Three myeloid populations expand in the airway after injury

The role of macrophages in epithelial repair was investigated using the naphthalene (NA)-induced bronchiolar epithelial injury model²¹, which is characterized by a rapid loss of club cells as shown by the decrease of CCSP expression in tissue sections (Fig. 1a, b) and of the mRNA encoding secretoglobin family 1A member 1 (Scgb1a1) in lung homogenates (Supplementary Fig. 1a). Epithelial regeneration was determined by the recovery of CCSP protein and Scgb1a1 mRNA expression, beginning after day 6 (d6, maximal proliferation of club cells), and returning to normal levels by d35 (Fig. 1a, b, Supplementary Fig. 1a). During this epithelial repair phase (d6–d35), we observed an accumulation of F4/80⁺ myeloid-derived macrophages near to the injured bronchial epithelium (brown staining, Fig. 1c). Total monocytes/macrophages in the bronchoalveolar lavage (BAL) also increased and peaked between d6 and d9, after which numbers declined to baseline (Fig. 1d). Macrophage expansion was associated with an early (d1–d3) increase in BAL fluid levels of IL-1α, IL-13, CCL2, and CXCL10²², important regulators of monocyte/macrophage function (Fig. 1e).

a–i WT C57BL/6 mice were untreated (naïve, N) or treated with naphthalene (NA) and analyzed at various days thereafter. a Bronchiolar epithelium regeneration after NA-induced injury, as assessed by immunofluorescence staining of CCSP in lung tissue sections. b Quantification of CCSP expression in lung tissue sections from naïve and NA-treated mice, expressed as percentage of fluorescence within bronchioles (150–400 µm diameter), at the indicated time-points after NA. c Immunohistochemical analysis of F4/80 expression (brown deposit) illustrating macrophage localization (black arrows) around the injured bronchiolar epithelium in lung tissue sections. d Quantification of the total number of cells in the bronchoalveolar lavage (BAL). e Levels of IL-13, CCL2, CXCL10, and IL-1α in BAL supernatants. f Monocyte/macrophage subsets (P1–P4). Inflammatory monocytes F4/80^low CD11b⁺ (P1), recruited macrophage F4/80^int CD11b⁺ (P2), resident macrophages F4/80^high CD11b⁻ (P3) and apoptotic macrophages Annexin V⁺ F4/80^low CD11b⁻ (P4) in the BAL are defined by their gates in (f). g Total cell numbers of P1–P3 subsets at the indicated time-points after NA administration. h Representative FACS profiles of BAL cells obtained on d6 after NA, illustrating the expression of CD206, FIZZ-1, YM1, and Arg-1 in P1–P3 BAL cell subsets, respectively. i Quantification of BAL macrophage proliferation as assessed by FACS analysis on P2 and P3 subsets, using Ki-67 staining. Data are from 8 (a–e) and 6 (g, i) mice, obtained in 3 independent experiments, and represented as mean ± SEM. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between NA-treated and naïve WT mice using one-way ANOVA, Bonferroni post-test. Scale bars in a and c = 100 µm.

The majority of cells within the airway lumen were myeloid-derived monocyte and macrophage populations, (Fig. 1f, Supplementary Fig. 1b). We further defined three distinct myeloid cell populations (Fig. 1f, g), namely F4/80^low CD11b⁺ Ly6C⁺ infiltrating inflammatory monocytes (P1 subset); F4/80^int CD11b⁺ monocyte-derived macrophages (P2 subset); F4/80^high SiglecF⁺ CD11c⁺ resident alveolar macrophages (P3 subset), and F4/80^low annexin V⁺ apoptotic macrophages (P4 subset) (Supplementary Fig. 1c). The absolute numbers of P1–P3 subsets peaked between d6 and 15 post NA (Fig. 1g), whereas 40% of F4/80^low CD11b⁻ annexin V⁺ apoptotic macrophages (P4 gate) were detected on d3 but declined by d9 (Fig. 1f). Further phenotyping found that P3 subset exhibited an enhanced AAM phenotype, when compared to P1 and P2 cells and expressed CD206, FIZZ-1, Arg-1, and YM1 (Fig. 1h, Supplementary Fig. 1d), and their counts increased approximately fourfold between d6 and d15 post-NA, when compared to naïve mice (Supplementary Fig. 1d, e). In addition, increased mRNA expression of the AAM markers, Itgax, Cd206, Retnla, Arg1, and Chil3 was observed in total BAL cells isolated from NA-treated mice, when compared to naïve (Supplementary Fig. 1f). NA-induced injury also triggered local macrophage proliferation between d3 and d21 (Fig. 1i), as evidenced by Ki-67⁺ P2 and P3 subsets (Supplementary Fig. 1g). Proliferation of F4/80⁺ macrophages was further confirmed by co-immunofluorescence (Supplementary Fig. 1h). Thus, macrophage expansion during epithelial repair involves a significant proliferation of monocyte-derived macrophages (P2) and resident alveolar macrophages (P3), as well as the recruitment of inflammatory monocytes (P1).

