The objective of this study was to evaluate the maturity of three inhibitory mechanisms (end-inhibition, side-inhibition, and cross-orientation inhibition) in the striate cortex of kittens at 4 weeks postnatal. To accomplish this, we made extracellular recordings from area 17 neurons while presenting visual stimuli consisting of sinusoidal luminance gratings or composites of gratings. We then compared data from kittens relating to various characteristics of each inhibitory mechanism with data from adults. We find that end-inhibition, side-inhibition, and cross-orientation inhibition are all present in kittens, and all show signs of maturity by 4 weeks postnatal. We conclude that the development of these inhibitory mechanisms occurs relatively early, and may coincide with the development of excitatory properties.

We have employed the tetrode technique, which allows accurate discrimination of individual neuronal spike trains from multiunit recordings, in order to examine the variation of orientation selectivity among local groups of neurons. We recorded a total of 321 cells from 62 sites in area 17 of halothane-anesthetized cats; each site contained between three to ten neurons that were estimated to be less than 65 μm away from the tetrode tip. For each cell, we determined the orientation tuning in response to moving bars. Of the cells tested, 8.4% were unresponsive, 22.7% had no preferential response to any particular orientation, while 68.8% were tuned. The average difference in preferred orientation between cell pairs recorded at the same site was 10.7 deg, but the variance in preferred orientation differences differed significantly among sites. Some clusters of cells exhibited the same or nearly the same orientation preference, while others had orientation preferences that differed by as much as 90 deg. Our data demonstrate that the tuning for orientation is more heterogeneously distributed at a local level than previous studies have suggested.

Immunocytochemical methods were used to visualize glutamate immunoreactivity in the cat retina and to compare its localization with that of aspartate, GABA, and glycine. The cellular and subcellular distribution of glutamate was analyzed at the light-microscopic level by optical densitometry and at the electron-microscopic level by immunogold quantification. The findings were consistent with the proposed role for glutamate as the neurotransmitter of photoreceptors and bipolar cells as particularly high concentrations of staining were found in synaptic terminals of these cells. Ganglion cells were also consistently stained. Aspartate was totally colocalized with glutamate in neuronal cell bodies but the synaptic levels of aspartate were much lower than for glutamate. In addition to the staining of photoreceptor, bipolar, and ganglion cells, glutamate immunoreactivity was also observed in approximately 60% of the amacrine cells. These cells exhibited colocalization with either GABA or glycine. The elevated levels of Glu in amacrine cells may reflect its role as a transmitter precursor in GABAergic cells and as an energy source for mitochondria in glycinergic cells.

The generation of compensatory eye movements in response to rotational head movements involves the transformation of visual-optokinetic and vestibular signals into commands controlling the appropriate eye muscles. Previously, it has been shown that the three systems (optokinetic, vestibular, and eye muscle) share a similar three-dimensional reference frame. In this report, we suggest that a peculiarity in the structure of the horizontal recti in pigeons demonstrates that the optokinetic system is organized with respect to the eye muscles rather than the vestibular canals. Measurements of the orientation of the plane for each of the lateral and medial recti were obtained. These were compared with the direction preferences of optokinetic neurons responsive to horizontal motion, namely “back” units in the nucleus of the basal optic root (nBOR), “forward” units in the pretectal nucleus lentiformis mesencephali (LM), and “vertical axis” (VA) Purkinje cells in the flocculus. The average direction preference of LM neurons excited in response to forward (temporal to nasal) visual motion, and VA Purkinje cells in response to optokinetic motion in the ipsilateral visual field was approximately parallel to the visual horizontal. This corresponded to the orientation of the medial rectus, which was also approximately parallel to the visual horizontal. The average direction preference of nBOR neurons excited in response to backward (nasal to temporal) visual motion, and VA Purkinje cells in response to optokinetic motion in the contralateral visual field was approximately 20–30 deg down from the visual horizontal. The orientation of the lateral rectus was also approximately 20–30 deg down from the visual horizontal. These data suggest that the incoming optokinetic signals are organized with respect to the outgoing extraocular muscle commands.

