Areg ^-/- mice (Areg^tm1Dle/J) were provided by Matias A. Avila (Hepatology Program, CIMA, Universidad de Navarra, Pamplona, Spain). Il1rl1 ^-/- mice (Il1rl1^tm1Anjm) were provided by Max Löhning (Experimental Immunology and Osteoarthritis Research, Department of Rheumatology and Clinical Immunology, Charité-Universitätsmedizin Berlin, Berlin, Germany). Foxp3^Cre x ST2^fl/fl mice (B6.Il1rl1^{tm1c(KOMP)Wtsi}xFoxp3^tm1(cre)Saka) (21) were provided by Astrid M. Westendorf (Infection Immunology, Institute of Medical Microbiology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany). C57BL/6 mice, FIR-tiger (Foxp3-IRES-mRFP [FIR]xIL-10-IRES-GFP [tiger]) (22) and all other transgenic mice were bred in the animal facility of the University Medical Center Hamburg-Eppendorf (Hamburg, Germany) according to the Federation of European Laboratory Animal Science Association guidelines. All mice received human care according to the criteria outlined in the “Guide for the Care and Use of Laboratory Animals”. The experiments were approved by the institutional review board (Behörde für Justiz und Verbraucherschutz, Hamburg, Germany) and carried out according to the German animal protection law. Mice were housed in IVC cages under controlled conditions (22°C, 55% humidity, and 12-hour day-night rhythm) and fed a standard laboratory chow (LASvendi, Altromin, Germany). We have used male mice for the in vivo and a mixture of male and female mice for the in vitro studies.

Mice intraperitoneally received recombinant murine (rm)IL-33 (0.3 µg/mouse; BioLegend, San Diego, CA) on three or four consecutive days. To induce immune-mediated hepatitis, ConA (7 mg/kg; Sigma-Aldrich, München, Germany) was intravenously injected and mice were analyzed 24 hours later. For adoptive transfer experiments, FIR-tiger mice were treated with rmIL-33 (0.3 µg/mouse) or PBS on three consecutive days. Thereafter, Foxp3⁺ (mRFP⁺) CD4⁺ Tregs were isolated by FACS. 3x10⁵ Tregs were transferred into C57BL/6 recipient mice, which received ConA (7 mg/kg) 24 hours later. Mice were analyzed one day after induction of hepatitis. Heart blood was drawn from individual mice and liver injury was quantified by automated measurement of plasma activities of alanine transaminase (ALT) using a COBAS Mira System (Roche Diagnostic, Mannheim, Germany).

Liver samples were fixed with 4% formalin and embedded in paraffin. 3 μm liver sections were cut, stained with hematoxylin and eosin (H&E) following standard protocols, and analyzed by light microscopy.

To expand hepatic ILC2s for isolation, C57BL/6 mice were treated with rmIL-33 (0.3 µg/mouse) on four consecutive days. Hepatic non-parenchymal cells were isolated by Percoll density gradient centrifugation. Single non-parenchymal cells were stained with lineage antibodies (anti-CD3, 145-2C11; anti-CD11b, M1/70; anti-CD45R/B220, RA3-6B2; anti-TER-119, TER-119; anti-GR1, RB6-8C5; all APC; all BioLegend), anti-Sca-1 (D7, Pacific Blue) and anti-ST2 (RMST2-2, PE-Cy7; both BioLegend) antibodies. Lineage-negative (lin^-) cells were enriched by MACS and ST2⁺ Sca-1⁺ lin^- ILC2s were isolated by FACS (BD FACSAriaTM III sorter; BD Biosciences, Heidelberg, Germany) (23).

CD4⁺ T cells were isolated from spleen and lymph nodes using the CD4⁺ T Cell Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions. For Treg isolation, CD4⁺ T cells were stained with an anti-CD25-PE antibody (PC61, BioLegend) and anti-PE MicroBeads (Miltenyi Biotec), and CD25⁺ CD4⁺ Tregs were isolated by MACS.

