Suppression of Retinal Neovascularization With an Antagonist to Vascular Endothelial Cadherin FREE

Ocular angiogenesis in response to tissue ischemia is a leading cause of vision loss in numerous clinical conditions.¹ Alterations in the retinal vasculature leading to new vessel formation are commonly seen in diabetic retinopathy, retinopathy of prematurity, and retinal vein occlusion. The formation of new vessels in the eye and elsewhere is a multistep process involving endothelial cell proliferation, migration, tubule formation, and subsequent maturation of the newly formed vessels.^2- 3 These new vessels are formed by the sprouting of endothelial cells from preexisting vessels, and their formation can be aided by the infiltration of bone marrow–derived progenitor cells that differentiate into endothelial or microglial cells.^4- 5 The migration and assembly of endothelial cells into tubelike structures is a vital step in the morphogenesis of new blood vessels. The formation and stabilization of these vascular structures are mediated by cell-cell contact and the subsequent engagement of homotypic adhesion molecules on adjacent endothelial cells.⁶

Endothelial cell junctions are populated by highly specialized families of cell adhesion molecules.^7- 9 Vascular endothelial cadherin (VE-cadherin) is localized at specialized cell junctions called adherens junctions and mediates calcium-dependent cell adhesion. The clustering of VE-cadherin ectodomains between adjacent endothelial cells begins the formation of stable adherens junctions between endothelial cells. In both normal and pathological angiogenesis, the formation and stabilization of cell-cell contact are prerequisites to lumen formation, generation of basement membranes, and inhibition of vascular regression.^6,10

In addition to mediating cell adhesion, VE-cadherin has been implicated in a variety of cell signaling pathways crucial to the endothelium.^11- 12 These signaling functions depend on the interaction of the VE-cadherin cytoplasmic domain with a number of accessory proteins, including β-catenin, plakoglobin, and α-catenin. Interaction of the catenin-cadherin complex with the cytoskeleton induces molecular signaling events that lead to contact inhibition, cell cycle arrest, and endothelial cell survival.^13- 16 In vasculogenesis, a null mutation in the VE-cadherin gene is embryonic lethal because of deficient vascular remodeling in the yolk sac and embryo proper. Cells devoid of VE-cadherin demonstrate an inability to organize into vascularlike structures, pointing to the central role played by VE-cadherin in developmental angiogenesis.¹¹

Recently, VE-cadherin has become a target for the inhibition of pathological angiogenesis.^17- 18 Inhibition of VE-cadherin may disrupt the transmission of intracellular signals induced by the angiogenic factor vascular endothelial growth factor (VEGF), and expression of VE-cadherin with a truncated cytoplasmic domain results in impaired vascular remodeling and maturation of the vascular network.¹¹ Because VE-cadherin regulates vascular permeability, the use of antibody-based antiangiogenic treatments risks eliciting widespread systemic vascular permeability increases.¹⁹ In this study, we have tested the antiangiogenic properties of a cyclic peptide inhibitor of VE-cadherin directed against the cell adhesion recognition sequence present on the VE-cadherin ectodomain. This peptide was found to significantly inhibit angiogenesis in a murine model of oxygen-induced retinal neovascularization. Our observations suggest that selective inhibition of VE-cadherin in newly forming vessels may be a useful target for therapeutic intervention in retinal neovascularization.

Mice of the C57BL/6J strain were bred at the University of New Mexico Animal Research Facility. All of the experiments were consistent with the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research. Retinal angiogenesis was induced using the oxygen-induced retinopathy model described by Smith et al.²⁰Age-matched mice housed only in room air served as controls. Experimental mice were injected intraperitoneally on postnatal (P) days P12 to P17 with 0.1 to 1.0 mg/kg of the VE-cadherin antagonist ADH100191-A (Ac-CDAEC-NH2) dissolved in sterile phosphate-buffered saline (PBS) (Adherex Technologies Inc, Durham, North Carolina). Some animals received an equivalent dose of a scrambled control peptide ADH100692-A (Ac-CAEDC-NH2). Eyes were collected at day P17 and analyzed histologically for the extent of retinal neovascularization.

