Age-Dependent Reduction in Asthmatic Pathology through Reprogramming of Postviral Inflammatory Responses

This article is featured in Top Reads, p.

Asthma is a common lung disease characterized by paroxysms of airway obstruction, airway hyperresponsiveness, and inflammation. In susceptible individuals, the disease arises when infectious agents, allergens, or irritants trigger a stereotypical pattern of airway pathology, which is then exacerbated through recurrent exposures to these agents over time. Histologic asthma phenotypes include airway mucous cell metaplasia, peribronchial infiltration of immune cells, and airway wall thickening (1). Although asthma can arise at any age, most patients first present in early childhood (<6 y old), often in association with a viral wheezing illness (2, 3). Childhood-onset asthma is often marked by a “type 2” pattern of airway inflammation that includes airway eosinophilia and the secretion of certain cytokines like IL-4, IL-5, and IL-13 (4).

The triggers that drive asthma exacerbations are ubiquitous, and asthma is commonly regarded as a lifelong disease. However, permanent symptoms are not inevitable for many patients, especially those diagnosed with asthma in early life (5). Roughly 50% of children diagnosed with asthma before the age of 6 will experience remission of their symptoms by early adulthood, and 20–25% of cases will achieve resolution of their airway hyperresponsiveness (6, 7). Population studies indicate that asthma exacerbation rates fall progressively as children age, reaching a nadir by early adulthood (8). Why some children with asthma “age out” of their exacerbations is poorly understood.

One clue about the origins of asthma remission is that the seasonality of asthma exacerbations wanes as children mature (9, 10). This observation might imply a developmental process that mutes host responses to seasonal triggers of asthma exacerbations, like viruses and allergens, and leads to a reduction in airway pathology elicited by these agents. Over time, clinical asthma remission might thus be achieved in developing children. Among seasonal asthma triggers, respiratory viruses play an outsize role in children, as they can be isolated from 80% of young patients presenting with an asthma exacerbation, regardless of the perceived exposure that precipitated the exacerbation (11). Although many viral illnesses exhibit an age-tropism in general, little is known about how common viral exposures shape the natural history of asthma exacerbations. In this study, we examined the relationship between age and the ability of common viruses to promote asthmatic lung pathology. Our results have implications for the natural history of childhood asthma and the origins of asthma remission.

This study was approved by the Washington University School of Medicine Human Research Protection Office. We extracted deidentified electronic medical record data from children 0–18 y old who visited the Washington University St. Louis Children’s Hospital emergency department (ED) or who were hospitalized with a chief diagnosis of asthma or allergic reaction from March 14, 2016 to March 29, 2021. Data collected included International Classification of Diseases, Ninth Revision and Tenth Revision codes (J45, J45.20, J45.21, J45.22, J45.30, J45.31, J45.32, J45.40, J45.41, J45.42, J45.50, J45.51, J45.52, J45.991, J45.902, J45.9, J45.99, J45.901, J45.909, T78.4, J39.3, J30.89, L50.0, J30.9, J30.2, J30.8, H10.11/H10.12, Y55.5/J1200, 493.x, 995.3/708.0/287.0, and V13.81/995.0), date of encounter, time of encounter, and age of patient at encounter. To examine respiratory viral burden, we obtained weekly tallies of virus identifications by the St. Louis Children’s Hospital Microbiology Laboratory from March 2016 to March 2021. Virologic assays included multiplex PCR testing and rapid Ag tests for influenza and respiratory syncytial virus (RSV). All electronic medical record data fitting the a priori–defined International Classification of Diseases, Ninth Revision and Tenth Revision codes were included in the analysis. Because lockdown measures officially began on March 23, 2020 in our region (12), we defined each preceding 12-mo period before lockdown as a “year” for the purpose of comparison.

C57BL/6J mice (4–8 wk old, 20–25 g) were sourced from The Jackson Laboratory and acclimatized at least 1 wk before use. We employed a range of ages spanning 5 wk to 18 mo for experiments. In some experiments, older mice (>9 mo) were sourced from retired C57BL/6J breeders obtained from the Washington University School of Medicine Microinjection Core/Gnotobiotic Facility. Unless stated, male mice were used for experiments. Mice were housed in a standard 12-h light/dark cycle, with unlimited food and water available. All experiments were approved by the Washington University School of Medicine Animal Care and Use Committees.

Sendai virus (SeV; VR-105; Sendai/52) and influenza A virus (IAV; A/WS/33 H1N1) were obtained from American Type Culture Collection. Viruses were plaque purified and propagated in embryonated chicken eggs as described (13). SeV at various doses or 5 PFU H1N1 IAV was intranasally administered to mice in 30 μl PBS as described (13, 14).