2x10⁴ CD25⁺ CD4⁺ Tregs were cultured with rmIL-2 (100 U/ml; R&D Systems, Minneapolis, MN) in presence of rmIL-33 (10 ng/ml; BioLegend), rmAREG (10 ng/ml; R&D Systems) or both (10 ng/ml each) for 1.5 days. 1.5x10⁴ hepatic ILC2s were cultured with rmIL-2 (1 U/ml) in presence of rmIL-33 (10 ng/ml), rmAREG (10 ng/ml) or both (10 ng/ml each) for 3.5 days.

CD4⁺ T cells were isolated from spleen and lymph nodes of C57BL/6 mice and stained with an anti-CD25-PE antibody and anti-PE MicroBeads. CD25^- CD4⁺ responder T cells were isolated by MACS and labeled with carboxyfluorescein succinimidyl ester (CFSE; CFSE Cell Division Tracker Kit, Biolegend) to assess T-cell proliferation. CD25⁺ CD4⁺ Tregs were isolated from spleen and lymph nodes of C57BL/6 and Areg ^-/- mice by MACS. 1x10⁵ responder CD4⁺ T cells were cultured with 1x10⁵ Tregs for 2.5 days. For responder CD4⁺ T-cell activation, co-cultures were done with Dynabeads (Dynabeads Mouse T-Activator CD3/CD28; ThermoFisher Scientific).

The percentage of inhibition was calculated as ( h i g h e s t   v a l u e   % p r o l i f e r a t e d   c e l l s − x ) h i g h e s t   v a l u e   % p r o l i f e r a t e d   c e l l s × 100 .

Cells were incubated with anti-CD16/32 antibody (93; BioLegend) prior to antibody staining in order to prevent unspecific binding. LIVE/DEAD Fixable Staining Kits (Thermo Fisher Scientific) were used to exclude dead cells. For cell surface analysis, cells were stained with fluorochrome-labelled antibodies listed in Supplementary Table S1 . For intracellular and intranuclear staining, cells were re-stimulated with phorbol myristate acetate (20 ng/ml) and ionomycin (1 µg/ml) for 4 hours with the addition of brefeldin A (1 µg/ml; all Sigma Aldrich) and monensin (2 µM; BioLegend) after 30 min. Thereafter, cells were fixed using the Transcription Factor Staining Buffer Set (ThermoFisher Scientific, Waltham, MA) and incubated in Permeabilization buffer with antibodies listed in Supplementary Table S2 . Data were acquired using a BD LSRFortessa II (BD Biosciences) and analyzed by FlowJo software (Tree Star, Ashland, OR, USA).

Plasma AREG levels were analyzed using the Mouse Amphiregulin ELISA DuoSets (R&D Systems) and the Infinite M200 plate reader (Tecan, Männedorf, Switzerland) according to the manufacturer’s instructions.

Total RNA was isolated from shock-frozen liver tissue using the NucleoSpin RNA Midi Kit (Macherey-Nagel, Düren, Germany). Genomic DNA was digested using the DNA-free Kit (ThermoFisher Scientific). 1 µg RNA was transcribed into cDNA using the Verso cDNA Synthesis Kit (ThermoFisher Scientific) and the Biometra TAdvanced thermal cycler (Analytik Jena, Jena, Germany). Quantitative RT-PCR was performed using the PowerUp SYBR Green Master Mix (ThermoFisher Scientific) and the QuantStudio 7 Flex (Applied Biosystems, Waltham, USA). Relative mRNA levels were calculated using the ΔΔCT method after normalization to the reference gene β-actin. Quantification was shown in x-fold changes to the corresponding control cDNA. Primers were designed for detection of exon overlapping amplicons and were obtained from Metabion (Martinsried, Germany). Sequences of the primers are listed in Supplementary Table S3 .

