Isolating and defining cells to engineer human blood vessels

Abstract

A great deal of attention has been recently focused on understanding the role that bone marrow‐derived putative endothelial progenitor cells (EPC) may play in the process of neoangiogenesis. However, recent data indicate that many of the putative EPC populations are comprised of various haematopoietic cell subsets with proangiogenic activity, but these marrow‐derived putative EPC fail to display vasculogenic activity. Rather, this property is reserved for a rare population of circulating viable endothelial cells with colony‐forming cell (ECFC) ability. Indeed, human ECFC possess clonal proliferative potential, display endothelial and not haematopoietic cell surface antigens, and display in vivo vasculogenic activity when suspended in an extracellular matrix and implanted into immunodeficient mice. Furthermore, human vessels derived became integrated into the murine circulatory system and eventually were remodelled into arterial and venous vessels. Identification of this population now permits determination of optimal type I collagen matrix microenvironment into which the cells should be embedded and delivered to accelerate and even pattern number and size of blood vessels formed, in vivo. Indeed, altering physical properties of ECFC‐collagen matrix implants changed numerous parameters of human blood vessel formation, in host mice. These recent discoveries may permit a strategy for patterning vascular beds for eventual tissue and organ regeneration.

Introduction

Much attention has been focused on determining the role of putative bone marrow‐derived endothelial progenitor cells (EPC) as cells that are mobilized into the circulation and are recruited to a site of tissue ischaemia, injury, inflammation, or all of the above, in the context of tumour development, to participate in the process of neoangiogenesis (1). Unfortunately at present, no specific cell surface marker or unique gene expression pattern has been identified to unambiguously identify EPC in mouse or human. Thus, the field is left with controversy and inability to compare and contrast putative EPC in normal and in different populations of diseased subjects or to compare study results from one paper to the next. While such ambiguity will likely continue to plague the field until a suitable marker is found, series of investigations have begun to clarify properties of putative EPC that better define their lineage of origin and their functions (2, 3, 4, 5). Ultimately, it is the function of various cell populations that defines them, better than their cell surface phenotype, and many so‐called EPC are found to be members of the haematopoietic lineage. Evidence to define cells as haematopoietic rather than endothelial in nature is presented in this concise overview.

In contrast to the vast number of papers elucidating roles of putative bone marrow‐derived EPC in cancer, cardiovascular disorders, autoimmune disease, trauma, chronic inflammatory disorders, or central nervous system disease (6, 7, 8, 9, 10), little focus has been placed on fully understanding roles, behaviour, or function of the rare circulating endothelial cells that may also be changed in number in many of these disorders (11, 12, 13, 14, 15). This overview will discuss known roles of vascular endothelial cells in repair of experimental vessel damage, deposition of proliferating circulating endothelial cells on implanted cardiovascular devices or prostheses, and evidence that some endothelial cell turnover exists at homeostasis. By comparing in vivo vessel‐forming functions of the rare circulating endothelial cell with proliferative potential and putative bone marrow‐derived EPC, one can readily assess the distinct roles that each may play for optimal vascular repair and regeneration or for developing novel human vascular constructs.

Defining sequence of vascular endothelial repair of local injury

In experimental animal models, insertion of a nylon thread or metal probe into a blood vessel can lead to denudation injury to the vascular endothelium that may not perturb underlying endothelial basement membrane (16, 17, 18). In this context, a well‐described sequence of events leads to complete restoration of the site with intact endothelium. Apparently, the first events include deposition of platelets on exposed basement membrane, increased migratory behaviour of endothelial cells adjacent to the injury site and endothelial cell spreading into the injury site. For small injuries, endothelial migration and spreading can result in closure of the previously denuded area within 24 h. Subsequently, endothelial cell proliferation of cells distal to the original edge of the denuding injury is initiated with eventual onset of proliferation of cells previously migrated into the original injury site. Extent of proliferation at the site of healing is significant, often leading to cell density several times that of the original population prior to injury. Over ensuing weeks, eventual remodelling of the endothelium results in return to original, normal endothelial cell density for the particular vessel under examination. Thus, the major cell type involved in resolution of the injury and return to the endothelial monolayer, is endogenous endothelium of the vessel injured.

