however, BM-derived cell integration into the endothelium varies among vascular beds [88–91]. The contribution of BM to the endothelium of injured tissue ranges from 1% to as high as 50% of vessels [75, 84–86]. These data strongly suggest that the magnitude of recruitment of EPC may be organ-specific and dependent on the extent of vascular injury and remodeling.

Recruitment of specific subsets of HSCs may be essential for the proper repair and incorporation into locally derived endothelium. Compelling evidence suggests that unique subsets of proangiogenic HSCs support angiogenesis postnatally not only by incorporating into the vascular lumen but by delivering bioavailable angiogenic factors including VEGF, matrix metalloproteases (MMPs) and angiopoietins to the neovessels [92– 98]. Monocyte precursors of EPC such as CD14+ cells contribute to neoangiogenesis by releasing MMP-9 [99] and MMP-12 [100].

Methods for Studying Stem/Progenitor Cell Behavior

Much of the work in characterizing the contribution of HSCs has been with the use of chimeric animals where the BM cells from a donor are labeled for tracking using either transgenic fluorescent proteins or an overexpressed protein that can be detected immunologically, such as LacZ. Similarly, male BM cells transplanted into female recipients can be detected by fluorescence in situ hybridization to the Y-chromosome. In any case, the donor cells are given to recipient animals whose own BM has been ablated either chemically or by high-dose irradiation.

Alternatively, either HSCs or EPCs may be administered directly into the circulation without BM ablation of the recipient in a method known as adoptive transfer. This technique has been extremely useful, specifically to examine the contribution of specific stem cell subpopulations to repair. It has been used therapeutically

with adoptive transfer of EPCs to restore blood flow and increase capillary density, resulting in decreased loss of limbs and recovery from myocardial ischemia [101–107]. EPCs can be ex vivoexpanded and then infused and have been shown to improve neovascularization in hind limb ischemia models [108] and improve ejection fractions and end-systolic volumes, indicating better cardiac function in myocardial infarction models [84, 106, 107]. Several clinical studies showed similar effects [109].

Factors Regulating Stem and Progenitor Cell

Involvement in Angiogenesis

Numerous hypoxia-regulated factors have been implicated in angiogenesis. VEGF is by far the most well studied [110–113]. Even minor states of hypoxia can promote VEGF expression through a family of hypoxia-inducible transcription factors that bind to a hypoxia response element in the VEGF promoter [114]. Six isoforms of VEGF exist including placental growth factor (PlGF). PlGF can stimulate angiogenesis in vivo [115], migration of endothelial cells in vitro, potentiate the effect of VEGF on permeability, and induce chemotaxis of monocytes [92, 116–119]. Other isoforms, VEGF-A and VEGF-B, are highly expressed in EPCs as compared to human umbilical vein endothelial cells and human microvascular endothelial cells [120]. VEGF receptors, VEGFR-1, VEGFR-2 and VEGFR-3 only bind certain isoforms of VEGF. Ligands for VEGFR-1 include VEGF-A, -B and PlGF; ligands for VEGFR-2 include VEGF-A, -C, -D, and -E, while ligands for VEGFR-3 are VEGF-C and -D. Thus, PlGF uniquely binds VEGFR-1, and VEGF-E uniquely binds VEGFR-2. By using these specific ligands, the activities of these receptors can be dissected [121, 122].

VEGFR-2expressionisupregulatedbyhypoxia and possibly by VEGF-A, and it is accepted as the receptor that mediates functional VEGF signaling

in endothelial cells [123]. The role of VEGFR-1 is less clear as it may function as a negative regulator of VEGFR-2 [124]. VEGFR-1 signaling may also be involved in migration of monocytes and endothelial cells induced by PlGF and VEGF-A, due to its ability to induce tissue factor [117]. PlGF regulates interand intramolecular crosstalk between VEGFR-1 and VEGFR-2 tyrosine kinases. Activation of VEGFR-1 by PlGF resulted in intermolecular transphosphorylation of VEGFR-2, thereby amplifying VEGF-driven angiogenesis through VEGFR-2 [125]. These studies show the complexity of the VEGF signaling mechanisms. Furthermore, aspects of signaling of these receptors may be context-dependent as well as cell type specific. Most of what is known about these receptors in HSCs is their surface expression as determined by flow cytometry analysis. Less is known about their characterization in vivo.

