis via upregulation of Foxp3+ Treg numbers and function.

Introduction

Liver disease and cirrhosis constitute the sixth most common cause of death in adults between the ages of 25 and 64 years.¹ Drug-induced hepatitis (DIH) is a leading cause of acute liver failure and the most common reason an approved medication is removed from the consumer market.¹ Various drugs can cause DIH, and current evidence suggests that susceptible individuals develop immune-mediated hepatitis after receiving halogenated anesthetics, anti-seizure medications, antibiotics, or non-steroidal anti-inflammatory drugs. Key mechanisms that initiate DIH have been uncovered, but the mechanisms that modulate disease severity remain elusive.

A major challenge to understanding the pathogenesis of DIH is its diverse presentations, which can range from toxic hepatitis-induced acute liver failure to immune-mediated hepatitis with autoimmune features. Even so, the post-translational modification of native proteins by reactive drug metabolites is generally accepted to be a key mechanism in the pathogenesis of immune-mediated DIH following halogenated anesthetics.² Furthermore, cytochrome P450 (CYP) 2E1 is a major autoantigen in this process.³ Interleukin (IL)-4-upregulated CD4+ T cells that recognize a post-translationally modified CYP2E1 and other liver proteins initiate immune-mediated, anesthetic DIH by attracting neutrophils, eosinophils and mast cells to the liver.^4,5 Induced FoxP3-expressing CD4+CD25+ T cells (Tregs)⁶ regulate severity in immune-mediated DIH, and the relative deficiencies of these cells in the spleen are thought to worsen immune-mediated DIH.

IL-33, an innate, IL-4-inducing, Th2-polarizing cytokine that plays a critical role in autoimmune disease, has been detected in patients with acute liver failure,⁷ but its roles in DIH are unclear. The serum levels of IL-33 are elevated in patients with hepatitis C⁸ and hepatitis B,⁹ possibly signifying a role in human liver diseases. However, studies in mice offer conflicting evidence regarding whether signaling through IL-33 and its receptor, ST2, is protective¹⁰ or pathogenic.¹¹ For example, in primary biliary cirrhosis, IL-33 repairs biliary cells, but signaling is carcinogenic.¹² Furthermore, IL-33 reduced liver damage in mice with concanavalin A immune-mediated hepatitis.^13–15 Cellular diversity may affect the outcomes after IL-33 expression. Thus, the roles attributed to IL-33 in hepatitis may depend heavily on the responding cells, triggering antigens, and accompanying innate and adaptive signals that predominate in the liver. Recently, IL-33 has been associated with the maturation and differentiation of ST2+Foxp3-expressing Tregs in autoimmune colitis.¹⁶ Additionally, IL-33 is thought to upregulate the expression of ST2, which is believed to increase the suppressive function of Tregs.^17–20 However, the significance of IL-33-induced ST2+ Treg maturation in immune-mediated DIH is unknown.

In this study, we investigated the role of IL-33 in immune-mediated DIH by injecting BALB/c mice and IL-33- and ST2-deficient (IL-33−/− and ST2−/−) mice with CYP2E1 epitopes altered by anesthetic metabolites to induce a model of DIH. We also studied IL-33 and IPEX, the human corollary of FoxP3, in patients with anesthetic hepatitis. Our goal is to uncover mechanisms that regulate the severity of DIH using mouse models that could uncover therapeutic targets or mechanisms that regulate the morbidity or mortality of DIH in patients.

Materials and methods

Human serum collection

All studies were approved by our Institutional Review Board. Human serum samples of surviving patients with anesthetic hepatitis (n = 32) were obtained from our anesthetic hepatitis-stored sera. In addition, control serum samples (n = 14) were obtained from healthy patients using complex antibodies.⁵ Briefly, blood was collected in serum separator tubes. The samples were allowed to clot and then were centrifuged (1000 × g, 4 °C) for 20 min. The serum samples were then divided into aliquots, snap frozen, and stored at −80 °C.

Mice

The procedures were approved by the Animal Care and Use Committee. Six- to eight-week old, female inbred BALB/c (wild-type) mice were supplied by The Jackson Laboratory (Bar Harbor, ME). Female IL-33−/− (BALB/c background) mice were obtained from Dr. Susumu Nakae (RIKEN Center for Developmental Biology, Kobe, Japan). Female ST2−/− (BALB/c background) mice were from Dr. Andrew McKenzie (MRC Laboratory of Molecular Biology, Cambridge, UK). All the mice were maintained under pathogen-free conditions in the animal facility.

Induction of hepatitis in mice

We induced hepatitis in BALB/c, IL-33−/−, and ST2−/− mice by immunizing them with a CYP2E1 epitope, GIIFNNGPTWKDIRRFSLTTL (JHDN5), covalently modified by trifluoroacetyl chloride (TFA) drug metabolites.⁴ Specifically, the mice were injected with 100 µg of TFA-JHDN5 (AnaSpec, Fremont, CA) emulsified in an equal volume of complete Freund’s adjuvant (CFA; Difco Bacto, Pittsburgh, PA) on days 0 and 7. Control mice received only CFA. On day 0, experimental and control mice were also immunized with 50 ng of pertussis toxin (List Biologicals, Campbell, CA). The mice were killed at 1, 2, or 3 weeks after the initial immunization (day 0). Each experiment was repeated three times and consisted of 5 mice per group.