The effects of changing retinal ganglion cell (RGC) density and availability of presynaptic sites on the development of RGC dendritic arbor in the developing chick retina were contrasted. Visual form deprivation was used to induce ocular enlargement and expanded retinal area resulting in a 20–30% decrease in RGC density. In these retinas, RGC dendritic arbors increased in a compensatory manner to maintain the inner nuclear layer to RGC convergence ratio in a way that is consistent with simple stretching; RGC dendritic arbors become larger with increased branch lengths, but without change in the total number of branches. In the second manipulation, partial optic nerve section was used to produce areas of RGC depletion of approximately 60% in the central retina. This reduction in density is comparable to the density of locations in the normal peripheral retina. In RGC-depleted retinas, dendritic arbor areas of RGCs in the central retina grow to match the size of normal peripheral arbors. In contrast to the expanded case, two measures of intrinsic arbor structure are changed in RGC-depleted retinas; the branch density of RGC dendrites is greater, and the relative areas of the two arbors of bistratified cells are altered. We discuss the potential roles of retinal growth, local RGC density, and availability of presynaptic terminals in the developmental control of RGC dendritic arbor.

The excitatory receptive-field centers of On-Off directionally selective (DS) ganglioncells of the rabbit retina correspond closely to the lateral extent of their dendritic arborizations. Some investigators have hypothesized from this that theories for directionalselectivity that entail a lateral spread of excitation from outside the ganglion cell dendritic tree, such as from starburst amacrine cells, are therefore untenable. We show herethat significant motion facilitation is conducted from well outside the classical excitatory receptive-field center (and, therefore, dendritic arborization) of On-Off DS ganglioncells for preferred-direction, but not null-direction moving stimuli. These results are consistent with a role in directional selectivity for cells with processes lying beyond the On-Off ganglion cell's excitatory receptive-field center. These results also highlight the fundamental distinction in retinal ganglion cell receptive-field organization between classical excitatory mechanisms and those that facilitate other excitation without producing directly observable excitation by themselves.

Biplexiform ganglion cells were labelled by retrograde transport of HRP in five species of marine fish from the neoteleost acanthopterygian orders Perciformes and Scorpaeniformes. Their forms and spatial distributions were studied in retinal flatmounts and thick sections. Biplexiform ganglion cells possessed sparsely branched, often varicose, dendrites that ramified through the inner nuclear layer (INL) to reach the outer plexiform layer (OPL), as well as conventional arborizations in the most sclerad part of the inner plexiform layer (IPL). Their somata were of above-average size and were displaced into the vitread border of the INL. Mean soma areas ranged from 99 ± 6 μm2 in Bathymaster derjugini (Perciformes) to 241 ± 12 μm2 in Hexagrammos stelleri (Scorpaeniformes), but were similar in each species to those of the outer-stratified alpha-like ganglion cells, whose dendritic trees occupied the same IPL sublamina. In the best-labelled specimens, biplexiform cells formed clear mosaics with spacings and degrees of regularity much like those of other large ganglion cells, but spatially independent of them. Biplexiform mosaics were plotted in three species, and analyzed by nearest-neighbor distance and spatial correlogram methods. The exclusion radius, an estimate of minimum mosaic spacing, ranged from 113 urn in Hexagrammos stelleri, through 150 μm in Ernogrammus hexagraminus (Perciformes), to 240 μm in Myoxocephalus stelleri (Scorpaeniformes). A spatial cross-correlogram analysis of the distributions of biplexiform and outer-stratified alpha-like cells in Hexagrammos demonstrated the spatial independence of their mosaics. Similar cells were previously observed not only in the freshwater cichlid Oreochromis spilurus (Perciformes) but also in the goldfish Carassius auratiis (Cypriniformes) which, being an ostariophysan teleost, is only distantly related. Thus, biplexiform ganglion cells may be regular elements of all teleost fish retinae. Their functional role remains unknown.

