Ocular versus Extraocular Neovascularization: Mirror Images or Vague Resemblances

The First ARVO/Pfizer Ophthalmics Institute meeting was held Friday and Saturday, April 29 and 30, 2005, in Fort Lauderdale, Florida, in conjunction with the 2005 ARVO Annual Meeting. The Institute meeting was funded by The ARVO Foundation for Eye Research through a grant from Pfizer Ophthalmics. The goal of the Institute is to develop strategies to improve research and clinical care in areas of ophthalmology related to preventable vision loss and blindness. The specific topic and organizer were selected by an independent selection committee which reviewed proposals received through an open call for proposals related to angiogenesis, neovascularization, and vasoproliferation. The proposal organizer solicited an international, interdisciplinary panel of expert speakers that comprised the First ARVO/Pfizer Ophthalmics Institute Working Group . One outcome of the Institute is publication of the research presented and discussions that followed to share possible new opportunities for research with the vision community and its partners in other disciplines. 

Overview

Angiogenesis plays an important role in many disease processes throughout the body. It is often assumed that new vessel growth is identical in all contexts, so that if the molecular pathogenesis is defined in one setting, it will apply to all angiogenic disease processes. However, there is mounting evidence suggesting that the molecular pathogenesis may differ in angiogenesis occurring in different environments and/or as part of different disease processes. The first ARVO/Pfizer Institute meeting explored how ocular angiogenesis is similar and how it differs from angiogenesis elsewhere in the body. 

Role of Vascular Endothelial Growth Factor in Angiogenesis

VEGF is an important stimulator of endothelial cell proliferation, migration, and tube formation in vitro. It is essential for developmental angiogenesis, and the absence of even a single allele is embryonically lethal. ^{1 2} During development, endothelial cells are completely dependent on VEGF for survival. In adult animals, endothelial cells are not completely dependent on VEGF, but it still plays an important role in physiologic angiogenesis in the ovary and uterus and many types of pathologic angiogenesis, including that occurring in tumors, retinal and choroidal neovascularization, rheumatoid arthritis, and psoriasis (for review, see Ref. ³ ). Because VEGF is a stimulator that ocular neovascularization shares with many other types of neovascularization, it is a key molecule that deserves considerable attention. 

General Background Information Regarding VEGF

The VEGF family consists of several gene products, including VEGF-A, -B, -C, and -D and placental growth factor (PlGF)-1 and -2 (see Supplement 1). When VEGF is used without further specification, it is assumed to refer to VEGF-A. The VEGF-A gene has eight exons, and alternative splicing of its mRNA results in four relatively abundant isoforms (206, 189, 165, and 121 amino acids in humans and 205, 188, 164, and 120 amino acids in mice) and a few rare isoforms. The different isoforms have different properties ⁴ : VEGF₁₈₉ and VEGF₂₀₆ bind very tightly to heparin, and so they are sequestered in the extracellular matrix (ECM). VEGF₁₆₅ is the predominant isoform. It binds heparin less tightly and therefore has some solubility, but is also associated with the ECM. VEGF₁₂₁ lacks a heparin-binding domain and is very soluble. Other active moieties of VEGF are generated by proteolytic degradation. Plasmin cleaves VEGF₁₆₅, resulting in an N-terminal 110-amino-acid peptide that lacks a heparin-binding domain. VEGF₁₁₀ can associate with VEGF₁₆₅ to form VEGF₁₁₀-VEGF₁₆₅ heterodimers and then with more digestion, VEGF₁₁₀ homodimers can result. Matrix metalloproteinase (MMP)-3 cleaves VEGF₁₆₅, resulting in an N-terminal 113-amino-acid peptide. Inflammation causes increased levels of these enzymes and therefore increased levels of the VEGF fragments. 

Endocrine gland–derived (EG)-VEGF is an angiogenic protein that is structurally unrelated to VEGF and is a selective mitogen for adrenal endothelial cells with no effect on human umbilical vein endothelial cells. ⁵ It is expressed in steroidogenic tissues, such as adrenal gland, ovary, testis, and placenta. Like VEGF, it can induce fenestrae in endothelial cells. It is a mediator of tissue-specific angiogenesis; it induces angiogenesis in the ovary, but if it is expressed in the cornea, there is no effect. Prokinetisin 2 (BV8) is highly homologous to EG-VEGF. ⁶ It is selectively expressed in testis, bone marrow, and circulating leukocytes; can induce hematopoiesis; and is a potent chemoattractant for monocytes. Other tissue-specific angiogenic factors may exist and could explain differences in the angiogenic process among different tissues. 

The Role of VEGF Isoforms

The role of the various VEGF isoforms has been determined in part from investigations in mice genetically engineered to express only one isoform ⁷ (see Supplement 2 online). Mice that express only VEGF₁₂₀ have normal organ development until late in gestation, and then problems related to vascular insufficiency develop that are due to abnormal density of blood vessels. There is a deficiency in the branching of blood vessels, and microvessels are approximately 60% larger in diameter than normal. Mice that express only VEGF₁₈₈ have ectopic branching, but mice that express only VEGF₁₆₄ are normal. Mice that express both VEGF₁₂₀ and VEGF₁₈₈, but not VEGF₁₆₄, are also normal. Thus, it appears that both soluble and matrix-binding VEGFs are needed and that these can be VEGF₁₆₄ alone or the combination of VEGF₁₂₀ and VEGF₁₈₈. It has been hypothesized that heparin-binding VEGFs are needed to set up VEGF gradients, which promote pathfinding for vessels and branching morphogenesis. 

What is the role of the heparin-binding domain in the context of disease? In xenograft tumor models, tumor growth is best when VEGF₁₆₄ or a combination of VEGF₁₂₀ and VEGF₁₈₈ are available. In some way, the heparin-binding domain provides a growth advantage. Alanine-scanning mutation analysis of the heparin-binding domain of VEGF shows that three amino acids are necessary for heparin binding and that the same three amino acids are essential for leukostasis. VEGF proteins mutated in one of these three amino acids bind to VEGFR2, but have diminished interaction with VEGFR1, the receptor that mediates the proinflammatory effects of VEGF. ^{8 9} So, it appears that the heparin-binding domain helps to mediate inflammation and amplify the pathologic effects of VEGF₁₆₄. 

VEGF Receptors

VEGF interacts with multiple receptors. ³ There are three primary receptors, designated VEGFR1, -R2, and -R3. VEGFR3 binds VEGF-C and -D and mediates lymphangiogenesis (see Supplement 3 online). VEGFR2 is the major mediator of mitogenesis of endothelial cells. ¹⁰ It binds VEGF-A and proteolytic fragments of VEGF-C and -D; therefore, although VEGF-C and -D function primarily in lymphangiogenesis, they can also contribute to angiogenesis. VEGFR1 binds VEGF-A and -B and PlGF. The role of VEGFR1 is context dependent. In embryos and some adult tissues, it acts as a decoy receptor that suppresses angiogenesis ^{11 12} ; and, in some adult tissues, it mediates VEGF signaling and is proangiogenic. ⁹ In the eye, VEGFR1 is proangiogenic, and its inhibition can suppress retinal or choroidal neovascularization. ¹³  

VEGF also interacts with neuropilins (see Supplement 4 online), which were first identified as receptors for semaphorins that function in axon guidance (for reviews, see Refs. ¹⁴ , ¹⁵ ). Neuropilin 1 binds VEGF₁₆₅, but not the other isoforms, and neuropilin 2 binds VEGF₁₆₅ and VEGF₁₄₅, but not VEGF₁₂₁. ^{16 17} Neuropilin 2 also binds PlGF2 and VEGF-C. In early development, neuropilin 1 is expressed in arteries and neuropilin 2 is expressed in veins. ¹⁸ The expression pattern of neuropilin 1 is similar to that of ephrin B2, but starts earlier. ¹⁹ The absence or blockade of either neuropilin 1 or -2 suppresses ocular neovascularization. ^{20 21}  

Neuropilins do not have kinase domains and must complex with plexins and/or VEGF receptors that have a kinase domain for intracellular signaling to occur. Since class-3 semaphorins bind neuropilins, there is an interaction between these semaphorins and particular VEGF isoforms that mediates signaling in endothelial cells. Purified semaphorin 3F causes human umbilical vein endothelial cells to contract, and it inhibits VEGF-induced proliferation. This effect is not a result of competitive inhibition of VEGF by semaphorin 3F, but rather appears to be an active signal that counteracts the effects of VEGF. It also counteracts FGF2-induced proliferation. This effect of semaphorin 3F is blocked by siRNA directed against neuropilin 2, but not siRNA directed against neuropilin 1. Therefore, semaphorin 3F is a potential therapeutic agent for suppression of VEGF-induced neovascularization. There are also some stimulatory semaphorins. Semaphorin 4D is membrane bound and is released by proteolysis. It binds to its receptor, plexin B1, which then forms complexes with c-Met, the receptor for hepatocyte growth factor, and stimulates angiogenesis in some settings. Semaphorin 6D binds to plexin A1, which can form complexes with VEGFR2 and stimulate it. It is not known whether semaphorins 4B and/or -6D stimulate ocular angiogenesis. 

