Single-cell transcriptome profiling reveals vascular endothelial cell heterogeneity in human skin

Abstract

Vascular endothelial cells (ECs) are increasingly recognized as active players in intercellular crosstalk more than passive linings of a conduit for nutrition delivery. Yet, their functional roles and heterogeneity in skin remain uncharacterized. We have used single-cell RNA sequencing (scRNA-seq) as a profiling strategy to investigate the tissue-specific features and intra-tissue heterogeneity in dermal ECs at single-cell level. Methods: Skin tissues collected from 10 donors were subjected to scRNA-seq. Human dermal EC atlas of over 23,000 single-cell transcriptomes was obtained and further analyzed. Arteriovenous markers discovered in scRNA-seq were validated in human skin samples via immunofluorescence. To illustrate tissue-specific characteristics of dermal ECs, ECs from other human tissues were extracted from previously reported data and compared with our transcriptomic data. Results: In comparison with ECs from other human tissues, dermal ECs possess unique characteristics in metabolism, cytokine signaling, chemotaxis, and cell adhesions. Within dermal ECs, 5 major subtypes were identified, which varied in molecular signatures and biological activities. Metabolic transcriptome analysis revealed a preference for oxidative phosphorylation in arteriole ECs when compared to capillary and venule ECs. Capillary ECs abundantly expressed HLA-II molecules, suggesting its immune-surveillance role. Post-capillary venule ECs, with high levels of adhesion molecules, were equipped with the capacity in immune cell arrest, adhesion, and infiltration. Conclusion: Our study provides a comprehensive characterization of EC features and heterogeneity in human dermis and sets the stage for future research in identifying disease-specific alterations of dermal ECs in various dermatoses.

Keywords: metabolic diversity; single-cell RNA sequencing; skin arteriovenous markers; skin vessel heterogeneity; vascular endothelial cells.

Figure 1

Transcriptionally distinct subtypes in human dermal ECs via scRNA-seq. (A) Workflow. Skin tissues from 10 healthy donors were used for scRNA-seq. CD31⁺ cell enriched by MACS were further purified to select live CD31⁺ CD45^- cells by FACS. 10X Genomics Chromium scRNA-seq was used to profile the cells. (B) UMAP plot of 33,265 human dermal endothelial cells, colored by cluster. (C) Expressions of SEMA3G, PLVAP, SELE, ACKR1, FBLN2, LYVE1, and PROX1 in each cluster. y axis represents log-normalized expression. (D) Heatmap of single-cell expressions of the top-10 DEGs in each cluster. Cluster A: arteriole ECs; Cluster C1, C2: capillary ECs; Cluster P: post-capillary ECs; Cluster V: venule ECs; Cluster L1, L2: lymphatic endothelial cells. A.U.: arbitrary unit; DEGs: differentially expressed genes; ECs: vascular endothelial cells; FACS: fluorescence activated cell sorting; MACS: magnetic-activated cell sorting; scRNA-seq: single-cell RNA sequencing; UMAP: uniform manifold approximation and projection.

Figure 2

Heterogeneous expressions and functions of dermal ECs along the arteriovenous axis. (A) Zonational expression of representative genes across dermal vascular bed. ECs are rearranged in an arteriovenous order. y axis represents counts. (B-E) Immunofluorescence of IGFBP3, RBP7, SELE, MT2A and CD31. Images of CD31⁺ ECs in the superficial, intermediate, and deep plexus are zoomed in. The image is representative of 11 biological replicates (limbs, n = 4; trunk, n = 4; foreskin, n = 3). The white dotted line marks the interface between epidermis and dermis. Scale bars represent 20 µm. (F) Single-cell trajectory of dermal EC subtypes by pseudotime analysis. The white line traces the trajectory. ECs in the left panel are colored by clusters, and white numbers in black circles mark the branch nodes of the pseudotime trajectory. ECs in the right panel plot are colored by pseudotime. Color scale: yellow, 'leaf' of the trajectory; purple, 'root' of the trajectory. (G) GO pathway enrichment analysis of the marker genes of dermal EC clusters. The color represents -log p value, and the size indicates the rich factor. Cluster A: arteriole ECs; Cluster C1, C2: capillary ECs; Cluster P: post-capillary ECs; Cluster V: venule ECs. ECs: vascular endothelial cells; GO: gene ontology; UMAP: uniform manifold approximation and projection.

