Group 2 innate lymphoid cells (ILC2s) are implicated in host defense and inflammatory disease, but these potential functional roles need more precise definition, particularly using advanced technologies to better target ILC2s and engaging experimental models that better manifest both acute infection and chronic, even lifelong, disease. In this study, we use a mouse model that applies an improved genetic definition of ILC2s via IL-7r–conditional Rora gene targeting and takes advantage of a distinct progression from acute illness to chronic disease, based on a persistent type 2 immune response to respiratory infection with a natural pathogen (Sendai virus). We first show that ILC2s are activated but are not required to handle acute illness after respiratory viral infection. In contrast, we find that this type of infection also activates ILC2s chronically for IL-13 production and consequent asthma-like disease traits that peak and last long after active viral infection is cleared. However, to manifest this type of disease, the Csf1-dependent myeloid–macrophage lineage is also active at two levels: first, at a downstream level, this lineage provides lung tissue macrophages (interstitial macrophages and tissue monocytes) that represent a major site of Il13 gene expression in the diseased lung; and second, at an upstream level, this same lineage is required for Il33 gene induction that is necessary to activate ILC2s for participation in disease at all, including IL-13 production. Together, these findings provide a revised scheme for understanding and controlling the innate immune response leading to long-term postviral lung diseases with features of asthma and related progressive conditions.
This article is featured in In This Issue, p.875
The innate immune system is proving to have new cellular components that orchestrate tissue homeostasis under normal conditions and the response to environmental insults that lead to injury and disease. In particular, there has been a surge in reports on the role of innate lymphoid cells (ILCs) and the special role of group 2 ILCs (designated ILC2s) in the response to infectious and allergenic agents (1). Indeed, some of the first examples of ILC existence and behavior included the immune response to helminth infection and allergen challenge at gut and airway sites in mice (2–8) as a model for what might occur in corresponding human inflammatory diseases such as asthma (9–12). Together, the picture emerged for ILC2s as a key immune component for the development of allergic asthma (13–15). The ILC2 population was also reported to defend against injury due to respiratory viral infections, such as influenza A virus (IAV) (16), that is also linked to asthma (17).
However, these initial studies of ILC2s were challenged by the difficulty in specific definition of ILC2 function, generally based on genetic deletion. This challenge has continued into the latest round of studies using Rag2−/−Il2rg−/−, Bcl11bflox/–, or Plzf−/− mice (13, 18–25) and Il7ra-Cre mice without an additional cell-specific floxed target gene (26, 27). A recent advancement has been the combination of Il7ra-Cre crossed to Rora-flox transgenic mice to more selectively eliminate ILC2s (28, 29), taking advantage of requirements specific for ILC2s (30–32) and avoiding complications of Rorα deficiency at other cell sites (33). Thus, the application of Il7ra-Cre–Rora-flox mice was used successfully to define a role for ILC2s to partner with CD4+ Th2 cells in the type 2 immune response against helminth infection and with Th2 and dendritic cells in the immune memory response to allergen (28, 29). However, this technical advance has not been applied to the key issues of host defense and chronic inflammatory disease. Indeed, the role of ILC2s remains undefined in long-term models in general, leaving uncertainty for the pathological role for ILC2s alone and in combination with other immune and stromal cell populations (1).
To address these issues, the current study engaged mice that were bred as Il7rawt/Cre-Rorafl/fl mice to achieve more selective and efficient deletion of ILC2s. In addition, we employed a mouse model that manifests both acute illness and subsequent progression to chronic inflammatory disease postinfection with the natural mouse pathogen Sendai virus (SeV) (34–37). To date, the SeV model appears similar to the response to human pathogens, such as IAV and respiratory enterovirus (EV-D68) in mice (17, 37) and the type 2 immune response found in humans with lung disease due to asthma and chronic obstructive pulmonary disease (COPD) (34, 35, 38, 39). Together, the present approach offered opportunities not found in models that lack long-term disease and ILC2-specific targeting (40–42) and thereby provided for a series of unexpected findings to establish a distinct paradigm for ILC2 engagement in host defense and inflammatory disease in collaboration with neighboring niches of myeloid and epithelial cell populations.
Materials and Methods
Male and female wild-type (WT) C57BL/6J mice (000664) mice were obtained from The Jackson Laboratory. The Il7rawt/Cre-Rorafl/fl mice were bred by crossing Il7rawt/Cre and Rorafl/fl, generated as described previously (28, 29). The Rorafl/fl mice were kindly provided by A. McKenzie (Cambridge, U.K.), and the Il7ra-Cre mice were provided by Hans-Reimer Rodewald (German Cancer Research Center, Heidelberg, Germany). The Il13wt/gfp mice were initially generated as described previously (7) and were also kindly provided by A. McKenzie. The op/opT mice were generated as described previously (43) and were kindly provided by Nandini Ghosh-Choudhury (University of Texas at San Antonio). These mice were crossed to WT mice to generate heterozygous wt/opT mice and littermate wt/wt control mice as described previously (36). The Il33cherry/cherry mice were generated with insertion of an mCherry reporter cassette between exons 4 and 5 of the Il33 gene in embryonic stem cells from C57BL/6J mice. Levels of IL-33 expression were checked with goat anti-mouse IL-33 Ab (AF3626; R&D Systems) for Western blotting of lung tissue samples, as described previously (44). All mouse strains were maintained on a C57BL/6J genetic background, and except where indicated, were studied as the gene-targeted strain compared with littermate control mice housed under the same conditions.
