B10Q mice (C57/Bl10.Q/rhd) were originally from Jan Klein (Tübingen University, Tübingen, Germany) (19), and maintained in the medical inflammation research lab led by Rikard Holmdahl (rhd). B10Q.Ncf1 ^* mice indicate B10Q mice with a mutation in the Ncf1 gene (m1j), in which the NOX2 complex function is impaired but not deleted (19, 20). B10Q.Ncf1 ^*.MN strain transgenically expresses Ncf1 under the control of the CD68 promotor in a B10Q mouse having the Ncf1 ^m1j mutation, thus expressing Ncf1 in macrophages only. The above mice were maintained in SPF animal feeding equipment of the Experimental Animal Center of Xi’an Jiaotong University Health Science Center. In this study, female mice aged 8 to 12 weeks were used. All animal experiments followed ARRIVE guidelines and were approved by the Institutional Animal Ethics Committee of Xi’an Jiaotong University.

Female B10Q, B10Q.Ncf1 ^* and B10Q.Ncf1 ^*.MN mice were used to induce AIPI model essentially following previously described protocols (10). A colloid emulsion was prepared by mixing 0.1 mg/ml OVA (Sigma) and Imject™ Alum (Pierce, Thermo Fisher Scientific) in equal volume. The mice were sensitized by intraperitoneal injection of the OVA emulsion (200 μl per mouse) at days 0 and 7. From day 14, the mice were challenged with 1% OVA aerosol for 30 min per day for 7 consecutive days. On day 21, blood, bronchoalveolar lavage fluid (BALF) and lung tissues were collected.

Lung tissues from the lower left lobe were fixed with 4% paraformaldehyde (SinoPharm, China), embedded in paraffin (Hualing, Shanghai, China), sectioned, and then stained with hematoxylin and eosin (H&E, Solarbio, China), periodic acid Schiff (PAS, Solarbio, China) and Masson (Solarbio, China) (10). Peritracheal inflammatory cell infiltration was scored by an investigator with no knowledge of the sample identities, according to the number of layers of infiltrating cells, as previously described (10). Score 0: no peritracheal infiltrating cells. Score 1: sporadic infiltrating cells around the tube, less than one layer; Score 2: peritubule infiltrating cells reached 1-2 layers; Score 3: peritubule infiltrating cells up to 2-3 layers; Score 4: peritubule infiltrating cells with 3-5 layers; Score 5: more than 5 layers with peritubule infiltrating cells. Both the portion of goblet cell staining of airway mucus and the degree of airway fibrosis were scanned using Image Pro Plus software and normalized with relative airway area.

To quantify the levels of OVA-specific IgE and OVA-specific IgG1 in serum, a 96-well plate was coated with OVA (5 μg/ml, Sigma) overnight and blocked using 1% bovine serum albumin (BSA, Amresco). Diluted sera were added for 2 hour incubation, followed by rat anti-mouse IgE-biotin (clone No. 23G3, SouthernBiotech) or rat anti-mouse IgG1-biotin (clone No. LO-MG1-2, Bio-Rad) antibodies, and developed with streptavidin-horseradish peroxidase (HRP, R&D systems, Bio-techne). The reaction was visualized by using the substrate ABTS (Roche) and read at 405 nm using a microplate reader (Thermo Electron Corporation) (10).

Selected cytokines and chemokines including chemokine (C-X-C motif) ligand (CXCL) 1 (range: 9.9-7220 pg/ml), tumor necrosis factor (TNF)-α (0.81-590 pg/ml), IL-12p70 (14.95-10900 pg/ml), IL-1β (76.74-55940 pg/ml), vascular endothelial growth factor (VEGF, 1.48-1080 pg/ml), IL-4 (20.44-14900 pg/ml), IL-5 (1.62-1180 pg/ml), IL-6 (8.41-6130 pg/ml), IL-10 (4.6-3350 pg/ml), IL-13 (16.69-12170 pg/ml), IL-17 (14.12-10290 pg/ml), interferon (IFN)-γ (4.39-3200 pg/ml), IL-33 (27.59-20110 pg/ml), IL-25 (46.82-34130 pg/ml), C-C motif chemokine ligand (CCL) 11 (4.24-3090 pg/ml) were quantified in serum samples of mice using a 15-plex Luminex panel (LXSAMSM-15, R&D Systems). The sample was diluted twice and the plate was detected by Luminex X-200.

