Airway Epithelial Cells Are Crucial Targets of Glucocorticoids in a Mouse Model of Allergic Asthma

This article is featured in In This Issue, p.1

Asthma is a chronic lung disease that is characterized by inflammation, structural changes, and hyperresponsiveness of the airways. During the last decades, the prevalence of asthma has greatly increased, currently affecting up to 300 million people worldwide. The majority of patients suffer from an allergic variant of the disease (1) that is triggered by an IgE-driven immune response directed against inhaled Ags and leads to various symptoms, such as wheezing, coughing, and breathing difficulties. The pathomechanism of allergic asthma involves a complex interplay between the immune system and parenchymal cells of the lung, including the airway epithelium (2). Inhaled allergens are phagocytosed by macrophages and dendritic cells (DCs) presented on MHC class II molecules and initiate the differentiation of Th2 cells and a humoral immune response. Following class switching, Ag-specific B cells secrete IgE, which causes degranulation of mast cells. Cytokines, such as IL-4, IL-5, and IL-13, are produced by Th2 cells, type 2 innate lymphoid cells (ILC2s), and airway epithelial cells, and they trigger pathological events, including airway wall remodeling, bronchial hyperresponsiveness, and goblet cell metaplasia (3). Once the immune response has been initiated, eosinophils become the major effector cells that are responsible for airway dysfunction. In addition to the importance of immune cells in allergic asthma, there is evidence for a prominent role for airway epithelial cells in this disease (4–6). In particular, alveolar type 2 (AT2) cells produce a variety of cytokines, including IL-33, and thereby orchestrate inflammation in the airways, for instance by activating ILC2s (7–9). In line with this crucial function, several genes associated with an increased susceptibility to asthma are expressed by airway epithelial cells (10, 11). Taken together, the pathogenesis of allergic asthma involves a complex interplay between cell types of hematopoietic and nonhematopoietic origin, and many of its details are not clear.

A number of new drugs are being tested for the clinical management of asthma, in particular therapeutically active mAbs; nevertheless, synthetic glucocorticoids (GCs) often remain the measure of choice (12–15). Locally and systemically applied GCs lead to an efficient relief of disease symptoms and are available as derivatives with various activity profiles. However, adverse effects and the development of GC resistance complicate their use. Thus, a better understanding of the mode of GC action is important and may allow for improved therapeutic usage of GCs in the future.

Analysis of mice with an impaired GC receptor (GR) dimerization interface (16) revealed that intact gene regulation in immune cells is vital for many of the anti-inflammatory activities of GCs (17, 18). Experiments with conditional knockout mice further helped to define cell type–specific GR functions in the pathogenesis and treatment of contact dermatitis, multiple sclerosis, sepsis, and rheumatoid arthritis and revealed that T cells and myeloid cells are the predominant GC targets in these diseases (19–22). In this study, we set out to define the therapeutically relevant cell types and mechanisms in the management of allergic asthma by GCs and demonstrate that the airway epithelium, in particular AT2 cells, plays an important role in this highly prevalent inflammatory lung disease.

All mice were bred under specific pathogen–free conditions in our animal facilities at the Universities of Göttingen and Ulm. Mice of both genders were used at 8–12 wk of age. GR^wt (Nr3c1) and GR^dim (Nr3c1^tm3GSc) mice were on a BALB/c background (16, 23). GR^flox (Nr3c1^tm2GSc), GR^lysM [Nr3c1^tm2GScLyz2^tm1(Cre)Ifo], GR^lck [Nr3c1^tm2GScTg^{(Lck-cre)1Cwi}] and GR^lcklysM [Nr3c1^tm2GSc Tg^{(Lck-cre)1Cwi}Lyz2^tm1(Cre)Ifo] mice (24), as well as GR^cd19 [Nr3c1^tm2GScCD19^tm1(cre)Cgn], GR^cd11c [Nr3c1^tm2GScTg^{(Itgax-cre)1-1Reiz}], and the corresponding GR^flox mice were on a C57BL/6 background (20). GR^spc [Nr3c1^tm2GScSftpc^{tm1(cre/ERT2)Blh}] mice on a C57BL/6 background were generated by crossing GR^flox mice with Sftpc-CreER^T2 [Sftpc^{tm1(cre/ERT2)Blh}] mice kindly provided by Dr. B. Hogan (Duke University, Durham, NC) (25). To induce Cre-mediated recombination in GR^spc mice, tamoxifen (Sigma-Aldrich, Taufkirchen, Germany) was dissolved in EtOH and sunflower oil at a 1:20 ratio and administered by oral gavage at a concentration of 20 mg/ml in 150 μl three times every other day (9 mg in total). All experiments were conducted according to Lower Saxony state regulations for animal experimentation and were approved by the responsible authority (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit, Oldenburg, Germany).