Mice were treated with broad-spectrum antibiotics to deplete and normalize their microbiomes as described (15). Three weeks before SeV infection, water provided to mice was replaced for 2 wk by either control sugar solution (20 g/l grape Kool-Aid) or high-dose VNAM solution (sugar solution plus 0.5 g/l vancomycin, 1 g/l neomycin, 1 g/l ampicillin, and 1g/L metronidazole; Sigma-Aldrich). Then, for the week prior to and 1 wk after SeV infection, mice received a low-dose VNAM solution (25 g/l sugar solution plus 0.35 g/l vancomycin and 0.75 g/L metronidazole; Sigma-Aldrich) or control sugar solution. Stool samples were obtained prior to VNAM treatment, on the day of virus infection, and on the day VNAM treatment ended. To confirm microbiome depletion, stool DNA was extracted from fresh samples using a commercial kit (Qiagen), and DNA content per milligram stool was determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific).

Stool DNA obtained from juvenile and mature mice was extracted using a commercial kit (Qiagen). Normalized fecal DNA obtained from juvenile and mature mice was subjected to V4-16S rRNA PCR as previously described (16) using custom Illumina adapted primers. Briefly, these primers were designed to support combinatorial based demultiplexing of samples based on their position within a 96-well plate. The forward primer contains a 5′ Illumina-specific i7 sequence, followed by a plate-specific barcode (shown in italics), a primer pad site, a linker sequence, a row-specific sequence (underlined) (17), and then the V4-16S 515F sequence (18) on the 3′ end: 5′-CAAGCAGAAGACGGCATACGAGAT-XXXXXXXX-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCT-AC-XXXXXXXX-GTGYCAGCMGCCGCGGTAA-3′.

The reverse primer contains a 5′ Illumina i5 sequence, followed by a primer pad site, a linker sequence, a column-specific sequence (underlined), and then a 3′ V4-16S 806R sequence (19): 5′-AATGATACGGCGACCACCGAGATCTACAC-TCTTTCCCTACACGACGCTCTTCCGATCT-GT-XXXXXXXX-GGACTACNVGGGTWTCTAAT-3′.

Following PCR, V4-16S amplicons were quantitated, pooled in an equimolar ratio, and then purified (Ampure XP beads; Beckman Coulter) before sequencing on an Illumina MiSeq instrument using 2 × 250-bp chemistry. Following demultiplexing, reads were trimmed, and then amplicon sequence variants were generated using the large-data adapted pipeline for DADA2 version 1.10.1 in R (20). Sequences were classified to the highest taxonomic resolution possible using the RDPClassifier and a minimum bootstrap support of 80% (21). Analysis of amplicon sequence variant abundance was performed in R version 3.5.3 using phyloseq version 1.26.1 (22). Bray-Curtis distances were used to perform principal coordinates analysis on mouse microbiota over the treatment interval.

Mice were intranasally treated with clodronate liposomes like prior reports (23). Briefly, isoflurane-anesthetized mice were intranasally administered 40 μl high-potency clodronate liposomes (20 mg/ml) or control liposomes (FormuMax) 4 d and 2 d prior to SeV infection.

RNA extraction and quantitative PCR (qPCR) were done according to standard protocols (13). At indicated times postinfection, lungs were manually excised and immersed in RNAlater solution (Applied Biosystems). Total RNA was isolated from lung homogenates using RNeasy Mini Kits (Qiagen). Reverse transcription was carried out using the Applied Biosystems High Capacity cDNA Reverse Transcription Kit. Real-time qPCR was performed on the ABI 7500 Fast Real-Time PCR system. We employed commercial primers (Integrated DNA Technologies) for qPCR analysis of Muc5ac, Tbp, Il13, Mmp12, Alox12e, Chi3l3, Retnla, Arg1, Il33, and Trem2. SeV qPCR analysis was conducted as described (13).

Bulk RNA sequencing (RNAseq) was performed at the Washington University Genome Technology Access Center using standard protocols. RNA from whole lung homogenates were converted to cDNA libraries for sequencing using the RiboZero Kit (Sigma-Aldrich). To obtain bronchoalveolar lavage (BAL) cell RNA, BALs from n = 9 mice were combined into 3 pools for each age group, and the cells were pelleted by centrifugation at 3000 × g for 5 min. Cell pellets were resuspended in RLT buffer, and RNA was extracted using the RNeasy Kit (Qiagen). BAL cell RNA libraries were generated using the SMARTer Kit (Clontech Laboratories). Sequencing was carried out on a NovaSeq6000 instrument (Illumina). RNAseq reads were aligned to the Ensembl release 76 assembly with STAR version 2.0.4b. Gene counts were derived from the number of uniquely aligned unambiguous reads by Subread:featureCount version 1.4.5. Transcript counts were produced by Sailfish version 0.6.3. Sequencing performance was assessed for total number of aligned reads, total number of uniquely aligned reads, genes and transcripts detected, ribosomal fraction known junction saturation, and read distribution over known gene models with RSeQC version 2.3. All gene-level and transcript counts were then imported into the R/Bioconductor package EdgeR, and trimmed mean of M values normalization size factors were calculated to adjust for sample differences in library size. Genes or transcripts not expressed in any sample were excluded from further analysis. The trimmed mean of M values size factors and the matrix of counts were imported into R/Bioconductor package Limma (24), and weighted likelihoods based on the observed mean-variance relationship of every gene/transcript were then calculated for all samples with the Voom function (25). Performance of the samples was assessed with a Spearman correlation matrix and multidimensional scaling plots. Gene/transcript performance was assessed with plots of residual SD of every gene to their average log-count with a fitted trend line of the residuals. Generalized linear models with dispersion estimates were then created to test for gene/transcript-level differential expression. Differentially expressed genes and transcripts were then filtered for false discovery rate–adjusted p values ≤0.05. To examine age-associated differential gene expression in BAL samples, log₂-normalized Voom-adjusted counts per million (CPM) data were analyzed for statistical significance by the Kruskal–Wallis one-way ANOVA by ranks.