Epithelial regeneration requires resident lung macrophages

To determine the contribution of alveolar macrophages to bronchial repair, myeloid cells were depleted by administering Clodronate (CL)-containing liposomes²³ to NA-treated mice on days 2, 5, and 8 (Fig. 2a). CL-treated mice exhibited an incomplete bronchial re-epithelialization (Fig. 2b, c), that was associated with ~90% reduction of BAL macrophages, when compared to vehicle containing liposomes, including reduced P2 and P3 numbers, while P1 counts were unaffected by CL treatment (Fig. 2d). In addition, a significant reduction in the expression of AAM markers was seen (Supplementary Fig. 2a, b). Adoptive transfer experiments were performed to decipher which macrophage subset contributed to repair. Given their significant increase during the repair phase (Fig. 1g), P3’s (i.e., F480^high CD11b⁻ CD11c⁺ CD206⁺) were sorted to ≥98% purity from the airways of NA-injected mice, and transferred into recipient, macrophage-depleted, NA-treated animals, 72 h after the final CL administration (d11, Fig. 2a, Supplementary Fig. 2c). Reconstitution of P3 resident macrophages completely rescued repair in macrophage-depleted mice, (Fig. 2b, c), suggesting that these cells are required for club cell regeneration. We hypothesized that P2’s would also contribute to repair by their ability to differentiate into mature P3 cells^9,24,25, therefore, GFP⁺ P2 cells (F4/80^int CD11b⁺ cells), were transferred as described above, to recipient macrophage-depleted mice and tracked for their surface markers post-NA injury (Fig. 2a). Interestingly, immature, GFP⁺ monocyte-derived macrophages acquired markers of P3 resident macrophages, namely CD11c and SiglecF, within 1 week of transfer to recipient mice; these maturation markers were still evident 2 weeks post injury (Fig. 1e). Thus, P2’s are a transient population which upon injury can readily switch into P3 cells and contribute to replenishment of the P3 resident pool enabling continued repair. This observation is consistent with bleomycin-induced lung injury where 50% of alveolar macrophages were found to be monocyte-derived after fibrosis had resolved²⁶, in addition to adoptive cell transfer studies demonstrating that CD11c⁺ alveolar macrophages originate from blood monocytes in response to conditional macrophage depletion²⁴.

a Schematic for clodronate (CL)-mediated macrophage depletion during naphthalene treatment (NA + macrophage depletion). NA-treated macrophage-depleted mice were then either given no cells or intratracheally adopted with GFP⁺ P2 or P3 cells (NA + P3/AAM adoptive transfer). b Levels of the Scgb1a1 mRNA in total lung homogenates from NA-treated mice ± macrophage depletion. c Assessment of club cell regeneration after NA-induced injury in lung tissue of NA-treated WT mice (Control) or NA-injected mice treated with clodronate liposomes (Depleted) or AAM adoptively transferred into depleted mice (see Fig. 2a). CCSP immunofluorescence staining was performed at d35. Graph on the right represents CCSP quantification in lung tissue sections, expressed as percentage of fluorescence within bronchioles (150–400 µm diameter). d Total BAL cells subdivided into P1–P3 gates as defined in (Fig. 1f) determined by counting and flow cytometry from NA-treated mice ± macrophage depletion. e Representative flow plots illustrating the percentage of adoptively transferred GFP⁺ P2 cells that switched into CD11c⁺ SiglecF⁺ cells after 1 week in the lungs of depleted mice. Graph on the right represents the quantification of CD11c⁺ SiglecF⁺ GFP⁺ cells. Data are from 6 to 10 (b), 3 (c), 6 (d), and 3 to 7 (e) mice, obtained in three independent experiments and represented as mean ± SEM. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between macrophage-depleted and NA-treated WT mice using one-way ANOVA, Bonferroni post-test. ^#P < 0.05, ^##P < 0.01 and ^###P < 0.001 between AAM adoptively transferred and macrophage-depleted WT mice using one-way ANOVA, Bonferroni post-test. Scale bar in c = 50 µm.

RNASeq confirms club cell regeneration pathways in P3 cells

RNAseq analysis was performed on flow-sorted P1–P3 subsets 6d post-NA treatment. Pairwise comparisons between the different subsets revealed 3751 differentially expressed genes (DEG) during monocyte differentiation (P2 vs. P1), while there were 6244 DEG between P2 and P3 subsets (Supplementary Fig. 3a, Supplementary Datas 1–4). Thus, based on their transcriptome, resident (P3) and recruited macrophages (P2) are functionally distinct. Computational analyses were applied to decipher key functional differences between these myeloid populations. Principal component analysis (PCA) identified transcriptomic clustering of the distinct myeloid subsets and highlighted the similarities between naïve and post injury P3 subsets (Fig. 3a). Additional hierarchical clustering identified three distinct expression patterns within the P1–P3 subsets, named clusters I–III (Fig. 3b, Supplementary Data 5). Notably, cluster I included 374 highly expressed genes in P1’s, which were slightly downregulated upon differentiation into P2 cells and further repressed within P3’s. Cluster II exhibited an expression profile of 33 genes which were highly upregulated in P2’s compared to P1 and P3 subsets. Upon P2’s converting into P3 cells, 299 genes associated with alveolar macrophage homeostasis²⁷ became more prominent in cluster III compared to cluster I and II (Fig. 3b). Pathway enrichment analysis was performed to understand the biological processes associated with each gene cluster. Genes upregulated during the P2–P3 cellular switching were associated with efferocytosis, cell cycle, biosynthetic processes, secretion of ECM components and growth factors, pathways all involved in repair. In contrast, downregulated genes were mostly associated with development, complement activation, inflammatory responses and signal transduction (Fig. 3b, Supplementary Data 6).

