Eur J Clin Invest 2012
Background The role of bone marrow–derived cells in stimulating angiogenesis, vascular repair or remodelling has been well established, but the nature of the circulating angiogenic cells is still controversial.
Design The existing literature on different cell types that contribute to angiogenesis in multiple pathologies, most notably ischaemic and tumour angiogenesis, is reviewed, with a focus on subtypes of angiogenic mononuclear cells and their local recruitment and activation.
Results A large number of different cells of myeloid origin support angiogenesis without incorporating permanently into the newly formed vessel, which distinguishes these circulating angiogenic cells (CAC) from endothelial progenitor cells (EPC). Although CAC frequently express individual endothelial markers, they all share multiple characteristics of monocytes and only express a limited set of discriminative surface markers in the circulation. When cultured ex vivo, or surrounding the angiogenic vessel in vivo, however, many of them acquire similar additional markers, making their discrimination in situ difficult.
Conclusion Different subsets of monocytes show angiogenic properties, but the distinct microenvironment, in vitro or in vivo, is needed for the development of their pro-angiogenic function.
Over the past decade, numerous reports have cited bone marrow–derived circulating cells to play a major role in the establishment of angiogenesis. Indeed, the ability to induce the growth of new blood vessels from pre-existing vessels, defining angiogenesis, has been particularly attributed to subpopulations of mononuclear cells existing in the adult bone marrow and circulating in peripheral blood. The important roles of such circulating angiogenic cells (CAC), sometimes mistakenly referred to as endothelial progenitor cells (EPC) in tumour and ischaemia-induced angiogenesis have been described (reviewed in ref. [1,2]). Therefore, they have been regarded as promising candidates for both therapeutic angiogenesis modulation in multiple diseases such as acute myocardial infarction, atherosclerosis and ischaemic retinopathies as well as tumour angiogenesis suppression . CAC were originally regarded as EPC, a term coined by Asahara et al.  after their first isolation and culture of naïve circulating blood cells enriched with the endothelial/stem marker CD34+ and expressing VEGFR2 . Since then, numerous reports have indicated, however, that these CD34+ cells are in fact of myeloid origin and do not present endothelial precursors , but rather act in a paracrine manner (reviewed in ref ). Controversy still continues regarding the true nature of CAC as pro-angiogenic monocytes  in particular because myelomonocytic cells share several characteristics with endothelial cells  (Table 1). Moreover, monocytes are widely described in the literature to be cells, which can develop endothelial-like characteristics  improving postischaemic revascularization in contrast to mature endothelial cell injection . Current consensus defines 2 major classes of angiogenic cells after in vitro culture, the (CAC, early EPC), which appear after a few days (early) in culture after adherence to fibronectin, and the endothelial colony forming cells (ECFC, or ‘outgrowth endothelial cells’ (OEC) or late EPC) that appear after 2 weeks (late) of culture on collagen . A third type of pro-angiogenic cells cultured in vitro has been suggested to be the CFU-Hill EPC  from re-plated peripheral blood mononuclear cells (PBMC), but these have been shown to be composed of mixed colonies of monocytes and T-lymphocytes . In conclusion, based on marker selection (Table 1), it is now generally accepted that the CAC are of myeloid origin and therefore comprise true angiogenic monocytes. Yet, the exact role of monocytes/macrophages and their subpopulations in neovascularization, vascular repair and reperfusion have been debated in recent experimental data . This review will discuss the role of monocytes contributing in angiogenesis and other types of vascular remodelling and repair and will discuss the existence of a monocytic cell type capable of promoting angiogenesis without incorporation in the vasculature.
