Role of Bone Marrow-Derived Cells in Angiogenesis: Focus on Macrophages and Pericytes

Abstract

Tumor growth relies on the formation of new blood vessels to receive an adequate supply of oxygen and nutrient. This process is facilitated by both the remodeling of the pre-existing vasculatures and the recruitment of the progenitor/stem cells originated from bone marrow-derived cells (BMDCs). Evidences from both animal studies and human trials have reported that these tumor-associated BMDCs differentiate into a series of stromal cells including macrophages and pericytes, and regulate tumor angiogenesis in various aspects. Macrophages constitute a large portion of the BMDCs infiltrated in the tumor microenvironment, and have been shown to disrupt the balance of pro- and anti-angiogenic signalings by the secretion of various cytokines. Pericytes, mainly derived from the subpopulation of PDGFRβ⁺ BMDCs, can provide both pro-survival signaling and mechanical support to maintain the newly formed endothelium via the direct interactions with endothelial cells. In the current review, we summarize the recruitment mechanisms of BMDC-derived macrophages and pericytes within tumor microenvironment, and also review the contribution of these cells to the different aspects of angiogenesis, with particular emphasis on their therapeutic implications as potential targets for anti-tumor strategies.

Introduction

Tumor is not a solo island, but rather an ensemble performance of various cell types in the tumor microenvironment. Tumor cells play a dominant role with versatile stromal cells which are recruited and activated to support malignant progression. The tumor microenvironment mainly consists of endothelial cells, infiltrated immune cells, pericytes, fibroblasts, and extracellular matrix [1]. It has been well established that in order to grow beyond a certain size, tumors require blood supply to deliver nutrients and oxygen [2]. The tumor vasculature is regarded to derive from both the pre-existing vessels via endothelial cell migration and proliferation (angiogenesis) and recruitment of progenitor cells from bone marrow (vasculogenesis) [3, 4]. Accumulated studies have shown that the dynamic crosstalk between endothelial cells and other stromal cells has a substantial contribution to tumor angiogenesis [5]. Although tumor cells are normally the chief regulators to drive cellular events including angiogenesis, evidences have emerged for the fundamental roles of bone-marrow-derived cells (BMDCs), especially macrophages and pericytes, in this process [6]. Both macrophages and pericytes are recruited to the tumor mass by a broad spectrum of tumor-derived chemoattractants. By interacting with tumor cells, macrophages are stimulated to produce multiple cytokines and matrix-degrading enzymes, which directly regulate the endothelial cell functions and new blood vessel formation [7]. Pericytes are involved in the process of vascular maturation, which provide both structural support and survival signals for endothelial cells, generating mature and functional vasculatures [8].

The essential roles of macrophages and pericytes in tumor progression have been extensively reviewed [1, 6, 7]. Here we present a comprehensive overview about macrophages and pericytes as pivotal players in tumor angiogenesis, along with the regulatory mechanisms underlying their recruitment and activation. A better understanding of their roles may provide valuable insights into therapeutic implications, which will shed new light on the development of efficient anti-cancer therapies.

Macrophage and Angiogenesis

Macrophages represent up to 50 % of the tumor mass and play vital roles in the regulation of angiogenesis and tumor progression. After recruitment and activation, macrophages secret a broad spectrum of growth factors, cytokines, chemokines, and matrix-degrading enzymes, which are directly involved in the endothelial cell function and facilitate endothelial cell migration via extracellular matrix remodeling. Current therapeutics, such as radiation, chemotherapy, and the anti-angiogenic agent bevacizumab, have all shown tumor refractoriness. One underlying mechanism is the enhanced infiltration of macrophages following these treatments, which disrupt the balance of pro- and anti-angiogenic signalings in tumors. Therefore, new therapeutic strategies targeting the recruitment, activation and functions of macrophages will have considerable potential in compensating conventional therapies.

Monocyte Recruitment

Macrophages in tumors, usually termed tumor associated macrophages (TAM), originate from recruited BMDCs. They are initially released from the bone marrow as immature monocytes, circulate in the blood, and extravasate into tumors, where they start to differentiate into TAM. BMDCs recruitment is driven by chemoattractants secreted by both tumor and stromal cells [9].