SDF-1 is the principal chemokine responsible for the localization of HSCs to the BM niche and subsequent mobilization to the circulation. Together with VEGF, SDF-1 not only stimulates the migration of mature endothelial cells but also acts as the main chemoattractant to promote homing and tissue invasion of endothelial and progenitor cells [126]. SDF-1 expression is increased in response to tissue ischemia and its expression is regulated by VEGF. We demonstrated that SDF-1 is elevated in the vitreous fluid of diabetic patients and correlates with vitreous VEGF levels and with retinopathy severity [127]. BlockingSDF-1preventsrecruitmentofHSCsand EPCs to the retina [65] and choroid [68] following injury of these areas, and thus prevents development of neovascularization. Overexpression of SDF-1 promoted neovascularization of ischemic tissues [128].

Picomolar concentrations of SDF-1, similar to those found in the vitreous of patients with proliferative diabetic retinopathy (PDR), increase CD34+ cell migration [129] and promote nondiabetic CD34+ cell differentiation into endothelial cells by increasing VEGFR-2 surface expression

[Grant, unpubl. studies]. Exposure of CD34+ cells to SDF-1 at a concentration of 0.1 ng/ml results in a rapid increase in VEGFR-2 expression with a gradual return to baseline over a 6-hour period and no change in VEGFR-1 expression. In contrast, a high concentration of SDF-1 (100 ng/ ml) results in a sustained increase in VEGFR-2 expression and no change in VEGFR-1 levels. This suggests that SDF-1 is mediating its effects in CD34+ cells via VEGFR-2 activation, whereas pigment epithelium-derived factor inhibits VEGFR-2-induced angiogenesis via VEGFR-1 [130].

Insulin-like growth factor (IGF)-1 is a potent antiapoptotic protein and promotes angiogenesis in different models [131, 132]. Urbich and Dimmeler [106] and Urbich et al. [133] found that IGF-1 mRNA was highly expressed in CD34+ cells when compared to mature endothelial cells or CD14+ monocytes, which produce approximately 10-fold less IGF-1 mRNA. IGF-1 is needed for survival of EPC populations in culture [133]. IGF-1 is regulated by a series of binding proteins (BPs); IGFBP is the most abundant BP in serum. Finally, Liu et al. [134] showed using in vitro cell proliferation assays that the addition of exogenous IGFBP-3 to cultures of purified CD34+/– CD38–Lin– cells stimulates the proliferation of primitive hematopoietic cells with CD34+CD38– phenotype, suggesting that IGFBP-3 is capable of expanding primitive human blood cells. Our data show that IGFBP-3 stimulates migration, tube formation and differentiation of CD34+ cells into endothelial cells in a dose-dependent manner [135].

The expression of IGF-1 in EPC has been shown. Urbich et al. [136] analyzed the expression profile of cytokines in human peripheral blood-derived EPC, human umbilical vein endothelial cell, human microvascular endothelial cell and CD14+ monocytes by microarray technology. These authors found that IGF-1 mRNA was highly expressed in EPC when compared to mature endothelial cells or CD14+ monocytes

which produce approximately 10-fold less IGF- 1 mRNA. These results suggest that progenitor cells may promote neovascularization-releasing factors, which act in a paracrine manner to support local angiogenesis and mobilize tissue-re- siding progenitor cells. Hanley et al. [137] identified specific targets of IGF-1 within human fetal BM (FBM). These authors found that IGF-1 stimulated the expansion of primitive multilineage CD34+CD38– hematopoietic progenitor cells and increased yields of several hematopoietic subpopulations, including CD34+CD38+CD10+ lymphoid progenitor cells. Additionally, IGF-1 had direct effects on FBM stromal elements, inducing the expansion of myeloid-like CD45+CD14+ FBM stromal cells and enhancing production of the hematopoietic cytokine interleukin-3 by fi- broblast-like CD45-CD10+ FBM stromal cells.

In addition, Kim et al. [133] demonstrated that AC133–CD14+ cells from human umbilical cord blood are able to develop endothelial phenotype with expression of endothelial-specific surface markers and form cordand tubular-like structures in vitro. The AC133-CD14+ cells were grown in medium supplemented with fetal bovine serum, VEGF, basic fibroblast growth factor and IGF-1. After 14 days, the cells formed cordand tubular-like structures, and showed a strong increase in the endothelial marker P1H12 over time. In addition, CD14 decreased, and CD45 did not change. The cells also expressed endothelial markers von Willebrand’s factor, platelet/endothelial cell adhesion molecule-1 (CD31), VEGFR-1, VEGFR-2, eNOS and VE-cadherin, but did not express Tie-2 after 7 days of culture. Finally, Liu et al. [134] showed in in vitro cell proliferation assays that the addition of an exogenous IGFBP-3 to cultures of purified CD34+/– CD38–Lin– cells stimulates the proliferation of primitive hematopoietic cells with CD34+CD38– phenotype, suggesting that IGFBP-3 is capable of expanding primitive human blood cells.

