Morphogenesis of retinal ganglion cells during formation of the fovea in the Rhesus macaque

Glia of the human retina

The human retina contains three types of glial cells: microglia and two types of macroglia, astrocytes and Müller cells. Macroglia provide homeostatic and metabolic support to photoreceptors and neurons required for neuronal activity. The fovea, the site of the sharpest vision which is astrocyte- and microglia-free, contains two populations of Müller glia: cells which form the Müller cell cone in the foveola and z-shaped Müller cells of the foveal walls. Both populations are characterized by morphological and functional differences. Müller cells of the foveola do not support the activity of photoreceptors and neurons, but provide the structural stability of the foveal tissue and improve the light transmission through the tissue to the photoreceptors. This article gives overviews of the glia of the human retina and the structure and function of both Müller cell types in the fovea, and describes the contributions of astrocytes and Müller cells to the ontogenetic development of the fovea.

The primate fovea: Structure, function and development

A fovea is a pitted invagination in the inner retinal tissue (fovea interna) that overlies an area of photoreceptors specialized for high acuity vision (fovea externa). Although the shape of the vertebrate fovea varies considerably among the species, there are two basic types. The retina of many predatory fish, reptilians, and birds possess one (or two) convexiclivate fovea(s), while the retina of higher primates contains a concaviclivate fovea. By refraction of the incoming light, the convexiclivate fovea may function as image enlarger, focus indicator, and movement detector. By centrifugal displacement of the inner retinal layers, which increases the transparency of the central foveal tissue (the foveola), the primate fovea interna improves the quality of the image received by the central photoreceptors. In this review, we summarize ‒ with the focus on Müller cells of the human and macaque fovea ‒ data regarding the structure of the primate fovea, discuss various aspects of the optical function of the fovea, and propose a model of foveal development. The "Müller cell cone" of the foveola comprises specialized Müller cells which do not support neuronal activity but may serve optical and structural functions. In addition to the "Müller cell cone", structural stabilization of the foveal morphology may be provided by the 'z-shaped' Müller cells of the fovea walls, via exerting tractional forces onto Henle fibers. The spatial distribution of glial fibrillary acidic protein may suggest that the foveola and the Henle fiber layer are subjects to mechanical stress. During development, the foveal pit is proposed to be formed by a vertical contraction of the centralmost Müller cells. After widening of the foveal pit likely mediated by retracting astrocytes, Henle fibers are formed by horizontal contraction of Müller cell processes in the outer plexiform layer and the centripetal displacement of photoreceptors. A better understanding of the molecular, cellular, and mechanical factors involved in the developmental morphogenesis and the structural stabilization of the fovea may help to explain the (patho-) genesis of foveal hypoplasia and macular holes.

Retinal microstructures are altered in patients with idiopathic infantile nystagmus

PURPOSE

To compare segmented retinal layer thicknesses between patients with idiopathic infantile nystagmus (IIN) and controls.

METHODS

This retrospective case-control study included 66 patients with IIN and 66 age-matched controls. The retinal layers were examined using spectral domain optical coherence tomography with autosegmentation. Central foveal thickness (CFT), outer nuclear layer (ONL), and outer segment length (OSL) thickness were measured at the fovea center. Mean values for retinal nerve fiber layer, ganglion cell inner plexiform layer (GCIPL), inner nuclear layer, outer plexiform-outer nuclear layer (OPNL) thicknesses were calculated at two measurement points (nasal and temporal hump points at the macula area).

RESULTS

There were no significant between-group differences in age, gender, or refraction error. The CFT was thicker in the IIN group compared with the control group (225.0 μm vs. 217.8 μm, P = 0.017) and OSL was shorter in IIN than in controls (40.0 μm vs. 43.7 μm., P < 0.001). The ONL thickness at the central fovea was not statistically different between the two groups. At the nasal and temporal position where the ganglion cell density was thickest, the GCIPL thickness was thinner in the IIN group compared to the controls (99.5 μm vs. 102.8 μm, P = 0.010). The GCIPL thickness was negatively correlated with logMAR visual acuity (Spearman's rho = -0.502, P < 0.001).

CONCLUSIONS

The foveal pit was shallower, OSL was shorter, and the GCIPL thicknesses at macular humps were decreased in the patients with IIN compared with that of controls. The faulty development of the macula may be related to unknown pathophysiologic mechanism during fovea maturation in IIN or continuous eye movement itself interrupt fovea development.