Bone marrow (BM) chimeras were generated using GR^wt and GR^dim mice acting as donor or host. Recipient mice were subjected to total body irradiation 24 h prior to BM transfer (8.5 Gy) at a rate of 1 Gy/min using a RS 225 X-Ray Research System (Gulmay Medical Systems, Camberley, Surrey, U.K.) operated at 200 kV and 15 mA and with 0.5-mm Cu filtration. The next day, BM was isolated from femur and tibia of the donor mice. A cell suspension was prepared in PBS using a 40-μm cell strainer and injected into the irradiated mice i.v. at a concentration of 4 × 10⁷ cells in a total volume of 100 μl. Antibiotic therapy consisting of 25 μg/ml neomycin sulfate (Sigma-Aldrich) was started on the day of irradiation and lasted for 3 wk. Reconstitution of the immune system and the identity of the different types of chimeric mice were confirmed 6 wk after BM transfer by genotyping peripheral blood and tail biopsies by PCR using primers located in exons 3 and 5 of the GR gene, followed by digestion of the PCR product with BsrGI and agarose gel electrophoresis, as described (16).

Allergic airway inflammation (AAI) was induced in mice using chicken egg OVA, as previously described (26), and served as a model of allergic asthma (27). Mice were sensitized on days 0, 7, 14, and 21 by i.p. injections of 10 μg of OVA (Sigma-Aldrich), together with 2 mg of the adjuvant alum (Alhydrogel 2%; InvivoGen, Toulouse, France), in a total volume of 200 μl of PBS. Control mice received only alum. On days 28 and 29, mice were challenged i.n. with 250 μg of OVA solubilized in 20 μl of PBS under isoflurane anesthesia. Half of the immunized mice were additionally injected i.p. with dexamethasone (Dex; Ratiopharm, Ulm, Germany) at a dose of 10 mg/kg body weight 1 h after each challenge step. Two days after the second challenge, mice were sacrificed with a lethal dose of CO₂, and bronchoalveolar lavage fluid (BALF), serum, spleen, and lung were collected for further ex vivo analyses.

Lung function was assessed in mice treated with OVA and Dex as described above, and the responsiveness of the airways to increasing concentrations of methacholine (Sigma-Aldrich) was recorded using flexiVent (Scireq, Montreal, QC, Canada) (28, 29). Mice were anesthetized by i.p. injection of 1 mg/kg urethane and subsequently received 0.125% curare to stop the respiration (both from Sigma-Aldrich). Thereafter, a PE-90 (1.27 mm OD, 0.86 mm ID) cannula was inserted into the trachea, and the mice were connected to a computer-controlled piston pump and quasi-sinusoidally ventilated to achieve a mean lung volume close to spontaneous breathing. The use of the MSX (Multi-Subject Extension) allowed us to assess lung function in up to four mice in parallel. After baseline measurement, mice were challenged for 10 s with saline aerosol and, at 4.5-min intervals, with methacholine at increasing doses (from 25 to 800 mg/ml). For each dose, the peak response was calculated as the mean of the three maximal or minimal values and was used for calculation of resistance (expressed as cm H₂O × s/ml), elastance (expressed as cm H₂O/ml), and compliance (expressed as ml/cm H₂O) as a measure of airway hyperresponsiveness (AHR).

