CD157+ vascular endothelial stem cells represent a conserved subpopulation with an angiogenic gene expression profile Authors: Tomohiro Iba, Taku Wakabayashi, Rie Ito, Ai Sugawara, Satoshi Fujimura, Mika Sawane, Kazuaki Yoshioka, Aya Matsui, Jun-ichi Morishige, Naoto Nagata, Yukinobu Ito, Masafumi Horie, Daichi Maeda, Rica Tanaka, Hitoshi Ando, Nobuyuki Takakura, Hisamichi Naito Published: 2026 ABSTRACT SummaryEndothelial cells expressing the CD157 antigen (CD157+ ECs) contribute to vascular regeneration and maintenance in adult tissues, but their molecular identity is not fully defined. Here, we show that CD157-positive ECs in mouse and human tissues share conserved transcriptional profiles enriched for angiogenesis-associated genes. Regulon analysis revealed a gene-regulatory network in which the NFAT pathway contributes to vascular network formation. Integration of mouse and human scRNA-seq datasets revealed human EC clusters with gene expression profiles resembling mouse CD157-positive ECs. The clusters were localized in the large vessel intima and expressed known stem-like EC markers such as BST1 (CD157), PROCR (CD201), and ABCG2. Functionally, human CD157+ ECs isolated from muscle exhibited greater proliferative capacity than CD157− ECs. Cell-cell interaction analysis suggested active communication between CD157-positive ECs and surrounding cells, via the CXCL12-CXCR7 axis. Our findings identify a conserved gene signature for CD157+ ECs with potential relevance for vascular regeneration. Introduction Blood vessels form an essential systemic network that maintains homeostasis through the delivery of oxygen and nutrients to tissues and the removal of metabolic waste. Endothelial cells (ECs), which line the inner surface of blood vessels, play an essential role in these functions. Under hypoxic conditions, ECs become activated and initiate angiogenesis (the formation of new blood vessels from pre-existing ones) to restore oxygen supply (Potente et al., 2011). This process is tightly regulated by signaling pathways such as VEGF and Notch and involves dynamic changes in EC states (Blanco and Gerhardt, 2013). Several research groups, including ours, have recently proposed that peripheral blood vessels may contain EC stem/progenitor cell populations that contribute to angiogenesis and vascular regeneration, particularly following severe endothelial injury (Chambers et al., 2021; Naito et al., 2020a). We previously identified CD157 (Bst1) as a cell surface marker that distinguishes stem cell-like ECs (also known as vascular endothelial stem cells [VESCs]) from other ECs (Wakabayashi et al., 2018). CD157, a glycosylphosphatidylinositol-anchored membrane protein with ADP-ribosyl cyclase activity, is involved in NAD+ metabolism, generating cyclic ADP-ribose and releasing nicotinamide. It is mainly expressed in bone marrow stromal and certain myeloid cell populations and has functions in cell adhesion, migration, and immune regulation (Ortolan et al., 2019). However, its expression in vascular ECs is not widely recognized. Paneth cells in the intestinal stem cell niche also express CD157, suggesting a potential role in regulating stem cell microenvironments (Yilmaz et al., 2012). Nevertheless, the precise characteristics of CD157+ VESCs and the extent of their conservation across different organs remain unclear, and are even less well understood in humans. One approach that may provide a deeper understanding of the VESC characteristics is single-cell RNA sequencing (scRNA-seq). In vascular biology, scRNA-seq is now widely used to analyze endothelial heterogeneity under physiological and pathological conditions (Kalucka et al., 2020; Pan et al., 2024). Yet, few scRNA-seq analyses have focused specifically on EC stem/progenitor populations. Here, we performed scRNA-seq analyses of ECs isolated from adult mouse liver, skeletal muscle, and lung, as well as from adult human skeletal muscle, focusing on CD157-positive ECs. In mice, CD157-positive ECs were enriched in a venous-like cluster and shared a gene expression signature associated with angiogenesis. Regulon analysis suggested involvement of the nuclear factor of activated T-cells (NFAT) pathway. Fresh human lower limb muscle samples showed an EC subcluster with a gene expression profile similar to that of mouse CD157+ ECs. Human CD157+ ECs had a higher proliferative capacity and were primarily localized to large vessels. Integration of publicly available scRNA-seq datasets suggested conserved intercellular communication patterns between CD157-positive ECs and neighboring cells. These findings suggest that CD157+ ECs share common transcriptional and functional features across tissues and species and may localize to specific vascular niches. Results CD157-positive ECs reside in a specific cluster with conserved gene expression across tissues in mice We investigated the gene expression profiles of CD157+ ECs in different tissues using fluorescence-activated cell sorting (FACS) to isolate CD31+CD45− EC fractions from mouse liver, hindlimb muscle, and lung, followed by scRNA-seq analysis of these populations (Figures 1A and 1B). Vascular ECs were distinguished from contaminating cells by the expression of well-established cell type markers. Cells expressing Cdh5, Pecam1 (CD31), and Emcn, and negative for Ptprc (CD45) and Prox1 were classified as vascular ECs (Trimm and Red-Horse, 2023; Wakabayashi and Naito, 2023).Figure 1Single-cell RNA-seq identifies CD157-positive ECs in the mouse vasculature of liver, muscle, and lung(A) Schematic diagram showing the experimental workflow.(B) Representative flow cytometry dot plots of cells obtained from mouse liver, hindlimb muscle, and lung. The red gate indicates CD31+CD45− cells isolated for single-cell analysis.(C) UMAP plots of scRNA-seq data from sorted cells isolated from mouse liver (12,972 cells), hindlimb muscle (11,989 cells), and lung (7,911 cells).(D) Bubble plots comparing the expression levels of selected signature genes across clusters of CD31+CD45− cells.(E) Feature plots of tissue-specific EC markers projected onto UMAP plots of CD31+CD45− sorted cells.(F) Immunofluorescence images of the mouse liver, hindlimb muscle, and lung stained for CD31 (green) and CD157 (red). Single-channel images are shown on the left. Scale bars, 100 μm. Single-cell RNA-seq identifies CD157-positive ECs in the mouse vasculature of liver, muscle, and lung (A) Schematic diagram showing the experimental workflow. (B) Representative flow cytometry dot plots of cells obtained from mouse liver, hindlimb muscle, and lung. The red gate indicates CD31+CD45− cells isolated for single-cell analysis. (C) UMAP plots of scRNA-seq data from sorted cells isolated from mouse liver (12,972 cells), hindlimb muscle (11,989 cells), and lung (7,911 cells). (D) Bubble plots comparing the expression levels of selected signature genes across clusters of CD31+CD45− cells. (E) Feature plots of tissue-specific EC markers projected onto UMAP plots of CD31+CD45− sorted cells. (F) Immunofluorescence images of the mouse liver, hindlimb muscle, and lung stained for CD31 (green) and CD157 (red). Single-channel images are shown on the left. Scale bars, 100 μm. In the liver, 12,972 cells were initially detected. Lymphatic ECs (Prox1 and Pdpn), pericyte/smooth muscle cells (SMCs) (Pdgfrb and Acta2), cholangiocytes (Krt8), and hepatocytes (Alb) were excluded from further analysis. As a result, 11,739 liver ECs were analyzed, and unbiased clustering identified nine clusters. We validated the cluster identity of vascular ECs using canonical marker genes and the top 10 differentially expressed genes (DEGs) and classified the vascular ECs into two portal vein/hepatic artery EC clusters (clusters 0 and 1 marked by Cd34, Sox17, Gja5, and Ly6a), one zone 1 liver sinusoidal EC cluster (cluster 2 marked by Ntn4 and Msr1), three zone 2 liver sinusoidal EC clusters (clusters 3, 4, and 5 marked by Stab1, Stab2, and Mrc1), one zone 3 liver sinusoidal EC cluster (cluster 6 marked by Cdh13, Thbd, and Fabp4), one central vein EC cluster (cluster 7 marked by Wnt2, Wnt9b, and Rspo3), and one proliferating EC cluster (cluster 8 marked by Mki67 and Top2a) (Figures 1C, 1D, and S1A) (Su et al., 2021). In the hindlimb muscle, 11,989 cells were initially detected. Lymphatic ECs (Prox1 and Pdpn), pericytes/SMCs (Acta2 and Pdgfrb), fibroblasts (Fn1 and Col1a1), muscle cells (Actn3, Des, and Tnnc2), and Schwann cells (Sox10, Mpz, and Mbp) were removed, leaving 9,220 ECs for analysis. Unbiased clustering of vascular ECs identified seven clusters classified into one arterial EC cluster (cluster 0 marked by Efnb2, Gja4, and Gja5), five capillary EC clusters (clusters 1, 2, 3, 4, and 5 marked by Rgcc and Sparc), and one venous EC cluster (cluster 6 marked by Nr2f2, Ephb4, and Ackr1) (Figures 1C, 1D, and S1B). In the lung, 7,911 cells were initially detected. Lymphatic ECs (Prox1 and Pdpn), type II epithelial cells (Sftpc, Sftpa1, and S100g), type I epithelial cells (Akap5, Hopx, and Cav1), blood cells (Ptprc), fibroblasts (Fn1 and Col1a1), and pericytes/SMCs (Pdgfrb and Acta2) were excluded from downstream analyses, leaving 6,636 ECs from the lung for analysis. Unbiased clustering of vascular ECs identified six clusters classified into one arterial EC cluster (cluster 0 marked by Efnb2, Gja4, and Gja5), three capillary EC clusters (clusters 1, 2, and 3 marked by Rgcc and Sparc), one venous EC cluster (cluster 4 marked by Nr2f2, Ephb4, and Slc6a2), and one aerocyte (aCap) cluster (cluster 5 marked by Car4) (Figures 1C, 1D, and S1C). Based on the classification of the single-cell data, analysis of the distribution of CD157-positive cells across EC subpopulations showed that they accounted for 2.0% of total vascular ECs in the liver, 3.9% in the limb muscle, and 8.0% in the lung (Figure 1E). In the liver, these cells were most abundant in the central and portal vein clusters, with fewer detected in the hepatic artery cluster. In the limb muscle and lung, they localized predominantly to venous-like EC clusters. These findings suggest a preferential association of CD157-positive ECs with larger vessel regions rather than capillaries. Consistently, immunofluorescence staining of mouse tissues confirmed their localization to the endothelium of relatively large vessels in the liver, muscle, and