The neurotrophins are trophic and mitogenic factors critical for the development of specific classes of neurons in the central and peripheral nervous systems. In the retina, BDNF and NT-3 have been shown to promote the survival of differentiated ganglion cells (Rodriguez-Tebar et al., 1989; De La Rosa et al., 1994). NT-3 has also been demonstrated to support the survival of amacrine cells and facilitates the differentiation of retinal neurons in culture (De La Rosa et al., 1994). Here, we examine immunohistochemically the expression of BDNF and NT-3 proteins, their cognate receptors, trk B and trk C, respectively, and the p75 neurotrophin receptor in the developing chick retina. At E8, the earliest stage of retinal development examined, all of these proteins exhibit diffuse expression throughout the width of the retina, with the strongest reactivity in the innermost layers. A gradual restriction in expression to ganglion cells and amacrine cells, the staining of which is most prominent at E15, is followed by a downregulation of expression with the strongest immunoreactivity persisting in the ganglion cell layer. Overlapping patterns of expression throughout embryonic development indicate a colocalization of the neurotrophins and their receptors, although NT-3 and p75 alone are present in the inner plexiform layer and only p75 is observed in the outer plexiform layer. Although some of the immunoreactivity for BDNF, NT-3, and their receptors in retina may reflect trophic mechanisms operating in association with the optic tectum and isthmo-optic nucleus, the colocalization of ligands and receptors in retina strengthens the assertion that these neurotrophins function locally during development.

There is evidence that neurons in medial superior temporal area (MST) respond to rotation in depth of textured planes. MST neurons project to the ventral intraparietal area (VIP) and the question arises whether VIP neurons are responsive to rotation in depth as well. In the present study on awake monkeys, we have simulated movement of a flat board, covered with dots, by a computer. The two-dimensional images corresponded to the projection of structured planes rotating around a fronto-parallel axis. In the literature this stimulus is called fanning. Fanning effectively induced responses in VIP neurons. Most often the responses were nearly as strong as for translation, expansion/contraction, or rotation, indicating that there was no special sensitivity for rotation in depth. For neurons, sensitive to expansion, the response to fanning could often be explained by the positioning of the expanding part of the fanning stimulus over the area which was most responsive to expansion. For neurons which were direction selective to translation, the optimal direction of fanning was usually the same as the preferred direction for translation. It is concluded that VIP neurons may be sensitive to movement of structured planes but they are not specialized for the detection of such movement.

Inside-out patches containing cGMP-gated channels were excised from catfish rod or cone outer segments and held under voltage clamp. The net cGMP-dependent currents elicited by saturating and subsaturating concentrations of cGMP at ±30 mV were measured and the dependence of current upon cGMP concentration was determined. The apparent affinity of the channel for its ligand was estimated by fitting these data with the Hill equation. The concentration of cGMP required to give half the maximum current (K1/2) in rod and cone channels at +30 mV was ~28 μM and ~37 μM, respectively, and was weakly voltage dependent. Thus, cone channels have an intrinsically higher K1/2 than rod channels. For both types of channel, the Hill coefficient was ~2.3. In the presence of calcium-calmodulin, the apparent affinity of the rod channel for cGMP decreased by about twofold, but the apparent affinity of the cone channels was unaffected. These results indicate that the open probability of the cone channel for its ligand cannot be modulated by calmodulin. This represents the first significant departure between rod and cone photoreceptors in mechanisms used by phototransduction and suggests that the β subunit of the cone channel must be different from that of the rod channel.

Previous tests of the linearity of spatiotemporal summation in cat simple cells have compared the responses to moving sinusoidal gratings and to gratings whose contrast was modulated sinusoidally in time. In particular, since a moving grating can be expressed as a sum of modulated gratings, the response to a moving grating should be predictable (assuming linearity) from the responses to modulated gratings. However, these simple linear predictions have shown varying degrees of failure (e.g. Reid et al., 1987, 1991), depending on the directional selectivity of the neurons (Tolhurst & Dean, 1991). We demonstrate here that the failures of these linear predictions are, in fact, explained by the contrast-normalization model of Heeger (1993). We concentrate on the ratio of the measured to predicted moving grating responses. In the context of the contrast-normalization model, calculating this ratio turns out to be particularly appropriate, since the ratio is independent of the precise details of the linear front-end mechanisms ultimately responsible for directional selectivity. Hence, the contrast-normalization model can be compared quantitatively with this ratio measure, by varying only one free parameter. When account is taken both of the expansive output nonlinearity and of contrast normalization, the directional selectivity of simple cells seems to be dependent only on linear spatiotemporal filtering.

