Difference in Ischemic Regulation of Vascular Endothelial Growth Factor and Pigment Epithelium--Derived Factor in Brown Norway and Sprague Dawley Rats Contributing to Different Susceptibilities to Retinal Neovascularization

BN and SD rats were purchased from Harlan (Indianapolis, IN). Care, use, and treatment of all animals in this study were in strict agreement with the Statement for the Use of Animals in Ophthalmic and Vision Research from the Association for Research in Vision and Ophthalmology, as well as the guidelines set forth in the care and use of laboratory animals by the Medical University of South Carolina.

Induction of retinal neovascularization was performed as described by Smith et al. (13), with minor modifications. Briefly, newborn rats were randomly assigned to experimental and control groups. At postnatal day 7, rats in the experimental group were exposed to hyperoxia (75% O₂) for 5 days (postnatal days 7–12) and then returned to normoxia (room air) to induce retinal neovascularization. Control rats were kept in room air.

Rats at postnatal days 12, 14, 18, and 22 were anesthetized with 10 mg/kg xylazine plus 75 mg/kg ketamine i.p. and perfused with fluorescein via intraventricle injection of 50 mg/ml of 2 × 10⁶-molecular-weight fluorescein isothiocyanate-dextran (Sigma, St. Louis, MO) as described by Smith et al. (13). The animals were immediately killed. The eyes were enucleated and fixed with 4% paraformaldehyde in PBS for 10 min. The retina was then separated from the eyecup and fixed with 4% paraformaldehyde for 3 h. Several incisions were made to the retina, which was flat-mounted on a gelatin-coated slide. The vasculature was then examined under a fluorescent microscope (Axioplan2 Imaging, Carl Zeiss, Jena, Germany). Both the total retinal area and the area of the avascular regions were measured using a computerized image-analysis system (Scion, Frederick, MD), averaged within each group and analyzed using Student’s t test.

Retinal neovascularization was quantified as previously described (22). Briefly, the eyes of eight rats from each group at postnatal day 18 were enucleated, fixed with 10% formaldehyde, sectioned, and then stained with hematoxylin and eosin. The nuclei of vascular cells on the vitreal side of the retina were counted under a light microscope in a double-blind study. Ten sagittal sections from each eye were examined, and cell numbers were averaged in each group of animals. The average number of preretinal vascular nuclei was compared with the control group using Student’s t test.

The retinas were dissected from rats at postnatal day 12, 14, and 16 and homogenized by sonication. The insoluble pellet was removed by centrifugation, and the protein concentration of the supernatant was measured with the BioRad protein assay (BioRad Laboratories, Hercules, CA). Then, 100 μg soluble protein was resolved by SDS-PAGE (12% polyacrylamide gel for PEDF, 15% polyacrylamide gel for VEGF) and electro-transferred to a Hybond ECL nitrocellulose membrane (Amersham International, Piscataway, NJ). The membrane was blotted with a polyclonal anti-PEDF antibody (23) or an anti-VEGF antibody (Santa Cruz Biotechnology, Santa Cruz, CA). The concentrations of primary antibodies are 1:500 for the anti-VEGF antibody and 2.4 μg/ml purified IgG for the anti-PEDF antibody. The membranes were stripped and reblotted with an antibody specific to β-actin (1:5,000 dilution; Sigma) as described previously (23).

The retinas from four animals at postnatal day 12, 14, and 16 were dissected and pooled for RNA isolation. Total RNA was isolated from the retina using Trizol reagents (Gibco-BRL Life Technologies, Gaithersburg, MD) according to the protocol recommended by the manufacturer. Then, 15 μg total RNA from each sample was loaded for Northern blot analysis using a rat PEDF cDNA probe as described previously (23). The RNA blot was stripped and rehybridized sequentially with a VEGF cDNA probe and a labeled oligonucleotide probe specific to the 18S ribosomal RNA (23). The RNA levels were semiquantified by densitometry and normalized by 18S RNA levels. For each analysis, two independent experiments were performed, and the results averaged.

As shown by fluorescein angiography in flat-mounted retinas, both BN and SD rats maintained under normoxia developed superficial and deep vascular layers in the retina that extended from the optic disc to the periphery. The vessels formed a fine radial branching pattern in the superficial retinal layer and a polygonal reticular pattern in the deep retinal layer (Figs. 1 and 2). Small avascular areas were observed only at the earliest stage analyzed (postnatal day 12); at later stages (postnatal days 18–22), the avascular areas disappeared, and the entire retina was covered by a mature capillary network (Fig. 1). There was no difference in the time courses of retinal vascular development between age-matched BN and SD rats from postnatal days 12 to 22 under normoxia. These time courses are also similar to that observed in mice (13).

