Origin, prospective identification, and function of circulating endothelial colony-forming cells in mice and humans

Abstract

Most circulating endothelial cells are apoptotic, but rare circulating endothelial colony-forming cells (C-ECFCs), also known as blood outgrowth endothelial cells, with proliferative and vasculogenic activity can be cultured; however, the origin and naive function of these C-ECFCs remains obscure. Herein, detailed lineage tracing revealed murine C-ECFCs emerged in the early postnatal period, displayed high vasculogenic potential with enriched frequency of clonal proliferative cells compared with tissue-resident ECFCs, and were not committed to or derived from the BM hematopoietic system but from tissue-resident ECFCs. In humans, C-ECFCs were present in the CD34bright cord blood mononuclear subset, possessed proliferative potential and in vivo vasculogenic function in a naive or cultured state, and displayed a single cell transcriptome sharing some umbilical venous endothelial cell features, such as a higher protein C receptor and extracellular matrix gene expression. This study provides an advance for the field by identifying the origin, naive function, and antigens to prospectively isolate C-ECFCs for translational studies.

Keywords: Endothelial cells; Vascular Biology.

Figure 1. Some murine circulating ECFCs form functional blood vessels in vivo and have the ability to self-renew.

(A) Kinetics of emergence of ECFCs in C57BL/6J mouse blood; 3–6 mice per time point. (B) Schematics of lineage tracing using Tie2TT mice. (C) Representative TT⁺ EC colonies derived from PB of Tie2TT mice (155 TT⁺ EC colonies out of 177 colonies from 15 pups). (D) Schematic of collagen plug transplantation (all) and cell injection to hind limb muscle (left and middle) using PB-derived cells (left), PB without OP9 culture (middle), and EC colony from clonal EC culture (right) of Tie2TT mouse (P2). TT⁺ vessels can be digested and replated on OP9, resulting in secondary TT⁺ EC colonies (left). (E) PB-derived (with in vitro OP9 coculture) TT⁺ vessels are inosculated with host vasculature 2 weeks after transplantation (shown by systemic IB4 i.v. injection); 5 successes out of 5 recipients. (F) Uncultured CEC-derived blood vessels (TT⁺) are shown 4 weeks after collagen plug transplantation using PB of Tie2TT (P2); 4 successes out of 4 recipients. Scale bars: 200 μm in C, E, and F. CEC, circulating endothelial cells.

Figure 2. Abundance of murine C-ECFCs with clonal proliferative potential and their vessel-forming capacities with and without prior culture, comparable to human C-ECFCs.

(A) Schematic of single-cell colony-forming assay using CD45^–Ter119^–CD31⁺TT⁺ cells in Tie2TT mice (P2). Right panel shows a representative picture of an EC colony. TT⁺ EC colonies were confirmed at least 4 times. (B) Quantitation of the frequency of ECFCs from PB, lung, or heart-derived CD45^–Ter119^–CD31⁺TT⁺ cells in Tie2CreTT mice (P2). n = 3–4. Data are shown as the mean ± SD. **P < 0.01. Tukey-Kramer post hoc test. (C) Uncultured and cultured human CB CD34⁺CD45^– cells (MACS sorted) can form functional blood vessels in vivo after transplantation; 4 successes out of 4 recipients of uncultured cells. (D) Representative TT⁺ vessel of uncultured and cultured Tie2TT PB injection to hind limb muscle; 2 successes out of 4 recipients of uncultured cells. Scale bars: 200 μm (C), 50 μm (D).

Figure 3. Murine C-ECFCs are derived from resident ECs.

(A and B) Quantitation of labeling efficiency of endothelial and hematopoietic lineage, and resultant EC colonies in each tissue/organ from Tie2TT and Cdh5TT mice. Data are shown as the mean ± SD; colonies were counted in at least 3–4 independent experiments. (C) Representative TT⁺ EC colonies derived from PB of Cdh5TT mice (52 TT⁺ EC colonies out of 56 colonies, see also B). (D) Schematics of lineage tracing using hematopoietic specific Flt3TG mice. (E) Representative TT⁺ EC colonies (left) and GFP⁺ HC colonies (right) derived from PB of Flt3TG mice (99 TT⁺ EC colonies out of 99 colonies, see also F). (F) Percentage of GFP- and TT-expressing cells in the fraction of lung EC, lung HC, BM KSL, PB HC, PB B220⁺, PB CD11b⁺, and PB CD3⁺. All EC colonies derived from peripheral MCs are TT⁺. Data are shown as the mean ± SD; GFP or TT expression was evaluated in at least 7 independent experiments. Scale bars: 200 μm. KSL, c-Kit⁺Sca1⁺Lineage^– cell.