There is increasing evidence that GABAC receptors are composed of GABA ρ subunits. In this study, we compared the properties of native GABAC receptors with those of receptors composed of a GABA ρ subunit. A homologue of the GABA ρ gene was cloned from a white perch (Roccus americana) retinal cDNA library. The clone (perch-s) has an open reading frame of 1422 nucleotide base pairs and encodes a predicted protein of 473 amino acids. It is highly homologous to GABA ρ subunits cloned from human and rat retinas. The receptors (perch-s receptor) expressed by this gene in Xenopus oocytes show properties similar to those of the GABAC receptors present on white perch retinal neurons. GABA induced a sustained response that had a reversal potential of –27.1 +minus; 3.6 mV. The EC50 for the response was 1.74 +− 1.25 μM, a value similar to that reported for GABAC receptors. Pharmacologically, the responses were bicuculline insensitive and not modulated by either diazepam or pentobarbital as is the case for GABAc receptors. There were, however, some distinct differences between native GABAc and perch-s receptors. I4AA acts as a partial agonist on perch-s receptors whereas it is strictly an antagonist on native GABAC receptors. Picrotoxin inhibition is noncompetitive on perch-s receptors, but both competitive and noncompetitive on GABAC receptors. We conclude that GABAC receptors are formed by GABA p subunits but that native GABAc receptors probably consist of a mixture of GABA ρ subunits.

To study the relationship between ocular dominance columns (ODCs) and axonal projections of individual layer 6 pyramidal neurons in the primary visual cortex, neurons were intracellularly labeled with biocytin in live slices prepared from macaque monkeys that had received an intravitreal injection of tetrodotoxin (TTX). The TTX injection indirectly causes a decrease in cytochrome oxidase (CO) expression in the cortical ODCs corresponding to the treated eye (Wong-Riley & Carroll, 1984). Sections from slices with labeled layer 6 neurons were double stained for biocytin and CO, to allow visualization of neuronal processes as well as ODCs. Twenty-seven layer 6 pyramidal neurons in ODC-labeled slices were analyzed. These neurons were classified according to the criteria of Wiser and Callaway (1996). Eight of these are class I neurons, which have dense axonal projections to the monocular layer 4C. The remaining 19 are class II neurons which project primarily to the binocular layers outside 4C. Among class I neurons, two have dense axonal arbors in layer 4Cα (type Iα), one in layer 4Cβ (type Iβ), and two throughout the depth of layer 4C (type IC). None of these neurons have ODC-specific axonal arbors. The remaining three class I neurons have focused axonal projections in layers 4Cβ and 4A (type IβA). All three appear to have axonal arbors predominantly within their home ODC in layer 4C. The axonal arbors of class II neurons do not appear to relate to ODCs in any specific fashion.

These experiments were designed to test the idea that the optic layer in the tree shrew, Tupaia belongeri, is functionally distinct and provides a link between the visuosensory superficial and the premotor intermediate layers of the superior colliculus. First, cells in the optic layer were intracellularly labeled with biocytin in living brain slices. Compared to cells in the adjacent lower part of the superficial gray layer, which have apical dendrites that ascend toward the tectal surface, optic layer cells have dendritic fields that are restricted for the most part to the optic layer itself. The differences in dendritic-field location imply that superficial gray and optic layer cells have different patterns of input. The axons of optic layer cells terminate densely within the optic layer and, in addition, project in a horizontally restricted fashion to the overlying superficial gray and subjacent intermediate gray layers. This pattern also is different from the predominantly descending interlaminar projections of lower superficial gray layer cells. Next cells in the intermediate gray layer were labeled in order to examine the relationships between optic layer cells and these subjacent neurons that project from the superior colliculus to oculomotor centers of the brain stem Neurons in the upper part of the intermediate gray layer send apical dendrites into the optic layer and therefore can receive signals from the superficial gray layer either directly, from descending axons of lower superficial gray layer cells, or indirectly, through intervening optic layer cells. In contrast, lower intermediate gray layer cells have more radiate dendritic fields that are restricted to the intermediate gray layer. Thus, these lower cells must depend on descending projections from optic or upper intermediate gray layer cells for signals from the superficial gray layer. Together, these results support the idea that the optic layer is a distinct lamina that provides a link between the superficial and intermediate gray layers. They also are consistent with the traditional view that descending intracollicular projections play a role in the selection of visual targets for saccades.