Transcriptional Regulation of VEGF Expression

Hypoxia inducible factor (HIF)-1 is a transcription factor that regulates a large battery of genes, including VEGF, that help tissues respond to oxygen deprivation ^{22 23} (see Supplement 12 online). HIF-1 is a heterodimer composed of HIF-1α and -1β. It binds to hypoxia response elements (HREs) in promoters and recruits the coactivators P300 and CREB-binding protein (CBP) to increase gene expression. ²⁴ HIF-1 causes increased expression of several proangiogenic factors, including, but not limited to, VEGF-A and -C, VEGF receptor 1, PlGF, platelet-derived growth factor (PDGF)-B, angiopoietin (Ang)-1, Ang2, and CXCR4. 

The levels of the HIF-1α subunit are tightly regulated by the cellular oxygen concentration via oxygen-dependent hydroxylation of two proline residues in HIF-1α. This reaction is catalyzed by a group of enzymes called prolyl hydroxylases (for review, see Ref. ²⁵ ). These enzymes use molecular oxygen as a substrate and have a high K _m for oxygen, so that oxygen is the rate-limiting substrate under physiological conditions. Under normoxic conditions, these residues are hydroxylated, and this hydroxylation is necessary for the binding of the Von Hippel-Lindau (VHL) tumor-suppressor protein, the recognition component of an E3 protein ligase that ubiquitinates HIF-1α and targets it for degradation by the proteosome. Under hypoxic conditions, oxygen becomes limiting, the proline residues are not hydroxylated, VHL does not bind, and HIF-1α accumulates within the cell. There is also an asparagine residue in the transactivation domain that is hydroxylated by an asparagine hydroxylase called factor-inhibiting HIF (FIH)-1. ²⁶ This hydroxylation prevents the interaction of the transactivation domain with the coactivators. Under hypoxic conditions, this hydroxylation does not occur, and the binding of the coactivators can occur. Thus, both the half-life of HIF-1 and its specific activity as a transcription factor are regulated in an oxygen-dependent manner. 

In addition to regulation by oxygen, a whole series of cytokines and growth factors can also increase HIF-1α levels, but rather than affecting degradation, they increase the synthesis of the protein. Insulin-like growth factor (IGF)-1 stimulates VEGF production through HIF-1α. ²⁷ The levels of HIF-1α protein correlate with VEGF mRNA levels, but there is no effect on HIF-1α mRNA levels, because the regulation occurs at the posttranscriptional level. The basis for the regulation of HIF-1α protein synthesis is through the stimulation of the PI3-kinase and mitogen-activating protein (MAP) kinase pathways that ultimately regulate the activity of eIF-4E, which is a critical regulator of cap-dependent mRNA translation. The signals transduced through these pathways stimulate translation of a select subset of mRNAs within the cell, including HIF-1α mRNA. ²⁸ This pathway is relevant to diabetic retinopathy, because there is activation in the diabetic retina of the Akt kinase and increased levels of HIF-1α and VEGF. Pharmacological blockade of the IGF-1 receptor dramatically reduces the levels of activated Akt, HIF-1α, and VEGF. Thus, there are at least two mechanisms for regulation of HIF-1α levels: hypoxia and growth factors. 

HIF-1α is essential for normal embryonic development. HIF-1α-knockout mice arrest in development at embryonic day (E)8.5 and die by day 10. ²⁹ The vasculature is established, but then it regresses. When wild-type mouse embryonic stem (ES) cells are exposed to 1% oxygen for 24 hours, there is strong induction of VEGF, Ang1, PlGF, and PDGF-B mRNA expression. ³⁰ In HIF-1α-null ES cells, there is a dramatic reduction in the expression of each of these mRNAs. In primary cultures of cardiac fibroblasts, cardiac myocytes, arterial smooth muscle cells, and arterial endothelial cells, expression of VEGF, Ang1, PlGF, and PDGF- B were compared when cultured at 1% versus 20% oxygen for 24 hours. Each of these cell types responds in a different manner in terms of the regulation of these genes. VEGF is uniformly upregulated, but the others are not. So, the response to hypoxia is cell type–dependent. This is a source of differences among tissues in angiogenic responses to hypoxia and growth factors that modulate HIF-1 levels. 

HIF-1α plays a role in the induction of VEGF in oxygen-induced ischemic retinopathy. ³¹ There are basal levels of HIF-1α at P7 in normoxia, and these levels are reduced when mice are exposed to hyperoxia. When mice are removed from the hyperoxic environment and their retinas become relatively hypoxic, HIF-1α levels increase and remain elevated for several days, during which time Vegf mRNA levels increase. Mice in which the HRE is deleted from the Vegf promoter (Vegf ^δ/δ mice) ³² do not have hypoxia-induced retinal neovascularization, indicating that HIF-1 plays an essential role. ³³ Because the pathophysiology of the mouse model of ischemic retinopathy closely mimics that of human ischemic retinopathies, HIF-1 is an important therapeutic target in those disease processes. Choroidal neovascularization from the laser-induced rupture of Bruch’s membrane is also markedly reduced in Vegf ^δ/δ mice, raising the possibility that HIF-1 may also be a target in patients with choroidal neovascularization. 

Other Mechanisms by Which VEGF Activity Is Modulated

VEGF gradients appear to be needed for ordered vascular growth or regrowth. When such gradients are disrupted, pathologic sprouting, which is complicated by hemorrhage and scarring, occurs. VEGF gradients can be disrupted by an imbalance among heparin-binding and soluble VEGF isoforms or by excessive proteolytic activity that can cleave heparin-binding domains and result in excessive amounts of soluble VEGF (see Supplement 2 online). There is specialization within vascular sprouts so that endothelial cells at the tip of sprouts, “tip cells, ” respond to VEGF by migrating toward high levels, whereas other cells within sprouts that are not at the tip, “stalk cells, ” respond to VEGF by proliferating ³⁴ (see Supplement 20 online). This specialization allows directed growth along VEGF gradients and provides another example of differential responsiveness to VEGF, even among endothelial cells in very close proximity. The molecular basis of this differential responsiveness is not yet understood, although it has been noted that tip cells express high levels of VEGFR2 relative to stalk cells. There are other differences as well, because tip cells produce PDGF-B and δ-like 4 notch ligand, whereas stalk cells do not. The manner in which tip and stalk cells are formed is an active area of research. A model of lateral inhibition involving notch signaling has been proposed (see Supplement 6 online). 

Directional Cues in Addition to VEGF in Vessel Formation

Gaining control over VEGF gradient formation and differentiation of sprout cells into tip and stalk cells could help to provide a means to replace pathologic angiogenesis with ordered vascular regrowth that remedies ischemia. However, it is likely that much more knowledge is needed before that goal is achieved, because new insights gained from embryonic vascular development suggest that vascular guidance is an extremely complex process that has many parallels with axon guidance (for review, see Ref. ³⁵ , and see Supplement 15 online). In addition to VEGF, four classes of molecules have been shown to provide directional cues: (1) netrin binding to deleted in colorectal carcinoma (DCC) or Unc5, (2) semaphorins binding to neuropilins and plexins, (3) ephrins binding to Ephs, and (4) slits binding to roundabout (Robo). In vessels, the primary Robo expressed is Robo4. RNAi knockdowns in developing zebrafish have helped to unravel how these ligand-receptor pairs act to achieve ordered vascular growth in early embryos. Although there may be similarities in vascularization of particular organ systems later in development, there are also likely to be differences that further increase complexity. 