Total RNA was extracted from lung tissue of mice using TRIzol reagent (Invitrogen) and cDNA was synthesized using the first strand cDNA synthesis Kit (Fermentas). Agilent Stratagene Mx3005P performed real-time quantitative polymerase chain reaction (RT-qPCR) with FastStart Universal SYBR Green Master (Roche) to assess cytokine expression. Mouse β-actin gene was used as a housekeeping gene. The primers used in this experiment are listed in the Supplementary Table 1 .

On day 0 and day 1, female mice with three genotypes were treated with IL-33 (0.5 μg/mouse, SinoBiological) (21) or IL-25 (1.5 μg/mouse, SinoBiological) (22) intranasally, respectively. On the 3rd day, mice were sacrificed and the whole lungs were taken and digested for flow cytometry assay.

Lung tissues were minced and digested by using collagenase type I (1 mg/ml, Sigma) and DNase (25 μg/ml, Sigma) for 1 h. Cells from lung tissue or BALF were first stained with LIVE/DEAD Fixable Violet stain kit (Thermo Fisher), and labeled with the following monoclonal antibodies for the membrane antigen staining: anti-CD45.2-BV785, anti-CD11b-APC (allophycocyanin), anti-CD11c-APC-Cy7, anti-F4/80-FITC, anti-Ly6G-Pacific blue, anti-Siglec F-PerCP (peridinin chlorophyll protein)-Cy5.5, anti-γδ T cell receptor (TCR)-PE (phycoerythrin), anti-TCRβ-Alexa Fluor 700, anti-B220-PE-Cy7, anti-CD4-PE-Cy5, mouse lineage antibody cocktail-PerCP-Cy5.5, anti-ST2-BV421 and anti-CD68-Pacific blue. For cytokine staining, cells were pre-treated with phorbol 12-myristate 13-acetate (PMA, 100 ng/ml, Sigma), ionomycin (1 μg/ml, Sigma) and brefeldin A (10 μg/ml, Selleck) for 3 h, fixed and permeated by using BD Cytofix/Cytoperm after membrane staining, and labeled with the following monoclonal antibodies for intracellular staining: anti-IL-5-PE, anti-IL-13-PE-Cy7, anti-IFN-γ-Pacific blue. Mouse lineage antibody cocktail could react with cells from the major hematopoietic lineages, such as T cells, B cells, monocytes/macrophages, NK cells, erythrocytes, and granulocytes, which includes clone 145-2C11 (recognizes CD3e), M1/70 (recognizes CD11b), RA3-6B2 (recognizes B220), TER-119 (recognizes Ly-76), and RB6-8C5 (recognizes Ly-6G and Ly-6C). The detailed antibody information is shown in Supplementary Table 2 . Samples were analyzed using an Agilent NovoCyte 3005 flow cytometer. Immune cell gating strategy and the fluorescence-minus-one (FMO) staining controls are shown in Supplementary Figures 1, 2 .

On day 0, IFN-γ antibody or isotype IgG (200 μg per mouse, Bio X Cell) was injected intravenously to B10Q.Ncf1^* mice. Then on day 0 and day 1, mice were treated with IL-33 (0.5 μg per mouse, SinoBiological) intranasally. On the third day, the whole lung was isolated, and flow cytometry was performed after digestion.

Graphpad Prism software was used for statistical analysis and graph construction. The quantitative data are expressed by mean ± SEM. The differences were statistically analyzed using one‐way ANOVA with post‐hoc comparison (Fisher’s) test among three experimental groups or the student’s t test between two experimental groups. A P value of less than 0.05 is considered significant.