Histology

Liver sections (5 µm thick) were fixed in 10% neutral buffered formalin and stained with H&E (Histoserv; Gaithersburg, MD). The histology scores are determined first at low power (×40) as an average of 2 views and confirmed at ×64. The tissue sections were scored as follows: Grade 0, no inflammation or necrosis; Grade 1, minor lobular inflammation with no necrosis; Grade 2, lobular inflammation involving <50% of the section; Grade 3, lobular inflammation involving >50% of the section; and Grade 4, inflammation with necrosis.

IL-33 blocking experiment

BALB/c mice were immunized as described above. In addition, each mouse was administered 3.6 µg of anti-IL-33 (diluted in sterile filtered 1× PBS) by intraperitoneal injection on days 0, 7, 14, and 20.^21,22 Experiments were run in duplicate with 2–3 mice per group.

ELISA

Human IL-33 and IPEX ELISA

The IL-33 levels in human serum samples were measured using the Quantikine ELISA kit (R&D Systems, Minneapolis, MN) according to the manufacturer’s instructions. The plates were read at 450 nm. The IPEX levels were measured in the same samples using the Human Foxp3 (IPEX) ELISA kit (Aviva Systems, San Diego, CA) according to the manufacturer’s instructions.

Mouse tissue ELISA

After 2 or 3 weeks, the liver and spleen samples (1 g) from each mouse was homogenized in 1 mL of RPMI (Sigma-Aldrich)/2% fetal bovine serum (FBS; Invitrogen Corporation, Carlsbad, CA) until smooth and then were centrifuged as previously described.⁴ The supernatant was snap frozen and stored at −80 °C until ready for use. The cytokine and chemokine levels were measured using Quantikine ELISA kits, as per kit instructions. The levels of cytokines were measured at 450 nm and were standardized by converting the levels to pg/g of tissue.

Flow cytometry

Splenocyte isolation

Splenocytes were isolated 2 weeks after the initial TFA-JHDN5 immunization. They were re-stimulated with media ± CYP2E1 to investigate immune responses to the autoantigen JHDN5, immune responses to the epitope of the autoantigen or TFA-OVA (10 μg/mL), and immune responses to the trifluoroacetyl chloride (TFA) drug metabolite, followed by incubation for 72 h at 37 °C in 5% CO₂ and 95% air (humidified). After incubation, the cells were pooled by stimulation of antigen and washed, and the supernatant was collected and used for cytokine/chemokine analyses by ELISA as well as cellular analysis by flow cytometry.

Immune cell isolation from the liver

Liver cells, isolated as previously described,⁴ and splenocytes⁵ were collected at baseline, day 14, or day 21 and were washed with PBS/2% FBS. Next, 1 × 10⁶ cells were incubated with FcR blocking reagent (Miltenyi Biotec, Gaithersburg, MD) and were stained with 1:100 dilutions of CD4-FITC, CD25-PE, and CD45-PerCP (all from Miltenyi Biotec) for 30 min. Next, the cells were stained intracellularly with FoxP3-APC, fixed with 250 µL of fixation buffer (BD Pharmingen, San Jose, CA), and analyzed by flow cytometry. Additional cells were assessed for ST2 expression.

Gating strategy

Live cells were identified using the Live/Dead Fixable Aqua Dead Cell stain kit (Thermofisher). We determined CD4+CD25+FoxP3+ Tregs by first gating on liver cells and splenocytes that were CD45+ (Miltenyi, PerCP, clone RA3-6B2). Cells that were CD45+ were then gated to show CD4+ cells (Miltenyi, FITC, cloneGK1.5). From the CD4+ liver cells and splenocytes, the percentages of CD25+ (Miltenyi, PE, clone 7D4) FoxP3+ (Miltenyi, APC, clone 3G3) cells in addition to those expressing ST2+ (BD Biosciences, BV421, clone U29-93) were determined by analysis. We normalized the percentages to numbers of cells using total cell counts in the liver or spleen (Supplemental Fig. 2D).

Treg isolation for co-culture experiments

We isolated Tregs from naive mice using the mouse CD4+CD25+ Regulatory T Cell Isolation Kit (Miltenyi Biotec). Briefly, liver and spleen CD4+ cells were identified by negative selection. Pooled samples of cells were stained with CD25-PE, purified with LD columns and then purified twice with LS columns to achieve the highest purity. Finally, they were stained with CD4-FITC. Tregs utilized for co-culture (1:100) were not stained. All materials used for isolating and staining Tregs were purchased from Miltenyi Biotec.

Proliferation assays

Splenocytes from BALB/c, IL-33−/−, or ST2−/− mice were incubated in 2 µL of 5 µM CFSE (eBioscience, San Diego, CA) in the dark for 10 min and then were washed with PBS/2% FBS at 1300 rpm. Next, CFSE-labeled cells were cultured at 5 × 10⁶ per well in six-well plates at 37 °C for 24 h with 0.5 µg of anti-mouse CD3e (eBioscience, clone 145-2C11) and 0.25 µg of anti-mouse CD28 (eBioscience, clone 37.51). The splenocytes were then washed twice with PBS/2% FBS at 1300 rpm and stained with CD4-APC (1:100; eBioscience) for 30 min at 4 °C. In some experiments, IL-33−/− splenocyte cultures were supplemented with IL-10 (0–10 ng). After incubation, the cells were washed and fixed with 250 µL of fixation buffer until they were analyzed by flow cytometry. Proliferating cells were evaluated using histograms, where the peaks of proliferating cells were analyzed between strains or after treatment with IL-10. A left shift of the peak demonstrated more proliferation, while a right shift demonstrated less proliferation. The percentages of proliferating cells were also analyzed.

Gating strategy

Live cells were identified using the Live/Dead Fixable Aqua Dead Cell stain kit (Thermofisher). We identified CFSE+CD4+ (APC; clone GK1.5) splenocytes within the live cell population.

