Single-Cell RNA-Sequencing and Metabolomics Analyses Reveal the Contribution of Perivascular Adipose Tissue Stem Cells to Vascular Remodeling

Abstract

Objective: Perivascular adipose tissue (PVAT) plays a vital role in maintaining vascular homeostasis. However, most studies ascribed the function of PVAT in vascular remodeling to adipokines secreted by the perivascular adipocytes. Whether mesenchymal stem cells exist in PVAT and play a role in vascular regeneration remain unknown. Approach and Results: Single-cell RNA-sequencing allowed direct visualization of the heterogeneous PVAT-derived mesenchymal stem cells (PV-ADSCs) at a high resolution and revealed 2 distinct subpopulations, among which one featured signaling pathways crucial for smooth muscle differentiation. Pseudotime analysis of cultured PV-ADSCs unraveled their smooth muscle differentiation trajectory. Transplantation of cultured PV-ADSCs in mouse vein graft model suggested the contribution of PV-ADSCs to vascular remodeling through smooth muscle differentiation. Mechanistically, treatment with TGF-β1 (transforming growth factor β1) and transfection of microRNA (miR)-378a-3p mimics induced a similar metabolic reprogramming of PV-ADSCs, including upregulated mitochondrial potential and altered lipid levels, such as increased cholesterol and promoted smooth muscle differentiation.

Conclusions: Single-cell RNA-sequencing allows direct visualization of PV-ADSC heterogeneity at a single-cell level and uncovers 2 subpopulations with distinct signature genes and signaling pathways. The function of PVAT in vascular regeneration is partly attributed to PV-ADSCs and their differentiation towards smooth muscle lineage. Mechanistic study presents miR-378a-3p which is a potent regulator of metabolic reprogramming as a potential therapeutic target for vascular regeneration.

Keywords: adipose tissue; homeostasis; regeneration; stem cells; vascular remodeling.

Figure 1.

Characterization of perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs). A and B, Representative histograms of flow cytometry analysis of cultured PV-ADSCs and the percentage of indicated phenotypic markers (n=3). Gating is set for the IgG control to be between 0.5% and 1%. C, Immunofluorescent staining of the adipose tissue surrounding the mouse aorta with CD29, Sca-1 (stem cell antigen 1), PECAM1 (platelet and endothelial cell adhesion molecule 1), PLIN-1 (perilipin-1), and 4′,6-diamidino-2-phenylindole (DAPI; n=3). Arrows indicate the CD29⁺/Sca1⁺/PECAM1⁻/PLIN1⁻ cells. The border between End (endothelium), Med (media), Adv (adventitia), and perivascular adipose tissue (PVAT) is drawn with dashed line. Scale bar, 50 μm. Ctrl indicates IgG isotype control.

Figure 2.

Single-cell RNA-sequencing (scRNA-seq) reveals 2 distinct clusters in perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs). A, Sorting gate for scRNA-seq was set as live nucleated cells (Syto16⁺/DAPI⁻), single cells, CD45⁻/CDH5⁻ cells and CD29⁺/Sca1⁺ cells. B, Scatter plot obtained from multidimensional scaling (MDS) analysis showed 2 distinct clusters. Color scale, log2(gene counts). C, Expression of representative genes for cluster 1 (Cd36) and cluster 2 (Tgfbr2). D, Heatmap of marker genes in each cluster. Color scale, log2(gene counts). Full list of genes was in Table I in the online-only Data Supplement. E and F, Gene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis of the top 200 genes (by P value) upregulated in Cluster 1 (E) and Cluster 2 (F). Full list of GO terms and KEGG pathways was in Table II in the online-only Data Supplement. G, Enrichment of gene sets including Myogenesis (Molecular Signature) and TGF (transforming growth factor)-β signaling (KEGG). H, Expression heatmap of frequently used mesenchymal stem/stromal cell (MSC) markers. Color scale, log2(gene counts). I, Expression of frequently used MSC markers in the 2 clusters was shown as violin plot. Cl 1 indicates cluster 1; Cl 2, cluster 2; DAPI, 4′,6-diamidino-2-phenylindole; FDR, false discovery rate; GOMF, gene ontology molecular function; NES, normalized enrichment score; and PPAR-γ, peroxisome proliferator-activated receptor-γ.

