All animal experiments were approved by the Gothenburg County Regional Ethical Committee (permit number 126/14 and 2459/19). WT C57BL/6J and C57BL/6JRj mice were purchased from Charles River (Sulzfeld, Germany) and Janvier Labs (Le Genest-Saint-Isle, France), respectively. Rag1^−/− mice and WT C57BL/6 mice, used in some experiments, were obtained from in-house breeding (University of Gothenburg, Sweden). Rag2^−/−Il2rg^−/− mice were purchased from Taconic (Germantown, NJ, USA). Age and sex-matched mice at 7–12 weeks of age were used in all experiments. Mice were housed in pathogen-free conditions and were given food and water ad libitum.

To induce allergic airway inflammation mice were administered 10 μg papain (Sigma-Aldrich) intranasally on days 1–3 and samples were collected 24 hours (h) after final exposure as previously described (28). In some experiments mice received an intraperitoneal single dose of 25 μg anti-mouse IL-5 or isotype control (BD Biosciences, San Jose, CA, USA) 1 h before the first exposure. In the kinetic study, mice received an intranasal single dose of 10 μg papain and samples were collected at 3, 6, 12, 24, and 48 h. Control mice received phosphate buffered saline (PBS) vehicle.

IL-33-induced airway inflammation was performed as previously described (25, 29). In brief, WT, Rag1^−/−, and Rag2^−/−Il2rg^−/− mice were exposed to 1 μg recombinant murine IL-33 (rmIL-33, PeproTech, Rocky Hill, NJ, USA) by intranasal administration on day 1, 3, and 5. Samples were collected 24 h after the final exposure. Control mice received PBS vehicle.

Bronchoalveolar lavage (BAL), blood, and bone marrow were collected in this study. A detailed description of the sample collection is provided in Johansson et al. (29). Cells in BAL were processed for differential cell count analysis. Furthermore, cell-free BAL and serum were processed for mediator analysis by ELISA. Bone marrow cells were isolated from left and right femurs by flushing with wash buffer (2% fetal bovine serum, Sigma-Aldrich, in PBS). The samples were filtered through a 100 μm cell strainer (CellTrics®, Sysmex, Goerlitz, Germany) and red blood cells were lysed (0.1 mM EDTA in distilled water/0.15 M NH₄Cl, Sigma-Aldrich/Merck Chemicals) by 10 minutes (min) incubation on ice. The cells were further processed for flow cytometric analysis, ex vivo stimulation, and differential cell count as previously described (25).

Approximately, 10,000–50,000 cells were used for slides (425 × g, 6 min, Shandon Cytospin 3 centrifuge) and stained with Hemacolor® Rapid stain (Merck, Darmstadt, Germany) according to the manufacturer's protocol. Eosinophils were assessed by histological examination as previously described (29).

Bone marrow cells from PBS exposed WT mice were seeded at a concentration of 2.5 x 10⁶/ml in complete cell culture medium: RPMI-1640 (HyClone™; GE Healthcare Life Sciences, South Logan, UT, USA), 10% fetal bovine serum (Sigma-Aldrich), 2 mM L-glutamine (HyClone), 100 U/ml penicillin, 100 μg/ml streptomycin (HyClone), 1 mM sodium pyruvate (Sigma-Aldrich). Cells were stimulated with rmIL-33 (100 ng/ml) for 24 h or kept in complete culture medium as control. Monensin (BD GolgiStop™, BD Biosciences) was added to all samples (4 μl/6 ml) during the last 3 h of the incubation. Newly produced IL-5 by ILC2s (SSC^loLin⁻CD45⁺CD127⁺ST2⁺) was measured by intracellular flow cytometry. Bone marrow cells from IL-33 and PBS exposed WT mice, PBS exposed Rag1^−/− and Rag2^−/−Il2rg^−/− mice were cultured as described above (2.5 × 10⁶/ml in complete cell culture medium) with or without rmIL-33 (100 ng/ml) for 44 h and cell-free culture supernatants were collected for measurement of secreted IL-5 by ELISA.

