Female-specific enhancement of eosinophil recruitment and activation in a type 2 innate inflammation model in the lung

Abstract

A sex disparity in asthma prevalence and severity exists in humans. Multiple studies have highlighted the role of innate cells in shaping the adaptive immune system in chronic asthma. To explore the sex bias in the eosinophilic response, we delivered IL-33 to the lungs of mice and delineated the kinetics by which the inflammatory response was induced. Our data demonstrate that females recruited more eosinophils capable of responding to IL-33. Eosinophil activation occurred selectively in the lung tissue and was enhanced in females at all time points. This increase was associated with increased ex vivo type 2 cytokine and chemokine production and female-specific expansion of group 2 innate lymphoid cells lacking expression of the killer-cell lectin-like receptor G1. Our findings suggest that the enhanced eosinophilic response in females is due, firstly, to a greater proportion of eosinophils recruited to the lungs in females that can respond to IL-33; and secondly, to an enhanced production of type 2 cytokines in females. Our data provide insight into the mechanisms that guide the female-specific enhancement of eosinophil activation in the mouse and form the basis to characterize these responses in human asthmatics.

Graphical Abstract

eosinophil, group 2 innate lymphoid cells, IL-33

Introduction

Asthma is a chronic inflammatory airway disease characterized by bronchial hyperreactivity, airway inflammation, smooth muscle hypertrophy, and goblet cell hyperplasia [1]. Both the incidence and severity have been increasing steadily over the last two to three decades; there are now more than 300 million asthma sufferers worldwide, corresponding to 15 million disability-adjusted life years lost annually [2]. While the prevalence of asthma is higher in women than in men, this pattern is the opposite in children [3, 4].

The most common phenotype of asthma is allergic asthma, with a type 2 mediated immune response [1]. In murine models of asthma, multiple studies have shown a more pronounced type 2 immune response in females compared to males [5–8]. The type 2 immune response involves CD4⁺ T helper 2 (Th2) cells, group 2 innate lymphoid cells (ILC2s), IgE-secreting B cells, eosinophils (Eos) and mast cells, amongst others [9].

Eos are specialized effector cells of the innate immune system that are recruited to tissues, mainly in helminth infections and allergic reactions [10]. Emerging evidence has highlighted phenotypic plasticity within the terminally differentiated Eos population [11, 12]. For example, in an ovalbumin (OVA)/alum murine asthma model, upon OVA challenge, two Eos populations present in the lungs are distinguished through their expression of CD11c. CD11c^med Eos, with an activated phenotype, express low levels of CD11c, and transition into the airways where they are recovered in the bronchoalveolar lavage fluid. On the other hand, Eos lacking CD11c (CD11c^– Eos) are unable to make this transition [12, 13]. Similarly, CD101^hi Eos in the lungs of house dust mite (HDM)-treated mice differentially express multiple proinflammatory genes compared to CD101^low Eos [11]. While large numbers of CD11c^– Eos are present in the lungs of STAT6 knock-out (KO) mice treated with OVA/IL-33, Eos expressing CD11c are almost completely absent in these mice [13]. In human asthmatics, stable mucus plugs in the airways, containing abundant Eos, are associated with airflow obstruction and lower forced expiratory volume in 1 s (FEV1) [14]. Oxidants generated by Eos peroxidase (EPX) promote mucus plug formation and correlate with increased disease severity [14]. In addition, Charcot–Leyden crystals, found in dense mucus plugs in humans with severe asthma, are associated with activated airway Eos [15], and upon transfer in murine models of asthma promote type 2 inflammation [16]. Levels of blood and sputum Eos correlate with poor disease control and prognosis in asthmatics [17]. Moreover, in ST2 KO and IL-13 KO murine models, Eos can induce airway inflammation, upstream of Th2 cells [18, 19].

ILC2s are lineage-negative innate lymphoid cells that produce type 2 cytokines IL-4, IL-13, and IL-5 [20, 21]. Murine models comparing Rag2 KO mice (lacking T and B cells) to IL2rg KO mice (lacking all lymphoid cells, including ILC2s) show that ILC2s are capable of inducing type 2 responses independent of Th2 adaptive immunity [22]. New data show that the sex disparity in asthma is likely in place long before the onset of Th2 adaptive immunity [23]. In addition to data showing androgens inhibit ILC2 differentiation and proliferation [23–25], Kadel et al. showed that females have larger numbers of ILC2s lacking killer cell lectin-like receptor G1 (KLRG1) expression, contributing to the greater ILC2 numbers in female mice [23].