Pooled BAL cell samples from healthy juvenile (6 wk old, n = 10) and mature (12 mo old, n = 9) mice were subjected to single-cell RNAseq (scRNAseq) at the Washington University Genome Technology Access Center using standard protocols. Single-cell barcoded cDNA libraries were generated using a Chromium microfluidics platform (10X Genomics) and the Chromium Single Cell 3′ GEM, Library and Gel Bead Kit v3 (10X Genomics). Sequencing was carried out on a NovaSeq 6000 instrument (Illumina). Sequencing data were processed using Cell Ranger version 3.1, starting with demultiplexed FASTQ files and using GRCm38-2.0.1 as the reference genome. Processed scRNAseq data were analyzed using R version 3.6.2. and the SEURAT v3 package as described (26) (available at https://satijalab.org/seurat/). We used previously published marker genes to type cell clusters identified on UMAP analysis (27, 28).

Lungs were fixed by inflation with 10% buffered formalin under 20 cmH₂O of pressure, paraffin-embedded, and stained with periodic acid–Schiff (PAS) or subjected to immunofluorescence staining using standard techniques (13). PAS-stained slides were scanned using a Hamamatsu Photonics NanoZoomer at the Alafi Neuroimaging Laboratory at Washington University, and PAS-positive pixels were counted using Aperio ScanScope Software as described (29). For immunofluorescence experiments, we used Abs directed against SeV (Abcam; catalog number ab33988; 1:500 dilution), followed by Alexa Fluor–conjugated secondary Abs and DAPI counterstain (Invitrogen). Stained slides were visualized using an Olympus BX51 microscope and a Retiga 2000R camera system (QImaging). We used a double-blind design to quantitate SeV infection in lung histological sections. After masking sample identification, 5 microscopic fields were randomly selected at ×10 magnification per lung section. SeV⁺ and total cell counts were quantified using ImageJ software.

BAL analysis was performed as described (13). Mice were euthanized, and their tracheas were cannulated with a 20-gauge flexible catheter, 1 to 2 mm distal to the larynx. The lungs were then slowly lavaged in two passes with a total of 1 ml sterile PBS. For BAL cell counts, 50 μl BAL suspension was adsorbed to frosted glass slides using a Cytospin 4 centrifuge (Thermo Fisher Scientific) and treated with Diff-Kwik stain (Thermo Fisher Scientific). Slides were then digitally scanned at ×20 original magnification using a NanoZoomer HT imaging system (Hamamatsu Photonics). To quantify cell number, scanned slides were analyzed via ImageScope software (Leica) using the “Nuclear v9” algorithm, which counts nuclei in histologically stained samples. For BAL cell differential, sample identifications were blinded, and 5 random ×40 original magnification fields were captured as TIFF files using ImageScope software. Cells in each field were then scored as either lymphocytes, macrophages, or neutrophils by a separate investigator also blinded to the sample identification. A minimum of 50 cells were counted per biological specimen. BAL protein concentration was quantified using Bradford reagent (Sigma-Aldrich) according to standard protocols. BAL cytokine levels were analyzed at the Washington University Immunomonitoring Core Facility on a FLEXMAP 3D instrument (Luminex) and using the Mouse Cytokine and Chemokine Panel 1A (Affymetrix), according to the manufacturer’s instructions.

Airway resistance (R_I) was measured using a Buxco Elan RC mouse volume-controlled ventilator as described (13). Briefly, mice were anesthetized by i.p. injection of ketamine/xylazine mixture, and then, a tracheostomy was performed using a 22-gauge flexible catheter. Mice were then ventilated and given progressively doubled concentrations of methacholine through an in-line nebulizer. Inspiratory resistance measurements were averaged over 3-min intervals after each dose. Mice were euthanized after lung mechanics measurements.

Single lung cell suspensions were prepared using Liberase TM and DNase in RPMI 1640 (Roche) essentially as described (13). Cells were stained using panels of metal-conjugated Abs or fluorophore-conjugated Abs for mass and flow cytometry, respectively (Supplemental Table I). Mass cytometry was carried out at the Washington University Immunomonitoring Laboratory using a CyTOF2/Helios instrument (Fluidigm) and analyzed using CytoBank software as described (30). Flow cytometry was performed at the Washington University Siteman Flow Cytometry Core using an Attune NxT instrument (Thermo Fisher Scientific) and was analyzed using FlowJo software (BD Biosciences). See Supplemental Fig. 3A and 3B for representative gating strategies for mass and flow cytometry, respectively. For flow cytometry, 50,000 events/sample were targeted.