A number of chemokines (CXC, CC, C, and CX3C) have been shown to participate in BMDCs recruitment via a family of seven-transmembrane G protein–coupled receptors [10, 11]. Among them, C-C chemokine ligand 2 (CCL2, also known as monocyte chemoattractant protein-1, MCP-1), is the most prominent player, which was first described to stimulate the monocyte chemotaxis via its receptor CCR2 in 1983 [12, 13]. Overexpression of CCL2 has been found to be associated with increased monocyte recruitment in various human cancers including ovarian, prostate, lung, breast, esophagus, gastric cancers, melanoma, and gliomas [14–21]. In animal studies, CCL2 has been demonstrated to induce angiogenesis and tumor progression via macrophage recruitment in melanoma, gastric, breast, and prostate tumors [22–26]. Other chemokines involved in monocyte recruitment are CCL3 (macrophage inflammatory protein-1α, MIP-1α), CCL4 (MIP-1β), CCL5 (RANTES), CCL7 (MCP-3), CXCL8 (Interleukin 8, IL-8), and CXCL12 (stromal cell derived factor 1, SDF-1) [27]. The role of CXCL12/CXCR4 axis in monocyte recruitment has been reported in irradiation-induced glioblastoma recurrence [28]. Moreover, certain chemokines, such as CCL2 and CCL5, can activate monocytes and amplify their recruitment by stimulating the production of molecules for monocyte migration. For instance, CCL5 can promote the expression of CCL2, CCL3, CCL4, CXCL8, and the chemokine receptor CCR1 in human monocytes, which in turn accelerates their recruitment [29]. Matrix metalloproteinase-9 (MMP-9) and urokinase-type plasminogen activator receptor (uPAR) are up-regulated by both CCL2 and CCL5 as well, which are responsible for the degradation of basement membrane and extracellular matrix (ECM) components to facilitate monocyte migration [30].

In addition to chemokines, several cytokines such as vascular endothelial growth factor (VEGF), placental growth factor (PlGF), and colony-stimulating factor-1 (CSF-1) have been well documented in recruiting monocytes to tumors [31]. The pro-angiogenic factor VEGF-A has been shown to modulate monocyte recruitment through both vascular endothelial growth factor receptor 1 (VEGFR-1) and VEGFR-2 in orthotropic pancreatic tumors [32, 33]. VEGF-induced skin carcinogenesis also depends on the recruitment and alternative activation of macrophages [34]. PlGF is another important cytokine involved in monocyte recruitment. PlGF-induced macrophage infiltration has been demonstrated to confer resistance to anti-VEGF cancer therapy [35]. PlGF attracts monocytes to tumors via its functional receptor VEGFR-1. The underlying mechanism lies in the induction of tumor necrosis factor alpha (TNF-α) via the activation of nuclear factor of activated T cells 1 (NFAT1) in myelomonocytic cells [36]. Either suppression of NFAT1 or neutralization of TNF-α in myelomonocytic cells significantly prohibits the PlGF-induced recruitment. Meanwhile, the calcium-binding proteins S100A8 and S100A9 have been observed to participate in monocyte chemotaxis in lung carcinomas [37]. Furthermore, monocytes can also be attracted by fibronectin, fibrinogen, and other factors produced by the cleavage of ECM proteins induced by proteases [38].

Activation and Polarization of Macrophages

Macrophages constitute extremely heterogeneous populations. Following recruitment and extravasation to tumor mass, monocytes are “educated” by the tumor microenvironment and differentiate into distinct types of macrophages, normally identified as M1 (classically activated) and M2 (alternatively activated) phenotypes.

