A T cell–myeloid IL-10 axis regulates pathogenic IFN-γ–dependent immunity in a mouse model of type 2–low asthma

Background Although originally defined as a type 2 (T2) immune-mediated condition, non-T2 cytokines, such as IFN-γ and IL-17A, have been implicated in asthma pathogenesis, particularly in patients with severe disease. IL-10 regulates TH cell phenotypes and can dampen T2 immunity to allergens, but its functions in controlling non-T2 cytokine responses in asthmatic patients are unclear. Objective We sought to determine how IL-10 regulates the balance of TH cell responses to inhaled allergen. Methods Allergic airway disease was induced in wild-type, IL-10 reporter, and conditional IL-10 or IL-10 receptor α (IL-10Rα) knockout mice by means of repeated intranasal administration of house dust mite (HDM). IL-10 and IFN-γ signaling were disrupted by using blocking antibodies. Results Repeated HDM inhalation induced a mixed IL-13/IL-17A response and accumulation of IL-10–producing forkhead box P3–negative effector CD4+ T cells in the lungs. Ablation of T cell–derived IL-10 increased the IFN-γ and IL-17A response to HDM, reducing IL-13 levels and airway eosinophilia without affecting IgE levels or airway hyperresponsiveness. The increased IFN-γ response could be recapitulated by IL-10Rα deletion in CD11c+ myeloid cells or local IL-10Rα blockade. Disruption of the T cell–myeloid IL-10 axis resulted in increased pulmonary monocyte–derived dendritic cell numbers and increased IFN-γ–dependent expression of CXCR3 ligands by airway macrophages, which is suggestive of a feedforward loop of TH1 cell recruitment. Augmented IFN-γ responses in the HDM allergic airway disease model were accompanied by increased disruption of airway epithelium, which was reversed by therapeutic blockade of IFN-γ. Conclusions IL-10 from effector T cells signals to CD11c+ myeloid cells to suppress an atypical and pathogenic IFN-γ response to inhaled HDM.

Background: Although originally defined as a type 2 (T2) immune-mediated condition, non-T2 cytokines, such as IFN-g and IL-17A, have been implicated in asthma pathogenesis, particularly in patients with severe disease. IL-10 regulates T H cell phenotypes and can dampen T2 immunity to allergens, but its functions in controlling non-T2 cytokine responses in asthmatic patients are unclear.
Objective: We sought to determine how IL-10 regulates the balance of T H cell responses to inhaled allergen.
Asthma comprises a broad spectrum of clinical phenotypes with diverse and incompletely understood causes. 1,2 Dissection of mechanisms in patients with severe asthma is of particular importance because these patients are inadequately treated with conventional therapy regimens and contribute disproportionally to health care costs. 1 Allergic asthma is typically a type 2 (T2) immune-mediated condition characterized by IL-4, IL-5, and IL-13 production from CD4 1 T H 2 cells and innate lymphoid cells, eosinophilic airway inflammation, and increased circulating IgE levels. 3 However, non-T2 cytokines have also been implicated in asthma pathogenesis, including IL-17A, 4,5 IL-6, 6 and IFN-g. 7-9 IFN-g and IFN-g-inducible type 1 (T1) immune gene expression is increased in the airways of a proportion of patients with severe asthma, 7-10 and IFN-g drives corticosteroid-refractory airway hyperresponsiveness (AHR) in a murine severe asthma model. 9 Importantly, patients with T1-high asthma coexpress T2 gene signatures to varying extents, 8 suggesting that interplay between distinct T H cell phenotypes might contribute to asthma pathology.
The anti-inflammatory cytokine IL-10 is a key regulator of the balance between T H subsets during immune responses. 11 IL-10 was originally described as a T H 2-derived cytokine with the potential to suppress T H 1 function by acting on antigen-presenting cells (APCs). [12][13][14][15] However, IL-10 can also suppress T2 and IL-17-dependent type 17 (T17) immune responses and be produced by T H 1, T H 2, and T H 17 effector T (Teff) cells, as well as forkhead box P3 (FoxP3) 2 and FoxP3 1 regulatory T (Treg) cells, B cells, and several myeloid cell subsets. 11,16,17 Therefore IL-10 can act through diverse mechanisms to limit pulmonary immunopathology in different contexts. 18 T H cell-derived IL-10 is a key mediator of induced tolerance to allergens 19,20 and resolution of allergic inflammation. 21 Conversely, impaired IL-10 production by T H cells has been reported in patients with severe asthma, 22,23 and genetic variants affecting IL-10 expression have been associated with asthma incidence and severity. 24,25 However, despite these associations, the mechanisms by which IL-10 regulates diverse immunologic phenotypes of asthma are incompletely understood. 18 Mechanistic studies in asthma frequently use murine models of allergic airway disease (AAD) typically involving peripheral sensitization to a model allergen followed by airway allergen challenge. 26 These protocols bypass the APC network of the lungs, which is essential for sensitization to inhaled allergens, 27 and generate a highly polarized T2 response that does not reflect the heterogeneous immunologic phenotypes observed in human asthma. 1,8,26 In contrast, sensitization through repeated allergen inhalation drives more complex immune responses, including T1 and T17 cytokines, in a manner dependent on activation of pulmonary innate immunity by endogenous or exogenous adjuvants, such as LPS [28][29][30] or the bacterial messenger cyclic-di-GMP. 9 To date, mechanistic studies of IL-10 function in patients with AAD have largely focused on T2-high peripheral sensitization/ airway challenge models and shown IL-10 to limit T H 2 cell survival, eosinophilic inflammation, and AHR. [31][32][33] However, studies of IL-10 regulation of non-T2 immunity in patients with AAD are lacking.
In this work we showed repeated inhalation of the common aeroallergen house dust mite (HDM) to induce a mixed T2/T17 immune response and elicit IL-10 production from Teff cells. T cell-derived IL-10 was shown to suppress a pathogenic IFN-g response to HDM by signaling to myeloid cells, limiting damage to the airway epithelium. These findings provide new insight into the context-specific nature of IL-10 function and regulation of non-T2 immunity in asthmatic patients.

