Plasticity Between ILC2 Subsets and Amphiregulin Expression Regulate Epithelial Repair in Biliary Atresia

Mice

All animal studies were performed in strict accordance with the National Institutes of Health guidelines for use and care of live animals and were approved by the Institutional Animal Care and Use Committees of Cincinnati Children’s Hospital Medical Center. Wild-type (WT) BALB/c mice were purchased from Charles River Laboratories and Il4ra^−/− and Stat6^−/− mice (on a BALB/c background) were obtained from The Jackson Laboratory. The Rag2^−/−gc^−/− mice (on a C57BL/6 background) were purchased from Taconic Laboratories, and Il13^−/− mice on a BALB/c background²² were bred and backcrossed to C57BL/6 background to the 6^th generation in our facilities (Cincinnati Children’s Hospital Medical Center). The experimental animals were cohoused, acclimated, and maintained in microisolator cages in a pathogen-free facility equipped with a 12-hour dark light cycle and temperature and humidity maintained between 65–75°F and 40–60%, respectively. Mice had free access to water and sterilized chow (Purina LabChows Rodent Laboratory Chow 5010; Protein: 23.0%, Fat: 4.5%, Fiber: 6.0%) and were monitored daily by veterinary and study staff to assure humane conditions. For in vivo experiments, IL-33 was injected intraperitoneally to mice daily with specified dose and timing. The number of mice in each experiment is presented in the figure legends.

Antibodies and reagents

The following fluorochrome-conjugated anti-mouse antibodies were obtained from Biolegend: Lineage cocktail (anti-CD3 (17A2), CD4 (RM4–5), B220 (RA3–6B2), CD49b (DX5), CD11b (M1/70), CD11c (N418), F4/80 (BM8), Gr-1 (RB6–8C5), FcεR1α (MAR-1), Ter119 (TER-119)), ST2 (DIH4), KLRG1 (2F1/KLRG1), CD44 (IM7), CD62L (MEL-14), IL-4 (11B11), and IL-17 (TC11–18H10.1). Anti-mouse IL-25R (170220) was purchased from R&D Systems. Anti-mouse IL-5 (T21) and IFNg (XMG1.2) were purchased from BD Biosciences. Anti-mouse IL-13 (eBio13A) was purchased from eBioscience. Recombinant mouse AREG (262-AR) and anti-mouse IL-2 antibody (MAB-702) were from R&D Systems. Recombinant mouse IL-33 (0.1 or 1 μg injected intraperitoneally) was from BioLegend. Initial experiments were conducted with 1 μg; however, similar cholangiocyte proliferation was achieved with as little as 0.1 μg administered to adult mice.³ Anti-AREG antibody (200 μg/ mouse) was purchased from R&D Systems. Biotin Anti-BrdU antibody was purchased from Abcam for immunohistochemistry staining. The Rabbit IgG Vectastain ABC Kit (Vector Laboratories) was used for blocking, antibody incubations, and signal amplification according to the manufacturer’s instructions. Purified anti-CK19 antibody was from Abcam and anti-PanCK antibody was purchased from Dako. DyLight 488–conjugated donkey anti-goat and DyLight 594–conjugated streptavidin antibodies were obtained from Jackson Immunoresearch. ProLong^™ Gold antifade mountant (ThermoFisher Scientific) with DAPI was used for nuclear staining.

Non-parenchymal mononuclear cell isolation from liver and bile duct

Hepatic mononuclear cells were obtained by subjecting the livers to dissociation using gentleMACS Dissociator (Miltenyi Biotec) according to the manufacturer-specified and previously described protocols.^3,22 The resulting cell suspension was strained through a 40-μm cell filter to obtain single-cell suspensions and then subjected to density centrifugation using 33% Percoll in DMEM (ThermoFisher Scientific) at 836 g, 20°C for 25 minutes. The cell pellets were then washed twice in complete DMEM containing 10% FBS (Corning) followed by red cell lysis using RBC Lysis Buffer (Millipore Sigma) and used in flow cytometry or adoptive transfer experiments. To isolate immune cells from extrahepatic bile ducts, tissues were dissected, disrupted into small pieces and digested with Collagenase IV (Sigma) at 37 °C for 40–60 min. Tissue pieces were strained into single cells, and the leukocytes were purified by centrifugation using 40% Percoll (Sigma) in DMEM cell culture media, followed by RBC Lysis Buffer (Sigma-Aldrich) treatment for flow cytometry analysis.

