Age-related macular degeneration (AMD) is the leading cause of blindness in people over 55 years of age, and diabetic retinopathy (DR) in people under 55 years of age (Klein et al., 1994; Williams et al., 2004). Both disease conditions are characterized by new blood vessel growth—choroidal neovascularization in AMD and retinal neovascularization in DR (Freund, 1993; Speicher et al., 2003; Williams et al., 2004; Zarbin, 2004). There is ample evidence that α_V integrins are involved in ocular angiogenesis. Proangiogenic growth factors, including VEGF and FGF, are up-regulated in AMD and DR and they stimulate α_V expression. In the well established mouse model of oxygen-induced retinopathy (OIR), or retinopathy of prematurity (ROP) model, α_V integrins and the ligand osteopontin are overexpressed in neovascular endothelial cells during the peak time of retinal vessel growth (Takagi et al., 2002). The α_V integrins bind to their extracellular matrix ligands via a specific arginine-glycine-asparagine (RGD) binding site. Cyclic peptides mimicking the RGD binding motif have been developed as potent antagonists to α_V integrins, and when administered via s.c. (Friedlander et al., 1996; Chavakis et al., 2002), intraperitoneal, periocular (Luna et al., 1996), or topical (Riecke et al., 2001) routes, these cyclic peptides have been shown to inhibit retinal neovascularization in the mouse OIR model. In the laser-induced choroidal neovascularization model (rats), a well accepted model for AMD, integrins α_Vβ₃ and von Willebrand factor are expressed on endothelial cells of new vessels after photocoagulation, but they are not expressed in normal choroidal vessels (Kamizuru et al., 2001). In this model, intravitreal injection of a cyclic RGD peptide significantly reduced the development of choroidal neovascularization (Yasukawa et al., 2004). Additional evidence has emerged from the use of the extracellular region of brain-specific angiogenesis inhibitor 1, which is known to exert antiproliferative activity via functional blockade of α_Vβ₅ in endothelial cells. Subconjunctival injection of brain-specific angiogenesis inhibitor 1 (via gene delivery) or an α_Vβ₅ antibody showed significantly reduced neovascularization in a rabbit model of laser-induced choroidal neovascularization (Yoon et al., 2005). In humans, α_Vβ₃ and α_Vβ₅ are not expressed in normal retinal tissue, but they have been observed at elevated levels in vascular cells in the eyes of patients with DR (Friedlander et al., 1996; Luna et al., 1996), whereas α_Vβ₅ was primarily observed at high levels in ocular tissues in patients with AMD (Friedlander et al., 1996). Increased expression of fibronectin on the inner limiting membrane and α_Vβ₃ in the neovessels in eyes of patients with proliferative diabetic retinopathy has also been described previously (Ljubimov et al., 1996).

Angiogenesis inhibitors targeting VEGF (VEGF aptamer, Macugen, and VEGF monoclonal antibodies) have made a significant impact on the treatment of DR and AMD. A large body of work has been published on the complex relationship between VEGF and α_V integrins. Antagonists of both α_Vβ₃ and α_Vβ₅ block VEGF-R2 phosphorylation and VEGF-stimulated adhesion, proliferation, and migration of endothelial cells (Terai et al., 2001; Tsou and Isik, 2001). It is interesting to note that in the monkey eye, VEGF induces an angiogenic phenotype in pericytes that accompanies endothelial cells in newly formed vessel sproutings. In a similar manner, in the monkey eye, after exposure to VEGF, α_Vβ₃ and α_Vβ₅ are highly up-regulated in migrating cells from pre-existing and newly formed vessels (Witmer et al., 2004). These studies offer compelling evidence that α_Vβ₃ function is necessary for VEGF-mediated effects. It is important to note that VEGF and FGF are up-regulated in AMD and proliferative DR. Therefore, α_V antagonists may offer broader coverage for these diseases. Furthermore, in both AMD and diabetic retinopathy, VEGF is not only an important angiogenic signal but also a key regulator of retinal vascular permeability. Vascular permeability involving the breakdown of the blood-retinal barrier is driven by VEGF, and it is a key element in loss of visual acuity (Patel et al., 2003). Although inhibitors of VEGF have been shown to reduce retinal vascular permeability, this effect has not yet been demonstrated with α_V integrin antagonists.