METHODS
Detailed methods, including additional experimental procedures, are presented in the Methods section in this article's Online Repository at www.jacionline.org.

Experimental animals
All animal work was performed in accordance with the Animals (Scientific Procedures) Act 1986 at Imperial College London. Female C57BL/6J mice were purchased from Charles River Laboratories (Margate, United Kingdom). 10BiT IL-10 reporter mice, 34 Il10 fl/fl mice 35 crossed to CD4-Cre, 36 and Il10ra fl/fl mice 37 crossed to CD11c-Cre mice were obtained from the Francis Crick Institute (London, United Kingdom) and maintained at Charles River Laboratories. Littermate control mice and comparable proportions of male and female mice were used for all transgenic experiments. No sex-dependent differences were observed in experiments.

Induction of AAD and cytokine blockade
Mice were anesthetized with isoflurane and administered 25 mg (total protein) of HDM (Citeq Biologics, Groningen, The Netherlands) in 25 mL of PBS (Thermo Fisher Scientific, Waltham, Mass) intranasally 5 times per week for 3 weeks. Control mice received 25 mL of PBS. HDM endotoxin content was 1.1 3 10 3 EU/mg. In IL-10 receptor a (IL-10Ra) blockade experiments 50 mg/50 mL of 1B1.3a (aIL-10R) or RTK2071 rat IgG 1 isotype UltraLEAF antibodies (BioLegend, San Diego, Calif) were administered intranasally 24 hours before the first allergen challenge and twice weekly thereafter. In IFN-g blocking experiments 250 mg/250 mL of XMG1.2 (aIFN-g) or HPRN rat IgG 1 isotype InVivoMab antibodies (Bio X Cell, West Lebanon, NH) were injected intraperitoneally 4 times on alternating days over the final 8 days of the HDM protocol. Allergen treatments were initiated at 6 to 9 weeks of age, and analysis was performed 24 hours after the final challenge.

Measurement of airway function
Lung resistance was measured in response to increasing concentrations of methacholine (Sigma-Aldrich, St Louis, Mo), as described previously, 38 by using the flexiVent apparatus and software (SCIREQ, Montreal, Quebec, Canada). Resistance was calculated by using flexiVent software with the following equation: Pressure 5 ðResistance 3 FlowÞ 1 ðElastance 3 VolumeÞ1Fitting constant:

Epithelial disruption scoring
A previously published scoring strategy 39 was modified for application to AAD. Hematoxylin and eosin-stained sections were scored by a blinded investigator and independently verified by a second blinded investigator. All complete airways (minimum of 5) on a section were scored on a 0-to 8-point scale, and mean scores were reported. Scoring was as follows: 1, minor ruffling of up to 50% of the airway epithelium; 2, ruffling of 50% or more of the airway epithelium; 11/12 for severe morphological change (hypertrophy or vacuolation of cells) in up to 50% or 50% or more of the airway; and an additional 11 to airways with scores of 4 with severe loss of epithelial uniformity. Additional scores of 11, 12, or 13 were given for shedding of 1% to 25%, 25% to 50%, or greater than 50%, respectively, of epithelial cells.

Statistical analysis
Analyses were performed with Prism software (version 7; GraphPad, La Jolla, Calif). Mann-Whitney U tests or Kruskal-Wallis tests with Dunn post hoc tests were used for single and multiple comparisons, respectively.

