Novel Cell-to-Cell Communications Between Macrophages and Fibroblasts Regulate Obesity-Induced Adipose Tissue Fibrosis

Abstract

Recent evidence has shown that adipose tissue eventually develops fibrosis through complex cellular cross talk. Although advances in single-cell transcriptomics have provided new insights into cell diversity during this process, little is known about the interactions among the distinct cell types. In this study, we used single-cell analytical approaches to investigate cell-to-cell communications between macrophages and fibroblasts in the adipose tissue of diet-induced obese mice. Spatial transcriptomics was used to understand local cellular interaction within crown-like structures (CLS), a characteristic histological feature of adipose tissue in obesity driving inflammation and fibrosis. Macrophages and fibroblasts were divided into several subclusters that appeared to interact more intensely and complexly with the degree of obesity. Besides previously reported lipid-associated macrophages (LAMs), we found a small subcluster expressing macrophage-inducible C-type lectin (Mincle), specifically localizing to CLS. Mincle signaling increased the expression of oncostatin M (Osm), suppressing collagen gene expression in adipose tissue fibroblasts. Consistent with these findings, Osm deficiency in immune cells enhanced obesity-induced adipose tissue fibrosis in vivo. Moreover, OSM expression was positively correlated with MINCLE expression in human adipose tissue during obesity. Our results suggest that Osm secreted by Mincle-expressing macrophages is involved in dynamic adipose tissue remodeling in the proximity of CLS.

Article highlights: Adipose tissue fibrosis is a complex and dynamic process that involves many cell types, such as macrophages and fibroblasts. Crown-like structures, which drive inflammation and fibrosis in obesity, are excellent targets for single-cell and spatial transcriptomics. We found novel cell-to-cell communications between macrophages and fibroblasts in adipose tissue from diet-induced obese mice, particularly during the fibrotic phase. We elucidated the role of the macrophage-inducible C-type lectin-oncostatin M axis in obesity-induced adipose tissue fibrosis.

Figure 1

Morphological features of Mincle-expressing macrophages. A: Experimental protocol for preparation of Mincle-expressing macrophages from epididymal adipose tissue of wild-type mice fed the HFD for 11 weeks. Representative FACS plots showing the gating strategy for CD45-positive, F4/80-positive, and Mincle-positive macrophages. Mincle-positive macrophages were determined using isotype controls. APC, allophycocyanin; PE, phycoerythrin. B: Percentage of Mincle-expressing macrophages (red) among living cells. Green and gray indicate non–Mincle-expressing macrophages and other cell types, respectively. Values are mean ± SEM (n = 9). C: Giemsa staining of sorted macrophages. As control cells, CD45-positive and F4/80-positive macrophages were sorted from epididymal adipose tissue of wild-type mice fed the SD. Scale bars, 50 μm. D: Cell adhesion rate of sorted macrophages was calculated as the number of cells attached to other cells per total cells. Values are mean ± SEM (n = 4). **P < 0.01. E–H: Electron microscopy of adipose tissue macrophages. Scale bars, 1 μm. E and F: Mincle-expressing macrophages in epididymal adipose tissue from HFD-fed obese mice. E: Arrowheads indicate adhesion sites. F: Red shading indicates exosome budding sites. Yellow shading indicates endosomes. Purple shading indicates sites in the process of endocytosis. G: Adipose tissue macrophages from SD-fed lean mice. H: Non–Mincle-expressing macrophages from HFD-fed obese mice.

Figure 2

Macrophage and fibroblast subclusters associated with obesity-induced adipose tissue fibrosis. A: mRNA expression in epididymal adipose tissue from wild-type mice fed the SD or HFD for 8 weeks (HFD8, inflammatory phase) or 16 weeks (HFD16, fibrotic phase). Values are mean ± SEM (n = 4–5). B: Sample preparation for scRNA-seq analysis. SVF was obtained from epididymal adipose tissue of lean and obese mice as indicated (n = 3 mice/group). C: Two-dimensional FItSNE representation of 33 clusters of SVF. Each number indicates the respective cell cluster. D: FItSNE representation of macrophages and fibroblasts. E: Violin plots showing expression of genes characterizing macrophage and fibroblast clusters. F: The cell number of macrophages, fibroblasts, and others. Values are mean ± SEM (n = 3 mice/group). The data of macrophages were analyzed by ANOVA, followed by the Tukey test. **P < 0.01 vs. SD. G: FItSNE representation of 11 subclusters of macrophages. Each number indicates the respective cell cluster. H: Changes in macrophage cell composition. Values are mean ± SEM. I: FItSNE representation of seven subclusters of fibroblasts. Each number indicates the respective cell cluster. J: Changes in fibroblast cell composition. Values are mean ± SEM (n = 3 mice/group). The data of fibroblast subcluster 3 were analyzed by ANOVA, followed by the Tukey test. *P < 0.05 vs. SD. K: Heat map of top two differentially expressed genes defining the subclusters indicated in panel I. L: Specific IPA pathways in fibroblast subcluster 3. The z score represents the activation (orange) or inhibition (blue) state of each canonical pathway.