Macrophage phenotypes are classified according to the Th1/Th2 dichotomy. M1 macrophages exhibit pro-inflammatory features and induce T-helper 1 (T_H-1) response, which are activated by interferon γ (IFN-γ) alone, or in concert with microbial stimuli (e.g., lipopolysaccharide, LPS) and cytokines (e.g., TNF and granulocyte-macrophage colony-stimulating factor, GM-CSF) [39]. In general, they are characterized by IL-12^high, IL-23^high, IL-10^low expression, efficient production of effector molecules (e.g., reactive oxygen intermediates, ROIs) and inflammatory cytokines (e.g., IL-1β, TNF, IL-6), and cytotoxicity to phagocytized microorganisms and neoplastic cells [39]. Consequently, M1 macrophages are regarded as soldiers to protect the host from infections and tumors by activating immune responses. In contrast, M2 macrophages exert an immunosuppressive phenotype and release cytokines to enhance T_H-2 response. Various forms of M2 macrophages are differentiated from monocytes when they are stimulated with CSF-1, IL-4, IL-13, immune complexes in association with either IL-1 receptor (IL-1R) or toll like receptor (TLR) ligands, and IL-10 together with glucocorticoids [40]. The hallmarks of M2 macrophages are IL-12^low, IL-23^low, IL-10^high, IL-1ra^high, decoy IL-1RII^high expression, secretion of CCL17, CCL22, and transforming growth factor beta (TGF-β), high expression of mannose, scavengers and galactose-type receptors, and poor antigen-presenting capability. Another important characteristic is the variation of the arginine metabolism which is shifted toward the production of ornithine and polyamine rather than citrulline and nitric oxide (NO) via arginase [40, 41]. As a consequence, M2 cells act as labors of the host, which regulate inflammatory responses, adaptive immunity, scavenging debris, angiogenesis, tissue remodeling, and tissue repair [41].

The phenotype of TAM has been described to diversify along with the tumor development. At the sites of chronic inflammation or in regressing and non-progressing tumors, M1 cells are often predominate, which can switch to an M2 phenotype as tumors become malignant [42, 43]. In aggressive tumors, TAMs are commonly lack in producing NO, ROIs and inflammatory cytokines (e.g. IL-12, IL-1b, TNF-α, and IL-6), which resemble M2 macrophages [44–46].

The transition of TAM phenotypes is associated with their locations within tumors [47]. TAMs preferentially accumulate in avascular and necrotic areas where they are exposed to hypoxia. Activation of hypoxia-inducible factor 1α (HIF-1α) and HIF-2α in TAM has been shown to influence the localization and function of TAM by up-regulating the chemoattractants including VEGF, CXCL12, and the receptor CXCR4 [48, 49]. Loss of HIF-1α in a mammary tumor model has been found to suppress the expression of arginase-1 which controls the arginine metabolism during macrophage polarization [50]. Meanwhile, the NF-κB activity is progressively modulated in tumors during the transition from benign to advanced stages [41, 51]. Although NF-κB is constitutively activated in tumor cells and may be promoted by TAM and other microenvironment cues, defective and delayed NF-κB activation in TAM has been found in advanced tumors [51]. Inactivated NF-κB results in down-regulation of inflammatory cytokines, such as IL-12 and TNF-α, which restrain the M1 macrophages and enhance their transition to an M2 phenotype.

On the other hand, the expression pattern of cytokines in the tumor microenvironment also plays a central role in the orientation and differentiation of macrophages. Both tumor cells and TAM can secret the immunosuppressive cytokines IL-10, TGF-β, and prostaglandin E2 (PGE2) [44]. IL-10 has been identified to reduce a plethora of pro-inflammatory cytokines and chemokines in macrophages, which is dependent on the activation of signal transducer and activator of transcription 3 (STAT3). IL-10 also synergizes with IL-4 to induce the expression of arginase-1 in macrophages [52]. Moreover, IL-10 can orient the differentiation of monocytes to mature macrophages instead of dendritic cells [53], and drive the macrophage activation to an M2 phenotype favoring angiogenesis and tumor progression. Additionally, tumor cell-derived PGE2, the principal metabolic product of cyclooxygenase-2 (COX-2), enables TAM to express the M2-typical molecule IL-10 [54]. The aberrant expression of COX-2 has been shown to result in increased abundance of PGE2 in colorectal cancer [55, 56]. COX-2 inhibition represses arginase-1 expression and up-regulates CXCL1 in murine intestinal tumor model, which is responsible for the repolarization of TAM from M2 to M1 phenotype [57]. Recently, another pro-tumorigenic factor adrenomedullin (ADM) has been found to be primarily secreted by TAM in melanoma, which is stimulated by co-culture with melanoma cells. Most importantly, ADM can polarize macrophages to an M2 phenotype in an autocrine manner and consequently promote melanoma progression [58]. Therefore, the crosstalk between tumor cells and their microenvironment is critical for the polarization of TAMs, which fundamentally regulates angiogenesis and tumor progression.

Roles of Macrophages in Tumor Angiogenesis

TAMs have a profound influence on tumor angiogenesis. Clinical investigations have indicated the correlation between high TAM level and poor prognosis in a variety of human cancers, such as breast, prostate, uterine cervical, uterine endometrial, liver, lung, bladder, kidney, brain, ewing sarcoma, and oral cancers [59–69]. Intriguingly, macrophage infiltration is intimately associated with tumor angiogenesis in these cancers.

