Search

Midget ganglion cells in the foveal slope, parafovea, near periphery and far periphery of human and monkey retinas have been studied by electron microscopy (EM). Five human foveal ganglion cells were reconstructed and found to share input from seven midget bipolar cells. The OFF center ganglion cells were in a one to one relationship with their midget bipolar cells. But the ON center cells received input from two to three midget bipolar cells, of which one was dominant in terms of numbers of ribbon synapses directed at the midget ganglion cell dendrites. In the human parafovea every midget ganglion cell received input from only one midget bipolar cell (previously published, Kolb and DeKorver, 1991). At 4 mm of eccentricity, the near peripheral ON midget ganglion cell received input from three midget bipolar cells and thus from three cones. In far peripheral retina (12 mm) the ON midget ganglion cell received input from three to four midget bipolar cells. The peripheral midget bipolar cells probably contacted three cones each: therefore between nine and 12 cones could have input to such midget ganglion cell. The relationship of the increasing dendritic field size and increasing convergence of cones to the midget ganglion cells with eccentricity from the fovea is discussed in terms of color processing and resolution.

We have examined amacrine cells that are calretinin-immunoreactive (-IR) in the macaque monkey retina with the aim of classifying them into morphological and functional subtypes. There are calretinin-IR cells in the fovea and throughout the retina. Their highest density is reached at 1.0 mm from the foveal pit (10,500 cells/mm2) and falls to 2,600/mm2 by 10 mm of eccentricity. Nearest-neighbor statistics for the calretinin-IR cell body distribution indicate a nonregular pattern, with a regularity index of 1.4–1.6. There is an increase or “bump” of cell density 3.5–4.0 mm from the foveal pit, corresponding to the rod photoreceptor density peak. Based on morphological differences, there appear to be three types of amacrine cell that are calretinin-IR. To determine the types, we doubly immunolabeled retinas, from fovea to periphery, for calretinin-IR in combination with other calcium binding proteins and inhibitory amino acid neurotransmitters. Labeling with parvalbumin and calretinin antibodies indicated that 70% of the amacrine cells were solely calretinin-IR, and 30% contained parvalbumin-IR as well. In the same way, 70% of the calretinin-IR amacrine cells colocalized calbindin, but 30% were only calretinin-IR. Among the calretinin/calbindin-colocalized cells, there were small-field and wide-field types. Double labeling with antibodies to calretinin and γ-aminobutyric acid (GABA) and to calretinin and glycine revealed the majority to be glycine-IR, but some were GABA-IR. The glycine-IR population consists mainly of AII amacrine cell types, but clearly another non-AII type is involved. The non-AII glycine-IR population resembles a small- to medium-field diffuse type. The calretinin-IR wide-field type is GABAergic and corresponds to an A19 type. The central, rod-free, fovea contains the calretinin-IR, non-AII glycine-IR type and the calretinin-IR, GABAergic type only. To learn more concerning the circuitry of the calretinin/glycine-IR, non-AII amacrine cell type in isolation from AII amacrine cells, we concentrated on the rod-free fovea, where AII amacrine cells are absent. We performed a serial section electron microscopy (EM) study on four calretinin-IR cells. They were involved with cone pathway circuitry. They got input from ON and OFF midget bipolar cells, reciprocated synapses to these bipolar cells, and provided synapses to ON-center ganglion cells. Thus we have obtained new information on a cone pathway amacrine cell of the central monkey fovea that is involved in the midget system. National Institute of Health; Grant number EY03323 (H.K.); Research to Prevent Blindness grant (Department of Ophthalmology, University of Utah School of Medicine).

The distribution of endothelial nitric oxide synthase immunoreactivity (eNOS-IR) was investigated in the retinas of all phylogenetic vertebrate classes by using a monoclonal eNOS antibody. Confocal light microscopy showed immunoreactive labeling in Müller cells of fish, frog, salamander, turtle, chicken, rat, ground squirrel, and monkey retina. In vascularized retinas (rat, monkey), astrocytes and some blood vessels were also stained. Furthermore, eNOS-IR was localized to axon terminals of turtle and fish horizontal cells. These observations are the first to show the presence of eNOS-IR in Muller glia and horizontal cell structures of the vertebrate retina. Supported by grants EY03323 and EY04855 and a grant from Research to Prevent Blindness to the Department of Ophthalmology, Moran Eye Center.

We reported previously that an antibody to the α isoform of protein kinase C (PKC) immunostained rod bipolar cells and bipolar cells that could be blue-cone (B-cone)-specific in postmortem human retina (Kolb et al. (1993) Vis. Neurosci. 10:341–351). In addition, we showed that antibodies to the β isoform of PKC immunostained cone system bipolar, amacrine, and ganglion cells. Since the fixation of the human material was poor, we were unable to make positive identifications of the specific cell types that were immunoreactive, particularly in the case of PKC-β antibodies. Thus, herein we have repeated the study on well-fixed monkey foveal retina. PKC-α immunoreactivity (IR) was restricted to a single type of cone bipolar cell that contacted only a minority of the cone pedicles at central invaginating contacts of ribbon triads. This bipolar type shares some morphological characteristics of B-cone-specific bipolar cells of primate retina. PKC-β immunoreactivity was found in cone bipolar cells that made primarily basal contacts with cone pedicles and had axon terminals in sublamina a of the inner plexiform layer (IPL). Immunoreactivity also occurred in a type of cone bipolar that made central element contacts and had axon terminals in sublamina b of the IPL. Some ganglion cells, particularly those postsynaptic to flat midget bipolar cells also exhibited PKC-β-IR. One type of amacrine with an 8 μm diameter cell body showed strong PKC-β-IR. It was postsynaptic to cone bipolar cells in both sublamina a and b and presynaptic to bipolar axons, other immunoreactive amacrine cells, and ganglion cell dendrites and bodies. The other amacrine cell type showed less strong PKC-β-IR, large-bodied (12–15 μ cell body diameter), and probably diffuse in branching pattern. The latter interacted with the intensely immunoreactive amacrines, bipolars, and ganglion cells. By comparison to cat and primate retinas where morphology and physiology of many retinal neurons are well documented, we suggest that PKC-β may be specific to flat midget, flat diffuse, and invaginating diffuse cone bipolar cells and to at least two amacrine cells. Some of these neural types are proposed to be involved in OFF-center cone pathways in the monkey retina. Microsc Res Tech 36:57–75, 1997. © 1997 Wiley-Liss, Inc.

Intracellular recording and horseradish peroxidase (HRP) staining of amacrine cells in the isolated arterially perfused cat retina have revealed examples of small-field cells that hyperpolarize to light. Two were examined in detailed electron microscopic reconstructions to determine patterns of synaptic relationships within the inner plexiform layer (IPL). The cells were morphologically similar to A8 and A13 types as described in Golgi-impregnated material (Kolb et al. [1981] Vision Res. 21:1081-1114). Both types received ribbon synaptic input from rod and cone bipolar cells. The latter input was numerically predominant, occurred in both a and b sublaminae of the IPL, and arose from at least three cone bipolar types. Reciprocal synapses were evident between A13 cells and cone bipolar cells. Amacrine input occurred throughout the dendritic tree of both A8 and A13 types, and numerically exceeded bipolar cell input for A13. Gap junctions between stained, and similar-appearing unstained dendritic profiles were observed for both amacrine types. In addition, A8 engaged in gap junctions with cone bipolar profiles in sublamina b which also provided ribbon input. Synaptic output for both amacrine types occurred primarily upon amacrine and ganglion cells in sublamina a. Both cells were presynaptic upon single OFF-center beta ganglion cells running through the middle of their dendritic trees. Mixtures of rod and cone signals were found in the centrally evoked hyperpolarizations of each type. Center mechanism space constants of such types ranged from 100 to 400 microns, with antagonistic surround in 1 of 5 cases. Dopamine (250 microM) reduced receptive field space constants by one-third in one case. The synaptic organization and potential circuitry implications of these cone system-dominated amacrine types are compared and contrasted to the better-known AII and A17 types previously described for the rod system.