a–e Transcriptional profiling of myeloid cell subsets flow-sorted from naïve and NA-treated mice on d6. a Principal component analysis (PCA) of the transcriptomes of flow-sorted monocytes (P1), monocyte-derived macrophages (P2), and resident AAMs (P3) performed on all expressed genes identified from a generalized linear model to perform an ANOVA-like test for differential expression between any conditions in the dataset (FDR step-up procedure q-value < 0.05. b k-means clustering of all identified genes revealed core gene signatures specific to P1 (Cluster I), P2 (Cluster II), and P3 (Cluster III). The pathway enriched processes associated with each cluster are shown on the bottom of the heatmap. Scale bar on the bottom denotes relative log₂ differences in gene expression for each row. c Heatmaps showing the top relative expression of DEG that are associated with macrophage ability to modulate the epithelial cell niche. Those functions include growth factors secretion, extracellular matrix (ECM) remodeling and immune response regulation between P2 and P3 cells. d, e Score plot illustrating the signal transduction pathways upregulated in Cluster I (d), highlighting the overrepresentation of IL-33 signaling and MyB88/NFkb axis in the different subsets of myeloid cells (e).

Because epithelial progenitor behavior is regulated by the local environment^28,29, we hypothesized that both recruited and resident macrophages may modulate the club cell niche. The expression of genes associated with epithelial repair including growth factors, ECM remodeling and inflammation between P2 and P3 subsets were compared (Fig. 3c, Supplementary Fig. 3b, Supplementary Data 7). Although P2 cells displayed higher expression of the alveolar regenerating factor Wnt11³⁰, P3 cells were enriched with a larger variety of growth factors required for club cell regeneration, such as Plet1, Nrg4, Gdf15, Nrp2, Ereg, and Mreg^31,32,33. P3’s also displayed higher expression of the epithelial niche ECM components, Ctsk³⁴, Mnt2³⁵, and Krt79³⁶ (Fig. 3c, Supplementary Fig. 3b). Interestingly, P2 cells were more profibrogenic and proinflammatory, similar to those reported in lung fibrosis²⁶, than their P3 counterparts that exhibited more of an immunomodulatory phenotype as seen by the higher expression of Cd200r³⁷ (Fig. 3c). Lastly, we examined the enriched signal transduction pathways within P1 and P2 subsets associated with cluster I in depicting myeloid cell development and differentiation pathways, (i.e., complement activation, innate and immune responses, transcription, cytoskeleton remodeling, and chemotaxis). In addition to the upregulation of IL-4 and IL-13 signaling, we found significant evidence for IL-1 and IL-33 pathway activation, including overexpression of genes encoding Il1rl1 (ST2) and Il1racp, (IL-1RAcP; the co-receptor required for IL-1R1 and ST2 signaling), and subsequent downstream NF-κB activation via the genes encoding the heterodimeric signaling complex MyD88/IRAK1/IRAK2/TRAF6 which activates NF-κB transcription factor¹⁶ (Fig. 3d, e).

AAM-mediated epithelial repair requires the IL-33-ST2-axis

Evidence for IL-33 pathway activation is consistent with our previous report of ST2⁺ macrophages modulating inflammation during influenza infection¹⁵, thus we examined ST2 function on macrophages during NA-induced repair. Interestingly, Il1rl1 transcripts were increased in total BAL cells and both ST2⁺ P2 and ST2⁺ P3 cells were evident in the lung post-injury (Fig. 4a). Using a commercial monoclonal antibody, we found that ST2 expression appeared to be much higher on P2 subset vs. relatively dim P3 cells (Fig. 4b, left panel). ST2 was not detected on P1 cells (Supplementary Fig. 3c), but ST2 mRNA expression was observed in P1 and P2 subsets 6 days post injury (Fig. 3e). Commercial ST2 antibodies exhibit significant background staining and are best used for highly expressing cells, thus, we attempted to confirm receptor levels using cells from ST2-GFP reporter mice^15,38. Increased expression of GFP-ST2 was detected on P2 macrophages, (Fig. 4b, right panel), however, the weak fluorescence emission signals from these reporters were not sensitive enough to capture the lower ST2 levels detected on P3 cells by flow cytometry. This observation is consistent with our RNAseq data where Il1rl1 mRNA levels are high in P2 subset while Il1racp mRNA, is relatively low, in contrast, P3 cells have minimal Il1rl1 mRNA expression but very high Il1racp mRNA levels (Fig. 3e). Thus, ST2 expression may be transient during the differentiation of immature P1 cells to mature P3 macrophages following NA-induced injury. This hypothesis is supported by a previous report demonstrating that ST2 expression during endothelial cell differentiation was growth dependent³⁹.

