Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Workflow for mapping circulating endothelial cells (CECs) to their tissue of origin through construction of a comprehensive scRNAseq human endothelial cell (EC) reference atlas. Schematic representation showing a two-dimensional scRNAseq UMAP projection of the transcriptional manifold of (from left to right): CECs of unknown origin collected from 417 individuals; ECs from the Tabula Sapiens dataset; ECs from 17 healthy tissue datasets. Each dot represents a metacell, with colors representing EC type. These latter reference atlases, comprising 191,175 across 23 organs, enabled the development of a quantitative scoring framework for EC type, state, and tissue identity, into which the CEC model was projected to infer vascular type, state, and tissue of origin. Created with BioRender.com . B. Boxplots showing the log₂-transformed expression of endothelial subtype scores (lymphatic vessel endothelial cells [LVEC] versus blood vessel endothelial cells [BVEC], arterial, venous, capillary ECs) across individual cells, grouped by their original cell-type annotation. C. Heatmap showing literature-derived vascular endothelial type markers across the endothelial Tabula Sapiens metacell atlas. The top color bars indicate endothelial type (top bar; color as in ) and tissue of origin (bottom bar; additional color-code provided separately). The right color bar indicates marker type, with colors matching the corresponding cell type in ). D. 3D scatter plot of absolute BVEC subtype scores (arterial, venous, capillary) per metacell. Dashed circles highlight intermediate groups co-expressing two BVEC subtypes. E. Heatmap of differentially expressed, tissue-specific marker genes in capillary endothelial metacells from the endothelial Tabula Sapiens atlas. The top color bar indicates tissue of origin. F. Immunohistochemistry of human cardiac tissue. The image demonstrates the co-localization of Troponin T3 and the endothelial marker PECAM1 along the cardiac vasculature. Cardiomyocytes are stained for Troponin T2 (white), and nuclei with DAPI (blue).

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Transformation Assay, Expressing, Derivative Assay, Marker, Immunohistochemistry, Staining

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Scatter plots of gene expression applied to filter endothelial cells from the Tabula Sapiens dataset. Each dot represents a metacell. Orange dots represent metacells consistent with an endothelial cell identity that exhibit high expression of CDH5 and PECAM1 , and were filtered to exclude non-endothelial cells. B. Two-dimensional UMAP projection of the transcriptional manifold of Tabula Sapiens endothelial cells. Metacells containing at least 75% cells from the same tissue were annotated by the tissue, while a mixture of tissues was annotated as multi-tissue ECs. C. Lymphatic score per cell across tissues in the endothelial Tabula Sapiens atlas, based on original single-cell annotations. Each dot represents a single cell. D. Kernel density estimation (KDE) plot of lymphatic versus venous scores in salivary gland endothelial cells. The distribution reveals two main populations: one expressing both lymphatic and venous markers, and another enriched for venous signature only, suggesting misannotation of the latter as lymphatic. E. Lymphatic vessel endothelial cells (LVEC) and blood vessel endothelial cells (BVEC) score expression at the cell level for salivary gland lymphatic annotated cells. Each dot represents a cell. Despite potential misannotations, salivary gland LVECs show significantly higher LVEC scores compared to BVECs. F. Scatterplot showing BVEC versus LVEC score expression in the endothelial Tabula Sapiens model. Each dot represents a metacell.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Gene Expression, Expressing

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Scatter plot BVEC versus HSP score expression across metacells in the endothelial Tabula Sapiens model. Colors indicate endothelial type and state. Each dot represents a metacell. B. Heatmap of novel endothelial type markers across metacells in the Tabula Sapiens atlas. Genes were selected for high expression and correlation with endothelial type. The top bar indicates endothelial type (top row, color-coded as in Extended Data Fig. 2A) and tissue of origin (bottom row, color-code provided separately). The lower panel shows extended type scores, computing both literature-based and novel markers. C. Bar plot showing the fraction of arterial, capillary, and venous endothelial cells across tissue.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Expressing

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Heatmap of differentially expressed tissue-specific marker genes across arterial endothelial metacells from the endothelial Tabula Sapiens atlas. The top bar indicates the tissue of origin. B. Heatmap of differentially expressed tissue-specific marker genes across venous endothelial metacells from the endothelial Tabula Sapiens atlas. The top bar indicates the tissue of origin. C. Scatter plot of CDH5 versus TNNT3 expression in cardiac endothelial cells from Kanemaru, K., Cranley, J., Muraro, D. et al . High TNNT3 expression is observed in the capillary cardiac endothelial cells. D. Scatter plot of CDH5 versus TNNT3 expression across all metacells in the full Tabula Sapiens atlas, including cardiomyocytes. Each dot represents a metacell. Orange dots represent cardiac endothelial metacells. E. Two-dimensional UMAP projection of the transcriptional manifold of endothelial cells derived from the tissue atlases. F. Heatmap of literature-based vascular-type endothelial markers across metacells in the tissue atlases. The top annotation indicates endothelial type (top bar) and tissue of origin (lower bar).

