Regulation of Retinal Vascular Endothelial Growth Factor and Receptors in Rabbits Exposed to Hyperoxia

Abstract

purpose. To evaluate the molecular responses of vascular endothelial growth factor (VEGF) and its receptors to dexamethasone (Dex) and celecoxib (Cel) during hyperoxia and during hyperoxia followed by recovery in room air, in newborn rabbit retinas.

methods. Newborn rabbits at 3 days postnatal age (n = 96) received room air or oxygen (80%–100%) for 4 days, during which they were administered saline (Sal), Dex, vehicle (Veh), or Cel (n = 12/treatment group). Six animals from each group were killed immediately after hyperoxia (or room air) and the remainder exposed to room air for 5 days. Retinal mRNA expression of VEGF₁₂₁, VEGF₁₆₅, VEGF receptor-1 (VEGFR-1, or Flt-1), and VEGFR-2 (or KDR/Flk-1) was determined.

results. Hyperoxia resulted in increased retinal expression of mRNA of the VEGF splice variants in the groups treated with Sal, Dex, and Veh, whereas a decrease in VEGF₁₂₁ was noted in the Cel-treated group. In contrast, retinal Flt-1 receptor mRNA was markedly increased in the Cel-treated group only, whereas retinal VEGFR-2 (KDR/Flk-1) receptor mRNA was suppressed in all the treatment groups. Hyperoxia followed by recovery in room air resulted in a minimal decrease in expression of retinal Flt-1 mRNA in the Sal and Dex groups. Cel treatment abolished its expression.

conclusions. The findings of increased retinal expression of VEGF mRNA in the newborn rabbit in response to hyperoxia are most likely due to species differences. Selective targeting of VEGF₁₂₁ and Flt-1 mRNA by Cel may represent one regulatory pathway for their anti-inflammatory effects. Further studies are needed to evaluate the therapeutic benefits of cyclooxygenase (COX)-2 inhibitors for the treatment and/or prevention of diseases associated with neovascularization.

Vascular endothelial growth factor (VEGF)-A is a potent endothelial cell mitogen and angiogenic factor involved in embryonic development. ^{1 2} Alternative splicing of VEGF-A mRNA generates at least five variants with 121, 145, 165, 189, or 206 amino acids. ^{3 4 5 6} VEGF₁₄₅, VEGF₁₈₉, and VEGF₂₀₆ are tightly heparin-bound to the cell surface, whereas VEGF₁₆₅ and VEGF₁₂₁ are diffusible ⁴ and predominate in most tissues. ³ Of the two diffusible isoforms, VEGF₁₆₅ is the dominant isoform and has some heparin-binding capability, whereas VEGF₁₂₁ has no heparin-binding activity. ⁶ Although the exact role for each isoform remains unclear, recent investigations have demonstrated that the splice variants VEGF₁₂₁ and VEGF₁₆₅ are highly involved in angiogenesis and may constitute up to 98% of total overexpression of VEGF. ⁴ VEGF-A exerts its actions by binding with high affinity to specific receptors, namely VEGF receptor-1 (VEGFR-1, or Flt-1) and VEGFR-2 (or KDR/Flk-1), which are restricted to endothelial cells. ⁷ Whereas both VEGF₁₂₁ and VEGF₁₆₅ bind with equipotent affinity to their receptors, Flk-1 appears to be the dominant signaling receptor, mediating mitogenesis and permeability, ⁸ whereas the much weaker Flt-1 appears to be involved in VEGF autoregulation. ^{9 10}  

VEGF has been shown to be associated with numerous pathologic conditions associated with angiogenesis, including the pathogenesis of retinopathy of prematurity (ROP), a developmental vascular disease that is the leading cause of blindness in very-low-birthweight (VLBW) infants. ¹¹ Recent studies have shown that exposure of the VLBW newborn to high concentrations of oxygen causes retinal ischemia and vascular thrombotic occlusion, followed by arrest of the normal progression of vascularization of the retina. ^{12 13} This vaso-obliteration may be associated with a concurrent suppression of VEGF and its receptors, as demonstrated in the adult rat lung and the adult mouse retina. ^{14 15} Conversely, studies in porcine brain microvessels and mice and kitten retinas have shown that VEGF is highly induced by hypoxia. ^{16 17 18} It therefore seemed reasonable to assume that during recovery in room air there is an overproduction of retinal VEGF. ^{17 18 19} Induction of VEGF is also associated with generation of reactive oxygen species ²⁰ and inflammatory cytokines and prostanoids. ^{21 22} Recent data have demonstrated that the inducible isoform of prostaglandin synthetase, cyclooxygenase (COX)-2 correlates with expression of VEGF ²³ and may contribute to tumor angiogenesis. ^{24 25 26} Selective COX-2 inhibitors, such as dexamethasone (Dex) suppress hypoxia-induced VEGF mRNA in ovine lungs ²⁷ and rat glioma cells. ²⁸ Further, Dex significantly reduces the incidence of ROP in human newborn infants, ²⁹ mice, ^{30 31} and rabbits. ³²  

