Survey of the morphology of macaque retinal ganglion cells that project to the pretectum, superior colliculus, and parvicellular laminae of the lateral geniculate nucleus.

In common with other vertebrates, the primate retina contains a number of different ganglion cell types that project to different regions in the brain. We wanted to determine how the different ganglion cell types, distinguished morphologically, mapped to these regions of the brain. We injected a fluorescent dye into one of three regions of a macaque brain: the superior colliculus (SC), the pretectal region, and the parvicellular laminae of the lateral geniculate nucleus. By means of an in vitro preparation, the retrogradely labelled ganglion cells were intracellularly injected with horseradish peroxidase, so as to reveal their dendritic morphology. When the dendritic-field diameters of the injected cells were plotted against retinal eccentricity, each of the three regions was found to receive input from a distinctive population of cells. The pretectal projection was dominated by cells with large dendritic fields. The SC projection was composed of a number of distinct types, with smaller dendritic fields. Parasol cells project to SC but are extremely rare. In addition to midget ganglion cells, the parvicellular laminae receive inputs from at least two additional groups. Parvicellular bistratified (PB) cells have bistratified dendritic fields, slightly larger than those of parasol cells. Parvicellular giant (PG) cells have dendritic-field diameters larger than that of any parasol cell, ranging from 250 microns to greater than 850 microns--the largest of any primate ganglion cells. In contrast to the retinal projections of the cat, in which a specific ganglion cell type can project to different regions of the brain, each of the regions in this survey appears to receive inputs from its own distinct group of ganglion cells.

Spatial density and distribution of choline acetyltransferase immunoreactive cells in human, macaque, and baboon retinas.

Whole-mounted human, macaque, and baboon retinas were labelled with an antiserum to human choline acetyltransferase (ChAT), by the immunoperoxidase technique. Previous work in nonprimate species has shown that these cells correspond to the starburst amacrine cells. Labelled somata were disposed on either side of the inner plexiform layer, and their processes formed two narrow zones within it. In human retinas, the ratio of labelled somata in the ganglion cell layer (GCL) to those in the inner nuclear layer (nominal Sb/Sa ratio) was about 60/40 at all locations, similar to that found in nonprimate mammalian species. The density of labelled cells in the human GCL ranged from 1,000 to 1,150 mm-2 near the fovea to 300 to 400 mm-2 in the periphery. Labelling tended to be more erratic in macaque retinas. Nevertheless the Sb/Sa ratio was as high as 70/30 and spatial densities were similar to those of humans. The overlap factor in macaque retinas outside the nasal quadrant was about 10 at all retinal eccentricities, based upon dendritic-field sizes from a Golgi study. About each labelled soma there was a region 20 to 120 microns in diameter in which the probability of the occurrence of other labelled somata was lower than elsewhere. No such nonrandomness was found between labeled cells in the GCL and those in the amacrine cell layer. The packing factor was about 0.3 in well-labelled regions, independent of retinal position or spatial density. Published data on ChAT-labelled cells in rabbit and rat show a similar value. This invariance is consistent with the hypothesis that this nonrandomness is a residual consequence of somal contiguity at an early developmental stage.

The density recovery profile: a method for the analysis of points in the plane applicable to retinal studies.

The density recovery profile is a plot of the spatial density of a set of points as a function of the distance of each of those points from all the others. It is based upon a two-dimensional point autocorrelogram. If the points are randomly distributed, then the profile is flat, with a value equal to the mean spatial density. Thus, any deviation from this value indicates that the presence of the object represented by the point alters the probability of encountering nearby objects of the same set. Increased value near an object indicates clustering, decreased value near an object indicates anticlustering. The method appears to be unique in its ability to provide quantitative measures of the anticlustered state. Two examples are presented. The first is based upon a sample of the distribution of the somata of starburst amacrine cells in the macaque retina; the second is based upon the distribution of the terminal enlargements on the dendrites of a single macaque ganglion cell that projects to the superior colliculus. In both cases, the density recovery profile is initially lower than the mean density, and increases up to the plateau at the value of the mean density. Two useful measures can be derived from this profile: an intensive parameter termed the effective radius, which quantifies the extent of the region of decreased probability and is insensitive to random undersampling of the underlying distribution, and an extensive parameter termed the packing factor, which quantifies the degree of packing possible for a given effective radius, and is insensitive to scaling. An extension of this method, applicable to correlations between two superimposed distributions, and based upon a two-dimensional point cross-correlogram, is also described.