Data were analyzed using the GraphPad Prism software (GraphPad software, San Diego, CA). For comparisons between two groups, a non-parametric Mann-Whitney U test and for more than two groups, a one-way ANOVA with Tukey’s post-hoc test were used. Data were shown as medians. A p value of less than 0.05 was considered statistically significant with the following ranges *p< 0.05, **p< 0.01, ***p< 0.001, and ****p< 0.0001.

We analyzed the phenotype of hepatic ST2⁺ Foxp3⁺ Tregs in immune-mediated hepatitis to assess their impact on liver disease pathology. ST2⁺ Tregs have been described as a tissue-associated Treg subset crucial for immunosuppression and organ homeostasis (10, 24). In line with this, we determined an elevated frequency of ST2⁺ Foxp3⁺ Tregs in naïve liver compared to spleen ( Figure 1A , Supplementary Figure 1A ). Hepatic ST2⁺ Foxp3⁺ Tregs showed enhanced expression of the tissue residency markers CD103 and CD69 compared to ST2^- Foxp3⁺ Tregs, while they expressed less CCR7, a chemokine receptor required for lymph node migration ( Figure 1B , Supplementary Figure 1B ). Phenotypically, hepatic ST2⁺ Foxp3⁺ Tregs showed stronger expression of their key transcription factor Foxp3, the activation marker killer cell lectin like receptor G1 (KLRG1) and inducible T cell co-stimulator (ICOS), of the proliferation marker Ki-67, and of the inhibitory molecules cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), IL-10, programmed death ligand 1 (PD-L1) and T cell immunoreceptor with Ig and ITIM domains (TIGIT) than ST2^- Foxp3⁺ Tregs in liver homeostasis ( Figures 1C , Supplementary Figure 1B ). In immune-mediated hepatitis, liver injury and formation of necrotic lesions ( Figure 1D ) were associated with an elevated frequency of hepatic ST2⁺ Foxp3⁺ Tregs ( Figure 1E ). Expression of Foxp3 was not further increased in ST2⁺ and ST2^- Foxp3⁺ Tregs in liver inflammation ( Supplementary Figure 1C ). Both, hepatic ST2⁺ and ST2^- Tregs were activated in immune-mediated hepatitis and enhanced expression of inhibitory molecules. In fact, by calculating the fold change in frequencies, we found stronger activation of ST2^- Tregs compared to ST2⁺ Tregs in immune-mediated hepatitis (KLRG1: ST2^- 3.4 ± 1.0, ST2⁺ 1.2 ± 0.1; ICOS: ST2^- 4.9 ± 1.2, ST2⁺ 2.3 ± 0.2). Expression of inhibitory molecules was also stronger induced in ST2^- Tregs (CTLA-4: ST2^- 1.7 ± 0.2, ST2⁺ 1.3 ± 0.1; IL-10: ST2^- 2.5 ± 0.9, ST2⁺ 1.1 ± 0.5; PD-L1: ST2^- 9.7 ± 8.8, ST2⁺ 3.6 ± 0.7; TIGIT: ST2^- 2.3 ± 0.8, ST2⁺ 1.5 ± 0.1). We believe that due to the elevated basal activation and inhibitory molecule expression of ST2⁺ Tregs, they do not become activated as strong as ST2^- Tregs in immune-mediated hepatitis. However, activation and inhibitory molecule expression were still elevated in ST2⁺ compared to ST2^- Tregs ( Figure 1F ). Thus, ST2⁺ Foxp3⁺ Tregs constituted an activated effector Treg subset in the murine liver.