In contrast, the hyperoxia-treated BN and SD rats showed clearly different time courses in the development of retinal neovascularization. In hyperoxia-treated BN rats at postnatal day 12 (the first day after the return of the animals from hyperoxia to normoxia), retinal arteries were highly constricted, and large avascular regions in the retina were observed. Vascular networks were developed only in small areas in the peripheral retina (Fig. 1). Under the same conditions, SD rats at postnatal day 12 showed less constriction of the retinal arteries and larger areas in the retina covered by vascular networks (Fig. 1), suggesting that the retinal vascular development is inhibited more significantly by hyperoxia in pigmented BN rats than in SD rats.

From postnatal days 14 (2 days after the return from hyperoxia to normoxia) to 22 (10 days after the return), BN rat retinas showed tortuous and dilated radial vessels and the gradual appearance of neovascularization (Fig. 1). The most severe retinal neovascularization was found at postnatal day 18 (6 days after the rats were returned to normoxia), with a typical pattern of pathological retinal neovascularization, including neovascular tufts, avascular regions, microaneurism, and hemorrhage (Figs. 1 and 2A). Although the hyperoxia-treated SD rats showed tortuous and dilated blood vessels at postnatal day 14, they lacked the typical pattern of neovascular tufts, microaneurism, and hemorrhage (Fig. 1). From postnatal days 18 to 22, the retinal vasculature in SD rats gradually became normal, with only dilated vessels being observed (Figs. 1 and 2C). These results demonstrated that BN rats developed more severe retinal neovascularization than SD rats after the hyperoxia treatment.

The area of avascular regions in the retina has been shown to be a characteristic of ischemia-induced retinal neovascularization and to be correlated with the severity of retinal neovascularization in rat and mouse models (18,20,24). Therefore, the avascular area was measured and expressed as a percentage of the total retinal area in hyperoxia-treated BN and SD rats at postnatal day 18. Retinal flat-mounts showed a significantly larger avascular area in BN rats than in SD rats (P < 0.01, n = 5) (Figs. 3A–C).

Retinal neovascularization was also quantified by counting preretinal vascular cells, as described by Smith et al. (13). At age postnatal day 18 (6 days after the rats were returned to normoxia), BN rats showed an average of 56 ± 11 (mean ± SE) preretinal vascular cells/section (n = 8) (Fig. 3D). At the same time point (postnatal day 18), SD rats showed less neovascularization, with 17 ± 4 (n = 8) preretinal cells per section (Fig. 3E). Statistical analysis demonstrated that BN rats developed significantly more preretinal neovascular cells than albino SD rats under the same conditions (P < 0.01, n = 8) (Fig. 3F). Taken together, Figs. 1, 2, and 3 demonstrated that pigmented BN rats are more susceptible to ischemia-induced retinal neovascularization than albino SD rats.

PEDF was used as a representative angiogenic inhibitor. In BN rats, after exposure to 75% O₂ for 5 days (postnatal day 12), retinal PEDF levels were elevated by 3.8-fold over that in age-matched normal controls, which correlated with the enlarged avascular areas in the retina at this time point (Fig. 1). After the return from hyperoxia to normoxia for 4 days (postnatal day 16), PEDF was decreased to ∼50% of the control level. The decreased PEDF levels correlated with the progression of retinal neovascularization (Figs. 1 and 4).

In hyperoxia-treated SD rats, however, retinal PEDF levels were not increased at postnatal day 12 when compared with levels in the control animals, which may explain why the avascular area was not significantly enlarged (Figs. 1 and 4). Retinal PEDF levels showed only a slight decrease at postnatal day 12 that recovered to control levels by postnatal day 16, when PEDF reached its lowest levels in BN rats (Fig. 4).

In the same hyperoxia-treated BN rats at postnatal day 12, VEGF was reduced by threefold, when compared with the age-matched normal controls (Fig. 5). At postnatal day 14 (2 days after the return to normoxia), VEGF started to increase, and by postnatal day 16, it reached a level that was fivefold more than the control (Fig. 5).

In hyperoxia-treated SD rats, however, VEGF levels were increased by only 1.9-fold at postnatal day 12 (Fig. 5). The VEGF levels decreased to the normal control level by postnatal day 14 and remained at the normal level thereafter (Fig. 5), indicating that the induction of VEGF expression by ischemia occurs and disappears earlier and is less significant in SD rats than in BN rats.

