Sputum macrophage diversity and activation in asthma: Role of severity and inflammatory phenotype

Macrophages control innate and acquired immunity, but their role in severe asthma remains ill‐defined. We investigated gene signatures of macrophage subtypes in the sputum of 104 asthmatics and 16 healthy volunteers from the U‐BIOPRED cohort.


| INTRODUC TI ON
Asthma is a heterogeneous disease driven by diverse inflammatory mechanisms, which presents in various clinical forms. Severe asthma is a clinically defined subset of asthma that remains partly or totally unresponsive to known asthma treatments. 1 Both recruited inflammatory and resident airway structural cells participate in the inflammatory and remodelling processes associated with asthma phenotypes. 2 Recently, type-2-related and non-type-2-related molecular phenotypes or transcriptomic-associated clusters (TACs) have been described. 3 TACs were derived from clustering of differentially expressed genes identified in the sputum of asthmatic patients and provide mechanistic pathways beyond eosinophilic subtyping.
Macrophages within the lung play a role in the control of both innate and acquired immunity, tissue homeostasis, angiogenesis and metabolism. 4 Hallmarks of macrophages are their plasticity and diversity.
Two populations of macrophages coexist in the lung: (a) tissue-resident macrophages (TR-Mφ) that arise during embryogenesis are associated with proliferation and defined by Krebs cycle, amino acid and fatty acid (FA) metabolic processes and (b) recruited macrophages that originate postnatally from circulating monocytes and are involved in immune signalling and inflammation and defined by glycolysis and arginine metabolism. 5 Tissue damage or infection results in TR-Mφ activation with production of inflammatory mediators, together with monocyte recruitment. 4,6 Cytokines drive macrophage polarization (diversity) with two classical subtypes according to their functional properties: classically activated M1 driven by IFN-γ, LPS, GM-CSF or TNF-α exposure and alternatively activated M2 macrophages induced by IL-4, IL-10, IL-13, glucocorticoids or TLR ligands. 4 While M1 macrophages have pro-inflammatory properties and are associated with Th1 responses, M2 macrophages control inflammation through the high expression of IL-10 and are involved in Th2-mediated immunity. 4,7 The role of macrophages in asthma pathogenesis is unclear because of the limited methods available to characterize them. 6 Current data suggest that the presence of M1 macrophages is associated with disease progression and airway remodelling, M2 macrophages are associated with type-2 asthma 4,8 and the role of TR-Mφ remains unknown. However, human monocyte-derived macrophages stimulated with 28 different stimuli produced 49 distinct transcriptomic modules highlighting the spectrum of macrophage diversity. 9 We hypothesized that analysis of these 49 macrophage transcriptomic signatures will indicate activation of distinct macrophage subsets in severe asthma. We assessed this using gene set variation analysis (GSVA) using sputum cells from the U-BIOPRED (Unbiased BIOmarkers in PREDiction of respiratory disease outcomes) asthma cohort 10 with validation in the ADEPT (Airway Disease Endotyping for Personalized Therapeutics) cohort. 11 This is the first comprehensive analysis of airway macrophage subtypes in asthma.

K E Y W O R D S
asthma, gene set variation analysis, macrophage subtypes, sputum, tissue-resident

G R A P H I C A L A B S T R A C T
This study investigates gene signature of macrophage phenotypes in sputum of asthmatic and healthy controls. Several specific macrophage subsets identified by gene signatures are highly activated in severe granulocytic asthma and involved distinct inflammatory pathways (eg tumour necrosis factor [TNF] and Toll-like receptor [TLR] signalling pathways). Macrophage activation and numbers are reduced in severe granulocytic asthma highlighting defective innate immunity. Abbreviations: MMA, mild/moderate asthmatics; SAns, severe asthmatics nonsmokers; SAsm, severe asthmatics smokers; TLR, Toll-like receptor; TNF, tumor necrosis factor TA B L E 1 Characteristics of study participants from U-BIOPRED cohort providing sputum transcriptomics

| Study population and design
We studied data from participants of the U-BIOPRED adult cohort 10 who underwent sputum cell transcriptomic analysis. 3 This included 104 asthmatics: mild-moderate asthma (n = 20), nonsmoking severe asthmatics (n = 61) and current or ex-smokers with severe asthma (n = 23), and 16 healthy volunteers (Table 1). All participants gave written informed consent. For the asthmatic patients, the analyses were done also according to the presence of airflow limitation defined as FEV1/FVC < 0.70 and FEV1 < 75% of predicted. 3 A description of the analytical plan is provided in Figure S1.