The morphology of horizontal cells in ox, sheep, and pig retinae as observed after Lucifer Yellow injections are described and compared with the descriptions of Golgi-stained cells by Ramón y Cajal (1893). Horizontal cells in the retinae of less domesticated species, wild pig, fallow and sika deer, mouflon, and aurochs were also examined. All these retinae have two types of horizontal cell; their morphologies are in common, although with some familial differences. Their basic appearance is as Cajal described; except in one important respect, a single axon-like process could not be identified on the external horizontal cells. It is concluded that external horizontal cells of artiodactyls correspond to the axonless (A-type) cells of other mammals. Cajal's internal horizontal cells have a single axon which contacts rods. This type corresponds to the B-type cells of other mammalian retinae. Artiodactyl A- and B-type horizontal cells differ from those of many other mammals in that the B-type dendritic tree is robust and the A-type dendritic tree is delicate. Historically, this morphological difference between orders of mammals has led to some confusion. The comparisons presented here suggest that the morphological types of primate horizontal cells can be integrated into a general mammalian classification.

Brn-3a, Brn-3b, and Brn-3c are highly homologous POU-domain transcription factors that are expressed in subsets of retinal ganglion cells. From each of the mouse Brn-3 genes, a DNA segment ranging in size from 4.6 to 13.4 kb and located immediately upstream of the start site of translation was joined to a human placental alkaline phosphatase (AP) reporter cDNA. Following the introduction of each construct into the mouse germline, a total of 19 transgenic lines were obtained, of which 16 expressed the AP reporter in the retina. Unexpectedly, at least 14 of the 16 expressing lines showed AP activity in subsets of amacrine cells, and these subsets typically differed among mouse lines injected with the same construct. Transgene expression was also found in ganglion cells in four lines and bipolar cells in seven lines. In all cases AP activity was confined to cells in the inner nuclear layer and the ganglion cell layer. The expression of Brn-3 transgenes in multiple cell types in the inner retina is reminiscent of earlier experiments in which visual pigment transgenes were found to be expressed in multiple cell types in the outer retina. Taken together, these observations suggest that anatomically and/or functionally related retinal neurons contain partially overlapping transcriptional regulatory specificities.

The peptide, neurotensin, is found in a class of amacrine cells synapsing chiefly withother amacrine cells in the chicken retina (Li & Lam, 1990; Watt et al., 1991). Toinvestigate the possible effects of neurotensin, we have used Ca2+ imaging to measure cytosolic Ca2+ concentrations in cultured chick amacrine cells. Following a delay of about 2 min, neurotensin (300 nM) induced oscillations in Ca2+ concentration that typically had a period of 2 min and peak values of about 300 nM when averaged over the cell body. The phospholipase C inhibitors U-73, 112 and 4′-bromophenacyl bromide terminated oscillations induced by neurotensin but the protein kinase inhibitors H7 and staurosporine did not inhibit oscillations, increasing their frequency instead. In the absence of external Ca2+, neurotensin induced only a single Ca2+ transient, much briefer than when external Ca2+ was present. Together these results suggest that neurotensin activates phospholipase C, thereby producing IP3 that triggers Ca2+ release from an internal store. Although this released Ca2+ contributes to periodic Ca2+ peaks, the majority of cytosolic Ca2+, even in the first peak, comes from Ca2+ influx across the plasmalemma.

Glutamate, when puffed near the synaptic terminals, elicits a current in rod photoreceptors. The current is strongly dependent upon both the intracellular and extracellular chloride concentration: its reversal potential follows the predicted Nernst potential for a chloride permeable channel. The glutamate-elicited current also requires the presence of extracellular sodium. This glutamate-elicited current is pharmacologically like a glutamate transporter: it is elicited, in order of efficacy, by L-glutamate, L-aspartate, L-cysteate, D-aspartate, and D-glutamate, all shown to activate glutamate transport in other systems. Furthermore, it is reduced by the glutamate transport antagonists dihydrokainate (DHKA) and D, L-threo-3–hydroxyaspartate (THA). THA, when applied alone, elicits a current similar to that elicited by glutamate. The current cannot be activated by the glutamate receptor agonists kainate, quisqualate, NMDA and APB, nor can it be blocked by the glutamale receptor antagonists CNQX and APV. Thus, the current does not appear to be mediated by a conventional glutamate receptor. Taken together, the ionic dependence and pharmacology of this current suggest that it is generated by glutamate transporter coupled to a chloride channel.