a Relative mRNA expression of Il1rl1 as determined by qPCR for total BAL macrophages. b FACS quantification as percentage of ST2 expression on P2 and P3 cells of total cells in BAL. Results were confirmed using GFP ST2 reporter mice in the right panel. c Single-cell RNA-sequencing analysis of mouse naïve bronchial epithelial lineages showing t-SNE plots of club cell lineages as identified by Scgb1a1 and Scgb3a1. Clusters 0, 1, 5, and 7 represent club-epithelial lineages subsets. Gene expression plots demonstrating expression of IL-33 in a proportion of club cells (Cluster 1). High expression of Foxj1 in Cluster 2 identifies club-derived ciliated cell. Other clusters (3, 4, and 6) are club-derived mesenchymal lineages. Scale bar to the right denotes normalized gene expression level of marker’s gene for each cell. d Representative sections from histological staining of CCSP (red), IL-33 (white), β-tubulin (green), and nuclear content (DAPI, blue) in lung sections. White arrows indicate IL-33-containing club cells. e soluble ST2 (sST2) levels in the BAL supernatants. f Levels of IL-33 in homogenized lung samples. Data from n = 7 (a, d), 8 (e), and 9 (f) mice are representative of at least three independent series of experiments and show mean ± SEM. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between NA-treated and naïve (N) WT mice using one-way ANOVA, Bonferroni post-test. Scale bar in d = 50 µm. t-SNE (t-distribution stochastic neighbor embedding) plots in graph c show data from one experiment (n = 1).

Consistent with the precedence for IL-33 release during injury^14,15,40,41 and IL-33 expression being reported in subsets of lung epithelial progenitors post infection¹⁴, single cell-RNAseq analysis of mouse bronchial epithelial cells, (Epcam⁺ Scgb1a1⁺), demonstrated that a proportion of club cells (33% of Cluster 1, Fig. 4c) constitutively expressed Il33 mRNA. To confirm club cell expression at homeostasis and decipher whether there was a loss of IL-33 signal following NA-induced desquamation, we optimized an established immunohistological method using a validated antibody^15,42. Indeed, we showed that club cells expressed IL-33 protein at baseline and a small subset of these were also positive for this cytokine following NA treatment (Fig. 4d, left and middle panel respectively, red/white cells). In addition, we observed a significant reduction in IL-33 immunoreactivity within red-labeled club cells after NA, (Fig. 4d, middle panel; NA-resistant club cells maintain IL-33 expression, white cells d9 post NA; ciliated cells are green), levels of which were restored upon epithelial regeneration (Fig. 4d, right panel, d35 post NA). Since IL-33 is susceptible to oxidation⁴³ and technically challenging to detect, we investigated the levels of soluble ST2 (sST2), a decoy receptor for IL-33, produced following IL-33 release and thought to predict activation of this pathway^44,45. Increased amounts of sST2 were found between d1 and d2 post NA while lung IL-33 levels, (i.e., presumably intracellular), remained elevated until d6 and d15, respectively (Fig. 4e, f). These data, along with the RNAseq above (Fig. 3d, e), prompted us to examine whether the IL-33-ST2 pathway was required for AAM-mediated epithelial repair. Strikingly, NA-treated mice lacking the ST2 receptor, (Il1rl1^−/−), exhibited a severe defect in epithelial repair as shown by significantly lower CCSP expression when compared to wild-type (WT) mice (Fig. 5a, b, Supplementary Fig. 4a). We questioned whether this incomplete re-epithelialization resulted from an impaired differentiation and/or proliferation of club cells. The proportion of CCSP^high (terminally differentiated) and CCSP^low (potential to proliferate) and Ki-67⁺CCSP^low cells within each bronchiole were quantified by immunofluorescence. In WT mice, NA injury induced the proliferation of NA-resistant club cells (Ki-67⁺CCSP^low) that covered 25% of bronchial epithelium (d6–9; Supplementary Fig. 4b). After differentiation, mature club cells (CCSP^high) re-epithelized approximately 45% of the bronchial epithelium (d35, Supplementary Fig. 4b, left panel). Intriguingly, absence of ST2 significantly downregulated club cell proliferation and differentiation after injury, as evidenced by the decreased percentages of Ki-67⁺CCSP^low cells (10%; Supplementary Fig. 4b, d6, d9, right panel) and mature CCSP^high (20%; Supplementary Fig 4b, d35, middle panel). Incomplete epithelial repair in ST2^−/− mice was associated with reduced accumulation of all myeloid subsets (P1–P3) within the airway compartment, despite the increased expression of the monocyte chemokines CCL2 and CXCL10 (Fig. 5c, d, Supplementary Fig. 4c, d).

