Molecular Mediators of Angiogenesis

Abstract

Angiogenesis, or the formation of new blood vessels from the preexisting vasculature, is a key component in numerous physiologic and pathologic responses and has broad impact in many medical and surgical specialties. In this review, we discuss the key cellular steps which lead to the neovascularization of tissues, and highlight the main molecular mechanisms and mediators in this process. We include discussions on proteolytic enzymes, cell/matrix interactions, pertinent cell signaling pathways, and end with a survey of the mechanisms which lead to the stabilization and maturation of neovasculatures.

I. Introduction

The study of angiogenesis has clinical applications in numerous fields including peripheral and coronary vascular disease, oncology, hematology, wound healing, dermatology, ophthalmology, and many others. Experimental evidence demonstrates that the promotion of angiogenic processes can have potential therapeutic benefit in ischemic tissue, while the study of anti-angiogenic therapies have yielded several FDA approved additions to chemotherapeutic regimens in the field of oncology.¹ In wound physiology, neovascularization is a key component in the adaptive repair response, promotes prompt coverage of thermal and traumatic wounds, and is required for the survival of grafts and flaps. In addition, inadequately vascularized tissue increases the risk for bacterial and fungal infections which complicate wound management, increase hospital admissions and clinic visits, prolong hospital stays, and add to the cost of health care for patients. For these reasons, a clear and fundamental understanding of the molecular mechanisms which regulate angiogenesis should allow for the development of targeted pro- and anti-angiogenesis therapies which would provide for benefits in the management of complicated wounds, the development of biologic tissue substitutes, and the treatment of pathologies across numerous medical and surgical specialties.

This review focuses on the relevant growth factors, cell-matrix interactions, matrix remodeling enzymes, and pertinent intracellular signaling pathways involved in modulating angiogenesis. We will also discuss the cellular and molecular processes which contribute to the prevention of uncontrolled growth and regression of newly developed vessels.

II. The Process of Angiogenesis

Overview

New blood vessel formation, or neovascularization, occurs either by angiogenesis or by vasculogenesis. Vasculogenesis is the creation of new vessels de novo from precursor cells, such as angioblasts, which differentiate into endothelial cells (ECs), form lumens, and create primitive blood vessels. In contrast, angiogenesis is the formation of new capillaries from the preexisting vasculature. Arteriogenesis, or collateralization, results from the hypertrophy and luminal distention of preexisting vessels in response to mechanical stresses caused by redirected blood flow from occluded or stenosed distal vessels, mediated in part by mechanosensitive signaling in vascular wall cells and by macrophage-derived biomolecular signals.² While a clinically important process, arteriogenesis is not the formation of new blood vessels and will not be discussed.

The process of angiogenesis involves a complex and dynamic interaction between ECs and the corresponding extracellular environment. In vivo, angiogenesis occurs either by the sprouting of vascular ECs from pre-existing capillary endothelia into the surrounding tissues, or by intussusception (aka non-sprouting angiogenesis), which involves the division of capillaries by tissue pillars into two or more daughter vessels.^{3, 4} This review will focus primarily on sprouting angiogenesis, to which our usage of the term “angiogenesis” will refer. However, the distinctions made between sprouting and non-sprouting angiogenesis, angiogenesis and vasculogenesis, and the individual processes described below, may be somewhat arbitrary, as they all likely occur in concert and are regulated by related cellular and molecular mechanisms.

Endothelial Cell Sprouting

In a normal blood vessel, a 100-200 ηm thick basement membrane (BM) lies immediately deep to the EC monolayer in the arterial intima. Composed mainly of laminins, type-IV collagen, type-VIII collagen, and proteoglycans, the BM must be degraded prior to EC invasion into the surrounding extracellular matrix (ECM), which itself is comprised of fibrillar collagens, elastins, and various other ECM proteins.⁵ ECs, in response to angiogenic stimuli, convert from a quiescent to a synthetically active phenotype characterized by a high mitotic index and increased capacity for migration and matrix proteolysis. These activated ECs are capable of disrupting the tight junctions, adherens junctions, and gap junctions which exist between neighboring intimal ECs and perivascular cells, and invade into the BM and surrounding ECM.^{3, 6} Once freed from the capillary intima and in the extravascular space, ECs proliferate and migrate towards chemotactic and angiogenic stimuli in a 3-D extracellular environment and form new angiogenic sprouts (Fig 1). ⁷

Figure 1.

Diagrammatic description of the steps involved in angiogenesis. A) A stable vascular structure comprised of an endothelial monolayer surrounded by mural tissue (i.e. basement membrane, pericytes, and adventitia); B) Destabilized vessel demonstrating the disruption of mural tissue by endothelial cells; C) Endothelial cell invasion into perivascular tissue and subsequent sprout formation; D) Endothelial cell lumenogenesis and tube formation with inosculation (not shown) to other pre-existing vascular structures; E) Stabilization of the capillary sprout by recruitment of pericytes and deposition of basement membrane.

Lumenogenesis and Tubulogenesis

Lumen formation (and subsequent tube formation) can be considered a hallmark of angiogenesis as it is a relatively specific behavior attributed mainly to cells of the epithelium and endothelium. It is the genetically programmed capacity of ECs to create luminal compartments within multicellular chains which allows for the flow of blood from the pre-existing vasculature to the neovasculature, without which new capillary networks would be unable to perform their central function of oxygen and nutrient transport to normal or pathologic tissues. Indeed, one of the earliest demonstrations of angiogenesis in vitro by Folkman and Haudenschild provided evidence that, in the presence of “direct” angiogens, ECs in culture form capillary-like structures with rudimentary luminal compartments.⁸