Likewise, the pro-angiogenic role of TAMs has been demonstrated by multiple experimental murine models. In oncogene-induced mouse mammary tumors (MMTV-PyMT), macrophages are recruited to the tumor mass, which accelerate the angiogenic switch and increase tumor malignancy [70]. Accordingly, depletion of macrophages in MMTV-PyMT tumors reduces the vascular density by 50 % and retards the malignant transition. Meanwhile, genetic overexpression of CSF-1 has been shown to enhance macrophage recruitment which results in enforced vascularization [71, 72]. Using a mouse dorsal skinfold chamber model, macrophages are found to promote angiogenesis via the induction of VEGF [73]. Furthermore, removal of circulating monocytes by clodronate liposomes inhibits both TAM density and angiogenesis in Lewis lung carcinoma xenografts [74]. These studies demonstrate that TAMs play significant roles in vascular remodeling during the malignant progression.

How do macrophages stimulate tumor angiogenesis? Several mechanisms should be involved, such as secretion of a plethora of pro-angiogenic factors and trans-differentiation into endothelial cells. Specifically, interactions with tumor cells and the microenvironment substantially support the role of TAM in angiogenesis.

Numerous growth factors including VEGF, basic fibroblast growth factor (bFGF), epidermal growth factor (EGF), platelet-derived growth factor (PDGF), and TGF-α, can be produced by TAMs. These factors are well-documented pro-angiogenic molecules and mostly expressed when TAMs are exposed to hypoxia [75]. Among them, VEGF is a key determinant for the regulation of TAM in angiogenesis. Up-regulation of VEGF in TAM has been revealed to increase vascularization and malignancy in glioblastoma, renal cell carcinoma, T-cell lymphoma, and breast cancer [76–78]. VEGF has been well defined to promote endothelial cell proliferation, migration and survival [79], amplify the attraction of TAMs to the hypoxic regions, and induce the expression of tissue matrix to enhance angiogenesis [80]. In addition, hypoxia tightly regulates the expression of various pro-angiogenic chemokines in macrophages, including CXCL12, CCL2, CXCL8, CXCL1, CXCL13, and CCL5. CXCL12 has been reported to chemoattract endothelial cells by activating HIF-1α via CXCR4 [81]. CXCL5 mediates pancreatic tumor angiogenesis through activation of protein kinase B (Akt), extracellular signal-regulated kinase (ERK), and STAT via CXCR2 in human endothelial cells [82]. CXCL8 also results in increased neovascularization in several cancer types [83].

Several inflammatory cytokines participate in macrophage-associated angiogenesis as well, among which TNF-α is a major one. Although TNF-α is postulated to cause tumor necrosis, new blood vessel formation has been found to be elicited by exogenous application of TNF-α [84]. Neutralization of TNF-α in the conditioned media of activated TAM impedes its angiogenic activity [85], indicating the essential role of TNF-α as an angiogenic molecule derived from macrophages. Meanwhile, TAMs are important sources of IL-8 and IL-6. Tumor angiogenesis induced by the conditioned media of macrophages is noted to be markedly abolished by anti-IL-8 antibody [86]. IL-6 is implicated in the regulation of vasculogenesis and can synergize with other cytokines to promote the generation of new blood vessels [87]. Moreover, recent studies have shown that TAM-derived ADM promotes angiogenesis in a paracrine manner via the endothelial nitric oxide synthase (eNOS) signaling pathway [58]. Intriguingly, the expression of TNF-α, IL-8, IL-6, and ADM can be significantly elevated in macrophages when co-cultivated with tumor cells, which demonstrates that the interaction between tumor cells and macrophages have significant effects on endothelial cell function and tumor angiogenesis.

Secretion of tissue matrix-degrading enzymes and inhibitors by TAMs is also responsible for its pro-angiogenic function. TAMs are capable of producing MMPs (MMP-2, MMP-7, MMP-9, and MMP-12), serine and cysteine proteinases (urokinase, tissue plasminogen activator, and cathepsin B), and plasminogen activator inhibitors to regulate the disruption and re-construction of extracellular matrix [6], which provide support for blood vessels and mediate endothelial cell migration. Furthermore, degradation of tissue matrix can release many angiogenic factors sequestered in tissue matrix and therefore facilitate angiogenesis [88].

