Ординатура / Офтальмология / Английские материалы / The Retina and its Disorders_Besharse, Bok_2011

neuromodulator dopamine is a key player in this rodmediated pathway:

1.intraocularly injected dopamine triggers light-adaptive cone contraction and RPE pigment granule dispersion in animals maintained in the dark, and

2.dopamine antagonists block light-induced cone contraction and RPE pigment granule dispersion in animals moved from darkness to light.

Further evidence that dopamine plays a central role in paracrine regulation of cone retinomotor movements is provided by pharmacological studies using intraocular injection in vivo, isolated retinas in culture, and isolated cone inner–outer segment fragments in vitro. In each of these preparations, light and/or dopamine promote cone contraction, whereas darkness and/or the adenylate cyclase stimulator, forskolin, promote cone elongation. Pharmacological profiles indicate that dopamine is acting

through D4-like or other D2 family dopamine receptors. This class of dopamine receptors causes inhibition of adenylate cyclase and, thus, would be expected to lower cone cytoplasmic cAMP levels, consistent with observations that experimentally increasing cAMP produces dark-adaptive movements.

The ability of light and dopamine to regulate retinomotor movement in isolated cone inner–outer segment fragments indicates that these agents can act directly on cones. Thus, light can trigger cone contraction in isolated cones, even though action spectra indicate that in vivo light acts through a rod-mediated pathway. These studies also indicate that D4 dopamine receptors are present on cone inner and/or outer segments. In teleost retinas, the source of dopamine is the interplexiform cell. Since projections of this inner retinal cell extend only into the outer plexiform layer, released dopamine would have to diffuse over tens of microns to reach receptors on the inner

segments in vivo. The very high affinity of D4 receptors (dissociation constant (Kd) in the nM range) may facilitate this paracrine mode of signaling.

Light is much more effective than dopamine in triggering myoid elongation in rod fragments, and light and circadian regulation of rod movement was unaffected by treatment with 6-hydroxydopamine, which kills dopaminergic interplexiform cells. Light activation of myoid elongation in rod inner–outer segment fragments in vitro requires relatively high light intensities (20% bleach). The photoreceptive mechanism responsible for this activation can apparently count photons accurately for light pulse durations up to at least 10 min, suggesting that the critical factor in light activation of rod elongation is quantum catch, rather than duration of the light stimulus. Similar high-intensity thresholds and photon counting have been reported to mediate entrainment of circadian rhythms in several species.

The observation that dopamine plays a role in circadian as well as light regulation of cone retinomotor movement is suggested by results obtained by intraocular injection of dopamine agonists and antagonists in animals maintained in DD. Cone myoid contraction can be induced at midnight by intraocular injection of dopamine or D4 receptor agonists. Partially contracted cones of DD animals at subjective midday can be induced to contract fully by intraocular injection of dopamine or D4 agonists, or to elongate by injection of the D4/D2 family antagonists. The predawn cone contraction observed in DD animals in response to circadian signals can be completely eliminated by intraocular injection of the D4/D2 antagonist shortly before the expected time of light onset. These observations suggest that circadian regulation of cone myoid length is mediated by endogenous dopamine, acting through D4/D2 family receptors.

Consistent with these findings, retinal dopamine release is higher in the light than in the dark, and dopamine release is higher in subjective day than in subjective night in DD in many species. These diurnal and circadian cycles of dopamine release are inversely correlated with diurnal and circadian cycles of cAMP and protein phosphorylation levels in photoreceptors. Recent findings suggest that dopaminergic amacrine cells contain an autonomous circadian clock that drives dopamine release and metabolism. Although dopamine plays a critical role in regulating fish retinomotor movements, it is nonetheless unlikely to be the sole circadian regulator, since both light-induced and circadian cone movements persist (at reduced amplitude) in fish retinas after lesion of interplexiform cells by 6-hydroxydopamine.

Adenosine

This neuromodulator has an opposite effect to that of dopamine on cone retinomotor movements. In isolated cone inner–outer segment fragments, adenosine A2 receptor agonists activate and A2 antagonists inhibit cone myoid

elongation. Since A2 receptors are positively coupled to adenylate cyclase, adenosine effects on cone fragments are consistent with observations that elevating cAMP triggers dark-adaptive retinomotor movements. These findings indicate that A2 receptors must be present on cone inner (and/or outer) segments. Since cones elongate at subjective dusk in fish maintained in DD, and since adenosine stimulates cone elongation, adenosine could provide an endogenous circadian signal for expected dark onset in addition to the decrease in dopamine release occurring at that time.