We studied the morphology, photic responses, and synaptic connections of ON-OFF amacrine cells in the cat retina by penetrating them with intracellular electrodes, staining them with horseradish peroxidase, and examining them with the electron microscope. In a sample of seven cells, we found two different morphological types: the A19, which ramifies narrowly in stratum 2 (sublamina a) of the inner plexiform layer, and the A22, which ramifies mostly in stratum 4 (sublamina b) but extends some dendrites to sublamina a. Both of these cell types have axon-like processes that extend >800 μm from the conventional dendritic arbor. ON-OFF amacrine cells in our sample had receptive fields (1.7 ± 0.3 mm diameter) that were broader than their dendritic arbors (425 ± 35 μm diameter) and that extended over the region of axon-like processes. In addition, we found many features in common with ON-OFF amacrine cells in poikilotherm vertebrates: a broad receptive field without surround antagonism, two sizes of spike-like events, narrow dynamic range (1 log unit intensity), and excitatory postsynaptic potentials at light on and light off. Two A19 amacrine cells were examined in the electron microscope: most synaptic inputs (93 and 76%, respectively) to either cell were from amacrine cells, with minor inputs from cone bipolar cells. Synaptic outputs were to bipolar, amacrine, and ganglion cells, including the OFF-α cell. © 1996 Wiley-Liss, Inc.

Connections of the three human horizontal cell (HC) types with overlying cone pedicles have been studied via electron microscopy (EM). Because blue cones (B-cones) can be recognized on distinctive morphological criteria, we could determine their presence by light microscopy (LM) in the mosaic overlying HC dendritic trees. Then we could confirm the presence or absence of dendritic contacts to B-cone pedicles by examining EM serial sections and making reconstructions of examples of the three HC types. Three HI cells have been reconstructed. Their dendritic terminals ended as lateral elements of ribbon synapses in green and red cone pedicles (G- and R-cones) primarily. B-cone pedicles in HI cell dendritic fields received no more than one or two contacts. Six reconstructed HII cells were found to contact all the pedicles within their dendritic field. However, their dendrites reached especially for B-cone pedicles and innervated them with disproportionately large numbers of terminals compared with G- and R-cones. HII axons appeared to contact B-cones exclusively. The four reconstructed HIII cells were found to avoid completely B-cones in their dendritic fields. Data have been collected on synaptic ribbon lengths at HI and HII lateral elements in the B-cone as compared with G- and R-cone pedicles. HII dendritic terminals end almost exclusively at the smaller ribbons and HI dendrites at the larger ribbons. The number of dendritic terminals provided by the three HCs to G- and R-cone pedicles as compared B-cone pedicles has been more accurately quantitated than was possible in the LM analysis (accompanying paper). New findings on the morphology of B-cone pedicles in peripheral retina have revealed that 1) B-cone pedicles end further vitread in the outer plexiform layer (OPL) than other cone pedicles, thereby forming a sublayer of the OPL neuropil, here named OPLb, in comparison to OPLa, where the G- and R-cone pedicles end; 2) B-cone pedicles have very few telodendrial connections; and 3) in peripheral retina (probably beyond 8 mm from the fovea to the ora serrata), they are bi- or trilobed, with each lobe containing separate synaptic invaginations. The vitread position and unique morphology of B-cone pedicles appear to relate directly to the unique morphology and unusual connectivity patterns of both their B-cone-specific bipolar and B-cone-related horizontal cell, the HII cell.(ABSTRACT TRUNCATED AT 400 WORDS)

The relationship of primate horizontal cells (HC) to cone pedicles was assessed by superimposing the cone inner segment mosaic upon Golgi-impregnated HC dendritic terminal clusters in a light microscope (LM) study. The HI, HII, and HIII types of HC were identified, hand-drawn, photographed, and analyzed by computer graphics methods. Blue cone (B-cones) inner segments and their projected pedicles were distinguished from red (R-cones) and green (G-cones) cones on morphological criteria. Thus the inclusion or avoidance of B-cone pedicles by the various HC types dendritic terminal clusters establishes whether there is any color specificity to their connections. In addition, we made counts of the number of dendritic terminals in the clusters going to cone pedicles in the various HCs' dendritic fields and plotted these against distance the cone pedicles lay from the cell body. In this way we could evaluate the weighting of spectral type of cone input. In general, the three HC types made the majority of their dendritic contacts with cones lying closest to their cell bodies at the center of their dendritic fields. However, HI and HIII cells, with their distinct terminal clusters, did not contact all the centrally located cones uniformly. They either avoided completely (HIII cells) or made only sparse dendritic connections (HI cells) with certain cones. The avoided or sparsely innervated cones were identified as B-connes. HII cells, on the other hand, with their more profuse and diffusely branched dendrites, appeared to contact all overlying cone pedicles and, in contrast to HI and HIII cells, directed a relatively larger number of dendrites to B-cone positions. Axon terminals of HIII cells appeared to contact B-cones exclusively. © 1994 Wiley-Liss, Inc.

Golgi-impregnated horizontal cells (HCs) as viewed in whole mount human retinas have been studied by light microscopic (LM) techniques. Impregnated HCs have been drawn by camera lucida and by the Eutectics neuron tracing method to provide quantitative data on dendritic tree sizes, dendritic tree shapes, and dendritic terminals for statistical treatment and cluster analysis. In addition, fractal analyses of HC dendritic branching patterns have been performed. Three significantly different HCs can be classified on both subjective and objective morphological criteria in central and peripheral human retina. In the fovea all HCs are so small that it is difficult to achieve a clear separation of the subtypes, although they can be distinguished by the experienced observer. HI types are the classic HCs of Polyak (The Retina, Chicago: University of Chicago Press, 1941) with distinct dendritic terminal clusters going to cones and a fan-shaped axon terminal consisting of large numbers of rod-destined terminals. HII cells have profusely branched, overlapping dendrites, with poorly defined terminals going to cones and a short curled axon bearing small terminals also going to cones. The HIII types exhibit larger diameter, more asymmetrically shaped dendritic trees and 30% more dendritic terminal clusters than HI cells at any location on the retina. Many HIII cells appear to emit a process from the cell body in the inner nuclear layer (INL) that descends into the outer strata of the inner plexiform layer (IPL). The axon of the HIII cell may end in a loosely organized, sprawling arborization. Fractal dimensions of the horizontal cells also show significant differences between the three groups. HII cells exhibit the highest fractal dimension followed by HI and HIII cells with lower and lowest fractal dimensions, respectively. The fractal dimension of HII cells of rhesus monkey, as determined from drawings by other authors in other publications, are the same as HII cells of human retina.

A variety of intracellular recording and staining techniques has been used to establish structure-function and, in some cases, structure-function-neurochemical correlations in fish, turtle, and cat retinae. Cone photoreceptor-horizontal cell connectivity has been studied extensively in the cyprinid fish retina by intracellular staining with horseradish peroxidase (HRP) and subsequent electron microscopy. The available data suggest that horizontal cell dendrites around the ridge of the synaptic ribbon are postsynaptic, whilst finger-like extensions ("spinules") of lateral dendrites function as inhibitory feedback terminals. An interesting feature of this interaction is its plasticity: the feedback pathway is suppressed in the dark and becomes potentiated by light adaptation of the retina. Intracellular recordings and stainings of ganglion cells in both turtle and cat retinae have been possible. Prelabelling of ganglion cells by retrograde transport of rhodamine from the tectum allows ganglion cells to be stained under visual control, and their synaptic inputs determined by electron microscopy. Such studies have been extended to double labelling by using autoradiography or postembedding immunohistochemistry to identify the neurotransmitter content of the labelled cell and/or the neurotransmitter(s) converging upon it. It is envisaged that further applications of intracellular staining followed by double- or even triple-labelling will continue to enhance greatly our understanding of the functional architecture of the vertebrate retina.