All mice were maintained and cohoused in a barrier facility using cages fitted with microisolator lids. Animal husbandry and experimental procedures were approved by the Animal Studies Committees of Washington University School of Medicine in accordance with the guidelines from the National Institutes of Health. SeV was obtained from American Type Culture Collection (Sendai/52 Fushimi strain, ATCC VR-105) and prepared and titrated by plaque-forming assay and quantitative PCR assay, as described previously (34). Mice were infected with SeV (2.6 × 105 PFU) as described previously (37). Dosing was performed intranasally using SeV in 30 μl of PBS or an equivalent amount of UV-inactivated virus or PBS alone under ketamine/xylazine anesthesia at 6–9 wk of age. Results from male and female mice were pooled because no significant differences were found between sexes, as reported initially (45) and confirmed recently (37) and in the present experiments (data not shown). Viral titers for stock solutions and lung infections were monitored by PCR assay using primers defined previously (37). All animals underwent daily cage-side observation for clinical behavior and food consumption.
Single-cell suspensions were generated from minced lung tissue that was subjected to collagenase (Liberase TM Research Grade; Roche), hyaluronidase (Sigma-Aldrich), DNase I (Sigma-Aldrich), and Dispase II (Roche) digestion for 45 min at 37°C and then treated with ACK buffer (Lonza) to remove RBCs. Following FcR blockade, lung cell suspensions were incubated with labeled Abs and were sorted using a Sony SY3200 Synergy high-speed cell sorter. The following Abs were used: anti-mouse CD31 (clone MEC 13.3; BD Biosciences), anti-mouse CD45 (clone 30-F11; BD Biosciences), anti-mouse EpCAM (clone G8.8; BioLegend), anti-mouse F4/80 (clone BM8; eBiosciences), anti-mouse Ly-6G (clone 1A8; BD Biosciences), anti-mouse CD11c (clone HL3; BD Biosciences), anti-mouse Siglec-F (clone E50-2440; BD Biosciences), anti-mouse CD11b (clone M1/70; BD Biosciences), Mouse Lineage Mixture (BD Biosciences), anti-mouse 90.2 (clone 53-2.1; BD Biosciences), anti-mouse suppressor of tumorigenicity 2 (ST2; clone DIH9; BioLegend), anti-mouse CD25 (clone PC61; BD Biosciences), anti-mouse CD117 (clone 2B8; BD Biosciences), anti-mouse CD278 (clone 7E.17G9; BD Biosciences), anti-mouse 6A/E (clone E13-161.7; BD Biosciences), anti-mouse MHC class II (clone M5/114.15.2; eBiosciences), anti-mouse CD3e (clone 145-2C11; BD Biosciences), anti-mouse NK1.1 (clone PK136; BD Biosciences), and anti-GFP (clone 1A12-6-18; BD Biosciences). Flow cytometry results were plotted and analyzed using FlowJo software (Tree Star).
Cells were fixed and permeabilized, and lung sections were incubated with citrate-based Ag unmasking solution for Ag retrieval as described previously (38). Immunostaining was performed using the following Abs: chicken anti-GFP (Abcam), rat anti-mouse F4/80 (clone CI:A3-1; Abcam), rabbit anti-human GATA-3 (clone D13C9; Cell Signaling Technology), goat anti-mouse IL-13 (R&D Systems), and goat anti-mouse IL-33 (R&D Systems). Ab binding was visualized using Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary Ab (Life Technologies). All slides were also counterstained with Prolong Gold with DAPI (Life Technologies) and then imaged by immunofluorescent microscopy using a Leica DM5000 B microscope. Staining was quantified in whole lung sections using a NanoZoomer S60 slide scanner (Hamamatsu Photonics) and ImageJ software.
Real-time quantitative PCR assay
RNA was purified from lung homogenates and cells using TRIzol (Invitrogen) and was converted to cDNA using a High-Capacity cDNA Archive Kit (Life Technologies). Target mRNA was quantified by real-time PCR assay, using specific fluorogenic probes and primer sets and the Fast Universal PCR Master Mix system (Applied Biosystems). The forward and reverse primers and probes were as follows (in that order): for SeV-NP, 5′-GGCGGTGGTGCAATTGAG-3′, 5′-CATGAGCTTCTGTTTCTAGGTCGAT-3′, and 5′-AGCTCTAGACAATGCC-3′; for Il13, 5′-GGTGCCAAGATCTGTGTCTC-3′, 5′-CCACACTCCATACCATGCTG-3′, and 5′-AAGACCAGACTCCCCTGTGCAAC-3′; for mucin 5AC (Muc5ac), 5′-TACCACTCCCTGCTTCTGCAGCGTGTCA-3′, 5′-ATAGTAACAGTGGCCATCAAGGTCTGTCT-3′, and 5′-TATACCCCTTGGGATCCATCATCTACA-3′; for Arg1, 5′-AGTGTTGATGTCAGTGTGAGC-3′, 5′-GAATGGAAGAGTCAGTGTGGT-3′, and 5′-ACAGTCTGGCAGTTGGAAGCATCT-3′; for Il33, 5′-TCATGTTCACCATCAGCTTCT-3′, 5′-GTGCTACTACGCTACTATGAGTC-3′, and 5′-ACCGTCGCCTGATTGACTTGCA-3′; for Il12b, 5′-TGTCCTCAGAAGCTAACCATC-3′, 5′-TCCAGTCCACCTCTACAACA-3′, and 5′-ACGTCTTTCTCCAGCTCCCACATG-3′; for Gata3, 5′-CCTTATCAAGCCCAAGCGAA-3′, 5′-GTCCCCATTAGCGTTCCTC-3′, and 5′-TGTCCCTGCTCTCCTTGCTGC-3′; for Rora, 5′-TGGAGACAAATCGTCAGGAAT C-3′, 5′-GACAGGAGTAGGTGGCATTG-3′, and 5′-TGG TGT CATTAC GTG TGA AGG CTG C-3′; for Il1rl1, 5′-AAT CCT CCA TAC AAC CAC ACA A-3′, 5′-GACATCAGCCAAGAAGTGAGAG-3′, and 5′-AAGTAT TGCCTGTTCAGCTTGCTTTGG-3′; and for Areg, 5′-GTCACTATCTTTGTCTCTGCCA-3′, 5′-CCTCCTTCTTTCTTCTGTTTCTCC-3′, and 5′-AGTATCGTTTCCAAAGGTGCACTGTGA-3′. Samples were assayed with the 7300HT or QuantStudio 6 Fast Real-Time PCR System and analyzed using Fast system software (Applied Biosystems). All real-time PCR data were normalized to the level of GAPDH mRNA. Values were expressed as fold change, based on the ΔΔCt method as described previously (35), with the exception of Fig. 5C, in which Il13 mRNA was quantified by copy number using an Il13-expressing plasmid as an internal standard (38).
