Location of eosinophils in the airway wall is critical for specific features of airway hyperresponsiveness and T2 inflammation in asthma

Abstract

Background

Eosinophils are implicated as effector cells in asthma, but the functional implications of the precise location of eosinophils in the airway wall is poorly understood. We aimed to quantify eosinophils in the different compartments of the airway wall and associate these findings with clinical features of asthma and markers of airway inflammation.

Methods

In this cross-sectional study, we utilised design-based stereology to accurately partition the numerical density of eosinophils in both the epithelial compartment and the subepithelial space (airway wall area below the basal lamina including the submucosa) in individuals with and without asthma and related these findings to airway hyperresponsiveness (AHR) and features of airway inflammation.

Results

Intraepithelial eosinophils were linked to the presence of asthma and endogenous AHR, the type that is most specific for asthma. In contrast, both intraepithelial and subepithelial eosinophils were associated with type 2 (T2) inflammation, with the strongest association between IL5 expression and intraepithelial eosinophils. Eosinophil infiltration of the airway wall was linked to a specific mast cell phenotype that has been described in asthma. We found that interleukin (IL)-33 and IL-5 additively increased cysteinyl leukotriene (CysLT) production by eosinophils and that the CysLT LTC₄ along with IL-33 increased IL13 expression in mast cells and altered their protease profile.

Conclusions

We conclude that intraepithelial eosinophils are associated with endogenous AHR and T2 inflammation and may interact with intraepithelial mast cells via CysLTs to regulate airway inflammation.

Shareable abstract (@ERSpublications)

Intraepithelial eosinophils are a specific feature of asthma, associated with endogenous airway hyperresponsiveness and type 2 inflammation, and may interact with intraepithelial mast cells via cysteinyl leukotrienes to regulate airway inflammation https://bit.ly/3e1QFo1

Introduction

Eosinophils are implicated in asthma pathogenesis [1–3] and have been associated with type 2 (T2) inflammation based on correlations with immunostaining for eosinophils in the airway tissue, induced sputum and peripheral blood [4, 5]. However, no prior study has used quantitative morphometry such as design-based stereology to precisely quantify the density of eosinophils within the different compartments of the airway wall [6]. Eosinophil location within the airway wall may influence asthma pathogenesis and regulation of T2 inflammation. Intraepithelial eosinophils may be of particular interest, since the epithelium produces specific cytokines that serve to modulate innate immune cell function and regulate T2 inflammation [7, 8]. Furthermore, recent studies have demonstrated increased numbers of mast cells within the airway epithelium in asthma [9, 10], but little is known about the relationship between mast cells and eosinophils in the airway epithelium.

Design-based stereology, a form of quantitative morphometry, avoids many biases inherent to two-dimensional methods and is currently regarded as the gold standard for accurate localisation and determination of the numerical density of cells per volume of a reference space [6]. Studies using design-based stereology have revealed differences in the number of airway smooth muscle cells [11], the density of goblet cells [12], thickness of the basement membrane [13] and intraepithelial mast cell density [10].

We recently used design-based stereology to demonstrate a shift in mast cells from the lamina propria and submucosal space (henceforth described as subepithelial space) to the epithelium in subjects with asthma and identified associations between this shift in mast cell location, T2 inflammation and airway hyperresponsiveness (AHR) [9]. Here, we used the same cohort of patients and the same methodology to precisely localise and enumerate eosinophils in the epithelial and subepithelial compartments and relate the numerical density of eosinophils in the different compartments to features of asthma, AHR and T2 inflammation. This patient cohort is ideal for studying clinical features of asthma and airway inflammation as these individuals were characterised for distinct forms of AHR, including direct AHR, which is sensitive but nonspecific for asthma, and indirect/endogenous AHR, which is specific for asthma, but less sensitive overall [14, 15]. Moreover, these patients were not using controller therapies at the time of sample collection or phenotypic characterisation. Additionally, we modelled the T2 microenvironment of the asthmatic epithelium to examine interleukin (IL)-33 and IL-5 effects on eosinophil function and we further explored the influence of mediators derived from the epithelium and eosinophils on mast cell phenotype and T2 gene expression.

Methods

A comprehensive description of the methods is available in the supplementary material. The University of Washington (Seattle, WA, USA) institutional review board approved the study and participants provided written informed consent.