Lineage negative cell purification and culture

Lin⁻ cell subsets were isolated from livers using the mouse Lineage Cell Depletion Kit (Miltenyi Biotec) according to the manufacturer’s instructions.

Flow cytometry analysis

Intracellular staining and flow cytometry analysis of immune cells from the hepatobiliary tissues was performed according to previously described protocols.³ Cellular phenotyping was performed using cell surface and intracellular markers as described in Antibodies and Reagents. Mononuclear cells were resuspended at a concentration of 1×10⁶-1× 0⁷ cells/ml in FACS buffer (PBS containing 0.1% sodium azide and 2% FBS) and added to a 96-well microtiter plate (Corning) at a volume of 200 μl/well. Cells were then preincubated with 1:200 dilution of anti-mouse FcγII/III receptor mAb CD16/CD32 (2.4G2) for 15 minutes at 4°C to block nonspecific antibody binding to Fc receptors. Surface staining for specific markers was performed by incubating cells with fluorochrome-conjugated antibodies at 4°C for 30 minutes. Intracellular cytokine staining was performed using the Cytofix/Cytoperm (with GolgiPlug) Fixation/Permeabilization kit (BD Biosciences) according to the manufacturer’s instructions. Cells were washed twice with staining buffer, incubated in 100 μl of BD Cytofix/Cytoperm solution for 20 minutes at 4°C and then stained using antigen-specific antibodies. The fluorescent signals were detected using either the FACSCanto II or LSRFortessa flow cytometer (BD Biosciences). Flow cytometric data was analyzed using FlowJo software (Tree Star Inc.).

Histology and immunohistochemistry staining

Livers and extrahepatic bile ducts were harvested from experimental mice and subjected to fixation in 10% formalin for 24–48 hours. After fixation, tissues embedded in paraffin were cut in longitudinal sections and subjected to hematoxylin and eosin (H&E) staining for histopathologic analysis. Immunohistochemistry and immunofluorescence staining was performed using biotinylated anti-BrdU (Abcam), anti-mouse Amphiregulin (AF989; R&D Biosciences), and anti-Cytokeratin-19 (CK19) antibodies (M-17; Santa Cruz Biotechnology). For BrdU epitope staining, citric acid buffer (pH=6) was used for antigen retrieval. In sections undergoing immunohistochemical staining, the color was developed using the DAB peroxidase substrate kit (Vector Laboratories) followed by hematoxylin counterstain. For immunofluorescence studies, sections were stained using rabbit PanCK or goat CK19 (1:100) antibodies and DyLight 488–conjugated donkey anti-goat antibodies (1:1000), followed by biotinylated anti-BrdU antibody (1:200) overnight at 4°C, and DyLight 594–conjugated streptavidin (1:1000). ProLong Gold antifade reagent with DAPI was used to stain the nuclei in the cover glass for the slides. Images were captured using an Olympus BX51 microscope with appropriate filters. Proliferation of cholangiocytes was determined by counting CK19⁺ and CK19⁻ cells within well-oriented EHBDs at 400X magnification, followed by counting the BrdU⁺ cells. The percentage of BrdU-positive CK19⁺ epithelial cells, was determined for each animal on the EHBD sections as previously reported³, with at least 500 cells counted per genotype for each animal.

Adoptive transfer experiments

For adoptive transfer experiments, total hepatic mononuclear cells from a pool of 3 to 4 WT or Il13^−/− mice (6- to 8-week-old) after IL-33 injections were subjected to flow sorting to obtain Lin⁻ST2⁺ ILC2s cells as previously described.³ A total of 2–4× 10⁴ purified Lin⁻ST2⁺ cells from WT or Il13^−/− littermate livers were resuspended in 100 μl PBS and injected intraperitoneally into Rag2^−/−gc^−/−mice; 4 days later, donor ILC2 cells were analyzed by flow cytometry. Proliferation was assessed by analyzing BrdU uptake in paraffin-embedded sections of recipient extrahepatic bile ducts using dual immunofluorescence staining.

