Notch 2 signaling contributes to intestinal eosinophil adaptations in steady state and tissue burden following oral allergen challenge

Abstract

Eosinophils not only function as inflammatory effectors in allergic diseases, but also contribute to tissue homeostasis in steady state. Emerging data are revealing tissue eosinophils to be adaptive cells, imprinted by their local tissue microenvironment and exhibiting distinct functional phenotypes that may contribute to their homeostatic vs. inflammatory capacities. However, signaling pathways that regulate eosinophil tissue adaptations remain elusive. Notch signaling is an evolutionarily conserved pathway that mediates differential cell fate programming of both pre- and postmitotic immune cells. This study investigated a role for notch receptor 2 signaling in regulating eosinophil functions and tissue phenotype in both humans and mice. Notch 2 receptors were constitutively expressed and active in human blood eosinophils. Pharmacologic neutralization of notch 2 in ex vivo stimulated human eosinophils altered their activated transcriptome and prevented their cytokine-mediated survival. Genetic ablation of eosinophil-expressed notch 2 in mice diminished steady-state intestine-specific eosinophil adaptations and impaired their tissue retention in a food allergic response. In contrast, notch 2 had no effect on eosinophil phenotype or tissue inflammation within the context of allergic airways inflammation, suggesting that notch 2–dependent regulation of eosinophil phenotype and function is specific to the gut. These data reveal notch 2 signaling as a cell-intrinsic mechanism that contributes to eosinophil survival, function, and intestine-specific adaptations. The notch 2 pathway may represent a viable strategy to reprogram eosinophil functional phenotypes in gastrointestinal eosinophil-associated diseases.

Notch 2 signaling underlies eosinophil survival, function, and intestine-specific adaptations; targeting notch 2 may represent a viable strategy to reprogram eosinophil functional phenotypes in gastrointestinal eosinophil-associated diseases.

1. Introduction

Eosinophils are granulocytes that in healthy individuals home into tissue sites including the thymus, lung, adipose, and gastrointestinal (GI) tract, wherein they contribute to local and systemic immune, tissue, and metabolic homeostasis.^1–6 In contrast to their homeostatic functions, eosinophilia is a common clinical indicator of T helper 2 inflammatory diseases including allergic asthma and eosinophilic GI disorders, wherein infiltrating eosinophils are implicated in tissue damage, epithelial barrier disruption, goblet cell hyperplasia, and fibrosis.^7–11 Tissue adaptations shaped by local microenvironmental cues support functional diversity among eosinophil phenotypes^12–15 that may at least partially explain their dichotomous homeostatic vs. pathologic roles. This may be especially relevant within the GI tract, which is home to the highest density of eosinophils in health. To meet the challenges of balancing nutrient uptake while maintaining homeostasis at a host-microbial interface, resident GI immune cells, including eosinophils, exhibit unique adaptations.^16–18 Supported by the local cytokine milieu, intestinal eosinophils exhibit extended longevity relative to blood or lung eosinophils¹⁹ and are imprinted with tissue-specific transcriptional and functional adaptations.^14,15 Moreover, the intestinal eosinophil compartment is dynamic, responding to acute or chronic inflammatory insults with qualitative shifts in functional states.^15,20 Delineating cell-intrinsic mechanisms that regulate eosinophil tissue-adaptive programs is an unmet need, and important consideration in next-generation eosinophil-targeting therapeutic strategies nuanced to selectively eliminate pathologic, while sustaining homeostatic, eosinophil functional states.³

Activation of notch receptors drives evolutionarily conserved pathways that iteratively regulate cell-fate decisions throughout organogenesis and hematopoiesis.²¹ In mammals, notch signaling continues to shape cellular differentiation and functional plasticity of mature cells, including lymphocytes,^22–24 monocytes,²⁵ and dendritic cells.²⁶ Mature human and mouse eosinophils express notch receptors 1 and 2, along with 4 of the 5 notch ligands (i.e. jagged 1, jagged 2, DLL1, and DLL4).²⁷ Notch 1 is implicated in eosinophil motility in both human cells and mouse models.²⁸ Notch 2–dependent signaling in eosinophils had not yet been explored.

Here, we investigated a role(s) for notch 2 signaling in regulating core functions and tissue adaptations of eosinophils. In vitro, we determined transcriptional and functional impacts of pharmacologic inhibition of notch 2 in human blood eosinophils stimulated with interleukin (IL)-5 family cytokines. To explore functioning in vivo, we generated mice with eosinophil-targeted genetic deletion of notch 2 and evaluated lung and intestinal tissue eosinophils in steady state and within the context of allergic airway or intestinal inflammation.

2. Methods

2.1. Human eosinophil isolation

Studies involving human samples were approved by the Beth Israel Deaconess Medical Center Institutional Review Board, and informed consent obtained from all subjects. Eosinophils were isolated from peripheral blood of healthy donors by negative selection using an eosinophil enrichment cocktail (StemCell Technologies) and gravity flow through LD columns (Miltenyi Biotec) as described.²⁹ Purity (cytospin) and viability (trypan blue exclusion) were confirmed to be >98%.