The VEGF-to-PEDF ratio was calculated as a measure of the balance between angiogenic stimulators and inhibitors (Fig. 6). In BN rats, the VEGF-to-PEDF ratio progressively increased from a level that was less than normal to 10-fold higher than the controls in the period between postnatal days 12 and 16. Consistently, the most aggressive progression of neovascularization was observed in BN rats during this period (Fig. 1). In contrast, hyperoxia-treated SD rats at postnatal day 12 showed a 2.6-fold increase in the VEGF-to-PEDF ratio, which may account for the slight retinal neovascularization at the early stage. This ratio decreased to the control level by postnatal day 16 (Fig. 6).

PEDF and VEGF mRNA levels were examined in the retina by Northern blot analysis. In hyperoxia-treated BN rats, the decreased PEDF mRNA and increased VEGF mRNA levels relative to respective controls were detected at postnatal day 12 (Fig. 7). At postnatal day 14, the PEDF mRNA was decreased by 2-fold to the lowest level, whereas the VEGF mRNA level peaked at 2.5-fold over the control level at the same time point (Fig. 7). Consistent with our previous findings, maximal changes in the VEGF and PEDF mRNA appeared 48 h before their protein changes (23)

In SD rats, only slightly decreased PEDF mRNA levels were observed at postnatal days 12 and 14 (25 and 14% of control levels, respectively). By postnatal day 16, the mRNA level of PEDF recovered to the control level (Fig. 7). The VEGF mRNA change was also less significant than that in BN rats, with only a 40% increase over the control at postnatal day 12, followed by a decrease to control levels by postnatal day 14 (Fig. 7). The smaller changes in PEDF and VEGF mRNA levels in SD rats, when compared with BN rats, were consistent with their protein level (Figs. 4 and 5).

To determine whether SD rats have higher basal levels of angiogenic inhibitors and lower levels of angiogenic stimulators than BN rats, the basal levels of PEDF and VEGF in the retina were measured in normal BN and SD rats using Western blot analysis. Rats maintained under normoxia at postnatal day 16 were selected for this analysis because the most significant differences in PEDF and VEGF levels between hyperoxia-treated SD and BN rats were observed at this time point (Figs. 5 and 6). After normalization by β-actin levels, both PEDF and VEGF levels in SD rats were ∼50% of those in BN rats (Fig. 8). Lower levels of both PEDF and VEGF in SD rats results in no change in the VEGF-to-PEDF ratio compared with the ratio in BN rats.

Ischemia-induced retinal neovascularization is a commonly used model for studies of ROP and proliferative diabetic retinopathy. The present study demonstrated that BN rats have a higher susceptibility to ischemia-induced retinal neovascularization than SD rats. Retinal neovascularization is more severe and has a longer duration in BN rats than in SD rats after the hyperoxia treatment, suggesting that these differences might be strain-specific. The different susceptibilities were correlated with the differences in ischemic regulation of angiogenic stimulators and angiogenic inhibitors in the retina.

Both rats and mice have been used as an oxygen-induced retinopathy model. However, previous evidence showed that it was more difficult to induce retinal neovascularization in SD rats compared with C57 mice, using constant hyperoxia followed by exposure to normoxia (13,20). Although retinal neovascularization can be induced in mice by exposure to constant 75% oxygen for 5 days followed by exposure to normoxia, albino SD rats require cycling between hyperoxia and normoxia to induce typical retinal neovascularization (18,20). Therefore, most previous studies using constant hyperoxia were limited to mice. The present study showed that pigmented BN rats develop typical retinal neovascularization after exposure to constant hyperoxia, to an extent comparable to that in mice. Therefore, differences between SD rats and mice in susceptibility to retinal neovascularization may not be ascribed to different species. Our results demonstrate that pigmented BN rats are as efficient a neovascularization model as mice. Because of the large size of rat eyes compared with mouse eyes, using rats for ischemia-induced retinal neovascularization model offers some advantages, e.g., easy intravitreal injection and retinal examination. Therefore, the findings that BN rats are highly sensitive to neovascularization will broaden the application of this animal model for studies of retinal neovascularization, especially in searching for therapeutic antiangiogenic reagents.