We have examined sex-specific differences present before the onset of Th2 adaptive immunity and found that both Eos recruitment and activation are increased in mice upon delivery of recombinant IL-33 to the lung, in a manner that is dramatically enhanced in female mice [13]. Our goal here was to address whether the enhanced Eos response in female mice was linked to ILC2 expansion and/or type 2 cytokine release. We also examined more precisely the site of Eos activation and we delineated the kinetics by which ILC2s expanded, type 2 cytokines were increased, and Eos recruited and activated in the lungs of male and female mice.

Materials and methods

Mice

Female and male WT BALB/c mice (6–8 week-old; originally from The Jackson Laboratory) were bred in-house under pathogen-free conditions at the Research Institute of the McGill University Health Centre (RI-MUHC). Mice were kept in cages supplemented with water and irradiated food. All animal studies were approved by McGill University Animal Care Committee and performed following the Canadian Council on Animal Care guidelines.

IL-33-induced airway inflammation

A brief state of anesthesia was achieved by placing mice in a closed chamber with 3.5% isoflurane and 1 l/min oxygen. To induce airway inflammation, mice were treated intranasally with 50 μg OVA (Worthington, Lakewood, NJ) and 0.5 μg IL-33 (eBioscience, San Diego, CA or Biolegend, San Diego, CA) in a volume of 30 μl phosphate-buffered saline (PBS) according to the timepoints specified in each experiment. As a control, mice were treated with OVA only. Mice were sacrificed using a lethal dose of sodium pentobarbital followed by cardiac puncture, lung harvest and/or bone harvest (see below).

Lung digestion

Lung tissue was collected in RPMI-1640 media, minced, and incubated for 30 min at 37^oC and 5% CO₂ in a cocktail of digestive enzymes containing 200 μg/ml DNase I (Sigma–Aldrich, Oakville, ON), 1 mg/ml hyaluronidase 1a (Life Technologies, Carlsbad, CA), 250 μg/ml collagenase XI (Life Technologies) and 100 μg/ml LiberaseTM (Sigma–Aldrich, Oakville, ON) [26]. Afterward, cells were washed with RPMI-1640 medium with 5% fetal bovine serum (FBS) and 1% penicillin/streptomycin. Red blood cells (RBC) were either lysed using ammonium–chloride–potassium (ACK) lysis buffer or by inducing a brief hypotonic shock in distilled water followed by dilution into phosphate-buffered saline (PBS). Afterward, cells were filtered through a 70 μm strainer.

Bone marrow processing

The tibias and femurs of each mouse were collected in PBS. The bones were placed in 70% ethanol for 1 min and then transferred into PBS. The epiphysis of each bone was cut to expose the bone marrow, and the bones were placed in bone marrow isolation tubes. Briefly, these tubes were prepared by placing a 200 μl pipette tip in a 1 ml pipette tip. The ends of both tips were cut and placed in a 1.5 ml microcentrifuge tube, containing 50 μl PBS. The tubes were centrifuged at 2500 × g for 5 s after which ACK lysis buffer was added to lyse RBC.

Peripheral blood processing

Blood (0.5–1 ml) was collected via a cardiac puncture in EDTA-coated syringes. Collected blood was diluted with an equal volume of PBS and carefully layered on top of 4.5 ml Histopaque-1083 (Sigma–Aldrich) in 15 ml SepMate tubes (Stemcell Technologies, Vancouver, BC). After centrifugation, the layer containing Eos was collected and washed twice with PBS.

Lung explant culture

Two million live cells from the total lung, processed as above, were cultured in either saline or 20 ng/ml IL-33. Cells were incubated for 48 h in RPMI-1640 medium with 10% FBS, 1% penicillin/streptomycin, 1 mM sodium pyruvate, 1 mM non-essential amino acids, and 55 μM 2-mercaptoethanol at 37^oC and 5% CO₂. Supernatants were collected for Multiplex ELISA quantification.

Multiplex ELISA

Levels of IL-17, IL-13, IL-5, IL-4, CCL11, CCL22, CCL17, and INF-γ were quantified by mouse custom Q-Plex array (Quansys Biosciences, Logan UT). Plates were read via Q-View Imager LS and analyzed with Q-View Software (Quansys Biosciences). Samples below the limit of detection were assigned a value equal to 50% of the detection limit.