NanoString

A NanoString Pan Cancer Immunology Panel (774 genes) was carried out using the Johns Hopkins Deep Sequencing and Microarray Core facility as per kit instructions. Briefly, RNA counts from liver samples of BALB/c and IL-33−/− mice were measured using nCounter and analyzed for differences with the analysis package provided.

Statistical analysis

Histology scores, antibodies, cytokines, chemokines, and in vitro assays were analyzed between groups using the Mann–Whitney U-test. Correlation analysis between IL-33 and IPEX in human sera was carried out using logistic regression (GraphPad Prism Version 6 for Windows, GraphPad Software, Inc. San Diego, CA). A p-value <0.05 was considered statistically significant.

Results

Hepatic IL-33 and ST2 levels are upregulated in BALB/c mice with experimental, immune-mediated DIH

Our model of DIH is characterized by a splenic (initiation) phase 2 weeks after immunizations in which IL-4-driven cytokines and antigen-specific CD4+ T cells are detected in the spleen, followed by a hepatic (effector) phase, in which the CD4+ T cells have migrated to the liver. The hepatic phase peaks at 3 weeks, when primarily neutrophils, as well as mast cells, T cells, B cells, and macrophages are detected.⁵ In this study, we measured, IL-33 and ST2 levels in the liver and spleen, measured liver tissue cytokines, and scored each liver for inflammation after 2 and 3 weeks. Hepatic IL-33 (p < 0.001) and ST2 (p < 0.001) levels in BALB/c livers were elevated 2 weeks after immunization compared with those in control mice (Fig. 1a), suggesting that IL-33 and ST2 may play a role in the pathogenesis of DIH. Although, the histology showed no difference in hepatitis between the immunized and control groups at 2 weeks (Fig. 1b), the immunized group exhibited elevated levels of hepatic IL-4 (p < 0.001), IL-6 (p < 0.01), IL-7 (p < 0.05), and IL-17 (p < 0.05) cytokines (Fig. 1c), which play key roles in the pathogenesis of DIH.⁵ Interestingly, splenic IL-33 and ST2 were unchanged at this time point (not shown), suggesting liver-specific roles for these cytokines in DIH. Three weeks after immunization, the hepatic IL-33 and ST2 levels remained elevated (both p < 0.05 vs. control; Fig. 1d). Additionally, hepatitis was apparent by histology (Fig. 1e), and the levels of IL-4 and IL-6 were elevated compared with those in control BALB/c mice (both p < 0.05 vs. control; Fig. 1f). We also observed that, in BALB/c mice with DIH, the hepatic IL-33 levels had increased from the 2-week levels (p < 0.05), whereas hepatic ST2 levels were five times lower than those at the 2-week time point (p < 0.0001). Thus, the upregulation of hepatic IL-33 and ST2 occurred before the development of DIH; however, IL-33 was further increased in conjunction with hepatitis, while hepatic ST2 was decreased, suggesting a possible uncoupling of IL-33 and ST2 in DIH.

Fig. 1.

IL-33 and ST2 levels are upregulated in the livers of BALB/c mice with DIH. BALB/c mice were injected with 100 µg of TFA-JHDN5 emulsified in an equal volume of CFA on days 0 and 7. Control mice received only CFA. On day 0, the mice were also immunized with 50 ng of pertussis toxin. Liver samples from each mouse were homogenized in RPMI/2% FCS as described in Methods section and then were analyzed for cytokine expression as per kit instructions and standardized by converting the levels to pg/g of tissue. Liver sections (5 µm thick) were fixed in 10% neutral buffered formalin and stained with H & E. After 2 weeks, the hepatic IL-33 and ST2 levels were increased in BALB/c mice immunized with TFA-JHDN5 (a, p < 0.001); representative H & E section demonstrating no difference in hepatitis at this time point (b, ×64 magnification). However, IL-4 (p < 0.001), IL-6 (p < 0.01), IL-7 (p < 0.05), and IL-17 (p < 0.05) levels were increased in TFA-JHDN5-immunized mice, while MIP-2 levels were reduced (p < 0.05) (c). In sharp contrast, after 3 weeks, hepatic IL-33 and ST2 levels remained increased in TFA-JHDN5-immunized mice (d, p < 0.05); representative H&E section demonstrating significant hepatitis in the form of granulocytic (blue) infiltration at this time point (e, ×64 magnification). After 3 weeks, hepatic IL-4 and IL-6 levels remain elevated (f, p < 0.05). Each experiment was repeated three times and consisted of five mice per group. *p < 0.05, **p < 0.01, ***p < 0.001

In the IL-33 blocking experiments, we found that anti-IL-33 reduced the hepatitis scores from 2.8 ± 1.1 to 1.6 ± 0.6 in BALB/c mice (p < 0.05; Fig. 2a, b). However, treatments caused the mice to appear ill, with a hunched posture, ruffled fur, and minimal spontaneous movements. In addition, the IL-33 levels did not decrease in the livers of these mice, regardless of whether treatments occurred at the initiation (splenic) or effector (hepatic) phase of the disease (Fig. 2e, f). Interestingly, blocking experiments increased splenic IL-33 during the initiation phase (Fig. 2c, d), suggesting that anti-IL-33 treatments might increase the morbidity in BALB/c mice, and our anti-IL-33 treatments did not induce complete blockade of IL-33.

Fig. 2.

