Genesis of the retinal pigment epithelium in the macaque monkey

The development of the retinal pigment epithelium (RPE) was studied in rhesus monkey (Macaca mulatta) fetuses, neonates, and juveniles exposed to a pulse of (³H-TdR) between embryonic day (E) 25 and postnatal day (P) 204 and examined at short and long intervals after the injection of the isotope. The RPE develops from the outer layer of the optic cup which by E45 consists of a multistratified epithelium. The outer layer appears immature near the retina's edge and gradually becomes monostratified and more mature centrally. Even at this early stage, all cells contain pigmented melanosomes, although peripherally the pigment is limited to the apical portion of the cells. Examination of autoradiograms from animals allowed to survive for several postnatal months shows that monkey RPE cell genesis begins just after E27, increasing to a peak frequency of 0. 38 cells/mm at E43. Between E30 and E85 the density of radiolabelled cells varies within a restricted range of from 0.2 to 0.4 cells/mm (mean = 0.25 ± 0.09). From the density of radiolabelled cells, and data on the overall density of RPE cells in the juvenile retina, we determined the labelling index. During the first half of gestation, between 0. 38% and 0. 99% (mean = 0. 65 ± 0. 22) of RPE cells are generated during the short interval of isotope availability after pulse injection. Approximately 5% of RPE cells were generated by E33, and 50% by E71. After E85, RPE cytogenesis begins gradually to decrease, and 95% of the cells have been generated by the time of birth. Continued, very low density (0. 01 cells/mm) cytogenesis in the RPE is seen at P17, and persists at least until seven months postnatally. RPE cell genesis begin near the fovea, and proceeds towards the periphery. Cell division largely ceases in both foveal and perifoveal regions by E56, at which time labelled cells first begin to appear peripheral to the equator. Besides the timing differences, RPE genesis in the central retina differs from that in the peripheral retina in that it proceeds at a higher rate, and lasts for a shorter time period. A prolonged postnatal period of low density RPE cell genesis persists in both central and peripheral retina. Comparison of the pattern of expansion of the area containing radiolabelled cells in the RPE and neuroretina demonstrates a remarkable spatial and temporal correspondence. Close analysis suggests that at any point on the retina, the last cells are generated in the neuroretina slightly before the last cells in the RPE. © 1995 Wiley-Liss, Inc.     

Histogenesis of the goldfish retina

The order of production of retinal cells was studied in embryos of the goldfish, Carassius auratus using ³H-thymidine autoradiography. Cell division ceases first in the neuroepithelium at the fundus of the eye between embryonic stages 19 and 20 and gradually becomes restricted to the retinal margin by stage 24. In the fundus the cell whose nuclei will reside in the inner layer of the retina stop dividing earlier than the cells whose nuclei will reside in the outer retinal layers. Thus the ganglion cells in the fundus of the retina are produced first and the receptors and horizontal cells last.     

Biphasic retinal neurogenesis in the brush-tailed possum, Trichosurus vulpecula: further evidence for the mechanisms involved in formation of ganglion cell density gradients.

We investigated cell generation in the retina of the brush-tailed possum (Trichosurus vulpecula) by using tritiated (3H)-thymidine labelling of newly generated cells. Animals aged between postnatal day (P) 5 and 85 each received a single injection of 3H-thymidine. Following autoradiographic processing, maps of labelled cells were constructed from retinal sections. Retinal cell generation takes place in two phases, the first is concluding in the retinal periphery at P53 as the second is seen to commence in midtemporal retina. In the first phase, cells in central retina are generated earlier than those in peripheral regions. In the second phase, cells complete their final division in midtemporal retina first and in the periphery last. Cells generated in the first phase comprise virtually all cells in the ganglion cell layer, amacrine cells, horizontal cells, and cones. Ganglion cells are produced at a slightly earlier stage than displaced amacrine cells, horizontal cells, or cones. Amacrine cells in the inner nuclear layer are the final cells produced in the first phase. When ganglion cells and amacrine cells are pooled, their combined rate of production matches that of the other cell types. These data indicate that the ratio of displaced amacrine cells: horizontal cells: cones: combined ganglion cells and amacrine cells does not change throughout development. However, the ratio of ganglion cells:macrines changes steadily as development proceeds to favour amacrine cells. In the second phase, sparse numbers of nonganglion cells in the ganglion cell layer and large numbers of bipolar and Müller cells are produced along with all rods. The two phases in the possum are similar to those seen in the wallaby, the quokka. However, fewer cells are added in central retina in the possum than in the quokka and cell addition continues for a more extended period in the periphery in the possum. We suggest that this difference in cell addition could account for the development of a more pronounced visual streak of retinal ganglion cells in the possum than in the quokka. A comparison of possum retinal cell generation with that of other marsupials adds support for the "homochrony theory."     

Identity of cells produced by two stages of cytogenesis in the postnatal cat retina.

