Epithelial overexpression of IL-33 induces eosinophilic esophagitis dependent on IL-13

Abstract

Background:

Eosinophilic esophagitis (EoE) is an increasingly common inflammatory condition of the esophagus; however, the underlying immunologic mechanisms remain poorly understood. The epithelium-derived cytokine IL-33 is associated with type 2 immune responses and elevated in esophageal biopsies from EoE subjects.

Objective:

We hypothesized that overexpression of IL-33 by the esophageal epithelium would promote the immunopathology of EoE.

Methods:

We evaluated the functional consequences of esophageal epithelial overexpression of a secreted and active form of IL-33 (saIL-33) in a novel transgenic mouse, EoE33. EoE33 mice were analyzed for clinical and immunologic phenotypes. Esophageal contractility was assessed. Epithelial cytokine responses were analyzed in three-dimensional organoids. EoE33 phenotypes were further characterized in ST2^−/−, eosinophil-deficient, and IL-13^−/− mice. Finally, EoE33 mice were treated with dexamethasone.

Results:

EoE33 mice displayed ST2-dependent, EoE-like pathology and failed to thrive. Esophageal tissue remodeling and inflammation included basal zone hyperplasia, eosinophilia, mast cells, and Th2 cells. Marked increases in levels of type 2 cytokines, including IL-13, and molecules associated with immune responses and tissue remodeling were observed. Esophageal organoids suggested reactive epithelial changes. Genetic deletion of IL-13 in EoE33 mice abrogated pathological changes in vivo. EoE33 mice were steroid responsive.

Conclusion:

IL-33 overexpression by the esophageal epithelium generated immunopathology and clinical phenotypes resembling human EoE. IL-33 may play a pivotal role in the etiology of EoE by activating the IL-13 pathway. EoE33 mice are a robust experimental platform for mechanistic investigation and translational discovery.

Graphical Abstract

Capsule Summary:

Overexpression of IL-33 in the mouse esophagus elicits clinical and pathologic features of EoE. These features do not require eosinophils and are dependent on IL-13 signaling.

Introduction

Eosinophilic esophagitis (EoE) is a chronic, immune-mediated disease characterized by type 2 inflammation of the esophagus. Recent studies have highlighted the importance of the epithelium in disease susceptibility and the pathophysiology of EoE.¹ Indeed, mutations or dysregulation in multiple genes regulating proteolytic activity and epithelial turnover have been implicated in EoE.^{2, 3} Dysregulation of these pathways likely results in impaired barrier function and increased production of type 2 cytokines. Notably, an epithelial-derived cytokine, IL-33, is elevated in esophageal biopsies from EoE subjects.^4–6 IL-33 is sequestered in the nucleus and generally released to the extracellular milieu by cellular damage or tissue injury. The IL-33 receptor, ST2 (IL-1RL1), is expressed by multiple effector cell types involved in EoE, including eosinophils, mast cells, basophils, group 2 innate lymphoid cells (ILC2s), and Th2 cells.^{7, 8} Nonetheless, the role of IL-33 in EoE is unclear.

To understand the pathophysiology of EoE, several mouse models are noteworthy. For example, epicutaneous sensitization with ovalbumin (OVA) or Aspergillus antigen and airway challenge with the same antigen induced IL-5-dependent esophageal eosinophilia.⁹ Epicutaneous sensitization with OVA plus an adjuvant, vitamin D analog MC903, and intragastric challenge with OVA promoted development of EoE, which was dependent on TSLP.¹⁰ Alternatively, intratracheal administration of IL-13 protein¹¹ or transgenic overexpression of IL-13 in the lungs, or from the esophageal epithelium, induced esophageal eosinophilia and structural changes reminiscent of human EoE.^{12, 13} Finally, overexpression of IL-5 in the esophagus combined with oxazolone sensitization/oral challenge¹⁴ or oxazolone sensitization/oral challenge alone¹⁵ also achieved marked esophageal pathology. While these previous studies have been successful in recapitulating certain features of EoE, limitations include the use of model antigens and ectopic overexpression/administration of cytokines or a hapten.