BALF was obtained by cannulation of the trachea and flushing the lung three times with 1 ml of 0.1% BSA (26). Prior to Ab staining, BALF was treated with erythrocyte lysis buffer (20 mM Tris/HCl, 155 mM NH₄Cl [pH = 7.2]). Blood was collected by cardiac puncture and similarly subjected to erythrolysis. Cell counts were determined using a Neubauer hemocytometer. FcR blockade was performed by incubation with TruStain fcX (anti-mouse CD16/32; clone 93). Cell suspensions were stained with mAb mixtures containing anti-mouse CD4 (clone RM 4-5), anti-mouse CD3 (clone 17 A 2), anti-mouse CD8α (clone 53-6.7), anti-mouse F4/80 (clone cI:13-1), anti-mouse CD11b (clone M1/70), anti-mouse Gr-1 (clone RB6-8C5), and anti-mouse Siglec-F (clone E-502440). All Abs were from BioLegend (Uithoorn, the Netherlands) or BD Biosciences (Heidelberg, Germany). Data were acquired on a FACSCanto II flow cytometer (BD Biosciences) and analyzed using FlowJo software (TreeStar, Ashland, OR).

Flat-bottom 96-well plates were coated with 50 μg/ml OVA in 0.1 M Na₂CO₃ (pH  = 9.5) overnight at 4°C (26). After washing with 0.05% Tween-20 and blocking with 10% FCS for 1 h, serum samples were added to the plate and incubated for 1 d at 4°C. For IgE detection, serum was preincubated with Protein G PLUS agarose (Santa Cruz Biotechnology, Heidelberg, Germany) to remove excess IgG. OVA-specific Igs were detected using HRP-coupled anti-IgG1–, anti-IgG2a–, and anti-IgE–specific Abs (SouthernBiotech, Birmingham, AL). After incubation at room temperature for 1 h, the plate was washed, and the color reaction was developed by adding tetramethylbenzidine and H₂O₂. Absorption was measured at 450 and 570 nm using a PowerWave 340 microplate reader (BioTek, Winooski, VT).

Spleens were passed through a 40-μm cell strainer and washed in 0.1% BSA. Erythrocytes were lysed, and cells were plated in round-bottom 96-well plates at a concentration of 3 × 10⁵ cells per well. Splenocytes were cultured in RPMI 1640 medium with 10% FCS, 100 U/ml penicillin, and 100 μg/ml streptomycin (all from Invitrogen, Karlsruhe, Germany) and were restimulated ex vivo with 10 μg/ml OVA for 72 h at 37°C and 5% CO₂. Unstimulated cells served as controls and were used to calculate the proliferation index. For quantification of IL-2 and IL-4 levels, a 50-μl aliquot was taken from each well and analyzed using commercially available ELISA kits, according to the manufacturer’s instructions (BioLegend). To analyze proliferation, 37 kBq of [³H]thymidine (Hartmann Analytics, Braunschweig, Germany) was added to each well and incubated for another 16 h. The labeled DNA was collected onto Filtermat A glassfiber filters using a MicroBeta FilterMate-96 Harvester and was encapsulated by a MeltiLex solid scintillator. The incorporated radioactivity was determined using a MicroBeta² β-Scintillation counter (all from PerkinElmer, Rodgau, Germany).

Lungs were flushed extensively with 0.1% BSA after cannulation of the trachea to remove the infiltrating leukocytes. Subsequently, total RNA was isolated with an RNeasy Mini Kit (QIAGEN, Hilden, Germany) and reverse transcribed into cDNA using an iScript cDNA Synthesis Kit (Bio-Rad, Munich, Germany), according to the manufacturers’ instructions. For relative quantification of gene expression, an RT-PCR reaction was performed using the ABI 7500 Real-Time PCR System (Applied Biosystems, Darmstadt, Germany) in combination with Power SYBR Green PCR Master Mix from the same company. Detection of individual transcripts was achieved using the following primer combinations: IL-5 (5′-TGC TGA AGG CCA GCG CTG AAG-3′ and 5′-GGG ACA GGA AGC CTC ATC GTC TCA T-3′), IL-13 (5′-CCC CTG TGC AAC GGC AGC AT-3′ and 5′-CGG GGA GGC TGG AGA CCG TA-3′), IL-33 (5′-TCA ATC AGG CGA CGG TGT GGA-3′ and 5′-AAG GCC TGT TCC GGA GGC GA-3′), MCP-1 (5′-AGC ACC AGC CAA CTC TCA CT-3′ and 5′-CGT TAA CTG CAT CTG GCT GA-3′), Occludin (5′-CCT CCA CCC CCA TCT GAC TA-3′ and 5′-CTT CAG GCA CCA GAG GTG TT-3′), and Claudin 5 (5′-GAG ATC CTG GGG GCA CTA GA-3′ and 5′-TGC CCT TTC AGG TTA GCA GG-3′). Values were normalized to HPRT (5′-GTC CTG TGG CCA TCT GCC TA-3′ and 5′-GGG ACG CAG CAA CTG ACA TT-3′), which served as a housekeeping gene, and evaluated using the ΔΔ cycle threshold method.