Participants and study design

We conducted a cross-sectional study to examine the association between the numerical density of eosinophils in the different airway wall compartments (epithelial and subepithelial spaces) and features of asthma including AHR and T2 airway inflammation. The subepithelial space included the airway wall area extending below the basal lamina including the lamina propria and the submucosal space. We used endobronchial biopsy samples from a repository collected at the University of Washington designed to examine differences between mild to moderate asthmatics with and without exercise-induced bronchoconstriction (EIB), and non-asthmatic controls [16]. Participants underwent assessment of direct AHR via methacholine challenge testing and endogenous AHR in the form of exercise-induced bronchoconstriction (EIB) via dry-air exercise challenge. All asthmatics had a positive methacholine challenge. The severity of EIB was quantified by the forced expiratory volume in 1 s (FEV₁) over 30 min after exercise (called the area under the FEV₁ time curve) and by the maximum fall in FEV₁ after exercise challenge (max fall in FEV₁). Healthy controls had negative methacholine and exercise challenge tests. Induced sputum and research bronchoscopy with endobronchial biopsies were conducted on separate days [17]. Subjects with asthma who were receiving controller therapies (e.g. an inhaled corticosteroid) discontinued such therapy for ⩾2 weeks before any study procedures or sample collections.

Immunohistochemistry and design-based stereology

Murine monoclonal anti-eosinophil peroxidase (EPX; clone MM25–82.2) was used to localise eosinophils in endobronchial tissue. The physical dissector method (design-based stereology) was conducted, as described previously [9, 10]. We used two complementary metrics; the primary metric was the numerical density of eosinophils per reference volume and the secondary metric was the numerical density of eosinophils per surface area of the basal lamina, which enumerates the eosinophils located above or below the basal lamina. The secondary metric aimed to control for potential confounding related to changes in the reference volume associated with the presence of asthma. Supplementary figure E1 provides an overview of the stereology methods.

Quantitative PCR

TaqMan-based PCR of induced sputum cells [9, 18] quantified the expression of selected genes relevant to airway inflammation; IL4, IL5, IL13, CMA1 (chymase), TPSAB1 (tryptase) and CPA3 (carboxypeptidase A3).

Isolation and assessment of cysteinyl leukotriene formation by peripheral blood eosinophils

Blood samples were obtained from individuals with a physician diagnosis of asthma and/or allergic rhinoconjunctivitis. Granulocytes were isolated from peripheral blood by density gradient centrifugation followed by negative immunomagnetic selection of eosinophils. The eosinophils were treated with human IL-5 and/or IL-33 for 20 min followed by immediate measurement of cysteinyl leukotriene C₄ (LTC₄) levels by ELISA.

Assessment of the effects of cysteinyl leukotrienes and IL-33 on mast cell T2 gene expression and protease phenotype

Laboratory of Allergic Disease-2 (LAD2) mast cells [19] were treated with human IL-33 and/or LTC₄ for 4 h and in some conditions, a cysteinyl leukotriene (CysLT)₁ receptor antagonist (LT₁RA) MK571 (5 μM) and/or a CysLT₂ receptor antagonist (LT₂RA) HAMI3379 (5 μM) (Cayman Chemical, Ann Arbor, MI, USA) were added before treatment. TaqMan-based PCR analysis quantified the expression of IL13, CMA1, TPSAB1 and CPA3 genes relative to the endogenous control gene.

Statistical analysis

Statistical methods are described in figure legends and detailed in the supplementary materials.

Results

Intraepithelial eosinophils are a distinctive feature of asthma and endogenous AHR

Adequate endobronchial biopsy samples for stereology were available from 10 healthy controls, 12 EIB-negative asthmatics (EIB⁻), and 18 EIB-positive asthmatics (EIB⁺). Study participant characteristics are summarised in table 1. Using stereology to evaluate the airway wall overall, we identified a modest increase in the density of eosinophils in the airway wall in subjects with asthma compared to healthy controls that did not reach statistical significance (p=0.08). However, subjects with asthma had a higher eosinophil density in the epithelium than healthy controls (figure 1a). In contrast, there was no significant difference in the subepithelial eosinophil density between individuals with asthma and healthy controls (figure 1c).

