| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
INVITED REVIEW
Departments of 1Integrative Physiology and 2Dermatology, University of Iowa, Iowa City, Iowa
 |
ABSTRACT |
|---|
 TOP ABSTRACT GRANTSREFERENCES |
|---|
endothelial progenitor cell; endothelium; tumorigenesis; stromal derived factor 1; monocyte
If bone marrow-derived (BMD) cells can differentiate into ECs, that they might have therapeutic potential is obvious. Two early studies (90, 151) demonstrated that transplantation of subsets of BMD cells can indeed stimulate vascular growth. However, it was puzzling that implantation of only a small number of cells could dramatically increase vascular growth, particularly when relatively few injected cells could be found in the endothelium. It was first thought that limitations in our abilities to detect the BMD cells were responsible for the paradox, but this hypothesis has not been substantiated. Instead, it has become clear that, even when BMD cells do not differentiate into vascular cells, they exert dramatic proangiogenic effects. In fact, in most therapeutic settings, it appears that it is their paracrine actions that provide the greatest benefits, and it is these effects that most BMD cell-based therapies seek to exploit. Still, it appears that, even when BMD cells do not differentiate into ECs, those cells that most potently stimulate vascular growth are those that can themselves take on at least some endothelial-like characteristics. There is great interest in the use of BMD cells as exogenous therapeutic agents, and many reviews on the topic can be found in the literature.
Though differentiation of endogenous BMD cells into ECs may be rare in some in vivo situations, in others, BMD cells can contribute enormously to the endothelium. This review will focus on endogenous bmdEC biology in endothelial maintenance, repair, and pathology. To begin with, we will attempt to define an EC before reviewing the recent history of adult bmdECs and their precursors. We will explore the data demonstrating the existence of hemangioblasts and angioblasts in the adult circulation, and then we will try to make sense of the confusing data on EC precursor phenotype. Next, the physiological significance of bmdECs in the endothelium in vascular homeostasis, repair, and cancer will be examined. Finally, we will present an overview of some of the physiological and molecular factors that regulate BMD cell recruitment to and integration into the endothelium.
What is an Endothelial Cell?
In our 1997 publication, (5) we found that peripheral blood mononuclear cells (PBMCs) enriched for CD34+-expressing cells could differentiate into endothelial-like cells. After 7 days in culture on fibronectin, the majority of cells expressed CD31 (platelet endothelial cell adhesion molecule-1) and/or Tie-2, and the fraction of cells expressing CD34 and VEGF receptor 2 (VEGFR2) increased. The cells also made nitric oxide (NO) synthase III (eNOS) mRNA and produced NO after stimulation with VEGF or acetylcholine in a dose-dependent manner. In addition, they took up acetylated low-density lipoprotein (acLDL), bound Ulex lectin, and formed tubelike structures in vitro. Thus the cultured CD34+-enriched cells exhibited multiple antigenic and functional characteristics that suggested that they were ECs.
Unfortunately, during the course of this study and in many subsequent studies, it became clear that it is not simple to identify an EC. We struggled to find an antigen or characteristic that was unique to ECs. Our fluorescence-activated cell sorting analysis and many before and since demonstrated that circulating cells express the putative EC antigens CD31, VEGFR2, and the angiopoietin receptor Tie-2. Moreover, monocyte/macrophages bind Ulex lectin and take up acLDL (113). The only property that we could identify that was specific to ECs and not other circulating cells was expression of eNOS and release of NO in response to acetylcholine and VEGF (5, 68).
Three subsequent works brought this problem into sharper focus. In all three studies, the authors studied the potential of a different subset of BMD cells, CD14+ PBMCs, to differentiate into ECs. While they showed a clear progression from a monocytic to EC phenotype in their CD14+ cell cultures, expression of monocytic antigens persisted in the EC-like cells (46, 68, 153) (Table 1) [this was summarized in a review (154)]. Two more recent studies demonstrate even greater overlap in EC-monocyte phenotypes (147, 188), and one concludes that acLDL uptake, Ulex binding, CD31, CD105 (endoglin), and CD144 (vascular endothelial cadherin) expression are inherent features of monocytes, making them phenotypically indistinguishable from putative ECs (147). Additional studies confirmed that, like CD34, VEGFR2, CD144, and Tie-2 are expressed by a subset of circulating BMD cells (5, 153, 154).
|
A recent study (113) suggests that there may be a way to definitively characterize ECs, particularly those derived from the bone marrow. The phenotypic and morphological overlaps of dendritic cells, macrophages, and ECs of BMD cell origin were compared. There was a broad overlap in both antigenic expression and functional assays, but three characteristics were unique to ECs. First, consistent with our findings, eNOS expression was confined to ECs. Second, only ECs integrated into tubelike structures formed by human umbilical vein ECs (HUVECs). Finally, only ECs significantly stimulated tube formation by HUVEC. Thus, these three criteria appear to be the best by which to judge the EC phenotype in vitro at this time, regardless of what other "non-EC" antigens they may express.
In many ways, it is simpler to identify an EC in vivo, since we have a primary functional assay on which to rely; is it lining a blood vessel? Many argue that this localization along with expression of one or two "EC antigens" is not adequate to characterize a BMD cell as an EC in vivo. Yet it is precisely the requirement that has been used for decades and continues to be used for cells in the vessel wall. That is, if a cell of which the origin is unknown lines a vessel and expresses vWF and CD144, it is generally accepted that the cell is an EC. However, if the cell is later identified as a BMD cell, its characterization as an EC is brought into question. If we are to object to the characterization of the bmdEC, we must then question every putative EC and not apply a double standard. After all, the other cells lining the vessel wall could have originated in the adventitia, media, or mesenchyme, for example, without our knowledge.
Hemangioblasts and EC Stem Cells
Studies by a variety of laboratories using multicell bone marrow transplants have shown that angioblasts reside in the bone marrow in both small and large mammals (4, 9, 31, 85, 87, 117, 159, 202). However, multicell transplants do not distinguish between the existence of a distinct angioblast population and true hemangioblasts. In the embryo, ECs arise from hemangioblasts, a common endothelial/hematopoietic precursor. Since embryonic hemangioblasts are phenotypically similar (if not identical) to adult HSCs, it seems logical that EC-generating activity in BMD cells would reside in the HSC compartment. In a search for the hemangioblast, single cell transplants of HSC by three investigators were unable to show that bone marrow contains hemangioblasts (11, 124, 187). However, in three other studies (9, 62, 102), single cell transplants of nonadherent c-kit+sca-1+lin– cells (which are enriched for HSC in the mouse) resulted in reconstitution of the bone marrow with bone marrow cells that produced progeny that contributed to the endothelium. It is somewhat surprising that. despite these three independent investigations, skepticism of the existence of adult hemangioblasts persists. Some attribute the EC-producing activity to nonhematopoietic "tissue-committed stem cells" or "pluripotent stem cells" resident in the bone marrow (99). Of course, there is no reason to suppose that such nonhematopoietic stem cells cannot coexist with hemangioblasts, and, indeed, there is evidence for their existence (99).