IL-33 is the only known ligand of ST2 and has been described as a potent activator of Tregs (4). We demonstrated that treatment with exogenous IL-33, which did not induce liver injury and tissue damage ( Figure 2A ), selectively increased the frequency of hepatic ST2⁺ Foxp3⁺ Tregs in C57BL/6 wild-type (WT) mice and their activation and expression of inhibitory molecules. ST2⁺ Foxp3⁺ Tregs did not enhance expression of Foxp3 in response to exogenous IL-33. We determined elevated Foxp3 expression by Foxp3⁺ Tregs from IL-33-treated WT compared to Il1rl1 ^-/- mice, probably due to lack of ST2⁺ Tregs in Il1rl1 ^-/- mice ( Figures 2B, C , Supplementary Figure 2A ). Moreover, Foxp3⁺ Tregs of IL-33-treated WT mice were stronger activated and showed enhanced expression of inhibitory molecules than Foxp3⁺ Tregs from IL-33-treated Il1rl1 ^-/- mice ( Figure 2D ). Thus, targeting the IL-33/ST2 axis by application of exogenous IL-33 selectively induced ST2⁺ Treg activation and expansion in the liver. In addition, culture of WT Tregs in presence of IL-33 enhanced their expansion, activation and expression of inhibitory molecules, whereas Tregs from Il1rl1 ^-/- mice were not stimulated by IL-33 in vitro. Similar to the in vivo data, IL-33-stimulated Il1rl1 ^-/- Foxp3⁺ Tregs showed decreased Foxp3 expression compared to WT Foxp3⁺ Tregs in vitro ( Supplementary Figures 2B, C ). We further showed that ST2⁺ but not ST2^- Foxp3⁺ Tregs from WT mice were activated by IL-33 in vitro ( Supplementary Figure 2D ).

We and others have shown that ST2-deficient (Il1rl1 ^-/-) mice developed more severe liver injury and tissue damage than WT mice ( Figure 2E ) (20), suggesting an immunoregulatory role of endogenous IL-33 in immune-mediated hepatitis. Liver pathology was aggravated in Il1rl1 ^-/- mice despite the absence of hepatic ILC2s ( Supplementary Figure 3 ) and an elevated frequency of Foxp3⁺ Tregs in the inflamed liver ( Figure 2F ). However, Il1rl1 ^-/- mice lacked ST2⁺ Foxp3⁺ Tregs and the remaining Foxp3⁺ Treg population showed reduced Foxp3 expression, activation and expression of inhibitory molecules in immune-mediated hepatitis compared to Foxp3⁺ Tregs in WT mice ( Figure 2F ). Thus, endogenous IL-33 selectively activated hepatic ST2⁺ Foxp3⁺ Tregs and lack of this Treg subset in Il1rl1 ^-/- mice resulted in an overall reduced Treg activation in immune-mediated hepatitis.

To assess the immunosuppressive capacity of Tregs activated by exogenous IL-33, we isolated Foxp3⁺ Tregs from IL-33-treated FIR-tiger mice and adoptively transferred them into WT mice one day before induction of immune-mediated hepatitis. As control, naïve Foxp3⁺ Tregs from PBS-treated FIR-tiger mice were transferred. We showed reduced liver injury and formation of necrotic lesions after transfer of IL-33-activated Foxp3⁺ Tregs ( Figures 3A, B ). This was associated with decreased expression of the pro-inflammatory cytokines interferon (IFN)γ and tumor necrosis factor (TNF)α by hepatic CD8⁺ and CD4⁺ T cells ( Figure 3C ). In contrast, transfer of naïve Foxp3⁺ Tregs did not alter disease pathology and immune cell phenotype ( Figures 3A-C ), demonstrating the high immunosuppressive potential of Tregs that were pre-activated through exogenous IL-33 in immune-mediated hepatitis.

IL-33 has been described to induce AREG expression in Tregs and ILC2s thereby supporting Treg function and tissue repair (6, 12). Indeed, we showed increased AREG expression in hepatic ST2⁺ Foxp3⁺ Tregs and ILC2s upon treatment with exogenous IL-33 ( Supplementary Figure 4A ). IL-33 also enhanced expression of AREG in ST2⁺ Foxp3⁺ Tregs ( Supplementary Figure 4B ) and hepatic ILC2s ( Supplementary Figures 5A, B ) in vitro.