Light microscopic immunocytochemistry using anti-choline acetyltransferase (ChAT) was performed to stain putative cholinergic amacrine cells in turtle retina. ChAT-immunoreactive somata lie in the inner nuclear (INL) and ganglion cell (GCL) layers. Three types of amacrine cells were found according to the location of their somata and their dendritic stratification pattern in the inner plexiform layer (IPL). Type I amacrines lie in the row of cells closest to the INL/IPL limits and they branch along the s1/s2 border of the IPL. Type II amacrines are displaced to the GCL and they ramify along the s3/s4 border of the IPL. Type III amacrines lie in the middle of the INL, 2–3 rows away from the IPL limits and their dendrites appear to be bi- or tri-stratified in s1 and s3-s4 of the IPL. The turtle ChAT-IR amacrines are thus similar to the types described in chicken retina. A regular, non-random mosaic formed by stained type II amacrine cells was observed in the GCL. Their density in mid-central retina was 750 cells/mm 2 , tapering off to 393 cells/mm 2 in peripheral retina. Our study indicates that a pair of cholinergic amacrine cell types in turtle retina is arranged in mirror-image symmetry contributing to sublamina “a” and sublamina “b” of the IPL, like in other vertebrate retinas.

The turtle retina has been shown to have a variety of different morphological ganglion cell types as well as distinct physiological ganglion cell types. The major projection of the retina to the brain in nonmammalian vertebrates is to the optic tectum. In this study, we address the question of which retinal ganglion cell types project to the optic tectum in the turtle.Fluorescent rhodamine-labeled microspheres were used to trace the retinal ganglion cell projection to the superficial layers of the optic tectum. The fluorescent ganglion cell somata, retrogradely marked by transport from the contralateral optic tectum, were impaled with micropipettes containing rhodamine-horseradish peroxidase solution and this dye was iontophoresed into the cells under visual control.Most of the morphological ganglion cell types described in Golgi studies (Kolb, 1982; Kolb et al., 1988) were stained. Thus, the small cell types G1, G2, G3, G5, G6, and G7; the medium-sized types G10, G11, G12, G13, and G14; and the large-sized types G15, G16, G19, G20, and G21 project to the optic tectum in the turtle. We have added a new type, G2a, which proves to have some differences from the original G2 in branching pattern. We were unable to stain the small type G4, the medium-sized types G8 and G9, and the large cell types G17 and G18; this suggests that they might not project to the superficial layers of the dorsolateral optic tectum, at least, in the turtle.

Horizontal cells are second-order neurons in the vertebrate retina, receiving direct excitatory input from cone and rod photoreceptors. They appear in several morphological types; the most general classification is into A-type without axons and B-type with axons. Horizontal cells in all vertebrates receive direct input from photoreceptors but also interact with each other via gap junctions. They form large syncytia, each composed of cells with similar morphological and physiological properties. These syncytia provide the means for lateral spread of light signals within the retina, and form the basis for the surround component of the receptive fields of bipolar cells and ganglion cells. Thus, horizontal cells play a major role in the first stage of spatial information processing. Horizontal cells are also presynaptic to photoreceptors, exerting a negative feedback effect. This pathway underlies color opponency in horizontal cells. Such horizontal cells are called chromaticity cells (C-types), and they play a major role in the genesis of color information processing.

The major neurons and neural circuits that are involved in the transmission of color signals through the human retina to produce the color and spatially opponent P cell or midget ganglion cell responses are described. The older findings of single cone to midget bipolar connectivity is reviewed, and the single midget bipolar cell to midget ganglion cell connectivity as revealed by a recent serial section electron microscope study is described in detail. Our present knowledge concerning the discrimination of the blue-cone subtype from the other longer wavelength cones in the human at the outer plexiform layer is summarized, and our most recent findings concerning horizontal cell connectivity to the different spectral types of cones are discussed. Finally, a hypothetical pathway is proposed for color-opponent surrounds of midget ganglion cells using both horizontal cells at the outer plexiform layer and amacrine cell pathways at the inner plexiform layer.

Cone photoreceptor pedicles from midperipheral regions of the human retina (6 mm from the foveal center) have been studied by light and electron microscopy. Three areas of cone pedicle mosaic were serially thin-sectioned, in the tangential plane, from the inner border of the outer plexiform layer to the emergence of the cone axons from the cone pedicles. Semithin sections were then collected from the cone axon level through the cone cell bodies to the cone inner segment level. Two hundred twenty-one cone pedicles were followed by this means to their respective inner segments. Eight percent of the cone pedicles were from cones with inner segment characteristics of the blue cones. All 221 cone pedicles were reconstructed by tracing images from electron micrographs. The cone pedicle locations, surface areas, telodendrial projections, and synaptic ribbons could then be measured by morphometry and analyzed by statistical methods. Some selected cone pedicles were reconstructed by computer graphics methods. The cone pedicles identified as belonging to the blue cone type could be distinguished from the surrounding longer wavelength types on the following morphological criteria: 1) they were smaller (50% the area of the surrounding pedicles), 2) they contained shorter synaptic ribbons, 3) they exhibited essentially no telodendrial contact to neighboring cone pedicles, 4) they were positioned slightly more vitread in the outer plexiform layer than neighboring pedicles, and 5) their irregular occurrence in the cone mosaic coincided with the distribution criteria established in our previous paper (Ahnelt et al: J. Comp. Neurol. 255:18-34, 87) for putative blue sensitive cones in midperipheral human retina.

Using an antibody against tyrosine hydroxylase on newborn to 30-day kitten retinas, we have been able to follow the development of the dopaminergic amacrine cells of the cat retina by light-microscopical investigations of retinal wholemounts. The Type 1 or large Toh+ amacrine cells described by others (Oyster et al., 1985; Törk & Stone, 1979) and named A18 from a Golgi study (Kolb et al., 1981), is at birth (P1) an immature neuron with a small cell body and two or three simple thick radiating dendrites stratifying in stratum 1 with many of the dendrites ending in enlarged growth cones. With increasing postnatal age, the cell body size increases from 12.5 μm diameter to reach 15.5 μm diameter at P30. The dendritic fields also increase in size and complexity. At P1, cells of the central area exhibit dendritic appendages which then develop progressively until at P13 (after eye opening) they are part of rudimentary rings and by P30 the dendritic plexus of Toh+ dendrites and rings in stratum 1, typical of the adult cells, are complete. Toh+ stained processes with growth cones that run deep in stratum 5 of the inner plexiform layer and processes passing to the outer plexiform layer first become apparent at P1 in cells of central inferior retina but not till after P13 are these processes clearly expressed. At P1, the total number of Toh+ Type 1 cells is approximately 4000 and this number remains unchanged to the adult retina. However, the retina increases in size over the Pl–P30 stage and thus the mean density of Type 1 Toh+ cells decreases from 30/mm2 at P1 to 18/mm2 at P30. The maximum density of Type 1 Toh+ cells occurs in central retina 2–4 mm superior temporal to the area centralis, corresponding to the maximum rod photoreceptor concentration. A second type of small Toh+ amacrine cell can be visualized at P1. This Type 2 cell is characterized by a much smaller cell body than Type 1 cells (9 μm diameter), and with faintly stained dendrites located in stratum 3 of the inner plexiform layer. During later postnatal days, Type 2 cells gradually become unstainable and only few are still seen in far peripheral retina by P23. Type 2 Toh+ cells form a total population of 40,000 cells at P1 with their highest density occurring in peripheral retina. By P13, they cannot be seen in central retina and are reduced to a total population of cells staining for the antibody of 7400 cells in far peripheral retina. Their density decreases from 233/mm2 at P1 to 0/mm2 at P30. The transiently staining population of Type 2 Toh+ immunoreactive cells probably correspond to the small THI-CA-cell seen in rat retina (Nguyen–Legros et al., 1983), and the second type of dopaminergic amacrine cell seen in macaque retina (Mariani & Hokoc, 1988). This work was supported by Grant NNSF 3861004 from China, to H.-H. Wang, a grant from the Generalitat Valenciana, Spain, to N. Cuenca, and Grant EYO3323 to H. Kolb.