SDF-1 and IGF-1 released from EPCs which have already been recruited into the ischemic

tissue may promote vascular remodeling of resident cells [104, 138]. The release of factors from EPCs involved in neovascularization is a dynamic process, and it is very likely that the expression pattern of angiogenic factors by EPCs may regulate their differentiation and may change during different EPC activities such as homing versus vascular incorporation in ischemic tissue.

The role of monocyte chemoattractant protein (MCP)-1 and its receptor (CCR2) in repair has also been examined. Sakai et al. [139] showed that human peripheral CD14+ cells contribute directly to fibrogenesis by an MCP-1/CCR2-dependent amplification loop. These authors investigated the effect of MCP-1 on the expression of MCP-1, CCR2, transforming growth factor-β1 (TGF-β1) and type I collagen in circulating human CD14+ cells. They found that the stimulation of CD14+ cells with MCP-1 increased mRNA and protein levels of TGF-β1 and a pro-α1-chain of type I collagen. Similarly, the expression of MCP-1 and CCR2 was enhanced by the stimulation with MCP-1 in doseand time-dependent manners. Umland et al. [140] showed that CD34+ BM cells stimulated by TNF-α also show enhanced secretion of MCP-1. Awad et al. [108] demonstrated that at least some progenitor-induced healing is probably mediated through increased sensitivity to VEGF and increases in MCP-1, and possibly modulation of angiopoietins. These authors showed that injection of CD14+ and CD34+ cells into mice improved healing and vessel growth associated with the expression of VEGF and MCP- 1 proteins. Nakajima et al. [141] found that in the pathogenesis of multiple sclerosis (MS), the CD14+CCR2+ blood monocytes may play an important role in the shift from active disease to a state in which MS is in remission. These authors found that expression of CCR2 and CD14 on the monocytes in the MS patients was markedly decreased, and there was a significant negative correlation between the Th1/Th2 ratio [CD4+CXCR3+ cells (Th1), CD4+CCR4+ cells (Th2)] and the CCR2 and CD14 expression on

monocytes. However, despite all these studies, the magnitude and temporal sequence of MCP- 1 expression in relation to tissue injury and regeneration following ischemic injury remains unknown. Shireman et al. [142] found that the transient increases and selective tissue distribution of MCP-1 during early inflammation and muscle regeneration, in a mouse model of femoral artery excision, support the hypothesis that this cytokine participates in the early reparative events preceding the restoration of vascular perfusion following ischemic injury.

CD34+ and CD14+ Cells in Diabetes

CD34+ Cells from Diabetic Patients Have Impaired Migration

As discussed previously, one marker that has been extensively used to identify the origin of human EPCs among hematopoietic cells is CD34 [143–147]. There is considerable disagreement in the literature as to whether CD34 is found on HSCs or whether it is expressed by more differentiated HSC progeny such as EPCs [145, 148– 151]. Angiogenesis can be amplified by injection of CD34+ cells [77]. Among those who feel that CD34 is expressed by less differentiated stem cells, it has been hypothesized that the presence of CD34 may represent an activated state of the stem cell [152, 153]. These data indicate that CD34+ cells are involved in stem/progenitor cell identification and angiogenesis; however, the precise mechanisms of (inter)action have yet to be determined.

Defective CD34+ function is associated with diseases such as diabetes [88, 90, 154–163]. Diabetes is associated with reduced mobilization of CD34+ cells from the BM, reduced numbers of CD34+ in the circulation, reduced migration of CD34+ cells into areas of ischemia, reduced incorporation of these cells into capillaries and reduced differentiation into endothelial cells [83, 161, 162, 164, 165]. Blood glucose control also

correlates with CD34+ cell counts, with better control associated with higher numbers [161, 162]. However, diabetic CD34+ cell growth defects are not reversed by cultivation in normoglycemic medium, suggesting that the impairment of CD34+ cells is not reversible by glucose correction alone [162].