DQ-OVA assay was conducted as in prior reports (31). Isoflurane-anesthetized mice were intranasally treated with 30 μg DQ-OVA (Thermo Fisher Scientific) in 30 μl sterile PBS. Two hours later, mice were euthanized, and flow cytometry of whole lung cell suspensions was conducted as above using an Ab mixture (Supplemental Table I). A representative gating strategy is presented in Supplemental Fig. 3C.

For statistical testing of flow cytometric, qPCR, and BAL cytokine data, Student t tests were performed on log-normalized data as recommended (32) using Microsoft Excel. For survival curve analysis, the log-rank test was used. UMAP analysis was performed using the SEURAT package in R. Visualization of t-distributed stochastic neighbor embedding (viSNE) analysis of mass cytometry data was performed using CytoBank software. Principal component analysis (PCA) of BAL cytokines and mass cytometry data was performed in Microsoft Excel using a freeware add-in (available at http://wak2.web.rice.edu/bio/Kamakura_Analytic_Tools.html). Functional enrichment analysis was conducted using DAVID v6.8 software (33, 34) (available at https://david.ncifcrf.gov/home.jsp). For correlating cytokine levels to airway mechanics, we calculated the Pearson correlation coefficient to rank-ordered values and converted this to a p value using a Student two-tailed t test. All morphometric analyses were double-blinded.

This study did not generate new reagents. Bulk and scRNAseq data (Figs. 9 and 10) are available through the National Institutes of Health Gene Expression Omnibus (accession number GSE185260, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi). R code for analyzing scRNAseq data in SEURAT can be found at the following link: https://www.dropbox.com/s/ovikd97763kn1y5/Hazan%20et%20al_ScRNA%20analysis%20R%20code.pdf?dl=0.

On a population level, rates of asthma exacerbations decline as children mature (8). To examine the role of viruses in this phenomenon, we analyzed how societal measures imposed during the ongoing coronavirus disease 2019 (COVID-19) pandemic affected the age-gradient in pediatric asthma exacerbations (Fig. 1). In response to COVID-19, states enacted lockdown measures beginning in the spring of 2020, including face mask use, social distancing, school closures, and the suspension of public venues. Although the intent of these lockdown measures was to limit the spread of severe acute respiratory syndrome (SARS) coronavirus 2, they had the side effect of suppressing common respiratory viruses (35). Recent studies have leveraged COVID-19 lockdown as a “natural experiment” to infer the role of viral infection in various diseases (36–38), but have not previously considered disease timing. We retrospectively analyzed ED visits for asthma at our institution, comparing the 12-mo period following the introduction of lockdown measures to 4 y prior to lockdown (see Materials and Methods). COVID-19 lockdown was highly successful in suppressing respiratory virus positivity in our area, reducing yearly rates of positive tests by 70.2% (p < 0.001, Mann–Whitney U test). In the 4 y prior to lockdown, ED visits for asthma exhibited a stereotypical age distribution (8), in which exacerbation frequency was maximal in 3- to 4-y-old children and then linearly declined with older age (Fig 1A). Strikingly, in the year after lockdown, the age distribution of ED asthma visits was uniform (Fig. 1A). Additionally, the effectiveness of lockdown in lowering asthma exacerbations was inversely proportional to age (Fig. 1B). This contrasts with ED visits for allergic reactions, which showed a similar age distribution before and after lockdown (Fig. 1C). Additionally, the ability of lockdown to suppress ED presentations was independent of age (Fig. 1D). Because COVID-19 lockdown was a viral mitigation intervention, these data suggest a causal role for common respiratory viruses in driving the traditional age-gradient in pediatric asthma exacerbations. Further, they suggest that the ability of common respiratory viruses to provoke clinically apparent asthma exacerbations might wane as children mature.

To investigate directly how age impacts lung pathology relevant to asthma exacerbations, we infected mice across a span of ages with SeV (Fig. 2A). SeV is a parainfluenza virus used extensively in young rodents for modeling postinfectious lung remodeling relevant to asthma, including chronic mucous cell metaplasia, type 2 inflammation, and airway hyperresponsiveness (13, 39–47). A key advantage of SeV for our study is that the lung pathology produced by this virus persists long-term as opposed to other agents (44). Although juvenile (6-wk-old) mice infected with SeV developed the expected chronic airway inflammation and mucous cell metaplasia by 49 d postinfection (DPI), these phenotypes were greatly attenuated in older mice (Fig. 2B, 2C). Across a range of SeV doses, induction of inflammatory mucus as measured by Muc5ac gene expression was significantly lower in mature mice compared with juvenile mice, and this age effect was sex-independent (Fig. 2D, 2I). Mature age also suppressed the induction of multiple type 2 inflammatory signatures by SeV, including IL-13, IL-33, and M2 macrophage gene expression, factors all previously identified as critical for postviral pathology in juveniles (40) (Fig. 2F). To map an age window for postviral lung pathology, we infected mice spanning from 4 wk to 16 mo old with SeV. At 4 wk old, mouse lungs are in the alveolarization stage of development, which in humans occurs primarily between 2 and 8 y of age, whereas a 16-mo-old mouse is akin to late middle age in humans (48–50). Postviral induction of Muc5ac diminished in mice >3 mo, an age threshold analogous to early adulthood in humans (49) (Fig. 2G). Mature mice also failed to develop airway hyperresponsiveness to methacholine by SeV 49 DPI, as normally seen in juveniles (Fig. 2H). To test the generalizability of these findings, we infected mice with a strain of H1N1 IAV previously shown to provoke chronic lung pathology in young mice (Fig. 3A) (14). Although IAV infection produced similar levels of acute viral RNA expression in juvenile and mature mice, production of chronic lung pathology occurred selectively in juvenile mice as assessed by histologic severity, airway hyperresponsiveness, and mucous cell metaplasia (Fig. 3B–F).

