Cellular and Molecular Neurobiology, Vol. 11, No. 5, 1991

Adenosine in Vertebrate Retina: Localization, Receptor Characterization, and Function Christine Blazynski 1'3 and Maria-Thereza R . Perez 2 Received May 30, 1990; accepted December 14, 1990

KEY WORDS: retinal neuromodulator; adenosine receptors; adenosine; retina.

SUMMARY 1. The uptake of [3H] adenosine into specific populations of cells in the inner retina has been demonstrated. In mammalian retina, the exogenous adenosine that is transported into cells is phosphorylated, thereby maintaining a gradient for transport of the purine into the cell. 2. Endogenous stores of adenosine have been demonstrated by localization of cells that are labeled for adenosine-like immunoreactivity. In the rabbit retina, certain of these cells, the displaced cholinergic, G A B A e r g i c amacrine cells, are also labeled for adenosine. 3. Purines are tonically released from dark-adapted rabbit retinas and cultured embryonic chick retinal neurons. Release is significantly increased with K ÷ and neurotransmitters. The evoked release consists of adenosine, A T P , and purine metabolites, and while a portion of this release is Ca z÷ dependent, one other component may occur via the bidirectional purine nucleoside transporter. 4. Differential distributions of certain enzymes involved in purine metabolism have also been localized to the inner retina. 5. Heterogeneous distributions of the two subtypes of adenosine receptors, A1 and A2, have been demonstrated in the mammalian retina. Coupling of receptors to adenylate cyclase has also been demonstrated. 1Departments of Biochemistry and Molecular Biophysics, Anatomy-Neurobiology, and Ophthalmology and Visual Sciences, Washington University School of Medicine, 660 South Euclid Avenue, Box 8231, St. Louis, Missouri 63110. a Department of Ophthalmology, University of Lund, S-22185 Lund, Sweden. 3To whom correspondence should be addressed. 463 0272-4340/91/1000-0463506.50/0 © 1991 Plenum Publishing Corporation

464

Blazynski and Perez

6. Adenosine A1 receptor agonists significantly inhibit the K+-stimulated release of [3H]-acetylcholine from the rabbit retina, suggesting that endogenous adenosine may modulate the light-evoked or tonic release of ACh.

INTRODUCTION

The vertebrate retina has been extensively studied with an aim to identify the neurochemical agents involved in signal transmission. Consistent with what has been learned from other parts of the central nervous system, certain biogenic amines, such as dopamine and acetylcholine (ACh), as well as the amino acids, 7-aminobutyric acid (GABA), glycine, and glutamate/aspartate, are well established as retinal neurotransmitters (Ehinger and Dowling, 1987; Massey and Redburn, 1987). In addition, a number of neuroactive peptides have also been identified over the past years (Watt et al., 1985). The complex pattern of neuronal output to the brain is thus likely to result from the interaction between these numerous neurotransmitter systems. However, it is reasonable to assume that, in addition to the substances listed above, other neuroactive compounds are involved in the processing and/or modulation of visual information. More recently, an increasing emphasis has been given to the purine nucleoside, adenosine, as a possible neuroactive agent. Considerable evidence has accumulated indicating a neuromodulatory role for adenosine in the brain, complementary to its function as a common cellular metabolite (Phillis and Wu, 1981; Dunwiddie, 1985). A large proportion of adenosine is formed by the dephosphorylation of AMP by a 5'-nucleotidase which can be found in both neurons and glia (Kreutzberg et al., 1978; Fastbom et al., 1987a). However, the intracellular levels of adenosine in all cell types are apparently kept very low under basal conditions, due mainly to its rapid phosphorylation to adenine nucleotides and, to some extent, to its degradation to inosine by adenosine deaminase (Henderson, 1979). One factor found to stimulate the formation of adenosine is an increase in neuronal activity. The elevation is, in this case, likely to reflect mainly an acceleration of the hydrolysis of ATP which occurs in order to supply energy to the activated cells. The increased intracellular adenosine is released by the cells in amounts which are apparently sufficient to modify neuronal excitability, and this provides the framework for the hypothesis that adenosine has its most important role in neuroprotection. Consistent with this is the fact that the release of adenosine and its derivatives can be stimulated by depolarization as demonstrated for a number of nervous tissue preparations (Hollins and Stone, 1980; Jonzon and Fredholm, 1985). It is still unresolved whether ATP is the major purine released, then metabolized to adenosine, or whether adenosine is released as such. Both situations have been found in different preparations and thus one cannot generalize as to which is released. The major release mechanism requires calcium in most systems studied, but calcium-independent release has also been demonstrated (Bender et al., 1981; Wu et al., 1984; MacDonald and White, 1985).

Adenosine in the Retina

465

The effects elicited by adenosine occur via interaction with specific receptors whose activation results in responses as diverse as inhibition of transmitter release, depression of electrical activity, vasodilation, modulation of adenylate cyclase, and stimulation of glycogenolysis (Dunwiddie and Fredholm, 1984; Magistretti et aL, 1986). By the combined action of these effects, adenosine itself thus serves to reduce neuronal activity, and thereby energy expenditure, and promotes an increase in blood and glucose supplies, allowing the restoration of energy stores. Adenosine receptor antagonists, such as the methylxanthines, applied in the absence of exogenous adenosine, increase neuronal activity (Dunwiddie et al., 1981), indicating the presence of a continual purinergic tone. In addition, the biochemical as well as the electrophysiological effects of adenosine are potentiated by including adenosine uptake inhibitors (Motley and Collins, 1983), further indicating the involvement of extracellular sites. Two types of extracellular adenosine receptors have been characterized and defined as A1 (Ri) and A2 (Ra), originally based on their ability to inhibit or stimulate adenylate cyclase activity, respectively (van Calker et al., 1979; Londos et al., 1981). Recent studies have shown that the A~ receptor is coupled to a regulatory GTP-binding protein which can be associated with adenylate cyclase (inhibitory, Gi) but may also be linked to other effector systems such as ion channels (Michaelis et al., 1988; Schubert, 1988). The A1 receptor has a high affinity (Kd in the nanomolar range) for N6-substituted derivatives such as phenylisopropyladenosine (PIA) and cyclohexyladenosine (CHA) and exhibits a high stereospecificity for the isomers of PIA (R-PIA>>S-PIA). At the low-affinity A2 receptor (K~ in the micromolar range), 5'-Nethylcarboxamidoadenosine (NECA) is more potent than R-PIA, CHA, and 2-chloroadenosine (CADO), and a poor stereospecificity for the isomers of PIA is noted (Williams, 1987). An intracellular adenosine receptor (P-site) which is not sensitive to methylxanthines is also found t o b e negatively coupled to adenylate cyclase. However, the depressant actions of adenosine are consistent with activation of the extracellular sites. Both radioligand binding assays and receptor autoradiography have revealed the presence of adenosine receptors in distinct regions of the CNS (Fastbom et al., 1987b; Jarvis et al., 1989). The pattern of distribution of these sites is found to parallel, to a great extent, that determined from functional studies. Adenosine receptors are found both pre- and postsynaptically and may thus control neuronal activity at different levels. Regulation of cAMP formation, inhibition of neurotransmitter release, possibly through modulation of Ca 2+ and K + fluxes across the plasma membrane, and a direct effect on the excitability of the postsynaptic elements appear to be some of the mechanisms underlying the effects observed with adenosine. A particular localization of adenosine receptors on terminals of excitatory neurons has been demonstrated, which is consistent with a role for adenosine as a regulator of excitatory input via, for instance, modulation of transmitter release (Goodman et al., 1983; Geiger, 1986). A high-affinity, carrier-mediated transport system, thought to participate in synaptic clearance of adenosine, has been demonstrated in both neurons and glia

466

Blazynski and Perez

(Wu and Phillis, 1984; Geiger et al., 1988), and at least one class of transport site has been found to be heterogeneously distributed throughout the brain (Geiger and Nagy, 1984; Deckert et al., 1988). Recently, indicators of purinergic neurons have been reported such as the immunohistochemcial localization of adenosine and of the degradative enzyme, adenosine deaminase (ADA), with the latter showing a distribution that compares well with that of adenosine uptake sites (Nagy et al., 1984, 1985). In the retina, evidence consistent with a functional role of adenosine is accumulating. The aim of this review is to summarize these data and examine if a neurotransmitter/modulator role for adenosine is evident.

