Hemangioblasts, angioblasts, and adult endothelial cell progenitors


  • Gina C. Schatteman,

    Corresponding author
    1. University of Iowa, Department of Exercise Science, University of Iowa, Exercise Science, Iowa City, Iowa
    • University of Iowa, Department of Exercise Science, University of Iowa, Exercise Science, FH 412, Iowa City, IA 52242
    Search for more papers by this author
    • Fax: (319) 335-6966

  • Ola Awad

    1. University of Iowa, Department of Exercise Science, University of Iowa, Exercise Science, Iowa City, Iowa
    Search for more papers by this author


After decades of speculation, proof of embryonic hemangioblasts finally emerged a few years ago. Surprisingly, at about the same time, evidence for adult hemangioblasts began to appear, and recent single-cell bone marrow transplants have confirmed their existence. Embryonic and adult hemangioblasts appear to share antigenic determinants, including CD34, ACC133, and VEGFR2, although their phenotype may be plastic. They also respond to similar factors, prominent among them vascular endothelial growth factor (VEGF). In the adult, hemangioblasts reside principally in the bone marrow, although they may subsequently leave that niche to reside in nonhematopoietic tissues. A number of studies indicate that these cells or their progeny may be a significant source of endothelial cells in adult pathologic and nonpathologic vascularization, and may participate in vascular repair. In addition to hemangioblasts, a more differentiated source of endothelial cell progenitors may be present in the blood, namely, monocytes or monocytic-like cells. The relative importance of the two cell types in vivo is not clear, though endothelial cells derived from the two sources may not be identical, and hemangioblasts seem to provide a stimulus for differentiation of the monocytes. Treatment with exogenous bone marrow-derived cells can promote neovascularization, accelerate restoration of blood flow to ischemic tissues, and improve cardiac function after infarct. Hence, there is great hope that either alone, in combination with angiogenic factors, or as gene therapy vectors, we can harness these cells to treat ischemic and vascular diseases in the relatively near future. Anat Rec Part A 276A:13–21, 2004. © 2004 Wiley-Liss, Inc.


For decades it was commonly thought that, embryologically, endothelial cells and hematopoietic stem cells are derived from the same precursor, the hemangioblast. Proof of this concept remained elusive, but work at the beginning of the last decade provided further evidence that the two cells were related and suggested that vascular endothelial growth factor (VEGF) might be the key to definitively establishing such a relationship. First, in situ hybridization showed that mRNA of flk-1 (VEGFR2), a murine VEGF receptor, was expressed by endothelial cells at all stages of mouse development, including the blood islands in the yolk sac of day 8.5–10.5 embryos, in which many endothelial cell progenitors originate (Millauer et al., 1993). In addition, flt-1 (VEGFR1), a second mouse VEGF receptor, was expressed in populations of embryonic cells from which endothelium is derived, including early yolk sac mesenchyme (Peters et al., 1993). In the quail, quek1 and quek2 function as VEGF receptors. Quek1 is expressed in the mesoderm beginning with the onset of gastrulation, whereas quek2 message is first detected on QH1-expressing endothelial cells. The expression pattern is consistent with quek1 being a marker of the hemangioblast and of quek2 being an indicator of a partially or fully differentiated endothelial cell (Eichmann et al., 1993). A subsequent similar analysis of mouse embryos suggested that the differentiation of hemangioblasts from pluripotent epiblast may be characterized by expression of flk-1 mRNA (Flamme et al., 1995).

Deletion of VEGFR2 by gene targeting provided further support for the existence of a common hematopoietic and endothelial cell progenitor. Both endothelial and hematopoietic cells are absent in homozygous null mice, suggesting that VEGFR2 is expressed by the hemangioblast and essential for their further differentiation along both lineages (Shalaby et al., 1995, 1997). To test this, Eichmann et al. (1997) isolated VEGFR2+ cells from the mesoderm of chicken embryos at the gastrulation stage. Using blocking antibodies against avian VEGFR2, they showed that a single VEGFR2+ cell can give rise to both hematopoietic and endothelial cell colonies, and that the developmental decision appears to be regulated by the binding of two different VEGFR2 ligands. Moreover, their data suggested that in the absence of the VEGFR2, neither hematopoietic nor endothelial progenitors can survive, and that VEGF is required for differentiation of endothelial cells, while a second VEGFR2 ligand was important for hematopoietic cell development.

