Localization of VEGF Receptor-2 (KDR/Flk-1) and Effects of Blocking It in Oxygen-Induced Retinopathy

Abstract

purpose. Vascular endothelial cell growth factor (VEGF) has been implicated in vascular development and in proliferative retinopathies. The goal of this study was to examine the immunohistochemical localization and relative levels of VEGF receptor-2 (KDR) in canine retina during postnatal vasculogenesis and during angiogenesis in oxygen-induced retinopathy (OIR) and to investigate the effects of neutralizing KDR on these processes.

methods. Eyes from normal dogs ranging from 1 to 22 days of age and age-matched oxygen-treated animals were snap frozen for immunohistochemical analysis with antibodies against human KDR. To examine the effects of blocking KDR, 6-day-old air-reared control and oxygen-treated animals were surgically implanted with slow release polymer pellets containing control IgG or anti-KDR. Material eluted from pellets was assessed using a binding assay (measures binding to soluble KDR) to determine the kinetics of anti-KDR release and endothelial cell proliferation to measure bioactivity. Animals were killed at 22 days of age and tissues examined with adenosine diphosphatase (ADPase) histochemical staining of blood vessels.

results. KDR immunoreactivity was only weakly associated with developing retinal vessels and was not observed in angioblasts throughout normal postnatal development. Immunoreactivity was very strong in reforming retinal vessels and intravitreal neovascularization in oxygen-treated animals. Anti-KDR had no effect on vessel morphology or growth in air-reared control animals. In oxygen-treated animals, anti-KDR significantly inhibited revascularization of the retina (P = 0.005) and formation of intravitreal neovascularization compared with control IgG pellet eyes (P < 0.04).

conclusions. KDR/Flk-1 was only weakly associated with normal developing primary retinal vessels but was strongly expressed by proliferating endothelial cells in reforming retinal vessels and intravitreal neovascularization after hyperoxic insult. Anti-KDR antibody delivered by slow-release pellets had no effect on normal vasculogenesis, but it inhibited the formation of intravitreal neovascularization and retinal vessel development in OIR. The study suggests that blocking KDR may be beneficial for treating pathologic angiogenesis in adult tissue.

Retinal blood vessel assemblage in the neonatal dog occurs by a process of vasculogenesis, a term referring to the formation of blood vessels by differentiation and organization of mesenchymal precursors or angioblasts. ¹ This process is thought to be driven by the release of a diffusable vasoformative factor by inner retinal cells in response to their increasing metabolic needs. ² Patz ³ and Ashton et al. ⁴ have suggested that relative oxygen deficiency is probably one of the normal stimuli for retinal vessel growth. In the premature human and newborn dog, sustained breathing of high oxygen produces a progressive constriction of the developing retinal vessels that eventually results in vaso-obliteration, or the irreversible closure of capillaries and degeneration of vasoformative cells. ^{5 6} Because the choriocapillaris is unaffected by hyperoxia, ⁶ the diffusion of oxygen from choroid is thought to be sufficient to satisfy the relatively low metabolic demands of inner retina and removes the stimulus for blood vessel growth. ³ The proliferative phase of oxygen-induced retinopathy occurs after withdrawal from the high oxygen environment, when it is thought that the inner retina, previously hyperoxygenated, becomes hypoxic as a result of decreased diffusion of oxygen from the choroidal blood and the obliteration of retinal vessels. This results in the apparent release of vasoformative factors that stimulate an abnormal overgrowth of blood vessels in inner retina that subsequently invade the vitreous, where the neovascularization leaks macromolecules and is prone to hemorrhage. Unlike normal vasculogenesis, however, these new vessels form by proliferation and migration of endothelial cells or by a process of angiogenesis. ⁷  