Mature C57BL/6J mice differ physically from juveniles in several respects, including divergent microbiome composition (51, 52). Because the microbiome can regulate inflammatory responses (15), we examined its role in SeV-induced chronic lung pathology. We added a broad-spectrum antibiotic mixture (VNAM) to the drinking water of juvenile and mature mice (Fig. 4A). VNAM treatment strongly suppressed stool DNA content and microbial diversity at the time of SeV infection (0 DPI, Fig. 4B, 4C). Although juvenile and mature mice could be differentiated from one another by stool 16S DNA sequencing at baseline, VNAM treatment eliminated clear-cut age differences during and after SeV infection (Fig. 4D). SeV RNA expression was not affected by VNAM during the acute viral illness in either age group (Fig. 4E). At SeV 49 DPI, VNAM treatment also did not alter age-related differences in airway hyperresponsiveness, Muc5ac, IL33, and the M2 marker Trem2 (Fig. 4F, 4G).

To summarize, we found that the ability of viral triggers to induce asthmatic traits was robust in juvenile mice but waned with physical maturity. The age sensitivity of virus-induced lung pathology could not be explained by animal adiposity or by gut microbiome composition.

To identify points of divergence between juvenile and adult mice across the 7-wk course of the SeV model, we first examined indices of the initial viral bronchiolitis (Fig. 5). During acute SeV infection weight loss, viral RNA expression, BAL cell count, and cell differential were statistically similar in juvenile and mature mice and followed similar kinetics (Fig. 5A–F). In contrast to the above similarities, airway mechanics measurements at SeV 5 DPI (the peak of acute illness) revealed a greater methacholine sensitivity in juvenile mice than mature mice despite similar levels of viral RNA expression (Fig. 5G). This ventilatory mechanics difference diminished at subsequent time points, and by SeV 12 DPI, airway hyperresponsiveness had normalized in both age groups (Fig. 5H, 5I). In the context of acute bronchiolitis, we wondered if increased methacholine sensitivity might reflect the distribution of viral infection in small airways. We therefore counted SeV-infected airway cells in acutely infected mice and found that the percentage of SeV⁺ cells per airway was greater in juvenile mice than mature mice at 5–8 DPI (Fig. 5J, 5K). To summarize, although general indices of respiratory illness and viral load were similar across age groups, SeV-infected juvenile mice tended to have a wider distribution of virus-infected airway cells, which was associated with transient airway mechanical differences between the age groups. A key insight was that age-specific differences emerged during early SeV infection.

To investigate age-specific inflammatory patterns in response to SeV, we profiled BAL cytokines in juvenile and mature mice at various time points postinfection (Fig. 6). We also correlated BAL cytokines to invasively measured airway mechanics at SeV 49 DPI. Viewed as a heat map, SeV promoted a generally biphasic pattern of cytokine expression with a major peak occurring during the acute viral illness and a lesser peak associated with lung remodeling at 49 DPI (Fig. 6A). PCA resolved the cytokine profiles into three loose clusters (Fig. 6B). Cluster I was enriched for cytokines differentially expressed in juvenile mice at SeV 49 DPI (red lettering), cluster II was enriched for cytokines preferentially expressed in mature mice at the same time point (blue lettering), and, in cluster III, juvenile-overexpressing and mature-overexpressing cytokines were equally represented (Fig. 6B). Of the 21 BAL cytokines exhibiting age-dependent expression at SeV 49 DPI, 17 of these (underlined symbols) significantly correlated with R_I, a measure of obstructive airway disease (p < 0.05) (see Materials and Methods and (Fig. 6B). Strikingly, cytokines overexpressed in juvenile mice at 49 DPI positively correlated with R_I, whereas cytokines overexpressed in mature mice negatively correlated with R_I (Fig. 6C). Thus, age promoted specific patterns of lung cytokines after SeV infection that strictly reflected outcomes in terms of airway physiology. We noted that in many cases, the age-specific cytokine expression seen at SeV 49 DPI was preceded by a similar trend during the acute phase of infection (Fig. 6D–H). Also of note, juvenile-specific cytokine expression at SeV 49 DPI was significantly enriched for type 2 cytokines (3.2-fold enrichment; p = 0.027 Fisher exact test), including IL-4, -5, and -13 (Fig. 6B, 6G, 6H). As with airway mechanics (Fig. 5G), age-specific type 2 cytokine responses emerged early, if transiently, during acute SeV respiratory illness.