At about the same time, a unique precursor population was identified in mouse embryonic stem cell-derived embryoid bodies (Choi et al., 1998). In response to VEGF, a single such precursor cell gave rise to blast colonies and adherent cells, of which the latter eventually differentiated into endothelial cells. Kinetic analysis showed that the blast cells develop early and are lost quickly during embryoid body development. These two studies offered definitive proof that at least some cells in the embryo can function as hemangioblasts. They also identified embryonic subpopulations that act as hemangioblasts and indicated a role for VEGF and VEGFR2 in hemangioblast development and/or differentiation.

A subsequent study in VEGFR2–/– mouse embryos showed that normal numbers of hematopoietic progenitors are present at 7.5 days postcoitum, even though by 8.5 days postcoitum the embryos are deficient in these cells, supporting the idea that VEGFR2 plays a direct role in both normal hematopoietic and endothelial cell development (Schuh et al., 1999). Additionally, VEGF induces multipotent VEGFR2-expressing embryonic cells to differentiate into endothelial cells, whereas PDGF (platelet derived growth factor) induces mural cell formation (Yamashita et al., 2000). Further, in vitro, VEGFR2–/– embryonic stem cells differentiate normally into hematopoietic and endothelial cells but give rise to a greatly reduced number of blast colonies, reflecting reduced hemangioblastic potential. These data suggest that hemangioblasts arise independently of VEGFR2 in the mouse, but that their subsequent migration and expansion require a VEGFR2-mediated signal.


While the unproven idea that embryonic hemangioblasts exist was well accepted prior to the late 1990s, the general consensus was that such cells were not present in the adult. As a corollary, adult neovascularization was generally thought to occur via angiogenesis, i.e., proliferation and migration of differentiated endothelial cells, and not through the well-described embryonic process of vasculogenesis, wherein vascular progenitors differentiate and coalescence to form blood vessels (Risau, 1995). Yet, as early as 1932, a description of “Capillary-Like Formations in Tissue Cultures of Leukocytes” was published (Hueper and Russell, 1932). One year later “the development of organized vessels in cultures of blood cells” and in 1950 “in vitro blood vessel” formation from bone marrow of adult chickens were described (Parker, 1933; White and Parshley, 1950). In addition, reports of fallout endothelialization, i.e., endothelialization of grafts by blood-derived cells, have been appearing almost as long as synthetic arterial grafts have been in use (Stump et al., 1963; Mackenzie et al., 1968; Scott et al., 1994; Shi et al., 1994). Elegant studies by Sauvage and colleagues showed that when blood flow is present, implanted synthetic arterial grafts can be reendothelialized even when endothelial cell ingrowth from the anastomoses and the vaso vasorum are apparently prevented (Shi et al., 1994; Wu et al., 1995). Another study found that when bone marrow cells were infiltrated into synthetic vascular grafts, endothelialization occurred more rapidly than in uninfiltrated controls (Noishiki et al., 1996). All of these studies pointed to the presence of hemangioblasts, endothelial cells, or their precursors in the blood and/or bone marrow and provided strong evidence for adult vasculogenesis.

In the late 1990s new studies appeared that brought this earlier work to the forefront and challenged old ideas about neovascularization in the adult. To study the effects of VEGF on reendothelialization of arteries, Isner and colleagues had denuded carotid arteries bilaterally and then delivered VEGF protein locally via a double balloon catheter to one of the carotid arteries. VEGF markedly increased reendothelialization of the treated artery, but surprisingly, it also improved reendothelialization of the untreated contralateral artery (Asahara et al., 1995). The small amounts of VEGF administered could not have appreciably elevated systemic levels of VEGF even if none of the VEGF remained at the site of administration. This suggested that cells in adult blood might be responding to locally high concentrations of VEGF in the vicinity of the balloon that in some way induced them to reendothelialize the contralateral artery. Since fallout endothelialization appeared to occur principally in conditions that would be expected to induce VEGF, this seemed reasonable. Putative circulating reendothelializing cells could be fully differentiated endothelial cells or adult angioblasts. Earlier reports describing circulating endothelial cells indicated that the cells were both unhealthy and rare, arguing against the former (Sbarbati et al., 1991; George et al., 1992; Sinzinger et al., 1996).