Oral α_V antagonist therapy would represent a novel approach that would be complementary to other treatment modalities, such as VEGF antagonists, which do not affect angiogenesis induced by other growth factors (e.g., FGF) and which are not available as oral agents. We report on an orally bioavailable α_V integrin antagonist that markedly inhibits both retinal neovascularization in the ROP model in mice and retinal vascular permeability in diabetic rats.

Materials and Methods

Materials. Integrins α_vβ₃ and α_vβ₅ and CytoMatrix cell adhesion strips were purchased from Millipore Bioscience Research Reagents (Temecula, CA). Anti-human vitronectin IgG rabbit polyclonal antibody was from Calbiochem (San Diego, CA). Binding assays were visualized using VectaStain ABC peroxidase kit reagents (Vector Laboratories, Burlingame, CA). Acridine orange and vitronectin were purchased from Sigma-Aldrich (St. Louis, MO). Immulon-2 microtiter plates from Dynex Technologies (Chantilly, VA) were used in all binding assays. Human α_IIbβ_IIIa from Enzyme Research Laboratories Inc. (South Bend, IN), human fibrinogen from American Diagnostica (Greenwich, CT), and the biotin-X-NHS biotinylation kit from Calbiochem (San Diego, CA) were reagents for the α_IIbβ_IIIa binding assay. Binding assay plates were read using a microplate reader from Molecular Devices (Sunnyvale, CA). Calcein acetoxymethyl ester was from Invitrogen (Carlsbad, CA). Adhesions assay plates were read on a Cytofluor 2300 from Millipore Corporation (Billerica, MA). Human microvascular endothelial cells (HMVECs) were from Cascade Biologics (Portland, OR), rat endothelial cells were from Vec Technologies (Rensselaer, NY), and human umbilical cord endothelial cells (HUVEC) were from Cambrex Charles City, Inc. (Charles City, IA). Human K-562 and HT-29 cells were purchased from American Type Culture Collection (Manassas, VA). Chemotaxis assay plates were from NeuroProbe (Gaithersburg, MD). JNJ-26076713 (Fig. 1) was synthesized by the methods described in the literature for compound 29d (Ghosh et al., 2004; Kinney et al., 2007), and it was used as the HCl salt (C₂₉H₃₈N₄O₃·2.4 HCl·1.2 H₂O). Fresh solutions of the drug were made up daily to minimize cleavage of the central amide or oxidation to the quinoline.

Animals. All animal procedures were in strict accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the University of Florida Institutional Animal Care and Use Committee, Joslin Diabetes Center Animal Care, and Use Committee and the Johnson and Johnson Animal Care and Use Committee. Timed pregnant C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME) for the ROP model and pharmacokinetic studies. Male Long-Evans rats (Taconic Farms, Germantown, NY) with initial weights between 180 and 250 g were used for the retinal vascular permeability model. All procedures involving Sprague-Dawley rats were from Charles River Laboratories, Inc. (Wilmington, MA), weighing 250 to 350 g, were used in the pharmacokinetic studies.

Integrin Binding. Human α_vβ₃ at a concentration of 1 μg/ml dissolved in Tris buffer (20 mM Tris, 1 mM CaCl₂, 1 mM MgCl₂, 10 μM MnCl₂, and 150 mM NaCl) was immobilized on Immulon-2 96-well plates overnight at 4°C. Plates were washed and treated with blocking buffer (3% BSA in Tris buffer) for 2 h at 37°C. Plates were then rinsed two times in Tris buffer containing 0.3% BSA and 0.2% Tween 20 (assay buffer). Five minutes before the addition of 4 nM vitronectin, JNJ-26067613 was added to wells in duplicate. After a 3-h incubation at 37°C, plates were washed five times in assay buffer. An anti-human vitronectin IgG rabbit polyclonal antibody was added, and plates were incubated for 1 h at room temperature. Vectastain ABC peroxidase kit reagents using a biotin-labeled anti-rabbit IgG were used for detection of bound antibody. Optical density was read at 490 nM on a microplate reader. IC₅₀ values were determined using a four-parameter-fit logistics model. To determine competition for α_vβ₅ binding, the method was identical to that for human α_vβ₃ except that human α_vβ₅ at a concentration of 1 μg/ml was immobilized on the 96-well plates (Luci et al., 2004). JNJ-26076713 was tested at half-logarithmic doses from 0.1 to 100 nM in duplicate.