One of the angiogenic factors that is a potential candidate for the vasoformative factor during retinal development and oxygen-induced retinopathy is vascular endothelial cell growth factor (VEGF), a potent mitogen specific for some endothelial cells. ^{8 9} It stimulates collagenase production, ¹⁰ is upregulated by hypoxia, ¹¹ increases vascular permeability, and is angiogenic in vivo. ¹² Two of the more widely studied cell surface receptors for VEGF are VEGF receptor (VEGFR)-1 (or Flt-1) ¹³ and VEGFR-2 (or Flk-1) in mouse and KDR in human. ¹⁴ Activation of KDR/Flk-1 is thought to be responsible for stimulation of mitogenesis, whereas Flt-1 is responsible for chemotaxis. ¹⁵ VEGF has been implicated in retinal vasculogenesis ¹⁶ and in vasculogenesis in other organ systems. ^{17 18} VEGF has also been shown to be upregulated in animal models of oxygen-induced retinopathy. ^{19 20 21 22 23}  

In this study, we examined the localization and relative levels of KDR/Flk-1 in the normal developing canine retina and in the canine model of oxygen-induced retinopathy (OIR), using monoclonal antibodies specific for KDR. These well-characterized antibodies inhibit activation of KDR, compete on an equimolar basis with VEGF for binding to KDR, and inhibit signaling and mitogenesis of endothelial cells in response to VEGF. ^{24 25} We also examined the effects of blocking KDR in the eyes of normal air-reared control animals and in oxygen-treated animals after return to room air, by implanting in the vitreous a slow-release polymer (Elvax; DuPont Industrial Polymers, Wilmington, DE) containing anti-KDR–specific antibodies. 

Methods

Immunohistochemistry

One-day-old dogs were exposed to 95% to 100% oxygen for 4 days and killed in oxygen or returned to room air. Triplicate animals were killed at 1, 5, 8, 15, and 22 days of age by an intraperitoneal overdose of pentobarbital sodium. Animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Oxygen-treated animals were compared with age-matched, room air-reared control animals. Eyes were snap frozen in optimal cutting temperature (OCT) compound (Miles Inc., Elkhart, IN) using isopentane cooled with dry ice and sectioned at −20°C. Fellow eyes from each animal included in the immunohistochemistry studies were used in previously published morphometric studies in which the retinas were incubated for adenosine diphosphatase (ADPase) enzyme histochemistry and then flat embedded. ^{6 7} Enzyme histochemical localization of menadione-dependent α-glycerophosphate dehydrogenase (M-α-GPDH), a marker for vascular precursors and immature endothelial cells, was performed on unfixed 12-μm-thick sections, as previously reported. ²⁶ Immunohistochemical localization of von Willebrand factor (vWf) and VEGF receptor-2 (KDR) was performed using a streptavidin peroxidase technique, as described previously. ²⁷ Rabbit anti-human vWf (Accurate Chemical Co., Westbury, NY) was used at a concentration of 0.2 μg/mL to label formed blood vessels. The monoclonal anti-KDR antibody (6.64; ImClone Systems, Inc., New York, NY) was used at 75 μg/mL and the chimeric antibody (c-p1C11; ImClone Systems) at 2.5 μg/mL. The chimeric antibody was made as described previously. Briefly, a single-chain antibody, p1C11, was chimerized to create a full-length IgG antibody consisting of the mouse variable heavy and light chains fused with human heavy- and light-chain constant regions. The chimeric p1C11 IgG (c-p1C11) retained its specificity for KDR and demonstrated a higher affinity of binding for KDR than the parent p1C11 antibody. ²⁴ Primary antibody incubations were performed at 4°C for 20 hours. Localization of vWf and M-α-GPDH was compared with the localization of KDR. Control sections were incubated with protein concentration-matched nonimmune IgG (Zymed Laboratories, South San Francisco, CA) or antibody preadsorbed overnight at 4°C with 10 molar excess soluble KDR (ImClone Systems, Inc.). ^{24 25} Sections from room air–reared control animals and oxygen-treated experimental animals in each age group were processed simultaneously in the same reagents so that reasonable comparisons in staining intensity could be made. A minimum of 18 sections from each animal were processed and examined for KDR localization. 