With the accumulating evidence for blood as a source of adult angioblasts, in collaboration with Isner and colleagues, we sought to identify a possible adult angioblast population. Though hemangioblasts had not been definitively identified at the time, they did appear to share many antigenic characteristics with adult hematopoietic stem cells. Among these, CD34 is an antigen routinely used to enrich for human hematopoietic stem and primitive progenitor cells that is expressed in blood islands, on early endothelial cells in the embryo, and on activated endothelial cells in the adult (Siena et al., 1989; Fritsch et al., 1995; Young et al., 1995; Krause et al., 1996). Using CD34 as a starting point, we demonstrated that CD34+-enriched peripheral blood mononuclear cells (PBMCs) can differentiate into endothelial cells in vitro. The differentiated cells express endothelial nitric oxide synthase (eNOS) mRNA, CD31, and tie-2; take up acetylated low-density lipoprotein (acLDL); bind Ulex lectin; form blood island and tube-like structures; and produce NO in response to acetylcholine and VEGF in a dose-dependent manner (Asahara et al., 1997) (Fig. 1). Moreover, intravenous infusion of fluorescently tagged CD34+ PBMCs, but not CD34 PBMCs, into mice and rabbits recovering from hindlimb ischemia resulted in integration of the labeled cells into the newly forming vasculature (Fig. 2). Interestingly, infusion of genetically tagged VEGFR2+ PBMCs also led to vascular integration of the infused cells. Both the CD34+ and VEGFR2+ cells that incorporated into the vasculature expressed the endothelial cell antigens CD31 and tie-2 (Asahara et al., 1997).

Figure 1.

Schematic of likely adult hemangioblast differentiation pathway. Hemangioblasts appear to have the same phenotype as the classical hematopoietic stem cell. While this phenotype is plastic, the figure indicates cell surface antigens commonly found on the cells. As the hemangioblasts destined to become endothelial cells differentiate, they downregulate AC133 expression and probably express myeloid lineage markers. A subset of myeloid progenitors will differentiate subsequently and assume a typical monocytic phenotype, which includes expressing CD14+ and downregulating CD34+. Whether all endothelial cell progenitors must proceed through the monocytic phase of differentiation is unclear. Ultimately, under the appropriate stimulus, CD34CD14+ monocytes differentiate into endothelial cells.

Figure 2.

Integration of blood-derived cells into vascular structures. The top two panels show integration of injected fluorescent red DiI-labeled human CD34+-enriched cells into the fluorescent green Bandeiraea simplicifolia lectin B4-labeled neovasculature of a recovering mouse ischemic limb. The bottom two panels illustrate CD14+-enriched cells forming vascular structures in vitro and in vivo. The phase contrast micrograph (left) shows formation of capillary-like structures by CD14+-enriched cells plated on fibronectin for 21 days in two-dimensional culture. The lower right panel shows integration of DiI -labeled human CD14+-enriched cells into Bandeiraea simplicifolia lectin B4-labeled neovasculature of a recovering mouse ischemic limb as in the panels above. However, integration of CD14+ cells was observed only, as in this example, when unlabeled CD34+ cells were co-injected with the labeled CD14+ cells.

Confirming the presence of circulating endothelial cells or progenitors, in the following year, a study using a canine bone marrow/PBMC transplantation model unequivocally demonstrated fallout endothelialization (Shi et al., 1998). A Dacron aortic graft, impervious to ingrowth from perigraft tissue and the vaso vasorum, was implanted 6–8 months after marrow/PBMC transplantation. When the graft was retrieved 12 weeks later, reendothelialization of the graft had occurred, but all graft endothelial cells were donor derived. In the same study, the authors cultured 95% pure CD34+ cells from human bone marrow, umbilical cord, fetal liver, and mobilized peripheral blood in the presence of VEGF and other factors. Cells from all four sources differentiated into cells that expressed CD34+, von Willebrand factor (vWF), and VEGFR2, and incorporated acLDL.