Parasol and midget ganglion cells of the primate retina.

We have intracellularly filled the dendritic arbors of 996 midget and parasol ganglion cells with horseradish peroxidase (HRP) in macaque and baboon retinas. Only minor differences in the properties of these cell groups were found between species. Ninety of these cells were cut from their retinas, embedded in methacrylate, and transversely sectioned. According to their depth of stratification, there are two types of parasol cells (termed a-parasol and b-parasol), and two types of midget ganglion cells (a-midget and b-midget). Each of these four types stratifies at a different level within the IPL. The dendritic fields of midget ganglion cells lie either near the border of the ganglion cell layer (GCL) or near the border of the amacrine cell layer (ACL). The dendrites of the two types of parasol cells stratify closer to the center of the IPL, where they divide it into three approximately equal parts. There was no vertical overlap in the dendritic fields of a-parasols and b-parasols; they were always separated by at least 1 micron. The border between the a- and b-sublaminae of the IPL, defined in terms of this narrow gap between the stratification of the two parasol cell types, lies approximately at the center of the IPL. The dendritic-field thickness for each of these types, on average, is no greater than 30% of the IPL thickness. At a similar location, there is no significant difference between the dendritic-field diameters of the two parasol types or between those of the two midget types. As previously reported (Perry et al.: Neuroscience 12:1101-1123, '84) the dendritic fields of both parasol and midget ganglion cells are smaller in the nasal retina than at a position in the temporal retina equidistant from the fovea. Because dendritic-field diameters prove to depend upon local ganglion-cell density, the scatter in these diameters as a function of retinal eccentricity is due in part to the asymmetric distribution of ganglion cells. We have devised a measure, termed equivalent eccentricity, that allows data points of cells from regions having the same local ganglion-cell density to be plotted at the same position on this scale. The use of this measure, rather than eccentricity per se, significantly reduces the scatter of dendritic-field diameters. The dendritic-field diameters of parasol cells within the nasal quadrant of the retina are not fully brought into line with those of cells lying elsewhere in the retina.(ABSTRACT TRUNCATED AT 400 WORDS)

Starburst amacrine cells of the primate retina.

A group of readily recognized amacrine cells were observed in Golgi-impregnated and flat-mounted macaque, baboon, and human retinas. These cells had roughly-circular or oval dendritic fields that were narrowly stratified within the inner plexiform layer (IPL). Most of these cells stratified in the inner half (sublamina b) of the IPL, and they had their somata in the ganglion-cell layer; a few stratified in the outer half (sublamina a) of the IPL and had their somata in the amacrine-cell layer. Typically, a single dendrite issued from the soma, and, after passing for 10 microns or so, gave rise to five or more radiate processes. As these processes neared the edge of the dendritic field they branched, turned, and became varicose. Most showed no evidence of an axon, although a few had a short process extending inward, toward the optic-fiber layer. Dendritic-field diameters were about 100 microns near the fovea and increased to about 350 microns in the peripheral retina. Mean somal diameter also increased slightly from near the fovea (7.8 microns) to the periphery (8.7 microns). Although the primate cells are smaller, and there are some minor differences in the form of the dendritic fields, these cells appear to be morphologically equivalent to the starburst amacrines of the rabbit retina, whose counterparts have also been observed in the retinas of rats and cats. Presuming that these cells correspond to the choline acetyltransferase immunoreactive primate cells described by Mariani and Hersh (J. Comp. Neurol. 267:269-280, '87), their overlap factor is about ten for the type whose somata lay in the ganglion-cell layer and about 0.25 for those whose somata lay in the amacrine-cell layer.

Central projections of cat retinal ganglion cells.