TABLE 1.

Baseline characteristics of the study groups

	Control	Individuals with asthma		p-value
		EIB−	EIB+	All three groups	EIB+ versus EIB−
Subjects	10	12	18
Age (years)	30.4±12.7	26.4±8.7	24.8±4.95	0.34	0.64
Gender: male	2 (20)	3 (25)	6 (33.3)	0.62	0.63
Ethnicity: white	7 (70)	11 (91.7)	15 (83.3)	0.41	0.51
BMI (kg·m−2)	22±2.5	22.4±2.7	24±3.8	0.22	0.37
FEV1 (% pred)	96.5±11.3	90.7±9.75	89±11.2	0.21	0.90
FVC (% pred)	95.7±13.5	96.2±8.9	103.5±9.9	0.09	0.16
FEV 1 /FVC ratio	0.87±0.06	0.80±0.09	0.73±0.09	<0.001	0.03
Methacholine PC 20	>8±0	1.8±1.3	0.6±1.5	<0.001	0.04
Exercise challenge
Max fall in FEV1 (%)	1.7±2.1	2.3±2.6	27±9.2	<0.001	<0.001
AUC30	−7.4±58.1	−13.4±69.5	602.5±287.5	<0.001	<0.001
T2GM	−0.35±0.18	−0.56±0.89	0.49±0.97	0.01	0.02
Sputum eosinophil count (cells·mL−1)	16960±19495	9880±7637	24586±33912	0.30	0.27
Sputum eosinophils (%)	5.41±10	1.98±1.77	2.25±2.01	0.24	0.99

Data are presented as n, mean±SD or n (%), unless otherwise stated. EIB: exercise-induced bronchoconstriction; BMI: body mass index; FEV₁: forced expiratory volume in 1 s; FVC: forced vital capacity; PC₂₀: provocative concentration causing a 20% fall in FEV₁; AUC₃₀: area under the FEV₁ time curve; T2GM: T2 gene mean.

FIGURE 1.

Intraepithelial eosinophils, but not subepithelial eosinophils, are increased in individuals with asthma with endogenous airway hyperresponsiveness. a) Intraepithelial eosinophils were higher in individuals with asthma compared to healthy controls (Ctrl); b) intraepithelial eosinophils are increased in subjects with exercise-induced bronchoconstriction (EIB) compared to healthy controls; c) subepithelial eosinophils were present in healthy controls and there was no statistically significant difference between healthy controls and individuals with asthma; d) subepithelial eosinophils were not different between EIB⁺ and healthy controls. Significance was assessed using the Mann–Whitney U-test (two-group) or the Kruskal–Wallis test with multiple comparisons (three-group).

Although there were no significant associations between eosinophil location and baseline lung function, individuals with asthma and eosinophil infiltration of their airway epithelium tended to have lower FEV₁ and lower FEV₁/forced vital capacity (FVC) ratio compared to those without eosinophil infiltration of their airway epithelium (p=0.06 and p=0.24, respectively). Further refinement of the asthma group into subjects with and without EIB demonstrated increased intraepithelial eosinophils in EIB⁺ asthmatics compared to healthy controls, but no differences in intraepithelial eosinophils between EIB⁻ asthmatics and healthy controls, or between EIB⁺ and EIB⁻ groups (figure 1b). While there was a trend towards higher subepithelial eosinophils in EIB⁺ asthmatics when compared to healthy controls or EIB⁻ asthmatics, results were not statistically significant (figure 1d). Additional analyses correlating the numerical eosinophil densities in the epithelial and subepithelial compartments and the severity of direct AHR (assessed by the provocative concentration of methacholine causing a 20% fall in FEV₁) and endogenous AHR (assessed by the max fall in FEV₁) are presented in table 2. When healthy controls were excluded from the regression analysis, neither intraepithelial nor subepithelial eosinophil densities were associated with the severity of direct AHR; however, we observed a consistent association between intraepithelial eosinophil density and the severity of endogenous AHR in the form of EIB (table 2). The relationship between intraepithelial eosinophil density and severity of EIB was stronger for our secondary metric that assesses the number of eosinophils located within the epithelium above the basal lamina, irrespective of the epithelial volume. Collectively, these results indicate that intraepithelial eosinophils are a specific feature of asthma and are related to the presence and severity of endogenous AHR in the form of EIB. Representative two-dimensional images of eosinophils in the airway wall from each of the groups are shown in figure 2.