Neonatal mouse model of experimental biliary atresia and treatments

Experimental biliary atresia was induced in newborn BALB/c mice by intraperitoneal injection of 1.5 × 10⁶ plaque-forming units (pfu) of RRV in 20 μl volume within the first 24 hours of birth; control newborn mice were injected with a similar volume of 0.9% NaCl (saline) solution. Newborn mice that died within the first 2 days after RRV or saline injections or that were neglected by their mothers were excluded from the study. For IL-33 studies, 0.1 μg of murine IL-33 was injected 24 hours after RRV infection and then daily for 6 days until day of life 7; control mice received an equivalent amount of protein (1% BSA). For blocking the AREG signaling pathway, infected neonatal mice who received IL-33 were injected i.p. with 80 mg/kg of the EGFR inhibitor (pEGFRi). Il4a^–/– newborn mice similarly challenged with RRV and receiving IL-33 were injected with IL-2/anti-IL-2 complex (1.5 μg IL-2 plus 15 μg anti-IL-2 mAb) until day 7 of life. The recombinant mouse IL-2 was incubated with mouse anti-IL-2 mAb for 15 min at room temperature before injecting into the neonatal mice.

Bone marrow transplant experiments

Bone marrow was isolated from the femur and tibia of 5- to 6- week-old Thy1.1⁺/Il4ra^+/+ or Thy1.2⁺/Il4ra^−/− mice. Donor bone marrow from Il4ra^+/+ or Il4ra^−/− mice (0.5–1 × 10⁷ cells in 100 μl PBS) was injected intravenously into sub-lethally irradiated (450 cGy) Il4ra^−/− mice. After 5 weeks, cells from recipient livers were analyzed by flow cytometry. Cholangiocyte proliferation was analyzed with BrdU uptake by immunofluorescence staining after 4 daily consecutive IL-33 injections.

RNAseq and survival data of patients with biliary atresia

Bulk liver RNAseq data of 121 patients with biliary atresia was retrieved from previous deposition in the Gene Expression Omnibus database under accession number GSE122340. All research was conducted the accordance with both the Declarations of Helsinki and Istanbul, approved with the institutional review committee, and written consent was given in writing by all subjects. The RPKM values (reads per kilobase of transcript per million reads mapped) were used for analyses. For calculating IL-33 + IL-2 scores, z-score normalizations of the raw RPKM values of IL-33 and IL-2 were performed separately across all the patients. The sum of the normalized levels of IL-33 and IL-2 were defined as the IL-33 + IL-2 scores. Survival data was obtained from previous publication.¹⁴ The survival data refers to the time of liver transplant-free survival after hepatoportoenterostomy (HPE).

Estimation of relative cell abundance of specific immune populations in patients with biliary atresia

Estimation of the relative cell abundance of nILC2s, iILC2s, and other targeted immune populations was performed by the single sample Gene Set Enrichment Analysis (ssGSEA) implemented in the GSVA-R package. RPKM values were used as the input. ssGSEA is rank-based method that calculates an enrichment score for each sample.³³ The enrichment score was calculated by a sum of the difference between weighted Empirical Cumulative Distribution Functions of the marker genes and the other genes in an expression profile. The enrichment score represents the abundance of cells. The marker genes for nILC2, iILC2, and other immune populations were obtained from previous publications and reported in Table S1.

Statistical analysis

All in vitro experiments were performed at least in triplicate. Student’s t-test (two-tailed) was used to analyze the significance of data comparison with a significance set at P < 0.05, except where indicated otherwise. 1-way or 2-way ANOVA was used for comparison between 3 or more groups, with a significance of P < 0.05. Values are expressed as Mean ± SD. Spearman correlation analyses and plots were performed with corrplot package in R (version 3.6.1). Kaplan-Meier survival curves were generated using GraphPad Prism (GraphPad Software).