Tissue-Specific Signaling Systems for Vessel Formation

Differences among organ systems in vascularization are demonstrated by the finding that norrin acts as a tissue-specific ligand for frizzled 4 (Fz4) to help direct retinal vascular development ³⁶ (see Supplement 8 online). Mutations in norrin result in Norrie disease, in which retinal vascular development is incomplete, resulting in large areas of avascular retina, neovascularization, scarring, and retinal detachment. Heterozygosity for mutations in Fz4, a presumptive Wnt receptor, results in familial exudative vitreoretinopathy (FEVR), in which retinal vascular development is perturbed, but often not as severely as in Norrie disease, resulting in areas of avascular retina, neovascularization, and a spectrum that ranges from focal tractional retinal detachments to severe detachments and blindness. Norrin binds to Fz4 with high affinity and specificity and induces downstream signaling, indicating that, despite the lack of any structural homology to Wnts, norrin functions as a ligand for Fz4. The mechanism by which the activation of Fz4 by norrin influences retinal vascular development is not yet known, but null phenotypes suggest that it helps to orchestrate growth of retinal vessels. This signaling system is not completely unique to the eye, because it is also essential for maintaining the vascularization of the inner ear, but is not known to function in any other organs. It is important to determine whether other organs, or groups of organs, have their own specific signaling systems that influence the vasculature. This interaction is another example of how the vasculature in different tissues can differ in important ways. It is not yet known whether norrin/Fz4 signaling has any function in the retinal vasculature of adults or plays a role in the pathophysiology of ocular neovascularization in adults, but if so, useful therapeutic targets may emerge that may have greater specificity than targets related to more generalized signaling systems. 

Tie Receptors and the Angiopoietins

Tie1 and Tie2 receptors are selectively expressed on vascular endothelial cells and are required for embryonic vascular development. ^{37 38} Mice with targeted disruption of Tie1, die between E14.5 and birth, with severe edema and hemorrhage, suggesting that Tie1 signaling promotes vascular integrity. ^{38 39} Mice with targeted disruption of Tie2 or mice expressing a dominant-negative Tie2 mutant, die between E9.5 and E10.5 and show a lack of development of the endothelial lining of the heart, a lack of remodeling of the primary capillary plexus to form more complex higher-order branching vessels, and a failure of vascular invasion of neuroectoderm. ³⁷ The first binding partner identified for Tie2, Ang1, binds with high affinity and initiates Tie2 phosphorylation and downstream signaling. ⁴⁰ Ang1 is a critical Tie2 agonist, because mice deficient in Ang1 die around E12.5 and show vascular defects similar to but less severe than those in Tie2-deficient mice. ⁴¹ The second Tie2-binding partner identified, Ang2, binds with high affinity, but does not stimulate phosphorylation of Tie2 in cultured endothelial cells. ⁴² In vitro, Ang2 acts as a competitive inhibitor of Ang1, because it decreases Ang1 binding to Tie2 and Ang1-induced phosphorylation. With regard to embryonic vascular development, Ang2 also acts as an Ang1 antagonist, because transgenic mice that overexpress Ang2 have a phenotype similar to Ang1-deficient mice. ⁴²  

Until recently, the ligand(s) for Tie1 has been unknown, but new information suggests that the angiopoietins may interact with Tie1. ⁴³ Ang1 does not bind to the Tie1 extracellular domain in solution, but if Tie1 is overexpressed on the surface of cells, addition of recombinant Ang1 (CompAng1) results in phosphorylation of Tie1. When Tie1 and Tie2 are coexpressed, they form heteromeric complexes, and CompAng1 induces greater phosphorylation of Tie1 than it does when Tie1 is expressed alone. The kinase domain of Tie2 must be intact for this enhancement of Tie1 activity to occur, which suggests that Tie2 participates in Tie1 signaling. In normal endothelial cells, Ang1 (or Ang4) stimulates phosphorylation of Tie1 and Ang2 blocks the Ang1-induced phosphorylation. Therefore, Ang1 and Ang4 are agonists for both Tie1 and Tie2, and Ang2 is an antagonist for both; they are important modulators of vessel maturation. 

Effects of VEGF in Various Tissues

Transgenic mice can be generated by using tissue-specific promoters to express VEGF or related molecules in a tissue and studying the effects. This can be combined with the tetracycline-inducible promoter systems (tet-on or tet-off) to control the temporal aspects of expression. Transgenic mice that overexpress PlGF or VEGF in the dermis under control of the keratin-14 (K14) promoter show enhanced acute inflammation ⁴⁴ (see Supplement 17 online). In a delayed-type hypersensitivity model, mice are made allergic, and a subsequent challenge leads to inflammation and edema of the ear. Swelling of the ear is measured to quantify inflammation. In a normal mouse, the initial inflammation peaks at 48 hours and resolves within a week. In PlGF overexpressors, there is increased inflammation and edema, but just as in normal mice, these do not persist. In contrast, in VEGF overexpressors, there is persistent inflammation. 

In homozygous K14/VEGF transgenic mice, spontaneous inflammatory lesions develop at 6 months of age that look very much like psoriasis with inflammation, induration, and scaly skin. Histologically they are indistinguishable from psoriasis, with hyperplastic epidermal keratinocytes that form fingerlike projections, inflammation with CD4 cells in the dermis and CD8 cells in the epidermis, and prominent angiogenesis. Thus, chronic overexpression of VEGF in the skin is sufficient to stimulate angiogenesis and a psoriasis phenotype. VEGF is also necessary in the maintenance of the disease, because treatment with a VEGF antagonist restores normal skin within 2 weeks. 

The tet-off system and a heart-specific promoter have been used to initiate VEGF expression in the hearts of adult mice ⁴⁵ (see Supplement 5 online), causing an influx of cells into the heart. When doxycycline was administered, halting the production of VEGF, the cells dissipated. These cells surrounded blood vessels and were demonstrated to be hematopoietic cells, because they stained for the pan-hematopoietic marker CD45. In some experiments, mice were given whole-body irradiation to eliminate all endogenous bone marrow cells, and then the bone marrow was reconstituted by transplantation of cells labeled with LacZ or GFP. Initiation of VEGF expression in the heart caused an accumulation of labeled cells around cardiac blood vessels. The cells were not incorporated into the endothelium, but rather remained in a perivascular location. Cell-type–specific stains demonstrated these cells to be monocytes that express CXCR4 and respond to stromal-derived factor (SDF)-1. When VEGF levels are increased, SDF-1 expression is induced around blood vessels, leading to the accumulation of monocytes, which contribute to the angiogenic response. 

Expression of VEGF in airway epithelium of adult mice using the Clara cell 10-kDa promoter and the tet-on system resulted in sprouting of new vessels from venules within 3 to 5 days of starting doxycycline. ⁴⁶ The vessels regressed when VEGF overexpression ceased. Regression of the newly formed vessels was accompanied by an influx of inflammatory cells. 

The rhodopsin promoter was used to express VEGF in the retina of transgenic mice (Rho/VEGF mice) ⁴⁷ (see Supplement 21 online). It initiates expression at P7, and by P10 endothelial cells begin to migrate from the deep capillary bed into the outer nuclear layer. By P14 there are vessels extending from the deep capillary bed into the subretinal space. At later time points, there is an extensive network of new vessels throughout the subretinal space. It is remarkable that only the deep capillary bed, and not the other vessels in the retina or choroid, participate in the sprouting. The deep capillary bed develops later than the other capillary beds, suggesting the possibility that a developmentally regulated gene product influences the process. To determine whether the timing of the onset of VEGF expression is critical in the determination of the phenotype, the tet-on system was used with one of two retina-specific promoters, the rhodopsin promoter or the interphotoreceptor retinoid binding protein (IRBP) promoter, resulting in two double-transgenic lines (Tet/opsin/VEGF and Tet/IRBP/VEGF), with doxycycline-inducible expression of VEGF in the retina. ⁴⁸ In adult Tet/opsin/VEGF or Tet/IRBP/VEGF mice, the retinas are completely normal until the mice are treated with doxycycline; and then, within 4 to 5 days, there is neovascular sprouting from the deep capillary bed. Thus, the phenotype is the same as that in Rho/VEGF mice in which VEGF expression starts at P7, in that only the deep capillaries respond to VEGF; however, the level of expression of VEGF is higher in the double-transgenic mice, resulting in severe neovascularization that ultimately causes retinal detachment. The endogenous IRBP promoter initiates expression during embryonic life, so Tet/IRBP/VEGF mice are able to express VEGF in the retina at P0 if doxycycline is given to nursing mothers who deliver it to the pups in breast milk. In litters treated between P0 and P10, there is a neonatal phenotype consisting of sprouting from the superficial capillary bed and dilation of the superficial vessels. ⁴⁹ When the doxycycline is started at P4, both the neonatal and adult phenotype occur between P10 and P14; the neonatal phenotype occurs in the peripheral retina, both occur in the midperipheral retina, and the adult phenotype occurs in the central retina. Because development occurs from the center to the periphery, the central retina is at a later developmental stage than the peripheral retina. This supports the model that a developmentally regulated permissive factor controls the neovascularization response. 