Adenosine effects on cone movements could be a local effect, reflecting photoreceptor metabolic activity. Since an adenosine transporter is present in photoreceptors, increases in adenosine in photoreceptor inner segments are likely to be accompanied by an increase in adenosine level in the subretinal space. The increased sodium/ potassium adenosine triphosphatase (Naþ/Kþ-ATPase) activity in photoreceptor inner segments occurring in darkness is likely to be accompanied by an increase in intracellular adenosine levels, and consequently, increased release of adenosine into the subretinal space. The binding of released adenosine to photoreceptor A2 receptors might then enhance dark adaptation of the photoreceptors by stimulating adenylate cyclase, thereby reinforcing the dark signal. Other studies have shown that adenosine triphosphate (ATP) is released across the apical membrane of the RPE into the subretinal space in response to various triggers. This ATP is dephosphorylated into adenosine by extracellular enzymes (ectoenzymes) on the RPE apical membrane. Regulation of adenosine release and ectoenzme activity in response to light signals could also alter the balance of purines in subretinal space and, thus, influence retinomotor movements in cones.

Regulation of Retinomotor Movements in RPE Cells by Paracrine Messengers

RPE pigment migration in frogs has an action spectrum most closely resembling that of the rod photopigment. These observations suggest that for RPE, as for cones, a rod-mediated pathway triggers light-adaptive retinomotor movement. Pigment position is not affected by light in isolated sheets of teleost RPE, further demonstrating that light acts indirectly through a photoreceptor-mediated, paracrine pathway.

In isolated sheets of teleost RPE in culture, dopamine and D2 family agonists induce pigment granule dispersion, while treatments that elevate cAMP induce pigment granule aggregation. In isolated RPE cells aggregated in cAMP, pigment dispersion is induced by microinjection of PKA inhibitors, suggesting that continuous phosphorylation of PKA targets is required to maintain the aggregated state. Since aggregation can also be induced by the phosphatase inhibitor okadaic acid, it seems likely that PKA

and phosphatases are simultaneously active in RPE cells, and that their relative activities are altered by light and dark signals from the retina.

Surprisingly, underivatized cAMP is just as effective as membrane-permeant cAMP analogs at activating pigment granule aggregation in isolated RPE sheets. Washout of exogenous cAMP induces dispersion. This cAMP directly enters the RPE cell through organic ion transporters is suggested by observations that ATP and adenosine are ineffective at triggering aggregation and that organic ion transport inhibitors block cAMP – but not forskolininduced pigment aggregation. Recently, it has been shown that isolated RPE takes up cAMP in a saturable manner. Thus, it seems clear that cAMP from the subretinal space can actually enter RPE cells through organic ion channels to activate pigment aggregation. These findings indicate that an increase in cAMP in the subretinal space in the dark would activate pigment granule aggregation in RPE cells. cAMP efflux has been shown to be associated with increased intracellular cAMP accumulation in many cell types, including pinealocytes. Thus, diurnal cycles of cAMP levels in retinal photoreceptors might be expected to produce increased cAMP efflux into the subretinal space at night. This extracellular cAMP in the subretinal space could then function as a retina-to-RPE signal for darkness.

cAMP and intracellular calcium often interact in cellular signaling pathways, either antagonistically or synergistically. Pigment granule migration is regulated by calcium in several types of dermal chromatophores. However, increasing extracellular or intracellular calcium has no effect on pigment granule position in either aggregated or dispersed isolated RPE cells. Furthermore, intracellular calcium levels are unaffected when RPE pigment granule motility is triggered by cAMP or cAMP washout. These findings suggest that RPE pigment granule movements are regulated directly by cAMP, that is, by phosphorylation and dephosphorylation of PKA targets, and that calcium does not act downstream of cAMP in the regulatory process.

Carbachol (an acetylcholine analog) triggers dispersion in isolated fish RPE. The carbachol receptor acts through a pathway commonly linked to calcium mobilization and carbachol-induced pigment granule dispersion is blocked by the calcium chelator 1,2-bis(o-aminophenoxy) ethane-N,N,N 0,N 0-tetraacetic acid (BAPTA), suggesting that a rise in intracellular calcium plays a role upstream of cAMP in carbachol-induced dispersion.

Functions and Significance of

Retinomotor Movements

What role retinomotor movements play in retinal function has been a topic of inquiry and speculation for more

than a century. Direct experimental demonstration that retinomotor movements affect vision has been difficult to achieve, primarily because of the difficulty in interfering with retinomotor movements without compromising other aspects of retinal function. Nevertheless, several likely functions for retinomotor movements have been suggested. Occurring generally in species that lack pupillary movements, rod and RPE retinomotor movements provide an alternative mechanism to pupillary movements for shielding rods from full bleach in bright light while permitting optimal exposure for dim-light vision. Repositioning the rod and cone photoreceptors likewise provides an efficient mechanism for optimizing space by positioning the rods first in line to detect light at the focal plane across the entire retinal surface under dim-light conditions, and then moving cones to this position for bright-light vision. These movements permit the entire retinal surface to be used alternately for rod and cone vision. Some have noted that each photoreceptor type is elongated when it is expected to be silent, and suggested that the cable properties of the elongated myoid might contribute to attenuation of the signal from outer segment to synapse. The great morphological variation in retinomotor movements perhaps reflects the optimization of one or the other of these functions in different species.