IL-33 may play conflicting roles in BALB/c mice with DIH. BALB/c mice were untreated or injected with CFA ±100 µg of TFA-JHDN5 on days 0 and 7, in addition to 50 ng of pertussis toxin on day 0. TFA-JHDN5-treated mice also received 3.6 µg of anti-IL-33 (diluted in sterile filtered 1× PBS) by intraperitoneal injection on days 0, 7, 14, and 20 (Initiation Phase) or on days 14 and 20 (Effector Phase). All mice were killed after 3 weeks. Liver sections (5 µm thick) were fixed in 10% neutral buffered formalin and stained with H&E. The tissue sections were scored as described in Methods section. Liver samples were analyzed for cytokine expression as described in Methods section. Anti-IL-33 treatment started during the initiation phase (init.) reduced DIH measured using inflammation/injury scores in BALB/c mice (2.8 ± 1.1 vs. 1.6 ± 0.6, (p < 0.01). Anti-IL-33 treatments begun during the initiation phase (init.) reduced injury (1.6 ± 0.6) compared to treatments begun during the effector phase (eff.) (2.9 ± 1.0, p < 0.05). In addition, mice immunized with CFA + TFA-JHDN5 exhibited more inflammation/injury (2.8 ± 1.1) than CFA-immunized (1.3 ± 0.3, p < 0.01)) and unimmunized mice (0.1 ± 0.0 p < 0.001) (a, median ± SD). Representative H&E liver section (×64) demonstrating the absence of granulocytic (blue) infiltration in mice treated with anti-IL-33 during the initiation phase (b). Splenic IL-2 (p < 0.05) and IL-4 (p < 0.05) levels were increased following anti-IL-33 administration during the initiation phase (c). There were no differences in splenic cytokine expression following anti-IL-33 administration during the effector phase only on days 14 and 20 (d). In sharp contrast, the hepatic IL-17 (p < 0.05) and IL-33 (p < 0.01) levels were increased following anti-IL-33 administration during the initiation phase (e). However, the hepatic IL-2 and IL-4 levels were reduced following anti-IL-33 administration during the effector phase (f, p < 0.05). The experiments were run in duplicate with 2–3 mice per group. *p < 0.05, **p < 0.01

IL-33 offers protection in DIH by upregulating IL-10 as well as the number and function of FoxP3+ Treg cells

To help clarify the role of IL-33 in DIH, we immunized IL-33-deficient (IL-33−/−) mice. Surprisingly, we found no difference in the hepatitis scores between IL-33−/− and BALB/c mice (Fig. 3a). To determine whether there were differences in the cellular composition of hepatitis between BALB/c and IL-33−/− mice, we measured the number of CD4+ T cells and B cells in the hepatic infiltrate. We previously detected elevated CD4+ T cells in BALB/c mice in this model. In this study, we detected significantly more hepatic CD4+ T cells in BALB/c than in IL-33−/− mice after 2 weeks (p < 0.05); however, there were significantly fewer of these cells in BALB/c mice than in IL-33 −/− mice after 3 weeks (p < 0.01, Supplemental Figs. 1A and 1C), suggesting more CD4+ T-cell infiltration in IL-33−/− mouse livers. In sharp contrast, there were no significant differences in the levels of hepatic B cells between BALB/c and IL-33−/− mice at 2 or 3 weeks (Supplemental Figs. 1B and 1D). To determine whether there were differences in granulocytic infiltration, we measured the chemokine levels in the livers of BAL/c and IL-33−/− mice. We also detected no difference in KC between BALB/c and IL-33−/− mice at 2 weeks, suggesting that neutrophil signaling was similar at this time point²³ and supports our findings of no difference in the inflammation/injury scores at 2 weeks (Supplemental Fig. 1E). Interestingly, KC was higher in BALB/c than in IL-33−/− mice at 3 weeks (P < 0.001), likely supporting an earlier signaling role for this chemokine in IL-33−/− mice or suggesting that the hepatic infiltrate in BALB/c mice is more neutrophilic in BALB/c mice, while the infiltrate in IL-33−/− mice has more CD4+ T cells (Supplemental Fig. 1F). Additionally, at 2 and 3 weeks, BALB/c mice demonstrated increased MIG compared with IL-33−/− mice (p < 0.05), supporting Th1 inflammation, which has been shown to be protective in this model.²³ Moreover, the lack of MIG could support the increased morbidity in IL-33−/− mice. (Supplemental Figs. 1E and 1F). Despite evidence for similar levels of hepatitis, nearly 20% of IL-33−/− mice died before 4 weeks (p < 0.05, Fig. 3c), and we could not detect necropsy in these mice. Thus, decreased survival was observed when IL-33 was diminished, revealing a critical immune-regulatory role for this cytokine and the possibility of dual roles in the liver. In addition, we detected reduced levels of hepatic IL-10 (p < 0.05, Fig. 3d) and Treg-associated RNA (such as Foxp3), which is used to generate a Treg score (Fig. 3e), in the surviving mice, suggesting that, in DIH, IL-33 may decrease mortality and offer protection through the induction of Foxp3+ Tregs, similar to what has been reported in Con A hepatitis¹⁸ and autoimmune colitis.¹⁶

Fig. 3.

