In adults, the vasculature is normally quiescent, due to the dominant influence of endogenous angiogenesis inhibitors over angiogenic stimuli. However, blood vessels in adults retain the capacity for brisk initiation of angiogenesis, the growth of new vessels from pre-existing vessels, during tissue repair and in numerous diseases, including inflammation and cancer. Because of the role of angiogenesis in tumor growth, many new cancer therapies are being conducted against tumor angiogenesis. It is thought that these anti-angiogenic therapies destroy the tumor vessels, thereby depriving the tumor of oxygen and nutrients. Therefore, a better understanding of the molecular mechanisms in the process of sprouting angiogenesis may lead to more effective therapies not only for cancer but also for diseases involving abnormal vasculature. It is widely believed that after birth, endothelial cells (EC) in new blood vessels are derived from resident EC of pre-existing vessels. However, evidence is now emerging that cells derived from the bone marrow may also contribute to postnatal angiogenesis. Most studies have focused initially on the contribution of endothelial progenitor cells in this process. However, we have proposed a concept in which cells of the hematopoietic lineage are mobilized and then entrapped in peripheral tissues, where they function as accessory cells that promote the sprouting of resident EC by releasing angiogenic signals. Most recently we found that hematopoietic cells play major roles in tumor angiogenesis by initiating sprouting angiogenesis and also in maturation of blood vessels in the fibrous cap of tumors. Therefore, manipulating these entrapment signals may offer therapeutic opportunities to stimulate or inhibit angiogenesis. (Cancer Sci 2006; 97: 568–574)
Vascular development consists of two different steps, vasculogenesis and angiogenesis. Vasculogenesis is the process by which endothelial precursors, the angioblasts, are committed from mesodermal cells and form a primitive vascular plexus and larger organized vessels in the embryo. In contrast, angiogenesis involves vascular growth and maturation by sprouting and remodeling of pre-existing vessels.(1) In both processes, bidirectional signaling between endothelial cells (EC) and the surrounding mesenchymal cells is critical.(2) Several molecules have been isolated that regulate the processes of vasculogenesis and angiogenesis, and are involved in maintaining the integrity of vessels by recruitment and formation of the periendothelial layer or by mediating interactions between arteries and veins.(2–5) Among them, two receptor tyrosine kinase subfamilies are characterized by their largely endothelial-specific expression. One family includes Flt-1/VEGFR1, Flk-1/KDR/VEGFR2 and Flt-4/VEGFR3, all of which are members of the vascular endothelial growth factor (VEGF) receptor family. The critical roles of Flt-1, Flk-1 and Flt-4, as well as those of VEGF, have been demonstrated by analysis of genetically engineered mice mutants.(6–9) The other family of angiogenesis-related molecules includes tyrosine kinase with Ig and EGF homology domain 1 (TIE1)/TIE and TIE2/tunica interna endothelial cell kinase (TEK). The onset of embryonic expression of these receptors seems to follow that of the VEGF receptors (VEGFR).(10) Targeted mutation of TIE1 and TIE2 demonstrated that these receptors, like VEGFR, play a critical role in embryonic vascular formation.(11–13) Embryos deficient in TIE2 or TIE1 fail to develop structural integrity of the vasculature, resulting in hemorrhage at E9.5 and 13.5. Compared with the early defect in vasculogenesis seen in VEGF or VEGFR mutant embryos, mice lacking TIE1 or TIE2 exhibit defects at a later stage in angiogenesis and vascular remodeling as well as in vascular integrity.
Blood vessels constructed by vasculogenesis are immature. For the maturation of vessel structure, mural cells (MC), such as pericytes or smooth muscle cells, attach to EC. This association between EC and MC is induced initially by platelet-derived growth factor (PDGF)-BB or PDGF-DD produced from EC, which promote recruitment of MC to EC. Subsequently, adhesion between EC and MC is promoted by angiopoietin-1 (Ang-1), a ligand for TIE2 on EC, constitutively produced by MC.(13–17) Therefore, under normoxia, blood vessels are stable structurally (Fig. 1A). Once hypoxia occurs, Ang-2 production is usually upregulated in EC under the transcriptional control of hypoxia inducible factor-1α.(18) Ang-2 is an antagonist of Ang-1 and inhibits the binding of Ang-1 to TIE2. Therefore, under tissue hypoxia, MC dissociate from EC upon inactivation of TIE2, which may lead to migration of EC into the region where new vessel formation is required (Fig. 1B). Whereas this sequence is currently widely accepted, the mechanism whereby EC sense and migrate into the restricted region has not been clarified. In the process of angiogenesis, we have found that hematopoietic stem cells (HSC) and hematopoietic cells (HC) induce construction of the network structure of blood vessels(19) and stabilize the new vessels (Y. Yamada and N. Takakura, unpublished data, 2005). In this article, we present an overview of the functions and contributions of HSC and HC in angiogenesis.