We have demonstrated that CD34+ cells isolated from diabetic individuals have defective migration in response to SDF-1 [129]. We have since studied the migration of CD34+ cells isolated from patients with type 1 or type 2 diabetes in response to VEGF and IGF-1 and have found that the diabetic CD34+ cells have defective migration to these factors. These data suggest that the defect in migration of diabetic CD34+ cells is a generalized defect to all hypoxia-regulated factors. We have characterized the mechanism of this defect in diabetic CD34+ cells by measuring intracellular, bioavailable NO using diaminoflu- orescein-FM [129]. Diabetic CD34+ cells inherently have diminished NO compared to CD34+ cells isolated from healthy controls. This does not appear to be the case in CD14+ cells of diabetic and nondiabetic origin, and these cells migrate to MCP-1.

Human nondiabetic blood-derived CD34+ cells promoted revascularization of skin wounds in mice with type 1 diabetes [160]. In a nude mouse model of hind limb ischemia, exogenous nondiabetic blood-derived CD34+ cells profoundly accelerated blood flow restoration in type 1 diabetic mice [163]. Lambiase et al. [165] demonstrated that reduced numbers of CD34+ cells with impaired chemotactic and proangiogenic activity exist in type 1 diabetics and that when infused result in reduced formation of collateral vessels.

CD14+ Monocytes Participate in Capillary Formation

Traditionally, monocytes were considered a homogenous class of blood mononuclear cells, behaving mostly as acute-phase phagocytes and as

precursors of tissue macrophages. With the recent progress in the understanding of their true heterogeneity in the blood, the very notion of ‘monocyte’ [166], as well as the whole ‘mononuclear phagocyte system’ seems to have outlived its usefulness [167]. In particular, monocytes and their descendants in culture were repeatedly shown to acquire endothelial properties when exposed to appropriate growth factors [168, 169]. In vivo, incorporation of monocyte descendants into neovessels is more and more accepted [170, 171].

Moldovan and coworkers demonstrated that monocytes and macrophages participate in neovascularization by staging a pattern for development of new capillaries. To better understand this process, they developed in vitro and in vivo models of extracellular matrix invasion by monocytes, identified either by their origin, or by the F8/40 marker [172].

These cells form tubular, low-density domains (tunnels) in the Matrigel, are often at the ‘tips’ of new capillaries and pave the way for subsequent vascular maturation by providing a conduit to revascularization. In vitro, they confirmed both formation of tunnels and the adoption of a cylindrical shape by many cells, consistent with a transcellular lumen [173]. The polarized matrix dissolution and stepwise development of mac- rophage-generated intracellular vacuoles, culminating with formation of lumen is remarkably similar to lumen formation in endothelial cells. Moreover, macrophages in their in vivo experimental model also formed a lumen and generated branching patterns, supporting the recent suggestion that macrophages could control the branching of capillaries [99].

The many different metabolic perturbations typically associated with diabetes including excess free fatty acids, insulin resistance, oxidative stress, PKC activation and others may impact EPC behavior. In addition, vascular basement membranes including those of the retinal capillaries are heavily modified by advanced glycation

end products’ crosslink formation [174, 175] and the impact of these changes on EPC behavior has not yet been characterized.

It is well-recognized that cellular phenotype and response to exogenous factors are highly dependent on receptor expression; however, receptor expression in EPCs has been mostly used to classify the EPC population rather than characterize cell function. VEGF receptors, VEGFR-1 and VEGFR-2, and the SDF-1 receptor CXCR4 are expressed on CD34+ and CD14+ cells. However, quantitation of receptor number in health and disease and regulation of receptor expression by their appropriate ligands remains largely unknown. The differential interplay between VEGF receptors, CXCR4 and other growth factor receptors in these two cell populations will determine their ability to differentiate into endothelial cells. In addition, these recep- tor-ligand interactions will regulate the production of cytokines by activated CD34+ and CD14+ cells and orchestrate their complex behavior in vascular remodeling.

Diabetic CD34+ cells are defective and less able to repair ischemic regions associated with acellular capillaries. These defects could manifest themselves as reduced attachment, differentiation and invasive potential by CD34+ cells. The repair of injured vessels will require the EPCs to first attach, migrate through any thrombus/matrix in the region and finally differentiate into endothelium. The incorporation of EPCs into preexisting vessels relies not only on growth factor gradients and recruitment factors but also on appropriate interaction with the underlying vascular basement membrane that is exposed after endothelial cell death.