IL-33 upregulates IL-10, splenic FoxP3+ Tregs and survival in BALB/c mice with DIH. BALB/c and IL-33-deficient (IL-33−/−) mice were injected with 100 µg of TFA-JHDN5 emulsified in an equal volume of CFA on days 0 and 7, in addition to 50 ng of pertussis toxin on day 0. The mice were either killed at 3 weeks or followed for 5 weeks. After 3 weeks, the liver and spleen tissue cytokines were measured as described in Methods section, RNA copies were measured using Nanostring™, and CD45+ (PerCP), CD4+ (FITC), CD25 (PE), and FoxP3+ (APC) regulatory T cells (Tregs) were measured by flow cytometry. There was no difference in the levels of DIH by histology between BALB/c (3.0 ± 0.8) and IL-33−/− (3.0 ± 0.8) mice (median ± SD) (a). Representative histological sections demonstrating similar levels of inflammation/injury at 3 weeks between BALB/c and IL-33−/− mice (b, ×64 magnification). IL-33−/− mice demonstrated increased mortality by 3 weeks compared with BALB/c mice (c, p < 0.05). IL-33−/− mice (109.9 ± 49.6 pg/g) showed reduced levels of IL-10 in their livers compared with BALB/c mice (154.4 ± 74.3 pg/g) (d, median ± SD, p < 0.05). IL-33−/− mice demonstrated reduced Treg RNA copies that were utilized to generate a Treg score (6.5 ± 0.5) compared with BALB/c mice (7.4 ± 0.3) (e, p = 0.08). Absolute Treg numbers were lowered in the liver from IL-33−/− mice when compared to BALB/c mice (f, p < 0.05). Absolute Treg numbers were increased in the spleen from IL-33−/− mice (g, p < 0.05) and lowered in the liver from IL-33−/− mice when compared to BALB/c mice (g, p < 0.05). Experiments were run in duplicate with 2–3 mice per group *= p < 0.05

To determine whether IL-33 deficiency is associated with a reduction in FoxP3 expression by Tregs, we measured these cells at baseline, and at 2 and 3 weeks after immunization in both the liver and spleen because the splenic Treg population plays a key role in decreasing the severity of DIH.⁶ We found that hepatic Tregs were lowered in IL-33−/− mice at 2 weeks (p < 0.05). Additionally, splenic Tregs were increased in IL-33−/− mice at baseline (p < 0.05) but were reduced at 2 weeks (p < 0.05) compared with those in BALB/c mice (Fig. 3f, g). Thus, our findings suggest that IL-33 deficiency is associated with reduced Treg numbers and percentages at 2 weeks (Supplemental Fig. 2C) of induced Foxp3+ Tregs. We also found that most CD25+ cells in the liver and spleens of BALB/c but not in those of IL-33−/− mice are Foxp3+. Because CD25 is an activation marker, this observation could imply that there are more activated T cells in the spleens and liver of IL-33−/− mice.

To discover whether IL-33 deficiency was associated with reduced Treg function in DIH, we first compared CD4+ T-cell proliferation in IL-33−/− and BALB/c mice after direct T-cell receptor (TCR) stimulation. CD4+ T-cell proliferation has been associated with the initiation of DIH. Nevertheless, the proliferation of CD4+ T cells in response to direct TCR stimulation was similar in IL-33−/− and BALB/c mice (Fig. 4a, b), suggesting that IL-33 is not required for CD4+ T-cell proliferation. However, when we co-cultured CD4+ T cells from BALB/c or IL-33−/− mice with Tregs from IL-33−/− or BALB/c mice, we found that BALB/c Tregs but not IL-33−/− Tregs reduced the proliferation in both strains (p < 0.05, Fig. 4c–e). In fact, co-culture of IL-33−/− or BALB/c CD4+ T cells with Tregs from IL-33−/− mice increased the proliferation induced by direct TCR stimulation (p < 0.001, Fig. 4f). Our findings support our contention that IL-33 regulates the maturation and function of induced FoxP3+ Tregs and suggests that this role may be an important effect of IL-33 in DIH. However, whether the IL-33-ST2 interaction is the critical component of FoxP3+ Treg maturation and function in DIH remains unclear.

Fig. 4.

IL-33 upregulates splenic FoxP3+ Treg function but not CD4+ T-cell proliferation in BALB/c mice with DIH. CFSE-labeled splenocytes from BALB/c, IL-33−/−, or ST2−/− mice were stimulated with 0.5 µg of anti-mouse CD3e and 0.25 µg of anti-mouse CD28 for 24 h at 37 °C, stained with CD4-APC (1:100) for 30 min at 4 °C and then were analyzed by flow cytometry. Representative histogram of proliferating CFSE-positive CD4+ T cells from BALB/c and IL-33−/− mice (a). Summary of CFSE-positive CD4+ T-cell proliferation responses from BALB/c, IL-33−/− and ST2−/− mice demonstrating no differences in the proliferation responses among BALB/c, IL-33−/−, and ST2−/− mice. b Representative histograms of CD4+ T-cell proliferation in splenocytes from BALB/c and IL-33−/− mice following TCR stimulation (c and d, top panels), following TCR stimulation + Tregs from BALB/c mice (c and d, middle panels) and following TCR stimulation + Tregs from IL-33−/− mice (c and d, lower panels). The addition of Tregs from BALB/c mice reduced CD4+ T-cell proliferation in splenocyte cultures from BALB/c (p < 0.05) and IL-33−/− (p < 0.001) mice (e). The addition of Tregs from IL-33−/− mice increased CD4+ T-cell proliferation in splenocyte cultures from BALB/c (p < 0.001) and IL-33−/− mice (f). Experiments were run in duplicate with 2–3 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001

ST2−/− mice and BALB/c mice have similar levels of splenic Foxp3+ Tregs but reduced levels of DIH