We have used immunostaining techniques to reveal a new type of amacrine cell that is immunoreactive for choline acetyltransferase (ChAT), the acetylcholine synthesizing enzyme, in the Ground Squirrel (Spermophilus beecheyi) retina. Cryostat sections and double immunostained wholemount preparations were examined by confocal microscopy. This new ChAT type III cell is distinct in morphology and neurotransmitter content from the well know ‘starburst’ amacrine cells (types I and II) that are so well represented in the ground squirrel retina [J. Comp. Neurol. 365 (1996) 173–216]. The type III cell colocalizes glycine with the acetylcholine and does not appear to be GABAergic or exhibit calcium-binding proteins like the well-known starburst type. As well, type III cells do not occur as a mirror-symmetric pair with normally placed and displaced varieties. The type III cell is probably a small field amacrine type branching broadly in upper sublamina b of the inner plexiform layer, and is most likely A6 of the Ground Squirrel retina [J. Comp. Neurol. 365 (1996) 173–216]. Type III cells are ideally placed in the architecture of the Ground Squirrel retina to influence ON directionally selective ganglion cell types. Supported by Research to Prevent Blindness to HK and DGESIC PB98-0972 to NC.

Unlike in birds and cold-blooded vertebrates' retinas, the photoreceptors of mammalian retinas were long supposed to be morphologically uniform and difficult to distinguish into subtypes. A number of new techniques have now begun to overcome the previous limitations. A hitherto unexpected variability of spectral and morphological subtypes and topographic patterns of distribution in the various retinas are being revealed. We begin to understand the design of the photoreceptor mosaics, the constraints of evolutionary history and the ecological specialization of these mosaics in all the mammalian subgroups. The review discusses current cytological identification of mammalian photoreceptor types and speculates on the likely "bottleneck-scenario" for the origin of the basic design of the mammalian retina. It then provides a brief synopsis of current data on the photoreceptors in the various mammalian orders and derives some trends for phenomena such as rod/cone dualism, spectral range, preservation or loss of double cones and oil droplets, photopigment co-expression and mono- and tri-chromacy. Finally, we attempt to demonstrate that, building on the limits of an ancient rod dominant (probably dichromatic) model, mammalian retinas have developed considerable radiation. Comparing the nonprimate models with the intensively studied primate model should provide us with a deeper understanding of the basic design of the mammalian retina.

In this study, we discriminated the various types of horizontal cell in the turtle retina on their content of neuroactive substances. Double label immunocytochemistry was performed on sectioned and wholemount retina using antisera to neural- and endothelial-nitric oxide synthase (nNOS, and eNOS), calretinin (CR), calbindin (CB), gamma-aminobutyric acid (GABA) and choline acetyltransferase (ChAT). H1 cells and their axon terminals label with CR, CB and GABA. Only H1 axon terminals label with eNOS. H2 cells contain CB, CR, nNOS and GABA maybe in their dendrites. H3 cells label only with nNOS. The localization of nNOS in the H2 and H3 cells is a novel finding. None of these antibodies labels H4 cells. The photoreceptor subtypes have been differentiated by different intensity of labeling with CB. The accessory member of the double cone is less intensely labeled with CB than the principal member and rods and blue cones do not appear to label at all. ChAT-IR is located in terminal boutons of H1 and H2 horizontal cells and H1 axon terminals and these boutons contact rods and all spectral types of cones. Clearly, GABA is present in H1 horizontal cells and may be used in neurotransmission between horizontal cells and possibly for feedback pathways to photoreceptors. The evidence of nNOS immunoreactivity in H2 and H3 horizontal cells, combined with available physiological evidence, suggests that NO may be involved in electrical coupling and/or modulation of synaptic input to these types of cells. Furthermore, our results raise the possibility that cholinergic synaptic transmission may occur from horizontal cell processes to photoreceptors in the outer plexiform layer of the turtle retina. Supported by grants EY04855 (H.K.) and a grant to the Department of Ophthalmology from Research to Prevent Blindness. A fellowship from the Secretaría de Estado de Educación Universidades, Investigación y Desarrollo of Spain supported N.C.

Immunocytochemistry was used to reveal a population of bipolar cells that contain γ-atrial natriuretic peptide 1–25 (γ-ANP) in turtle retina. This same antibody was also used in rat retina as a comparative control. The retinas were examined by both conventional light microscopy and confocal microscopy with double-labeling to determine whether protein kinase C-α-like immunoreactivity (PKC-α-LI) was colocalized with the γ-ANP-LI. Some thick sections of turtle retina immunostained with only the γ-ANP antibody were also examined by electron microscopy. In rat, a subpopulation of bipolar cells with axons terminating close to the ganglion cell layer was labeled. Double-labeling experiments indicated that the γ-ANP-LI and PKC-α-LI were colocalized in rat retina, and thus all the bipolar cells with γ-ANP-LI were rod bipolar cells. In turtle, the γ-ANP antibody labeled certain bipolar cells that were characterized by bistratified axon terminals arborizing on the borders of strata S2/3 and S3/4 in the inner plexiform layer (IPL). Double labeling with PKC-α antibody indicated that bipolar cells with γ-ANP-LI were not the same bipolar cell types with PKC-α-LI. Thus, γ-ANP-LI appears to be a new marker for a distinct type of bipolar cell in turtle retina. At the ultrastructural level, the γ-ANP-LI was visible throughout the cytoplasm of the bipolar cells from dendrites to axon terminals. In the outer plexiform layer (OPL), labeled dendrites contacted photoreceptor pedicles almost exclusively at narrow-cleft basal junctions, but infrequently formed the central element at a photoreceptor ribbon synapse. In the IPL, axon terminals with γ-ANP-LI made ribbon synapses onto a combination of amacrine and ganglion cells. Since narrow-cleft basal junctions and photoreceptor ribbon-related junctions are known to be associated with ON-center bipolar cells in turtle, and since the axon terminals of bipolars with γ-ANP-LI stratify primarily in the ON-strata of the IPL, we suggest that these cells are likely to be ON-center cells. It is possible that the γ-ANP may be involved in regulating the activity of Na+/K+ ATPase or in the modulation of cGMP levels.

The purpose of this study was to investigate more fully the shape and content of ribbons and synapses to second-order neurons in the short-wavelength cone (S-cone, blue cone) pedicle and to learn more concerning the uniqueness of the S-cone system in the primate retina. A piece of well-fixed peripheral human retina (10 mm, 35° nasal to the fovea) was serially thick sectioned in the tangential plane from the level of the outer segments to the tops of the cone pedicles. Then serial electron microscope (EM) sections were collected through the whole depth of the pedicle-occupying region into the neuropil of the outer plexiform layer (OPL). The resultant EM micrograph montages of a large field of cone pedicles were perused, and S-cone pedicles were identified. Serial micrographs of a single S-cone pedicle, picked out of the montages, were digitized and reconstructed by computer three-dimensional methods. The S-cone pedicle arose from a slightly oblique axon and projected 0.5-1 μm more vitread in the OPL than other cone pedicles. It was bilobed in shape, with synaptic invaginations and ribbons in both lobes. No cone-contacting telodendria projected from the S-cone pedicle itself, but a small number of neighboring cones sent telodendria to its surface to make small gap junctions. Neighboring rod spherules also made small gap junctions. Four robust bipolar cell dendrites, most likely from S-cone-specific bipolar cells, made synapses at ribbons and basal (distal) junctions. A small number of other bipolar cell dendrites made narrow-cleft basal junction only. The majority of lateral elements were thought to be from HII horizontal cells, and a minority from HI horizontal cells. We conclude that the S-cone pedicle has a unique morphology and connectivity to second-order neurons that makes it quite different from the other two longer wavelength cone systems, and we speculate on the consequences for color processing in the visual system in general. J. Comp. Neurol. 386:443-460, 1997. © 1997 Wiley-Liss, Inc.