2.2. Transcriptome analysis

Eosinophils from n = 4 donors were isolated from peripheral blood as described above. A total of 3 × 10⁶ eosinophils per condition were pretreated with anti-Notch 2 antibody (NRR2, gift from Chris Siebel, Genentech) or human IgG1 isotype control for 30 min, then left unstimulated or stimulated with IL-3 (1.5 ng/mL) for 4 h. Total RNA was isolated with QIAzol (QIAGEN) and filtered through a RNeasy column (QIAGEN) prior to quality analysis (TapeStation with Agilent Analysis Software A.02.02). Gene expression profiling was performed on the NanoString nCounter XT instrument using NanoString PanCancer Pathways 770 gene panels (XT-CSO-PATH1-1, NanoString Technologies). Data were analyzed using nSolver 4.0 software (MAN-C0011-04). Supplemental gene set enrichment analysis was performed with Metascape v3.5.20240101 (http://metascape.org).³⁰

2.3. Eosinophil in vitro survival

Eosinophils were isolated from peripheral blood as described previously and cultured overnight in RPMI complete media (RPMI 1640 supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin) alone or supplemented with 1.5 ng/mL of IL-3, IL-5, or granulocyte-macrophage colony-stimulating factor (GM-CSF), ± anti-Notch 2 neutralizing antibodies at the indicated concentrations. Following overnight culture, eosinophil viability was evaluated by flow cytometry using anti-Annexin V antibodies and propidium iodide dye. Viable eosinophils are reported as the frequency of AnnexinV^–PI^– cells.

2.4. Mice

Animal studies received prior approval from the Institutional Animal Care and Use Committees from Beth Israel Deaconess Medical Center or the University of Colorado School of Medicine. C57BL/6J, B6.129S-Notch2^tm3Grid and ROSA^mT/mG mice were originally obtained from the Jackson Laboratory and subsequently maintained in-house. eoCRE knock-in mice³¹ were a kind gift from Drs. James Lee and Elizabeth Jacobsen, Mayo Clinic Arizona. EoN2 mice were generated by crossing eoCRE to B6.129S-Notch2^tm3Grid. EoN2 mice exhibiting hemizygous Cre expression (i.e. EoCre^+/−N2^fl/fl) were used in experiments to ensure retention of unperturbed EPX expression from 1 allele. ROSA^mT/mG encodes a 2-color fluorescent Cre-reporter with membrane targeted expression. In the absence of Cre, tdTomato fluorescence expression is widespread in all cells. In the presence of Cre recombinase, GFP expression replaces tdTomato. Eosinophil GFP⁺ reporter mice were generated by crossing eoCRE or EoN2 mice to ROSA^mT/mG to generate Eo-mTmG and EoN2-mTmG lines, respectively.

2.5. Allergic airway inflammation

A total of 6 micrograms (protein weight) Dermatophagoides pteronyssinus house dust mite (HDM) extract (Greer Laboratories) was instilled intranasally in 35 microliters sterile phosphate-buffered saline (PBS) to anesthetized mice at time points indicated. Intranasal HDM was administered to anesthetized mice, which were held in the supine position for 10 s following inhalation. Mice were euthanized 4 d after the final intranasal challenge. Bronchoalveolar lavage was performed using 3 mL total volume of cold HBSS^−/− supplemented with 0.5% bovine serum albumin prior to excising a portion of the left lobe for formalin-fixed paraffin-embedded histology.

2.6. Allergic intestinal inflammation

Mice were sensitized by intraperitoneal injections on days 0 and 14 with 50 mg chicken egg ovalbumin (OVA) (grade VI; Sigma-Aldrich) mixed with 0.4 mg alum (Imject Alum; Thermo Fisher Scientific) in 100 mL sterile PBS. Mice received intragastric (i.g.) challenges 3 times per week beginning on day 21 and continuing for 1 to 3 wk, as indicated. Challenge inoculations consisted of 50 mg of OVA (grade V; Sigma-Aldrich) in 250 mL sterile PBS. Control mice were sensitized with 0.4 mg alum in PBS and challenged with PBS.

2.7. Cell recoveries

Eosinophil quantifications were determined by flow cytometry of whole blood obtained by cardiac punctures immediately after euthanasia. Bone marrow (BM) cells were recovered by flushing tibias and femurs. For analysis of lung tissue cells by flow cytometry, lungs were perfused with 10 mL cold PBS injected through the right ventricle prior to tissue excision and digestion. Excised lungs were minced, digested with 175 U/mL collagenase IV (Worthington Biomedical Corporation), mechanically disrupted, and passed through 40 micron filters to achieve single-cell suspensions as described.³² For intestinal tissue digests, whole small intestine (entire organ excised, from approximately 0.25 cm below stomach to 0.25 cm above cecum) was disaggregated as described.³³ Briefly, intestinal fragments were incubated in HBSS^−/− supplemented with 1 mM HEPES, 2.5 mM NaHCO₃, 1 mM DTT, and 10% fetal bovine serum at 37 °C with agitation, followed by vortexing to release the intraepithelial (IE) cell fraction. Tissue fragments were further incubated with EDTA-containing buffer to remove epithelial cells before digestion with 1 U/mL collagenase I. Digested tissue was passed through a 70 micron strainer to achieve a single-cell suspension and collected as the lamina propria (LP) fraction. Enriched viable IE and LP cells were collected from the interphase after centrifugation over a 44%:67% Percoll gradient. Intestinal lavage cells were recovered by flushing the whole length of the intact small intestine with 10 mL cold RPMI supplemented with 3% fetal bovine serum prior to digestion.