Lungs were fixed at room temperature in Roti-Histofix 4% (Carl Roth, Karlsruhe, Germany) overnight before they were dehydrated and embedded in paraffin (26). Five-micron-thick tissue sections were prepared and stained with H&E, according to standard protocols. Photomicrographs were acquired using an Olympus BX-41 microscope (Hamburg, Germany).

For immunohistochemical staining, 2-μm-thick tissue sections were incubated in EnVision FLEX Target Retrieval Solution, Low pH for staining of CD3, CD8, CD68, Foxp3, and Gr-1 or High pH for staining of Muc5b (Dako Agilent Technologies, Santa Clara, CA). This was followed by incubation with primary Abs against CD3 (1:2000; Santa Cruz Biotechnology), CD8 (1:2000; Bioss Abs, Woburn, MA), CD68 (1:200; Abcam, Cambridge, U.K.), Foxp3 (1:1000; Abcam), Gr-1 (1:200; BD Biosciences), or Muc5B (1:50; GeneTex, Irvine, CA) for 30 min at room temperature. Polymeric secondary Abs coupled to HRP (ImmPRESS HRP Polymer Detection Kit; Vector Laboratories, Burlingame, CA) and DAB (Dako Agilent Technologies) were used to visualize the sites of immunoreactivity. Counterstaining was done with hematoxylin. Photomicrographs were acquired using a Leica Axio Scope A1 microscope (Leica Microsystems, Wetzlar, Germany).

Statistical analysis was performed by one-way ANOVA using GraphPad Prism software (GraphPad, La Jolla, CA). Outlying sample exclusion was done with GraphPad Prism Outlier Calculator (GraphPad). All data are depicted as mean ± SEM; *p < 0.05, **p < 0.01, ***p < 0.001.

To test whether an intact GR dimerization interface was required for successful GC therapy of allergic asthma, we initially used GR^dim mice carrying the point mutation A458T in the DNA-binding domain (16). GR^wt and GR^dim mice were sensitized with OVA i.p., followed by intranasal challenge with the same Ag to induce AAI, a major hallmark of allergic asthma (Fig. 1A). Half of the immunized mice also received Dex by i.p. injection as a model of systemic GC therapy, and mice treated only with adjuvant served as controls. Histological analysis of the lung revealed massive leukocyte infiltration, increased mucus production based on Muc5B immunostaining, and marked bronchoconstriction after AAI induction, which was prevented by concomitant treatment with Dex in GR^wt mice but not GR^dim mice (Fig. 1B, 1C). Immunization with OVA also increased the abundance of CD3⁺ and CD8⁺ T cells, Foxp3⁺ regulatory T cells, CD68⁺ monocytes/macrophages, and Gr-1⁺ neutrophils in the lung (Fig. 2, Supplemental Fig. 1). GC therapy inhibited this cellular influx in GR^wt mice, whereas GR^dim mice did not respond to Dex treatment and appeared to contain more neutrophils in the lung under these conditions (Fig. 2, Supplemental Fig. 1).

To confirm our observations, BALF was collected and analyzed by flow cytometry. In line with the histological data, induction of AAI resulted in strongly increased leukocyte numbers in the airways (Fig. 3). The infiltrate in both genotypes was dominated by eosinophils, with minor populations of macrophages, neutrophils, and T cells (Fig. 3). Although GR^dim and GR^wt mice developed AAI to a similar degree, GR^dim mice were refractory to the repressive effect of Dex (Fig. 3). In agreement with our previous observation, GC therapy increased neutrophil numbers in BALF of GR^dim mice, although statistical significance was not reached (Fig. 3). Taken together, our findings indicate that an intact dimerization interface of the GR is required for the beneficial effect of GC treatment on the development of AAI in mice.

