Horizontal cells of the turtle retina. I. Light microscopy of golgi preparations

In Golgi preparations of turtle retina, four types of horizontal cells were observed and their morphological characteristics determined in vertical thick sections, whole mount preparations, and reconstructions from serial 1-μm sections. H1 consists of a nucleated, stellate cell body (H1CB) and an irregular, tuberous axon terminal (H1AT) connected by a slender axon. Both parts of these cells make contact with receptor cells. H1CB's appear to correspond to “L2-type cells” while H1AT's correspond to “L1-type cells” described in the physiological literature. H2 and H3 are axonless stellate cells which are similar to one another in vertical profile and may occasionally appear similar in horizontal view. In general, the dendritic tree is more densely branched and the density of receptor cell contacts is higher for H2 than for H3. H2-type cells may correspond to “R/G C-type cells.” H4 is also an axonless stellate cell type which is smaller than H2 or H3 at equivalent retinal locations. The dendritic fields of H1CB's vary widely, but systematically, in size and shape over the retina. Their size is inversely related to receptor cell density, and the shape of the dendritic tree varies from roughly circular in the central area to elliptical in the periphery of the retina.     

Interaction between the horizontal cells of the turtle retina]

The interaction between horizontal cells in the turtle retina was tested by means of two microelectrodes, polarizing and recording ones impaling two cells at different distances between each other. The direct electrical coupling is shown to exist between L-cells of the same type (type I - with big receptive fields and type II - with small receptive fields). The value of this coupling changes with the conditions of illumination as well as with the level of the membrane potential. This can be accounted for by the known properties of subsynaptic and nonsynaptic membranes of the horizontal cells. There is no direct electrical coupling between L-cells of different types. However strong hyperpolarization of L-cells of type I by extrinsic current or by a light annulus evokes a depolarization in L-cells of type II. This indirect interaction between L-cells also dependent on the conditions of illumination may be explained by a mechanism of feedback between the horizontal cells and photoreceptors. Polarization of L-cells of both types has no effect on the cells of C-type.     

Color-specific interconnections of cones and horizontal cells in the retina of the goldfish

In Golgi preparations of goldfish retina we have observed three types of horizontal cell which receive exclusively from cones and one which receives exclusively from rods. The cone horizontal cells were designated H1, H2 and H3, in order of increasing dendritic spread, increasing separation from the outer synaptic layer, decreasing size of perikaryon, and decreasing density of cone contacts. Slender appendages with knobby terminal enlargements project horizontally from the perikarya and larger dendrites of both rod and cone horizontal cells. We determined patterns of cone inputs to Golgi-impregnated horizontal cells by analyzing serial 1 μm sections with the light microscope. The probable inputs, in terms of visual pigments in the cones which contact them, are: H1, red + green + blue; H2, green + blue; H3, blue. Analysis of previously published work suggests (1) that H1 cells generate monophasic or L-type responses, H2 cells generate biphasic or C₁-type responses, and H3 cells generate triphasic or C₂-type responses; (2) that H1 cells receive direct functional input at least from red-sensitive cones, H2 cells from green-sensitive cones, and H3 cells from blue-sensitive cones; and (3) that H1 cells constitute pathways from cones to H2 cells, and H2 cells constitute pathways from cones and H1 cells to H3 cells. The precise location and route of the transfers, from H1 to H2 and from H2 to H3, are not yet known.     

Retinal structure in the smooth dogfish, Mustelus canis: light microscopy of photoreceptor and horizontal cells