BIOCHEMICAL A N D PHARMACOLOGIC EVIDENCE Receptor Coupling Neurotransmitter-mediated modulation of second messenger systems is often used to identify the presence of receptors in tissues. Effects of adenosine on retinal adenylate cyclase are summarized in Table I and provide consistent evidence for the presence of both subtypes of receptors. In all species studied, adenosine- or adenosine analogue-mediated effects were significantly reduced or abolished by the inclusion of methylxanthines. For example, the addition of 0.5 m M I B M X to cultures of chick embryo retinas significantly shifted the dose-response curves for CADO and adenosine (Paes de Carvalho and de Mello, 1982). Table I. Retinal Preparation Chick embryo retina

Cultured chick embryo retina Rabbit retina Rabbit retina homogenates Mouse retinal homogenate Bovine retinal membrane

Adenosine Receptor Coupling to Adenylate Cyclase Ligand

Adenosine, R-PIA, S-PIA, CADO Adenosine, CADO CADO, CHA Adenosine, CADO Adenosine, CADO, R-PIA NECA, CHA, R-PIA, CADO NECA, CADO

Effective concentrations

Adenylate cyclase modulation

Reference a

100 nM-100/aM 10/aM-1 mM

Inhibition Stimulation

1

1 nM-10 mM 100 nM-1 mM

Inhibition Stimulation

2 3

>10/~M

Stimulation

4, 5

0.1 nM-5 nM >10 #M

Inhibition Stimulation

6 6

R-PIA, CHA

0.1-10 nM

Inhibition

7

CPA, R-PIA, NECA NECA, MECA b

0.1-10 nM > 10 nM

Inhibition Stimulation

7 7

Paes de Carvalho and de Mello (1985); (2) Paes de Carvalho (1990); (3) Paes de Carvalho and de Mello (1982); (4) Schorderet (1989); (5) Blazynski e t al. (1986); (6) Blazynski (1987); (7) Blazynski (unpublished observation). b 5'-N-Methylcarboxamidoadenosine.

a (l)

Adenosinein the Retina

467

Indirect evidence consistent with endogenous or tonic regulation of adenylate cyclase activity has been reported. In the presence of the phosphodiesterase (PDE) inhibitor R020-1724, A D A treatment of cultured chick embryonic retina reduced cAMP levels by over 80%, whereas exposure to dipyridamole, an inhibitor of adenosine uptake, increased the level of cAMP measured in the presence of R020-1724 by approximately 25% (Paes de Carvalho and de Mello, 1982). Also, when isolated mouse retinas were incubated in media containing ADA (1 U/ml), cAMP was significantly decreased (Blazynski, unpublished observation). A2-receptor-mediated increases in cAMP levels of incubated retinas have been detected in rabbit, but not in other mammalian species (Blazynski et aL, 1986). Large, nonphysiological concentrations of adenosine were required to observe this effect. However, when PDE was inhibited, lower doses of agonist significantly stimulated cAMP formation (Schorderet, 1989). A nonxanthine PDE inhibitor may be required when testing for adenosine receptor-mediated changes in retinal cAMP. Moreover, in light of the evidence for tonic enzyme modulation, it would also be recommended that retinas be pretreated with A D A to lower endogenous adenosine levels. This is supported by the report that treatment of cultured embryonic chick retinal cells with A D A increased the sensitivity of the response to CADO, as evidenced by a shift in the dose-response curve (Paes de Carvalho and de Mello, 1982). As listed in Table I, biphasic dose-response effects on adenylate cyclase in rabbit retinal homogenates illustrate the presence of both subtypes of extracellular adenosine receptors. These effects can be seen provided that homogenates are first pretreated with ADA, that adenylate cyclase is submaximally activated (forskolin, calmodulin), and that GTP concentrations are reduced to 5 ~M (Blazynski, 1987; unpublished observation). Increased concentrations of GTP decreased the ability of adenosine analogues to inhibit forskolin-activated adenylate cyclase in rabbit retinal homogenates. This is consistent with the known effects of GTP to decrease the affinity of the A1 receptor for its agonists (Goodman et al., 1982; Yeung and Green, 1983; Lohse et al., 1984; Cooper et al., 1985). In a recent report by Osborne (1989), adenosine, CADO, and NECA failed to modulate rabbit retinal adenylate cyclase. The assay medium lacked GTP, and effects on the unstimulated enzyme were studied, creating suboptimal assay conditions. As illustrated in Fig. 1, under the appropriate assay conditions, CHA inhibited forskolin-activated adenylate cyclase activity in mouse retinal homogenates. A2 receptor-mediated increases in cAMP, antagonized by methylxanthines, have been determined in cultured human RPE cells as well as in human RPE membrane preparations (Friedman et al., 1989). NECA stimulated the accumulation of cAMP in intact cultured RPE cells, with 100 nM NECA inducing a significant increase in cAMP levels, while 1/~M NECA was needed to detect a significant increase in human RPE membrane preparations. The maximal cAMP increase was 16-fold at 10 ~tM NECA. PIA and cyclopentyladenosine (CPA) also elicited increases in cAMP at concentrations higher than those needed to observe the NECA-mediated stimulation, thus confirming the presence of true A2

468

Blazynski and Perez

100

T /

80

/

JJ / / j/"

.~ 60
R-PIA > NECA > IBMX, indicative of A1 receptor binding. Using the Arselective agonist [3H]PIA, characterization of receptors in bovine retinal membranes revealed saturable, high- and low-affinity binding sites (Ko of 0.134 and 22 nM, respectively) with relatively low densities (Woods and Blazynski, 1991). Total binding was reduced by incubation with Gpp(NH)p.

469

Adenosine in the Retina P H Y S I O L O G I C A L EVIDENCE Endogenous and Accumulated Purines

Endogenous purines in retina consist primarily of ATP (70%), with a very minor fraction identified as the nucleoside, adenosine (2%) (Perez et aL, 1986, 1988). When rabbits were sacrificed at 4 hr following an intraocular injection of [3H]adenosine, over 85% of the radioactivity was associated with nucleotides, and only 1% with adenosine; the ratios were not significantly different from that observed for endogenous purines. The low intracellular level of adenosine is consistent with what is found in most cells, where purine transport is followed rapidly by phosphorylation, thus maintaining low adenosine levels. Some qualitatively different results have been found for cultures of chick embryo retinal neurons (Paes de Carvalho et al., 1990). Accumulated radioactivity (after 15 min of incubation) coeluted with inosine (51%), adenosine (32%), and hypoxanthine, uric acid, and nucleotides (10%). One possible explanation for this difference is that in these cells, the activity of adenosine deaminase is greater than that of adenosine kinase. However, while this has been found to be the case in some tissues, t h e information available for CNS preparations indicates that phosphorylation is the preferred intracellular metabolic route. Besides, inosine as well as its metabolites is known to diffuse readily out of cells, so that it is unexpected to see such high levels intracellularly. Also, unless precautions were taken to prevent the breakdown of nucleotides to form adenosine and then inosine (e.g., by chilling the preparation prior to extraction), the metabolites found in the cultured cells may derive from hydrolysed ATP. Purine nucleoside transporters are reported to be rather non-selective and readily transport inosine or uridine (Jarvis, 1988). This does not appear to be the case for the cultured chick embryonic retinal cultures since treatment with A D A in the presence of [3H]adenosine reduced the accumulated radioactivity by 90%, and the addition of deoxycoformycin, an A D A inhibitor, reversed this effect. The uptake of adenosine in the cultured cells was temperature sensitive, blocked by the transport inhibitors, dipyridamole and nitrobenzylthioinosine, and insensitive to sodium substitution. However, in the rat retina, nucleoside uptake was found to be strongly sodium dependent (Shaeffer and Anderson, 1981). Thus, differences in the biochemical characteristics of the transporters and subsequent metabolic reactions involving adenosine may be species related or due to a lack of glial cells in the neuronal cultures. Glia are known to transport and release adenosine in the brain, and in one report, retinal glia has been demonstrated to accumulate adenosine autoradiographically (Ehinger and Perez, 1984). In a sense, what the aforementioned results tell us is that retinal cells contain detectable amounts of purines, which, at least in the rabbit retina, appear mainly in the form of adenine nucleotides, and that this pool can be labeled by exogenously supplied [3H]adenosine. Considering the fact that all energyrequiring processes in cells utilize ATP, it is not surprising that cells contain the nucleotide in reasonable amounts under normal conditions and that an enzymatic apparatus is present to maintain this. The fact that retinal cells take up adenosine