Stem cells, under normal circumstances, maintain their own population, undiminished in function and size (i.e., are capable of long-term self-renewal), and furnish daughter cells that are themselves or give rise to mature cell types that have characteristic morphologies and functions (156). The mere presence of hemangioblasts does not guarantee that the cells are true stem cells. To demonstrate self-renewal, serial transplantation experiments are required. With the use of cells derived from the single cell-transplanted mice for serial transplants, the true stem cell nature of sca-1+lin– hemangioblasts was revealed (9, 62).
Clearly, it is not simple to verify the presence of hemangioblasts in adult human bone marrow, but there is evidence of their existence. When human CD34+ cells were xenotransplanted into mice to reconstitute the hematopoietic system, subsequent retinal injury in the animals resulted in human neovascular formation (27). However, since these were not single cell transplants, this is not definitive proof of hemangioblasts. In an in vitro study, clones were generated from single cell cultures of CD34+VEGFR2+ cells. Sibling cells replated in unicellular culture produced either hematopoietic or endothelial colonies, depending on the specific culture conditions, providing further evidence of adult human hemangioblasts (134). The self-renewal capacity of these cells was evaluated by culturing single CD34+VEGFR2+ cells for three 3-mo extended long-term culture rounds (i.e., 9 mo). Hemangioblast activity was observed in a subset of these long-term cultures, suggesting that at least some of these putative adult human hemangioblasts are stem cells (134).
As noted above, when attempting to isolate adult angioblasts or hemangioblasts, most investigators have relied on HSC antigens to do so. Initially, attention focused on CD34+ cells, which are enriched for HSCs(26, 96) and VEGFR2+ cells, because the antigen is expressed on embryonic angioblasts and many HSCs (43). Since then, almost every known HSC antigen has been used to enrich BMD cell for EC precursors. These include CD34 (human), sca-1 (mouse Ly 5.2), CD117 (c-kit), VEGFR2, and CD133 (AC133) (5, 53, 107, 128, 132, 140). There has been a great deal of controversy associated with which of these antigens or combination of them delineates the definitive profile for an adult EC precursor. Yet the question seems one that is impossible to resolve, at least with the selected antigens, since different studies often identify mutually exclusive subsets of cells. More recently, CXCR4, the receptor for the chemokine stromal derived factor 1 (SDF-1 or CXCL12) has been thrown into the mix. The SDF-1/CXCR4 axis is thought to play a key role in adult stem cell recruitment (as discussed below) and may prove to be an essential marker of EC precursors (24, 32, 199). However, because many differentiated cell types express this receptor, it may be necessary but is certainly not sufficient to identify bmdEC precursors (142).
In a 1983 review article, Schofield (156) wrote that "...stem cell properties do not reside in one specific cell type in the population but, when necessary, cells other than those normally playing the stem cell role, can have stem cell function imposed upon them by the appropriate microenvironment... .The postulate is offered that there are no cells which are intrinsically stem cells but that a range of cells in a tissue possess stem cell potential to a greater or lesser extent." Evidence supporting this hypothesis comes from work with mesenchymal stem cells. Mesenchymal stem cells from the five different mouse strains differed in their media requirements for optimal growth, rates of propagation, and presence of the surface epitopes CD34, sca-1, and vascular cell adhesion molecule 1 (133).
The ability to modulate antigenic phenotype in response to the environment is common to all cells. Just like an EC may regulate levels of VEGFR2 in response to shear stress, or Tie-2 and CD34 when activated, an EC precursor may modulate expression of CD133, CD34, or other proteins in response to environmental changes. At least some HSCs are initially CD34 negative, subsequently becoming CD34 positive (51). This concept of antigenic plasticity seems to have eluded the consciousness of many EC precursor biologists for a number of years, though, more recently, the notion of antigenic plasticity seems to have gained fairly wide acceptance. As a result, the frequency of heated discussions over the "true" antigenic phenotype of EC stem and progenitor cells has been reduced, and far less effort has been expended of late to characterize the definitive EC stem cell.
Cell Cycle and EC Stem Cells
Presumably to maintain their longevity and to minimize mutations due to replication errors, stem cells are normally relatively quiescent. Historically, assays such as long-term retention of the nuclear labels [3H]thymidine or 5-bromo-2'-deoxyuridine-5'-triphosphate (BrdU) in skin, oral mucosa (14), hair follicles (30, 121), intestine (137), cornea (29), and bone marrow (205) have been used to identify "label-retaining cells." The retention of the nuclear label is thought to indicate replicative quiescence, and, accordingly, cells that retain the label are thought to represent stem cells. These methods are useful for studying the cells in situ but are of less value in isolating the cells because the tissues need to be fixed to detect the nuclear label. To isolate stem cells, FACS sorting is commonly applied to cells after Hoechst 33342, rhodamine 123, or pyronine Y dye labeling. (16, 59, 61, 186). For poorly understood reasons, these dyes tend to be actively excluded from stem cells, and dye exclusion is used to select cells called side population (SP) cells. SP cells are enriched for HSCs and tissue-specific stem cells and are seven- to 10-fold enriched in G0/G1 cells relative to the general leukocyte population, demonstrating a correlation between phase of the cell cycle and the undifferentiated cell phenotype (16, 59, 61, 186). Selecting stem cells on the basis of on their cell cycle status has important ramifications with respect to their activity. For example, injection of 100 HSCs in G0/G1 phase rescued 90%, whereas the same number of stem cells in S/G2/M only rescued 25% of lethally irradiated mouse recipients (49). Furthermore, that an inability to maintain quiescence is associated with stem cell exhaustion has been demonstrated in HSC lacking the p21 (cyclin-dependent kinase inhibitor). These cells are forced to enter the cell cycle, increase their proliferation, eventually leading to stem cell exhaustion (25). Similarly, stem cells in mice lacking the Atm (ataxia telangiectasia mutated) gene actively divide, and this loss of quiescence appears to contribute to stem cell exhaustion and bone marrow failure (84).