AHR is another hallmark of allergic asthma, which prompted us to study the role of an intact GR dimerization interface for the ability of GCs to resolve airflow limitation after methacholine exposure. To this end, GR^wt and GR^dim mice were treated with OVA similarly as with the previous experiment, and half of the immunized mice also received Dex (Fig. 1A). AHR was observed in GR^wt and GR^dim mice, given that increased resistance and elastance of the airways developed with ascending doses of methacholine, whereas compliance concomitantly decreased (Fig. 4). Although Dex administration improved the airflow in GR^wt mice after methacholine challenge, the same treatment had no effect in GR^dim mice (Fig. 4). These data indicate that the GR point mutation A458T confers resistance to Dex application in AHR therapy, reconfirming a crucial role for an intact GR dimerization interface in the control of allergic asthma by GCs.

In search of the mechanisms by which GCs repress AAI, we first studied the role of Ag-specific B cells. Mice were subjected to the same experimental procedure as described above (Fig. 1A) and analyzed for serum levels of Ag-specific IgE, IgG1, and IgG2a. Production of OVA-specific Abs was strongly increased after AAI induction, whereas Dex did not have an effect on their levels in GR^wt nor GR^dim mice (Fig. 5A). Second, we tested peripheral T cell function after AAI induction. Following Ag restimulation of splenocytes in vitro, T cell proliferation and IL-2 and IL-4 production were increased in immunized mice, but Dex treatment had no impact on these T cell parameters in either genotype (Fig. 5B). Collectively, our findings suggest that amelioration of AAI after GC therapy can not be explained by modulation of Ag-specific B or T cells.

To define the cell types responsible for the beneficial effects of GCs in the treatment of AAI, we used a panel of cell type–specific GR-knockout mice (20, 24). Initially, we analyzed mice lacking the GR in T cells (GR^lck), myeloid cells (GR^lysM), or both (GR^lcklysM). AAI was induced as in previous experiments (Fig. 1A), and the effect of Dex treatment was compared between mutant mice and GR^flox littermate controls. Quantification of inflammatory cells in BALF revealed similar levels upon AAI induction for all mutant mouse strains and a comparable reduction in these numbers when mice were also treated with Dex (Fig. 6A). To test for a role of other immune cells in mediating the therapeutic effects of GCs, we subsequently analyzed GR^cd19 mice that carry a deletion of the GR in B cells or GR^cd11c mice that are devoid of the GR in DCs. However, enumeration of leukocytes in BALF after induction of AAI indicated that, in this case too, the GR was dispensable for therapeutic efficacy of Dex (Fig. 6B). Notably, measurement of OVA-specific IgG1 Ab levels confirmed that immunization had been successful in all mouse strains (Supplemental Fig. 2A, 2B). Collectively, the presence of the GR in individual leukocyte subsets is not required for GC therapy of AAI.

To obtain further evidence for a possible role of immune cells in mediating the therapeutic effects of GCs, we compared mice in which the GR dimerization interface was disrupted in the entire hematopoietic system or in the radioresistant nonhematopoietic compartment (Fig. 6C). For the generation of chimeric mice by BM transplantation, we used GR^dim mice instead of mice completely lacking the GR, because a germline deletion of the GR is not viable (30), and the A458T point mutation was sufficient to confer resistance of AAI to GC therapy (Figs. 1, 3). GR^dim BM was transplanted into irradiated GR^wt mice (GR^dim → GR^wt); conversely, GR^wt BM was transferred into GR^dim recipients (GR^wt → GR^dim). Furthermore, autologous transplantations were conducted as controls (GR^wt → GR^wt and GR^dim → GR^dim). After reconstitution of the immune system, AAI was induced in the different chimeric mice as described above (Fig. 1A), and half of each of them were treated with Dex. Mice receiving only the adjuvant served as controls. Expectedly, the two control groups behaved similarly as nontransplanted GR^wt and GR^dim mice (compare Fig. 6C with Fig. 3). Arguing against a role for the hematopoietic system in mediating the effects of GCs, we found that GR^dim → GR^wt chimeric mice, which have a mutant GR in the hematopoietic system but a wild-type GR in all other organs, were fully treatable with Dex (Fig. 6C). In sharp contrast, GR^wt → GR^dim chimeric mice, in which the GR point mutation A458T was restricted to radioresistant cells of presumed nonhematopoietic origin, did not respond to GC therapy (Fig. 6C). Analysis of OVA-specific IgG1 confirmed successful immunization of all groups of chimeric mice (Supplemental Fig. 2C). Based on these findings, we conclude that cells of nonhematopoietic origin are essential for the repression of AAI by Dex.