Although the involvement of VEGF in mediating vasoproliferation is currently under investigation, very little attention has been directed toward the role of its receptors. Considering the ability of Dex to effectively suppress VEGF overproduction in animal studies and possibly reduce the risk of ROP in humans, there may be a role for selective COX-2 inhibitors. Because the use of Dex is associated with known short- and long-term adverse side effects, such as suppression of organ and body growth, we designed a pilot study to investigate the molecular responses of VEGF and its receptors to Dex and celecoxib (Cel, the novel selective COX-2 inhibitor), during hyperoxia and during hyperoxia followed by recovery in room air (hyperoxia-recovery), in the newborn rabbit retina. 

Materials and Methods

All experiments were conducted according to the guidelines of the Institutional Review Board and Institutional Animal Care and Use Committee, of Long Beach Memorial Medical Center, and animals were managed according to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Animals were treated humanely, according to the guidelines outlined by the Association for the Assessment and Accreditation of Laboratory Animal Care (AAALAC) and the Guide for the Care and Use of Laboratory Animals (National Research Council). Animals were killed according to the guidelines of the American Veterinary Medical Association. ³³  

Experimental Design.

Timed-pregnant New Zealand rabbits carrying fetuses of known gestational age (28–29 days) were purchased from a local breeder (The Rabbit Source, Ramona, CA). The pregnant rabbits were placed in appropriate cages with nesting boxes until delivery (30–32 days gestation). Once the rabbit pups were born, they were undisturbed for 3 days, to allow for stabilization with the doe. At 3 days postnatal age, six animals were killed and their retinas removed for mRNA expression of the VEGF isoforms and their receptors. The remaining pups were pooled to minimize interlitter variability and randomly assigned to (1) 80% to 100% oxygen for 4 days (n = 24); (2) 80% to 100% oxygen for 4 days followed by recovery in room air for 5 days (n = 24); (3) room air for 4 days (n = 24); or (4) room air for 9 days (n = 24). Within each group, the animals were further randomized to receive (1) dexamethasone sodium phosphate (Merck Sharp Dohme, Rahway, NJ) at 1 mg/kg per day in normal saline to a volume of 0.05 mL, intramuscularly (IM); (2) an equivalent volume of saline (Sal), IM; (3) 5 mg/kg per day Cel (G. D. Searle), suspended in vehicle (Veh) to a 0.5-mg/mL suspension, through an orogastric feeding tube; or (4) an equivalent volume of Veh (0.5% methylcellulose and 0.1% polysorbate 80 in water) through an orogastric feeding tube. Administration of drug or Veh occurred daily during the first 4 days of hyperoxia or normoxia. During hyperoxia (80%–100%), the pups were placed in a Plexiglas chamber (Plas Laboratories, Lansing, MI) equipped with outlets for (1) flow of oxygen and (2) output and input flow for removal of carbon dioxide (CO₂) and return of oxygen, respectively, using a pump (Model DOA-V191-AA; Gast Manufacturing Co., Benton Harbor, MI). A desiccant (Drierite; W. A. Hammond Drierite Co., Ltd., Xenia, OH) and a CO₂ absorbent (Baralyme; Allied Health Care, St. Louis, MO) were used in line between the chamber and the pump to remove moisture and CO₂, respectively. Oxygen was heated and humidified using a heater-humidifier (Humidifier-Nebulizer Heater; Inspiron, Cucamonga, CA) before it entered the chamber. An oxygen analyzer (model 5524; Ventronics, Temecula, CA) was placed in the bottom of the chamber for monitoring oxygen concentrations. 

At the time of randomization, no more than eight pups were placed with each doe (room air) or in the oxygen chamber (hyperoxia). For those animals exposed to hyperoxia, the doe was placed in the chamber with the pups for 1 to 2 h/d in the early morning around 6:30 AM, until the feeding was completed. This method of feeding was very successful, resulting in little mortality. Those pups exposed to room air remained in the nesting boxes where the doe would enter in the early morning to feed them. Rabbits usually feed their pups once per day. ³⁴ The animals randomized to hyperoxia-recovery were returned to their nesting boxes for an additional 5 days of room air. At the end of each experimental time, the pups were killed and their retinas removed for extraction of total RNA. 

Isolation of Total RNA.