Despite the relevance of cell cycle to stem cell identification and activity, an association between cell cycle status and putative EC stem cells has generally been overlooked. However, SP cells with angioblastic (and probably hemangioblastic) activity have been identified in BMD cells in both in vitro and in vivo studies (85). Furthermore, in an elegant study, De Falco et al. (35) demonstrated that differentiation into ECs was reduced in mitogen-stimulated c-kit+ BMD cells, whereas cytokine withdrawal or the overexpression of the cyclin-dependent kinase inhibitor p16 (INK4) restored differentiation, suggesting a link between cell cycle and EC precursor differentiation.
The inattention to the cell cycle in part may explain the often paradoxical results of different investigators; that is, expression of a particular antigen in one case is indicative of an EC progenitor, whereas its lack is in another. These data imply that relying on multiple characteristics such as size, cell cycle, cell adhesion, and other functional characteristics, rather than cell surface markers alone, may be more useful when isolating and characterizing EC progenitors. For example, selecting cells that express CD34 or VEGFR2 (but not necessarily both), which are in G0/G1 and of a particular size, may be more indicative of the EC stem cell phenotype than simple CD133+VEGFR2+CD34+ cell or similar cell selection.
EC Progenitors
Often, stem cells generate one or more intermediate cell types before fully differentiated cells are formed. For example, progeny of HSC include lymphoblastic and myeloid progenitors, each of which is capable of short-term self-renewal and of differentiating into a variety of white blood cells (165). These cells are far more abundant than their relatively rare stem cell parents and initially arise when a stem cell divides asymmetrically to produce a stem cell (which returns to a quiescent state) and a progenitor (Fig. 1). The progenitors can further replicate and then differentiate or they can differentiate directly, after which their differentiated progeny replicate. It seems likely that some form of EC progenitor exists, though through which scenario hemangioblasts give rise to bmdECs is not known (Fig. 1).
|
Innumerable studies have been and are being carried out in which whole bone marrow or total PBMCs are studied under the tacit assumption that the "active" component is the HSC-like cells (6, 89, 90, 172). This lack of enrichment for any subpopulation makes identification of EC progenitors necessarily retrospective. That is, cells that have taken on an EC phenotype are identified as EC progenitors, but what is actually being assayed are not EC progenitors but, rather, ECs derived from BMD cells. Moreover, while in some cases a careful and thorough characterization of the putative ECs has been performed, others have been less rigorous. Commonly, uptake of acLDL in combination with binding of Ulex lectin has been used to identify EC progenitors; however, as noted above, this is also a phenotype characteristic of monocyte/macrophages. The failure to thoroughly characterize cell populations has led to confusion within the field and created skepticism about the significance of the work among scientists outside the field.
Recently, it has been suggested that not all bmdEC progenitors are hematopoietic cells. CD45 is considered to be a pan-leukocyte marker; however, it has been found recently that both CD45+ and CD45–sca-1+lin– cells are present in the mouse bone marrow. Whereas CD45+sca-1+lin– cells are capable of reconstituting the bone marrow, CD45–sca-1+lin– cells are not (99). However, within this CD45–sca-1+lin– population are cells that express "tissue-specific" antigens (98, 194). That is, despite the fact that they express no other hematopoietic markers, they express the "stem cell" antigen sca-1 and the stem cell recruitment antigen CXCR4. Each also expresses a tissue-specific antigen for cardiac muscle, endothelial cells, liver, or other cell type. This has led to speculation that these represent a cadre of rare nonhematopoietic tissue-specific stem cells in the bone marrow (99).
Among cells that do take on an EC phenotype, two distinct phenotypes have been described: late and early outgrowth cells (53, 109). Early outgrowth cells, which are probably derived primarily from monocytic cells, proliferate modestly, express both monocytic and EC antigens, and die after a few weeks in culture (42, 46, 143). Late outgrowth cells, on the other hand, represent a small subset of adherent PBMCs that initially proliferate slowly (53, 82) but ultimately proliferate more rapidly and can be expanded from 20 to 1019 cells in 6 wk (109). That is, they act like classical primitive hematopoietic progenitor cells. Indeed, considering their immense replicative potential, one might even consider them stem cells. Unlike early outgrowth cells, these cells do not express CD1a or CD14 and are ultimately indistinguishable from microvascular endothelial cells (53, 109). Late outgrowth cells may be enriched in cells that express some combination of CD133, CD34, and VEGFR2 (42, 53, 58, 132, 197), though it is not clear that the lack of any of these antigens precludes a cell from being a late outgrowth cell (58, 109, 196). Moreover, the majority of cells in human CD34+ and mouse sca-1+ cell cultures are early outgrowth cells, so these antigens do not in and of themselves identify late outgrowth cells.
Despite the clear in vitro phenotypic distinction between early and late outgrowth BMD cells, the in vivo significance of this difference remains unclear. Early outgrowth cells are principally monocyte derived (46, 68, 143, 153, 154, 188), and, as such, their EC identity is somewhat ambiguous. Nevertheless, they do take on the key characteristics of ECs, including expression of eNOS (68). Furthermore, in vivo, monocytes are recruited to sites of injury where they traverse the vessel wall to enter the tissue. It is likely that at least some of these monocytes remain in the endothelium as ECs. Certainly, there is abundant evidence that they can intercolate among cells of the tunica media, differentiating into smooth muscle-like cells (146). Since the endothelium is normally a relatively quiescent tissue, these cells, which lack a high replicative potential, might be well suited to maintaining vascular homeostasis by repairing minor vessel damage.
On the other hand, late outgrowth BMD cells may also represent a major source of EC progenitors in vivo. Like the rapidly self-renewing mesenchymal stem cells (105, 164), such a rapidly proliferating cell could be advantageous at times of extensive vascular injury. Thus early outgrowth/monocytic progenitor cells may play a role in patching the vasculature, whereas late outgrowth/primitive progenitors may be more important during neovascularization. In fact, human aortic ECs apparently come in two flavors; one subset exhibits a high replicative potential similar to late outgrowth cells, whereas the other has a very limited ability to proliferate (81). Whatever the normal in vivo significance of early and late outgrowth cells, this distinction may have immense implications as we move toward using BMD cell in clinical applications.
Making New Blood Vessels With bmdECs
If BMD cells are a source of ECs, then how are the bmdECs used to make new blood vessels? Angiogenesis occurs in both the adult and embryo and is a process wherein differentiated ECs proliferate and migrate as the extracellular matrix is degraded at sites along existing blood vessels, allowing sprouts to form new tubes (22). Initially, capillaries are formed, some of which go on to recruit smooth muscle cells and mature into arterioles. It has long been thought that this was the only process through which adults form new blood vessels, and it appears that adult bmdECs contribute to this process. We found that capillary-like structures are formed in CD14+ BMD cell cultures through vacuolization and coalescence of multiple cells, a process normally associated with angiogenesis (150). In addition, numerous papers (4, 5, 68) have reported the sporadic integration of BMD cells into EC tubes in vitro and blood vessels in vivo, again consistent with a role for BMD cells in angiogenesis.