Eyes from control and experimental animals were fixed overnight in neutral buffered formalin, 10%, and embedded in paraffin. Serial sections (6 μm) parallel to the optic nerve were mounted on gelatin-coated slides. After deparaffinization, sections were treated with trypsin, 0.0125%, at 37°C for 10 minutes. Sections were blocked with normal goat serum, 10%, and stained for type IV collagen (Rockland Inc, Gilbertsville, Pennsylvania). The sections were mounted with Vectashield (Vector Laboratories, Burlingame, California) containing 4′,6-diamidino-2-phenylindole to stain the nuclei. Nuclei associated with type IV collagen–positive vessels on the vitreous side of the inner limiting membrane of the retina were counted in every third section using a Zeiss inverted fluorescence microscope (Carl Zeiss MicroImaging, Inc, Thornwood, New York). Slides were masked prior to counting.

At day P17, eyes were removed and fixed with paraformaldehyde, 2%, in PBS for 2 hours at 4°C. The retina was dissected free and incubated in cold ethanol, 70%, for 20 minutes followed by cold PBS and Triton X-100 (Sigma-Aldrich, Inc, St Louis, Missouri), 1%, for an additional 30 minutes. The retinas were incubated overnight at 4°C in PBS containing 5-mg/mL isolectin GS-IB4 conjugated to Alexa Fluor 488 (Invitrogen Corp, Carlsbad, California), washed extensively with PBS, and mounted on glass slides. Images of the vasculature were obtained at ×10 magnification and pseudocolored.

Bovine retinal microvascular endothelial cells (VEC Technologies, Inc, Rensselaer, New York) were grown on fibronectin-coated cell culture dishes in MCDB-131 media supplemented with fetal bovine serum, 10%, 10-ng/mL epidermal growth factor, 1-μg/mL hydrocortisone, 0.2-mg/mL EndoGro (VEC Technologies, Inc), and 0.09-mg/mL heparin sodium. Passages 3 through 10 were used for all of the experiments.

Endothelial cells were plated onto fibronectin-coated 24-well plates at a density of 20 000 cells/well. Cells were stimulated with VEGF (40 ng/mL) in the presence or absence of the VE-cadherin antagonist (1 mg/mL) and a control peptide (1 mg/mL). After 48 hours of incubation, cell proliferation was quantitated using the Vybrant R cell proliferation assay kit (Invitrogen Corp).

Cell migration assays were performed using Falcon cell culture inserts (BD Biosciences, San Jose, California) containing a membrane with 3.0-μm pores. Bovine retinal microvascular endothelial cells (2.5 × 10⁴ cells) were plated onto membranes coated with fibronectin in serum-free MCDB-131 media. The lower chamber contained serum-free media with or without 40-ng/mL VEGF added as a chemoattractant. The upper chamber of the cell culture insert received specified doses of the VE-cadherin antagonist (1 mg/mL), control peptide (1 mg/mL), or VE-cadherin neutralizing antibody (100 ng/mL). The cells were allowed to migrate for 18 hours, and the cells on the upper surface of the membrane were removed with a cotton swab. The cells attached to the lower surface of the membrane were fixed with methanol, 100%, for 5 minutes and rinsed with PBS. The membranes were excised and then mounted on glass slides with Vectashield containing 4′,6-diamidino-2-phenylindole to stain the cell nucleus. The number of migrated cells was determined at ×10 magnification in 4 peripheral fields and 1 central field. The number of migrated cells was averaged from 5 fields per membrane.

Bovine retinal microvascular endothelial cells (2 × 10⁵/mL) were seeded into growth factor–reduced Matrigel matrix (BD Biosciences) and allowed to gel on glass coverslips. Gels were incubated in complete media containing VEGF (40 ng/mL) with either control peptide (1 mg/mL) or VE-cadherin antagonist (1 mg/mL) for 48 hours. Gels were fixed in paraformaldehyde, 4%, and examined at ×20 magnification.