Several studies have suggested that monocytes can trans-differentiate into endothelial cells under certain circumstances [89], which may contribute to tumor angiogenesis. However, the exact role of monocyte differentiation in tumor angiogenesis and the underlying mechanisms are still obscure and merit further investigations.

Therapeutic Implications of Macrophages

Experimental and clinical studies have revealed that TAMs are of great significance in angiogenesis, tumor progression, and poor prognosis, which have represented a promising therapeutic target in neoplastic diseases. The comprehension of its recruitment, activation, and pro-angiogenic functions has prompted the development of anti-cancer therapeutics by either impeding their accumulation at tumor sites or retarding the efficacy of TAM-derived angiogenic mediators.

Since monocyte recruitment is a pre-requisite for the accumulation of TAMs, monocyte chemoattractants have represented valuable therapeutic targets in neoplasia, especially for CCL2, CXCL12, PlGF, and CSF. Gene-transfer of a dominant negative CCL2 mutant inhibits TAM recruitment and tumor angiogenesis [90]. Suppression of CXCL12 delays glioblastoma recurrence after irradiation by attenuating macrophage recruitment which participates in tumor re-vascularization [28]. Combined blockade of integrin-α4β1 plus CXCL12 or IL-1β also potently inhibits TAM recruitment and suppresses tumor progression [91]. Neutralization of PlGF has been shown to inhibit lung metastasis by blocking the recruitment of myeloid cells [92]. Inhibition of NFAT1 and TNF-α in macrophages prevents PlGF-induced monocyte recruitment to the tumor mass [36]. Transgenic expression of CSF-1 in mammary epithelium accelerates the carcinoma progression and increases lung metastasis, suggesting that agents targeted CSF-1/CSF-1R activity will have important therapeutic effects [93].

As well as recruitment, activation and polarization of TAM from M1 to M2 phenotype have been regarded to fundamentally modulate tumor growth and metastasis. Hence, restoration of an M1 phenotype in TAMs may provide therapeutic benefits. The M2-typical factors such as IL-10, TGF-β, ADM, and PGE2 are all valuable targets of TAM polarization. In fact, available information has shown that combination of CpG plus an anti-IL-10 receptor antibody switches infiltrated macrophages from M2 to M1 and triggers tumor rejection [94].

Considering the role of TAM in angiogenesis, agents that antagonize the pro-angiogenic factors secreted by TAM will be effective for tumor repression. For instance, antagonists of VEGFR-1 and CXCR4, which respectively neutralize the function of VEGF and CXCL12, may reduce macrophage-associated angiogenesis. Suppression of ADM has been demonstrated to impede angiogenesis and tumor growth in melanoma. Disruption of the MMPs secretion patterns from TAM is probably to prevent the degradation of extracellular matrix, thus inhibits angiogenesis and tumor cell invasion.

Taken together, these findings suggest that TAM can serve as a promising target for cancer therapy. Novel agents aimed at inhibiting the recruitment, activation, and pro-angiogenic functions of macrophages will have great potential for the improvement of current anti-tumor therapeutics.

Pericyte and Angiogenesis

Angiogenesis is a dynamic process that involves the remodeling of existing vessels and the formation of new vessels [3, 95]. The newly formed functional blood vessels are composed of two distinct cell types: endothelial cells and pericytes [96–98]. Endothelial cells form a single cell layer that lines all blood vessels, while pericytes either are associated with the endothelial cell tube, or form a single, often discontinuous, cell layer around it [96, 99, 100]. Nowadays, a series of endothelial cell-targeted inhibitors including endostatin, bevacizumab, have been widely used as anti-tumor drugs in the clinic, indicating the indispensible role of endothelial cells in tumor angiogenesis [2, 101]. However, more and more reports have shown that targeting endothelial cells alone is not as effective as previously expected [96, 102, 103]. One well-accepted explanation for this mystery is that the anti-angiogenic inhibitors mainly target the immature endothelium lacking pericytes coverage, while show limited effect on the pericyte-associated mature vessel [103, 104]. Hence the treatment of endothelial-targeted drugs could specifically repress the immature vessels but promote the process of vascular normalization, which has been demonstrated as a positive factor for the delivery of oxygen and nutrient, facilitating tumor progression [105, 106]. Thus pericytes have emerged as an important therapeutic target for anti-angiogenic therapy. However, the detailed mechanisms of the derivation of pericytes as well as their crucial roles in angiogenic process are still not completely understood [107–109].