Photoreceptor and horizontal cells in retinas of the smooth dogfish were impregnated by the rapid Golgi method. Cones as well as rods are present; there are hundreds of rods per cone. The rod synaptic endings are small spherules which send out basal processes (telodendria). The cone synaptic endings are large pedicles which also extend telodendria. Horizontal cells of three distinctive varieties are segregated in vertically separate layers. Horizontal cells of the first (external) row are thick and cuboidal. They send out processes which enter into the invaginations in rod spherules. These cells and their processes cover a circular to elliptical field measuring 75–125 by 125–200 μm. They probably contact every rod in this field. Horizontal cells of the second (intermediate) row are flattened and stellate. Their processes also enter into the rod spherules. These cells cover a circular to elliptical field slightly larger than that covered by cells of the first row. Although they appear not to contact every rod in their field, the fields of adjacent cells of this type overlap. Every rod, therefore, probably contacts one horizontal cell of the second row as well as one of the first row. Horizontal cells of the third (internal) row are stellate, with a few long cylindrical horizontal processes. Finer vertical processes go from these to contact cone pedicles. Probably each cone contacts one such process, and each third-row horizontal cell contacts on the order of ten cones. Although somewhat uncertain, the size, shape, and degree of overlap of the fields covered by third-row cells are probably not very different from those of first- and second-row cells. The vertical sequence of horizontal cells which contact rods or cones in dogfish is inverted from that in teleosts.     

Electron microscopic study of synaptic contacts between photoreceptors and HRP-filled horizontal cells in the turtle retina

The synaptic contacts between photoreceptors and horizontal cells in the retina of the turtle (Geoclemys) were studied. Horizontal cells were classified into three types according to their intracellularly recorded spectral responses: luminosity, biphasic chromaticity, and triphasic chromaticity horizontal cells (LHC, BHC, and THC). These cells were then iontophoretically filled with horseradish peroxidase (HRP). The various types of photoreceptors located within the dendritic field of the HRP-filled horizontal cells were identified either as rods or as one of the three chromatic types of cones; the latter were identified by the presence and colors of their oil droplets in the inner segments. The synaptic contacts between photoreceptors and labeled horizontal cells were then investigated by light and electron microscopy of serial sections on three LHCs, two BHCs, and one THC. LHCs made synaptic contacts with about 100 photoreceptors, including rods and three chromatic types of cones; two-thirds of these photoreceptors contacted the cell body and the remaining its axon terminal. BHCs contacted about 30 cones; red-, green-, and blue-sensitive cones in the ratio of 3:4:1. THC contacted 20 cones; red- and blue-sensitive cones in the ratio of 2:1. The dendritic processes of HRP-filled horizontal cells were found as the lateral element of the ribbon synaptic complexes. The present finding suggests that the responses of BHC and THC to red flashes are fed directly from red-sensitive cones in addition to the feedback pathway of cone-horizontal cell connections in the previous studies.     

Local nonuniformities in the syncytium of the horizontal cells of the retina in the turtle

Electrical coupling between horizontal cells of the turtle retina was investigated by means of two microelectrodes (current and recording ones) penetrating neighbouring cells at a fixed distance from each other. The morphological coupling was revealed by means of fluorescent dye Lucifer Yellow. The electrical coupling was confirmed between elements of similar type (L1--axonal terminals, or L2--cell bodies, or R/G type cells) and no coupling was found between elements of different types, though L1 and L2 are directly connected through thin axons. In the L1 syncytium the electrical coupling at small (less than or equal to 50 microns) but fixed distances between microelectrodes could differ several times depending on the minimal displacement of microelectrodes. This local nonuniformity of coupling can be explained on the basis of structural nonuniformities in the L1 (axon terminal) network. It is unlikely however that the structural nonuniformities can influence the functional properties of horizontal cell network when the retina is stimulated adequately (by light).     

Dopaminergic mechanisms underlying the reduction of electrical coupling between horizontal cells of the turtle retina induced by d-amphetamine, bicuculline, and veratridine

Previous studies have shown that dopamine, bicuculline, or d-amphetamine reduce the electrical and dye-coupling between the axon terminals of the horizontal cells of the turtle retina (see Piccolino et al., 1984). In the present study we observed similar effects following the application of veratridine. The actions of all these drugs were prevented by dopamine antagonists acting on D1 receptors such as flupenthixol and SCH 23390. However, in contrast to dopamine, the actions of d-amphetamine, bicuculline, and veratridine were attenuated or abolished by pharmacological agents (such as 6-OH-dopamine, alpha-methyl-p-tyrosine, or reserpine) known to reduce the release of dopamine from dopaminergic neurons. Moreover, the actions of veratridine and bicuculline were prevented by tetrodotoxin, indicating that one or more neurons in the dopamine pathway are spike-generating. We conclude that d-amphetamine, bicuculline, and veratridine reduce electrical coupling between the axon terminals of the turtle horizontal cells by promoting the release of endogenous dopamine from the dopaminergic amacrine cells previously identified (Witkovsky et al., 1984). Electron-microscopic observations revealed that 6-OH-dopamine selectively attacked this population of amacrine cells. No degenerating terminals were found adjacent to the horizontal cell axon terminals. On this basis, we postulate that dopamine reaches the horizontal cell by diffusion through the extracellular space.     