Pellet Preparation and Implantation

Ethylene-vinyl acetate copolymer pellets (Elvax; DuPont Industrial Polymers) were prepared using a previously described procedure. ^{28 29} In brief, 50 μL of 10% ethylene-vinyl acetate copolymer in methylene chloride was added to 450 or 900 μg of lyophilized antibody in sterile glass vials to make three pellets of 150 or 300 μg, respectively. After vortexing, the solution was pipetted onto a glass dish and dried in a laminar flow hood. Once dry, the polymer was rolled, compressed, and cut into three pellets of equal size. The pellets were then dipped once in 10% ethylene-vinyl acetate copolymer to coat them and then air dried again in the laminar flow hood. Control pellets were made with nonimmune mouse IgG1 (Zymed) instead of anti-KDR antibody. 

To determine the kinetics of anti-KDR antibody release from the copolymer, pellets were soaked in 1 mL PBS at 37°C. Each day the PBS was collected and replaced. The viability of the anti-KDR antibodies was assessed with a binding assay (BIAcore assay; Biacore, Uppsala, Sweden), as described previously. ²⁴ This assay measures the amount of antibody that binds to soluble KDR receptor that has been immobilized onto a chip (Biacore). From a 300-μg pellet, 33μ g of antibody was released in the first 24 hours, with a gradual linear decline in release over the next 21 days (data not shown). The eluted material also inhibited the VEGF-stimulated proliferation of human umbilical vein endothelial cells (HUVECs) and canine retinal microvascular endothelial cells (data not shown). ³⁰  

Three litters of newborn purebred beagles were exposed to 95% to 100% oxygen continuously for 4 days and then abruptly returned to room air for 1 day before pellet implantation. Three animals from these litters (six animals per litter), served as a room air–reared control subjects and were not exposed to oxygen, but received intravitreal pellets to examine the effect of KDR inhibition on normal vasculogenesis. Oxygen-exposed animals were allowed to recover from oxygen exposure for 24 hours before anesthesia and surgery. Animals were anesthetized with halothane/O₂ by face mask initially, then intubated and maintained on halothane/O₂ throughout the surgery. Proparacaine was administered topically to each eye. The lateral canthus was clamped with a hemostat for 1 minute to minimize bleeding and a canthotomy performed. The eye was sterilely draped and a superotemporal conjunctival peritomy was created at the limbus. Limbal vessels were cauterized with dry-field cautery. A 20-gauge needle was used to make a limbal incision, which was enlarged to 1.5 mm. The needle was then inserted into the vitreous to create a channel through the vitreous gel for the pellet to slide into. The angle between the iris and needle was kept at 60°. After the pellet was soaked in normal saline, it was inserted into the vitreous by forceps, again maintaining a 60^o angle. After the pellet was implanted, a 10-0 nylon suture was used to close the incision and the conjunctiva, and the canthotomy was closed with 4-0 silk. Topical 0.3% gentamicin sulfate solution (Bausch & Lomb, Tampa, FL) was applied after surgery and the eyes were treated with neomycin, polymyxin B sulfate, and bacitracin antibiotic ointment daily (Fougera, Melville, NY). 

Dogs were killed at 22 days of age by an intraperitoneal overdose of sodium pentobarbital. ADPase retinal and vitreous preparations were made as described previously. ³¹ Briefly, after enucleation, the anterior portion of the eyes were removed. The vitreous was removed in toto, and retinas were dissected from the RPE. The entire vitreous body and dissected retina were then placed in 2% paraformaldehyde in 0.1 M sodium cacodylate buffer (pH 7.4) at 4°C for 20 hours. After washing, the tissues were incubated for the histochemical demonstration of ADPase activity, as described previously. ³² Morphometric analysis (vascular area measurements) of ADPase retinas was performed before sectioning, using digitized images collected by photomicroscope (Photomicroscope II; Carl Zeiss; Oberkochen, Germany) equipped with a charge-coupled device (CCD) camera (Hamamatsu, Hamamatsu City, Japan) and a computer (Macintosh Iici; Apple, Cupertino, CA) with NIH Image software program (ver. 1.44; provided in the public domain by the National Institutes of Health, Bethesda, MD, and available at http://rsb.info.nih.gov/nih-image/). Wet vitreous ADPase preparations were analyzed in 0.1 M cacodylate buffer using the system described earlier. 