To determine selectivity against α_IIbβ_IIIa, wells of a 96-well Immulon-2 microtiter plate were coated with 10 μg/ml RGD affinity-purified human α_IIbβ_IIIa in 10 mM HEPES, 150 mM NaCl, and 1 mM MgCl₂, pH 7.4 (500 ng/well α_IIbβ_IIIa final), and then the pate was incubated overnight at 4°C. The next day, the wells were blocked with 5% BSA in the above-mentioned buffer at room temperature for 2 h. The assay plate was washed five times with modified Tyrode's buffer (150 mM NaCl, 12 mM NaHCO₃, 2.6 mM KCl, 2.5 mM HEPES, 1 mM MgCl₂, and 1 mg/ml BSA, pH 7.4). Biotinylated fibrinogen was prepared using human fibrinogen according to directions from the biotin-X-NHS biotinylation kit. Fifty microliters of the compound/biotinylated fibrinogen mix was transferred to the assay plate and incubated at room temperature for 2 h. After incubation, the assay plate was washed five times with modified Tyrode's buffer. Reactions were visualized as described above. JNJ-26076713 was tested at half-log doses from 0.1 to 30 μM in duplicate.

Cell Adhesion. The ability of JNJ-26076713 to inhibit human, mouse, and rat endothelial cell adhesion to vitronectin was evaluated. HMVECs, passages 3 to 9 (Cascade Biologics); mouse endothelial cells isolated from normal mouse lungs; and rat endothelial cells (Vec Technologies), all at 70 to 90% confluence, were trypsinized from flasks and diluted in Dulbecco's phosphate-buffered saline (PBS) with 0.1% BSA (assay buffer). Cells were labeled with 5 μM calcein acetoxymethyl ester in assay buffer for 30 min at 37°C. Cells were washed three times in assay buffer. JNJ-26076713 (50 μl) was added at 2× concentration to each well of the plate containing CytoMatrix vitronectin-coated cell adhesion strips and tested in duplicate. Labeled endothelial cells (50 μl; 5 × 10⁵/ml) were transferred into wells and allowed to adhere for 1 h at 37°C and 5% CO₂. Plates were washed three times in assay buffer. Cells were lysed for 15 min in 100 μl of 1 M Tris, pH 8.0, with 1% SDS. Plates were read at 485 excitation/530 emission (Cytoflour 2300). IC₅₀ values were determined using a four-parameter fit logistics model. JNJ-26076713 was tested at half-log doses ranging from 0.001 to 1 μM (final concentration in well) in duplicate.

To test for integrin selectivity, the ability of JNJ-26076713 to inhibit K562 cell (α₅β₁-mediated) adhesion and HT29 cell (α_Vβ₆-mediated) adhesion to CytoMatrix fibronectin-coated adhesion strips was evaluated. The methods were similar to those for endothelial cells. The compound was initially tested in duplicate at 0.1, 0.3, 1, 3, 10, and 50 μM in duplicate in all adhesion assays. In all cell types but the K562 cells, JNJ-2607613 was ultimately tested at half-logarithmic doses ranging from 0.001 to 3 μM in duplicate.

Cell Migration. FGF-, serum-, and VEGF-induced migration studies are described. For FGF-induced migration, vitronectin (0.4 μg/well) was added to the bottom wells of a chemotaxis assay plate. HUVECs, 45 μl of a 2 × 10⁶ cells/ml solution and 5 μl of compound, were added to wells of a 96-well plate and incubated for 10 min at room temperature. The cell mix (25 μl) and FGF₂ (10 ng/ml) were added to top wells of the holding filter of the chemotaxis assay plate, and the plate was incubated overnight in cell culture incubator. Media and nonmigrated cells were removed from the top filter (using a scraper). The migration filter was washed and fixed in 1.0% formaldehyde, and cell membranes were permeated with 0.2% Triton X-100 for 15 min. The migration filter was then stained with rhodamine phalloidin for 30 min in the dark, washed, and air-dried before measuring fluorescence at 530-nm excitation/590-nm emission. All concentrations were run in triplicate and repeated three times.