If large areas of endothelium should be removed and/or there is damage to underlying basement membrane or an artificial vascular grafting material is implanted, greater influx of circulating cells ensues. In this instance, a host of haematopoietic cells along with platelets readily attaches to these areas of damage or to any grafted artificial material (19, 20, 21, 22, 23). In some instances, these deposited blood cells from circulating blood are soon replaced by migrating and spreading endogenous endothelium. In other cases, colonies of replicating endothelial cells grow on the exposed area and replace attached blood cells (24, 25, 26). In yet further circumstances, endothelial cells never replace the blood cells and in time, these areas develop a fibrous, non‐thrombogenic covering. At present, we still do not understand how to augment repair of extensive vascular injuries or implanted artificial materials, to permit complete endothelial monolayer generation.

In those examples where a vessel is artificially ablated or severed, and distal tissue becomes ischaemic and hypoxic, a series of cellular events may permit recovery of blood flow to that area via generation of adequate collateral vessels and regeneration of microvasculature. The earliest cells to be recruited to the site are haematopoietic cells; neutrophils and macrophages that must ‘clean up’ apoptotic and/or necrotic cellular debris, provide innate immune surveillance, and begin recruitment of new blood vessels from nearby ones to regain adequate tissue perfusion and recovery of tissue homeostasis (27, 28). A similar series of events occurs in areas of tumour growth where local ischaemia and hypoxia induced by overgrowth of malignant cells stimulate the tumour to secrete molecules that recruit myeloid cells to begin the process of neoangiogenesis and tumour vascular sequestration away from normal tissues. It is in these types of vascular injury or areas of tumour vascular ingrowth that the earliest evidence has been put forward for existence of bone marrow‐derived EPC. The majority of studies examining rescue from vascular injury or induction of tumour microvasculature, have been conducted in rodent models and it remains unclear whether results of these studies would be directly translatable to human vascular injury or tumour growth.

Methods to define human EPC

In the human system, putative EPC have been identified using three general approaches including, growth of peripheral blood or marrow mononuclear cells on fibronectin‐coated dishes in specific growth factor‐containing media, flow cytometric detection of a panel of cell surface molecules to define a population subset, and use of several in vitro colony‐forming cell assays to identify and enumerate progenitor cells. Each of these approaches has been recently described and discussed in detail. For the purposes of this overview, each method and limitations encountered in relying on that single test for EPC identification will be highlighted.

Perhaps the simplest method involves collecting low‐density mononuclear cells (MNC) from human peripheral blood and plating them in dishes coated with fibronectin, in culture medium containing endothelial growth factors and foetal calf serum (29, 30). After 4–5 days, non‐adherent cells are removed and adherent cells examined for ability to bind acetylated low‐density lipoprotein (AcLDL) and Ulex europaeus agglutinin 1 (a plant lectin), two properties thought to discriminate EPC from other peripheral blood cells. However, recently published data suggest that platelets contaminate most MNC preparations and presence of platelets in this culture milieu can result in transfer of platelet plasma membrane proteins to any adherent cells also attached to the culture matrix [including certain proteins thought to be endothelial specific (31)]. In addition, monocytes are enriched from MNC when plated on fibronectin‐coated dishes and in presence of endothelial growth factors are known to express a variety of proteins typically thought to be specific of endothelial cells [von Willebrand factor, endothelial nitric oxide synthase, CD31, CD144 and vascular endothelial growth factor 2 receptor (KDR)] (32, 33, 34, 35). Thus, this straightforward method of adherent MNC growth in vitro does not promote emergence of endothelial lineage progenitors.

A variety of cell surface antigens has been proposed as markers for putative EPC identification, although none is unique nor specific for such a cell [reviewed in (1)]. First descriptions of human EPC attempted to utilize cell surface markers that might be expressed by both haematopoietic and endothelial cells, to search for a putative circulating angioblast precursor (36, 37). While this rationale was based on known recognition of close temporal and spatial emergence of blood and endothelial cells during embryogenesis, it has often led to difficulty in discriminating putative human EPC populations from well‐recognized haematopoietic stem and progenitor cell subsets, mature monocyte subsets, dendritic cells and B lymphoid cells with proangiogenic activity, already fully described phenotypically and functionally. Such ambiguity has led to confusion over the type of cell termed EPC, is defining. Thus, until a specific cell surface EPC marker is identified, one must functionally validate phenotypic subsets of cells to confirm their lineage of origin and specific roles in neoangiogenesis before applying the term EPC.