The area of retinal vasculature and the area of preretinal neovascularization in eyes with anti-KDR pellets was compared with the fellow eyes using the Wilcoxon matched-pairs signed rank test. The mean area of retinal vasculature and the mean area of preretinal neovascularization for the group of eyes receiving anti-KDR pellets versus the group receiving control pellets was compared using the paired t-test. For both analyses, P < 0.05 was considered significant. 

Results

Immunolocalization

Both antibodies against KDR/Flk-1 gave a similar localization at all time points, and only the results from c-p1C11 immunohistochemistry are shown. In normal 1-day-old animals, very low-level KDR/Flk-1 immunoreactivity was associated with formedvessels both at the border of vascularized retina and in the more mature posterior retina (Fig. 1) . Angioblasts in advance of formed blood vessels, as visualized by M-α-GPDH staining, were not KDR immunoreactive. In contrast, prominent immunolabeling was observed within fenestrated endothelial cells of the choriocapillaris and the capillaries of the ciliary processes. In some cases, weak labeling of neural retina was observed. 

The normal 5-day-old air-reared control animals had a pattern of KDR/Flk-1 immunoreactivity identical with that observed in the 1-day-old dogs (data not shown). The pattern and relative levels of KDR/Flk-1 immunoreactivity was similar in the 5-day-old animals killed in oxygen, except that the number of weakly labeled retinal vessels was reduced as a consequence of oxygen-induced vaso-obliteration. ⁶ KDR/Flk-1 immunoreactivity in 8-day-old animals appeared much more pronounced in the reforming vasculature of the oxygen-treated animals, both at the border of vascularized retina and more posteriorly (Fig. 2) . We have previously demonstrated a large increase in both the number of proliferating endothelial cells and the number of blood vessels in inner retina when the vasculature reforms after vaso-obliteration. ⁷ The primary vasculature in the 15-day-old air-reared control animal had spread to the far periphery, and very-low-level KDR/Flk-1 immunoreactivity was associated with these formed vessels. In contrast, KDR/Flk-1 immunoreactivity in the 15-day-old oxygen-treated animals was highly elevated in blood vessels, both at the border of vascularized retina and throughout the posterior retina (Fig. 3) . KDR/Flk-1 immunolabeling was also prominent in intravitreal neovascularization, which was first evident at this age in the oxygen-treated animals. When anti-KDR antibody was preincubated with soluble KDR before applying it to tissue sections, immunolabeling of intraretinal and extraretinal neovascularization was completely eliminated. 

The relative intensity of KDR/Flk-1 immunoreactivity observed in inner retinal capillaries of oxygen-treated animals was similar to that observed in the fully formed choriocapillaris throughout postnatal development (Fig. 4) . In the normal 22-day-old animals, weak KDR/Flk-1 immunoreactivity was asso-ciated with both the formed primary and developing secondary retinal capillaries. Again, the relative level of immunoreactivity in retinal vessels was less than that observed in the fully formed choriocapillaris (data not shown). Compared with the control subjects, KDR/Flk-1 immunoreactivity was enhanced in retinal vessels of the oxygen-treated 22-day-old dogs, both at the edge of reforming vasculature, which was retarded compared with the air control, and in more posterior areas and intravitreal neovascularization (data not shown). Intravitreal neovascularization was observed from the optic nerve to the edge of surviving vasculature in the OIR 22-day-old animals. 

The results of our KDR/Flk-1 immunolocalization study can be summarized as follows. The endothelial cells of normal developing retinal vessels expressed relatively low levels of the VEGF receptor KDR/Flk-1. This low level of expression was observed throughout the developmental period. During the period of postnatal vasculogenesis retinal angioblasts did not express detectable KDR/Flk-1 immunoreactivity. In dogs subjected to hyperoxic insult and returned to room air, proliferating endothelial cells of reforming retinal vessels and forming intravitreal neovascularization expressed high levels of KDR/Flk-1. This high level of expression was similar to the constitutive expression normally seen in the endothelium of fenestrated ocular capillaries. 