Bone marrow–derived cells in angiogenesis
Studies in mouse and human have shown that bone marrow–derived cells (BMDCs) are first recruited from bone marrow and then circulate to sites of vascular injury  to play an essential role in tumour angiogenesis . Thus, BMDCs, as a provider of hematopoietic stem and progenitor cells, and in particular myeloid progenitors, play an important role in angiogenesis as source of several potent actors of this process. A particular sequence of events determines neovascularization induced by HSC-derived CAC, starting with the recruitment of a certain type of cells from the bone marrow to the circulation in response to hypoxia-, stress-, damage-related signals or pro-angiogenic factors like VEGF, SDF-1 and MCP-1 . However, a recent study underlines the importance of common myeloid progenitors and also a granulocyte progenitor subset in developing pro-angiogenic characteristics. After being conditioned in vitro, the subsequent myeloid progenitor-derived pro-angiogenic cells showed improved in vivo action compared to HSC-derived cells . Likewise, certain types of circulating cells with a myeloid origin, which respond to these injury-related factors, may be recruited and participate in vascular repair process . Once arrived in the injured or calling site, the specific microenvironment may have the ability to directly act on bone marrow–derived pro-angiogenic cells by retaining and conditioning them [2,18]. In nonpathological conditions, Grunewald et al.  elegantly showed that VEGF-overexpressing tissues recruited circulating BMDCs and next retained them through the local induction of SDF1α, in close proximity to angiogenic vessels. Ziegelhoeffer et al.  demonstrated that BMDCs do not incorporate into adult growing vasculature but possibly function as supporting cells. Also during tumour angiogenesis, the BMDC are commonly of myeloid origin, expressing GR-1, F4/80 and CD11b, but not endothelial markers stressing a supportive role over transdifferentiation . Similarly, during endothelial repair after vascular injury, BMDC do not seem to be involved in re-endothelialization of the wound . Of note, a recent study using a selected population of BM-CD31+ cells showed higher angiogenic and neo-vasculogenic capacity compared to CD31− cells, in vitro and in mice, after the induction of hindlimb ischaemia . Nevertheless, the frequency of incorporation or transdifferentiation of these cells still remained low (3,3% of total endothelial cells) . Interestingly, monocyte surface marker CD31 frequently used in CAC/EPC characterization is shared with endothelial cells, because it participates in endothelial transmigration.
Specific role of monocytes in angiogenesis and arteriogenesis
In animal studies, monocytes were shown to accumulate within the vessel wall of growing collaterals, which suggests that monocytes contribute to arteriogenesis. Indeed, monocytes adhere to the vascular wall during both arteriogenesis and tumour angiogenesis [22,23]. In vivo and ex vivo studies with transgenic mouse models have shown that bone marrow–derived myeloid cells like monocytes and macrophages have important roles in the regulation of blood vessel formation in tumours . Interestingly, Rohde et al.  showed that recruited monocytes do not form vascular networks, but instead phenotypically mimic endothelial cells. Of note, monocytes/macrophages share phenotypic features and surface markers with endothelial cells , and they can obtain endothelial-like properties after angiogenic stimulation as well as function as angioblasts [25,26]. Because monocytes and macrophages express endothelial cell markers at a later stage after recruitment in vivo and the same phenomenon is observed during generation of CAC in vitro, the question arises regarding the exact nature of these cells. In a nonpathological model of parabiosis, however, Kim et al.  conclusively showed that monocytes expressing both CD31 and F4/80 are recruited to ischaemic sites to contribute to angiogenesis, while they do not express specific endothelial or EPC markers. In vivo injection of a mixed population of in vitro differentiated myeloid CAC and endothelial progenitor ECFC synergistically improves neovascularization , which may unify the discrepancies on the in vivo action of these different populations and their ability to incorporate into the injured endothelium or the neovessels. Indeed, the various cytokines and pro-angiogenic factors differentially produced by ECFC (like MMP2) and CAC (like MMP9 and IL-8) may account for this synergy during neovascularization, and the conditioned medium of CAC may indeed induce ECFC proliferation in vitro  and their ability to incorporate into the injured endothelium or the neovessels.