2.8. Tissue staining and imaging

For wild-type (WT) and EoN2 mice, eosinophils were quantified from FFPE tissues stained with antibodies against major basic protein (rat anti-mouse MBP, clone MT-14.7.3, from Dr. Elizabeth Jacobsen, Mayo Clinic Arizona) following antigen retrieval as described.³² Blinded slides encompassing whole longitudinal length (lung) or 5 to 6 circumferential rings (jejunum) were manually counted. For anatomic consistency, for all approaches jejunal rings were excised from the center of the small intestine. For immunofluorescence analysis of jejunal tissue from EoN2-mTmG reporter mice, frozen sections and FFPE blocks were generated from adjacent tissue rings. Frozen sections were mounted in DAPI-containing mounting medium and imaged directly. Due to loss of endogenous fluorescence during formalin fixation, FFPE sections were stained with anti-MBP antibodies following antigen retrieval as described previously, followed by Alexa Fluor 488–conjugated secondary antibody. Images were acquired using an Olympus IX83 microscope and CellSense software (Olympus).

2.9. Flow cytometry

CD45 (clone 30-F11) and Siglec F (clone E50-2440) were from BD Biosciences. Ly6g (clone RB6-8C5) was from Invitrogen. CD11c (clone N418), CD11b (clone M1/70), CLEC4a4 (clone 33D1), PDL1 (clone I0F.9G2), and CD80 (clone 16-10A1) were from BioLegend. LIVE/DEAD fixable aqua dead cell stain kit was from Invitrogen. Tissue eosinophils were identified as follows: from BM and blood, live, CD45 ⁺ SSC^int-hiSiglecF⁺; from lung tissue, CD45 ⁺ SSC^hiSiglecF ⁺ CD11c^−/lo; and from intestinal tissues, live, CD45 ⁺ SSC^hiFSC^lo-intSiglecF ⁺ CD11b ⁺ Ly6g⁻. For intracellular flow cytometry, cells were permeabilized by preincubating in fluorescence-activated cell sorting (FACS) buffer (HBSS^−/− + 0.5% bovine serum albumin) supplemented with 0.1% saponin for 5 min at 4 °C prior to incubation with staining antibodies in the continued presence of 0.1% saponin. Following staining, cells were washed in fresh saponin-containing FACS buffer to remove unbound intracellular antibodies, then stored in FACS buffer without saponin until analysis.

2.10. Mouse tissue RNA extraction and quantitative polymerase chain reaction

Whole tissue RNA was obtained from flash frozen tissue excised from the center of the jejunum using the RNeasy kit (Qiagen). Briefly, tissue was disrupted in RLT lysis buffer using a handheld homogenizer, and precipitated RNA extracted using RNeasy columns. Reverse transcription was performed using the high-capacity reverse transcription kit (Applied Biosystems). Quantitative polymerase chain reaction was performed using a CLDN1 TaqMan assay primer set (Thermo Fisher Scientific; Mm01342184_m1).

2.11. FITC-dextran permeability assays

Mice were fasted for 4 h, gavaged with 4 kDa FITC-dextran (Sigma-Aldrich), and sacrificed 2 h after gavage. Concentration of FITC-dextran in plasma collected by cardiac puncture was measured using a Synergy H1 Hybrid Reader and quantified by comparing with a standard curve.

2.12. Statistical analysis

All data points reflect biological replicates. Unpaired Student's t tests were used to compare the means between groups unless otherwise noted. Statistical analyses were performed using Microsoft Excel or GraphPad Prism (GraphPad Software, v10.1.1).

3. Results

3.1. Notch 2 mediates IL-5 family cytokine-dependent survival of human eosinophils

We previously reported constitutive expression of notch 2 receptors on freshly isolated human blood eosinophils.²⁷ Ligand engagement of notch receptors elicits sequential alpha and gamma secretase–dependent cleavages that generate intermediary receptor fragments before culminating in release of the notch intracellular domain, which functions as a transcriptional activator (Fig. 1A). A polyclonal antibody against notch 2 detected bands of expected sizes for the full-length receptor and cleavage fragments in lysates of freshly isolated blood eosinophils from healthy donors (Fig. 1B). These data suggest that notch 2 receptors are constitutively active in human blood eosinophils. Similar to localization of eosinophil-expressed notch 1,²⁷ a monoclonal antibody recognizing an extracellular epitope revealed a higher level of notch 2 expression intracellularly, relative to the cell surface, of blood eosinophils (Fig. 1C).

Fig. 1.