Type 2 inflammation is a trait common to virus- and allergen-triggered asthma exacerbations. Therefore, we asked to what extent age impacts airway pathology induced by Alternaria alternata, a common aeroallergen in severe asthma (Fig. 7A) (53). Unlike viral infection, Alternaria extract induced extensive peribronchial airway inflammation, mucous cell metaplasia, and Muc5ac expression in both juvenile and mature mice (Fig. 7B–D). Additionally, gene expression of the type 2 cytokines IL-33, IL-5, and IL-13 was statistically similar in juvenile and mature mice (Fig. 7C). In contrast, Alternaria induced less eosinophil infiltration, lung consolidation, and airway hyperresponsiveness in adults compared with juveniles, although adults did develop pathology (Fig. 7E–G). We concluded that age affected allergic triggers of asthmatic pathology differently than viral triggers: whereas postviral pathology was categorically selective for the juvenile age bracket, allergic triggers were age sensitive in only a subset of type 2 inflammatory markers and could dependably induce asthmatic traits across age groups. Such a difference is clinically plausible because asthmatic children with coincident atopy are much less likely to go into clinical remission as they mature (54, 55).

To clarify immunological differences between juvenile and mature mice, we used mass cytometry to examine whole lung leukocyte composition in the presence or absence of SeV infection (Fig. 8, Supplemental Fig. 1). PCA revealed that juvenile and mature lungs were most divergent in terms of their leukocyte composition prior to SeV infection (0 DPI, Fig. 8A). At 0 DPI, mature mouse lungs exhibited greater frequencies of PD-1⁺ and CD44⁺ T cells compared with juvenile lungs (Supplemental Fig. 1A, 1B). Leukocyte infiltration kinetics into SeV-infected lungs could be grouped into three general patterns: a “U” or “S”-shaped pattern of infiltration (cluster I), a peak at SeV 5 DPI (cluster II), or a peak at 12 DPI (cluster III) (Fig. 8B). Although these trafficking patterns were similar across age groups, the overall prevalence of specific cell types differed with age (Fig. 8C, Supplemental Fig. 1C–J). B-cells and neutrophils were more prevalent in mature mice over discrete intervals during SeV infection, whereas eosinophils accumulated progressively only in juvenile mice (Fig. 8D–F). Alveolar macrophages (AMs) stood out among the major cell types in exhibiting age differences both at baseline and at SeV 49 DPI (Fig. 8G). We therefore analyzed AMs in further detail.

Using viSNE analysis of mass cytometry data (56), we noticed that AMs from juvenile and mature mice congregated in distinct clusters that were strongly influenced by differential surface expression of MHC class II (MHC-II) protein, both at baseline and at SeV 49 DPI (Fig. 9A, 9B). In the literature, mouse AMs are described as expressing low levels of MHC-II in the basal state (57), but this was based on observations from juvenile animals. We found that AMs developed MHC-II^Hi status progressively with age, achieving > 60% prevalence by 7 mo old in the absence of infection or other clear activating stimulus (Fig. 9C). Analysis of permeabilized cells by mass cytometry indicated that MHC-II upregulation was a selective feature of AM aging, although some upregulation in interstitial macrophages was noted as well (Fig. 9D). Previous studies suggest that adenoviral infection can induce MHC-II expression in juvenile AMs (58). Although we found that SeV infection can similarly induce MHC-II expression on AMs in juvenile mice, the frequency of MHC-II^hi AMs in juvenile mice was significantly lower than those seen in mature mice, even at baseline (Fig. 9E). Interestingly, MHC-II expression did not correlate with increased Ag-processing activity as measured by the DQ-OVA assay (Fig. 9F). This made us wonder if the MHC-II expression we observed reflected a broader reprogramming of AM function with age rather than a retasking of these cells for Ag presentation.

To examine the transcriptional landscape of AM aging, we performed bulk RNAseq analysis of BAL samples from naive mice, which are largely composed of AMs. We observed age-dependent expression of H2-aa and H2-eb1, genes that encode the 1A/1E MHC-II proteins detected by flow cytometry (Fig. 9G). Also upregulated was Ciita, a master regulator of MHC-II gene expression (59) (Fig. 9G). To put these findings in context, 1024 genes (2.5% of identified annotations) demonstrated age-dependent expression in BAL samples from naive mice using a cutoff of p < 0.05, with a slight majority (55%) demonstrating increased expression in the juvenile (<3 mo old) compared with the mature (>7 mo old) age range (Fig 9H). However, from an informatic perspective, all Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways found to be enriched in the total set of 1024 genes were entirely driven by the subset for which expression is higher with maturity (Fig. 9I, blue-shaded terms). These KEGG terms included “asthma” and multiple pathways connected to Ag presentation, all of which include MHC-II genes in their functional annotation definitions (Fig. 9I). Thus, a major informatic feature of age-associated gene expression in BAL cells is the acquisition of MHC-II gene expression with aging. Finally, we asked whether MHC-II upregulation in AMs was cell autonomous or driven by environmental context. Interestingly, in vitro culture of AMs from mature donors led to a rapid downregulation of H2-aa expression, but expression levels were not affected in AMs from juvenile mice (Fig. 9J). Similarly, in vitro culture of AMs downregulated Pla2g2d expression, a gene that exhibited age-dependent expression in our BAL dataset (Fig. 9K). These data suggest that RNA signatures of AM aging are not cell autonomous and may be driven by the local microenvironment.