CD14+ monocytes and the role of in vitro selection and differentiation
In hindlimb ischaemia models, freshly isolated CD14+ monocytes isolated from the bone marrow or peripheral blood did not show any improvement of perfusion in vivo [28,29] except in pathological conditions such as after injection in diabetic mice . However, selection and in vitro conditioning of these same cells markedly improve their pro-angiogenic capacity. The existence of a population of monocytes isolated from peripheral blood, which can differentiate into several mesenchymal cell types after 1-week culture on fibronectin (i.e. CAC) , further underlines the importance of the microenvironment, like matrix composition. These monocyte-derived multi/pluripotent cells (MOMCs) were not observed when CD14+ monocytes were cultured on tissue culture plastic, thus need fibronectin to become spindle-shaped and to develop a unique molecular phenotype characterized as CD14+/CD45+/CD34+ . Indeed, in the presence of pro-angiogenic factors, these pluripotent monocytes differentiate into endothelial-like cells. They can also form endothelial structures in Matrigel in response to angiogenic stimuli and take part of the vasculogenesis and angiogenesis in xenograft tumour . Interestingly, the conditioning medium of CD14− cells cultured on fibronectin is sufficient to restore the ability of a purified CD14+ culture to form MOMCs. Over the past years, it has become clear that CD14+ cells can generate CAC that display pro-revascularizing characteristics. In contrast, CD14− cells would lead to the development of ECFC after long-term culture . Urbich and colleagues  showed that fibronectin-adherent cells derived from CD14+ or CD14− mononuclear cells incubated on fibronectin-coated dishes in endothelial growth medium display endothelial marker proteins and markedly improve postischaemic revascularization in nude mice in contrast to freshly isolated CD14− or CD14+ mononuclear cells. Likewise, peripheral blood-derived CD14+ monocytes activated by the monocyte chemoattractant MCP-1 in vivo or in vitro adhered to the injured endothelium and differentiated into endothelial-like cells by losing their hematopoietic markers . Even if the freshly isolated CD14+ monocytes were reported to share surface markers with human microvascular endothelial cells, they did not express the characteristic endothelial markers vWF and VEcadherin . However, Schmeisser et al.  demonstrated that CD14+ monocytes expressing both VEGFR2 and macrophage markers formed tubular- and cord-like structures under angiogenic stimulation in vitro. Interestingly, it has been recently shown that the secretion of Ang1 by monocytes upon contact with endothelial cells stimulates endothelial Tie2 expression and promotes endothelial survival in vitro . Furthermore, Ang1 secreted by monocytes could stimulate other Tie2-expressing cells to induce vasculogenesis . These studies raise the question whether the pro-angiogenic microenvironment is sufficient to model all monocytes into CAC or whether intrinsic pro-angiogenic ability is present in a specific monocyte subset.
