Thrombospondin-1 promotes fibro-adipogenic stromal expansion and contractile dysfunction of the diaphragm in obesity

Abstract

Pulmonary disorders affect 40%-80% of individuals with obesity. Respiratory muscle dysfunction is linked to these conditions; however, its pathophysiology remains largely undefined. Mice subjected to diet-induced obesity (DIO) develop diaphragm muscle weakness. Increased intradiaphragmatic adiposity and extracellular matrix (ECM) content correlate with reductions in contractile force. Thrombospondin-1 (THBS1) is an obesity-associated matricellular protein linked with muscular damage in genetic myopathies. THBS1 induces proliferation of fibro-adipogenic progenitors (FAPs) - mesenchymal cells that differentiate into adipocytes and fibroblasts. We hypothesized that THBS1 drives FAP-mediated diaphragm remodeling and contractile dysfunction in DIO. We tested this by comparing the effects of dietary challenge on diaphragms of wild-type (WT) and Thbs1-knockout (Thbs1-/-) mice. Bulk and single-cell transcriptomics demonstrated DIO-induced stromal expansion in WT diaphragms. Diaphragm FAPs displayed upregulation of ECM and TGF-β-related expression signatures and augmentation of a Thy1-expressing subpopulation previously linked to type 2 diabetes. Despite similar weight gain, Thbs1-/- mice were protected from these transcriptomic changes and from obesity-induced increases in diaphragm adiposity and ECM deposition. Unlike WT controls, Thbs1-/- diaphragms maintained normal contractile force and motion after DIO challenge. THBS1 is therefore a necessary mediator of diaphragm stromal remodeling and contractile dysfunction in overnutrition and a potential therapeutic target in obesity-associated respiratory dysfunction.

Keywords: Extracellular matrix; Metabolism; Muscle biology; Obesity; Skeletal muscle.

Figure 1. The diaphragm FAP population is heterogeneous and altered by obesity.

(A) t-Distributed stochastic neighbor embedding (t-SNE) plot: Mononuclear cells from pooled diaphragms of 6-month high-fat diet–fed (HFD-fed) and age-matched control diet–fed (CD-fed) male C57BL/6J mice. n = 2 mice per group. (B) t-SNE plots: FAP marker genes. (C) t-SNE plots: FAP subpopulations from pooled CD and HFD samples. (D) Heatmap showing transcripts enriched in FAP subclusters. (E) Heatmap showing genes enriched in total FAP populations from CD and HFD samples. Bolded gene names are enriched in the FAP2 subcluster. (F) t-SNE and violin plots showing FAP subclusters and Thy1 expression in CD and HFD samples. (G) Violin plots indicating cluster-specific Thy1 expression. (H) Percentage of Thy1-expressing among FAPs from CD and HFD samples.

Figure 2. High-fat diet feeding causes THBS1-dependent FAP population expansion.

(A) Primary FACS-isolated FAPs treated with THBS1 (5 μg/mL) or DMEM vehicle (VEH) and subjected to Ki67 immunocytochemistry with phalloidin (PHAL) counterstain. Scale bar: 50 μm. Arrowheads indicate Ki67⁺ nuclei. The bar graph indicates percentage Ki67⁺ cells. n = 2 unique experiments per group with 4–7 replicates per experiment. (B) Primary FAPs treated as indicated in A and subjected to fibronectin (FN) immunocytochemistry. Representative images from 2 unique experiments with 3–4 replicates per experiment. Scale bar: 50 μm. (C) Analysis of FAPs from costal diaphragm tissue of wild-type (WT) and Thbs1^–/– (KO) mice fed a control diet (CD) or high-fat diet (HFD) for 6 months. Left panels show representative flow cytometry plots. FAPs are positive for Sca-1 and negative for CD31, CD45, and integrin α₇ (Intα7). Right panel shows bar graph quantifying FAPs/mg tissue. Each sample contains 2 whole costal diaphragms. n = 3–6 samples (6–12 mice) per group. (D) Immunohistochemistry for PDGFRα, with wheat germ agglutinin (WGA) counterstain, in diaphragm samples from WT and KO mice fed CD or HFD for 6 months. Scale bar: 100 μm in main panel, 50 μm in inset. Arrowheads in inset indicate FAPs, defined as PDGFRα staining surrounding a DAPI⁺ nucleus. Bar graph indicates the quantification of PDGFRα⁺ cells/mm² tissue cross-sectional area (CSA). n = 5–8 mice per group. Statistical analysis with t test for individual comparisons, 2-way ANOVA for multiple variable comparisons. Error bars indicate mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3. Thbs1 is required for obesity-induced FAP subpopulation shifts.

(A) Velocity plots demonstrating temporal relationships between FAP subpopulations in wild-type (WT) mice fed control diet (CD) or high-fat diet (HFD) for 6 months and Thbs1^–/– (KO) mice fed HFD for 6 months. Arrow directions represent the trajectory of differentiation between subpopulations. Thickness of arrow indicates a rate of change. Pie charts indicate percentage of Thy1-expressing cells in each group. Stacked bar graph shows proportion of individual FAP subpopulations in each group. n = 2 mice per group. (B) Heatmap indicating the expression of FAP2 marker genes in each group. (C) Heatmap showing genes specifically enriched in the WT HFD group. (D) Gene Ontology terms (Biological Processes) specifically enriched in WT HFD FAPs, with corrected P values.