CD4 1 Teff cells are a major IL-10-producing population after repeated allergen inhalation
To facilitate the study of IL-10 regulation of non-T2 immunity in asthmatic patients, we first established a complex T H phenotype mouse AAD model using repeated administration of intranasal HDM for 3 weeks (Fig 1, A). This protocol elicited substantial accumulation of pulmonary IL-13 1 and IL-17A 1 T H cells, with more IL-17A 1 than IL-13 1 cells and a modest increase in numbers of T H cells producing IFN-g (see Fig E1, A, in this article's Online Repository at www.jacionline.org). This protocol induced AHR (data not shown) and eosinophilic airway inflammation with minor neutrophilia (see Fig E1, B). Cells capable of IL-10 production (IL-10-competent cells) were identified in this model by using surface CD90.1 staining in 10BiT IL-10 reporter mice, 34 first modifying a published myeloid cell flow cytometry strategy 40 to distinguish airway macrophages (AMs), granulocytes, dendritic cell (DC) subsets, and a mixed population of interstitial macrophages (IMs) and inflammatory monocytes and macrophages (IMMs; see Fig E1, C). Consistent with published findings, 41,42 a population of IL-10-competent IMs/IMMs, likely resident IMs, was present in lungs of control (PBS-treated) mice (Fig 1, B-E), which expanded during HDM-driven AAD (Fig 1, B, and see Fig E1, D). In contrast, IL-10 reporter signal was largely absent from AMs at steady state and minimally increased during AAD, either in postlavage lung tissue (Fig 1, B, and see Fig E1, D) or in bronchoalveolar lavage (BAL) fluid (data not shown). Ten percent to 50% of HDM-elicited CD64 1 Ly6C variable monocyte-derived dendritic cells (moDCs) were IL-10 competent, a substantial population in terms of absolute numbers, whereas IL-10-competent type 2 conventional dendritic cells (cDC2s) were less abundant in lung tissue (Fig 1, B, and see Fig E1, D and E). These results indicate that resident IMs and recruited monocyte-derived cells are the major myeloid cell sources of IL-10 after repeated HDM inhalation.
Among lymphoid cells, minimal IL-10 signal was detected in CD4 2 T cells (CD45 1 CD3 1 ) or B cells (CD45 1 CD3 2 CD19 1 ) in lungs (see Fig E1, E) or draining mediastinal lymph nodes (mLNs) of 10BiT mice (data not shown). However, CD4 1 T cells emerged as the major pulmonary IL-10-producing lymphocyte population during established allergic inflammation (Fig 1, C, and see Fig E1, D). Nearly all IL-10-competent CD4 1 T cells had a CD44 hi CD62L 2 CD69 1 effector memory T-cell phenotype, with a minority expressing the tissue residency marker CD103 or the hallmark Treg transcription factor FoxP3 (Fig 1, D). Ex vivo phorbol 12-myristate 13-acetate (PMA) and ionomycin stimulation and intracellular cytokine staining confirmed around 5% to 15% of lung CD4 1 T cells to be IL-10 producers (see Fig E1, F) and also showed more than half of these cells to coexpress either of the predominant HDM-elicited effector cytokines, IL-13 or IL-17A (Fig 1, E). Thus HDM inhalation largely drives IL-10 production from Teff cells rather than a discrete Treg cell population.
the T cell-derived IL-10 elicited by means of HDM inhalation influenced the balance of T H subsets in this model using CD4 conditional IL-10 knockout (Il10 DCD4 ) mice 35 and ex vivo PMA and ionomycin stimulation and intracellular cytokine staining of T H cells. As expected, HDM-elicited IL-10 1 T H cells were completely ablated in Il10 DCD4 mice (Fig 2, A). However, in contrast to previous reports of IL-10 signaling disruption in T2-high AAD models, 31,32 pulmonary IL-13 1 T H cell numbers were not increased in Il10 DCD4 mice relative to those in control mice after repeated HDM inhalation. Instead, IL-17A 1 and IFN-g 1 T H cells were more abundant in lung tissue, including a double-positive population that was much less abundant in IL-10-replete mice (Fig 2, B, and see Fig E2, A and B, in this article's Online Repository at www.jacionline.org). Accordingly, levels of IFN-g and IL-17A protein and mRNA were increased in lungs of HDM-treated Il10 DCD4 mice and in supernatants of lung cell suspensions restimulated with HDM (Fig 2, C-E), suggesting that increased cytokine levels in vivo were attributable to allergen-specific T cells. In contrast, IL-13 protein concentrations were reduced in lungs of HDM-treated Il10 DCD4 mice (see Fig E2, C), as was expression of the T2 cytokine genes Il4, Il5, and Il13 (see Fig E2, D and E). Alignment of relative T H cytokine expression (2 D-Ct ) values in lung tissue supported a skewing effect in knockout mice, with a reduction in the relatively high T2 cytokine expression observed in HDM-treated Il10 DCD4 controls accompanied by increased Il17a and Ifng levels (see Fig E2, E).
T H responses were also examined in mLNs to determine whether changes were restricted to lung tissue. Production of both IFN-g and IL-17A production was increased in Il10 DCD4 mice after HDM restimulation of mLN cells, but only IL-17A 1 and not IFN-g 1 T H cells were overrepresented in mLNs of knockout mice when analyzed directly ex vivo (see Fig E2, F and G). Thus T cell-derived IL-10 is required to restrain IFN-g and IL-17A responses to inhaled HDM, with enhancement of the IFN-g response being most pronounced at the site of allergen exposure in the lungs.
Absence of T cell-derived IL-10 results in an eosinophil-low AAD phenotype with more severe epithelial pathology We next determined whether the skewed T H response to HDM in Il10 DCD4 mice was associated with changes to the hallmark pathophysiologic features of allergic asthma. HDM-treated Il10 DCD4 mice had comparable lung resistance in response to inhaled methacholine and serum IgE levels to littermate control mice (Fig 3, A and B), indicating that the heightened IFN-g/IL-17A responses in these mice did not impair allergic sensitization or prevent AHR. Serum IgG 1 concentrations were slightly increased in Il10 DCD4 mice, whereas little to no IgG 2a induction was observed in either group (see Fig E3, A, in this article's Online Repository at www.jacionline.org). J ALLERGY CLIN IMMUNOL VOLUME 145, NUMBER 2 HDM drove cellular inflammation in the airways and lung tissue of both genotypes, with significantly more BAL fluid cells in Il10 DCD4 mice compared with Il10 fl/fl control mice (see Fig E3, B and C). Despite this overall increase, BAL fluid eosinophilia was significantly reduced in Il10 DCD4 mice accompanied by increased neutrophil numbers, together skewing the eosinophil/neutrophil ratio toward an eosinophil-low phenotype (Fig 3, C and D). Numbers of CD4 1 T cells expressing IFN-g and IL-17A were also increased in BAL fluid of Il10 DCD4 mice and were largely absent in control mice (Fig 3, E). IL-13 1 T H cells were more abundant in BAL fluid of knockout mice, although at lower numbers than IFN-g 1 or IL-17A 1 cells (see Fig E3, D). Airway mucus was increased in both groups after HDM exposure but was less substantial in Il10 DCD4 mice (Fig 3, F), which was congruent with reduced expression of the mucin genes Muc5ac and Muc5b (Fig E3, E). In addition, expression of genes indicative of early airway repair/remodeling responses, fibronectin (Fn1) and type III collagen (Col3a1), but not type I collagen (Col1a1), was also dampened in HDM-treated Il10 DCD4 mice (see Fig E3, E).
These changes prompted us to further examine the airway epithelium for imbalances in damage and repair by using composite scoring of disruption of the epithelial layer. More severe epithelial disruption was observed in HDM-treated Il10 DCD4 mice than control mice (Fig 3, G and H), along with increased airway levels of albumin (Fig 3, I), which is suggestive of leakage of serum proteins across the epithelial barrier, and uric acid, which is associated with epithelial damage and stress (Fig 3, J). 43,44 Although neutrophils were more abundant in the airways of Il10 DCD4 mice (Fig 3, C) and can contribute to airway epithelial damage in certain contexts, 45 no evidence of infiltration of damaged epithelium was observed in lung sections (data not shown). Collectively, our data show that cell-derived IL-10 restricts atypical immunity to inhaled HDM and limits airway epithelial pathology.
Local pulmonary IL-10 signaling is required to suppress the IFN-g response to HDM We next investigated whether the effects of T cell-specific IL-10 knockout could be recapitulated by panblockade of IL-10 signaling in cells residing in or migrating from lung tissue. Low-dose blocking antibody to the unique IL-10Ra subunit of the IL-10 receptor (aIL-10R) was administered intranasally throughout the HDM protocol to preferentially block pulmonary IL-10 signaling (Fig 4, A). Local IL-10R blockade caused remarkably similar effects to T cell-specific IL-10 knockout on pulmonary T H cell numbers, increasing IL-17A 1 and IFN-g 1 T H cell numbers, including double-positive cells (Fig 4, B, and see Fig E4, A, in this article's Online Repository at www.jacionline.org). IL-10Ra blockade also increased IL-17A 1 , but not IFN-g 1 , T H cell abundance in mLNs (see Fig E4, B), as observed in Il10 DCD4 mice (see Fig E2, G). Accordingly, lung IFN-g mRNA and protein levels were This altered inflammatory phenotype was accompanied by increased BAL fluid albumin and uric acid concentrations (Fig 4, E and F), which is suggestive of more severe airway damage. This parity between the results of pan-IL-10Ra blockade in the lungs and T cell-specific IL-10 knockout suggests that T cells are the functionally dominant IL-10 source acting locally in lung tissue to limit atypical immunologic and pathologic responses to inhaled HDM.