While several mechanisms for lumen formation have been suggested, the most widely investigated mechanism is the process of intracellular vacuolization (or intracellular canalization). The earliest observations of angiogenesis described the presence of “seamless” EC lined capillary lumens in vivo which were lacking in apparent cross-sectional EC-EC junctions.^{9, 10} Since then, several in vitro and in vivo studies have provided for a mechanistic model of lumen formation consistent with this observation. Mediated by the α₂β₁ integrin and members of the Rho GTPase family, ECs undergo pinocytosis, leading to the formation of intracellular vacuoles which coalesce and form one larger intracellular lumen.^11-16 This would explain the appearance of “seamless” EC lined capillary lumens lacking apparent junctional contacts among several lumen lining cells. The association of caveolin-1 expression, a protein involved in receptor mediated endocytosis and the formation of caveolae (cellular invaginations which often precede vacuole formation), with EC lumenogenesis is an interesting observation in light of this mechanistic hypothesis.¹⁷ Concurrently, cytoplasmic projections have been suggested to sense and form junctional contacts with neighboring ECs to thereby form more complex multicellular capillary tubes.¹⁵

While intracellular vacuolization is the most widely studied model of lumen formation in ECs, there are descriptions of other mechanisms. These include intussusception, or the insertion of tissue pillars into the newly forming capillaries,¹⁸ autophagy by lysosomes within individual ECs which leave behind luminal structures,¹⁹ trans-luminal ingrowth of cytoplasmic filopodial projections which create a network of luminal structures,¹⁶ as well as the exocytosis of vacuoles between two (or more) ECs which are sprouting in close apposition to one another to form intercellular lumens (intercellular canalization).^{11, 13} Other groups have proposed that apoptosis of centrally located ECs within a cluster of many ECs can lead to tubular neovascular structures.^{20, 21} This is consistent with data that demonstrates deficiencies in embryologic lumen formation and microvascular development in transgenic mice which conditionally express the anti-apoptotic protein Bcl-2 in ECs.²²

It is important to note that these processes are often described or observed using in vitro assays of angiogenesis, and that the translation of these finding to the modeling of physiologic lumen formation in vivo is challenging.²³ Nonetheless, these models of angiogenesis are invaluable in attempting to understand the basic cellular and molecular mechanisms regulating lumenogenesis and angiogenesis in general. It is also important to note that while the above mechanisms for lumen formation have been proposed as separate entities, they may in fact all be contributing to EC lumen formation concurrently, and the relative importance of any one particular mechanism may be dependent on factors such as the location of the cells undergoing lumenogenesis (i.e. the sprouting tip vs. the trunk/stalk of the sprout).²⁴ It has also been proposed that heterogeneous populations of ECs with different functional characteristics can contribute to lumen and tube formation simultaneously. Meyer et. al. suggested a model of angiogenic tube formation which relies on three different populations of ECs: 1) those with phagocytic functions which create matrix channels and later undergo apoptosis, 2) those which form vacuoles which coalesce and get exocytosed to the intercellular space leading to a primitive lumen which is subsequently remodeled, and 3) those which are added into the capillary tube at the final stage of angiogenesis and aid in lumen growth and remodeling.²⁵

A further consideration is the temporal relationship of lumen formation to sprouting. Whether one precedes the other, or, as suggested above, these processes occur simultaneously, is not entirely clear. There is experimental evidence to suggest both possibilities. Davis and Camarillo have demonstrated that vacuoles can be visible in ECs which are actively sprouting,¹⁴ while other studies demonstrate that branching and new sprout formation can occur in the absence of lumen formation.^{26, 27}

Inosculation

Inosculation is the fusion or anastomosis of two vascular lumens or luminal segments to form one continuous lumen. Although an understudied process in angiogenesis, some literature from the field of skin graft vascularization can be extrapolated to provide some insights into this later step of angiogenesis. In 1980, Tsukada demonstrated that the preservation of the subcutaneous vascular network of a skin graft facilitated free tissue transfer, and that this was associated with both the formation of a neovasculature, as well as the inosculation between the preexisting host and donor vasculature.²⁸ By analyzing the establishment of donor/host vascular continuity in grafted skin, studies have demonstrated that skin neovascularization with complete inosculation occurs between three to seven post-operative days following skin grafting.^{29, 30} Immunohistochemical staining of human skin transplanted onto athymic mice has documented recipient EC invasion into the graft between days 3 and 21 via neovascularization with blood flow commencing on days 3-7. Of note, recipient beds were not seen to have donor EC ingrowth.³¹ The molecular mechanisms involved in inosculation, however, are still relatively unknown, and further study of this phase of angiogenesis is required.

III. Molecular Mediators of Angiogenesis

Growth Factors and Cytokines

Among the most commonly described angiogenic growth factors and cytokines include vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), tumor necrosis factor-alpha (TNF-α), transforming growth factor-beta (TGF-β), and the angiopoietins (Ang). Sources of these growth factors include endothelial cells, fibroblasts, smooth muscle cells, platelets, inflammatory cells, and cancer cells. Some growth factors, including fibroblast growth factor-2 (FGF-2, or basic FGF), are sequestered in the ECM, and serve as reservoirs for sustained cell-demanded growth factor release. The local concentrations of specific growth factors in the extracellular milieu may be impacted by physiologic or pathologic responses such as those seen during injury response, chronic inflammatory states or cancer. Platelet derived growth factor (PDGF), also an important mediator of angiogenesis, will be discussed later in the context of neovessel stabilization and maturation.

Vascular endothelial growth factor

VEGF is one of the most well studied growth factors involved in EC migration, mitogenesis, sprouting, and tube formation. Upregulated VEGF and VEGF-receptor (VEGF-R) mRNA has been detected in the tips of invasive angiogenic sprouts, and antibody blockade of VEGF significantly decreases microvessel outgrowth.^{32, 33} This heparan sulfate binding molecule exists in several isoforms resulting from alternative splice variants from a single gene product. The soluble or membrane bound form binds most commonly to 2 transmembrane receptor tyrosine kinases, VEGFR-1 (Flt-1) and VEGFR-2 (KDR/Flk-1).^{34, 35} A third transmembrane receptor neuropilin-1 (Nrp-1) binds the VEGF-A splice variant of VEGF₁₆₅, is involved in capillary morphogenesis, and appears to be required for EC filopodial tip directionality during angiogenesis.³⁶ VEGF also promotes von Willebrand factor release, integrin expression, interstitial collagenase expression, plasminogen activator (PA) and plasminogen activator receptor (PA-R) expression, and increases both vascular permeability and fenestration, which accounts for its original name, vascular permeability factor. Heparin and heparan sulfate proteoglycan binding contributes to VEGF/VEGF-R binding, stabilizes the active conformation of VEGF, prevents its degradation, and has been shown to greatly increase the angiogenic response of ECs to VEGF in a type-I collagen matrix.³⁷