Biochemical studies provide evidence that the pathway from visual cortex to the superior colliculus (SC) utilizes glutamate as a neurotransmitter. In the present study, we have used immunocytochemistry, visual cortex lesions, and retrograde tracing to show directly by anatomical methods that glutamate or a closely related analog is contained in corticocollicular neurons and terminals. A monoclonal antibody directed against gamma-L-glutamyl-L-glutamate (gamma glu glu) was used to localize glutamate-like immunoreactivity in both the superior colliculus (SC) and visual cortex (VC). Unilateral lesions of areas 17–18 were made in four cats to determine if gamma glu glu labeling was reduced in SC by this lesion. WGA-HRP was injected into the SC of 10 additional cats in order to determine if corticocollicular neurons were also labeled by the gamma glu glu antibody. A distinctive dense band of gamma glu glu immunoreactivity was found within the deep superficial gray and upper optic layers of SC where many corticotectal axons are known to terminate. Both fibers and cells were labeled within the band. Immunoreactivity was also found in cells and fibers throughout the deep layers of SC. Measures of total immunoreactivity (i.e. optical density) in the dense band were made in sections from the SC both ipsilateral to and contralateral to the lesions of areas 17–18. A consistent reduction in optical density was found in both the neuropil and in cells within the dense band of the SC ipsilateral to the lesion. A large percentage of all corticocollicular neurons that were retrogradely labeled by WGA-HRP also contained gamma glu glu. These results provide further evidence that the corticocollicular pathway in mammals is glutamatergic. The results also suggest that visual cortex ablation alters synthesis or storage of glutamate within postsynaptic SC neurons, presumably as a result of partial deafferentation.

This study sought to characterize the tracer coupling of regenerated amacrine cells in the retina of the goldfish and assess the integration of regenerated neurons into existing retinal circuits. Regeneration of new neurons from injury-induced progenitors was stimulated by surgically excising a small rectangular piece of retina. Several months after regeneration was complete, intracellular injections of Neurobiotin, a gap junction-permeant tracer, were made into single regenerated amacrine cells or nonregenerated (extant) amacrine cells lying outside the regenerated patch. Two groups of amacrine cells were injected: those that in normal retina are tracer coupled and a single type (the radiate amacrine cell) that is not. The data show that regenerated amacrine cells are tracer coupled to each other and to their homologous counterparts outside the patch of regenerated retina. Regenerated radiate cells possess morphologically abnormal dendrites, but these processes can extend out of regenerated retina into surrounding normal retina. Similarly, the dendrites of extant radiate cells, severed by the original lesion, can regenerate into the patch of regenerated retina. These results indicate that in the goldfish retina the cell-specific junctional circuitry present in normal retina is re-created in the regenerated retina, and suggest that regenerated neurons are functionally integrated into the existing retina.

We recorded from single cells in area V2 of cynomolgus monkeys using standard acute recording techniques. After measuring each cell's spatial and temporal properties, we performed several tests of its chromatic properties using sine-wave gratings modulated around a mean gray background. Most cells behaved like neurons in area V1 and their responses were adequately described by a model that assumes a linear combination of cone signals. Unlike in V1, we found a subpopulation of cells whose activity was increased or inhibited by stimuli within a narrow range of color combinations. No particular color directions were preferentially represented. V2 cells showing color specificity, including cells showing narrow chromatic tuning, were present in any of the stripe compartments, as defined by cytochrome-oxidase (CO) staining. An addition of chromatic contrast facilitated the responses of most neurons to gratings with various luminance contrasts. Neurons in all three CO compartments gave significant responses to isoluminant gratings. Receptive-field properties of cells were generally similar for luminance and chromatically defined stimuli. We found only a small number of cells with a clearly identifiable double-opponent receptive-field organization.