However, with the recognition of the existence of adult angioblasts, the possibility of adult vasculogenesis arises. Vasculogenesis is a key means of vessel formation in the embryo and involves the formation of a primary embryonic vascular network from in situ differentiating angioblasts (145). Studies of BMD cells indicate that this process can occur in the adult as well and that BMD cells can participate. We showed the formation of blood island-like structures by CD34+-enriched BMD cells in vitro (5); however, more importantly, the formation of blood vessels in situ from BMD cells was subsequently reported in vivo (174), both of which are suggestive of adult vasculogenesis.
Though some small neovessels may expand into collaterals in the adult, a series of studies by Schaper and colleagues suggest that collaterals are typically formed by remodeling of existing smaller vessels in a process termed "arteriogenesis" (183). In a very thorough study, this group found no evidence of BMD cell participation in arteriogenesis (208). However, these authors found no integration of BMD cells into any vessels, including capillaries, so it remains an open question whether BMD cells might be found to contribute to arteriogenesis in a different model.
There are two additional ways in which blood vessels can be formed. Intussusception, which occurs in both the embryo and adult, is the formation of a vascular network by division of larger vessels into smaller through focal insertion of a tissue pillar or by longitudinal foldlike splitting of a vessel (18, 37). This process is difficult to study, and to our knowledge, no one has looked at intussuception in vessels of bone marrow origin. However, it is a process associated with vessel development in tumors, and, as discussed above, many studies have found that BMD cells are involved in tumor vessel formation (130, 131). Finally, a new process, the "drilling" of tunnels by monocytes, was described only recently (3, 119). In this process, F4/80+ monocyte/macrophages pattern the tissue space by degrading the extracellular matrix, thereby facilitating the penetration of EC precursors and other cells. In addition, the F4/80+ cells can themselves form lumens and line the resulting tubes, apparently rapidly establishing a local microcirculation. The F4/80+ cells may also stimulate differentiation of progenitors in the tube. The origin of the ECs that ultimately come to line these drilled tunnels is not yet known.
Significance of bmdECs
Probably the first evidence that BMD cells might have a role in maintaining the vasculature came as early as 1963 from the observation of "fall-out" endothelialization, that is, deposition and differentiation of circulating cells on synthetic vascular grafts (169). That these cells were of bone marrow origin was convincingly demonstrated by Sauvage and colleagues using canine bone marrow chimeras (160). However, in vivo proof of ECs of bone marrow origin in humans came much later through study of chronic myelogenous leukemia patients carrying the BCR/ABL fusion gene. In patients who received bone marrow transplants from nonleukemic donors, both BCR/ABL positive and donor-derived ECs were found in the patients' myocardial and bone marrow endothelium (65, 66, 100). Engraftment of BMD cells into the bone marrow endothelium was particularly high after marrow transplant, with as many as 20–30% of ECs being of BMD cell origin. Moreover, in patients who suffered a leukemic relapse, the fraction of ECs lacking the BCR/ABL mutation decreased correspondingly (100).
Analysis of blood vessels from patients with sex-mismatched bone marrow transplants also has shown that BMD cells contribute to the endothelium in a variety of tissues (76, 87, 124, 138, 177). However, quantitation of the contribution of bmdECs to the endothelium has produced widely differing results, so the significance of bmdECs remains controversial. In a female-to-male heart transplant Y-chromosome labeling revealed that 20% of arterioles and 15% of capillaries were principally male derived in the infarcted heart (138), and in another, 16% of intramural and subepicardial cells were host derived (177). On the other hand, two of other groups reported that <0.1% of ECs in the heart are bmdECs (76, 124). In one of these studies, although the fraction increased after infarct, the highest percentage of bmdECs in the infarcted region among five patients was found to be 1.6% (76).
All of these human studies involved pathology, so it is important to consider how the situation might change in its absence. Even during normal physiological ovarian and uterine neovessel growth, BMD cells were identified throughout the endothelium though a quantitative assessment was not performed (4). Since new blood vessel growth rarely occurs in the adult, and turnover of the quiescent endothelium is thought to be low (56, 57), one would expect little integration of BMD cells into the quiescent vasculature. To investigate this, the endothelium of skin and gut of transplant recipients was examined, and a mean frequency of 2% donor-derived ECs were detected (87). Though obviously, since the patients required a transplant, this experiment had its limitations; nevertheless, it looked at vessels in uninvolved tissues. BMD cell contribution to the pancreatic endothelium was reported as 2.2% (117), and, in the quiescent vasculature of skin, aorta, and brain in mice, 0.2% to 1.4% of ECs were of bone marrow origin 90 days after transplant (31). This distribution was tissue dependent and varied depending on the antibody used to detect the EC. On the other hand, others failed to detect any BMD cell in uninjured tissue in the heart, liver, and colon in a mouse bone marrow transplant model (41, 85). Together, these studies suggest that the contribution of BMD cells to the quiescent endothelium is tissue dependent, and with consideration of the low mitotic index of the endothelium, BMD cells may make a major contribution to the maintenance of endothelial homeostasis in some vascular beds.
With the onset of neovascularization due to hypoxia, injury, or other stimuli, one might anticipate that the contribution of BMD cells to the endothelium would increase, and this has been observed repeatedly. For example, 8.3–11.2% of total ECs in the neovessels of an implanted angiogenic sponge and ischemic limb were found to be of bone marrow origin (31). Similarly, the contribution of BMD cells to skin graft neovascularization was 
15–20% of new ECs (21). In addition, when a gradient of hypoxia was created in skin wounds in mice, recruitment of BMD cells followed the gradient, and the greatest number of cells homed to and integrated into vessels of the most ischemic tissue (174). Also, the proportion of bmdECs in the pancreas of mice after streptozotocin treatment (which is toxic to the pancreas) was 17.8% (117).
In contrast, engraftment of bmdECs was only 1.6% in the peritubular endothelium of the kidney (40). When supra physiological levels of growth factors are introduced, even greater BMD cell recruitment can be seen. In mice implanted subcutaneously with fibroblast growth factor-2-impregnated Matrigel plugs, 26.5% of ECs in the neovasculature were bmdECs (123). Transplantation of VEGF-impregnated pellets in the cornea resulted in 17.7% of ECs in the neovasculature being of bone marrow origin (123). In another study using the same model, whereas only 7.3% was reported in the absence of simvastatin, drug treatment increased the percentage to 25.7% (112). What the many studies demonstrate, some which report no contribution of BMD cells to the endothelium (187, 208), is that the relative contribution of bmdECs to the growing vasculature depends on the nature of the stimulus as well as the tissue.