Its role in angiogenesis has made VEGF a popular target in both pro- and anti-angiogenesis therapies, and anti-VEGF/VEGF-R antibodies and receptor tyrosine kinase inhibitors have become agents of chemotherapeutic regimens in lung and colon cancer. ^{1, 38} However, its utility as a targeted drug or gene therapy has also been associated with some potential and observed side effects including gastro-intestinal toxicity, hypothyroidism, proteinuria, coagulation disorders, neurotoxicity, impaired wound healing, and excessive chaotic neovascularization.^{39, 40} It is probable that these potential side effects are partially attributable to a potentially narrow therapeutic range, making accurate therapeutic concentrations for both pro- and anti-angiogenesis therapy critical.⁴¹ Thus, the clinical applicability of VEGF (or any angiogen) therapy can likely greatly improve with a better understanding of its absorbance and release kinetics from ECM.⁴²

The above concept is especially true for therapeutic neovascularization strategies. For example, the attachment of VEGF to ECM scaffolds, such as fibrin or collagen, has been shown to promote sustained and controlled release of growth factor, and is associated with more organized vascular networks as compared to the chaotic and highly branching structures that commonly result from the local administration of their soluble counterparts.^{43, 44} Recently, VEGF-R binding and vascular ingrowth induced by a fibrin binding variant of VEGF₁₂₁ demonstrated low continuous receptor binding and increased subcutaneous vascular invasion with eventual pericyte recruitment compared to wild type controls.⁴⁵

Fibroblast Growth Factor

FGF is another pro-angiogenic growth factor, which is both stored in the vascular basement membrane to serve as a reservoir supply, and is upregulated during active angiogenesis. The two most commonly studied forms are FGF-2 or basic FGF (bFGF) and FGF-1 or acidic FGF (aFGF) which bind most commonly to the receptor tyrosine kinases FGFR-1 or FGFR-2.⁴⁶ In vitro, FGF binding of FGFR-1 increases EC migration and promotes capillary morphogenesis when cultured on collagen gels,⁴⁷ and activates signaling pathways mediated in part by protein kinase-C (PKC), phospholipase A₂, ^{48, 49} and numerous others. The observation that FGF-2 enhances endogenous VEGF production, and that VEGF is required for the FGF-2-induced expression of placental growth factor demonstrates the existence of cross talk and synergism between FGF and other growth factor pathways.^50-55 In addition, FGF-2-mediated proteolysis of matrix components (via up-regulation of urokinase receptors), and induction of the synthesis of collagen, fibronectin and proteoglycans by ECs, demonstrate its effects on ECM remodeling during angiogenesis.^{47, 56}

The successful maintenance of microvasculatures engineered by the exogenous delivery of FGF, or VEGF, is likely dependent on the local concentration of total growth factor present in the culture environment. ⁵⁷ Proof of this concept is provided by studies by Uriel, et. al. which demonstrated that persistent low levels of exogenously delivered FGF-1 promoted long-term survival of morphologically healthy appearing microvasculatures in vitro compared to higher levels of FGF-1.⁵⁸ The observed regression of engineered microvasculatures created by the sustained delivery of angiogens (i.e. VEGF, FGF) in the absence of physiologic demand highlights a potential cautionary paradigm in growth factor therapy of angiogen supply and cellular demand.⁴⁵ Unlike VEGF₁₆₅, however, FGF binds with a high degree of affinity to ECM proteins such as fibrin and with a high degree of saturation without a cross-linking molecule (i.e. heparan sulfate proteoglycan), which can potentiate its angiogenic potential.^{59, 60} This observation, coupled with the ability of ECs to produce and release FGF during angiogenesis,⁶¹ makes controlling local concentrations of FGF for therapeutic neovascularization purposes challenging. Strategies for sustained or targeted delivery of FGF and designer FGF mutants with biologic properties relatively more beneficial than those of wild type FGF have been investigated to partially address these challenges.^62-65

Angiopoietin

The angiopoietins are a family of extracellular ligands that bind to the primarily EC-specific Tie receptors. Similar to other growth factor receptors, they represent a group of receptor tyrosine kinases which function primarily during vascular remodeling and angiogenesis.^{66, 67} Ang-1 and Ang-2 bind the same receptor, Tie-2, but elicit very different responses. Ang-1 acts as an agonist, causing rapid receptor transphosphorylation, leading to subsequent downstream activation of the protein kinase B/Akt/FKHR (FOX01) pathway to promote EC survival and to inhibit Ang-2 expression.⁶⁸

This latter effect is significant, as Ang-2 is perhaps the most well studied signaling molecule involved in EC activation and the initiation of angiogenesis. The early role of Ang-2 in angiogenesis is evidenced by its dramatic upregulation in ECs after exposure to angiogenic stimuli such as hypoxia and VEGF.^69-75 The release of Ang-2 from Weibel-Palade bodies of vessel ECs induces an autocrine response, which, in the presence of persistent pro-angiogenic signals such as VEGF, acts as a competitive antagonist of Ang-1 to prevent Tie-2 receptor phosphorylation and signaling, is involved in the disruption of the endothelial monolayer, and promotes EC survival, vessel sprouting and angiogenesis.^{66, 76-78} Ang-2 has been shown to both promote the formation of neovessels during cerebral ischemia and to act as a critical regulator of postnatal vessel remodeling in the ischemic mouse hind limb.^78-80 Elevated Ang-2 levels in some tumors and direct Ang-2 retinal injections are associated with decreased pericyte coverage of vessel capillaries, which is consistent with a model of Ang-2-mediated dynamic vessel instability which can lead to increased neovascularization in the presence of other angiogenic stimuli.^{79, 81-83} A clear understanding of the role of angiopoietins remains elusive and difficult to define however, as their highly context dependent functions vary with the relative levels of Ang-1, Ang-2 , VEGF, and cellular sources of angiogens under specific physiologic and pathologic conditions.