To determine the role of Ncf1 in lung inflammation, B10Q.Ncf1 ^* mice were compared with B10Q littermates. We did not find any significant difference in lung tissue structure, or change in the composition of immune cell populations in naïve mice ( Figure S3A–C ). Then we investigated the AIPI model using OVA as an antigen, to induce symptoms resembling asthma ( Figure 1A ). We found that the total number of immune cells in the BALF were lower in B10Q.Ncf1 ^* mice ( Figure 1B ). Particularly, the number and the cell proportion of eosinophils were both reduced in B10Q.Ncf1 ^* mice compared with the wild-type littermates ( Figures 1B, C ). Thus, a functional Ncf1 was important to allow an allergic inflammation in the lung.

Ncf1 is highly expressed in phagocytic cells, including macrophages and neutrophils (23). We next addressed the role of Ncf1 in macrophages, due to their central regulatory role in both innate and adaptive immunity. A transgenic mouse strain, B10Q.Ncf1 ^*.MN (MN) was used, which transgenically expresses physiologic levels of functional Ncf1 with the help of a CD68 promotor (24). We confirmed CD68 expression in BALF cells by flow cytometry and found that it was highly expressed on residential alveolar macrophages in lung, but not on infiltrating interstitial macrophages, neutrophils, eosinophils or dendritic cells (DC), suggesting that alveolar macrophages of B10Q.Ncf1 ^*.MN mice had functional Ncf1 ( Figure 1D ).

We then induced AIPI in the three strains, wild-type, Ncf1 ^* and MN mice. Histological analyses were performed on lung tissues showing the extent of inflammatory infiltration around the airway, mucus secretion of airway epithelia and airway fibrosis ( Figure 2A ). H&E staining showed that the degree of peri-airway inflammatory infiltration in Ncf1 deficient mice was significantly lower than that in wild-type mice, and MN tended to compensate for this difference ( Figure 2B ). However, Ncf1 deficiency had no significant effect on the production of airway mucus and the airway fibrosis, although a trend in the PAS staining could be observed ( Figures 2C, D ). At the same time, we found that Ncf1 deficient mice showed no difference in the production of OVA-specific IgE and IgG1 compared with wild-type, although MN mice showed higher antibody levels ( Figure 2E ).

The inflammatory cell profiles of cells infiltrating in lungs were studied by flow cytometry. The results of staining of BALF cells are visually represented via a cluster heat map, which revealed a clustering of the inflammatory profiles of Ncf1 deficient mice versus the wild-type mice, and with the MN mice in between ( Figure 3A ). When Ncf1 was deficient, the total number of cells and various inflammatory cells in BALF were significantly decreased compared with wild-type ( Figure 3B ), and the proportion of eosinophils was also decreased ( Figure 3C ). Clearly, the replenishment of Ncf1 in alveolar macrophages could compensate for the above differences to a certain extent ( Figures 3B, C ). In addition, the population of immune cells in lung tissue was also analyzed, and the result of lung tissue was very similar to that of BALF, with reduced numbers of total immune cells and eosinophils in Ncf1 deficient mice ( Figures 3D, E ). Taken together, the results suggest that expression of a functional Ncf1 allows the development of lung inflammation, which is mainly depended on Ncf1 expressing alveolar macrophages.

To address the mechanism whereby Ncf1 regulated airway inflammation, we determined the cytokine profiles of lung tissue and serum. The mRNA expression of cytokines and chemokines in lung tissue was detected by RT-qPCR ( Figures 4A, B ), and their protein concentrations in serum were detected by Luminex assay ( Figures 4C, D ). The results showed that the mRNA expression of Ifng and Tnfa in lung tissue, and the protein concentrations of IFN-γ and IL-12 in serum were increased in B10Q.Ncf1 ^* mice, as compared with the wild-type mice, and their expression was decreased in MN mice ( Figures 4B, D ). In addition, the mRNA expression of Il10, Il33, and Vegf in lung was increased, whereas the protein concentration of IL-13 in serum was decreased, in Ncf1 deficient mice, while MN reversed these changes to some extent ( Figures 4B, D ). Overall, Ncf1 deficiency was associated with an enhanced Th1-type of cytokine response, an effect that could be completely neutralized by expression of Ncf1 in macrophages. It raises the possibility that activation of the NOX2 complex in macrophages enable these cells to suppress a Th1 type of inflammatory response and thereby allowing a pathogenic Th2 type of reaction, explaining why expression of Ncf1 in macrophages promotes an allergic type of inflammation in AIPI.