Multi-color flow cytometry

One million live cells for Eos and two million live cells for ILC2s were plated in low adherence, round-bottom 96 well plates. ILC2s were incubated at 37^oC and 5% CO₂ with 0.665 μl/ml Golgi Stop (BD Biosciences, Franklin Lakes, NJ). Then, cells were incubated in the dark at 4^oC with eFlour789 viability dye (eBioscience) for 20 min. Next, cells were washed twice with PBS and incubated with anti-CD32/16 (BD Biosciences) to block Fc receptors. ILC2s were stained using EF-450-Thy1.2 (Life Technologies), PeCy7-CD127 (Biolegend), PerCP-eF710-ST2 (Life Technologies), BV605-KLRG1 (Biolegend), BUV395-CD45.2 (BD Biosciences) and a combination of PE-conjugated antibodies against CD3ε (Biolegend), CD11c (Biolegend), CD11b (Life Technologies), CD49b (BD Biosciences), CD45R (Life Technologies), TCRγδ (BD Biosciences), Ly6G (Biolegend), and FcεRIα (Biolegend). Eos were stained with BUV395-CD45.2 (BD Biosciences), AF700-Ly6G (Biolegend), AF488-CD11c (Life Technologies), BV480-SiglecF (Life Technologies), BV711-ST2 (BD Biosciences), PE-Vio770-CD101 (Miltenyi Biotec, Bergisch Gladbach, Germany), and PE-SiglecF (BD Biosciences). Cells were washed twice and fixed with IC fixation buffer (eBiosciences) for >1 h. ILC2s were permeabilized the next day with Perm/Wash (BD Biosciences) and stained intracellularly with AF488-IL-13 and APC-IL-5. For in vivo staining of circulating leukocytes, 3 μg of BV650-CD45.2 was injected intravenously (IV) in 200 μl PBS via the tail vein. Mice were euthanized 3 min later by injection of a lethal dose of sodium pentobarbital, after which lungs were flushed by injecting 5 ml PBS into the right ventricle and Eos stained as above. All samples were acquired using BD LSRFortessa (Immunophenotyping Core Facility, RI-MUHC) flow cytometer. Analysis was performed using FlowJo V10 (FlowJo LLC, Ashland OR). Positive and negative populations were defined using fluorescence mine one (FMO) controls.

Cell sorting and Eos Diff-Quick staining

Lung cells were obtained from eight mice treated with OVA and IL-33 as above. Eos were stained using Eos staining cocktail as above and CD11c^– Eos and CD11c^med Eos were sorted using a BD FACSAria Fusion Flow Cytometer by personnel at the RI-MUHC Immunophenotyping Core Facility. Sorted CD11c^– Eos and CD11c^med Eos were spun (Cytospin^TM) on microscope glass slides and Diff-Quick staining was performed. Ten images for each population were taken using an Olympus BX50 microscope at 400×. Four blinded individuals successfully identified CD11c^– Eos and CD11c^med Eos at original magnification, based on the shape of nuclei and presence (or not) of cytoplasmic vacuoles [12].

Statistical analysis

All analyses and graphs were generated using Prism, version 9 (GraphPad Software, San Diego, CA). Data were analyzed by one-way or two-way ANOVA followed by multiple comparisons using Tukey’s post hoc test. A P-value less than 0.05 was considered significant. Outcomes are presented as mean ± SEM. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001, ^****P < 0.0001. Grubb’s test with α = 0.05 was employed to remove outliers.

Results

Females recruit more Eos and neutrophils capable of responding to IL-33

We have previously shown that in response to acute administration of OVA/IL-33, more Eos are recruited to and activated in the lungs of female mice compared to males and that Eos activation is STAT6 dependent [13]. As part of our goal to better understand sex differences in Eos responses in IL-33-treated wild-type mice, we examined whether Eos in males and females differed in the expression of ST2, and thus their ability to respond to IL-33. Male and female mice were treated daily for each of 2 days with OVA alone or with IL-33 prior to being euthanized 96 h after the second dose (Fig. 1A). OVA was included to approximate a natural in vivo response, in which IL-33 is released in the presence of an antigen. CD11c^– Eos and CD11c^med Eos were defined as CD45⁺Ly6G^–/intSiglecF⁺CD11c^– and CD45⁺Ly6G^–/intSiglecF⁺CD11c^med, respectively, distinguishing them by their level of CD11c expression (Supplementary Fig. S1). As expected [13], in response to OVA/IL-33 there were larger numbers of both CD11c^– Eos (Fig. 1B) and CD11c^med Eos (Fig. 1C) in the lungs of females compared to males. ST2⁺ CD11c^– Eos increased in response to OVA/IL-33 treatment, in a manner that was enhanced in females (Fig. 1D), a response that was also observed in CD11c^med Eos (Fig. 1E). Interestingly, ST2⁺ neutrophils were increased only in females in response to OVA/IL-33 treatment (Fig. 1F). The median fluorescence intensity (MFI) of ST2 on CD11c^– Eos was also increased selectively in females (Fig. 1G). These data show that the sex bias in Eos activation following delivery of OVA/IL-33 to the lung can be linked, at least in part, to larger numbers of Eos with the ability to respond to IL-33 in females.