Transforming Growth Factor-β

The TGF-β superfamily is comprised of over 30 structurally similar growth factors and include the three TGF-β isotypes 1-3. They are secreted in an inactive latent form which requires cleavage of the latency associated peptide domain by proteases, often under acidic conditions. In vivo, TGF-β promotes ECM deposition and integrin receptor upregulation,⁸⁴ is one of the most important cytokines for granulation tissue formation,⁸⁵ and modulates wound EC proliferation, migration and capillary tube formation.⁸⁶ While in vitro results have demonstrated that TGF-β can directly inhibit angiogenesis in a receptor specific manner,⁸⁷ its ability to recruit inflammatory cells in vivo, which act as a source of pro-angiogenic signals, or its ability to modulate other angiogenesis pathways, may account in part for its pro-angiogenic effects.^{86, 88} Its more well established role in mediating EC-pericyte interactions and vessel stabilization will be discussed below.

Tumor Necrosis Factor-alpha

TNF-α is an inflammatory cytokine, secreted primarily by activated macrophages during inflammation and immune response, but has also been associated with mast cells, ECs, fibroblasts, myocytes, adipose tissue, and neuronal tissue. It has a broad array of functions ranging from stimulation of granulocyte-macrophage-colony stimulating factor (GM-CSF) and interleukin-1 (IL-1), and has been suggested to act on ECs both directly, by inducing cell differentiation, and indirectly in vivo, by stimulating the production of angiogenic factors from other cells.⁸⁹

Cell-matrix interactions

The proteolysis and remodeling of the ECM is critical in all phases of angiogenesis, affecting EC migration, invasion into the perivascular tissue, and the morphologic formation of luminal structures. The regulation of the activity of these enzymes is highly complex and is impacted by the micro-environmental context within which it occurs. We will discuss the main tools of matrix remodeling, and attempt to highlight the bidirectional nature of cell-matrix interactions.

Matrix metalloproteinases (MMPs) and other proteases

Also call matrixins, MMPs are a family of proteases that degrade ECM proteins and are critical in vascular remodeling, cellular migration, and sprout formation.⁹⁰ The observation that there is significant upregulation of MMP activity in ECs during inflammation, wound healing and tumor growth reflect their important role in both physiologic and pathologic angiogenesis.⁹¹ Protein inhibitors of MMP-2 and MMP-9 have been shown to greatly attenuate the migration of both ECs and tumor cells and have been proposed as therapeutic targets in oncology.⁹² Knockout mice deficient in MMP-2 demonstrate greatly reduced melanoma tumor angiogenesis and tumor progression.⁹³ The anti-tumor effects of targeted MMP inhibition is likely due to the dual benefits of inhibiting tumor angiogenesis and metastasis, as these processes both require significant ECM degradation.^{94, 95} Disappointing clinical results with MMP inhibition in cancer therapy, however, highlights the complex nature of MMP activity in tumor and angiogenesis progression.⁹⁶

In humans, these zinc-containing endopeptidases may be either secreted or membrane bound (MT-MMP).⁹¹ The secreted forms are typically secreted as inactive pro-peptides which are often activated by a cascade of cell surface protease activity, including that of the MT-MMPs, which cleaves the pro-peptide to the active form. Increased expression of surface urokinase receptors in activated ECs, for example, bind the inactive proenzyme of urokinase plasminogen activator (uPA) (aka urokinase) and convert it to the active form. Active urokinase then promotes the conversion of plasminogen to plasmin.⁹⁷ Plasmin then activates MMPs that degrade the extracellular matrix.^{98, 99} As such, the inhibition of uPA by plasminogen activator inhibitors (PAI-1 and 2) can affect not only the direct degradation of ECM but also of the activation of latent MMPs,¹⁰⁰ and PAI expression in ECs or pericytes can act to modulate angiogenesis.¹⁰¹

In addition to contributing to the proteolytic activation of MMPs, other proteases like uPA, tPA, and plasmin contribute to significant ECM degradation with relative specificity. Plasmin, a serine protease, cleaves both ECM (i.e. fibronectin and laminin), as well as coagulation cascade proteins, and is the enzyme primarily responsible for fibrin degradation, a significant component of the injury response which often is a signal for neovascularization.¹⁰² Plasmin transversely cuts fibrin fibers and after prolonged exposure results in a broad range of progressively smaller fragments, which may themselves modulate the angiogenic response.^103-105 uPA is a secreted enzyme that binds with high affinity to its surface receptor, uPA receptor (uPAR), on a variety of cell types, and upregulated expression of both uPA and uPAR is seen in actively migrating ECs.^{56, 99, 106} In vitro, TNF-α potentiates VEGF- or FGF-2- induced EC tube formation in fibrin matrices by the mediation of cell surface u-PA expression.¹⁰⁷ Blocking uPA or uPAR, or the presence of anti-plasmin compounds such as ε-aminocaproic acid or aprotinin reduces cellular invasion and matrix degradation,¹⁰⁸ and the upregulation of PAI in an athymic mouse prostate cancer model has been shown to decrease both the density of tumor-associated microvasculatures and the rates of liver and lung metastases.¹⁰⁹

While the presence of growth factors can modulate the activity and expression of ECM proteolytic pathways, it is clear that the interaction between growth factors and proteases is bidirectional. Proteolytic growth factor activation or release from ECM reservoirs can propagate the angiogenic response by allowing for a sustained cell-demanded local concentration of active pro- (or anti-) angiogenic stimuli within the tissue. In addition, MMP-2 has been shown to directly cleave the extracellular domain of the FGF-1 receptor, which releases a soluble extracellular receptor domain that binds FGF-1 and may modulate angiogenesis in vivo.¹¹⁰ Pathologic dysregulation of this intrinsic positive feedback loop between protease and growth factor activity can have significant clinical consequences as seen during the neovascularization of tumors.¹¹¹