TABLE 2.

Relationship between the location of eosinophils within the airway wall and features of airway hyperresponsiveness^#

	Methacholine PC20				Severity of EIB (AUC30)				Max fall in FEV1
	All subjects		Asthma group		All subjects		Asthma group		All subjects		Asthma group
	r2	p-value	r2	p-value	r2	p-value	r2	p-value	r2	p-value	r2	p-value
Intraepithelial eosinophils
Volume (cells·mm−3)	0.18	<0.01	0.06	0.19	0.17	0.01	0.10	0.09	0.21	<0.01	0.13	0.06
Basal lamina (×102 cells·mm−2)	0.11	0.04	0.05	0.20	0.27	<0.01	0.21	0.01	0.28	<0.01	0.22	0.01
Subepithelial eosinophils
Volume (cells·mm−3)	0.11	0.03	0.07	0.15	0.06	0.11	0.02	0.40	0.12	0.03	0.07	0.16
Basal lamina (×102 cells·mm−2)	0.08	0.06	0.06	0.17	0.03	0.22	0.01	0.52	0.03	0.22	0.01	0.50

Bold type represents statistical significance. PC₂₀: provocative concentration causing a 20% fall in forced expiratory volume in 1 s (FEV₁); EIB: exercise-induced bronchoconstriction; AUC₃₀: area under the FEV₁ time curve.

^#:assessed using a simple linear regression with the methacholine PC₂₀, the severity of EIB over the first 30 min after exercise challenge (AUC₃₀), and the max fall in FEV₁ representing the dependent variables.

FIGURE 2.

Representative two-dimensional images of eosinophils within the airway wall. Murine monoclonal anti-eosinophil peroxidase was used to localise eosinophils in endobronchial tissue from a) healthy controls, b) asthma patients without exercise-induced bronchoconstriction (EIB⁻) and c) asthma patients with EIB (EIB⁺). d) A 100× image of an intraepithelial eosinophil. Scale bars=50 μm.

T2 inflammation is associated with eosinophils in both the epithelial and subepithelial compartments

To examine the relationship between airway eosinophil location and T2 inflammation, we used the T2 gene mean (T2GM) as a surrogate for T2 inflammation. T2GM combines the expression of IL4, IL5 and IL13 in induced sputum cells into a single metric. Subjects were categorised as T2-high if their T2GM was two standard deviations above that of healthy controls within our study population [9, 18]. Across the full study population, T2-high individuals had higher densities of eosinophils in the airway wall overall, the subepithelial and the epithelial compartments when compared to T2-low individuals (figure 3a–c). Similar results were obtained when only participants with asthma were analysed (figure 3d–f).

FIGURE 3.

Intraepithelial and subepithelial eosinophils are increased among T2-high subjects. Across the full study population, T2-high individuals had a significantly higher numerical density of eosinophils in a) the total airway wall, b) the subepithelial space and c) the intraepithelial compartment. Among individuals with asthma, T2-high individuals had a significantly higher density of eosinophils in d) the airway wall, e) the subepithelial space and f) the intraepithelial compartment. Among the T2 genes, the expression of h) the interleukin (IL)5 gene in the airways was most closely related to the presence of intraepithelial eosinophils, while the differences were not as strong for g) IL4 and i) IL13 expression. Significance was assessed using the Mann–Whitney U-test.

In relation to the expression of individual T2 genes, subepithelial eosinophil density was associated with IL4, IL5 and IL13 expression in induced sputum cells, while intraepithelial eosinophil density was only significantly associated with IL5 gene expression (table 3). To further delineate this relationship, we assessed differences between T2 genes and individuals with asthma based on the presence of intraepithelial eosinophils. Asthmatics with intraepithelial eosinophils had higher IL5 gene expression (figure 3h) in induced sputum when compared to those without intraepithelial eosinophils. The expression of IL4 and IL13 were not significantly different between these two groups (figures 3g and i).

TABLE 3.