470

Blazynski and Perez

could simply reflect as well the presence of a mechanism to supply cells with substrate. However, in view of the known extracellular effects of adenosine, it may also be assumed that retinal cells take up adenosine as a means of removing it from the extracellual space. Tonic and K+-Evoked Purine Release

The composition of the material tonically released by superfused rabbit retinas and by cultured embryonic chick retinal neurons also reveals some differences. In the samples obtained from cultures incubated with [3H]adenosine, most of the radioactivity comigrated with inosine (62%), and less with adenosine (10%) (Paes de Carvalho et al., 1990). On the other hand, in the superfusates from rabbit retina prelabeled with [3H]adenosine, radiolabel coeluted primarily with hypoxanthine and xanthine (62%), and much less with adenosine (3%) and inosine (6%) (Perez et al., 1986). The composition of endogenous purines tonically released from the rabbit was the same as that of radioactive compounds. It is not possible to establish whether the differences found between preparations could be the result of distinct sampling conditions or if they reflect differences in the activity of the extracellular enzymes in the two preparations. Nevertheless, it should be noted that, in both cases, the superfusates consisted mainly of products of adenosine breakdown, as is found for most nervous tissue preparations. The depolarization of the rabbit retina in vitro and of cultured chick embryonic retinal neurons has been found to induce a rapid and reversible increase in the outflow of purines (Perez et al., 1986, 1988; Paes de Carvalho et al., 1990). For the chick, 50 mM K + increased the absolute level of radioactivity released by three-fold (Paes de Carvalho et al., 1990), whereas for the rabbit retina, depolarization with 23 or 43 mM K ÷ resulted in an 80% increase in radioactivity released. It should be noted that depolarization did not alter the proportions of purine metabolites released as compared to tonic release (Perez et al., 1988). This suggests that upon depolarization, the release rate is simply increased. Whereas K ÷ stimulation increased the release of radioactivity by 80% compared to basal, only a 20% increase in release of endogenous purines occurs (Perez et al., 1988). It can thus be hypothesized that the radioactive purines are more readily mobilized, indicating different purine compartments. In an attempt to discern whether any released purines originated as ATP, a~,fl-methylene ADP plus GMP was added to the culture or superfusion medium to inhibit 5'-nucleotidase. For both preparations, no shift in the composition of tonically released purines was seen. This was also true for stimulated release from chick cultures, but stimulated release from superfused rabbit retina was altered by this treatment, indicating that some of the released purines were in the form of adenine nucleotides. A similar strategy was used to test if A D A compromised the detection of purines, using the A D A inhibitor, EHNA. In both chick cultures and rabbit retina, increases in radioactivity associated with adenosine were detected. While it appears that stimulation-evoked ATP release does not occur from the cultured chick preparation, at least a small part of the release from rabbit retina originates as ATP.

Adenosine in the Retina

471

The K+-evoked increase in purine release in both preparations was blocked when Ca 2+ was removed from the medium. However, when rabbit retinas were superfused with 43mM K + (as opposed to 13.6 or 23.6mM), some release persisted, suggesting that under these strongly depolarizing conditions, release occurs in a nonvesicular manner. This calcium-independent release does not appear to be due to cell death or lysis, since baseline levels of release are seen after repeated stimulations (Perez et al., 1988). Yet in chick cultures, the release evoked by 5 0 m M K + was dependent on extracellular calcium. Again, this difference may be explained by the fact that glial cells are absent from the culture preparation. However, it must also be considered that upon depolarization, the resultant increased intracellular sodium may drive a bidirectional facilitative transporter (LeHir and Dubach, 1984, 1985a; 1985b; Bender and Hertz, 1986; Jarvis, 1988). Dipyridamole, a transport inhibitor, partially antagonized the K+-evoked release, consistent with a role for the transporter. Light-Evoked Release of Purines While continuous illumination had no effect on radioactive efflux, 50-msec flashes (2 Hz) resulted in a slow, persistent increase in the release of purines from isolated rabbit retinas (Perez et al., 1988). The preparation consisted of an everted eyecup so that photopigment regeneration could occur. It is interesting, then, that only flashing-light stimulation resulted in a slow, delayed increase in radioactivity. It could be suggested that the increase may be mediated by cells responding to light off, or receiving input from off-cells, but this remains to be demonstrated. Transmitter-Modulated Purine Release The excitatory amino acids glutamate and asparate, as well as the more selective agonists kainic acid (KA), quisqualic acid (QQ), and N-methyl-D,Lasparate (NMDA), evoked increases in radioactive overflow associated with purine metabolites (Perez and Ehinger, 1989a). KA, QQ, and N M D A evoked much larger increases in radioactivity released compared to the naturally occurring amino acids. At 25/~M, only the response to KA was completely blocked by removal of calcium. AP-5 (2-amino-5-phosphonovaleric acid) and AP-7 (2-amino-5-phosphonoheptanoic acid), NMDA receptor selective antagonists, were quite effective at inhibiting the release of purine-associated radioactivity. Carbachol, GABA, dopamine, glycine, and serotonin all stimulated purine release at high concentrations; specific antagonists inhibited these responses. Thus, it appears that purines can be released by activation of specific receptors. The chain of events initiated in a cell by the activation of receptors is likely to require energy, which is obtained by ATP hydrolysis and formation of purine metabolites and their subsequent release. Further, in view of the data obtained with calcium-free and sodium-free medium (which abolished the NMDA-evoked release of purines), it seems clear that part of the purine release is associated with a nonvesicular mechanism, most probably via the purine

472

Blazynski and Perez

nucleoside transporter. High exogenous concentrations of excitatory amino acid agonists cause massive depolarizations (either directly by increasing Na ÷ conductance or by increasing intracellular Na + as a result of transmitter cotransport), which ultimately results in increased ATP hydrolysis by the Na ÷/K ÷ ATPase. Evidence that adenosine derivatives are continuously released by the rabbit retina has been presented (Perez and Ehinger, 1989a). An increase in the outflow of purines was seen upon return to control medium after exposure to the NMDA receptor antagonists AP-5 and AP-7, suggesting that under normal conditions, the levels of excitatory amino acids in the extracellular space slightly modulate purine release. Antagonist removal, by allowing depolarization of postsynaptic cells, increases ATP metabolism and purine release. The inhibitory neurotransmitter antagonists (e.g., bicuculline, strychnine, and haloperidol) evoked Ca 2÷independent increases in purine release. Furthermore, the bicuculline-evoked release was abolished by AP-7, consistent with the GABA antagonist's action to disinhibit glutamate containing neurons. A problem with the latter data is that GABA also increased purine efflux. One would have to assume that the GABA-mediated release of purines is due to its specific action on some other inhibitory transmitter system in the retina.