lung (Figure 1F). We next examined the common features of the CD157-positive ECs in each tissue by analyzing highly expressed genes in these cells. Compared to other ECs, the CD157-positive ECs in large vessel regions (clusters 0, 1, and 7 in the liver, cluster 6 in the hindlimb, and cluster 4 in the lung) exhibited differential expression profiles, defined by a fold-change threshold >1.5 for upregulated and <0.667 for downregulated genes, with 203 upregulated and 230 downregulated genes in the liver, 253 and 183 in the hindlimb muscle, and 138 and 100 in the lung (Figure 2A). Gene ontology (GO) analysis of the biological processes associated with these gene sets revealed enrichment of genes in pathways related to angiogenesis, wound healing, and leukocyte migration in all three tissues (Figure 2B). Across tissues, we identified 25 common DEGs that were highly expressed in CD157-positive ECs. Although without statistical significance, we found potential enrichment signals implicating VEGF receptor signaling pathways (Figures 2C, 2D, and S2). These results suggest shared gene expression profiles in CD157-positive cells in all three tissues and conserved angiogenic features regardless of tissue origin.Figure 2CD157-positive ECs from different mouse tissues share common gene expression profiles and pathways(A) Volcano plots showing gene expression changes between CD157-positive ECs in venous clusters and other ECs in each indicated tissue. Cutoff values were set at a fold change of 1.5 and a p value of 0.01. The numbers below the “upregulated genes” and “downregulated genes” labels indicate the number of genes exceeding these thresholds. Dots labeled with gene names represent the top 10 genes most highly expressed in CD157-positive ECs.(B) GO pathway analysis of genes characteristically upregulated in CD157-positive ECs in each tissue. GO analysis was performed using the upregulated genes identified in Figure 2A.(C) Venn diagram of upregulated genes detected in the three indicated tissues.(D) Violin plots comparing CD157-positive ECs in venous clusters with remaining ECs, showing the expression of 25 upregulated genes shared across the three tissues. CD157-positive ECs from different mouse tissues share common gene expression profiles and pathways (A) Volcano plots showing gene expression changes between CD157-positive ECs in venous clusters and other ECs in each indicated tissue. Cutoff values were set at a fold change of 1.5 and a p value of 0.01. The numbers below the “upregulated genes” and “downregulated genes” labels indicate the number of genes exceeding these thresholds. Dots labeled with gene names represent the top 10 genes most highly expressed in CD157-positive ECs. (B) GO pathway analysis of genes characteristically upregulated in CD157-positive ECs in each tissue. GO analysis was performed using the upregulated genes identified in Figure 2A. (C) Venn diagram of upregulated genes detected in the three indicated tissues. (D) Violin plots comparing CD157-positive ECs in venous clusters with remaining ECs, showing the expression of 25 upregulated genes shared across the three tissues. CD157-positive ECs exhibit transcriptomic commonality across tissues and aggregate into a conserved vascular subpopulation Previous scRNA-seq studies have shown marked heterogeneity and tissue-specific gene expression in vascular ECs. Integration of single-cell datasets across tissues typically reveals few shared clusters (Kalucka et al., 2020). Reanalysis of publicly available mouse scRNA-seq datasets revealed scarce numbers of CD157-positive ECs in the liver, brain, and heart under physiological conditions, whereas they were abundant in the lung and ischemic heart (Figure S3) (Li et al., 2019, 2024; Profaci et al., 2024; Rodor et al., 2022; Su et al., 2021). In contrast, our tissue-specific analyses in mice indicated that CD157-positive cells represent a conserved EC subpopulation. Therefore, we investigated whether our dataset, which included a larger number of these ECs, contained subpopulations shared across multiple tissues. Unbiased integration of single-cell data from each tissue, comprising 27,595 ECs in total, revealed tissue specificity for most ECs, but a population composed of cells derived from all tissues was also present (Figure 3A). Cluster classification of the integrated dataset showed that this mixed-origin EC population (hereafter referred to as mixed-ECs) formed distinct clusters characterized by venous and arterial-like endothelial features (Figures 3B and 3C). The remaining ECs were classified according to their tissue origin. ECs of the liver were classified into a central vein EC cluster (Wnt2 and Wnt9b), liver sinusoidal EC cluster (Lyve1 and Robo4), and arterial/portal vein EC cluster (Cd34, Gja5, and Ly6a). The limb muscle ECs were classified as a capillary EC cluster (Rgcc). The lung ECs were classified into a capillary EC cluster (Sparc and Tek), venous EC cluster (Slc6a2), and aCap cluster (Car4) (Figures 3C and S4A). This integrated clustering framework revealed a predominance of the CD157-positive ECs within these mixed-EC clusters (Figure 3D). However, the CD157-positive ECs from the lung were not restricted to the mixed-EC clusters, but were also broadly distributed within the lung venous cluster and, to a lesser extent, within the capillary EC cluster.Figure 3CD157-positive ECs are enriched in a mixed-EC cluster and share common gene expression profiles in mice(A) UMAP plot of 27,595 ECs integrated from three mouse tissues, color-coded by tissue of origin. The region outlined by the red box indicates an area where cells from all three tissues are intermixed.(B) UMAP plots of integrated ECs. The right image shows a magnified view of a cluster composed of cells derived from all three tissues.(C) Bubble plots comparing the expression levels of selected signature genes across clusters defined in Figure 3B.(D) Feature plots of Bst1 (CD157) expression projected onto the UMAP plots of integrated ECs. The feature plots within the black box on the right show Bst1 expression in mixed-EC cluster, followed by separate projections of Bst1 expression onto the UMAPs of the integrated dataset, split according to tissue.(E) Volcano plots showing gene expression changes between CD157-positive cells within the mixed-origin cluster and all other ECs. Genes with p values below 0.05 were considered significant; red dots indicate upregulated genes in CD157-positive cells, and blue dots indicate downregulated genes. The numbers below the “upregulated genes” and “downregulated genes” labels indicate the number of significantly changed genes. Horizontal line: p = 0.05; vertical lines: fold-change >1.5 (upregulated) or <0.67 (downregulated).(F) Volcano plots showing gene expression changes between CD157-negative cells within the mixed-origin cluster and all other ECs. Genes with p-values below 0.05 were considered significant; red dots indicate upregulated genes in CD157-negative cells, and blue dots indicate downregulated genes. The numbers below the ‘Upregulated Genes’ and ‘Downregulated Genes’ labels indicate the number of significantly changed genes. Horizontal line: p = 0.05; vertical lines: fold-change >1.5 (upregulated) or <0.67 (downregulated).(G) Heatmap showing the expression of genes that met the fold-change cutoff of >1.5 detected in Figure 3E.(H) GO pathway analysis of upregulated genes in CD157-positive ECs within the mixed-EC cluster.(I) GO pathway analysis of upregulated genes in CD157-negative ECs within the mixed-EC cluster.(J) Regulon specificity score (RSS) plots showing the ranking of regulons specifically active in CD157-positive ECs within the mixed-EC cluster. CD157-positive ECs are enriched in a mixed-EC cluster and share common gene expression profiles in mice (A) UMAP plot of 27,595 ECs integrated from three mouse tissues, color-coded by tissue of origin. The region outlined by the red box indicates an area where cells from all three tissues are intermixed. (B) UMAP plots of integrated ECs. The right image shows a magnified view of a cluster composed of cells derived from all three tissues. (C) Bubble plots comparing the expression levels of selected signature genes across clusters defined in Figure 3B. (D) Feature plots of Bst1 (CD157) expression projected onto the UMAP plots of integrated ECs. The feature plots within the black box on the right show Bst1 expression in mixed-EC cluster, followed by separate projections of Bst1 expression onto the UMAPs of the integrated dataset, split according to tissue. (E) Volcano plots showing gene expression changes between CD157-positive cells within the mixed-origin cluster and all other ECs. Genes with p values below 0.05 were considered significant; red dots indicate upregulated genes in CD157-positive cells, and blue dots indicate downregulated genes. The numbers below the “upregulated genes” and “downregulated genes” labels indicate the number of significantly changed genes. Horizontal line: p = 0.05; vertical lines: fold-change >1.5 (upregulated) or <0.67 (downregulated). (F) Volcano plots showing gene expression changes between CD157-negative cells within the mixed-origin cluster and all other ECs. Genes with p-values below 0.05 were considered significant; red dots indicate upregulated genes in CD157-negative cells, and blue dots indicate downregulated genes. The numbers below the ‘Upregulated Genes’ and ‘Downregulated Genes’ labels indicate the number of significantly changed genes. Horizontal line: p = 0.05; vertical lines: fold-change >1.5 (upregulated) or <0.67 (downregulated). (G) Heatmap showing the expression of genes that met the fold-change cutoff of >1.5 detected in Figure 3E. (H) GO pathway analysis of upregulated genes in CD157-positive ECs within the mixed-EC cluster. (I) GO pathway analysis of upregulated genes in CD157-negative ECs within the mixed-EC cluster. (J) Regulon specificity score (RSS) plots showing the ranking of regulons specifically active in CD157-positive ECs within the mixed-EC cluster. Comparison of the characteristics of CD157-positive ECs in the mixed-EC clusters to those of all other vascular ECs identified 290 upregulated and 86 downregulated genes (Figure 3E). A similar comparison of CD157-negative ECs in the mixed-EC clusters and other vascular ECs identified 241 upregulated and 108 