The retinas from both eyes of each animal were examined under light microscopy (to ensure exclusion of neural tissue), removed, and immediately placed in 1.0 mL ice-cold extraction reagent (TRIZol; Gibco BRL, Eggenstein, Germany). The samples were immediately frozen and placed in −80°C until assay. Total cellular RNA in the retina was extracted by homogenization using a polytron homogenizer (Brinkman Instruments, Inc., Westbury, NY). The homogenates were centrifuged at 12,000 rpm for 20 minutes at 4°C. The supernatant was collected and kept at room temperature for 5 minutes, after which 200 μL/mL chloroform was added. The mixture was shaken vigorously for 15 seconds and centrifuged at 12,000 rpm for 20 minutes at 4°C. The aqueous phase was removed, and the RNA was precipitated with isopropanol and collected by centrifugation at 12,000 rpm for 10 minutes at 4°C. The supernatant was discarded, and the RNA pellet was washed with 75% ethanol, centrifuged at 8000 rpm at 4°C for 5 minutes, air dried, and solubilized in diethylpyrocarbonate (DEPC)-treated water. The integrity of the RNA was determined by gel electrophoresis in 1% agarose gel stained with ethidium bromide (EtBr). The purity of the RNA was assessed by the ratio of absorbance at 260 and 280 nm. The total RNA concentration was estimated by spectrophotometric measurements at 260 nm, assuming 40 μg/mL RNA equals one absorbance unit. The total RNA yield was diluted in DEPC-treated water to 1 μg/μL total RNA for all samples. 

Reverse Transcription-Polymerase Chain Reaction.

Two micrograms total RNA was reverse transcribed to cDNA using Moloney murine reverse transcriptase. Amplification of cDNA was performed using specific sense and antisense primers for rabbit β-actin, VEGF, and the receptors Flt-1 and KDR/Flk-1 with Taq DNA polymerase (AmpliTaq; PE Biosystems, Foster City, CA). The sense and antisense primers were prepared by Life Technologies (Gaithersburg, MD). The primer sequence for β-actin sense was 5′-CCT TCC TGC GCA TGG AGT CCT GG-3′, and the primer sequence for β-actin antisense was 5′-GGA GCA ATG ATC TTG ATC TTC-3′. The sense primer sequence for VEGF was 5′-CAG TGA ATT CGA GAT GAG CTT CCT ACA GCA C-3′, and the antisense primer sequence was 5′-CCT GGA ATT CTC ACC GCC TCG GCT TGT CAC-3′. The PCR products from these primers showed amplification of two VEGF splice variants corresponding to VEGF₁₂₁ (110 bp) and VEGF₁₆₅ (242 bp). The PCR cycle profile, which was performed using a model 480 DNA thermal cycler (Perkin Elmer-Cetus, Norwalk, CT) was 94° for 30 seconds, 59° for 1 minute, and 72° for 1 minute, for 35 cycles. ³⁵ The sense and antisense primers for Flt-1 were 5′-CTG-ACT CTC GGA CCC CTG-3′ and 5′-TGG TGC ATG GTC CTG TTG-3′, respectively (735 bp), and the sense and antisense primers for KDR/Flk-1 receptor were 5′-TGG CTC ACA GGC AAC ATC 3′ and 5′-CTT CCT TCC TCA CCC TTC-G-3′, respectively (819 bp). The PCR cycle profile was 95 for 5 seconds, 56° for 5 seconds, and 72° for 1 minute, for 40 cycles, followed by 72° for 10 minutes. ³⁶ The alternative primer sequences for VEGF splice variants used to confirm our findings, were taken from Cherng et al. ⁵  

Densitometric Scanning.

Gel electrophoresis of the PCR products was performed on 1.5% agarose gels stained with EtBr. The intensities of the bands were measured with the use of an imager and software (GelDoc 1000 Darkroom Imager and Molecular Analyst; BioRad Laboratories, Hercules, CA). The PCR fragments were identified according to their molecular mass using a DNA mass ladder (PE Biosystems). The amount of DNA in each specimen was quantitated by the integrated density of the product bands within a closed rectangle, which was then normalized to the density of the β-actin bands. The data are expressed as mean VEGF₁₂₁, VEGF₁₆₅, Flt-1, or KDR/Flk-1-to-β-actin ratio ± SEM. 

Statistical Analysis.

Statistical analyses were accomplished on computer (Instat; GraphPad, San Diego, CA). An unpaired Student’s t-test was used to examine differences between the treated and control groups (Dex versus Sal; and Cel versus Veh) for the VEGF splice variants and its receptors. An unpaired t-test was also used to examine differences between room air and hyperoxia or room air and hyperoxia-recovery, for each group (Sal, Dex, Veh, and Cel). Data are expressed as the mean ± SEM. P < 0.05 was considered significant. 