Pulmonary myeloid cell dysregulation in the absence of T cell-derived IL-10
Because IL-10 can suppress T H 1 cell responses by acting on APCs 14,15 and IFN-g can show positive feedback to APCs to perpetuate T H 1 immunity, we investigated whether cross-talk between lung macrophages, DCs, and T H cells was dysregulated during the imbalance of IFN-g and IL-10 observed in HDM-exposed Il10 DCD4 mice. We first validated integrin CD11c as a specific marker of AMs, the major pulmonary macrophage subset, and DCs 40 in the lung. AMs were confirmed to be the predominant CD11c 1 population in BAL fluid and lung tissue after HDM treatment, followed by DCs (see Fig E5, A, in this article's Online Repository at www.jacionline.org), which were predominantly CD64 1 CD11b 1 CD103 2 moDCs in both the Il10 fl/fl and Il10 DCD4 groups. CD4 1 T cells could be observed in close (< _10 mm) proximity to CD11c 1 AMs/DCs in live ex vivo lung tissue of HDM-treated mice, and these interactions were more frequent in Il10 DCD4 mice, which is consistent with the increased T-cell numbers seen in these animals (Fig 5, A, and see Fig E5, B), suggesting that local myeloid-T-cell cross-talk might be altered in the absence of IL-10. Consistent with macrophage dysregulation, expression of the activation marker MHC class II was increased on AMs and pulmonary CD11c 2 IMs/IMMs in HDM-treated Il10 DCD4 mice (Fig 5, B, and see Fig E5, C), despite numbers of these cells being unchanged from control mice (see Fig E5, D and E). MHC class II expression was not increased on cDC2s or moDCs, the critical DC subsets for HDM-driven immunity, 27 from Il10 DCD4 mice and was in fact decreased on cDC2s (see Fig E5, F). However, although pulmonary cDC2 numbers were comparable, moDCs were more abundant in lung tissue and BAL fluid of HDM-treated Il10 DCD4 mice (Fig 5, C, and see Fig E5, G-I), suggesting that moDCs might locally promote HDM-driven inflammation in these mice, as previously proposed. 27 Neither DC population was more abundant in mLNs of Il10 DCD4 mice, with a slight decrease in the proportion of cDC2s (see Fig E5, J). These data suggest that pulmonary macrophages and moDCs are dysregulated in the absence of T cell-derived IL-10 and might contribute to the atypical immune response to HDM in these mice.