The Origin of Pericytes

Within the tumor microenvironment, activated endothelial cells proliferate after activation, and form new endothelium, or endothelial progenitor cells are recruited by the factors such as VEGF, PlGF, and contribute to de novo vasculogenesis. However, the origin of pericytes remains controversial [96, 110–114]. Several studies have indicated that tumor-associated fibroblasts could differentiate into pericyte-like cells which further sustain the immature vessels [115]. It has also been demonstrated that Flk1⁺ cells can differentiate into endothelial cells or pericytes upon stimulation by VEGF or PDGF-B, respectively [116]. This suggests that endothelial cells and pericytes share a common progenitor cell. More recently, it has been well documented that pericytes are mainly derived from BMDCs [111, 112, 117]. PDGFRβ⁺ pericyte progenitor cells are also discovered to be bone-marrow derived [118]. Tumors are explored to recruit endothelial cell precursors expressing hematopoietic stem cell markers including Sca1, CD11b and c-Kit from bone marrow [113]. In addition, bone marrow-derived mesenchymal stem cells (MSCs) have also been indicated as the pericyte progenitor cells. Salven and his colleagues have reported that BM-derived peri-endothelial vascular mural cells are persistently detected at sites of VEGF-induced angiogenesis [119], which suggests that the major contribution of bone marrow to angiogenic processes is not only by preexisted endothelial cells, but also comes from progenitors for pericytes. Although these studies all provide sufficient evidences to support their individual discoveries, it is worth noting that the aforementioned pericytes express variant markers. For instance, alpha smooth muscle actin (α-SMA) is a well-accepted marker protein for pericytes [108], however, the expression of α-SMA in tumor-associated pericyte is controversial. Bergers and his colleagues have shown that α-SMA co-localizes with PDGFRβ, while Rajantie et al. have reported that the subpopulations of nerve/glial antigen 2 (NG2) proteoglycan positive cells cannot detectably express the pericyte marker α-SMA [119]. One plausible explanation is that the pericytes recruited to the newly formed blood vessels are heterogeneous within the tumor tissue. Unlike the well-controlled physiological environment, the pathological tumor microenvironment is more complicated. To fulfill the oxygen demand, dozens of functional blood vessels should be generated once the tumors reach a size of 1–2 mm in diameter [3, 95]. Thus a series of recruitment- or differentiation-related cytokines/chemokines may be largely synthesized and secreted under hypoxic condition, generating an overwhelming effect on the local and distant environment. Not only stromal cells are in situ differentiated into pericytes, but also the BMDCs are recruited to the endothelium [115]. Therefore, it is conceivable that the tumor-associated pericytes exhibit different expression patterns according to the different derivations. Till now, only little is known about the variability of pericyte markers in tumors, as most studies have used only a single marker, usually α-SMA or NG2 [120, 121]. Published reports have suggested that the level of pericyte coverage on vessels in different tumors ranges from extensive to little or none [120]. These controversial reports might be therefore based on the different marker expression by the pericytes with variant origins.

Pericyte-endothelial Cell Interactions

Although the role of pericytes in tumor pathophysiology is still not completely identified, the mechanism of pericyte-endothelial cell interactions within tumor microenvironment has been largely reported [96–98, 107, 109]. Factors including platelet-derived growth factor BB (PDGF-BB), angiopoietin-1 (Ang-1), sphingosine 1-phosphate (S1P) and transforming growth factor beta (TGF-β) are exclusively reported to regulate the recruitment and attachment of pericytes to endothelium [96, 107, 108].