Effects of Anti-KDR on Normal Vasculogenesis and OIR

Animals with eyes implanted with anti-KDR or control IgG copolymer pellets (Elvax) were killed at 22 days, and the area of vascularized retina and intravitreal neovascularization was measured in ADPase-incubated tissue by computer-assisted morphometric analysis. Eyes were examined before death for any signs of postsurgical inflammation. After the anterior segment was removed, the eye was examined for proper pellet placement. Of the 12 oxygen-treated animals, three demonstrated post surgical inflammation or had poor pellet placement (pellet touching lens, iris, or retina) in one or both eyes and were therefore eliminated from the study. Compared with eyes receiving control IgG, anti-KDR had no effect on radial growth of primary vasculature, density of capillaries, or development of the secondary capillary bed in three normal air-reared animals. Statistical analysis showed no significant difference in area of vasculature between the control IgG pellet eyes and the anti-KDR pellet eyes (P > 0.5). Figure 5 shows a representative air-reared normal animal that received a control pellet in one eye and an anti-KDR pellet in the fellow eye. Both retinas were fully vascularized, and the vascular pattern was remarkably similar at 22 days of age. 

In oxygen-treated animals, anti-KDR pellet eyes had less area of retinal vasculature and less area of intravitreal neovascularization than fellow eyes, which received control pellets (Fig. 6) . As has been reported previously, the proliferative response varied considerably between animals. ³¹ However, normally there is remarkable bilateral symmetry of disease, with an average of 9% difference in vascular area between eyes of oxygen-treated dogs. ³¹ Figure 7A shows the individual retinal area measurements in a group of nine oxygen-treated animals that received anti-KDR pellets in one eye and control pellets in the fellow eye. The greatest inhibition of both retinal vascular growth (57% inhibition) and intravitreal neovascularization (61% inhibition) occurred in an eye that received a pellet with 150 μg of antibody 6.64. The least inhibition (9% inhibition of retinal vasculature and 11% inhibition of intravitreal neovascularization) was observed in an eye that received a pellet with 150 μg 6.64 antibody. When the paired eyes were compared by the Wilcoxon matched-pairs signed rank test, the difference in area of retina vascularized between the eyes receiving anti-KDR pellets and control pellets was significant (P = 0.0039). Similarly, when the area of preretinal neovascularization in fellow eyes was compared by Wilcoxon matched-pairs signed rank test (Fig. 7B) , the difference was significant (P = 0.0078). There was no significant difference in the effect between the 150- and 300-μg pellets or between antibody 6.64 and c-p1C11, and the data were therefore combined for analysis of the groups. As a group, the anti-KDR pellet eyes had an average of 32% less retinal vasculature area and 45% less intravitreal neovascular area than control pellet eyes. Statistical analysis using the paired t-test revealed that the intravitreal neovascular area was significantly less in the anti-KDR pellet eyes compared with control pellet eyes (P = 0.04) and the retinal vascular area was also found to be significantly less (P = 0.005) in the anti-KDR treated eyes. Cross-sectional analysis of retinas (Fig. 8) revealed no inflammation, no impaired neurogenesis, and no cellular toxicity in animals in this surgically successful implant group (eyes with pellet placed properly and no postsurgical inflammation). 

Discussion

Hypoxia caused by the increasing oxidative demands of developing retinal neurons has long been considered to be a driving force in initiation of retinal vasculogenesis. ^{3 4} Postnatal retinal vasculogenesis is arrested when neonatal mice, rats, cats, and dogs are exposed to high levels of inspired oxygen. When these animals are returned to a normoxic environment, inner retinal hypoxia becomes exaggerated and proliferative OIR develops, which in the dog mimics many aspects of human retinopathy of prematurity (ROP). ³¹ Considerable evidence implicates the hypoxia-inducible angiogenic factor VEGF as a mediator of both normal retinal vasculogenesis and proliferative OIR in several mammalian species. ^{16 19 20 21 22 23} Although there is some disagreement about which retinal cell type or cell types are responsible for VEGF production during development, most investigators have described a temporal and spatial relationship between retinal VEGF expression patterns and blood vessel formation in normal development. Moreover, increased expression of VEGF has been reported during the proliferative phase of OIR in these species. ^{20 21 23}  