Summary

Although retinomotor movements are most pronounced in lower vertebrates, their study contributes to a broader understanding of diurnal and circadian regulation of photoreceptor and RPE physiology. Retinomotor movements provide excellent models for investigating the roles of cytoskeletal elements in cell shape change and intracellular transport. Studies of retinomotor movements have called attention to the importance of cyclic changes in intracellular cAMP levels in the diurnal and circadian regulation of other aspects of photoreceptor physiology and metabolism. The recognition that light induction of cone and RPE movements depends on a rod-mediated paracrine pathway has contributed to recognition of the role that dopamine plays in light and circadian signaling. Since retinomotor movements exhibit properties of light and circadian regulation, characteristic of many other aspects of retinal physiology, they provide excellent experimental models for understanding light and circadianregulatory processes.

See also: Circadian Metabolism in the Chick Retina; The Circadian Clock in the Retina Regulates Rod and Cone Pathways; Circadian Regulation of Ion Channels in Photoreceptors; Limulus Eyes and Their Circadian Regulation; Neurotransmitters and Receptors: Dopamine Receptors.

Further Reading

Ali, M. A. (1975). Retinomotor responses. In: Ali, M. A. (ed.) Vision in Fishes, pp. 313–355. New York: Plenum Press.

Arey, L. B. (1915). The occurrence and significance of photomechanical changes in the vertebrate retina – a historical survey. Journal of Comparative Neurology 25: 535–554.

Back, I., Donner, K. O., and Reuter, T. (1965). The screening effect of the pigment epithelium on the retinal rods in the frog. Vision Research 5: 101–111.

Blaxter, J. H. S. (1975). The eyes of larval fish. In: Ali, M. A. (ed.) Vision in Fishes, pp. 427–443. New York: Plenum Press.

Burnside, B. (1988). Photoreceptor contraction and extension: Calcium and cAMP regulation of microtubuleand actin-dependent changes in cell shape. In: Lasek, R. J. (ed.) Intrinsic Determinants of Neuronal Cell Form, pp. 323–359. New York: Alan R. Liss.

Burnside, B. (2001). Light and circadian regulation of retinomotor movement. Progress in Brain Research 131: 477–485.

Burnside, B. and Dearry, A. (1986). Cell motility in the retina. In: Alder, R. and Farber, D. B. (eds.) The Retina, Part 1, pp. 151–206. New York: Academic Press.

Burnside, B. and King-Smith, C. (2009). Retinomotor movements. In: Squire, L. R. (ed.) Encyclopedia of Neuroscience. Oxford: Academic Press. http://www.sciencedirect.com/science/referenceworks/ 9780080450469 (accessed July 2009).

Burnside, B. and Nagle, B. (1983). Retinomotor movements of photoreceptors and retinal pigment epithelium: Mechanisms and regulation. Progress in Retinal Research 2: 67–109.

Dearry, A. and Burnside, B. (1989). Regulation of cell motility in teleost retinal photoreceptors and pigment epithelium by dopaminergic D2 Receptors. In: Redburn, D. and Morales, H. P. (eds.) Extracellular and Intracellular Messengers in the Vertebrate Retina, pp. 229–256. New York: Alan R. Liss.

Douglas, R. H. (1982). The function of photomechanical movements in the retina of the rainbow trout (Salmo gairdneri). Journal of Experimental Biology 96: 389–403.

McNeil, E. L., Tacelosky, D., Basciano, P., et al. (2004). Actindependent motility of melanosomes from fish retinal pigment epithelial (RPE) cells investigated using in vitro motility assays. Cell Motility and the Cytoskeleton 58: 71–82.

Pozdeyev, N., Tosini, G., Li, L., et al. (2008). Dopamine modulates diurnal and circadian rhythms of protein phosphorylation in photoreceptor cells of mouse retina. European Journal of Neuroscience 27: 2691–2700.

Wagner, H. J., Kirsch, M., and Douglas, R. H. (1992). Light dependent and endogenous circadian control of adaptation in teleost retinae. In: Ali, M. A. (ed.) Rhythms in Fishes, pp. 255–292. New York: Plenum Press.

Glossary

Disinhibition – A synaptic interaction in which the inhibitory input to a neuron is itself inhibited, thereby relieving the neuron of inhibitory control. Endocannabinoids – Natural chemicals in the body whose actions on metabotropic cannabinoid receptors are mimicked by the active component of marijuana.