Notch 2 signaling mediates a cytokine-driven transcriptome and prolonged survival of human eosinophils in vitro. (A) Schematic illustrating notch 2 receptor cleavage fragments. (B) Western blot of postnuclear lysates of blood eosinophils from 4 independent donors, probed with polyclonal antibodies against notch 2. (C) Representative histogram from flow cytometry of nonpermeabilized (left) and saponin-permeabilized (right) blood eosinophils using a monoclonal antibody (Ab) recognizing a notch 2 extracellular domain. Shaded histogram, isotype control. (D) Relative expression of notch 2 messenger RNA in blood eosinophils following 4 h stimulation in medium alone (NS, non-stimulated) or 1.5 ng/mL of IL-3. Data compiled from n = 3 to 4 independent donors. (E) Heat map and (F) enriched ontology cluster analysis of top differentially expressed genes (DEGs) in IL-3–stimulated eosinophils pretreated with isotype control (Cntrl) or anti-notch 2 (aN2) neutralizing antibodies. Eosinophils isolated from n = 4 independent donors. (G) Dose response of aN2-dependent impairment of IL-3–mediated survival in overnight culture. (H) Viability of blood eosinophils following overnight culture in media supplemented with 1.5 ng/mL IL-3, IL-5, or GM-CSF, ± 2 mg/mL aN2. Eosinophils isolated from n = 4 independent donors. *P ≤ 0.05; paired t tests. Cartoon in panel A created with BioRender.com. FL = full length; M = marker; N1 = notch receptor 1; NICD = notch intracellular domain; NTM = notch transmembrane domain; 293 = HEK293T cell lysate.

We next investigated a role for notch 2 in eosinophil activation. The IL-5 family of cytokines (i.e. IL-5, GM-CSF, and IL-3) signal through the common beta chain to elicit largely overlapping responses that regulate eosinophil survival, priming, and activation.³⁴ Due to the centrality of this cytokine family to eosinophil expansion and activation, biologics targeting IL-5 signaling are at the forefront of therapeutic eosinophil-depletion strategies.³⁵ IL-3 exerts a relatively prolonged response compared with GM-CSF or IL-5, in part due to higher relative stability of the IL-3Ra chain³⁶; therefore, stimulation with IL-3 was chosen for initial studies. Compared with nonstimulated eosinophils, IL-3 augmented notch2 messenger RNA expression (Fig. 1D) and elicited an activated transcriptome indicative of a proinflammatory and prosurvival phenotype (Supplementary Fig. 1), including induction of JAK/STAT signaling regulators (e.g. SOCS1, SOCS2), upregulation of antiapoptotic genes (e.g. BCL2, BCL2L1), and downregulation of proapoptotic genes (e.g. FOXO4), within 4 h of stimulation. Of note, these data are similar to a previous report utilizing a longer (48 h) stimulation.³⁷ Pretreatment with a notch 2 inhibitor (NRR2) dysregulated the IL-3–induced transcriptome; the top 60 differentially expressed genes in IL-3 stimulated eosinophil cultures ± notch 2 inhibition are shown in Fig. 1E. Hierarchical clustering of differentially expressed genes revealed predicted protein interactive “neighborhoods” related to apoptosis, senescence, and autophagy. Likewise, pathways related to cell survival and apoptosis were among the strongest hits predicted by enriched ontology cluster analysis (Fig. 1F), suggesting that notch 2 regulates IL-3–induced eosinophil survival.

Building from the transcriptional findings, we directly tested a role for notch 2 in eosinophil survival. While IL-3 maintained >95% viability in overnight cultures, neutralization of notch 2 dose-dependently diminished the capacity of IL-3 to rescue eosinophils from apoptosis (Fig. 1G, H). Although notch receptor 1 is also constitutively expressed by human blood eosinophils²⁷ and mediates GM-CSF priming-induced migration,²⁸ regulation of IL-3–mediated survival was specific to notch 2, as a neutralizing antibody to notch 1 had no effect on eosinophil viability (Fig. 1G). Notch 2 regulation of eosinophil survival was not restricted to IL-3–mediated effects, as notch 2 neutralization similarly abrogated GM-CSF– and IL-5–mediated survival (Fig. 1H). Collectively, these data suggest that notch 2 signaling is active in circulating human eosinophils and contributes to their in vitro activation and survival downstream of stimulation with IL-5 family cytokines.

3.2. Notch 2 is dispensable for steady-state eosinophil development and tissue distribution in vivo

eoCRE mice express Cre recombinase downstream of the eosinophil peroxidase (EPX) promotor.³¹ To investigate a role for notch 2 in vivo, we crossed eoCRE mice with mice expressing loxP site–floxed notch2,³⁸ thereby generating a line of mice (hereafter EoN2) expressing notch 2^−/− eosinophils. To characterize the eosinophil compartments in these mice, we evaluated eosinophils from the blood, BM, lungs, and small intestinal tissues by flow cytometry (see Supplementary Fig. 2 for tissue eosinophil gating strategies). We confirmed absence of notch 2 expression on >95% of side scatter high SiglecF⁺ BM cells (Fig. 2A). The minor population of SiglecF ⁺ Notch2⁺ BM cells may represent a noneosinophil myeloid precursor,³⁹ immature SiglecF ⁺ EPX⁻ eosinophil precursors,⁴⁰ or immature EPX⁺ precursors exhibiting remnant notch 2 remaining from an earlier differentiation state.^41,42 The lower side scatter characteristic of this subpopulation (see Fig. 2Ai) indicative of immature granule development supports these latter interpretations. Importantly, blood eosinophils from EoN2 mice were uniformly notch 2 deficient (Fig. 2B), confirming that successful deletion of notch 2 in mature eosinophils egressed from BM into circulation.

Fig. 2.