Matrix remodeling proteins also appear to be involved in the formation of luminal and tubular structures during the morphogenic differentiation of ECs in a manner that is specific to the type of MMP involved, with the membrane-bound forms appearing to be more critical in these events than secreted MMPs.^{112, 113} This has been supported by experiments utilizing specific tissue inhibitors of metalloproteinases (TIMPs) which demonstrate relative substrate specificity. The inhibition of MT1-MMP, but not MT3-MMP, for example, by TIMP-2 or by siRNA knockdown, inhibits EC tube formation in a 3-D model of angiogenesis.^{15, 114} This is consistent with results of an experiment in which aortic rings from MT1-MMP knockout mice displayed attenuated neovessel formation within a 3-D collagen matrix.¹¹⁵ These results should be viewed cautiously, however, as numerous factors (e.g. EC source), can produce vastly different responses from study to study.¹¹⁶ The mechanistic explanation for how MMPs contribute to lumenogenesis is unclear, but it has been proposed that MMPs create “vascular guidance tunnels”, which serve as molds in the ECM that aid in vascular morphogenesis, pericyte recruitment, and basement membrane assembly.¹⁵

The extracellular matrix

The ECM is a highly important modulator of angiogenesis during oncogenesis, wound healing, and development. For example, the differentiation of ECs grown as a monolayer in vitro into branching tubule structures has been shown to require the coverage of cells with collagen and the establishment of appropriate cell-matrix interactions.¹¹⁷ Angiostatic molecules like thrombospondin-1 (TSP-1) and interferon-inducible protein-10 have been shown to be less effective in inhibiting VEGF-induced EC proliferation when cultured on type-I collagen, an observation that has intuitive implications in the clinical use of pro- or anti-angiogenesis drugs.¹¹⁸ Laminin, a non-collagen hetero-trimeric (3 chains: α, β, and γ) glycoprotein found in basement membrane and granulation tissue, has been shown to increase EC-ECM attachment and migration in vitro,⁸⁶ is produced throughout the length of EC sprouts and has been shown to limit proliferation and lumen size,¹¹⁹ and is associated with breast and brain cancer neovascularization and membrane invasion (i.e. laminin-8 and laminin-10).^{120, 121} Fibronectin, often an injury response protein, is actively unfolded by ECs and serves as an ECM scaffold for cellular matrix deposition and neovascularization,¹²² while combinations of other healing proteins like hyaluronic acid, type I collagen and fibrin can have differential angiogenic effects compared to when they are present alone.^{123, 124}

While the composition of the ECM is clearly an important factor in the regulation of angiogenesis, the structure of the ECM is also key. Studies have demonstrated in numerous cell types that cell culture in 3-D environments vs. 2-D environments can significantly impact the expression of differentiation markers, diminish cell proliferation, and affect the expression of proteins like MMP-2.^{125, 126} Physical characteristics such as fiber thickness, density, orientation, cross-linking, and pore size have all been shown to affect the number, length, and cross-sectional area of angiogenic sprouts.^{5, 127, 128} In vitro, ECs have been shown to demonstrate decreased cell migration and invasion with increased capillary morphogenesis in fibrin as the rigidity is decreased by altering pH, NaCl concentration, and thrombin content. ^{108, 129, 130} In chick chorioallantoic membrane (CAM) angiogenesis assays, significantly more sprout invasion is associated with a reduction in fibrin substrate concentration, and the physical characteristics of the ECM were found to be as important as the presence or absence of growth factors such as FGF and VEGF.¹²³

Integrins

Among the most important cell surface receptors involved in cell-matrix interactions, the integrins are transmembrane heterodimeric receptors composed of an α and β subunit which engage ECM ligands to transduce environmental signals to cytoskeletal alterations and intracellular signaling pathways which affect phenotypic modulation, motility, mechanotransduction, and cell proliferation,^131-135 and thus impact a broad range of angiogenic processes.^{26, 136-145} In addition, downstream integrin signaling events mediate the cellular processes involved in angiogenesis, as exemplified by focal adhesion kinase (FAK) knockout mice which demonstrate altered EC morphogenesis during blood vessel assembly.¹⁴⁶

These receptors have the capacity to act in a ligand-specific manner producing specific cellular responses depending on the particular ECM or integrin subunits engaged,^{131, 147} with the α_v and β₁ subunits, specifically, proposed to be particularly important in morphologic differentiation.⁸⁶ Some evidence suggests that the disruption of cell-matrix interactions by α_vβ₃ and α_vβ₁ integrin receptor blockade actually tends to decrease EC proliferation and to promote the differentiation of ECs to large and multicellular tubules in a matrix specific manner.¹⁴⁸

Not surprisingly, the expression and activity of the EC proteolytic machinery is mediated by integrin-dependent signaling pathways. Observed colocalization of active cell surface MMP-2 with the α_Vβ₃ integrin in angiogenic blood vessels and melanoma cells likely reflects the tight coordination between motility and matrix degradation in invasive cells, once again highlighting the bidirectional nature of cell-matrix interactions.¹⁴⁹ The α₂β₁ integrin receptor, but not α₃β₁integrin receptor, has been shown to be required for type I collagen-induced gelatinase A (MMP-2) activation, in a mechanism that also requires MT1-MMP.¹⁰⁵ This, and other studies, demonstrate the highly complex proteolytic cellular responses which are dependent on the nature of the ECM-integrin interaction identified by the specific ECM and corresponding integrin subtype engaged.¹⁵⁰