Ionotropic receptors – Membrane proteins that, when activated by specific ligands, directly alter membrane conductance.

IPSC – Inhibitory post synaptic current counteracts excitatory input to a neuron, for example by hyperpolarization or induction of a shunt current.

Metabotropic receptors – Membrane proteins that, when activated by specific ligands, indirectly alter a wide variety of cellular properties through G-protein-coupled enzyme cascades.

Introduction

Gamma-aminobutyric acid (GABA) is the major inhibitory amino acid transmitter in the retina. It is overwhelmingly represented in lateral inhibition, being most prominent in one class of horizontal cell and in numerous subtypes of amacrine cell. GABAergic transmission requires the synthesizing enzyme glutamic acid decarboxylase (GAD) and the degradative enzyme GABA transaminase (GABA-T) that may or may not be found in the same cells. The physiological actions of GABA are effected by receptors that may be broadly defined as proteins that bind to and respond to the presence of GABA. Under this scheme, there are three functional types of GABA receptor: (1) a vesicular transporter that concentrates cytoplasmic GABA into synaptic vesicles, (2) a membrane transporter that translocates GABA from the extracellular space into glia or neurons, and (3) plasma membrane receptors that mediate the cell’s response to synaptically released GABA. This article highlights the properties and functions of these major types of the GABA receptor. Numerous sources on the history, physiology, and molecular biology of these receptor types are provided. Table 1 illustrates the major types of the GABA receptor with representative agonists, antagonists, and most common

cellular locations in the outer plexiform layer (OPL) and inner plexiform layer (IPL). The pharmacology of GABA transporters (GATs) is less well developed than it is for the synaptic receptors and continues to be an area of intensive research.

Vesicular Transporters

Cytoplasmic GABA is concentrated into synaptic vesicles by a vesicular inhibitory amino acid transporter (VIAAT) that uses Hþ-antiport activity to drive the uptake of GABA or glycine into synaptic vesicles. VIAAT is the only member of the solute carrier 32 (SLC32) family of Hþ-coupled amino acid transporters; it is not related to the vesicular transporters for glutamate, monoamines, or acetylcholine. VIAAT, referred to as the vesicular GAT (VGAT), when applied to GABAergic neurons, is essential for the vesicular release of GABA and glycine. The existence of a common vesicular transporter for GABA and glycine is consistent with reports of a small percentage of amacrine cells that co-localize and likely release both GABA and glycine.

VIAAT was localized by immunohistochemistry (IHC) in zebrafish, mouse, rat, cat, and human retinas. In the inner retina, synaptic boutons throughout all layers of the IPL contain VIAAT immunoreactivity (IR), consistent with data showing that virtually all amacrine cells in the retina are either GABAergic or glycinergic. Data from salamander, cat, and human suggest the existence of subpopulations of bipolar cells that likely release both GABA and glutamate. For example, certain OFF cone bipolar cells in the cat retina contain not only VIAAT-IR and GAD65-IR, but also vesicular glutamate transporter-IR (VGLUT-IR). In the outer retina, VIAAT-IR is present in horizontal cell dendrites in zebrafish, mouse, rat, and human. The presence of GABAA receptors on the photoreceptor terminals of some mammals supports the notion that GABA is synaptically released in the OPL. VIAATIR in the OPL most likely includes not only the horizontal cell dendrites, but also the boutons of GABAergic and glycinergic interplexiform cells. These processes, however, have a relatively low density and are unlikely to be confused with the more numerous horizontal cell dendrites. Electron microscopy shows that VIAAT-immunoreactive horizontal cell dendrites innervate rods and cones in mouse and rat, suggesting that A-type and B-type horizontal cells of the mammalian retina are capable of vesicular GABA release. In fish horizontal cells, the subcellular

localization of VIAAT-IR has not been determined by electron microscopy; there are no data to indicate whether VIAAT-IR is restricted to the H1 GABAergic horizontal cells, innervating only cones, or distributed among all four types, innervating both rods and cones.

In the outer retina of mammals and nonmammals, the classification of neurons as GABAergic can be difficult. The presence of VIAAT alone does not necessarily indicate the presence of GABA for vesicular release. Other indicators, such as the presence of glutamate decarboxylase (GAD), the enzyme that catalyzes decarboxylation of glutamate to form GABA, or GABA uptake, are needed to support a GABAergic identity. For example, in most nonmammalian species, VIAAT-positive horizontal cell dendrites do not contain synaptic vesicles or synaptic specializations. Rather, a sodium-dependent GAT appears to facilitate GABA release from horizontal cells. To complicate matters further, the presence of GABA or GAD in an outer retinal neuron does not necessarily stipulate that GABA is synaptically released. For example, although GAD-IR and GABA-IR have been found in cone terminals in primate, lizard, and toad retinas, VIAAT-IR has

not been localized to photoreceptor terminals in any species. Lastly, although horizontal cells of some mammals contain VIAAT-IR, GABA-IR, and GAD-IR, neither GABA uptake nor GATs have been described in the horizontal cells of any mammal.