ILC2 subtypes predict cholangiocyte proliferation and transplant-free survival in children with biliary atresia

To investigate the clinical implications of ILC2 subpopulations in human infants with biliary atresia, we first performed Spearman correlation analysis among nILC2 and iILC2-associated genes (IL33, IL2, and AREG) and the expression of KRT19 (keratin-19) representing cholangiocytes in an RNAseq dataset from the livers of 121 patients at the time of biliary atresia diagnosis (reported previously in reference¹⁴). The abundance of nILC2s and iILC2s was estimated with single sample gene set enrichment analysis (ssGSEA) using marker genes for nILC2s and iILC2s³⁴ (Table S1), respectively. ssGSEA is a rank-based method that calculates an enrichment score for each sample and indicates the abundance of cells.³³ Spearman correlation and linear regression analyses confirmed positive correlations between nILC2s and the expression of IL33, IL33 + IL2, AREG, and with the epithelial compartment represented by KRT19 (Figure 1A and B; p < 0.001). Negative correlations were detected with iILC2s (Figure 1B). Natural ILC2s are also positively associated with Th2 cells (rho: 0.63; p < 0.001), neutrophils (rho: 0.68, p < 0.001), macrophages (rho: 0.46, p < 0.001), and eosinophils (rho: 0.38, p < 0.001; Figure S1).

To further examine the association between nILC2s, iILC2s, and cholangiocyte proliferation (as determined by the abundance of KRT19 expression), we stratified the patients according to levels of nILC2s and iILC2s (Figure 1C). There was a gradation of expression level of KRT19, with the expression peaking in the high-nILC2s/low-iILC2s groups, and the lowest level in the low-nILC2s/high-iILC2s group (Figure 1C). Importantly, the low-nILC2/high-iILC2 cohort had the worst transplant-free survival in patients with biliary atresia after hepatoportoenterostomy (P < 0.05 compared to the low-nILC2/high-iILC2 cohort), Figure 1D). These findings were specific to nILC2s as there was no difference in survival between high-nILC2/high-iILC2 and high-nILC2/low-iILC2 cohorts (P = 0.1657). From this human data, we hypothesized that IL-33 and/or IL-2 driven expansion of nILC2s promotes epithelial proliferation and repair. To test this hypothesis, we probed IL-33 and IL-2- dependent ILC2 inter-lineage plasticity and resultant epithelial proliferation in both healthy mice and experimental biliary atresia.

IL-33 activates ILC2 subtypes and increases nILC2-expression of AREG with resultant epithelial proliferation

ILC2s undergo substantial expansion in the liver after IL-33 administration³, but the population of hepatic ILC2 subtypes is unknown. Previous reports identified Lin⁻ST2⁺KLRG1^int nILC2s in the lung and fat-associated lymphoid tissues at steady state, with Lin⁻ST2⁻IL-25R⁺KLRG1^hi iILC2s transiently emerging as ILC progenitors that are mobilized during inflammatory states and able to adopt either an nILC2 or ILC3-like phenotype.^31,35 To examine the number and distribution of these ILC2 subtypes in the liver, we administered 1 μg of IL-33 intraperitoneally (i.p.) daily for four days to BALB/c mice. Natural ILC2s (nILC2s) were identified in the livers of naïve mice under homeostatic conditions, while iILC2s only emerged after administration of IL-33 (Figure 2A). Interestingly, KLRG1 expression increased in the nILC2 population after IL-33 administration (Figure 2A), representing an intermediary ILC2 population with an activated phenotype (Lin⁻ST2⁺KLRG1^hi, denoted “activated-nILC2s”). Expression of additional activation markers (CD44 and CD25) were increased and CD62L expression was decreased in this population (Figure S2A).

