Localization of Adenosine A2a Receptor in Retinal Development and Oxygen-Induced Retinopathy

Abstract

purpose. To investigate the association of adenosine A2a receptors (A2aR) with retinal vasculogenesis and angiogenesis that occurs in the canine model of oxygen-induced retinopathy (OIR).

methods. One-day-old dogs were exposed to 100% oxygen for 4 days and killed in oxygen (5 days old) and at 3, 10, 17, and 23 days after exposure to hyperoxia. Room air control animals were killed at 1, 5, 8, 15, 22, and 28 days of age. Immunolocalization of A2aR was performed on frozen sections, and reaction product density was quantified using microdensitometry. Cell types were identified in serial sections using antibodies against von Willebrand factor (endothelial cells) and GFAP (astrocytes), and enzyme histochemistry for menadione-dependentα -glycerophosphate dehydrogenase (M-α-GPDH) (to label angioblasts and developing blood vessels).

results. A2aR immunoreactivity was associated with forming blood vessels and angioblasts in the nerve fiber layer (NFL) of peripheral retina. As development progressed, vascular labeling decreased, whereas labeling of neuronal elements increased. In OIR, A2aR immunoreactivity in the NFL was reduced after exposure to hyperoxia and significantly elevated in the inner retina throughout vascularized retina and in advance of forming vasculature in all oxygen-treated animals returned to room air. A2aR immunoreactivity was also prominent in fronds of intravitreal neovascularization.

conclusions. A2aR immunoreactivity was associated with developing retinal vessels. As development progressed, vascular-associated A2aR labeling decreased and, concomitantly, labeling of neuronal elements increased. A2aR immunoreactivity was significantly elevated at the edge of forming vasculature in all animals returned to room air after hyperoxia and in intravitreal neovascular formations. These results provide additional evidence for the importance of A2aR and its ligand adenosine in retinal vascular development and in the vasoproliferative stage of canine OIR.

The purine nucleoside adenosine is a product of adenosine triphosphate (ATP) catabolism. Adenosine is a modulator of synaptic transmission ^{1 2} and a potent vasodilator. ^{3 4} In retina, adenosine dilates arterioles ^{5 6 7} and serves an autoregulatory role in mediating compensatory dilation in response to hypoxia, ⁸ ischemia, ⁹ hypotension, ⁸ and hypoglycemia. ¹⁰ Adenosine is chemotactic and/or mitogenic for some endothelium and angiogenic on the chorioallantoic membrane angiogenesis assay. ^{11 12 13} We have demonstrated that adenosine stimulates migration of adult retinal microvascular endothelial cells and formation of tubes in vitro, two events that are required in vasculogenesis. ¹⁴  

The newborn dog retina is 60% vascularized at birth, equivalent to a 7-month gestation human fetus in terms of retinal vascular development. Blood vessel assemblage in the dog retina occurs by a process of vasculogenesis, a term referring to de novo formation of vasculature from mesenchymal precursors or angioblasts. ^{15 16} We have demonstrated in the companion article in this issue that adenosine immunoreactivity is associated with vasculogenesis in the dog. Furthermore, the source of adenosine appears to be the ectoenzyme 5′ nucleotidase (5′N), which is transiently expressed on inner Muller cell processes during development. ¹⁷ When the neonatal dog is exposed to hyperoxia, vasculogenesis ceases, and vaso-obliteration occurs. ¹⁸ In Lutty et al., ¹⁷ (companion article) we demonstrate that adenosine levels and 5′N activity decrease in the vaso-obliterative stage of the canine model of oxygen-induced retinopathy (OIR). When animals are returned to room air, angiogenesis occurs in the nerve fiber layer and neovascularization invades the vitreous. During this stage in the dog, 5′N activity and adenosine levels increase markedly. ¹⁷  

Therefore, it appears that adenosine produced by Muller cells may stimulate normal retinal vasculogenesis and angiogenesis in the canine model of OIR. However, in order for adenosine to be vasogenic, angioblasts and immature endothelial cells would need to express adenosine receptors. Several subclasses of adenosine receptors have been identified. The two major classes are A1 and A2. In general, A1 receptors are associated with neuronal elements, whereas A2 receptors are often associated with vasculature. ^{19 20} Two A2 receptor subclasses are known, A2a and A2b. The A2a receptor binds adenosine with higher affinity than the A2b receptor. ^{19 21}  

In this study we sought to determine if adenosine A2a receptors were present in angioblasts and developing blood vessels and examine what, if any, changes in distribution occurred during development. We also investigated the association of A2a receptors with angiogenesis in the dog model of OIR. 

Methods

One-day-old purebred beagles were exposed to 100% oxygen for 4 days and killed in oxygen or returned to room air as previously described. ^{18 22} Animals were killed at 5, 8, 15, 22, and 28 days of age by an intraperitoneal overdose of sodium pentobarbital. One eye from triplicate oxygen-treated animals were compared to triplicate age-matched, room air–reared control animals (except at 28 days of age when only one was analyzed in each group). Eyes from three normal 1-day-old and two adult beagles also were included in the analysis. From the fellow eye of each animal, the retina was incubated for ADPase and flat-embedded, and results from these fellow eyes have been previously reported by McLeod et al. ^{18 22 23} Which eye was snap-frozen for histochemistry was randomized and should not be of consequence because severity of OIR in dog is remarkably bilateral. ²³ Animals were handled in accordance with the tenets of the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. 