Bovine retinal microvascular endothelial cells were seeded onto fibronectin-coated transwell inserts (0.4-μm pore size) (BD Biosciences) in a modified Boyden chamber. Transendothelial resistance was measured every day until the transendothelial resistance of the monolayer read more than 200 mΩ, at which point the cells were regarded as confluent. At the start of the experiment, 5 μL of 40K fluorescein isothiocyanate–dextran (40 mg/mL) was added to the top chamber. Inserts were treated with either the VE-cadherin antagonist (1 mg/mL) or control peptide (1 mg/mL). Aliquots of media were removed from the bottom chamber after 24 hours and read at 485 nm in a spectrofluorimeter (PerkinElmer, Waltham, Massachusetts).

Cells were seeded at low density onto glass coverslips coated with fibronectin and allowed to attach and spread. The cells were incubated for 3 hours with the VE-cadherin antagonist or control peptide, fixed with an ice-cold solution of methanol and acetic acid (3:1) for 2 minutes, and washed with PBS. Coverslips were blocked with goat serum, 10%, and incubated with either anti–β-catenin antibody (BD Biosciences) or anti–VE-cadherin (Alexis Corp, San Diego, California) for 1 hour at room temperature. Coverslips were rinsed 3 times for 10 minutes in TRIS-buffered saline and polysorbate 20, 0.1%, incubated with a fluorescently labeled secondary antibody, washed, mounted with Vectashield, and examined by fluorescence microscopy.

All of the statistical analyses were performed using GraphPad Prism version 4.0 software (GraphPad Software Inc, La Jolla, California). Data were analyzed by either t test or 1-way analysis of variance followed by Bonferroni multiple comparison test.

Animals with oxygen-induced retinopathy were treated with a VE-cadherin antagonist peptide. The mice received an intraperitoneal injection of either the antagonist (0.1, 0.5, or 1 mg/kg) or control peptide (1 mg/kg) from days P12 to P16. Eyes were collected on day P17 and the angiogenic response was quantitated as described. A significant 54% reduction in the degree of retinal neovascularization was found in mice treated at the highest dose (1 mg/kg) of the antagonist (Figure 1) (P < .01). Other doses at 0.1 or 0.5 mg/kg did not show any significant inhibition of retinal neovascularization. No obvious toxic effects on the normal vasculature were observed in retinas of animals receiving the VE-cadherin antagonist or control peptide (Figure 2).

The mechanism by which the VE-cadherin antagonist inhibits retinal angiogenesis was investigated using isolated bovine retinal endothelial cells. Individual cells were suspended in a 3-dimensional extracellular matrix and allowed to form tubes in the presence or absence of the VE-cadherin antagonist and control peptide. In cultures without treatment, the cells organized themselves into extensive branching networks typical for these cells grown in a Matrigel-containing matrix (Figure 3). Cells treated with the VE-cadherin antagonist exhibited far fewer interconnected tubules and organized primarily as small groups of elongated cells. Cells treated with control peptide looked similar to untreated cells.

The final angiogenic step of tube formation and stabilization is preceded by proliferation and migration of cells derived from either the existing vessels or bone marrow–derived endothelial progenitor cells. We next examined whether the disruption of tubule formation seen in antagonist-treated cultures might be due in part to a migration defect in the endothelial cells. The migration of bovine retinal endothelial cells was examined in vitro as described earlier. Cells plated onto fibronectin-coated migration inserts and stimulated with VEGF showed extensive migration after 18 hours. When the VE-cadherin antagonist was added at the time of plating, the cells showed a significant 46% reduction in the extent of migration (P < .001) (Figure 4). No significant effect was seen when cells were incubated in the presence of the control peptide (P = .32) or an anti–VE-cadherin antibody (P = .45).