An avian retinopetal nucleus, the isthmo-optic nucleus (ION), is known to receive predominant inputs from the ipsilateral optic tectum. We injected biocytin into the ION in the Japanese quail, and retrogradely labeled tectal neurons projecting to the isthmo-optic (IO) neurons, or the tecto-IO neurons, with an extraordinary Golgi-like quality. Somata of the tecto-IO neurons were located in layer 9 of the tectum. The tecto-IO neurons did not have apical dendrites extending into superficial retino-recipient layers (layers 2–7), but had descending dendrites ramified in layers 9–12. They also possessed short ascending dendrites ramifying in the upper half of layer 9. This dendritic morphology suggests that main input to the tecto-IO neurons may not be of retinal origin. The tecto-IO neurons were spatially arranged in a regular pattern. Distances between neighboring tecto-IO neurons were 50–100 μm. The dendrites of each tecto-IO neuron were not widely dispersed in the horizontal plane, and were confined in a vertically oriented column of 100–200 μm diameter. They possessed axon collaterals extending horizontally in layers 12 and 13. The estimated total number of the tecto-IO neurons was approximately 7000–10,000, which is almost identical to the total cell number of the IO neurons. To label a small number of the tecto-IO terminals, biocytin was injected into a confined area of the optic tectum. The tecto-IO fibers densely arborized in a restricted space of the ION, which is comparable to the dimension of dendritic arborization of individual IO neurons. It is suggested that single tecto-IO neurons may make contact with single IO neurons. IO neurons are known to make synaptic contact with single target cells (association cells of Cajal) in the retina (Uchiyama & Ito, 1993; Uchiyama et al., 1995). The arborization pattern of the tecto-IO neurons' dendrites indicates that the tecto-IO neurons receive very local information in sensory and sensorimotor coordinate space. The morphology of the tecto-IO terminals suggests that the spatially confined information that drives the tecto-IO neurons is sent in parallel to single IO neurons, and then further to single retinal association cells of Cajal. Cellular-level restricted projection patterns of the tecto-isthmo-retinal system may reflect its function for centrifugal control of retinal function.

We have developed two brain slice preparations for studying tectofugal visual pathways in the chick: conventional, 400-μm slices (“thin slices”), and “thick slices” which encompass the rostral pole of the optic tectum and the contralateral optic nerve. Stimulation was delivered with a bipolar electrode positioned in stratum opticum in thin slices and in the contralateral optic nerve in thick slices. While the latter preparation provided a means of exclusively and unambiguously activating retinal afferents, several lines of evidence also indicated that the evoked field potentials in thin slices were chiefly consequent to retinal afferent excitation: (1) the similarity of evoked field potentials in thin slices to those in thick slice preparations; (2) their precise localization in retinorecipient layers as shown by prelabeling from retina with FITC-coupled cholera toxin; (3) transmission delays appropriate for retinal afferents as established with the thick slice preparation; (4) patterns of labeled afferents resulting from applications of Dil crystals to slices fixed after recording; and (5) the similarity in transmitter pharmacology between thin and thick slice preparations. Pharmacological manipulations carried out with bath-applied antagonists indicated that glutamate is the principal retinotectal transmitter. The broadly active glutamate receptor blocker, kynurenic acid, reversibly eliminated the postsynaptic component of the field potential as confirmed with 0 Ca2+ saline. A complete block was also effected by the non-NMDA antagonists CNQX and DNQX. The specific NMDA antagonist, APS, caused a smaller and variable reduction in response amplitude. The GABA antagonist, bicuculline, caused a prolongation of the monosynaptic field epsp in retinorecipient layers and an enhancement of the long-latency, negative wave in cellular layers below, supporting a late, excitation-limiting role for this inhibitory transmitter.

Three interneurons were recorded from and then injected with horseradish peroxidase in the parvocellular laminae of the squirrel monkey's (Saimiri sciureus) dorsal lateral geniculate nucleus. They were then examined using the electron microscope for their synaptic contacts, both the afferent contacts onto their dendrites and their presynaptic dendritic contacts onto presumptive projection (relay) neuron dendrites. The somata of these interneurons were small (mean = 178 μm2), but the dendritic trees were large compared with those of projection neurons. All three interneurons had similar synaptic patterns onto their dendrites with about equal numbers of retinal, cortical, and GABAergic contacts. The distribution of these contacts was more uniform compared with the same types of contacts made onto projection neurons. The presynaptic dendrites were observed to contact only the dendrites of presumptive projection neurons, and these contacts were nearly all in the form of geniculate triads. None of the three interneurons displayed an axon. The receptive fields of these interneurons were similar to those of projection cells, but were larger and had center-response signs that were the opposite of the projection neurons around them (e.g. OFF center for the dorsal part of the parvocellular mass where ON-center projection neurons reside). The squirrel monkey data provides additional evidence that one aspect of the laminar pattern observed in the parvocellular pathway of the primate's dLGN might be related to a segregation of projection neurons of one center-response sign with interneurons of the opposite center-response sign.

