CAI Is a Potent Inhibitor of Neovascularization and Imparts Neuroprotection in a Mouse Model of Ischemic Retinopathy

Abstract

purpose. This study was performed to characterize the effects of an antimetastatic and antiangiogenic molecule, carboxyamido-triazole (CAI), on retinal neovascularization in a mouse model.

methods. Neonatal mice were subjected to 75% to 85% oxygen from postnatal day (PND)-7 to -12 and then were abruptly placed in room air. CAI (100 mg/kg) or vehicle control polyethylene glycol-400 (PEG-400) was given daily from PND-14 to -16, and mice were killed on PND-17 to form group A. In group B, CAI (100 mg/kg) or PEG-400 was given daily from PND-17 to -19, and mice were killed on PND-20.

results. A 92% inhibition of neovascular cell nuclei on light microscopy was observed in mice treated with CAI in group A (P < 0.0001). Fluorescein-perfusion demonstrated a similar profound inhibition of neovascular frond formation in CAI-treated mice in group A. In group B, after neovascular fronds had already formed, CAI administration reduced neovascular cell nuclei by 72% (P < 0.001). Fluorescein perfusion studies confirmed that CAI induced regression of neovascular fronds. Similar amounts of posterior retinal ischemia were observed in all mice at both PND-17 and -20. In group A and B animals, CAI increased immunoreactivity of a cellular survival factor, Bcl-2, decreased TUNEL-positive cells, and after CAI treatment the normal morphology of the inner retina remained intact.

conclusions. CAI almost completely abolished retinal neovascularization in group A, and neovascular fronds involuted after treatment with CAI in group B. Thus, CAI is a potent inhibitor of ischemia-induced neovascularization and also imparts retinal neuroprotection after ischemic injury.

Pathologic ocular neovascularization is the proximal cause of the severe visual loss observed with both age-related macular degeneration (AMD) and diabetic retinopathy (DR), the two most common causes of legal blindness in Europe and the United States. AMD affects one in three people older than 70 years. ¹ Choroidal neovascularization (CNV) affects approximately 10% to 15% of individuals with AMD and is responsible for most of the irreversible visual loss that leads to legal blindness. ¹ The abnormal new blood vessels have a predilection to form beneath the foveal center (subfoveal CNV), which almost invariably causes significant visual loss. Although the Macular Photocoagulation Study demonstrated that subfoveal CNV can be eradicated with laser, the best corrected visual acuity after 12 months is usually worse than 20/200, whether the subfoveal CNV is treated or observed. ^{2 3 4 5} More recently, a less destructive laser modality, photodynamic therapy (PDT), demonstrated a modest relative preservation of central vision and retinal function, compared with no treatment. One year after PDT treatment the average vision was 20/160 compared with 20/200 in eyes that were not treated by PDT. ^{6 7 8 9 10} This is the first report of relative preservation of vision in the presence of subfoveal CNV; however, these results are far from ideal. Many other therapeutic interventions are currently being studied that include transpupillary thermoplasty (TTT), submacular surgery, foveal translocation, radiation therapy, and antiangiogenic factors. ^{11 12 13 14} Diabetic retinopathy is the leading cause of severe visual loss among working-age adults (28–64 years) in Europe and the United States, ^{15 16 17} and pathologic neovascularization is responsible for most of the severe vision loss observed. 

Many angiogenesis-modulating factors are under investigation for the treatment of a variety of human diseases that include metastatic cancer and coronary artery, cerebral, and peripheral vascular diseases as well as neovascular eye disease. ^{13 18 19 20 21 22 23 24 25 26} Antiangiogenic therapy has been most well studied in the treatment of metastatic cancer, as there are currently well over 100 agents that are in various phases of investigation. These agents can be divided into many general categories that include matrix metalloproteinase inhibitors (MMPIs), ^{25 27 28 29} agents that block the action of integrins, ³⁰ agents that block endothelial cell division, ^{22 31} and agents that antagonize the function of vascular endothelial growth factor (VEGF) or other cytokines. ^{19 32 33 34 35} Relative retinal ischemia induces VEGF expression, which then acts as an essential mediator of neovascularization in ischemic retinopathies such as diabetic retinopathy and retinal vein occlusion. ³⁶ In both animal models and human disease, VEGF is significantly upregulated in neovascular growths, and blockade of VEGF has been shown to inhibit pathologic angiogenesis. ^{32 36 37 38} In addition, relative posterior segment ischemia appears to play a significant role in formation of CNV and glaucomatous optic neuropathy. ^{38 39 40 41 42} Therefore, we have become interested in the development of molecules that block the downstream effects of relative posterior segment ischemia, as putative inhibitors of pathologic neovascularization and vision loss. 