During contact inhibition, the recognition and engagement of cell adhesion recognition sequences in VE-cadherin ectodomains on adjacent cells can lead to inhibition of cellular proliferation and responsiveness to growth factors.^11,21- 22 We thus investigated whether the proliferation of isolated bovine retinal endothelial cells might be disrupted by the VE-cadherin antagonist. Cells were grown in serum-free conditions and stimulated to proliferate by the addition of VEGF in the presence or absence of the VE-cadherin antagonist. After 48 hours of incubation, cellular proliferation was quantitated using the Vybrant R cell proliferation assay kit. In the presence of the VE-cadherin antagonist, cells demonstrated a significantly reduced rate of proliferation as compared with that in untreated cells or cells incubated with the control peptide (Figure 5).

To elucidate the molecular effects of a contact inhibition–like signal by the VE-cadherin antagonist, cells were stained for β-catenin. The β-catenin protein is localized to the intracellular aspect of the cell membrane when VE-cadherin proteins are engaged in a confluent monolayer of cells (Figure 6). Immunocytochemical analysis of β-catenin in a sparse culture of cells treated with the control peptide showed a lack of β-catenin in completely isolated cells but the presence of β-catenin in vesiclelike structures in the cytoplasm and at the membrane where cells are in contact with one another. In cells treated with the VE-cadherin antagonist, β-catenin was dramatically upregulated and appeared to be localized to the peripheral edges and processes of the cell.

To test the specificity and selectivity of the antagonist peptide, confluent monolayers of endothelial cells were treated with the antagonist peptide and the staining pattern of VE-cadherin and monolayer permeability were studied. Treatment of confluent monolayers with the antagonist had no effect on the structural integrity of the monolayers as demonstrated by continuous VE-cadherin staining along cell borders (Figure 7). The functional integrity of the monolayers was also unaffected by the antagonist as the permeability of fluorescein isothiocyanate–dextran across the monolayer was comparable between antagonist peptide–treated, control peptide–treated, and untreated cells (Figure 7).

This study addresses the role of the endothelial cell adhesion molecule VE-cadherin in the regulation of new vessel formation in a model of oxygen-induced retinal angiogenesis. Animals treated with a peptide antagonist that binds to the extracellular domain of VE-cadherin exhibited a 54% reduction in the neovascular response as compared with animals treated with a control peptide. This degree of inhibition was within the range of inhibition seen with other agents, including inhibitors of proteinases,^23- 25 cyclooxygenase 2,²⁶ integrins,²⁷ and VEGF.²⁸

The action of the antagonist was investigated using isolated bovine retinal endothelial cells. The antagonist appeared to be directed toward actively growing endothelial cells as it failed to disrupt an existing monolayer of endothelial cells with homotypically engaged adhesion proteins. Numerous stages of the angiogenic process could be disrupted by the antagonist, including cell migration, cell proliferation, and tubule formation. These endothelial cell behaviors were disrupted concomitant with a redistribution of the accessory adhesion protein β-catenin, suggesting that the antagonist may act as a decoy receptor. The antagonist likely binds to unengaged VE-cadherin molecules on the cell surface, simulating cell-cell interactions. A VE-cadherin antibody was unable to inhibit endothelial cell behavior as the antibody binds to regions of VE-cadherin other than the cell recognition sequence, thus failing to initiate a decoy response.

The structural and functional integrity of the microvasculature is partially dependent on the formation of adhesive junctions between adjacent endothelial cells of the capillary. A major component of the adherens junction in endothelial cells is the VE-cadherin protein. Like other members of the cadherin family, VE-cadherin is a transmembrane protein containing an extracellular domain composed of 5 subunits and a cytoplasmic domain that associates with the actin cytoskeleton. Homophilic interactions of VE-cadherin on adjacent cells and interactions of VE-cadherin with the cytoskeleton are facilitated and stabilized by additional proteins, including β-catenin, α-catenin, plakoglobin, and p120.^21- 22 In addition to providing for cell-cell adhesion, VE-cadherin has been shown to regulate numerous endothelial cell behaviors, including intercellular permeability, cell proliferation, and cell survival.^11,14,19,29