Our unexpected finding that the hematopoietic compartment was dispensable for successful GC treatment of AAI provoked the question about which other cell types might be responsible. Thus, we turned our attention to the lung parenchyma, which mainly consists of epithelial cells, fibroblasts, smooth muscle cells, and endothelial cells. AT2 cells, a subtype of airway epithelial cells in the alveoli, are best known to fulfill immunological functions, such as Ag presentation and cytokine production. Hence, we disrupted the GR in these cells by treating inducible GR^spc mice with tamoxifen (25, 31). Sftpc-CreER^T2–knock-in mice that we used to generate this mouse model express a modified Cre recombinase under the control of the surfactant protein C promoter and were shown to mediate gene recombination in 84% of AT2 cells (25). AAI was induced in GR^flox and GR^spc mice by immunization and challenge with OVA as previously (Fig. 7A). This procedure resulted in enhanced leukocyte infiltration into the lung, bronchoconstriction, and augmented mucus production in the airways (based on Muc5B staining), regardless of the genotype (Fig. 7B, 7C). In addition, the numbers of CD3⁺ and CD8⁺ T cells, Foxp3⁺ regulatory T cells, CD68⁺ monocytes/macrophages, and Gr-1⁺ neutrophils in the lung were increased (Fig. 8, Supplemental Fig. 3). Flow cytometric analysis of BALF revealed massive eosinophilia and confirmed the influx of macrophages, neutrophils, and T cells into the lung following AAI induction (Fig. 9). Analysis of OVA-specific IgE and IgG1 serum levels affirmed that both mouse strains had been successfully immunized (Supplemental Fig. 4). Most importantly, the repressive effect of Dex on leukocyte infiltration into the lung, which was observed in GR^flox mice as expected, was strongly compromised, although not fully abolished, in GR^spc mice (Figs. 7B, 9). In line with these data, GC treatment reduced the influx of eosinophils, macrophages, neutrophils, and major T cell subsets into the lung of GR^flox mice but not GR^spc mice (Figs. 8, 9, Supplemental Fig. 3). Furthermore, bronchoconstriction and mucus hyperproduction were resolved in GR^flox mice after GC therapy, whereas they largely persisted in GR^spc mice (Fig. 7C). Taken together, AT2 cells play a critical role in the therapeutic effects of Dex in AAI, although other parenchymal cell types, such as endothelial cells or smooth muscle cells, are presumably targets of GC activity as well.

Because GR^dim and GR^spc mice were similarly found to be unresponsive to Dex therapy for AAI, we speculated that control of gene expression in airway epithelial cells was a critical mechanism of GCs in repressing lung inflammation. To address this issue, AAI was induced in both mouse models and treated with Dex, as in previous experiments (Figs. 1A, 7A). To restrict gene expression analysis to parenchymal cells, RNA was isolated from lungs that were flushed extensively to remove infiltrating leukocytes. Initially, we used quantitative RT-PCR to analyze cytokines and chemokines that are known to be associated with asthma pathogenesis and to be produced by airway epithelial cells. As expected, expression of IL-5, IL-13, IL-33, and MCP-1 was enhanced after AAI induction in all strains of mice (Fig. 10). However, repression by Dex was not observed in GR^dim or GR^spc mice (Fig. 10). Furthermore, there was a tendency toward decreased expression of the tight junction proteins occludin and claudin 5 after Dex treatment in the lungs of GR^wt and GR^flox mice but not in the two mutant mouse strains (Fig. 10). Although statistical significance was not always reached, our data strongly suggest that lack of AAI responsiveness to GCs after disrupting the GR dimerization interface or deleting the GR in AT2 cells was linked to the inability of Dex to regulate gene expression in airway epithelial cells (Fig. 11).