Notch 2 signaling is dispensable for eosinophil development and basal tissue distribution in mice. Notch 2 expression on (A) BM and (B) blood eosinophils from WT and EoN2 mice. Shaded histograms indicate isotype control. In Ai, Notch 2 expression on SiglecF⁺ BM cells from EoN2 mice is re-displayed as a countour plot versus SSC. (C–F) Quantifications of eosinophils from (C) BM, (D) lung, (E) small intestine, and (F) blood from WT and EoN2 mice. In panels D and E, frequencies were determined by flow cytometry of digested tissues and total eosinophil numbers were counted from tissue sections. Each circle represents data from an individual mouse. Bars represent mean ± SD. *P ≤ 0.05; Student's t test. CVU = crypt-villus unit; Eos = eosinophils; SSC = side scatter.

EoN2 mice exhibited normal numbers and tissue distributions of eosinophils within BM (Fig. 2C), lung tissue (Fig. 2D), and small intestine (Fig. 2E). An increase in the basal frequency of blood eosinophils was observed in EoN2 mice (Fig. 2F). Of note, a similar trend was observed in lung tissue that did not reach statistical significance (see Fig. 2D). These data demonstrate that notch 2 is dispensable for steady-state eosinophil development and tissue homing to lung and gut.

3.3. Eosinophil-expressed notch 2 is not required for eosinophilic inflammation in allergic airways

In mice, eosinophils resident within the naïve lung actively dampen type 2 responses.⁶ In contrast, allergic airway inflammation is characterized by IL-5–dependent infiltration of inflammatory eosinophils that actively promote type 2 immunity. Inflammatory eosinophils are phenotypically distinguished from their basal tissue-resident counterparts by upregulation of Siglec F and induction of CD11c, aggregate around blood vessels and airways, and cross the epithelial barrier into bronchoalveolar spaces.^6,43 To investigate a role for notch 2 in eosinophil recruitment and activation within allergic airways, we subjected EoN2 mice and their WT littermates to repetitive intranasal administrations of HDM over the course of 3 wk (Fig. 3A). HDM inhalation elicits an inflammatory response characterized by high levels of IL-4, IL-5, and IL-13 that correlate with infiltration of T helper 2 lymphocytes and eosinophils into lung tissue and bronchoalveolar spaces, respectively.⁴⁴ Inhaled HDM elicited robust lung tissue inflammation (Fig. 3B) and bronchoalveolar lavage (BAL) eosinophilia (Fig. 3C) in both WT and EoN2 mice. Likewise, BAL eosinophils from both lines expressed equivalent levels of surface Siglec F and CD11c (Fig. 3D). Flow cytometry confirmed complete absence of notch 2 expression by BAL eosinophils from inflamed EoN2 mice (Fig. 3E). These data demonstrate that notch 2 is dispensable for eosinophil tissue recruitment, in situ induction of an inflammatory phenotype, and accumulation within allergic airways.

Fig. 3.

Notch 2 signaling is dispensable for induction of eosinophilic allergic airway inflammation. (A) Timeline for experimental HDM-induced allergic airway inflammation. (B) Hematoxylin and eosin staining of lung tissue from challenged WT and EoN2 mice. (C) Frequencies (left) and total numbers (right) of eosinophils within BAL fluid. (D) Geometric mean fluorescence intensities (gMFIs) of Siglec F and CD11c expression on BAL eosinophils from WT and EoN2 mice following allergen inhalation challenge. (E) Notch 2 expression by eosinophils recovered from the BAL fluid of HDM-challenged mice. Bars represent mean ± SD. i.n. = intranasal.

3.4. Notch 2 signaling contributes to small intestine–specific eosinophil imprinting

Although eosinophils emerge from BM as terminally differentiated cells, the intestinal microenvironment further imprints organ-specific adaptations, evidenced by distinct transcriptional and surface receptor profiles, compared with eosinophils within circulation or other organs.^{14,15,19,45–47} Gene Ontology analyses suggest enhanced notch signaling in GI eosinophils.¹⁴ Specifically, small intestinal (compared with BM) eosinophils exhibited greater than 10-fold increases in expression of notch1 and notch2, notch ligands DLL1 and DLL4, notch transcriptional regulators Rbpj and Maml, and the canonical notch target Hes1 (Supplementary Fig. 3, derived from GSE185070, associated with Diny et al.¹⁴ Notch 2 signaling is a tissue-specific determinant of cellular functional differentiation,^22–25 including splenic and intestinal dendritic cells.²⁶ In light of enhanced expression of notch 2–related signaling molecules in intestinal eosinophils and a recognized role for notch 2 in intestinal immune cell adaptations, we queried whether notch 2 might contribute to an eosinophil intestine-adapted phenotype.