Different subsets of circulating monocytes
The circulating monocyte population may be divided into three phenotypically different subsets . In human, the heterogeneity in CD14 and CD16 expression profiles is the base of the identification of the monocytes subpopulations. The two primary subpopulations originally described are the inflammatory/classical (CD14++CD16−) (more than 85% of monocytes) and the nonclassical/resident monocytes (CD14+CD16++). These subsets are believed to correspond to Ly6Chi and Ly6Clo monocytes in mice, respectively. The intermediate phenotype (CD14++CD16+) has only recently been defined because of its overlap with the 2 other subsets. Although originally the nonclassical monocytes, which patrol the endothelium , were thought to be the most potent pro-angiogenic subset, recent analysis of gene and protein expression profiles of the intermediate-type subpopulation highlighted strong pro-angiogenic/CAC characteristics especially in this subset . FACS sorting of the three different subpopulations showed that the intermediate monocytes indeed expressed higher surface protein levels of Tie2, VEGFR2 and endoglin (CD105) compared to the other two subsets. However, nonclassical monocytes showed the highest CD31 surface expression as well as enhanced gene expression for adhesion molecules. Shantila et al.  reported that the CD14+ CD16+ CCR2+ intermediate subset is the monocyte population that expresses the highest levels of Tie2, VEGFR2, VEGFR1 and CXCR4. Nevertheless, in vivo data from a mouse model of postischaemic hindlimb revascularization underlined different conclusions. Although nonclassical monocytes (Ly6Clo or 7/4lo in mice) were suggested to develop pro-angiogenic activity as they produced higher levels of VEGF in the mouse ischaemic myocardium , the adoptive transfer of selected BM-derived inflammatory monocytes (CXCR3lo-Gr1+) significantly increased postischaemic perfusion, whereas nonclassical monocyte (CXCR3hi-Gr1−) transfer did not . A similar study showed that transfer of the nonclassical subset only improved postischaemic arteriogenesis, not angiogenesis, as assessed by an increased angiographic score based on collateral vessel formation . Of note, in these in vivo studies, the specific action of intermediate monocytes was not yet assessed. However, the fact that the recruitment and infiltration of both classical and nonclassical monocytes subsets to the site of ischaemia are dependent on CCL2/CCR2 signalling [17,43] may account for a putative role of intermediate monocytes, as they do express the MCP1 receptor CCR2, in contrast to nonclassical monocytes .
Macrophages as pro-angiogenic tissue monocytes
Once emigrated to the target tissue, circulating monocytes adhere to and differentiate into macrophages. Depending on the microenvironment, macrophages are polarized into distinct subtypes, which are in vitro defined as the pro-inflammatory, anti-bactericidal M1 developing in response to inflammatory conditions (LPS and IFN-gamma) or the alternative/repair M2-type developing in response to anti-inflammatory conditions (IL4 or IL10). M2-like macrophages have been shown to develop pro-angiogenic functions in vitro and to promote vascular remodelling particularly in tissue repair [44,45]. Interestingly, recent microarray expression analysis studies demonstrated the close genetic and immunophenotypic relationships between CAC and M2-like macrophages, leading to a novel denomination of these cells as myeloid angiogenic cells (MACs) [46,47]. In the same way, VEGF-overexpressing macrophages develop CAC features in vitro and improve postischaemic cardiac neovascularization in vivo [48,49]. However, in in vivo models of hindlimb ischaemia, their role seems to be preponderant in collateral formation, more than angiogenesis . While inflammatory monocytes are recruited and differentiate into macrophages in inflammatory lesions, resident macrophages by contrast seem to proliferate and promote arteriogenesis locally, even if originating from circulating monocytes . The manipulation of blood monocyte concentration also strongly affects postischaemic collateral formation . Still, mouse macrophages (F4/80+) have been linked to angiogenesis in vivo. Direct evidence was brought by Anghelina et al.  who showed that F4/80+ monocytes/macrophages infiltrated Matrigel in vitro and in vivo and created so-called cell columns, which support the active development of new vessels possibly by precursor cells. Especially the production of matrix metalloproteinases (MMP) has been suggested to be crucial in both the de novo vascularization in a Matrigel plug assay  and in tumour angiogenesis . The adoptive transfer of selected BM-derived inflammatory monocytes significantly increased postischaemic hindlimb perfusion, despite the fact that the transferred macrophages (F4/80+) contributed only a small proportion (1%) of total macrophages in the ischaemic muscle [43,54]. They were believed to locally produce MCP1 for a secondary recruitment of monocytes thus leading to maximal angiogenesis . In contrast, monocytes with a M2-like phenotype after ischaemia-conditioning or when isolated from SHIP−/− mice (inducing a preferential M2-like polarization) seem to be ineffective in promoting postischaemic perfusion .