The organization of the primary visual cortex (VI) of the common marmoset (Callithrix jacchus) was studied both physiologically and by means of transneuronal labelling of geniculocortical afferents. We addressed the question whether monocular deprivation (MD) could stabilize segregation into ocular dominance (OD) columns, which are not seen in normal adult marmosets but are present in juvenile animals (Spatz, 1979, 1989). Properties of neurons in normal marmosets closely resembled those of other New-World and Old-World monkeys and orderly tangential progressions of preferred orientation were observed. However, in contrast to species that display well-defined OD columns, neurons of layer 4 in VI of normal adult marmosets received balanced inputs from the two eyes. Early MD (even though followed by prolonged binocular experience into adulthood) resulted in a reduction of cell size in laminae of the lateral geniculate nucleus with input from the deprived eye and a dramatic overall shift in ocular dominance towards the non-deprived eye in the cortex. However, isolated clusters of cells dominated by the deprived eye were found in both layers 4 and 6. Injection of lectin-conjugated horseradish peroxidase (WGA-HRP) into the deprived eye revealed elongated patches of terminal label, about 350 μm wide, in flat-mounted sections through layer 4. Afferent segregation was sharper and more regular in the region of VI representing parafoveal visual space than in that representing the fovea. Our findings support the notion that all Old-World and New-World monkeys possess the capacity for segregation of geniculocortical afferents into OD columns.

To examine the role of visual area V4 in pattern vision, we tested two monkeys with lesions of V4 on tasks that required them to discriminate the orientation of contours defined by several different cues. The cues used to separate the contours from their background included luminance, color, motion, and texture, as well as phase-shifted abutting gratings that created an “illusory” contour. The monkeys were trained to maintain fixation on a fixation target while discriminating extrafoveal stimuli, which were located in either a normal control quadrant of the visual field or in a quadrant affected by a lesion of area V4 in one hemisphere. Comparing performance in the two quadrants, we found significant deficits for contours defined by texture and for the illusory contour, but smaller or no deficits for motion-, color-, and luminance-defined contours. The data suggest a specific role of V4 in the perception of illusory contours and contours defined by texture.

Fast, multiple-site optical recording of voltage-sensitive dye (VSD) signals and intracellular microelectrode recordings were combined to characterize visually evoked neuronal responses in the visual cortex of the pond turtle, Pseudemys scripta. By using an in vitro, eye-brain preparation stained with the merocyanine oxazolone voltage-sensitive dye, NK-2495 or a close analog, NK-2761, large VSD signals relatively free of vibrational noise could be recorded in single trials following a stroboscopic light flash to the contralateral eye. VSD signals recorded from the same cortical location in repeated trials exhibited considerable variability in the onset, duration, and amplitude of secondary depolarizations. Because of this variability, secondary depolarizations were largely absent in signal-averaged responses. Superposition of VSD signals with intracellular recordings obtained from cortical pyramidal cells revealed a close correspondence between their signal waveforms. The two signals were virtually identical in their onset, initial rate of rise, and time-to-peak. At later periods (>500 ms), the correspondence was less close, especially for large cortical depolarizations. Some of this disparity could be attributed to contamination of the VSD signal by a large intrinsic optical response. A second contribution was a failure of the VSD signal to register asynchronous regenerative effects occurring in single pyramidal cells. It is suggested that the close correspondence between the microelectrode and optical recordings in the early phase of the response may reflect the organization of pyramidal cells into clusters that receive virtually identical synaptic inputs.