The expression of VEGF receptors Flt-1 and KDR/Flk-1 has also been examined in the retina of animal models of ROP during normal vasculogenesis and in OIR. ^{16 33 34} In general, most groups have shown that KDR/Flk-1 is expressed to some degree by endothelial cells of retinal vessels during normal development; however, there are some reports of strong nonvascular cell-associated Flk-1 expression in rodents. ^{33 34 35} In some cases, we also observed weak labeling of neural retina in the dog. The results of our KDR/Flk-1 immunolocalization study in normal dog retina are similar to the in situ hybridization results reported by Stone et al. ¹⁶ in cat, who found that endothelial cells of formed vessels express the most KDR/Flk-1. The validity of relatively weak retinal vascular endothelial cell–associated KDR/Flk-1 immunoreactivity that we observed in normal retina throughout the postnatal developmental period is corroborated by the observation that the fenestrated endothelium of the ciliary processes and the choriocapillaris of these eyes both exhibited strong immunoreactivity and therefore served as built-in positive controls. High constitutive expression of KDR/Flk-1 has been described in fenestrated capillary endothelium of other organs and, recently, in the human choriocapillaris. ^{36 37} Moreover, studies demonstrate that KDR/Flk-1 is either nondetectable or expressed at extremely low levels in nonproliferating, nonfenestrated endothelium. ¹⁴  

An important finding of this study was that KDR/Flk-1 was not demonstrable in retinal angioblasts during normal postnatal vasculogenesis but could be detected (although weakly) in endothelial cells of formed vessels (i.e., those expressing vWf). This is of particular interest, considering that this VEGF receptor is considered an early marker for angioblasts in other organs during embryogenesis, ^{38 39 40} and there is failure of vasculogenesis and blood island formation in KDR/Flk-1–knockout mice. ⁴¹ There are several possible explanations for this apparent discrepancy. Immunohistochemistry may not be sensitive enough to reliably demonstrate low-level KDR/Flk-1 and, therefore, angioblastic protein levels of this receptor may be below the limits of detection using this system. However, the in situ hybridization studies of Stone et al. ^{16 23} have described similar distribution of KDR/Flk-1 expression in developing retinal vessels of rat and cat, suggesting that KDR/Flk-1 may not be associated with retinal vasculogenesis. Perhaps a more plausible explanation is that angioblasts proliferate during embryogenesis, ^{38 39 40} whereas in postnatal canine retina the angioblasts are postmitotic. Previous studies from our laboratory have shown that the primary vasculature of the postnatal canine retina forms by differentiation and migration of angioblasts with little or no cellular proliferation involved. ^{1 7} Furthermore, results from the present study failed to show any inhibitory effect of anti-KDR antibody on normal retinal vasculogenesis in air-reared control animals. It is possible that the KDR antibody was unable to penetrate the retina in normal control animals and inhibit vasculogenesis, particularly the development of the deeper secondary capillary bed that is thought to form by proliferation. ¹ It is also conceivable that KDR antibody inhibited vasculogenesis early in development, but this effect may have been transient and, therefore, was not evident at 22 days of age. 

Despite the absence of inhibition of normal vasculogenesis in the dog by anti-VEGFR-2, studies in rodents have shown that small molecule inhibitors of VEGFR-2 in rodents inhibit or delay vasculogenesis. ^{42 43 44} However, the compounds used in those studies were not VEGFR-2 specific. Our results suggest that VEGF acting through the KDR/Flk-1 receptor may not be required for normal postnatal vasculogenesis in the canine retina. This seems reasonable in that KDR/Flk-1 is believed to be responsible for VEGF-induced mitogenesis ^{15 45} and the canine retinal vasculature forms with very limited mitosis. ⁷ Obviously, further studies are needed to examine the contribution of the Flt-1 receptor to vasculogenesis in this model, because Flt-1 may control VEGF-induced angioblast migration ⁴⁶ that occurs during their coalescence to form the inner retinal vasculature. ¹ The small molecule inhibitors reported to delay retinal vascular development inhibit both VEGFR-1 and VEGFR-2 signaling. ^{42 43 44}  