Interaction of VEGF with Angiopoietins

Several lines of evidence indicate that Ang2 is a developmentally and hypoxia-regulated permissive factor for VEGF-induced neovascularization in the retina. Expression of Ang2 in the retina increases during the first week after birth, peaks around P8, during development of the deep capillary bed, and then decreases in adults. ⁵⁰ Ang2/LacZ knockin mice show the spatial pattern of Ang2 expression in the retina. Between P0 and P7, there is Ang2 expression along the surface of the retina, and at P8 there is intense expression in the region of the deep capillary bed within horizontal cells that is maintained at a reduced level in adults. ⁵¹ In mice with oxygen-induced ischemic retinopathy, at P12, when the mice are removed from hyperoxia, the retina becomes hypoxic. Within hours, ectopic expression of Ang2 occurs at the surface of the retina. Within a few days, new vessels sprout from superficial vessels, and there is intense expression of Ang2 within and around the sprouts. Homozygous Ang2 knockouts show very poor development of the superficial capillary bed and almost no development of the deep capillary bed, indicating that Ang2 is essential for normal retinal vascular development. They also show persistence of the hyaloid vessels. 

Double transgenic Tet/opsin/ang2 and Tet/IRBP/ang2 mice with inducible expression of Ang2 have also helped to elucidate the effects of Ang2 in the eye. ⁵² During retinal vascular development, increased expression of Ang2 causes an increase in the density of the deep capillary bed at P12 that normalizes by P18. Expression of Ang2 in adult mice has no effect on normal retinal vessels. In mice with ischemic retinopathy, increased expression of Ang2 during the hypoxic period, P12 to P17, results in a marked increase in the amount of retinal neovascularization. In this model, VEGF levels reach a peak, plateau between P17 and P19, and then decline. Onset of Ang2 expression at P20 results in rapid regression of neovascularization. In Rho/VEGF transgenic mice or mice with choroidal neovascularization due to laser-induced rupture of Bruch’s membrane, high-level expression of Ang2 results in the regression of neovascularization. Thus, it is not just the presence of VEGF that seems to modulate the effect of Ang2, it is the ratio of Ang2 to VEGF. If the level of Ang2 is far above the level of VEGF, then new vessels regress, whereas mature vessels are unaffected. In Tet/opsin/ang2 mice, in the absence of doxycycline, injection of an adenoviral vector that expresses VEGF results in neovascularization of the cornea and iris, but no neovascularization of the retina except for a few sprouts in the area of needle penetration (due to a combination of VEGF and the injury). ⁵³ In the presence of doxycycline, which causes expression of Ang2, injection of the adenoviral vector expressing VEGF results in florid retinal neovascularization that originates from both the superficial and deep capillaries. 

Thus, retinal vessels require expression of both VEGF and Ang2 for neovascularization to occur. There is constitutive expression of Ang2 in the region of the deep capillary bed, and so if VEGF levels are increased alone, new vessels will sprout only from the deep capillary bed, but if Ang2 and VEGF are coexpressed, new vessels sprout from all capillary beds. HIF-1 increases expression of both Ang2 and VEGF in the retina, and as one would predict from the findings summarized so far, intravitreous injection of an adenoviral vector expressing a constitutively active form of HIF-1α causes sprouting of new vessels from all capillary beds in the retina. ³⁰ The situation is even more complicated in the choroid, because another permissive factor in addition to Ang2 is necessary for VEGF to induce choroidal neovascularization. ⁵⁴  

As already noted, normal vessels in the retina and choroid are not responsive to the regression-promoting effects of Ang2, but new vessels are responsive; high levels of Ang2 relative to the levels of VEGF are needed. This finding suggests that Ang2 may be a useful therapeutic agent for the promotion of regression of pathologic retinal or choroidal neovascularization, but its use alone would run the risk of stimulating neovascularization if levels of VEGF were sufficiently high. Therefore, it seems most prudent to use Ang2 or another Tie2 antagonist in combination with a VEGF antagonist. 

Transgenic mice with increased expression of Ang1 in skin under control of the K14 promoter exhibit a moderate increase in the number and a large increase in the diameter of dermal vessels. ⁵⁵ Transgenic mice with increased expression of VEGF in skin show a large increase in leaky dermal vessels, and double transgenic mice with coexpression of Ang1 and VEGF show an additive effect on angiogenesis, but the vessels do not leak spontaneously and are resistant to inflammation-induced leakage. ⁵⁶ Overexpression of Ang1 in the retina of transgenic mice suppresses the development of retinal or choroidal neovascularization. ⁵⁷ Ang1 also suppresses neovascularization and retinal detachment in double-transgenic mice with high-level expression of VEGF in the retina. ⁵⁸ Therefore, unlike the situation in skin where Ang1 is proangiogenic, Ang1 is a potent antiangiogenic agent in the retina—another example of a major difference between vascular beds in different tissues. Unlike Ang2, expression of Ang1 after neovascularization is already established does not cause regression of the new vessels, although it suppresses further growth. Therefore, increasing expression of Ang1 in the eye may be a good strategy for prevention of neovascularization. 

Endogenous Inhibitors of Angiogenesis

Various proteins in addition to Ang2 promote the regression of new vessels, including pigment epithelium–derived factor (PEDF), ⁵⁹ thrombospondin 1 (TSP1), ⁶⁰ angiostatin, ⁶¹ endostatin, ⁶² and many more (see Supplement 11 online). They may function as part of an endogenous system that regulates and turns off the angiogenesis that is necessary for wound repair. PEDF is a secreted glycoprotein and a member of the serpin family that lacks serine protease inhibitor activity. It is a multifunctional protein that has neurotrophic as well as angioinhibitory activity. Different parts of the PEDF molecule mediate these two different activities, and peptides corresponding to these regions retain the corresponding activity of native PEDF. ⁶³ PEDF is not essential for development, because PEDF-knockout mice are viable, but PEDF may function to regulate microvessel density in some organs. ⁶⁴ Knockout mice show increased microvessel density in the retina, prostate, and pancreas, but not in the kidney. This finding is somewhat controversial, because another PEDF knockout did not show increased microvessel density in the retina or pancreas (Renard et al. IOVS 2005:46:ARVO E-Abstract 3258). Whereas the role of endogenous PEDF in the eye is uncertain, several studies have demonstrated that intraocular injection of viral vectors that express PEDF or injection of recombinant PEDF suppresses retinal or choroidal neovascularization. ^{59 65 66 67} However, high doses of PEDF may have paradoxical effects because, whereas continuous infusion of low doses suppressed choroidal neovascularization, high doses were stimulatory. ⁶⁸ TSP1 is a large (180 kDa), secreted glycoprotein that, like PEDF, has multiple functions, including inhibition of angiogenesis, neuroprotection, and axon guidance. ⁶⁹ Unlike PEDF, the receptor through which TSP1 mediates its effects on endothelial cells is known. It is CD36, which is also a multiligand scavenger receptor. ⁷⁰ TSP1-knockout mice show increased vascular density during retinal vascular development. ⁷¹  

Both PEDF and TSP1 selectively induce apoptosis in endothelial cells participating in neovascularization, but have no effect on endothelial cells in mature vessels. It appears that proliferating endothelial cells express the cell death receptor Fas, which makes them susceptible to Fas ligand, which is induced by PEDF and TSP1. ⁷² Another possible determinant regarding the occurrence of apoptosis or survival of endothelial cells is NFATc2, a member of the nuclear factor of activated T cells (NFAT) family of transcription factors related to the Rel/NF-κB family (for review, see Ref. ⁷³ ). Activation of NFATc2 by dephosphorylation has been implicated in VEGF-induced corneal neovascularization. ⁷⁴ NFAT activation is disrupted by PEDF and TSP1, which cause rephosphorylation of NFAT by c-Jun N-terminal kinase (JNK) kinases. These kinases are activated by PEDF, but only in activated endothelial cells. VEGF or FGF2 cause a mild activation of JNK kinases that is greatly augmented by PEDF. Thus, in activated but not quiescent endothelial cells, PEDF and TSP1 induce Fas ligand and inactivate NFATc2, both of which contribute to induction of apoptosis. 

PEDF and TSP1 are proteins that have domains with different activities that are fully functional in the context of the intact proteins. Several proteins, such as angiostatin and endostatin, have domains that acquire antiangiogenic activity only after proteolytic cleavage. To develop reagents that can recognize proteolyzed or denatured forms of collagen, but not the native triple helical form, subtractive immunization has been used to generate monoclonal antibodies ⁷⁵ (see Supplement 9 online). One antibody, HU426, recognized a cryptic site in collagen IV that is identifiable in tumor vessels, but not normal vessels, and that stimulates angiogenesis by binding to αvβ3 integrin on endothelial cells. ⁷⁶ This collagen IV fragment also binds αvβ3 on tumor cells to confer a growth advantage by downregulating expression of TSP1 and insulin-like growth factor–binding protein (IGFB)-4. 