In the cancer field, TAM, originating from circulating monocytes in the tumour microenvironment and attracted by locally produced chemotactic factors, express many genes that are characteristic of the M2-like phenotype . Signals produced in the tumour microenvironment are believed to stimulate the pro-angiogenic function of these cells. They indeed have been shown to accumulate in VEGF-releasing tumour hypoxic areas . Evidence for the importance of TAM in the regulation of tumour angiogenesis has come from a study by Lin et al. . They used a transgenic mouse model of breast cancer, in which the angiogenic switch and progression to malignancy were regulated by infiltrated macrophages in the mammary tumour. The fact that TAM are capable of inducing the formation of a vascular network suggests that they may produce local factors to promote tumour angiogenesis . TAM have been shown to express a number of angiogenesis modulating factors comparable to those expressed by Tie2-expressing monocytes (TEM) and CD14+ monocytes in vitro, . Moreover, TAM-conditioned medium may induce neovascularization in vivo and endothelial proliferation in vitro . Thus, the specific microenvironment of tumours highly enriched in cytokines and pro-angiogenic growth factors may directly act on monocyte/macrophages to modify their polarization towards an M2-like phenotype, which will further amplify the angiogenic capacity of the tumour. Among pro-angiogenic factors, angiopoietins (Ang) that are encountered at high levels in tumour sites may specifically activate a certain type of monocyte/macrophages characterized by the expression of the Ang1 and Ang2 receptor Tie2.
TIE-2 expressing monocytes/macrophages (TEMs)
Cells that express the angiopoietin receptor Tie2 seem major actors of tumour angiogenesis [58,59]. TEM were first identified in the mouse peripheral blood and were found in close proximity to nascent tumour vessels. They were not incorporated in the vasculature, however, suggesting a paracrine stimulation of tumour blood vessel growth [24,58,59,60,61]. Their surface markers (Tie2+ Sca-1+ CD11b+) distinguished them from the majority of TAM . Comparably, human peripheral blood also contains TEM, which have a pro-angiogenic activity. The co-injection of human glioma cells and CD14+ TIE2+ cells from human peripheral blood in nude mice promoted tumour vascularization, while CD14+ TIE2− cells did not [59–61]. The selective ablation of TEM by activation of a suicide gene significantly induced tumour regression [58,63]. TEM have also been detected, surrounding neovessels during hepatic regeneration in mice , and their role has been recently highlighted in angiogenesis-related endometriosis . Although Tie2+ cells represent only 1–2% of the PBMC population, they can contribute upto 20% of the locally recruited monocyte population [60,64] and belong to the resident/nonclassical subset of monocytes . It should be noted that this study did not yet follow Tie2 expression in the recently defined intermediate monocytes. However, Murdoch et al.  observed that TEM develop phenotypic characteristics of both nonclassical and classical monocytes, although with a clear predominance of the nonclassical phenotype. Both studies converged on low numbers of Tie2-expressing cells in the classical monocyte population. In contrast, TEM were defined by Venneri et al.  as being both CCR2-negative and VEGFR2-negative cells, which discriminates them from the nonclassical/intermediate monocyte population. Indeed, in vitro data showed that resident monocytes that are enriched for Tie2 migrated towards Ang-2, in contrast to classical monocytes . Although Tie2 remains weakly expressed by circulating monocytes, it is upregulated upon their homing to tumours and their differentiation into a subset of perivascular macrophages . The tumour microenvironment may play a direct role in the recruitment and activation of TEM, as activated endothelial cells are a major source of Ang2  and hypoxia stimulates Tie2 expression in vitro . Ang2, and not Ang1, acts as a chemoattractant for TEM and enhances their pro-angiogenic activity in vitro [60,64,66]. Interestingly, the angiogenic potency of Tie2-overexpression on progenitor cells was studied through the use of a cartilage oligomeric matrix protein COMP-Ang1 (Ang1 recombinant chimera). Kim et al.  observed that in vitro incubation with COMP-Ang1 was effective in priming G-CSF-mobilized peripheral blood stem cells (mobPBSC), but not PBMC, to overexpress Tie2. This increased expression was accompanied by an increase in adhesion and pro-angiogenic functions of the primed mobPBSC in vitro as well as in vivo in two different models of ischaemia-induced revascularization. By selectively enhancing Tie2 expression, angiopoietins secreted in the microenvironment may profoundly act on a certain population of cells, that is, the Tie2-expressing circulating cells, (either TEM or other myeloid progenitors) and stimulate them towards an active and effective pro-angiogenic phenotype.