Interestingly, systemic administration of IL-33 to mice produced pathologic changes in the esophagus consistent with EoE⁵ but the study was unable to evaluate the chronic outcomes of IL-33-mediated inflammation. Another study in mice suggests tape stripping induced epicutaneous release of IL-33 and predisposed to esophageal eosinophilia, dependent on ST2. However, the effects of local IL-33 expression in the esophagus were not assessed.¹⁶ Transgenic overexpression of IL-33 could address these limitations. However, several unique features of IL-33 need to be considered: 1) IL-33 is expressed largely in mucosal tissues and only by certain cell types (e.g. esophageal basal epithelial cells in EoE⁶); 2) IL-33 may be constitutively expressed but is sequestered in the cell nucleus⁸; and 3) systemic overexpression of IL-33 causes lethal systemic inflammation.^{17, 18}

The objective of this study was to examine the role of epithelial-derived IL-33 in EoE pathogenesis. To address the hurdles associated with IL-33 expression, we developed a novel mouse model in which a secreted and active form of IL-33 is overexpressed by esophageal epithelial cells. These mice (i.e. EoE33) failed to thrive and showed esophageal pathology and functional changes resembling human EoE. Importantly, they showed elevated esophageal levels of IL-13 and genetic deletion of IL-13 abrogated clinical and pathologic phenotypes.

Methods

Gene construct

A transgene expressing secreted and active IL-33 (saIL-33) was created by combining a cDNA fragment for the mouse IL-2 secretory signal sequence corresponding to the N-terminal 20 amino acids (a.a.) of IL-2 with another cDNA fragment encoding an active form¹⁹ of mouse IL-33 (a.a. 109–266). Further details are included in Supplementary Methods.

Transfection experiments

HEK-293 cells were transfected with pcDNA3.1+saIL-33 or pcDNA3.1 vector (V79020, Thermo Fisher Scientific, Waltham, MA) alone. HEK-293 cells were stained for IL-33 or collected for Western blot. Conditioned media was collected for eosinophil culture experiments. Bone marrow-derived eosinophils²⁰ were cultured in conditioned media for 24 h and assessed for IL-13 production by ELISA (R&D Systems, Minneapolis, MN). Further details are included in Supplementary Methods.

Mice

Protocols and studies involving mice were performed in accordance with National Institutes of Health guidelines and approved by the Mayo Clinic Institutional Animal Care and Use Committee. All mice used in these studies were age- and sex-matched on a C57BL/6J background. See Supplementary Methods for dexamethasone treatment. EoE33 transgenic mice were generated by the Special Animal Services Laboratory at Mayo Clinic Arizona (directed by Dr. Doyle) following a standard protocol.²¹ Further details are included in Supplementary Methods.

Esophageal muscle tension measurements

To assess esophageal contractility, we adapted a method previously described for tracheal tension assessment.²² Further details are included in Supplementary Methods.

Hematological assays

See Supplementary Methods.

Esophageal tissue characterization and histological analysis

Esophageal diameters were assessed using a digital micrometer (Mitutoyo, Kawasaki, Japan). Tissue was processed and stained as described previously.^{23, 24, 25} The EoE Histologic Scoring System (EoEHSS) was adapted for mouse histology and H&E-stained sections of esophagus were analyzed by a gastrointestinal pathologist in a blinded manner.²⁶ Further details are included in Supplementary Methods.

Flow cytometry analyses of immune cells in esophageal tissue

Three to five whole esophagi per group were pooled. Further details are included in Supplementary Methods.

RNA isolation, sequencing, and analysis

Following tissue extraction, whole esophagi were collected in RNeasy (Qiagen Inc. Venlo, Netherlands), immediately flash frozen in liquid nitrogen and shipped on dry ice to GENEWIZ (Azenta Inc., Burlington, MA) for bulk RNA extraction, quality control assessment, Illumina HiSeq mRNA library preparation, sequencing, and differential gene expression analysis. mRNA was enriched using polyA selection and sequenced using a 2×150 paired end configuration. The raw sequencing data was trimmed (Trimmomatic v.0.36)²⁷ before alignment and mapped to the ENSEMBL (www.ensembl.org) Mus musculus GRCm38 reference genome (STAR aligner v.2.5.2b)²⁸. Unique reads that fell within exonic regions were quantified as gene hit counts (featureCounts Subread package v.1.5.2 https://subread.sourceforge.net/) and used for differential expression analysis. Gene hit counts were normalized and compared using DESeq2.²⁹ Differences in normalized mean gene expression between EoE33 and wild-type (WT) groups (4 biological replicates each) were quantified as log2 [fold change (FC)] values. To assess statistical significance, p-values were generated using the Wald test and adjusted using the Benjamini-Hochberg/FDR correction (padj). Genes with a |log2FoldChange|>1 and padj<0.05 were considered statistically significant differentially expressed genes (DEGs). A volcano plot was generated using log2 fold change and padj values to visualize DEGs (selected genes are highlighted). Exploratory analyses, data visualization, as well as pathway and over-representation analyses were performed using iDEP.96³⁰, ShinyGO³¹, and g:Profiler.³² Over-representation-based pathway enrichment analysis was performed to identify gene ontology (GO) terms enriched with either up or downregulated DEGs. Heat maps of DEGs of interest were generated to visualize expression data across samples for each gene. Data are deposited in the Gene Expression Omnibus (GEO; www.ncbi.nlm.nih.gov/geo) database under accession number GSE238122.