BMD Cells in the Tumor Vasculature
A key issue in BMD cell physiology is their potential role in tumor growth. Do they potentiate tumor vascular growth, and if so, are they a target for tumor therapy? A number of studies have examined this question in mice with bone marrow transplants and failed to detect bmdECs in tumors. After tamoxifen induction, bmdECs were not found in the endothelium of either normal or tumor (i.e., LLC and B6RV2 tumor) vasculature in wild-type tumor-bearing mice that received tamoxifen inducible Cre-SCL-lacZ bone marrow transplants (60) (SCL is expressed in ECs). Similarly, De Palma et al. (36) failed to detect bmdECs in TS/A carcinomas tumors in mice transplanted with Tie2-, CMV-, or PGKp-green fluorescent protein (GFP) transgenic bone marrow cells. This and two additional studies that failed to detect bmdECs among tumor ECs did, however, find BMD cells so closely associated with the endothelium that distinguishing between these perivascular cells and ECs was only possible through molecular ablation or high-resolution serial microscopy (141, 182). Thus it may be that some studies that suggest BMD cells have differentiated into ECs are actually observing their conversion to pericytes.
Of course, as with all other aspects of bmdEC biology, a number of studies suggest the contrary, that bmdECs are major contributors to tumor endothelium. bmdECs in the tumor endothelium were first demonstrated when MCA38 colon cancer cells were implanted subcutaneously into mice with bone marrow cells derived from mice with a VEGFR2-LacZ or Tie-2-LacZ transgene (4). X-Gal staining revealed numerous labeled cells in the vasculature of the developing tumor. Since then, studies have documented bmdECs in a variety of tumors. In mice with NXS2 neuroblastomas, 5% of CD31/CD34-expressing cells were BMD cells (33), and ECs of donor origin were "readily detectable" in Tg.AC line 43 carcinomas (159). The contribution of BMD cells to the endothelium was a more modest 1.3% (range 0.6–1.9%) in mammary adenocarcinomas (41), and in B6RV2 lymphoma cells, <1% of tumor ECs were of BMD cell origin, even when the tumor overexpressed VEGF or if VEGFR2 was constitutively activated in the BMD cells (102). However, in one surprising study of MCA/129 fibrosarcomas in mice, 50% of tumor neovessels were reported to be derived from BMD cells (52). The contribution of bmdECs to tumors in humans was documented when six different human tumors were examined, and bmdEC contributions to tumor endothelium ranged from 1 to 12.1% (136).
Id1+/–Id3–/– mutant mice have a diminished angiogenic capability and are tumor resistant (115). However, in Id1+/–Id3–/– mutant mice that received a wild-type bone marrow (BM) transplant, tumor growth was similar to that of wild-type animals, suggesting that tumor growth is dependent on BMD cells in at least some cases. Furthermore, bmdECs were localized in both tumors and implanted Matrigel plugs in the BM transplant mice. Interestingly, the type of tumor affects BMD cell recruitment to the vasculature, since when 10 different human tumor cell lines were transplanted into nude mice, bmdECs in the tumor endothelium ranged from 0 to 29.6%, depending on tumor type (149). The most significant predictor of bmdECs in the tumor endothelium (r = 0.89) was the amount of stroma in the tumor. If this correlation is found to hold over a variety of tumors in humans, it may be useful as a predictor of the potential responsiveness of a particular cancer to anti-BMD cell therapy.
A number of other factors also govern bmdEC contributions to the tumor vasculature (Table 2). Wild-type bone marrow was transplanted into Id1+/–Id3–/– mice carrying a TRAMP mutation that induces prostate tumors of multiple grades. Whereas in poorly differentiated tumors up to 14% of blood vessels contained bmdECs, BMD cells contributed few ECs in well-differentiated tumors, suggesting that the role of bmdECs in tumors may be influenced by the grade of the tumor (106). Additionally, formation of blood vessels by BMD cells appears to differ in primary versus metastatic tumors. When mammary carcinoma cells were implanted into the mammary fat pad, 1.5% of blood vessels were of bone marrow origin. This number increased more than 39-fold (58.4%) in brain metastases of the tumors. Similarly, whereas only 4.5% of vessels derived from BMD cells in adenocarcinomas implanted subcutaneously, >10% were of bone marrow origin in lung metastases (39). Genetic background can alter bmdEC recruitment, as the same adenocarcinoma transplanted into C57Bl/6 or FVB mice resulted in a more than 10-fold increase in bmdECs in tumors in the latter relative to the former (39). Consistent with this is the finding that, at 7 days in culture, BMD cells from FVB mice produced four times more ECs than did those from C57Bl/6 mice (113).
|
Not surprisingly, investigations have already begun into the possibility of using BMD cells to target tumors with anti-tumor agents, and in two studies, tumor growth was significantly inhibited in the mice carrying the genetically manipulated bone marrow (33, 48). In another, where Tie-2-expressing BMD cells were ablated by using a "suicide gene," blood vessel and tumor growth was significantly inhibited (33, 36, 48).
The wars over the significance of BMD cells in the endothelium and in other tissues are likely to continue for many years to come. In discussions over the past few years, a huge emphasis has been placed on potential artifacts in detection methods. Yet, whereas reevaluation suggests that there may have been flaws in quantitation and that, in some instances, cells that appear to be ECs may, in fact, be pericytes, that BMD cells can differentiate into ECs and other cells of the vascular wall seems clear. By careful analysis, we can sort out these technical problems. What ultimately will prove more difficult to understand are the fundamental biological changes in BMD cell behavior due to subtle methodological differences that can lead to apparently contradictory results. For example, both the type of stem cell engrafted into the bone marrow as well as the accessory cells transplanted with it may affect the nature of the stem cell engraftment. Furthermore, even if two different investigative teams both reconstitute the bone marrow with, for example, a single c-kit+sca-1+lin– cell, differences in data collected from the recipients may vary since engraftment is affected by cell cycle, and cell cycle can be modulated by the method of isolation (139, 180). Since strain differences can affect BMD cell behavior, it is likely that epigentic effects that arise in different mouse colonies will also produce functional changes. Yet another significant biological issue that is almost uniformly ignored is that GFP is toxic to cells and can alter their program of gene expression (110, 203). Given all of these variables, we should perhaps be less dismayed by the divergent results than amazed that there is any coherent picture emerging at all.
Is Fusion Relevant?