Figure 1.

Female lungs recruit more ST2 + Eos and neutrophils post OVA/IL-33 treatment. (A) BALB/c mice were treated twice daily with OVA or OVA/IL-33 and euthanized 96 h after the second treatment. (B and C) Absolute count of CD11c^– Eos and CD11c^med Eos in the lung. Data are from one experiment with four mice per group, representative of two independent experiments and are presented as mean ± SEM. (D–F) Quantification of ST2 + CD11c^– Eos, CD11c^med Eos and neutrophils. Data are from two experiments with six to eight mice per group and presented as mean ± SEM. (G) MFI of ST2 on total CD11c^– Eos. Data are from one experiment representative of two independent experiments, presented as mean ± SEM. Two-way ANOVA, Tukey’s post hoc test. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001

OVA/IL-33 induces an exaggerated cytokine producing KLRG1^– ILC2 response in females

To further explore the mechanisms driving the increased Eos activation in females in response to OVA/IL-33, we delineated the kinetics by which ILC2s expand and produce cytokines in response to OVA/IL-33. We focused on ILC2s, as they are considered early and potent responders to IL-33 in the lung. Males and females were treated with a single dose of OVA/IL-33 and euthanized 12 h, 24 h, and 48 h post treatment. As a negative control, mice were treated with OVA only and euthanized 48 h later. ILC2s were defined as lineage^–CD45⁺Thy1.2⁺CD127⁺ST2⁺ (Supplementary Fig. S2). Total ILC2 expansion began only 48 h post OVA/IL-33 administration, with no significant differences between males and females (Fig. 2B). While there were no sex differences in the number of KLRG1⁺ ILC2s (Fig. 2C), KLRG1^– ILC2s were significantly increased in females compared to males in the control group, as well as 48 h post OVA/IL-33 stimulation (Fig. 2D). Intracellular cytokine staining revealed that IL-5 and IL-13 producing KLRG1⁺ ILC2s expanded equally at 48 h in both sexes (Fig. 2G), with more KLRG1⁺ ILC2s producing only IL-13 in males (Fig. 2F), and more KLRG1⁺ ILC2s producing only IL-5 in females (Fig. 2E). Conversely, the KLRG1^– ILC2 response was largely female-specific, with the expansion of IL-13 and IL-5 producing KLRG1^– ILC2s occurring as early as 12 h post OVA/IL-33 stimulation in females only (Fig. 2H–J). Although an increase in the number of ILC2s was not detectable before 48 h, an increased proportion of both KLRG1⁺ and KLRG1^– ILC2s produced IL-13 and IL-5 as early as 12 h post OVA/IL-33 stimulation (Supplementary Fig. S3). Together, these data indicate that: (i) early after stimulation with IL-33, ILC2s produce cytokines prior to increasing in number; and (ii) cytokine-producing KLRG1^– ILC2s may be, in part, responsible for the enhanced Eos responses in female mice.

Figure 2.

Cytokine-producing KLRG1^– ILC2s are increased in females in response to OVA/IL-33. (A) BALB/c mice were treated once with OVA or with OVA/IL-33 and euthanized 12 h, 24 h or 48 h after the treatment. (B–D) Absolute count of total, KLRG1⁺ and KLRG1^– ILC2s in the lung. (E–J) Absolute count of KLRG1⁺ and KLRG1^– ILC2s that produce IL-13 and/or IL-5. Data are a combination of two independent experiments with eight mice per group. Data are presented as mean ± SEM. Two-way ANOVA, Tukey’s post hoc test. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001. #: significantly different than all other groups within the same sex

Type 2 cytokines and chemokines are greatly increased in ex vivo lung cultures in female mice