Carboxyamido-triazole (CAI) is an antiangiogenic factor that is undergoing clinical trials for the treatment of various human cancers. ^{43 44 45 46 47} CAI, originally developed as a coccidiostat, was subsequently shown to have potent antiproliferative and antimetastatic effects in many animal models. ^{48 49 50 51 52 53} CAI decreases intracellular calcium by inhibition of non–voltage-gated calcium channels, the predominant calcium channel type present on endothelial cells. ^{48 50 54 55 56 57} This reduction of intracellular calcium diminishes endothelial cell proliferation and division. In addition, CAI antagonizes expression of many proangiogenic cytokines, which includes VEGF, as well as their downstream intracellular effects. ^{50 56 58 59} Furthermore, CAI inhibits the growth and metastasis of many tumors that are associated with local tumor ischemia, increased VEGF expression, and pathologic angiogenesis. ^{51 52 60 61 62} These data suggest that CAI may be a candidate as an inhibitor of pathologic ocular neovascularization. Therefore, we tested CAI in a mouse pup model of hyperoxia-induced ocular neovascularization, in which posterior retinal ischemia mediates pathologic angiogenesis. We now report that CAI was a potent inhibitor of ocular angiogenesis in this model. CAI nearly completely abolished formation of neovascular fronds and caused a dramatic and significant regression of preformed neovascular fronds. Immunohistochemistry and light microscopy studies established that this event was associated with upregulation of intracellular Bcl-2, diminished TUNEL-positive cells, and relative protection of normal cellular morphology in the ischemic retina. 

Methods

Mouse Model, CAI Administration, and Tissue Preparation

A neonatal mouse model of hyperoxia-induced retinal neovascularization was accomplished by placing postnatal day (PND)-7 mice in 75% to 85% oxygen for 5 days. On PND-12, the mice were abruptly placed in room air (21% oxygen). ⁶³ Approximately 10 to 12 litters of C57BL/6J mice were used with three to five mice per litter. The mice were housed in a standard 12-hour light–dark cycle. The protocol was approved by the University of Iowa Hospital Animal Care Department, and conformed to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Two different administration schemes were performed and animals were divided into groups A and B, according to the schedule of treatment. The mice were given the previously studied dose of CAI (100 mg/kg per day) or an equivalent volume of the vehicle control, polyethylene glycol-400 (PEG-400) by oral gavage. In group A, CAI was administered from PND-14 to -16, and the mice were killed early on PND-17. In group B, the same doses were used, however CAI or PEG-400 was dispensed on PND-17 through PND-19. The mice were then killed early on PND-20. After death, eyes were enucleated, fixed, and viewed by light microscopy. Some animals were perfused with fluorescein-dextran, molecular weight 2 × 10⁶, before enucleation. 

Neovascular Cell Quantitation

After enucleation, eyes were immersion fixed at 22°C in 4% freshly prepared paraformaldehyde in 0.1 M phosphate-buffered saline (PBS; pH 7.4) with light agitation. The tissue was then washed several times over 4 to 6 hours in 0.1 M PBS, routinely embedded in paraffin, and sectioned at 4 μm. The counted sections were approximately 30 μm apart to avoid the counting of the same cell on multiple sections. Nuclei from three to four sections on either side of the optic nerve were counted, and the average number of nuclei per section was recorded. The observer for the cell counts was masked. In group A, 10 eyes were counted for both CAI-and vehicle-treated control animals. In group B, 10 and 8 eyes were counted for control vehicle- and CAI-treated animals, respectively. Retinal lengths for sections were measured in nine animals; four in group A (two vehicle control, two CAI treated), four in group B (two vehicle control, two CAI treated), and one normal PND-17 animal. The retinal lengths were equal among all group A animals and in the PND-17 animal. Retinal lengths in group B were overall slightly longer than those in group A, but among group B animals, the retinal lengths were equal in vehicle- and CAI-treated animals. 

Confocal Immunofluorescence Analysis

Paraffin-embedded sections of 4 μm were cut, hydrated, and pretreated by boiling on 0.01 M citrate buffer (pH 6.0) for 10 minutes at 95°C. The sections were then cooled and blocked in 0.1 M PBS with 5.0% normal donkey serum and 1.0% bovine serum albumin. After they were blocked, the sections were incubated overnight in a rabbit anti-Bcl-2 antibody (Santa Cruz Biotech, Inc., Santa Cruz, CA) diluted 1:100 at 4°C. After they were washed in PBS, sections were incubated for 1 hour in a mixture of donkey anti-rabbit-CY3 (1:2000; Jackson ImmunoResearch, Inc., West Grove, PA) and a nuclear counterstain (YO-PRO1; Molecular Probes, Inc., Eugene, OR). Sections were mounted and imaged with a scanning confocal head (MRC 1024; Bio-Rad, Hercules, CA) coupled to an upright microscope (BX50; Olympus, Lake Success, NY) with an Ar-Kr laser, using the 488- and 568-nm excitation lines (University of Vermont Microscope Imaging Facility). Images were acquired on computer (LaserSharp software ver. 3.2; Bio-Rad), merged, and formatted in image management software (Photoshop; Adobe, Mountain View, CA). 

Quantitation of Bcl-2

A semiquantitative comparison of Bcl-2 immunofluorescence intensity was accomplished by batch staining and sampling by confocal microscopy for Bcl-2 (CY3) and a nuclear counterstain (YO-PRO1; Molecular Probes, Inc.). Samples were imaged on the confocal microscope with a 60× objective lens (NA 1.4; PlanApo, Carl Zeiss Meditec, Thornwood, NY) and 568- and 488-nm excitation lines of an Ar-Kr laser. In each field, the microscope was focused to maximize the number of inner retinal cells optically sectioned through the middle of the nucleus. All confocal imaging parameters were identical for each imaged field with four nonoverlapping fields of ∼190 mm² captured in each eye. Gray-scale images (512 × 512 pixels) were transferred to a computer (Power Macintosh G5; Apple Computer, Cupertino, CA) running NIH Image (ver. 1.63) for image analysis. For each field studied, the Area-Measure tool was used to determine the mean pixel intensity (range, 0–255 gray-scale levels) of the Bcl-2 immunofluorescence in a randomly-picked 20 pixel area of the inner plexiform layer (IPL). The final corrected intensity was calculated for each cell after subtraction of the field background fluorescence. The probabilities were calculated by Student’s t-test. 