Modulation of Acetylcholine (ACh) Release In the rabbit retina, the A1 receptor agonists R-PIA, CHA, and CPA as well as the mixed A 1 - A 2 agonist, NECA, significantly attenuated the K+-evoked release of [3H]ACh. Specific antagonists blocked the action of these compounds, illustrating extracellular receptor specificity (Perez and Ehinger, 1989b). More recently, R-PIA was demonstrated to be more potent than NECA in reducing the release of ACh, and S-PIA was shown to be inactive, consistent with the involvement of A1 receptors (Perez and Ehinger, submitted for publication). A1 receptors have been localized to the inner retina in all species studied (Braas et al., 1988; Blazynski, 1990). Displaced amacrine cells in the rabbit retina, which express adenosine-like immunoreactivity, are also labeled by the cholinergic marker, DAPI (Blazynski, 1989a, b). Conceivably, adenosine and ACh are coreleased, and adenosine can further modulate ACh release via activation of A1 receptors, thus controlling input to ganglion cells. Adenosine and a specific agonist were recently shown to decrease significantly the light-evoked optic nerve response in the superfused cat eye (Niemeyer and Friih, 1989; Blazynski et al., 1989a). The activation of A1 receptors may thus directly reduce the responsiveness of ganglion cells and/or, as described, decrease the release of transmitters, such as ACh.

Modulation of Glycogen Hydrolysis Adenosine has been observed to elicit an increase in glycogen hydrolysis in incubated rabbit retinal slices; concentration-dependent increases in glycogen breakdown were observed, with an ECs0 of 15/~M adenosine (Osborne, 1989). Both theophylline and 8-phenyltheophylline (10/tM) completely antagonized the agonist-induced increase in glycogenolysis, indicating the presence of specific

473

Adenosinein the Retina

extracellular receptors. The rank order of potency for stimulation of glycogen hydrolysis was reported as CADO, adenosine >S - PIA > R-PIA > NECA > CHA, different from that reported for either the A1 or the A2 receptors determined from pharmacological, biochemical, or physiological studies (see Introduction), and may correspond to a unique type of adenosine receptor. Electrophysiology

Evidence that adenosine receptors influence electrical activity in the normal retina has been provided recently by Niemeyer and colleagues (Niemeyer and Friih, 1989; Blazynski et al., 1989a) in studies employing the isolated, arterially perfused cat retina. Adenosine (30/~M) induced a reversible rise of the standing potential within less than 2 min, followed by a marked depression of the light peak (evoked by a 60-sec pulse of diffuse white light). The c-wave of the electroretinogram (ERG) was also increased. These effects indicate the involvement of receptors localized to the retinal pigmented epithelium (RPE). A 2 receptors on RPE have been demonstrated biochemically (Friedman et al., 1989) and autoradiographically (Blazynski, 1989d). Thus, it is possible that adenosine, acting to stimulate adenylate cyclase, influences the light peak generated at the basal membrane of the RPE. In an earlier study, high concentrations of theophylline, which would be expected to result in phosphodiesterase inhibition and elevation of cAMP levels, abolished the light peak (Dawis et al., 1987). The entire role of adenosine in the RPE remains to be elucidated considering that its function may not be restricted to modulation of light responses. As proposed by Friedman et al. (1989), other cAMP-dependent processes such as phagocytosis and outer segment digestion, metabolite transport between RPE and choroid, and RPE cell movement and chemotaxis might also be mediated by adenosine receptors. The ERG b-wave of dark-adapted cat retinas, stimulated with a 620-nm pulse of red light (matched with a 400-nm pulse for the rod system), was significantly increased by adenosine in a concentration-dependent manner (Blazynski et al., 1989a; Niemeyer and Frtih, 1989). It was suggested that the increase in b-wave amplitude resulted from an elevation of glucose levels induced by adenosine (Osborne, 1989). In addition, adenosine (2.5/~M), markedly depressed the ON-, plateau-, and OFF-components of the optic nerve response. This effect was antagonized by micromolar concentrations of IBMX (G. Niemeyer, personal communication). Thus, it appears that adenosine receptors also modulate input onto ganglion cells. Unlike what is seen in the RPE, this effect of adenosine may be mediated via AI receptors. This class of receptors has been localized to the inner plexiform layer (Braas et al., 1987; Blazynski, 1990). By decreasing cAMP levels and/or by modulating Ca 2+ and K + fluxes, adenosine may reduce excitability of ganglion cells and may also decrease excitatory transmitter release. Adenosine has also been shown to modulate the gating of ion channels in the inner segments of some tiger salamander cone photoreceptors (Barnes and Hille, 1989). The voltage-dependent Ca 2+ current (Ica) and a calcium-activated C1current (Ia~ca~) of enzymatically isolated cone photoreceptors were depressed by

Blazynskiand Perez

474

high levels of adenosine. Thus at high concentrations, exogenous adenosine can regulate conductances in photoreceptors, perhaps participating in the modulation of calcium dependent processes such as transmitter release and cyclase activity.

MARKERS OF PURINERGIC TRANSMISSION

The localization of uptake of specific neurotransmitter radioligands into defined neurons, as well as a cytochemical demonstration or localization of immunoreactivity to a neurotransmitter or an enzyme involved in its synthesis or degradation, has been cited in many systems as evidence for the presence of a particular neurotransmitter system. Such results have been reported for putative markers for purinergic neurons in retinas obtained from a variety of species. Uptake

Autoradiographic localization of the uptake of [3H]-adenosine has been studied in retinas from cat (Blazynski et al., 1989b), rabbit (Ehinger and Perez, 1984; Blazynski, 1987; Blazynski et al., 1989b), mouse (Blazynski et al., 1989b), ground squirrel (Blazynski, et al., 1989b), goldfish and carp (Ehinger and Perez, 1984), and chicken (Perez and Bruun, 1987). For all species, as illustrated in Figure 2, the distribution of cells that accumulate [3H]-adenosine is quite similar. Uptake into cell bodies localized in the ganglion cell layer (GCL) and the inner nuclear layer (INL) is apparent across species. In the GCL, the labeled cells appeared to have the highest density of silver grains associated with the nucleus, while in the INL the cells tended to have very little cytoplasm and were more uniformly labeled. In rabbit retina, using a dry fixation technique, silver grains (corresponding to uptake of radioligand) were detected at the external limiting membrane as well as the proximal portion of the GCL (Ehinger and Perez, 1984). In addition, a concentration of radially oriented silver grains were observed running through the retina. This is consistent with the known morphology of Miiller cells, a type of glial cell. Surprisingly, this pattern of labeling was not observed when tissue was immersion fixed. The autoradiographic localization of uptake of the ADA-resistant adenosine analogues [3H]CHA and [3H]PIA has also been determined for rabbit (Ehinger and Perez, 1984; Perez and Ehinger, 1986; Blazynski, 1987; Blazynski et al., 1989b) and carp (Ehinger and Perez, 1984) retinas. In the rabbit, these radioligands labeled populations of cell bodies similar to those labeled by [3H]adenosine. In both goldfish and carp retinas (Ehinger and Perez, 1984), the former two ligands did not specifically label any population of retinal cells. Adenosine-like Immunoreactivity Although adenosine is a ubiquitous molecule, it has proved possible to localize cells and/or their processes that apparently store high levels of adenosine. Using an antiserum generated against a derivative of adenosine [laevulinic acid, modified on the ribose moiety (Newby and Sala, 1982)], but highly specific for