Plasma Membrane Transporters

GATs are members of the SLC6 family that belongs to the Naþ/Cl–-dependent neurotransmitter transporter superfamily. GATs are arranged in 12 transmembrane domains. Molecular cloning studies identified three subtypes of high-affinity GAT (GAT-1, GAT-2, and GAT-3) and one lower-affinity betaine/glycine transporter that also transports GABA (BGT-1). GATs are responsible for the clearance of GABA from the extracellular space and, conceivably, could be present on the presynaptic neuron, postsynaptic neuron, or surrounding glia. In general, GAT-1 is present in neurons of the inner retina, GAT-3 in Mu¨ller’s cells, and GAT-2 in the retinal pigmented epithelium (RPE) and ciliary epithelium of the

Müller cell

Cl−

GABAA

Succinic semialdehyde

T-GABA

GABA

GAT-1

mammalian retina (Figure 1). Overall, there is excellent agreement in the cellular distributions of 3H-GABA uptake, GABA-T, which degrades GABA, and GATs in amacrine cells across species. However, in regard to horizontal cells and Mu¨ller’s cells, there are notable species differences and inconsistencies.

Neuronal Localization

In a wide variety of mammals and nonmammals, from humans to teleost fish, GAT-1 is the predominant GAT/type in amacrine cells, displaced amacrine cells, and interplexiform cells. This distribution is consistent with the patterns of GAD-IR, GABA-IR, and 3H-GABA uptake. Mammalian Mu¨ller’s cells are labeled intensely with GAT-3 and, to a lesser extent, by GAT-1, consistent with observations that Mu¨ller’s cells contain high levels of GABA-T and are the initial site for 3H-GABA uptake. For the most part, nonmammalian Mu¨ller’s cells neither take up 3H-GABA nor display GAD-IR or GABA-IR. One exception is the skate retina, in which Mu¨ller’s cells express GAT-1 and GAT-3, display a GAT current, and take up 3H-GABA. Another exception is the salmon retina, in which Mu¨ller’s cells have GABA-IR, but do not contain GAT-1. Cultured chick Mu¨ller’s cells contain GAT-1 and GAT-3, while bullfrog Mu¨ller’s cells contain GAT-1

and GAT-2 and display a GAT current. The implication of these findings in chick and bullfrog is not clear, given the absence of supporting evidence from 3H-GABA uptake or GABA-T immunohistochemical studies. Given the high density of GAT on chick and bullfrog Mu¨ller’s cells, including the end feet at the inner margin of the retina, it is surprising that 3H-GABA uptake has never been demonstrated in Mu¨ller’s cells of these species, regardless of how the 3H-GABA was administered.

Although horizontal cells in many mammals contain GABA-IR and GAD-IR, they do not take up 3H-GABA. As would be expected, GATs have not been localized to any mammalian horizontal cells. The evidence for a class of GABAergic horizontal cell is overwhelming for nonmammals. Horizontal cells of fish, amphibians, birds, and reptiles take up 3H-GABA with great avidity. GAT-1, present in the inner retina, has not been localized to horizontal cells of any species. Studies on the localization of GAT-2 and GAT-3 in nonmammalian horizontal cells are sporadic, with studies usually focusing on one or the other. Exceptions are goldfish and zebrafish in which GAT-2 and GAT-3 were localized to a subtype of the cone horizontal cell, the H1 luminosity type. GAT-3 was not found in horizontal cells of skate, salamander, or bullfrog. GAT-2, absent in bullfrog horizontal cells, was not studied in skate, salamander, or salmon.

Except for goldfish and zebrafish, the lack of horizontal cell labeling by GAT in other nonmammals could be due to the general use of antibodies against the C-terminus of rat GAT-2 and GAT-3. There is a strong homology in the GAT amino acid sequence in the animal kingdom. Torpedo GAT-3 shows a 77% identical amino acid sequence with the mammalian GAT-3. Immunoblots of the goldfish brain with GAT-3 show a band between 60 and 75 kDa, consistent with the expected weight of 71 kDa. However, regardless of the overall homology or the sequence homology at the C-terminus, the site of antibody production is more critical in determining the validity of the labeling. Except for fish, the negative finding of GAT-2 and GAT-3 on horizontal cells in nonmammals has not been resolved. A curiosity is that 3H-muscimol, a GABAA agonist, is transported avidly by amacrine cells in mammals and nonmammals, but not to any great extent by horizontal cells in any species, including fish and bird. This is a further indication of the difference in the type of GAT present on amacrine cells and nonmammalian horizontal cells.