Results

Pooling and randomization resulted in 5% to 15% mortality, with the deaths of 10 pups, equally distributed among the groups. On examining the mRNA expression of VEGF, we were surprised to find that hyperoxia evoked a significant increase in the VEGF splice variants in the Sal-, Dex- and Veh-treated groups, a response that contradicted our hypothesis and previously reported data in mice and kittens. ^{17 18} To confirm our findings, we retested the expression of VEGF mRNA with a different sense and antisense primer sequence that had been previously validated in rabbits by Cherng et al. ⁵ Despite distinct differences in the density of VEGF gene expression, similarities in response patterns to hyperoxia and hyperoxia-recovery were noted. The results obtained in the present study (Figs. 1 2 3 4 5 6) are based on the more rabbit-specific VEGF primer sequence taken from Watkins et al. ³⁵  

Response of Retinal VEGF to Dex and Cel during Hyperoxia.

Normal retinal mRNA expression of the VEGF splice variants as measured in rabbits exposed to room air (4–7 days postnatal age) was compared with levels found in age-matched retinas of rabbits exposed to 80% to 100% oxygen for 4 days. Consistent with data from Watkins et al., ³⁵ VEGF₁₆₅ and VEGF₁₂₁ exhibited differential expression of mRNA (Fig. 1) . In keeping with earlier reports, exposure to hyperoxia resulted in increased ³⁵ or unchanged ³⁷ expression of the mRNA of both VEGF splice variants in the Sal- (VEGF₁₆₅: 1.5 ± 0.2 vs. 2.3 ± 0.2, P < 0.007; VEGF₁₂₁: 1.16 ± 0.2 vs. 1.73 ± 0.2, P < 0.01), Dex- (VEGF₁₆₅: 1.27 ± 0.1 vs. 1.7 ± 0.2, P < 0.05; VEGF₁₂₁: 0.71 ± 0.2 vs. 0.64 ± 0.05, NS), and Veh- (VEGF₁₆₅: 1.26 ± 0.3 vs. 2.20 ± 0.22, P < 0.05; VEGF₁₂₁: 0.76 ± 0.2 vs. 1.54 ± 0.21, P < 0.05) treated groups, whereas an opposite response was seen in the Cel-treated group (VEGF₁₆₅: 1.64 ± 0.2 vs. 1.22 ± 0.16, NS; VEGF₁₂₁: 1.09 ± 0.1 vs. 0.33 ± 0.1, P < 0.001). In the hyperoxia-exposed groups, Cel-treatment resulted in decreased expression of VEGF₁₆₅ (P < 0.01) and VEGF₁₂₁ mRNA (P < 0.001) compared with that in the Veh-treated group. 

Total Retinal VEGF_(165+121) during Hyperoxia and Hyperoxia-Recovery.

Total VEGF mRNA was calculated as the sum of the VEGF₁₆₅ and VEGF₁₂₁ splice variants. Despite similarities in the response patterns of the control groups (Sal and Veh) during hyperoxia, there were distinct differences between the effects of Dex and Cel. Figure 2 shows that hyperoxia induced expression of total VEGF mRNA in the Dex-treated (1.9 ± 0.1 vs. 2.73 ± 0.14, P < 0.05) group, whereas Cel treatment resulted in a marked decline in expression of total VEGF mRNA (3.73 ± 0.4 vs. 1.56 ± 0.22, P < 0.001). Compared with Veh, Cel treatment during exposure to room air induced expression of total VEGF mRNA (P < 0.01), a response that was reversed in hyperoxia (P < 0.05). 

During hyperoxia-recovery, expression of total VEGF mRNA declined in the Sal-treated group exposed to hyperoxia-recovery (1.39 ± 0.1, P < 0.001) compared with room air only (2.4 ± 0.1). In contrast, expression of total VEGF mRNA increased in the Veh- (3.24 ± 0.1, P < 0.0001) and Cel- (2.1 ± 0.9, P < 0.01) treated groups exposed to hyperoxia-recovery compared with the room air groups (Veh: 0.9 ± 0.2; Cel: 0.9 ± 0.2). Furthermore, comparing the hyperoxia-recovery groups, Cel treatment resulted in significantly lower expression of total VEGF mRNA compared with Veh treatment (P < 0.01). 

Percentage of Total Retinal VEGF mRNA during Hyperoxia and Hyperoxia-Recovery.