A region of dorsomedial frontal cortex (DMFC) has been implicated in planning and executing saccadic eye movements; hence it has been referred to as a supplementary eye field (SEF). Recently, activity related to executing smooth-pursuit eye movements has been recorded from the DMFC, and microstimulation here has been shown to evoke smooth eye movements. This report documents neuronal activity present in smooth-pursuit tasks where the predictability of target motion was manipulated. The activity of many neurons in the DMFC reached a peak when a predictable change in target motion occurred. Furthermore, the peak activity of some cells was systematically shifted by manipulating the duration of the target event, indicating that the network these neurons were in could learn the temporal characteristics of new target motion. Finally, the activity of most neurons tested was greater when target motion was predictable than when it was unpredictable. The results suggest that the DMFC participates in planning smooth-pursuit eye movements based on past stimulus history.

The distribution of the carbohydrate epitope CD 15, a putative cell adhesion molecule, was studied in adult vertebrate retinas by light-microscopic immunohistochemistry. Except for Old World primates, in which no immunoreactivity was detectable, all other species expressed the epitope on retinal interneurones. Subpopulations of stratified amacrine cells were found in all species with the exception of bats and marmoset monkeys, and bipolar cells were immunoreactive in frogs and all amniotic animals. Ganglion cells were labelled in urodelian, in all sauromorphian, as well as in some mammalian species. In some species, the distribution of immunoreactive neurones was correlated to areas of retinal specialization such as the visual streak in frogs and the dorsotemporal field in birds. In these parts of the retina with enhanced visual acuity, more CD 15 glycosylated bipolar cells were found than in other parts. Among mammals, labelled bipolar cells were found exclusively in species with cone-dominated retinas. This comparative study shows that CD 15 expression is consistently membrane associated in morphologically defined subsets of amacrine, bipolar, and ganglion cells throughout the vertebrate phylum. Its distribution pattern was found to depend more on the visual behavior of a given species (cone-dominated or rod-dominated retina) than on phylogenetic proximity between species.

The integrative properties of starburst amacrine cells in the rabbit retina were studied with compartmental models and computer-simulation techniques. The anatomical basis for these simulations was provided by computer reconstructions of intracellularly stained starburst amacrine cells and published data on dendritic diameter and biophysical properties. Passive and active membrane properties were included to simulate spiking and nonspiking behavior. Simulated synaptic inputs into one or more compartments consisted of a bipolar-like conductance change with peak and steady-state components provided by the sum of two Gaussian responses. Simulated impulse generation was achieved by using a model of impulse generation that included five nonlinear channels (INa, ICa, Ia,. Ik. Ik.Ca). The magnitude of the sodium channel conductance change was altered to meet several different types of impulse generation and propagation behaviors. We studied a range of model constraints which included variations in membrane resistance (Rm) from 4,000 Ω.cm2 to 100,000 Ω.cm2, and dendritic diameter from 0.1 to 0.3 μm. In a separate series of simulations, we studied the feasibility of voltage-clamping starburst amacrine cells using a soma-applied, single-electrode voltage clamp, based on models with and without dendritic and somatic spiking behavior. Our simulation studies suggest that single dendrites of starburst amacrine cells can behave as independent functional subunits when the Rm is high, provided that one or a small number of dendrites is synaptically co-activated. However, as the number of co-activated dendrites increases, the starburst cell behavior becomes more uniform and independent dendritic function is less prevalent. The presence of impulse activity in the dendrites raises new questions about dendritic function. However, dendritic impulses do not necessarily eliminate independent dendritic function, because dendritic impulses commonly fail as they propagate toward the soma, where they contribute EPSP-like responses which summate with conventional synaptic currents.

In the rabbit retina, the nuclear dye, 4,6, diarnidino-2-phenylindole (DAPI), selectively labels a third type of amacrine cell, in addition to the previously characterized type a and type b cholinergic amacrine cells. In this study, these “DAPI-3” amacrine cells have been characterized with respect to their somatic distribution, dendritic morphology, and neurotransmitter content by combining intracellular injection of biotinylated tracers with wholemount immunocytochemistry. There are about 100,000 DAPI-3 amacrine cells in total, accounting for 2% of all amacrine cells in the rabbit retina, and their cell density ranges from about 130 cells/mm2 in far-peripheral retina to 770 cells/mm2 in the visual streak. The thin varicose dendrites of the DAPI-3 amacrine cells form a convoluted dendritic tree that is symmetrically bistratified in S1/S2 and S4 of the inner plexiform layer. Tracer coupling shows that the DAPI-3 amacrine cells have a fivefold dendritic-field overlap in each sublamina, with the gaps in the arborization of each cell being occupied by dendrites from neighboring cells. The DAPI-3 amacrine cells consistently show the strongest glycine immunoreactivity in the rabbit retina and they also accumulate exogenous [3H]-glycine to a high level. By contrast, the All amacrine cells, which are the best characterized glycinergic cells in the retina, are amongst the most weakly labelled of the glycine-immunopositive amacrine cells. The DAPI-3 amacrine cells costratify narrowly with the cholinergic amacrine cells and the On-Off direction-selective ganglion cells, suggesting that they may play an important role in movement detection.