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Marker, Expressing, Derivative Assay

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Expression of (clockwise) arterial, capillary, and venous endothelial type scores across tissues and models. Each dot represents a metacell. The boxplot color corresponds to the scored endothelial type. B. Heatmap of differentially expressed tissue-specific marker genes across arterial endothelial metacells from the tissue atlases. The top bar indicates the tissue of origin. C. Heatmap of differentially expressed tissue-specific marker genes across capillary endothelial metacells from the tissue atlases. The top bar indicates the tissue of origin. D. Heatmap of differentially expressed tissue-specific marker genes across venous endothelial metacells from the tissue atlases. The top bar indicates the tissue of origin.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Expressing, Marker

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Experimental workflow for creating a scRNA-seq atlas of circulating endothelial cells (CECs). Created with BioRender.com . B. Scatter plot showing expression of canonical endothelial markers PECAM1 and CDH5 across the circulating CD34⁺ metacell manifold. A small subset of cells co-expresses both markers, consistent with circulating endothelial cell identity. C. Two-dimensional UMAP projection of the transcriptional manifold of 597 CECs. Colors indicate the annotated state of each metacell, with small dots representing individual cells assigned to that metacell. D. Heatmap of literature-based vascular type endothelial markers and computed scores across circulating endothelial metacells. The top color bar indicates the annotated endothelial type. E. Boxplots of arterial (left), capillary (center), and venous (right) blood vessel endothelial cells subtype scores in the endothelial Tabula Sapiens (TS) reference atlas and CECs. Each comparison shows subtype-positive and subtype-negative cells within both datasets, showing that CECs consistently exhibit a lower mature vascular subtype relative to tissue-resident endothelial cells. F. Boxplot of angiogenic scores across three groups of endothelial metacells: Tabula Sapiens (TS) angiogenic endothelial cells, TS blood vessel endothelial cells (BVECs), and CECs. CEC metacells show a higher angiogenic score compared to non-angiogenic BVECs and lower than angiogenic ECs. G. Scatter plots of tissue-specific marker gene expression in circulating endothelial metacells (blue) and tissue-matched endothelial metacells from the endothelial Tabula Sapiens reference atlas (orange). Each dot represents a metacell. While circulating endothelial metacells show lower expression of tissue markers, some retain detectable expression, suggesting a possible origin from corresponding mature tissues. The panel shows markers for (clockwise): muscle, pancreas, and heart. H. Histograms of tissue-specific score expression (heart, liver, brain, muscle, and skeletal muscle) across individual circulating endothelial cells, downsampled to 2¹¹ UMIs per cell. While most cells show low expression, a subset displays elevated tissue-specific scores, suggesting a potential tissue-of-origin signature.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Expressing, Comparison, Marker, Gene Expression

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Scatter plot of blood vascular endothelial cells (BVEC) versus lymphatic vascular endothelial cells (LVEC) expression scores across metacells from the endothelial Tabula Sapiens model (gray dots) and from the circulating endothelial cell (CEC) model (blue dots), supporting a BVEC origin for CECs. B. Scatter plot of CDH5 versus S100A10 expression in CEC metacells. A subset of CECs co-express high levels of endothelial and stress-associated markers, suggesting stress-related activation. C. Boxplot of CD34 expression across CECs, Tabula Sapiens (TS) angiogenic ECs, TS BVECs, and circulating CD34 + hematopoietic stem cells (HSCs). CD34⁺ peripheral blood sorting captures diverse endothelial populations, including angiogenic ECs. D. Scatter plot of CDH5 versus heat shock protein (HSP) scores in circulating endothelial metacells. Low HSP expression indicates no distinct HSP-positive cell population. E. Scatter plot of cell cycle scores across CEC metacells, showing two metacells with high M-phase scores, indicative of active cycling. F. Histograms of tissue-specific score expression across circulating endothelial cells. Each histogram shows a different tissue with a subset of cells exhibiting elevated tissue-specific marker expression, supporting that tissue origin.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Expressing, Activation Assay, Marker

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Experimental workflow for creating a scRNA-seq model of cultured circulating endothelial cells (CECs). Created with BioRender.com . B. Phase-contrast microscopy images of cultured CECs after three weeks, showing the characteristic cobblestone morphology of endothelial monolayers. C. Phase-contrast microscopy images of cultured CECs in a Matrigel tube formation assay, CECs organize into capillary-like network structures within six hours of incubation. D. Two-dimensional UMAP projection of the transcriptional manifold of the cultured circulating endothelial cells model, with colors representing the annotated states of each metacell. E. Heatmap of differentially expressed genes across cultured CEC metacells. The top color bar indicates the annotated metacell state. F. Boxplots of arterial, capillary, venous, lymphatic, and angiogenic endothelial subtype scores across metacells from Tabula Sapiens (TS), CEC, and cultured CEC datasets. The boxplot color corresponds to the endothelial type shown in . G. Boxplots of PROCR (left) and CD34 (right) expression levels across endothelial cells from the endothelial Tabula Sapiens (TS) reference atlas, CEC, and cultured CECs at 3 and 6 weeks, showing an increase in PROCR and a decrease in CD34 expression during culture.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Cell Culture, Microscopy, Tube Formation Assay, Incubation, Expressing

