Alveolar Macrophages were the Major Cell types in Patients with ACO
To better understand the pathogenesis of ACO, we analyzed the existing scRNA-Seq dataset generated from human lung tissues of a patient with ACO and two transplant donors without ACO34. By using the Leiden algorithm, an automated algorithm designed to effectively cluster cells in scRNA-seq data39, we identified a total of 12 cell clusters within dataset (Fig. 1, A). These cell clusters were further annotated as different cell types by using the SingleR algorithm with manual adjustments. Cell markers were these differentially expressed across different cell types identified by the FindAllMarkers function from the Seurat package36. A total of nine different cell types were identified within the dataset, which included monocytes/macrophages, T cells, NK cells, AT2 alveolar type II cells, endothelial cells, airway epithelial cells, B cells, fibroblasts, and mast cells (Fig. 1, B). The top-ranking differentially expressed genes for each cell type were presented in Fig. 1, C and Figure E1). For example, complement-related genes (C1QA, C1QB), CD68, and APOC1 were highly expressed in monocytes/macrophages (Figure E1, A), while genes like SFTPB, SFTPA1, and SFTPA2 were uniquely and significantly expressed in AT2 airway epithelial cells ( Figure E1, B). Tryptase genes (TPSAB1, TPSB2) were highly expressed in mast cells ( Figure E1, C). To identify the distribution of different cell types in the samples from ACO and control group, we performed cell proportion analysis as defined by relative percentages (Fig. 1, D). Of these, monocytes/macrophages were a predominant cell type, constituting more than 50% of the cells in the different groups being analyzed. Furthermore, the proportion of monocytes/macrophages was significantly higher in ACO patient compared to control group (p < 0.001). Next, we focused on those monocytes/macrophages specifically and re-clustered them by using the Leiden algorithm. A total of seven clusters were identified (Fig. 1, E). These cell clusters were further annotated as different sub-types of monocytes/macrophages (Fig. 1, F), including alveolar macrophages (AM), cycling cells, interstitial macrophages (IM), and monocytes. The top-ranking differentially expressed genes for each cell type were presented in Fig. 1, G and Figure E2). Similar to monocytes/macrophages, the subtype AMs express the complement-related genes C1QA, C1QB ( Figure E2, A). Additionally, several other genes RBP4, CD9, SERPING1, and CES1 are also significantly expressed in AMs. The proliferation markers PCLAF, TOP2A and MKI67 were highly expressed in cycling cells ( Figure E2, B). IMs were in an intermediate state in the UMAP plot and expressed high levels of molecules such as LGMN, RNASE1, and CCL2 ( Figure E2, C). While CFP, FCN1, and S100A8,were highly expressed in monocytes ( Figure E2, D). Taken together, the results suggest that monocytes/macrophages in ACO are major cells that may drive airway inflammation.
Decreased cellular Senescence in Monocytes/Macrophages of Patients with ACO
Cellular senescence has gained considerable attention across various diseases, including respiratory conditions such as asthma29. To determine whether senescence contributes to the development of ACO, we specifically assessed the enrichment of genes associated with senescence within lung monocytes/macrophages. Employing the AUCell algorithm, we computed enrichment scores against a senescence-related gene set known as SenMayo41. Among all the subtypes, AMs/monocytes showed the higher enrichment sore as assessed by the density of AUCell enriched SenMayo senescence scores (Fig. 2, A). Next, we investigated the distribution of ACO and control group among monocytes/macrophages (Fig. 2, B) and enrichment scores of SenMayo senescence in ACO and control group (Fig. 2, C). As illustrated in Fig. 2, C, the senescent signatures are enriched in the control group as compared to ACO group. Among them, AMs were a predominant cell type, constituting more than 50% of the cells among all these subtypes (Fig. 2, D). The proportion of AMs was significantly higher in ACO patients as compared to control group (p < 0.001). Additionally, cell cycle arrested in the G1 phase is one of the important features of cellular senescence60. We found that monocytes/macrophages that were in the G1 phase of the cell cycle were more abundant in the control group compared with ACO group (Fig. 2, E). These findings were further supported by the expression of cellular senescence markers, CDKN1A (p21) and CDKN2A (p16)61. The expression of CDKN1A was much lower in the ACO group as compared to the control group (Fig. 2, F). In contrast, no significant change was observed for CDKN2A (Fig. 2, G). These findings suggest a lower prevalence of cellular senescence within lung monocytes/macrophages among patients with ACO.