TUNEL Assay and Quantitation

For determination of the apoptotic activity of the retina, a modified TUNEL staining protocol was used to mark cells fluorescently with fragmented DNA strands. A kit (Dead-End TUNEL Kit; Promega Corp., Madison, WI) was used with the initial steps of the protocol performed according to the manufacturer’s suggestions. Modification of the kit (TdT-based end-labeling with biotin-UTP as the labeled nucleotide) included incubation with streptavidin-CY3 (1:2000; Jackson ImmunoResearch, Inc.) in the final labeling step. Sections were counterstained with 0.5 μg/mL Hoechst 33258; mounted in aqueous medium (Aqua-Polymount; Polysciences, Inc., Warrington, PA); and imaged on an epifluorescence microscope (Universal; Carl Zeiss Meditech) using a 63× objective lens (PlanApo; Carl Zeiss Meditec) and captured with a color charge-coupled device (CCD) camera (spot-RT; Diagnostic Instruments, Inc., Sterling Heights, MI). TUNEL-positive cells were quantitated on multiple sections in multiple animals from both control vehicle- and CAI-treated animals in groups A and B. A total of 5,000 to 10,000 cells were analyzed in each group in both the early- and late-intervention animals, and the cells were counted in a masked fashion. The probabilities were calculated by the Fisher exact test. 

Results

Effect of CAI on Neovascular Formation in Group A Animals

Light microscopy was performed on sections of multiple organs of animals treated with CAI or the vehicle control and showed no evidence of drug-related damage, which supports previously reported toxicity data. Representative eyes that were harvested from animals treated with CAI or PEG-400 vehicle are shown in Figure 1 . Group A consisted of eyes from mice treated daily for 3 days with CAI or PEG-400 beginning on PND-14. Multiple tufts of abnormal vessel-associated cells were easily detected in vehicle control eyes. Vascular lumens within the inner nuclear layer (INL) were observed to traverse toward these neovascular tufts (Figs. 1A 1B) . In contrast to the active retinal neovascularization found in eyes of control animals, eyes of mice that received CAI showed a statistically significant and profound reduction in the number of neovascular nuclei (Figs. 1C 1D 1E) . The capillaries in the posterior retina of eyes of CAI-treated mice usually appeared normal (Figs. 1C 1D , arrowheads). However, peripheral retinal vessels sometimes appeared dilated, with some abnormal endothelial cells (Fig. 1E) . Figure 2 shows quantitation of nuclei associated with angiogenic fronds. Cell nuclei associated with vessels that penetrated the internal limiting membrane were counted. Neovascular nuclei were observed in all sections of eyes of animals that received vehicle control alone, whereas many sections of eyes of animals that were treated with CAI did not contain any neovascular cell nuclei. A statistically significant reduction (92%, P < 0.0001; 10 eyes per group) was attained with CAI administration. 

Fluorescein perfusion demonstrated a similar profound inhibition of neovascular fronds for CAI-treated mice in group A (Fig. 3) . Figures 3A and 3B demonstrate the normal, rich capillary pattern in PND-17 mice that were not exposed to a hyperoxic environment. In contrast, a marked dropout in the capillary pattern are found in animals exposed to transient hyperoxia (Figs. 3C 3D 3E 3F) . The peripheral 65% of the retina was vascularized in both vehicle control and CAI-treated animals (not shown). This degree of retinal vascularization is within the range previously reported for the rodent retinal neovascularization models. ^{64 65 66} Extensive fronds of abnormal vessels were found that arose from the peripheral retina of mice treated only with PEG-400 (Fig. 3D , arrowheads). In addition, posterior retinal neovascularization was observed (Fig. 3C , open arrow denotes neovascularization and closed arrow shows the optic nerve head). Finally, the abnormal vasculature shown in Figures 3C and 3D , was associated with leakage of fluorescein dye and higher background fluorescence. This association between dye leakage and neovascularization is a well-established finding in clinical fluorescein angiography. Although a similar amount of capillary dropout and ischemia was observed in mice after CAI treatment, no posterior or peripheral neovascularization was detected (Figs. 3E 3F) . Instead, some dilated and tortuous capillaries were found in the peripheral retina, which supports the light microscopy analysis. Moreover, the background or fluorescein leakage was minimal in CAI-treated eyes, which argues against occult neovascularization on these slides. 