Protein isolation and quantification

Esophageal protein concentrations were normalized and assessments were made by bead-based multiplex assay or ELISA. Serum levels of food-specific antibodies were determined by ELISA as described previously.³³ Further details are included in Supplementary Methods.

Three-dimensional (3D) esophageal organoids

Esophageal epithelial cells were isolated from EoE33 and wild-type (WT) mice to generate 3D organoids as described previously.^{23, 34} Organoids were subjected to histological analyses as described.^{23, 34} Further details are included in Supplementary Methods.

Statistical methods

Data was assessed for normality and Student’s t test or Mann-Whitney U test was used to compare group means or medians. Experimental groups contained n=3–12 mice per group. All experiments were repeated to verify reproducibility. Statistical comparisons and plots were generated using GraphPad Prism (version 9, GraphPad Software, San Diego, CA).

Results

A novel gene expression construct generated secreted and active IL-33

We developed a novel secreted and active IL-33 (saIL-33) transgene construct (Figure 1A) by combining the IL-2 secretory signal peptide gene sequence with a sequence encoding an active 19-kDa c-terminal IL-33 fragment (a.a.109–266). To examine saIL-33 expression and function in vitro, the saIL-33 (pCDNA3.1+saIL-33) construct was transfected into HEK-293 cells. IHC and Western blot showed saIL-33 expression in transfectants with pcDNA3.1+saIL33, but not in those with an empty vector control. Cells transfected with saIL-33 revealed cytoplasmic IL-33 staining and the expected 19-kDa molecular mass that was comparable to commercial recombinant IL-33 (a.a. 109–266) (Figure 1B, C). To verify extracellular IL-33 secretion and its activity, bone marrow-derived mouse eosinophils were incubated with conditioned media from the transfected HEK-293 cells. Conditioned media from the saIL-33 transfectants induced eosinophil IL-13 production. This response was dependent upon IL-33 receptor (ST2) expression in eosinophils (Figure 1D).

Figure 1: A novel gene construct generated secreted and active IL-33.

(A) Secreted and active IL-33 (saIL-33) gene construct with a human CMV promoter (hCMV). (B) Immunocytochemical staining of IL-33 (red) on HEK-293 after transfection with either saIL-33 or empty vector. (C) Western blot of IL-33 in HEK-293 cell lysates after transfection with saIL-33 (lane 2) or empty vector (lane 3). Recombinant IL-33 (rIL-33) (5 ng, lane 1) was used as a positive control. B-actin was used as a loading control. (D) ELISA of eosinophil-derived IL-13 following in vitro culture with conditioned media from transfected (vector vs. saIL-33) HEK-293 cells. Data is shown as mean ± SEM. **p<0.01

Squamous epithelial expression of saIL-33 generated a novel mouse model of EoE

To investigate the role of IL-33 in EoE in vivo, we developed a novel mouse model in which the saIL-33 transgene was expressed selectively in the esophageal epithelium (i.e. EoE33). To achieve this, we fused the saIL-33 transgene to the EBV ED-L2 promoter that is primarily active in the upper aerodigestive stratified squamous epithelia³⁵ (Figure 2A). Immunohistochemical analysis of EoE33 mice showed increased expression of IL-33 protein in the epithelium of the esophagus as compared to WT control mice (Figure 2B, Supplementary Figures 1, 2). H&E staining showed epithelial hyperplasia, elongation of papillae, and eosinophilic inflammation in the esophagus of EoE33 mice (Figure 2B). EoE33 mice crossed with ST2^−/− mice showed comparable expression of IL-33 in the epithelium; however, they did not show pathological changes. ELISA of esophageal homogenates confirmed increased production of IL-33 in EoE33 mice (16.6 ng vs. 0.4 ng, p<0.0001) (Figure 2C). Increased IL-33 was also observed in the oropharynx and forestomach consistent with transgene expression patterns of the ED-L2 promoter reported in previous studies (Figure 2C).^{14, 35} IL-33 was detectable in the serum of EoE33 mice (115 pg/mL vs. undetected) (Supplementary Figure 3A). However, peripheral blood leukocyte counts were similar between EoE33 and WT mice (Supplementary Figure 3B).

Figure 2: Expression of functional saIL-33 by the mouse esophageal epithelium.