We also examined the production of several cytokines and chemokines from lung cells harvested from mice treated in vivo as in Fig. 2A and cultured ex vivo with saline or IL-33 for 48 h. While culturing the lung cells with saline reflects the cytokine production by the immune cells recruited in vivo, culturing with IL-33 allows us to investigate cytokine production in response to restimulation without the possibility of additional cell recruitment. Lung cells from OVA/IL-33 treated females produced more IL-5 than cells from males when cultured with saline (Supplementary Fig. 4A). Moreover, compared to saline, ex vivo IL-33 stimulation dramatically increased IL-5 production, with more IL-5 produced by cells from females compared to males (Fig. 3A) though this difference was significant only at the 12 h time point. No sex differences were detected in IL-13 production from IL-33-stimulated lung cells, though cells from female mice tended to produce more IL-13 compared to their male counterparts at all time points (Fig. 3B). The culture of cells from OVA/IL-33 treated females produced more CCL11 and CCL22 than males (Figs. 3C and D), though the amounts produced were no greater than the saline cultures (Supplementary Fig. S4C and SD). Larger amounts of CCL17 were produced by cells from female mice only at the 48 h time point (Fig. 3E, Supplementary Fig. S4E). Of note, the only cytokine/chemokine that was increased in cells cultured from OVA/IL-33 treated males (compared to OVA-treated controls), was IL-13 and this occurred only in saline-cultured lung cells (Supplementary Fig. S4B) whereas the majority of mediators were increased in cells from female mice treated with OVA/IL-33. Levels of IL-4, IL-17, and IFN-γ did not differ at any time point or between cells from males vs. females (not shown). Altogether, these data suggest that the enhanced immune response in females is associated with increased production of type 2 cytokines and chemokines.

Figure 3.

Type 2 cytokines and chemokines in lung cells cultured with IL-33 are increased only in female mice. Mice were treated as in Fig 2A, and lungs were harvested for ex vivo IL-33 re-stimulation. Levels of IL-13, IL-5, IL-4, CCL11, CCL22, and CCL17 in supernatants. Data are from one experiment with four mice per group, representative of two independent experiments. Data are presented as mean ± SEM. Two-way ANOVA, Tukey’s post hoc test. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001. #: significantly different than all other groups within the same sex

Eos activation occurs selectively in the lungs

To determine the location of Eos activation in OVA/IL-33 treated mice and whether this differed in males vs. females, we examined the kinetics by which Eos upregulated two markers of activation, CD11c [19] and CD101 [18] in three different compartments: bone marrow, blood, and lungs, using the same model described in Fig. 2A. Firstly, we show that Eos in the bone marrow were significantly increased only 48 h after OVA/IL-33 stimulation, and only in females (Fig. 4A). Females had significantly more Eos in the bone marrow than males across all groups (Fig. 4A). The frequency of Eos in the blood remained unchanged, with a trend for females to have more Eos than males (Fig. 4B). In the lungs, the total number of Eos was increased 12 h post OVA/IL-33 treatment in females, but delayed in males, being greater only 24 h post OVA/IL-33 treatment (Fig. 4C). OVA/IL-33 treated females had significantly more Eos than males at all time points (Fig. 4C). Levels of CD101, which is upregulated on Eos in mice exposed to the common household allergen, HDM [11], was increased and linked to CD11c expression, as only Eos present in the lung, but not the bone marrow or blood, upregulated CD11c and CD101 (Figs. 4D and E, (Supplementary Figs. S5A and B). We confirmed that the majority of the CD101^hi Eos co-expressed CD11c (Fig. 5A). Consistent with the morphological changes expected in activated Eos [11, 12], the nuclei in flow-sorted lung CD11c^med Eos were more segmented (Fig. 5B), compared to CD11c^– Eos where the nuclei were ring-shaped (Fig. 5C). Altogether, these data provide evidence that in response to IL-33, Eos is recruited to the lung at earlier times and to higher levels in female mice compared to males. Once recruited to the lung, Eos upregulates markers of activation, CD11c and CD101.

Figure 4.

Upregulation of CD11c and CD101 occurs selectively in the lungs. Mice were treated as in Fig. 2A. (A) Bone marrow Eos presented as the frequency of live. Data are a combination of two independent experiments with six to eight mice per group. (B) Blood Eos presented as the frequency of live. Data are from one experiment with 3–4 mice per group, representative of two independent experiments. (C) Lung Eos absolute count. Data are a combination of two independent experiments with seven to twelve mice per group. (D and E) Representative histograms of the expression of CD11c and CD101 in CD11c^– Eos and CD11c^med Eos at 48 h in different compartments (bone marrow, blood, and lung). Control is CD11c^– Eos from OVA-treated mice. For data in A–C, two-way ANOVA, Tukey’s post hoc test. Data are presented as mean ± SEM. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001. #: significantly different than all other groups within the same sex

Figure 5.