We previously reported differential levels of expression of surface receptors, including CD11c and CD80, on LP and IE-associated intestinal eosinophils from naïve WT mice,⁴⁸ and demonstrated the utility of relative CD11c expression to track eosinophil versatility in situ.²⁰ Using single-cell transcriptomics, Gurtner et al.¹⁵ revealed a subset of “active” eosinophils with bactericidal activities that exhibit a distinct transcriptome and surface proteome characterized by dual expression of CD80 and PD-L1. Fate mapping in that study demonstrated in situ maturation of the “active” subset, whose expression appeared restricted to the GI tract. Finally, Wang et al.⁴⁶ reported a population of Clec4a4⁺ eosinophils unique to the GI tract that exhibit an immunoregulatory gene signature compared with Clec4a4⁻ eosinophils. We assessed each of these markers to evaluate imprinting of small intestinal eosinophils and found significantly lower frequencies of CD11c^int-hi (Fig. 4A, B), PD-L1 ⁺ CD80⁺ (Fig. 4A, C), and Clec4a4⁺ (Fig. 4A, D) cells among LP eosinophils in EoN2 mice compared with WT littermates. Litter- and cage-mate WT and EoN2 mice were compared directly due to known effects of vendor and rearing variables on eosinophil function.⁴⁹ To confirm that eosinophil repertoire alterations were due to loss of notch2, rather than to the eoCRE background, small intestinal eosinophils from the eoCRE^+/− parent line were compared with their WT littermates. Frequencies of small intestinal eosinophils expressing CD11c, CLEC4a4, or CD80 were not different between eoCRE^+/− and WT littermates (Supplementary Fig. 4). Unexpectedly, a non–statistically significant reduction in PD-L1–expressing small intestinal eosinophils was observed in eoCRE^+/− mice relative to their WT littermates, although loss of notch 2 lowered the frequency of PD-L1⁺ eosinophils even further (EoCre^+/− vs. EoN2, Fig. 4E). These data demonstrate that notch 2 signaling underlies aspects of the adapted phenotype(s) of intestinal eosinophils.

Fig. 4.

Notch 2 signaling partially underlies steady-state small intestine–specific eosinophil adaptations. (A) Histograms of CD11c, PD-L1, CD80, and CLEC4a4 expression on small intestinal eosinophils from WT and EoN2 mice. Solid gray histograms indicate isotype control. (B–D) Frequencies among small intestinal LP eosinophils of (B) CD11c^int-hi, (C) PD-L1 ⁺ CD80⁺, and (D) CLEC4a4⁺ cells. *P ≤ 0.05; Student's t test. (E) Direct comparison of frequencies of PD-L1⁺ small intestinal LP eosinophils from WT mice, the EoCre^+/− parent line, and EoN2 mice. *P ≤ 0.05, ****P ≤ 0.0001; 1-way analysis of variance with Tukey's multiple comparisons post hoc test. Bars are mean ± SD. ns = not significant.

3.5. Notch 2–deficient eosinophils fail to infiltrate villi and undergo rapid luminal egress following oral allergen challenge

Having demonstrated notch 2–dependent regulation of eosinophil intestinal imprinting, we next investigated effects of notch 2 deletion on their phenotype, localization, and function within the context of GI allergic inflammation. Mouse models of OVA-alum sensitization and oral OVA challenges (Fig. 5A) have been used to model intestinal eosinophilic inflammation^32,50,51 and as a preclinical model of eosinophilic gastroenteritis.⁵² Despite equivalent numbers of eosinophils at baseline (see Fig. 2E), eosinophil quantifications from jejunal tissue histology of EoN2 mice exhibited fewer eosinophils after repetitive oral allergen challenges compared with their WT littermates (Fig. 5B). Lower numbers of jejunum eosinophils in allergen-challenged EoN2 mice were entirely accounted for by the absence of villus-associated eosinophils, a discrepancy further exacerbated with additional oral allergen challenges (Fig. 5Bi, C, D). Therefore, loss of notch 2 signaling in eosinophils reduced their overall tissue burden following oral allergen challenge and prevented their allergen-driven aggregation within villi.

Fig. 5.

Notch 2–deficient eosinophils fail to aggregate within villi following oral allergen challenge. (A) Schematic of food allergen challenge model. (B) Total eosinophils (MBP⁺ cells) quantified from jejunal histology sections 1 d after 3, 6 or 9 i.g. OVA challenges are expressed as total eosinophils counted per crypt-villus unit. (Bi) Data from (B) re-expressed differentiating numbers of LP eosinophils within villi from those localized to LP at the level of the crypts. (C) Cartoon denoting how areas were demarcated in quantifying LP eosinophils within villi vs those at the level of the crypts. (D) Representative images of MBP⁺ eosinophils (DAB-Ni) within jejunal tissue from WT and EoN2 mice after 6 i.g. OVA challenges. Scale bar = 50 mM. Bars are mean ± SD. *P ≤ 0.05; Student's t test. d = day; Eos = eosinophils; i.p. = intraperitoneal.

Three factors work in concert to regulate numbers of tissue eosinophils within the context of allergic inflammation (i.e. recruitment, survival, and retention). Similar induction of peripheral eosinophilia (Fig. 6A) and equivalent numbers of eosinophils localized to LP surrounding crypts (quantified in Fig. 5Bi) suggest similar eosinophilopoiesis and recruitment, respectively, in allergen challenged WT and EoN2 mice. Histologic analysis of intestinal tissues after allergen challenge did not reveal overt evidence of eosinophil cytolytic cell death (i.e. deposition of cell-free granules was not observed) in either strain (Fig. 5D). Viability of small intestinal eosinophils immediately after recovery was >90% in both WT and EoN2 mice, with eosinophils from EoN2 mice exhibiting a statistically significant survival advantage over their WT counterparts (Fig. 6B). Taken together, we found no evidence for impaired viability of intestinal tissue eosinophils from EoN2 mice. Unexpectedly, within 24 h of allergen challenge, eosinophils were observed between epithelial cells in EoN2 (but not WT) mice (Fig. 6C). Flow cytometric analysis of IE leukocytes confirmed a transient spike in eosinophils 24 h after allergen challenge in EoN2 mice (Fig. 6D), and intestinal lavage fluid showed a coincident increase in luminal eosinophils (Fig. 6E), together suggestive of rapid transepithelial egress of tissue eosinophils. Intriguingly, eosinophil egress temporally aligned with heightened jejunal expression of CLDN1, encoding a tight junction protein (Supplementary Fig. 5A), which may suggest that transepithelial migration (TEM) of eosinophils shapes the local epithelial barrier by strengthening epithelial tight junctions, as was previously described for neutrophils.^53,54 In support of this interpretation, allergen-challenged EoN2 mice exhibited a modest improvement in overall epithelial barrier function compared with allergen-challenged WT littermates (Supplementary Fig. 5B). Collectively, these data demonstrate that notch 2 signaling prolongs retention of intestinal eosinophils within the context of allergic intestinal inflammation.