Figure 3

Potential cellular interaction between Mincle-expressing macrophages and fibrosis-associated fibroblasts. A: CellChat analysis of macrophages and fibroblasts. Circle plots displaying communication networks between macrophages and fibroblasts, with the width of edges representing the strength of the communication. HFD8, HFD for 8 weeks; HFD16, HFD for 16 weeks. B: Violin plot showing the expression levels of the cells expressing Mincle (Clec4e). C: Dot plots showing the expression levels and percentages of the cells expressing genes defining macrophage subclusters 4 and 10. Color intensities and dot sizes indicate expression levels and percentages, respectively, of the cells expressing the indicated genes. D: Heat map showing the summary of the signaling pathways identified to the outgoing (ligand) and incoming (receptor) signals between macrophages and fibroblasts from wild-type mice fed the HFD for 16 weeks (fibrotic phase). The color bar represents the relative signaling strength of a signaling pathway across cell types. The signaling pathways indicated by the triangles are those we focused on. E: Significant ligand-receptor pairs sending signals from macrophage subclusters 4 or 10 to fibroblast subcluster 3 in wild-type mice fed the HFD for 16 weeks (fibrotic phase). Color intensities and dot sizes indicate communication probability and P value, respectively. F: FItSNE representation of the genes identified as potentially involved in the CellChat analysis.

Figure 4

Spatial transcriptomic analysis of obesity-induced adipose tissue inflammation and fibrosis. A: UMAP plots of the Visium spot transcriptome clusters showing eight clusters of epididymal adipose tissue from wild-type mice fed the SD (lean) or HFD for 36 weeks (obese, fibrotic phase). B: Spatial visualization of each cluster. The green dotted rectangles show the tissue images, including blood vessels. The light blue solid rectangle shows the histological image including CLS). Arrowheads indicate CLS. Scale bars, 1 mm. C: Violin plots showing the expression levels of the genes that characterize each Visium spot. UMAP plots of the Visium spot (D) and spatial visualization of each cluster (E) excluding clusters 6 and 7 that were histologically rich in blood vessels and expressed high levels of vascular endothelial markers. APC, allophycocyanin; PE, phycoerythrin; SSC, side scatter. Scale bars, 1 mm. F: Dot plots showing the expression levels and percentages of the cells expressing Mincle and the genes identified as potentially involved in the CellChat analysis. Genes highly expressed in macrophages and fibroblasts are shown in red and green, respectively. Color intensities and dot sizes indicate expression levels and percentages, respectively, of the cells expressing the indicated genes. G: Representative images of Mincle (cyan) and Acta2 (magenta) in situ hybridization in epididymal adipose tissue from HFD-fed obese mice (fibrotic phase). Nuclei were visualized by DAPI staining (blue). Scale bars, 100 μm.

Figure 5

Mincle-expressing adipose tissue macrophages express Osm in diet-induced obesity. A: Expression of Mincle and the genes identified as potentially involved in the CellChat analysis. Peritoneal macrophages were stimulated with TDM, an exogenous Mincle ligand, for 24 h. Veh, vehicle. Values are mean ± SEM (n = 3). B: Expression of Osm. RAW264, peritoneal macrophages, and bone marrow–derived macrophages (BMDMs) were treated with TDM for 8 h. Values are mean ± SEM (n = 3). C: Osm and its receptor Osmr expression in adipocytes and SVF from epididymal adipose tissue of lean wild-type mice. Values are mean ± SEM (n = 3). D: FACS gating for preparation of CD45(−) Col1a2/EGFP(−) (including adipocyte progenitor cells, endothelial cells, etc.), CD45(−) Col1a2/EGFP(+) (including activated fibroblasts), and CD45(+) F4/80(+) (including macrophages) cells from epididymal adipose tissue of lean Col1a2/EGFP-transgenic mice, which express EGFP exclusively in collagen I–producing cells (left). mRNA expression in each cell type (right). Values are mean ± SEM (n = 4). E: Expression of Mincle and Osm in Mincle-expressing and nonexpressing macrophages from epididymal adipose tissue of HFD-fed obese mice. Values are mean ± SEM (n = 5). F: Representative F4/80 and Osm immunostaining in serial sections of epididymal adipose tissue from HFD-fed obese mice (fibrotic phase). *Indicates CLS. Scale bars, 50 μm. G: Representative F4/80 (green) and Osmr (red) immunostaining in epididymal adipose tissue from HFD-fed obese mice (fibrotic phase). Nuclei were visualized by DAPI staining (blue). Images were acquired in Z-stack mode (7.45-μm thickness) and projected into two dimensions using maximum intensity projection. *Indicates CLS in merged image. Scale bars, 50 μm. H: Correlation analysis of gene expressions of MINCLE and OSM in human subcutaneous adipose tissue (n = 43). I: Correlation analysis of gene expression of MINCLE and OSM in human visceral adipose tissue (n = 8, Gene Expression Omnibus DataSet: GSE235696). RPM, reads per million mapped reads, represents normalized gene expression data. *P < 0.05; **P < 0.01. Spearman correlation coefficient (ρ) and P values are shown on the plot (H and I).