(A) The Epstein–Barr virus-derived promoter, ED-L2, was used to drive expression of saIL-33 from esophageal epithelium to generate “EoE33” mice. (B) Representative hematoxylin & eosin (H&E (top) and IL-33 immunohistochemistry staining (brown; bottom) of esophageal cross sections from wild-type (WT), EoE33, and EoE33 × ST2^−/− mice. (C) IL-33 ELISA of whole tissue homogenates from EoE33 vs. WT mice. n = 4 mice per group. Data is shown as mean ± SEM. **** p<0.0001. Scale bar = 200 μm. Inset scale bar = 100 μm.

EoE33 mice exhibited clinical and pathologic features of EoE

EoE33 mice failed to thrive compared to WT littermates, a clinical feature observed in some pediatric patients with EoE³⁶ (Figure 3A). In addition, esophageal thickening^{37, 38} was grossly apparent in EoE33 mice (Figure 3B) throughout the length of the esophagus. Histologic and flow cytometric assessment revealed epithelial and submucosal infiltration of eosinophils (Figure 3C, Supplementary Figure 4). Further examination by IHC staining for EPX showed extensive eosinophil infiltration and occasional eosinophilic microabscesses.²⁶ In addition, Masson Trichrome staining revealed fibrotic changes and Ki-67 staining confirmed epithelial hyperplasia (Figure 3C). These findings were also observed throughout the length of the esophagus and consistent across sexes (Supplementary Figure 5). In humans, EoE is diagnosed and monitored by histologic analysis of tissue biopsies. We used the histologic scoring system, EoEHSS²⁶, to assess esophageal biopsies from EoE33 and WT mice. EoE33 mice showed esophageal histopathology closely resembling human disease (Figure 3D, E, Supplementary Figure 6), including basal zone hyperplasia (BZH), dilated intercellular spaces (DIS), and lamina propria fibrosis (LPF).

Figure 3: EoE33 mice exhibited features pathognomonic of eosinophilic esophagitis.

(A) Representative image and weights of eight-week-old male wild-type (WT) and EoE33 littermates. (B) Whole esophagi and esophageal diameter measurements from WT and EoE33 littermates. (C) Representative hematoxylin & eosin (H&E), eosinophil peroxidase (EPX (red)), Masson’s trichrome (MT), and Ki-67 (brown) staining of esophageal cross sections from WT and EoE33 mice. (D) H&E-stained esophageal biopsies from WT and EoE33 mice were assessed using the EoE Histologic Scoring System (EoEHSS). Esophageal cross sections were scored for both degree (grade) and extent (stage) of pathology. Mean EoEHSS component scores and composite scores are shown for EoE33 mice (WT scores = zero). EI, eosinophil infiltration, EA; eosinophil abscess, SL; eosinophil surface layering, DIS; dilated intracellular spaces; BZH; basal zone hyperplasia, and LPF; thickened lamina propria fibers. (E) H&E-stained esophageal biopsies from WT and EoE33 mice were assessed for peak eosinophils per high-power field (eos/hpf) in the epithelium and stroma. (F) Esophageal muscle tension in response to methacholine (MCh) in WT vs. EoE33 mice. n = 3 – 12 mice per group. Data is shown as mean ± SEM. **** p<0.0001, **p<0.01, *p<0.05. Scale bar = 100 μm. ND – not detected.

Esophageal contractility is required for peristalsis and efficient transport of ingested materials. EoE patients may present with dysphagia and food impaction.³⁹ We observed concentration-dependent increases in muscle tension in response to methacholine (MCh) in WT esophagi (change in tension from baseline of 16.3% (10⁻⁶ MCh) to 34.7% (10⁻⁴ MCh), p<0.01). However, esophageal muscle tension was significantly increased in EoE33 esophagi as compared to WT mice (Figure 3F). Together these data suggest that IL-33 overexpression promotes structural and functional changes in the esophagus and histopathology consistent with human EoE.