Bone marrow cells were resuspended in 2% mouse serum (Dako, Glostrup, Denmark) and antibodies for surface receptors were added (30 min, 4°C). Cells were washed and fixed (BD CellFix™, BD Biosciences, Erembodegem, Belgium) for 15 min in the dark at room temperature (RT). Cells were washed and analyzed on a BD FACSVerse™ Flow Cytometer running BD FACSuite™ software (BD Biosciences). Collected data were analyzed by FlowJo Software (tree Star Inc, Ashland, OR, USA). Linage negative cells were determined as CD45⁺CD3⁻CD45R/B220⁻CD11b⁻TER-119⁻Ly-G6/Gr1⁻CD11c⁻CD19⁻NK-1.1⁻FceR1⁻.

Eosinophil progenitors, mature eosinophils, and ILC2s were defined as SSC^loCD45⁺CD34⁺IL5Rα⁺, SSC^hiCD45⁺CD34⁻IL5Rα^loCCR3+Siglec-F+, and SSC^loLin⁻CD45⁺CD127⁺CD25⁺ST2⁺, respectively. Antibodies used are listed in Table S1. The IL-33 receptor (ST2) expression was estimated by mean fluorescence intensity (MFI) values. Relative MFI (rMFI) equals MFI of monoclonal antibody divided by MFI of corresponding fluorescence minus one (FMO) control.

Cultured bone marrow cells were stained with surface antibodies (as described above) and fixed with 4% Paraformalaldehyde (Sigma-Aldrich) in PBS for 15 min at RT in the dark. All solutions used before fixation were supplemented with Monensin (BD GolgiStop™, 4 μl/6 ml). Cells were permeabilized with 0.1% saponin (Sigma-Aldrich) in Hank's balanced salt solution (HyClone). Anti-IL-5- or isotype control antibodies were added and cells were incubated for 40 min at RT in the dark and washed before flow cytometric analysis.

Cytokine and chemokine quantification was performed using mouse ELISA Kits (DuoSet®, R&D Systems, Minneapolis, MN, USA) according to the manufacturer's instructions. IL-5 was measured in cell-free bone marrow culture supernatants and serum. CCL24/Eotaxin-2 and IL-33 was measured in cell-free BAL. Absorbance or luminescence was measured on a Varioskan™ LUX multimode microplate reader (ThermoFisher Scientific Vantaa, Finland). Samples below detection limit was set to zero.

The statistical analysis was performed with GraphPad Prism 8 Software (GraphPad Software Inc, La Jolla, CA, USA). Kruskal-Wallis test was used to determine the variance among more than two groups. If significant variance was found, the nonparametric Mann-Whitney U test was used for analysis between two independent groups. Paired Student's t-tests were applied for the analysis of in vitro studies. Statistical significance was defined as ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and, ^****P < 0.0001.

Investigations of the requirement of ILC2s in IL-33-mediated bone marrow eosinophilopoiesis were carried out using Rag1^−/− mice which have an intact innate immune system, but lack mature T and B cells. Naïve WT and Rag1^−/− mice were challenged with rmIL-33 (Figure 1A) which resulted in increased eosinophil numbers in both bone marrow (Figures 1B,C) and BAL (Figures 1B,D) compared to control mice that received PBS vehicle (25). Furthermore, IL-33 challenge caused elevated levels of the eosinophil chemokine CCL24/Eotaxin-2 in BAL in both mouse strains where higher levels of CCL24/Eotaxin-2 were observed in Rag1^−/− mice compared to WT mice (Figure 1E). Mature eosinophils were significantly increased in the bone marrow of mice exposed to IL-33 (Figure 1F) whereas the relative number of eosinophil progenitors (Figure 1G) remained unchanged. Importantly, WT and Rag1^−/− mice responded with a similar increase of eosinophils after IL-33 challenge, suggesting that adaptive immunity is dispensable in both IL-33-driven eosinophil production and recruitment of eosinophils from the bone marrow. Thus, in the current study by using Rag1^−/− mice, we confirm our previous findings suggesting that bone marrow ILC2s support eosinophil development in WT mice in IL-33-induced inflammation (25).