An unanticipated by-product of studies on BMD cells is a renewed interest in cell fusion. There is little evidence at this point that fusion is particularly significant in the endothelium, but it can occur, so it is worth a discussion of the topic. Also, it is important to be aware that fusion of BMD cells can be significant in certain cell types, such as hepatocytes and skeletal muscle, which might impact the availability or behavior of BMD cells to repair the vasculature in tissues in which fusigenic cells reside. The interest in fusion arose principally as a result of two studies published in 2002 showing that embryonic stem cells could fuse with neural and BMD cells in vitro (176, 201). However, embryonic stem cells are notoriously fusigenic, and the rarity of the fusigenic events was such that they could not possibly account for the widespread phenotypic changes seen in adult BMD cell cultures. Despite this, these unexpected findings led some to dismiss the relevance of adult stem/progenitor cells.
Unfortunately, these reports set the bmdEC field back considerably, but they also forced adult stem cell biologists to consider the possibility of heterokaryon formation. It is not uncommon in some adult tissues (e.g., liver and skeletal muscle) for cells of similar phenotype to fuse to form a syncytium with intact and distinct nuclei. Heterokaryons, on the other hand, are formed when two phenotypically distinct cells fuse and retain both nuclei and, until recently, had been observed only in the presence of nonphysiological stimuli (69). However, now numerous reports have verified that BMD cell heterokaryons form in many tissues (47, 67, 70, 185, 191, 192, 206). The background physiology of the mouse affects BM contributions to injured tissue and may dictate whether fusion or frank differentiation occurs (20, 38). Heterokaryons tend to be prevalent when the endogenous cells are defective, that is, where a functional benefit might be derived from fusion (47, 70, 185, 191). In situations where BMD cells are recruited to repair acute damage in otherwise healthy tissue, fusion rarely occurs (70, 101). Nevertheless, BMD cell heterokaryon formation has been detected in neurons and cardiac muscle even in the absence of injury (2).
There is little evidence for significant fusion of BMD cells in the endothelium, suggesting that fusion occurs principally in fusigenic cell types and rarely in others. In one report, only 8% of bmdECs appeared to be derived from fusion (206), whereas, in two other reports, no BMD cell-EC fusion events were detected (9, 87). Still, the extent of heterokaryon formation is variable within the same tissue type; it appears to be unique to each muscle (17), so it remains possible that BMD cells could form heterokaryons with higher frequency in some vascular beds under specific conditions.
What may become the greatest legacy of the fusion controversy is the identification of synkaryons in vivo (2, 126). Synkaryons arise when phenotypically distinct cells fuse not only their cytoplasm but their nuclei as well and, subsequently, undergo a reduction division. However, it is important to note that, once again, these events were observed when two fusigenic cell types were involved. Thus, whether or not this will prove to relevant to the endothelium or any other vascular wall cells remains to be seen.
What may be a more relevant fusion related event in the endothelium is mitochondrial transfer. Spees et al. (166) found that mesenchymal stem cells can rescue cells with dysfunctional mitochondria. This occurs by transfer of mitochondria from healthy to dysfunctional cells. If mitochondrial transfer also occurs between BMD cells and ECs, it might be a way for circulating cells to rescue ECs of which the mitochondria have been damaged because of overexposure to reactive oxygen species or other toxins.
Physiological and Pathophysiological Factors Governing BMD Cell Function
Local and systemic pathological changes, including vascular injury (58, 172), myocardial infarction (116, 162, 194), tissue hypoxia (34), choroidal injury (178), exercise (103, 148), and increased risk for cardiovascular disease and stroke (171, 184), among others, can affect the number of putative EC progenitors in the circulation. The recruitment of these cells into the circulation appears to be tightly regulated. Whereas CD34+, CD117+, and CXCR4+ PBMCs were elevated approximately two- to threefold in patients after myocardial infarct (which might just reflect generalized mobilization of BMD cells), c-met+ cells were elevated eightfold (194). That is, c-met+ cells were specifically recruited into the circulation. Whether stem cells with tissue-specific antigenic repertoires reside in and move out of the bone marrow when signaled or whether generic stem cells modulate their phenotype before or after leaving the bone marrow in response to a particular type of stimulus is not known. In any case, the retrieved total PBMCs produce more ECs in culture after these stimuli. This has been attributed to mobilization, but it, too, may simply indicate that circulating cells have modulated their phenotype in response to the stimulus.
One of the most potent stimuli recruiting BMD cells to any tissue is ischemia. Ischemia leads to an influx of inflammatory cells that release cytokines, proangiogenic factors, matrix proteases, and other molecules. Along with the influx of inflammatory cells, BMD cells expressing surface markers of EC precursors, including sca-1, Tie-2, Flt-1, vWF, VEGFR2, and CD133, are recruited in response to hindlimb ischemia, myocardial infarction, stroke, vascular trauma, pancreatic damage, and other ischemic events (58, 85, 117, 127, 172, 207). When a gradient of hypoxia was created in skin wounds in mice, recruitment of BMD cells followed the gradient (174). Not only did the greatest number of cells home and integrate into vessels of the most ischemic tissue, a significant proportion of the vessels in the ischemic tissue were of BMD cell origin.
Cells in the blood and vasculature may promote EC precursor quiescence. Addition of CD34–CD14– PBMCs (which are largely depleted of EC precursors) to cultures of CD14+ PBMCs significantly reduced the responsiveness of the CD14+ cells to proangiogenic growth factors (190). In addition, treatment with conditioned media from both coronary smooth muscle and HUVEC reduced EC number in CD14+ PBMC cultures. On the other hand, once the cells leave the vasculature, tissue-resident cells seem to have the opposite effect because culture in cardiomyocyte-conditioned medium increased the number of CD14+ PBMC-derived EC (190).
The number of putative EC precursors in the circulation as well as their colony-forming ability has been investigated as a biomarker for disease in a number of studies, and, generally, there is an inverse correlation with disease progression and the number of circulating cells and colony-forming units (44, 75, 95, 129, 155, 171, 184, 193). On the other hand, acute pathologies and pregnancy can transiently stimulate release of BMD cells into the circulation (116, 170, 194, 198.) These studies quantitate a variety of different putative EC precursors, and the precise choice of BMD cell studied may greatly influence the findings, since a recent study showed that there is variability in the correlation between any subset of circulating cells relative to another (55). For example, numbers of circulating CD34+CD133+ and CD34+VEGFR2+CD133+ cells are correlated, but these do not correlate with CD34+VEGFR2+ cell numbers. Moreover, the only correlations between circulating EC precursors and colony-forming units found were negative. Studies in Type 2 diabetic patients may give some insight into this complicated picture. Most but not all reports suggest that the number of circulating EC precursors is decreased in people with diabetes (44, 79, 104, 114, 151, 184). When Type 2 diabetic patients with, without, with either, or with both peripheral vascular disease or retinopathy were studied, the fraction of different subsets of putative EC precursors varied according to the type of complication. For example, fewer circulating CD34+ BMD cells were found in patients with diabetic retinopathy than in those without this complication (45). On the other hand, the CD34+VEGFR2+/CD34+ cell ratio was higher in this group than in any other, whereas only patients with peripheral artery disease had a reduced number of CD34+VEGFR2+ cells. In another study, the number of circulating CD34+CD133+ cells was governed by the SDF-1 allelic phenotype when diabetes was controlled but not in patients with high glycosylated hemoglobin (HbA1c) levels (79). Given this complicated picture, it is difficult to imagine that the number of a particular subfraction of circulating BMD cells will ultimately prove to be a useful predictive clinical parameter since so many different physiological factors can impact the numbers.