Effect of CAI on the Neovascular Response in Group B Animals

Group B animals were allowed to undergo ischemic retinal damage and neovascular formation before intervention with CAI treatment or PEG-400 vehicle. CAI was highly effective in reducing existent neovascular damage. Figures 4A and 4B (arrowheads) depict the active neovascularization observed in PND-20 mice treated only with the vehicle. In contrast, mice treated with CAI from PND-17 through PND-19, showed a marked reduction in abnormal vascular nuclei (Figs. 4C 4D 4E) . Moreover, when neovascularization was observed, it often appeared as a small fibrotic nodule of cells (Figs. 4D 4E) . The arrowhead in Figure 4E identifies a cell with a fragmented nucleus, suggestive of either cellular death or necrosis. Figure 5 presents the quantitation of neovascular nuclei counted in group B animals. CAI treatment led to a 72% reduction of neovascular cell nuclei, (P < 0.001; 10 vehicle-treated control eyes and 8 CAI-treated eyes). 

Fluorescein perfusion studies of representative eyes in group B are shown in Figure 6 . A large frond of abnormal new vessels that emanated from the optic disc (Fig. 6A , arrowhead). An area of peripheral neovascularization is shown in Figures 6B and 6C (arrowheads). As observed in group A, similar amounts of posterior capillary dropout were found on both the control vehicle- and CAI-treated eyes, so that approximately 65% of the peripheral retina was vascularized in all group B animals (data not shown). Also, similar to group A, dilated and tortuous peripheral retinal vessels were observed (Fig. 6E , arrowheads). Neovascularization detected in CAI-treated eyes appeared fibrotic, nodular, and without significant dye leakage (Fig. 6F) . These are well-established signs of regressed neovascularization in clinical fluorescein angiography. The characteristics of the regressed vessels on fluorescein perfusion corroborated the light microscopy analysis and were distinct from the active, lacy appearance of the peripheral neovascular fronds found in control animals (Fig. 6C) . These data demonstrate that CAI can both inhibit the formation of abnormal neovascular structures and cause previously formed neovascular fronds to regress. Thus, CAI demonstrated potent angiogenic inhibition in this mouse hyperoxia model of retinal neovascularization. 

Effect of CAI on Bcl-2 Protein in the IPL, on TUNEL-Positive Cells, and on INL Nuclear Appearance during Relative Retinal Ischemia

In addition to effectively preventing formation and causing involution of neovascularization, CAI treatment regulated apoptosis and cell survival in the ischemic retina. A relatively low level of Bcl-2 antigen in the normal PND-17 mouse retina, and a normal morphology of the inner retina, where nuclei of the INL have a relatively consistent pattern of heterochromatin was observed (data not shown). Group A animals treated with PEG-400 alone possessed a similar low level of Bcl-2 (Fig. 7A) . The nuclei were counterstained with a counterstain (YO-PRO1; Molecular Probes, Inc.). A light micrograph of a serial section close to the area shown in Figure 7A is shown in 7B, where an arrow delineates a cluster of abnormal cells in the INL with vacuolations and pyknotic nuclei. These findings indicate that the low Bcl-2 expression, while similar to that in normal (nonhypoxic) PND-17 retinas, is inappropriately low in the setting of hypoxic stress to these cells. Administration of CAI on the hypoxic background led to a small increase in the relative amount of Bcl-2 protein in the inner retina (Fig. 7C) . When the level of Bcl-2 immunoreactivity was quantified by confocal microscopy, there were no significant difference between PEG-400–and CAI-treated animals; however, these data represent a relatively small sample size (see Fig. 9 ). Light microscopic analysis of the inner retina sectioned nearby demonstrates the maintenance of the normal cellular architecture observed in nonhypoxic PND-17 mice (Fig. 7D) , suggesting a neuroprotective effect of CAI in group A animals, which was also observed in group B animals. There is almost no detectable Bcl-2 immunoreactivity in group B animals treated with PEG-400 alone (Fig. 8A) . The green arrows indicate active-appearing neovascularization in Figures 8A and 8B . Light microscopy of a serial section in the vehicle control animals again showed an abnormal appearance of the INL, with cellular vacuolation and pyknotic nuclei (Fig. 8B , black arrows). CAI markedly intensified Bcl-2 immunoreactivity in group B animals (Fig. 8C) . Confocal microscopy and fluorescence quantitation demonstrated a greater than twofold increase in Bcl-2 immunoreactivity after administration of CAI (P = 0.002; Fig. 9 ). The normal histologic appearance of the inner retina is also recovered in the PND-20 animals after CAI treatment (Fig. 8D) . 

In addition, CAI decreased TUNEL-positive cells in the inner and outer nuclear layers in both group A and B animals. Figure 10 shows representative sections. At postnatal day 17, group A TUNEL-positive cells were rarely observed in animals not exposed to oxygen (Fig. 10A) or after transient hyperoxia and treatment with CAI (Fig. 10C) . However, TUNEL-positive cells were observed in both the inner and outer nuclear layers after exposure to transient hyperoxia with the vehicle alone (Fig. 10B , arrowheads). Similarly, in group B, TUNEL-positive cells were observed in animals with vehicle control (Fig. 10D , arrowheads), but after CAI administration TUNEL-positive cells were diminished by approximately twofold (Fig. 10E , Fig. 11 ). Multiple TUNEL-positive cells were observed in pancreatic tissue from adult Zucker diabetic fatty (ZDF) rats (positive control, not shown). Quantification of TUNEL from group A animals show that TUNEL-positive cells after CAI administration were increased twofold, compared with normal PND-17 mice, however animals treated with PEG-400 alone had six- to sevenfold more TUNEL-positive cells than did animals treated with CAI alone (P = 0.001; Fig. 11 ). Administration of CAI to group B animals caused a twofold reduction in TUNEL-positive cells (P = 0.008; Fig. 11 ). Thus, CAI treatment not only increased the relative amount of a cellular survival factor in the retina after ischemic damage, but also permitted recovery of a relative normal inner retinal morphology. The decrease in TUNEL-positive nuclei observed after CAI treatment further corroborates the neuroprotective effects of this molecule. 