Figure 3

Tissue-specific heterogeneity of ECs from 10 different tissues. (A) UMAP visualization of 10,417 ECs from 10 different tissues/organs with cells colored by their origin. The numbers in brackets after each tissue/organ name indicate the cell numbers in the corresponding tissue/organ. The dotted line circles ECs originated from dermis. (B) Circular heatmap of average expressions of top-10 DEGs in ECs from 10 tissues/organs. (C) Signature gene expressions of dermal ECs. (D) GO annotation enrichment analysis of DEGs in dermal ECs. x axis represents the rich factor, the circle size represents the -log p value, and the color represents the -log q value. DEGs: differentially expressed genes; ECs: vascular endothelial cells; GI: gastrointestinal; GO: gene ontology; SAT: subcutaneous adipose tissue; UMAP: uniform manifold approximation and projection.

Figure 4

Metabolic heterogeneity in ECs across tissues and within dermis. (A) KEGG analysis of metabolic pathways across ECs from different tissues/organs. The circle size represents the rich factor, and the color represents the -log p value. (B) Heatmap of expressions of representative metabolic genes among ECs from different tissues. (C) Heatmap of expressions of oxidative phosphorylation and glycolytic genes across 5 dermal EC clusters. (D) GSEA analysis of the dataset of dermal ECs against the oxidative phosphorylation geneset. (E) GSEA analysis of the dataset of dermal ECs against the glycolysis/gluconeogenesis geneset. A positive enrichment score on the y axis indicates positive correlation with cluster A and a negative enrichment score indicates a negative correlation. Cluster A: arteriole ECs; Cluster C1, C2: capillary ECs; Cluster P: post-capillary ECs; Cluster V: venule ECs. ECs: vascular endothelial cells; FDR: false discovery rate; GI: gastrointestinal; GSEA: gene set enrichment analysis; KEGG: kyoto encyclopedia of genes and genomes; NES: normalized enrichment score; SAT: subcutaneous adipose tissue.

Figure 5

The expressions of HLA-II genes in dermal capillary ECs. (A) Expressions of selected cytokines and receptors, and HLA-II molecules in ECs from 10 different tissues/organs. (B) Antigen-presenting scores of cells in 5 dermal EC subtypes. The mean score is labeled with a dotted line. Cluster C2 with the highest score is indicated by a hash key. (C, D) Immunofluorescence of HLA-DR, HLA-DQ, and CD31. Images of CD31⁺ ECs in superficial, intermediate, and deep plexus are zoomed in. The image is representative of 11 biological replicates (limbs, n = 4; trunk, n = 4; foreskin, n = 3). The white dotted line marks the interface between epidermis and dermis. Scale bars represent 20 µm. Cluster A: arteriole ECs; Cluster C1, C2: capillary ECs; Cluster P: post-capillary ECs; Cluster V: venule ECs. A.U.: arbitrary unit; ECs: vascular endothelial cells; HLA-II: human lymphocyte antigen class II; UMAP: uniform manifold approximation and projection.

Figure 6

Susceptible cellular adhesion site in dermal post-capillary venules. (A) Gene expressions of representative chemokines and receptors, integrins, and adhesion molecules in ECs from 10 different tissues/organs. (B) Dot plot for expressions of adhesion molecules in dermal EC subtypes. The color represents scaled average expression of adhesion molecules in each subtype, and the size indicates the proportion of cells expressing adhesion molecules. (C) Adhesion scores of cells in 5 dermal EC subtypes. The mean score is labeled with a dotted line. Cluster P with the highest score is indicated by a hash key. (D, E) Representative immunofluorescence images of ICAM1, SELP, and CD31 in human skin. Images of CD31⁺ ECs in superficial, intermediate, and deep plexus are zoomed in. The image is representative of 11 biological replicates (limbs, n = 4; trunk, n = 4; foreskin, n = 3). The white dotted line marks the interface between epidermis and dermis. Scale bars represent 20 µm. Cluster A: arteriole ECs; Cluster C1, C2: capillary ECs; Cluster P: post-capillary ECs; Cluster V: venule ECs. A.U.: arbitrary unit; ECs: vascular endothelial cells; UMAP: uniform manifold approximation and projection.