Proteolytic degradation can generate both stimulators and inhibitors of angiogenesis, so alteration of proteolytic activity may have different effects, depending on the setting, which may explain the initially puzzling observation that overexpression of tissue inhibitor of metalloproteinases (TIMP)-1 inhibits angiogenesis in some tissues, but stimulates it in the retina. ⁷⁷ In 1994, when it was found that mutations in tissue inhibitor of metalloproteinases (TIMP)-3 that generate unpaired cysteine residues at the C terminus of the protein cause Sorsby fundus dystrophy (SFD), ⁷⁸ it was widely assumed that the pathogenic mutations increase the susceptibility of Bruch’s membrane to vascular invasion by compromising the ability of TIMP3 to block proteolytic activity (see Supplement 10 online). However, expressed mutant TIMP3 proteins retain proteolytic activity, arguing against this hypothesis. ⁷⁹ The mutant protein accumulates in the ECM in vitro and accumulates in Bruch’s membrane in vivo, perhaps contributing to pathogenesis, ^{79 80 81} but it is not clear whether this is a nonspecific effect that would be mimicked by a large excess of any protein in Bruch’s membrane, or whether there are intrinsic, gain-of-function properties of the accumulated mutant TIMP3 that are contributory. Excess mutant TIMP3^S156C in the ECM of fibroblasts alters their morphology in a way that excess wild-type TIMP3 does not, but it is not clear how this relates to pathogenicity in vivo. ⁸⁰ TIMP3 has antiangiogenic activity unrelated to its ability to block metalloproteinases, ⁸² but loss of this activity does not result in choroidal neovascularization or other features of SFD, as suggested by Timp3-knockout mice, which do not reveal a retinal degeneration–choroidal neovascularization phenotype. A strategy that may provide important insights in the future is the generation of mouse models in which the Cre-Lox system is used to substitute different types of Timp3 mutations for the wild-type allele. 

Stimulators Other than VEGF

Members of the VEGF family other than VEGF-A also stimulate neovascularization and excessive vascular permeability, and like VEGF-A, their roles in various tissues may vary. PlGF may have particular importance in the retina, because knockout of PlGF or neutralization with an anti-PlGF antibody substantially suppresses retinal neovascularization. ⁹ This is a crucial issue that may have therapeutic implications, because VEGF antagonists vary in their selectivity. Based on data generated from animal models, one would predict that agents that block all VEGF family members would be more efficacious than more selective antagonists in patients with choroidal neovascularization and/or macular edema or retinal neovascularization due to ischemic retinopathies, but results of clinical trials are needed before this can be stated with confidence. 

Pericytes also contribute to neovascularization. Their recruitment, proliferation, and survival are stimulated by PDGF-B (see Supplement 6 online). PdgfB-knockout mice die during embryonic development, with severe vascular defects and an almost complete lack of pericytes. ⁸³ To assess the role of PDGF-BB in adult animals, conditional PdgfB knockouts were generated by engineering LoxP sites around the PdgfB gene and using a Tie1 promoter to express Cre recombinase in endothelial cells. ⁸⁴ The efficiency of recombination was assessed by quantitative RT-PCR for PdgfB mRNA in isolated capillary fragments. The recombination rate was found to vary and in some mice was fairly low, whereas in others it was as high as 90%. The mice with very low levels of PdgfB mRNA in capillary fragments survived into adulthood and were crossed with a LacZ reporter strain (XlacZ4) in which the pericytes are labeled. In control mice, there was heavy staining in arteries and veins, whereas in mice with low PdgfB mRNA in capillary fragments, there was a marked reduction in coverage of arteries and veins with smooth muscle cells. Retinopathy developed in some of these mice, in which some areas of the retina were normal, some showed dilated vessels and capillary dropout, and some showed retinal neovascularization breaking through the internal limiting membrane into the vitreous cavity. Areas of retina showing proliferative changes in the inner retina often showed evidence of traction and rosette formation in photoreceptors and neovascularization growing into the vitreous. This phenotype is similar to that in patients with proliferative diabetic retinopathy. The severity of retinopathy correlated with the amount of pericyte loss. The threshold for development of retinopathy appeared to be reduction of pericyte coverage by approximately 50%. 

Transgenic mice that express high levels of PDGF-B in the retina show excessive migration and proliferation of pericytes, endothelial cells, and glial cells, resulting in severe proliferative retinopathy and retinal detachment. ^{85 86} Therefore, either deficiency or an excess of PDGF-B can result in retinopathy. 

It is likely that other proangiogenic factors in addition to the VEGF and PDGF family members participate in the stimulation of angiogenesis, but they may play larger roles in some tissues and diseases than others. The cytokines IL8 and TNFα are highly expressed in the setting of inflammation and are capable of stimulating angiogenesis. They have been implicated, along with VEGF, in stimulation of angiogenesis in rheumatoid arthritis (see Supplement 19 online). The histology of normal synovium shows that it is very loose tissue, but rheumatoid synovium is much denser, because it is packed with blood vessels often surrounded by leukocytes. 

Inflammation and Neovascularization

As noted earlier, increased expression of VEGF in some organs, such as skin and heart, results in an influx of mononuclear cells that contributes to the angiogenic response. The proinflammatory effects of VEGF are mediated through VEGFR1, but the exact mechanisms involved are not clear. In the heart, high levels of VEGF induce expression of SDF-1 in a perivascular location, and this expression recruits monocytes that express CXCR4, the receptor for SDF-1. The monocytes release stimulators of angiogenesis that augment the effects of VEGF and other stimulators produced by heart cells. Inhibitors of CXCR4 or SDF-1 reduce VEGF-induced recruitment of monocytes and thereby reduce the load of angiogenic stimulators in the tissue. Thus, CXCR4 and SDF-1 provide additional targets for therapeutic intervention. Although a mononuclear influx has not been recognized as a prominent feature of increased expression of VEGF in the retina, participation by mononuclear cells has not been excluded. Furthermore, CXCR4 antagonists have been shown to suppress the development of neovascularization in Rho/VEGF transgenic mice, mice with ischemic retinopathy, and mice with rupture of Bruch’s membrane (Lima e Silva R, et al. IOVS 2005;46:ARVO E-Abstract 1412). Inflammation seems to play a substantial role in ischemic retinopathy, because mice depleted of leukocytes show reduction in ischemia-induced retinal neovascularization. ⁸⁷  

The promoter for the CXCR4 gene has an HRE, and levels of CXCR4 are increased by HIF-1. Another family of transcriptional regulators, Id (inhibitors of differentiation) proteins, may also participate in the regulation of CXCR4 and other proteins that influence angiogenesis (see Supplement 13). Id proteins are dominant-negative helix-loop-helix (HLH) transcription factors. Normally, HLH transcription factors bind as dimers primarily to E-boxes and often regulate genes needed for differentiation. Id proteins lack the basic domain, cannot bind DNA, and inhibit transcriptional activation (for review, see Ref. ⁸⁸ ). Id1, -2, and -3 are expressed ubiquitously, whereas Id 4 is expressed in neurons. Mice that lack both Id1 and -3 die, with defects in neurogenesis and angiogenesis in the brain. In xenograft tumor models, heterozygous double knockouts are unable to support the growth of any of three types of tumor cells due to defective angiogenesis. ⁸⁹ A possible explanation is that Id1 and -3 downregulate TSP1 and in the absence of Id1 and -3, TSP1 levels are increased, and angiogenesis is inhibited. ⁹⁰ In this sense, Id proteins are stimulators of angiogenesis. However, Id proteins can also act as inhibitors of angiogenesis, because they downregulate CXCR4. ⁹¹ Therefore, the effect of Id proteins on angiogenesis depends on which of these two opposite effects predominates in any particular setting. This context-dependent effect of Id proteins is another source of variability in angiogenic processes in different tissues and different diseases. 

As noted earlier, hypoxia is a major driving force in angiogenesis in several tissues and several diseases. Inflammation and other types of stress also upregulate VEGF and other proangiogenic molecules. This effect is mediated primarily by the NF-κB family of transcription factors, which consists of a group of both activating and inhibiting transcription factors (see Supplement 14). 