What is certainly at least as important as numbers of EC precursors is the function of the cells. The number of HSCs in the circulation does not decrease, nor is the ability of HSCs to reconstitute the bone marrow lost with age (92, 122). However, old cells cycle more frequently and home less efficiently than do young cells, indicating more subtle functional problems (122). We have found slightly fewer lin– cells in older C57Bl/6 mice, but have found no difference in their in vitro functions compared with young cells. In vivo, however, treatment of skin wounds with old lin– BMD cells profoundly inhibited vascular growth, whereas young cells stimulated it (152). These experiments were performed in young mice, so the dysfunction is intrinsic to the BMD cells. A reduction in the repair capacity of mesenchymal stem cells with age has also been observed (204). Cardiovascular dysfunction is associated with diabetes, and studies on bmdEC precursors have led to the consensus that the function of the cells is perturbed by diabetes (114, 163, 167, 173, 175). However, all EC precursors are not affected equally by the disease. Sca-1+lin– cells from diabetic mouse bone marrow are more susceptible to stress-induced cell death than are their nondiabetic counterparts when cultured in EC differentiation conditions (8). In contrast, no differences were observed in cultures where ECs were derived from whole bone marrow cultures. The sca-1+lin– cells were unable to stimulate restoration of blood flow in an ischemic limb and inhibited it in skin wounds (8, 167)
Molecular Factors Governing BMD Cell Function
Considering the pleiotropic effects of VEGF on ECs, it is not surprising that it can stimulate differentiation and proliferation of EC precursors (10). Less expected were data suggesting that VEGF can mobilize and recruit EC precursors from the bone marrow into tissues (6, 50, 63, 120), though one recent study was unable to detect mobilization (102). VEGF mobilization appears to be isoform specific, because VEGF165, but not VEGF189, induced a rapid mobilization of VEGFR2+ cells into the circulation (71). Like VEGF, the VEGF family member placental growth factor (PLGF) may also modulate recruitment or function of BMD cell EC precursors. Embryonic vessel growth is unaffected, but arteriogenesis is delayed in mice lacking PLGF. However, if the bone marrow of PLGF–/– mice is replaced by wild-type bone marrow, normal arteriogenesis is restored, suggesting that BMD cells stimulate vascular growth or stabilization through PLGF and VEGFR1, its receptor (157). In vitro, freshly isolated VEGFR1+ BMD cells failed to differentiate into ECs in one study (125), but in another (173), PLGF administration stimulated production of ECs from BMD cells by sixfold. Additional studies suggest that, by increasing signaling through VEGFR1, PLGF amplifies the responsiveness of ECs to VEGF (23). Together, these suggest that PLGF probably does not play a role in differentiation of the BMD cells to ECs but may induce proliferation in the partially differentiated cells. It may also act in some unspecified ways to stabilize larger vessels.
What has probably become the most surprising and complicated story with respect to molecular regulators of BMD cells is SDF-1. SDF-1 mobilizes hematopoietic stem and progenitor cells from the bone marrow (72, 195) as does the CXCR4 (the SDF-1 receptor) antagonist AMD3100 (108). SDF-1 is chemotactic for EC precursors (35, 120, 199), and in vitro data suggest it is required for recruitment of bmdEC precursors to sites of injury (24, 35). Studies with human into mouse bone marrow transplants support this conclusion, because blockade of CXCR4 abrogated human CD34+ BMD cell homing to the liver, whereas local injection of SDF-1 increased their homing (94). In contrast, overexpression of SDF-1 in the heart failed to recruit BMD cells even though AMD3100 blockade of CXCR4 diminished recruitment of BMD cells to the infarcted heart (1). Similarly, overexpression of SDF-1 in the absence of VEGF failed to recruit BMD cells to the liver, whereas blocking CXCR4 decreased BMD cells in the liver even in the presence of high levels of VEGF. Together, these data suggest that SDF-1 is not sufficient to recruit the BMD cells to the tissue in the absence of another signal, such as VEGF or one of its targets. Instead, SDF-1 may play a crucial role in sequestering BMD cells at the site of injury once they have arrived, because three studies have shown that the molecule is required for adhesion of BMD cells at sites of injury (24, 35, 63).
As noted above, hypoxia is a potent mobilizer of BMD cells, and hypoxia-inducible factor-1
(HIF-1) activity is induced by hypoxia (77). VEGF, in turn, is induced by HIF-1 which then up-regulates SDF-1 and CXCR4 (97, 111, 200). A positive feedback loop occurs because HIF-1 also induces SDF-1, and SDF-1 is a weak inducer of VEGF (93, 118). Thus it appears that VEGF and SDF-1 work in concert such that VEGF and SDF-1 induce each other, SDF-1 mobilizes BMD cells, VEGF (or a VEGF induced factor) recruits the cells, and SDF-1 sequesters the arriving BMD cells, and VEGF promotes the differentiation and proliferation of the sequestered cells. In vitro data suggest that SDF-1 may also promote the viability and differentiation of the BMD cells in the tissue (35) (Fig. 2). This pathway may not apply in all tissues, though, because overexpression of VEGF in the heart resulted in only a slight increase in the number of circulating BMD cells, whereas an eightfold increase in circulating BMD cells resulted from overexpression in the liver (63).
|
Unfortunately, we cannot just continuously infuse SDF-1 systemically and stimulate vascular growth, since SDF-1 binding leads to internalization of CXCR4 and downregulation of the receptors (32). Thus too much of a good thing can be problematic. We have found, however, that administration of two boluses of AMD-3100 on the day of and 2 days after induction of hindlimb ischemia potently stimulates revascularization of the limb (88). This suggests that, by modulating timing and dose, BMD cells can be efficiently mobilized, but CXCR4 receptors can be returned to the cell surface quickly enough to allow the cells to be recruited to the site of injury. What is more, CD34+ BMD cells mobilized by the drug are more potent stimulators of vascular growth than are unmobilized cells, suggesting that AMD-3100 either mobilizes a particularly proangiogenic type of CD34+ cell or in some way primes them (88).