For IL-33 production in lung tissues, mouse lungs were homogenized in RIPA buffer (Sigma-Aldrich) supplemented with protease inhibitor mixture (Roche) and 1 mM EDTA. For IL-33 production in bronchoalveolar lavage (BAL), mouse lungs were washed twice with 1 ml of PBS on ice. The samples from each mouse were pooled, spun to remove cells, and supplemented with protease inhibitor mixture and 1 mM EDTA. Levels of IL-33 were determined using the DuoSet ELISA Kit (R&D Systems).
All data are presented as mean and SEM and are representative of at least three experiments with at least five data points per experiment. An unpaired Student t test with Bonferroni correction as well as mixed-model repeated measures ANOVA with Tukey correction for multiple comparisons were used to assess statistical significance between means. In all cases, the significance threshold was set at p < 0.05.
ILC2s do not influence acute respiratory viral infection
Initial experiments established the presence of ILC2s in the lung at baseline using flow cytometry for forward scatter and side scatter (SS) combined with absence of immune cell lineage markers (CD11c, NK1.1, CD3e, CD45R/B220, CD11b, TER-119, and Ly-6G/6C) and presence of CD90.2 (alloantigen Thy-1.2) and ST2 (IL-33R subunit IL-1rl1) (Fig. 1A). This cell population also expressed CD25 (IL-2rα), CD117 (c-Kit), CD127 (IL-7rα), CD278 (ICOS), Ly-6A/E (Sca1), and MHC class II, based on additional flow cytometry analysis (Fig. 1B). Together, this profile for ILC2s is similar to reports using the comparable methods (16, 29, 46, 47). Moreover, this ILC2 population was no longer detected in the lungs of Il7rawt/Cre-Rorafl/fl mice that were designed to specifically delete ILC2s (Fig. 1C). ILC2 deficiency was consistent with previous work using this genetic approach (28, 29).
These findings allowed us to next test the effect of ILC2 deficiency on the outcome from respiratory viral infection. As introduced above, we used SeV, a natural mouse pathogen capable of replicating at high efficiency and causing acute infectious illness across outbred and inbred strains (including C57BL/6J). In this setting, we found no significant changes in the typical weight loss, viral RNA level and clearance, or immune cell accumulation in Il7rawt/Cre-Rorafl/fl mice compared with WT control mice (Fig. 1D–F). Similarly, we found no significant differences in ILC2-deficient mice in Il12b mRNA (Fig. 1G), a validated marker of acute inflammation after SeV infection (48). In addition, we detected no significant differences in amphiregulin (Areg) mRNA level in ILC2-deficient mice, except for a slight decrease at 12 d postinfection (Fig. 1G). However, by this time, repair of epithelial injury and clearance of infectious virus are already complete (37, 49). Thus, the results are consistent with normal recovery in ILC2-deficient mice as noted above (Fig. 1D–F). In addition, we detected attenuation of the usual increase in lung levels of Il13 mRNA at the time of peak viral titer (5 d postinfection) and not at other time points (Fig. 1G), consistent with a transient type 2 immune response found after viral infection and the role of ILC2s in IL-13 production during acute activation in the gut and lung (4, 5, 7, 8, 10).
ILC2s contribute to chronic lung disease after viral infection
We next studied the behavior and influence of ILC2s during the development of chronic lung disease that develops after clearance of active viral infection. For SeV infection, chronic disease becomes detectable at 21 d and maximal at 49 d postinfection, the key time points for study (34, 35, 37, 50). This strategy showed that the levels of ILC2s in the lung were increased markedly at 21 and 49 d after SeV infection in WT mice and were more pronounced than the increases at 5 d postinfection or the overall increases in total cell numbers that are typical of acute illness and chronic lung disease (Fig. 2A). In addition, ILC2s from the lung that were FACS purified showed an increased induction of ILC2 biomarkers Il13, Gata3, and Rora mRNA in response to IL-33 stimulation at 49 d after SeV infection compared with SeV-UV controls (Fig. 2B). As expected, the increase in ILC2s found in WT mice was markedly attenuated in Il7rawt/Cre-Rorafl/fl mice, based on flow cytograms at 49 d postinfection (Fig. 2C). Together, these findings provided for ILC2 accumulation and activation in concert with the development and progression of chronic lung disease after viral infection.