In addition to its direct effects on cellular responses, engagement and activation of integrin-dependent pathways mediate responses to growth factors via significant cross-talk with growth factor pathways, allowing for coordination of cell-matrix-growth factor interactions during angiogenesis.^151-153 FGF-2-induced EC migration seems to be mediated in part by the upregulation of the α_Vβ₃ integrin. ¹⁵⁴ Similarly, VEGF has been shown to upregulate the expression of the collagen binding integrins α₁β₁ and α₂β₁, and blockade of these receptors attenuates the angiogenic response to VEGF.¹⁴¹ Recently, it was shown that VEGF recruits c-src to the EC membrane, which subsequently tyrosine phosphorylates the β₃ integrin and thereby mediates interaction between VEGF-R2 and integrin β₃ after VEGF stimulation.¹⁵⁵

Intracellular Signaling Pathways

A detailed and extensive review of the intracellular signaling pathways involved in angiogenesis is well beyond the scope of this review. We will attempt to introduce some of the most pertinent pathways here, and have cited recent reviews for further study.^{7, 156}

Rho GTPases

The Rho family of guanine nucleotide (GTP) binding proteins (including Rho, Rac, and Cdc42), which are members of the Ras superfamily of GTPases, are intimately involved in growth factor signaling, cytoskeletal dynamics, cell motility, cell shape and spreading, cell phenotype and other processes which significantly impact angiogenesis.^{157, 158} VEGF-induced EC chemotaxis has been shown to require activated Rac, and type I collagen-induced migration requires Rac and Cdc42 activity.¹⁵⁹ Interestingly, while both Rac and Rho induce the formation of stress fibers, special actinomyosin filaments arranged to generate the force required for movement, Rho activity may not be involved in either VEGF- or collagen-induced motility, and may even inhibit the migratory response to extracellular matrix.¹⁵⁹ Intracellular signaling molecules phosphatidylinositol 3-kinase (PI3-K), Akt, protein kinase B (PKB), p38 MAPK, FAK, and Rho-associated-kinase (ROCK) all provide molecular linkages among VEGF-R2-mediated Rho GTPase signal transduction pathways in EC migration.^{7, 157}

In addition to its effects on EC migration, several studies have demonstrated that Cdc42 plays a crucial role in EC lumenogenesis.¹² Consistent with this, recent evidence demonstrated that overexpression of Cdc42 GTPase activating protein (Cdc42 GAP), a protein which catalyzes GTP hydrolyzation and consequently decreases the active conformation of Cdc42, attenuates EC lumen formation while siRNA knockdown of Cdc42 GAP increased lumenogenesis.¹⁶⁰ In vitro, ECs with lower levels of Cdc42 GAP expression have been shown to be more likely to form tubes and lumens in comparison to ECs with a higher level of expression.¹⁶⁰ Although the exact mechanistic role of Cdc42 in EC lumen formation is still unclear, this and other downstream GTPase pathway proteins demonstrate potentially promising therapeutic targets for pro or anti-angiogenesis therapies.¹⁶¹

Protein Kinase C (PKC)

PKC is an intracellular kinase that is an important mediator of endothelial cell migration, proliferation and capillary tube formation. Decreasing PKC activity has been shown to attenuate EC attachment to and migration along ECM coated surfaces, without significantly affecting cell spreading.¹⁶² Its multifaceted role in VEGF signaling is demonstrated by the inhibition of VEGF-induced cell proliferation, MAPK phosphorylation, migration, endothelial tube formation, matrix adhesion, and vessel formation in vitro and in vivo in cells treated with antisense PKC-α oligonucleotides.¹⁶³

Notch

Signaling through Notch, a cell membrane receptor, is an evolutionarily conserved pathway which impacts cell fate and differentiation determination,¹⁶⁴ and is involved in regulating angiogenesis.^{165, 166} Notch receptors and their Delta-like-4 (Dll4) ligand appear to be central components in restricting the formation of tip cells in response to VEGF within angiogenic sprouts, which limits the branching and sprouting patterns of angiogenic networks¹⁶⁷ and limits tube diameter by inhibiting EC proliferation.¹⁶⁸ Blockade of Dll4 in tumor-associated ECs leads to the increased vascularity and branching of tumors, although these microvasculatures demonstrate poor function and are associated with decreased tumor growth.^{169, 170}

Cell-Cell Interactions

Perivascular cells such as pericytes (aka mural cells), SMCs, and adventitial fibroblasts, are associated with the endothelium of capillaries and are known to be involved in the later stages of angiogenesis during vessel maturation and stabilization (discussed below). However, these primarily mesenchymally derived cells surrounding the endothelium also appear to have a significant role in mediating the early events of active angiogenesis.^171-174 The co-culture of ECs with pericytes, SMCs and fibroblasts in vitro have demonstrated that the secretion of soluble factors such as VEGF or the deposition of ECM proteins such as fibronectin elicit angiogenic activities in ECs such as increased plasminogen activity, proliferation, motility, and cordlike structure formation.^175-180 Evidence that these paracrine interactions may be regulated by the local extracellular environment is provided by the observed endothelin-1-mediated activation of EC mitogenesis by fibroblasts,¹⁷⁶ and the production of Ang-2 and VEGF in mesangial cells (renal pericytes) under hypoxic conditions.⁶⁹ In vivo observations of pericytes found “leading” angiogenic sprout tips and lining vascular tubes connected to tumor vasculatures which lack ECs in vivo suggest a more direct role for pericytes as guiding cells during angiogenesis.^{181, 182} These data are consistent with observations that under hypoxic stimulation, VEGF-secreting perivascular cells appear to be the first vascular cells to invade the corpus luteum parenchyma during ovulatory angiogenesis.¹⁸³ In our own lab, we have observed that ECs in both direct and indirect co-culture with SMCs form sprouts which extend to greater lengths, form more robust networks, and persist over longer time periods in fibrin hydrogels compared to ECs cultured alone (unpublished results). Thus, while the later stabilization and maturation of the microvasculature by pericytes is well established, the likelihood that they promote angiogenesis given the proper pro-angiogenic environmental context suggests a more broad role for these cells which warrants further study.¹⁸⁴