Since amacrine cells are important interneurons of the inner retina and their activity may be detected in certain waveforms of the electroretinogram, this paper reviews their morphologies, classification, mosaics, neurotransmitter content, neural circuitry and physiological responses to light. Nine different amacrine cell types of cat, rabbit and human retinas are presently quite well studied in terms of the aforementioned aspects and are described in detail in this paper.

Pseudo-3-dimensional models and surface renderings of a small group of neighboring cone pedicles in the turtle retina, in which the 5 different spectral types are represented, have been made from scanned serial electron micrographs using a Macintosh personal computer, Autocad and Studio Pro software. The resultant computer generated images clarify the morphological differences between the spectral types and show how they are related to each other via telodendrial connections and different levels of ending in the neuropil of the outer plexiform layer (OPL). The double cone pedicles end highest in the neuropil and the single red and green pedicles lie slightly lower into the OPL with the single blue and ultraviolent (UV)-sensitive cone pedicles coming in from obliquely angled axons to assume positions between the others much more vitread in the OPL. Telodendria interconnect the double and single red and green cones but not the blue and UV-sensitive cones.

The paper reviews recent work on the morphological identification of mammalian short-wavelength-sensitive (S-) cones using LM- and EM-Golgi studies and its correlation with histochemical and immunhistochemical labeling techniques. S-cones comprise less than 10% of the total cone population. The S-cone inner segments are more slender yet longer, projecting further into subretinal space, and their synaptic terminals may be distinguished by criteria such as radial position, smaller size and scarcity of telodendrial contacts to neighboring cones.

In the vertebrate retina, light, acting upon visual pigment molecules in the photoreceptor outer segment, is transduced into electrical signals, which then modulate the flow of neurotransmitter from photoreceptors to second-order neurons, the bipolar and horizontal cells in the outer plexiform layer (OPL). The photoreceptor neurotransmitter, glutamate, binds to receptor complexes located on bipolar-cell and horizontal-cell postsynaptic membranes, directly or indirectly gating ion channels which mediate either ON-center or OFF-center responses. Thus, information concerning brightness, darkness and color is transmitted proximally. Information is also transformed in this process, yielding second-order neurons with unique electrophysiological signatures. Horizontal cells integrate signals laterally over wide spatial regions and mediate antagonistic surround responses of bipolar cells.

The cat retina has been a model retina for anatomical, physiological and pharmacological studies to elucidate its functional organization for at least a quarter of a century now. Over most of this time Hisako Ikeda has been at the forefront of research on ganglion cell function in this retina in both normal, abnormal and developmental conditions. Thus, it is fitting that in this paper which is to be amongst contributions from colleagues and friends of Hisako Ikeda I describe my research on the anatomical organization of the cat retina. This research fits very well with her physiological findings and represents our mutual interests in understanding how cat ganglion cell receptive fields are organized.

In this paper we describe the morphology of the different spectral types of photoreceptor pedicles in the outer plexiform layer (OPL) of the Pseudemys turtle as studied by light (LM) and electron microscopy (EM). Tangential serial thick sections were cut from the oil droplet region to the level where the axons emerge from cell bodies and then serial thin sections through the axons and the entire pedicles were collected and examined. Thus photoreceptor pedicles could be identified by tracing cells from their oil droplets to their synapses in the OPL. Double cone pedicles consisted of closely applied pairs with the principal member's pedicle wrapping around the accessory member's pedicle. Each pedicle was approx. 104 microns 2 in area and contained 12 and 8 synaptic ribbons respectively. Rods comprised 8% of the pedicles in the field, were small (84 microns 2), contained closely packed synaptic vesicles, and on average, 9 long ribbons. Single red and green cone pedicles could not be told apart without following them from their oil droplets, however, both were about the same size (106-127 microns 2) and contained 10-12 ribbons. Blue cone pedicles were small and round (80 microns 2) and arose from short oblique axons giving off from large, greenish, clear oil droplet-containing cell bodies (13% of the cone population). The least common pedicle types (5% of the cone population) were identified tentatively as UV cones because they originated from small, clear oil droplet-containing cell bodies. UV cones had spherical pedicles, elongated in the vertical axis, that arose from extremely long, angled axons. Their very small pedicles (64 microns 2) exhibited characteristic "horns" that projected from the top sides of the pedicle. Both putative UV and blue cone pedicles ended more vitread in the OPL than other pedicles and contained only 5-6 and 8-10 ribbons respectively. Understanding the ultrastructural features that distinguish the different types of photoreceptor pedicle will allow us to begin a study of spectral connections to second order neutrons in the turtle OPL in the future.

Two well-stained, horseradish peroxidase-filled varieties of putative ON-OFF directionally selective ganglion cells, G14a and G15, that project to the dorsolateral optic tectum (Guiloff and Kolb [1992a] Vis. Neurosci. 8:295–313) were studied qualitatively and quantitatively. Both were bistratified ganglion cells with one tier of dendrites; in the OFF sublamina and the other in the ON sublamina of the inner plexiform layer (IPL). The cells were serially sectioned and examined for synaptic inputs by electron microscopy. Portions of the dendritic trees were also analyzed after postembedding immunocytochemistry for neurotransmitter candidates gamma aminobutyric acid (GABA), glycine, choline acetyltransferase (ChAT), and glutamate in presynaptic neurons. Both G14a and G15 are dominated by amacrine cell inputs and have only minor bipolar cell involvement. Probably at least two different types of bipolar cell are presynaptic. Both ganglion cells receive some GABA-Positive (GABA+) amacrine inputs and G14a receives ChAT+ amacrine inputs. Glycine+, and glutamate+ inputs could not be detected in either cell. The GABA+ inputs appeared to be regionally arranged in the dendritic trees. The general distribution of amacrine and bipolar inputs to the two tiers of dendrites in both cell types appeared to be asymmetrical, both along the radial extent of the dendritic trees and within the depth of the IPL. Our data support some aspects of the current models for directional selectivity. We suggest candidate bipolar and amacrine cells that could have input to these ganglion cells. Since many of the putative presynaptic amacrine cells coincide with directionally selective types recorded and stained, by other authors, we propose that in turtle retina directional selectivity arises in neurons presynaptic to the ganglion cells. © 1994 Wiley-Liss, Inc.

Monoclonal antibodies to the three isozymes of protein kinase C (PKC) (α, β, and γ) were applied to postmortem human retina. Immunostaining was done on wholemount, or cryostat-sectioned retina, and visualized after ABC/DAB procedures by light (LM) and electron (EM) microscopy.The PCK-α antibody stained rod bipolar cells throughout the retina. EM analysis confirmed they were PKC-α-immunoreactive (IR) on their characteristic dendritic and axonal synaptology. Putative blue cone bipolar cells with wide-field axon terminals, stratifying in s5 of the inner plexiform layer (IPL), were also PKC-α-IR, and EM showed them to engage in narrow-cleft ribbon junctions in blue cone pedicles.The PKC-β antibody stained cone bipolar cells, many amacrine cells, and most ganglion cells. Cone bipolar cells were stained all the way into the foveal center: both midget and diffuse varieties were included. The IPL was densely PKC-IR and individual neurons could not be identified on stratification patterns. EM of the outer plexiform layer (OPL) revealed that both flat and invaginating cone bipolar types were IR and that IR axon terminals were presynaptic in all strata of the IPL. The occurrence of PKC-β-IR bipolar axons in stratum 2 of the IPL suggests that OFF-center as well as ON-center types were included.The PKC-γ antibody gave inferior staining compared with results from the other two antibodies; however, two varieties of wide-field monostratified amacrine cell and a large-bodied ganglion cell type were discernible.PKC in one form or another appears to be a second messenger used in neurotransmission by both rod and cone systems and ON- and OFF-center systems in the human retina.