Discussion

An ideal candidate molecule for the treatment of retinal and/or choroidal neovascularization should fulfill several criteria: (1) It must be relatively nontoxic; (2) it must have reasonable bioavailability in ocular tissues; (3) it should have some physiologic basis to be studied; and (4) it must not adversely interfere with other physiological neovascular responses such as wound healing and coronary vascular remodeling. 

CAI Characteristics and Bioeffects

We selected CAI to study for three major reasons. First, CAI is an antiangiogenic agent, and the biochemical effects mediated by CAI inhibit many processes that are critical in the pathophysiology of human ocular neovascularization and retinal ischemia. Second, CAI has been safely administered systemically in most of more than 300 patients. Third, CAI is a relatively small molecule that crosses the blood–brain barrier, ⁴⁵ and should possess favorable local ocular diffusion characteristics. 

CAI exerts antiproliferative and antimetastatic effects in many animal models of cancer by blocking tumor angiogenesis. ^{51 52} This molecule mediates antiangiogenic properties in many animal models. ^{48 51 52 54 67} CAI virtually abolished formation of ischemia-driven neovascularization in this rodent model. VEGF, which appears to be an important mediator of human ocular neovascularization, also is involved in the angiogenic cascade of this model, so that VEGF expression is induced, and introduction of VEGF antisense RNA or blocking antibodies into the eye causes regression of neovascularization. ^{13 32 33 68 69 70} In this mouse model, VEGF blockade by antisense mRNA or antibody caused an approximate 50% reduction of neovascular formation, whereas CAI administration led to an approximate 90% reduction (Fig. 12) . CAI inhibits expression of other angiogenic cytokines in addition to VEGF, including HIF-1α, MMP-2, c-fos, and VL30, and modulates intracellular events in the proangiogenic pathway, including IP3 metabolism and tyrosinase kinase activity. ^{50 54 59 67 71} Although there are some dissimilarities in the processes of retinal and CNV such as the different extracellular environment, integrin expression, and relative ischemia, ^{38 72 73 74 75} the mechanisms are likely to be closely enough related so that a molecule that inhibits expression of many angiogenic cytokines such as CAI may be effective in the treatment of both these neovascular processes. In addition, CAI negatively affects many proangiogenic processes in cultured RPE and choroidal endothelial cells, so that it decreases proliferation and attachment of both these cell types, as well as cellular migration in response to fibronectin. Further, CAI inhibits VEGF- and bFGF-mediated secretion of MMP-2 from choroidal endothelial cells. ⁷⁶ Hence, CAI appears to be an attractive candidate molecule to treat pathologic ocular neovascularization in human disease, based on its high efficacy in this animal model of retinal neovascularization, its antiangiogenic properties demonstrated in other animal models, the antiproliferative effects on cultured RPE and choroidal endothelial cells, and its inhibitory effects on expression of multiple angiogenic cytokines. 

CAI has undergone testing in multiple clinical oncology trials. ^{43 45 46 47 77 78} Long-term administration of up to 3 years has shown it to be well tolerated. ^{43 45 46 47 77 78 79} It is orally bioavailable and reaches a plateau steady state serum concentration within 2 to 3 weeks of daily treatment. ^{44 45} CAI has been shown to have limited systemic toxicity at circulating concentrations at or above those necessary to reduce angiogenic activity. ^{43 45 46 47 77 78 79} Major side effects requiring dose modification are rare and include mild and reversible cerebellar ataxia in less than 1%, rapidly reversible sensory peripheral neuropathy in less than 3%, and exacerbation of depression in 1% to 3% of patients. ^{43 45 46 47 77 78 79} Two cases of loss of vision occurred in phase I study of CAI dispensed in PEG-400 gelatin capsules. ⁴⁵ Retinal examination was not described as abnormal, and many different mechanisms can contribute to vision loss associated with metastatic disease. Currently, more than 300 patients are receiving daily CAI, some for more than a year, with only those two cases of completely reversible vision loss. CAI does not appear to affect normal vasculogenesis, as in this mouse model, the relative peripheral retinal vascularization was not affected by CAI administration. Furthermore, the healing from lacerations to emergency surgery in humans is not altered when CAI is administered, which implies that CAI does not affect the physiological neovascularization associated with wound healing. ^{43 44 45}  

Local Ocular Bioavailability of CAI

Drug delivery to the posterior segment of the eye can be accomplished by four different methods: systemic delivery, topical drops, intraocular or intravitreal delivery, and transscleral delivery, such as periocular injection or implants. ⁸⁰ If the sclera is relatively permeable to a drug, then multiple schemes can be developed to deliver a sustained, high concentration of that agent to the posterior segment, while avoiding potential complications of intraocular surgery or frequent injections such as cataract formation, glaucoma, retinal detachment, and endophthalmitis. ^{80 81 82 83}  