Fig. 6.

Intestinal eosinophils undergo rapid egress into the lumen following food allergen challenge of EoN2 mice. (A) Frequencies of blood eosinophils and (B) viability of LP eosinophils recovered from whole small intestinal preparations following food allergen challenge of WT or EoN2 mice. (C) Immunofluorescence images of villi within jejunum of EoN2 eosinophil reporter mice (Ci, longitudinal view of a crypt-villus unit; Cii, cross-section of a villus) reveal eosinophils (arrows) between epithelial cells of EoN2 mice 1 d after 6 i.g. allergen challenges. In Ci, FFPE sections were stained with anti-MBP followed by Alexa Fluor 488 secondary antibodies to label eosinophils (green) and mounted in DAPI-containing media to label nuclei (blue). In Cii, frozen sections were mounted in DAPI containing media and imaged directly to visualize Cre-expressing eosinophils (GFP⁺, green). Non–Cre-expressing cells express endogenous Tomato Red (red). Scale bar = 20 mM. (D) Quantification of IE eosinophils recovered from small intestines of WT or EoN2 mice at baseline, or 1 or 4 d after final allergen challenge. (E) Frequency of eosinophils among CD45⁺ leukocytes recovered from small intestinal lavages of EoN2 mice at baseline, or 1 or 4 d after final allergen challenge. *P < 0.05; #P < 0.05 vs. baseline. Eos = eosinophils; Vi = villus.

4. Discussion

Here, we demonstrated that intrinsic notch 2 signaling underlies IL-5 family cytokine-mediated survival and intestinal adaptations of human and mouse eosinophils, respectively. Blood eosinophils from healthy donors expressed receptor cleavage fragments suggestive of constitutive notch 2 activation in circulation. Ex vivo, notch 2 inhibition altered their cytokine-activated transcriptome and survival. In mice, genetic ablation of eosinophil-expressed notch 2 was associated with a lower proportion of intestine-resident eosinophils exhibiting tissue-specific adaptations. Downstream of chronic oral allergen exposure, notch 2^−/− eosinophils failed to accumulate within villi or form periepithelial clusters, but rather underwent early egress into the intestinal lumen, thereby lowering postchallenge eosinophil tissue burdens. Taken together, these data reveal, for the first time, notch 2 signaling as a cell-intrinsic modulator of eosinophil phenotype and function in humans and mice.

Our findings of normal steady-state numbers and tissue distributions of eosinophils in EoN2 mice suggest that notch 2 is dispensable for eosinophilopoiesis and basal tissue homing. This observation is intriguing because global blockade of notch signaling was previously shown to augment eosinophil development from human umbilical cord blood cells, and eosinophils derived under those conditions exhibited impaired chemotactic responses.⁴² Interpreted together, these data may suggest that signaling through notch 1, which remains intact in EoN2 mice, is sufficient to regulate normal eosinophil development and chemotaxis in vivo. However, the elevated frequency of blood eosinophils in EoN2 mice may point to a more complicated story, with overlapping additive and/or redundant roles for notch 1 and 2 in restraining eosinophilopoiesis. Further studies are needed to fully appreciate temporal effects of notch 2 ablation on eosinophil development vs. mature cell functions.

Because notch 2 was required for IL-5–mediated survival of human eosinophils ex vivo, it was surprising to us that loss of notch 2 had no effect on eosinophilia within allergic airways, despite induction of IL-5 in this HDM model.⁴⁴ These data may reflect discrepancies between mouse and human eosinophils, redundant or different survival mechanisms in vivo compared with the in vitro monocultures, or differences between transient neutralization in mature cells vs. genetic ablation of notch 2. Of note, small intestinal eosinophils recovered from EoN2 mice after oral allergen challenge exhibited a small, yet statistically significant, improvement in viability compared with their WT littermates (Fig. 6B). Future studies enabling temporal inhibition of eosinophil-expressed notch 2 in WT mice will be needed to parse effects of transient neutralization vs. sustained loss of notch 2 and fully unravel a potential role for notch in eosinophil survival.