Modulation of cone horizontal cell activity in the teleost fish retina. II. Role of interplexiform cells and dopamine in regulating light responsiveness

Following the destruction of the terminals of the dopaminergic interplexiform cells by intraocular injections of 6-hydroxydopamine (6-OHDA), cone horizontal cells exhibited high light responsiveness in prolonged darkness and their responses to moderate and bright full-field flashes were as large as 60 mV. Furthermore, the light responsiveness of these cells in the 6-OHDA-treated retinas was not enhanced by background illumination. The application of dopamine (50 microM) by superfusion to 6-OHDA-treated retinas resulted in a decrease in light responsiveness and changes in response waveform of the cone horizontal cells. Twenty minutes following dopamine application the responses of the cone horizontal cells closely resembled the response of cells recorded in prolonged dark-adapted retinas. Dopamine caused similar changes in cone horizontal cells recorded in light-exposed retinas, but had no obvious effects on rod horizontal cells. The selective dopamine D1 receptor antagonist, Sch 23390, enhanced cone horizontal cell responsiveness when applied to prolonged dark-adapted retinas, mimicking background illumination. The light responsiveness of cone horizontal cells recorded after application of Sch 23390 was less than that for cells in retinas that had been exposed to background lights, but light responsiveness could not be further enhanced by background illumination. Another dopamine antagonist, (+)-butaclamol, was found to have effects similar to Sch 23390 on cone horizontal cells, but (-)-butaclamol, the inactive enantiomer, did not enhance the light responsiveness of these cells. The results suggest that the dopaminergic interplexiform cells play a crucial role in the regulation of cone horizontal cell responsiveness by prolonged darkness and background illumination. These cells may release dopamine tonically in the dark, which suppresses cone horizontal cell responsiveness. Background illumination may decrease dopamine release and liberate cone horizontal cells from the suppression.     

Horizontal cells of the pigeon retina

Two types of horizontal cells are seen in Golgi-impregnated retinas of the pigeon. Type I horizontal cells are compact, “brush-shaped,” and have an axon ending as an irregular spinous arborization. The majority of the dendrites terminate in the distal part of the outer plexiform layer (OPL) as clusters which contact cones, but some terminate as single expansions in the proximal part of the OPL. The axon terminal spines are found only in the distal part of the OPL and contact both rods and cones. Pigeon Type I horizontal cells are Caja's “brush-shaped” cells, and their axon terminals resemble Caja's “stellate” cells. Type II horizontal cells have irregular, wavy, multi-branched dendrites, appear horizontally flattened, and lack axons. The dendrites terminate in the proximal part of the OPL as isolated spines and contact only cones. The Type II horizontal cells of the pigeon have not been previously described in the avian retina.     

Telodendrial contact of HRP-filled photoreceptors in the turtle retina: pathways of photoreceptor coupling

Synaptic contacts of photoreceptors in the turtle retina were studied by intracellular injection of horseradish peroxidase (HRP) and electron microscopy. Both cone and rod photoreceptors radiated basal processes (telodendria) from their terminal endings. These telodendria ran laterally in the outer plexiform layer. The telodendria of cones gave rise to many fine branches that penetrated synaptic cavities of several neighboring cones. Tips of these branches terminated near the walls of synaptic cavities. Some of the telodendrial contact formed two types of basal junction: symmetrical and punctate. The distribution of cones that made telodendrial contacts with the HRP-filled cone were quantitatively investigated. Green-sensitive cones (n = 3) made telodendrial contacts with neighboring red- and blue-sensitive cones, blue-sensitive cones (n = 4) with red- and green-sensitive cones, and red-sensitive cones (n = 9) with red- and green-sensitive cones. In contrast to these cone connections, rod telodendria did not penetrate neighboring photoreceptors. Direct synaptic contacts were not found between rods and cones. Our results clarify the variety of cone couplings in turtle retina: the three chromatic classes of cones are selectively coupled by the basal junctions at the ends of telodendrial processes.     