An important role for VE-cadherin in angiogenesis has been previously reported.^7,10,30 Deletion of the VE-cadherin gene is lethal to mice during development due to severe vascular defects.³¹ Endothelial cells in these animals could adhere to one another but fail to undergo correct vascular morphogenesis (migration, branching, and tubule formation). Studies also demonstrate that pathological angiogenesis can be regulated by VE-cadherin function.^17- 18 In these studies, specific antibodies to VE-cadherin were shown to block tumor growth, coincident with a decrease in the extent of tumor vascularization. These effects were seen independent of changes in permeability or changes in cellular adhesion, suggesting that the effective antibodies were targeting the developing vasculature and had no effect on the existing normal vessels. Similar findings were reported in the current study for the activity of the VE-cadherin antagonist. The antagonist was directed to the cell adhesion recognition sequence that facilitates homotypic recognition, dimerization, and cell-cell adhesion. The studies suggest that targeting an epitope on the VE-cadherin protein that is accessible only in loosely organized cells (ie, endothelial cells during angiogenesis) may be an effective way to target only the angiogenic cells while sparing the normal vasculature. This is important for the antagonist to be considered for therapy as any unintended side effects on the normal existing vasculature would be absent.

The peptide used in this study was directed against the VE-cadherin cell adhesion recognition sequence and antagonizes its ability to form stable adherens junctions with another endothelial cell. We speculate that this antagonist activity delivers a decoy adhesion signal to the cell, thereby eliciting a range of cell behaviors aimed at quiescence rather than angiogenesis. This is supported by the findings of decreased proliferation and migration in response to the peptide as well as redistribution of β-catenin within the cell.

The contact inhibition–like signal induced by treatment with the antagonist might also decrease the responsiveness of endothelial cells to angiogenic cytokines and growth factors such as VEGF. Contact inhibition is mediated largely by the establishment of stable cadherin junctions between adjacent endothelial cells. Caveda et al¹⁴ demonstrated that cells plated onto a substrate containing the VE-cadherin ectodomain displayed decreased cell division and proliferation. Several signaling pathways have been identified with VE-cadherin–dependent cell cycle arrest and inhibition of cell proliferation. In actively growing cells, β-catenin is found in the nucleus where it activates the transcription of many proteins important in the regulation of cell proliferation.^13,32 The cytoplasmic and membrane localization of β-catenin in response to antagonist stimulation seen in our study might prevent the transcription of several proteins required to orchestrate cell cycle progression. In angiogenic cells, VE-cadherin is associated with the VEGF receptor and Src, leading to phosphorylation of VE-cadherin's cytoplasmic tail and induction of cell proliferation facilitated by ERK/MAPK signaling. In addition, the VEGF receptor and VE-cadherin association induces cell survival through a PI3-kinase-Akt signaling pathway.¹¹ The delivery of a decoy signal might inhibit these molecular events and thus decrease cellular proliferation in endothelial cells, even in the presence of an angiogenic stimulus like VEGF.

Data from this study support the hypothesis that VE-cadherin plays an important role in the development of pathological ocular angiogenesis. The findings identify VE-cadherin as a potentially useful therapeutic target for diseases such as diabetic retinopathy, retinopathy of prematurity, and retinal vein occlusion.

Correspondence: Arup Das, MD, PhD, Division of Ophthalmology, Department of Surgery, MSC10-5610, 1 University of New Mexico, Albuquerque, NM 87131-0001 (adas@unm.edu).

Submitted for Publication: January 14, 2008; final revision received April 4, 2008; accepted April 8, 2008.

Financial Disclosure: None reported.

Funding/Support: This work was supported by grant RO1 EY12604 from the National Institutes of Health. Adherex Technologies Inc provided the VE-cadherin antagonist and control peptide.

Additional Contributions: Gina Menicucci, MS, and Dana Wilging, BS, provided expert technical assistance.