Relationship between the location of eosinophils within airway wall and type-2 gene expression in induced sputum cells^#

	Full study population				Individuals with asthma
	Intraepithelial eosinophil volume		Subepithelial eosinophil volume		Intraepithelial eosinophil volume		Subepithelial eosinophil volume
	r2	p-value	r2	p-value	r2	p-value	r2	p-value
T2GM	0.13	0.07	0.25	0.008	0.10	0.17	0.23	0.03
IL4 gene	0.06	0.21	0.22	0.01	0.07	0.26	0.25	0.02
IL5 gene	0.15	0.05	0.24	0.01	0.11	0.15	0.22	0.04
IL13 gene	0.11	0.09	0.15	0.04	0.07	0.24	0.13	0.13
Intraepithelial mast cells (cells·mm−3)	0.09	0.08	0.08	0.09	0.05	0.28	0.05	0.24
TPSAB1 gene	0.14	0.06	0.21	0.02	0.17	0.07	0.27	0.02
CPA3 gene	0.18	0.03	0.19	0.02	0.17	0.07	0.20	0.05
CMA1 gene	0.007	0.68	0.03	0.42	0.008	0.72	0.05	0.36
Sputum eosinophils (cells·mL−1)	0.002	0.75	0.60	<0.001	0.004	0.73	0.69	<0.001

T2GM: T2 gene mean; IL: interleukin; TPSAB1: tryptase; CPA3: carboxypeptidase A3; CMA1: chymase.

^#:assessed using a simple linear regression with the intraepithelial eosinophil volume and subepithelial eosinophil volume representing the independent variables.

Induced sputum eosinophils are not associated with intraepithelial eosinophils

Sputum eosinophilia has been considered a surrogate marker for airway tissue eosinophilia in subjects with asthma. As we conducted three-dimensional quantitative analysis and partitioned eosinophils into their precise location, we were able to refine this relationship. We found that sputum eosinophil concentration was associated with the subepithelial eosinophil density and not with intraepithelial eosinophil density (table 3). This relationship between the subepithelial eosinophil density and sputum eosinophil concentration persisted when the analysis was limited to participants with asthma (table 3).

Airway eosinophilia is associated with expression of specific mast cell genes in the airway epithelium

Previous studies have identified a unique subpopulation of mast cells residing in the airway epithelium of subjects with asthma that is associated with AHR and T2 inflammation. This subpopulation of mast cells has higher expression of carboxypeptidase A3 (CPA3), persistent expression of tryptase (TPSAB1), and low expression of chymase (CMA1) [9, 20]. Here, we assessed the association between airway eosinophil location and intraepithelial mast cells and sputum mast cell protease expression. We observed no association between the numerical densities of intraepithelial or subepithelial eosinophils and intraepithelial mast cells (table 3); however, we noted a distinctive relationship between both subepithelial and intraepithelial eosinophils and the expression of mast cell-specific protease genes in induced sputum. The numerical density of subepithelial eosinophils was significantly associated with expression of both TPSAB1 and CPA3 but not with CMA1 (table 3). Intraepithelial eosinophil density was associated with CPA3 expression, and to a lesser extent with TPSAB1 expression but not with CMA1 expression (table 3). When individuals with asthma were analysed separately, a consistent association between subepithelial eosinophil density and the expression of the TPSAB1 and CPA3 was observed, and there was less of an association between the density of intraepithelial eosinophils and mast cell-specific protease genes (table 3). However, the median TPSAB1 and CPA3 expression was generally higher among asthmatics with intraepithelial eosinophils when compared to those without intraepithelial eosinophils (figure 4a and b). This relationship did not hold true for CMA1 (figure 4c).

FIGURE 4.

Expression of a specific set of mast cell genes was associated with the presence of intraepithelial eosinophils. a) Tryptase (TPSAB1) gene expression was significantly elevated in individuals with asthma and intraepithelial eosinophils compared to those without intraepithelial eosinophils. b) Carboxypeptidase A3 (CPA3) gene expression was significantly higher in individuals with asthma and intraepithelial eosinophils, but c) chymase (CMA1) gene expression was not different between subjects with asthma, with or without intraepithelial eosinophils. Statistical significance was assessed using the Mann–Whitney U-test.