In the neonatal dog, oxygen-induced vaso-obliteration is accompanied by a severe degeneration of endothelial cells. ⁶ Consequently, when the animals are returned to room air, proliferation of endothelial cells is required to revascularize the retina. ⁷ Therefore, unlike normal vasculogenesis, revascularization after hyperoxic insult is primarily an endothelial cell–proliferative event. The marked increase in KDR/Flk-1 immunoreactivity associated with intraretinal and intravitreal neovascularization during the proliferative phase of OIR is coincident with the high proliferative activity of the endothelial cells in these vessels. ⁷ By implanting slow-release pellets containing anti-KDR antibody in the eyes of oxygen-treated animals, we were able to partially inhibit the growth of intravitreal neovascularization. Achieving only a partial inhibition of neovascularization may be due to our failure to deliver an optimal dose of anti-KDR antibody. However, the percentage of inhibition of neovascular growth achieved in this study was similar to the degree reported in other animal models of OIR, in which different VEGF-related inhibition strategies were used. ^{47 48 49} The incomplete inhibition by any of these strategies supports the view that other factors may be involved. Retinal revascularization in anti-KDR-implanted oxygen-treated eyes was also markedly inhibited because blood vessel growth in posthyperoxic retina is dependent on endothelial cell proliferation. Previous studies exploring pharmacologic therapies designed to inhibit some component of the VEGF system in animal models of OIR have seemingly failed to address this deleterious side effect. ^{47 49}  

In summary, the low level of KDR expression in normal developing retinal vessels and the absence of an inhibitory effect of anti-KDR antibody on normal blood vessel growth correlates with the low vasoproliferative activity associated with vasculogenesis. This suggests that KDR–VEGF interaction may not be essential for vascular development in the neonatal canine retina. The high level of KDR expression in retinal blood vessels and intravitreal neovascularization after hyperoxic insult is coincident with the high endothelial cell proliferation associated with oxygen-induced retinopathy. Blocking KDR after hyperoxic insult partially inhibits the formation of intravitreal neovascularization, but also has the undesirable effect of inhibiting retinal revascularization, a process dependent on endothelial cell proliferation. Therefore, any form of antiangiogenesis therapy may not be a viable therapeutic approach for treating developmental diseases such as ROP unless neovascularization can be specifically targeted. However, anti-KDR therapy may be useful in treating pathologic angiogenesis in adult tissues. 

 Supported by National Institutes of Health Grants EY09357 (GAL) and EY01765 (Wilmer Institute), the Brownstein Foundation (New York, NY), and a grant from ImClone Systems, Inc. GAL is an American Heart Association Established Investigator and a recipient of a Research to Prevent Blindness Lew Wasserman Merit Award.

 Submitted for publication April 16, 2001; revised September 6, 2001; accepted October 18, 2001.

 Commercial relationships policy: E (LW, ZZ); F (all others).

 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.

 

Serial sections showing the edge of vascularized retina in a 1-day-old air-reared control animal. (A–C, long arrow) Most anterior edge of formed vasculature. In the 1-day-old retina, formed blood vessels, as demonstrated by vWf immunolabeling (A), had low-level KDR immunoreactivity (B). Angioblasts in advance of formed blood vessels were not KDR immunoreactive but were labeled with the angioblast marker M-α-GPDH (C, short arrows, arrowheads). (D) Nonimmune IgG control section. (A, B, D) 3-Amino-9-ethyl-carbazole (AEC) reaction product; original magnification, ×130.

Figure 2.

 

Serial sections showing posterior retina in an 8-day-old air-reared control animal (top) and in an 8-day-old oxygen-treated animal 3 days after return to room air (bottom). (A, B, D, E, long arrow) Blood vessels in the inner retina. Blood vessels (vWf immunolabeled) of the 8-day-old air-reared control animal (A) had weak KDR immunoreactivity (B), whereas blood vessels in oxygen-treated animal (D) had much more KDR immunoreactivity (E). (C, F) Control nonimmune IgG sections. AEC reaction product; original magnification, ×50.