To pursue any possible function of ILC2s in the development of chronic lung disease, we next compared WT to Il7rawt/Cre-Rorafl/fl mice for signs of this disease at 49 d after SeV infection. Consistent with ILC2 accumulation and activation with disease, we found significant attenuation of hematoxylin+ cell staining (a sign of focal immune and epithelial cell accumulation) in lung sections from Il7rawt/Cre-Rorafl/fl compared with WT mice at 49 d after SeV infection (Fig. 2D, 2E). Similarly, we found attenuation of increased periodic acid-Schiff (PAS)+ staining (a sign of focal mucus production) in lung sections from Il7rawt/Cre-Rorafl/fl compared with WT mice (Fig. 2F, 2G). In concert with these observations, we also detected reduction of the usual increases in Il13, Muc5ac, and Arg1 mRNA levels (as signs of a type 2 immune response and mucus production) in the lungs of Il7rawt/Cre-Rorafl/fl compared with WT mice at 49 d after SeV infection (Fig. 2H). Together, the findings established a significant role for ILC2s in the development of chronic lung disease after viral infection. However, we also recognized that the degree of attenuation in Il7rawt/Cre-Rorafl/fl mice was incomplete in comparison with Il13−/− mice for all phenotypes [i.e., hematoxylin+ staining (Fig. 2D, 2E), PAS+ staining (Fig. 2F, 2G), and Il13, Muc5ac, and Arg1 mRNA levels (Fig. 2H)]. These results indicated that ILC2 deficiency alone cannot fully block the development of chronic lung disease after viral infection, and the remaining effect might be due at least in part to a separate cellular mechanism for IL-13 production in this setting.
ILC2s and macrophages contribute to IL-13 production
Our previous work indicated that lung macrophages (but not mast cells, basophils, neutrophils, dendritic cells, B cells, CD4+ T cells, CD8+ T cells, NK cells, or NKT cells) can also be a source of increased IL-13 production after viral infection, based on PCR assay for Il13 mRNA and immunostaining for IL-13, and thereby contribute to the development of chronic lung disease after SeV infection (34, 36). To more precisely define the cellular source of IL-13, in this study, we studied IL-13–GFP reporter mice, using heterozygous Il13wt/gfp mice that manifest the same degree of acute illness and chronic lung disease as WT control (Il13wt/wt) mice after SeV infection (data not shown). Accordingly, we found a similarly increased percentage of SSlow lineage−ST2+CD90.2+ ILC2s in lungs of Il13wt/wt and Il13wt/gfp mice at 49 d after SeV infection compared with SeV-UV control, based on flow cytograms from these conditions (Fig. 3A). In addition, we found a marked increase in GFP+ ILC2s at 49 d after SeV infection compared with SeV-UV in lungs of Il13wt/gfp mice, and both of these values were increased compared with background levels found in WT mice (Fig. 3A). Similarly, we found high-level IL-13 expression marked by GFP fluorescence in ILC2s from the lungs of Il13wt/gfp mice compared with Il13wt/wt mice, and this signal was significantly increased at 49 d after SeV infection compared with SeV-UV control, based on mean fluorescence intensity (MFI) (Fig. 3B, 3C).
We also checked for GFP expression in lung macrophage populations, based on flow cytometry schemes that separated these populations into alveolar macrophages (Ly-6G−CD11c+Siglec-F+F4/80+CD11b−) and tissue macrophages (Ly-6G−CD11c−Siglec-F−F4/80+CD11b+) and further separated tissue macrophages into interstitial macrophages and tissue monocytes, based on differences in cell size (Fig. 4A, 4B). In contrast to results with ILC2s, we found no detectable GFP+ population of macrophages (alveolar or tissue subsets) at 49 d after SeV infection compared with SeV-UV control using detection of fluorescence directly from GFP or anti-GFP mAb (Fig. 4C, 4D). As expected from our previous work (34), we also found no detectable IL-13–GFP signal in NKT cells (SSlowCD3e+NK1.1+) or T cells (SSlowCD3e+NK1.1−) (Fig. 4C, 4D). Similarly, we found no significant IL-13–GFP signal in eosinophils (Ly-6G−CD11c−Siglec-F+F4/80−CD11b+) (Fig. 4C, 4D) that contribute to IL-13 production in other models (51). Together, these findings again show that IL-13–expressing ILC2s were recruited and activated during the chronic lung disease that develops after SeV infection but did not as yet identify alternative cell sources of IL-13 production.
To investigate this issue, we developed additional approaches that might better define any role for lung macrophages as a source of IL-13 production in this model (34, 36), especially relative to ILC2s and in relation to macrophage subsets. Accordingly, we combined FACS of alveolar and tissue macrophages with a PCR-based assay for Il13 mRNA to establish cell-type sites of IL-13 expression. Similar to the results from the IL-13–GFP reporter approach, we found that the level of Il13 mRNA per cell at baseline (without SeV infection) and after induction at 5 and 49 d after SeV infection was highest in lung ILC2s compared with macrophages, in which only low levels could be detected in tissue macrophages at 49 d postinfection (Fig. 5A). To further define the site of IL-13 production, we also determined IL-13 protein levels in BAL fluid and lung tissue. This approach demonstrated increases in IL-13 levels in both compartments at 5 and 49 d after SeV compared with SeV-UV in WT mice, and these increases were attenuated in ILC2-deficient Il7rawt/Cre-Rorafl/fl mice (Fig. 5B). However, despite the selective increase in Il13 mRNA in ILC2s, the loss of ILC2s did not fully block IL-13 production in the lung or release into the airway/alveolar space after SeV infection. This finding again raised the possibility for another cellular source of IL-13 in the lung, with the tissue macrophage as a particular candidate.
To address this issue, we compared Il13 mRNA levels in the chief cell sources (tissue macrophages and ILC2s) as mRNA copy number per cell and per lung. Using this comparison in our analysis of Il13 mRNA per cell, we again found a relatively low level of Il13 mRNA in tissue macrophages compared with ILC2s (Fig. 5C). However, when we expressed Il13 mRNA as levels per lung (where lung level equals level per cell × number of cells per lung), we recognized that more abundant tissue macrophages are the main site of Il13 mRNA expression at 49 d after SeV infection (Fig. 5C). In fact, the fold increase and overall level of Il13 mRNA in the lung at 49 d after SeV infection was significantly higher for tissue macrophages than ILC2s (Fig. 5C). The comparative analysis for eosinophils (gated as shown in Fig. 4A) showed Il13 mRNA levels similar to tissue macrophages, but the eosinophil number was not sufficient to provide a substantial contribution to Il13 expression in the lung (Fig. 5C). As introduced above, the larger number of lung macrophages was also accompanied by increased postviral induction of Il13 gene expression, with 21-fold increase in tissue macrophages compared with 5.2-fold increase in ILC2s and 5.1-fold increase in eosinophils.