An OFF-center alpha and an OFF-center beta ganglion cell in cat retina, which had been recorded from and intracellularly stained with horseradish peroxidase (HRP) were examined by serial section electron microscopy. We counted synapses and identified presynaptic neurons to the HRP-stained cells in 20 μm radial slices through the centers of their dendritic trees. Presynaptic amacrine and bipolar cells were identified on cytological criteria known from previous studies. The OFF-beta cell with a 62 μm dendritic arbor, restricted to S1 and S2 (sublamina a) of the inner plexiform layer (IPL), received 38% bipolar and 62% amacrine cell synapses. The bipolar input was from both cb1 and cb2 cone bipolar types. Input from three distinct amacrine cell types occurred upon the dendrites, namely from: (1) AII amacrine lobular appendages, (2) large pale amacrine profiles (possibly A2 or A3 cells), and (3) small, dark amacrine types (possibly A8 cells). Large pale amacrine profiles (possibly A13) were found on the cell body and apical dendrite in sublamina b of the IPL. In addition, several amacrine profiles synapsed directly on the sides and base of the cell body in the ganglion cell layer. We estimate that the complete dendritic tree of this beta cell received about 1,000 synapses contributed by 12–14 bipolar cells, 7–10 AII amacrines and 28–41 other amacrine cells. The OFF-alpha cell had a dendritic tree size of 680 × 920 μm. A 250 μm length of two major dendrites stratifying narrowly in S2 of the IPL was reconstructed. Amacrine cells provided most of the synaptic input (80%). This input came from: (1) AII amacrine lobular appendages, (2) amacrines exhibiting large, pale synaptic profiles (possibly A2 or A3 cells), (3) pale amacrines with large mitochondria and a few neurotubules (unknown type), and (4) densely neurotubule-filled amacrine profiles (possibly A19 cells). A large pale amacrine cell type (possibly A13) provided synaptic input to the cell body as a serial synaptic intermediary with rod bipolar cells. Cone bipolar synapses were from only one type of cone bipolar, the cb2 type and formed 20% of the total synaptic input. We estimate that a minimum of 142 bipolar cells, 256 AII amacrine cells and 1,011 other amacrine cells, altogether providing 6,000–10,000 synapses, converged on the dendritic tree of this OFF-alpha cell. © 1993 Wiley-Liss, Inc.

A light microscope study using postembedding immunocytochemistry techniques to demonstrate the common neurotransmitter candidates gamma-aminobutyric acid (GABA), glycine, glutamate, and tyrosine hydroxylase for dopamine has been done on human retina. By using an antiserum to GABA, we found GABA-immunoreactivity (GABA-IR) to be primarily in amacrine cells lying in the inner nuclear layer (INL) or displaced to the ganglion cell layer (GCL). A few stained cells in the INL, which are probably interplexiform cells, were observed to project thin processes towards the outer plexiform layer (OPL). There were heavily stained bands of immunoreactivity in strata 1, 3 and 5 of the inner plexiform layer (IPL). An occasional ganglion cell was also GABA-IR. By using an antiserum to glycine, stained cells were observed at all levels of the INL. Most of these were amacrines, but a few bipolar cells were also glycine-IR. Displaced amacrine cells and large-bodied cells, which are probably ganglion cells, stained in the GCL. The bipolar cells that stained appeared to include both diffuse and midget varieties. The AII amacrine cell of the rod pathway was clearly stained in our material but at a lower intensity than two other amacrine cell types tentatively identified as A8 and A3 or A4. Again, there was stratified staining in the IPL, with strata 2 and 4 being most immunoreactive. An antiserum to glutamate revealed that most of the neurons of the vertical pathways in the human retina were glutamate-IR. Rod and cone photoreceptor synaptic endings labeled as did the majority of bipolar and ganglion cells. The rod photoreceptor stained more heavily than the cone photoreceptor in our material. While both midget and diffuse cone bipolar cell types were clearly glutamate-IR, rod bipolars were not noticeably stained. The most strongly staining glutamate-IR processes of the IPL lay in the outer half, in sublamina a. The antiserum to tyrosine hydroxylase (TOH) revealed two different amacrine cell types. Strongly immunoreactive cells (TOH1) had their cell bodies in the INL and their dendrites ramified in a dense plexus in stratum 1 of the IPL. Fine processes arising from their cell bodies or from the stratum 1 plexus passed through the INL to reach the OPL but did not produce long-ranging ramifications therein. The less immunoreactive amacrines (TOH2) lay in the INL, the center of the IPL or the GCL and emitted thick dendrites that were monostratified in stratum 3 of the IPL.

In this study we used serial section electron microscopy and three-dimensional reconstructions to examine four midget ganglion cells of the human retina. The four cells were located in the parafoveal retina 2.5 mm or 8° from the foveal center. Both type a (with dendritic trees in distal inner plexiform layer) and type b (with dendritic trees in proximal inner plexiform layer) midget ganglion cells have been studied. These culls have dendritic trees of 7–9 μm diameter, and their complete dendritic trees in the neuropil of the inner plexiform layer can be analyzed, as well as the bipolar cell axon terminals having synaptic input, by a study of 100–150 serial ultrathin sections. Type a midget ganglion cells appear to be in a one-to-one relationship with fiat midget bipolar cell axon terminals ending in distal inner plexiform layer. Type b midget ganglion cells are in a one-to-one synaptic relationship with invaginating midget bipolar cell axon terminals in proximal inner plexiform layer. The midget bipolar cells primarily involved with the midget ganglion cells do not contact other ganglion cell dendrites. In other words, midget bipolar cells appear to be in exclusive contact with single midget ganglion cells in the human retina. The midget ganglion cells receive most of their input from their associated midget bipolar cells in the form of ribbon synapses at dyads or monads (55–81 ribbons total), although ribbonless synapses are seen occasionally. In all four midget ganglion cells reconstructed, one or two other bipolar cell axon terminals, presumed to be from wide-field bipolar types, provide 1–3 ribbon synapses each. The number of amacrine synapses upon a midget ganglion cell's dendritic tree is approximately equal to the number of bipolar ribbon inputs (43%–56% bipolar ribbons: 44%–57% amacrine synapses). We assume from our knowledge of response characteristics of ganglion cells in other mammalian retinas (Nelson et al., '78: J. Neurophysiol. 41:427–483), that the type a midget ganglion cell and its exclusive connectivity with a flat midget bipolar cell forms a single cone connected OFF-center pathway, whereas the type b midget ganglion cell with its exclusive connectivity to an invaginating midget bipolar cell forms a single cone connected ON-center pathway, through the retina to the brain.

Immunocytochemical staining procedures using the HRP-complexed antibody to protein kinase C (PKC) have been carried out on the turtle retina. Wholemounts and frozen sections of retina have been studied by light microscopy to evaluate PKC immunoreactivity after stimulation of the retina with light and neurotransmitters known to be active in the vertebrate retina. The most dramatically stained sites are cone synaptic pedicles and bipolar cells under all conditions. Ganglion cells stain weakly under certain conditions. Applying the antibody to a 'control' retina under dark adapted conditions results in uniform background staining of both hyperpolarizing and depolarizing bipolar pathways, while stimulating the retina with K+ under dim light conditions results in discretely stained bipolar cells and a prominent band of staining in stratum 4 of the inner plexiform layer. Stronger stimulation of bipolar cells with their terminals contributing to strata 3 and 4 and the continuous dominant band in stratum 4 can be elicited with incubation of the retina in neurotransmitter agonists, GABA and dopamine. Incubation with dopamine, in particular, brings out the putative dopaminergic amacrine cell. The only condition in which a strong band in stratum 2 can be demonstrated is under stimulation with a flashing bar of spot of light. Thus K+ and neurotransmitter stimulation elicit PKC staining in neurons contributing to the ON or depolarizing sublamina of the IPL, while intermittent flashing light stimulus is required to elicit PKC staining in the OFF or hyperpolarizing sublamina of the IPL.