CAI is a smaller molecule than most of the antiangiogenic molecules that are currently under study to treat pathologic ocular neovascularization. We have shown that CAI diffuses across human sclera efficiently, with kinetics similar to fluorescein, rhodamine, and dexamethasone (Cruysberg et al., manuscript submitted). After a single 5-mg peribulbar injection in humans, dexamethasone has been detected at 13 ng/mL in the vitreous cavity, ^{84 85} which results in a vitreous concentration of approximately 0.10 to 0.13 μM. The serum half-life of CAI is four to five times longer than dexamethasone. Attainable and bioeffective circulating concentrations of CAI are in the 1- to 10-μM range so that we postulate based on scleral permeability and serum half-life that CAI may be effectively delivered transsclerally to the subretinal space and the vitreous cavity. Thus, a variety of delivery schemes can be designed to provide relatively high concentrations of CAI in the posterior segment, while minimizing intraocular and systemic toxicity. 

Relative Potency of CAI in Rodent Models of Ischemia-Induced Retinal Neovascularization

More than 25 potential antiangiogenic compounds have been tested in the mouse or rat pup model of oxygen-induced retinopathy. Figure 12 summarizes the relative efficacy of many of these compounds. ^{25 29 32 68 69 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103} Because of species and oxygen variation in the mouse and rat models, small differences probably exist in the relative neovascular stimuli, and the different assessment systems used to quantitate neovascularization may yield mild variances in data analysis. 

The relative number of neovascular cell nuclei obtained in control animals in this study is comparable to the number measured by other investigators. The 92% neovascular inhibition observed in group A animals is almost 10% greater than the most potent antiangiogenic effect measured to date (Fig. 12) . Moreover, most compounds that include anecortave and VEGF-blocking antibodies reduce abnormal vessel formation in the range of 45% to 67%. ^{25 32 68} One of the more potent compounds measured, a cyclic RGD peptide integrin antagonist, demonstrated angiogenic inhibition of 75% when administered similar to group A animals. However, no significant antiangiogenic effect was observed when this integrin antagonist was administered according to the group B schema, ⁸⁶ in contrast to the 72% reduction of existing neovascular cell nuclei caused by CAI. Therefore, CAI can be considered a promising agent when compared with other antiangiogenic factors in rodent models of ischemia-induced retinal neovascularization. 

Neuroprotection and Posterior Segment Disease

Injury to the inner retina has been associated with many vision-threatening diseases, such as diabetic retinopathy and glaucoma. ^{39 104 105 106 107 108} Therefore, identification and characterization of factors that may protect the inner retina is important for the development of novel treatments. Activation of ionotropic glutamate (N-methyl-d-aspartate; NMDA) receptors on inner retinal cells is a crucial pathophysiological event associated with inner retinal and optic nerve damage observed in glaucoma. ^{39 41 108} Receptor activation leads to calcium influx, generation of free radicals, and, ultimately, cellular damage and death. ^{39 40 41 108 109} Brimonidine, an NMDA calcium channel inhibitor that effectively treats glaucoma, inhibits inner retinal cell death and up regulates Bcl-2 in a rat model of inner retinal ischemia. ⁴⁰ Similar to brimonidine, CAI inhibits calcium influx and increased the amount of Bcl-2 protein in the inner retina. Many mechanisms of retinal cell damage and death have been associated with calcium influx, which makes CAI a logical agent for consideration. ^{39 110 111 112} Calcium influx also has been implicated in apoptosis associated with photoreceptor degenerations. ^{110 111 112} CAI administration reduced TUNEL-positive cells in group A by six- to sevenfold (P = 0.001) and in group B by twofold (P = 0.008). Fewer TUNEL-positive cells were observed in this report than after optic nerve crush injury; however, a similar number of TUNEL-positive cells has been observed by other investigators in the mouse transient hyperoxia model. ¹⁰⁵ Taken together, reduction of TUNEL-positive cells, protection and/or rescue of normal cellular architecture, and increase of a cellular survival factor fulfill important criteria to demonstrate neuroprotection. PEDF is the only other factor that has been shown to have retinal neuroprotective and antiangiogenic properties, and these properties have been demonstrated by methods similar to those reported in the current study. ^{26 87 113 114} The protection of retinal cells from ischemic injury coupled with the reversal of the abnormal cell nuclear appearance observed after ischemic damage indicates that the beneficial effects of CAI on the ischemic retina are not limited to simple angiogenic inhibition, and to date, PEDF is the only other molecule that has demonstrated similar beneficial effects. 