Finally, two different colony‐forming cell assays have been developed to enumerate putative EPC. Plating low‐density MNC on fibronectin‐coated dishes and culturing for 48 h, then replating non‐adherent cells leads to emergence of a distinct colony of round cells overlying a layer of stellate shaped cells adherent to the culture dish (Fig. 1a). Conditions for this assay have been optimized by Hill et al. (38) and putative EPC counted in this assay referred to as colony‐forming unit‐Hill (CFU‐Hill). This assay is commercially available and has been utilized to quantify circulating putative EPC in numerous studies. Another assay calls for plating low‐density MNC on type 1 collagen‐coated dishes, and employing specific commercially available tissue culture medium to allow outgrowth of endothelial cells with colony‐forming cell (ECFC) ability (Fig. 1b). This takes 1–3 weeks depending on source of MNC (umbilical cord blood versus adult peripheral blood, respectively) (39, 40). A hierarchy of clonal proliferative potential is displayed by these ECFC with some colonies growing to more than 10 000 progeny from a single endothelial cell plated 14 days earlier. Cord blood ECFC display high telomerase activity and vigorous in vivo human vessel formation when suspended in matrix and implanted into immunodeficient mice. Adult peripheral blood cells display lower frequency of ECFC with high proliferative potential, although in vivo, ECFC form human vessels on implantation, in the mouse model (5). ECFC are rare in circulating blood of adult humans with frequency of approximately 2 ECFC/10⁸ MNC (5). Circulating concentration of ECFC changes in several human cardiovascular and ophthalmological disorders.

Figure 1.

  Colony‐forming unit‐Hill (CFU‐Hill) and endothelial colony‐forming cell (ECFC) colonies as they appear in vitro. (a) Representative phase‐contrast image of CFU‐Hill colony, which can be identified as an aggregate of phase‐contrast bright round cells with stellate‐shaped phase‐contrast dark‐appearing adherent cells emerging from the base of the round cell aggregate. (b) ECFC depicted with edges of the flattened colony highlighted by arrows, and cells within the colony forming cobblestone pattern. Scale bars represent 500 μm. Image is reproduced with permission from Yoder et al. (5).

Evidence that some resident and circulating endothelial cells display proliferative potential

While the general consensus purports that endothelial cell turnover in systemic blood vessels is low in adult subjects, ample evidence has also been presented to suggest that endothelium in some blood vessels is easily detectable. In young experimental animals, endothelial replication rates have been reported to be as high as 60% in certain focal areas of the aorta (41). Experimental injury to aortic endothelium (direct denudation) or disorders such as hypertension, hyperlipidaemia and endotoxaemia all lead to increase in endothelial replication in rodent models (42, 43, 44, 45, 46, 47). Most evidence suggests that proliferating cells are retained in the endothelial monolayer; however, some circulating endothelial cells thought to be sloughed from vascular endothelium may also have proliferative potential. Evidence of this behaviour can be exemplified if implanted nylon vascular graft material is suspended within the vascular stream. Rather than being seeded directly by endothelial cells contiguous to implanted grafts, islands of endothelium can be detected; the most likely source of these is circulating endothelial cells (26). As human adult and cord blood vascular endothelial cells have been determined to possess clonal hierarchical proliferative potential similar to circulating ECFC derived from cord blood and adult peripheral blood, it is plausible that circulating ECFC may be derived from vascular endothelium (39). Only further study will permit detailed clarification of the relationship between resident and circulating ECFC. Nonetheless, both resident and circulating ECFC in rats, humans, cows and pigs display the ability to form blood vessels in vivo on implantation.

Modulation of physical properties of ECFC‐collagen matrix implants alter in vivo human vessel formation

We and others have recently reported that ECFC seeded in type I collagen/fibronectin matrices form functional vessels when implanted into the flank of immunocompromised mice (5, 48, 49, 50). A variety of factors influence formation and persistence of ECFC‐derived vessels in vivo including cell passage number, source (adult or cord blood) and presence of perivascular cells (51). Physical properties of type I collagen matrices including fibril density and stiffness – are known to influence endothelial cell capillary morphogenesis in vitro, but extent to which matrix physical properties influence ECFC‐derived in vivo vessel formation, has only been recently addressed (52).