Serum-induced endothelial cell migration assays were performed in 24-well Transwell chambers with a polystyrene membrane (6.5 mm in diameter, 10 mm in thickness, and a pore size of 8 mm). Subconfluent 24-h cell cultures (HUVECs) were harvested with trypsin-EDTA, washed twice, and resuspended in their respective serum-free medium containing 0.1% BSA. Cells (100,000/500 ml) were added to the top chamber in the presence or absence of various concentrations of compound. To facilitate chemotactic cell migration, 750 ml of medium containing 2% serum was added to the bottom chambers and the plate, and the plate was placed in a tissue culture incubator. Migration was terminated after 4 to 8 h by removing the cells on the top with a cotton swab, and then the filters were fixed with 3% paraformaldehyde and stained with crystal violet. The filters were dissolved in 10% acetic acid for 30 min. The absorbance was measured at 590 nm. The data were normalized to percentage of the vehicle control, which was considered as 100%, and each point is the mean of three Transwell filters (±S.D.).

VEGF-induced endothelial cell migration assays were also carried out in 24-well Transwell chambers. In brief, the underside of the membrane was coated with vitronectin (1 mg/ml) for 60 min at room temperature, and then the membrane was blocked with a solution of 1% BSA/PBS at room temperature for 60 min. Membranes were washed with PBS and air-dried. Subconfluent 24-h cell cultures (HUVECs) were harvested and resuspended in serum-free media containing 0.1% BSA. Cells (100,000/500 ml) containing different concentrations of compound were added to the top chamber. Then, 750 ml of medium containing 0.1% BSA and 50 ng/ml VEGF was added to the bottom chambers, and the plate was placed in a tissue culture incubator. Migration was terminated after 4 h by removing the cells in the top chamber with a cotton swab. Filters were treated identically to those in the serum-induced assays. Studies were repeated three times. Migration in vitronectin-coated Transwell chambers without VEGF stimulation was set as 100%.

All groups were analyzed using analysis of variance (ANOVA) followed by unpaired t tests (GraphPad Prism; GraphPad Software Inc., San Diego, CA).

Chick Chorioallantoic Membrane Model. Ten-day-old embryos were incubated at 37°C with 55% relative humidity. Light was used to define an avascular region. The CAM was dropped, and a window, approximately 1.0 cm², was cut in the shell over the dropped CAM, allowing for direct access to the underlying CAM. FGF₂ was used as a standard proangiogenic agent to induce new blood vessel branches on the CAMs of 10-day-old chick embryos. Sterile filter disks absorbed with 1 μg/ml FGF₂ dissolved in PBS were placed on growing CAMs. At 24 h, JNJ-26076713 was added directly (topically) at 0.1, 1, and 10 μg/CAM (n = 7/dose). CAM tissue directly beneath the filter disk was resected from embryos treated 48 h prior with the JNJ-26076713. Tissues were washed three times with PBS, and then they were examined under a stereomicroscope. Digital images of CAM sections adjacent to filters were collected using a 3-charge-coupled device, color video camera system, and they were analyzed with Image-Pro Plus software (Media Cybernetics, Inc., Bethesda, MD). The number of vessel branch points contained in a circular region equal to the area of a filter disk was counted for each section (Ribatti et al., 2000).

Pharmacokinetic Studies in Rats and Mice. Rats and mice (n = 4) were dosed i.v. at 2 mg/kg and by oral gavage at 10 mg/kg with JNJ-26076713 formulated in 5% dextrose in water (D5W). Blood samples were collected at 5 (i.v. only), 15, and 30 min and at 1, 2, 4, 7, and 24 h after dosing into heparinized (lithium) tubes. Blood samples were centrifuged for cell removal, and two 200-μl plasma samples were stored at –70°C for later analysis. Blank species-specific plasma was used for preparation of standard curves. Acetonitrile (150 μl) was added to 50 μl of plasma to precipitate proteins. Samples were centrifuged, and the supernatant was removed for analysis by liquid chromatography-tandem mass spectrometry. A ratio of 3 parts acetonitrile to 1 part plasma was maintained for all studies when <200 μl of plasma was available.

OIR Model. In the mouse model of oxygen-induced retinopathy (Grant et al., 2004), mice at postnatal day 7 were placed with their nursing dams in a 75% oxygen atmosphere for 5 days. Upon return to normal air, these mice develop retinal neovascularization, with peak development occurring 5 days (postnatal day 17) after their return to normoxia. On day 12, the mouse pups received 30, 60, or 120 mg/kg JNJ-26076713 by gavage twice daily for 5 days. Compound solutions were made up fresh in D5W, pH 2.0. After the fifth day of treatment (after return to normoxia), the animals were euthanized, and the eyes were removed and fixed in 4% paraformaldehyde. Then, they eyes were embedded in paraffin. Three hundred serial sections (6 μm) were cut sagittally through the cornea parallel to the optic disc. Every 30th section was placed on slides and stained with hematoxylin and eosin. This resulted in 10 sections from each eye being scored in a masked manner using light microscopy to count endothelial nuclei extending beyond the inner limiting membrane into the vitreous as described previously (Smith et al., 1994). The efficacy of treatment was calculated as the percentage of average nuclei per section for both eyes comparing treated versus untreated animals. All groups were analyzed using ANOVA followed by Tukey's multiple comparison test (GraphPad Prism). Blood was removed from a subset of mice 12 to 14 h after the last dose of compound, and plasma samples were frozen for analyses of drug levels. No samples were taken from the 120-mg/kg dose group.