Visual pathway for the optokinetic reflex in infant macaque monkeys

The horizontal optokinetic nystagmus (hOKN) in primates is immature at birth. To elucidate the early functional state of the visual pathway for hOKN, retinal slip neurons were recorded in the nucleus of the optic tract and dorsal terminal nucleus (NOT-DTN) of 4 anesthetized infant macaques. These neurons were direction selective for ipsiversive stimulus movement shortly after birth [postnatal day 9 (P9)], although at a lower direction selectivity index (DSI). The DSI in the older infants (P12, P14, P60) was not different from adults. A total of 96% of NOT-DTN neurons in P9, P12, and P14 were binocular, however, significantly more often dominated by the contralateral eye than in adults. Already in the youngest animals, NOT-DTN neurons were well tuned to different stimulus velocities; however, tuning was truncated toward lower stimulus velocities when compared with adults. As early as at P12, electrical stimulation in V1 elicited orthodromic responses in the NOT-DTN. However, the incidence of activated neurons was much lower in infants (40-60% of the tested NOT-DTN neurons) than in adults (97%). Orthodromic latencies from V1 were significantly longer in P12-P14 (x = 12.2 ± 8.9 ms) than in adults (x = 3.51 ± 0.81 ms). At the same age, electrical stimulation in motion-sensitive area MT was more efficient in activating NOT-DTN neurons (80% of the tested cells) and yielded shorter latencies than in V1 (x = 7.8 ± 3.02 ms; adult x = 2.99 ± 0.85 ms). The differences in discharge rate between neurons in the NOT-DTN contra- and ipsilateral to the stimulated eye are equivalent to the gain asymmetry between monocularly elicited OKN in temporonasal and nasotemporal direction at the various ages.

The fovea regulates symmetrical development of the visual cortex

The foveal region contains the highest cell density in the human retina; consequently a disproportionately large area of the visual cortex is dedicated to its representation. In aniridia and albinism the fovea does not develop, and the corresponding cortical representation shows a reduction in gray matter volume. In albinos there are chiasmatic irregularities in the hemispheric projections, which are not found in aniridics. Here, we ask whether the anomalies in central retinal development, present in albinism and aniridia, have a wider impact on the architecture of the visual cortex. The length, depth, and topology of the calcarine fissure is analyzed in albino, aniridic, and normal subjects. These measures are compared between groups and between the cortical hemispheres within each subject. We show that the calcarine fissure, where the primary visual cortex is represented, is abnormally short in those lacking a fovea. Moreover, surface reconstructions of the calcarine fissure revealed marked interhemispheric asymmetries. The two groups could not be distinguished on the basis of their cortical features, and we therefore interpret the abnormalities in cortical architecture in terms of the absence of the fovea, the common retinal feature found in both groups.

VEGF expression by ganglion cells in central retina before formation of the foveal depression in monkey retina: evidence of developmental hypoxia

In macaque monkeys the foveal depression forms between fetal day (Fd) 105 and birth (Fd 172 of gestation). Before this, the incipient fovea is identified by a photoreceptor layer comprising cones almost exclusively, a multilayered ganglion cell layer (GCL), and a "domed" profile. Vessels are absent from the central retina until late in development, leading to the suggestion that the GCL in the incipient fovea may be transitorily hypoxic. Vascular endothelial growth factor (VEGF), expressed by both glial and neuronal cells and mediated by the hypoxia-inducible transcription factor (HIF)-1, is the principal factor involved in blood vessel growth in the retina. We examined VEGF expression in macaque retinas between Fd 85 and 4 months postnatal. Digoxygenin-labeled riboprobes were generated from a partial-length human cDNA polymerase chain reaction fragment, detected using fluorescence confocal microscopy, and quantified using Scion Image. High levels of VEGF mRNA were detected in astrocytes associated with developing vessels. We also detected strong expression of VEGF mRNA in the GCL at the incipient fovea prior to Fd 105, with peak labeling in the incipient fovea that declined with distance in nasal and temporal directions. By Fd 152 peak labeling was in two bands associated with development of the inner nuclear layer (INL) capillary plexus: in the inner INL where Müller and amacrine cell somas are located, and in the outer INL where horizontal cells are found. The findings suggest that at the incipient fovea the GCL is hypoxic, supporting the hypothesis that the adaptive significance of the fovea centralis is in ensuring adequate oxygen supply to neuronal elements initially located within the avascular region.