The mechanism underlying retinal neovascularization is not completely understood. However, there is evidence that a delicate balance between angiogenic stimulators and angiogenic inhibitors plays a key role in regulating angiogenesis (2,25). Under certain hypoxic conditions, as found in diabetic retinopathy, the angiogenic stimulators are overproduced, whereas the angiogenic inhibitors are decreased (2,9,23). The subsequent disruption of the balance between antiangiogenic and angiogenic factors results in neovascularization in the eye. VEGF is a primary angiogenic stimulator (6,26–28), and PEDF has been identified as a major angiogenic inhibitor in the eye (8,9). Thus, we used the VEGF-to-PEDF ratio as representative of the balance between angiogenic stimulators and inhibitors to better understand the molecular basis responsible for the different susceptibilities to ischemia-induced retinal neovascularization between different rat strains. Our results showed that VEGF increased gradually from postnatal days 12 to 16 in hyperoxia-treated BN rats, reaching a peak with a fivefold increase over the control (Fig. 5). In the same animals, PEDF gradually decreased by 2-fold, when compared with their age-matched controls (Fig. 4), resulting in a substantially increased VEGF-to-PEDF ratio (10-fold) (Fig. 6). In SD rats undergoing the same treatment, the VEGF-to-PEDF ratio was only slightly higher (2.6-fold) than the normal controls at these early stages. After postnatal day 14, VEGF levels quickly dropped, whereas PEDF recovered to control levels, resulting in a normal VEGF-to-PEDF ratio. The gradual increase of the VEGF-to-PEDF ratio in BN rats from postnatal days 12 to 16 correlated with the progression of retinal neovascularization during this period, whereas a quick decline in the VEGF-to-PEDF ratio in SD rats in the same period corresponded to an early regression of neovascularization. The roles of other angiogenic stimulators and inhibitors in the different susceptibilities to retinal neovascularization in these two rat strains remain to be revealed, as there are multiple angiogenic factors in the retina.

To determine whether different basal levels of VEGF and PEDF in normal BN and SD rats are responsible for their different sensitivities to ischemia-induced retinal neovascularization, we measured retinal VEGF and PEDF levels in age-matched normal SD and BN rats. The result showed that both VEGF and PEDF levels are higher in BN rats than in SD rats at postnatal day 16, resulting in no significant change in the VEGF-to-PEDF ratio in normal animals of these two strains. This finding suggests that the different susceptibilities to retinal neovascularization may not arise from the basal balance between VEGF and PEDF in these rats. Instead, the different regulation of VEGF and PEDF in these two rat strains in response to ischemia may be responsible for different susceptibilities to ischemia-induced retinal neovascularization.

Both VEGF and PEDF levels are known to be regulated by oxygen concentration. The different responses of the VEGF-to-PEDF ratio in hyperoxia-induced retinopathy in SD and BN rats suggest a quantitative difference in the regulatory pathways of VEGF and PEDF expression under hypoxia in these two strains of rats. Our results have demonstrated that both VEGF and PEDF mRNA levels are changed in hyperoxia-treated animals, suggesting that the regulation occurs at the RNA level. Under ischemia, VEGF induction is known to be mediated by the HIF-1 pathway (29). The strain difference in the induction of VEGF is thus likely to be ascribed to a different response of the HIF-1 pathway to hypoxia. In contrast, the regulation of PEDF by hypoxia is unclear. The factors accounting for the strain difference in PEDF suppression in these two rat strains remain to be identified.

The present study showed that retinal neovascularization is more severe and enduring in BN than in SD rats after the same hyperoxia treatment. These results demonstrate that pigmented BN rats are more susceptible to ischemia-induced retinal neovascularization than albino SD rats. Strain differences have been observed previously in other neovascularization animal models. Different strains of inbred mice have shown a 10-fold difference in response to VEGF- or basic fibroblast growth factor-stimulated angiogenesis in the corneal micropocket assay (30). In a VEGF-induced choroidal neovascularization model, the pigmented BN and Long-Evans rats demonstrated more severe neovascularization when compared with the albino Wistar-Kyoto and Lewis rats (31). It is interesting that this report and our results seem to suggest a linkage between pigmentation and the sensitivity to choroidal or retinal neovascularization, although it is unknown how pigmentation is associated with susceptibility to retinal neovascularization. In the ischemia-induced retinopathy model, animals from postnatal days 7 to 18 are used. At this stage, the eyes had not yet opened. Furthermore, previous evidence has shown that light intensity does not affect ischemia-induced retinopathy in rats (32). Therefore, different levels of light-related stress in the retina of pigmented and albino rats are unlikely to be responsible for different sensitivities to ischemia-induced retinal neovascularization in pigmented BN and albino SD rats. Instead, genetic factors are more likely to be responsible for the different regulation of VEGF and PEDF and, subsequently, the susceptibilities to retinal neovascularization. These two strains of rats will be useful for the identification of genetic factors accounting for the resistance or susceptibility to retinal neovascularization.

This study was supported by National Institutes of Health Grant EY12600 and EY12231 (to J.X.M.) and EY09741 (to C.E.C.), a research grant from the Juvenile Diabetes Foundation International (to J.X.M.), and an unrestricted grant from Research to Prevent Blindness, New York, to the Department of Ophthalmology, Medical University of South Carolina.