CD34+ cells of patients with type 2 diabetes show impaired adhesion to the endothelium, decreased proliferation and aberrant tubule formation [161]. Murine Sca-1+ HSCs dramatically improved vascularization of skin wounds in obese type 2 diabetic Leprdb but not in congenic lean nondiabetic C57Bl/6 mice [155]. Moreover,

when skin wounds of Leprdb mice were treated with Leprdb-derived Sca-1+ HSC-enriched BM cells, wound vascularization was severely inhibited [155]. Awad et al. [156] demonstrated that the obese type 2 diabetes syndrome induces intrinsic defects in CD34+ EPCs but not in CD14+ monocytic cells. The defects in CD34+ cells were evident in vitro by decreases in CD34+ cell-de- rived endothelial cells after stress and in vivo in nondiabetic mice by the reduction in vascular growth in skin wounds and exacerbation of ischemia-induced tissue damage in limb muscle. The behavior of BM cells in diabetic and nondiabetic environments may differ [154, 155], and there may be negative synergism between the diabetic environment and diabetic BM-derived cells.

MCP-1 is the primary chemokine that induces CD14+ cell migration. Interestingly, blocking p38 MAPK in CD14+ cells promoted endothelial differentiation [176]. Therapy with CD14+ cells improved healing and vessel growth, although not as rapidly or effectively as CD34+ cells. Cell treatments with either cell type modulated local expression of VEGF, MCP-1 and angiopoietin. Most importantly, in diabetes CD14+ cells are not hindered in their angiogenic activity as are CD34+ cells [108]. Intramuscularly injected freshly isolated CD14+ cells, CD34+ cells, or the combination of the two increased arteriolar density and promoted muscle salvage in the diabetic mouse ischemic hind limb. All cell treatments also accelerated blood flow restoration, but with different kinetics. Western analysis showed distinct patterns of proangiogenic factor expression in CD34+ and CD14+ cell-treated limbs [108]. CD34+ cells isolated from umbilical cord blood and exposed to VEGF showed increased expression of CD14 and rapid differentiated into endothelial cells in vitro. These studies suggest that when CD34+ cells differentiate towards the more mature CD14+ cell, they become less vulnerable to the adverse conditions associated with diabetes.

CD34+ and CD14+ EPCs and the Retina

Data support that CD14+ and CD34+ cells participate in neovascularization, are affected by the disease state of the retina, and are likely to participate in both normal homeostasis and pathology in retinal vasculature. The unique responses of CD34+ and CD14+ cells may be dependent on distinctive VEGFR-1/VEGFR-2 interactions and further modified by exposure to SDF-1 acting through its receptor CXCR4. Our and other studies suggest that the VEGFR-1/VEGFR-2 interaction may affect the angiogenic phenotype. We have already demonstrated that in the most hypoxic regions of the retina, the new vascular tufts are composed exclusively of EPCs; however, we have not characterized the EPC population forming these tufts.

CD14+ cells are a heterogeneous class of progenitors that can generate dendritic cells, macrophages, fibroblasts and endothelial cells as part of vascular maintenance. CD34+ cells can become CD14+ cells, but CD34+ cells usually become endothelial cells without transitioning through a CD14+ phenotype. CD34+ cells can assist the CD14+ cells in acquiring full endothelial function (fig. 2) [108, 177]. In the context of repair and maintenance of the retinal vasculature, if any ischemia or vascular injury occurs in a nondiabetic individual, CD34+ cells would quickly be recruited to the ischemic/injured retinal vasculature to promote repair of any injured endothelium. In diabetes, this does not occur, and acellular capillaries, extracellular matrix tubes with no cellular components, develop instead. In this context, we also postulate that in the event of proper repair, i.e. re-endothelialization and re-perfusion of ischemic retina by CD34+ cells, minimal CD14+ cell contribution would occur. The CD14+ response to vascular repair is predominately initiated when the CD34+ cell response is impaired.

Numerous studies demonstrate that CD14+ cells can differentiate into endotheli- al-like cells [173, 178–182] and participate in

neovascularization in experimental models [179, 182]. These studies suggest that these cells need priming to differentiate into endothelial cells and promote vascular growth. This should not be surprising as monocytes require activation to perform virtually every function with which they are associated [108]. Diabetes is associated with inflammation implicated in the pathogenesis of macrovascular complications [183]. There is a growing body of evidence that the ability of BM-derived cells to promote vascular growth is

altered by diabetes, although exactly which BM cells are impaired and the precise nature of the impairment remains unknown.

In diabetic retinopathy, monocytes contribute to capillary occlusion and nonperfusion [184]. Leukostasis of circulating monocytes promotes endothelial apoptosis [184, 185]. The contribution of monocytes to pathological retinal neovascularization, however, has not been studied. Circulating CD14+ monocytes change towards an increased inflammatory phenotype