Neurons involved in the transfer of rod signals to the ganglion cells in the retina of the cat have been recorded from and stained with horseradish peroxidase (HRP) and their synaptic connections determined by electron microscopy. The single morphological type of rod bipolar cell responds with a sustained hyperpolarization to light and in turn drives at least five morphologically different types of amacrine cells, each of which has a unique response pattern. Two amacrines respond with either a transient (AII) or a sustained (A17) depolarization to light, while three amacrines give transient (A8) or sustained (A6, A13) hyperpolarizations. Circuitry whereby rod signals reach both on-centre and off-centre ganglion cells is discussed.

We have performed a serial-section electron microscopic study of the inner plexiform layer (IPL) of the retina of the turtle Pseudemys scripta elegans. A qualitative and quantitative assessment of the neuropil of the IPL has been done from photomontages taken from the linear visual streak area and peripheral retina. Counts of conventional, ribbon, serial, and reciprocal synapses, of ganglion cell dendrites, and of profiles containing large, dense-cored vesicles were made in five equal-thickness strata in each montage. Averages of these different features were plotted for each stratum in the linear visual streak and compared with peripheral retina. The trend was for stratum 2 to have the highest overall absolute number of amacrine and bipolar interactions, and also of serial synapses, both in the linear visual streak and in peripheral regions. Stratum 4 tended to have the second-highest number of synapses. The total number of synapses for the entire thickness of the IPL, regardless of stratification, is higher in the streak than in the periphery. The total amacrine-to-bipolar-synapse ratio in the IPL is the highest of any vertebrate studied to date (11.0 in the streak and 14.5 in the periphery) but the number of synapses/micron 2 was found to be similar to that reported for other vertebrates. Amacrine-to-amacrine synaptic contacts greatly outnumber other types of synapses; amacrines constitute the principal input to ganglion cells, whereas bipolar output is mainly onto amacrines. The trend for higher numbers of synaptic interactions in strata 2 and 4 of the streak region of the turtle IPL can be correlated with the branching of small-field amacrine and ganglion cells described in Golgi studies (Kolb: Philos. Trans. R. Soc. Lond. B 298:355-393, '82). In peripheral retina, branching of large-field amacrines and a lower number of bipolar pathways may account for the trend for larger numbers of amacrine synapses in strata 2 and 4. Profiles having large, dense-cored vesicles tend to occur most frequently in strata 1 and 5, which correlates well with the stratification in the IPL of the processes of immunoreactive amacrine cells described in other studies.

Signals from rod bipolar cells of cat retina are processed by a variety of rod amacrine cells before finally arriving at ganglion cells. Three of these rod amacrine cells (AII, A13, and A17 ) have been studied at the physiological and anatomical levels; the results suggest that each carries out a unique visual function: AII cells appear to quicken the response time of the rod system in the mid-scotopic range, while A17 cells may increase the light-gathering area of rod bipolars near visual threshold. Stimulation of A13 cells may disinhibit ganglion cells, thus heightening their responsiveness at low levels of illumination.

Quantitative light microscopic techniques have been applied to photoreceptors, ganglion cells and dopaminergic amacrine cells in the retina of the North American opossum Didelphis virginiana. This marsupial has a retina with a central areal specialization where ganglion cells reach a maximum density of 2900/mm2 and cone photoreceptors are concentrated. However, the retina is very much rod dominated and there is a ring of peak rod density approximately 2 mm around the area centralis (a.c.). At the a.c. the cone to rod ratio is 1:50 while in peripheral retina it is 1:120. Dopaminergic amacrines, revealed by induced fluorescence (Fa/Glu) techniques are extremely numerous in opossum retina reaching a peak density at the a.c. and distributed in a high density plateau covering much of tapetal superior retina. Opossum retina is compared to mammalian retinas, particularly cats, with the aim of understanding the evolution of mammalian retinal circuits.

Immunocytochemical staining procedures with the HRP-complexed antibody to substance P have been carried out on the turtle retina. Examination by light microscopy of wholemount retinas has allowed us to evaluate the morphology and distribution of the substance P immunoreactive cell types. Two amacrine cell types and two or more ganglion cell types are stained in our hands. Type A amacrines are tri-stratified wide-field amacrines. They have their major dendrites in S1 and S3 of the inner plexiform layer and they emit fine dendrites from the major dendrites that end in varicose boutons in S5 on and around cell bodies in the ganglion cell layer. Some of the dendrites in S1 radiate out in axon-like fashion for 1 mm across the retina. The type B amacrine cells are small to medium-field in dendritic extent. They have smaller cell bodies than type A and a single or, at most, two primary dendrites that pass directly to S3 before branching profusely into an intricate net-like dendritic field. The ganglion cells that are stained with substance P antibodies appear to be of several types but their exact morphologies are in doubt because only portions of their major dendrites are stained. Substance P immunoreactive axons are clearly seen to project from the cell bodies to the optic nerve head and axons are stained in the optic nerve itself. The substance P-stained ganglion cells occur in an irregular distribution that reaches a peak density in an elongated band parallel to and 1 mm below the visual streak. The type B amacrine cells reach a maximum density in the visual streak and are distributed in a highly regular mosaic decreasing in density in elliptical isodensity contours from the visual streak. In contrast the type A amacrine cells are rare or absent in the streak, being located in an irregular mosaic in peripheral retina.

The horizontal cells of the rabbit retina have been studied by light microscopy of Golgi-impregnated whole-mount retinas. The two types of horizontal cell of the rabbit retina are similar to the horizontal cells of the cat retina in most respects. However, the majority of the A-type horizontal cells of the rabbit have asymmetrical dendritic fields compared to the circular, symmetrical dendritic fields of this cell type in the cat. The A-type horizontal cells of the superior edge of the linear visual streak in the rabbit retina are the most strikingly asymmetric and most of them are elongated and oriented in a direction approximately parallel to the linear visual streak. Like H1 axon terminals of the turtle retina the oriented, elongated A-type horizontal cells of the rabbit visual streak region may play a role in the neurocircuitry which underlies orientation sensitive ganglion cells.

The structure of the human, but mainly of the rhesus monkey, retina as examined by Golgi-staining techniques is described and interpreted on evidence from both light and electron microscopy. One type of rod bipolar cell and two types of cone bipolar cell are recognized. The rod bipolar is exclusively connected to rods. The midget bipolar is postsynaptic to only one cone but each cone is also presynaptic to a diffuse cone (flat) bipolar. Such flat bipolar cells are in synaptic relationship with about seven cones. No other bipolar cell types have been found. The brush bipolar of Polyak is interpreted as probably a distorted rod bipolar, while Polyak’s centrifugal bipolar is a misinterpretation of the morphology of diffuse amacrine cells. When presumptive centrifugal bipolars were observed they appeared to be a developmental stage of amacrine cells. In the outer plexiform layer two types of horizontal cell have been defined. Each type of horizontal cell has a single axon and two kinds of horizontal cell axon terminals are recognized. In the inner plexiform layer there are two main classes of amacrine cells: the stratified amacrines and the diffuse amacrines. Each class of amacrine has a wide variety of shapes. Polyak’s midget ganglion cell is confirmed and his five other kinds of ganglion cell are classified into diffuse and stratified ganglion cells according to the level at which their dendrites branch within the inner plexiform layer. A fuller summary is given by the diagram and in the legend of figure 98, p. 174. A new type of midget bipolar is described in the Appendix (p. 177).