IL-33 and IL-5 have an additive effect on eosinophil production of cysteinyl leukotrienes

We modelled the effects of the asthmatic airway epithelial microenvironment on eosinophil production of LTC₄, representing new CysLT formation. Eosinophils within the airway epithelium are exposed to epithelial-derived cytokines such as IL-33 and to T2 cytokines including IL-5. Consistent with prior work [21, 22], IL-5 caused a modest but significant increase in eosinophil production of LTC₄ (p=0.02). IL-33 caused a dose-dependent increase in LTC₄ production (figure 5a). Together, these two cytokines showed an additive effect on LTC₄ production by eosinophils (figure 5b and c).

FIGURE 5.

Cysteinyl leukotriene C₄ (LTC₄) production by peripheral blood eosinophils is augmented by the combined effects of interleukin (IL)-33 and IL-5 cytokines. a) IL-33 caused a dose-dependent increase in LTC₄ production by peripheral blood eosinophils leading to a marked increase in the level of LTC₄ at the highest concentration (n=4 per condition). b) IL-5 at a concentration of 10 ng·mL⁻¹ increased the production of LTC₄ from peripheral blood eosinophils pre-treated with a low dose (1 ng·mL⁻¹) of IL-33 (n=4 per condition). c) A higher dose of IL-5 caused a marked increase in the LTC₄ production from peripheral blood eosinophils pre-treated with a higher dose of IL-33 (10 ng·mL⁻¹) (n=4 per condition). Mean values and sem bars are shown. Differences between multiple conditions were assessed by one-way ANOVA with multiple comparisons with p-values adjusted to control for false discovery rate using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli.

Cysteinyl leukotrienes act in concert with IL-33 to regulate mast cell phenotype and T2 gene expression

Our prior work demonstrated that IL-33 induces a strong T2 signal in mast cells and that T2 gene expression is associated with reduced CMA1 expression in mast cells [9]. We hypothesised that IL-33 would act along with CysLTs generated by eosinophils and/or mast cells in this local environment to regulate mast cell expression of IL13. There was little induction of IL13 expression in LAD2 mast cells after exposure to a lower concentration of IL-33 (1 ng·mL⁻¹) in comparison to a higher IL-33 concentration (10 ng·mL⁻¹) (figure 6a). LTC₄ alone caused a modest induction of IL13 expression in mast cells (figure 6b). Importantly, a low concentration of IL-33 combined with LTC₄ increased mast cell IL13 expression (figure 6b). Using receptor antagonists, we demonstrated that the induction of IL13 expression by treatment with both IL-33 and LTC₄ was largely dependent upon the CysLT₁ receptor, but not the CysLT₂ receptor (figure 6c).

FIGURE 6.

Cysteinyl leukotriene C₄ (LTC₄) and interleukin (IL)-33 coregulate IL-13 and granular proteases expression in LAD2 mast cells. a) IL-33 resulted in a modest increase in IL13 gene expression in mast cells at the highest concentration (n=3 per condition). b) LTC₄ alone caused a modest increase in IL13 gene expression in mast cells, but the combination of LTC₄ with a low concentration of IL-33 resulted in a marked upregulation of IL13 gene expression in mast cells (n=3 per condition). c) The combined effect of LTC₄ and IL-33 on IL13 gene expression was abrogated when mast cells were pre-treated with a cysteinyl leukotriene 1 receptor antagonist (LT₁RA), but not by a cysteinyl leukotriene 2 receptor antagonist (LT₂RA) (n=3 per condition). d–f) LTC₄ in combination with IL-33 reduced the expression of mast cell granular proteases d) tryptase (TPSAB1), e) carboxypeptidase A3 (CPA3) and f) chymase (CMA1), with the most pronounced effects on CMA1 where the combination resulted in undetectable levels of this gene (n=3–4 per condition). g) The suppressive effect of LTC₄/IL-33 on CMA1 expression was partially blocked with the LT₁RA, but not the LT₂RA. h and i) The effects of LTC₄ on CMA1 expression were blocked entirely by the LT₁RA, but not the LT₂RA, while the effect of IL-33 was partially blocked by the LT₁RA (n=3–4 per condition). Mean values and sem bars are shown. Differences between multiple conditions were assessed by one-way ANOVA with multiple comparisons with p-values adjusted to control for false discovery rate using the two-stage linear step-up procedure of Benjamini, Krieger and Yekutieli.