A small percentage ( 10%) of bipolar cells in some retinas may be GABAergic. In salamander, the vast majority of bipolar cells that contain GAT-1-IR also contain GABAIR, as well as VGAT and GAD. Two types of OFF bipolar cell in zebrafish contain GABA and likely correspond to the OFF bipolar cells that display a GAT current. However, in primate retina, the variable reports of GABA-IR in bipolar cells are not supported by data localizing GAD, GAT-1, GAT-2, GAT-3, or GABA-T. These findings are significant because of the possibility that some bipolar cells release GABA and glutamate as neurotransmitters.

A GABA/taurine transporter was identified in bullfrog at the apical surface of the RPE. It was suggested that the RPE could take over the clearance of GABA released from horizontal cells in the distal retina of nonmammals because there are no reports that Mu¨ller’s cells transport 3H-GABA. Despite the reports of GATs in bullfrog Mu¨ller’s cells, this seems to be an attractive idea that would be supported by the demonstration of GABA-T localization and GABA-T activity in the nonmammalian RPE.

Function

The major function of the GATs is to clear GABA from the extracellular space to either re-enter the vesicular pool or enter the tricarboxylic acid cycle after GABA-T action (Figure 1). There are at least three other consequences of the transporter activity. The first is that GABA transport is electrogenic with two Na2þ, one Cl–, and one GABA per cycle, resulting in one-net inward positive charge. The depolarization that accompanies GABA uptake could modulate neuronal excitability by triggering Ca2þ influx and the subsequent release of Ca2þ from intracellular stores. For example, a nonsubstrate blocker of GABA transport

(NO-711) reduced cellular edema, presumably by blocking the ionic influx that accompanies GABA uptake. This strategy has proven useful in treating seizures and could apply to protecting retinal ganglion cells following ischemia or an excitotoxic release of glutamate from bipolar cells. As a second consequence of transporter activity, neuronal depolarization or a breakdown of the Naþ/Cl–/GABA gradient can cause the GATs to operate in the reverse direction, thus releasing GABA. For example, an excitotoxic assault on the chick retina increases extracellular GABA by reversal of the GAT. This calcium-independent GABA release has been suggested to be the normal mode of release in horizontal cells, glia, and, perhaps, in starburst amacrine cells. This form of GABA release does not require adenosine triphosphate (ATP), except to maintain the ionic gradients. Third, the activity of GATs can limit the spillover of GABA in the extracellular space and regulate inhibition. This has been shown in the inner retina, in which inhibition of GATs by NO-711 enhances the light-evoked inhibitory postsynaptic currents (IPSCs) at GABAC receptors on bipolar cells, but has no effect on GABAA receptors on ganglion cells. The difference in response is due to the observation that GABAA receptors desensitize, whereas GABAC receptors do not. Thus, the GAT selectively modulated the transmission of signals from bipolar cells to ganglion cells. In all these cases, the effect of GAT activity is to regulate excitability.

Synaptic Receptors

There are three general classes of synaptic GABA receptors: two ionotropic and one metabotropic. These are differentiated by ligand-binding affinities and molecularcloning techniques. GABAA receptors are heteropentameric structures that form a chloride channel using structurally related subunits a1–6, b1–4, g1–3, d, e, and p. The most common pentamers are composed of abg subunits, with d often substituting for g. The subunit composition determines receptor pharmacology and kinetics. For example, GABAA receptors mediate the effects of benzodiazepines (BDZ), and the g subunit is necessary for BZD sensitivity, while the type of a subunit determines BZD affinity. The b subunit affects channel properties and BZD efficacy. Substitution of an a4 or a6 subunit for an a1 subunit eliminates BZD sensitivity as does substitution of the d subunit for a g subunit. GABAA receptors are blocked by bicuculline, the nonselective chloride channel blocker, picrotoxin, and can be modulated allosterically by barbiturates, BZDs, ethanol, and steroids. The action of the GABAA receptor tends to be phasic. The ionotropic GABAC receptor also is a chloride channel that is composed mostly of homoligomeric r subunits (r1, r2, and r3). GABAC receptors are insensitive to bicuculline, more sensitive to GABA than GABAA receptors, and can also be blocked by picrotoxin.

GABAC receptors do not desensitize; therefore, their effect is tonic. Metabotropic GABAB receptors are guanine nucleotide-binding protein-coupled receptors and regulate potassium or calcium channels. GABAB receptors are activated by baclofen, antagonized by phaclofen, insensitive to bicuculline and picrotoxin blockade, and less sensitive to GABA than GABAA receptors.