To further assign production of IL-13 to macrophages, we also performed immunostaining for IL-13 protein production in lung tissue and FACS-isolated cells. Tissue immunostaining for IL-13 itself might not discriminate sites of cell production versus cell binding or uptake. We overcame this difficulty by staining for GFP expression in Il13wt/gfp mice. This approach showed cells with tissue macrophage morphology and location and dim staining as well as cells with ILC2 morphology and bright staining for IL-13–GFP, with no background signal with GFP-only staining (Fig. 5D). Cells identified as ILC2s by GFP-staining intensity and morphology were also found to immunostain positive for GATA3 (Fig. 5E), as a marker of type 2 lymphocytes. Further, this approach showed that cells with macrophage morphology and location were not stained for GATA3 (Fig. 5E) but, instead, were costained for IL-13–GFP and F4/80 in lung sections from Il13wt/gfp mice at 49 d after SeV infection but only for F4/80 in SeV-UV control mice (Fig. 5F). These results were similar to macrophage immunostaining for IL-13 in previous work (34–36). In this study, again, we identified cells with lymphoid morphology that were stained even more brightly for IL-13–GFP but were not costained for F4/80 (Fig. 5F), consistent with identification as IL-13–expressing ILC2s. Image analysis quantitation of tissue immunostaining indicated that GFP+F4/80+ cells (consistent with IL-13–expressing tissue macrophages) were far more abundant than GFP+F4/80− cells (consistent with IL-13–expressing ILC2s) (Fig. 5G). As introduced above, we also checked for IL-13 expression in FACS-isolated tissue macrophages, in this case by immunostaining with anti-mouse IL-13 Ab, which can define IL-13 production in preparations of purified cells. In this study, again, we demonstrated that tissue macrophages defined by flow cytometry and morphology characteristics were immunostained for IL-13 when obtained from the lungs at 49 d after SeV but not after SeV-UV control (Fig. 5H). Together, these results provided evidence for marked induction of Il13 gene expression in tissue macrophages during chronic lung disease after viral infection and for this population along with ILC2s to represent prominent sites of IL-13 production. This combined contribution might therefore lead to an increased concentration-dependent effect of IL-13 to better explain continued IL-13–dependent lung disease despite ILC2 deficiency (Fig. 2D–H).
ILC2 activation depends on myeloid–macrophage input to ST2
To further address macrophage contribution to IL-13 production, we next incorporated a genetic strategy for targeted Csf1 deficiency that would markedly downregulate the myeloid-macrophage lineage and, in turn, tissue macrophage levels. For this approach, we started with Csf1op/opT (op/opT) mice that carry an osteocalcin-driven Csf1 transgene (T) that rescues the osteopetrosis but not the macrophage defect in Csf1op/op (op/op) mice (43). We then generated heterozygous Csf1wt/opT (wt/opT) mice that still manifest a significant decrease in the level of lung tissue macrophages (i.e., interstitial macrophages plus tissue monocytes) and attenuation of chronic lung disease (36). Comparison of wt/opT to littermate WT (wt/wt) mice can therefore provide an indication of tissue macrophage contribution to phenotype, including IL-13 expression. In this setting, we found no significant changes in the typical weight loss, viral RNA level and clearance, or immune cell accumulation in wt/opT mice compared with wt/wt mice (Fig. 6A–C). These results indicated that downregulation of the myeloid–macrophage lineage did not significantly influence acute illness after viral infection, similar to the case for ILC2 deficiency.
In contrast to the results for acute illness, we detected informative differences in the development of chronic disease in wt/opT mice. In particular, we found that the increased levels of Il13 mRNA were localized selectively to both subsets of tissue macrophages and that Il13 mRNA induction was markedly downregulated in wt/opT compared with wt/wt mice (Fig. 6D). The IL-13 target Arg1 mRNA was upregulated in both alveolar and tissue macrophages in corresponding WT mice (i.e., wt/wt mice), and this effect was fully blocked in wt/opT mice (Fig. 6D). These results were consistent with our earlier observations that tissue macrophages were sensitive to Csf1-dependent downregulation (36), in this case leading to decreases in Il13 and Il13-dependent gene expression.
To further define macrophage control of IL-13 production, we also determined the levels of ILC2 recruitment and activation in the lungs of wt/opT mice compared with wt/wt mice. Unexpectedly, we found first that the increased percentage of SSlow Lin−CD90.2+ST2+ ILC2s at 49 d after SeV in wt/wt mice was attenuated in wt/opT mice without a significant change in baseline percentages after SeV-UV (Fig. 6E). Quantitation of flow cytometry analysis showed the usual increase in total cell number after SeV infection in wt/opT mice (Fig. 6F) that was similar to wt/wt mice (Fig. 2A). We also detected an increase in ILC2 levels at 5 d after SeV infection in wt/opT (Fig. 6F), again similar to wt/wt control mice (Fig. 2A). However, the marked increases in ILC2 levels in the lung at 21 and 49 d after SeV infection found in wt/wt mice (Fig. 2A) were no longer significant in wt/opT mice (Fig. 6F), recognizing that ST2 expression can represent cell number and activation as contributions to cell function. We did not find a significant difference in ILC2 levels in wt/opT compared with wt/wt mice (Fig. 6F versus Fig. 2A) at any timepoint in SeV-UV controls, indicating that there was no defect in ILC2 development and migration to the lung tissue under baseline conditions.