To further investigate whether MHC-II upregulation with aging was a specific feature of AMs, we performed scRNAseq on pooled BAL samples from naive juvenile (6-wk-old) and mature (12-mo-old) mice (Fig. 10). UMAP analysis revealed that juvenile and mature mice harbor similar populations of cell types that tightly cocluster (Fig. 10A). There were subtle differences in the cell differential, with mature BALs containing greater amounts of lymphocytes (Fig. 10B). We noted a higher cellular yield and a higher prevalence of mitotically active (KI-67⁺) AMs in BALs from juvenile mice compared with mature mice, consistent with prior reports (Fig. 10B) (60). The age-dependent upregulation of H2-aa seen by bulk RNAseq was confirmed by single-cell analysis and mapped broadly across the entire AM cluster in mature mice (Fig. 10C). Examining global AM gene expression by scRNAseq revealed a distinct group of genes upregulated in mature compared with juvenile AMs that were predominantly MHC-II pathway genes (Fig. 10D, blue circles). Age-regulated expression of these genes was specific to AMs and was not observed in the other cell types detected in BALs (Fig. 10E, arrow). Thus, scRNAseq confirmed differential MHC-II gene expression to be a prominent and selective informatic feature of AM aging in healthy mice. We should note that a recent report compared single-cell gene expression in 4–6-mo-old versus 18–24-mo-old AMs sorted by flow cytometry but did not observe differences in MHC-II gene expression aside from CD74 (also found in our data set) (61). However, this analysis left out juvenile mice in which the major differences in MHC-II gene expression occur and employed cells exposed to the stress of tissue dissociation and cell sorting.

We considered the possibility that AMs from mature animals might originate from bone marrow–derived monocytes as opposed to the embryonic yolk sac, and this alternate origin might explain differences in MHC-II expression. However, AMs from healthy mature mice exclusively expressed markers connoting embryonic but not bone marrow origin (Fig. 10F) (23). Altogether, our data demonstrate substantial AM programming with aging in terms of the number of these cells in the lung, global gene expression, Ag-processing potential, and MHC-II class gene expression. Importantly, gene expression changes to AMs plateau by 7 mo of age, indicating the phenomenon is one of organism maturation rather than senescence or old age.

To examine the significance of mature AMs in postviral lung pathology, we used clodronate liposomes to selectively deplete these cells prior to SeV challenge (23) (Fig. 11A). Intranasal delivery of clodronate liposomes substantially reduced AM abundance and normalized differences in AM number between juvenile and mature mice (Fig. 11B, 11C). In contrast, control liposomes did not affect age differences in AM number, although they had a slightly depressing effect on AM abundance in both age groups (Fig. 11C). AM depletion in mature mice prior to SeV infection led to a marked increase in postviral lung disease that approached that of juvenile mice in terms of lung pathology, methacholine responsiveness, and mucous cell metaplasia (Fig. 11D–F). In contrast, AM depletion had only minimal effects on postviral pathology in juvenile mice (Fig. 11D–F). The effect of AM depletion could not be explained by a differential increase in acute SeV severity in mature mice, because survival, weight loss, and SeV RNA expression remained age-independent in clodronate-treated mice (Supplemental Fig. 2A–C). Additionally, clodronate treatment did not induce lung pathology in isolation (Supplemental Fig. 2D). These results indicate that AMs damp postviral lung pathology specifically in mature mice, coincident with a transcriptional reprogramming of these cells with aging.

In this study, we detail an age-dependent host response to respiratory viruses that dictates whether an acute viral infection will result in lung pathology relevant to asthma. The age threshold for the decline in postviral pathology in mice occurred at the juvenile-to-adult age transition. This phenomenon mirrors the decline in asthma exacerbation rates seen in childhood asthma remission. Systematic comparison of SeV-infected juvenile and mature mice revealed several points of divergence, all of which could contribute mechanistically to the age effect and are the subject of ongoing research. Mature age was marked by a lower propensity toward type 2 inflammation after viral challenge, a change in the distribution of infection within the bronchial tree, and alterations in both AM number and gene expression. Finally, our analysis of COVID-19-era clinical data strongly supports a connection between the aging out of asthma exacerbations and viral responses in children.

To our knowledge, our report represents the first mechanistic evidence connecting the host response to common respiratory viruses with childhood asthma remission, a phenomenon in which asthma control and asthma exacerbation rates gradually improve as some patients mature (55). When asthma arises in childhood, it usually presents as an acute wheezing illness, often triggered by common respiratory viruses like RSV, rhinovirus C, influenza, and parainfluenza (62). Rates of asthma remission in children tend to plateau ∼10–15 y after disease onset, but the likelihood that remission will ever occur correlates inversely with age of asthma debut (63). This epidemiological relationship suggests that the pathophysiology usually associated with early-life asthma symptoms—viral infection—may also be connected to the process of remission. What our data add is a characterization of how age affects postviral pathology using controlled mouse models, as well as an affirmation of clinical relevance through a novel analysis of asthma clinical activity during the COVID-19 pandemic.