| Microarray analysis of sputum transcriptomes
Sputum induction 12 and transcriptomic analysis has been described previously. 13 In brief, sputum cells were obtained from plugs obtained following inhalation of hypertonic saline. Gene expression profiling was performed on extracted RNA using microarrays. 3 A comparison of sputum gene expression profiles from asthmatics with high (≥1.5%) or low (<1.5%) sputum eosinophilia and the healthy volunteer group identified 508 differentially expressed genes. Batch/technical effects, age, sex and administration of oral corticosteroid were adjusted for as covariates.
Unsupervised hierarchical clustering based on Euclidean distance on these 508 genes identified 3 transcriptomic-associated clusters (TACs): TAC1 was associated with Th2 inflammation, sputum eosinophilia, more severe asthma, high rate oral corticosteroid dependency, frequent exacerbations and severe airflow obstruction, TAC2 with IFN-γ, TNF-α and inflammasome-signalling pathways, sputum neutrophilia, high prevalence of eczema and serum C-reactive protein levels and TAC3 with paucigranulocytic inflammation, metabolic pathways and better preserved lung function. 3 These groups are considered as reflective of different molecular phenotypes of asthma in the U-BIOPRED cohort and were used in our analyses.

| Gene set variation analysis (GSVA)
Gene set variation analysis is a gene set enrichment method that estimates the variation in a pathway activity across a sample population in an unsupervised manner. It is used to detect changes in pathway or gene signature activity over a sample population and indicates differences within populations and between groups of subjects. 13,14 GSVA calculates sample-wise enrichment scores (ES) across the whole data set with a range from −1 to +1.

| Macrophage signatures
Forty-nine distinct modules previously published 9 15 and a TR-Mφ signature adapted from a murine model 16 were also examined (Table S1). Although some of the stimuli were the same, the module genes were often different reflecting altered cellular states resulting from distinct pathways characterizing the modules and highlighting macrophage heterogeneity ( Table 2). Validation of the module signatures was performed using GSE90010 comparing purified MDMs populations against alveolar macrophages (AM) from idiopathic pulmonary fibrosis (IPF) patients (Table S2).

| Pathway analysis
The functions and pathways for each gene signature were assessed using KEGG 2019, WikiPathways 2019 and Reactome 2016 databases within Enrichr 17 (Table 2), and for the pooled activated genes by STRING (www.strin g-db.org).

| Macrophage signatures according asthma severity
In initial analysis to determine whether macrophage modules were selective for macrophages, we demonstrated that 41/49 module signatures were similarly enriched in AM and MDMs from healthy volunteers and IPF patients (Table S2). Two of the signatures were significantly decreased while 6 of the signatures were significantly enriched in AM compared with monocyte-derived macrophages.
The total number and percentage of macrophages in sputum was reduced by ~50% in severe asthma but not in mild-moderate asthma (Table 1). However, not all macrophages subtypes were decreased according to GSVA analysis. Figure 1 (Table 2).
Only module 14, associated with the phago-lysosome fusion, lymphocyte migration and fat cell proliferation, was significantly enriched in severe asthma smokers vs severe asthma nonsmokers (P = .02). In contrast, representative GSVA plots of the enrichment of module 18 (FA) show significant down-regulation in severe asthma compared to mild-moderate asthma and healthy volunteers ( Figure 2C). There were no significant differences in the enrichment of TR-Mφ, M1 or M2 14,15 or in sputum across asthma severity using GSVA ( Figure 2D-F).
These data highlight the plasticity of macrophages in that macrophage subtypes or states distinguished by WGCNA modules provide additional signals compared to validated M1 and M2 signatures. Overall, the data indicate reduced macrophage numbers and activation status in severe asthma compared to mild-moderate asthma and healthy volunteers suggesting an attenuated innate immune response in these patients. However, there is a limited subtype-specific activation of macrophages in severe asthma compared to mild-moderate asthma.
The M2-validated signature was significantly correlated with TAC3 and the TR-Mφ signature with TAC3, to phago-lysosome activity and to oxidative phosphorylation (OXPHOS) ( Table S5). More detailed information about TR-Mφ functions is provided in the (Table S6).
Overall, these data suggest that patients with severe granulocytic asthma have a reduced innate immune response as indicated by the

| Macrophages signatures in airflow obstruction
The same modules 15, 21 and 29 that were overexpressed in severe asthma and associated with inflammation driven by innate immunity were correlated with airflow obstruction ( Figure S3). In contrast, most M1-and M2-related modules were inversely associated with obstruction. The TR-Mφ signature did not correlate with airflow obstruction ( Figure S4).  Figure S5). We therefore pooled the activated genes within each module 9 for severe asthma and for each TAC and examined the pathways and gene ontology terms associated with these groups to determine whether specific pathways or driver mechanisms were present.