The relative percentage of each splice variant in the retina was calculated with the sum of the two splice variants as the denominator. The proportion of VEGF₁₆₅ and VEGF₁₂₁ in the rabbits exposed to room air was comparable with that in hyperoxia, whether treated with Sal, Dex, or Veh (Fig. 3) . However, a different pattern emerged with Cel treatment. The proportion of total VEGF mRNA that was composed of VEGF₁₆₅ was significantly greater in hyperoxia (80% ± 2%, P < 0.0001) than in room air (60% ± 2%) or in Veh-treated animals exposed to hyperoxia (42% ± 2%, P < 0.001). In contrast, in the Cel-treated groups, the proportion of total VEGF mRNA that was composed of VEGF₁₂₁ was significantly less in the hyperoxia-exposed rabbits (21% ± 2%, P < 0.01) than in those exposed to room air only (40% ± 2%) and than in Veh-treated animals exposed to room air and hyperoxia (38% ± 0.45%, P < 0.001). 

During hyperoxia-recovery, the proportion of total VEGF mRNA that was composed of VEGF₁₆₅ was slightly lower in the Dex-treated hyperoxia group (74% ± 2%, P < 0.05) than in the room air group (82% ± 2%). In contrast, Dex treatment resulted in a higher percentage of VEGF₁₂₁ in the hyperoxia-recovery group (26% ± 2%, P < 0.01) than in the room air group (18% ± 1%, P < 0.05) compared with Sal treatment that resulted in a slightly lower percentage in the hyperoxia-recovery group. In the animals exposed to room air only, Dex treatment increased the percentage of VEGF₁₆₅ (P < 0.05), whereas VEGF₁₂₁ decreased (P < 0.01) compared with that in the Sal-treated group. Cel treatment dramatically decreased the percentage of VEGF₁₆₅ in the hyperoxia-recovery group (58% ± 3%, P < 0.001) compared with the room air group (79% ± 2%). Conversely, Cel treatment increased the percentage of VEGF₁₂₁ in the animals exposed to hyperoxia-recovery (42% ± 3%, P < 0.001) compared with those exposed to room air (21% ± 2%). In the animals exposed to hyperoxia-recovery, Cel treatment caused a significant decrease in both VEGF₁₆₅ (P < 0.01) and VEGF₁₂₁ (P < 0.01) compared with Veh. 

Response of Retinal Flt-1 and KDR/Flk-1 Receptor during Hyperoxia.

Figure 4 shows notable differences in the response patterns between the Flt-1 and KDR/Flk-1 receptors. In the Sal-treated animals, hyperoxia decreased Flt-1 and KDR/Flk-1 receptor mRNA expression (0.36 ± 0.05, P < 0.01 and 0.45 ± 0.15, P < 0.05, respectively) compared with room air (0.75 ± 0.1 and 0.72 ± 0.2, respectively). Dex treatment appeared to spare Flt-1 receptor mRNA expression during hyperoxia, but KDR/Flk-1 receptor mRNA was notably more suppressed (0.79 ± 0.12, P < 0.05) than with room air (1.38 ± 0.2). In the animals exposed to room air only, Dex-treatment caused marked reductions in Flt-1 receptor mRNA (P < 0.01) and significant elevations in KDR/Flk-1 receptor mRNA (P < 0.05) than did Sal treatment. A quite different pattern was evident in the groups that were gavaged with Veh and Cel. Compared with room air, hyperoxia had no adverse effect on Flt-1 receptor mRNA in the Veh-treated animals; however, a marked elevation was noted in the Cel-treated animals (1.08 ± 0.2 vs. 0.21 ± 0.1, P < 0.001). Hyperoxia caused a markedly subnormal yet equivalent suppression in KDR/Flk-1 receptor mRNA expression in the Veh-treated (0.08 ± 0.01, P < 0.01) and Cel-treated (0.09 ± 0.01, P < 0.01) animals compared with room air (0.74 ± 0.2 and 0.89 ± 0.2, respectively). Cel treatment caused a significant increase in Flt-1 receptor mRNA expression, compared with Veh treatment (P < 0.05) during hyperoxia. 

Response of Retinal VEGF to Dex and Cel during Hyperoxia-Recovery.