PDGF-BB

Platelet-derived growth factor is a mitogenic peptide growth factor, and PDGF family consists of 5 members: PDGF-AA, PDGF-BB, PDGF-CC, PDGF-DD, and PDGF-AB, sharing 3 receptors named PDGFR αα, PDGFR ββ, and PDGFR αβ [122–124]. Among these five dimers, PDGF-AA, PDGF-BB, and PDGF-CC serve as pro-tumorigenic factors [125], while only PDGF-BB has been implicated to be involved in vascular maturation [122]. Endothelial cells, triggered by hypoxia, generate a chemotaxis gradient that recruits the PDGFRβ⁺ pericytes to the surface of endothelium. In addition, PDGF-BB also stimulates the proliferative activity of pericytes and induces mural cell fate of undifferentiated mesenchymal stem cells [118, 126]. Ablation of PDGFRβ in mice leads to severe vascular leakage and pericyte apoptosis [127]. Endothelial cells in the PDGFB deficient mice are unable to attract PDGFRβ⁺ pericytes progenitors, thus fail to form mature blood vessels, suggesting the essential role of PDGF-BB/PDGFRβ signaling in the formation of functional vasculatures [127, 128]. Intriguingly, Lee and his colleagues have reported that overexpression of PDGF-BB in tumor cells dramatically increase the pericyte coverage [129]. Moreover, Song et al. have also shown that tumor-derived PDGF-BB increases tumor pericyte coverage by activation of stromal-derived factor 1 alpha (SDF-1α), which forms a chemotaxis gradient to compensate the PDGF-BB signaling and induce pericytes recruitment [8].

Angiopoietin

The second important mediator for vascular maturation is the angiopoietin (Ang) system [130, 131]. The Ang family consists of several members including Ang-1, Ang-2, Ang-3 (murine specific), and Ang-4 (human specific) and two tyrosine kinase receptors (Tie1 and Tie2) [132]. Among them, constitutive Ang-1/Tie2 signaling is required to maintain the quiescent vasculature [132, 133]. Ang-1, predominantly secreted by pericytes, can bind with Tie-2 on endothelial cells in a paracrine pattern, enhance the pericyte-endothelial interactions, repress the proliferation and migration of endothelial cells, and further promote the maturation of newly formed blood vessels [130, 131]. Moreover, Kobayashi et al. have observed that hepatocyte growth factor (HGF), a mediator of pericyte cell motility, is up-regulated by Ang1 stimulation [134], suggesting a compensate recruitment signaling that cooperates with PDGF-BB/PDGFRβ.

Ang-2, released from Weibel-Palade bodies of the endothelial cells, acts as an antagonist of Ang-1/Tie2 signaling [135]. When endothelial cells are activated [136] by tumor-derived pro-angiogenic factors, Ang-2 is largely secreted and competes the binding of Ang-1 to Tie2, leading to the reduction of pericyte association and the remodeling of the pre-existed vasculatures [135]. Moreover, transgenic mice overexpressing Ang-2 show dense vascular networks with reduced pericyte coverage in the retina [137]. On the other hand, Ang-2 can also act as an agonist of Tie2 in an autocrine manner, and activate the downstream pathways including Pl3K/Akt, thus function as a promoter of angiogenesis [138]. Moreover, Ang-2 also destabilizes the vasculature in distant organs to facilitate the formation of tumor pre-metastatic niche. Huang et al. have discovered that primary tumor secretes pro-metastasis factors including TGF-β and TNF-α, which can interfere with quiescent pulmonary capillaries and lead to the lesion formation by up-regulating Ang-2 and MMPs [139]. All these reports suggest that Ang-2 is a potent target for the anti-tumor therapies.

TGF-β

The differentiation of pericyte progenitor cells around the immature blood vessels depends on TGF-β [140]. Genetic ablations of TGF-β and its receptors including activin-receptor-like kinase 1 (alk1), alk5, TGF-β receptor II (tβrII), and type III TGF-β receptor, all lead to embryonic vasculature defects and lethality [141–143]. It has been reported that TGF-β triggers different endothelial responses via either ALK1/Smad1/5 or ALK5/Smad2/3 [140, 144]. When the ALK1 pathway is activated, the TGF-β response is shifted toward proliferation [140]. In contrast, ALK5 signaling induces the differentiation of endothelial cells, and further facilitates pericytes association to the endothelium. Several studies have highlighted the indispensable role of TGF-β for pericyte differentiation. It has been demonstrated that TGF-β is essential for the differentiation of 10 T1/2 cells to a pericyte-like phenotype. Intriguingly, similar with PDGF-BB and Ang-1, TGF-β can also stimulate the secretion of MCP-1 in endothelial cells via Smad3/4, which mediates the TGF-ß-induced pericyte migration toward endothelial cells [136].