downregulated genes (Figure 3F). Of these, 69 genes upregulated in CD157-positive ECs and 30 in CD157-negative ECs exhibited fold changes >1.5. Heatmap analysis confirmed high expression of these genes in the mixed-EC cluster. Notably, Procr and Abcg2, both recognized as VESC markers (Lin et al., 2024; Yu et al., 2016), were upregulated in the CD157-positive mixed-ECs (Procr; FC 1.97, Abcg2; FC 1.73) (Figure 3G). Pathway enrichment analysis revealed associations with angiogenesis, wound healing, and leukocyte migration (Figures 3H and S4B). Comparison of CD157-positive and CD157-negative cells within the mixed-EC population showed higher expression of most of these pathway-associated genes in CD157-positive cells (Figures S4C and S4D). No specific pathways were identified for the upregulated genes in CD157-negative ECs (Figure 3I). Regulon analysis of CD157-positive ECs identified several key transcriptional regulators involved in vascular regulation, including Nfatc1, Foxc1, and Creb3l1 (Figure 3J) (Binet and Sapieha, 2015; Seo et al., 2006; Suehiro et al., 2014). These findings suggested that CD157-positive ECs are enriched within mixed-EC clusters and may share a common vascular regulatory system. The NFAT pathway is essential for CD157+ EC network formation To validate these findings, and because Procr is a known VESC marker, we examined CD157 and Procr expression by FACS and identified four EC populations across all three tissues: CD157+Procr−, CD157−Procr+, CD157−Procr−, and CD157+Procr+ (Figure 4A). Immunofluorescence staining confirmed CD157 and Procr protein expression patterns consistent with the FACS analysis in all tissues (Figure 4B). Culture of liver-derived fractions showed that CD157+Procr+ ECs had the highest proliferative capacity (Figures 4C and 4D).Figure 4Mouse CD157+Procr+ ECs exhibit high proliferative potential and NFAT contributes to vascular network formation(A) Dot plots showing CD31+CD45− cells from the mouse liver, hindlimb muscle, and lung analyzed for CD157 and Procr expression.(B) Immunofluorescence images of mouse liver, hindlimb muscle, and lung stained for CD31 (green), CD157 (white), and Procr (red). Enlarged views of the areas outlined with white dashed lines in the upper images are shown in the lower images. Scale bars, 100 μm (upper) and 50 μm (lower).(C) Endothelial colony formation assay using 3,000 mouse liver-derived CD157+Procr+, CD157+Procr−, CD157−Procr+, and CD157−Procr− ECs cultured on OP9 feeder cells. Cells were stained with anti-CD31 antibody. Enlarged views of the areas outlined with red dashed lines in the upper images are shown in the lower images. Scale bars, 5 mm (upper) and 1 mm (lower).(D) Bar plots showing the relative quantification of CD31+ areas formed by each EC subpopulation after 10 days of culture, as shown in Figure 4C. Dots represent different wells (n > 8). Data from five independent isolation experiments. Data are presented as mean ± SEM. ∗∗∗p < 0.0005, ∗∗p < 0.005, ∗p < 0.05.(E) Endothelial colony formation under CsA treatment (1 μM), visualized by anti-CD31 immunostaining, using mouse liver-derived CD157+Procr+, CD157+Procr−, CD157−Procr+, and CD157−Procr- ECs. Scale bars,sss 5 mm.(F) Bar plots showing, from left to right, CD31+ area relative to the corresponding control; reduction in CD31+ area normalized to the reduction observed in CD157+Procr− cells; and reduction in CD31+ area relative to the corresponding control, after 10 days of culture with or without CsA, as shown in Figure 4E. Dots represent individual wells (n > 3). Data are derived from three independent experiments. Data are presented as mean ± SEM. ∗∗p < 0.005, ∗p < 0.05. Mouse CD157+Procr+ ECs exhibit high proliferative potential and NFAT contributes to vascular network formation (A) Dot plots showing CD31+CD45− cells from the mouse liver, hindlimb muscle, and lung analyzed for CD157 and Procr expression. (B) Immunofluorescence images of mouse liver, hindlimb muscle, and lung stained for CD31 (green), CD157 (white), and Procr (red). Enlarged views of the areas outlined with white dashed lines in the upper images are shown in the lower images. Scale bars, 100 μm (upper) and 50 μm (lower). (C) Endothelial colony formation assay using 3,000 mouse liver-derived CD157+Procr+, CD157+Procr−, CD157−Procr+, and CD157−Procr− ECs cultured on OP9 feeder cells. Cells were stained with anti-CD31 antibody. Enlarged views of the areas outlined with red dashed lines in the upper images are shown in the lower images. Scale bars, 5 mm (upper) and 1 mm (lower). (D) Bar plots showing the relative quantification of CD31+ areas formed by each EC subpopulation after 10 days of culture, as shown in Figure 4C. Dots represent different wells (n > 8). Data from five independent isolation experiments. Data are presented as mean ± SEM. ∗∗∗p < 0.0005, ∗∗p < 0.005, ∗p < 0.05. (E) Endothelial colony formation under CsA treatment (1 μM), visualized by anti-CD31 immunostaining, using mouse liver-derived CD157+Procr+, CD157+Procr−, CD157−Procr+, and CD157−Procr- ECs. Scale bars,sss 5 mm. (F) Bar plots showing, from left to right, CD31+ area relative to the corresponding control; reduction in CD31+ area normalized to the reduction observed in CD157+Procr− cells; and reduction in CD31+ area relative to the corresponding control, after 10 days of culture with or without CsA, as shown in Figure 4E. Dots represent individual wells (n > 3). Data are derived from three independent experiments. Data are presented as mean ± SEM. ∗∗p < 0.005, ∗p < 0.05. Given the enrichment of NFAT-related transcriptional activity in CD157-positive ECs, we next examined the role of NFAT in these populations by inhibiting the NFAT pathway in each EC subpopulation with cyclosporin A (CsA). Although CsA suppressed network formation in all EC subpopulations, the reduction in network area was significantly greater in CD157+Procr+ and CD157+Procr− ECs than in the other two fractions (Figures 4E and 4F). These results suggest that inhibiting the highly proliferative CD157+ population results in a more pronounced reduction in vascular network formation, while also supporting a broader role for NFAT signaling across EC fractions. Identification of stem-like vascular ECs in humans VESCs have rarely been characterized in humans, and their defining features are largely unknown. To identify and characterize these cells in humans, we isolated CD31+CD45− cells from human lower limb muscle and performed scRNA-seq analysis. Unbiased clustering of the 4,958 sorted cells revealed 10 clusters (Figures 5A and 5B). Based on the expression of well-established cell type markers, cells expressing CDH5, PECAM1 (CD31), and EMCN but negative for PROX1 were classified as vascular ECs, whereas lymphatic ECs (PROX1, PDPN, and LYVE1), muscle cells (TNNC2), and pericytes/SMCs (PDGFRB, RGS5, and ACTA2) were identified as non-EC populations. As a result, 4,779 vascular ECs were identified and further classified into seven EC clusters. In this dataset, a small number of CD157-expressing cells was detected in capillary and venous EC populations (Figure 5C). Assessment of interindividual variability using scRNA-seq analysis of ECs isolated from an independent human lower limb muscle sample revealed a similar distribution pattern in this independent dataset (Figures S5A and S5B).Figure 5Identification of a human VESC-enriched cluster(A) UMAP plots of scRNA-seq data from sorted CD31+CD45− cells isolated from human lower limb muscle (4,958 cells).(B) Bubble plots comparing the expression levels of selected signature genes across clusters in the CD31+CD45− cells isolated from human lower limb muscle.(C) Feature plots of CD157 (BST1) and arterial, capillary, and venous EC markers projected onto the UMAP shown in Figure 5A.(D) Endothelial colony formation assay using 5,000 human lower limb-derived CD157+ and CD157− ECs cultured on OP9 feeder cells. Cells were stained with anti-CD31 antibody. Enlarged views of the areas outlined by red dashed lines in the left images are shown in the right images. Scale bars, 1 mm (left) and 200 μm (rightss).(E) Bar plots showing the relative quantification of CD31+ areas formed by CD157+ and CD157− ECs after 10 days of culture, as shown in Figure 5D. Dots represent individual areas (n = 6). Data are derived from two independent experiments. Data are presented as mean ± SEM. ∗∗p < 0.005.(F) Immunofluorescence images of human soleus muscle showing large vessels and small vessels/capillaries stained for CD31 (white), CD157 (green), αSMA (red), and DAPI (blue). Enlarged views of the regions outlined by white dashed lines in the upper images are shown in the lower images. Arrows indicate CD157+ ECs. Scale bars, 100 μm (upper) and 50 μm (enlarged lower images).(G) UMAP plots from reanalysis of publicly available human EC scRNA-seq datasets.(H) Bubble plots comparing the expression levels of selected signature genes across clusters in Figure 5G.(I) Feature plots of CD157 (BST1) and arterial, capillary, and venous EC markers projected onto the UMAP shown in Figure 5G.(J) Pie chart showing the number of CD157-positive cells in each cluster in Figure 5G, and bar graph showing the proportion of CD157-positive cells in each cluster.(K) Heatmap representation of cluster enrichment per tissue in Figure 5G.