In immune-mediated hepatitis, liver Areg mRNA and serum levels of AREG were elevated ( Figure 4A ), and we determined enhanced expression of AREG in hepatic ST2⁺ Foxp3⁺ Tregs and ILC2s ( Figure 4B ). In Il1rl1 ^-/- mice, Foxp3⁺ Tregs showed reduced expression of AREG compared to Foxp3⁺ Tregs from WT mice, suggesting that endogenous IL-33 induced AREG expression in ST2⁺ Tregs in liver inflammation ( Supplementary Figure 4C ). In Areg ^-/- mice, ST2⁺ Foxp3⁺ Tregs and ILC2s did not express AREG ( Supplementary Figure 4C ). Interestingly, Areg ^-/- mice developed more severe liver injury and tissue damage than WT mice ( Figure 4C ). Hepatic Il33 mRNA expression was not altered in Areg ^-/- compared to WT mice in immune-mediated hepatitis ( Supplementary Figure 4C ), indicating that the elevated liver injury shown in Areg ^-/- mice did not correlate with increased Il33 mRNA induction. Lack of AREG resulted in a reduced frequency of hepatic ST2⁺ Foxp3⁺ Tregs in immune-mediated hepatitis ( Figure 4D ), whereas their phenotype was not altered compared to ST2⁺ Foxp3⁺ Tregs from WT mice. Notably, Areg ^-/- ST2⁺ Foxp3⁺ Tregs up-regulated expression of Foxp3 and the anti-apoptotic protein B cell lymphoma 2 (Bcl-2) during liver inflammation ( Supplementary Figure 4D ), probably to increase their stability in absence of AREG. Functionally, we showed an impaired capacity of Areg ^-/- CD25⁺ Tregs to inhibit activation and subsequent proliferation of CD25^- CD4⁺ T cells in vitro ( Figure 4E ).

The frequency of hepatic ST2⁺ Foxp3⁺ Tregs was also reduced in Areg ^-/- mice treated with exogenous IL-33. Again, their phenotype was comparable to ST2⁺ Foxp3⁺ Tregs from IL-33-treated WT mice, but they showed elevated expression of Foxp3 as well as Ki-67 ( Supplementary Figure 4E ), indicating stronger proliferation. In vitro, IL-33 induced a lower expansion of ST2⁺ Foxp3⁺ Tregs from Areg ^-/- mice, and the IL-33-induced expression of inhibitory molecules was reduced in Areg ^-/- compared to WT ST2⁺ Foxp3⁺ Tregs ( Supplementary Figure 4F ).

We further demonstrated similar frequencies of hepatic ILC2s in Areg ^-/- and WT mice in immune-mediated hepatitis. However, lack of AREG resulted in an enhanced ILC2 activation and IL-13 expression ( Figure 4F ). In contrast, activation and type 2 cytokine expression of ILC2s were not altered in Areg ^-/- compared to WT mice upon treatment with exogenous IL-33 ( Supplementary Figure 5C ). Thus, AREG exerted an immunoregulatory function in immune-mediated hepatitis.

To further assess the effect of AREG on immune cell phenotype, we cultured Tregs and hepatic ILC2s in presence of AREG. Exogenous AREG induced an activation and expansion of ST2⁺ Foxp3⁺ Tregs. They increased expression of inhibitory molecules, whereas Bcl-2 expression was not altered. Moreover, AREG stimulated its own expression in ST2⁺ Foxp3⁺ Tregs. In contrast, ST2^- Foxp3⁺ Tregs were not activated by AREG ( Figure 5A ). We did also not observe an effect of AREG on activation of Foxp3⁺ Tregs from Il1rl1 ^-/- mice ( Supplementary Figure 6A ), demonstrating that exogenous AREG selectively activated ST2⁺ Foxp3⁺ Tregs.