Photoreceptors

The source of GABA in the OPL of nonmammals is a type of horizontal cell, so-called H1 or luminosity cells. In mammals, GABA can be released from a type of interplexiform cell as well as from horizontal cells. GABAA receptors were localized to the OPL by in vitro autoradiography (ARG) of 3H muscimol binding, GABAA receptor subunit-specific IHC, and single-cell electrophysiology. 3H-BZD-binding sites were localized by in vitro ARG in the OPL of rat, monkey, and human, but were not present in the OPL of fish, salamander, or bird. There is evidence for GABAA receptors on rod terminals in goldfish and rat, and on the cone terminals in goldfish, tiger salamander, bullfrog, turtle, chicken, rat, mouse, pig, and cat. Most evidence shows b2/3 subunits on photoreceptor terminals. Data for a subunits are inconsistent, with negative results from in situ hybridization and positive results with IHC. In contrast to the localization of g1 and g2 subunits to salamander cone terminals by IHC, negative results with 3H-BZD binding indicate an absence of g subunits. This discrepancy could be due either to the nonspecific nature of IHC using antibodies that are not against salamander g subunits or to low sensitivity of in vitro ARG using photoaffinity labeling of BZD ligands.

GABA-induced responses in cones of all species are reduced by bicuculline, indicating GABAA receptors. In addition, there is a component of GABA-induced responses in mammalian (pig and mouse) cones that is resistant to bicuculline, but sensitive to antagonists of GABAC receptors, indicating participation of GABAA and GABAC receptors. GABAC receptors have not been identified on photoreceptors in nonmammals as yet. In mouse, both GABAA and GABAC components of the response to GABA are potentiated by pentobarbital, suggesting heteromeric channels of GABAC r1 and GABAA receptor subunits. In bullfrog, GABAB receptors contribute to the GABA response along with GABAA receptors. GABAergic negative feedback onto cones is well established and involves a suppression of presynaptic Ca2þ currents. The function of GABA feedback onto photoreceptors is still controversial; it may have limited participation in the formation of the cone receptive field surround. More likely, the modulation of voltage-gated Ca2þ currents could set the gain at the cone synapse to maintain sensitivity to changing light conditions. The nondesensitizing nature of GABAC receptors on mammalian cones may compensate for

the lower level of GABA release that is available from horizontal cells and interplexiform cells in the OPL of mammals compared to the high GABA content of H1 horizontal cells in nonmammals.

Horizontal Cells

The information regarding GABA receptors on horizontal cells is based primarily on data obtained with electrophysiology and, to a lesser extent, on immunohistochemical evidence. In addition, there is considerable species variability. GABAA and GABAB receptors have been localized by IHC on horizontal cells of some mammals and nonmammals, while GABAC receptors have been localized to horizontal cells of perch and goldfish. GABAB receptormediated responses have not been reported for horizontal cells in any species. GABAA responses have been recorded from horizontal cells in catfish, salamander, mouse, rabbit, and cat, while GABAC responses are prominent in horizontal cells in fish (goldfish, perch, and catfish) and salamander. Skate and zebrafish horizontal cells reportedly show neither GABAA nor GABAC responses. In goldfish, GABAA responses are present in isolated chromaticity horizontal cells, while GABAC responses are present in the cell body and axon terminal of the GABAergic H1 horizontal cells. In contrast, in perch, GABAC responses are restricted to the rod horizontal cells. The GABAA response in isolated mouse horizontal cells is enhanced by the BZD, diazepam, and pentobarbital, consistent with the localization of 3H-BZD-binding sites in the mammalian OPL by ARG. GABAC receptor-mediated responses of GABAergic horizontal cells are complicated by the presence of the electrogenic GAT current. H1 horizontal cells release GABA at some steady rate under ambient illumination. They are depolarized by decrements in light intensity, resulting in increased GABA release by the transporter. During the hyperpolarizing response to light onset, the reuptake of GABA will depolarize the horizontal cells, reducing the light response and establishing a positive-feedback loop that allows for further GABA release from the H1 horizontal cells. In turn, the released GABA, via GABAA receptors, will depolarize the other cone and rod horizontal cells. In addition, the GABAC response shows a run-up with repetitive application of GABA, further depolarizing the horizontal cells. However, GABAA and GABAC receptors are suppressed by Zn2þ that is co-released with glutamate. This dampening effect by Zn2þ is especially important to downregulate the tonic GABAC response.