The central projections of different groups of cat retinal ganglion cells were studied following small iontophoretic injections of horseradish peroxidase (HRP) into physiologically characterized sites. Analysis was restricted to labeled cells in the upper periphery of the nasal retina, contralateral to the injection site. Injections were made to the A lamina and C lamina of the dorsal lateral geniculate nucleus (LGNd-A,C), the geniculate wing (LGNd-W), the ventral lateral geniculate nucleus (LGNv), the pretectum (PT), and the superior colliculus (SC). The dendritic fields of alpha, beta, and epsilon cells were well labeled by the procedures we employed. A group, termed "g1," had somal sizes within the range of the smaller beta and epsilon cells, but dendritic morphologies distinct from either class. The g1 group may consist of a number of types, but our material provided no basis for further distinguishing them. Many cells were observed that had smaller somas; all had thin axons, and few had dendritic fields that labeled to any significant extent. We were not able to further distinguish these cells, and refer to this group, which may include a number of types, as "g2" cells. From the peripheral nasal retina, alpha cells project to LGNd-A, LGNd-C, PT, and SC. Beta cells project to LGNd-A, LGNd-C, and PT. Epsilon and g1 cells project to the LGNd-C, LGNd-W, LGNv, PT, and SC. We determined the total spatial density of cells in the region of the retina analyzed, using a Nissl-stained preparation. We then estimated the relative fraction of cells in each of the above groupings by injecting HRP throughout a cross section of the optic tract. Multiplying this relative fraction by the total spatial density gave an estimate of the spatial density of each of these groupings. From the spatial density of cells labeled from the injection site, we were able to estimate the fraction of cells of each retinal grouping that project to each of the zones investigated. By these calculations, almost all alpha cells from the upper nasal retina project to LGNd-A and LGNd-C; most project to SC, and about a third to PT. Beta cells, by contrast, project almost exclusively to LGNd-A, with about 10% going to LGNd-C, and about 1% to the PT. The great majority of epsilon cells, if not all, project to LGNd-W, and up to half of this population also project to the other zones noted above.(ABSTRACT TRUNCATED AT 400 WORDS)

The retinal projection to the cat pretectum.

Retinal ganglion cells were labeled retrogradely by localized injections of HRP into different regions of the pretectum, tectum, and optic tract in 26 cats. Retinal projection zones in the pretectum were labeled anterogradely in the same cats by intravitreal injections of 3H-proline. This allowed the HRP injection sites to be located with respect to the retinal termination zones. The form of the projection zones from retina to pretectum was determined from serial reconstructions of either coronal or horizontal sections. The zones are best distinguished in horizontal sections, where they are seen as four roughly parallel strips on either side of the brain. They are more-or-less parallel to the anterior border of the tectum, and appear to traverse the entire width of the retinal projection to the tectum. Each zone is similar in form for the ipsilateral and contralateral projections, although the contralateral projection is thicker and denser. Binocular injections of 3H-proline showed that the projections from the two eyes were in register and did not interdigitate. Cells labeled by HRP injections in the anteromedial end of the pretectum were concentrated in the lower nasal quadrant of the contralateral retina, and the lower temporal quadrant of the ipsilateral retina. Posterolateral injections labeled cells in the upper quadrants. There is thus a rough retinotopic mapping along the elongated axis of the pretectum. When the distributions of ganglion cells labeled by HRP injections to different parts of the pretectum are combined, they show a concentration in both the visual streak and area centralis, and thereby reflect, at least qualitatively, the relative spatial distribution of the entire ganglion-cell population. About 85% of the retinal projection to the pretectum is contralateral. For all of the HRP injections, the spatial density of labeled cells was always low, accounting for no more than 3% of the total spatial density of ganglion cells in any retinal region. Several types of ganglion cells were labeled following injections to most regions of the pretectum; these included alpha, beta, and epsilon cells, as well as small-bodied cells showing a variety of morphologic forms. Alpha cells were labeled mainly from the anterolateral end of the pretectum, but other cell types were labeled from all injected regions. In the peripheral retina, 2% of the labeled cells were alpha cells, 32% were beta cells, 19% were epsilon cells, and the remaining 47% were small cells whose dendrites only occasionally filled to any significant extent.(ABSTRACT TRUNCATED AT 400 WORDS)

Parasol and midget ganglion cells of the human retina.

Golgi-impregnated ganglion cells were studied in two flat-mounted human retinas. A number of different morphologic forms were observed, and those showing a thickly branching dendritic field with terminals that stratified within a narrow zone of the inner plexiform layer were selected for further investigation. When the dendritic field diameter of these cells was plotted against distance from the fovea, the scatter diagram showed two distinct clusters. At any given eccentricity, there was no overlap between the cell group with large dendritic fields and the group with small dendritic fields. Those with the larger dendritic fields also tended to have larger somas and thicker axons than the group with the smaller dendritic fields. The dendritic fields of both groups tended to be elongated, and the orientation and degree of this elongation were quantified by determining the best-fitting ellipse for each dendritic field. The degree of elongation was independent of eccentricity. The orientation of the dendritic field (major axis of the ellipse) of a cell did not appear to be independent of its position on the retina. To test whether the major axes tended to be directed toward any particular point on the retina, the positions of the cells on the retinal flat mount were transformed to relative positions on the retinal hemisphere, and the orientations of the dendritic fields were expressed in a spherical coordinate system for this hemisphere. A search was made for the position on the hemisphere which minimized the mean square deviation of the orientations from this point. The group with the large dendritic fields showed a significant tendency to be radially oriented toward a specific location on the retinal hemisphere, and that location lay within a few degrees of the fovea. Leventhal and Schall ('83) have reported a similar finding for ganglion cells of the cat retina. For the group with small dendritic fields, the retinal location that minimized the mean square deviation was also near the fovea; however, the set of orientations showed no greater tendency for mutual alignment than did a randomized set. The cell group with the large dendritic fields appears to correspond to Dogiel's (1891) type II cells, to Polyak's ('41) parasol cells, to the A cells of the monkey retina described by Leventhal et al. ('81), observed following HRP injection to the magnocellular layer of the LGN, and to the P alpha cells of the monkey retina, observed by Perry and Cowey ('81), following HRP uptake by cut axons of the optic nerve.(ABSTRACT TRUNCATED AT 400 WORDS)