Link between hematopoiesis and angiogenesis
Hematopoiesis is closely linked to angiogenesis as HSC and EC have common progenitors, the hemangioblasts, and interact with each other.(20) Definitive HSC adhere closely to EC at several sites in the embryo, including the yolk sac,(21) omphalomesenteric and vitelline arteries, and the dorsal aorta.(16,20–24) In addition, some stromal cell lines that are able to support hematopoiesis have been characterized as EC.(25) These observations suggest a close interaction between hematopoiesis and vascular development. This relationship is evolutionarily conserved. Several primitive species, such as earthworms and fruit flies, have a rudimentary network of blood channels that are not lined by EC but rather are lined by hemocytes, a cell type belonging to the hematopoietic lineage.(26,27) So far, it is difficult to probe the crosstalk mechanism between EC and HC in vivo. To address this issue, we established a culture system that supports angiogenesis and hematopoiesis using the para-aortic splanchnopleural (P-Sp) region (the pre-aorta-gonad-mesonephros region), a site where definitive HSC are committed from hemang-ioblasts.(16,23,24) When P-Sp explants from the E9.5 embryos were cultured on OP9 stromal cells,(16) platelet and endothelial cell adhesion molecule (PECAM-1; CD31)+ EC formed a sheet-like structure (vascular bed) and subsequently formed a network in the periphery of the endothelial sheet (Fig. 2A,a). The development of HC was observed in this culture system. HSC, which were observed initially in the peripheral edge of the vascular bed, migrated into the vascular network area and proliferated (Fig, 2A,b). These observations suggest that hematopoiesis and angiogenesis (vascular network) may occur concurrently and we hypothesized that HC may regulate angiogenesis.
Acute myeloid leukemia 1 (AML1)-deficient mice provide a tool to analyze the interaction between hematopoiesis and angiogenesis. Disruption of AML1 leads to failure in the development of definitive hematopoiesis and lethality at E12.5.(28,29) Mutant embryos exhibit hemorrhage in the ventricles of the central nervous system, in the vertebral canal, within the pericardial space, and in the peritoneal cavity. AML1-deficient mice(30) showed that extensive vascular branching and remodeling into large and small vessels occurred normally in the head region at E11.5, as observed in wild-type (Wt) animals. However, the number of small capillaries sprouting from the anterior cardinal vein in mutant embryos was fewer than that observed in Wt. In mutant embryos, less branching of capillaries was observed in vessels of the pericardium and in the vitelline artery of the yolk sac. Strikingly, severe defective angiogenesis was observed in the fetal liver of mutant embryos (N. Takakura, unpublished data, 2000). In mice, liver development occurs at E10.5 and, subsequently, HSC migrate into the fetal liver, and expand and differentiate into various hematopoietic lineages in the fetal liver. Hematopoiesis changes into the definitive type in the fetal liver instead of the primitive type that develops in the yolk sac. Within this time frame, the very fine capillary structure that is the main environment of hematopoiesis in the fetal liver should be established. From the phenotype of AML1-deficient embryos, it can be deduced that HC are the major source for induction of very fine capillary structure in the fetal liver.
To analyze the interaction between HC and EC, we studied the development of EC in P-Sp cultures from AML1 mutant embryos. As depicted in Fig. 2B, in the absence of HC (AML1−/–), EC formed a sheet-like structure but did not form network-like structures. In contrast, explants from Wt embryos developed vascular beds and networks. To test the hypothesis that HC promote angiogenesis, a HSC-enriched population of cells from the bone marrow of normal mice was added to P-Sp cultures of AML1 mutant embryos. As expected, the addition of HSC rescued defective angiogenesis in AML1 mutant embryos. (AML1−/– + HSC).