Adenosine in the Retina

475

adenosine, discrete populations of neurons in both brain (Braas et al., 1986) and retina (Braas et al., 1987; Blazynski et al., 1989b) have been described. Retinas from several mammalian species have been examined, as illustrated in Fig. 2. For all species examined, strong labeling of cell bodies localized in the GCL and INL have been observed. Braas et al. (1987), using a glutaraldehyde fixation, found substantial labeling of the inner plexiform layer (IPL) of mammalian retinas. Adenosine-like immunoreactivity was observed in the outer retina of monkey and human eyes, but not in any other species. Blazynski et al. (1989b), found that when eyecups were fixed by immersion in 0.1% glutaraldehyde and 4% paraformaldehyde, this intense pattern of label in the IPL of rabbit, mouse, and ground squirrel retinas was observed only with less dilute antibody concentrations. Pretreatment of tissue sections with A D A reduces the extent of labeling by the antisera (Braas et al., 1987; Blazynski et al., 1989b), indicating that stores of adenosine are indeed labeled. However, most accumulated adenosine assayed in whole retina is metabolized to ATP (Perez et al., 1986) and at least part of the K+-evoked release consists of ATP (Perez et al., 1986). Thus the label corresponding to adenosine-immunoreactive cells is somewhat perplexing unless one assumes that these neurons contain purines in a different proportion than that found in homogenates from whole retina. In general, the pattern of labeling of both adenosine-like immunoreactivity and adenosine uptake is very similar, consistent with a role for adenosine transport to clear stimulation- or basally released purine nucleosides. However, it must be pointed out that cultures of embryonic chick retinal cells reveal colocalization of [3H]adenosine uptake and adenosine-like immunoreactivity in all photoreceptors, but in only 25% of other neurons (Paes de Carvalho et al., 1990). Uptake into posthatched chick photoreceptors has been reported (Perez and Bruun, 1987). Localization of Enzymes Associated with Adenosine Metabolism The distribution of reaction product formed by the enzyme 5'nucleotidase, which hydrolyzes high-energy phosphates from nucleotides, has been described (Scott, 1967; Kreutzberg and Hussain, 1984). Histochemical staining was present throughout the retina, but graded in intensity over various retinal layers. The heaviest label was observed over the outer segment layer, and decreased from layer to layer toward the inner retina. However, the activity appeared to demarcate the boundaries between retinal layers. Immunoreactivity toward the enzyme adenosine deaminase (ADA), which deaminates adenosine to form inosine, has been localized in discrete populations of cells and processes in both brain and retina (Senba et al., 1986). As demonstrated in Fig. 2, ADA-like immunoreactivity in rat retina has a distribution that is virtually identical to that seen for adenosine-like immunoreactivity. Cell bodies in both the GCL and the INL are labeled, as well as processes that ramify in particular sublamina in the IPL. The density of cell bodies in the GCL was not altered following optic nerve transection, nor did the dye fast blue retrogradely label any A D A positive cell bodies. These results are consistent with the identification of these cells as predominantly amacrine cells. Yet intraocular injections of the nuclear dye 4,6-diamino-2-phenylindole [DAPI; known to label

Blazynski and Perez

4'76

,,,0

~,~

"~ ~

~

o

~N •

U

o •~

"~ .. = ~ ' 6 ,~"~ = ~ ~.-

~ = ,.-,

Z o.

Q 3

K

LT! 11h

I I

'~

~'~" ~ '~ -*

~

= ~

~

~ ~--., ~ ~ ~ : ~

so ~ 0 z

Im

i

T

I

ii

~

~

~I

~

©

,~

O N ',~ ~.~ ~--~

~ ~

O ," . ~ 0 "~~

~

~ ,~

L "

~ ~.o • ~ ~_A.

6

~ ,-~

~ ..,.0., = ~ ~.~

Adenosine in the Retina

477

cholinergic amacrine cells in rabbit retina as reported by Tauchi and Masland (1984)] failed to label the ADA-positive cells in the rat retina. However, a systematic study of the uptake of DAPI by particular neurons in the rat retina has not been reported; thus this result may not eliminate the possibility that the cells labeled by adenosine deaminase-like immunoreactivity are displaced cholinergic cells.

Receptor Localization Localization of adenosine A1 receptors has now been reported for a number of mammalian retinas, as illustrated in Fig. 3. For human, monkey, and rat retinas, [3H]PIA was used to label receptor binding sites; specific binding was displaced with an excess of cold R-PIA (Braas et al., 1987). For rabbit and mouse eyecups, tissue sections were labeled with both [3H]PIA and [3H]CHA (Blazynski, 1989a; Blazynski et al., 1989b). With the exception of the monkey retina, the highest density of binding sites was observed over the IPL. For the monkey, specific label was detected over the entire inner retina. All species exhibited a low but significant degree of specific binding in the outer retina, with the exception of the rat retina. Thus, in general among the mammalian species, there was some agreement on the distribution of A1 receptor binding sites. Binding sites for the mixed A1-A2 agonist, [3H]NECA, have also been localized in rabbit and mouse eyecups (Fig. 3). The distribution of these binding sites is markedly different from that observed using highly specific m I receptor agonists. Whereas a sparse distribution of silver grains was observed over the inner retina, a high density of silver grains was localized over the sclera and choroid and the RPE, as well as the outer and inner segment layers of photoreceptors. The binding associated with the scleral-choroid border was nonspecific, while specific binding was observed in both the inner and the outer retina, with the highest degree of specific binding determined in the layers comprising the RPE and OS-IS. IBMX (3-isobutyl-5-methylxanthine; 100 gtM) or CHA (100/~M) displaced [3H]NECA binding in the inner retina as well as the outer retina. However, neither the antagonist nor a large excess of the highly selective A1 agonist displaced [3H]NECA binding to the extent of that seen using cold NECA, and thus this residual binding probably comprises non A~-Aa receptor binding. Since NECA is known to bind at both A1 and A2 receptors, tissue incubations were performed with a low concentration of the A~ specific agonist R-PIA (10 nM). Under these conditions, the distribution of binding sites for [3H]NECA in the outer retina remained unaltered for both rabbit and mouse eyecups, while virtually no binding was observed over the inner retina. Tissue sections were also treated with N-ethylmaleimide to uncouple A1 receptors from Gi and, thereby, reduce agonist affinity. This resulted in a loss of binding only in the inner retina. Thus, the binding in the inner retina corresponded to ligand binding to A 1 receptors, while the binding in the outer retina is A2 receptor specific. The localization of A2 receptors in the outer retina is problematical, since

478

Blazynski and Perez

A 1 RECEPTOR BINDING

PL

(a)

"

~

Human

Monkey

Rabbit

=

_

Rat

'~

Mouse

[3H]-NECA BINDING

NL

PL

LL

4_ (b~

oc Rabbit

Mouse

Fig. 3. Distribution of binding sites for Al-selective and mixed A1-A 2 radioligands. (a) A 1 receptor binding has the highest density localized over the inner plexiform and ganglion-ceU layers (Braas et al., 1987; Blazynski, 1990). (b) Binding of the A1-A 2 ligand [3H]NECA has the highest density over the outer retina. Sparse binding in other layers was displaced by including 10 nM PIA (Blazynski, 1990).

there appears to be no available source of adenosine in the region. However, the cytochemical localization of 5'-nucleotidase activity to the synapses between photoreceptors and horizontal cells (Kreutzberg and Hussain, 1984; Scott, 1967) as well as to the cytoplasmic surface of outer segments provides a mechanism for adenosine formation from nucleotides. It is notable that adenosine has been shown to inhibit rhodopsin phosphorylation (Donner and Hemila, 1985) by acting as a competitive inhibitor of rhodopsin kinase [K; of 4/~M (Palczewski, 1985)].