Short-term potentiation of off-responses in turtle horizontal cells.

Depolarizing responses to light off were studied in turtle horizontal cells using intracellular recording in the everted eyecup preparation. In many cells the off-response showed two components (fast and slow) which could overshoot beyond the steady-state dark level. The peak amplitudes of the fast and slow components increased with increasing duration of the light stimulus. A similar enhancement of the off-responses could also be produced by repetitive stimulation with brief flashes. However, the degree of enhancement produced by repetitive stimulation was greater than could be produced by increasing stimulus duration, and the latency of the onset of depolarization was longer, suggesting that the enhancement produced by repetitive stimulation involves an additional mechanism. Dramatic enhancement of the off-response by stimuli which did not affect the on-response during light indicates that the off-response may contain information not present in the on-response. The fast component of the off-response was suppressed to a greater degree than other components by reducing extracellular calcium or in the presence of 500 microM cobalt, suggesting that this component may involve a calcium current.     

Synaptic contacts between bipolar and photoreceptor cells in the retina of Callionymus lyra L

A combined light and electron microscopic study of Golgi-impregnated retinas of the marine teleost Callionymus lyra L. revealed mixed bipolar cells (M types) contacting rods and cones and pure cone bipolar cells (C types). Five types of mixed bipolar cells can be differentiated on the basis of their synaptic contacts. Two out of the five mixed bipolar cell types contact double cones, single cones, and rods (mixed, dark, pale, single [Mdps and midget-Mdps]). Their endbuds make narrow cleft junctions, with each type of photoreceptor, and in addition, two endbuds end centrally in the synaptic ribbon complexes of the dark and pale double-cone pedicles. Three types of mixed bipolar cells contact only double cones and rods. The endbuds of one type (mixed, dark, pale, ribbon [Mdpr]) end centrally in the synaptic ribbon complexes of the dark and pale double-cone pedicles as well as of the rod spherules. The endbuds of two types (Mdp and midget-Mdp) make wide cleft junctions in dark and pale double-cone pedicles and in rod spherules. All pure cone bipolar cell types contact cones exclusively with narrow cleft junctions. Four types are seen: a type that contacts predominantly pale double-cone pedicles but also a few dark double-cone pedicles (Cp), a type that is connected with dark and pale double-cone pedicles in about equal numbers (Cdp), a type that makes synaptic contacts with pale double-cone pedicles and single-cone pedicles (Cps), and a type that is connected with both types of double cones and to single-cone pedicles (Cdps). A resemblance between the ultrastructural features of mixed bipolar cell synapses in Callionymus and in Carassius auratus is noted.     

Horizontal cells in the retina of the rabbit

The light responses, morphology, and connections of horizontal cells (HCs) were studied in the retina of the rabbit using intracellular recordings and the injection of visible markers. Two types of HCs were identified, axonless and axon-bearing HCs. Axonless HCs and the somatic end of axon-bearing HCs respond to white light of varying intensity with graded hyperpolarizations; both display a transient superimposed on the sustained hyperpolarization at stimulus initiation and a small rod aftereffect at the cessation of high intensity stimuli. Anatomically, both are connected to cones, but their responses also suggest rod influence. Both summate stimuli from a retinal area which is much larger than their respective fields. However, only axonless HCs transfer a fluorescent, low molecular weight dye to adjoining, homologous cells. The axon terminal of axon-bearing HCs has response properties different from those of the cell body: the transient at stimulus initiation is absent; furthermore, at high levels of illumination, the rod aftereffect becomes equal in amplitude to the primary hyperpolarization. Anatomically, it is connected to rods, but its responses also suggest cone influence. Its receptive field approximates in diameter its anatomical spread and it does not transfer fluorescent dye to its neighbors.     