Asthma is one of the most prevalent inflammatory diseases worldwide, it severely affects the patients’ quality of life, and it puts health care systems under financial pressure (1, 2). Intensive research efforts have led to improved therapeutic regimens; however, despite these gains, many challenges remain (7, 14). Synthetic GCs continue to be one of the most effective medications to achieve long-term control of clinical symptoms (13–15, 32), but complications, such as GC resistance and growth retardation in children, limit their application (33). A major caveat of the therapeutic usage of GCs is their broad activity profile, which is a consequence of the GR’s ubiquitous expression. The fact that the GR is expressed throughout the body explains the adverse effects, but it also raises the question about which cell types are crucial for therapeutic efficacy. Furthermore, GCs regulate transcription and also exert nongenomic effects, for instance on the PI3K pathway (34), the cytoskeleton (35), or TCR signaling (36, 37). However, the relevance of these activities is unknown; a better understanding of the mode of GC action might ultimately allow asthma care to be more specific and tolerable in the future.

Essentially every cell type in the body is a potential target of GC action, and our results now provide provocative new insight about which of them are essential for the inhibition of AAI. Unexpectedly, airway epithelial cells, rather than immune cells, turned out to be responsible, given that neither deletion of the GR in T cells, macrophages, granulocytes, B cells, or DCs, nor disruption of its dimerization interface in cells of the entire radiosensitive hematopoietic system compromised the ability of Dex to repress AAI. Conversely, mice carrying the A458T point mutation in radioresistant cells of presumed nonhematopoietic origin or a GR deletion in AT2 cells were fully or at least largely resistant to GCs, highlighting a crucial role for the GR in airway epithelial cells for AAI therapy. With regard to the observation that partial repression of leukocyte infiltration into the lung was still observed in GR^spc mice, one has to keep in mind that other cell types of the lung parenchyma, such as endothelial cells, smooth muscle cells, or bronchial epithelial cells, might also contribute to therapeutic efficacy. Furthermore, Cre recombination in GR^spc mice occurs in only 84% of AT2 cells, leaving some of them GC responsive (25). Although the importance of airway epithelial cells in allergic asthma is well established (38), the dispensability of immune cells for the treatment of AAI by GCs still came as a surprise. Allergic asthma, both in men and mice, involves differentiation of Ag-specific Th2 and B cells, which is assisted by macrophages and DCs and causes infiltration of eosinophils into the airways. Nevertheless, repression of any of these activities via the cell-intrinsic action of the GR in immune cells is apparently sufficient to reduce airway inflammation. The argument that disruption of the GR in single-leukocyte subtypes might not suffice to abolish the beneficial effects of GC therapy is refuted by our results obtained in chimeric mice transplanted with GR^dim BM. In this model, the majority of hematopoietic cells lack an intact GR dimerization interface, with a few exceptions, such as mast cells and tissue-resident macrophages. Nevertheless, the chimeric mice still responded to Dex treatment. These data strongly support the argument that immune cells are not essential targets for the therapeutic efficacy of GCs in AAI.

Surprisingly, Dex treatment of GR^dim mice not only failed to reduce neutrophil infiltration into the airways, it even appeared to increase their numbers in the lung. Thus, it is intriguing to speculate that GC therapy in the absence of an intact GR dimerization interface might shift the asthma phenotype from a predominantly eosinophilic form to one with more involvement of neutrophils. However, it remains unclear whether the different responsiveness of eosinophils and neutrophils to Dex is a cell-intrinsic effect or is due to the regulation of chemoattractive factors or adhesion molecules.

Another important finding of our study is that AAI and AHR, two hallmarks of allergic asthma, were refractory to GC therapy in GR^dim mice carrying the point mutation A458T that disrupts the GR dimerization interface (16). Transcriptional control by the GR has clearly more facets than can be distinguished by this mutation. For instance, recent studies suggested that the GR can bind DNA and regulate transcription, even in the absence of dimerization (39–41). However, because modulation of at least a subset of GC target genes is abolished in GR^dim mice (42, 43), this model still allowed us to assess the importance of gene regulation in allergic asthma in mice. Consistent with a role for transcriptional control, we found that leukocyte infiltration into the airways and the eosinophilia were inhibited after Dex treatment in GR^wt mice but not in GR^dim mice. Moreover, GCs were unable to prevent bronchoconstriction or mucus hyperproduction or to improve airway flow after methacholine exposure in the absence of an intact GR dimerization interface. These findings indicate that genes that are sensitive to the A458T point mutation need to be regulated to ensure successful GC therapy in the used OVA-induced model of allergic asthma. Similar findings were made previously for the anti-inflammatory activity of the GR in other diseases. For instance, Dex application to GR^dim mice failed to resolve inflammation in mouse models of contact allergy, arthritis, and acute lung injury (19, 20, 44). In contrast, clinical symptoms were ameliorated after Dex treatment in a model of multiple sclerosis (37). Thus, for some medical indications, it appears desirable to identify selective GR ligands that favor GR dimerization (45). Considering the finding that a monomeric GR was responsible for induction of osteoporosis, such a strategy might even make it possible to avoid some adverse effects of GCs (46).