(L) Proportion of CD157-positive cells by tissue of origin. Identification of a human VESC-enriched cluster (A) UMAP plots of scRNA-seq data from sorted CD31+CD45− cells isolated from human lower limb muscle (4,958 cells). (B) Bubble plots comparing the expression levels of selected signature genes across clusters in the CD31+CD45− cells isolated from human lower limb muscle. (C) Feature plots of CD157 (BST1) and arterial, capillary, and venous EC markers projected onto the UMAP shown in Figure 5A. (D) Endothelial colony formation assay using 5,000 human lower limb-derived CD157+ and CD157− ECs cultured on OP9 feeder cells. Cells were stained with anti-CD31 antibody. Enlarged views of the areas outlined by red dashed lines in the left images are shown in the right images. Scale bars, 1 mm (left) and 200 μm (rightss). (E) Bar plots showing the relative quantification of CD31+ areas formed by CD157+ and CD157− ECs after 10 days of culture, as shown in Figure 5D. Dots represent individual areas (n = 6). Data are derived from two independent experiments. Data are presented as mean ± SEM. ∗∗p < 0.005. (F) Immunofluorescence images of human soleus muscle showing large vessels and small vessels/capillaries stained for CD31 (white), CD157 (green), αSMA (red), and DAPI (blue). Enlarged views of the regions outlined by white dashed lines in the upper images are shown in the lower images. Arrows indicate CD157+ ECs. Scale bars, 100 μm (upper) and 50 μm (enlarged lower images). (G) UMAP plots from reanalysis of publicly available human EC scRNA-seq datasets. (H) Bubble plots comparing the expression levels of selected signature genes across clusters in Figure 5G. (I) Feature plots of CD157 (BST1) and arterial, capillary, and venous EC markers projected onto the UMAP shown in Figure 5G. (J) Pie chart showing the number of CD157-positive cells in each cluster in Figure 5G, and bar graph showing the proportion of CD157-positive cells in each cluster. (K) Heatmap representation of cluster enrichment per tissue in Figure 5G. (L) Proportion of CD157-positive cells by tissue of origin. We next isolated CD31+CD45−CD157+ ECs from human lower limb muscle to evaluate their proliferative capacity (Figure S5C). The proliferative capacity was higher in these cells than in CD157− ECs, and they formed colony-like clusters resembling those formed by mouse CD157+ ECs (Figures 5D and 5E) (Wakabayashi et al., 2018). CD157+ ECs were detected in larger vessels surrounded by multilayered αSMA-positive SMCs, whereas CD157 was not detected in capillary ECs, including capillaries covered by a αSMA-positive cell monolayer (Figure 5F). Therefore, although CD157+ ECs in the human dataset did not form the clearly segregated cluster observed in mouse tissues, the functional and anatomical data showed similarities between both species. Reanalysis of publicly available multi-organ EC scRNA-seq datasets also identified a subset of CD157+ ECs enriched within venous clusters, consistent with our findings in mice (Figures 5G–5K) and originating from multiple organs (Figure 5L) (Barnett et al., 2024). Comparative scRNA-seq analysis of mouse and human ECs indicates conservation of CD157-positive ECs Based on these findings, we hypothesized that stem-like ECs exist in the human vasculature. We performed cross-species integration to characterize human CD157-positive ECs and assess their conservation with mouse CD157-positive ECs. Recognizing that integration between mouse and human datasets is often challenging and does not consistently achieve the same level of alignment obtained in mouse-to-mouse dataset integration, we integrated the mouse and human datasets by converting mouse gene symbols to their human orthologs. We selected only the genes common to both species and applied multiple integration methods for clustering (Figure S6A). Among the methods tested, Harmony integration yielded the most balanced mixing of mouse and human cells while preserving arterial, venous, and capillary classifications (Figure S6B). Therefore, this method was used for the downstream analyses. The Harmony-integrated data were classified into 18 clusters (Figures 6A and 6B) and further classified into three venous-like EC clusters (clusters 0, 1, and 2 marked by ACKR1 and NR2F2), two post-capillary venule clusters (clusters 3 and 4 marked by POSTN), five capillary clusters (clusters 5–9 marked by RGCC and SPARC), two arterial-like EC clusters (clusters 10 and 11 marked by EFNB2, GJA4, and GJA5), two lymphatic EC clusters (clusters 12 and 13 marked by PROX1, PDPN, and LYVE1), two pericyte/SMCs clusters (clusters 14 and 15 marked by PDGFRB, RGS5, and ACTA2), one muscle cell cluster (cluster 16 marked by TNNC2), and one fibroblast cluster (cluster 17 marked by FN1 and COL1A1). In this integrated dataset, both mouse and human CD157-positive cells accumulated in clusters 0, 1, and 2, although their numbers were limited in the human dataset (Figures 6C, 6D, and S6C). Despite their limited number, the CD157-positive cells accumulated within the venous EC cluster after integration. PROCR- and ABCG2-positive cells were also predominantly enriched in the same clusters; however, they were also detected to some extent in other clusters. Likewise, NFATC1 was widely expressed but exhibited approximately twice the expression frequency in these clusters than in the other clusters.Figure 6Comparative scRNA-seq analysis of mouse and human ECs reveals a conserved CD157-enriched EC cluster and candidate human markers(A) UMAP of integrated dataset combining human lower limb muscle CD31+CD45− cells (4,958 cells) and mouse hindlimb muscle CD31+CD45− cells (12,681 cells), totaling 17,639 cells.(B) Bubble plots comparing the expression levels of selected signature genes across cell clusters from the integrated dataset shown in Figure 6A.(C and D) Feature plots of VESC-associated genes projected onto species-split UMAPs of the integrated dataset: mouse (C) and human (D).(E) Volcano plots showing DEGs for the human-derived components of clusters 0, 1, and 2 from the integrated dataset. Genes with a p value < 0.05 and a fold change (FC) > 1.25 were considered significant. Among the upregulated genes, the top five with the highest fold change and strong cluster specificity (defined as expression in <20% of cells outside the target cluster) are shown. If fewer than five genes met these criteria, all qualifying genes are displayed.(F) Results of GO enrichment analysis for DEGs identified in clusters 0, 1, and 2 from the integrated dataset (p value < 0.05 and FC > 1.25).(G) Feature plots showing the expression of five genes—CPE, EFEMP1, GPM6A, CDH11, and NRG3—that are highly expressed in cluster 0, as presented in Figure 6E.(H) In situ hybridization for CD31 together with GPM6A, CDH11, or NRG3 in human soleus muscle. CD31 (Cy5, magenta) and target genes (Cy3, green) are shown. ECs in large blood vessels and capillaries are visualized. Higher magnification views of the regions outlined by black dotted boxes are shown on the right. Red and green arrowheads indicate CD31 and target gene signals, respectively. Two independent patient samples were analyzed. Scale bars, 50 μm. Comparative scRNA-seq analysis of mouse and human ECs reveals a conserved CD157-enriched EC cluster and candidate human markers (A) UMAP of integrated dataset combining human lower limb muscle CD31+CD45− cells (4,958 cells) and mouse hindlimb muscle CD31+CD45− cells (12,681 cells), totaling 17,639 cells. (B) Bubble plots comparing the expression levels of selected signature genes across cell clusters from the integrated dataset shown in Figure 6A. (C and D) Feature plots of VESC-associated genes projected onto species-split UMAPs of the integrated dataset: mouse (C) and human (D). (E) Volcano plots showing DEGs for the human-derived components of clusters 0, 1, and 2 from the integrated dataset. Genes with a p value < 0.05 and a fold change (FC) > 1.25 were considered significant. Among the upregulated genes, the top five with the highest fold change and strong cluster specificity (defined as expression in <20% of cells outside the target cluster) are shown. If fewer than five genes met these criteria, all qualifying genes are displayed. (F) Results of GO enrichment analysis for DEGs identified in clusters 0, 1, and 2 from the integrated dataset (p value < 0.05 and FC > 1.25). (G) Feature plots showing the expression of five genes—CPE, EFEMP1, GPM6A, CDH11, and NRG3—that are highly expressed in cluster 0, as presented in Figure 6E. (H) In situ hybridization for CD31 together with GPM6A, CDH11, or NRG3 in human soleus muscle. CD31 (Cy5, magenta) and target genes (Cy3, green) are shown. ECs in large blood vessels and capillaries are visualized. Higher magnification views of the regions outlined by black dotted boxes are shown on the right. Red and green arrowheads indicate CD31 and target gene signals, respectively. Two independent patient samples were analyzed. Scale bars, 50 μm. To account for potential bias arising from differences in cell numbers between human and mouse datasets, we also performed cross-species integration of mouse hindlimb cells (12,681 cells) with CD31+CD45− cells from one independent human sample (12,915 cells; UMAP shown in Figure S5A) and, separately, with two human samples (17,873 cells; UMAP shown in Figure S5B). Again, a subset of human CD157-positive ECs co-localized within clusters enriched for mouse CD157-positive ECs (Figures S6D and S6E). Another subset of human CD157-positive ECs localized to a venous/capillary-like cluster characterized by ACKR1 expression, but this was not detected in the mouse dataset. To minimize complexity introduced by multi-sample integration and batch correction, we performed downstream analyses using an integrated dataset comprising a single human sample and the mouse dataset and evaluated the gene expression signatures of clusters containing CD157-positive ECs (Figure 6E; Table S1). Comparison with the mouse CD157-positive EC markers defined in Figure 3G revealed limited overlap between the three clusters (Figure S6F), suggesting modest direct gene-level similarity. However, examination of cluster-specific pathways revealed that cluster 0 was enriched for angiogenesis-, vasculogenesis-, and wound healing-related pathways, resembling the enrichment observed in the mouse CD157-positive EC population (Figure 6F). Therefore, we focused on cluster 0 and confirmed that its top five signature genes (CPE, EFEMP1, GPM6A, CDH11, and NRG3) were predominantly expressed in this cluster (Figure 6G). Further verification of the expression patterns of these genes in mice datasets revealed a partial co-expression of Cpe, Efemp1, and Gpm6a with CD157 in limb muscle ECs and in the integrated dataset. However, Cdh11 was scarcely expressed in the capillary cluster, and Nrg3 was not detected in either dataset (Figure S6G). Although not all genes expressed in cluster 0 overlapped with those in the mouse CD157-positive cluster, we investigated the spatial distribution of ECs of this cluster by performing in situ hybridization of human lower limb muscle tissue for NRG3, GPM6A, and CDH11. These genes were not detected in capillaries but were expressed in ECs associated with relatively large vessels, consistent with the distribution of CD157+ ECs (Figure 6H). These findings suggest the presence of human ECs with gene expression profiles