| Drivers of macrophage activation according to severity and TAC characterization
Combining the macrophage gene signatures enriched in severe asthma and performing PPI network analysis identified several pathways important in severe asthma including innate immunity, leucocyte activation, neutrophil degranulation, IL-4/IL-13 signalling, TLR and IL-17 signalling and leukotriene synthesis (Table 3, Figure 5). The pathways associated with macrophages activated in TAC1 patients highlighted cholesterol and steroid biosynthesis pathways along with pathways associated with Th2 activation and defence responses ( Figure S6, Table S8). In addition, the PPI networks identified as being preferentially activated in TAC2 patients included immune and inflammatory pathways linked to inflammasome and neutrophil activation ( Figure S7, Table S9). Finally, TAC3-associated PPI pathways reflect metabolic and mitochondrial function in addition to cytokine signalling and viral response pathways ( Figure S8, Table S10). These data expand the pathways associated with TACs described previously and highlight the importance of macrophages in severe asthma pathophysiology.  TNF receptors including TNFR1 and   2, TL1A, OX40, 4-1BB, CD30, CD40, LIGHTR, LTβR, GITR, BAFF, APRIL and TWEAK which all stimulate NF-κB. 24,25 In addition, OX40

| D ISCUSS I ON
stimulation promotes T-cell proliferation and survival 26,27 by the regulation of IL-2 production. 28 The third module enriched in severe asthma is module 21 that is associated with eicosanoids biosynthesis via the 5-LOX pathway.
These macrophages are stimulated by palmitic acid, a pro-inflammatory factor for monocytes and associated with increased mitochondrial ROS production. 29 CysLT promotes Th2 polarization, activation for migration, ROS and pro-inflammatory cytokines production (via NF-κB activation) and phagocytosis. 29 Conversely, the resolution of inflammation occurs through the interaction of oxidized The importance of macrophages in mediating asthmatic inflammation is becoming increasingly clear. 38 In mice, pharmacological suppression of alternatively activated YM1 + M2 macrophages using the galactin-3 pathway inhibitor cynaropicrin resulted in reduced eosinophilic lung inflammation and less collagen deposition around airways and a shift towards neutrophilic inflammation and worse lung function. 44 These data revealed an important role for M2 macrophages generally in airway remodelling and the role of specific M2-like macrophages associated in airway remodelling and AHR should be further studied.
This is the first study using gene signatures to extensively analyse the enrichment and activation of macrophage subtypes in asthma according to severity, molecular phenotype and the presence of airflow limitation. To the best of our knowledge, it is the first time that the enrichment of TR-Mφ has been analysed in asthma where it plays a more important role in the paucigranulocytic inflammatory phenotype. Another strength of this study is the relatively large data set, the validation in another severe asthma cohort (ADEPT) F I G U R E 5 Protein-protein interaction (PPI) network analysis of module genes significantly enriched in macrophages activated in severe asthma compared to mild-moderate asthma and healthy controls. Pink: arachidonic acid metabolism; light blue: leucocyte activation and neutrophil degranulation; green: IL-4 and IL-13 signalling; dark blue: NF-κB signalling; light brown: cellular response to stimuli; red: immune system; and yellow: Toll-like receptor cascades. All genes in each module that was differentially expressed were pooled and used in this analysis and the relative accessibility of the sputum compartment for future comparative analysis. Nevertheless, there are several limitations to the study including the failure to obtain functional data from these macrophage modules and the need to isolate or define these macrophage subtypes using another approach such as cell sorting. It is possible that the results may reflect, at least in part, the varying proportional make-up of the immune cells present across the various subgroup comparisons, and future studies should use single cell sequencing or other approaches to address this. Furthermore, analysis of BAL and tissue macrophages may provide additional insights in the pathogenesis of severe asthma.
In summary, macrophage numbers and activation status are generally decreased in severe asthma suggesting an innate immune defect in this population. Enrichment was found only in three modules involved in the regulation of inflammatory response by TLR/ TNF/NF-κB activation, IL-2 production and leukotriene biosynthesis. EFPIA companies' in-kind contribution (www.imi.europa.eu). We acknowledge the contribution of the whole U-BIOPRED team as listed in the Appendix S1. Angelica Tiotiu thanks Collège Lorrain de Pathologie Thoracique for supporting her Fellowship at Imperial College London. We thank Professor Louise Donnelly for her comments on this manuscript.

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflict of interest.

AUTH O R CO NTR I B UTI O N S
AT, NZK, YB and SP made substantial contributions to the acquisition and analysis of the data, and PH, YG, KFC and IMA made substantial contributions to the conception and interpretation of the work. AT and IMA drafted the initial manuscript, and all authors provided substantial input into the revision and interpretation of the manuscript.
All authors approved the final version for submission and accept responsibility for the accuracy and integrity of the work.

DATA AVA I L A B I L I T Y S TAT E M E N T
The transcriptomic data are deposited in the GEO (Gene Expression