Retinal mRNA expression of the VEGF splice variants was measured in rabbits exposed to room air for 4 to 12 days postnatal age and compared with levels found in age-matched retinas of rabbits exposed to 80% to 100% oxygen for 4 days followed by recovery in room air for 5 days (hyperoxia-recovery). In the groups receiving IM injections of Sal or Dex, there were similarities in the response of VEGF₁₆₅ (Fig. 5) . In both groups of animals, VEGF₁₆₅ mRNA expression declined substantially in response to hyperoxia-recovery (Sal: 0.99 ± 0.11, P < 0.001; Dex: 1.35 ± 0.2, P < 0.01) compared with room air (1.86 ± 0.1 and 2.31 ± 0.2, respectively). In the Sal-treated groups, VEGF₁₂₁ mRNA was only slightly subnormal with exposure to hyperoxia-recovery compared with room air only (0.31 ± 0.03 vs. 0.54 ± 0.08, P < 0.05). Dex treatment caused no significant deficits in VEGF₁₂₁ mRNA expression comparing exposure to room air and hyperoxia-recovery; whereas when both room air groups were compared, a significant reduction was noted in the Dex-treated (P < 0.05) versus Sal-treated animals. A more consistent response pattern was observed in the groups that were gavaged with Veh and Cel. The Veh-treated animals that were initially exposed to hyperoxia demonstrated notable increases in VEGF₁₆₅ on recovery in room air (2.02 ± 0.6, P < 0.05) compared with the animals exposed to room air only (0.62 ± 0.1). No difference in VEGF₁₆₅ was detected between the Cel-treated room air and hyperoxia-recovery groups. VEGF₁₂₁ was visibly elevated in the Veh-treated hyperoxia-recovery group compared with the Veh-treated room air group, but no significance was achieved. Expression of VEGF₁₂₁ mRNA increased in the Cel-treated hyperoxia-recovery group (0.99 ± 0.27, P < 0.05) compared with the Cel-treated room air group (0.19 ± 0.02). 

Response of Retinal Flt-1 and KDR/Flk-1 Receptor during Hyperoxia-Recovery.

Figure 6 shows abundant expression of Flt-1 receptor mRNA in all groups except the Veh-treated group. At this postnatal age, expression of KDR/Flk-1 was not detected. Similar to the Sal-treated hyperoxia group (Fig. 4) , Flt-1 receptor mRNA expression was suppressed (0.53 ± 0.08, P < 0.05) in the hyperoxia-recovery group compared with the Sal-treated room air (0.83 ± 0.08) group. Dex treatment decreased Flt-1 receptor mRNA expression during hyperoxia-recovery (0.3 ± 0.06, P < 0.05) compared with room air (0.57 ± 0.1) and Sal treatment (P < 0.05) in the groups exposed to hyperoxia-recovery. Cel treatment caused strikingly subnormal levels (undetectable) of Flt-1 mRNA expression in the hyperoxia-recovery group compared with room air. 

Discussion

To our knowledge the present pilot study is the first to examine and compare the effects of Dex (a corticosteroid) and Cel (a nonsteroidal anti-inflammatory agent) on mRNA expression of two retinal VEGF splice variants and their receptors, in response to hyperoxia and hyperoxia-recovery in room air. Based on previously reported data in rats and mice, we hypothesized that hyperoxia would suppress mRNA expression of the two retinal VEGF isoforms, VEGF₁₂₁ and VEGF₁₆₅, whereas, after recovery in room air, there would be an overexpression of retinal VEGF mRNA. Contrary to our hypothesis, we found that, during hyperoxia, the mRNA expression of the VEGF splice variants increased in all the newborn rabbit groups, except in the Cel-treated group, which demonstrated a marked reduction in VEGF₁₂₁ mRNA (Fig. 1) . Notwithstanding, the most significant finding was the additive effect of hyperoxia and Cel on the retinal Flt-1 receptor mRNA, which was abundantly expressed, compared with expression in the other treatment groups (Fig. 4) . Another surprising finding was the decrease in the mRNA expression of the VEGF isoforms, particularly VEGF₁₂₁, after hyperoxia-recovery in all except the Cel-treated animals, suggesting that the VEGF response to hyperoxia in rabbits may be time dependent. 

In a recent study in which the rabbit model was used for hyperoxia-induced retinal vaso-obliteration, ³² it was demonstrated that exposure of 3-day-old rabbits to hyperoxia for 4 days followed by recovery in room air for 5 days, results in neovascularization and ROP, an effect that is significantly attenuated by Dex. ³² Our hypothesis, experimental design, and the Dex dose of 1.0 mg/kg per day during the 4-day hyperoxia (or room air) phase were based on that previously reported observation. On examining our hypothesis that newborn rabbits exposed to hyperoxia would show decreased retinal expression of VEGF mRNA, we found that our results were in sharp contrast to previously reported data. ^{17 18 19}  

However, there is precedence for the increased expression of VEGF mRNA in newborn rabbits exposed to hyperoxia. Watkins et al. ³⁵ reported increased expression of the mRNA of both VEGF₁₂₁ and VEGF₁₆₅ in newborn rabbit lungs after 9 days of hyperoxia, whereas VEGF₁₈₉ was significantly decreased. In the same experiment, normal VEGF ratios were established after 5 days of recovery in room air. Further studies by Maniscalco et al. ³⁷ demonstrated decreased lung VEGF mRNA in newborn rabbits after only 9 days of hyperoxia, whereas data on adult rat lungs exposed to 95% oxygen for 24 and 48 hours demonstrated decreased VEGF mRNA. ¹⁴ The investigators speculated that differences in the timing of oxygen delivery, age of the animals, species differences, and increased tolerance of newborn rabbits to hyperoxia, may be responsible for the differential responses. Clearly, our data are in support of this speculation. 

