Anita E. Hendrickson

The autoradiographic demonstration of axonal connections in the central nervous system.

Most of the published work on the phenomenon of axoplasmic transport has been directed towards determining the site of synthesis and the chemical nature of materials involved, the mechanisms responsible for their movement, and the rates at which they are transported (see refs. 1, 23, 44, 49, 53 for reviews). However, following the suggestion of Taylor and Weiss 68, several workers, notably Lasek et al. 45, have shown that the transport of radioactively labeled materials can be used to trace axonal connections autoradiographically. This approach has thus far been applied to the study of the central connections of certain peripherally located neurons (specifically the dorsal root ganglion cells 45, the ganglion cells of the retina 19,a°,aa,5°,65 and the olfactory receptorsT1). Although this method does not appear to have been used for tracing pathways arising and terminating within the central nervous system, it should be possible to do this by locally injecting radioactively labeled precursors of proteins or other macromolecules into the brain or spinal cord. Indeed, there are several reasons for thinking that this method may offer a number of advantages over other currently available techniques4L Nearly all experimental neuroanatomical techniques (including the various cell degeneration methods, the reduced silver techniques for impregnating degenerating axons such as the Nauta-Gygax method and its recent variants and the electron microscopic examination of degenerating axons and their terminals) are dependent upon a sequence of pathological changes following injury to the nervous system. This essential feature underlies a number of the problems encountered in the use and interpretation of these methods. Among the most significant difficulties is the fact that experimental lesions are usually non-specific, and destroy not only the neurons within the intended focus but also the fibers passing through that region and the axons terminating within it. Since degeneration inevitably alters the normal morphol-

A qualitative and quantitative analysis of the human fovea during development

The anatomical development of the human fovea has been sampled from 22 weeks gestation to adulthood, using both qualitative and quantitative methods. The foveal depression continues to deepen after birth until 15 months, due to the migration of the cells of the inner retina toward the periphery. Before birth the rod-free zone or foveola is over 1000 microns in diameter, but it becomes progressively narrower after birth because of a centralward migration of cones. It reaches the adult diameter of 650-700 microns by 45 months of age. Postnatally, foveolar cone development is characterized by maturation, elongation, and an increase in packing density. Foveolar cone diameter changes markedly after birth, going from 7.5 microns at 5 days postnatal to 2 microns by 45 months. During this time the foveolar cone develops both its outer segment and basal axon process (fiber of Henle). This combination of elongation and centralward migration results in an increase of foveolar cone density from 18 cones/100 microns at 1 week postnatal to 42 cones/100 microns in the adult. Measures of foveola width and cone diameter reach the adult stage of development at 45 months of age, but the two important visual factors of outer segment length and cone packing density still are only half the adult values at 45 months of age.

An autoradiographic and electron microscopic study of retino-hypothalamic connections

SummaryRetino-hypothalamic connections have been studied autoradiographically in the rat, guinea pig, rabbit, cat and monkey following the intravitreal injection of 3H-leucine or 3H-proline, and electron microscopically following unilateral eye removal in the guinea pig and monkey. In each of the species examined evidence has been found for a direct projection from the retina to the suprachiasmatic nucleus, but to no other region of the hypothalamus. The projection to the suprachiasmatic nucleus is always bilateral (even in the albino guinea pig, in which all other components of the retinal projection are crossed) but from grain counts in our autoradiographs it appears that the input to the contralateral nucleus is about twice as heavy as that on the ipsilateral side. Most of the retinal fibers appear to terminate within the ventral part of the nucleus where they form asymmetric synapses either upon small dendritic branches or dendritic spines. The possible role of this retinal projection to the suprachiasmatic nucleus in mediating a variety of light-induced neuroendocrine responses is discussed.

The Morphological Development of the Human Fovea

The development of the human fovea has been traced from 22 weeks gestation to 45 months postpartum using aldehyde-fixed, plastic embedded, serially-sectioned normal retinas. Five anatomical indicators of foveal maturity were used in this study: the shape of the foveal curvatures; the presence of the transient layer of Chievitz; the width of rod-free zone in the central retina; the width and length of the individual foveal cones; and the number and thickness of layers of nuclei within the fovea. The future fovea is identifiable at 22 weeks by the presence of a thick layer of ganglion cells and a photoreceptor layer containing only cones. By 1 week after birth, there is a shallow foveal depression, but the thick cones still lack outer segments and are only 1 cell deep in the fovea. The inner nuclear layer contains a thick transient layer of Chievitz. As judged by these anatomical criteria and compared to normal adult foveas similarly processed, the human fovea reaches maturity between 15 and 45 months of age.