Developmental changes in astrocyte density in the macaque perifoveal region

We studied astrocyte density both in the perifoveal region and in extrafoveal regions within the same distance of the optic disc (OD) over a time period from foveal pit formation (embryonic day E112) until 2 months after birth. The study was prompted by earlier observations that the adult macaque displays an almost astrocyte-free region around the fovea which, however, at birth is occupied by astrocytes. Thus, we wanted to determine if the perifoveal region is invaded by astrocytes during early development to the same degree as other regions in the central retina, and how the reduction in density can be explained. From the earliest age we studied (embryonic day 112), less astrocytes were found in the perifovea than in other regions equidistant from the OD. In addition, the number of astrocytes steadily declined both in the perifovea and outside until birth. During the first week after birth, there was a further dramatic decline in perifoveal astrocyte density. Double-labelling with glial fibrillary acidic protein (GFAP) immunocytochemistry and the TUNEL method showed that during the whole observation period astrocytes undergo DNA fragmentation and presumably die. However, the rate of TUNEL-positive astrocytes did not significantly differ between perifovea and other regions equidistant to the OD, and at no time did we find a significant peak of apoptosis rate. Thus, the reduction in perifoveal astrocyte density cannot be explained by missing invasion or by selectively elevated apoptosis rates in the foveal and perifoveal regions. Alternative hypotheses are discussed.

The human fetal retinal nerve fiber layer and optic nerve head: a DiI and DiA tracing study

The organization of the primate nerve fiber layer and optic nerve head with respect to the positioning of central and peripheral axons remains controversial. Data were obtained from 32 human fetal retinae aged between 15 and 21 weeks of gestation. Crystals of the carbocyanine dyes, DiI or DiA, and fluorescence microscopy were used to identify axonal populations from peripheral retinal ganglion cells. Peripheral ganglion cell axons were scattered throughout the vitreal-scleral depth of the nerve fiber layer. Such a scattered distribution was maintained as the fibers passed through the optic nerve head and along the optic nerve. There was a rough topographic representation within the optic nerve head according to retinal quadrant such that both peripheral and central fibers were mixed within a wedge extending from the periphery to the center of the nerve. There was no indication that the fibers were reorganized in any way as they passed through the optic disc and into the nerve. The present results suggest that any degree of order present within the fiber layer and optic nerve is not an active process but a passive consequence of combining the fascicles of the retinal nerve fiber layer. Optic axons are not instructed to establish a retinotopic order and the effect of guidance cues in reordering fibers, particularly evident prechiasmatically and postchiasmatically, does not appear to be present within the nerve fiber layer or optic nerve head in humans.

A tension-based theory of morphogenesis and compact wiring in the central nervous system

Many structural features of the mammalian central nervous system can be explained by a morphogenetic mechanism that involves mechanical tension along axons, dendrites and glial processes. In the cerebral cortex, for example, tension along axons in the white matter can explain how and why the cortex folds in a characteristic species-specific pattern. In the cerebellum, tension along parallel fibres can explain why the cortex is highly elongated but folded like an accordion. By keeping the aggregate length of axonal and dendritic wiring low, tension should contribute to the compactness of neural circuitry throughout the adult brain.

Normal and pathological mechanisms in retinal vascular development

Angiogenesis is a complex biologic process that occurs normally in development and in turnover and remodeling of mature vascular networks. Pathological angiogenesis and neovascularization occur in association with retinal and ocular ischemic diseases, in retinopathy of prematurity and other developmental disorders, and in tumor growth and metastasis. We describe current understanding of cellular and molecular mechanisms of retinal vascular development, highlighting aspects that relate to eye diseases, that provide sites of therapeutic intervention in ophthalmology and that are potential avenues for research.

Transience of astrocytes in the newborn macaque monkey retina

We investigated the morphology and spatial distribution of retinal astrocytes in newborn and early postnatal macaque monkeys. As in adults, retinal astrocytes in neonatal animals were closely associated with ganglion cell axons and blood vessels. However, in contrast with adults, astrocytes transiently occupied the fovea and perifoveal region in newborns and, to a lesser degree, also in early postnatal animals. The density of the perifoveal astrocytes rapidly declined during the first 2-3 months of life. The results are discussed in relationship to foveal development.