Bipolar Cells

Bipolar cells form functionally diverse groups that are differentiated on the bases of rod and cone input as well as ON or OFF response. On anatomical grounds, bipolar

cells can receive GABAergic input from horizontal cells and interplexiform cells in the OPL and from a multitude of amacrine cell types in the IPL. Species differences among mammals and nonmammals and the effects of dissociation techniques on GABA receptors have resulted in considerable variability in the data. However, GABAA receptors are generally more prominent on dendrites in the OPL and the ratio of GABAA to GABAC receptors on axon terminals in the IPL is high for cone bipolar cells and low for rod bipolar cells. In mammals, IR to GABAA receptors (a1, b2/3, and g2) and the GABAC r subunit is present at nonoverlapping, extrasynaptic sites on the dendrites of rod and cone bipolar cells. The contribution of GABAA and GABAC receptors to GABA-evoked IPSCs is balanced in dendrites of isolated mouse rod bipolar cells, whereas, in the ferret retinal slice, GABAA receptors overwhelmingly dominate the response to GABA puffed at dendrites in all types of the bipolar cell. Dendrites of bipolar cells are insensitive to GABA in salamander and display GABAA properties in goldfish and GABAC properties in zebrafish. In bullfrog, all bipolar cells show GABAA and GABAC sensitivity at dendrites, though GABAA receptors dominated, particularly for on bipolar cells.

In the IPL, GABA-evoked IPSCs are carried mostly by tonic GABAC receptors in rod bipolar cells and by phasic GABAA receptors in cone bipolar cells (Figure 1). This difference in GABA receptor type likely corresponds to the rapid kinetics of the photopic pathway and slower kinetics of the scotopic pathway. This is illustrated in the ferret retina in which activation of GABAC receptors more effectively inhibits the output to AII and A17 amacrine cells than GABAA receptors. When co-expressed ON bipolar cell axons, GABAA and GABAC receptors are differentially distributed. For example, in the ON mixed rod/cone bipolar cell of goldfish, GABAC receptors are clustered at the distal (vitreal) margin of the terminal, while GABAA receptors are more proximal to the cell body, near the connecting axon. Unlike the photoreceptors, there is a very high density of GABA receptors on glutamatergic bipolar cell terminals. Zn2þ, likely to be coreleased with glutamate, suppresses the amplitude of responses mediated by GABAC receptors more so than GABAA receptors. The effects of zinc on the kinetics of GABAA receptors are variable and likely due to differences in GABAA subunit composition. The hyperpolarizing effect of GABA on all bipolar cells strongly inhibits glutamate release. This powerful inhibition is buffered by Zn2þ and other endogenous modulators including dopamine, neuropeptides, and endocannabinoids that control the transmission of signals to amacrine and ganglion cells.

Amacrine Cells and Ganglion Cells

At least two-thirds of the amacrine cells in all species utilize GABA as the primary neurotransmitter; GABAergic

amacrine cells have additional secondary transmitters including acetylcholine, glycine, dopamine, and a wide selection of neuropeptides. Their processes are present in all layers of the IPL and they make extensive feedback contact with bipolar cells, serial contacts with other amacrine cells, and feed-forward contacts with ganglion cells. This makes for an inordinately complicated series of nested inhibitory circuits. In all species, GABAA receptors are expressed postsynaptically on amacrine cells and ganglion cells. GABAB receptors are located presynaptically on amacrine cells and postsynaptically on ganglion cells (Figure 1). GABAC receptors are found on ganglion cells in salamander, but apparently not in other species, and only rarely on amacrine cells. GABAA receptors on subtypes of GABA-receptive amacrine and ganglion cells are characterized by specific subunit compositions. Cholinergic amacrine cells are the only retinal neurons shown, as yet, to express the d subunit, the presence of which should eliminate BZD sensitivity. The various a subunits aggregate at different synaptic sites. For example, the a2 subunit is found on cholinergic amacrine cells, while the a4 subunit is found on ganglion cells. Also, for the most part, GABAA receptors in the inner retina are enhanced by barbiturates and BZDs, indicating the presence of the g subunit. In general, the effect of GABAA stimulation of amacrine cells is to suppress the transient component of the light response. The application of GABA tends to suppress the spontaneous activity and light responses of ganglion cells without having a great effect on the center-surround organization. However, GABAA antagonists abolish directional and orientation selectivity in ganglion cells, presumably by interacting with GABAA receptors on bipolar cells and other amacrine cells.

In general, the activation of GABAB receptors on amacrine cells and ganglion cells makes the light response more transient and reduces spike frequency in ganglion cells. In rabbit retina, baclofen facilitates the light-evoked release of ACh. GABAB receptors are found at limited but discrete locations on the dendrites of starburst and other types of amacrine cells. A feedback circuit involving the disinhibition of glycine onto starburst amacrine cells could account for the facilitatory effect of baclofen. GABAergic modulation of ganglion cell activity involves a complicated interaction of the subtypes of GABAA, GABAC, and GABAB receptors that are present ON bipolar and amacrine cells and act on chloride channels and a variety of calcium channels. While GABAA influence is rapid in onset and offset and likely participates in phasic inhibition, GABAB and GABAC influences are slower and likely participate in tonic inhibition. As GABAB receptors are less sensitive to GABA than GABAA or GABAC receptors, their influence may be more prominent during excessive GABA release that could occur with intense stimulation or excitotoxic release of glutamate from bipolar cells.