Eyes were snap-frozen, serial-sectioned, and stained using streptavidin peroxidase immunohistochemistry ²⁴ and enzyme histochemistry to identify different cell types. Anti-von Willebrand factor (vWf; Accurate Chemical Co., Westbury, NY, 1:20,000 dilution) was used to identify endothelial cells in formed vessels. Anti-glial fibrillary acidic protein (GFAP; Dako, Carpenteria, CA, 1:60,000 dilution) was used to identify astrocytes. Primary antibodies were incubated overnight at 4°C and peroxidase reaction product developed with 3-amino-9-ethyl carbazol (Sigma Chemical Co., St. Louis, MO). ²⁴ The enzyme histochemical reaction for menadione-dependent alpha glycerophosphate dehydrogenase (M-α-GPDH) was used to identify angioblasts and immature endothelial cells. ²⁵ Localization of these antibodies and M-α-GPDH activity was compared to localization with anti-adenosine A2a receptor (A2aR) (Chemicon International, Temecula, CA, 1:1000 dilution). Control sections were incubated overnight at 4°C with nonimmune IgG (vWf, GFAP) or with A2a antibody that had been preadsorbed with the peptide (20-fold excess peptide by weight) used to generate the A2aR antibody (Chemicon International). The slides of serial sections from each eye were used in the following order: anti-vWf, anti-A2aR, nonimmune IgG or blocking control, M-α-GPDH, and anti-GFAP. The series was repeated three times for each eye. 

Microdensitometric measurements of inner retina were performed on triplicate slides from each eye from ora serrata to 7 mm posterior to quantify A2aR reaction product density and determine its relationship to developing blood vessels. Digital images of inner retina were captured using a charge coupled device (CCD) camera (Hamamatsu, Hamamatsu City, Japan) and a Macintosh IIci computer (Cupertino, CA) with NIH Image version 1.47 software. Three separate measurements in the inner retina were made at 1-mm steps from ora serrata to 7 mm posterior and precisely at the edge of vasculature in each section, three sections per animal (a total of 2106 measurements), using the density plot profile function of the software. The sample with the most reaction product (15-day-old, oxygen-treated) was used to set the gain and offset on the video system. The background was set near zero on the grayscale (central vitreous cavity). The darkest structure in the nerve fiber layer was used to set the values nearest 255 (upper limit of grayscale). This assured that all density measurements were made in the range of 0 to 255 arbitrary units (histogram optimization). Once the illumination, gain, and offset were set for a group of animals, all images were captured under identical conditions. Density plot profiles were generated using rectangular field selections (75 μm wide and 200μ m high) through the inner retina. The background density of the vitreous was subtracted from the peak density of each plot, which coincided with the nerve fiber layer for A2aR. Therefore, a single observer made three measurements every 1 mm from ora serrata in triplicate slides from each animal, and there were three animals in each group except at 28 days. Direct microdensitometric comparisons were made on all sections from an age-matched control and oxygen-treated animal that were incubated at the same time in the same reagents so that immunohistochemical conditions were identical. It was not possible to make comparisons between all animals in the groups except at the edge of the vasculature, an area of interest, because of the variability in location of the edge of the vasculature in each animal, especially oxygen-treated animals, ²³ even though the superior lobe was always used for analysis. The mean density and SEM were calculated for each region or structure from nine density plot profiles (three per region per section), and statistical analysis of the data were performed using the two-tailed Student’s t-test. 

Results

One Day of Age

The inner retinal vasculature was several millimeters from the ora serrata in the 1-day-old normal dog, as determined with vWf immunolabeling (Fig. 1A ). vWf localization permitted the border of vascularized retina to be clearly delineated. Angioblasts, as indicated by M-α-GPDH enzyme histochemistry, ²⁵ were present throughout the inner retina from ora serrata to disc (Figs. 1E 1F) . M-α-GPDH enzyme histochemistry also labeled forming vessels (Fig. 1E) and, to a lesser extent, vessels in more central retina (Fig. 1F) . A2aR immunoreactivity was associated with formed vasculature and with angioblasts in advance of the vasculature (Figs. 1C 1D) . A2aR localization was abolished by preincubation of the antibody with peptide used as antigen (results not shown). Astrocytes, as indicated by GFAP labeling, were present in areas of posterior vascularized inner retina (Fig. 1H) but were not present at or anterior to the edge of forming vasculature (Fig. 1G) . A2aR localization in the 1-day-old retina most closely resembled the localization of M-α-GPDH. However, because the enzyme is mitochondrial ²⁶ and the receptors are associated with cytoplasmic membrane, the appearance of the two reaction products was not identical. 