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Boxplots of arterial, capillary, venous, lymphatic, and angiogenic endothelial scores across metacells from endothelial Tabula Sapiens (TS), circulating endothelial cells (CEC), and the different subgroups of cultured CEC, showing a progressive decrease of mature type endothelial cells from resident endothelial cells to CECs and through culture. Boxplot colors represent the corresponding endothelial type and state.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Cell Culture

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Experimental workflow for creating a scRNA-seq model of cultured CECs following Notch signaling pathway activation and inhibition. Created with BioRender.com . B. Phase-contrast microscopy images of cultured CECs. Notch signaling activation (right) showing elongated morphological changes compared to standard adherent culture conditions (left). C. Two-dimensional UMAP projection of cultured CEC transcriptional manifold following notch signaling alteration, with colors representing annotated metacell state. D. Boxplots of CD34 (left) and PROCR (right) expression levels in cultured circulating endothelial cells under control conditions (day 0 and day 5), Notch inhibition, and Notch activation. Notch signaling alteration reverses the previously observed trends, with CD34 expression increasing and PROCR expression decreasing. E. Density plots showing angiogenic and blood vascular endothelial cell (arterial, capillary, venous) subtype scores in cultured circulating endothelial cells under different conditions. Notch activation significantly increases angiogenic and capillary scores compared to control. F. Gene Ontology enrichment analysis of Notch-activated circulating endothelial cells. The top ten enrichment programs, ranked by FDR significance, are shown. The analysis was based on differentially expressed genes (Notch activation vs. control) and on the top ten correlated genes from the 20 strongest DE genes (duplicates removed). Results were obtained using STRING v11.5 (Szklarczyk et al., 2021; https://string-db.org ). G. String network analysis of Notch-activated circulating endothelial cells. The analysis was based on differentially expressed genes (Notch activation vs. control) and on the top ten correlated genes from the 20 strongest DE genes (duplicates removed). Interaction evidence was derived from experiments, databases, co-expression, neighborhood, gene fusion, and co-occurrence (minimum score 0.4). Disconnected nodes were omitted; node halos indicate gene expression values, and edge thickness reflects interaction support. Analysis performed with STRING v11.5 (Szklarczyk et al., 2021; https://string-db.org ).

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: Cell Culture, Activation Assay, Inhibition, Microscopy, Expressing, Control, Derivative Assay, Gene Expression

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Two-dimensional UMAP projection of the transcriptional manifold of endothelial cells from the Chen, Y., Pal, B., Lindeman, G.J. et al . breast cancer dataset, showing metacells colored by type and state (left) and by the fraction of cells originating from breast cancer tissue. B. Boxplots of angiogenic cell fraction distributions across individuals in normal breast tissue versus breast cancer tissues (log scale). Only patients with at least 50 cells were included. C. Boxplots of angiogenic cell fraction distributions across individuals in normal breast tissue versus breast cancer tissues, with separation to cancer type (log scale). Only patients with at least 50 cells were included. D. Two-dimensional UMAP projection of the transcriptional manifold of endothelial cells from the Kumar et al. breast cancer dataset, showing metacells colored by annotated endothelial state (left) and by the fraction originating from contralateral non-tumor tissue (right). E. Boxplots of angiogenic cell fraction distributions across normal breast tissue versus contralateral breast cancer tissue (log scale). Only patients with at least 50 cells were included. F. Endothelial cell state composition per patient. Each vertical bar represents one patient, with colors indicating proportions of annotated cell states. Only patients with at least 50 cells are shown.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques:

Journal: bioRxiv

Article Title: Deciphering the Origin and Plasticity of Circulating Endothelial Cells: A Model for Systemic Angiogenesis Programs in Health and Disease

doi: 10.1101/2025.08.26.672338

Figure Lengend Snippet: A. Composition of endothelial cell states at the single-patient level from Chen, Y., Pal, B., Lindeman, G.J. et al. Each column represents a patient, with colors indicating the relative proportion of annotated endothelial type and states. Only patients with at least 50 cells are shown. B. Histogram showing the distribution of contralateral breast tissue patient samples across angiogenic endothelial cell fractions. C. Boxplots of angiogenic cell fraction distributions across normal breast tissue of healthy individuals versus normal contralateral breast tissue of breast cancer patients (log scale). Patients with fewer than 500 cells were excluded.

Article Snippet: Two differential conditions were applied: (1) a non-direct Notch activation, using CultEC growth medium excluding the kit-supplied VEGF, supplemented with only 2% FBS, vascular endothelial growth factor (VEGF; 100 ng/mL, Cat. no. 130-109-383, Miltenyi Biotec), and Activin A (25 ng/mL, Cat. no. 130-115-008, Miltenyi Biotec)., (2) A direct Notch inhibition, using CultEC growth medium supplemented with 10% FBS and the γ-secretase inhibitor RO4929097 (1 μM, Cat. no. sc-364602, Santa Cruz Biotechnology).

Techniques: