Current concepts on endothelial stem cells definition, location, and markers

Abstract Ischemic vascular disease is a major cause of mortality and morbidity worldwide, and regeneration of blood vessels in perfusion‐deficient tissues is a worthwhile therapeutic goal. The idea of delivering endothelial stem/progenitor cells to repair damaged vasculature, reperfuse hypoxic tissue, prevent cell death, and consequently diminish tissue inflammation and fibrosis has a strong scientific basis and clinical value. Various labs have proposed endothelial stem/progenitor cell candidates. This has created confusion, as there are profound differences between these cell definitions based on isolation methodology, characterization, and reparative biology. Here, a stricter definition based on stem cell biology principles is proposed. Although preclinical studies have often been promising, results from clinical trials have been highly contradictory and served to highlight multiple challenges associated with disappointing therapeutic benefit. This article reviews recent accomplishments in the field and discusses current difficulties when developing endothelial stem cell therapies. Emerging evidence that disputes the classic view of the bone marrow as the source for these cells and supports the vascular wall as the niche for these tissue‐resident endothelial stem cells is considered. In addition, novel markers to identify endothelial stem cells, including CD157, EPCR, and CD31low VEGFR2low IL33+ Sox9+, are described.


| DEFINING ENDOTHELIAL STEM CELLS
Development studies in mice have established the origin of blood vessel formation to occur with the appearance of angioblast precursors derived from early mesoderm, which leads to the formation of the primitive vascular network within the extraembryonic yolk sac at E6.5. 1 These angioblasts that form blood islands also migrate to colonize embryonic tissues. During subsequent vasculogenesis, extensive remodeling of the primitive vasculature leads to formation of a functional vascular network consisting of specialized arterial, venous, lymphatic, and hemogenic endothelial cell subtypes. Colonization of the bone marrow occurs later by hematopoietic stem cells arising from the hemogenic endothelium at E16.5. 2 The close physical, temporal, and spatial association between hematopoietic and endothelial cells during embryonic development has given rise to theories such as the existence of a common bipotent progenitor known as the hemangioblast. While the existence of the hemangioblast is still debated, 3 this relationship between endothelial and hematopoietic cells is also shown in the hemogenic endothelium, which has been defined as a specialized endothelial cell population that gives rise to hematopoietic cells. The shared expression of markers, such as CD34 and VEGFR-2 (in human); or sca1 and flk-1 (in mouse), in cells with capacity for hematopoietic and endothelial differentiation led to the proposal for an adult hemangioblast circulating in peripheral blood. 4 These cells have been studied as putative circulating endothelial progenitor cells (EPCs) with vasoreparative properties. While in preclinical models, transplantation of these EPCs promoted angiogenesis in ischemic tissues, 5 follow on clinical trials have not shown similar efficacy. 6 There is significant controversy defining EPCs, as the term has been broadly used to describe cells that are non-endothelial in nature, such as hematopoietic stem cells (HSCs) and myeloid angiogenic cells (MACs). 7 Furthermore, studies have demonstrated that blood monocytic cells can acquire endothelial markers in culture by uptake of platelet extracellular vesicles. 8 Following basic stem cell biology principles, endothelial stem cells must exhibit capacities for self-renewal and lineage differentiation into organ-specific specialized endothelial cells. Evidence has confirmed a higher cell turnover rate in adult cardiac tissue for endothelial cells (>15% per year), when compared with mesenchymal cells (<4% per year) or cardiomyocytes (<1% per year). 9 Similarly, multi-isotope imaging mass spectroscopy and electron microscopy has revealed significant endothelial cell turnover in brain, liver, and pancreas, with blood vessels being described as mosaics structures consisting of young and aged cells. 10 Furthermore, in the adult bone marrow, a small population of Apelin expressing endothelial cells is critical for vascular regeneration after irradiation. 11 All these data suggest that endothelial homeostasis and maintenance is preserved by a selfrenewing subpopulation of cells with stem/progenitor properties.
During development, initially homogeneous embryonic endothelial precursors acquire distinct identities to support organ relevant function. This endothelial cell heterogeneity in different tissues and organs have been described at the transcriptome and translatome level. 12,13 In addition, endothelial cells can adopt tissue-specific characteristics during organ development and regeneration. 14 This endothelial specialization according to organ microenvironment underscores the plasticity of some endothelial cells, which also suggests the existence of an endothelial stem cell. Hematopoietic stem cells are the quintessential example for a precise stem cell definition, not only by molecular markers but with robust evidence for lineage differentiation and selfrenewal that goes beyond the lifespan of the HSC donor. 15 This scientific stringency is currently missing when defining endothelial stem cells. Therefore, we propose that an endothelial molecular identity coupled with self-renewal and differentiation capacity into various endothelial cell phenotypes should be essential requirements to define endothelial stem cells. While the terms endothelial stem and progenitor are currently used interchangeably, future studies are needed to formulate models for endothelial differentiation hierarchy and distinguish stem from progenitor cells. Until evidence for precise molecular identities is established, we hypothesize that endothelial stem cells must retain the capacity to differentiate into arterial, venous, lymphatic, and capillary endothelial cells as required. This can be demonstrated by their molecular plasticity and adoption of organ specific endothelial characteristics, akin to induced pluripotentderived endothelial cells. 14 Endothelial progenitors, on the other hand, are expected to be committed to targeted differentiation into organ specialized endothelium, suggesting that an endothelial progenitor cell niche exists within highly vascularized organs.
Endothelial colony-forming cells (ECFCs) are a subtype of endothelial progenitor with remarkable clonogenic potential and postnatal vascularization ability in vivo. 16 ECFCs are consistently isolated from human cord blood and there is an established consensus for their definition. 17 ECFCs have been shown to significantly contribute to vascular regeneration of ischemic tissues such as heart, brain, retina, and limbs 18 ; and they are on a translational pathway toward a cell therapy. 19 A systematic review of controlled preclinical animal studies using human ECFCs concluded that while the potential clinical application for ECFCs is evolving rapidly, the wider implementation of already established standardized ECFC characterization will enable more rapid and effective transition to clinical trials. 20

| LOCATION OF ENDOTHELIAL PROGENITORS
EPCs were originally thought to reside in the bone marrow 21 ; however, recent studies in patients with sex-mismatched bone marrow transplantations have shown that these endothelial vasoreparative cells do not originate from bone marrow. 22 ECFCs isolated from venous wall and peripheral blood of male patients, who had previously received bone marrow transplants from female donors, displayed a XY genotype, which negates bone marrow as the ECFC source. In agreement with the idea for a vascular niche, ECFCs have been isolated from human saphenous vein. 23 Evaluation of transcriptomic and proteomic profiles in human peripheral blood-derived ECFCs suggested that they represent an intermediate endothelial cell population between human coronary artery endothelial cells and human umbilical vein endothelial cells. 24 While these studies suggests that ECFCs reside within macrovessels, these progenitors have also been isolated from microvessels of human placenta. 25 Interestingly, microvascular ECFCs show greater vessel forming capacity than macrovascular ECFCs in a Matrigel graft implant model. Furthermore, there is evidence that ECFCs are also resident in white adipose tissue (WAT).
ECFCs isolated from WAT stromal vascular fraction displayed comparable angiogenic properties to umbilical cord and peripheral bloodderived ECFCs. 26 Although evidence suggests ECFCs arise from a vascular niche, their origin and specific location as vascular wall resident endothelial cells in vivo remains to be proven.  Figure 1). This study demonstrated that EVPs, but not TA or D cells, exhibited intrinsic stem/progenitor cell properties, including in vivo self-renewal potential. 29 Unbiased single-cell RNA sequencing from mouse aortas confirmed this endothelial hierarchy, and identified Sox9, Il33, and PDGFRA as molecular markers for EVP, while CD31 and Sox18 identified differentiated cells. 30 Similarly, highly proliferative and vasoreparative CD157 + CD200 + cells were found in large adult murine vessels, which were also identified as tissue resident endothelial stem cells. 31 Interestingly, EVPs highly expressed Pdgfra, a mesenchymal gene which indicates their potential to differentiate into endothelial and mesenchymal cell types. In agreement with this theory, genetic lineage tracing studies and single cell RNA sequencing in the murine model of hindlimb ischemia has shown that a subset of fibroblasts can acquire endothelial genes. 32 The emergence of CD144 + cells within FSP (fibroblast-specific protein)-1+ fibroblasts in ischemic limbs was mediated by Toll-like receptor-3 and nuclear factor κB. On the contrary, in the mouse model of myocardial ischemia, lineage tracing experiments pointed that pre-existing endothelial cells and not fibroblasts, were responsible for the observed neovascularization. 33 These studies highlight the importance of fibroblasts in promoting vascularization, and further investigations are warranted to confirm the existence of a bipotent mesenchymal stem cell or fibroblast with capacity for mesenchymal to endothelial transition.
An unambiguous molecular definition of endothelial stem cells phenotype and status is lacking. However, based on current evidence, it can be considered that these stem cells lie dormant in healthy tissues, but with the capacity to rapidly respond by proliferating, differentiating into endothelium and thereby contributing to vascular homeostasis and repair. Whether they engage with repair by directly differentiating into endothelium or evoke a more complex pathway involving transient amplifying cells is unknown. Clearly further research is needed since the process of vascular repair is complex and expected to involve not only endothelial stem/progenitor cells but also differentiated endothelial cells, mural cells, immune cells, and platelets.

| CLINICAL TRANSLATION CHALLENGES FOR ENDOTHELIAL STEM/ PROGENITOR CELLS
Since the early 90s, there has been considerable interest in endothelial stem/progenitor cells as a potential cell therapy for ischemic diseases. Unfortunately, meta-analysis of clinical trials for ischemic heart disease 34 and critical limb ischemia, 6  action. In summary, the rapid move of endothelial progenitor cell therapies from the labs into the clinics, without having well-defined cells with a proven mechanism of action has negatively impacted the results. We propose some strategies to dispel these challenges (Table 1)   EPCs circulating in human peripheral blood has been often used for clinical studies, and these circulating CD34 + VEGFR-2 + cells are considered a measure of the endogenous reparative capacity of the cardiovascular system. 35 A recent investigation on endothelial progenitor enumeration in blood from patients with COVID-19 or undergoing bioprosthetic total artificial heart implantation, has employed mass cytometry and imaging flow cytometry to demonstrate that circulating CD19 À CD34 + cells are negative for VEGFR-2. Furthermore, VEGFR-2 was only expressed in CD19 + B-cells and CD14 + monocytes. Imaging cytometry showed that events which were CD19 À CD34 + VEGFR-2 + represent cellular particles, fragments, or debris. 36 These results have underscored the need to redefine circulating endothelial progenitors, in line with technological advances, such as mass cytometry, which increases accuracy for molecular definition of cells.