CD11c^– Eos and CD11c^med Eos have distinctive morphologic nuclear features. Mice were treated as in Fig. 1A. (A) Representative flow cytometry panels comparing CD11c and CD101 expression on total lung Eos. FMO controls are included as small panels to the right. (B and C) CD11c^– Eos and CD11c^med Eos were flow sorted and spun on microscope glass slides and Diff-Quick stained. (B) CD11c^med Eos exhibiting segmented nuclei and multiple cytoplasmic vacuoles. (C) CD11c^– Eos exhibiting round nuclei. Panels are representative of 10 images for each population taken at 400×

More in-depth analysis of Eos in the lungs showed that CD11c^– Eos were increased by 12 h in females (Fig. 6A) and then activated by 24 h (Fig. 6B). Although not significant, while the count of CD11c^– Eos dropped from 24 h to 48 h post OVA/IL-33 treatment in females, this decrease was met with a reciprocal increase in CD11c^med Eos (Fig. 6B), providing further evidence that the two populations are not recruited independently, and that activation occurs in the lung. Again, males had fewer recruited CD11c^– Eos and activated CD11c^med Eos (Fig. 6A and B), both of which were significantly increased only at 24 h.

Figure 6.

CD11c^– Eos and CD11c^med Eos localize to the lung parenchyma only following OVA/IL-33 treatment. (A and B) CD11c^– Eos and CD11c^med Eos in the lung were quantified. Data are a combination of two to three experiments with five to twelve mice per group and are presented as mean ± SEM. (C) Representative flow cytometry panels showing Eos located in the vasculature (CD45.2 IV positive) or tissue (CD45.2 IV negative). The first panel represents Eos from an OVA-treated mouse that did not receive an IV-injected anti-CD45 antibody. All others received anti-CD45.2 antibodies. (D and E) Comparison between CD11c^– Eos and CD11c^med Eos located in the lung circulation vs. lung parenchyma in males and females. Data are from one experiment (males) or representative of two (females) using five mice per group. OVA group is a combination of males and females. Lines showing statistical significance in D and E represent a comparison of parenchymal populations only. Two-way ANOVA (for A and B) and One-way ANOVA (for D and E), Tukey’s post hoc test. ^*P ≤ 0.05, ^**P ≤ 0.01, ^***P ≤ 0.001, ^****P ≤ 0.0001

To further isolate the site of Eos activation in the lungs, we performed a technique that allows one to discriminate between Eos localized to the lung parenchyma (and/or air spaces) and Eos present in the vasculature, including the capillary network of the lung. In fact, Anderson et al. have shown that perfusing the lungs does not clear leukocytes localized to the circulation of the lung, highlighting the importance of performing intra-vascular staining before collecting lungs [27] to confirm tissue localization. Thus, male and female mice were treated with OVA or OVA/IL-33 and 48 h later, fluorochrome-conjugated CD45.2 antibody was IV injected 3 min before collecting lungs and blood. We confirmed that all Eos isolated from blood were labeled with IV-injected antibodies (data not shown) and that none of the alveolar macrophages was labeled (Supplementary Fig. 5C). Remarkably, almost all Eos in the mice treated with OVA only (Fig. 6C, 2nd panel) were localized to the lung vasculature and only upon OVA/IL-33 treatment did a small proportion of the CD11c^– Eos move into the lung parenchyma (Fig. 6C, 3rd panel). A similar number of parenchymal CD11c^– Eos was present in males and females upon OVA/IL-33 treatment (Fig. 6D). On the other hand, >80% of the activated CD11c^med Eos in OVA/IL-33 treated mice escaped being labeled, providing evidence that they were localized to the lung parenchyma and/or airways (Fig. 6C, 4th panel). More parenchymal CD11c^med Eos were present in females upon OVA/IL-33 treatment (Fig. 6E). No differences were found between males and females when comparing the circulatory Eos populations (data not shown). These data indicate that nearly all Eos in untreated mice is localized to the lung vasculature and that upon OVA/IL-33 administration, they are recruited into the lung parenchyma and activated, in a manner that is enhanced in females.

Discussion

In this study, we have demonstrated that female mice exhibit a greater magnitude of inflammation in response to OVA/IL-33, characterized by enhanced type 2 cytokine and chemokine release, and an earlier and greater eosinophilic response. Each of these responses was lower in males, providing evidence that the magnitude of these responses is sex-specific. Two factors were responsible for the enhanced Eos recruitment and activation in females. First, females recruited a larger number of Eos, and secondly, a greater proportion of these Eos had the ability to respond to IL-33. In humans, a rare loss-of-function mutation in IL1RL1—which encodes ST2—disrupts the IL-33/ST2 axis and is associated with decreased blood Eos and protection against asthma [28]. Similarly, other single nucleotide polymorphisms in the IL1RL1 gene are associated with an increase in the levels of soluble ST2, preventing its signaling and protecting against asthma in humans [29]. Therefore, it is possible that the increased prevalence and severity of asthma observed in females [3], is partially explained by the greater proportion of Eos that express ST2.