Over the next several years, various reports indicated that populations of cells expressing the hematopoietic stem cell antigens CD34 (human), sca-1 (mouse), c-kit, VEGFR2, and AC133 are all enriched for endothelial cell progenitors as are side population (SP) cells (Asahara et al., 1997; Gehling et al., 2000; Peichev et al., 2000; Jackson et al., 2001; Orlic et al., 2001; Quirici et al., 2001) (Fig. 1). Among these, only AC133 is not also expressed on subsets of differentiated endothelial cells. Thus, the ability of AC133+ cells to differentiate into endothelial cells in vitro and in vivo provided compelling evidence not only for the existence of endothelial cell progenitors, but also for a subset of AC133+ cells representing adult hemangioblasts (Gehling et al., 2000). Nevertheless, definitive proof of the existence of adult hemangioblasts was provided only recently through single-cell bone marrow transplants in mice (Grant et al., 2002). After reconstitution the bone marrow with a single EGFP (enhanced green fluorescent protein) -expressing Sca-1+ cell, subsequent induction of retinal neovascularization led to significant differentiation and integration of EGFP-expressing cells into the retinal vasculature.


Though these studies demonstrated that at least some endothelial cells could be marrow derived, there was concern that circulating endothelial cells might confound interpretation of in vitro data. A study of transplant patients that had received sex-mismatched marrow showed that cultures of blood cells enriched for the endothelial cell antigen P1H12 had a mixed phenotype. Recipient-derived cells, presumably circulating endothelial cells, proliferated rapidly and were numerous initially, but as they died, the more slowly growing donor-derived marrow cells became the dominant cell in the cultures (Lin et al., 2000). Since P1H12 is rarely if ever expressed on endothelial cell progenitors, and a consistent feature of most even moderately purified endothelial cell progenitor cultures is their initial slow proliferation rate, it is unlikely that in the absence of a procedure that actively selects for circulating endothelial cells, they have a significant impact on the cultures.

Initially, most attention focused on hematopoietic stem cell-related cells as the sole source of endothelial cell progenitors. However, a variety of studies began to emerge that suggested that other cells might serve the same purpose. First, findings that brain and muscle-derived cells could function as hematopoietic stem cells indirectly suggested that they might also act as endothelial cell progenitors (Bjornson et al., 1999; Jackson et al., 1999). Interestingly, recent work showed that the muscle-derived progenitors actually originate in the marrow and come to reside in the muscle, where they lose their blood-like phenotype, including expression of the pan-hematopoietic antigen CD45 (Issarachai et al., 2002; McKinney-Freeman et al., 2002). The finding that brain cells transdifferentiate in the adult remains controversial, but murine neuronal stem cells injected into stage 4 chick embryos can contribute to tissues of ectodermal, mesodermal, and endodermal origins, suggesting that they are indeed highly plastic (Clarke et al., 2000).

Second, a report suggested that tumors could make their own vascular channels, and that these channels were lined with tumor-derived cells (Maniotis et al., 1999). We and our colleagues recently demonstrated that transplantation of human tumor cells into a mouse recipient resulted in a tumor vascular-like structure that was wholly human cell derived, but that was functionally connected to the external murine vasculature. Moreover, tumor cells appeared to have escaped the tumor and integrated into the surrounding murine vasculature (Hendrix et al., 2001, 2002).

Hematopoietic stem cells do not differentiate directly into lymphocytes, erythrocytes, and other cells. Instead, they follow an orderly sequence of differentiation steps that branch to form a variety of blast and progenitor cells that ultimately terminally differentiate into the various blood components. One of these lineage progressions leads to formation of myeloid progenitor cells that in turn can differentiate into monocytes. Monocytes are a highly plastic cell type that can differentiate into dendritic cells, various types of tissue macrophages, and Kupffer cells (Wisse, 1974; Eiermann et al., 1989; Hausser et al., 1997; Randolph et al., 1998).

Beginning in 1967 several morphological studies suggested that monocytes could differentiate into endothelial cells to form intrathrombus capillaries (Tsapogas et al., 1967; Prathap, 1972; Feigl et al., 1985; Leu et al., 1987). Recent evidence that monocytes also act as endothelial progenitors cells comes from data showing that they can drill channels in the myopathic heart and that monocytes could take on an endothelial cell-like phenotype in vitro (Fernandez Pujol et al., 2000; Moldovan et al., 2000; Harraz et al., 2001; Schmeisser et al., 2001). Endothelial cells derived from cultured monocytes are phenotypically distinct from those derived from culture of the more primitive hematopoietic stem cells (i.e., AC133+ or CD34+ cells). Monocyte-derived endothelial cells simultaneously express endothelial cell antigens, including CD31, Tie-2, and vWF, as well as the macrophage/monocyte antigens CD68, CD80, CD45, and CD36 (Fernandez Pujol et al., 2001; Schmeisser et al., 2001; Schmeisser and Strasser, 2002). However, the cells also make the enzyme eNOS, suggesting they can perform physiological functions associated with endothelial cells and appear to downregulate expression of monocytic antigens with time in culture (Fernandez Pujol et al., 2001; Harraz et al., 2001) (Fig. 1).