Similar to the results for ILC2 levels, we found that the usual induction of Il13 and Arg1 mRNA found in ILC2s from the lungs of wt/wt mice were fully blocked in wt/opT mice at 49 d after SeV infection (Fig. 6G). Further, the levels of Gata3 and Il1rl1 mRNA (the latter encoding for the IL-33R designated ST2) were markedly downregulated in ILC2s from wt/opT mice at 49 d after SeV or SeV-UV infection (Fig. 6G). In contrast to Il1rl1 mRNA, levels of ST2 protein on ILC2s were upregulated at 49 d after SeV infection, and this effect was fully blocked in wt/opT mice, based on cell counts (Fig. 6H) or MFI (Fig. 6I). Similar induction of ST2 expression after SeV infection was also found on tissue macrophages, based on flow cytometry analysis by histogram and cytogram (Fig. 7) albeit on a smaller subset of this cell population compared with broad expression across the ILC2 population (Fig. 6H). Nonetheless, both findings suggested a basis for enhanced ST2-dependent activation of ILC2s and tissue macrophages for IL-13 production and for the unexpected role of Csf1-dependent myeloid cells in controlling ILC2 accumulation and activation as a distinct mechanism for chronic lung disease after viral infection. Moreover, the loss of myeloid cell input resulted in a targeted decrease in Il1rl1/ST2 expression and the consequent downregulation of the type 2 immune response as marked by Il1rl1, Gata3, and ST2 levels after viral infection in wt/opT mice.
ILC2 activation via ST2 depends on IL-33 production
To further develop this new disease mechanism, we next sought to define the basis for increased ST2 expression and the impact on ST2 signaling function after viral infection. As introduced above, we recognized that ST2 signal transduction was required for induction of Il13 gene expression, and this signal depended on IL-33 as an ST2 receptor ligand in the present and previous mouse models, in which mucus production is markedly attenuated by anti-ST2 mAb or Ilrl1 or Il33 gene targeting (35, 52). To better define this issue, we generated IL-33–deficient mice in which a cherry reporter gene cassette was inserted between exons 4 and 5 of the Il33 gene (Fig. 8A, 8B). Accordingly, homozygous Il33cherry/cherry mice were unable to generate any detectable IL-33 protein in the lung at 49 d after SeV infection or SeV-UV infection (Fig. 8C). Nonetheless, the acute illness after SeV infection was no different in Il33cherry/cherry mice compared with Il33wt/wt mice, based on cage-side observation, body weight loss, viral RNA level, and histopathology (Fig. 8D–F). These results are also similar to our previous findings using Il33Gt/Gt mice that show no differences from WT mice in these same parameters after SeV infection (35).
Despite similarities in acute illness, the development of chronic lung disease marked by the usual increases in inflammatory cells and PAS+ airway mucous cells in lung sections (Fig. 9A, 9B) and in levels of Il13 and Arg1 mRNA in lung tissue (Fig. 9C) were all significantly attenuated in Il33cherry/cherry mice compared with Il33wt/wt mice at 49 d after SeV infection. In addition, lung levels of Gata3 and Il1rl1 mRNA were decreased in Il33cherry/cherry mice after SeV and SeV-UV infection (Fig. 9C). The same pattern of downregulated gene expression was found in ILC2s from Il33cherry/cherry mice (Fig. 9D) in concert with decreased levels found in ILC2s from wt/opT mice (Fig. 6G). In addition, we found that the increased percentage of ILC2s in the lungs of Il33wt/wt mice at 49 d after SeV infection was also significantly downregulated in Il33cherry/cherry mice, based on flow cytometry analysis for SSlow Lin−CD90.2+ ST2+ ILC2s (Fig. 9E). Quantitation of this analysis confirmed that the marked increases in the numbers of ILC2s in the lungs at 21 and 49 d after SeV infection in Il33wt/wt mice were significantly decreased in Il33cherry/cherry mice (Fig. 9F). This analysis also demonstrated a slight decrease in ILC2 levels in Il33cherry/cherry compared with Il33wt/wt mice at baseline in SeV-UV controls (Fig. 9F), consistent with the relatively small effect of IL-33 on ILC2 development in the bone marrow and migration to the lung tissue under baseline conditions (53). In follow-up to the attenuation of ILC2 levels, we also observed that the increased levels of ST2 on ILC2s at 49 d after SeV infection in Il33wt/wt mice were no longer found in Il33cherry/cherry mice, based on cell counts (Fig. 9G) and MFI (Fig. 9H). Together, these findings indicated that IL-33 production was a suitable candidate for Csf1-dependent myeloid cell control of ILC2 accumulation and activation for chronic lung disease after viral infection.
IL-33 production depends on myeloid–macrophage lineage input
We next aimed to determine whether, in fact, myeloid cells could control IL-33 production as a mechanism to regulate the development of disease after viral infection. As a first step, we combined a flow cytometry scheme for separating CD31−CD45−EpCam+ lung epithelial cells (Fig. 10A) with our scheme for lung macrophages (Fig. 4A) to quantitatively track the cell site of IL-33 expression. This combined analysis showed that Il33 mRNA was almost entirely confined to lung epithelial cells at baseline levels without infection and at increased levels at 49 d after SeV infection (Fig. 10B). This finding was consistent with the predominant site of IL-33 expression in the mouse lung being localized to alveolar epithelial type 2 (AT2) cells, based on cell and tissue morphology and coexpression with AT2 cell markers such as surfactant protein C (Sftpc) (35, 52, 54, 55). Indeed, we also found increases in IL-33+ immunostaining localized to cells with AT2 cell morphology and location in wt/wt mice at baseline (Fig. 10C). No such immunostaining was found in Il33-deficient cherry/cherry reporter mice as a sign of specificity for IL-33 detection. This approach also showed that IL-33+ immunostaining was increased in wt/wt mice at 49 d after SeV infection compared with SeV-UV infection, and this increase was attenuated in wt/opT mice (Fig. 10C). These findings were confirmed as significant with quantitation of IL-33+ cell levels in lung tissue (Fig. 10D). In addition, we found that the usual increase in Il33 mRNA and the consequent increase in Il13 mRNA at 49 d after SeV infection in the lungs of wt/wt mice were no longer detectable in wt/opT mice (Fig. 10E). Together, these results are consistent with an effect of Csf1-dependent myeloid cells to promote epithelial cell expression of IL-33 (in lung epithelial cells) that in turn drives downstream IL-13 production and feed-forward ST2 expression (in ILC2s and tissue macrophages) that are all critical for the development of chronic lung disease after viral infection (Fig. 11).