Despite comparable total numbers of steady-state resident GI eosinophils, frequencies of eosinophils exhibiting markers of intestinal adaptations were significantly reduced in EoN2 compared with WT mice, suggesting that notch 2 underlies at least some aspects of intestinal imprinting. No phenotypic differences were observed between lung eosinophils from WT and EoN2 mice, suggesting that notch 2 regulation of tissue adaptations is intestine specific. This observation is reminiscent of notch 2–dependent shaping of intestinal dendritic cell functional adaptations.²⁶ Although notch 2 is expressed by eosinophils across tissue compartments, enrichment of notch transcriptional effectors (MAML, Rbpj) and gene targets (Hes1) in small intestinal eosinophils¹⁴ suggests that notch signaling might be uniquely relevant to shaping their fates within the GI tract. Moreover, notch ligands are differentially expressed throughout the GI tract on epithelial, endothelial, and immune cells and therefore may provide spatially compartmentalized cues.⁵⁵

Surprisingly, the eoCRE^+/− parent strain exhibited a reduced frequency of PD-L1⁺ eosinophils within the small intestine relative to WT littermates. The biological basis for this deficiency is unknown. The eoCRE line was generated through homologous recombination insertion of Cre recombinase into the open reading frame of Epx. Despite also inserting an internal ribosome entry site in an effort to achieve a dicistronic messenger RNA encoding both functional EPX and Cre, enzymatic activity and protein expression assays reveal expression of EPX is lost in homozygous, and diminished in hemizygous, eoCRE mice.³¹ Therefore, it cannot be ruled out that EPX might directly or indirectly contribute to PD-L1 expression by small intestinal eosinophils. Despite the basal reduction in PD-L1 expression in the parent strain, notch 2 deletion reduced PD-L1 expression even further on intestinal eosinophils (Fig. 4E).

Although our study is limited in that it does not dissociate basal defects in intestinal imprinting from allergen-driven eosinophil effector functions, EoN2 mice exhibited a reduced eosinophil tissue burden compared with WT littermates following repetitive food allergen challenges reflective of a selective absence of villus-associated eosinophils. Infiltration of eosinophils into villi and their subsequent proximation to villus and crypt epithelium is considered a pathologic feature of eosinophilic GI disorders,⁵⁶ and murine eosinophils localized to the luminal third of villi are enriched for an inflammatory phenotype.^15,57 Therefore, the selective absence of villus eosinophils in allergen-challenged EoN2 mice would likely be a therapeutically desirable outcome. Notch 2–dependent regulation of mast cell and basophil tissue localizations and function within inflamed intestines have also been reported.^58,59 These data collectively suggest therapeutic interventions targeting notch 2 signaling may elicit multipronged effects that simultaneously regulate eosinophils, basophils, and mast cells, thereby diminishing their collective effector functions in intestinal allergic inflammation.

In contrast to WT littermates, intestinal eosinophils from EoN2 mice underwent rapid egress into the gut lumen following allergen ingestion. We previously reported TEM of a (comparatively) smaller population of CD11c^−/loCD11b^hi eosinophils in WT mice 4 d after allergen challenge (also seen here in Fig. 6D, day 4).²⁰ Absence of CD11c distinguished these cells from the resident pool of CD11c^hiCD11b^hi IE-associated eosinophils⁴⁸ and may imply that TEM is a default pathway for eosinophils that fail to fully adapt. This interpretation aligns with our data here, wherein defective eosinophil intestinal adaptations in EoN2 mice (including impaired CD11c expression, see Fig. 4A) are accompanied by their exacerbated and rapid exodus. In this paradigm, whether CD11c itself might function as a “retention signal” through interactions with extracellular matrix or basement membrane ligands, or its absence is simply an indicator of failed adaptation, remains to be determined.

Further studies are also needed to unravel the biological consequences of eosinophil TEM and their subsequent presence within gut lumen. Temporal coincidence of eosinophil egress with CLDN1 upregulation and lower permeability tempts speculation that eosinophil TEM contributes to a stronger epithelial barrier within localized microdomains, as previously reported for transmigrating neutrophils via stabilization of hypoxia-inducible factor.⁵³ Thus, TEM may represent a resolving function of GI eosinophils within the context of allergic diseases. It is also unclear if or for how long eosinophils remain within the luminal space, or what function, if any, they provide there. The relatively low number and high viability of intestinal lavage eosinophils in our model argue against in situ cytolytic degranulation and/or extrusion of DNA traps, as has been described for activated granulocytes accumulated within the colonic lumen in ulcerative colitis.⁶⁰

In summary, this study uncovers notch 2 as an intrinsic regulator of eosinophil function and phenotype in both humans and mice. Our finding that notch 2 underlies tissue-specific adaptations and allergen-driven functions of intestinal eosinophils provides a valuable new tool in the delineation of organ-specific adaptations and new insights into molecular underpinnings of eosinophil specialization and function in health and disease. Moreover, these findings may support consideration of notch 2 signaling as a therapeutic target to reprogram eosinophil phenotypes in primary eosinophil-associated GI diseases.

Author contributions

S.A.S. and C.L.O. contributed to study conceptualization, investigation, data curation, and manuscript writing. L.D.L. contributed to investigation and manuscript review and editing. E.H., L.L., and K.K. contributed to investigation. L.A.S. conceptualized and oversaw the study and contributed to data curation and manuscript writing.

Supplementary material

Supplementary materials are available at Journal of Leukocyte Biology online.

Funding

This work was supported by National Institutes of Health R01AI168134 and a Senior Research Award from the Crohn's and Colitis Foundation, both to L.A.S.