partially resembling those of mouse CD157+ VESCs and support the existence of a conserved subpopulation localized to large blood vessels. Bioinformatic analysis reveals interactions between CD157-positive ECs and pericytes/SMCs in mouse We further performed bioinformatic analysis to evaluate potential interactions between CD157-positive ECs and surrounding cells. Stem-like cells are known to form niches within tissues and to engage in distinct cell–cell interactions with surrounding cells (Hsu and Fuchs, 2012). We explored the microenvironment associated with CD157-positive ECs by collecting publicly available mouse scRNA-seq datasets encompassing all cell types from the liver, limb, and lung (Reyfman et al., 2019; Schaum et al., 2018; Xiong et al., 2019). We then integrated these whole-tissue datasets (48,972 cells) with our vascular EC dataset (32,872 cells), totaling 81,844 cells, to analyze cell-cell interactions between CD157-positive ECs and surrounding cell types. The vascular EC population clustered in a manner consistent with our EC-only dataset, including a mixed-EC cluster, while the surrounding cells clustered according to their respective cell types (Figures 7A–7C). Cell-cell interaction analysis using CellPhoneDB revealed that CD157-positive ECs within the mixed-EC cluster engaged in numerous interactions with surrounding cells, particularly pericytes/SMCs (Figure 7D). Given that these interactions exhibited the highest number of signaling events, we also used CellChat to analyze pericyte/SMC-derived signals toward CD157-positive ECs. The predominant signals were associated with the extracellular matrix (i.e., collagen and laminin), followed by CXCL signaling (Figure 7E).Figure 7CD157-positive ECs exhibit specific cell-cell interaction signals in mice(A) UMAP of our integrated dataset of mouse ECs from three tissues combined with all cell types from a public dataset covering the same three tissues (total: 81,844 cells).(B) Bubble plots showing the expression of selected marker genes used to define each cluster in Figure 7A.(C) Feature plots of selected genes projected onto the integrated UMAP shown in Figure 7A.(D) Heatmap showing the results of cell-cell interaction analysis using CellPhoneDB between the mixed-origin EC cluster and surrounding cell populations from the liver, hindlimb muscle, and lung. The x axis represents the sender cell populations, and the y axis represents the receiver cell populations. The numbers indicate the total number of detected ligand-receptor interactions.(E) Circos plot showing ligand-receptor interactions from the pericyte/SMCs cluster to CD157-positive ECs within the mixed-EC cluster, as analyzed by CellChat. Interactions are grouped by signaling family.(F) Heatmap showing communication probabilities between sender-receiver pairs for CXCL signaling networks. The black dotted box indicates signals sent from surrounding cells to EC subpopulations.(G) Violin plots showing the expression levels of CXCL signaling-related genes across clusters.(H) Immunofluorescence images of mouse liver, hindlimb muscle, and lung stained for CD31 (green), Cxcr7 (red), and CD157 (white). Scale bars,ss 50 μm. CD157-positive ECs exhibit specific cell-cell interaction signals in mice (A) UMAP of our integrated dataset of mouse ECs from three tissues combined with all cell types from a public dataset covering the same three tissues (total: 81,844 cells). (B) Bubble plots showing the expression of selected marker genes used to define each cluster in Figure 7A. (C) Feature plots of selected genes projected onto the integrated UMAP shown in Figure 7A. (D) Heatmap showing the results of cell-cell interaction analysis using CellPhoneDB between the mixed-origin EC cluster and surrounding cell populations from the liver, hindlimb muscle, and lung. The x axis represents the sender cell populations, and the y axis represents the receiver cell populations. The numbers indicate the total number of detected ligand-receptor interactions. (E) Circos plot showing ligand-receptor interactions from the pericyte/SMCs cluster to CD157-positive ECs within the mixed-EC cluster, as analyzed by CellChat. Interactions are grouped by signaling family. (F) Heatmap showing communication probabilities between sender-receiver pairs for CXCL signaling networks. The black dotted box indicates signals sent from surrounding cells to EC subpopulations. (G) Violin plots showing the expression levels of CXCL signaling-related genes across clusters. (H) Immunofluorescence images of mouse liver, hindlimb muscle, and lung stained for CD31 (green), Cxcr7 (red), and CD157 (white). Scale bars,ss 50 μm. CXCL signaling plays key roles in angiogenesis, cell migration, and inflammatory cell recruitment (Mehrad et al., 2007). Based on communication probability scores, CXCL signaling activity was approximately 2-fold higher between CD157-positive ECs in the mixed-EC cluster and pericytes/SMCs than in other EC clusters (Figure 7F; Table S2). The CD157-positive ECs also highly expressed Ackr3 (Cxcr7), a gene regulated by Nfatc1, whereas pericytes/SMCs expressed Cxcl12, the corresponding ligand (Figure 7G). Immunofluorescence staining confirmed CXCR7 expression in CD157+ ECs in mouse liver, hindlimb muscle, and lung tissues (Figure 7H), while in situ hybridization demonstrated CXCL12 expression in perivascular regions, including mural cells (Figures S7A and S7B). Together, these results suggest that CD157+ ECs express CXCR7 and may interact with surrounding cells, including pericytes/SMCs, through CXCL signaling within the microenvironment. Discussion An EC population with stem-like features has been proposed previously (Dight et al., 2022), and CD157+ ECs have been shown to display proliferative and vessel-forming capacity (Wakabayashi et al., 2018; 2022). However, scRNA-seq-based characterization of these cells has been limited, and their transcriptomic features have remained unclear. Here, we used scRNA-seq to analyze multiple mouse tissues and identified a common gene expression profile shared between CD157-positive ECs within mixed-EC clusters. These cells expressed Abcg2 and Procr and were enriched for angiogenesis-related pathways, consistent with a stem-like endothelial phenotype. FACS analysis identified CD157+Procr+ ECs as a subset with elevated proliferative capacity, indicating heterogeneity within the CD157+ EC population. Co-expression of CD157 and Procr may mark a proliferative, stem-like EC subset. Mechanistically, these cells were linked to NFAT signaling, as its inhibition reduced proliferation and network formation. NFAT functions downstream of VEGF and is involved in angiogenic regulation, suggesting that it may contribute to the stem-like properties in CD157+ ECs. However, NFAT signaling alone does not sufficiently account for their stem-like characteristics, as NFAT activity broadly regulates endothelial proliferation and function across EC populations. In humans, we identified a VESC-enriched cluster (hVESC) with high proliferative capacity and a gene expression profile resembling that of mouse CD157-positive ECs. Similar to the mouse population, this cluster included cells expressing CD157, PROCR, and ABCG2, although their distribution differed between species. CD157 expression was more restricted in humans, whereas PROCR was more broadly detected but concentrated within the hVESC cluster. ABCG2 was widely expressed in mice but more confined in humans. NFATC1 was broadly expressed across ECs, with relatively higher levels in the hVESC cluster. Together, these observations indicate that stem cell-associated features are conserved but not identical across species, and that both interspecies and organ-specific differences require further clarification. Functionally, Abcg2 acts as a drug efflux transporter. It is expressed in various stem cell populations and underlies the side population phenotype used to enrich tissue stem cells, including VESCs (Goodell et al., 1996; Iba et al., 2019; Naito et al., 2012; 2016). Its drug efflux activity may contribute to the maintenance of a protective stem cell microenvironment. CD157 has been linked to cellular energy metabolism, although its role in stemness remains unclear (Covarrubias et al., 2021), raising the possibility that it influences the metabolic environment supporting stem-like properties. Procr has anticoagulant and cytoprotective functions and has been reported to influence EC cycle progression, suggesting a potential role in regulating proliferative capacity (Chambers et al., 2025; Mohan Rao et al., 2014). The roles of newly identified genes in the hVESC cluster likewise remain to be characterized. Integration of our EC datasets with publicly available whole-tissue transcriptomic data enabled assessment of potential interactions between EC subclusters and surrounding cell types. This analysis suggested that CD157+ ECs may engage in communication with neighboring cells and indicated a potential role for the CXCL12–CXCR7 signaling axis. We confirmed CXCR7 expression in CD157+ ECs by immunostaining. CXCL12–CXCR7 signaling has been reported to be activated by VEGF in tumor ECs and to influence angiogenic properties (Yamada et al., 2015), raising the possibility that VESCs may be primed to initiate angiogenesis in response to stimulation. While this analysis relies primarily on bioinformatic inference and requires further experimental validation, it may provide a useful framework for exploring the functional properties of endothelial subclusters. More broadly, a limitation of the present study is that our conclusions are largely based on transcriptomic analyses and remain to be validated in vivo. Cross-species integration of human and mouse datasets remains methodologically challenging and may introduce potential biases. Thus, findings derived from these integrated datasets, including the cell-cell interaction analyses and the inferred functions of newly identified genes, require cautious interpretation and warrant further experimental confirmation. Our experiments using human CD157+ ECs also did not directly establish in vivo relevance, nor did they fully address inter-individual variability or potential organ-specific differences. Isolation of CD157+ ECs by FACS was also less efficient in human samples than in mice, indicating that technical optimization may be necessary. Nonetheless, we believe that the identification and characterization of human VESCs, including CD157+ ECs, represent an important step toward a better understanding of human vascular biology. Moreover, although multiple signaling pathways were identified, their relative importance in mediating interactions between VESCs and neighboring cells and in maintaining the niche remains to be defined. Future studies will be required to clarify how these pathways relate to the stem-like properties and proliferative capacity of VESCs. In summary, we have demonstrated that CD157-positive ECs exhibit conserved features across multiple tissues. Our scRNA-seq analysis revealed that these cells express high levels of pro-angiogenic factors, even under homeostatic conditions, and that their counterparts are present in human tissues. These findings support the concept of a vascular endothelial stem-like population with potential relevance for vascular regeneration or disruption therapy. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Hisamichi Naito (naito.hisamichi@gmail.com). Materials availability This study did not generate new unique reagents. Data and code availability •The scRNA-seq data are available at the Gene Expression Omnibus (GEO) under accession number GSE303731.•All original code has been deposited at GitHub and is publicly available at https://github.com/Ivalyne/scRNAseq_CD157VESCs.•Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. •The scRNA-seq data are available at the Gene Expression Omnibus (GEO) under accession number GSE303731. The scRNA-seq data are available at the Gene Expression Omnibus (GEO) under accession number GSE303731. •All original code has been deposited at GitHub and is publicly available at https://github.com/Ivalyne/scRNAseq_CD157VESCs. All original code has been deposited at GitHub and is publicly available at https://github.com/Ivalyne/scRNAseq_CD157VESCs. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments We thank Ms. M. Ishida, Ms. Y. Mitani, Ms. C. Hirose, Ms. T. Yuhi, Ms. M. Oura, and Ms. Kikuta for their technical assistance. We also appreciate the technical support provided by Kanazawa University Environmental Stress Research Center for single-cell RNA library preparation and by Biken Biomics for sequencing. We acknowledge the following financial support for the research presented in this article: the 10.13039/100009619Japan Agency for Medical Research and Development (PRIME JP21gm6210009, FORCE JP24gm4010023, Research Program on Hepatitis JP25fk0210129), the MEXT/JSPS KAKENHI (25K18773, 24K02221, 22H05063, 22H05060, 22K15523, 21K20741, and 20H03435), JST FOREST Program (JPMJFR206A), the Takeda Science Foundation, the Kato Memorial Fund for Incurable Disease Research, and the Kanazawa University SAKIGAKE project. Author contributions H.N. conceived the project; T.I., T.W., and H.N. performed the single-cell analysis; T.I., T.W., K.Y., Y.I., and M.H. conducted the cellular and histological experiments; M.S. provided support for the analysis of human samples; A.M., J.M., and N.N. provided additional experimental support; T.I. performed all data analyses; R.H., A.S., and S.F. provided resources; T.I., T.W., and H.N. wrote the manuscript, and H.N. edited the final version; D.M., R.T., and H.A. provided critical input; N.T. and H.N. supervised the research. All authors read and approved the final manuscript. Declaration of interests N.T. serves as a Scientific Advisor for Revascular Bio. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work, the authors used ChatGPT to improve language and readability. After using this tool, the manuscript was further edited by a professional English editing service. The authors reviewed and revised the text as needed and take full responsibility for the content of the publication. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesPurified Anti-Mouse CD16/CD32 (2.4G2)CYTEC BiosciencesCat# 70-0161-U100; RRID: AB_2621487Brilliant Violet 421 anti-mouse CD31BioLegendCat# 102423; RRID: AB_2562186CD45 Monoclonal Antibody (30-F11), FITCThermo Fisher ScientificCat# 11-0451-82; RRID: AB_465050APC anti-mouse CD157 (BST-1)BioLegendCat# 140208; RRID: AB_10901172PE anti-mouse CD201 (EPCR)BioLegendCat# 141504; RRID: AB_10895909Human TruStain FcX (Fc Receptor Blocking Solution)BioLegendCat# 422302; RRID: AB_2818986APC anti-human CD31BioLegendCat# 303116; RRID: AB_1877152Brilliant Violet 421 anti-human CD45BioLegendCat# 304032; RRID: AB_2561357BST-1/CD157, Human (FITC conj.)MBLCat# D036-4; RRID: AB_590906Purified Rat Anti-Mouse CD31BD BiosciencesCat# 553370; RRID: AB_394816Anti-Alpha Smooth Muscle ActinBioLegendCat# 904601; RRID:AB_2565041PE anti-human/mouse CXCR7BioLegendCat# 331104; RRID: AB_1227619CD31 (PECAM-1) (F8M3S) Rabbit Monoclonal AntibodyCell Signaling TechnologyCat# 95462; RRID: AB_3739873Alexa Fluor 647-AffiniPure Fab Fragment Goat Anti-Rabbit IgGJackson ImmunoResearch LabsCat# 111-607-008; RRID: AB_2632470Goat anti-Mouse IgG2a Cross-Adsorbed Secondary Antibody, Alexa Fluor 568Thermo Fisher ScientificCat# A-21134; RRID: AB_2535773Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 488Thermo Fisher ScientificCat# A-21121; RRID: AB_2535764Goat anti-Rat IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488Thermo Fisher ScientificCat# A-11006; RRID: AB_2534074Goat anti-Rat IgG (H + L) Secondary Antibody, BiotinThermo Fisher ScientificCat# 31830; RRID: AB_228355Goat Anti-Mouse IgG Antibody (H + L), BiotinylatedVector LaboratoriesCat# AI-9200-1.5; RRID: AB_3107016Biological samplesHuman soleus muscleJuntendo University HospitalN/AChemicals, peptides, and recombinant proteinsCyclosporine (Sandimmun®)Novartis PharmaNDC 00078-0109DAPIThermo Fisher ScientificCat# D1306Propidium Iodide SolutionBioLegendCat# 421301Dispase IIThermo Fisher ScientificCat# 17105041CollagenaseWakoCat# 034-22363Collagenase Type IIWorthington BiochemicalCat# LS004176Collagenase type XISigma-AldrichCat# H3506Soybean trypsin inhibitorWorthington BiochemicalCat# LS003571Lyophilized elastaseWorthington BiochemicalCat# LS002292Hyaluronidase type ISigma-AldrichCat# H4034ACK Lysing BufferLonzaCat# 10-548EPercollCytivaCat# 17089102Debris Removal SolutionMiltenyi BiotecCat# 130-109-398Bovine serum albuminSigma-AldrichCat# A9418-10GMinimum Essential Medium Eagle-alpha modificationSigma-AldrichCat# M8042-500MLFetal bovine serumNichirei BiosciencesCat# 175012L-glutamineThermo Fisher ScientificCat# 25030081Penicillin/streptomycinThermo Fisher ScientificCat# 15140122RPMI-1640Sigma-AldrichCat# R8758VEGF 165, Human, RecombinantPeproTechCat# 100-20-10ug2-mercaptoethanolThermo Fisher ScientificCat# 21985023EGM-2 Endothelial Cell Growth Medium-2 BulletKitLonzaCat# CC-3162Opal 480 Reagent PackAkoya BiosciencesCat# FP1500001KTOpal 570 Reagent PackAkoya BiosciencesCat# FP1488001KTOpal 620 Reagent PackAkoya BiosciencesCat# FP1495001KTNew Hematoxylin Type GMuto Pure ChemicalsCat# 30161VECTASTAIN Elite ABC KitVector LaboratoriesCat# PK-6100optimal cutting temperature (OCT) compoundSakura FinetekCat# 45833Critical commercial assaysRNAscope 2.5 Duplex Reagent KitAdvanced Cell DiagnosticsCat# 322430Chromium Next GEM Single Cell 3′ Kit v3.110x GenomicsCat# PN-1000269Deposited dataMouse and Human Single-Cell RNA seq dataThis studyGEO: GSE303731Mouse Liver Single-Cell RNA seq dataSu et al. (2021)GEO: GSE147581Mouse Liver Single-Cell RNA seq dataLi et al. (2024)GEO: GSE225786Mouse Liver Single-Cell RNA seq dataXiong et al. (2019)GEO: GSE129516Mouse Brain Single-Cell RNA seq dataProfaci et al. (2024)GEO: GSE267591Mouse Heart Single-Cell RNA seq dataLi et al. (2019)GEO: GSE132880Mouse Lung Single-Cell RNA seq dataRodor et al. (2022)GEO: GSE154959Mouse Lung Single-Cell RNA seq dataReyfman et al. (2019)GEO: GSE121611Mouse Single-Cell RNA seq dataSchaum et al. (2018)GEO: GSE109774Human Vascular Atlas Single-Cell RNA seq dataBarnett et al. (2024)https://www.vascularcellatlas.org/Experimental models: Cell linesMouse: OP9Riken BRCRCB1124Experimental models: Organisms/strainsC57BL/6Japan SLCN/ASoftware and algorithmsImageJ/Fiji v2.16Schindelin et al. (2012)https://imagej.net/10x Genomics Cell Ranger v7.1.0Zheng et al. (2017)https://www.10xgenomics.com/Seurat v4.4.0Hao et al. (2021)https://satijalab.org/seurat/tidyverse v2.0Wickham et al. (2019)https://tidyverse.org/clusterProfiler v4.7.1.002Wu et al. (2021)https://bioconductor.org/packages/clusterProfiler/Scanorama v1.7.4Hie et al. (2019)https://github.com/brianhie/scanoramaSCENIC v1.3.1Aibar et al. (2017)https://github.com/aertslab/SCENICharmony v1.2.0Korsunsky et al. (2019)https://CRAN.R-project.org/package=harmonyCellPhoneDB v5.0.1Troulé et al. (2025)https://github.com/ventolab/CellphoneDBCellChat v1.6.1Jin et al. (2021)https://github.com/jinworks/CellChatFlowJo softwareBD Bioscienceshttps://www.flowjo.com/Photoshop CCAdobe Systemshttps://www.adobe.com/R v4.2.2R Core Team, 2022https://www.r-project.org/reticulate v1.44.1GitHubhttps://rstudio.github.io/reticulate/org.Hs.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/org.Hs.eg.db/org.Mm.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/org.Mm.eg.db/Orthology.