a Co-immunofluorescence staining of CCSP and Ki-67 in lung tissue sections from naïve (N) and NA-injected wild-type (WT) and ST2^−/− mice, at indicated time-points. b Quantification of CCSP in lung tissue sections, expressed as percentage of fluorescence within bronchioles (150–400 µm diameter). c Absolute numbers of P1–P3 subsets. d IL-13 and CCL2 levels in BAL supernatants. e Quantification of IGF-1 and HGF in BAL macrophage lysates. f Quantification of Ki-67⁺ AAMs in BAL samples from mice. g–j Schematic representation (g) of the procedure used for P3/AAM adoptive transfer into ST2^−/− mice on d9 after NA administration (ST2^−/− + AAM adoptive transfer), and immunofluorescence staining of CCSP in lung tissue sections (j). Quantification of CCSP in lung tissue sections (h), expressed as percentage of fluorescence within bronchioles (150–400 µm diameter), and Scgb1a1 mRNA levels in lung homogenates (i). Data from n = 10 (a, b), 5 (c), 8 (d), 5 (e), 5 (f), 10 (h, j) and 8–23 (i) mice are representative of at least 2–3 independent experiments and represented as mean ± SEM. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between NA-treated ST2^−/− and NA-treated WT mice using one-way (b, e, f, h, and i) and two-way (c, d) ANOVA, Bonferroni post-test. ^#P < 0.05, ^##P < 0.01 and ^###P < 0.001 between AAM adoptively transferred and NA-treated ST2^−/− mice using one-way ANOVA, Bonferroni post-test. Scale bars in a and j = 50 µm.

Consistent with decreased numbers of AAMs, we found significantly lower BAL levels of IL-13 and of the AAM-associated mediators CCL17, breast-regression protein-39 (BRP-39), CXCL12, as well as a reduction in the epithelium-regenerating factors, IGF-1 and HGF in BAL macrophage lysates from ST2^−/− mice (Fig. 5d, e, Supplementary Fig. 4e, f). ST2 deficiency also resulted in reduced numbers of Ki-67⁺AAMs (Fig. 5f) and lower pro-inflammatory cytokine levels in BAL macrophage lysates post NA, factors of which are critical for macrophage polarization (Supplementary Fig. 4e). To demonstrate whether P3 AAMs could also reverse the phenotype in ST2-deficient mice, we reconstituted NA-treated ST2^−/− mice with P3 cells isolated from NA-treated WT mice (Fig. 5g). Transfer of WT AAMs completely restored bronchiolar re-epithelialization in ST2^−/− mice (Fig. 5h–j).

ST2 controls self-renewal and differentiation of macrophages

To understand which biological processes in lung macrophages were regulated by ST2, myeloid cell subsets were flow-sorted from NA-treated WT and ST2^−/− mice on d6 and their transcriptome analyzed using RNAseq. Pairwise analysis between the P1, P2, and P3 WT vs. ST2^−/− cell populations revealed that 396, 467, and 63 genes were differentially regulated in each cell subset respectively (Supplemental Datas 8–10). Interestingly, the ST2 pathway regulated differential gene expression in each of these three myeloid subsets, which was reflected by the low overlap between the DEG (Fig. 6a, Supplementary Fig. 5a). Although the P1 subset presented a considerable number of DEG (396), pathway enrichment analysis revealed a significant alteration in cytokine-mediated signal transduction processes, such as MIF, TNF, and IL-8 signaling (Supplementary Fig. 5a), whereas, pathway enrichment analysis of the DEG within the P2 and P3 populations identified 71 and 19 ST2-dependent pathways, respectively. Those pathways downregulated in P2 ST2^−/− cells post-NA injury were predominantly involved in cell self-renewal and differentiation processes, such as DNA synthesis/replication/repair, transcription/translation, metabolism (tricarboxylic acid (TCA) cycle, RNA, and proteins) and phagocytosis (Fig. 6b, Supplemental Data 11). Further, ST2^−/− P3 cells exhibited drastic defects in processes controlling cell cycle, cell–matrix adhesion and migration (NCAM signaling), metabolism (RNA, protein, and glycan), vesicle-mediated transport and ECM remodeling following NA-treatment (Fig. 6c, Supplemental Data 12). More specifically, P3 cells exhibited an imbalance in ECM remodeling in favor of increased expression of proteinases, such as Ctse, Adam3, and Mmp19, while epithelial niche components, Spon2, Sdc2, Sdc4, Eln, Ogn, Serpinb2, and Arg1 were significantly decreased (Fig. 6d). In addition, a dysregulated profile of growth factors promoting angiogenesis was observed, (i.e., upregulation of Thbs1-3, Vgf, and Pdgfb), as opposed to epithelial regeneration (i.e., decreased expression of Areg, Ereg, and Mreg). Furthermore, ST2 deficiency induced a dysregulated immune response in P2 and P3 subsets given the decreased expression of genes encoding Il1a, Il1b, Il6, and Tnfa, whereas genes implicated in inflammasome activation, such as Cd300lb⁴⁶, P2yr10⁴⁷, and Nlrp12⁴⁸ were upregulated in both subsets (Fig. 6d, Supplemental Data 13 and 14). These findings suggest that loss of ST2 may have markedly downregulated the differentiation of P2 cells into resident P3 subset. Consequently, cells that switched into P3 phenotype exhibited significant defects in their maturation and repairing function.