More recently, Yu and Cringle ³⁸ demonstrated that the oxygen supply to the inner limiting membrane of the rabbit retina is different from the rat retina. They examined and compared the responses of adult rat, rabbit, and guinea pig retinas to stepwise increases in inspired oxygen. They found that the oxygen levels at the retinal surface remain relatively constant in the retinas of rats but increase dramatically in rabbit retinas, suggesting that avascular or partially vascularized retinas respond differently to hyperoxia than fully vascularized retinas. Based on these findings, we believe that differences in retinal vascularization and differences in choroidal and retinal oxygen supply in the newborn rabbit could be the most plausible explanation for our findings. Therefore, although the neonatal rabbit may have partially vascularized retinas at birth (similar to the human premature infant), it may be resistant to hyperoxia and therefore may not be a suitable model for the study of ROP. Despite this limitation in our model, the novel finding that Cel, the specific COX-2 inhibitor, selectively suppressed VEGF₁₂₁ during hyperoxia, provides evidence of its possible therapeutic benefits. Hyperoxia followed by recovery in room air did not result in an overexpression of retinal VEGF mRNA. Instead, an appreciable decrease was observed in the Sal- and Dex-treated groups, whereas levels in the Veh- and Cel-treated groups returned to room air values. This response further substantiates that the newborn rabbit retinal vasculature may be resistant to the effects of hyperoxia. 

Expression of the VEGF receptors analyzed in our study is exclusive to endothelial cells and is essential for their survival. Because the VEGF ligand is coordinately expressed with its receptors during hypoxia, ³⁹ it seems logical that hyperoxia would have similar suppressive effects on its receptors. A decrease in the abundance of the expression of retinal VEGF receptor mRNA could suggest apoptosis in endothelial cells. ²⁷ It was interesting to note that hyperoxia (which also caused increased expression of VEGF mRNA) resulted in a dramatic decrease in the abundance of retinal Flt-1 receptors in the Sal-, Dex-, and Veh-treated groups, whereas in the Cel-treated group, there was a relative predominance. This novel finding is of particular significance, because the Flt-1 (in contrast to the KDR/Flk-1) receptor also undergoes alternate splicing to produce a membrane-bound (Flt-1) and a soluble (s)Flt-1 receptor variant. ^{9 40} The soluble variant isoform appears to have a suppressive effect on VEGF, whereas membrane-bound Flt-1 induces VEGF, suggesting a dual, autoregulatory role for Flt-1. Recent work has shown that Flt-1–knockout mice exhibit disorganization of blood vessels and overgrowth of endothelial cells, ⁴¹ a phenomenon similar to neovascularization in ROP. The major positive signal transducer mediating endothelial proliferation and vascular permeability is KDR/Flk-1. ⁹ However, Flt-1 binds strongest to VEGF, and there are overwhelming data supporting its role in regulating VEGF overproduction. This negative regulation suggests that Flt-1 can be considered an optional therapy for control of tumor progression. The ability of Cel to dramatically upregulate Flt-1 receptor mRNA at a time when VEGF is supraelevated, indicates a beneficial role for the selective COX-2 inhibitor in diseases associated with VEGF overexpression resulting in neovascularization. It was notable that Dex-treatment did not have a similar significant impact on Flt-1 receptor mRNA during hyperoxia. 

The role of KDR/Flk-1 is very well defined. Binding of VEGF₁₆₅ to KDR/Flk-1 elicits strong signal transduction that mediates vasculogenesis and angiogenesis. ² The present results clearly demonstrate that hyperoxia results in downregulation of the receptor. The corollary to this, of course, is decreased vascularization, despite the increased expression of VEGF₁₆₅. The rapid response of KDR/Flk-1 receptor to hyperoxia suggests that it may be more sensitive to the suppressive effects of hyperoxia than VEGF itself in this model of partially vascularized retinas. It was interesting to note that at 12 days postnatal age, KDR/Flk-1 receptor mRNA was not detected in any of the groups, further emphasizing that rabbit retinas are partially vascularized and do not develop postnatally. 