Retinal ganglion cells: properties, types, genera, pathways and trans-species comparisons.

This article attempts to develop an empirical foundation for the notion of 'cell type', and to use this notion to clarify the issues involved in the classification, pathways, and trans-species comparisons of retinal ganglion cells.

Ganglion cells of the cat accessory optic system: morphology and retinal topography.

The ganglion cells of the cat's retina form classes that are distinct in their cell morphology, retinal distribution, central projections, and functional properties. By means of the retrograde transport of horseradish peroxidase injected into the accessory optic nuclei of the cat midbrain, we have characterized the retinal ganglion cells projecting to these nuclei. The retinal projection is virtually completely crossed to the medial terminal nucleus and to the lateral terminal nucleus. This appears to be true for the dorsal terminal nucleus as well, although difficulties of the technique limit our findings for this region. No differences were found in either the spatial distribution, or the somal size distribution, or the morphological characteristics of the ganglion cells projecting to these three nuclei. In spatial distribution, these cells are concentrated in the area centralis and visual streak and show no evidence of a nasotemporal division. Morphologically, they have small to medium-sized somas and delicate, sparsely branching dendrites. They do not appear to belong to any of the morphological cell type thus far identified.

Retinal ganglion cell classes in the Old World monkey: morphology and central projections.

Labeled ganglion cells were studied in whole-mount retinas of Old World monkeys after electrophoretic injections of horseradish peroxidase into physiologically characterized sites. A number of different morphological classes have been identified, each of which has a distinctive pattern of central projection. Since different functional classes of primate retinal ganglion cells also have distinctive patterns of central projection, correspondences between functional and morphological cell types have been inferred. There prove to be parallels between morphological types of cat monkey ganglion cells.

Visual suppression from nondominant eye in the lateral geniculate nucleus: a comparison of cat and monkey.

We have studied the suppression of firing in single LGN cells of cat and monkey in response to visual stimulation of the nondominant eye. In the cat LGN most of the cells of each of the main laminae show this nondominat suppression. X cells having their dominant input from the ipsilateral eye were suppressed to a significantly greater degree than any other cell type in the cat LGN. In the monkey LGN nondominant suppression was absent in all 19 X-like cells studied, whereas 6 of 21 Y-like cells showed nondominant suppression. Thus nondominant suppression is present in the magnocellular laminae of the monkey LGN, where the Y-like cells are found, but appears to be absent from the parvocellular laminae, where the X-like cells are found.

Identification, classification and anatomical segregation of cells with X-like and Y-like properties in the lateral geniculate nucleus of old-world primates.