Though we may not fully understand them, SDF-1 and its cognate receptor CXCR4 clearly play important roles in BMD cell differentiation and ability to induce vascular growth. The recent discovery that the orphan receptor RDC1 (CXCR7) is also a receptor for SDF-1 will undoubtedly make sorting out their precise functions even more difficult (10). In addition, blocking CXCR4 is more effective in Balb/c than in C57Bl/6 mice with respect to inhibiting tumor vascular growth, indicating that genetic contributions will further complicate matters (64).
Another factor that may be critical in the HIF-1/SDF-1 pathway is monocyte chemoattractant factor-1 (MCP-1). Monocytes are attracted to sites of injury, and once there themselves secrete MCP-1, which, in turn, stimulates HIF-1 activity (78). Responsiveness of monocytes to VEGF is reduced in people with diabetes, resulting in fewer monocytes being recruited to sites of injury, the failure of which may lead to lower HIF-1 activity and reduced engagement of the VEGF/SDF-1 system (189) (Fig. 2). Consistent with this, we have found that implantation of CD34+ or CD14+ BMD cells in diabetic ischemic limbs results in elevated levels of MCP-1 in the tissue, increased vascular growth, and accelerated blood flow restoration (7). The attracted monocytes may also have an important role in laying out the scaffold on which neovessels will form (3, 119).
Transforming growth factor-
1 (TGF-1) is a pleitropic factor that can regulate the balance between proliferation and differentiation of hematopoietic cells and can modulate vascular growth (12, 135). We have shown that TGF-1 leads to rapid cell death of total CD34–, but not CD34–CD14+, BMD cells in culture (190). Furthermore, despite high levels of TGF-1 in the conditioned medium of HUVEC, the medium stimulates production of ECs from CD34– BMD cells. EC stemlike precursors are also affected. Treatment of CD34+ cells with SDF-1 reduces the proportion of cells expressing CXCR4, but combination treatment with TGF-1 and SDF-1 inhibits the loss of CXCR4 by cells (12). Correspondingly, chemotactic and adhesion responses to SDF-1 that were lost on prolonged SDF-1 treatment were restored after coincubation with SDF-1 and TGF-1. Thus, though TGF-1 may not be considered a primary controller of EC precursor function, it is likely to be playing a role in the background, modulating the effects of other factors.
Erythropoietin (Epo) is a potent stimulator of vascular growth (86, 144), and, like SDF-1, it is strongly induced by hypoxia. One study in mice suggests that Epo increases the number of circulating CD34+VEGFR2+ BMD cells (73), whereas, in another study of 28 patients with congestive heart failure, long-term Epo treatment failed to increase the number of CD34+, CD34+CD45+, CD34+CD133+, CD34+VEGFR2+, or CD34+CD133+VEGF-R2+ cells in the circulation (54). Epo clearly has direct effects on EC precursors, stimulating proliferation of CD14+ BMD cells and their differentiation into ECs as well as their adhesion to vascular ECs (54, 73, 190). Additionally, in culture conditions used in our laboratory, addition of Epo to the culture medium reduces the serum requirement for maximal growth from 20 to 10%. (G. C. Schatteman and C. Jao, unpublished observations).
Finally, the angiopoietins have a role in mobilization and function of EC precursors. Angiopoietin-1 appears to regulate initial commitment to the EC lineage, whereas angiopoietin-2 induces BMD cell EC expansion in vitro and stimulates formation of neovessels by BMD cells in vivo (74). We have found that complete blockade of Tie-2 leads to rapid EC precursor death in culture (G. C. Schatteman and C. Jao, unpublished observations). In vivo, CD34+ BMD cell tend to localize near sites of angiopoietin-1 and -2 immunoreactivity. However, in brain tumors, though CD34+ cells continued to be associated with angiopoietin-2 (>80%), few (<20%) colocalized with angiopoietin-1, suggesting that angiopoietin-2 activity may create a proangiogenic environment that enhances the recruitment of BMD cells into the vasculature of the tumor (181). Like VEGF, angiopoietin-1 can mobilize VEGFR2+ cells, but the mobilization is delayed relative to response to VEGF (71, 120). It has also been reported that Tie-2+ but not Tie-2– BMD cells can differentiate into ECs in culture (125). It is worth noting that, although many investigators view VEGFR2 as the "best" indicator of an EC precursor, whereas 80% of Tie-2+ BMD cells take on an EC-like phenotype in culture, only 6% of the freshly isolated cells are VEGFR2+ (125). In contrast, 55% of VEGFR2+ cells express Tie-2.
Where Do We Go From Here?
The biology of EC precursors derived from the bone marrow is not well understood, but considering that the scientific community only became fully engaged in their study in about 1999, the progress has been substantial. We have discovered that 1) eNOS is a good marker of the EC phenotype; 2) there are indeed adult hemangioblasts; 3) there are two distinct classes of EC precursors in the bone marrow; 4) some BMD cells may fuse (though this is probably rare in the endothelium); 5) systemic physiology can profoundly alter EC precursor function; 6) ischemia is a key inducer of neovascularization via BMD cells; 7) EC precursors can exacerbate tumor growth, particularly metasteses; and 8) classical factors that regulate hematopoiesis and vascular growth also control of EC precursor function. We have also relearned the lessons that we should not assume that a positive finding in one setting will be applicable to another and that a negative finding in one study does not mean that something never happens.
As in all biological processes, EC precursor function is governed by a plethora of systemic and local factors. Because we are on the verge of a new imaging era, in which it may be possible to follow single cells in vivo in real time, many of the thorniest fate mapping problems may become tractable in the foreseeable future. It may also make it feasible to study BMD cells in multiple strains of animals simultaneously since a single animal can serve for multiple time points. Understanding molecular regulation of the cells, as always, may be much more problematic. Tissue culture cannot recapitulate the in vivo molecular environment, and this is compounded by the fact that most investigators use static two-dimensional culture systems. Nevertheless, a number of key regulators have been identified and provide us a starting point. Also, now that we understand that there is no single EC precursor, we can focus our effort on understanding the significance of these phenotypic variations and use them to our advantage in better understanding EC precursor biology.
|
GRANTS |
|---|
|
TOPABSTRACTGRANTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTGRANTSREFERENCES |
|---|
plays a critical role in stem cell recruitment to the heart after myocardial infarction but is not sufficient to induce homing in the absence of injury. Circulation 110: 3300–3305, 2004.1 modulates responses of CD34+ cord blood cells to stromal cell-derived factor-1/CXCL12. Blood 106: 485–493, 2005.