Five Days of Age

The peripheral retina was still not fully vascularized in normal 5-day-old dog (Fig. 2A ). A2aR immunoreactivity was associated with forming blood vessels at the border of vascularized retina (vWf positive, Fig. 2A ) and in angioblasts anterior to the vascular border (Fig. 2C) . In the 5-day-old dog killed in the isolette after 4 days exposure to hyperoxia, the retinal vasculature was highly constricted and the majority of the capillaries were obliterated (Fig. 2B) . ¹⁸ A2aR immunoreactivity was present in cells of the avascular and vascularized retina in the 5-day-old, oxygen-treated animal (Fig. 2D) , but the relative amount was reduced compared to the room air control animal (Fig. 2E) . Only a representative pair of animals is shown in Figure 2E and other graphs of densitometric values and distance from the ora serrata, when air control and oxygen-treated animals are compared, because of the differences in distance between the ora serrata and the edge of the vasculature in animals, especially those exposed to oxygen. However, the shapes of the curves and trends in reaction product densities were similar for the triplicate animals in each group. At the edge of the vasculature, the reaction product density in the oxygen-treated animals was reduced by as much as 14% compared to air control animals (Fig. 2F) . Using microdensitometry, there was a 52.1% increase in GFAP reaction product density at the edge of the vasculature in oxygen-treated animals compared to controls which was significant (P = 0.0013) (results not shown). 

Eight Days of Age

The inner retinal vasculature was close to the ora serrata in the 8-day-old control animal (Fig. 3A ). A2aR immunoreactivity was present in formed vessels and in advance of the edge of the vasculature (Fig. 3C) . Angioblasts were present in advance of the edge of formed vessels (Fig. 3E) , but astrocytes trailed the forming vasculature (Fig. 3G) . Because of vaso-obliteration, radial growth of the vasculature in oxygen-treated animals was considerably retarded (Fig. 4A ), as demonstrated by vWf localization (Fig. 3B) . Angioblasts were present in advance of the formed vasculature (Fig. 3F) and a few astrocytes were present at the edge of the vasculature in oxygen-treated animals (Fig. 3H) . There was a significant 156.5% increase in GFAP reaction product at the edge of the vasculature in oxygen-treated animals (P < 0.0001). A2aR immunoreactivity was elevated in oxygen-treated animals compared to room air controls, and this was most obvious in advance of formed vasculature (Fig. 3D) . Microdensitometry demonstrated that A2aR reaction product was elevated in all areas of retinal vasculature in oxygen-treated animal, except in far periphery (Fig. 4A) , and that this difference in inner retinal immunoreactivity was quite significant at the edge of the vasculature (P < 0.0001), when data from all animals at this age were analyzed (Fig. 4B) . 

Fifteen Days of Age

The inner retinal vasculature had almost reached the ora serrata in the 15-day-old control animals (Fig. 5 A), and a secondary, deep capillary network had begun forming more posteriorly (results not shown). A2aR immunoreactivity and M-α-GPDH activity were still prominent in inner retina (Figs. 5C 5E) . Radial growth of the retinal vasculature was still severely retarded in the oxygen-treated animal (Fig. 6A ), and intravitreal neovascular formation was present. Astrocytes were present at the edge of (Fig. 5H) and in advance of the vasculature, and GFAP reaction product at the edge was 2.2-fold greater in oxygen-treated animals than controls (P < 0.001) (results not shown). A2aR immunoreactivity was greatly elevated at the edge of the vasculature where numerous M-α-GPDH+ angioblasts were present in advance of formed vessels (Figs. 5D 5F) . Microdensitometric analysis demonstrated that A2aR was elevated in the oxygen-treated animal in all areas except at the ora serrata (Fig. 6A) . The highest level of A2aR immunoreactivity observed in this study was in the 15-day-old, oxygen-treated animals. This was evident when the values of all animals at the edge of the vascular border were compared (Figs. 2F 4B 6B 8F) . 

Twenty-two Days of Age

At 22 days of age, a large sinusoidal vein occupied the region nearest the ora seratta in all three air control animals (Fig. 7A ). This is a normal feature of developing canine retinal vasculature at this age. Vasoformative cells (M-α-GPDH positive) were present where the secondary capillary network forms in the inner nuclear layer (Fig. 7E) . Some astrocytes also were present at ora serrata at this age (Fig. 7G) . A2aR immunoreactivity was still present at the ora serrata, but it was most prominent in the area where the secondary capillary network forms (Fig. 7C) . In oxygen-treated animals, the peripheral 4 mm of retina was still avascular (Fig. 7B) , but M-α-GPDH–positive angioblasts were quite prominent (Fig. 7F) . A2aR immunoreactivity was very high in this region and appeared coincident with the location of M-α-GPDH–positive angioblasts (Figs. 7D 7F) . Although this region of inner retina was devoid of vasculature, numerous astrocytes were present in this area (Fig. 7H) . 

At the edge of the vasculature in oxygen-treated, 22-day-old animals, A2aR immunoreactivity and M-α-GPDH activity were high in cells from the internal limiting membrane to outer plexiform layer (Figs. 8B 8C ). Numerous astrocytes were present, giving the appearance of astrogliosis (Fig. 8D) . Microdensitometry demonstrated that A2aR immunoreactivity was elevated in all areas of the inner retina of oxygen-treated animal, and this difference was not significant at the edge of the vasculature (P = 0.29) (Figs. 8E 8F) . Posteriorly, intravitreal neovascular formations were prominent. 