Survival-Promoting Activity of VEGF

VEGF not only drives vascular sprouting and growth, it is also an important survival factor for endothelial cells and plays an important role in the maintenance of blood vessels. ⁹² Gaining control over vascular sprouting is critical in the prevention of neovascularization, whereas targeting the maintenance of new vessels is critical in the treatment of angiogenic blood vessels that are present at the start of treatment. Endothelial cells within new vessels tend to be more dependent on VEGF for survival than those in normal vessels, but this difference is relative (see Supplement 16). In fact, there are differences regarding VEGF dependence among various types of new vessels and also among mature vessels in different tissues. ^{93 94 95} Many normal fenestrated capillaries, which are abundant in endocrine glands and the gastrointestinal tract, are dependent on VEGF for survival. The choriocapillaris consists of fenestrated capillaries, and their response to long-term blockade of VEGF should be investigated. The vasculature of some tumors, such as renal cell carcinoma, is dependent on VEGF for survival; therefore, renal cell carcinoma is a particularly good target for VEGF antagonists. Treatment with VEGF antagonists in tumor animal models results in regression of tumor vessels, but the proportion of vessels that regress varies among tumor types. ⁹⁶ Despite the dependency of some normal capillaries on VEGF, in general, tumor vessels appear to have greater sensitivity to VEGF antagonists than normal vasculature. ⁹⁴ As vessels regress, they leave behind basement membrane sheaths and pericytes. Stopping the VEGF antagonist can be followed by explosive regrowth of tumor vessels along the sheaths. By contrast, treatment with antagonists of PDGF-B kills pericytes on some tumor vessels (but not normal vessels), and this effect is followed by the death of endothelial cells and vascular regression, suggesting that pericytes provide survival signals. For blood vessels in some tumors, the combination of antagonists for both VEGF and PDGF-B may be more efficacious than using a VEGF antagonist alone. ⁹⁷ Thus, the microenvironment of new vessels, including ECM (which provides signaling through integrins and also binds growth factors) and the surrounding cells, is a source of survival cues that make some vessels less dependent on VEGF. However, the interaction between new vessels, ECM, and pericytes is quite complex, in that the mere presence of basement membrane and pericytes on blood vessels does not guarantee their survival when VEGF is withdrawn or antagonized. ^{46 93 94 95}  

Although VEGF-independent survival signals are much better established in mature normal vessels that in tumor vessels, VEGF antagonists cause regression of a percentage of normal fenestrated capillaries in some organs. ⁹⁴ For instance, treatment with VEGF antagonists for 3 weeks results in regression of roughly 50% of capillaries in the thyroid. Roughly 20% of tracheal capillaries regress during treatment, and this is a possible explanation for why many patients treated with systemic blockade of VEGF experience hoarseness or a change in voice quality. When the VEGF blockade is stopped, most capillaries regrow, and permanent sequelae have not been identified. Another organ that may contain a substantial proportion of VEGF-dependent vessels in adults is the lung (see Supplement 18). Although newly developed retinal vessels are dependent on VEGF for survival, VEGF dependence has not been demonstrated for any mature ocular vessels. However, a systematic study of the effects of long-term VEGF inhibition, particularly on fenestrated capillaries of the choriocapillaris is needed. Whereas retinal neovascularization appears to be dependent on VEGF, choroidal neovascularization appears to be less so. Patients treated for a year with repeated intraocular injections of pegaptanib, an aptamer that specifically binds VEGF₁₆₅, did not experience regression of choroidal neovascularization; instead there was continued growth, but at a slower rate than in placebo-treated patients. ⁹⁸ In mouse models, combined antagonism of VEGF and Tie2 causes regression of choroidal neovascularization and accelerates regression of retinal neovascularization. ⁵² Blockade of Tie2 most likely disrupts survival signals other than those from VEGF. 

New vessels, like normal vessels, are likely to gain new sources of survival signals as they mature. This ability is likely to differ, depending on the microenvironment. The tumor environment is so abnormal that tumor vessels have multiple abnormalities, which may include relatively high and prolonged VEGF dependency. ⁹⁹ Choroidal new vessels that penetrate Bruch’s membrane are rapidly surrounded by hyperplastic retinal pigmented epithelial (RPE) cells, which may be a source of survival signals not available in other types of neovascularization. Differences in local populations of cells within tissues could be a major source of tissue-specific features of angiogenesis. 

Vascular Regression

Certain vascular beds are needed at one stage in development and not at later stages and are eliminated by programmed vascular regression. Understanding the molecular signals that regulate physiologic vascular regression may provide insights that will allow development of regression-inducing treatments of pathologic neovascularization (see Supplement 7). Three related vascular beds in the newborn rodent eye—the pupillary membrane, the tunica vasculosa lentis, and the hyaloid vessels—undergo programmed vascular regression and provide useful model systems. In mice, the hyaloid vessels regress between P3 and P9. Study of the hyaloid system is facilitated by the ability to visualize the vessels in vivo and the ability to analyze the entire vasculature en bloc in wholemounts. 

There are a small number of resident macrophages in the pupillary membrane—300 to 400 macrophages for a vessel network that contains 6000 to 7000 vascular endothelial cells. It is well established that macrophages play a passive role in recognizing and engulfing apoptotic cells. They simply engulf cells after there has been a macrophage-independent signal for cell death, but there is also evidence suggesting that macrophages may play an active role in inducing apoptosis. Some of that evidence was obtained by observations in PU.1 mutant mice. PU.1 is a transcription factor that plays a role in hematopoiesis. Deleting the PU.1 gene eliminates mature tissue macrophages, and PU.1 mutant mice have persistent hyaloid vessels, indicating that macrophages are essential for hyaloid vessel regression. ¹⁰⁰  

The canonical Wnt pathway is also necessary for hyaloid vessel regression. The receptor complex for Wnts consists of the frizzled proteins and LRP5 or LRP6 coreceptors. After Wnt ligands bind to the receptor complex, signal transduction events result in the stabilization of β-catenin, which then complexes with transcription factors of the LEF-TCF class to regulate gene expression. Wnt signaling has central roles in development and is activated in tumorigenesis. It is also essential for hyaloid vessel regression, since mice with a mutation in LRP5, show no hyaloid vessel regression because of a failure of cell death pathways. ¹⁰¹ Mutations in LEF1 give a similar phenotype. These data suggest that the canonical Wnt pathway is necessary for hyaloid regression. Wnt signaling causes cell-cycle entry, which enhances susceptibility to apoptosis. Ang2 contributes by causing withdrawal of survival signals after cell-cycle entry. ¹⁰² Macrophages contribute by producing Wnt7B, and pericytes produce Ang2. At present, the working model for regression of the hyaloid vasculature is the following. Macrophages deliver Wnt7B locally to vascular endothelial cells by cell–cell contact, and Wnt7B stimulates the endothelial cells to enter the cell cycle. Expression of Ang2 by pericytes causes withdrawal of survival signals; and, without those signals, endothelial cells that enter G₁ undergo programmed cell death. This model is consistent with findings in models of ocular neovascularization. ⁵² Increased expression of Ang2 causes regression of retinal or choroidal neovascularization in which endothelial cells are proliferating, but it has no effect on quiescent endothelial cells in mature vessels. 

Vascular Targeting

New vessels undergo maturation over time by acquiring survival signals from the ECM and surrounding cells. As this happens, a substantial percentage of endothelial cells in the new vessels may exit the cell cycle and become less susceptible to the regressive effects of Ang2 or other agents. Despite the relative maturation of the new vessels, their environment may still be abnormal and may promote altered gene expression (see Supplement 22). This differential gene expression can be exploited for selective targeting of the new vessels, an approach commonly called vascular targeting. ^{103 104} Vascular targeting agents are made up of two parts: a targeting moiety that is usually an antibody or growth factor that binds to a specific marker and an effector moiety that induces the therapeutic effect, typically damage or occlusion of new vessels. Some markers that have been used successfully for vascular targeting of tumors include (1) angiogenesis and remodeling markers, such as VEGF receptors, α_vβ₃ integrin, the extra domain B (ED-B) of fibronectin, endoglin, MMP2 and -9, aminopeptidase N (CD13), and prostate-specific membrane antigen (PSMA); (2) cell adhesion molecules, such as vascular cell adhesion molecule (VCAM)-1, E-selectin, and VE cadherin; and (3) phosphatidylserine (PS), a lipid that is normally located on the internal surface of cell membranes and is flipped to the external surface in response to oxidative stress. Annexins and β2-glycoprotein I, which bind to the exposed PS, can also serve as tumor vessel markers. Some effectors that have been used successfully in rodent model systems include (1) toxins such as diphtheria toxin, ricin, and gelonin; (2) coagulation-inducing proteins such as tissue factor; (3) the cytokines IL2, IL12, and TNFα; (4) apoptosis-inducing factors, such as the RAF-1 gene product and the mitochondrial membrane disrupting peptide; (5) radioisotopes, both α and β emitters; (6) liposome-encapsulated effectors such as doxycyclineorubicin; and (7) photosensitizers such as SnChe₆. 