Polymerization of ECFC‐containing, rat tail type I collagen solutions at concentrations of 0.5–3.5 mg/ml yielded in vitro tissue constructs with significantly different physical properties (52). As expected, this increase in collagen concentration yielded matrices with 2‐fold increase in fibril volume fraction and no change in fibril diameter (Fig. 2). These observed concentration‐dependent changes in fibril microstructure result in increase in matrix stiffness as reflected by shear storage modulus (G′) and compressive modulus (E _c) values derived from oscillatory shear and unconfined compression testing respectively (Fig. 3). Upon implantation of these matrices (containing a constant number of human cord blood ECFC), subcutaneously into the flanks of immunodeficient mice for 14 days, recovered implants displayed great differences in overall remodelling, including overall size of gross explant and neovascularization pattern. In general, at lower collagen concentrations, matrices were extensively remodelled and greatly reduced in size, whereas at higher concentrations, little apparent implant remodelling was detectable (Fig. 4). These differences in implant remodelling caused us to develop a more thorough approach to quantify human vessel formation in vivo as merely measuring vascular number per unit area would not reflect discrepancies in overall explant size (52). We have reported that dose‐dependent increase in collagen concentration was associated with significantly greater total vascular area occupied by the human blood vessels in vivo and that significant difference in size distribution of blood vessel diameters was observed (Fig. 5). These data indicate that modulating physical properties, namely fibril density and matrix stiffness, of implanted ECFC‐collagen matrix constructs, exerts significant effect on number and size of human blood vessels formed in vivo. While not the aim of these studies, future work will focus on how modulation of these and other relevant collagen matrix physical properties alters ECFC gene expression during human blood vessel formation.

Figure 2.

  Three‐dimensional matrix fibril density varies with collagen concentration. Fibril density increased linearly with increasing collagen concentration from 0.5 mg/ml (a) to 2.5 mg/ml (b) as imaged by confocal reflectance microscopy (scale bar = 10 μm). Image is reproduced with permission from Critser et al. (52).

Figure 3.

  Three‐dimensional matrix physical properties vary with collagen concentration. Shear storage modulus (G′ measured in Pascal [Pa] – a measure of stiffness) increased with greater collagen concentration (a), δ significantly decreased with greater collagen concentration (indicative of shear response being dominated by solid collagen fibril phase of the matrix) (b), and compressive modulus (E _c) significantly increased with greater collagen concentration (indicative of increased compressive resistance) (c). Asterisks highlight significant differences P < 0.05. Image is reproduced with permission from Critser et al. (52).

Figure 4.

  Histochemical analysis of explanted cell matrices. Removal of matrices after 14 days revealed significant differences in size and apparent vascularity of the matrices. Representative histological sections from control matrix with no added ECFC (a), or ECFC suspended in gels of 0.5 (b), 1.5 (c), 2.5 (d), or 3.5 (e) mg/ml collagen concentration. Given that the same amount of gel was implanted in each animal, significant differences in extent of remodelling are apparent. Scale bar represents 1 mm. Image is reproduced with permission from Critser et al. (52).

Figure 5.

  Analysis of red blood cell‐containing blood vessel areas within explanted matrices. Light micrographs of explant histological sections reveal percentage of human CD31‐expressing vessels as distribution of various area subsets (a) and total human CD31 vascular area as a function of varying collagen concentration matrices (b). Asterisks indicate significant differences P < 0.05 among the different collagen matrices. Image is reproduced with permission from Critser et al. (52).

Summary

Understanding mechanisms of blood vessel repair is critical for defining new treatments for cardiovascular disorders. Increasing evidence suggests that varieties of haematopoietic cells play critical roles in angiogenesis, although do not directly participate in vasculogenesis. Only rare circulating ECFC and resident ECFC of human blood vessels display true vasculogenic activity in vivo when implanted with several types of matrix molecules. Future studies focusing on combined administration of various haematopoietic subsets and ECFC should permit optimized cell therapeutics for treating ischaemic and hypoxic tissues or for generating complex vascularized artificial tissues and organs for use in human subjects.