Adenosine in the Retina

479

Colocalization of Purinergie Markers with Other Transmitter Markers A comparison of cells that accumulate [3H]adenosine and are labeled for 7-aminobutyric (GABA)-like immunoreactivity has been reported for rabbit and chicken retinas (Perez and Bruun, 1987). For both species, colocalization of the two markers was observed in a subset of conventionally placed amacrine cells (within the INL) and in certain cell bodies in the GCL; however, the extent of colocalization varied between the two species. For example, in the chicken retina, in the proximal portion of the INL (nearest the IPL) approximately 60% of the cells that were labeled by either of the two markers were double-labeled. In the rabbit retina, on the other hand, only 15% of cells labeled by either marker were double labeled. GABA-Iike immunoreactivity and [3H]adenosine uptake colocalized in only 10% of the cell bodies in the GCL in the rabbit and 35% in the chicken. For the rabbit retina, it has now been demonstrated (Blazynski, 1989b) that approximately 50% of the cell bodies that are labeled by adenosine-like immunoreactivity in the GCL also accumulate the nuclear dye 4,6-diamino-2phenylindole (DAPI). At 16 weeks following unilateral optic nerve transection, when most ganglion cells have been eliminated, the number of cell bodies in this layer that are labeled for both adenosine-like immunoreactivity and DAPI accumulation is identical to that in the control eye. In rabbit retina, choline acetyltransferase-immunoreactive neurons are located along both borders of the IPL, with two bands of immunoreactive processes present in the IPL. Uptake of [3H]choline into retinal neurons labels the same population of cell bodies and processes as does DAPI (Tauchi and Masland, 1984; Masland et al., 1984). Thus, the cell bodies labeled for DAPI in the GCL are displaced cholinergic amacrine cells, and the cells labeled for both adenosine-like immunoreactivity and DAPI accumulation are displaced cholinergic amacrine cells. It has also been reported that these cells are additionally labeled for immunoreactivity to the enzyme glutamic acid decarboxylase, a marker for GABAergic neurotransmission (Blazynski, 1989b), in agreement with the colocalization of GABA-like immunoreactivity and adenosine uptake (Perez and Bruun, 1987). Mapping of A1 receptors to the IPL of mammalian retina is consistent with the known morphology of processes of amacrine cells as well as ganglion cells. Present in this same region is immunoreactivity to adenosine deaminase (Fig. 2). For rabbit retina, colocalization of adenosine in cholinergic, GABAergic amacrine cells has been demonstrated (Perez and Bruun, 1987; Blazynski, 1989b), and the stimulation-evoked release of adenosine (Perez et al., 1988) can be hypothesized to act presynaptically to inhibit further release of acetylcholine and/or GABA. As presented earlier, adenosine does indeed inhibit K+-evoked release of acetylcholine. CONCLUSIONS Biochemical, pharmacological, and anatomical studies have provided convincing evidence for the presence of adenosine receptors, both A1 and A2, in the

480

Blazynski and Perez

retina. An uptake system for adenosine has also been demonstrated, which transports and retains purines within distinct populations of cells in the inner retina, providing a mechanism for transmitter inactivation, as does also the immunohistochemical evidence for the presence of A D A in the IPL (Senba et al., 1986). Purine release experiments demonstrate that adenosine and ATP are released from the rabbit and chick retina and that A D A is active in the retina (Paes de Carvalho et al., 1990; Perez et al., 1988), indicating that adenosine is present in the extracellular space. In the rabbit retina, part of the transmitterevoked release did not occur in a Ca2+-dependent manner and could be inhibited by the bidirectional transport inhibitor, dipyridamole, indicating that release may occur via a vesicular mechanism and via the transporter. The localization of adenosine-like immunoreactive neurons to the same populations of cells that are labeled for [3H]adenosine uptake is disturbing since it has been demonstrated that in extracts from whole retina very little accumulated purine is in the form of adenosine. It may be the case that the purine composition of a small subset of cells is different; inhibition of A D A with E H N A is consistent with these hypotheses, as perhaps does the fact that the increased purine release is composed primarily of the more recently accumulated tritiated purines. For adenosine to play a neuromodulatory or transmitter role in retinal processing, one would expect that its levels in superfusates would change as a function of light stimulation. A slow persistent release was seen only after the end of a 10-min flashing stimulation. Thus, an immediate light-evoked release of purines was not observed. An important criterion in establishing identity as a transmitter/modulator is thus missing. What evidence does, then, exist that endogenous adenosine may influence retinal physiology? At this point, very little. What is known is that agonists exert effects on electrical responses in RPE, inner retina, and ganglion cells and that the K÷-evoked release of [3H]ACh from the rabbit retina is inhibited via the activation of A1 receptors. This is strongly suggestive of a physiological role. Relatively high concentrations of agonists were required to elicit electrophysiological effects, and perhaps pretreatment with A D A might have increased the sensitivity of the response. The colocalization of adenosine-like immunoreactivity with cholinergic and GABAergic markers in displaced amacrine cells in the rabbit retina is also suggestive of some functional role but, by no means, provides definitive data. Due to the localization of A1 receptors to the IPL, interactions with other transmitter systems should also be examined. What remains to be demonstrated is that A 1 agonists inhibit the light-evoked release of ACh, while antagonists increase its release. What is evident from experiments studying the transmitter- and K÷-evoked release of purines is that, under strongly depolarizing conditions, purine release is augmented. The evoked release is dependent partly on extracellular calcium but a portion of this release is apparently mediated via the purine transporter. The action of excitatory amino acid agonists and most antagonists would be to depolarize cells, either directly or by some secondary mechanism such as inhibition of G A B A input. The resulting sodium influx and subsequent reactions initiated intracellularly, accompanied by increased Na÷/K+-ATPase activity to

Adenosine in the Retina

481

restore the sodium gradient, will ultimately increase intracellular levels of purine metabolites, which can then be transported from the cell. Thus there is implied a relationship between purine release and neurotransmission, although it does not reflect release of a transmitter substance as such. The purinergic modulation of neuronal activity will be determined by the location and activation of specific adenosine receptors. Levels of adenosine can reach high concentrations extracellularly; adenosine can then act at receptors to initiate responses that ultimately reduce the energy demands of cells, allowing a restoration of energy stores (Dragunow and Faull, 1988). Reduction of cellular activity (e.g., by an As-mediated decrease in ACh release) and increases in blood flow and glucose supplies are some of the global effects ascribed to adenosine in neuroprotection. This may indeed be one role that adenosine plays in the retina under extreme conditions. Finally, the observation of A2 receptors in the outer retina and RPE is quite perplexing since neither uptake nor stores of adenosine are detected in this region. It is conceivable that these receptors are present to participate in a protective response. Thus, adenosine released from the inner retina or choroid in pathological states, or after an ischemic episode, may bind and activate receptors leading to increases in cAMP. The increased intracellular cAMP levels in RPE and photoreceptors resulting from adenosine binding to the A2 receptors may be involved in metabolic regulation in these cells. Alternative roles for these receptors remain to be demonstrated.

ACKNOWLEDGMENTS

The authors wish to thank especially Drs. R. Paes de Carvalho, G. Niemeyer, and K. Palczewski for their generosity in providing unpublished data. A special acknowledgment is also given to Drs. S. Barnes and N. Osborne for providing the authors with manuscripts prior to publication. The authors also acknowledge support from the National Eye Institute (EY02294, EY00258), a grant from the Lucille P. Markey Trust (C.B.), and grants from the Crown Princess Margareta's Committee for the Blind, the Helfrid Lorentz Nilsson Foundation, the Swedish Society for Medical Research, the Retinitis Pigmentosa Foundation, the Faculty of Medicine at the University of Lund, and the Swedish Medical Research Council (Project 14X-2321; M.-T.R.P.)

REFERENCES

Barnes, S., and Hille, B. (1989). Ionic channels of the inner segment of tiger salamander cone photoreceptors. J. Gen. Physiol. 94:719-743. Bender, A. S., and Hertz, L. (1986). Similarities of adenosine uptake systems in astrocytes and neurons in primarycultures. Neurochem. Res. 11:1507-1524. Bender, A. S., Wu, P. H., and Phillis, J. W. (1981). The rapid uptake and release of [3H]adenosine by rat cerebral cortical synaptosomes.J. Neurochem. 36:651-660.