Dysregulated AMs overproduce T H 1 cell chemokines
We next sought to better understand pulmonary macrophage dysregulation in Il10 DCD4 mice, focusing on AMs because of their superior numbers and decisive role in regulating immunity to HDM. 46 AMs were sorted from BAL fluid of HDM-treated Il10 DCD4 mice and control mice (see Fig E5, K), and gene expression was examined. In line with the altered cytokine environment, AMs from Il10 DCD4 mice showed decreased expression of the IL-4/IL-13-induced gene Chil3/Ym1 and concomitant increases in the IFN-g-inducible nitric oxide synthase Nos2/iNOS and Irf5 (Fig 5, D), a transcriptional regulator 47 of classical macrophage activation that dictates AM phenotype during AAD. 46 We have previously found AMs to be major producers of the IL-4/IL-13-inducible chemokines CCL8 and CCL24 in models of AAD (Branchett et al, unpublished results), 48 which contribute to T2 immunity. 49,50 Gene expression analysis in sorted AMs showed expression of these T2 chemokines to be decreased, whereas levels of the IFN-g-inducible CXCR3 ligands Cxcl9 and Cxcl10, which are associated with amplification of T1 immune responses, 51 were increased with HDM treatment in AMs from Il10 DCD4 mice (Fig 5, E, and see Fig E5, L). These CXCR3 ligands were also more highly expressed in lungs of Il10 DCD4 mice (Fig 5, F). Collectively, these data suggest that dysregulated AMs contribute to feedforward amplification of T1 immunity in the absence of T cell-derived IL-10 through upregulation of IFN-g-inducible CXCR3 ligands.  T cell-derived IL-10, we next investigated whether IL-10 signals directly to these cells to suppress this phenotype. CD11c-Cre mice, which allow conditional gene deletion in AMs, lymphoid tissue DCs, and the majority of lung DCs, 52,53 were crossed to Il10ra fl/fl mice 37 to generate Il10ra DCD11c conditional knockouts and exposed to HDM for 3 weeks. Il10ra DCD11c mice had comparable increases in HDM-elicited IFN-g 1 and IL-17A 1 pulmonary T H cells with those observed in Il10 DCD4 and local IL-10R blockade experiments (Fig 6, A, and see Fig E6, A, in this article's Online Repository at www.jacionline.org), along with significantly more IFN-g (both protein and mRNA) in lung tissue (Fig 6, B, and see Fig E6, B) and supernatants of HDM-stimulated lung cells (see Fig E6, C). In contrast, IL-17A levels were unchanged in these assays (Fig 6, B, and see Fig E6, B and C), whereas IL-13 protein concentrations were lower in lungs of Il10ra DCD11c mice (see Fig E6, D). These results indicate that, similarly to local IL-10R blockade (Fig 4), conditional IL-10R knockout more markedly enhanced the IFN-g response to HDM than the IL-17A response, which was already substantial in mice with intact IL-10 signaling in our HDM model. Il10ra DCD11c mice also had a reduced airway eosinophil/ neutrophil ratio (Fig 6, C, Il Il Il Il Il Il Il Il Il Il Il Il I Il I I I I Il Il I I I I Il I I I I I I I I I I I I I I I I I Il I I I I I I I Il I I I I I Il Il I I I I I I I I I I I I Il Il Il Il Il I Il I I Il I I I I I 10  of IFN-g 1 and IL-17A 1 cells in this compartment were not significantly altered (see Fig E6, F). HDM-treated Il10ra DCD11c mice also displayed pulmonary APC dysregulation, which is apparent from increased MHC class II expression on AMs and IMs/IMMs and greater moDC numbers in lung tissue (Fig 6, D and E), along with increased pulmonary Cxcl9 and Cxcl10 levels (Fig 6, F), which is suggestive of increased CXCR3 ligand expression during the imbalance of IFN-g and IL-10 signaling to myeloid cells. Importantly, the IFN-g-high phenotype in Il10ra DCD11c mice was accompanied by increased markers and histologic evidence of airway damage (Fig 6, G-I, and see Fig E6, G). Overall, the similarity of these results to those in Il10 DCD4 mice suggests that T cell-derived IL-10 signals to CD11c 1 myeloid cells, regulating T H 1 cell-recruiting chemokines, HDM-elicited IFN-g responses, and airway epithelial damage.
IFN-g drives pathogenic airway damage during AAD in the absence of T cell-derived IL-10 Finally, anti-IFN-g (aIFN-g) was administered systemically to Il10 DCD4 mice and control mice throughout the final week of HDM exposure to assess whether IFN-g was responsible for the increased HDM-driven airway damage observed after disruption of the T H -CD11c 1 myeloid cell IL-10 axis (Fig 7, A). Therapeutic aIFN-g administration did not affect pulmonary cytokineproducing T H cell numbers (see Fig E7, A, in this article's Online Repository at www.jacionline.org) but partially corrected the altered expression of IL-4/IL-13 and IFN-g signature genes in lungs of Il10 DCD4 mice, particularly the T H 1 chemokines Cxcl9 and Cxcl10 (Fig 7, B), indicating successful suppression of IFN-g signaling. Therapeutic aIFN-g reduced the heightened MHC class II expression on AMs and lung moDC numbers in Il10 DCD4 mice (Fig 7, C and D), which is consistent with these myeloid cell dysfunction phenotypes being IFN-g dependent. Although therapeutic aIFN-g did not restore eosinophilia in Il10 DCD4 mice, it reduced numbers of neutrophils (see Fig E7, B and C) and IFN-g 1 CD4 1 T cells (Fig 7, E) in BAL fluid to levels comparable with those in Il10 fl/fl control mice, suggesting that IFN-g promotes airway inflammation in these mice. Importantly, airway albumin and uric acid levels were diminished by aIFN-g treatment, which was congruent with an improved epithelial pathology score (Fig 7, F-H). These data confirm that high levels of IFN-g arising during HDM-driven AAD in the absence of intact pulmonary IL-10 signaling contribute to airway pathology.  (Fig 6, C), percentages of BAL fluid AMs and lung IMs/IMMs expressing MHC class II (Fig 6, D), and numbers of moDCs in lung tissue (Fig 6, E). F, Fold changes in Cxcl9 and Cxcl10 mRNA expression in homogenized lung tissue. G and H, Concentrations of albumin and uric acid in BAL fluid. I, Composite airway epithelial disruption scores of hematoxylin and eosin-stained lung sections. Data are pooled from 2 experiments and show medians and individual replicates (n 5 6-12 per group). *P < .05, **P < .01, ***P < .001, and ****P < .0001.