Figure 3.

 

Serial sections from a 15-day-old oxygen-treated animal immunostained with anti-vWf (A, D), anti-KDR antibody that was preincubated overnight with PBS (B, E), or anti-KDR antibody preincubated overnight with a 10-M excess of soluble KDR (C, F). Areas from the border of vascularized retina (A–C) and a more posterior region with intravitreal neovascularization (D–F). Double arrows indicate multiple layers of capillaries in (A, B) and preretinal neovascularization (D, E). Preincubation of antibody with soluble KDR completely eliminated both retinal vascular and intravitreal neovascular immunostaining (C, F). AEC reaction product; original magnification, ×50.

Figure 4.

 

Comparison of KDR immunoreactivity in retinal blood vessels and choriocapillaris in identical sections from a normal 15-day-old air-reared control animal (A–D) and in an oxygen-treated 15-day-old animal (E–H). Arrows: blood vessels, in all micrographs. In the air-reared control animal, primary and forming secondary retinal vessels (A) and choriocapillaris in a nonpigmented tapetal region (B) were demonstrated with vWf immunoreactivity. Compared with KDR immunoreactivity in normal retinal blood vessels (C), the labeling of choriocapillaris in the same air-reared control section was much more intense (D). In an oxygen-treated animal, multilayered capillaries were demonstrated by vWf immunostaining in the inner retina in this region (E) which express strong KDR immunoreactivity (G). The choriocapillaris in this tapetal region (F) of the same animal also showed strong KDR immunoreactivity (H). AEC reaction product; original magnification, ×150.

Figure 5.

 

ADPase-incubated retinas from a 22-day-old air-reared control animal that received a control pellet in one eye (left) and an anti-KDR pellet in the fellow eye (right). The primary vasculature (top) in both eyes spread to the far periphery (arrowhead, ora serrata). In both eyes, vascular remodeling in the primary or superficial vascular network occurred posteriorly (bottom), where a secondary capillary network formed. Dark-field illumination en bloc; original magnification, ×70.

Figure 6.

 

ADPase-incubated retinas (top) showing retinal blood vessels and vitreous bodies (bottom) showing dense, matlike intravitreal neovascularization in a 22-day-old oxygen-treated animal. One eye (left) received a control pellet and the fellow eye (right) an anti-KDR pellet. ADPase-positive blood vessels in the retinas and the excised vitreous preparations appeared white, and their area was measured by image analysis. Both areas of retinal and vitreal neovascularization were larger in the control pellet–treated eye than in the anti-KDR pellet–treated eye.

Figure 7.

 

Retinal vascular area (A) and intravitreal neovascularization area (B) in a group of oxygen-treated 22-day-old animals that received control pellets in one eye and anti-KDR pellets in the fellow eye. (A) Retinal vascular areas and (B) the intravitreal neovascularization areas from paired eyes. Wilcoxon matched-pairs signed rank test demonstrated that the differences between paired eyes in retinal vascular area (P = 0.0039) and in intravitreal neovascularization area (P = 0.0078) were significant.

Figure 8.

 

ADPase-incubated temporal retinal arcades from the eyes of a 22-day-old oxygen-treated animal that received an anti-KDR pellet in one eye (A, C, E) and a control pellet in the fellow eye (B, D, F). Although retinal vessel growth was inhibited in the anti-KDR pellet–treated eye (A), compared with the control pellet–treated eye (B), both eyes showed an overgrowth of capillaries at the border of vascularized retina (C, D). Cross sections (E, F) taken from the border of vascularized retina shown in (C, D) demonstrated the absence of an inflammatory response or retinal toxicity resulting from the pellets. (A–D) ADPase-incubated flat-embedded retinas in dark-field illumination; (E, F) periodic acid-Schiff– and hematoxylin-stained sections; original magnification, (C, D) ×70; (E, F) ×190.

The authors thank Carol Merges, Rumi Taomoto, and Tsuyoshi Otsuji for excellent technical assistance.