Gene Expression

An emerging strategy for understanding the molecular regulators governing many aspects of angiogenic differentiation is the study of genetic expression patterns in activated angiogenic ECs. An analysis of genes expressed in ECs which invade and form tubules in a fibrin rich matrix vs. ECs which remain in a monolayer demonstrated differential regulation of a number of genes.¹⁶⁰ In vivo, genetic analyses of tumor-associated ECs, which tend to be more angiogenic than ECs from normal vessels, have identified potential angiogenesis markers, termed tumor endothelial markers (TEMs).^{160, 185, 186} The identity and function of many of these genes, are largely still unknown and under study, although their roles in EC survival, integrin signaling, migration, proliferation, and matrix remodeling have been implicated.^{160, 187-189} The analysis of TEM expression patterns, and genetic expression patterns in general, offers a unique and exciting methodology which may provide for real translational and therapeutic benefits in a wide variety of patients, and may even provide potential utility in the assessment of cancer prognosis.¹⁹⁰

IV. Stabilization and Maturation of the Neovasculature

Molecular and cellular feedback is critical in preventing pathologic angiogenesis. This is achieved by the preferential upregulation of anti-angiogenic stimuli and downregulation of the above pro-angiogenic processes. The net result is a final product of stable capillary tubes lined with a quiescent, functional monolayer of ECs demonstrating junctional contacts, low in both mitotic index and apoptotic turnover, supported by surrounding pericytes.

The extracellular matrix

A primary mediator of this final stage of angiogenesis is the composition of the ECM surrounding the endothelial sprouts.¹⁹¹ Type IV collagen, a major basement membrane protein, has been shown to both promote the lengthening of neovessels and to prevent their regression and continued growth in a dose-dependent manner.¹⁹² It has been shown that the degradation of basement membranes rich in type-IV collagen releases several anti-angiogenic proteins such as arresten (from collagen’s α1 chain), and canstatin (α2 chain), which suggests a mechanism for the almost immediate negative regulation of angiogenesis after its induction.^{193, 194} Alpha-statin and kinostatin, products of MMP mediated fibrinogen and HMW-kininogen degradation, respectively, possess anti-angiogenic activities and likely regulate the angiogenic response to injury.^{91, 96} Thrombospondin is a platelet and inflammatory cell derived protein found within the ECM that is secreted in response to injury¹⁹⁵ which has been shown to inhibit EC proliferation and migration, induce apoptosis, and downregulate MMP activity.^{102, 196} For these reasons, thrombospondin and other endogenous inhibitors of angiogenesis, like endostatin, have become attractive targets for cancer treatment, but have unfortunately met with limited clinical success.^{197, 198}

Regulation of proteolysis

Given the importance of proteolytic activity and the impact of ECM degradation in modulating angiogenesis, the regulation of MMPs and other proteases is a critical mode of angiogenesis control. Tissue inhibitors of metalloproteinases (TIMPs) are proteins that regulate MMP activity and can act to both activate and inhibit angiogenesis, reflecting the complicated pro- and anti-angiogenic nature of protease activity during angiogenesis.^{90, 91, 199} In addition to directly inhibiting MMPs, TIMP-3 prevents VEGF binding to VEGF-R2 and serves to inhibit VEGF induced signaling and angiogenesis.²⁰⁰

Pericytes

The recruitment and incorporation of pericytes are key events which mediate the final steps required for the stabilization and maturation of the neovasculature.⁶ Pericytes are primarily thought to be derived from cells of mesenchymal origins such as SMCs, fibroblasts or other progenitor cells and possess markers such as α-actin, desmin, PDGF-β receptor, and NG2 which are common to those cells.^{6, 173, 201} Several lines of evidence support the role of pericyte incorporation for the stabilization of neovessels and the regulation of uncontrolled angiogenic growth. Early studies showed that mice lacking the ability to recruit pericytes during embryonic vascular development demonstrated increased vessel microaneurysm formation and rupture.²⁰² Generally speaking, abnormal pericyte coverage and function in the microvasculatures of certain types of tumors is associated with increased EC sensitivity to VEGF withdrawal^203-207 and microvasculatures with disrupted EC/pericyte interactions are associated with more dynamic remodeling.²⁰⁸

Direct contact via gap junctions, adhesion plaques, and adherens junctions, as well as indirect paracrine mechanisms between ECs and pericytes, allow for significant interactions which impact cellular differentiation, migration, proliferation, and quiescence.^209-212 Among the paracrine mediators, PDGF is thought to be critical in the recruitment and proliferation of pericytes during angiogenesis. PDGF is a mitogen and chemoattractant for fibroblasts, SMCs, and other mesenchymal cells which is present in four known isoforms A, B, C, and D. Two known PDGF transmembrane receptor tyrosine kinases known as the α and β subunit bind each chain with relative specificity (the α receptor binds the A, B, and C chains primarily while the β receptor binds the B and D chains). Because PDGF exists as a hetero- or homodimer, ligand binding causes receptor hetero- or homodimerization (i.e. PDGF-BB and PDGF-AB can bind α/α, α/β, or β/β, while PDGF AA binds α/α exclusively) and subsequent cell signaling.^213-215 As PDGF-β is the primary receptor present on SMCs and other pericytes, PDGF-B/β signaling is especially critical in the proper recruitment of pericytes to capillaries. Disrupted PDGF-B/β signaling is associated with the attenuated ability to recruit PDGF-β receptor positive pericytes and precursors from the perivascular tissue and the bone marrow and is associated with significant vascular abnormalities in physiologic and pathologic angiogenesis.^{202, 205, 216-221}

The source of PDGF during this process is likely from actively angiogenic ECs themselves, which suggests an inherent negative feedback mechanism to prevent uncontrolled neovascularization.⁶ Proliferative ECs in vitro are potent stimulators of pericyte/SMC growth and migration, in part mediated by PDGF-BB, although this effect is dependent on the time in co-culture, the presence or absence of direct contact, and the deposition of endothelial-specific ECM proteins.^222-230 The importance of EC-specific PDGF-B production in pericyte recruitment in vivo was demonstrated in studies which showed that EC-specific knockout of PDGF-B led to significant vascular abnormalities with pericyte loss in mice.^231-233