DISCUSSION
Pulmonary immunity must be tightly regulated by anti-inflammatory cytokines to minimize tissue damage that might compromise respiratory function. 18 Asthma is thought to occur because of an imbalance between immune activation and regulation in the respiratory tract, prompting efforts to boost regulatory mechanisms by immunotherapy to treat allergic diseases. 54 The cytokine IL-10 can regulate diverse T H cell responses to limit immunopathology, 11 including T2 immunity in murine models of asthma. 31,32 However, there is increasing evidence of heterogeneous asthma phenotypes in human subjects in whom dysfunctional T2 immunity is not the sole immunologic feature, and it is not known how such T2-low/non-T2 immunity is regulated in the setting of asthma.
Using complex mouse models of mucosal HDM sensitization and conditional gene deletion, we have examined the interactions between IL-10 and key proinflammatory cytokines in the allergic lung. Although IL-10 can be produced by several different cell types in the respiratory tract, 18 we have shown IL-10-producing CD4 T cells, principally FoxP3 2 Teff cells, to be elicited by repeated HDM exposure and that T cell-derived IL-10 signals to CD11c 1 myeloid cells to regulate pathogenic IFN-g responses to inhaled allergen. Parity between our results using specific knockout of T cell-derived IL-10 and global blockade of pulmonary IL-10 signaling suggest that CD4 1 Teff cells are the functionally dominant IL-10-producing cells for limiting IFN-g responses in the lungs in patients with HDM allergy. However, it remains possible that non-T2 responses to allergen can be regulated by IL-10-producing myeloid cells. Indeed, others have reported that adoptive transfer of IL-10-producing IMs can reduce neutrophilic T17 immune responses to HDM in IL-10 knockout mice. 55 Our findings align with several mechanistic studies showing CD4 1 T cells to be the functionally dominant IL-10 source in dampening protective or pathogenic immunity to respiratory tract infection [56][57][58] and likely reflect negative feedback of Teff cell responses by intrinsic IL-10, restricting APC function to limit atypical and pathogenic IFN-g responses to inhaled HDM.
Although knockout of T cell-derived IL-10 increased IFN-g levels in the lung after HDM inhalation, it did not alter lung tissue IL-13 1 T H cell numbers and actually decreased pulmonary T2 cytokine levels and airway eosinophilia. This phenotype differs markedly from results in T2-high peripheral sensitization/airway challenge AAD models, in which disruption of endogenous IL-10 function augmented T2 immunity and AHR, 32,33 whereas lack of direct IL-10-signaling to T cells specifically enhanced T H 2 cell survival and downstream allergic inflammation. 31 This disparity with previous reports likely reflects engagement of distinct immunologic pathways by different allergens or sensitization routes. Previous studies have highlighted the importance of inhaled endotoxin for driving such differences because endotoxin can facilitate generation of mixed T2/T17 airway responses at lower concentrations 28 and T1 responses at greater concentrations. 30 Therefore we propose that T cell-derived IL-10 acts to limit T1 responses to the endogenous endotoxin present in HDM preparations, such that an IFN-g response is unleashed in which IL-10 signaling is disrupted during repeated airway HDM exposure. Differences between the effects of IL-10 signaling disruption in our HDM inhalation experiments and the increased T H 2 cell survival observed by Coomes et al 31  C-E, Flow cytometric data showing percentages of BAL fluid AMs expressing MHC class II (Fig 7, C), numbers of moDCs in lung tissue (Fig 7, D), and IFN-g 1 CD4 1 T cells in BAL fluid (Fig 7, E). peripheral sensitization HDM model might also result from the latter study specifically disrupting IL-10 signaling to T cells, whereas we instead targeted a T-cell-myeloid cell signaling axis. Our data underscore the concept that the effects of a particular cytokine in vivo depend on its cellular source and cross-talk with other context-specific signals, which in turn depend on the nature of the inflammatory stimulus. Therefore it is important to evaluate cytokine function in diverse models of AAD, particularly those such as ours in which sensitization occurs through the physiologically relevant airway route in the absence of systemic adjuvant.