EoE33 mice exhibited type 2 inflammation in the esophagus

We performed bulk RNA sequencing (RNA-seq) on whole esophagus to elucidate the mechanisms involved in development of esophageal pathology in EoE33 mice. We identified 4,603 DEGs in EoE33 vs WT esophagi; 2,116 genes were upregulated and 2,487 were downregulated (Figure 4A). Gene ontology analysis showed that upregulated genes in EoE33 mice were enriched in immune response, defense response, innate immune response, and adaptive immune response processes while genes downregulated in EoE33 mice were enriched in epithelial differentiation, cell junction and extracellular matrix organization, and muscle contraction (Figure 4B and Supplementary Table 1). We further focused on DEGs enriched in immune and tissue remodeling pathways previously associated with human EoE^{15, 40–44} (Figure 4C, Supplementary Tables 2, 3). Type 2 cytokines/chemokines were significantly elevated, including Il13 [log2(FC) = 6], Ccl11 [Eotaxin-1; log2(FC) = 5], Alox15 [log2(FC) = 7], eosinophil associated Rnase2a [log2(FC) = 9], and mast cell associated protease Mcpt1 [log2(FC) = 7]. Notably, the tissue remodeling-associated protease Mmp12 was increased [log2(FC) = 4] while epithelial junction molecules desmoglein and filaggrin were downregulated [Dsg1a and Flg, log2(FC) = 6 and 7, respectively]. These data suggest activation of type 2 immunity, increased eosinophils and mast cells, enhanced tissue remodeling, and epithelial barrier dysfunction in EoE33 mice.

Figure 4: EoE33 mice exhibited esophageal type 2 inflammation and tissue remodeling.

Bulk RNA-sequencing was performed on whole esophagi from eight-week-old male EoE33 and wild-type (WT) mice. (A) Volcano plot of differentially expressed genes (DEGs) between EoE33 and WT mice. Genes of interest are labeled. (B) Gene ontology (GO) analysis of up and downregulated DEG lists. Dot size represents the number of genes related to each biological process. (C) Heatmaps show normalized expression values for DEGs associated with immune and remodeling changes seen in EoE. The 20 largest absolute log2(fold change) values in each subset are bolded. (D) Type 2 cytokine levels in whole esophageal homogenates of WT vs. EoE33 mice. (E) Chloroacetate esterase (CAE) staining and mean CAE+ cells per esophageal cross section in WT vs. EoE33 mice. Arrows indicate CAE+ cells. (F) Immunohistochemistry staining for CD4 (brown) and CD4+ cells per esophageal cross section in WT vs. EoE33 mice. (G) Flow cytometry plot and quantification of esophageal Th2-type CD4+ T cells in WT vs. EoE33 mice. Frequencies shown as percentage of live, CD45+, CD3+, CD4+ population. n =−4 – 8 mice per group. Data shown as mean ± SEM. ***p<0.001, **p<0.01, *p<0.05. Scale bar = 100 μm.

Consistent with the bulk RNA-seq data, increased protein levels of type 2 cytokines including IL-4, IL-9, and IL-13 as well as eotaxin-1 were observed in EoE33 esophageal homogenates by bead-based multiplex assay and ELISA (IL-13) (Figure 4D). Increased mast cells in the EoE33 esophagus were observed by CAE staining (Figure 4E). Moreover, increased infiltration of CD4⁺ T cells, expressing ST2 and CD69, were observed in EoE33 compared to WT mice (Figure 4F, G and Supplementary Figure 7). Together, these data show increased Th2-type inflammation in EoE33 esophageal tissue, which may explain the marked increase in mast cells and eosinophils.

Patients with EoE show increased serum levels of IgG4 antibodies to foods, including milk, wheat, egg, and soy^{45, 46}, suggesting that they develop humoral immune responses to food components. Therefore, we examined serum levels of antibodies against food antigens in EoE33 mice by ELISA. Standard rodent chow (PicoLab Rodent Diet 20 cat#5053, Purina LabDiets, St. Louis, MO) contained wheat, corn, and soy as the major ingredients. Titration curves showed that EoE33 mice have increased serum IgG1^{47, 48} antibodies to wheat, but not to corn or soy (Figure 5).

Figure 5: EoE33 mice developed a food-specific humoral response to wheat.

Serially diluted serum from eight-week-old wild-type (WT) and EoE33 littermates was measured for antibodies against wheat, corn, and soy protein extracts (components of rodent chow) by ELISA. n = 4 mice per group. Data is shown as mean ± SEM. *p<0.05. O.D. – optical density.