The number of bone marrow ILC2s in Rag1^−/− mice were significantly increased when compared to WT controls (Figures 2A,B) at baseline. Thus, IL-33 challenge resulted in a significant increase of ILC2s in WT bone marrow only, and not in bone marrow from Rag1^−/− mice (Figure 2A). Moreover, ILC2s from both mouse strains responded to IL-33 challenge with increased ST2 expression (Figure 2C). Of note, 100% of bone marrow ILC2s (SSC^loLin⁻CD45⁺CD127⁺CD25⁺) in PBS and IL-33 challenged mice were ST2⁺, while IL-33 challenge further increased the ST2 receptor expression (Figure 2C). Furthermore, these results were consistent with an approximate 2-fold upregulation of ST2 expression evaluated after IL-33 stimulation of WT and Rag1^−/− bone marrow cells ex vivo (Figure 2D). In addition, stimulation with IL-33 in ex vivo cultures generated high levels of IL-5⁺ ILC2s in both mouse strains (Figures 2E,F), which suggests that bone marrow ILC2s contribute to IL-33-induced eosinophil development in vivo independent of adaptive immunity.

To investigate eosinophil development and the downstream effects of IL-33 in allergic inflammation WT and Rag1^−/− mice were subjected to a model of allergic inflammation using the protease allergen, papain (Figure 3A). Increased numbers of eosinophils in both bone marrow (Figure 3B) and BAL (Figure 3C) were detected in WT and Rag1^−/− mice following papain challenge. Furthermore, both WT and Rag1^−/− mice demonstrated elevated levels of CCL24/Eotaxin-2 and IL-33 in BAL following papain challenge (Figures 3D,E) allowing mature eosinophils to migrate toward the airways. A marginally decreased relative number of eosinophil progenitors was observed (Figure 3F and Figure S1A) whereas an approximate 2-fold increase of mature eosinophils was found in papain-challenged WT and Rag1^−/− mice compared to saline exposed control mice (Figure 3G and Figure S1B). We have previously reported a higher amount of ST2⁺ mature bone marrow eosinophils in IL-33 induced inflammation (25). We next investigated the amount of ST2⁺ mature bone marrow eosinophils during allergic inflammation and whether adaptive immunity is required for the generation of ST2-expressing eosinophils. An increase of ST2⁺ mature eosinophils was detected in WT and importantly, elevated levels of ST2⁺ mature eosinophils were also detected in papain challenged Rag1^−/− mice (Figure 3H and Figure S1C). Altogether, these results indicate that adaptive immunity is not required for induction of papain-driven eosinophilic inflammation in the bone marrow where IL-33 is an important mediator.

An increased relative number of bone marrow ILC2s was observed in WT mice airway challenged with papain compared to PBS exposed control mice whereas Rag1^−/− mice had a higher number of bone marrow ILC2s already at baseline (i.e., PBS exposed control) (Figure 4A). Bone marrow ILC2s from both WT and Rag1^−/− mice upregulated ST2 expression after papain challenge, which is crucial for the activation of ILC2 effector functions (Figure 4B). Moreover, high ST2 expression on ILC2s demonstrated a positive correlation with both increased bone marrow eosinophilia (Figure 4C) and an increased number of ST2⁺ mature eosinophils (Figure 4D). Collectively, these data indicate that IL-33-responsive bone marrow ILC2s contribute to allergen-induced bone marrow and airway eosinophilia.

We next performed a time course to identify when bone marrow ILC2s respond to the inhaled protease allergen papain. WT mice were exposed to a single dose of papain intranasally (Figure 5A). Elevated levels of eosinophils in bone marrow were detected at 48 h (Figure 5B) whereas a higher relative number of BAL eosinophils was seen at 24 h (Figure 5C). In line with increased eosinophil levels in the airways there was a significant increase of ST2 expression on ILC2s in the bone marrow (Figure 5D). Additionally, high levels of BAL CCL24/Eotaxin-2 (Figure 5E) and a trend toward elevated IL-33 levels in BAL (Figure 5F) were detected at this time point. At 48 h, the ST2 expression on bone marrow ILC2s was significantly lower and CCL24/Eotaxin-2 and IL-33 levels in BAL showed a trend toward decreased levels. At 72 h the inflammation in the airways was resolved where BAL eosinophil levels were similar to control mice (Figure 5C). In line with these data, the ST2 expression on bone marrow ILC2s was significantly decreased along with lower BAL CCL24/Eotaxin-2 and IL-33 levels. These data suggest that eosinophils are no longer recruited to the airways at 72 h post airway challenge. Collectively, our results suggest that bone marrow ILC2s act rapidly in response to an inhaled allergen by upregulating its ST2 receptor expression in the bone marrow at the onset of eosinophil development.