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/Orthology.eg.db/gplots v3.1.3.1CRANhttps://CRAN.R-project.org/package=gplotspheatmap v1.0.12CRANhttps://CRAN.R-project.org/package=pheatmapEnhancedVolcano v1.16GitHubhttps://github.com/kevinblighe/EnhancedVolcanoloomR v0.2.1.9000GitHubhttps://github.com/mojaveazure/loomRSCopeLoomR v0.13GitHubhttps://github.com/aertslab/SCopeLoomRAngioTool 2.0GitHubhttps://github.com/jbendtsen/AngioTool-Batchktplots v2.4.0GitHubhttps://github.com/zktuong/ktplotsOtherRNAscope™ Probe- Hs-GPM6AAdvanced Cell DiagnosticsCat# 470601RNAscope™ Probe- Hs-CDH11Advanced Cell DiagnosticsCat# 448381RNAscope™ Probe- Hs-NRG3Advanced Cell DiagnosticsCat# 893971-C1RNAscope™ Probe- Hs-PECAM1-O1-C2Advanced Cell DiagnosticsCat# 487381-C2RNAscope™ Probe-Mm-Bst1-O1-C1Advanced Cell DiagnosticsCat# 1868681-C1RNAscope™ Probe-Mm-Pecam1-C2Advanced Cell DiagnosticsCat# 316721-C2RNAscope™ Probe-Mm-Cxcl12-C4Advanced Cell DiagnosticsCat# 422711-C4 REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesPurified Anti-Mouse CD16/CD32 (2.4G2)CYTEC BiosciencesCat# 70-0161-U100; RRID: AB_2621487Brilliant Violet 421 anti-mouse CD31BioLegendCat# 102423; RRID: AB_2562186CD45 Monoclonal Antibody (30-F11), FITCThermo Fisher ScientificCat# 11-0451-82; RRID: AB_465050APC anti-mouse CD157 (BST-1)BioLegendCat# 140208; RRID: AB_10901172PE anti-mouse CD201 (EPCR)BioLegendCat# 141504; RRID: AB_10895909Human TruStain FcX (Fc Receptor Blocking Solution)BioLegendCat# 422302; RRID: AB_2818986APC anti-human CD31BioLegendCat# 303116; RRID: AB_1877152Brilliant Violet 421 anti-human CD45BioLegendCat# 304032; RRID: AB_2561357BST-1/CD157, Human (FITC conj.)MBLCat# D036-4; RRID: AB_590906Purified Rat Anti-Mouse CD31BD BiosciencesCat# 553370; RRID: AB_394816Anti-Alpha Smooth Muscle ActinBioLegendCat# 904601; RRID:AB_2565041PE anti-human/mouse CXCR7BioLegendCat# 331104; RRID: AB_1227619CD31 (PECAM-1) (F8M3S) Rabbit Monoclonal AntibodyCell Signaling TechnologyCat# 95462; RRID: AB_3739873Alexa Fluor 647-AffiniPure Fab Fragment Goat Anti-Rabbit IgGJackson ImmunoResearch LabsCat# 111-607-008; RRID: AB_2632470Goat anti-Mouse IgG2a Cross-Adsorbed Secondary Antibody, Alexa Fluor 568Thermo Fisher ScientificCat# A-21134; RRID: AB_2535773Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, Alexa Fluor 488Thermo Fisher ScientificCat# A-21121; RRID: AB_2535764Goat anti-Rat IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 488Thermo Fisher ScientificCat# A-11006; RRID: AB_2534074Goat anti-Rat IgG (H + L) Secondary Antibody, BiotinThermo Fisher ScientificCat# 31830; RRID: AB_228355Goat Anti-Mouse IgG Antibody (H + L), BiotinylatedVector LaboratoriesCat# AI-9200-1.5; RRID: AB_3107016Biological samplesHuman soleus muscleJuntendo University HospitalN/AChemicals, peptides, and recombinant proteinsCyclosporine (Sandimmun®)Novartis PharmaNDC 00078-0109DAPIThermo Fisher ScientificCat# D1306Propidium Iodide SolutionBioLegendCat# 421301Dispase IIThermo Fisher ScientificCat# 17105041CollagenaseWakoCat# 034-22363Collagenase Type IIWorthington BiochemicalCat# LS004176Collagenase type XISigma-AldrichCat# H3506Soybean trypsin inhibitorWorthington BiochemicalCat# LS003571Lyophilized elastaseWorthington BiochemicalCat# LS002292Hyaluronidase type ISigma-AldrichCat# H4034ACK Lysing BufferLonzaCat# 10-548EPercollCytivaCat# 17089102Debris Removal SolutionMiltenyi BiotecCat# 130-109-398Bovine serum albuminSigma-AldrichCat# A9418-10GMinimum Essential Medium Eagle-alpha modificationSigma-AldrichCat# M8042-500MLFetal bovine serumNichirei BiosciencesCat# 175012L-glutamineThermo Fisher ScientificCat# 25030081Penicillin/streptomycinThermo Fisher ScientificCat# 15140122RPMI-1640Sigma-AldrichCat# R8758VEGF 165, Human, RecombinantPeproTechCat# 100-20-10ug2-mercaptoethanolThermo Fisher ScientificCat# 21985023EGM-2 Endothelial Cell Growth Medium-2 BulletKitLonzaCat# CC-3162Opal 480 Reagent PackAkoya BiosciencesCat# FP1500001KTOpal 570 Reagent PackAkoya BiosciencesCat# FP1488001KTOpal 620 Reagent PackAkoya BiosciencesCat# FP1495001KTNew Hematoxylin Type GMuto Pure ChemicalsCat# 30161VECTASTAIN Elite ABC KitVector LaboratoriesCat# PK-6100optimal cutting temperature (OCT) compoundSakura FinetekCat# 45833Critical commercial assaysRNAscope 2.5 Duplex Reagent KitAdvanced Cell DiagnosticsCat# 322430Chromium Next GEM Single Cell 3′ Kit v3.110x GenomicsCat# PN-1000269Deposited dataMouse and Human Single-Cell RNA seq dataThis studyGEO: GSE303731Mouse Liver Single-Cell RNA seq dataSu et al. (2021)GEO: GSE147581Mouse Liver Single-Cell RNA seq dataLi et al. (2024)GEO: GSE225786Mouse Liver Single-Cell RNA seq dataXiong et al. (2019)GEO: GSE129516Mouse Brain Single-Cell RNA seq dataProfaci et al. (2024)GEO: GSE267591Mouse Heart Single-Cell RNA seq dataLi et al. (2019)GEO: GSE132880Mouse Lung Single-Cell RNA seq dataRodor et al. (2022)GEO: GSE154959Mouse Lung Single-Cell RNA seq dataReyfman et al. (2019)GEO: GSE121611Mouse Single-Cell RNA seq dataSchaum et al. (2018)GEO: GSE109774Human Vascular Atlas Single-Cell RNA seq dataBarnett et al. (2024)https://www.vascularcellatlas.org/Experimental models: Cell linesMouse: OP9Riken BRCRCB1124Experimental models: Organisms/strainsC57BL/6Japan SLCN/ASoftware and algorithmsImageJ/Fiji v2.16Schindelin et al. (2012)https://imagej.net/10x Genomics Cell Ranger v7.1.0Zheng et al. (2017)https://www.10xgenomics.com/Seurat v4.4.0Hao et al. (2021)https://satijalab.org/seurat/tidyverse v2.0Wickham et al. (2019)https://tidyverse.org/clusterProfiler v4.7.1.002Wu et al. (2021)https://bioconductor.org/packages/clusterProfiler/Scanorama v1.7.4Hie et al. (2019)https://github.com/brianhie/scanoramaSCENIC v1.3.1Aibar et al. (2017)https://github.com/aertslab/SCENICharmony v1.2.0Korsunsky et al. (2019)https://CRAN.R-project.org/package=harmonyCellPhoneDB v5.0.1Troulé et al. (2025)https://github.com/ventolab/CellphoneDBCellChat v1.6.1Jin et al. (2021)https://github.com/jinworks/CellChatFlowJo softwareBD Bioscienceshttps://www.flowjo.com/Photoshop CCAdobe Systemshttps://www.adobe.com/R v4.2.2R Core Team, 2022https://www.r-project.org/reticulate v1.44.1GitHubhttps://rstudio.github.io/reticulate/org.Hs.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/org.Hs.eg.db/org.Mm.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/org.Mm.eg.db/Orthology.e.g.,.db v3.16Bioconductorhttps://bioconductor.org/packages/Orthology.eg.db/gplots v3.1.3.1CRANhttps://CRAN.R-project.org/package=gplotspheatmap v1.0.12CRANhttps://CRAN.R-project.org/package=pheatmapEnhancedVolcano v1.16GitHubhttps://github.com/kevinblighe/EnhancedVolcanoloomR v0.2.1.9000GitHubhttps://github.com/mojaveazure/loomRSCopeLoomR v0.13GitHubhttps://github.com/aertslab/SCopeLoomRAngioTool 2.0GitHubhttps://github.com/jbendtsen/AngioTool-Batchktplots v2.4.0GitHubhttps://github.com/zktuong/ktplotsOtherRNAscope™ Probe- Hs-GPM6AAdvanced Cell DiagnosticsCat# 470601RNAscope™ Probe- Hs-CDH11Advanced Cell DiagnosticsCat# 448381RNAscope™ Probe- Hs-NRG3Advanced Cell DiagnosticsCat# 893971-C1RNAscope™ Probe- Hs-PECAM1-O1-C2Advanced Cell DiagnosticsCat# 487381-C2RNAscope™ Probe-Mm-Bst1-O1-C1Advanced Cell DiagnosticsCat# 1868681-C1RNAscope™ Probe-Mm-Pecam1-C2Advanced Cell DiagnosticsCat# 316721-C2RNAscope™ Probe-Mm-Cxcl12-C4Advanced Cell DiagnosticsCat# 422711-C4 Experimental model and study participant details Mice All animal experiments were approved by the Animal Care and Use Committees of Osaka University and Kanazawa University and were performed in accordance with the Science Council of Japan Guidelines for Proper Conduct of Animal Experiments. C57BL/6J mice were obtained from the Japan SLC. Animals were maintained in a clean conventional condition at Kanazawa University on a 12-hour light/dark cycle. Human samples Human skeletal muscle specimens were collected from patients at Juntendo University in accordance with protocols approved by the Research Ethics Committee of Juntendo University School of Medicine (approval no. H16-0101) and the Medical Ethics Committee of Kanazawa University (approval no. 2022-001 [82639]). Written informed consent was obtained from all participants. All samples were de-identified prior to downstream analyses. The specimens were used for flow cytometry, endothelial cell isolation, scRNA sequencing, cell culture assays, immunofluorescence analysis, and in situ hybridization assays. Cell culture OP9 cells (RIKEN Cell Bank, Tsukuba, Japan) were maintained in Minimum Essential Medium Eagle-alpha modification (αMEM) (Sigma-Aldrich, Tokyo, Japan) supplemented with 20% fetal bovine serum (FBS) (Nichirei Biosciences, Tokyo, Japan), 2 mM l-glutamine (Thermo Fisher Scientific, Waltham, MA, USA), and 1% penicillin/streptomycin (p/s) (Life Technologies, Tokyo, Japan). Method details Mouse cell preparation and cell sorting C57BL/6J mice (8 to 10 weeks old) were purchased from Japan SLC (Hamamatsu, Japan). Cells were isolated as previously described (Naito et al., 2020b). Briefly, organs were excised, minced, and digested with dispase II (Thermo Fisher Scientific), collagenase (Wako, Osaka, Japan), and type II collagenase (Worthington Biochemical Corp., Lakewood, NJ, USA) at 37°C with shaking. The cell suspension was filtered through 40 μm strainers, and erythrocytes were lysed using ACK Lysing Buffer (Lonza Japan, Tokyo, Japan). Cells were then treated with Fc blocker (clone 2.4G2, CYTEC Biosciences, Fremont, CA, USA), followed by surface antigen staining with anti-CD31 (clone 390, BioLegend, San Diego, CA, USA), anti-CD45 (clone 30-F11, Thermo Fisher Scientific), anti-CD157 (clone BP-3, BioLegend), and/or anti-CD201 (clone RCR-16, BioLegend) antibodies. Propidium iodide (PI, BioLegend) was added prior to FACS analysis to enable exclusion of dead cells. The stained cells were analyzed and sorted using a FACSAria SORP (BD Biosciences, San Jose, CA, USA) or CytoFLEX SRT (Beckman Coulter, Brea, CA, USA). Data were analyzed using FlowJo software (BD Biosciences). Human cell preparation and cell sorting Soleus muscle tissue was obtained from patients undergoing lower limb amputation and the cut edge was used for analysis. Immediately after resection, the samples were washed with saline and stored at 4°C in phosphate-buffered saline (PBS) containing 4% FBS (Nichirei Biosciences). For in situ hybridization, tissue was immediately fixed in 10% formalin. For single-cell analysis, the tissue was minced into small pieces, digested with dispase II, and then further digested in an enzyme cocktail containing 3 mg/mL collagenase type II, 0.15 mg/mL collagenase type XI (Sigma-Aldrich), 0.25 mg/mL soybean trypsin inhibitor (Worthington Biochemical Corp.), 0.1875 mg/mL lyophilized elastase (Worthington Biochemical Corp.), and 0.24 mg/mL hyaluronidase type I (Sigma-Aldrich) in 4% FBS/PBS supplemented with 1 mM CaCl2. The cell suspension was filtered through 40 μm strainers, and erythrocytes were lysed using ACK Lysing Buffer. Cells were then treated with Fc blocker (BioLegend), followed by surface antigen staining with anti-CD31 (clone WM59, BioLegend) and anti-CD45 (clone HI30, BioLegend). The stained cells were gated on FSC/SSC to exclude debris, and doublets and PI-positive dead cells were excluded. CD31+CD45- endothelial cells (ECs) were then sorted using a CytoFLEX SRT. For cell culture assays, the tissue was minced into small pieces and digested with dispase II, collagenase and type II collagenase at 37°C with shaking. The resulting cell suspension was filtered through 40 μm strainers, and erythrocytes were lysed using ACK Lysing Buffer. Remaining debris was removed using Debris Removal Solution (Miltenyi