Data from our laboratory provide in vivo support for a role for monocytes as endothelial cell progenitors. We labeled CD14+ monocytes with the lipophilic dye, DiI, and injected the cells into ischemic mouse muscle. Two weeks later, essentially no labeled cells were present in the wall of the neovasculature of the muscle. When we repeated the experiment co-injecting unlabeled CD34+ PBMCs and labeled CD14+ cells, many DiI-labeled cells were observed in the neovessel walls (Harraz et al., 2001) (Fig. 2). Thus, CD34+ cells can provide cues for the differentiation or vascular integration of CD14+ endothelial cell progenitors (Fig. 3). In support of a synergistic interaction between the cells, unpublished data from our laboratory show that culture conditions that fail to support survival of either CD34+ or CD14+ mono-cultures support growth and differentiation into endothelial cells of co-cultures of the cells. CD34+ cells also secrete VEGF and angiopoietin-1, which, as discussed below, may be critical inducers of monocyte differentiation into the endothelial cell lineage (Bautz et al., 2000; Takakura et al., 2000; Majka et al., 2001).

Figure 3.

Interrelationship between hemangioblasts and monocytic-like cells. Hemangioblasts (AC133+/CD34+/VEGFR+) are rare, slowly proliferating cells. Their direct differentiation into endothelial cells, if it occurs, is likely a rare event. One mechanism through which they might generate large numbers of endothelial cells is utilizing the myeloid progenitor pathway. Myeloid progenitors differentiate into monocytes (CD14+) or monocyte-like cells, which can function as a source of endothelial cell progenitors. CD34+ cells, some of which may be hemangioblasts, appear to provide signals for the differentiation and integration into the vasculature of the monocytic-like endothelial cell progenitors. Several molecular stimuli can induce differentiation and/or vascular integration (or coalescence) of hemangioblast progeny, while others induce hemangioblasts to leave the bone marrow. Although for simplicity the figure indicates a sequential progression of differentiation followed by integration, it may be that the sequence of differentiation and integration is reversed or the events occur simultaneously.

Because of their schizophrenic phenotype, there has been skepticism as to whether these monocyte-derived cells represent true endothelial cell progenitors. Yet, in vivo data demonstrate that endothelial cells with a mixed phenotype are present in many tissues. For example, liver sinusoidal cells express the classical monocytic markers CD32, CD16, CD14, and CD80, activated microvascular cells can express class II major histocompatibility (MHC) molecules, and CD86 is expressed in vessels of patients with vascular neuropathy (Schmeisser and Strasser, 2002). Thus, CD14+ cells may represent a highly plastic reserve of endothelial cell progenitors that can modulate their phenotype according to local conditions. Whether they act as temporary placeholders until other endothelial cells can replace them, or if they remain and eventually differentiate into a more classical endothelial cell phenotype, remains unclear.


We are just beginning to understand factors that regulate endothelial cell progenitor function (Fig. 3). Among pro-angiogenic growth factors, basic fibroblast growth factor (FGF-2) was present in the earliest cultures of blood-derived endothelial cell progenitors through either direct addition or supplementation with FGF2 containing extracts (Asahara et al., 1997, 1999a; Shi et al., 1998; Fernandez Pujol et al., 2000; Gehling et al., 2000; Kalka et al., 2000b; Murohara et al., 2000; Schatteman et al., 2000). However, whether it is essential for the survival, growth, or differentiation of endothelial cell progenitors is unclear. Insulin-like growth factor-1 and insulin are also typically included in progenitor cultures, but they do not appear to be essential factors since we routinely culture and differentiate both CD34+ and CD14+ cells in heat-inactivated serum and do not supplement with either factor (Schatteman et al., 2000; Harraz et al., 2001).