Figure 6

Mincle-Osm axis regulates ECM remodeling in adipose tissue fibroblasts. A–G: Effect of Osm on expression of the genes related to fibrosis in cultured fibroblasts. Values are mean ± SEM (n = 3 to 6). A: Experimental protocol. The SVF was prepared from epididymal adipose tissue of HFD-fed obese mice (fibrotic phase), then divided into CD45/CD31/Ter119-negative and remaining fractions by a magnetic cell sorting system (MACS). APC, allophycocyanin. B: FACS analysis of the cells divided by MACS. C–G: mRNA expression of genes related to collagen synthesis, degradation, and cross-linking. Veh, vehicle. Adipose tissue–derived fibroblasts (C, E, and G) and mouse embryonic fibroblasts (MEFs) (D and F) were treated with 3 ng/mL recombinant TGF-β1 (C, D, G) or Osm (E, F, G) for 24 h. H–K: Conditioned media supplementation experiments. Values are mean ± SEM (n = 3–6). H: Experimental protocol. Adipose tissue–derived fibroblasts were cultured in conditioned media from TDM-stimulated RAW264 macrophages. Mincle-stimulated mRNA expression (I) and secretion (J) of Osm in RAW264 macrophages. Cells were treated with TDM for 8 or 24 h. K: mRNA expression of genes related to collagen synthesis, degradation, and cross-linking. Adipose tissue-derived fibroblasts were cultured with conditioned media from TDM-stimulated RAW264 macrophages for 24 h.*P < 0.05;**P < 0.01; #P < 0.01 vs. Osm 3 ng/mL (G).

Figure 7

Macrophage-derived Osm suppresses obesity-induced adipose tissue fibrosis. Bone marrow transplantation experiments using Osm-deficient (knockout [KO]) mice. Values are mean ± SEM (n = 13–15). A: Experimental protocol. The lethally irradiated wild-type (WT) mice received bone marrow cells from WT or KO mice fed the HFD for 24 weeks (fibrotic phase). B: Body, adipose tissue, and liver weights. C: Fasting blood glucose, serum triglyceride (TG), total cholesterol, AST, and ALT concentrations. D: Serum insulin concentrations. E: mRNA expression of genes related to inflammation, fibrogenesis, and adipogenesis in epididymal adipose tissue. Representative images of F4/80 immunostaining (F), Sirius red staining (G), and collagen III immunostaining (H) in epididymal adipose tissue and their quantitative evaluations. Scale bars, 100 μm. I: Histogram and average adipocyte diameters in the epididymal adipose tissue. *P < 0.05; **P < 0.01; n.s., not significant.

Figure 8

Mincle deficiency induces collagen expression in fibrosis-associated fibroblasts during obesity. A: Sample preparation for scRNA-seq analysis. The SVF was obtained from epididymal adipose tissue from wild-type (WT) or Mincle-deficient (Mincle knockout [KO]) mice fed the HFD for 15 weeks (fibrotic phase) (n = 2 mice/group). B: UMAP plot showing the clustering of each genotype. C: UMAP plot (left) showing 11 subclusters of macrophages, based on label transfer from the data set in Fig. 2G. Each number indicates the respective cell cluster. Changes in macrophage cell composition are shown in the graph (right). D: UMAP plot (left) showing seven subclusters of fibroblasts, based on label transfer from the data set in Fig. 2I. Each number indicates the respective cell cluster. Changes in fibroblast cell composition are shown in the graph (right). E: DESeq2 results of pseudo-bulk analysis of fibroblast subcluster 3 across two genotypes for genes associated with collagen synthesis, degradation, and cross-linking. The indicated genes with P values <0.05 of Mincle KO relative to WT are shown in red. FC, fold change.