Diabetes-Induced Retinal Vascular Permeability Model. Retinal physiological measurements were performed to investigate whether oral treatment with JNJ-26076713 in a prevention regime affected retinal physiology in 2-week diabetic Long-Evans rats. (Abiko et al., 2003). Treatment was initiated upon confirmation of diabetes after streptozotocin injection. Untreated Long-Evans rats served as nondiabetic controls. Forty-eight rats were distributed in four groups of 10 to 12 animals each as follows: nondiabetic rat + vehicle (D5W), nondiabetic rat + JNJ-26076713 (60 mg/kg b.i.d.), diabetic rat + vehicle (D5W), and diabetic rat + JNJ-26076713 (60 mg/kg b.i.d.). Treated rats were administered JNJ-26076713 in D5W, pH 2.0, twice daily (8:00 AM and 5:00 PM) by oral gavage). Control animals were treated in a similar manner using only vehicle. Animal condition was monitored daily. Blood glucose (BG) and body weight (BW) were monitored. Hematocrit was also tested to ensure that animals did not become dehydrated.

On day 13, a jugular vein catheter was surgically implanted. On day 14, rats underwent retinal leukostasis measurements using acridine orange leukocyte fluorography (Abiko et al., 2003). In brief, acridine orange was infused (750 μl/min) through the jugular vein catheter and after 20 min, retinal images from both eyes were obtained to assess static stained leukocytes. For retinal vascular permeability measurements, rats were infused with Evans blue on day 15 using a standard in vivo protocol (Xu et al., 2001). In brief, 2 h after Evans blue perfusion, the animals were sacrificed, and the retinas were removed for permeability measurements. The eyes were enucleated, and the retinas were carefully dissected away under an operating microscope. The weight of each retina was measured after drying for 4 h in a Speed-Vac (Thermo Electron Corporation, Waltham, MA). Albumin leakage into the retinal tissue was estimated via the measurement of extravasated Evans blue dye. Evans blue is extracted by incubating each retina in 0.24 ml of formamide for 18 h at 72°C. The extract was filtered through a 30,000-mol. wt. filter at a speed of 12,000 rpm for 120 min at 4°C. The absorbance of the filtrate was measured with a spectrophotometer at 620 and 740 nm, the absorption maximum and minimum for Evans blue in formamide. Retinal permeability is expressed as nanograms of Evans blue per gram of retina dry weight. An increase in this measurement indicates leakage of Evans blue into the retina and an increase in vascular permeability.

At the time of catheterization, a blood sample was withdrawn for determination of compound concentration. Group sizes allowed for demonstration of a 25 to 30% change, with a p value <0.05 at a power of 0.8 using t test or ANOVA. The retinal physiological response to the treatment was compared with the untreated control groups for both diabetic and nondiabetic rats.

Results

In Vitro Studies. JNJ-26076713 is a potent inhibitor of both human α_vβ₃ and α_vβ₅ binding to vitronectin, demonstrating IC₅₀ values of 2.3 ± 0.27 nM (n = 18) and 6.3 ± 0.83 nM (n = 11), respectively. In contrast, JNJ-26076713 was extremely weak in its ability to inhibit human α_IIbβ₃ binding, with an IC₅₀ of 4690 ± 170 nM (n = 2). Furthermore, JNJ-26076713 was 4.5- and 23-fold more potent, respectively, in inhibiting human endothelial cell adhesion to vitronectin compared with rat and mouse endothelial cell adhesion (Table 1). This species differentiation is relevant when considering the effects of JNJ-26076713 in mouse and rat in vivo models of efficacy. To further characterize integrin specificity, α_Vβ₆- and α₅β₁-mediated adhesion was evaluated. JNJ-26076713 did not inhibit human α₅β₁-mediated adhesion, whereas it was moderately active in inhibiting α_Vβ₆-mediated adhesion (Table 1).