in a
The effects of T cell-restricted IL-10 deletion on IFN-g production could be recapitulated by panblockade of local pulmonary IL-10R signaling through the airways or deletion of IL-10Ra from CD11c 1 AMs and DCs, suggesting that Teff cells signal through IL-10 to CD11c 1 APCs resident in or migrating from the lung to suppress the atypical IFN-g response to HDM. This is reminiscent of early studies demonstrating IL-10 suppression of T H 1 polarization through direct action on APCs in vitro and in nonallergic responses. 14,15,59 In contrast to the increases in IFN-g production, which were observed across all models of IL-10 perturbation, HDM-driven IL-17A production and airway neutrophilia were only robustly increased when T cell-derived IL-10 was deleted. Because IL-17A can promote neutrophilic inflammation in the lung, 60,61 it seems likely that T cell-derived IL-10 suppresses an HDM-induced IL-17A-neutrophil axis partially independently of IL-10Ra on CD11c 1 myeloid cells, possibly by direct signaling to IL-10R 1 T H 17 cells. 62 However, enhanced neutrophilia in Il10 DCD4 mice was partially reversed by aIFN-g, suggesting that neutrophilia in this context depends on multiple signals, including IFN-g.
Reciprocal cross-regulation of T H 2 and T H 17 responses in asthmatic patients has been suggested based on data in human cells and mouse models, 5 and it is possible that the increased IL-17A and decreased T2 cytokine expression observed in HDM-treated Il10 DCD4 mice reflects such a relationship downstream of IL-10 regulation. However, IL-17A and IL-13 responses can be increased in parallel in asthmatic lungs, 9 and IL-17A can enhance IL-13 signaling synergistically through signal transducer and activator of transcription 6. 63 In addition, the role of IL-17A in asthma pathogenesis is incompletely understood, with variable reports of association with disease parameters. 4,64,65 Indeed, although numbers of both IFN-g 1 and IL-17A 1 T H cells were increased in lungs of patients with severe asthma and a mouse model of severe asthma, only knockout of IFN-g, but not the IL-17A receptor, was sufficient to improve lung function in these mice. 9 Although T cell-restricted IL-10 deletion did not increase AHR in our model, we observed enhanced epithelial pathology and airway albumin and uric acid levels in which IL-10 signaling from T cells to CD11c 1 myeloid cells was disrupted, which was reduced by therapeutic IFN-g blockade, supporting a pathogenic role for IFN-g during atypical immunity to inhaled HDM in this context. Increased BAL fluid albumin concentrations can indicate exudation of plasma into the airways without decreased epithelial barrier integrity. 66 However, concomitant increases in epithelial denudation and loss of uniformity, along with the established epithelial damage-associated molecular pattern uric acid, 44 collectively support reduced epithelial integrity in our models of IL-10 disruption. IFN-g can directly disrupt the integrity of human sinus epithelium in vitro by promoting apoptosis and decreasing tight junction protein expression, 67,68 suggesting that epithelial disruption downstream of IFN-g in our studies might reflect direct effects of this cytokine. It is notable that CD4 1 T cells were present in increased numbers in the airway lumens of these mice, positioning them optimally for delivering cytokine to the apical side of the epithelium.
It remains to be determined precisely which CD11c 1 cells and in which temporal and spatial contexts receive IL-10 signals to suppress IFN-g responses to HDM. This is challenging to dissect with available tools because CD11c-Cre targets both AMs and DCs, 53,69 whereas LysM-Cre, which is used to broadly target myeloid cells, has variable efficacy in AMs 69 and affects off-target populations, such as neutrophils. 53 A recent study showed CD11c-Cre-driven but not LysM-Cre-driven Il10ra knockout to prevent induction of tolerance to aeroallergen by using subcutaneous immunotherapy, leading to the conclusion that DCs were the critical IL-10R-expressing cells. 70 However, this study used peripheral sensitization with an adjuvant, in which involvement of AMs and DCs will almost certainly differ from our mucosal sensitization model. Nonetheless, because T cell-derived IL-10 can suppress T H 1 responses by preventing IL-12 production from DCs, 59 it is possible that the absence of such T cell-DC cross-talk in mLNs contributes to the increased IFN-g responses observed in our models of IL-10 disruption.
Our data strongly support a role for local APCs in perpetuating the IFN-g response to HDM in which the T-cell-myeloid IL-10 axis is disrupted because AMs and moDCs, both important local influencers of HDM responses, 27,46 were dysregulated in these mice. Moreover, IL-10 disruption increased expression of the T H 1 cell chemokines CXCL9 and CXCL10 51 in lungs, derived at least in part from dysregulated AMs, although contribution from other key pulmonary chemokine producers, such as epithelial cells 71 and moDCs, 27 is also possible. Increased lung moDC numbers and CXCR3 ligand expression were both reduced by therapeutic IFN-g blockade, suggesting that they arise downstream of an initial IFN-g/IL-10 imbalance. Therefore we propose a model in which a T cell-myeloid IL-10 axis restrains IFN-g responses to inhaled HDM, disruption of which creates an initial IFN-g/IL-10 imbalance that unleashes a feedforward loop of dysregulated AMs, moDCs, and CXCR3 ligand production to amplify the IFN-g response. Reliance on local factors is supported by the fact that IFN-g 1 T H cells were more abundant in the lungs but not mLNs of Il10 DCD4 mice and that inhaled aIL-10R was sufficient to augment the IFN-g response to HDM. A CXCR3-dependent amplificatory mechanism also aligns with the association of airway CXCL10 expression with an IFN-g-high asthma phenotype in human subjects and mice. 7 Our data highlight the importance of defining the cellular source of IL-10 in vivo and examining the regulation of different T H cell subsets after allergen inhalation. Increases in both T2 and T17 cytokine production, as induced in our mouse model, have been observed in the lungs of patients with severe asthma, 9,72 as have increased IFN-g responses. 7-10 However, none of these studies examined IL-10 levels or function alongside proinflammatory cytokines, despite impaired IL-10 production being implicated in pediatric and adult patients with severe asthma. [22][23][24] For the first time, our data present a pathogenic mechanism arising from disequilibrium between IL-10 and IFN-g in the allergic lung and suggest that insufficient IL-10 signaling could contribute to increased IFN-g responses in patients with severe asthma who might otherwise display a mixed T2/T17 phenotype. Our data might explain in part how heterogeneous phenotypes can occur in certain patients after allergen exposure. Therefore further investigation of the balance between IFN-g and IL-10 in patients with severe asthma might provide novel insight into disease mechanisms and therapeutic opportunities.

Tissue processing
Mice were euthanized by exsanguination by cardiac puncture under terminal anesthesia and blood centrifuged for 5 minutes at 12,000g in Microtainer Serum Separator Tubes (BD, Franklin Lakes, NJ) to obtain serum. For standard analysis or BAL fluid, the airways were flushed 3 times with 0.4 mL of PBS, pooled fractions were centrifuged, cells were suspended in complete RPMI (cRPMI; RPMI 1640 plus 10% FBS and 1% penicillin and streptomycin; all from Thermo Fisher Scientific), and fluid was stored at 2808C for analysis. For AM sorting, airways were flushed 7 times with 1 mL of 5 mmol/L EDTA in PBS (Thermo Fisher Scientific).
Single-cell lung and mLN suspensions were obtained by passing tissue through 70-mm cell strainers and suspending in cRPMI. Left lungs were minced and incubated in 150 mg/mL collagenase D and 25 mg/mL DNase I (Roche, Basel, Switzerland) at 378C for 1 hour before dissociation and were subsequently subjected to erythrocyte lysis in ammonium chloride buffer. Right middle lung lobes for mediator analysis were flash-frozen in liquid nitrogen before homogenizing in HBSS (Thermo Fisher Scientific) in the presence of cOmplete protease inhibitor cocktail (Roche). Right postcaval lung lobes for RNA analysis were preserved in RNAlater (Sigma-Aldrich) until analysis. Superior right lobes for histology were inflated with PBS through a major airway and fixed in 10% buffered neutral formalin (Leica Biosystems, Wetzlar, Germany) before processing and staining with either hematoxylin and eosin or periodic acid-Schiff (PAS).