A lower prevalence of senescence observed in patients with severe asthma
We further validated the results through another independent cohort (IMSA)35 and analyzed the relationship between cellular senescence features and the severity of asthma. After stringent quality control, a total of 39 subjects were finally included for analysis: 6 healthy controls (HC), 17 mild/moderate asthma patients (MMA), and 16 severe asthma patients (SA). Definition of these subjects has been previously reported9. Consensus clustering analysis with the Consensus Cumulative Distribution Function (CDF) identified cellular senescence clusters for immune cells from IMSA. Consensus clustering analysis involves repeated clustering of the data with different values of k (from 2 to 5) within a dataset to identify stable and meaningful clusters (Fig. 3, A). Based on the consensus CDF (Fig. 3, B), it is evident that cellular senescence clusters 2 provides the most consistent and stable clustering solution for the IMSA dataset. According to the ssGSEA SenMayo Senescence score, the cluster 2 was divided into senescence clustering low and high group (P = 0.002, Fig. 3, C). Senescence clustering low group showed lower expression of senescence markers CDKN1A as compared to senescence high group (P = 0.011, Fig. 3, D). No difference was noted for CDKN2A between senescence clustering low and high group (P = 0.29, Fig. 3, E). In contrast, senescence low group showed higher expression of cell proliferation marker MKI67 (P = 0.0015, Fig. 3, F). Intriguingly, among the analyzed subjects consisting of HC, MMA, and SA participants, the large proportion of severe asthmatic patients was observed in senescence low group as compared to senescence high group (P < 0.05, Fig. 3, G). These findings indicate that cellular senescence may be negatively associated with the severity of asthma.
Airway CD206 + macrophages show a lower prevalence of senescence and increased cytokines
To further explore the relationship between cellular senescence and airway inflammation in asthma, we focused on the CyTOF dataset that targets lineage markers for the adaptive and innate immune systems35. Unsupervised cell clustering was performed in BAL fluids from HC, MMA, and SA participants using the FlowSOM algorithm62,63. Cell types in those clusters were determined by the surface marker staining intensities across t-SNE spaces. A total of 7 different cell clusters were annotated using a combination of surface marker genes, including B lymphocytes (CD19+), CD206− macrophages (CD11C+CD206−), CD206+ macrophages (CD11C+CD206−), CD4 T lymphocytes (CD3+CD4+), CD8 T lymphocytes (CD3+CD8+), γδ T lymphocytes (TCRγδ+), and NK cells (CD56+) (Fig. 4, A, B, and Figure E3). To identify the link between these airway immune cells and cellular senescence, the percentages of those identified cell populations were compared in senescence high and low groups. Among all these populations, a significant difference was found for CD206+ macrophages, which showed a higher proportion in senescence low in relative to senescence high group (Fig. 4, C). Interestingly, the senescence low group showed elevated levels of cytokines IL-5, IL-13, IL-10, and IL-22 in BAL fluids of the IMSA cohort (Fig. 4, D). Particularly, CD206+ macrophages showed increased expression of IL-4, IL-13, and IL-22 in senescence low group as compared those in senescence high group (Fig. 4, E). Taken together, these data suggest that CD206+macrophages are major airway immune cells associated with cellular senescence and cellular senescence is negatively correlated with cytokines in airway CD206+Macrophages.
Differentially expressed genes and pathways in Alveolar macrophages of ACO
Given the significance of CD206+macrophages in cellular senescence, we re-visited the ACO dataset and determined whether the specific subtype of monocytes/macrophages drive the difference in cellular senescence between ACO and control group. To explore this, we used trajectory analysis to infer the relationships and distances between different subtypes of monocytes/macrophages. Analyses were started with cycling cells, we identified a specific cluster that has the highest distance (pseudo-time) and is in the upper right corner of a two-dimensional scatter plot (Fig. 5, A). This cluster was mostly composed of cells from Leiden cluster 1 (Fig. 5, B) and was predominantly found in the control group (Fig. 5, C), suggesting that cells in Leiden cluster 1 may contribute to the difference in biological processes between ACO and control group. To further characterize the Leiden cluster 1, we employed the FindAllMarkers to investigate the highly expressed marker genes for cluster 1 against other cell clusters. A total of 47 genes were identified specifically for the Leiden cluster 1, such as SERPING1, CES1, HLA-DQA1, INHBA, FABP4, LGALS3BP, AKR1C3, CITED2, TERM1, FN1, EVL, ALDH1A1, and PDLIM1 (Fig. 5, D and see Table E2). Additionally, we checked the