Next, we examined the effects of IL-33 and CysLTs on expression of mast cell proteases. Either IL-33 or LTC₄ alone reduced mast cell expression of TPSAB1 and CPA3 and there was an additive effect on protease expression when IL-33 was combined with LTC₄. This effect was most pronounced for the expression of CMA1, which was reduced to undetectable levels in these cells (figure 6d–f) and was partially reversed by a CysLT₁ receptor antagonist (LT₁RA), but not by a CysLT₂ receptor antagonist (LT₂RA) (figure 6g). Furthermore, the effect of LTC₄ was completely inhibited by the LT₁RA, but not by the LT₂RA (figure 6h), while only a portion of the inhibitory effect of IL-33 on CMA1 expression was inhibited by the LT₁RA (figure 6i). The effects of IL-33 and LTC₄ on TPSAB1 and CPA3 expression were less pronounced and have more complex regulation including effects mediated through both CysLT receptors (supplementary figure E2). Collectively, these results indicate that much of the effect of IL-33 and LTC₄ on CMA1 expression is through LTC₄ and its metabolite LTD₄ binding to CysLT₁ receptor, and that IL-33 triggers LTC₄ release from mast cells. However, there is a component of suppression of CMA1 in mast cells by IL-33 that is independent of LTC₄ generation and autocrine stimulation of the CysLT₁ receptor.

Discussion

This is the first study to use a three-dimensional form of quantitative morphometry to precisely enumerate eosinophils in the different airway wall compartments and relate these findings to clinical features of asthma, T2 inflammation and induced sputum eosinophil concentration using a population of subjects characterised for AHR in the absence of controller therapies. We demonstrate that intraepithelial eosinophils are unique to asthma, associate with the presence of endogenous AHR and IL5 gene expression, and contribute to the regulation of mast cell phenotype and T2 inflammation.

Our assessment of the overall eosinophil density in the airway wall identified eosinophils in subjects with asthma and healthy controls with no statistical difference in eosinophil density between these groups. As prior studies have associated the number of eosinophils per area of the airway wall using two-dimensional methods [1, 23, 24], our study suggests that these results might have overestimated differences in the number of subepithelial eosinophils of asthmatics. By partitioning the density of eosinophils per area of reference volume into epithelial and subepithelial spaces, we demonstrated that eosinophils within the epithelium, but not the subepithelial space, were strongly linked to the presence of asthma. This is consistent with an earlier cross-sectional study that examined the number of eosinophils in bronchial biopsies using two-dimensional histology and found that intraepithelial eosinophils were not present in healthy controls, but were present in some asthmatic individuals, most commonly in those with severe asthma [1, 25].

Our finding relating intraepithelial eosinophils specifically to the presence of endogenous AHR in the form of EIB is of particular importance to understanding asthma immunopathogenesis because endogenous AHR results from mediators released from the airways and is a specific, but less sensitive, clinical feature of asthma [14] and EIB is a generalisable form of endogenous AHR that does not require allergic sensitisation [10]. In contrast, direct AHR represents the response to an exogenously administered bronchoconstrictor and is a less specific feature of asthma [15]. The precise mechanisms underlying endogenous AHR are incompletely understood [14], but our findings implicate an important role for intraepithelial eosinophils in the development of endogenous AHR in asthma. In line with this observation, a prior study found that allergen challenge, which can induce endogenous AHR in individuals with allergic asthma, induced a nonsignificant increase in intraepithelial, but not subepithelial, eosinophils [25]. Furthermore, another study revealed an increase in EPX immunostaining among individuals with moderate persistent asthma that was related to the length and branching of airway nerves that reside in the airway epithelium [26]. As eosinophils have been identified in close contact with airway nerves [26, 27], and CysLTs can activate airway nerves leading to smooth muscle contraction [28–30], the observed close association between intraepithelial eosinophils and endogenous AHR suggests a causal interaction.

Our results diverge from findings of a prior study of subjects with mild-to-moderate asthma that identified an association between endogenous AHR to mannitol and submucosal eosinophils based on two-dimensional histology and immunohistochemistry using immunoreactivity to eosinophil cationic protein as a surrogate for eosinophil density [24]. This finding may be confounded by eosinophil cytolysis, whereby these cells release their intact membrane-bound specific granules extracellularly at sites of eosinophil-associated diseases [31, 32], and therefore may not be as specific as our methodology, which counted eosinophil nuclei to identify only intact cells. Nonetheless, we observed a trend for higher subepithelial eosinophils in asthmatics with endogenous AHR, and thus such a relationship may exist especially in more severe disease states.