Numerous studies have provided evidence that the large majority of bone marrow–derived angiogenic cells are of myeloid origin and that their monocyte/macrophage phenotype is essential for their full revascularization capacity. There is solid evidence that circulating monocytes may acquire endothelial-like properties and express common surface markers when cultured in vitro under angiogenic conditions, but they remain quite distinct from EPC (Table 2). Understanding the mechanisms and functional role of monocytes in angiogenesis and their potency to differentiate into angiogenic cells is crucial to potentially identify a single angiogenic monocyte subset. Noteworthy is the fact that the presence of endothelial cell markers on these circulating myeloid cells is not a prerequisite for the cells to differentiate into functional angiogenic cells or to participate in angiogenic, arteriogenic or vascular repair processes in vivo. Instead, the pro-angiogenic conditions artificially applied in vitro or encountered in vivo at the site of ischaemia or in the tumour microenvironment may markedly affect recruitment, differentiation and activation of angiogenic monocytes. Various subsets of myeloid cells would have the capacity to respond to the calling pro-angiogenic factors. However, depending on the particular microenvironment (tumour vs. ischaemic hindlimb or cornea), only some specific cells are retained and stimulated to actively take part in the angiogenic process. Information on the specific monocyte subsets involved and the identity of the local programming cues is still scarce and lacks a clear conceptual framework. In conclusion, although different subsets of monocytes show angiogenic properties, they need differentiation in response to the microenvironment, in vitro or in vivo, to actively develop their pro-angiogenic function.
|Selected population||Proangiogenic conditioning||In vitro evidence||In vivo evidence||Matrigel||Ischaemia||Tumour|
|BMDCs||Unselected||−||Yes [15,18]||Yes ||Yes |
|HSC||+||No ||No ||No |
|Myeloid/granulocyte progenitor||+||Yes ||Yes ||Yes |
|MobPBSC||+||Yes ||Yes ||Yes |
|CD31+||−||Yes ||Yes ||Yes |
|EPCs||HSC–CEPC||−||Yes ||Yes |
|ECFC||+||Yes [1,27]||Yes [1,27]||Yes |
|EarlyEPC||+||Yes ||Yes [9,27,28]||Yes [9,27,28]|
|Monocytes||Unselected||+||Yes [7,25]||Yes [7,25,34]||Yes  Yes |
|F4/80+||−||Yes ||Yes [13,23]/Arteriog. No [28,29]/Yes||Yes ||Arteriog.  No [28,29]|
|MOMC||+||Yes ||Yes ||Yes |
|Classical||−||Yes [17,43] |
|Yes [17,43] |
|Nonclassical||−||No [17,43]/Arteriog.||No [17,43]/Arteriog.|
|CD31+||−||Yes ||Yes |
|TE Ms||TIE2+||−||Yes [58,60,62,63]||Yes [58,60,62]|
|Macrophages||Unselected||−||No /Arteriog.||Arteriog. |
|+||Yes [48,49]||Yes [48,49]|
|F4/80+||−||Yes ||Yes ||Yes |
|M1||−||Yes ||Yes |
|M2||−||Yes [44,45]||Yes [44,45]/No/Arteriog.||No /Arteriog.|
|TAMs||−||Yes ||Yes [56,57]||Yes [56,57]|
Department of Molecular Cell Biology and Immunology, VU University Medical Center, 1081 BT Amsterdam, the Netherlands (J. Favre, N. Terborg, A. J. G. Horrevoets).