It was interesting to note that IL-33 and IL-25 expression was increased or showed an increased trend in Ncf1 deficient mice, suggesting that Ncf1 might regulate epithelial derived cytokines. These epithelial cytokines are believed to play important roles in mediating immune response in lung, and intranasal treatment of mice with IL-33 or IL-25, has been shown to induce pulmonary inflammation (25). To more directly test the role of these cytokines we treated the mice intranasally with IL-33 or IL-25 ( Figure 5A ), and the immune cell populations in lung tissues were analyzed by flow cytometry. In the IL-33 induced model, unlike the AIPI model, the heap map revealed a similarity of the inflammatory profiles between Ncf1 deficient mice and MN mice ( Figure 5B ). The numbers of the total immune cells, eosinophils, T cells, B cells, DC, γδT cells and ILC2, were all decreased in the lungs of B10Q.Ncf1 ^* mice ( Figure 5C ). In particular, the proportions of eosinophils and ILC2 were decreased due to Ncf1 deficiency ( Figure 5D ). However, MN did not reverse the protective effect of Ncf1 deficiency, suggesting that Ncf1 was involved in this process through other types of cells.

On the contrary, Ncf1 appeared to play only a minor role in IL-25 stimulated lung inflammation. The total number of inflammatory cells, or the number of eosinophils and ILC2, as well as their proportion, did not change with Ncf1 deficiency, although macrophages, DC and γδT cells were slightly downregulated ( Figures S4A, B ). Our results suggest that only the lung inflammatory response induced by IL-33, but not IL-25, is regulated by Ncf1.

To explain how Ncf1 regulated IL-33 induced lung inflammation, we determined the cytokine production in this model. Since we found that Ncf1 deficiency could enhance Th1 response in the AIPI model, and αβT cells were also downregulated in the IL-33 induced model, we analyzed the IFN-γ production in CD4⁺ T cells in the lungs by flow cytometry. The results showed that the proportion of IFN-γ⁺ cells in CD4⁺ T cells was increased in B10Q.Ncf1 ^* mice compared with B10Q mice ( Figures 6A, B ). This result again underlined the importance of an enhanced Th1 response in the inflamed lungs regulated by Ncf1.

Next, we addressed whether ILC2 in the lung, known to be activated by IL-33, could be regulated by Ncf1. Therefore, the activation of ILC2 was evaluated through determining the IL-5 and IL-13 production. The results showed that the proportions of IL-5⁺ and IL-13⁺ cells within the ILC2 population, were decreased in B10Q.Ncf1 ^* mice as compared with their wild-type littermates, suggesting that Ncf1 deficiency deactivated ILC2, which could contribute to the disease protection ( Figures 6C–E ). Expression of Ncf1 in alveolar macrophages (i.e. in the MN mouse) did not influence the decreased number of IL5⁺ and IL13⁺ ILC2 lymphocytes ( Figures 6C–E ). Activation of ILC2 was also observed in the IL-25 induced model, but in this model no difference on cytokine production in CD4⁺ T cells and ILC2 could be observed ( Figures S4C, D ).

Since both Th1 enhancement and ILC2 deactivation were observed in Ncf1 deficient mice, to reveal the causality, we blocked the IFN-γ activity in the IL-33 induced model. This was done by treating B10Q.Ncf1 ^* mice with anti-IFN-γ neutralizing antibodies ( Figure 7A ). Compared with isotype IgG group, mice treated with anti-IFN-γ antibodies showed increased eosinophil numbers and proportion in the lungs, although there was no change of the ILC2 numbers ( Figures 7B, C ). The mice treated with anti-IFN-γ antibodies showed an increased proportion of IL-5⁺ ILC2, and the proportion of IL-13⁺ cells in ILC2 also showed an increased trend, suggesting that blocking IFN-γ pathway rescued the ILC2 activation ( Figures 7D, E ). Thus, our results indicate that Ncf1 deficiency enhances Th1 response, and thereby deactivates ILC2 lymphocytes, limiting the development of lung inflammation ( Figure 8 ).