Next, we examined whether IL-33-induced Treg maturation and function require ST2 in DIH. Essentially, all Foxp3+ Tregs in the liver and most in the spleen of IL-33−/− mice were ST2+ (Fig. 5a, b), suggesting that, in DIH, IL-33 does not induce the ST2 maturity marker on Foxp3+ Tregs. The levels of CD4+ T-cell proliferation in response to TCR stimulation did not differ significantly among ST2−/−, IL-33−/−, and BALB/c mice (Fig. 4b), suggesting that, analogous to IL-33, ST2 does not significantly contribute to the CD4+ T-cell proliferation that initiates DIH. We also found that, similar to prior studies in a model of colitis,¹⁶ there was no difference in the splenic Treg levels between ST2−/− and BALB/c mice with DIH (Fig. 5c). This finding could suggest that hepatitis in ST2−/− mice would be at the same level as in BALB/c mice, because lowered splenic and non-hepatic Tregs have been correlated with increased DIH in this model.⁶ In fact, ST2−/− mice exhibited significantly less DIH than BALB/c mice by histology (p < 0.05, Fig. 5e, f). Moreover, the expression levels of IL-5 and IL-17 were lower in the spleens of ST2−/− mice than in those of BALB/c mice (p < 0.05, Fig. 5g). These cytokines may have been downregulated by functioning as ST2-negative, Foxp3+ Tregs. In the liver, ST2−/− mice had fewer hepatic Tregs than BALB/c mice (p < 0.05, Fig. 5d), a finding that was also found in IL-33−/− mice. Analogous to IL-33−/− mice, IL-10 was reduced in the liver (p < 0.05, Fig. 5g), along with reduced Foxp3+ Tregs. When we directly assessed ST2−/− Treg function, we found that ST2−/− Tregs diminished CD4+ proliferation in BALB/c (p < 0.05) and ST2−/− mice (Fig. 5h); although, the reduction in CD4 + ST2−/− cultures did not reach statistical significance. Taken together, our findings suggest that, in DIH, FoxP3+ Treg function may not involve ST2; however, whether IL-33 plays a role in this function needs clarification.

Fig. 5.

In DIH, IL-33 does not induce the ST2 maturity marker in FoxP3+ Tregs. DIH was induced in BALB/c, IL-33−/−, and ST2−/− mice as described in Methods section. After 3 weeks, 74.7% of FoxP3+ Tregs in the spleens (a) and 91.2% of FoxP3+ Tregs in the livers (b) of IL-33−/− mice expressed the ST2 maturity marker, demonstrating that IL-33 did not induce this marker in FoxP3+ Tregs. There was no difference in the absolute numbers of splenic FoxP3+ Tregs between BALB/c and ST2−/− mice. However, splenic FoxP3+ Tregs were diminished in IL-33−/− mice (c, p < 0.05). The absolute numbers of hepatic FoxP3+ Tregs were significantly lower in IL-33−/− and ST2−/− mice than in BALB/c mice (d, p < 0.05). Inflammation/ injury scores were reduced in ST2−/− (1.9 ± 0.8) compared with BALB/c (2.9 ± 0.7) mice (e, median ± SD, p < 0.01). Representative H&E-stained liver sections (f, 5 µm thick, ×64 magnification) from BALB/c and ST2−/− mice showed reduced inflammation in the form clusters of granulocytes in ST2−/− mice. Consistent with reduced hepatitis, IL-5 and IL-17 levels were reduced in the spleen (g, p < 0.05). Consistent with reduced hepatic FoxP3+ Tregs, IL-10 levels were reduced in the liver (g, p < 0.05). The addition of Tregs from ST2−/− mice decreased CD4+ T-cell proliferation in splenocyte cultures from BALB/c (p < 0.001) and ST2−/− mice (h). Experiments were run in duplicate with 2–3 mice per group. *p < 0.05, **p < 0.01, ***p < 0.001

To clarify the roles of IL-33 in Foxp3+ Treg function in ST2−/− mice, we measured the IL-33 and ST2 levels in the livers and spleens of immunized ST2−/− mice and compared them with the levels found in IL-33−/− and BALB/c mice. We found no significant difference in the splenic IL-33 levels between ST2−/− mice and BALB/c mice, a finding that could support the similar numbers of Foxp3+ Tregs detected in ST2−/− and BALB/c spleens. In fact, IL-33 was slightly elevated in ST2−/− mouse spleens, which may have contributed to the reduction of DIH (Fig. 6a). IL-33 levels were significantly lower in the livers of ST2−/− mice than in those of BALB/c mice (Fig. 6b), a finding that again supports significantly reduced levels of Foxp3+ Tregs in this site. As expected, the levels of splenic and hepatic IL-33 were negligible in IL-33−/− mice (p < 0.001), although ST2 levels were higher than those in BALB/c mice (p < 0.001, Fig. 6a, b). These findings corroborate our detection of ST2+ Tregs in these mice (Fig. 5a, b). Thus, IL-33 in the spleen of ST2−/− mice may have supported the development of FoxP3+ Tregs seen in the spleens of these mice. Taken together, our findings suggest that ST2 may not be required for IL-33-promoted Treg levels or Treg function and supports a regulatory association between IL-33 and Tregs. To confirm that IL-10 by itself could not reconstitute Tregs in IL-33−/− mice, we supplemented splenocyte cultures stimulated with CD3e and CD28 with IL-10 (10 ng). We found that IL-10 did not reconstitute Treg numbers (Fig. 6c) and increased CD4+ T-cell proliferation in IL-33−/− mice.

Fig. 6.