Decrease of gap junction permeability induced by dopamine and cyclic adenosine 3':5'-monophosphate in horizontal cells of turtle retina

The axon terminals of the H1 horizontal cells of the turtle retina are electrically coupled by extensive gap junctions. Dopamine (10 nM to 10 microM) induces a narrowing of the receptive field profile of the H1 horizontal cell axon terminals, increases the coupling resistance between them, and decreases the diffusion of the dye Lucifer Yellow in the network formed by the coupled axon terminals. These actions of dopamine involve the activation of D1 receptors located on the membrane of the H1 horizontal cell axon terminals proper. Increases of the intracellular cyclic AMP concentration induced by either stimulating the adenylate cyclase activity with forskolin or inhibiting the phosphodiesterase activity with isobutylmethylxanthine, theophylline, aminophylline, or compound RO 20-1724 elicit effects similar to those of dopamine on the receptive field profile of the H1 horizontal cell axon terminals, on their coupling resistance, and on the diffusion of Lucifer Yellow in the axon terminal network. It is concluded that dopamine decreases the permeability of the gap junctions between the axon terminals of the H1 horizontal cells of the turtle retina and that this action probably involves cyclic AMP as a second messenger.     

Dark- and light-induced changes in coupling between horizontal cells in mammalian retina

Retinal horizontal cells exhibit large receptive fields derived from their extensive electrical coupling by means of gap junctions. The conductance of these gap junctions seems to be regulated by dopamine acting through a cAMP-mediated cascade. There is now abundant evidence that extracellular dopamine levels vary with changes in ambient light intensity, suggesting that changes in the dark/light adaptational state of the retina can modulate coupling between horizontal cells. We studied this question in the mammalian retina by determining the effects of ambient light levels, in the form of changing background light intensity, on the coupling profiles of A- and B-type horizontal cells in the rabbit. Changes in coupling were assessed by measurements of the space constants of the syncytium formed by horizontal cells and the intercellular spread of the biotinylated tracer Neurobiotin. Our results indicate that dark-adapted horizontal cells show relatively weak coupling. However, presentation of background lights as dim as one-quarter log unit above rod threshold resulted in increases in both the averaged extent of tracer coupling and space constants of A- and B-type horizontal cells. Coupling expanded further as background light intensities were increased by 1-1.5 log units, after which additional light adaptation brought about an uncoupling of cells. Coupling reached its minimum at light intensities about 3 log units above rod threshold, after which, with further light adaptation, it stabilized at levels close to those seen in dark-adapted retinas. Our results indicate that electrical coupling between mammalian horizontal cells is modulated dramatically by changes in the adaptational state of the retina: coupling is maximized under dim ambient light conditions and diminishes as the retina is dark or light adapted from this level.     

Heterogeneity of horizontal cells in the chicken retina

Despite numerous reports that different markers are expressed by horizontal cells in the avian retina, it remains unknown whether different types of horizontal cells can be defined by differences in their immunocytochemical profiles. The purpose of this study was to rectify this deficiency. We identified horizontal cells by indirect immunofluorescence with antibodies to calretinin, trkA, GABA, Prox1, AP2alpha, Pax6, islet1, and Lim1 + 2. We found two major groups of horizontal cells, those that express trkA and those that express calretinin. The trkA-immunoreactive (-IR) horizontal cells had small, round somata and robust, bulbous dendritic endings, whereas calretinin-IR horizontal cells had large, polygonal cell bodies and fine, diffuse dendritic endings, both contacting the calbindin-IR pedicles of double cones. Weak gamma-aminobutyric acid (GABA) immunoreactivity was observed only in a few of the trkA-IR horizontal cells, whereas the overlap of calretinin and GABA immunoreactivities was 100%. The majority of trkA-IR horizontal cells expressed islet1, and the majority of calretinin-IR horizontal cells expressed Lim1 + 2, AP2alpha, and Pax6. Islet1 immunoreactivity was observed in a small fraction of calretinin-IR/non-trkA-IR cells. In agreement with previous reports, we detected Prox1 immunoreactivity in all types of horizontal cells. These immunolabeling profiles suggest that there are four immunochemically distinct subtypes of horizontal cells in the postnatal chick retina, which may match the four types that have been observed in Golgi-impregnated pigeon and turtle retinas.