Figure 4. scRNA-Seq reveals EC-related clusters in circulating CBMCs.

(A) Schematic of scRNA-Seq assay using freshly isolated CBMCs and HUVECs, using Ficoll density centrifugation or liberase enzymatic digestion, respectively. CD34⁺CD235a⁻CD45⁻ cells are enriched by MACS. (B and C) Global UMAP plots of CBMCs and HUVECs and dot plots of scaled average expression of major canonical markers (columns) in all clusters of CBMCs (rows). (D) Violin plots of EC genes comparing EC marker-expressing clusters of CBMCs.

Figure 5. C16 is a candidate cluster enriched for C-ECFC population characterized by PROCR expression and may be derived from resident vascular ECs.

(A) SCENIC analysis result on CBMC clusters. Binary regulon activity matrix is shown with AUCell scores, indicating the activity of each TF regulon in each cell. Some significantly enriched or reduced TF regulons in C16 are shown at right (with related gene number). (B) Volcano plots comparing C16 versus C3, C4, C5, and C7. Representative genes are shown. PROCR is the only cell surface marker gene in the plot with a significant P value. (C) Integrated UMAP of CBMCs and HUVECs (left above). CBMC EC clusters and C16 highlighted (left below and right below), respectively. (D) Heatmap of mean expression of ECM genes and PROCR in annotated clusters. (E) Pseudotime analysis using CBMC ECs (C3, C4, C5, C7, C14, and C16) and HUVECs (w/o C12), showing integrated plot (top) and sample plots (2 in the middle). C16 (red) is highlighted below.

Figure 6. Schematic of in vitro EC colony-forming assay using fresh CB.

After Ficoll separation, cells were magnetically sorted with PROCR-APC and anti-APC beads, followed by culture using cells from each fraction: (F1) and (F2). Below: Cells were magnetically sorted with CD34 MultiSort Beads and CD34⁺ fraction then incubated with beads-releasing reagent. The beads-released CD34⁺ fraction was further sorted with PROCR-APC and anti-APC beads, leading to (F3) and (F4), and followed by placement in culture. Beads-unreleased CD34⁺ fraction (enriched with CD34^bright population) (F5) was also cultured.

Figure 7. Colony-forming potential of PROCR^hi and CD34^bright endothelial population in CBMC identifies C-ECFCs within CECs.

(A and B) Flow cytometry analysis of pre- and post-MACS for each fraction (F1–F2 in A, F3–F5 in B). Quantitation of the number of EC colonies per 100K cells (graph). Table (above, yellow; below, blue) shows seeding cell fraction (with cell number per well) and EC colony number in each experiment. E, experiment; NT, not tested.

Figure 8. Characterization of PROCR^hi and CD34^bright cells and overall schematic of human CBMCs classified by PROCR and CD34 expression.

(A) Fold changes of the expression of EC- and HC-related genes in CD34^bright/hi with PROCR^hi/lo fractions after FACS. Schematic of each fraction — (i) CD34^hi and PROCR^lo, (ii) CD34^hi and PROCR^hi, (iii) CD34^bright and PROCR^lo, and (iv) CD34^bright and PROCR^hi — in flow cytometry (right above). Data are shown as the mean ± SD. n = 3. Student’s t tests were used to compare 2 groups — CD34^hi (i) and (ii) versus CD34^bright (iii) and (iv). Tukey-Kramer post hoc test was used to compare multiple groups in PROCR expression. *P < 0.05; **P < 0.01. (B) Schematic of presumed roles from each fraction using CD34 and PROCR expression in flow cytometry based on the results of the EC colony contributing cell fraction from Figure 7A (yellow dots) and Figure 7B (blue dots).