1. All the cells (158) that we studied in the lateral geniculate nuclei of Macaca nemestrina and Macaca irus could be distinguished as either X-like or Y-like on the basis of their responses to tests developed to classify cat retinal and lateral geniculate nucleus cells. These tests include responses to stationary spots, fast moving wands and moving gratings. 2. Response latencies to electrical stimulation of the optic chiasm were determined for 130 cells; no X-like cell showed a latency shorter than 1-7 ms, no Y-like cell showed a latency longer than 1-6 ms. Primate lateral geniculate nucleus cells with X-like properties thus receive their excitatory input from retinal cells with slowly conducting axons and these most probably include the tonic ganglion cells described by Gouras (1968, 1969); Y-like lateral geniculate nucleus cells are driven by retinal cells with faster conducting axons, most probably including the phasic ganglion cells described by Gouras. 3. Wiesel & Hubel (1966) classified monkey lateral geniculate nucleus cells into four main types based on their receptive-field properties, as revealed by spectrally and spatially distinct stimuli. We find that all Type I and Type II cells show X-like properties; all type IV cells show Y-like properties. Type III consists of a subtype that show X-like properties, here termed Type IIIx, and a subtype that show Y-like properties, here termed Type IIIy. 4. The first cells encountered as the micro-electrode reached the lateral geniculate nucleus were always X-like. In some penetrations only X-like cells were encountered as the electrode moved downward through the lateral geniculate nucleus. In the remaining penetrations, after recording X-like cells through most of the lateral geniculate nucleus, Y-like cells were then encountered. No X-like cells were found below Y-like cells. thus these two classes of cells are anatomically segregated within the primate lateral geniculate nucleus. Electrode marking showed the borger between X-like and Y-like cells to correspond to the border between the paro- and magnocellular layers of the lateral geniculate nucleus. Thus X-like cells (i.e. Types I, II and IIIx) occur in the parvocellular layers, Y-like cells (i.e. Types IIIy and IV)in the magnocellular layers.

Cancellation of rod signals by cones, and cone signals by rods in the cat retina.

1. The interaction of rod and cone signals at the level of cat retinal ganglion cells was studied by a method of light exchange. Two spectrally distinct lights were exchanged in such a manner that the rate of photon catch by rods increased in a stepwise manner at the same moment that the cone rate decreased in the same manner, and vice versa. 2. Under any conditions of adaptation, where both rods and cones contributed to the ganglion-cell discharge, it was always possible to adjust the ratio of the magnitudes of the rod and cone stimuli so that no change in ganglion-cell discharge could be detected by listening to the recorded activity via a loudspeaker. We term this condition a silent exchange. 3. On the face of it, the condition of silent exchange arises when rod and cone signals are able to cancel one another, when made opposite in phase by the exchange situation. But was this silence due to a true cancellation of the signals from one photoreceptor type by those of the other type, or was it due to our failure to stimulate the photoreceptors adequately? In order to test whether rod signals can cancel those of cones we bleached both visual pigments and set our exchange apparatus to stimulate the two photoreceptors in the antagonistic manner described above. At first no response could be heard on exchange, for the thresholds of both rods and cones lay above that of our apparatus. But the cones soon recovered and a strong response was heard on exchange. With no change in our stimulating situation, this response diminished with time and silence was again restored. This restoration of silence could not be due to the cones alone, for with time their sensitivity could only further increase. It could only be the increasing sensitivity of the rods that quietened the cone signals. In agreement with this conculsion, the dark-adaptation curve of the rods showed that they became sensitive to our stimulus at the time that the cones began to be silenced. 4. By means of coloured backgrounds we have also shown the converse, namely that rods signals can be cancelled by those of cones.

Isolation of rod and cone contributions to cat ganglion cells by a method of light exchange.

1. The great majority of cat retinal ganglion cells are known to receive signals from rods and from a single (green) cone type. The centre region of the receptive fields of these cells was stimulated by a spot that changed back and forth from orange to white. By adjusting the intensity of the white spot relative to that of the orange a condition could be established at which the photon-catch rate of the rods remained unchanged during the orange-white exchange. At this intensity setting, termed the rod isolept, rods are thus unstimulated by the exchange, however intense, and the ganglion-cell response was found to be due entirely to the green cones. At another intensity setting of the white spot relative to the orange (cone isolept), the photon catch of the green cones remained unchanged during the exchange and ganglion-cell responses were found to arise entirely from the rods. 2. A neutral wedge in the combined exchange beam (but not in the steady background that covered the whole receptive field) regulated the size of the exchange stimulus and thus the magnitude of the ganglion-cell discharge heard from a loud speaker to the exchange. Exchange threshold was the wedge setting at which this change in firing rate could only just be heard. 3. At the cone isolept, cones remain unstimulated however intense the exchange stimulus, and the rod increment threshold curve was determined over its full range from absolute threshold up to saturation. Likewise, at the rod isolept, the cone increment threshold curve was determined over the same intensity range as for the rods. Rod saturation was found to occur at the point where the cone increment threshold curve began to rise from its absolute threshold level toward its Weber region. 4. The exchange approach also enabled both rod and cone dark-adaptation curves following a strong bleaching exposure to be obtained in the same experiment by moving successively between the cone and rod isolepts. At the cone isolept the time course of early rod dark adaptation could thus be determined when the rod threshold to flashing spots lay well above that of the cones.