Organoid culture suggested reactive epithelial alterations in the inflammatory milieu of EoE33 mice

BZH is linked to epithelial barrier dysfunction in the pathogenesis of EoE. BZH in EoE33 mice (Figure 3C, D, Figure 6A) was corroborated by a statistically significant expansion of Ki-67-positive proliferative cells with nuclear expression of Sox2, a marker of basal keratinocytes (Figure 6A, B). BZH in EoE33 mice was further evidenced by nucleated squames with cells retaining nuclei in almost all layers (para-keratinization) (Figure 6A), suggesting a consequence of rapid proliferation and differentiation that does not allow for the dissolution and disappearance of the nuclei. By contrast, WT squames underwent normal keratinocyte maturation with anucleated cells at the superficial cell layers (ortho-keratinization). It remains elusive how epithelial IL-33 contributes to BZH. To elucidate epithelial changes linked with saIL-33 transgene expression, organoid cultures were established from EoE33 and WT mice. H&E staining revealed a notable difference in terminal differentiation between EoE33 and WT organoids. Most EoE33 organoids displayed concentric nucleated squames compatible with para-keratinization while WT organoids displayed predominantly complete central keratinization compatible with ortho-keratinization (Figure 6C), recapitulating histologic features of the original epithelia. Additionally, a subset (30–40%) of EoE33 organoids displayed an expansion of basaloid cells relative to WT counterparts, suggesting increased proliferation as evidenced by increased Sox2-positive cells concurrent with elevated Ki-67 LI (Figures 6C, D). However, the majority of EoE33 organoids lost the sign of proliferative expansion of basaloid cells by passage 3 (P3) as suggested by phase-contrast images (Supplementary Figure 8A). Of note, the organoid culture condition is not permissive for non-epithelial cell populations such as immune cells and fibroblasts present in the original inflammatory tissues. Thus, epithelial transgenic saIL-33 expression may not be sufficient to maintain BZH-like phenotypes ex vivo. In fact, exogenously added recombinant IL-13, but not IL-33, induced BZH-like changes in WT organoids (Figure 6E).

Figure 6: EoE33 mice exhibited epithelial alterations in esophageal organoid cultures.

EoE33 and wild-type (WT) original esophageal epithelia (A) and organoids grown ex vivo (C-E) were subjected to morphological analyses. Representative hematoxylin & eosin (H&E) - stained and immunofluorescent (IF) images of esophageal epithelia (A) and organoids (C) with indicated genotypes. Esophageal epithelia were analyzed in longitudinal sections. Sox2 (red), Ki-67 (green), and DAPI (nuclei) are shown in IF. Bar graphs show Sox2 and Ki-67 labeling indices (LI) for original epithelia in (B) and organoids in (D). (E) WT esophageal organoids were treated with either IL-13 or IL-33 (10 ng/mL or 100 ng/mL) and representative H&E images are shown. Data is shown as mean ± SEM. Dashed line indicates basement membrane. Scale bar = 100 μm. ns – not significant.

Bulk RNA-seq of EoE33 and WT organoids maintained significant differences of gene expression profile at passage 2 (P2) (Supplementary Figure 8B and C). g:Profiler analysis revealed several EoE-relevant phenotypes such as elevated stress response and response to cytokines. However, EoE-relevant cytokines known to induce BZH-like changes in organoids⁴⁹ such as IL-13 were not differentially expressed in the EoE33 organoids compared with WT organoids. Moreover, g:Profiler analysis failed to detect other EoE-relevant changes such as decreased cell adhesion, loss of differentiation, and altered cell cycle progression. Together, these organoid data suggest that isolated epithelial cells from EoE33 mice maintain some characteristics of EoE, but non-epithelial cells are required to induce full expression of the BZH phenotype observed in mice.

EoE33 pathology was abrogated by IL-13 deficiency and improved with steroid treatment

The pathological changes (e.g., H&E, RNAseq, epithelial cell organoids) and esophageal inflammation in EoE33 mice are likely mediated by type 2 immune responses. To address this further, we first crossed EoE33 with ST2^−/− mice and found that inflammatory and remodeling changes were dependent on ST2 expression. Notably, eosinophils were detected throughout the esophagus including the outer muscle layers, dependent on ST2. We also noted the presence of MPO-positive cells in EoE33 mice, which was dependent on ST2 (Supplementary Figure 9). Next, we examined the contribution of eosinophils by crossing EoE33 with eosinophil-deficient ΔdblGATA mice. We found that in the absence of eosinophils, dilated intracellular spaces, basal zone hyperplasia, and fibrosis persisted (Figure 7A, B). Notably, increased CD4 staining was also comparable to EoE33 mice (Supplementary Figure 10). Next, we examined the contribution of IL-13 in the EoE33 model. IL-13 is considered a central mediator of EoE^{12, 50, 51} and was upregulated in EoE33 mice (Figure 4D). To examine the contribution of IL-13 in EoE33 mice, we crossed EoE33 with IL-13^−/− mice. Remarkably, pathologic changes were absent in IL-13-deficient EoE33 mice (Figure 7C, D, Supplementary Figure 11). However, a small number of eosinophils were detected in the esophagus of IL-13-deficient EoE33 mice by EPX IHC (Supplementary Figure 12).