Ang-1 produced from HSC regulates angiogenesis
As the results from the P-Sp culture system suggested that extrinsic signals from HC promote angiogenesis, we searched for factors that could mediate this process. During embryogenesis, CD45+c-Kit+CD34+ cells are defined as HSC.(31) We found that these HSC expressed Ang-1, which is essential for angiogenesis, and that mature HC in embryos did not express Ang-1. As mentioned above, Ang-1 produced from MC promotes cell adhesion between EC and MC. It is well known that another action of Ang-1 is chemo-attraction for EC.(32) Moreover, defective angiogenesis in the anterior cardinal vein and the pericardium observed in AML1 mutant embryos was similar to those observed in Ang-1 mutant embryos.(19) These findings suggested that Ang-1 expressed on the HSC-enriched populations promotes angiogenesis by promoting chemotaxis of EC. To test this hypothesis, we added HSC from E10.5 Ang-1 mutants to P-Sp cultures of AML1 mutant embryos. HSC-enriched cells from Ang-1 mutant embryos could not rescue the defective network formation of EC of an AML1 mutant culture (AML1−/– + HSC [Ang-1−/–]; Fig. 2B). Furthermore, Ang-1 also rescued the network formation of EC of an AML1 mutant culture (AML1−/– + Ang-1; Fig. 2B). Taken together, we confirmed that Ang-1 produced from HSC is important for the network formation of EC, at least in our P-Sp culture system.
HSC migrate into avascular areas and induce sprouting of EC by releasing Ang-1
To clarify how HSC participate in angiogenesis in vivo, we evaluated whether they are present in the head region where severe angiogenic defects are observed in the AML1 mutants. As expected, HSC were located near the vessels and EC seemed to migrate toward HSC in the neuronal layer. Based on the localization of HSC and EC, we hypothesized that HSC may promote migration of EC and network formation in the avascular area. By using microchemotaxis chambers, we confirmed that Ang-1 from HSC directed the migration of TIE2+ EC. Therefore, we concluded that HSC migrate into avascular areas, produce Ang-1, and thus promote migration of EC and network-like capillary formation (Fig. 3). This scenario may not be applicable to all situations in which angiogenesis takes place, because AML1 mutant embryos show normal angiogenesis except in tissues such as the anterior cardinal vein, capillaries in the neuronal layer, the pericardium, the peritoneum, the fetal liver and the yolk sac. However, for the migration of EC in the correct direction, HSC plays an essential role by working as a guidepost for EC and marking a route from pre-existing vessels to the ischemic region.
Trap for HSC in ischemic foci
Although our findings indicate that EC migrate toward HSC producing Ang-1, the fundamental mechanism of HSC migration from the intraluminal cavity into parenchyma at a precise point in a vessel is unclear. Recently, the identity of signals in the peripheral tissues of adults that retain proangiogenic HC mobilized from the bone marrow was reported.(33) In that report, the authors proposed that VEGF recruits a heterogeneous mix of myeloid cells and HSC from the bone marrow to the blood, and chemokine stromal-derived factor-1 (SDF-1)/CXC ligand (CXCL)12 traps and correctly positions a subpopulation of circulating proangiogenic HC around the growing vessels in tissues (Fig. 4). Thus, as observed in adults, entrapment of HSC may be regulated by some chemokines such as SDF-1 (Fig. 3).
In terms of the transmigration of HSC from the intraluminal cavity of blood vessels to parenchyma, peripheral CD34+ hematopoietic progenitors that express high levels of matrix metalloproteinase (MMP)-2 and MMP-9 may be involved in this mechanism.(34,35) Moreover, a recent study showed that mast cells in the skin release chymase, which activates pro-MMP-9 and is associated with stimulation of angiogenesis during squamous epithelial cell carcinogenesis.(36) Our preliminary data also show that embryonic HSC (CD45+c-Kit+CD34+ cells) express MMP-9 prominently (data not shown). In addition, these HSC express TIE2 and adhere to fibronectin (FN) following stimulation by Ang-1.(16) Taken together, these results suggest that HSC adhere to FN on EC near the ischemic region, digest the matrix, and transmigrate through the basement membrane of capillary EC. Therefore, we hypothesize that the production of FN on the intraluminal surface of EC is the initial step in the migration of HSC. Usually FN is not observed in the intraluminal part of EC in adults. However, FN is expressed on the luminal surface of vessels in some parts of the embryo and tumor tissue at the time of angiogenesis. An analysis of the molecular cues promoting high expression of FN on EC at the lumen of blood vessels may be required to better understand how vessel sprouting is initiated.