Although it is well known that neurotransmitters mediate neuron-to-neuron communication, it is becoming clear that neurotransmitters also affect glial cells. However, knowledge of neuron-to-glial signalling is limited. In this study, we examined the effects of the glutamate agonist N-methyl-D-aspartate (NMDA) on Müller cells, the predominant glia of the retina. Our immunocytochemical studies and immunodetection by Western blotting with monoclonal antibodies specific for the NMDAR1 subunit provided evidence for the expression by human Müller cells of this essential component of NMDA receptor-channels. Under conditions in which potassium currents were blocked, NMDA-induced currents could be detected in perforated-patch recordings from cultured and freshly dissociated human Müller cells. These currents were inhibited by competitive and non-competitive blockers of NMDA receptor-channels. Extracellular magnesium reduced the NMDA-activated currents in a voltage-dependent manner. However, despite a partial block by magnesium, Müller cells remained responsive to NMDA at the resting membrane potential. Under assay conditions not blocking K+ currents, exposure of Müller cells to NMDA was associated with an MK-801 sensitive inhibition of the inward-rectifying K+ current (IK(IR)), the largest current of these glia. This inhibitory effect of NMDA appears to be mediated by an influx of calcium since the inhibition of IK(IR) was significantly reduced when calcium was removed from the bathing solution or when the Müller cells contained the calcium chelator, BAPTA. Inhibition of the Müller cell KIR channels by the neurotransmitter glutamate is likely to have significant functional consequences for the retina since these ion channels are involved in K+ homeostasis, which in turn influences neuronal excitability.

We investigated the representation of color in cortical area V2 of macaque monkeys, and the association of color with other stimulus attributes. We measured the selectivity of individual V2 neurons for color, motion, and form. Most neurons in V2 were orientation selective, about half of them were selective for color, and a minority of cells (about 20%) were selective for size or direction. We correlated these physiological measurements with the anatomical location of the cells with respect to the cytochrome oxidase (CO) compartments of area V2. There was a tendency for color-selective cells to be found more frequently in the thin stripes, but color-selective cells also occurred frequently in thick stripes and inter-stripes. We found no difference in the degree of color selectivity between the different CO compartments. Furthermore, there was no negative correlation between color selectivity and selectivity for other stimulus attributes. We found many cells capable of encoding information along more than one stimulus dimension, regardless of their location with respect to the CO compartments. We suggest that area V2 plays an important role in integrating information about color, motion, and form. By this integration of stimulus attributes a cue-invariant representation of the visual world might be achieved.

The passive electrical properties of 17 ON-OFF retinal ganglion cells were derived from electrophysiological recordings. The parameters for each cells' equivalent model were obtained from the transient current responses to small step changes in clamp potential. Thirteen of the cells could be adequately approximated by a spherical soma connected to an equivalent dendritic cable. Estimates for the cell input conductance (GN), membrane time constant (τm), the dendritic-to-soma conductance ratio (ρ), and the normalized electrotonic length (L) were obtained (mean ± standard deviation, n = 13): GN = 580 ± 530 pS, τm = 97 ±72 ms, ρ = 2.8 ± 2.8, and L = 0.34 ± 0.13. Series resistance averaged 32 ± 11 MΩ The mean of the derived soma diameters was 18 ± 6 μm and the mean diameter and length of the equivalent cables were 1.4 ± 0.6 and 470 ± 90 μm, respectively. The average of the specific membrane conductances, 1.67 ± 1.08 S/cm2, corresponded to a membrane resistivity of 60 kΩ-cm2. Computer simulations of synaptic inputs were performed on a representative model, with an electrode at the soma and using the worst-case configuration, in which all synaptic inputs were confined to the tips of the dendrites. We draw three conclusions from the modeling: (1) Under voltage clamp, fast, spontaneous EPSCs would be significantly attenuated and slowed while the time course of the slower, light-evoked non-NMDA and NMDA EPSCs would be minimally distorted by dendritic filtering. (2) Excitatory synaptic reversal potentials can be accurately determined under voltage clamp. (3) In the absence of GABAergic and glycinergic inhibition, the efficacy at the soma of excitatory conductance changes is essentially independent of their dendritic location. The specific membrane resistivity appears to represent a good compromise between having a small membrane time constant and minimal EPSP attenuation.