Since AREG is a ligand of the epidermal growth factor receptor (EGFR), we analyzed EGFR expression by Tregs from WT and Il1rl1 ^-/- mice. The frequency of EGFR⁺ Foxp3⁺ Tregs was not altered in Il1rl1 ^-/- compared to WT mice. ST2⁺ and ST2^- Foxp3⁺ Tregs from WT mice did also not differ in their EGFR expression ( Figure 5B ), indicating an EGFR-independent effect of AREG on ST2⁺ Foxp3⁺ Tregs in vitro. We also showed that exogenous AREG did not induce an expansion of ST2⁺ Foxp3⁺ Tregs from Areg ^-/- mice. With the exception of TIGIT, AREG-deficient ST2⁺ Foxp3⁺ Tregs did not up-regulate expression of inhibitory molecules or Bcl-2 in response to AREG ( Figure 5C ).

Culture of hepatic ILC2s in presence of AREG resulted in a reduced ILC2 number and decreased their activation and IL-13 expression ( Figure 5D ). Moreover, exogenous AREG reduced the IL-33-induced activation of ILC2s and their expression of IL-13 and AREG ( Figure 5E ). In contrast, AREG did not affect the activation of WT or Areg ^-/- ST2⁺ Foxp3⁺ Tregs by IL-33 in vitro ( Supplementary Figures 6B, C ). Thus, exogenous AREG differentially regulated the activation of ST2⁺ Foxp3⁺ Tregs and hepatic ILC2s in vitro.

Previously, we have shown that treatment of WT mice with exogenous IL-33 before ConA challenge potently suppressed immune-mediated hepatitis (17). In contrast, liver injury and tissue damage were not improved in Il1rl1 ^-/- mice ( Figure 6A ), demonstrating that targeting the IL-33/ST2 axis by exogenous IL-33 ameliorated disease pathology. In WT mice, we observed reduced frequencies of hepatic CD4⁺ and CD8⁺ T cells expressing IFNγ and TNFα in immune-mediated hepatitis after pre-treatment with exogenous IL-33 ( Figure 6B ). The frequencies of hepatic ST2⁺ Foxp3⁺ Tregs and ILC2s were increased and both up-regulated expression of AREG ( Figure 6C ). Moreover, Foxp3 expression and activation of ST2⁺ Foxp3⁺ Tregs but not their expression of inhibitory molecules was further enhanced by treatment with exogenous IL-33 ( Supplementary Figure 7 ).

We analyzed whether exogenous IL-33 mediates its immunosuppressive effect via induction of AREG. Therefore, we treated Areg ^-/- mice with IL-33 before ConA challenge. However, ALT levels were not significantly altered in Areg ^-/- compared to WT mice ( Figure 6D ). To assess whether activation of ST2⁺ Tregs by exogenous IL-33 is critical for the immunosuppressive effect of this cytokine on liver disease pathology, we treated Foxp3^Cre x ST2^fl/fl mice (21) with IL-33 before induction of hepatitis. Foxp3^Cre+ x ST2^fl/fl mice developed more severe liver injury and tissue damage than Foxp3^Cre- x ST2^fl/fl mice after treatment with exogenous IL-33 ( Figure 6E ). This correlated with a strongly reduced frequency of hepatic ST2⁺ Tregs in Foxp3^Cre+ x ST2^fl/fl mice compared to Foxp3^Cre- x ST2^fl/fl mice ( Figure 6F ). Moreover, we showed decreased Foxp3 expression, activation, proliferation and inhibitory molecule expression by hepatic Foxp3⁺ Tregs as well as elevated expression of TNFα and IFNγ by hepatic CD4⁺ and CD8⁺ T cells in Foxp3^Cre+ x ST2^fl/fl mice in immune-mediated hepatitis after pre-treatment with IL-33 ( Figure 6F ). Thus, the immunosuppressive effect of exogenous IL-33 depends on ST2⁺ Treg activation.