Current Treatment and Clinical Trials of Antiangiogenic Molecules for Posterior Segment Ocular Neovascularization

Current approved treatment of both diabetic retinopathy (DR) and age-related macular degeneration (AMD) relies on either threshold or subthreshold laser energy. Threshold laser is inherently destructive, and there is a need for less destructive approaches. ^{2 3 4 5 115} Ocular photodynamic therapy (OPT) is approved for the treatment of CNV underneath the foveal center, but has demontrated only limited efficacy. Although OPT is an effective treatment for classic subfoveal CNV, the results are far from ideal. ^{6 7 8 9 10 11 116 117 118 119} Other treatments such as TTT and foveal translocation are less well-characterized and are probably less effective. ^{12 13 15 120 121 122} Four antiangiogenic agents are under clinical investigation, and the Phase II/III data for the treatment of CNV are encouraging. ^{28 34 35} The first of these molecules is the orally administered protein kinase C inhibitors used primarily to treat diabetic retinopathy. ¹²³ The latter three molecules are under investigation for treatment of neovascular AMD. Two agents target VEGF. ^{13 34 35 124} One molecule is an RNA aptamer, whereas the other is a blocking antibody fragment. ^{13 34 35 124} Both currently are administered by intravitreal injections at a frequency in the range of every 6 weeks, and neither possesses favorable transscleral delivery properties. Finally, anecortave acetate is an antiangiogenic steroid derivative, that is administered every 6 months by juxtascleral injection, ²⁸ and therefore there is only one compound under investigation for the treatment of posterior segment neovascularization that is delivered transsclerally. 

Conclusions

In earlier studies, CAI has been shown to be relatively well tolerated with systemic administration and also to possess characteristics favorable for transscleral delivery. In the current study, we demonstrate that CAI exerts a potent antiangiogenic effect to inhibit formation of neovascularization and to hasten regression of preformed neovascular fronds. Moreover, CAI either directly or indirectly protects the inner retina after ischemic damage. Thus, CAI fulfills the four criteria for candidate compounds to treat posterior segment neovascularization associated with common blinding diseases such as diabetic retinopathy, retinal vascular occlusion, and AMD and should be considered as a clinical antiangiogenic ophthalmic agent. 

 Supported by the University of Iowa Retina Research Foundation and an unrestricted grant from Research to Prevent Blindness (AJF).

 Submitted for publication October 10, 2003; revised February 27 and June 30, 2004; accepted July 7, 2004.

 Disclosure: A.J. Franklin, None; T.L. Jetton, None; C.L. Kuchemann, None; S.R. Russell, Pfizer (C); E.C. Kohn, None

 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.

 

Figure 1.

 

Light microscopy analysis of neonatal mouse retina in group A. Photomicrographs of PND-17 mice subjected to transient hyperoxia and treated with either PEG-400 vehicle control (A, B) or CAI (100 mg/kg per day from PND-14 through PND-16) (C–E). (A, arrow) A frond of neovascularization (×40) that is shown at higher magnification in (B). Note the abnormal vascular morphology extending through the internal limiting membrane. In contrast, mice treated with CAI had essentially no vascular cell nuclei penetrating the internal limiting membrane. (C, boxed area) Normal-appearing posterior retinal capillary shown in higher power (×100) in (D). Some peripheral vessels were large and associated with abnormal endothelial cells (E); however, cells rarely penetrated the internal limiting membrane.

Figure 2.

 

Average number of neovascular nuclei in PEG-400 vehicle- and CAI-treated mice in group A. Histogram demonstrates the marked decrease in neovascular cell nuclei, counted as nuclei associated with blood vessels that have extended through the internal limiting membrane. Numbers reflect the average number of neovascular cell nuclei found in 10 eyes of mice given either PEG-400 or CAI. Six to eight sections taken at different levels were counted for each mouse eye. CAI decreased the number of neovascular nuclei by 91% (P < 0.0001 by exact Wilcoxon rank sum).

Figure 3.

 

Fluorescein-perfused neonatal mouse retina in group A. PND-17 C57BL/6J mice were killed and perfused with fluorescein-dextran (MW 2 × 10⁶). (A, B) Flatmounted normal PND-17 retina demonstrating normal retinal vasculature near the optic nerve head (A, arrow) and in the retinal periphery (B, arrow). In contrast, mice exposed to transient hyperoxia had massive capillary closure in the posterior retina near the optic nerve head, whether treated with PEG-400 control alone (C, D) or with CAI (E, F). This amount of capillary nonperfusion is a strong angiogenic stimulus, and neovascularization was found both near the optic nerve head and in the peripheral retina of mice treated with the PEG-400 vehicle control (C, D). (C, solid arrow) optic nerve head; (C, D, open arrows) neovascular fronds. However, in mice treated with CAI (100 mg/kg per day from PND-14 through PND-16) (E, F), the neovascular response was attenuated (E, arrow) optic nerve head, and retinal capillaries were observed in the retinal periphery (F, arrow).

Figure 4.

 

Light microscopy of neonatal mouse retina in group B. Photomicrographs of PND-20 mice subjected to transient hyperoxia and treated with either PEG-400 vehicle control (A, B), or CAI (100 mg/kg per day from PND-17 through PND-19) (C–E). (A, B) Neovascular fronds were present in mice treated with vehicle control alone. Similar to the findings in group A, large active fronds that extended through the internal limiting membrane were observed (arrows). These were markedly reduced in CAI-treated eyes. Moreover, often when neovascularization was observed, it appeared more nodular (D, arrow). On higher magnification, some cells were shown to contain fragmented nuclei, which suggested that some cellular death was occurring. Magnification: (A–D) ×40; (E) ×100.

Figure 5.

 

Average number of neovascular nuclei in PEG-400 vehicle and CAI-treated mice in group B. As found in group A, CAI caused a significant inhibition of neovascular cell nuclei (72% decrease, P < 0.001, by exact Wilcoxon rank sum). Nuclei were counted as described for group A. These data demonstrate that CAI not only inhibited formation of neovascularization (group A), but also reduced the amount of neovascularization that was already present (group B).