HDM stimulation of cell suspensions
Lung or lymph node cells (2 3 10 5 ) were incubated at 378C in 200 mL of cRPMI containing 50 mg/mL HDM in 96-well round-bottom cluster plates. Supernatants were collected by means of centrifugation after 4 days.

Flow cytometry and cell sorting
Up to 2 3 10 6 cells were stained with Fixable Blue Live/Dead Stain (Thermo Fisher Scientific) at room temperature for 30 minutes, followed by antibodies to extracellular antigens at 48C for 20 minutes in the presence of mouse Fc Block (clone 2.4G2; BD Biosciences, San Jose, Calif). Cells were washed twice and fixed in eBioscience IC Fixation Buffer (Thermo Fisher Scientific) for 15 minutes at room temperature, except when staining for FoxP3, when the eBioscience Intracellular Fixation & Permablization Buffer Set was used. Where necessary, fixed cells were permeabilized with eBioscience Permeabilization Buffer and stained with antibodies to intracellular antigens at 48C for 20 minutes before washing twice. When subsequently staining for cytokines, cells were stimulated with PMA and ionomycin for 4 hours at 378C in the presence of Brefeldin A.

Precision-cut lung slices and confocal microscopy
Lungs were inflated with 2% TopVision Low Melting Point Agarose (Thermo Fisher Scientific) through a tracheal cannula, and once solidified, 200-mm transverse slices were prepared from left lungs with a Compresstome VF-300 vibrating microtome (Precisionary Instruments, Greenville, NC). Slices were incubated for 1 hour at 378C in cRPMI to remove agarose before staining with Brilliant Violet 421-CD4 (clone RM4-5), Alexa Fluor 647-CD11c (clone N418), and phycoerythrin-EpCAM (clone G8.8) antibodies (BioLegend) for 1 hour at 378C in cRPMI in the presence of Mouse Fc Block. Slices were washed in cRPMI and visualized in 24-well m-plates (Ibidi, Martinsried, Germany) with an inverted laser-scanning confocal microscope (SP5; Leica Microsystems, Wetzlar, Germany) to acquire 20-to 40-mm Z-stack images of airways. Images of 1024 3 1024 mm field of view were taken under a 20 3 objective. Image analysis was performed by using the colocalization tool in Imaris version 8.1 (BitPlane; Oxford Instruments, Abingdon, United Kingdom), with a cutoff of 10 mm for colocalization.

Airway mucus scoring
All complete airways (> _5) in PAS-stained sections were scored on a 0to 4-point scale, as previously described, E1 with 1 point awarded per 25% of airway epithelial cells positive for PAS and mean scores reported per mouse.

RNA extraction and quantitative PCR
Lung lobes or cell pellets were homogenized in RLT buffer plus 1% b-mercaptoethanol, and RNAwas extracted by using RNeasy Plus Mini (lung) or Micro (cell pellets) kits (Qiagen, Hilden, Germany). Lung RNA (1 mg) and approximately 0.1 mg of cell pellet RNA was converted to cDNA by using the Applied Biosystems High-Capacity Reverse Transcription Kit (Thermo Fisher Scientific) or GoScript Reverse Transcriptase System (Promega, Madison, Wis), respectively, in both cases by using random primers. Quantitative PCR was performed with a Viia7 instrument and TaqMan Fast Advanced Master Mix and the following off-the-shelf assays from Thermo Fisher Scientific: , and Cxcl16 (Mm00469712_m1). Threshold cycle (Ct) values were determined by using Viia7 software, and relative expression was calculated as 2 2DCt relative to the mean Ct of the housekeeping genes Gapdh and Actb. Data are presented either as unadjusted relative expression or fold changes from means of an appropriate negative control group.

Mediator analysis
IFN-g and IL-17A levels were measured by using Ready-Set-Go ELISA kits (Thermo Fisher Scientific), IL-13 levels were measured by using the mouse Quantikine ELISA Kit (R&D Systems, Minneapolis, Minn), and albumin levels were measured by using the Bethyl Laboratories (Montgomery, Tex) Mouse Albumin ELISA Kit, according to the manufacturers' instructions. IgE, IgG 1 , and IgG 2a were detected by using custom sandwich ELISAs consisting of purified capture antibodies (clones R35-72, A85-3, and R11-89, respectively) and biotinylated detection antibodies (clones R35-118, A85-1, and R19-15, respectively) standardized to purified IgE, IgG 1 , and IgG 2a k isotype control antibodies (all from BD Biosciences) and developed by using streptavidin-horseradish peroxidase (R&D Systems) and TMB substrate (Thermo Fisher Scientific). Uric acid was quantified by using the Amplex Red Uric Acid/Uricase Assay Kit (Thermo Fisher Scientific), according to the manufacturer's instructions.     Fig E5, A, shows median percentages 6 interquartile ranges pooled from 2 experiments (n 5 7-12 per group). Points in Fig 5, B, show means 6 SDs of 3 equally sized fields from each of 2 to 3 mice per group. All other data show medians and individual replicates and are pooled from 2 to 3 experiments (n 5 7-12 per group). *P < .05, **P < .01, and ***P < .001. ND, Not detected; ns, not significant.  (Fig E6, C) and homogenized lung tissue (Fig E6, D). E and F, Flow cytometric data showing numbers of eosinophils and neutrophils (Fig E6, E) and IFN-g 1 and IL-17A 1 CD4 1 T cells (Fig E6, F)