Topographic variations in W-cell input to cat superior colliculus

Retinal origin of orientation but not direction selective maps in the superior colliculus

Neurons in the mouse superior colliculus ("colliculus") are arranged in ordered spatial maps. While orientation-selective (OS) neurons form a concentric map aligned to the center of vision, direction-selective (DS) neurons are arranged in patches with changing preferences across the visual field. It remains unclear whether these maps are a consequence of feedforward input from the retina or local computations in the colliculus. To determine whether these maps originate in the retina, we mapped the local and global distribution of OS and DS retinal ganglion cell axon boutons using in vivo two-photon calcium imaging. We found that OS boutons formed patches that matched the distribution of OS neurons within the colliculus. DS boutons displayed fewer regional specializations, better reflecting the organization of DS neurons in the retina. Both eyes convey similar orientation but different DS inputs to the colliculus, as shown in recordings from retinal explants. These data demonstrate that orientation and direction maps within the colliculus are independent, where orientation maps are likely inherited from the retina, but direction maps require additional computations.

Tonotopic and heterotopic projection systems in physiologically defined auditory cortex

Combined physiological and connectional studies show significant non-topographic extrinsic projections to frequency-specific domains in the cat auditory cortex. These frequency-mismatched loci in the thalamus, ipsilateral cortex, and commissural system complement the predicted topographic and tonotopic projections. Two tonotopic areas, the primary auditory cortex (AI) and the anterior auditory field (AAF), were electrophysiologically characterized by their frequency organization. Next, either cholera toxin beta subunit or cholera toxin beta subunit gold conjugate was injected into frequency-matched locations in each area to reveal the projection pattern from the thalamus and cortex. Most retrograde labeling was found at tonotopically appropriate locations within a 1 mm-wide strip in the thalamus and a 2-3 mm-wide expanse of cortex (approximately 85%). However, approximately 13-30% of the neurons originated from frequency-mismatched locations far from their predicted positions in thalamic nuclei and cortical areas, respectively. We propose that these heterotopic projections satisfy at least three criteria that may be necessary to support the magnitude and character of plastic changes in physiological studies. First, they are found in the thalamus, ipsilateral and commissural cortex; since this reorganization could arise from any of these routes and may involve each, such projections ought to occur in them. Second, they originate from nuclei and areas with or without tonotopy; it is likely that plasticity is not exclusively shaped by spectral influences and not limited to cochleotopic regions. Finally, the projections are appropriate in magnitude and sign to plausibly support such rearrangements; given the rapidity of some aspects of plastic changes, they should be mediated by substantial existing connections. Alternative roles for these heterotopic projections are also considered.

Physiological properties of neurons in the optic layer of the rat's superior colliculus

We made intracellular recordings from 74 neurons in the optic layer of the rat superior colliculus (SC). Resting membrane potentials were -62.3 +/- 6.2 (SD) mV, and input resistances were 37.9 +/- 10.1 MOmega. Optic layer neurons had large sodium spikes (74.2 +/- 12.3 mV) with an overshoot of 12 mV and a half-amplitude duration of 0.75 +/- 0.2 ms. Each sodium spike was followed by two afterhyperpolarizations (AHPs), one of short duration and one of longer duration, which were mediated by tetraethylammonium (TEA)-sensitive (IC) or apamin-sensitive (IAHP) calcium-activated potassium currents, respectively. Sodium spikes were also followed by an afterdepolarization (ADP), which was only revealed when the AHPs were blocked by TEA or apamin. In response to hyperpolarizing current pulses, optic layer neurons showed an inward rectification mediated by H channels. At the break of the current pulse, there was a rebound low-threshold spike (LTS) with a short duration of <25 ms. The LTS usually induced two sodium spikes (doublet). Most optic layer neurons (84%) behaved as intrinsically bursting cells. They responded to suprathreshold depolarization with an initial burst (or doublet) followed by a train of regular single spikes. The remaining 16% of cells acted as chattering cells with high-frequency gamma (20-80 Hz) rhythmic burst firing within a narrow range of depolarized potentials. The interburst frequency was voltage dependent and also time dependent, i.e., showed frequency adaptation. Unmasking the ADP with either TEA or apamin converted all of the tested intrinsically bursting cells into chattering cells, indicating that the ADP played a crucial role in the generation of rhythmic burst firing. Optic layer neurons receive direct retinal excitation mediated by both N-methyl--aspartate (NMDA) and non-NMDA receptors. Optic tract (OT) stimulation also led to gamma-aminobutyric acid-A (GABAA) receptor-mediated inhibition, the main effect of which was to curtail the excitatory response to retinal inputs by shunting the excitatory postsynaptic current. Intracellular staining with biocytin showed that the optic layer neurons that we recorded from were mostly either wide-field vertical neurons or other cells with predominately superficially projecting dendrites. These cells were similar to calbindin immunoreactive cells seen in the optic layer. The characteristics of these optic layer neurons, such as prominent AHPs, strong shunting effect of inhibition, and short-lasting LTS, suggest that they respond transiently to retinal inputs. This is consistent with a function for these cells as the first relay station in the extrageniculate visual pathway.