Both mutant mouse strains that did not respond to Dex treatment of AAI (i.e., GR^dim and GR^spc mice), behaved similarly with regard to gene expression in the lung. This finding indicates that GR-dependent transcriptional regulation in AT2 cells constitutes a major mechanism that is responsible for the therapeutic benefit of GCs in allergic asthma, at least in mice, and raised the question about the relevant GR target genes. Our study examined several candidates. Among those, IL-33 is the most promising, because it is mainly produced by airway epithelial cells, which would be consistent with the critical role that we identified for AT2 cells in GC therapy (8, 9). IL-33 acts on Th2 cells, DCs, mast cells, and eosinophils (47), plays a key role in allergic airway diseases (48), and was found to be reduced in bronchial biopsies of asthmatic children treated with GCs (49). IL-5 and IL-13 are also good candidates, because analysis of asthma patients had revealed that both genes were expressed in the airway epithelium and that their levels correlated with GC responsiveness (50, 51). IL-5 is needed for eosinophil differentiation and survival, and its blockade appears to ameliorate asthma symptoms (7). IL-13 contributes to eosinophil recruitment and airway remodeling, and its targeting with a mAb in a clinical trial improved the disease in a subgroup of patients (52). However, the relative contribution of IL-5 and IL-13 produced by airway epithelial cells compared with Th2 cells and ILC2s is unknown. The same applies to MCP-1, which is produced by immune cells and airway epithelial cells alike, and is inhibited by GCs (53, 54). Considering that these mediators of allergic asthma are resistant to the repression of Dex in the lungs of GR^dim and GR^spc mice, it is conceivable that airway epithelial cells, in particular AT2 cells, are a more prominent source of cytokines and chemokines than suspected so far. Another important function of the airway epithelium is the formation of a barrier due to the presence of tight junctions (55). Occludin and claudin family members are components of tight junctions, and their levels are altered in allergic asthma (56). Intriguingly, GCs induced occludin and claudin 8 in cell culture systems (57, 58), whereas Dex repressed claudin 5 in a mouse model of allergic asthma (59). Thus, GR-dependent control of these genes appears to differ under basal and inflammatory conditions. Whether downregulation of tight junction components, as observed to a minor degree in our study, is a direct effect of Dex treatment or, rather, is indirectly caused by altered cytokine production or compensatory mechanisms remains to be determined.

To our knowledge, allergic asthma is the first example of an inflammatory disease in which GC targeting of nonhematopoietic cells is essential. In all hitherto studied mouse models of autoimmunity (multiple sclerosis, rheumatoid arthritis) (20, 21, 37), atopy (contact dermatitis) (19), transplantation (graft-versus-host disease) (60), and infection-related diseases (sepsis, acute lung injury) (22, 44, 61), GR expression in T cells or myeloid cells was required for responsiveness to endogenous or exogenous GCs. Therefore, it was surprising that nonimmune cells were most critical for the treatment of AAI with Dex. Taken together, we identified GR-dependent gene regulation in AT2 cells as a crucial therapeutic mechanism of GC action in allergic asthma (Fig. 11). Our findings suggest that selective targeting of GCs to this subtype of airway epithelial cells, such as by using liposomes or nanoparticles (62, 63), could open new avenues for improved treatment of asthma with fewer side effects and to overcome GC resistance. Although our findings are limited to the OVA-induced allergic asthma model and need to be validated using a clinically more relevant allergen, as well as human studies, we are confident that our concept will foster new research efforts aimed at improving therapeutic regimens for the treatment of asthma.

We thank Amina Bassibas and Jennifer Appelhans for expert technical assistance, Cathy Ludwig for language correction, and Dr. Brigid Hogan for providing Sftpc-CreER^T2 mice.

The authors have no financial conflicts of interest.