Figure 3.

Pseudotime trajectory of cultured perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs). A, Cell stages expressing early to late smooth muscle cell (SMC) markers. B, Pseudotime trajectory of cultured PV-ADSCs with DDRTree method for dimension reduction. Color scale, pseudotime. C, Cell ordering from different differentiation stages along the pseudotime trajectory. D, Heatmap of the top 1000 (by Q value) significantly changed genes in 3 gene modules. E, Expression of Ly6a, Cnn1, and Myh11 along the trajectory. Color scale, log gene expression. F, Gene ontology (molecular function) analysis of each gene module from (D). Full list was in Table III in the online-only Data Supplement. G, Expression of significantly changed genes from TGF (transforming growth factor)-β signaling pathway (KEGG) along pseudotime. Full list was in Table I in the online-only Data Supplement. Color scale, log gene expression. H, Tgfb2 expression and total level of significantly changed genes from TGF-β signaling along the pseudotime trajectory. Color scale, log gene expression. I, Expression of significantly changed transcription factors along the pseudotime trajectory. Full list was in Table I in the online-only Data Supplement. Color scale, log gene expression.

Figure 4.

Perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs) differentiate towards smooth muscle cells (SMCs) in vivo and in vitro. A, H&E staining of the vein graft harvested 4 wk after the transplantation. Neointimal area was calculated against injury-only control (n=4). Arrow indicates the boundary between neointima and media layer. B, Vein grafts harvested 1 wk after the transplantation were stained with RFP (red fluorescent protein; red) and SMC marker ACTA2 (smooth muscle α actin; green) (n=4). Arrows indicate cells positive for both markers. Dashed line indicates boundary of adventitia and neointima. Scale bar, 20 μm. C–E, In vitro cultured PV-ADSCs were differentiated towards SMCs in differentiation medium for 5 d (diff) and then harvested for quantitative polymerase chain reaction (qPCR; C; n=4), Western blot (D; n=3) and immunofluorescent staining (E; n=3) for SMC markers. Cells cultured without TGF (transforming growth factor)-β1 (ctrl) served as control. Scale bar, 20 μm. F, Chemotaxis of PV-ADSCs in response to an increasing gradient of SDF-1 (stromal cell-derived factor 1) in an 8.0-μm transwell system was identified by applying 1% crystal violet staining after 16-h incubation (n=3). Scale bar, 50 μm. G, H&E staining of subcutaneous Matrigel plug containing PV-ADSCs-derived SMCs and endothelial cells (ECs; diff) in comparison with control group (ctrl) (n=4). H, Immunostaining of the tube-like structure in Matrigel plug with ACTA2 and PECAM1 (platelet and endothelial cell adhesion molecule 1; n=4). Scale bar, 50 μm. Data are presented as mean±SD. *P<0.05, **P<0.01. Adv indicates adventitia. CNN1 indicates calponin; DAPI, 4′,6-diamidino-2-phenylindole; and TAGLN, smooth muscle protein 22-α.

Figure 5.

Metabolic reprogramming of perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs) during smooth muscle cells (SMC) differentiation. A and B, Cellular metabolite abundance was detected with nuclear magnetic resonance (NMR) system in undifferentiated PV-ADSCs and PV-ADSCs differentiated for 1 d (n=8). A, Heatmap for metabolite levels. B, Orthogonal projection to latent structures. C, The oxygen consumption rate (OCR) measured in PV-ADSCs cultured with or without TGF (transforming growth factor)-β1 for 1 d (n=3). D, Live PV-ADSCs cultured in differentiation medium or basal growth medium for 1 and 4 d were stained tetramethylrhodamine, methyl ester, perchlorate (TMRM) to detect mitochondrial potential (n=3). E and F, Cellular metabolite abundance in undifferentiated PV-ADSCs or PV-ADSCs differentiated for 4 days was determined with gas chromatography-mass spectrometry (GC-MS) system (n=3). Heatmaps of metabolites involved in tricarboxylic acid (TCA) cycle (E) or lipid metabolism (F) were shown. G, Gene expression of genes important in lipid metabolism in PV-ADSCs treated with or without TGF-β1 for 2 d (n=4). H, Expression of Cebpb along the pseudotime trajectory. Color scale, log expression of gene. I, Crat mRNA level in PV-ADSCs treated with or without TGF-β1 for 2 d (n=4). Data are presented as mean±SD. *P<0.05, **P<0.01, and ***P<0.001. Anti-A/Rot indicates anti-mycin A/Rotenone; FCCP, carbonyl cyanide-4-phenylhydrazone; MFI, mean fluorescence intensity; and SRC, spare respiratory consumption.