Sphingosine-1-Phosphate

Sphingosine-1-phosphate (S1P) is a secreted sphingolipid that mediates cell communication via the G-protein coupled receptors (known as EDG-1) [145, 146]. S1P regulates the cell junction changes, and influences cell migration, proliferation, and survival [147]. Deletion of the s1p1 gene in mice causes mid/late-gestational lethality with defects in the pericyte coverage of vessels. In addition, tissue-specific knock-out of S1P₁ in endothelial cells resembles the global knock-out phenotype [146]. However, pericyte-specific deletion did not cause embryonic lethality, suggesting that only the signaling of S1P₁ in endothelial cells could modulate pericyte coverage of immature vessels. Regarding the signaling that endothelial cell-derived S1P affects pericyte recruitment, it has been reported that when interacting with its receptors S1P₁ and S1P₃, S1P could enhance the N-cadherin-based adherence junctions by the induction of small GTPase Rho- and Rac-dependent surface translocation of N-cadherin, thereby strengthening the association between endothelial cells and pericytes [147].

Pericyte and Anti-angiogenic Therapy

During the initiation of angiogenesis, the basement membrane is disrupted by MMPs. Activated pericytes detach from the vessel wall, and migrate into the perivascular spaces [108]. In this pathological condition, partial dissociation of pericytes contributes to increased tumor vascular permeability [148, 149]. In addition, although the initial endothelium formation may not involve the pericyte coverage, pericytes are indicated to localize at the growing front of angiogenic endothelium and guide newly formed vessels [102]. Therefore it is conceivable that pericytes play crucial roles in the angiogenesis process and could serve as a potent target for anti-tumor therapy.

The main mechanism of anti-angiogenic agents is vascular normalization [105, 106, 150]. This treatment could dramatically change the balance of the pro- and anti-angiogenic factors, and also help the immature tumor vasculature undergo maturation by specifically targeting the non-pericyte covered endothelial cells [105, 106, 151]. For instance, anti-angiogenic treatment based on blockade of VEGF induces the regression of tumor vessels and tumor growth, leading to the increased pericyte coverage and vessel normalization [152]. It has also been reported that after treatment of RIP1-TAG-2 tumors with AG-013737 or VEGF-trap, surviving pericytes associate with endothelial cells more tightly. Since pericytes provide a scaffold for the rapidly re-growing of tumor vessels, pericytes have been indicated as putative targets in pharmacological anti-tumor therapy. Bergers et al. have showed that combined treatment of anti-PDGFR agents together with anti-VEGF significantly reduces pericyte coverage and increases the success of anti-tumor treatment in the RIP1-TAG2 mouse model [103]. Gleevec (Imatinib), a well-known chemic targeting PDGFRs and other receptor tyrosine kinases, does not show a marked anti-angiogenic efficacy alone, but indeed promotes the effects of VEGF inhibitors synergistically. Long-term blockade of PDGF signaling by anti-PDGFRβ antibody reduces the content of pericyte within the tumor tissue, and also increases the apoptosis of endothelial cells. In addition, both the receptor tyrosine kinase inhibitor SU6668 (targeting PDGFRβ) and AX102 (a novel selective PDGF-B blockade DNA aptamer) could cause the progressive reduction of pericytes in xenografts, thereby suppressing tumor growth [153, 154].

Concluding Remarks

It is evident that tumor infiltrating macrophages and pericytes originated by bone-marrow derived cells contribute to tumor angiogenesis. These infiltrating cells can release paracrine angiogenic factors, or provide permissive conditions, thus inducing the growth and maturation of the newly formed vasculatures. In future studies, much greater attention should be paid to better identification of BMDCs subpopulations that are critical for tumor angiogenesis. Considering the heterogeneity of tumor associated BMDCs, the differentiation mechanisms of these cell subsets and their exact roles in neovascularization processes including sprouting, vessel remodeling, and maturation should be challenging to dissect. In addition, signaling pathways modulating the interaction of endothelial cell-pericyte have currently been investigated and will provide valuable information on the paracrine mechanisms for capillary formation. These discoveries are of critical interest in clinical approaches for anti-angiogenic therapies. Screening drugs that allow regulation of the pericyte coverage will provide a potent tool capable of controlling vascular permeability. A stable functional microvasculature may also represent an important prerequisite for preventing tumor metastasis. However, it is worth noting that the pericyte coverage varies in different types of tumor. Whether the normalization of tumor blood vessels can always be achieved remains obscure. Further studies are needed to highlight further aspects of pericyte molecular biology and physiology.