Twenty-eight Days of Age

The retina was completely vascularized in the 28-day-old normal animal in that the primary vasculature (nerve fiber layer) had reached ora serrata and secondary capillaries (inner nuclear layer) had formed (Figs. 9A 9B ). In peripheral retina, A2aR immunoreactivity was most prominent in vessels of both the inner and outer vasculatures (Fig. 9C) . However, at the optic nerve head A2aR immunoreactivity was associated primarily with nerve fibers in inner retina and within the optic nerve head (Fig. 9D) . Binding of the A2aR antibody to vessels in peripheral retina and nerve fibers near and in the optic nerve was completely blocked by preincubation of the antibody with the peptide used as antigen (Figs. 9E 9F) . 

Figure 10 compares posterior retina in a 28-day old air control and in a 28-day-old, oxygen-treated animal with proliferative retinopathy. There were vWf-positive vessels present in the nerve fiber layer and in the inner nuclear layer in oxygen-treated animal, but there also was extensive intravitreal neovascularization (Figs. 10B) . A2aR immunoreactivity was associated with vessels in both air control and oxygen-treated animals, but the relative amount of A2aR immunoreactivity was greatly increased in the oxygen-treated animal (Figs. 10C 10D) . A2aR immunoreactivity also was present in the neovascular formations. M-α-GPDH activity was associated with the plexiform layers in both animals, but the activity was also prominent in retinal vessels and neovascular formations in the oxygen-treated animal (Fig. 10F) . Astrocytes were present at the inner aspect of both 28-day-old retinas, but GFAP immunolabeling demonstrated astrogliosis in the inner retina of oxygen-treated animal (Figs. 10G 10H) . Astrocytes were not associated with the intravitreal vessels but were present around the base of the neovascular feeder vessels (Fig. 10H) . 

The adult pattern of A2a localization was similar to the 28-day-old control animal, but the capillary staining was reduced in the adult. A2a immunoreactivity in the adult was most prominent in nerve fibers and the inner nuclear layer (results not shown). 

Neovascular Formations in Oxygen-Treated Animals

We have reported two types of intravitreal neovascular formations (immature and mature) in canine OIR, based on morphologic criterion. ²³ Mature formations consist of capillary-like vessels with well-differentiated endothelial cells and pericytes. Immature formations are highly cellular and consist of poorly differentiated cellular components with few canalized lumens. ²³ vWf immunoreactivity was present in both forms of intravitreal neovascularization (Figs. 11A 11B ). However, M-α-GPDH activity was 2.8-fold greater by microdensitometric analysis in immature formations than in mature (Figs. 11E 11F) . A2aR immunoreactivity was 2.5-fold greater by microdensitometric analysis in immature neovascular formations than in mature (Figs. 11C 11D) . 

In summary, adenosine A2a receptor immunoreactivity was associated with vasculature and vascular precursors, angioblasts, at all stages in retinal vascular development. A2a immunoreactivity was significantly elevated in oxygen-treated animals compared air controls at the edge of the vasculature at 8 and 15 days of age (Fig. 12) , the proliferative stage in oxygen-induced retinopathy. Very high levels of A2a were observed in intravitreal neovascular formations, especially immature formations (Fig. 11) . 

Discussion

High levels of adenosine are associated with areas of vasculogenesis in the normal neonatal dog retina and sites of angiogenesis in the canine model of OIR, as shown in Lutty et al. ¹⁷ (companion article). The source of the adenosine appears to be 5′nucleotidase on the Muller cell cytoplasmic membranes. The present study suggests that one of the adenosine receptors, A2a, is associated with angioblasts and endothelial cells in areas with elevated adenosine during vasculogenesis and angiogenesis in OIR. 

Normal Retinal Vasculogenesis

The present study demonstrated that the A2a subtype of adenosine receptor (A2aR) was associated with retinal vasculogenesis, both developing vessels and vascular precursors (angioblasts) being immunoreactive. Gidday and Park ⁷ demonstrated the presence of A2 receptors in neonatal retina using a functional assay that showed A2 receptors mediated arteriolar dilation. When vascular development was complete in canine retina, A2a receptors became associated with neuronal elements. This was quite apparent near the optic nerve head at 28 days of age where A2a receptors were associated with nerve fibers and only weakly associated with vessels which were relatively mature at this age (Fig. 9) . At this age, A2aR immunoreactivity in the rest of retina was associated most prominently with both capillary networks, the inner nuclear layer, and nerve fibers. This localization is comparable to the in situ hybridization for A2aR performed by Kvanta et al. ²⁷ in adult rat, which demonstrated that A1 receptor mRNA was most prominent in the ganglion cell layer, whereas A2aR was associated primarily with the inner nuclear layer. Blazynski and coworkers ^{28 29} also characterized the localization of adenosine receptors in adult retina of several species. They found, using radiolabeled agonists, that A1 receptors were prominent in inner retina of most mammals and that A2 receptors were most prominent in outer retina. Their A2 data were based on binding of N-ethylcarboxamido adenosine (NECA; binds A1 and A2 with different affinities), which bound mostly to outer segments. ^{28 29} This difference in results could be explained by the three different techniques used in these studies (immunohistochemistry, in situ hybridization, radioligand binding), specificity of reagents, and the fact that our study focused on neonatal dog retina and the prior localization studies were performed on adult retina from other species. 