lineage markers that are commonly used for monocytes/macrophages (Fig. 5, E). we found that PPARγ, FCGR1A, FCGR3A, MSR1 (CD205), and MRC1 (CD206) were enriched in Leiden cluster 1. Furthermore, we used AUCell scoring to analyze the biological characteristics of Leiden cluster 1 in comparison to other clusters. Especially, the KEGG enrichment analysis (Fig. 5, F and see Table E3) indicated that Leiden cluster 1 had significant enrichments in several pathways, such as antigen processing and presentation, Peroxisome Proliferator-Activated Receptor (PPAR) signaling pathway. Next, we investigated the differences in cellular senescence between ACO and control groups specially in Leiden cluster 1. Consistent with our previous analyses among all the monocytes/macrophages, the SenMayo senescence score was remarkably lower in ACO groups compared with control group (Fig. 5, G). This was further supported by the expression of cellular senescence marker CDKN1A (Fig. 5, H), which also showed reduced expression in the ACO group. No statistical difference was observed for CDKN2A (Fig. 5, I). Additionally, to further display the characteristics of cellular senescence, we identified differentially expressed genes related to oxidation-reduction, cytokines, and growth factors between ACO and control groups in Leiden cluster 1. Of those, TGFB1 was increased but several chemokines such as CXCL2, CXCL3, CXCL8, and CCL18 were decreased in ACO group compared with control group (Fig. 5, J and see Table E4). These results suggest correlations between cellular senescence and cytokine expression in alveolar macrophages of Patients with ACO. Furthermore, strong correlations were identified for CDKN1A expression levels and those identified genes in Leiden cluster 1 and traditional lineage markers for monocytes/macrophages (Fig. 5, K and see Table E5 in this article’s Online Repository at www.jaci.org). For example, CDNK1A expression was correlated with genes with either positive regulation (e.g., CD14, CD68, CD86, CD163, FCGR3A, MSR1, SERPING1, HLA-DQA1, INHBA, LGALS3BP, AKR1C, TREM1, FN1, and ALDH1A1) or negative regulation (e.g., MARCO, FCGR1A, ITGB2, and MRC1). Collectively, the results indicate that these identified genes and biological pathways may be involved in regulating cellular senescence in alveolar macrophages of patients with ACO.
PPARγ, a Key Regulating Factor of Cellular Senescence in Alveolar Macrophages
To delve deeper into the underlying regulatory mechanisms of biological changes, we first utilized the hdWGCNA analysis focused on Leiden cluster 1 to explore the expression network associated with senescence signatures. The algorithm discerned six gene modules: green, blue, red, brown, cyan, and tan (Fig. 6, A). Within these, CDKN1A (p21) was pinpointed in the brown module, and the module feature gene score was positively correlated with the difference in cellular senescence between ACO and control groups (R = 0.669, p < 0.0001, Fig. 6, B). Prominently, the top 30 hub genes with high correlation in this module encompassed those related to growth arrest and proliferation (e.g., GADD45B, PPP1R15A, and DUSP2), TP53-mediated cell senescence-associated heat shock proteins (DNAJA1, DNAJB1, and HSPA5), DNA damage and repair (e.g., MYL12A), and energy, protein, and lipid metabolism (e.g., GLUL, PLIN2, RPS4Y1, PDIA3, and SOD2). Furthermore, genes associated with SASP and its upstream regulators (e.g., CCL20, CXCL3, CXCL5, CXCL8, JUN, NFKB1A, and NFKBIZ) and those tied to antigen presentation, antibody response, and macrophage activation (e.g., C83, CD68, FCGR3A, and B2M) were also included in these hub genes (Fig. 6, C). These genes or biological processes are directly or indirectly involved in senescence processes of alveolar macrophages. Subsequent enrichment analyses of entire module genes showed that the brown module was mainly associated with macrophage markers and cellular senescence (Fig. 6, D), cytokine signaling in immune system (Fig. 6, E), and antigen processing and presentation and PPAR signaling (Fig. 6, F). These biological characteristics are closely related and mutually regulated. To pinpoint the central regulator of senescence in alveolar macrophages, we leveraged SCENIC to discern the most specific transcription factors for each Leiden cluster. Notably, for Leiden cluster 1, PPARγ emerged as the predominant regulator (Fig. 6, G). This was further supported by the transcription factor enrichment analysis with the brown module against the TRRUST database. PPARγ was identified to be one of the top 10 pivotal transcription factors modulating gene expression within this module (Fig. 6, H). Intriguingly, each of these transcription factors had the potential to regulate CDKN1A expression. Taken together, these findings indicate that the brown module had the core impact on cellular senescence, and PPARγ is one of the predominant regulators modulating the senescent signature of alveolar macrophages.