Figure 7: EoE33 mice were responsive to steroid treatment and dependent on IL-13, whereas eosinophils were not required.

(A) Hematoxylin & eosin (H&E)-stained sections and (B) EoE Histologic Scoring System (EoEHSS) scores of wild-type (WT), EoE33, and eosinophil-deficient EoE33 (EoE33 × ΔdblGATA) mice. EI, eosinophil infiltration, EA; eosinophil abscess, ns; not significant, SL; eosinophil surface layering, DIS; dilated intracellular spaces; BZH; basal zone hyperplasia, and LPF; thickened lamina propria fibers. (C) H&E-stained sections and (D) EoEHSS scores of WT, EoE33, and IL-13-deficient EoE33 (EoE33 × IL-13^−/−) mice. (E) H&E staining of representative esophageal cross sections from wild-type (WT) and EoE33 mice after dexamethasone treatment (dex) vs. saline (sal). (F) EoEHSS scores in dex-treated EoE33 mice vs. sal. n = 3 – 5 mice per group. Data is shown as mean ± SEM. Scale bar = 200 μm. *p< .05; **p< .01

Steroids are commonly used to treat type 2 inflammation and have shown efficacy in EoE.^{52, 53} We assessed steroid-responsiveness by treating EoE33 mice with dexamethasone. Dexamethasone-treated EoE33 mice showed histologic improvement of the esophageal mucosa (Figure 7E) and EoEHSS grade, notably, eosinophil-associated parameters (Figure 7F). Together, these results suggest that IL-13 is required for IL-33-mediated EoE-like pathology, which is sensitive to glucocorticoid treatment.

Discussion

Expression of IL-33 is increased in the esophageal epithelium of patients with active EoE, but the role of IL-33 in EoE pathogenesis is not well understood. To directly address this knowledge gap, we generated a novel mouse model (EoE33) in which a secreted and active form of IL-33 is overexpressed from esophageal epithelial cells. EoE33 mice showed extensive esophageal tissue remodeling, eosinophilic infiltration, failure to thrive, altered esophageal muscle contraction, increased mast cells and elevated type 2 cytokines. Notably, EoE33 mice showed increased esophageal infiltration of CD4+ T cells as well as circulating wheat-specific IgG1 antibodies. Importantly, wheat is the second most common food trigger for human EoE⁵⁴ and this food-specific antibody response in EoE33 mice is reminiscent of increases in wheat-specific IgG4 seen in humans.⁴⁵ Together, these data highlight the contribution of esophageal epithelial cell-derived IL-33 in type 2 immune responses and pathology in the esophagus.

Previous studies have employed a similar strategy to express full-length mouse IL-33 in vivo.⁵⁵ In contrast to our studies, expression of full-length IL-33 led to inflammation that was not entirely dependent on ST2 signaling. We verified that the immunopathology of EoE33 mice is fully dependent on the IL-33 receptor, ST2 (Figure 2, Supplementary Figure 9). IL-33 overproduction due to chromosomal duplication in humans results in EoE and autoimmunity.⁵⁶ The factors driving non-genetic IL-33 increases in human EoE are unclear, but several models have implicated epithelial oxidative stress, which may promote IL-33 induction.^{49, 57, 58} Indeed, recent studies with the detergent sodium dodecyl sulfate suggest this environmental factor can induce oxidative stress pathways and promote IL-33 expression.^{23, 59}

Previous studies of i.p. administration of active 19-kDa IL-33 found that eosinophilic inflammation was dependent on IL-13 whereas infiltration of other leukocytes was not.⁵ Differences in strain and route of exposure may account for this discrepancy. We noted the presence of MPO in the EoE33 esophagus by IHC, suggesting neutrophilia, consistent with previous studies of IL-33-mediated inflammation.^{17, 18} Interestingly, this neutrophilic infiltrate was dependent on IL-13 (Supplementary Figure 11). Esophageal neutrophilia is not a known feature of human EoE.⁶⁰ It is unclear, given the IL-33 expression observed in human patients with active EoE, if neutrophilia is unique to the mouse esophagus, is underappreciated in humans, or if this may reflect temporal differences in the EoE33 model and human EoE.