The T2 cytokines IL-4, IL-5 and IL-13 are implicated in the pathogenesis of eosinophilic or T2-high asthma. Although we observed a general association between airway T2 inflammation and eosinophils residing in the epithelial and subepithelial compartments, the strongest association was with IL5 expression. IL-5 has been linked to direct AHR [33]; however, mepolizumab, a monoclonal antibody against IL-5, failed to completely abrogate direct AHR [34]. Since mepolizumab only resulted in partial depletion of airway tissue eosinophils [35], a potential implication for the success of anti-IL-5 therapies is in their capacity to reduce the activation state of eosinophils, which could be augmented through other therapies that target eosinophil function such as therapies targeting IL-33 that we found had an additive effect with IL-5 in activating eosinophils [36].

In asthma, a shift of mast cells from the subepithelial to the intraepithelial compartment has been linked to an altered pattern of mast cell proteases, T2 inflammation and endogenous AHR [9]. Furthermore, the expression of T2 genes by mast cells is directly regulated by IL-33, but not other prominent epithelial-derived cytokines including thymic stromal lymphopoietin or IL-25 [9]. The association we found between both intraepithelial and subepithelial eosinophils and the higher expression of mast cell-specific genes in the airway, including TPSAB1 and CPA3, suggest that airway eosinophils might alter airway mast cell phenotype. This finding is of particular interest because prior work has demonstrated that both TPSAB1 and CPA3 expression in the airways is associated with endogenous AHR and T2 inflammation. By modelling the asthmatic epithelial microenvironment, we examined the potential mechanism for eosinophil-mediated alteration in mast cell protease expression. We demonstrate that IL-33 acts additively with IL-5 on eosinophils to generate LTC₄. We also show that LTC₄, in combination with IL-33, amplifies mast cell T2 gene expression and alters the mast cell protease profile by suppressing chymase expression. These findings provide insight into the unique population of intraepithelial mast cells that are identified in human asthma. They may also have functional implications, as chymase serves to inactivate IL-33 [37, 38]. As this regulation of mast cell function was substantially but not fully mediated through the CysLT₁ receptor, these results could explain some of the clinical efficacy of LT₁RAs, including the marked reduction in mast cell degranulation rather than simple antagonism of the end products of mast cell degranulation [39].

As induced sputum eosinophils are often used as a surrogate for T2 airway inflammation [40], our observation that induced sputum eosinophils were associated with eosinophils in the subepithelial space rather than the epithelium suggests that intraepithelial eosinophils may have increased adhesion, while subepithelial eosinophils can transit into the airway lumen [41, 42]. A limitation to this conclusion is the time lag of 2–10 days between the bronchoscopy and induced sputum collection. However, this was done to avoid any potential confounding due to hypertonic saline administration during sputum induction, which can induce the release of mediators of inflammation [43]. Prior studies have shown a modest relationship between the severity of EIB and induced sputum eosinophil concentration [44, 45], and that AHR persists despite near-complete depletion of sputum eosinophilia with anti-IL-5 blockade [46]. Consistent with our findings relating subepithelial eosinophils with T2 inflammation, a marker of steroid responsiveness, prior studies have related sputum eosinophils to the clinical response to inhaled corticosteroids [45]. However, since T2 inflammation persists in many subjects treated with inhaled corticosteroids [47], the relationships identified between eosinophils and mast cell gene expression identify immunopathological alterations in the airway wall that may be responsible for persistent T2 inflammation.

In conclusion, intraepithelial eosinophils are a unique feature of asthma and are related to features of endogenous airway AHR and T2 inflammation. Ex vivo modelling of the airway epithelial microenvironment suggests that intraepithelial eosinophils may promote AHR via increased LTC₄ production and regulate mast cell phenotype and function via LTC₄. These findings provide evidence that eosinophils and mast cells interact within the airway epithelial compartment in a cytokine milieu that includes IL-33, IL-5, and CysLTs and may cooperate to regulate features of T2 inflammation and AHR.