JNJ-26076713 dose-dependently inhibited HUVEC migration to serum. In addition, JNJ-26076713 was a potent inhibitor of FGF₂-induced HUVEC migration (Fig. 2A). In experiments in which VEGF was the migration stimulus, VEGF-induced HUVEC migration was increased more than 2-fold (222%). As demonstrated in Fig. 2, JNJ-26076713 produced concentration-dependent inhibition of HUVEC migration stimulated by VEGF, with an IC₅₀ value of 30 nM. IC₅₀ values calculated from the serum and FGF migration studies were 29 and 31 nM, respectively, indicative of the ability of JNJ-26076713 to be a potent inhibitor of endothelial migration induced by several physiological factors.

JNJ-26076713 demonstrated potent inhibition of angiogenesis in the CAM model as shown in Fig. 3. Even at the lowest dose of JNJ-26076713, 0.1 μg/CAM, a 72% inhibition of FGF-induced angiogenesis was observed. Figure 4 exemplifies the inhibitory effect in this model at both 0.1 and 10 μg/ml JNJ-26076713. The new vessel growth induced by FGF (Fig. 4B) is nearly eliminated after exposure to either dose of antagonist (Fig. 4, C and D).

Pharmacokinetics. JNJ-26076713 was administered either intravenously (2 mg/kg) or orally (10 mg/kg) to male rats and plasma compound levels measured. Derived pharmacokinetic parameters are shown in Table 2. Volume of distribution (1100 ml/kg) was slightly larger than rat total body water (668 ml/kg), suggesting that the compound is well distributed. The clearance was low compared with liver blood flow (55 ml/min/kg), with a resulting terminal elimination t_1/2 of 10 h. Oral bioavailability was 21%.

Derived pharmacokinetic parameters in mice are shown in Table 3. Plasma concentrations decreased in a biphasic manner after i.v. administration. JNJ-26076713 had a high volume of distribution in the mouse (5504 ml/kg) compared with mouse total body water of 725 ml/kg, and low clearance (7 ml/min/kg) compared with mouse liver blood flow of 90 ml/min/kg. The t_1/2 was similar to that in the rat after i.v. administration but somewhat shorter after oral administration. Oral bioavailability in mice was similar to rats. The pharmacokinetic profile indicated that oral, twice daily dosing would be suitable for adequate drug exposure for in vivo efficacy studies in either rats or mice.

In Vivo Disease Models. JNJ-26076713 significantly inhibited retinal neovascularization at all doses tested (p < 0.001). Inhibition was dose-dependent (Fig. 5), with a 33, 43, and 67% inhibition of neovascularization at 30, 60, and 120 mg/kg, respectively. Drug levels, representing trough levels, were 3.9 and 4.1 μM after the 30 and 60 mg/kg dosing, respectively, indicating excellent exposure levels based on plasma samples 12 to 14 h after the final oral dose. Blood samples were not taken in the 120-mg/kg dose group.

JNJ-26076713 inhibited the increase in retinal vascular permeability and leukostasis associated with diabetes (Fig. 6). JNJ-26076713 at 60 mg/kg b.i.d. demonstrated significant inhibition of retinal vascular permeability in streptozotocin diabetic rats. Furthermore, JNJ-26076713 treatment tended to reduce leukocyte adhesion (48% reduction representing the decrease in number of static leukocytes in diabetic-treated mice from vehicle-treated diabetic rats using the nondiabetic vehicle-treated mice as a baseline), albeit this reduction was not statistically significant. There were no significant changes in BW or plasma glucose levels associated with JNJ-26076713 treatment (Table 4). Levels of JNJ-26076713 in plasma samples drawn at either 1 to 3 or 4 to 6 h after the last oral dose indicated good compound exposure in both diabetic rats (4.6 ± 0.7 and 7.6 ± 1.0 μM, respectively) and nondiabetic rats (5.3 and 6.4 μM, respectively, from a single sample).