Biotec, Bergisch Gladbach, Germany) and the cells were further enriched by density gradient centrifugation using Percoll (Cytiva, Tokyo, Japan). The cells were then treated with Fc blocker, followed by surface antigen staining with anti-CD31 and anti-CD45 and anti-CD157 (clone RF3, MBL, Tokyo, Japan). The stained cells were gated on FSC/SSC to exclude debris, and doublets and PI-positive dead cells were excluded. CD31+CD45-CD157+ and CD31+CD45-CD157- ECs were then sorted using a CytoFLEX SRT. Single-cell RNA sequencing Single-cell capture and library preparation were performed using the Chromium Next GEM Single Cell 3′ Kit v3.1 with Dual Index (10x Genomics, Pleasanton, CA, USA), following the manufacturer’s instructions. Fresh mouse-derived and patient-derived single-cell suspensions were diluted to 1000 cells/μL in PBS containing 0.1% bovine serum albumin (BSA) (Sigma-Aldrich). The sequencing library was constructed according to the manufacturer’s instructions. Sequencing was performed on a DNBSEQ-G400 platform (MGI Tech, Shenzhen, China) by Biken Biomics Co., Ltd. (Osaka, Japan), using 28 bp × 100 bp paired-end reads with dual indexing. Sequenced reads were demultiplexed and mapped to the GRCm39 (mouse) or GRCh38 (human) reference genome (Ensembl v98) using CellRanger v7.1.0 (10x Genomics). Single-cell RNA-seq data analysis Single-cell RNA-seq data were processed and analyzed using the Seurat package (version 4.4.0). The percentage of mitochondrial unique molecular identifier (UMI) counts was calculated using the Seurat function PercentageFeatureSet. Cells with feature counts or mitochondrial gene UMI count ratios exceeding either twice the peak value or the 95th percentile in their respective histograms were excluded as part of quality control filtering. Gene expression matrices were normalized using the LogNormalize function with default parameters. Cell cycle regression was performed using the CellCycleScoring function, which assigns S and G2/M scores and classifies cells into G1, S, or G2/M phases. For data integration and batch correction, two complementary workflows were employed, and their results were compared for consistency and robustness. In the first workflow, data were integrated using an optimized Seurat standard RPCA integration pipeline, followed by batch correction using the Harmony algorithm (Korsunsky et al., 2019). Briefly, highly variable genes (top 2000) were identified using FindVariableFeatures, and integration features were selected across datasets using SelectIntegrationFeatures with default parameters. Unwanted variation—including mitochondrial UMI counts, RNA feature counts, total RNA counts, and cell cycle scores—was regressed out using ScaleData. Principal component analysis (PCA) was then performed using RunPCA. Integration anchors were identified using FindIntegrationAnchors with reduction = “rpca”, and datasets were integrated using these anchors. After integration, ScaleData was again used to regress out unwanted variation, and highly variable genes (top 3000) were re-identified using FindVariableFeatures. PCA was repeated with RunPCA, and the optimal number of dimensions was determined using a modified ElbowPlot approach (Harvard Chan Bioinformatics Core). Batch correction was performed using the Harmony algorithm via RunHarmony. Dimensionality reduction and clustering were performed using RunUMAP, FindNeighbors, and FindClusters. In the second workflow, data were normalized and technical and biological variation was regressed using SCTransform. Integration features were selected using SelectIntegrationFeatures, and the datasets were subsequently integrated using the Scanorama algorithm (version 1.7.4). For the integration of mouse and human single-cell RNA-seq data from muscle tissue, genes commonly expressed in both datasets were extracted. The mouse gene expression matrix was then converted to human gene symbols prior to integration. Specifically, genome-wide annotation packages for both mice and humans, along with an orthology mapping package, were used to convert mouse gene identifiers to their human orthologs. After conversion, the gene expression matrices from the two species were compared. Only genes detected in both datasets were retained for further analysis; all others were excluded. Gene ontology (GO) pathway enrichment analysis was performed using the ClusterProfiler package (version 4.7.1.002). Adjusted p-values were used to determine significance. Cluster-specific transcription factor activity was analyzed using the SCENIC package (version 1.3.1). Cell–cell interaction analyses were conducted using CellPhoneDB (version 5.0.1) and CellChat (version 1.6.1), and visualization of CellPhoneDB results was performed using the ktplots package (version 2.4.0). Cell culture and primary endothelial colony forming assay Primary mouse ECs were isolated and plated on the OP9 stromal cells in 24-well plates at a density of 3 × 103 cells/well for CD157 and CD201 EC culture experiments. The cells were cultured in RPMI-1640 (Sigma-Aldrich) supplemented with 10% FBS, 10 ng/mL vascular endothelial growth factor (VEGF) (PeproTech, Rocky Hill, NJ, USA), 0.1% 2-mercaptoethanol (Thermo Fisher Scientific), and 1% p/s. Primary human ECs were isolated and seeded onto OP9 stromal cells in 96-well plates at a density of 5 ×103 cells/well. Both CD157+ and CD157- EC fractions were cultured under identical conditions in complete EGM-2 medium supplemented with 10% FBS and 10 ng/mL VEGF. The cells were fixed for immunostaining after 10 days of culture. Immunostaining For confocal microscopy images, sections were prepared as previously reported (Wakabayashi et al., 2018). In brief, mouse liver, lung, and hind limb tissues, as well as human soleus muscle tissue, were collected, fixed in 4% paraformaldehyde/PBS (PFA/PBS), washed with PBS, embedded in optimal cutting temperature (OCT) compound (Sakura Finetek, Tokyo, Japan), sectioned (10 or 40 μm), and mounted on MAS-coated slide glass (#S9115, Matsunami Glass, Osaka, Japan). The mouse sections were stained with anti-CD31 monoclonal antibody (clone MEC13.3, BD Biosciences), anti-CD157 monoclonal antibody (clone BP-3, BioLegend), anti-CD201 monoclonal antibody (clone RCR-16, BioLegend), anti-αSMA monoclonal antibody (clone 1A4, BioLegend) and/or anti-Cxcr7 monoclonal antibody (clone 8F11-M16, BioLegend). Alexa Fluor 488-conjugated anti-rat IgG (Thermo Fisher Scientific), Alexa Fluor 594-conjugated anti-rat IgG (Thermo Fisher Scientific), and/or Alexa Fluor 568-conjugated anti-mouse IgG2a (Thermo Fisher Scientific) were used as secondary antibodies. The human sections were stained with anti-CD31 monoclonal antibody (clone F8M3S, Cell Signaling Technology, Danvers, MA, USA), anti-CD157 monoclonal antibody (clone RF3, MBL), and anti-αSMA monoclonal antibody (clone 1A4, BioLegend). Alexa Fluor 647-conjugated anti-rabbit IgG (Jackson ImmunoResearch Laboratories, West Grove, PA, USA), Alexa Fluor 568-conjugated anti-mouse IgG2a (Thermo Fisher Scientific) and Alexa Fluor 488-conjugated anti-rat IgG1 (Thermo Fisher Scientific) were used as secondary antibodies. The sections were visualized using the Andor Dragonfly 200 system (Oxford Instruments, High Wycombe, UK) equipped with Fusion Stitcher software and built on a Nikon ECLIPSE Ti2 inverted microscope (Nikon Corporation, Tokyo, Japan). Images were processed using Adobe Photoshop CC software (Adobe Systems, San Jose, CA, USA). All images shown are representative of more than three independent experiments. For immunohistochemistry of colony-formation assays, anti-CD31 antibody was used for staining and biotin-conjugated polyclonal anti-rat immunoglobulin (IgG) (Thermo Fisher Scientific) was used as the secondary antibody for mouse samples, whereas biotin-conjugated anti-mouse IgG antibody (Vector Laboratories, Newark, CA, USA) was used for human samples. Biotinylated secondary antibodies were developed using Elite-ABC kits (Vector Laboratories). Samples were visualized using an IX83 microscope equipped with an automated stage (Olympus Life Sciences, Tokyo, Japan). The vascular area was quantified using ImageJ, and the number of branches and average network lengths were calculated using AngioTool 2.0. In situ hybridization (ISH) assay For human tissue, in situ hybridization (ISH) was performed on formalin-fixed, paraffin-embedded (FFPE) sections using the RNAscope 2.5 Duplex Reagent Kit (Advanced Cell Diagnostics, Newark, CA, USA) according to the manufacturer’s protocol. Briefly, for human tissues, soleus muscle was collected from the cut edge of resected tissue from patients undergoing lower limb amputation. Samples were fixed in 10% neutral buffered formalin, dehydrated in ethanol, and embedded in paraffin. Sections (4 μm thick) were processed using standard pretreatment procedures. Probes specific for GPM6A mRNA (RNAscope™ Probe-Hs-GPM6A), CDH11 mRNA (RNAscope™ Probe-Hs-CDH11), NRG3 mRNA (RNAscope™ Probe-Hs-NRG3), and PECAM1 mRNA (RNAscope™ Probe-Hs-PECAM1-O1-C2) were obtained from the manufacturer and hybridized using the HybEZ™ Hybridization System (Advanced Cell Diagnostics). Signals were detected using green horseradish peroxidase (HRP)-based and red alkaline phosphatase (AP)-based chromogenic assays. Tissues were counterstained with New Hematoxylin Type G (Muto Pure Chemicals, Tokyo, Japan), and bright-field images were captured using a IX83 microscope. For mouse tissues, multiplex fluorescence ISH was performed on liver, hindlimb muscle, and lung sections using the RNAscope Multiplex Fluorescent Reagent Kit v2 (Advanced Cell Diagnostics) with probes targeting Pecam1 (RNAscope™ Probe-Mm-Pecam1-C2), Bst1 (RNAscope™ Probe-Mm-Bst1-O1-C1), and Cxcl12 (RNAscope™ Probe-Mm-Cxcl12-C4). Fluorescently labeled sections were imaged by confocal microscopy using a Dragonfly 200 system. Quantification and statistical analysis All experiments were performed with at least three biological replicates unless otherwise noted. All data are presented as means ± SEM. Statistical comparisons between two groups were conducted using the Steel–Dwass–Critchlow–Fligner test in the open-source package R (v4.2.2). P values of P < 0.05 (∗P < 0.05; ∗∗P < 0.005; ∗∗∗P < 0.0005) were considered to indicate statistical significance. Replicates are indicated in the figure legends. All graphs were generated using R.