Effectors are not always needed, because antibodies alone can induce an immune response that destroys the new vessels. Tarvacin is a chimeric anti-PS antibody with a murine Fv region that binds to PS in a β2-glycoprotein 1-dependent fashion, and to human IgG1-constant regions. Treatment of mice with a murine version of tarvacin markedly reduces tumor growth in multiple rodent models, especially when combined with irradiation or cancer chemotherapy. ^{105 106} Tarvacin recruits macrophages that damage the endothelium and cause occlusion of tumor vessels. Phase I clinical trials are in progress. Another approach is ligand-targeted liposomal chemotherapy. ¹⁰⁷ PEGylated liposomes, loaded with doxyrubicin and coated with Asn-Gly-Arg peptides that recognize CD13 on activated endothelial cells, are selectively internalized by tumor vessels, resulting in doxyrubicin-induced damage that causes vessel occlusion and regression. An approach with particular relevance to the eye is the use of a VEGF₁₂₁/recombinant gelonin (VEGF/rGel) fusion protein. ¹⁰⁸ The expression of VEGF receptors is markedly increased in the endothelium of tumor vessels and choroidal neovascularization. VEGF₁₂₁, a soluble isoform of VEGF, does not bind heparin and therefore has little nonspecific binding to tissue. It therefore localizes to regions of high VEGF receptor density. After intravascular injection, VEGF/rGel selectively localizes to choroidal new vessels and causes them to regress. Intraocular injection of 5 ng of VEGF/rGel, a dose 1000-fold lower than the systemic dose, causes selective destruction of retinal or choroidal new vessels with no damage to normal vessels. 

The ED-B domain of fibronectin has been identified as an excellent target for homing to new vessels ¹⁰⁹ (see Supplement 23). Fibronectin is a very abundant protein that normally occurs without this extra domain B. However, ED-B is inserted by alternative splicing whenever there is tissue remodeling. It is a small domain of 91 amino acids that is identical in mouse, rat, rabbit, dog, monkey, and human. In a normal adult, the ED-B domain occurs only in the endometrium in the proliferative phase and in some vessels in the ovary. A high-affinity antibody to ED-B (L19) specifically localizes to tumor vessels. L19 antibodies have been used in clinical trials to image solid tumors, and more than 30 derivatives of L19 have been produced (e.g., fused to TNFα, VEGFs, IL12, interferon γ, truncated tissue factor, IL2, drugs, enzymes, or photosensitizers). 

Identification of novel targets for vascular targeting is still a high priority. A variety of techniques have been used, such as in vivo screening of phage-displayed peptide libraries, ¹¹⁰ serial analysis of gene expression, ¹¹¹ proteomic analyses, ^{112 113} and in vivo biotinylation of accessible normal proteins versus target proteins that are specifically expressed in a particular tissue or disease process. ¹¹⁴ These techniques have uncovered several proteins that are differentially expressed on new vessels, as well as many proteins that are selectively expressed in normal vascular beds. It is clear from this work that there are differences in gene expression among vascular cells in different tissues, and it is likely that this contributes to differences in responses to angiogenic stimuli. 

Summary and Conclusions

There are several pieces of evidence that suggest that neovascularization differs depending on its location within the body and the underlying disease process. 

EG-VEGF and BV8 stimulate angiogenesis in some tissues and not others. They may be unique or there may be other tissue-specific stimulators of angiogenesis that have not yet been identified. Their existence indicates that the “formula” for angiogenesis may have different “ingredients” in different tissues. 

The norrin/Fz4 ligand-receptor pair controls organ-specific vascularization in the retina and inner ear, which suggests that a signaling system used throughout the body may have a structurally unrelated ligand in one or two tissues that use the system for a specific purpose—perhaps adding to local diversity in regulation of the vasculature. 

HIF-1 is a central player, because it upregulates several proangiogenic proteins, but it does not upregulate the same ones in all cells; therefore, it may have somewhat different effects in different tissues. 

Id proteins are transcription factors that downregulate the antiangiogenic TSP1 and the proangiogenic receptor CXCR4. Their effect on angiogenesis in a particular tissue or disease process varies depending on which of these two opposite effects predominates. 

The local environment influences gene expression in endothelial cells. Endothelial cells participating in neovascularization express proteins that are not expressed in endothelial cells of normal vessels, forming the basis for vascular targeting. Even normal vascular cells in different tissues express different proteins, which has led to the concept of “molecular ZIP codes.” Differences in gene expression underlie differences in cell response to various signaling molecules and are likely to contribute to different responses to proangiogenic and antiangiogenic molecules in different vascular beds. 

There are many examples of vascular cells in different tissues, or sometimes in the same tissue, that respond to the same stimulus in different ways. In some cases, the mechanism is known or suspected, and in other cases it is not.

1.  
   Cells can be programmed to respond to the same angiogenic stimulus in different ways. This programming is exemplified by tip and stalk cells, specialized endothelial cells that occur in close proximity and respond to VEGF in different ways.
2.  
   A receptor expressed in different cells may act differently. For example, in vascular endothelial cells, Ang2 blocks phosphorylation of Tie2; but when Tie2 is expressed in other cell types, Ang2 promotes phosphorylation of Tie2 rather than blocking it. It is not known whether such differences also exist among different types of endothelial cells.
3.  
   Increased expression of VEGF can stimulate sprouting of new vessels from some vascular beds, but not others. Permissive factors are needed for VEGF to induce sprouting, and in some vascular beds they are constitutively expressed.
4.  
   Increased expression of Ang1 promotes neovascularization in skin and suppresses it in the retina and choroid. The mechanism causing this difference is not known.
5.  
   Increased expression of TIMP1 blocks neovascularization in some tissues, but stimulates it in the retina. A possible explanation for the different effects of proteinases or proteinase inhibitors in different settings is that proteolytic cleavage of ECM or ECM-associated proteins can yield both stimulators and inhibitors of angiogenesis, and one or the other may predominate, depending on the specific makeup of the ECM in a tissue.
6.  
   Cell types that are unique to a certain tissue, such as the RPE, may influence new vessel growth or regression, adding to local differences.

The tissue-specific aspects of angiogenesis have several important implications. 

It should not be assumed that experiments in chick chorioallantoic membrane, the cornea, or tumor models predict what will happen with regard to retinal and choroidal neovascularization. 

Normal retinal vascular development is, at best, an imperfect model of retinal neovascularization in adults. Although these processes have some similarities, they also have many differences, and effects of drugs or gene products on retinal vascular development may not predict effects on retinal neovascularization. Likewise, just because one VEGF antagonist inhibits retinal vascular development and another one does not, it does not follow that the latter one is safer in adults. In several respects, mature retinal vessels in adults do not behave like newly developed retinal vessels in neonates. 

The potential for developmental stage-specific effects on ocular vessels should not be overlooked. Increased expression of VEGF in RPE cells during embryonic life results in thickening of the choroid due to increased developmental growth, but if VEGF is expressed in the RPE in adult animals, there is no phenotype. It appears that embryonic choroidal vessels are responsive to VEGF, but adult choroidal vessels are not. This is similar to the developmental window between P0 and P7 when the superficial capillaries of the retina are responsive to VEGF. 

Also, although increased expression of VEGF does not cause sprouting of new vessels from adult choriocapillaris, it does not mean that that VEGF does not provide survival signals to the choriocapillaris in adults. VEGF is essential for the maintenance of fenestrated capillaries in several organs, and since the choriocapillaris is fenestrated, the effects of long-term VEGF blockade should be studied. 

Caution should be exercised in designing clinical trials to investigate a drug for retinal or choroidal neovascularization based on clinical or preclinical results in other vascular beds. For example, based on the effects of interferon α2a in patients with cutaneous hemagiomas and results in a monkey model in which interferon α2a inhibited iris neovascularization, ¹¹⁵ it was hypothesized that interferon α2a would also inhibit choroidal neovascularization. However in a large trial, patients with neovascular AMD treated with interferon α2a did worse than patients treated with placebo. ¹¹⁶  

Although it is important to identify differences among different types of neovascularization, it is equally important to identify similarities. The central role of VEGF as a stimulator and survival signal in most types of neovascularization makes it a major therapeutic target. VEGF antagonists have been found to provide benefit for tumor angiogenesis and choroidal neovascularization, and several new VEGF antagonists are being tested for each indication. Another VEGF family member, PlGF, has been implicated as a stimulator of both tumor and ocular angiogenesis, and two other family members, VEGF-C and -D, function primarily as stimulators of lymphangiogenesis, but they can also stimulate angiogenesis. Thus, it is reasonable to attempt to neutralize all members of the VEGF family in the treatment of retinal and choroidal neovascularization. VEGF antagonists are not likely to be displaced in the treatment of choroidal neovascularization, but rather will serve as the foundation to which other drugs are added. Likely candidates are antagonists of Tie2, PDGF-B, and integrins, to eliminate survival signals for new vessels and, we hope, allow for regression. Because HIF-1 upregulates several angiogenic factors, it is also an appealing target, because antagonizing it should resemble combination treatment. Finally, unlike many organs, the eye affords good opportunities for local, sustained delivery. In addition to identifying molecular targets and developing good antagonists, a critical challenge for the future is to determine pharmacokinetics with different modes of administration and to optimize delivery. 