We examined contrast, direction of motion, and concentration dependencies of the effects of GABAergic and cholinergic antagonists, and anticholinesterases on responses to movement of On—Off directionally selective (DS) ganglion cells of the rabbit's retina. The drugs tested were curare and hexamethonium bromide (cholinergic antagonists), physostigmine (anticholinesterase), and picrotoxin (GABAergic antagonist). They all reduced the cells' directional selectivity, while maintaining their preferred-null axis. However, cholinergic antagonists did not block directional selectivity completely even at saturating concentrations. The failure to eliminate directional selectivity was probably not due to an incomplete blockade of cholinergic receptors. In a extension of a Masland and Ames (1976) experiment, saturating concentrations of antagonists blocked the effects of exogenous acetylcholine or nicotine applied during synaptic blockade. Consequently, a noncholinergic pathway may be sufficient to account for at least some directional selectivity. This putative pathway interacts with the cholinergic pathway before spike generation, since physostigmine eliminated directional selectivity at contrasts lower than those saturating responses. This elimination apparently resulted from cholinergic-induced saturation, since reduction of contrast restored directional selectivity. Under picrotoxin, directional selectivity was lost in 33% of the cells regardless of contrast. However, 47% maintained their preferred direction despite saturating concentrations of picrotoxin, and 20% reversed the preferred and null directions. Therefore, models based solely on a GABAergic implementation of Barlow and Levick's asymmetric-inhibition model or solely on a cholinergic implementation of asymmetric-excitation models are not complete models of directional selectivity in the rabbit. We propose an alternate model for this retinal property.

The antibody H386F revealed microglia in the retinae of several species: owl monkey, slow loris, galago, ferret, raccoon, and tree shrew. The shape, size, and density of labeled microglia were identical to those labeled by OX-42 and OX-41, two antibodies specific for microglia, in both galago and owl monkey. The labeled microglia varied little in retinal location. There was remarkably little variability in density, shape, number, and size of the abeled microglia between species. All labeled microglia were evenly distributed across, but restricted to, the nerve liber layer. Possible reasons for this restriction in location are discussed.

Retinitis pigmentosa (RP) is an inherited disease that causes primary degeneration of rod photoreceptors in the retina. Although the causal gene (e.g. rhodopsin) is thought to be expressed in all rods across the retina, the degeneration is typically nonuniform, with rods in the far periphery surviving significantly longer than those in the midperiphery and macula. Basic fibroblast growth factor (bFGF) is a putative survival factor for photoreceptors, and the characteristic regional pattern of rod cell survival in RP suggested that bFGF might be distributed nonuniformly in the human retina. We performed double-label immunocytochemistry on 15 normal human retinas, using anti-bFGF and other antibody markers for retinal neurons and glia. Immunoreactivity for bFGF was consistently absent from cones but was present in rods, populations of cone bipolar and amacrine cells, Müller glial cells, and astrocytes. In the macula, the percentage of bFGF-reactive rods was very low (~0.5%) but it increased in a central to peripheral gradient, accounting for up to ~88% of the rods in the far periphery. These findings suggest that a central to peripheral gradient of rod bFGF is present in normal human retina and may influence the pattern of photoreceptor degeneration in RP. The absence of bFGF in cones and the low number of bFGF-positive rods in the macula may correlate with the vulnerability of these cells in RP, age-related macular degeneration, and other retinal diseases.