Three surprising factors, oncostatin M, interleukin-4, and macrophage colony-stimulating factor (M-CSF), may prove to be potent stimulators of the monocyte-to-endothelial cell transition. CD14+ cells cultured in the presence of oncostatin M and interleukin-4 form a homogeneous population lacking dendritic or macrophage antigens and upregulate VE-cadherin and begin to express CD34 (Fernandez Pujol et al., 2001). In addition, although its role in adult endothelial cell progenitor differentiation has not been tested directly, M-CSF appears to play a role in directing dendritic cells into a more endothelial cell-like phenotype (Hausser et al., 1997). Moreover, in the embryo M-CSF is critical for endothelial cell differentiation in the aorta/gonad/mesenephros region (Minehata et al., 2002).

Not surprisingly, the best-studied potential endothelial cell progenitor regulator is VEGF, and it is included in essentially all progenitor cell culture media. VEGF can stimulate the progenitor to endothelial cell transition of both hematopoietic stem cell-like and monocytic endothelial cell progenitors, apparently by increasing both differentiation and proliferation (Shi et al., 1998; Asahara et al., 1999b; Fernandez Pujol et al., 2000). It is also chemotactic for the cells and induces vascular permeability, making it possible for progenitors to enter and migrate at sites of injury (Asahara et al., 1999b; Peichev et al., 2000; Waltenberger et al., 2000). In vivo, it appears to mobilize endothelial cell progenitors from the bone marrow (Asahara et al., 1999b; Kalka et al., 2000c; Hattori et al., 2001; Moore et al., 2001).

Since circulating AC133+ and CD34+ cells or, in the mouse, Sca-1+ cells are rare, it may be necessary to mobilize the cells in order for them to make a significant contribution to repair or regeneration of the vasculature. Consistent with this, treatment of mice with granulocyte macrophage colony-stimulating factor (GM-CSF), which potently mobilizes hematopoietic stem cells, increases neovascularization in response to injury (Takahashi et al., 1999). However, GM-CSF also nearly triples the number of circulating monocytes, so it cannot be concluded that mobilization of Sca-1+ cells is the cause of the increases in vascularization. Another cytokine, G-CSF, increases circulating CD34+c-kit+ endothelial cell progenitors in rats. Interestingly, VEGF and angiopoietin-1 also induce release of cells from the bone marrow and increase the number of circulating endothelial cell progenitors, as assessed by increased in vitro production of endothelial cells (Asahara et al., 1999b; Kalka et al., 2000a; Hattori et al., 2001; Moore et al., 2001). In vivo, both VEGF and GM-CSF increase the total numbers of bone marrow-derived endothelial cells, but whether this is simply because there are more blood vessels or instead due to an increase in the rate (i.e., percentage) of marrow-derived cell integration has not been examined in the adult. However, in the early postnatal period, VEGF stimulates a 10-fold increase in bone marrow-derived liver sinusoidal endothelial cells, with no concomitant increase in total sinusoidal endothelium, suggesting a VEGF-stimulated increase in the rate of progenitor cell integration (Young et al., 2002). Interestingly, in the heart, the increase in bone marrow-derived endothelial cells precisely parallels increases in vascularity, demonstrating tissue-to-tissue variability in progenitor function.


Over the last several years numerous investigators have reported integration of bone marrow cells into the vasculature after either bone marrow transplantation or systemic or local injection in humans, dogs, cats, rabbits, rats, and mice (for example, Asahara et al., 1997; Shi et al., 1998; Crosby et al., 2000; Ikpeazu et al., 2000; Kobayashi et al., 2000; Murohara et al., 2000). Integration occurs in a variety of vascular beds, but there appear to be differences in the rates of incorporation among them. The endothelium of mice, examined 16 weeks after bone marrow transplantation, displayed low rates of bone marrow incorporation in the brain, intermediate levels in the aorta, and higher levels in the skin where up to 2.5% of endothelial cells were marrow derived (Crosby et al., 2000). In the same mice, rates of incorporation were much higher in the neovessels of an implanted sponge and ischemic limb, ranging from 8.3 to 11.2%. In another study, 20–25% of capillaries in nude rats injected intravenously with human CD34+ cells after myocardial infract heart contained injected cells (Kocher et al., 2001). In mice receiving bone marrow transplants, endothelial progenitor cell engraftment was 3.3% of endothelial cells in the infarcted heart, and the cells were localized adjacent to the infarct (Jackson et al., 2001). Additionally, although not specifically quantitated, in a mouse model of retinal neovascularization after bone marrow transplant, large foci of neovascularization comprised primarily of bone marrow-derived cells were observed. Thus, bone marrow-derived endothelial cell progenitors can make physiologically significant contributions to neovascularization, but the extent of their participation probably depends on physiological context, tissue, and method of cell presentation.