In contrast to viruses, atopy is strong negative predictor of asthma remission and the resolution of asthmatic airway remodeling (7, 55, 63–65). As such, it is interesting that Alternaria-induced asthma phenotypes were less age-sensitive than those induced by SeV. Why developmental aging should impact virus–host responses differently than allergen-host responses is unclear but deserves further investigation. It is tempting to speculate that the dichotomy reflects an evolutionary need to limit lung injury in the face of yearly cycles of respiratory viral infection that, due to antigenic shifts, cannot entirely be forestalled by acquired immunity or immunologic tolerance.

Our data suggest an age-specific role for AMs in postviral lung pathology. AMs represent the major immune cell in the lung microenvironment at baseline and are a major source of IFN-α/β production early in RNA respiratory viral infections (66). Studies show that AM depletion modifies the severity of various acute viral infections, in some cases exacerbating illness (influenza, RSV, vaccinia, and SeV as shown in this study) (67–69) and in other cases reducing acute illness (mouse hepatitis virus 1 and mouse-adapted SARS coronavirus 1) (70, 71). Thus, AMs would seem a logical cell to influence acute host responses to viruses that later modulate chronic asthma phenotypes. Moreover, AMs were suggested to have age-specific functions in other contexts. In one report, depletion of AMs ameliorated mouse-adapted SARS pneumonia, but only in older mice (71). Future research will be needed to define specific aspects of AM reprogramming that are critical for mitigating postviral lung pathology.

Using unsupervised mass cytometric and transcriptomic approaches, we identified upregulation of MHC-II pathway genes as a major informatic signature of AM aging. The functional significance of this change to AMs is at present unclear and will be the subject of future research. Our observation is compatible with prior reports of MHC-II upregulation in AMs from aged mice (>22 mo old), particularly on CD11b⁺ AMs (72). Because MHC-II^Hi AMs displayed more efficient phagocytosis of Mycobacterium tuberculosis bacteria, the authors suggested senile AM changes could help explain why elderly patients are more susceptible to M. tuberculosis (72). However, our study adds a key observation: MHC-II upregulation on AMs is saturating by 7 mo of age in healthy mice and therefore correlates with the completion of adult maturation, not old age. The distinction is significant because published studies involving aging and immunologic responses usually report a comparison of extremes between “old” (generally >18 mo old) and “young” animals (variably defined as 4 wk to 4 mo old). However, aging occurs continuously across the lifespan of an organism, and there may be significant milestones relevant to disease between the extremes of age, milestones that could be missed or misinterpreted by a dichotomous approach to this biological variable. The phenomenon of asthma remission makes clear that clinically important changes to host responses occur across the childhood to adulthood aging landscape, and we show some of these can be recapitulated in mouse models. Research is needed to more precisely map how host responses evolve across the entire continuum of aging.

It is worth mentioning some limitations to our methodological approach. Although mice are frequently used as a model organism to study inflammatory airway pathology, how far results obtained from mouse models can be generalized to human asthma is controversial (73). The scope of our data is most applicable to type 2^Hi asthmatic endotypes found in children and does not address other mechanistically distinct forms of asthma found in adults, like Th17-predominant and obesity-associated disease. Our study employed predominantly male mice, and further analysis of postviral responses in female mice would be valuable. SeV-induced asthmatic pathology occurs independently of T cell function (42), and so, caution should be used in extrapolating our results to acquired immune readouts like IgE production. Although our study focused primarily on viral triggers because of their marked age sensitivity, our data do not exclude a role for allergen-induced pathways in aspects of asthma remission. Although VNAM treatment did not impact the age sensitivity of postviral pathology, our data do not fully exclude a role for the microbiome in mediating age-dependent effects in asthma, particularly with regards to atopic triggers. Experiments in germ-free mice would constitute logical next steps. Our data examine the pathologic effects of discrete exposures to viruses or allergens as occurs during asthma exacerbations, but the clinical arc of human asthma involves recurrent exposures to such triggers over time. It is worth pointing out the C57BL6/J background is inherently primed for type 2 inflammatory responses (74, 75), so viral infection represents a second “hit” in this strain of mice. Nevertheless, further experiments involving recurrent viral challenge at different stages of development and correlation with biomarker studies in human asthma remission cohorts represent possible next steps. All considered, however, our approach supplies a needed mouse model that can recapitulate aspects of childhood asthma remission, specifically the observed reduction in asthma exacerbations with older age. We anticipate it will facilitate future investigation of this poorly understood phenomenon.

In summary, we report an attenuation of postviral chronic lung pathology and asthmatic phenotypes with maturation in mice. AMs have a role in mediating these age-related inflammatory differences. Our findings have implications for the natural history of asthma in children.

We thank Robyn Haspel and Steven Brody for their editorial input.

The authors have no financial conflicts of interest.