Discussion

This report describes the in vitro and in vivo pharmacological activity of a novel α_V antagonist, JNJ-26076713, which was shown to inhibit both retinal neovascularization in a mouse retinopathy model and retinal vascular permeability increase in diabetic rats when delivered orally. These findings represent the first demonstration of therapeutic activity of an orally delivered α_V integrin antagonist in models of retinopathy. JNJ-26076713 was derived from the RGD binding motif, and it was optimized to yield a potent antagonist of human α_Vβ₃/α_Vβ₅, as demonstrated in both ligand binding assays and in cell adhesion and migration assays. Although somewhat less potent in mouse and rat α_Vβ₃/α_Vβ₅, its activity was sufficient for assessing efficacy in the retinopathy model in the mouse and the vascular permeability model in the rat. Admittedly, the variability of potency to inhibit endothelial cell adhesion among species was limited to one endothelial type per species and could be attributed to difference in the origin of the endothelial cell. However, we have not seen significant differences in inhibitory potency of JNJ-26076713 between human endothelial cell and human tumor cell adhesion in those tumor lines that express α_V integrins (data not shown). JNJ-26076713 also had good oral bioavailability and elimination half-life in mice and rats (this report) and in dogs and monkeys (data not shown), producing excellent plasma compound exposure and volume of distribution beyond the plasma compartment.

Integrins bind protein ligands that are components of the extracellular matrix, such as collagens, fibronectin, fibrinogen, vitronectin, osteopontin, and laminin. There are specific ligand-binding patterns for each integrin (Hodivala-Dilke et al., 2003; Mousa, 2003). The α_V integrins are expressed on several cell types, including osteoclasts, vascular smooth muscle cells, endothelial cells, and a variety of tumor cells (Mousa, 2002). The α_V integrins bind to their extracellular matrix ligands at the RGD binding site. Cyclic peptides and small molecules mimicking this RGD binding motif have been developed as potent antagonists (Eliceiri and Cheresh, 1998; Mousa, 2002, 2003; Rüegg and Mariotti, 2003; Williams et al., 2004), and they have shown efficacy in animal models of DR, AMD, osteoporosis, restenosis, arthritis, tumor growth, and metastasis. In addition, monoclonal antibodies to these integrins have also been demonstrated to be efficacious in these models (Miller et al., 2000; Wilder, 2002; Trikha et al., 2004; Cai and Chen, 2006). Cilengitide, an RGD mimetic peptide, and CNTO 95, a pan α_V monoclonal antibody, are currently in phase II clinical trials, whereas Vitaxin, an α_Vβ₃ monoclonal antibody, is in phase III trials (The Investigational Drug Database; Thomson Scientific, Philadelphia, PA) for treatment of cancer and a variety of solid tumors.

Monoclonal antibodies to α_V integrins and RGD mimetic peptides have demonstrated the capacity to diminish both retinal and choroidal neovascularization in animal models (Luna et al., 1996; Kamizuru et al., 2001; Riecke et al., 2001; Chavakis et al., 2002; Witmer et al., 2004; Yasukawa et al., 2004; Yoon et al., 2005). To date, there has been only one published report of a nonpeptide α_V small-molecule antagonist evaluated in an animal model of DR or AMD. SB-267268, administered intraperitoneally at 60 mg/kg b.i.d. reduced angiogenesis in the murine OIR model by approximately 50%. In addition, both VEGF and VEGF receptor-2 expression in the inner retina was attenuated with SB-267268 (Wilkinson-Berka Jennifer et al., 2006). JNJ-26076713 was shown to produce a similar extent of angiogenesis inhibition in the ROP model, but in contrast to SB-267268, it was efficacious after oral dosing. No other small or large molecule α_V antagonist delivered via any route has been shown to inhibit increases in retinal vascular permeability associated with diabetes. Breakdown of the blood retinal barrier is an early characteristic of DR, and it correlates to vision loss. Vascular permeability and leakage are prevalent in both AMD and DR. JNJ-26076713 was effective in preventing vascular permeability increase in diabetic rats and also tended to decrease leukostasis, although this effect was not statistically significant. Leukostasis can result from insulin resistance before the onset and in the absence of diabetes. However, alterations in retinal blood flow and vascular permeability are dependent on the diabetic state and hyperglycemia (Abiko et al., 2003). Therapies that can reduce retinal vascular permeability and prevent neovascularization could have improved outcomes in limiting loss of vision.