Clinically, the simultaneous targeting of pericytes and ECs by pro- or anti-PDGF and VEGF therapy, has been investigated in tissue engineering and oncologic research. Combination therapy has been shown in some studies to be more effective in disrupting tumor vascularity than either agent alone and, as such, combinatorial antitumor therapy is under investigation.^234-237 It is unclear, however, if the apparent improved anti-angiogenic effect observed in combination therapy is specifically attributable to PDGF blockade or to the broad non-specific effects of the receptor tyrosine kinase inhibitors used in most studies of combination therapy. Recently, it was shown that adenoviral expression of soluble anti-PDGF-β receptor provided added inhibition of tumor growth and angiogenesis only when sub-optimal levels of VEGF blockade were present.²³⁸ Clearly, however, PDGF signaling does impact angiogenesis in vivo, as the spatiotemporal delivery of PDGF with VEGF has been shown to induce denser and more mature appearing vascular networks within tissue engineered constructs, which is in part associated with the recruitment of α-actin positive (indicative of pericytes) cells into the neovascular networks.^{239, 240}

Although the expansion, spreading, and migration of the SMC/pericyte population on newly formed capillaries may be dependent on PDGF-B/β signaling, it does not appear to be sufficient or necessary for the phenotypic differentiation of undifferentiated mesenchymal cells to pericytes. Instead, this appears to be regulated by TGF-β,^{218, 241} as evidenced by observations that ECs direct 10T1/2 (a cell line commonly used to model pericyte precursor cells in vitro) to an SMC/pericyte phenotype via a TGF-β-mediated mechanism,^{228, 242} and that the phenotypic differentiation of SMCs in co-culture with ECs involves activation of the SMC PI3-K/Akt pathway, which may be mediated by TGF-β.^{243, 244}

With these above observation taken together, it has been proposed that two modes of pericyte incorporation occur simultaneously: 1) the differentiation of undifferentiated mesenchymal and progenitor cells into pericytes; and 2) the recruitment and co-migration of differentiated mesenchymal cells such as SMCs from a pool of available cells in the perivascular tissue into neovessels.²¹⁸

It is believed that once incorporated, a major function of pericytes is to mediate the quiescence of ECs. In vitro, ECs in direct contact with pericytes have lower levels mitosis compared to ECs cultured alone, a process which may require Rho GTPase activity in pericytes.^{229, 245, 246} Similarly, ECs in close contact with SMCs in a spherical aggregate of SMCs in collagen gels demonstrated decreased responsiveness to VEGF (in the absence of Ang-2), increased inter-endothelial junctions, decreased PDGF-B growth factor chain expression, and reduced apoptosis.²⁴⁷ ECs in co-culture with 10T1/2 cells also displayed a significantly higher resistance to the permeability of a tracer molecule biotin-dextran which was associated with tightening of EC-EC tight junctions and with the localization of tight junction proteins plakoglobin, ZO-1, ZO-2, and occludin.²⁴⁸

It has also been suggested that pericytes induce vessel stabilization by the production of survival factors. This is supported by several lines of evidence: 1) Neonatal vessels which are resistant to oxygen-induced degeneration are associated with pericytes which express TGF-β₁, which induces VEGFR-1 expression in ECs;²⁴⁹ 2) 10T1/2 precursor cells produce VEGF upon TGF-β-mediated conversion to pericytes when in direct contact with ECs;²⁵⁰ 3) Pericytes express Ang-1 in vitro and in vivo, a growth factor which is associated with basement membrane formation, EC quiescence and endothelial leak resistance, and which can mature vessels lacking pericytes;^{67, 251-253} and 4) Pericyte loss caused by PDGF inhibition in a rat model of retinopathy results in overexpression of VEGF/VEGFR2 in ganglion cells and other supporting cells.²⁵⁴

A final mechanistic hypothesis for the pericyte-mediated regulation of angiogenesis is that pericytes significantly modulate the ECM remodeling capacity of ECs by directly influencing the local ECM composition. Pericyte-conditioned media has been shown to limit both EC migration and branching in vitro by inducing the upregulation of PAI-1 in ECs.²⁵⁵ In addition, pericytes have been shown to produce TIMP-3 and induce TIMP-2 production in ECs, which specifically targets MT1-MMP, MT2-MMP and ADAM proteinases, with the net effect of decreasing angiogenic potential and tube regression.^{112, 256, 257} In addition, pericyte-mediated ECM deposition, specifically the deposition and synthesis of basement membrane, has also been proposed to promote EC quiescence and vessel stability.^{191, 258, 259} The importance of the basement membrane in vascular stability is demonstrated in laminin-8 knockout animals which demonstrate poorly developed microvasculature with defective basement membranes, noticeable deformities, and increased responsiveness to pro-angiogenic stimuli.²⁶⁰

While the mechanisms discussed above in the maturation and stabilization of existing and newly formed vessels are still topics of current study, it is becoming clear that this final stage of angiogenesis is a significant contributor to the persistence and function of pathologic and physiologic neovasculatures, and will likely provide for future therapeutic strategies.

V. Summary

We have attempted to describe the major cellular processes involved in the angiogenic development of new blood vessels and the molecular mediators involved in regulating these events. It should be noted that while we do not cover the biomechanical environment within which these events occur, the effects of shear stresses and cyclic distention resulting from pulsatile flow is an important consideration.^{261, 262}

In summary, the relationships among phenotypically variable vascular (and progenitor and stem) cells, ECM and ECM receptors, growth factors and growth factor receptors, proteases, and intracellular signaling pathways during blood vessel formation represent an infinitely complex set of orchestrated events which are tightly regulated by a myriad of macro and micro-environmental factors in an apparently perpetual state of dynamic interdependent flux. It is a challenge for investigators of such complex biologic systems to develop more intricate and sophisticated models in order to paradoxically apply a more precise and reductionistic investigatory approach. The ultimate benefit of this work will likely be to facilitate the discovery of new anti- or pro-angiogenic therapies that can specifically target pathologic tissues to impact the health of patients in numerous medical and surgical specialties.