Figure 6.

 

Fluorescein-perfused neonatal mouse retina. In group B, PND-20 C57BL/6J mice were killed and perfused with fluorescein-dextran (MW 2 × 10⁶), Similar to group A, mice exposed to transient hyperoxia had massive capillary closure in the posterior retina near the optic nerve head. (A) Severe optic-disc–associated neovascularization (arrow) was observed in animals treated with the vehicle control. (B, C) More peripherally, active, lacy-appearing neovascular fronds (arrows) were found. (D–F) In contrast, mice treated with CAI (100 mg/kg per day from PND-17 through PND-19) had significantly less neovascularization. Arrowhead in (D) denotes optic nerve; arrowhead in (E) depicts a peripheral capillary network. In addition, when neovascularization was observed, it appeared fibrotic, with vessel simplification, and did not demonstrate significant leakage (F, arrowhead).

Figure 7.

 

Immunolocalization of Bcl-2 in PND-17 and group A animals. Relatively low levels of Bcl-2 antigen were found in the normal PND-17 mouse. (A) Similarly, low levels of Bcl-2 immunoreactivity (Cy-3, red) were found in animals exposed to hyperoxia and treated with the vehicle control only. (B) Light micrograph of a serial section close to the area shown in (A). (B, arrow) Cluster of abnormal cells with vacuolations and pyknotic nuclei. (C) CAI treatment led to a mild increase of Bcl-2 immunoreactivity in the inner retina. (D) Moreover, light microscopy of serial sections demonstrated that the normal cellular architecture of the INL observed in PND-17 mice was preserved. (A, C) Nuclei counterstained with YO-PRO1. Magnification: ×63.

Figure 8.

 

Immunolocalization of Bcl-2 in PND-17 in group B animals. (A) There was very little detectable Bcl-2 immunoreactivity (Cy-3, red) in group B animals treated with the vehicle alone. (A, B, green arrows) Active-appearing neovascular frond. Light microscopy showed an abnormal appearance of the INL, with cellular vacuolation and pyknotic nuclei (black arrows) similar to that observed in the vehicle-treated group A animals. (C) Upregulation of the Bcl-2 antigen observed after CAI. Even at PND-20, the normal histology of the inner retina was retained with CAI treatment (D). (A, C) Nuclei counterstained with YO-PRO1. Magnification: ×63.

Figure 9.

 

Quantitation of Bcl-2 immunoreactivity in group A and B animals. Relative luminescence of Bcl-2 antigen was calculated after confocal microscopy. Confocal images from four nonoverlapping fields of ∼190 mm² were captured in each eye. Gray-scale images (512 × 512 pixels) were subjected to computer image analysis. For each field studied, the Area-Measure tool (NIH Image) was used to determine the mean pixel intensity in a randomly picked 20-pixel area of the IPL. Corrected intensities after subtraction of the field background fluorescence are depicted. There was only a mild increase in Bcl-2 immunoreactivity after CAI treatment in group A (P = 0.81, Student’s t-test). However, after CAI in group B, Bcl-2 immunoreactivity was increased almost twofold compared with the PEG-400 vehicle control animals (P = 0.002, Student’s t-test).

Figure 10.

 

CAI decreased TUNEL-positive cells in group A and B animals. There are almost no TUNEL-positive cells in the normal PND-17 mouse (A) or after transient hyperoxia and CAI treatment in group A animals (C). However, TUNEL-positive cells were observed in the inner and outer nuclear layers after transient hypoxia and vehicle alone in group A animals (B, arrows). Similar results were observed in group B animals. TUNEL-positive cells were found after vehicle alone (D, arrows), whereas they diminished after CAI (E). Nuclei counterstained with Hoechst 33258. Magnification: ×63.

Figure 11.

 

Quantification of TUNEL-positive cells in group A and B animals. TUNEL-positive cells were counted in multiple sections from multiple animals in both CAI- and control vehicle-treated animals, so that 5,000 to 10,000 cells were counted for each subgroup. Only a small number of TUNEL-positive cells were observed in normal PND-17 mice. Initiation of CAI treatment several days after hypoxic insult led to an approximate twofold increase in TUNEL-positive cells compared with normal untreated PND-17 mice (P = 0.04, Fisher exact test). However, there was a six- to sevenfold increase in TUNEL-positive cells in group A animals treated with the PEG-400 vehicle control compared with animals treated with CAI (P = 0.001). Later CAI administration in group B animals caused a twofold reduction in TUNEL-positive cells compared with the vehicle control (P = 0.008).

Figure 12.

 

Relative potency of antiangiogenic compounds in rodent models of retinal neovascularization. The calculated percentage of neovascular inhibition of many antiangiogenic compounds is shown. Reference source follows each listing. R, rat model; M, mouse model; N, number of neovascular cells; C, clock hours of neovascularization; S, retinopathy score; and cited reference number.

The authors thank Thomas Weingeist, Henry Edelhauser, Raj Maturi, John Penn, and Randall Funderburk for their insight into this project and thoughtful comments on the manuscript; David Woods for expertise and help with for the figures; and Dawn Oh for help with the statistical and data analyses.