Concern has existed regarding the safety of long-term systemic administration of α_V antagonists, including potential side effects such as inhibition of wound healing. Recent evidence has lessened this concern, because an orally active α_V antagonist L-845704 showed no serious adverse effects after 1 year of treatment in postmenopausal osteoporotic women (Murphy et al., 2005). Furthermore, infusions of monoclonal antibodies to α_Vβ₃, α_Vβ₅, or both, resulting in sustained systemic exposure, did not hinder wound healing in monkeys and humans (Martin et al., 2005; Zhang et al., 2007). Current standard of care for retinopathies uses intravitreal injection of therapeutics or laser photocoagulation. A safe and orally effective α_V antagonist may provide the means to either sustain or improve outcomes of laser therapy or intravitreal injections of either Macugen or Lucentis, and ultimately it might limit the need for these more invasive treatments and improve compliance.

The newest therapies for AMD and DR are antiangiogenic therapies targeting VEGF (Lucentis and Macugen). In part, angiogenesis involves adhesive interaction of endothelial cells with the extracellular matrix. JNJ-26076713 inhibits human, mouse, and rat endothelial cell adhesion; however, it is severalfold less potent in its effects on rat and mouse adhesion. Thus, the doses and plasma compound levels associated with efficacy in the rodent models overestimate the probable effective doses and plasma levels, given the greater potency in inhibiting human endothelial cell adhesion. JNJ-26076713 is also a potent inhibitor of endothelial cell migration induced by FGF, VEGF, and serum with nearly identical potencies. This activity may represent an advantage over VEGF therapies alone as treatments for AMD or DR. Growth factors and cytokines increase integrin expression, and they are also key elements in angiogenesis pathways. In the chick CAM, FGF specifically induces β₃ expression, whereas VEGF induces β₅ expression (Friedlander et al., 1996). It is noteworthy that VEGF, FGF, and other prominent growth factors such as insulin-like growth factor, which is present in serum, are up-regulated in AMD and proliferative diabetic retinopathy (Simo et al., 2006). Ligand occupancy of α_Vβ₃ has been reported to be a prerequisite for insulin-like growth factor to induce cell migration and proliferation (Clemmons and Maile, 2005).

JNJ-26076713 exhibited excellent exposure levels in both the mouse and rat models after twice-daily oral dosing. In the mouse ROP model, >1 μM levels were detected even at trough (12–14 h after the final oral dose). Moreover, JNJ-26076713 has been shown to have no adverse activity in animal models of cardiovascular and central nervous system safety (data not shown), and its selectivity was notable as indicated by lack of significant interaction with a panel of 51 receptors and ion channels and 152 kinases (data not shown). JNJ-26076713 demonstrated some capacity to inhibit α_Vβ₆-mediated adhesion. Although α_Vβ₆ expression is limited to the epithelium, where expression is typically low, it can be highly elevated in carcinomas and in chronic wounds (Häkkinen et al., 2004; Bates, 2005) Therefore, it is possible that modest inhibition of α_Vβ₆ should not affect the overall profile of JNJ-26076713.

Integrin α_V antagonists have been studied for several years, and their utility in several disease models have been well documented. Clinically, the most therapeutically active area is oncology. Most recently, there has also been a strong interest in targeting α_V integrins as imaging/targeting agents in cancer. (Lim et al., 2005). Merck successfully demonstrated clinical efficacy for their α_V antagonist in osteoporosis. However, most likely due to successful and less expensive pre-existing therapies, such as alendronate sodium (Fosamax), further development of α_V antagonists for this indication were not pursued.

AMD and DR represent a large unmet medical need, and they may represent an important new opportunity for α_V antagonists. Inhibiting the leakage in existing blood vessels and preventing abnormal angiogenesis with an orally bioavailable α_V antagonist would represent an excellent therapy for preventing the retinal or choroidal neovascularization associated with DR and AMD.

In conclusion, JNJ-26076713 is the first small-molecule α_V antagonist with oral efficacy in both inhibiting retinal neovascularization in a model of retinopathy and in retinal vascular permeability increase in a diabetic model. α_Vβ₃ and α_Vβ₅ integrins in particular have been implicated in retinal neovascularization, and they are probably the primary targets responsible for the pharmacological effects of JNJ-26076713 in these models. The potential of α_V antagonists in the treatment of AMD and DR is only beginning to be explored in the clinical setting. The availability of an oral therapeutic agent for the treatment of AMD and DR has the potential to reduce the frequency of the current invasive treatments and improve upon overall outcome. Therefore, JNJ-26076713 represents a potential therapeutic candidate with the convenience of oral bioavailability for the treatment of age-related macular degeneration, macular edema, and proliferative diabetic retinopathy.