Overrepresentation of the central visual field in the superior colliculus of the pigmented and albino ferret

We have examined the retinotopy in the superior colliculus of pigmented and albino ferrets using both anatomical and electrophysiological methods. While the distribution of contralaterally projecting retinotectal ganglion cells is characterized by the presence of an area centralis superimposed on a visual streak in both strains, the ipsilateral projection from temporal hemiretina is strongly reduced in albinos. In spite of the significantly altered retinotectal projection pattern, the collicular visual field map in the albino ferret reveals the same characteristics as in the pigmented animal with a strongly enlarged representation of the center of visual space. An areal comparison between retinotectal ganglion cell distribution and collicular areal magnification shows that the increase in areal magnification factor between the periphery and the representation of the central visual hemifield exceeds the corresponding increase in retinal ganglion cell density between peripheral retina and area centralis by a factor of three in pigmented and a factor of four in albino ferrets. The areal magnification factor of the representation of the retinal visual streak does not exceed the increase in retinotectal ganglion cell density. Thus, our results suggest that the representation of visual space in the superior colliculus of albino and pigmented ferrets does not simply follow the retinotectal ganglion cell density, but that there is an enhanced representation of the frontal central visual field. The possibility is discussed that the collicular visual field map may be determined either by both retinotectal and corticotectal projections or by the colliculus' intrinsic structure.

Retinotopic organization of the superior colliculus in relation to the retinal distribution of afferent ganglion cells

Sensory representations in the brain exhibit topographic variations in magnification. These variations have been thought to reflect regional differences in the density of innervation at the sensory receptor surface. In the primate visual cortex, for example, local magnification factors have been reported to be proportional to the corresponding densities of retinal ganglion cells. We sought to learn whether this principle also operates in a second major retinofugal pathway--the projection to the superior colliculus. In cats, we first used retrograde transport to determine the retinal distributions of the ganglion cells that project to the colliculus. Then, we compared the numbers of colliculopetal ganglion cells in selected retinal sectors to the areas of the corresponding collicular representations. Collicular areal magnification was not simply proportional to the density of afferent ganglion cells, being instead at least 5-fold greater than expected in the representation of the central visual field. These data imply that incoming retinal afferents are more widely spaced in the central regions of the tectal map than in the map's periphery. Such variations in afferent density appear to play as large a role as the distribution of ganglion cells in determining the metric of the collicular map.

The morphology of collicular and retinal axons ending on small relay (W-like) cells of the primate lateral geniculate nucleus

The lateral geniculate nucleus (LGN) of every primate examined contains a set of small relay cells in addition to separate sets of magnocellular and parvocellular relay cells. These small cells receive a direct retinal projection, and an indirect retinal projection via the superior colliculus (SC). Receptive-field analyses of the small LGN cells in the bush baby, a lorisiform primate, indicate that this cell class is composed of subclasses, similar in physiology to cat W cells. In an effort to identify some of these subclasses, we have examined the morphological features of retinal and collicular axonal arbors that end on small W-like cells in the LGN of the bush baby, Galago crassicaudatus. Small cells in this species are found in a prominent pair of koniocellular (K) layers as well as the interlaminar zones (ILZs). Retinal arbors were examined by bulk iontophoretic injection of horseradish peroxidase into the optic tract. Collicular arbors were filled via iontophoretic injection of biocytin into the superficial layers of the SC. Forty-eight axon arbors were completely reconstructed and quantitatively evaluated. Our findings show that retinal and collicular axon terminals differ in morphology on the basis of a number of criteria. Our analyses also suggest that retinal axons may have a stronger influence on K cells and collicular axons have a stronger influence of ILZ cells. The ramifications of these findings are provocative since these small LGN cells are known to project directly to the cytochrome-oxidase (CO) blobs within striate cortex. This relationship suggests that CO blob cells receive complex visual input not only from magnocellular and parvocellular LGN cells, but also from small cell pathways that are differentially influenced by retinal and collicular cells.

Evidence for intrinsic expression of enkephalin-like immunoreactivity and opioid binding sites in cat superior colliculus

We have investigated the cellular localization of opioid peptides and binding sites in the cat's superior colliculus by testing the effects of retinal deafferentation and intracollicular excitotoxin lesions on patterns of enkephalin-like immunostaining and opiate receptor ligand binding. In normal cats, enkephalin-like immunoreactivity marks a thin tier in the most dorsal stratum griseum superficiale, small neurons of the stratum griseum superficiale, and patches of fibers in the intermediate and deeper gray layers. Eliminating crossed retinotectal afferents by contralateral eye enucleation had little immediate effect on this pattern, although chronic eye enucleation from birth did reduce immunoreactivity in the superficial layers. By contrast, fiber-sparing destruction of collicular neurons by the excitotoxins N-methyl-D-aspartate and ibotenic acid virtually eliminated enkephalin-like immunoreactivity in the neuropil of the upper stratum griseum superficiale, presumably by killing enkephalinergic cells of the superficial layers. Such lesions did not eliminate the patches of enkephalin-like immunoreactivity in the deeper layers. In normal cats, opiate receptor ligand binding is dense in the stratum griseum superficiale, particularly in its upper tier, and moderately dense in the intermediate gray layer. Contralateral eye removal had no detectable effect on the binding pattern, but excitotoxin lesions of the colliculus dramatically reduced binding in both superficial and deep layers. Some ligand binding, including part of that in the upper stratum griseum superficiale, apparently survived such lesions. Similar effects were observed in the lateral geniculate nucleus: enucleation produced no change in binding, whereas excitotoxin lesions greatly reduced specific opiate binding. We conclude that in the superficial collicular layers, both enkephalin-like opioid peptides and their membrane receptors are largely expressed by neurons of intrinsic collicular origin. The close correspondence between the location of these intrinsic opioid elements and the tier of retinal afferents terminating in the upper stratum griseum superficiale further suggests that opiatergic interneurons may modulate retinotectal transmission postsynaptically.