Figure 4. Whole tissue transcriptomics highlights enrichment of stromal genes in wild-type versus Thbs1^–/– mice subjected to DIO.

(A) Volcano plot of whole costal diaphragm RNA-Seq demonstrating differentially expressed genes between wild-type (WT) and Thbs1^–/– (KO) mice fed high-fat diet (HFD) for 6 months (6m HFD). n = 3 mice per group. X axis indicates log fold-change (FC) in KO versus WT. Y axis indicates –log adjusted P value. (B) Heatmap integrating tissue-level RNA-Seq with scRNA-Seq data. Genes indicated are those enriched in WT and KO mice on bulk RNA-Seq (i.e., the points on the volcano plot in A). Cell types are those identified on scRNA-Seq (as shown in Figure 1A). Heatmaps show cell type–specific expression as defined on scRNA-Seq. (C) Enrichment plots demonstrating selected HALLMARK pathways differentially expressed between 6m HFD-fed WT and KO mice: epithelial mesenchymal transition (Hallmark EMT) and inflammatory response (Hallmark Inflamm resp). Heatmaps show the expression of leading-edge genes in individual samples. (D) qPCR analysis of selected genes performed on costal diaphragm tissue of 6m HFD WT and 6m HFD KO mice. n = 3–8 whole hemidiaphragm samples per group. Statistical analysis with t test. Error bars indicate mean ± SD. *P < 0.05, **P < 0.01.

Figure 5. Thbs1 ablation protects against diaphragm fibro-adipogenic remodeling.

(A) H&E-stained longitudinal diaphragm sections from wild-type mice fed control diet (WT CD) or HFD (WT HFD) for 6 months and Thbs1^–/– mice fed HFD for 6 months (KO HFD). White arrowhead indicates rib attachment point. Black arrowhead indicates central tendon attachment point. Scale bar: 600 μm. Representative samples from 5–7 mice per group. (B) Adipocyte size, adipocyte number/millimeter cross-sectional area (CSA), and percentage total CSA occupied by adipocytes in samples described in A. Values are the average of measurements made on 3 nonconsecutive 7 μm–thick sections encompassing the entire rib-to-tendon extent of muscle. n = 4–7 mice per group. Box indicates 25th–75th percentile, midline indicates median, and whiskers indicate minimum and maximum values. (C) Immunofluorescence staining of perilipin (PLN) and THY1 on adjacent 7 μm–thick longitudinal sections from animals described above. Representative images from analysis of 5–7 mice per group. Inset indicates THY1 staining of a nerve passing through the sample, representing an internal positive-staining control. Scale bar: 200 μm. (D) PLN and fibronectin (FN) staining on adjacent 7 μm–thick longitudinal sections from animals described above. Representative images from analysis of 5–7 mice per group. Scale bar: 200 μm. (E) Picrosirius red (SR) staining of 7 μm–thick longitudinal sections from animals described above. Bright-field (BF) and polarized light (POL) images: polymerized collagens fluoresce red/yellow under POL. Representative images from analysis of 5–7 mice per group. Scale bar: 200 μm. Statistical analysis with Kruskal-Wallis test for nonparametric multiple comparisons. *P < 0.05, **P < 0.01.

Figure 6. DIO challenge compromises diaphragm force and motion in wild-type but not Thbs1^–/– mice.

(A) Isometric specific force (Sp force) of wild-type (WT) and Thbs1^–/– (KO) mice (normalized to baseline, relative units) at baseline (0m) and following 6-month (6m) control diet (CD) or high-fat diet (HFD) feeding. n = 4–6 animals per group; 1–2 diaphragm strips per animal averaged. (B) Isometric specific force (absolute value) of samples from 6m HFD WT and KO mice. n = 6 animals per group; 1–2 diaphragm strips per animal averaged. (C) Correlation plot demonstrating the relationship between isometric specific force and percentage tissue cross-sectional area occupied by adipocytes in diaphragm strips subjected to isometric force testing. 6m HFD WT and KO mice, 8–9 individual muscle strips per group. (D) Image of single myofiber undergoing isometric force testing. White arrowheads indicate sutures affixing fiber to force transducer-servomotor apparatus. (E) Isometric specific force of single myofibers isolated from 6m HFD WT and KO mice (n = 4–5 animals per group; 4–5 fibers per animal). (F) Diaphragm ultrasound M-mode tracing with measured parameters labeled. X axis represents time; y axis represents displacement along the rostral-caudal axis. (G–I) Diaphragm motion parameters: amplitude (Amp), inspiratory velocity (Ins Vel), expiratory velocity (Exp Vel), normalized to baseline measured at 0, 2, 4, and 6 m. n = 8–9 animals per group. Statistical analysis with t test for individual comparisons and linear regression for correlational analysis. Error bars indicate mean ± SD. *P < 0.05, **P < 0.01.