482

Blazynski and Perez

Blazynski, C. (1987). Adenosine A 1 receptor-mediated inhibition of adenylate cyclase in rabbit retina. J. Neurosci. 7:2522-2528. Blazynski, C. (1989a). Identification and localization of adenosine receptors and adenosinergic neurons in mammalian retinas. J. Neurochem. 52:S148-$148 (abstr.). Blazynski, C. (1989b). Displaced cholinergic, GABAergic amacrine cells in the rabbit retina also contain adenosine. Vis. Neurosci. 3:425-431. Blazynski, C. (1990). Discrete distributions of adenosine receptors in mammalian retina. J. Neurochem. 54:648-655. Blazynski, C., Kinscherf, D. A., Geary, K. M., and Ferrendelli, J. A. (1986). Adenosine-mediated regulation of cyclic AMP levels in isolated incubated retinas. Brain Res. 366:224-229. Blazynski, C., Cohen, A. I., Friih, B., and Niemeyer, G. (1989a). Adenosine: autoradiographic localisation and electrophysiological effects in the cat retina. Invest. Ophthalmol. Vis. Sci. 30:2533-2536. Blazynski, C., Mosinger, J. L., and Cohen, A. I. (1989b). Comparison of adenosine uptake and endogenous adenosine-containing cells in mammalian retina. Vis. Neurosci. 2:109-116. Braas, K. M., Newby, A. C., Wilson, V. S., and Snyder, S. H. (1986). Adenosine-containingneurons in the brain localized by immunocytochemistry. J. Neurosci. 6:1952-1961. Braas, K. M., Zarbin, M. A., and Snyder, S. H. (1987). Endogenous adenosine and adenosine receptors localized to ganglion cells of the retina. Proc. Natl. Acad. Sci. USA 84:3906-3910. Campochiaro, P., Ferkany, J. W., and Coyle, J. T. (1985). Excitatory amino acid analogs evoke release of endogenous amino acids and acetyl choline from chick retina in vitro. Vis. Res. 25:1375-1386. Cooper, D. M. F., Young, S. M. H., Perez-Reyes, E., Owens, J. R., Fossom, L. H., and Gill, D. L. (1985). Properties required of a functional Ni, the GTP regulatory complex that mediates the inhibitory actions of neurotransmitters on adenylate cyclase. Adv. Cyclic Nucleotide Prot. Phos. Res. 9:75-86. Dawis, S., and Niemeyer, G. (1987). Theophylline abolishes the light peak in perfused cat eyes. Invest. Ophthalmol. Vis. Sci. 28:700-706. Deckert, J., Bisserbe, J.-C., Klein, E., and Marangos, P. J. (1988). Adenosine uptake sites in brain: Regional distribution of putative subtypes in relationship to adenosine A1 receptors. J. Neurosci. 8:2238-2249. deMello, M. C. F., Ventura, A. L. M., Paes de Carvelho, R., Klein, W. L., and deMello, F. G. (1982). Regulation of dopamine- and adenosine-dependent adenylate cyclase systems of chicken embryo retina cells in culture. Proc. Natl. Acad. Sci. USA 79:5708-5712. Donner, K., and Hemila, S. (1985). Rhodopsin phosphorylation inhibited by adenosine in frog rods: Lack of effects on excitation. Comp. Biochem. Physiol. 81A:431-439. Dragunow, M., and Faull, R. L. M. (1988). Neuroprotective effects of adenosine. Trends Pharmacol. Sci. 9:193-194. Dunwiddie, T. V. (1985). The physiological role of adenosine in the central nervous system. In International Review o f Neurobiology (J. R. Symthies, and R. J. Bradley, Eds.), Academic Press, New York, pp. 63-139. Dunwiddie, T. V., and Fredholm, B. B. (1984). Adenosine receptors mediating inhibitory electrophysiological responses in rat hippocampus are different from receptors mediating cyclic AMP accumulation. Naunyn-Schmiedeberg Arch. Pharmacol. 326:294-301. Dunwiddie, T. V., Hoffer, B. J. and Fredholm, B. B. (1981). Alkylxanthines elevate hippocampal excitability. Evidence for a role of endogenous adenosine. Naunyn-Schrniedeberg Arch. Pharmacol. 316:326-330. Ehinger, B., and Dowling, J. (1987). Retinal neurocircuitry and transmission. In Handbook o f Chemical Neuroanatomy, Vol. 5. Integrated Systems o f the CNS Part 1 (A. Bjorklund, T. Hokfelt, and L. W. Swanson, Eds.), Elsevier, Amsterdam, pp. 389-446. Ehinger, B., and Perez, M. T. R. (1984). Autoradiography of nucleoside uptake into the retina. Neurochem. Int. 6:369-381. Fastbom, J., Pazos, A., and Palacios, J. M. (1987a). The distribution of adenosine A 1 receptors and 5'-nucleotidase in the rat brain of some commonly used experimental animals. Neuroscience 22:813-826. Fastbom, J., Pazos, A., Probst, A., and Palacios, J. M. (1987b). Adenosine A~ receptors in the human brain: A quantitative autoradiographic study. Neuroscience 22:827-839. Friedman, Z., Hackett, S. F., Linden, J. and Campochiaro, P. A. (1989). Human retinal pigment epithelial cells in culture possess A2-adenosine receptors. Brain Res. 492:29-35. Geiger, J. D. (1986). Localization of [3 H]cyclohexyladenosine and [3 H]nitrobenzylthioinosine binding sites in rat striatum and superior coiliculus. Brain Res. 363:404-408. Geiger, J. D., and Nagy, J. I. (1984). Heterogeneous distribution of adenosine transport sites labelled