A2aR localization at the edge of the developing vasculature was very similar to localization with M-α-GPDH and vWf, suggesting that the A2aR-immunoreactive cells were angioblasts and endothelial cells of immature vessels (Fig. 1) . The staining pattern of M-α-GPDH and A2aR reaction products was somewhat different, but this may be due to M-α-GPDH being in mitochondria, ²⁶ whereas A2aR are located on the cytoplasmic membrane. The association of A2aR with vasculature in developing retina could have been expected because we have found that A2aR agonists stimulate migration and tube formation of adult retinal microvascular endothelial cells in vitro, two processes required in vasculogenesis. ¹⁷ We examined the distribution of astrocytes in this study because, in other species, astrocytes are thought to induce normal vessel development by producing vascular endothelial growth factor (VEGF) as they migrate in advance of developing vessels. ³⁰ In the dog, astrocyte spread toward periphery trailed vascular development (Fig. 1) , so A2aR-immunoreactive cells in advance of the vasculature and at the edge of the vasculature were not coincident with GFAP-positive astrocytes. In areas with more developed vessels, GFAP-positive astrocytes were present adjacent to the internal limiting membrane (ILM). A2aR localization in these areas, however, was present from the ILM to the ganglion cell layer. So it is possible that some of the labeling in the innermost part of retina can be attributed to astrocytes, but other cell types were also positive in the same areas because labeling extended out to the ganglion cells. There is evidence in other organ systems of adult animals that A2a receptors are present on smooth muscle cells and endothelial cells. ³¹ Smooth muscle localization may be present late in development of the dog retinal vasculature where A2aR localization appeared perivascular (Fig. 10) and not lumenal. 

Oxygen-Induced Retinopathy

A2aR also was associated with angiogenesis in the canine model of OIR. High levels were observed in the nerve fiber layer at the border of vascularized retina, where multiple layers of capillaries form in the oxygen-treated dog. ²³ Higher A2aR immunoreactivity was localized to immature intravitreal neovascular formations, which have poorly differentiated cellular components and few canalized lumens, than mature formations, which have well-differentiated endothelial cells and pericytes. ²³ This differential staining of A2aR in neovascular formations could be related to the less differentiated state of vasoformative cells in immature formations, which is reflected by their higher levels of the M-α-GPDH activity. The neovascular formations provide the strongest evidence for A2a receptors being associated with vasoformative cells and endothelial cells, because both immature and mature intravitreal neovascular formations were positive. We previously demonstrated that the immature formations consist of angioblast-like cells that subsequently differentiate into endothelial cells and pericytes. ²³  

A2a Receptors

Binding of adenosine to A2a receptors stimulates adenylate cyclase in tissues like brain. ^{32 33} Gidday et al. ³⁴ recently demonstrated that adenosine increases retinal blood flow by activating K_ATP channels, not by increasing cAMP via activating adenylate cyclase. However, adenosine and not A2a-specific agonists were used in the work of Gidday et al. Furthermore, the vasodilation and increase in blood flow that were measured in the studies just mentioned are modulated presumably by smooth muscle cells, so these studies may have assessed the effects of adenosine on smooth muscle cells. It may be that A2a signaling in endothelial cells and angioblasts is through adenylate cyclase. Evidence for this comes from the work of Takagi et al., ³⁵ who demonstrated that hypoxia induced cAMP elevation in retinal endothelial cells is blocked by A2a selective antagonists. If adenylate cyclase is activated via adenosine binding to A2aR on endothelial cells and angioblasts, there are many ramifications of adenylate cyclase stimulation in endothelial cells, including cell shape change, ³⁶ changes in junctional permeability, ³⁷ and angiogenesis. Elevated cAMP has been correlated with increased expression of VEGF mRNA in smooth muscle cells ³⁸ and with transcription of VEGF receptor FLT-1 in endothelial cells. ³⁹ VEGF has been implicated in both vascular development and angiogenesis in other models of OIR. ^{40 41 42 43}  

Fisher et al. ⁴⁴ were the first group to demonstrate that adenosine stimulated production of VEGF. Takagi and associates ³⁵ then suggested that hypoxia-stimulated upregulation of VEGF mRNA is via the cellular production of ADO. They demonstrated that when adenosine agonists binds to A2a receptors, production of cAMP is elevated, activation of protein kinase A occurs, and then VEGF production is induced. Ironically, Takagi et al. ⁴⁵ have also demonstrated that binding of A2aR with A2 agonists inhibits the production of the VEGF receptor KDR. This would suggest that the paucity of KDR that we have observed recently on canine angioblasts in developing retina may be related to the high levels of A2a on angioblasts and developing vessels reported in this article. ⁴⁶ This would not explain, however, the high levels of KDR and A2a associated with intra- and extraretinal neovascular formations in the canine model of OIR. ⁴⁶ These studies taken together suggest that adenosine and its A2a receptor might actually control the level of VEGF in hypoxic retina and expression of its receptors. 