In this study, we apply improved genetic technologies and a distinct mouse model to define the role of ILC2s in acute illness and chronic disease after respiratory viral infection. Relative to previous work, the current study provides a series of unexpected conceptual advances. In particular, we show that ILC2s are activated during respiratory viral infection with a natural mouse pathogen (SeV) but are not required to handle acute illness and recovery from this type of infection. In contrast, we find that this type of infection also activates ILC2s chronically for IL-13 production that is required for the development of long-term postviral lung disease with features of asthma and COPD. Notably, however, ILC2s require critical activities of Csf1-dependent myeloid cells at two levels: first, at a downstream level for significant additional IL-13 production; and second, at an upstream level for enhanced IL-33 expression that is essential for ILC2 participation at all in the development of disease. Together, these findings provide a revised scheme (as depicted in Fig. 11) that appears distinct from the conventional view of an innate immune response engineered for short-term versus long-term activation and function. This alternative paradigm thereby provides a new framework for understanding and controlling the innate immune response to viral infection and, likely, other stimuli of the type 2 immune response. In this study, we discuss several of our observations that are critical to this issue.
First, we address the possible function of ILC2s in host defense against respiratory viral infection. As noted in the 1Introduction, ILC2s were reported to be responsible for clinical deterioration via poor epithelial repair postinfection with IAV (PR8 strain), based on ILC2 depletion using anti-CD90.2 mAb in Rag−/− mice (16). More recently, another group found decreased survival after IAV (California/04/2009 strain) infection, based on ILC2 depletion with Il7ra-Cre–Rora-flox mice (56). The combined studies of IAV implicate an IFN-γ–IL-5–amphiregulin axis for the ILC2-dependent phenotype after IAV infection. Similarly, others find a role for IL-33–IL-33R signal transduction (that should activate ILC2s) in defense against IAV infection (16, 56). In contrast, we found no detectable effect of ILC2 depletion on recovery from acute illness (or induction of amphiregulin expression during tissue repair) and no effect of IL-33 or IL-33R deficiency on acute illness in the present or previous work on SeV infection (35). Whether the low number and/or activation of ILC2s and the consequent absence of function during acute illness found in our model is present in other conditions is uncertain. Nonetheless, there are no reported increases in susceptibility to severity of respiratory viral infections in humans treated with mAbs that block IL-5–IL-5R or IL-33–IL-33R signal transduction (57–59), and IFN-γ–deficient humans are limited to susceptibility to mycobacterial infection (60). Our results are also consistent with normal susceptibility to viral infection in ILC-deficient humans (61). Thus, it is possible that ILC2s might manifest a distinct role specific to IAV infection adapted to mice, recognizing that studies of this issue are ongoing.
Second, we address the key question of ILC2 function in chronic inflammatory disease. In this study, we took special advantage of the capacity of SeV infection to trigger long-term lung disease lasting the 1–2 y lifespan of mice (50), analogous to epidemiology data that links severe respiratory viral infection in infancy to lifelong lung disease in humans (62). Consistent with these observations, we find that ILC2s are activated persistently after viral infection, thereby providing a basis for contributing to chronic lung disease that develops long after infectious virus is cleared. This prolonged time course is distinct from reports of short-term ILC2 activation after other stimuli [e.g., allergen challenge and infections with helminths and respiratory viruses in mice (4, 5, 7, 8, 10, 41, 42) and recently in humans (63)]. These studies often focused on the ILC2 population as a source of IL-13 production and therefore a driver for IL-13–dependent disease that is characteristic of a type 2 immune response. In this study, we also find that the ILC2s exhibit high-level IL-13 expression in concert with the later development and progression of lung disease after viral infection. However, we also discover two distinct features of the ILC2-based disease paradigm.
One of those elements is the observation that ILC2s cannot fully account for IL-13 production that develops during postviral lung disease. This observation led us to identify tissue monocytes and interstitial macrophages (that together we designate as tissue macrophages) as an additional site for induction of Il13 gene expression during chronic lung disease after viral infection. The present approach incorporated ILC2-deficient mice to quantify the contribution to lung IL-13 levels. In addition, the strategy incorporated Il13-gfp transgene reporter mice to better localize the cell-type source of Il13 gene expression. Together, these strategies defined Il13 gene expression in ILC2s (at high levels per cell) and tissue macrophages (at relatively lower levels per cell as also found in eosinophils) using a combination of FACS with PCR-based assay and immunostaining for IL-13 expression and tissue immunostaining for GFP reporter expression to enhance our previous analysis of lung macrophages (34–36). Together, the approach significantly extends evidence of IL-13 expression in mouse and human macrophages from our laboratory and other laboratories in the context of a type 2 immune response (34, 36, 64, 65). In particular, this full battery of assays provides a quantitative basis for the relative contributions of ILC2s versus tissue macrophages for the key IL-13 cytokine in the context of chronic disease. However, it is not yet possible to fully separate cytokine contributions of macrophages versus ILC2s because of another previously unrecognized control for ILC2 function.