Pathways between striate cortex and subcortical regions in Macaca mulatta and Saimiri sciureus: evidence for a reciprocal pulvinar connection.

The efferent and afferent connections of striate cortex have been compared in Macaca mulatta and Saimiri sciureus monkeys after injecting striate cortex with a mixture of horseradish peroxidase and tritiated amino acids. Three reciprocal pathways were found in Macaca; they were between striate cortex and (i) all layers of the dorsal lateral geniculate nucleus; (ii) the inferior subdivision of pulvinar; and (iii) the lateral subdivision of pulvinar. In Saimiri the only recprocal pathway involved the dorsal lateral geniculate nucleus; the pulvinar received a heavy striate input but too few peroxidase-labeled neurons were found in pulvinar to demonstrate convincingly a projection to striate cortex. In both species the reticular and posterior nucleus of the thalamus and the superior colliculus receive striate input; in Saimiri the ventral lateral geniculate also receives a striate projection. In both species, neurons in the basal forebrain project to striate cortex. Striate cortex synaptic terminals in both dorsal lateral geniculate nucleus and pulvinar are topographically organized. Inferior and lateral subdivisions of pulvinar each contain a representation of central retina, and the fiber bundles separating the two subdivisions apparently mark the vertical meridian. The results of this experiment suggest that both retino-geniculate and retino-superior colliculus-pulvinar types of visual information may converge within the striate cortex.

Local circuit neurons in the rat ventrobasal thalamus—A gaba immunocytochemical study

The ventrobasal thalamus of seven rats was processed for immunocytochemistry using antisera to glutamate decarboxylase or gamma-aminobutyrate (GABA). Glutamate decarboxylase-stained sections showed a network of stained fibers and terminals but no stained cell bodies. GABA-stained sections had fewer stained fibers and terminals but did show a few stained cell bodies. Cell bodies were especially apparent when carbazole was used for a chromogen for the peroxidase-antiperoxidase visualization. The GABA-stained cells were found to be distributed throughout the ventrobasal complex, to have smaller soma cross-sectional areas than most other cells (81 +/- 34 microns vs 105 +/- 36 microns for all cells) and to make up 0.4 +/- 0.3% of the neuronal population of the ventrobasal complex. Injections of horseradish peroxidase into the somatosensory cortex (SI) retrogradely filled many neurons in the ventrobasal thalamus, but none of these labeled neurons were double labeled with GABA. These results indicate that the GABA-labeled cells probably represent a small population of local circuit neurons in the rat ventrobasal thalamus.

Distribution of the calcium-binding proteins parvalbumin and calbindin-D28k in the sensorimotor cortex of the rat

This study examined and compared the immunocytochemical distribution of the two calcium-binding proteins parvalbumin and calbindin-D28k in the primary motor and somatosensory areas of the rat neocortex. Parvalbumin-immunoreactive cells were found in all layers of the cortex except layer 1 and reached their peak density in the middle layers. The two cortical areas differed markedly in the number, cell size and morphology of immunoreactive cells. Parvalbumin-positive cells were more than twice as numerous in the somatosensory cortex compared to the motor cortex. In addition, the average size of their cell bodies was 25-30% larger in the somatosensory area. Parvalbumin cells in the motor area represented several classes of nonpyramidal cells, while the somatosensory cortex contained in addition many large cells with thick vertically oriented primary dendrites. Some of these cells resembled regular or inverted pyramidal neurons. Punctate neuropil labeling was much heavier in the upper layers of the somatosensory than in the motor cortex and was especially heavy in layer 4. Dense parvalbumin-positive perisomatic puncta surrounded large, unstained pyramidal cells in layer 5B of the motor cortex. Calbindin-D28k neuronal staining in both areas was confined to two populations. The most prominent was darkly labeled, small nonpyramidal cells confined to two bands in layers 2/3 and 5/6. There was also a lighter stained population composed of many pyramidal cells distributed throughout layers 2 and 3. In addition, the motor area contained a band of lightly stained, large pyramidal cells in layer 5B. Calbindin-D28k neuropil labeling was heaviest in layers 1 to 3. In contrast to parvalbumin, we found only minor differences in distribution, size and morphology of calbindin-D28k cell body or neuropil staining in the two cortical areas. Double-labeling immunocytochemistry showed that the large majority of immunoreactive cells contained only calbindin-D28k or parvalbumin, but a distinct population of multipolar cells in the upper layers of the somatosensory cortex contained both. The clear parcellation of parvalbumin immunoreactivity in the rat neocortex suggests that parvalbumin is preferentially associated with specific neuronal populations and terminals in the somatosensory cortex. The more general and homogeneous labeling of the upper layers of the cortex indicates that calbindin-D28k could be related to the relatively high density of calcium channels or N-methyl-D-aspartate receptors in the superficial layers of the rat cortex.