Figure 6.

Crat knockdown promotes smooth muscle cells (SMC) differentiation via upregulation of TGF (transforming growth factor)-β1. A–C, Perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs) treated with Crat siRNA with or without TGF-β1 for 2 d. A, Crat level was significantly downregulated after siRNA Crat transfection. Quantitative polymerase chain reaction (qPCR) also showed the induction of SMC markers (Cnn1, Tagln, and Acta2) with Crat knockdown. B, Protein level induction of SMC markers by Crat siRNA was demonstrated by Western blot. C, Immunostaining and statistical analysis showed the induction of SMC markers by Crat siRNA in medium without TGF-β1. D–F, TGF-β1 level after Crat siRNA transfection for 2 d was determined with qPCR (D), Western blot (E), and immunofluorescent staining (F). Results are cumulative (A and D) or representative (B, C, E, and F) of 3 independent experiments. Scale bar, 30 μm. Data are presented as mean±SD. *P<0.05, **P<0.01, and ***P<0.001. ACTA2 indicates smooth muscle α actin; CNN1, calponin; DAPI, 4′,6-diamidino-2-phenylindole; and TAGLN, smooth muscle protein 22-α.

Figure 7.

MicroRNA (miR)-378a-3p induces metabolic reprogramming and promotes smooth muscle cells (SMC) differentiation. A, The oxygen consumption rate (OCR) of perivascular adipose tissue-derived mesenchymal stem cells (PV-ADSCs) cultured in miR-378a-3p mimics or control for 2 d (n=3). B, Live PV-ADSCs treated with miR-378a-3p mimic for 2 d with or without TGF (transforming growth factor)-β1 were stained tetramethylrhodamine, methyl ester, perchlorate (TMRM) and analyzed for mean fluorescence intensity (MFI) with flow cytometry (n=3). C and D, Cellular metabolite abundance in PV-ADSCs treated with miR-378a-3p mimics for 1 d determined with gas chromatography-mass spectrometry (GC-MS) system (n=3). Heatmap of metabolites in tricarboxylic acid (TCA) cycle (C) or lipid metabolism (D). E, SMC marker mRNAs were induced by miR-378a-3p mimics after 2 d (n=3). F, Representative immunofluorescent staining (n=3) showed the upregulation of SMC markers by miR-378a-3p mimic treatment. G, Transfection of PV-ADSCs with miR-378a-3p inhibitor inhibited the level of SMC markers as shown by quantitative polymerase chain reaction (qPCR). The level of Crat was derepressed by miR-378a-3p inhibitor transfection (n=3). H, OCR and extracellular acidification rate (ECAR) were detected with Seahorse Mito stress tests in wild-type (WT) and miR-378a knockout (KO) ADSCs (n=3). I, WT and miR-378a KO ADSCs were treated with or without TGF-β1 for 2 d. Immunofluorescent staining showed the protein level of CNN1 (calponin; n=1). Scale bar, 100 µm. Data are mean±SD. *P<0.05 and **P<0.01. Anti-A/Rot indicates anti-mycin A/Rotenone; Crat, carnitine acetyltransferase; FCCP, carbonyl cyanide-4-phenylhydrazone; inh 378, miR-378a-3p inhibitor; inh ctrl, inhibitor control; mim 378, miR-378a-3p mimics; mim ctrl, mimics control; and SRC, spare respiratory capacity.