Golgi-impregnated retinae of rhesus monkeys have been examined by serial section electron microscopyto establish in a quantitative manner the neural connexions in the outer plexiform layer. The results have shown that there are two types of midget bipolar cell, here called the invaginating midget bipolar and the flat midget bipolar. Both types of midget bipolar are exclusive to a single cone. The invaginating midget bipolar has been found to fit a dendritic terminal process into every invagination in the cone pedicle base. The flat midget bipolar has dendritic terminals that make superficial contact on the cone pedicle base. There are twice as many dendritic terminals and points of contact with the cone pedicle on a flat midget bipolar top as com pared with an invaginating midget bipolar top. These observations, together with light microscope counts of the numbers of the two types of midget bipolars, suggest that there are two midget bipolars per cone. The diffuse cone bipolar (the flat bipolar) also makes superficial contacts on the cone pedicle base, and serial sections have shown that a flat bipolar contacts about six cones. Rod bipolars connect exclusively to rods and their dendritic term inals always end as one of the central processes that penetrate the invagination. Horizontal cell dendrites end exclusively in cone pedicles and their axon terminals end in rod spherules. The point of contact with both the types of receptor is as the lateral elements of the invaginations. A single small horizontal cell contacts about seven cones and a large horizontal cell contacts about twelve cones. The numbers of contacts per cone pedicle decrease from the centre to the periphery of the horizontal cell’s dendritic field, suggesting there is an overlap of four to six horizontal cells onto a single cone pedicle. The horizontal cell axon terminals are too numerous to assess in absolute numbers but there is only one terminal to a given rod spherule from any particular axon.

NEURONES which respond preferentially to slits of light of a given orientation have been found at the ganglion cell level of the retina1, as well as at higher centres of the visual system2–4. Present knowledge about the morphology of such cells and the organisation of the inputs to these cells does not account for the origins of their specialised receptive field properties. We report here that the horizontal cells impaled in the visual streak of the turtle retina (a specialised linear area centralis5 analogous to the fovea of higher vertebrates) respond preferentially to slits of light oriented parallel to the streak and thus parallel to the horizontal meridian of the animal's visual field5. Anatomical studies demonstrate that these horizontal cells are all elongated and oriented in the direction of the streak. Thus, our findings show that a class of neurones exists in the retina whose morphologies may form the basis for their specialised receptive field properties.

Neuronal types contributing to the inner plexiform layer of the cat retina are described based primarily on light microscopy of Golgi-impregnated retinal whole-mounts. Cells have been characterized on morphological criteria that include dendritic branching patterns, dendritic tree sizes, cell body sizes and stratification of processes in the inner plexiform layer. Nine different types of bipolar cell, 22 different types of amacrine cell and 23 different types of ganglion cell can be distinguished using one or more of these morphological criteria. The significance of the different morphological types of cells is discussed, particularly in relationship to the functional bisublamination of the cat inner plexiform layer.

Cone linear sensitivities to red and green stimuli were measured intracellularly in the dark- and light-adapted turtle retina. Test flashes of small diameter were used to minimize horizontal cell feedback. Light adaptation was achieved with either green or red background illumination. The ratio of cone sensitivities to red and green light depended on the color of the background light and differed from the ratio measured in the dark. Electron microscope studies of Golgi-stained turtle cones revealed direct synaptic connections between red and green cones mediated by cone telodendria. These data indicate that the red cone photoresponse is not univariant as has been previously supposed and suggest that mixing of signals from different spectral classes of cones can occur via direct excitatory connections between cones.

Neural connections in the outer plexiform layer of the Pseudemys turtle retina have been studied by electron microscopy of serial ultrathin sections. While the distinguishing features of the photoreceptors have been described elsewhere, in this paper we describe the patterns of connectivity between identified second order neurons and identified photoreceptors or amongst second order neurons themselves. Basal telodendria emitted from double cone pedicles interconnect the two members of the double cone. Three morphologically different types of junction are made between bipolar cells and cone pedicles. H1 horizontal cells can be distinguished from H2 horizontal cells and synapses occur between them. Axon terminals of H1 cells are presynaptic to H1 cell bodies. Photoreceptors, H1 cell bodies and H1 axon terminals engage in electrical junctions while chemical synapses occur from both types of horizontal cell to bipolar cells. On rare occasions, bipolar cell dendrites were seen to be presynaptic to other bipolar cell dendrites. The significance of some of these contacts for the electrophysiological findings on the OPL of the turtle retina is discussed.

Retinal interplexiform cells have processes that branch within both the inner and outer plexiform layers. Their morphology is described from Golgi-preparations of cat, rhesus macaque and squirrel monkey retinae. Comparisons are made with similar cells, known to be catecholamine-containing, which have been observed histofluorometrically in the teleost fish and New World monkeys. It is concluded that there may be more than one pharmacological type of interplexiform cell. In addition an inner nuclear layer plexus of fibres is described for the first time from Golgi-material of the squirrel monkey’s retina. Electron microscopy reveals that this plexus synapses within the inner nuclear layer on to bipolar and amacrine cells. It is compared with the catecholamine-containing inner nuclear layer plexus of New World monkeys.

Two types of Golgi-stained bipolar cells have been examined by light and electron microscopy to determine the ultrastructure of their synapses with photoreceptors and the spectral type of photoreceptor with which they connect in the retina of the turtle (Pseudemys scripta elegans). We have chosen bipolar cells that correspond in morphology to a center-hyperpolarizing type (H-bipolar) and a center-depolarizing type (D-bipolar) shown in Marchiafava and Weiler's paper (Proc. R. Soc. B 208:103-113, '80). The latter authors recorded intracellular responses and marked their cells with the fluorescent dye Procion yellow so that we have a clear picture of their morphology. These bipolars have been called B4 and B6, respectively, according to our recent classification scheme of Golgi-stained cells. Serial section electron microscopy of two B4 bipolars shows that they made wide-cleft, striated basal junctions with red and green cone pedicles. They connected to six or seven different cone pedicles. The three B6 bipolar cells studied made narrow-cleft, semi-invaginating basal junctions with cone pedicles. The dendrites of B6 bipolars did not become central elements at ribbon synapses although they invaginated toward the synaptic ribbon. Serial section electron microscopy indicated that B6 bipolars contacted three to five cone pedicles also of a red or green cone type. We suggest that in the case of these two varieties of bipolar cell in turtle retina, sign-conserving synapses with cones are wide-cleft basal junctions while sign-inverting synapses are narrow-cleft, semi-invaginating in morphology.

The inner plexiform layer of the mammalian retina has a bisublaminar organization determined by restricted branching of the terminals of cone bipolar cells and dendrites of class I (large) and class II (small) ganglion cells. Comparison of dendritic field diameters and receptive fiedl center sizes of large ganglion cells suggests that neural circuitry in sublamina a conveys "OFF"-center properties and connections in sublamina b "ON"-center properties to retinal ganglion cells.

The outer plexiform layer of the cat retina has been examined by electron microscopy of random and serial ultrathin sections in order that neural profiles might be positively identified and their synaptic relationships studied. Photoreceptors are interconnected by means of gap junctions as are the A horizontal cells. B horizontal cells and axon terminals do not appear to be engaged in any synapses apart from those with photoreceptors, while A horizontal cells make rare ‘junctions’ with cone bipolars only. Interplexiform cell processes probably account for all the conventional chemical synapses in the outer plexiform layer of cat retina.

Electron microscopy of Golgi preparations reveals that the dendritic terminals of horizontal cells in the cat's retina are connected only to cones. The axon terminals of these horizontal cells connect exclusively to rods. Dendritic and axonal terminals of horizontal cells penetrate the photoreceptor endings to become lateral elements of the invaginations. In addition this study suggests that each cone is in contact with more than one horizontal cell and that each rod connects to more than one axon terminal system. The horizontal cells of the cat are similar to those of the rhesus monkey in that the rod and cone connections are segregated to different portions of these cells.