Spatial and temporal expression of short, long/medium, or both opsins in human fetal cones.

Human cone photoreceptors are characterized by long (L), medium (M), or short (S) wavelength‐specific opsin. No reports have described the developmental pattern of human cone opsin expression, nor has the existence of human cones containing more than one opsin been tested. Single‐and double‐label immunocytochemistry and in situ hybridization have been used to determine the developmental pattern of opsin appearance and to investigate the presence of double‐labeled cones in sections and wholemounts of human fetal, neonatal, infant, and adult retina. S opsin protein appears in and around the fovea at fetal week (Fwk) 10.9, whereas L/M opsin first appears in the fovea at Fwk 14–15. S opsin mRNA and protein are consistently detected much farther into peripheral retina than L/M opsin, indicating that S appears before L/M opsin. S cones cover 90% of the retina by Fwk 19. L/M cones appear outside the central retina by Fwk 21.5 and reach the retinal edge by Fwk 34–37. The spatial pattern of mRNA expression closely matches that for protein, but mRNA appears slightly earlier than protein at a given retinal point, indicating that only short delays occur between mRNA expression and translation into protein. Cones containing both S and L/M opsin (S+L/M) appear around the fovea shortly after L/M opsin is expressed, are found in more peripheral retina at older ages, and decrease in number after birth. Some S+L/M cones are still detected in adult retina. Both S opsin protein and mRNA appear significantly earlier than L/M mRNA or protein across the human retina, suggesting that the two cone types differentiate under independent controlling factors. However, the presence of single cones containing both S and L/M opsin during development suggests that human cones can respond to the factors controlling expression of each opsin. J. Comp. Neurol. 425:545–559, 2000.

The development of parafoveal and mid-peripheral human retina.

The morphological development of parafoveal retina (1-1.5 mm from the foveal center) and the mid-peripheral (4 mm from the foveal center) human retina has been studied from fetal (F) 26 weeks to adulthood. At both retinal points, all layers and neuronal types are present at F26 weeks. In parafovea at F26 weeks photoreceptors have only a rudimentary inner segment and no outer segments. Short outer segments are present on both rods and cones at F36 weeks. By postnatal (P) 5-8 days the inner retina is relatively mature. Photoreceptors have elongated basal axons which cause the photoreceptor layer to become much thicker than in prenatal retina. At birth cone inner segments are untapered, but rod inner segments have already reached their adult width of 2 microns. Both rod and cone inner and outer segments are 30-50% of adult length. By 13 months both inner and outer retina are mature appearing, with the photoreceptors accounting for half the retinal thickness due to the elongation of the fibers of Henle. Cone outer segments elongate up to P5 years and rod outer segments to P13 years. At mid-peripheral or rod-ring retina outer segments are present on rods at F26 weeks and on cones at F36 weeks. At birth the inner retina is adultlike. The outer plexiform layer becomes thicker up to P45 months due to the elongation of fibers of Henle. At birth both rod and cone mid-peripheral inner segments are slightly longer and outer segments are 50% longer than in parafoveal retina. By P5 years mid-peripheral rod outer segments are slightly longer than in parafoveal retina, and this changes little thereafter. This anatomical study has found that the photoreceptors in peripheral rod-ring retina develop earlier than those in more central retina, and in turn parafoveal photoreceptors develop well in advance of foveal cones. This suggests that human neonates may utilize more peripheral retinal regions for some aspects of visual function before foveal cone vision becomes dominant.

Distribution of the glycine transporter glyt-1 in mammalian and nonmammalian retinae

We have examined the distribution of the glycine transporter glyt-1 in retinae of macaques, cats, rabbits, rats, and chickens. In all species, all glycine-containing amacrine cells expressed immunoreactivity for glyt-1, though the intensity of immunoreactivity for glyt-1 did not appear to directly correlate with the intensity of immunoreactivity for glycine in individual cells. A small subpopulation of glycine-immunoreactive displaced amacrine cells or ganglion cells also expressed glyt-1 in retinae from macaques, cats, chickens, and rats but not in retinae from rabbits. In addition, in all species examined, some displaced amacrine cells also contained glycine but did not express glyt-1. In monkeys, cats, and rats, populations of cells which we interpret as being glycine-containing interplexiform cells expressed glyt-1: these cells lacked a content of glutamate, suggesting they are not bipolar cells. The glycine-containing bipolar cells did not express glyt-1, suggesting that these cells probably acquired their content of glycine by other means such as via gap junctional connections with glycine-containing amacrine cells.