To determine whether IL-33-induced eosinophilia can develop in the absence of ILC2s, we challenged naïve lymphocyte deficient Rag2^−/−Il2rg^−/− mice with rmIL-33 (Figure 6A). No induction of eosinophils was detected by differential cell count analysis of the bone marrow (Figures 6B,C) and BAL (Figures 6B,D) samples from Rag2^−/−Il2rg^−/− mice. The relative number of eosinophil progenitors was lower in Rag2^−/−Il2rg^−/− bone marrow compared to WT (Figure 6E). The most dramatic difference was observed in the number of mature eosinophils which was severely impaired in the bone marrow of both IL-33 and PBS challenged Rag2^−/−Il2rg^−/− mice (Figure 6F). Together, these data suggest that innate lymphocytes are essential for the maintenance of eosinophils under homeostatic conditions as well as for the maturation of eosinophils in response to inflammatory signals.

Both systemic IL-5 and local IL-5 production have been suggested to regulate eosinophil development and maturation in the bone marrow (11, 21, 25). Next, we analyzed the concentration of IL-5 in serum from WT, Rag1^−/− and Rag2^−/−Il2rg^−/− mice. IL-33 challenge significantly increased levels of IL-5 in serum of WT and Rag1^−/− mice but not in serum of Rag2^−/−Il2rg^−/− mice (Figure 7A). Notably, the level of serum IL-5 was higher in IL-33 challenged Rag1^−/− mice compared to WT mice (Figure 7A), which might be associated with a greater number of ILC2s in the bone marrow of Rag1^−/− mice. Furthermore, bone marrow cultures from WT and Rag1^−/− mice stimulated with IL-33 accumulated IL-5 in the culture media, but no IL-5 was detected in cultures from Rag2^−/−Il2rg^−/− mice (Figure 7B). Interestingly, we found that bone marrow cultures generated from in vivo IL-33 challenged mice produced significantly more IL-5 upon IL-33 re-stimulation ex vivo (Figure 7C). This result suggests that in IL-33 challenged mice presenting an increased expression of ST2 on bone marrow ILC2s makes the cells more susceptible to IL-33, thus, potentially leading to a higher production of IL-5, which in turn drives eosinophil development.

To assess whether IL-5 is required in the development and maturation of bone marrow eosinophilia in Rag1^−/− and WT mice, mice were pre-treated with anti-IL-5 antibodies 1 h before the dose regimen of intranasal papain exposures (Figure 8A). Anti-IL-5 treated papain challenged Rag1^−/− and WT mice were unable to induce eosinophilia in the bone marrow (Figures 8B,D) and BAL (Figure 8C). Moreover, a decreased number of ST2⁺ mature eosinophils was seen in both mouse strains treated with anti-IL-5 (Figure 8E). Interestingly, a significant decrease of the ST2 expression on mature eosinophils was observed in both Rag1^−/− and WT mice (Figure 8F). These findings suggest link between IL-5 levels and ST2 expression on eosinophils in the bone marrow. The relative number of eosinophil progenitors (data not shown) and ILC2s were not affected by anti-IL-5 treatment (Figure S2). Collectively, these data show that papain induced eosinophilic inflammation is dependent on IL-5 and that IL-5 levels might regulate ST2 expression on bone marrow eosinophils which in turn drives the maturation of eosinophil within the bone marrow. In addition, the absence of IL-5-producing T cells in Rag1^−/− mice make bone marrow ILC2s a potential local source of IL-5.