Adenosine in the Retina

483

by [3H]nitrobenzylthioinosine in rat brain: An autoradiographic and membrane binding study. Brain Res. Bull. 13:657-666. Geiger, J. D., Johnston, M. E., and Yago, V. (1988). Pharmacological characterization of rapidly accumulated adenosine by dissociated brain cells from adult rat. J. Neurochem. 51:283-291. Goodman, R. R., Cooper, M. J., Gavish, M., and Snyder, S. H. (1982). Guanine nucleotide and cation regulation of the binding of [3H]cyclohexyladenosine and [3H]diethylphenylxanthine to adenosine Al-receptors in brain membranes. Mol. Pharmacol. 21:329-335. Goodmam, R. R., Kuhar, M. J., Hester, L., and Snyder, S. H. (1983). Adenosine receptors: Autoradiographic evidence for their location on axon terminals of excitatory neurons. Science 220:967-969. Henderson, J. F. (1979). Regulation of adenosine metabolism. In Physiological and Regulatory Functions of Adenosine and Adenine Nucleotides (H. P. Baer, and G. I. Drummond, Eds.), Raven Press, New York, pp. 315-322. Hollins, C., and Stone, T. W. (1980). Characteristics of the release of adenosine from slices of rat cerebral cortex. J. Physiol. 303:73-82. Jarvis, S. M. (1988). Adenosine transporters. In Receptor Biochemistry and Methodology: Adenosine Receptors (D. M. F. Cooper, and C. Londos, Eds.), Alan R. Liss, New York, pp. 113-123. Jarvis, M. F., Jackson, R. H., and Williams, M. (1989). Autoradiographie characterization of high-affinity adenosine A 2 receptors in the rat brain. Brain Res. 484:111-118. Jonzon, B., and Fredholm, B. B. (1985). Release of purines, noradrenaline and GABA from rat hippocampal slices by field stimulation. J. Neurochem. 44:217-224. Kreutzberg, G. W., and Hussain, S. T. (1984). Cytochemical localization of 5'-nucleotidase activity in retinal photoreceptor cells. Neuroscience 11: 857-866. Kreutzberg, G. W., Barron, K. D., and Schubert, P. (1978). Cytochemieal localization of 5'-nucleotidase in glial plasma membranes. Brain Res. 158:247-257. LeHir, M., and Dubach, U+ C. (1984). Sodium gradient-energized concentrative transport of adenosine in renal brush border vesicles. Pflugers Arch. 401:58-63. LeHir, M., and Dubach, U. C. (1985a). Uphill transport of pyrimidine nueleotides in renal brush border vesicles. Pflugers Arch. 404:238-243. LeHir, M., and Dubach, U. C. (1985b). Concentrative transport of purine nucleosides in brush border vesicles of the rat kidney. Eur. J. Clin. Invest. 15:121-127. Lohse, M. J., Lenschow, V., and Schwabe, U. (1984). Two affinity states of Ri adenosine receptors in brain membranes. Mol. Pharmacol. 26:1-9. Londos, C., Wolff, J., and Cooper, D. M. F. (1981). Adenosine as a regulator of adenylate cyclase. In Purine Receptors, Receptors and Recognition, Series B, Vol. 12 (G. Burnstock, Ed.), Chapman and Hall, London, pp. 289-323. MacDonald, W. F., and White, T. D. (1985). Nature of extrasynaptosomal accumulation of endogenous adenosine evoked by K + and verayridine. J. Neurochem. 45:791-797. Magistretti, P. J., Hof, P. R., and Martin, J.-L. (1986). Adenosine stimulates glycogenolysis in mouse cerebral cortex: A possible coupling mechanism between neuronal activity and energy metabolism. J. Neurosci. 6:2558-2562. Masland, R. H., Mills, J. W., and Hayden, S. A. (1984). Acetylcholine-synthesizing amacrine cells: Identification and selective staining by using radioautography and fluorescent markers. Proc. R. Soc. Lond. 223:79-100. Massey, S. C., and Redburn, D. A. (1987). Transmitter circuits in the vertebrate retina. Prog. Neurobiol. 28:55-96. Michaelis, M. L., Johe, K. K., Moghadam, B., and Adams, R. N. (1988). Studies on the ionic mechanism for the neuromodulatory actions of adenosine in the brain. Brain Res. 473:249-260. Motley, S. J., and Collins, G. G. S. (1983). Endogenous adenosine inhibits excitatory transmission in the rat olfactory cortex slice. Neuropharmacology 22:1081-1086. Nagy, J. I., LaBella, L. A., Buss, M., and Daddona, P. E. (1984). Immunohistochemistry of adenosine deaminase: Implications for adenosine neurotransmission. Science 224:166-168. Nagy, J. I., Geiger, J. D., and Daddona, P. E. (1985). Adenosine uptake sites in rat brain: Identification using [3H]nitrobenzylthioinosine and colocalization with adenosine deaminase. Neurosci. Lett. 55:47-53. Newby, A. C., and Sala, G. B. (1982). A new procedure for haptenizing adenosine leading to a more specific radioimmunoassay method. Biochem. J. 208:603-610. Niemeyer, G., and Friih, B. (1989). Adenosine and cyclohexyladenosine inhibit the cat's optic nerve action potential. ExP3erientia 45:A18 (abstr.). Osborne, N. N. (1989). [ H]Glycogen hydrolysis elicited by adenosine in rabbit retina: Involvement of A2-receptors. Neurochem. Int. 14:419-422.

484

Blazynski and Perez

Paes de Carvalho, R. (1990). Development of A1 adenosine receptors in the chick embryo retina. J. Neurosci. Res. 25:236-242. Paes de Carvalho, R., and de Mello, F. G. (1982). Adenosine-elicited accumulation of adenosine 3',5'-cyclic monophosphate in the chick embryo retina. J. Neurochem. 38:493-500. Paes de Carvalho, R., and de Mello, F. G. (1985). Expression of A1 adenosine receptors modulating dopamine-dependent cyclic AMP accumulation in the chick embryo retina. J. Neurochem. 44:845-851. Paes de Carvalho, R., Braas, K. M., Snyder, S. H., and Adler, R. (1989). Adenosine uptake and release in purified cultures of chick retinal neurons and photoreceptors. J. Neurochem. 52:s157-s157 (abstr.). Paes de Carvalho, R., Braas, K. M., Snyder, S. H., and Adler, R. (1990). Analysis of adenosine immunoreactivity, uptake and release in purified cultures of developing chick embryo retinal neurons and photoreceptors. J. Neurochem. 55:1603-1611. Palczewski, K., McDowell, J. H., and Hargrave, P. H. (1988). Purification and characterization of rhodopsin kinase. J. Biol. Chem. 2,63:14067-14073. Perez, M. T. R., and Bruun, A. (1987). Colocalization of [3HI-adenosine accumulation and GABA immunoreactivity in the chicken and rabbit retinas. Histochemistry 87:413-417. Perez, M. T. R., and Ehinger, B. (1986). Adenosine uptake and release in the rabbit retina. In Retinal Signal Systems, Degenerations and Transplants (E. Agardh, and B. Ehinger, Eds.), Elsevier Science, Amsterdam, pp. 163-172. Perez, M. T. R., and Ehinger, B. (1989a). Multiple neurotransmitter systems influence the release of adenosine derivatives from the rabbit retina. Neurochem. Int. 15:411-420. Perez, M. T. R., and Ehinger, B. (1989b). Adenosine inhibits evoked acetylcholine release from the rabbit retina. J. Neurochem. 52:S157 (abstr.). Perez, M. T. R., Ehinger, B. E., Linstrom, K., and Fredholm, B. B. (1986). Release of endogenous and radioactive purines from the rabbit retina. Brain Res. 398:106-112. Perez, M. T. R., Arner, K., and Ehinger, B. (1988). Stimulation-evoked release of purines from the rabbit retina. Neurochem. Int. 13:307-318. Phillis, J. W., and Wu, P. H. (1981). The role of adenosine and its nucleotides in central synaptic transmission. Prog. Neurobiol. 16:187-239. Schorderet, M. (1989). Receptors coupled to adenylate cyclase in isolated rabbit retina. Neurochem. Int. 14:387-395. Schubert, P. (1988). Physiological modulation by adenosine: selective blockade of A1 receptors with DPCPX enhances stimulus train-evoked neuronal Ca influx in rat hippocampal slices. Brain Res. 458:162-165. Scott, T. G. (1967). The distribution of 5'-nucleotidase in the brain of the mouse. J. Comp. Neurol. 129:97-113. Senba, E., Daddona, P. E., and Nagy, J. I. (1986). Immunohistochemical localization of adenosine deaminase in the retina of the rat. Brain Res. Bull. 17:209-217. Tapia, P., and Arias, C. (1982). Selective stimulation of neurotransmitter release from chick retina by kainic acid and glutamic acid. J. Neurochem. 39:1169-1178. Tauchi, M., and Masland, R. H. (1984). The shape and arrangement of the cholinergic neurons in the rabbit retina. Proc. R. Soc. Lond. 223:101-119. van Calker, D., Muller, M., and Hamprecht, B. (1979). Adenosine regulates via two different types of receptors, the accumulation of cyclic AMP in cultured brain cells. J. Neurochem. 33:999-1005. Watt, C. B., Su, Y. Y. T., and Lam, D. M. K. (1985). Enkaphalins in the vertebrate retina. In Progress in Retinal Research, Vol. 4 (N. N. Osborne, and G. J. Chader, Eds.), Pergamon Press, Oxford, pp. 221-242. Williams, M. (1987). Purine receptors in mammalian tissues: Pharmacology and functional significance. Annu. Rev. Pharmacol. Toxicol. 27:315-345. Woods, C., and Bazynski, C. ,(1991). Characterization of adenosine A1 receptor binding sites in bovine retinal membranes. Exp. Eye Res. (in press). Wu, P. H., and Phillis, J. W. (1984). Uptake by central nervous tissues as a mechanism for the regulation of extraceUular adenosine concentrations. Neurochem. Int. 6:613-632. Wu, P. H., Moron, M., and Barraco, R. (1984). Organic calcium channel blockers enhance [3H]purine release from rat brain cortical synaptosomes. Neurochem. Res. 9:1019-1031. Yeung, S. M., and Green, R. D. (1983). Agonist and antagonist affinities for inhibitory adenosine receptors are reciprocally affected by 5'-guanylylimidodiphosphate or N-ethylmaleimide. J. Biol. Chem. 258:2334-2339.