| MARKERS FOR ENDOTHELIAL PROGENITORS
Alongside flow cytometry-based enumeration of putative endothelial progenitors as CD34 + VEGFR-2 + , cell culture-based isolation of ECFCs is another well-accepted way to study human endothelial progenitors. We summarize and discuss molecular markers that have been associated with resident endothelial stem/progenitors (Figure 1 and Supporting Information Table S1).
Additional markers, CD146 and CD144 have been suggested as substitutes for VEGFR-2, or as additional markers to further refine the population. 37 In addition, c-Kit expression was associated with higher CD34 and VEGFR-2 levels. CD34 + c-Kit + have been proposed to be at the origin of endothelial recovery after total artificial heart transplantation. 36 Added to this, c-Kit expression has also been reported in quiescent endothelial stem cells residing in the vascular wall in mouse lung vasculature, expressing lin À CD31 + CD105 + Sca1 + CD117(c-kit) + . 38  VEGFR-2 co-receptor, NRP1, has been identified as a marker of an endothelial precursor, since its expression precedes that of CD31 and CD34 in mouse embryonic studies. 46 Following on from this, a protocol was developed to generate ECFC-like cells from induced pluripotent stem cells that were directly comparable to cord-blood derived ECFCs in vasculogenic potential. During the differentiation process, cells were sorted based upon NRP1 + CD31 + expression, and when expanded demonstrated high clonal proliferative potential and in vivo vessel formation that inosculated with host vasculature in the hindlimb ischemia model and oxygen induced retinopathy model. 47 As such, iPS-derived ECFCs could represent a robust and patient-specific source of ECFCs for cell therapy trials. Alternative approaches for rapid, consistent, and highly efficient differentiation of iPSCs into endothelial cells include the timely activation of transcription factor ETV2. 48,49 However, more research into safety and memory of host susceptibility to disease is important to make this cell therapy a reality.
The glycoprotein von Willebrand factor (VWF) is frequently used as an endothelial cell marker and its expression increases during iPS to endothelial cell differentiation. In addition, VWF regulates blood vessel formation. 50 VWF knockdown in ECFCs led to increased proliferation, migration, and in vitro angiogenesis. 51  capacity vs the CD157 À population. 53 In this protocol, ECFCs were supported using an OP9 stromal cell coculture approach as previously described. 54 Although these results are akin to human ECFCs, their existence in humans has not yet been proven. 31 Nevertheless, CD157 has recently been used to identify ECFCs with highly proliferative potential, generated following differentiation from human pluripotent stem cells (hPSCs Endogenous endothelial stem cells in the mouse aorta have been defined as CD31 low VEGFR2 À/low , which give rise to transit amplifying CD31 int VEGFR-2 À/low and definitive differentiated CD31 high VEGFR-2 high cells ( Figure 1). These endothelial stem cells were also characterized by the expression of IL33 and Sox9. 29  Characterization of side population cells in the mouse lung identified CD45 À CD31 + VEGFR-2 À cells as progenitors capable of differentiating into smooth muscle cells and endothelial cells. 62 In vitro studies determined that CD45 À CD31 + VEGFR-2 À cells differentiate into CD45 À CD31 + VEGFR-2 + cells which give rise to endothelial cells; suggesting VEGFR-2 expression as a sign of late endothelial progenitor commitment. The side population is characterized by their ability to efflux fluorescent dyes via ATP-binding transporters such as ABCG2; and a recent study on lung resident ABCG2 + cells has identified them as mesenchymal vascular progenitors with capacity to modulate adaptive repair angiogenesis in the lung. 63

| CONCLUSION
Stem cell biology is being transformed thanks to technological advances such as single cell RNA sequencing, mass cytometry, imaging cytometry, and cell-fate mapping microscopy. Use of these technologies has revealed shortcomings and inaccuracies in previous approaches to define and locate endothelial progenitors. Here, we emphasized that the endothelial phenotype, coupled with self-renewal and differentiation capacities, must be essential features to define an endothelial stem cell. We also highlight that bone marrow can no longer be accepted as the sole source for isolation of endothelial stem/progenitor cells, as there is evidence to support a vascular niche.
Careful attention should be drawn to research using mouse vs human cells, as some results may not be easily replicated. For example, while isolation of human ECFCs is optimized, most labs find the isolation of mouse ECFCs challenging. Lastly, various surface markers such as CD157 or EPCR have been proposed to identify tissue-resident endothelial stem cells, but more research is warranted to establish overlap and relationships among these markers. The absence of a robust and exhaustive marker to clearly identify endothelial progenitors alongside the inherent species-species variation and the impact of the disease on the cellular properties, highlights the need to define these cells using a profile of combination of markers, rather than a single marker, coupled with appropriate functional assays to establish an operational definition of the endothelial cell differentiation hierarchy.

CONFLICT OF INTEREST
The authors declared no potential conflicts of interest.

AUTHOR CONTRIBUTIONS
S.E.J.C., V.P., E.P.: data collection and assembly, data analysis and

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analyzed in this study. ORCID