In addition to these phenotypic changes, IL-33 increased the frequency and absolute number of all hepatic ILC2 subpopulations within hepatic mononuclear cells (HMNCs) (Figure 2B), with no difference in T cell frequency after IL-33 administration (Figure S2B). IL-33 administration also had the expected increase in IL-4, IL-5, and IL-13 expression by natural and activated-nILC2s (Figure S2C,D). Notably, IL-33 increased AREG expression in nILC2s and activated-nILC2s by up to 10-fold above the increases in iILC2s (Figure 2C,D). Given the higher absolute number and frequency of AREG-expression within nILC2s and activated-nILC2s, we focus on this combined population of natural ILC2s as the most relevant source of AREG within hepatic ILC2s.

To explore the functional relationship between AREG and epithelial proliferation, wildtype mice were treated with IL-33 ± pEGFRi (Canertinib, an inhibitor of all epidermal growth factor receptor signaling, including AREG).³⁶ The co-administration of pEGFRi reduced proliferation by 5-fold (Figure 3A,B), prevented the visual enlargement of EHBDs (Figure 3C) and ductal hyperplasia (Figure S3A) compared to control mice treated with IL-33 alone. Mice treated with pEGFRi without IL-33 demonstrated the expected decrease in ductal patency confirming this pathway has relevance under physiologic conditions (Figure S3B). Administration of recombinant AREG to directly determine the proliferative properties of IL-33 resulted in a 5-fold increase in cholangiocyte proliferation (Figure 3D). Taken together, these findings in healthy mice demonstrate that hepatic nILC2s, but not iILC2s, produce AREG which drives epithelial cell proliferation.

Fewer nILC2s and reduced AREG expression in experimental biliary atresia

To investigate the role of ILC2 plasticity and AREG to disease, we utilized a mouse model of biliary atresia. Cholangiopathy is induced by the i.p. administration of 1.5×10⁶ plaque-forming units (pfu) of rhesus rotavirus (RRV) into BALB/c mice within 24 hours of birth (Figure 4A). Analysis of ILC2 subpopulations and AREG expression in sham mice, infected mice, and infected mice also receiving IL-33 revealed that nILC2s (Figure 4B) and AREG-expressing cells (Figure 4C,D) were suppressed after RRV infection and that this phenotype was reversed by the administration of IL-33. Inhibiting the AREG pathway with pEGFRi in this model resulted in significantly reduced cholangiocyte proliferation in RRV-infected mice treated with IL-33 (Figure 4E). Proliferation of CK19⁻ cells are present in RRV-infected mice, most likely representing immune cells as previously reported.^8,15,37 These data demonstrate that reduced nILC2 and AREG expression in mice with experimental biliary atresia results in decreased epithelial proliferation, recapitulating the associations observed in human disease (see Figure 1A,B).

ILC2 plasticity and AREG expression depend on an intact IL-13/IL-4Rα/STAT6 circuit

Based on previous reports that IL-13 mediates the proliferative properties of IL-33 in epithelial cells^3,24, we hypothesized that the expression of AREG is downstream of IL-13 signaling. To test this hypothesis, we administered 1 μg IL-33 ± 10 μg AREG to Il13^−/− mice. While IL-33 treatment of Il13^−/− mice resulted in minimal proliferation, the administration of AREG significantly increased proliferation (Figure 5A,B). This result demonstrated that AREG is dependent on IL-13, perhaps downstream of the IL-4Rα subunit of the IL-13R and signals through STAT6.³⁸ To examine this possibility, we administered IL-33 to Il13^−/−, Il4ra^−/− and Stat6^−/− mice. IL-33 increased the number of ILC2s in the livers and EHBDs of knockout strains compared to WT controls (Figure S4A,B) with significant ILC2 phenotypic differences. In the knockout strains, there was a proportional decrease in nILC2s (Figure 5C), a shift toward an activated phenotype as shown by increased KLRG1 expression and IL-25R positivity (Figure S4C,D), and an increase in iILC2s (Figure 5D and Figure S4E). Ex vivo cytokine expression by natural and activated nILC2s is also altered with increased IL-13 and IL-17 compared to WT controls (Figure S4F). Taken together, these data suggest that the disruption of the IL-13/IL-4Rα/STAT6 circuit promotes a switch of nILC2s to an activated and inflammatory phenotype. Notably, expression of AREG is decreased in all ILC2 subtypes from knockout mice (Figure 5E).