In summary, we have demonstrated for the first time that Cel treatment during hyperoxia results in a significant reduction in retinal VEGF₁₂₁ (the VEGF splice variant involved in tumorigenesis) and in a marked increase in retinal expression of Flt-1 receptor mRNA (the receptor involved in regulating VEGF overproduction). Selective targeting of VEGF₁₂₁ and Flt-1 mRNA may represent one regulatory pathway for the anti-inflammatory effects of Cel. Given the evidence, we speculate that selective COX-2 inhibitors, such as Cel, may be considered to be promising therapeutic alternatives to corticosteroids for the treatment and/or prevention of diseases associated with neovascularization. Further studies are needed to examine the therapeutic effects in an animal model with fully vascularized retinas (such as rats), in the setting of ROP. 

 Presented in part at the Annual Meeting of the Society for Pediatric Research, Baltimore, Maryland, 2001.

 Supported by grants from the Long Beach Memorial Medical Center Foundation, Long Beach, California, and Pharmacia Corp., Chesterfield, Missouri.

 Submitted for publication November 15, 2001; accepted December 11, 2001.

 Commercial relationships policy: N.

 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.

 

Response of retinal VEGF to Dex and Cel during hyperoxia shown in a representative gel from a single sample of RT-PCR amplification of newborn rabbit retinal VEGF mRNA, yielding splice variants corresponding to VEGF₁₆₅ and VEGF₁₂₁. Top gels: β-actin; bottom gels: the two VEGF splice variants (top band: VEGF₁₆₅; bottom band: VEGF₁₂₁). (A) Histogram represents the Sal and Dex groups and (B) represents the Veh and Cel groups. Room air and hyperoxia groups of the (A; lanes 1 and 2) Sal-treated animals and (lanes 3 and 4) the Dex-treated animals and (B) of the (lanes 5 and 6) Veh-treated animals and (lanes 7 and 8) the Cel-treated animals, respectively. Data are the mean ratio ± SEM of VEGF mRNA to β-actin mRNA (n = 6 separate animals group). Retinas were taken from both eyes of each animal and pooled.

Figure 2.

 

Total retinal VEGF_165+121 during hyperoxia (A, B, top) and hyperoxia-recovery (A, B, bottom), measured by the sum of the splice variants in (A) Sal- and Dex- and (B) Veh- and Cel-treated retinas. Retinas were taken from both eyes of each animal and pooled. Samples from six separate animals in each group were analyzed. Data are expressed as the mean ratio ± SEM of VEGF_(165+121) to β-actin.

Figure 3.

 

Comparison of percentages of VEGF₁₆₅ and VEGF₁₂₁ in retinas taken from (A) Sal- and Dex- and (B) Veh- and Cel-treated newborn rabbits exposed to room air, hyperoxia (top, 80%–100% oxygen), or hyperoxia-recovery (bottom). Retinas were taken from both eyes of each animal and pooled. Data are expressed as the mean percentage of total VEGF mRNA ± SEM from six separate animals in each group.

Figure 4.

 

Response of retinal Flt-1 and KDR/Flk-1 receptor during hyperoxia, shown in a representative gel from a single sample of RT-PCR amplification of newborn rabbit retinal Flt-1 (A, B, top gel) and KDR/Flk-1 (A, B, bottom gel) receptor mRNA expression. Histograms represent (A) the Sal and Dex groups and (B) the Veh and Cel groups. Lanes and description of data are as described in Figure 1 .

Figure 5.

 

Response of retinal VEGF to Dex and Cel during hyperoxia-recovery, shown in a representative gel from a single sample of RT-PCR amplification of newborn rabbit retinal VEGF mRNA, yielding splice variants corresponding to VEGF₁₆₅ and VEGF₁₂₁. (A) Sal and Dex groups, (B) Veh and Cel groups. Top: β-actin; bottom: the two VEGF splice variants (top band: VEGF₁₆₅; bottom band: VEGF₁₂₁). In the histograms room air and hyperoxia-recovery groups (A) for (lanes 1 and 2) for the Sal-treated animals and (lanes 3 and 4) for the Dex-treated animals and (B) for (lanes 5 and 6) the Veh-treated animals and (lanes 7 and 8) for the Cel-treated animals, respectively. Data are the mean ratio ± SEM of VEGF to β-actin (n = 6 separate animals/group). Retinas were taken from both eyes of each animal and pooled.

Figure 6.

 

Response of retinal Flt-1 receptor during hyperoxia-recovery, shown in a representative gel from a single sample of RT-PCR amplification of newborn rabbit retinal Flt-1 receptor mRNA expression. (A) Sal- and Dex-treated groups; (B) Veh- and Cel-treated groups. Lanes and description of data are as described in Figure 5 .