Previous experiments in chickens have shown that dopamine released from the retina may be one of the messengers controlling the growth of the underlying sclera. It is also possible, however, that the apparent relationship between dopamine and myopia is secondary and artifactual. We have done experiments to assess this hypothesis. Using High Pressure Liquid Chromatography with electrochemical detection (HPLC-ED), we have asked whether changes in dopamine metabolism are restricted to the local retinal regions in which myopia was locally induced. Furthermore, we have measured the concentrations of biogenic amines separately in different fundal layers (vitreous, retina, choroid, and sclera) to find out how changes induced by “deprivation” (= removal of high spatial frequencies from the retinal image by translucent eye occluders which produce “deprivation myopia”) are transmitted through these layers. Finally, we have repeated the deprivation experiments after intravitreal application of the irreversible dopamine re-uptake blocker reserpine to see how suppression of dopaminergic transmission affects these changes. We found that (1) Alterations in retinal dopamine metabolism were indeed restricted to the retinal areas in which myopia was induced. (2) The retina was the major source of dopamine release with a steep gradient both to the vitreal and choroidal side. Vitreal content was about one-tenth, choroidal content about one-third, and scleral content about one-twentieth of that of the retina. (3) There was a drop by about 40% in vitreal dopamine, DOPAC (3,4-dihydroxyphenylacetic acid) and HVA (homovanilic acid) concentrations following deprivation which occurred already at a time where little changes could yet be seen in their total retinal contents. (4) Choroidal and scleral dopamine levels were not affected by deprivation, indicating that other messengers must relay the information to the sclera. (5) A single intravitreal injection of reserpine lowered dopamine and HVA levels in retina and vitreous for at least 10 days in a dose-dependent fashion and diminished or suppressed further effects of deprivation on these compounds. DOPAC levels continued to change upon deprivation even after reserpine injection (Fig. 3). Our results suggest that the release rates of dopamine from retinal amacrine cells can be estimated from vitreal dopamine concentrations; furthermore, they are in line with the hypothesis that there is an inverse relationship between dopamine release and axial eye growth rates. Although our experiments do not ultimately prove that dopamine has a functional role in the visual control of eye growth, they are in line with this notion.

The question of whether the expression of mutant opsin predisposes the retina to light damage was addressed using transgenic mice that express rhodopsin with three point mutations near the N-terminus of the molecule. The mutations involve the substitution of histidine for proline at position 23 (P23H), glycine for valine at position 20 (V20G), and leucine for proline at position 27 (P27L). These mice express equal amounts of mutant and wild-type transcripts, and develop a progressive photoreceptor degeneration that is similar to that seen in human retinitis pigmentosa (RP). The P23H mutation is associated with the most frequently occurring form of human autosomal dominant retinitis pigmentosa (ADRP) in the United States. Transgenic and normal littermates were exposed to illuminance of 300 foot-candles (ft-c) for 24 h, then placed in darkness for either 6 h, 6 days, or 14 days. Histological and biochemical techniques were used to evaluate the outer retina in light-exposed and control animals reared on 12-h light/12-h dark cycle. The results indicate that light exposure accelerates the pathological changes associated with the transgene expression. Compared with transgenic animals reared in ambient cyclic light, retinas from light-exposed mice had a reduced rhodopsin content, fewer photoreceptor cell bodies, and less preservation of retinal structure. Data obtained from normal mice did not differ for the lighting regimens used. These findings suggest that the expression of VPP mutations in the opsin gene predisposes the transgenic photoreceptors to be more susceptible to light damage. The data also suggest that reducing photic exposure may be beneficial to any patient with RP mediated by an opsin mutation.

The entire population of ganglion cells in the retina of the toad Bufo marinus was labeled by retrograde transport of a lysine-fixable biotinylated dextran amine of 3000 molecular weight. Synaptic connections between bipolar, amacrine, and ganglion cells in the inner plexiform layer were quantitatively analyzed, with emphasis on synaptic inputs to labeled ganglion cell dendrites. Synapses onto ganglion cell dendrites comprised 47% of a total of 1234 identified synapses in the inner plexiform layer. Approximately half of the bipolar or amacrine cell synapses were directed onto ganglion cell dendrites, while the rest were made mainly onto amacrine cell dendrites. Most of the synaptic inputs to ganglion cell dendrites derived from amacrine cell dendrites (84%), with the rest from bipolar cell terminals. Synaptic inputs to ganglion cell dendrites were distributed relatively uniformly throughout all sublaminae of the inner plexiform layer. The present study provides unambiguous identification of ganglion cell dendrites including very fine processes, enabling a detailed analysis of the types and distribution of synaptic inputs from the bipolar and amacrine cell to the ganglion cells. The retrograde tracing technique used in the present study will prove to be a useful tool for identifying synaptic inputs to ganglion cell dendrites from neurochemically identified bipolar and amacrine cell types in the retina.