Our data show that CD11c^med Eos co-expressed CD101, and that CD11c^– Eos and CD11c^med Eos exhibited the expected morphological appearance associated with activation [11, 12], namely CD11c^– Eos had ring-shaped nuclei, while those in CD11c^med Eos were segmented, as shown previously for activated Eos [11, 12]. While the precise immunological role of CD11c^– Eos and CD11c^med Eos is yet to be clearly defined, we know that CD101^hi Eos express proinflammatory genes [11] and that CD11c^med Eos transition into the airways [12, 13]. We speculate that activated Eos present in the airways contribute to the immunopathology (e.g. promoting mucus production) in asthma [14].

Our ex vivo cultures showed that total lung cells from female mice displayed a greater release of type 2 cytokines that could also contribute to the enhanced eosinophilic response in females. In particular, IL-5 and CCL11 contribute to Eos mobilization [30, 31], and both were significantly increased in ex vivo cultures of lung cells from females only. Because CCL11 gene transcription is STAT6-dependent [32], and our published data show that abundant Eos are present in the lungs of STAT6 KO mice [13], we speculate that CCL11 (as well as other STAT6-dependent Eos chemokines) play a minimal role in recruiting Eos to the lung in response to IL-33, and likely promote transit of Eos from the lung vasculature into the lung parenchyma. IL-13 levels were also increased in females post ex vivo IL-33 stimulation of total lung cells, in agreement with our hypothesis that the sex-specific differences are largely dependent on greater cytokine release in females.

Rothenberg et al. have shown that, in response to IL-33 stimulation of bone marrow-derived Eos, IL-4 is produced and, through an autocrine loop supports IL-33-induced Eos activation [33]. However, in our experiments, IL-4 was undetectable from total lung cells cultured with IL-33 (data not shown). In the previous study, bone marrow-derived Eos cultured with 100 ng/ml IL-33 produced ~2.5 ng/ml of IL-4, around 12-fold less than IL-13 [33]. In our study, we cultured whole lung cells, only a portion of which was Eos, and stimulated them with less IL-33 (20 ng/ml); thus, IL-4 production may have been below the limit of detection. Although IL-4 was undetectable, we have not yet excluded a role for IL-4 in Eos activation in this model.

While cytokine-producing KLRG1⁺ ILC2s were not uniformly greater in female mice, those lacking KLRG1 were consistently greater in females compared to males [23], suggesting that KLRG1^– ILC2s participated in the sex-specific differences we have identified. Notably, mice deficient in ILC2s recruit significantly fewer CD101^– and CD101⁺ Eos to the lung in response to HDM or Alternaria alternata [34]. However, the recruitment of Eos into the lungs was partially rescued by the delivery of recombinant IL-5, suggesting a role for ILC2-derived IL-5 in recruiting Eos into the lungs [34]. This aligns with our findings, where in vivo we observed increased IL-5-producing KLRG1^– and KLRG1⁺ ILC2s in females, and ex vivo we observed that total lung cells from female mice produced more IL-5, whether cultured with saline or IL-33.

ILC2s lacking KLRG1 might, in fact, exhibit a greater functional response as E-cadherin, which is present in lung epithelium, interacts with KRLG1 to inhibit ILC2s, at least in vitro [35, 36]. Nevertheless, it is unknown whether E-cadherin is expressed at equivalent levels on lung epithelium in females and males. For example, if E-cadherin is expressed at lower levels in females, KLRG1⁺ ILC2s might exhibit greater function. In fact, regulation of E-cadherin may play an essential role in the pathogenesis of asthma, at least in mice [37].

In addition to ILC2s, other innate immune cells, including Eos and neutrophils, could be responsible for the sex difference in cytokine production in the ex vivo cultures. IL-5 is found as a preformed cytokine in Eos, stored in crystalloid granules [38]. Upon endobronchial challenge, significant airway recruitment of IL5⁺ Eos is induced in humans with asthma [39]. Similarly, Eos express and release functional IL-13 in humans with atopic diseases, including asthma [40]. In mice, Eos-derived IL-13 promotes features of allergic asthma [19, 41]. Thus, the ability of Eos to release functional IL-5 and IL-13 suggests an autocrine pathway of recruiting and activating the lung Eos population. In addition, increases in T-cell recruiting chemokines—CCL22 and CCL17—occurred only in females, with the highest increase 48 h post OVA/IL-33 treatment, a timepoint that correlates with the greatest Eos activation. Models of Eos adoptive transfer in mice have shown that Eos can be upstream of CD4⁺ T cells, through Eos production of CCL22 and CCL17 [18, 19].