This second feature of ILC2 function in chronic disease is the newly identified requirement for additional myeloid cell participation separate from direct production of IL-13. Thus, we incorporated wt/opT to downregulate the Csf1-dependent myeloid cells (particularly tissue macrophages) and thereby define the relative contributions to IL-13 production. As introduced above, this approach resulted in marked downregulation of Il13 gene expression and IL-13 protein production and, in turn, complete attenuation of chronic asthma-like lung disease after viral infection. Unexpectedly, we also found that wt/opT mice additionally manifested a marked downregulation of ILC2 expansion and activation. The decreased activation included attenuation of ST2 expression via transcriptional and posttranscriptional regulation (the former based on decreased mRNA levels in wt/opT mice and the latter based on an increase in ST2 protein but not corresponding mRNA in ILC2s from WT mice and blockade of this increase in wt/opT mice). Comparable, albeit lower level, ST2 expression was also found in tissue macrophages after SeV infection, suggesting similar regulation of ST2 levels in this cell population. In that regard, the present results were also similar to the appearance of Trem-2 on tissue macrophages after SeV infection, again based on increases in protein with no change in corresponding mRNA levels (36). Together, the findings continue to highlight the role of posttranscriptional control of the type 2 immune response after viral infection but left unchecked how this control might occur.
This question about control mechanism was also resolved by unexpected findings. Thus, we generated a line of Il33-cherry transgenic reporter mice based upon the role of IL-33–dependent signal transduction in controlling IL-13 production (35). Analysis of this mouse line showed that Il33 gene expression was required for ILC2 accumulation and activation, perhaps as expected, but also revealed that Il33 gene function was required for the transcriptional increase in Il1rl1 mRNA and posttranscriptional upregulation of ST2 on the surface of ILC2s during the development of chronic lung disease after viral infection. Moreover, this essential Il33-based signal was lost in wt/opT mice, thereby establishing an unrecognized role for Csf1-dependent myeloid cells in controlling ILC2 participation in the disease. Given that the major site for induction of Il33 gene expression is localized to lung epithelial cells (particularly AT2 cells in this model and other models) (35, 52, 54, 55), the findings also implied that Csf1-dependent myeloid cell control of the epithelial cell–derived IL-33 was in turn responsible for ST2-dependent ILC2 activation. These findings are consistent with macrophage support of AT2 cell expansion found after helminth infection and other models of short-term tissue injury and repair (66–69). However, these previous models defined an effect of macrophages (often M2 macrophages) that is downstream of IL-13 production (67–69). Instead, in the present case, the Csf1-dependent myeloid cell effect is upstream of the epithelial cell to IL-33 production to ILC2 activation to IL-13 production axis that drives disease. We therefore propose that Csf1-dependent myeloid cells might represent an immune cell niche for lung epithelial (likely AT2) cells. This aspect of regulating ILC2 function appears to have no clear precedent, based on a review of previous work (70). Since this review was published, others have reported that Csf1-dependent macrophages are required for maintenance of Paneth cells and nearby stem cells in the intestinal epithelium (71), but full definition of the macrophage subset was not performed, and no cytokine biology or disease model was studied. The present possibility for myeloid lineage control of IL-33 expression in AT2 cells thereby significantly advances this concept.
Together, the present results provide a new framework as well as new questions for the development of progressive postviral lung disease in particular and chronic inflammatory disease in general. Given the distinct nature of our findings, the myeloid cell effect on lung epithelial cells cannot yet be attributed to tissue macrophages versus another subset of the Csf1-dependent myeloid cell lineage. This issue still needs to be defined, most likely by identifying molecular factors that control Csf1-dependent myeloid cell activation and effector function in this setting and performing the corresponding loss-of-function and reconstitution experiments. Similarly, we still need to determine the precise epithelial cell target population and how it is regulated by myeloid lineage cells during acute illness and chronic disease in the present model and in general. In that context, our study nonetheless provides a distinct cell and molecular axis from viruses to innate myeloid cells (a Csf1-dependent subset) to epithelial cells (likely an AT2 cell subset, at least in mice) to ILCs (an ILC2 subset) in the development of persistent inflammatory disease. In the present model, regulation of ST2 levels provides a key checkpoint for the development of long-term disease. It will therefore be of interest to revisit whether a similar molecular and/or cellular mechanism will also participate in the type 2 immune response in related settings (e.g., allergy and parasitic infection). These comparisons will also define additional molecular interactions that drive this axis and thereby provide a therapeutic target for correcting the consequent inflammatory disease without interfering with host defense. Nonetheless, these findings already better explain the cellular context for ILC2 function and the specialized need for myeloid cell cooperation for the development of a long-lasting type 2 immune response that allows progression from acute infectious illness to long-term lung disease. This type of postviral disease resembles immunopathologic features found in asthma and COPD as well late events in the course of severe respiratory virus infections in general. Thus, postviral lung disease can develop after severe infections because of respiratory syncytial virus, influenza virus, respiratory enterovirus, and coronavirus (CoV) in humans and in animal models, including CoV outbreaks causing severe acute respiratory syndrome and CoV disease 2019 (17, 37, 63, 72, 73). Therefore, defining pathogenesis and consequent therapeutic targets for progressive postviral lung disease is critical to present and major public health concerns.
We thank Di Wu, Xiaohua Jin, and Rose Tidwell and the staff in the Siteman Flow Cytometry Core and Pulmonary Morphology Core at Washington University for expert assistance.
This work was supported by grants from the National Institute of Allergy and Infectious Diseases (AI130591), National Heart, Lung, and Blood Institute (HL145242), and the Cystic Fibrosis Foundation.
Abbreviations used in this article:
alveolar epithelial type 2
chronic obstructive pulmonary disease
influenza A virus
innate lymphoid cell
group 2 ILC
mean fluorescence intensity
suppressor of tumorigenicity 2
M.J.H. declares that he is a member of the Data Safety Monitoring Board for AstraZeneca and is the founder for NuPeak Therapeutics. The other authors declare no competing financial interests.