Disruption of the IL-13/IL-4Rα/STAT6 pathway in vivo impairs epithelial cell proliferation in an AREG-dependent manner

To precisely determine if IL-13 and its IL-4Rα signaling are required for the proliferative properties of IL-33, we quantified BrdU⁺ cells in the EHBDs of Il13^−/−, Il4ra^−/−, and Stat6^−/− mice after IL-33 administration. Proliferation and epithelial hyperplasia were significantly decreased in knockout mice (Figure 6A and Figure S5). To investigate the specificity of this finding to ILC2s (rather than impaired signaling in epithelial cells or T cell-dependent effects), we reconstituted Rag^−/−gc^−/− mice, which are deficient in ILCs, B cells, and T cells, with ILC2s from WT or Il13^−/− mice by adoptive transfer (Figure 6B). The population of ILC2s in reconstituted Rag^−/−gc^−/− mice was documented by flow cytometry (Figure S6). Subjecting these recipient chimeric mice to IL-33, we found that cholangiocyte proliferation in EHBDs increased only in Rag^−/−gc^−/− mice reconstituted with ILC2s from WT mice (Figure 6C). Thus, ILC2s are necessary and sufficient to mediate IL-33-dependent cholangiocyte proliferation in the absence of T cells.

Next, we administered IL-33 i.p. into Il4ra^−/− mice after their bone marrow had been reconstituted with cells from WT or Il4ra^−/− mice (Figure 6D and Figure S7A,B). Quantifying BrdU-positive bile duct epithelial cells, we found that Il4ra^−/− mice reconstituted with WT donor cells (Il4ra^+/+ > Il4ra^−/−) had an expected increase in cholangiocyte proliferation, which was substantially suppressed in Il4ra^−/− mice receiving cells from Il4ra^−/− donors (Figure 6E). Notably, the administration of anti-AREG antibodies to Il4ra^−/− mice reconstituted with WT donor cells (Il4ra^+/+ > Il4ra^−/−) blocked the cholangiocyte proliferation induced by IL-33 (Figure 6F). These data demonstrate the independent role of ILC2s and of its own IL-13 in controlling the plasticity of the functional phenotypes. They also identify a hematopoietic source of IL-4Rα expression and AREG as a downstream effector molecule produced by nILC2s to promote proliferation of neighboring epithelial cells.

IL-2 restores the nILC2 phenotype and AREG expression in mice with defective IL-13 signaling

While nearly all ILCs are dependent on IL-7 for development³⁹, the combination of IL-2 and IL-7 converts iILC2s into nILC2s in vitro.³¹ Thus, we hypothesized that IL-2 is required for nILC2 stability in addition to IL-33. This hypothesis is supported by the human Spearman correlation analysis demonstrating a positive correlation of nILC2s with IL33 + IL2 but not IL2 alone (see Figure 1A,B). To determine the in vivo effect of IL-2, we repopulated Rag^−/−gc^−/− mice with ILC2s from WT or Il13^−/− mice and challenged mice with IL-33 ± IL-2/anti-IL-2 complex (which activates IL-2 signaling) (Figure 7A).⁴⁰ Treatment with the IL-2/anti-IL-2 complex restored AREG production (Figure S8) and cholangiocyte proliferation (Figure 7B,C) in mice receiving Il13^−/− nILC2s.

Lastly, we infected Il4ra^−/− neonatal mice with RRV, followed by daily administration of IL-33 ± IL-2/anti-IL-2 complex. RRV infection alone resulted in decreased expression of ILC2-related cytokines in mice compared to WT controls (Figure S9A). The IL-2/anti-IL-2R complex successfully restored the nILC2 cell population (Figure 7D), increased cholangiocyte proliferation (Figure 7E,F) and maintained the epithelial lining and patency of EHBD (Figure S9B). Altogether, the data underscored the key role of IL-2 in the regulation of ILC2 functional plasticity and its source of AREG as an important effector of epithelial proliferation induced by IL-33.