Several reports show that two separate Eos populations accumulate in the lungs of mice in type 2 inflammation models, whether innate or adaptive [11–13]. These populations are distinguished through their expression of CD11c or CD101. Yet, it has been unclear whether Eos activation occurs before entry into the lung. Reports from studies using intravascular staining techniques re-define the interpretation of various studies describing tissue-resident immune cells [27, 42]. Here, we show that the majority of CD11c^– Eos in control mice (in our study, those treated with OVA alone) were localized to the vasculature of the lung, and not the lung parenchyma. Similarly, the use of intravital imaging to localize lung Eos showed that all Eos in control mice were intravascular, while those in mice after OVA/alum sensitization and OVA challenge were localized, in part, within the lung parenchyma [43]. Though assessed at a single time point, our data show that, following OVA/IL-33 delivery, the majority of CD11c^med Eos, was localized to the lung parenchyma, providing evidence that Eos activation occurred selectively in the lungs. Given our previous data showing Eos activation was largely absent in STAT6 KO mice, and our current data that Eos express ST2 and thus have the ability to respond to IL-33, we speculate that Eos activation occurs in the lung under the control of IL-33, and possibly IL-13 and/or IL-4. Nevertheless, we consider the possibility that Eos activation begins during the process of migration through the endothelium and into the lung, in line with the small proportion of CD11c^med Eos labeled with the intravascular anti-CD45 antibody.

The biological mechanisms driving the observed sex differences in Eos recruitment and activation are likely attributed—at least in part—to the effects of sex hormones. For example, male androgens limit the expansion of ILC2 subpopulations [23, 24], diminishing the release of cytokines required for Eos recruitment and activation. In addition, we speculate that sex hormones could be involved in regulating the expression and/or activity of ST2, driving enhanced responses in females. These differences could also be epigenetically regulated, though evidence for modifications at the level of chromatin (whether hormonally regulated or not) has not been described. Moreover, male androgens promote Treg suppressive functions, though our prior work suggested that Tregs are not involved in this model [13, 44]. Another possibility is that levels of soluble ST2, which acts as a decoy for IL-33, contribute to enhanced inflammation in females.

Altogether, our data demonstrate that females exhibit several greater innate immune responses compared to males. Multiple studies have highlighted the critical role of innate cells in shaping the adaptive immune system in chronic asthma [45]. Therefore, new therapies incorporating innate immune responses will likely be more effective than those focused only on adaptive immunity. Identifying early components of the innate immune response may provide possible targets to treat and possibly prevent asthma.

Supplementary Data

Supplementary data is available at Clinical and Experimental Immunology online.

Abbreviations:

Abbreviations:
 
• AAM

  alternatively-activated macrophage

 
• ACK

  ammonium–chloride–potassium

 
• Eos

  eosinophil(s)

 
• EPX

  Eos peroxidase

 
• FBS

  fetal bovine serum

 
• FEV1

  forced expiratory volume in 1 s

 
• FMO

  fluorescence mine one

 
• HDM

  house dust mite

 
• ILC2s

  group 2 innate lymphoid cells

 
• KLRG1

  killer cell lectin-like receptor G1

 
• OVA

  ovalbumin

 
• PBS

  phosphate-buffered saline

 
• RBC

  red blood cells

 
• Th2

  T helper 2

 
• WT

  wild type.

Acknowledgements

We thank Dr Q.H. and his lab members for their helpful comments. We thank the immunophenotyping platform of the Research Institute of McGill University Health Centre for excellent support for flow cytometry services.

Ethical Approval

Animal studies were approved by the McGill University Animal Care Committee and performed following the guidelines of the Canadian Council on Animal Care.

Conflict of Interests

None declared.

Funding

This work was performed with the support of operating grants from the Canadian Institutes of Health Research (CHIR: PJT-162254) to EDF. LL was supported by a studentship from Les Fonds de la Recherche Québec—Santé (FRQS). The Meakins Christie Laboratories, RI-MUHC are supported in part by a Centre grant from the FRQS.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Author Contributions

R.K. and E.D.F. were responsible for writing the manuscript. R.K., E.D.F. and V.G. designed the experiments. R.K., V.G., H.A., L.L., N.A.G., and J.S. participated in animal handling, tissue collection, and data acquisition.