a–d Transcriptomic profiling of myeloid cells flow-sorted from ST2^−/− and WT mice 6 days after NA treatment. a Summary of differentially expressed genes (P < 0.01; FC > 2) in each pairwise comparison showing the total number of differentially expressed genes unique and shared between P1–P3. b, c Heatmaps and summary of the total number of differentially expressed genes (P < 0.01; FC > 2) in P2 (panel b) and P3 (panel c) between WT and ST2^−/−. The downregulated pathways associated with the gene differentially expressed in ST2^−/− for P2 and P3 are illustrated on the right side of the heatmaps. d Heatmaps showing the top relative expression of DEG that are associated with macrophage ability to modulate growth factors secretion, extracellular matrix (ECM) remodeling and immune response regulation between P2 (left panel) and P3 (right panel). Presented in red and blue the gene upregulated and downregulated in ST2^−/−, respectively. Scale bar on the bottom denotes relative log₂ differences in gene expression for each row.

To corroborate the RNAseq analysis and demonstrate an intrinsic role for ST2 on monocyte/macrophage function, we generated bone marrow-derived macrophages (BMDMs) from WT- and ST2-deficient mice. Successful cell cycle progression is necessary for effective monocyte/macrophage expansion and self-renewal^49,50, yet myeloid cells lacking ST2 exhibited a dramatic downregulation of genes involved in these processes. To determine whether ST2 regulates cell cycle of BMDMs, we examined proliferation responses to IL-33 by flow cytometry. WT BMDMs significantly proliferated in response to IL-33, i.e., a higher proportion of cells progressed into S and G2 phases when compared to unstimulated cells. However, BMDMs lacking ST2 were completely arrested within the G0/G1 phase of the cell cycle and unable to differentiate further (Fig. 7a, b). Notably, IL-13 did not impact the cell cycle under these same conditions (Supplementary Fig. 5b). We next questioned whether IL-33 modulated the expression of key genes involved in these pathways. Here, responses to IL-13 were also examined, since Lechner et al.⁵¹ reported the requirement of CCR2⁺ monocytes and IL-4/IL-13 signaling in the regenerating lung after pneumonectomy. Strikingly, IL-33 induced the transcription of mRNAs encoding the apoptosis inhibitors, Api5 and Bcl-xl, the self-renewal mediators, c-Maf, Klf4 (Kruppel-like factor 4), c-Myc, Ccnd1 (Cyclin D1), and the transcriptional effector E2f1, all of which are involved in S phase progression (Fig. 7c). Furthermore, the expression of Bcl-xl and c-Maf⁵⁰ was specifically upregulated by IL-33 but not affected by IL-13, which was consistent with IL-13 having no impact on cell cycle (Fig. 7c, Supplementary Fig. 5b). In contrast, antiproliferative markers, such as cyclin-dependent kinase inhibitor p27, the liver X receptors α (Lxrα), and transcription factor Mafb⁵² were solely expressed in ST2^−/− BMDMs. Importantly, IL-33 also specifically enhanced the expression of genes controlling macrophage fate, including the transcription factors, early growth response protein 1 (Egr1), Gfi1 (growth factor independence 1), Stat3 and Stat6 (AAM regulators) and Irf4 (IFN regulatory factor 4). Furthermore, IL-33 significantly increased the transcription of genes associated with AAM polarization and maturation, such as Cd206, Siglecf, Arg1, Fizz1, Chil3, Hgf, Tlr4, Il1rl1, Il1rAcp, and p53. It is worth noting that the transcripts encoding Il1a, Il1b, and Il6 were also upregulated by IL-33, but not modulated by IL-13. However, IL-33 sustained IL-13-mediated downregulation of genes encoding the proinflammatory macrophage markers, Irf8 (IFN regulatory factor 8) and Stat1, as well as the pro-inflammatory mediators Tnfa and Cxcl10 (Fig. 7c, Supplementary Fig. 5c).

a–c Bone marrow-derived cells were cultured in M-CSF alone (unstimulated) or cultured in the presence of M-CSF with varying concentrations of IL-13 and IL-33 for 3 days ex vivo. a Representative histograms from WT and ST2^−/− BMDMs showing DAPI staining after fixation. Cell cycle phases (G0/G1, S, and G2) are gated by nuclear DNA content. b Quantification of BMDMs in the three cell cycle phases and shown in (a) and disc graphs denote mean only. c Heat-map constructed from Fluidigm analysis of mRNA transcripts for the denoted genes from cultured WT and ST2^−/− BMDMs. Scale bar on the bottom denotes relative log₂ differences in gene expression for each row. Samples are bone marrow treated and cultured separately from three individual mice per genotype. All bar graphs show means ± SEM. Data from n = 3 mice are representative from two independent experiments. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between indicated IL-33-treated and unstimulated BMDMs. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between IL-33-stimulated ST2^−/− and WT BMDMs using two-way ANOVA, Bonferroni post-test.