Adenosine is also a potent vasodilator and A2aR, specifically, is an important modulator of vascular tone. ^{47 48} Binding of A2aR on endothelial cells and smooth muscle cells induces vasodilation by stimulating l-arginine transport and nitric oxide (NO) production by endothelial cells and smooth muscle cells. ^{49 50} Vasodilation is prominent during vascular development in dog and during the proliferative stage in canine OIR. Gidday and Park ⁷ have demonstrated that A2 receptors can specifically modulate vasodilation in the neonatal pig. Extreme vasodilation associated with increased adenosine and A2a receptors in oxygen-treated animals may contribute to the tortuousity of arteries and hemorrhage that we have observed in the canine model of OIR. ²³  

Dilatation is the normal vascular response to hypoxia in all organs except lung. Adenosine levels are elevated in most tissues during ischemia and hypoxia. The peripheral retina during vasculogenesis and the majority of the dog retina after vaso-obliteration are presumed to be hypoxic, and both the normal developing retinal vessels and reforming vasculature in the proliferative phase of OIR are extremely dilated and have elevated adenosine and elevated A2aR. Roth et al. ⁵¹ have demonstrated that adenosine levels increase rapidly in retina after induction of ischemia. If the retina is made ischemic and then reperfused, retinal function and structure are severely affected, i.e., ischemia/reperfusion injury. Administration of an A2a antagonist, not an A1 antagonist, can protect both retinal function and structure after ischemia and reperfusion. ⁵²  

In summary, adenosine A2a receptors were expressed in high levels by angioblasts and retinal vessels during development. As vessels mature, A2aR labeling of vasculature decreased, whereas neuronal element labeling increased. As demonstrated in the companion article, ¹⁷ Muller cells may stimulate vasculogenesis by producing adenosine via 5′-nucleotidase during the period in which angioblasts and endothelial cells express high levels of adenosine A2a receptors. A2a receptors also were associated with angiogenesis in the canine model of OIR. In general, adenosine A2a receptors were elevated in the inner retina during the vasoproliferative stage in the canine model of oxygen-induced retinopathy compared to room air control retinas, and the most prominent increase was at the edge of the vasculature in oxygen-treated animals. A2aR was more prominent in immature than mature intravitreal neovascular formations. This suggests that both adenosine and its A2a receptor are important in normal vasculogenesis and the angiogenesis that occurs during the vasoproliferative stage in the canine model of OIR. Now that very selective and potent antagonists of A2aR have been synthesized, ^{21 53} the A2a receptor may be a therapeutic target for controlling retinal angiogenesis in OIR. 

 Supported by National Institutes of Health Grants EY 01765 (Wilmer Institute) and EY09357 (GL) and the ROPARD Foundation. Gerard A. Lutty is an American Heart Association Established Investigator and the recipient of a Research to Prevent Blindness Lew Wasserman Merit Award.

 Submitted for publication November 24, 1998; revised June 28, 1999; accepted August 17, 1999.

 Commercial relationships policy: N.

 

Figure 1.

 

Serial sections showing the edge of developing vasculature (A, C, E, G) and 3.0 mm posterior to the edge of the vasculature (B, D, F, H) in a normal 1-day-old dog. In the left panels, the long thin arrows indicate the edge of the vasculature. In the right panels, the short bold arrows point to the same vessel in each serial section. In (A) and (B), the open arrows point toward the ora serrata and the inner plexiform layer (ipl) is labeled for orientation. (A) vWf immunoreactivity was present only in vessels with formed lumens at the peripheral edge of the vasculature (long thin arrow). (B) Vessels posterior to the edge immunolabeled with vWf (short bold arrow). (C) A2a receptor staining was present in newly formed vessels at the edge of the vasculature (long thin arrow) and in cells (arrowheads) in advance of the vasculature. (D) A2aR labeling in central retina was associated with blood vessels (short bold arrow) and other cells in the nerve fiber layer. (E) M-α-GPDH activity was present in developing vessels at the edge of the vasculature (long thin arrow) and in angioblasts (arrowheads) in avascular retina. (F) M-α-GPDH was weaker in well-established vessels, but intense staining was associated with cells throughout the inner retina. (G) No appreciable GFAP labeling was observed near the edge of the vasculature (long thin arrow). (H) GFAP labeling clearly demonstrated that astrocytes were present at the vitreoretinal border and surrounding formed blood vessels (arrow) in more posterior regions. AEC reaction product, (A, B, C, D, G, H); M-α-GPDH reaction product, (E, F). Magnification, ×80.

Figure 2.

 

Comparison of vWf and A2a immunolabeling at the edge of the vasculature in a 5-day-old air control (A, C) and in a 5-day-old animal after 4 days exposure to hyperoxia (B, D) and microdensitometric analysis (E, F). In (A) through (D), the long thin arrows indicate the edge of the vasculature. In (A) and (D), the open arrows point toward ora serrata and the inner plexiform layer (ipl) is labeled for orientation. (A) vWf labels formed blood vessels in normal inner retina. (B) In the animals exposed to oxygen, vaso-obliteration greatly reduced the number of viable vessels and surviving vessels were highly constricted as assessed by vWf labeling. (C) A2aR labeling was present in developing vessels at the edge (arrow) and in advance of the vessels in the air control retina. (D) A2aR labeling was reduced in inner retina of the animal exposed to hyperoxia. AEC reaction product in all. Magnification, ×80. (E) The mean density of A2aR immunoreactivity (relative grayscale values) is shown in two representative animals in areas sampled, indicated in millimeters from the ora serrata on the x-axis. A2aR immunoreactivity was reduced in the oxygen-treated animals (dashed line) in almost all areas. The data points represent the mean of the three measurements in each area of three sections from the animal ±SEM and, therefore, simply demonstrate the reproducibility of the technique. The vertical arrows indicate the edge of the vasculature. (F) Mean A2aR reaction product at the edge of the vasculature for all three animals in each group. A2aR was reduced in the oxygen-treated animals (open bar) compared to the control animals (shaded bar), but the difference was not significant. P = 0.16.