Hematopoietic cells regulate tumor angiogenesis
As mentioned above, we found that HSC function in a unique way in promoting angiogenesis. Recently, another cell of hematopoietic lineage was reported to induce angiogenesis. Myeloid progenitor cells expressing VEGFR-1 undergo extravasation and position around vessels in ischemic foci where they stimulate angiogenesis, also through release of angiogenic factors.(37,38) Tumor-associated macrophages, monocytes expressing TIE2, mast cells, T lymphocytes, neutrophils and platelets have similar activities.(39) We also reported that neuropilin-1 (NP-1) expressed on HC, including T and B lymphocytes, myeloid cells, erythroid cells and HSC, induces angiogenesis. NP-1 is known to be expressed on EC and works as a receptor for VEGF165. NP-1 doesn't contain a kinase domain or a binding domain for adapter proteins. However NP-1 acts as a coreceptor that enhances the function of VEGF165 through VEGFR-2. We found that NP-1 on HC binds serum VEGF and stimulates VEGFR-2/Flk-1 on EC exogenously, resulting in brisk proliferation of EC(40,41) (Fig. 3). Based on these findings, we investigated the association of HC in tumor angiogenesis. As indicated in Fig. 5, CD45+ HC were found to be located around blood vessels in tumor and blood vessel formation, and tumor formation was enhanced as HC migrated into the tumor. In our studies of the mouse tumor model, most CD45+ HC accumulating in tumors were CD11b/Mac-1-positive monocytes and macrophage lineage cells. However, CD4+ and CD8+ T lymphocytes, B220+ B lymphocytes, Gr-1+ granulocytes, c-Kit+ mast cells and c-Kit+Sca-1+ HSC were also located in tumors. We found that when such migration of HC was inhibited by bone marrow suppression by anti-c-Kit neutralizing antibody, the initiation of sprouting angiogenesis in tumors was suppressed.(42)
As reviewed in this article, HC are closely linked with physiological and pathological angiogenesis. Therefore, regulation of HC function in cancer may be a promising approach for inhibition of tumor angiogenesis. Indeed, in our recent report, we induced bone marrow suppression by administering an anti-c-Kit neutralizing antibody, and the resultant leukocytopenia inhibited sprouting angiogenesis from pre-existing vessels in colon26 tumors.(42) Thus, suppression of HC migration into tumors is effective for inhibition of tumor angiogenesis; however, systemic leukocytepenia is not desirable as it leads to a reduced ability to fight infection. Therefore, development of a new strategy is required for the suppression of HC migration locally around tumor areas, possibly by shutting down expression of the chemoattractants that induce migration of HC.
Recent studies have suggested that inflammation is closely associated with the process of tumor angiogenesis; many disease conditions such as chronic infection and chronic irritation that foster an inflammatory environment appear to contribute to development of cancers.(43) Evidently, the tumor environment is largely orchestrated by inflammatory cells such as neutrophils, macrophages and monocytes, lymphocytes, and mast cells, eosinophils and basophils. We have shown that HSC also contribute to tumor angiogenesis. Moreover, platelets may also contribute to angiogenesis. Therefore, most HC populations must be associating with tumor angiogenesis. In addition, tumor cells have co-opted some of the signaling molecules of the innate immune system, such as selectins, chemokines and their receptors, for invasion, migration and metastasis. Taken together, in angiogenesis and tumor metastasis involving HC, these insights are fostering new anti-inflammatory therapeutic approaches to cancer development and microenvironment regulation.
We thank our collaborators, Dr G. D. Yancopoulos (Regeneron Pharmaceuticals, Tarrytown, NY, USA), and Dr M. Satake and Dr T. Watanabe (Institute of Development, Aging, Cancer, Tohoku University, Sendai, Japan), and thank Ms M. Sato, Ms K. Ishida and Ms Y. Shimizu for technical assistance. This work was supported by grants-in-aid for scientific research from the Ministry of Education, Science, and Culture of Japan.