Appendix AA

Disclosures

R.M. Alani, None; K. Alitalo, ImClone (F), Dyax (F); P. Brooks, CancerVax Corp. (C, P); P.A. Campochiaro, Alcon (F, C, R), Alimera (F, C, R), Genentech (F, C, R), GlaxoSmithKline (F, C, R), Novartis (F, C, R), GenVec (F, R), Sirna (F, C, R), Merck (F), Regeneron (F, C, R), Protein Design Labs (C, R); P. Carmeliet, (P); R. de Martin, None; M. Detmar, None; N. Ferrara, Genentech (E); M. Fruttiger, None; M. Hellstrom, AngioGenetics Sweden AB (I, E, P); E. Keshet, None; A. Koch, Pharmacia (C), Signature Biologicals (C), Novartis Pharmaceuticals (C), Medarex, Inc. (C), Forest Laboratories, Inc. (C); R.A. Lang, None; D.M. McDonald, Archemix Corp. (F), Pfizer (F); J. Nathans, None; D. Neri, Philogen (F, I, C, P, R); G. Neufeld, None; G. Semenza, None; D.T. Shima, (OSI) Eyetech (E, P); P. Thorpe, Peregrine Pharmaceuticals, Inc. (I, C, P, R); R.M. Tuder, None; O.V. Volpert, Abbott Pharmaceuticals (F); B.H.F. Weber, None 

Appendix AB

Online Supplements

Supplement 1. Napoleone Ferrara: VEGF and EG-VEGF 

Supplement 2. David Shima: VEGF Gradients 

Supplement 3. Kari Alitalo: Lymphangiogenesis 

Supplement 4. Gera Neufeld: Neuropilins and Semaphorins 

Supplement 5. Eli Keshet: Tissue-Specific Expression or Knockdown of VEGF 

Supplement 6. Mats Hellstrom: Platelet-Derived Growth Factors (PDGFs) and Perivascular Cells 

Supplement 7. Richard Lang: Macrophages, Wnts, and Programmed Vascular Regression 

Supplement 8. Jeremy Nathans: Norrin and Fz4: A Ligand-Receptor Pair that Controls Capillary Growth in the Developing Mammalian Retina 

Supplement 9. Peter Brooks: Integrins and Extracellular Matrix 

Supplement 10. Bernhard Weber: Abnormalities in Vessel Formation in a Mouse Model of Timp3 Deficiency 

Supplement 11. Olga Volpert: Endogenous Protein Inhibitors of Angiogenesis 

Supplement 12. Gregg Semenza: Hypoxia and HIF-1 

Supplement 13. Rhoda Alani: Id1 Regulation of Angiogenesis 

Supplement 14. Rainer de Martin: Inflammation and NF-κB 

Supplement 15. Peter Carmeliet: Mechanisms in Vessel Pathfinding 

Supplement 16. Donald McDonald: Tumor Angiogenesis 

Supplement 17. Michael Detmar: Wound Repair and Angiogenesis in Skin 

Supplement 18. Rubin Tuder: Role of VEGF in Maintenance of the Pulmonary Microcirculation and the Etiology of Emphysema 

Supplement 19. Alisa Koch: Angiogenesis in Rheumatoid Arthritis 

Supplement 20. Marcus Fruttiger: Developmental versus Pathologic Retinal Neovascularization 

Supplement 21. Peter Campochiaro: Ocular Neovascularization 

Supplement 22. Philip Thorpe: Vascular Targeting Agents and Strategies 

Supplement 23. Dario Neri: Targeting Specific Vascular Beds 

²  Members of the ARVO/Pfizer Institute Working Group are listed on page 463.

 Supported by The ARVO Foundation for Eye Research.

 Submitted for publication November 1, 2005; revised November 22, 2005; accepted December 19, 2005.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

First ARVO/Pfizer Institute Working Group

Rhoda Alani Johns Hopkins University, Baltimore, MD

Kari Alitalo, Biomedicum Helsinki, Helsinki, Finland

Peter Brooks, New York University, New York, NY

Ruth Caldwell, Medical College of Georgia, Augusta, GA

Peter A. Campochiaro, Johns Hopkins University, Baltimore, MD

Peter Carmeliet, University of Leuven, Leuven, Belgium

Pier Paolo Claudio, Temple University, Philadelphia, PA

Robert D’Amato, Children’s Hospital, Harvard, Boston, MA

Arup Das, University of New Mexico, Albuquerque, NM

Rainer de Martin, University of Vienna, Vienna, Austria

Michael Detmar, Swiss Federal Institute of Technology, Zürich, Switzerland

Napoleone Ferrara, Genentech, San Francisco, CA

Robert N. Frank, Wayne State University, Detroit, MI

Marcus Fruttiger, Wolfson Institute for Biomedical Research, London, England

Antonio Giordano, Temple University, Philadelphia, PA

Maria Grant, University of Florida, Gainesville, FL

Hans-Peter Hammes, University of Heidelberg, Mannheim, Germany

Mats Hellstrom, AngioGenetics Sweden AB, Göteborg, Sweden

David Hinton, Doheny Eye Institute, Los Angeles, CA

Eli Keshet, Hadassah Hebrew University, Jerusalem, Israel

Alisa Koch, University of Michigan, Ann Arbor, MI

Richard Lang, Children’s Hospital Research Foundation, Cincinnati, OH

Donald McDonald, University of California, San Francisco, CA

Jeremy Nathans, Johns Hopkins University, Baltimore, MD

Dario Neri, Swiss Federal Institute of Technology, Zurich, Switzerland

Gera Neufeld, Israel Institute of Technology, Haifa, Israel

Jean Plouet, Institut des Vaisseaux et du sang, Paris, France

Gregg Semenza, Johns Hopkins University, Baltimore, MD

Nader Sheibani, University of Wisconsin, Madison, WI

David Shima, Eyetech Pharmaceuticals, Woburn, MA

Philip Thorpe, University of Texas Southwestern, Dallas, TX

Rubin Tuder, Johns Hopkins University, Baltimore, MD

Olga Volpert, Northwestern University, Chicago, IL

Elizabeth Wagner, Johns Hopkins University, Baltimore, MD

Bernhard Weber, University of Würzburg, Germany

Stanley Wiegand, Regeneron Pharmaceuticals, Tarrytown, NY.

Supplementary Materials

Supplements 1-23 - (349 KB) 

Supplement 1 - Napoleone Ferrara: VEGF and EG-VEGF 

Supplement 2 - David Shima: VEGF Gradients 

Supplement 3 - Kari Alitalo: Lymphangiogenesis 

Supplement 4 - Gera Neufeld: Neuropilins and Semaphorins 

Supplement 5 - Eli Keshet: Tissue-Specific Expression or Knockdown of VEGF 

Supplement 6 - Mats Hellstrom: Platelet-Derived Growth Factors (PDGFs) and Perivascular Cells 

Supplement 7 - Richard Lang: Macrophages, Wnts, and Programmed Vascular Regression 

Supplement 8 - Jeremy Nathans: Norrin and Fz4: A Ligand-Receptor Pair That Controls 

Supplement 9 - Peter Brooks: Integrins and Extracellular Matrix 

Supplement 10 - B. H. F. Weber, A. Janssen, H. Schrewe, E. Tamm, A. May, M. Seeliger: Abnormalities in Vessel Formation in a Mouse Model of Timp3 Deficiency 

Supplement 11 - Olga Volpert: Endogenous Protein Inhibitors of Angiogenesis 

Supplement 12 - Gregg L. Semenza: Hypoxia and HIF-1 

Supplement 13 - Rhoda M. Alani: Id1 Regulation of Angiogenesis 

Supplement 14 - Rainer de Martin: Inflammation and NF-kappaB 

Supplement 15 - Peter Carmeliet: Mechanisms in Vessel Pathfinding 

Supplement 16 - Donald McDonald: Tumor Angiogenesis 

Supplement 17 - Michael Detmar: Wound Repair and Angiogenesis in Skin 

Supplement 18 - Rubin Tuder: Role of VEGF in Maintenance of the Pulmonary Microcirculation and the Etiology of Emphysema 

Supplement 19 - Alisa Koch: Angiogenesis in Rheumatoid Arthritis 

Supplement 20 - Marcus Fruttiger: Developmental versus Pathologic Retinal Neovascularization 

Supplement 21 - Peter Campochiaro: Ocular Neovascularization 

Supplement 22 - Philip Thorpe: Vascular Targeting Agents and Strategies 

Supplement 23 - Dario Neri: Targeting Specific Vascular Beds