Figure 3.

 

Comparison of serial sections at the edge of the vasculature in an 8-day-old air-control animal (A, C, E, G) and in an oxygen-treated animal 3 days after return to room air (B, D, F, H). The long thin arrows indicate the edge of the vasculature, the open arrows point toward ora serrata and the inner plexiform layer (ipl) is labeled for orientation. (A) vWf labeling demonstrated that the edge of the vasculature is near the ora serrata (left edge). (B) vWf labeling was present in reforming vessels at the border of vascularized retina, but the edge was 2.1 mm from the ora serrata. (C) A2aR labeling was present at and in advance of the formed vessels in the control animal. (D) In the oxygen-treated animals, A2aR labeling was well in advance of the edge of the vasculature (arrow) and amount of the labeling in inner retina was greater than in control animals. (E, F) M-α-GPDH reaction product was present in vessels at the edge of the vasculature and in angioblasts in advance of the vasculature in both animals. (G) Astrocytes, as represented by GFAP labeling, had not advanced into the area associated with the edge of the vasculature in controls. (H) Compared to the control, there was more GFAP labeling in oxygen-treated animals at the vascular border. Magnification, ×80.

Figure 4.

 

Microdensitometric analysis of A2aR reaction product at 8 days of age. (A) The mean density of A2aR immunoreactivity (relative grayscale values) is shown in two representative animals in areas sampled, indicated in millimeters from the ora serrata on the x-axis. The mean density of A2aR immunoreactivity is elevated in the oxygen-treated animal (dashed line) in all areas except ora serrata and 1 mm posterior to it. The data points represent the mean of the three measurements in three sections in each area ±SEM and, therefore, simply demonstrate the reproducibility of the technique. The vertical arrows indicate the edge of the vasculature. (B) Mean A2aR reaction product at the edge of the vasculature for all three animals in each group. A2aR was significantly elevated in the oxygen-treated animals (open bar) compared to the control animals (shaded bar). *P < 0.0001.

Figure 5.

 

Serial sections showing the edge of the vasculature in a 15-day-old control animal (A, C, E, G) and in a 15-day-old animal 10 days after return to room air (B, D, F, H). The long thin arrows indicate the edge of the vasculature, the open arrows point toward ora serrata, and the inner plexiform layer (ipl) is labeled for orientation. (A) The primary retinal vasculature (vWf labeling) had almost reached the ora serrata in the control animal. (B) The edge of reforming vasculature (vWf labeling) was still distant from ora serrata in the oxygen-treated animal. (C) A2aR immunoreactivity was still prominent in inner retina at the edge of and in advance of the vasculature. (D) A2aR labeling was more pronounced in the oxygen-treated animal at the edge of the vasculature (arrow) and well in advance of it. (E) M-α-GPDH activity was present in blood vessels and angioblasts in the control. (F) Numerous M-α-GPDH–positive angioblasts were present in advance of the edge of the vasculature in the oxygen-treated animal. (G, H) Some GFAP-positive astrocytes were present at the edge of the vasculature in both animals. Magnification, ×80.

Figure 6.

 

Microdensitometric analysis of A2aR reaction product in inner retina of 15-day-old animals. (A) Comparison of mean A2aR reaction product density (relative grayscale values) in two representative animals at 15 days of age. A2aR reaction product was elevated in the oxygen-treated animal (dashed line) in all areas except at ora serrata. The data points represent the mean of the three measurements in three sections in each area ±SEM and, therefore, simply demonstrate the reproducibility of the technique. The edge of the vasculature is noted by vertical arrows. (B) A2aR reaction product density in all 15-day-old animals (n = 3 in each group) at the edge of the vasculature. The oxygen-treated group (open bar) had greatly elevated immunoreactivity compared to control group (shaded bar), and this was the highest level of A2aR reaction product observed in the study. *P < 0.0001.

Figure 7.

 

Comparison of retina at ora serrata in a 22-day-old air control (A, C, E, G) and oxygen-treated animal (B, D, F, H). The long thin arrows indicate the edge of the vasculature in the left panels, the open arrows point toward ora serrata and the inner plexiform layer (ipl) is labeled for orientation. (A) vWf labeling demonstrated a large dilated vein in the control animal at ora serrata. (B) The peripheral retina was still avascular in the oxygen-treated animal as demonstrated with vWf immunohistochemistry. (C) A2aR reaction product was decreased in inner retina compared to day fifteen (Fig. 5C) and reaction product is now present in the inner nuclear layer where the secondary retinal vasculature will develop. (D) A2aR labeling was prominent throughout the avascular peripheral inner retina of the oxygen-treated animal. (E) M-α-GPDH activity was still present in angioblasts of the inner retina but it also labeled cells in the inner nuclear layer where secondary capillaries will form (arrowhead). (F) M-α-GPDH–positive angioblasts were present throughout peripheral inner retina of the oxygen-treated animal. (G) A few astrocytes (GFAP+) were present at the ora serrata in the control animal. (H) GFAP+ astrocytes have migrated to ora serrata in the oxygen-treated animal. Magnification, ×80.

Figure 8.