Comparison of photoreceptor spatial density and ganglion cell morphology in the retina of human, macaque monkey, cat, and the marmoset Callithrix jacchus

We studied the relationship between the morphology of ganglion cells and the spatial density of photoreceptors in the retina of two Old World primates, human and macaque monkey; the diurnal New World marmoset Callithrix jacchus; and the cat. Ganglion cells in macaque and marmoset were labelled by intracellular injection with Neurobiotin or by DiI diffusion labelling in fixed tissue. Cone photoreceptor densities were measured from the same retinas. Supplemental data for macaque and data for human and cat were taken from published studies. For the primates studied, the central retina is characterised by a constant numerical convergence of cones to ganglion cells. Midget ganglion cells derive their input, via a midget bipolar cell, from a single cone. Parasol cells derive their input from 40-140 cones. Outside the central retina, the convergence increases with eccentricity. The convergence to beta cells in the cat retina is very close to that for parasol cells in primate retina. The convergence of rod photoreceptors to ganglion cells is similar in human, macaque, and marmoset, with parasol cells receiving input from 10-15 times more rods than midget cells. The low convergence of cones to midget cells in human and macaque retinas is associated with distinctive dendritic "clusters" in midget cells' dendritic fields. Convergence in marmoset is higher, and the clusters are absent. We conclude that the complementary changes in photoreceptor density and ganglion cell morphology should be considered when forming linking hypotheses between dendritic field, receptive field, and psychophysical properties of primate vision.

Quantitative analysis of synaptogenesis in the inner plexiform layer of macaque monkey fovea

Synaptogenesis has been tracked by using quantitative electron microscopic methods in the inner plexiform layer (IPL) of the developing Macaca monkey fovea from fetal day (Fd) 55 to Fd132. Vesicle-containing profiles were classified according to whether (1) they contained a ribbon indicating that they originated from a bipolar cell, or (2) the profile formed a junction. Group 2 was further subdivided by morphological characteristics into (2a) amacrine, (2b) bipolar, or (2c) unknown profiles. Ribbon-containing bipolar profiles are clearly identifiable at Fd55 when they occur at a density of 0.9/100 microns2. Bipolar synapses increase rapidly to 4.7/100 microns2 by Fd88, similar to their density at Fd132. Identifiable amacrine profiles forming a junction are rare at Fd55-68. By Fd88, amacrine synaptic density has jumped to 6.7/100 microns2 and continues to increase to 9.5/100 microns2 at Fd132. These quantitative data strongly suggest that, at the Macaca fovea, bipolar synaptogenesis both begins and ends before amacrine synaptogenesis. The large number of immature amacrine synaptic profiles and densities at Fd132 suggests that amacrine synapses continue to form after Fd132. This study confirms that cone-dominated monkey fovea has a different sequence of synaptogenesis than the rod-dominated peripheral retina (Nishimura and Rakic, [1985] J. Comp. Neurol 241:420-434). The data support the concept that synaptic developmental sequence is determined by the type of photoreceptor which dominates a particular retinal region or species. Bipolar ribbon synapses are observed in the outer half of the IPL at Fd55, are present in the inner IPL at Fd60, and then, with increasing age, are found throughout the IPL. This pattern strongly suggests that vertical OFF bipolar pathways form earlier than ON pathways in the IPL. In contrast, amacrine profiles are found throughout the IPL at the youngest ages, with an adult-like banding pattern present by Fd132.

Early axon outgrowth of retinal ganglion cells in the fetal rhesus macaque

Employing retinal explants and retrograde transport techniques, we studied the formation of the arcuate fascicles by examining the growth of the central retina, the emergence of the adult fiber layer pattern, and the projections of retinal ganglion cells in the central and peripheral retina. Sixty days prior to foveal pit formation, the distance from the incipient fovea to the optic disk was equal to the adult, even though the retinal area was only 8% of the adult. Arcuate fibers, at this age, were observed to avoid the incipient fovea, with no fascicles and few axons projecting over this region. A small population of 15.2% of the ganglion cells located within 2 mm of the incipient fovea possessed an axon with an aberrant trajectory that wound around and projected 50 to several hundred microns away from the optic disk, compared to only 3% at other retinal locations. The incidence of disorder decreased with increasing fetal age, establishing mature values in late fetal periods. These findings suggest that the area of the central retina does not increase after embryonic day 60 and that guidance factors are present that allow outgrowing ganglion cell axons to distinguish and avoid that portion of the retina that will become the fovea.