Lastly, the expression of AAM markers on WT and ST2^−/− BMDMs, following IL-33 and/or combined IL-13+IL-33 stimulation was evaluated. Both IL-33 and IL-13 induced significant ST2 (GFP) expression on CD206⁺ BMDMs, albeit the effect of IL-33 stimulation was much greater than that of IL-13; when these cytokines were added in combination, ST2 expression was additively enhanced compared to IL-33, or IL-13 alone (Fig. 8a, b). Although stimulated with IL-13, ST2^−/− BMDMs failed to acquire an AAM phenotype as evidenced by the diminished frequency of CD206⁺Arg-1⁺ cells (Fig. 8c, d). Similarly, lack of ST2 not only downregulated HGF secretion in IL-33-induced BMDMs but also markedly reduced HGF production upon IL-13 stimulation (Fig. 8e). In this setting, IGF-1 production was not affected by ST2 deficiency, whereas IL-13-induced BRP-39 release was entirely dependent on ST2 (Fig. 8f, g). IL-33 has been previously reported to amplify AAM polarization and chemokine production¹⁹, however, our data confirm and extend an intrinsic role for ST2 on macrophage function, as well as highlight an important synergy between IL-33 and IL-13, whereby they can both impact macrophage function in an additive or synergistic manner depending on the mediator examined (Fig. 7c, Fig. 8a–d). In short, we demonstrate an upstream role of IL-33-ST2 in macrophage differentiation and maturation where synergy with IL-13 may potentiate macrophage activation and their ability to produce growth factors associated with epithelial repair.

a–g Bone marrow-derived cells from WT Balb/c, ST2^−/− (Balb/c background), or ST2-GFP (Balb/c) mice were cultured in M-CSF alone (unstimulated), or cultured in the presence of M-CSF with varying concentrations of IL-13 and/or IL-33 for 6 days ex vivo. a Representative flow cytometric plots of CD206 and GFP expression from ST2 GFP bone marrow-derived macrophages (BMDMs). Numbers next to gate denote percentage of CD206⁺ ST2 GFP⁺ BMDMs. b Quantification of the frequency of CD206⁺ GFP ST2⁺ gated in (a). c Representative flow plots of Arginase-1 (Arg-1) and CD206 expression on WT and ST2^−/− BMDMs. Gates and numbers denote the percentage of CD206⁺ Arg-1⁺ BMDMs. d Quantification of the frequency of CD206⁺ Arg-1⁺ BMDMs from WT or ST2^−/− mice as gated in (c). e–g Quantification of HGF (e), IGF-I (f), and BRP-39 (g) from the supernatants of WT and ST2^−/− BMDMs. Samples are bone marrow treated and cultured separately from three individual mice per genotype. Bar graphs from n = 4 (b), 3 (d–g) mice, show mean ± SEM pooled from three independent experiments.