IL-33 levels correlate with FoxP3 Tregs in ST2−/− mice with DIH. We induced DIH in BALB/c, IL-33−/− and ST2−/− mice and compared IL-33 and ST2 levels in these mice. Analogous to FoxP3+ Treg expression, BALB/c and ST2−/− mice with DIH expressed similar levels of IL-33 in their spleens and higher levels of IL-33 than IL-33−/− mice (a). In support of an IL-33-Treg connection, the hepatic IL-33 levels were lower in ST2−/− mice with DIH than in BALB/c mice with DIH (p < 0.05), but IL-33 levels were higher in ST2−/− mice than those in IL-33−/− mice (p < 0.05). IL-33 levels were expectedly higher in BALB/c than in IL-33−/− mice (p < 0.0001) (b). We confirmed lowered splenic and hepatic IL-33 and ST2 levels in IL-33−/− and ST2−/− mice with DIH, respectively (p < 0.001, a, b). Elevated ST2 expression was detected in the liver and spleen of IL-33−/− mice compared with that in BALB/c (p < 0001) and ST2−/− mice (p < 0.0001) (a, b). We also confirmed that, in IL-33−/− mice, supplemental IL-10 did not increase the Treg numbers (c) and increased CD4+ T-cell proliferation in these mice (d, p < 0.01), and e). Each experiment was repeated 2–3 times and consisted of two mice per group. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Uncoupled IL-33-IPEX responses predominate in patients with anesthetic DIH

IL-33 has been detected in patients with acute hepatic failure,⁷ and autoimmune hepatitis has been found in patients with IPEX deficiency;^24,25 however, IPEX has not been studied in anesthetic-induced DIH. We analyzed sera from patients who developed DIH after exposure to halogenated anesthetics and found elevated IL-33 levels compared with that in control sera (p < 0.001; Fig. 7a). Detecting elevated IL-33 levels in these patients could represent non-specific immune hyperactivation or a pathway to downregulate immune responses, as suggested in a prior study.⁷ IPEX levels were lower in patients with DIH than in control subjects (Fig. 7b, p < 0.01). Lower IPEX levels could represent lower levels of the transcriptional regulator critical for the development and inhibitory function of regulatory T cells in these patients. Therefore, we used logistic regression to determine whether a correlation existed between IPEX and IL-33. We found that IPEX was correlated with IL-33 in the controls (p < 0.05; Fig. 7c), but this correlation was lost in patients with DIH. This loss of correlation may contribute to the immune responses seen in DIH, as we also observed uncoupled IL-33 and Treg responses in BALB/c mice. Our studies suggest that the IL-33-IPEX (Foxp3) correlation may play a significant role in anesthetic DIH independent of ST2.

Fig. 7.

Anesthetic hepatitis is associated with dysregulated IL-33 and IPEX responses. IL-33 levels were higher in sera from patients with anesthetic hepatitis (2.6 ± 1.4 pg/ml) than in control sera samples 0.6 ± 0.5 pg/ml (a, p < 0.001). IPEX levels were lower in sera from patients with anesthetic hepatitis (82.5 ± 35.5 pg/ml) than in control (111.8 ± 42.8 pg/ml l) sera samples (b, p < 0.01). Logistic regression analysis demonstrated a correlation between IL-33 and IPEX in control sera samples but not in sera samples from patients with anesthetic hepatitis (c, p < 0.05). IPEX levels were log transformed because a Gaussian distribution was not demonstrated

Discussion

We have uncovered new and critical roles for IL-33 in DIH severity and mortality. In mice, we showed that IL-33 and ST2 are both upregulated during the induction phase of DIH and that blocking IL-33 in vivo diminishes hepatitis but increases morbidity and mortality. Mice genetically deficient in IL-33 developed DIH after immunization and had significantly reduced IL-10, revealing an unexpected role for IL-33 in the maturation and function of CD45+CD4+CD25+Foxp3+ Tregs. Specifically, mice deficient in IL-33 did not exhibit upregulation of Tregs during DIH induction, and Tregs obtained from these mice did not exhibit downregulation of CD4+ T-cell proliferation after TCR stimulation. In addition, the deletion of IL-33 increased DIH-related mortality. Our studies suggest that IL-33 may act without its known receptor ST2 because mice deficient in ST2 showed upregulation of Tregs and IL-33 during the induction of DIH and had reduced levels of DIH. Clinically, we found that IL-33 levels were correlated with IPEX levels in control subjects but not in patients with anesthetic DIH, confirming dysregulated IL-33-IPEX expression in DIH. Taken together, our studies suggest surprising roles for IL-33 in DIH that are independent of ST2 and may influence DIH severity and mortality.

IL-33 plays key roles in the maturation and function of diverse cell types. IL-33 is produced by barrier cells in response to various signals such as cellular stress, infection, allergens, or self-antigens.^26,27 By activating innate non-lymphoid leukocytes, such as mast cells, NKT cells, and dendritic cells, as well as innate lymphoid cells, IL-33 drives the expression of Th2 cytokines.^26,28,29 Tissue-derived, extracellular IL-33 targets innate immune cells that express the ST2 receptor. CD45-positive dendritic cells and macrophages also produce IL-33 but to a lesser degree than tissue-derived IL-33 sources. IL-33 promotes dendritic cell maturation via ST2 receptors, thereby upregulating their expression of IL-6, MHC II, and CD86.³⁰ Thus, IL-33 may play a role in the upregulation of inflammatory responses. In contrast, IL-33 also upregulates the maturation and function of induced Foxp3+ Tregs, and this upregulation is believed to occur through the upregulation of ST2 expression in these cells.^17–20 These studies highlight key roles for IL-33 in the maturation and function of diverse immune cells that may result in either proinflammatory or anti-inflammatory responses.