Limitations of this study include the nature of the transgene. Unlike human EoE, where IL-33 may be increased only during active EoE⁶, the transgene is always expressed in an active form from the esophageal epithelium. More research is needed to understand the levels and processing of IL-33 in human EoE. In addition, it is unclear what effect constitutive production of active IL-33 may have on the esophagus during early development. Another limitation of the model is the transgenic IL-33 expression in the oropharynx and forestomach. It is unclear what relevance this may have to human disease and what effect this may have on the observed pathologies. Of note, many patients with EoE have comorbid oral allergy syndrome, which may be initiated by local alarmin release (IL-33, TSLP) and epithelial barrier dysfunction in the oropharyngeal mucosa.⁶¹ Finally, the ED-L2 promoter is active in the squamous epithelium, whereas increased IL-33 was observed in the basal epithelium of human patients with active EoE.⁶

We have established that IL-33 overexpression alone can induce marked disease features pathognomonic of EoE. Future directions include identifying the cellular source(s) of IL-13 and dissecting the roles of various cells (e.g. ILC2s, mast cells, T cells) and mediators driven by IL-33. To understand how food antigens contribute to various pathologies, EoE mice could be placed on an elemental diet. Further, it would be of great interest to investigate the induction and reversibility of disease pathology by temporally regulating IL-33 expression with an inducible transgene (e.g., tetracycline (tet)-inducible system²¹) — a direction we are actively pursuing. This would also facilitate investigations using an alternative promoter to drive IL-33 expression in the basal esophageal epithelium (e.g., Krt5 promoter-driven reverse tet transactivator).⁶²

Treatment of EoE33 mice with pharmacologic and genetic depletion strategies advances our understanding of EoE pathophysiology. Topical swallowed steroids are effective in most patients with EoE and their efficacy has been primarily attributed to their effects on tissue eosinophils. We demonstrated that EoE33 mice are steroid-responsive, but this effect appears to be independent of eosinophils. Indeed, we showed striking evidence that pathology persists in the complete absence of tissue eosinophilia. This observation is consistent with multiple clinical trials of biologic agents targeting eosinophils that have successfully reduced tissue eosinophils, but failed to achieve remission of clinical symptoms or other histologic endpoints.^63–65 Furthermore, IL-13 deficiency alone in our model abrogated all hallmark histologic features of EoE. This not only supports efforts targeting IL-13 in humans, but suggests that, at least in the esophagus, pathogenic IL-33 activities are primarily, or even exclusively, mediated through IL-13. Notably, the only FDA-approved therapy for treatment of EoE, dupilumab, targets the IL-4Rα chain, a shared receptor for IL-4 and IL-13.⁵⁰

In summary, we have generated a novel mouse model of EoE by overexpression of IL-33 from the esophageal epithelium. EoE33 mice exhibit robust pathology consistent with key features of human disease. This model establishes the potential of IL-33 involvement in EoE, identifies IL-13 as a critical downstream mediator of IL-33-driven esophageal pathology, and provides a platform for dissection of associated disease pathways and testing of therapeutic agents.

Key Messages:

• We have developed a novel and robust mouse model of eosinophilic esophagitis (EoE) dependent on tissue-specific transgenic expression of IL-33, termed EoE33. EoE33 mice exhibit hallmark disease features of EoE including failure to thrive, eosinophilic inflammation, and tissue remodeling.

• We provide convincing evidence in EoE33 mice that eosinophils are dispensable to EoE pathogenesis, and that IL-33 mediates its activities in the esophagus through IL-13.

• Our findings corroborate the disappointing results of eosinophil-targeted therapies (e.g., mepolizumab, benralizumab, lirentelimab) and the success of IL-13 blockade with dupilumab, the first FDA-approved therapeutic for EoE. In addition to these important insights, EoE33 mice fill a major unmet need, enabling further dissection of disease mechanisms and serving as a platform for testing therapeutic strategies.

Abbreviations:

a.a.

amino acid

AEC

absolute eosinophil count

BZH

basal zone hyperplasia

CAE

chloroacetate esterase

DIS

dilated intercellular spaces

DEC

dyskeratotic epithelial cells

DEG

differentially expressed gene

dex

dexamethasone

EA

eosinophilic abscess

EEsAI

eosinophilic esophagitis symptom activity index

EGID

eosinophilic gastrointestinal disease

EI

eosinophilic inflammation

EoE

eosinophilic esophagitis

eos/hpf

eosinophils per high-power field

EoEHSS

eosinophilic esophagitis histologic scoring system

EPX

eosinophil peroxidase

EREFS

endoscopic reference score

FC

fold change

H&E

hematoxylin and eosin

HPF

high-power field

i.p.

intraperitoneal

LPF

lamina propria fibrosis

MPO

myeloperoxidase

MT

Masson’s trichrome

O.D.

optical density

OFR

organoid formation rate

p(A)

poly(A)

sal

saline

SEA

surface epithelial alteration

SEM

standard error of the mean

SL

eosinophil surface layering

WT

wild type