 

Edge of the vasculature in a 22-day-old, oxygen-treated animal and microdensitometric analysis. The long thin arrows indicate the edge of the vasculature, the open arrows point toward ora serrata, and the inner plexiform layer (ipl) is labeled for orientation. (A) Dilated, vWf+ blood vessels were present at the border of vascularized retina. (B) A2aR reaction product was still very prominent at and in advance of the edge of the vasculature. (C) M-α-GPDH–positive angioblasts were abundant in inner retina around formed vessels and in advance of the edge of the vasculature. The forming outer plexiform layer was also positive for this enzyme. (D) Astrocytes (GFAP+) were prominent at and in advance of the edge of the vasculature. Magnification, ×80. (E) Microdensitometric analysis of two representative 22-day-old animals demonstrated that A2aR reaction product density (relative grayscale value) was elevated in all areas of the oxygen-treated animal (dashed line) compared to the control animal (solid line). The data points represent the mean of the three measurements in three sections in each area ±SEM and, therefore, simply demonstrate the reproducibility of the technique. Vertical arrows indicate the edge of the vasculature. (F) Mean A2aR values at the edge of the vasculature were elevated in oxygen-treated group (open bar) compared to the air control group (shaded bar), but the difference was not significant. n = 3 in each group. P = 0.29.

Figure 9.

 

Periphery retina (A, C, E) and optic nerve head (B, D, F) in a 28-day-old air-control animal. The large open arrows point toward ora serrata and the inner plexiform layer (ipl) is labeled for orientation. (A) Vascular development was complete in peripheral retina, as demonstrated by vWf immunolabeling of blood vessels in inner retina (straight arrow) and in the inner nuclear layer (curved arrow). (B) vWf labeling at the optic nerve head (small open arrow). (C) A2aR labeling was associated with blood vessels in inner retina (straight arrow) but also in the inner nuclear layer (curved arrow). (D) A2aR labeling at the optic nerve head (small open arrow) was most prominent in nerve fibers. (E, F) Binding of the A2aR antibody to blood vessels (arrows) and nerve fibers was blocked when the antibody is preincubated with the peptide used as antigen. Magnification, (A, C, E)× 80; (B, D, F) ×50.

Figure 10.

 

Comparison of central retina in a 28-day-old air control (A, C, E, G) and oxygen-treated (B, D, F, H) animals. The short bold arrows in (A), (C), and (E) indicate the same vessel in the nerve fiber layer, the curved arrows indicate blood vessels in the deep vasculature, the open arrows point toward ora serrata, and the inner plexiform layer (ipl) and vitreoretinal border (V) are labeled for orientation. (A) vWf labeling demonstrated the primary (short bold arrow) and secondary vasculatures (curved arrow). (B) In the 28-day-old, oxygen-treated animal, there were vWf-positive vessels in inner retina (arrowhead) and inner nuclear layer (curved arrow) and also in the vitreous (paired arrows). (C) A2aR labeling was mostly in perivascular nerve fibers in inner retina and diffusely present in the inner nuclear layer (curved arrow). (D) A2aR labeling in the oxygen-treated animal was diffuse but prominent in the nerve fiber layer and inner nuclear layer (curved arrow) and also present in the intravitreal neovascular formation (paired arrows). (E) M-α-GPDH+ capillaries were present in nerve fiber layer of the control animal but activity was also prominent in both plexiform layers and the secondary capillaries (curved arrow). Large retinal vessels have weak M-α-GPDH activity (short bold arrow). (F) M-α-GPDH activity in the oxygen-treated animal was present in both plexiform layers, inner and outer (curved arrow) retinal vasculatures, and in the intravitreal neovascular formation (paired arrows). (G) GFAP-positive astrocytes were present in innermost retina in the control retina. (H) GFAP labeling demonstrated inner retinal astrogliosis in this area with neovascularization, but astrocytes were only present at the base of the feeder vessels (horizontal arrow) and not within the neovascular formation. Magnification, ×80.

Figure 11.

 

Comparison of two forms of intravitreal neovascular formations (paired arrows) in the canine model of OIR: mature in a 28-day-old animal (A, C, E) and immature in a 15-day-old animal (B, D, F). The open arrows point toward ora serrata, the paired arrows indicate intravitreal neovascular formations, and the inner plexiform layer (ipl) and vitreoretinal border (V) are indicated for orientation. (A) vWf immunohistochemistry demonstrated the delicate capillary-like vessels of the mature neovascular formation (paired arrows). (B) vWf labeling of the immature formation (paired arrows) demonstrate that it is a highly cellular polyp-like structure with few canalized lumen. (C) There was limited A2aR labeling of the mature, well-differentiated form of neovascularization. (D) There was much more prominent A2aR labeling of the immature neovascular formation. (E) M-α-GPDH activity was low in the mature neovascular formation. (F) M-α-GPDH activity was very high in the immature formation. Magnification, ×80.

Figure 12.

 

Microdensitometric analysis at the edge of the vasculature at each time point for oxygen-treated animals (dashed line) compared to the control animals (solid line). Each data point represents the mean value from all three animals in each group at that time point ±SEM. The mean A2aR values at the edge of the vasculature were significantly elevated in oxygen-treated group compared to the air control group at 8 and 15 days of age. *P < 0.0001.