That endothelial cell progenitors are attracted to sites of injury and tissue ischemia is a consistent finding. In ischemic hindlimb models, injected endothelial cell progenitors are more common in the ischemic limb than the nonischemic limb, but this perhaps is not unexpected since inflammatory responses attract leukocytes of many kinds (Asahara et al., 1997; Takahashi et al., 1999; Schatteman et al., 2000). As with GM-CSF and VEGF, the total number of bone marrow-derived endothelial cells increases in ischemic tissues, but, again, it is not clear if this is because there are more endothelial cells in the tissue or an increased rate of incorporation. Interestingly, injected bone marrow-derived cells were localized in the myocardial infarct zone and not in the peri-infarct or noninfarcted zones of infracted rats. This suggests specific ischemia-induced homing of progenitors. In addition, it indicates that vascularization in the infarct region may occur via vasculogenesis, and via angiogenesis in less affected regions (Kocher et al., 2001). Hindlimb ischemia, myocardial infarction, and vascular trauma also mobilize bone marrow endothelial cell progenitors, and this mobilization correlates with increased vascularization in tissues undergoing neovascularization (Takahashi et al., 1999; Gill et al., 2001; Shintani et al., 2001a).


The potential applications of endothelial cell progenitors to the treatment of vascular and ischemic diseases are manifold. Numerous reports demonstrate their ability to promote vascularization and lead to functional improvements, including acceleration of the restoration of blood flow to ischemic limbs and improvement of myocardial function after infarct in animal models (Kalka et al., 2000b; Kobayashi et al., 2000; Murohara et al., 2000; Schatteman et al., 2000; Ikenaga et al., 2001; Jackson et al., 2001; Kamihata et al., 2001; Kawamoto et al., 2001; Kocher et al., 2001; Shintani et al., 2001b). Clinical trials are under way to test their therapeutic potential in a variety of settings, and recent data from these trials are encouraging. For example, in a 20-patient study wherein autologous bone marrow cells were infused 5–9 days after myocardial infarct, the size of the wall defect decreased and wall movement increased in the 10 treated patients relative to controls (Strauer et al., 2002). In a similar 20-patient study intracoronary bone marrow cell infusion improved ejection fraction and coronary reserve as well as other functional parameters after infarct (Assmus et al., 2002).

As with other pro-angiogenic therapies, it is important to consider the potential of exogenous bone marrow cells to exacerbate tumor growth. This is highlighted by the finding that bone marrow cells contribute to tumor neovascularization and, when modified to express angiogenesis inhibitors, can decrease tumor growth (Davidoff et al., 2001).

Endothelial progenitor cell dysfunction may be of clinical relevance. Cultured cells from type 1 diabetic patients yield fewer endothelial cells than those from nondiabetic subjects, and in a hindlimb ischemia model in nude mice, intramuscular injection of CD34+ cells failed to improve flow restoration in nondiabetic mice, but significantly accelerated its return in diabetic mice, suggesting functional impairment (Schatteman et al., 2000). Progenitor dysfunction may also be associated with type 2 diabetes (Tepper et al., 2002).

Bone marrow-derived endothelial cells are found in atherosclerotic plaques (Campbell et al., 2001; Sata et al., 2002). This is consistent with the critical role of monocyte/macrophages in the pathogenesis of the disease and suggests that dysregulation of endothelial cell progenitors could contribute to atherosclerosis. However, the number and migratory activity of circulating progenitors inversely correlates with risk factors for coronary artery disease (Vasa et al., 2001).

Clearly, these studies demonstrate that bone marrow cells have important physiological roles in both pathological and nonpathological vascularization. Learning to harness them could contribute significantly to our ability to prevent and treat a variety of disorders. However, cell-based therapies are not without risk and must be approached with caution. Nevertheless, it is likely that the cells by themselves, in combination with angiogenic factors, or as gene therapy vectors are likely to become important therapeutic tools in the not too distant future.