Our initial studies seemed to suggest proinflammatory roles for IL-33 in DIH. We found that IL-33 and its only known receptor, ST2, were upregulated during the initiation of DIH. This finding was surprising because we had previously thought the liver was relatively inactive during the initiation of DIH.²³ In the effector phase of experimental DIH, we first gained insight into the role of IL-33 in DIH. We detected rising IL-33 levels as the ST2 levels were significantly declined, demonstrating a negative regulatory response between IL-33 and its receptor. This latter finding could be explained by a negative feedback loop in which the effect of IL-33 could be reduced by downregulating ST2 expression.

To gain an understanding of the mechanistic role of IL-33 in DIH, we blocked IL-33 in vivo. Although anti-IL-33 reduced DIH, it also seemed to increase mortality in BALB/c mice. To understand these seemingly contradictory findings, we induced DIH in IL-33−/− mice and found that IL-33 may induce Foxp3+ Tregs, as was reported in Con A hepatitis. However, in sharp contrast to Con A hepatitis, we found that most Tregs in IL-33−/− mice expressed ST2, demonstrating that IL-33 does not induce ST2 expression in these Tregs in DIH. This latter finding was demonstrated in a prior study.¹⁹ Moreover, the disconnect between IL-33 and ST2 may suggest alternative pathways or receptors for IL-33 effects on Tregs.

In ST2−/− mice, which exhibited reduced levels of hepatitis, we found evidence for functioning Foxp3+ Tregs, supporting a possible ST2-independent mechanism for IL-33 induction of Treg maturation and function. ST2−/− spleens demonstrated higher levels of IL-33 than BALB/c mice, but this was not statistically significant (Fig. 6). Unsurprisingly, the splenic Treg levels in ST2−/− mice were comparable to those of BALB/c mice. Reduced levels of proinflammatory cytokines in the spleen could provide evidence of functioning Tregs. Additionally, because we showed no differences in CD4+ T-cell proliferation among any of the genotypes, our findings suggest that CD4+ T cells are functional and capable of initiating DIH, and Tregs detected in ST2−/− mouse spleens may have had greater suppressive function than those from BALB/c mice. The observed reduction in hepatic IL-10 and reduced IL-33 in ST2−/− mice support our finding that hepatic Tregs were reduced in ST2−/− similar to that in IL-33−/− mice; however, our prior studies showed that splenic and not hepatic levels of Tregs correlate with DIH.⁶ Thus, ST2−/− splenic Tregs are exposed to splenic IL-33, and these splenic Tregs access the liver through the bloodstream.

Our results indicate that IL-33 may not play a major role in the proinflammatory responses that initiate DIH. The induction of CD4+ T-cell proliferation by direct TCR stimulation did not differ in mice lacking IL-33 or ST2 compared with that in BALB/c mice. This finding is critical because CD4+ T cells initiate DIH. IL-4, which initiates DIH, did not differ among the mouse genotypes. Thus, our findings support a prior study that suggested that elevated IL-33 detected in patients with liver failure could represent a pathway to downregulate immune responses.⁷ This pathway may involve an IL-33-Treg maturation process that does not require ST2.

IL-10 was consistently lower in the livers of IL-33−/− and ST2−/− mice with DIH than in the BALB/c mice with DIH but did not differ in the spleens of the three genotypes. We propose that hepatitis was reduced in the ST2−/− mice because splenic IL-33 promotes Foxp3+ Treg function. Even so, we do not know the sources of IL-10 in our mice. The Foxp3+ Tregs may be a source of IL-10, as has been previously described.³¹ Alternatively, IL-33 might also upregulate dendritic cells and macrophages capable of producing IL-10.³¹ IL-33 might also induce regulatory B-cells capable of producing IL-10.³² We are currently investigating the source of IL-10 in all our strains after the induction of DIH.

We found a correlation between IL-33 and IPEX in control subjects that was lost in patients with anesthetic DIH, although IL-33 was significantly elevated in these patients. This finding in humans is supported by our finding in mice that IL-33 is required to reduce mortality from DIH and is in agreement with an earlier study showing that IL-33 was protective in acute liver failure.⁷ Even so, we must exercise caution because IL-33 also possesses proinflammatory effects, at least in experimental mice. Additionally, although ST2−/− mice have less hepatitis at 3 weeks, their hepatitis may rebound at a later time point because of lowered Tregs in the liver. Studies are ongoing to follow these mice long term.

Regardless of our findings and those of others, the roles of cytokines in the initiation of DIH remain controversial.³³ Cytokines may be responsible for the initiation of halothane-induced liver injury in knockout mice, but off-target effects, such as differences in diet, infection, and housing, should also be considered.³⁴ We suggest that IL-33 may address some mechanisms responsible for the severity of DIH. We also suggest that IL-33 promotes survival in DIH. Our future studies will further elucidate the exact mechanisms responsible for IL-33-induced Treg maturation and, more importantly, mechanisms responsible for IL-33-promoted survival from DIH.

Conflict of interest

Ms. Cottagiri declares no potential conflict of interest. Ms Nyandjo declares no potential conflict of interest. Mr. Stephens declares no potential conflict of interest. Mr. Mantilla declares no potential conflict of interst. Dr. Saito declares no potential conlict of interest. Dr. Mackay delarees no potential conflict of interest. Dr. Rose declares no potential conflict of interest. Dr. Njoku owns a patent on the epitope utilized in this manuscript. There are no financial relationships or compensations associated with this patent.

Electronic supplementary material

Supplementary Information The online version of this article (10.1038/s41423-018-0087-y) contains supplementary material.