IL-33 promotes ILC2 activation for macrophage-induced repair

ILC2s have also been implicated in epithelial repair following influenza-induced injury^41,53,54,55 and mediate lung immunity at barrier sites by actively responding to IL-33^{15,53,54,56,57}. Recently, Lechner et al.⁵¹ demonstrated that both ILC2-derived IL-13 and IL-4/IL-13 signaling within the hematopoietic compartment are required for optimal lung regeneration after pneumonectomy. Similarly, we hypothesized that IL-33 was upstream of IL-13 and that an ST2-ILC2 axis may orchestrate bronchial repair by providing IL-13 and regulating AAM polarization post NA injury. Thus, we performed the following experiments to validate our hypothesis. ILC2s were easily identifiable in the lungs of naïve and NA-treated mice as cells lacking CD3 and CD49b, as well as other lineage markers, except for CD90⁵⁵ (Supplementary Fig. 6a). Indeed, we observed a small, but significant increase in the frequency of lung ILC2s post NA but this did not reflect an appreciable increase in total ILC numbers (Fig. 9a, Supplementary Fig. 6a). Unlike other forms of lung injury, such as cigarette smoke exposure and viral infection^15,55, ILC2s from NA-treated mice maintained their type 2 phenotype, producing significantly more IL-13, GM-CSF, and IL-5, when compared to naïve mice (Fig. 9a–c, Supplementary Fig. 6b–d). We examined all cells over the course of injury, specifically IL-13 producing cells, that were present in the lung. In this setting, IL-13⁺ lung NK or total T cells, CD4⁺ and CD8⁺, numbers were unaltered (Supplementary Fig. 6e–h), supporting our hypothesis that ILC2s are the main source for IL-13 production post injury. Lung regulatory T cells (Treg cells) and ILC2s have been shown to mediate lung tissue repair following influenza-induced damage, the latter cells through the production of amphiregulin^41,58. Post injury, we found no significant differences in the number of lung-associated Treg cells (CD4⁺ CD25⁺ CD44⁺ cells) when compared to those from naïve mice (Supplementary Fig. 6i, j). Furthermore, we quantified the levels of IL-13 and amphiregulin in the supernatants of stimulated ILC2s, isolated from the lungs of NA-treated mice; while the amounts of IL-13 were markedly upregulated, we were unable to detect significant changes in amphiregulin levels at several timepoints post NA (Fig. 9c). Together, these data suggest that ILC2s were activated early following epithelial damage and responded by producing type 2 cytokines. To demonstrate that ILC2s contribute to AAM activation post-injury, lung GFP⁺ILC2s were adoptively transferred into Rag2^−/−/Il2rγc^−/− mice, which lack T, B, NK cells and ILCs, 2 days prior to NA (Fig. 9d). Successful transfer was determined by the presence of GFP⁺ILC2s in recipient mice (Supplementary Fig. 7a, b). When compared to WT NA-treated animals, Rag2^−/−/Il2rγc^−/− mice exhibited an altered epithelial repair, i.e., a decrease in lung expression of Scgb1a1 mRNA (Fig. 9e). Concomitantly, we observed a significant decrease in total BAL cell numbers (Fig. 9f, g), which associated with an altered distribution of ST2⁺ macrophage populations within the lung. This was accompanied by an increase in P1 cell numbers and a significant reduction in ST2⁺ P2 and ST2⁺ resident P3 macrophages (Fig. 9f–h, Supplementary Fig. 7c). Furthermore, mice that lacked ILC2s had dramatically decreased lung levels of IL-13 (Fig. 9i, j) and lower amounts of the AAM-associated markers, BRP-39 and CCL17, and epithelium growth factors, HGF and IGF-1 (Fig. 9k, Supplementary Fig. 7d). Reconstitution of ILC2s to the lungs of Rag2^−/−/Il2rγc^−/− mice restored club cell regeneration and myeloid cell populations, including the number of ST2⁺ resident macrophages, to levels observed in C57BL/6 NA-treated mice, similarly IL-13 levels and growth factors associated with AAM and repair were also restored (Fig. 9e–k, Supplementary Fig. 7c, d). Lastly, we showed that BMDMs derived from Rag2^−/−/Il2rγc^−/− mice were able to differentiate into Arg-1⁺ CD206⁺ cells expressing ST2 (Supplementary Fig. 7e, f) and importantly produce significant levels of CCL17, IGF-1, and HGF in response to combined IL-33 and IL-13 stimulation (Supplementary Fig. 7g). Thus macrophages are functional in Rag2^−/−/Il2rγc^−/−, (ILC-deficient), mice and cytokine production in ILC2 recipient Rag2^−/−/Il2rγc^−/− animals was not due to contamination of B, T, or NK cells. Collectively, these findings highlight the importance of ILC2-derived IL-13 and the synergistic role which this cytokine plays with IL-33 in myeloid cell differentiation and effective macrophage activation into the AAM phenotype, both of which are essential for epithelial repair. Notably, the IL-33 pathway appears to be upstream from ILC2s, being a potent activator of myeloid cells and required for maintaining IL-13 production post NA injury.

a–c Quantification of the frequency of ST2⁺ GATA-3⁺ expressing ILCs (a) found in the lung at days 0, 2, and 5 after NA-treatment; N = naïve. ILCs were isolated from the lungs of mice on days 0, 2, 5, and 6 post-NA-treatment and stimulated ex vivo, then flow-stained for IL-13 production and quantified as frequency (b). IL-13 and amphiregulin production were quantified also in stimulated ILC supernatants (c). d Schematic for the transfer of lung ILC2s into Rag2^−/−/Il2rγc^−/− recipient mice followed by naphthalene administration (Rag2^−/−/Il2rγc^−/− + ILC2). e Levels of the Scgb1a1 mRNA in total lung homogenates in WT C57BL/6 or Rag2^−/−/Il2rγc^−/− mice that adoptively received GFP⁺ ILC2s (Rag2^−/−/Il2rγc^−/− + ILC2). f Representative flow cytometric plots of P1-P3 subsets of BAL cells from WT C57BL/6 or Rag2^−/−/Il2rγc^−/− mice that adoptively received GFP⁺ ILC2s (Rag2^−/−/Il2rγc^−/− + ILC2). Numbers nears gates denote percentage. g Quantification of total P1–P3 cells in BAL using the gating strategy in (f). h Quantification of the total number of ST2-expressing recruited macrophages (ST2⁺ P2 recruited macrophages) in the BAL. i Levels of IL-13 in lung homogenates. j, k Levels of IL-13 (j) and IGF-1 and HGF in BAL supernatants (k). Data from n = 15 (a), 6 (b), 3 to 9 (c), 4 (e), 5 (f–h), 6 (i, j), and 8 (k) mice, show mean ± SEM pooled from three independent experiments. ^***P < 0.001 and ^****P < 0.0001 between NA-treated and naïve (N) WT mice using one-way ANOVA, Bonferroni post-test. ^*P < 0.05, ^**P < 0.01, ^***P < 0.001 and ^****P < 0.0001 between NA-treated Rag2^−/−/Il2rγc^−/− and WT mice using one-way ANOVA, Bonferroni post-test (a–c). ^#P < 0.05, ^##P < 0.01 and ^###P < 0.001 between AAM adoptively transferred and NA-treated ST2^−/− mice using one-way (g–k) and two-way ANOVA (e), Bonferroni post-test.