Cytogenesis in the postnatal cat retina was studied with the aid of 3H-thymidine autoradiography to identify the cell classes generated. Cells proliferate in two stages, which are separate spatially and temporally. Previous studies have shown that during Stage 1, cytogenesis occurs at high density at the ventricular surface of the retina, whereas Stage 2 occurs at low density in the inner retinal layers. At the ages studied, the progeny of Stage 1 cytogenesis are distributed in an annulus toward the margin of the retina, and those of Stage 2 occur central to the annulus, indicating that Stage 2 follows Stage 1. Cell genesis in Stage 1 appears to cease by P16; genesis in Stage 2 persists until between P21 and P30. The same cell classes (amacrine cells, bipolar cells, Müller cells, and rod photoreceptors) are generated during both Stages 1 and 2, but there are significant changes in their proportions both within and between stages. The proportion of the Stage 1 mitoses that form bipolar cells increases from 31% at P0 to 62% at P14. A corresponding decrease is observed in the proportion of rods (from 60% at P0 to 32% at P14). The proportion of cells generated during Stage 2 that become rods increases from 39% at P0 to 70% at P21, whereas the proportion of bipolar cells decreases from 50% at P0 to 23% at P21. Müller cells form a relatively constant proportion (8 to 15%) of the cells generated during both Stage 1 and 2. Thus at the end of Stage 1, mostly bipolar cells are generated; at the end of Stage 2, mostly rods are generated.     

Specification of retinal central connections in Rana pipiens before the appearance of the first post-mitotic ganglion cells

The order of production of retinal cells and the time when retinal cells become post-mitotic was studied in Rana pipiens embryos using ³H-thymidine autoradiography. Cell division stops first in the fundus of the retinal rudiment between embryonic stages 17 and 18 and gradually becomes restricted to the retinal margin. The ganglion cells in the fundus are among the first cells to become post-mitotic. The specification of the central connections of ganglion cells was studied by rotating the eye primordium at embryonic stages 16–21. After metamorphosis, the visual projection from the rotated eye to the contralateral optic tectum was mapped electrophysiologically and compared with the normal retinotectal map. In all cases, the visual projection map was rotated through the same angle as was indicated by the position of the choroidal fissure. It appears that ganglion cell connections with the tectum were specified by stage 17. These results indicate that ganglion cell central connections are specified before the first ganglion cells become post-mitotic.     

Generation and death of cells in the dorsal lateral geniculate nucleus and superior colliculus of the wallaby, Setonix brachyurus (quokka).

To study postnatal cell generation in primary visual centres of the quokka, tritiated thymidine was injected into pouch-young aged postnatal day (P)1-P85. Brains were examined at P100, just before eye-opening, when primary visual projections are essentially mature. Neurons in the dorsal lateral geniculate nucleus (dLGN) and superior colliculus (SC) were generated at P1-P10 and P1-P18 respectively. Peak numbers of labelled cells were seen at P3 and P5 in the dLGN and SC. Cell death was assessed in the dLGN and SC of young aged P10-P150. Low numbers of dying cells were seen in the dLGN throughout this period, with a small peak at P85. A more substantial peak of cell death was seen in the SC, also at P85. In the quokka, the time interval between the peaks of cell generation and of cell death in the dLGN and SC is 70-80 days, considerably longer than the interval of 40 days between birth and death of retinal cells.     

Timing and topography of cell genesis in the rat retina

To understand the mechanisms of cell fate determination in the vertebrate retina, the time course of the generation of the major cell types needs to be established. This will help define and interpret patterns of gene expression, waves of differentiation, timing and extent of competence, and many of the other developmental processes involved in fate acquisition. A thorough retinal cell "birthdating" study has not been performed for the laboratory rat, even though it is the species of choice for many contemporary developmental studies of the vertebrate retina. We investigated the timing and spatial pattern of cell genesis using 3H-thymidine (3H-TdR). A single injection of 3H-TdR was administered to pregnant rats or rat pups between embryonic day (E) 8 and postnatal day (P) 13. The offspring of prenatally injected rats were delivered and all animals survived to maturity. Labeled cells were visualized by autoradiography of retinal sections. Rat retinal cell genesis commenced around E10, 50% of cells were born by approximately P1, and retinogenesis was complete near P12. The first postmitotic cells were found in the retinal ganglion cell layer and were 9-15 microm in diameter. This range includes small to medium diameter retinal ganglion cells and large displaced amacrine cells. The sequence of cell genesis was established by determining the age at which 5, 50, and 95% of the total population of cells of each phenotype became postmitotic. With few exceptions, the cell types reached these developmental landmarks in the following order: retinal ganglion cells, horizontal cells, cones, amacrine cells, rods, bipolar cells, and Müller glia. For each type, the first cells generated were located in the central retina and the last cells in the peripheral retina. Within the sequence of cell genesis, two or three phases could be detected based on differences in timing, kinetics, and topographic gradients of cell production. Our results show that retinal cells in the rat are generated in a sequence similar to that of the primate retina, in which retinogenesis spans more than 100 days. To the extent that sequences reflect underlying mechanisms of cell fate determination, they appear to be conserved.     

Autoradiographic study of histogenesis in the mouse olfactory bulb. I. Time of origin of neurons and neuroglia

Time of origin (final cell division) of neurons and neuroglia of the mouse olfactory and accessory olfactory formations was determined by autoradiography. Animals were injected with thymidine-H³ at various developmental stages and killed at or near maturity. In the olfactory formation mitral cells (the largest neurons) arise first, mainly over the three day period from the eleventh day of gestation (E11) to E13, tufted cells chiefly from E13 to E18, and granule cells (the smallest neurons) mainly from E18 to postnatal day 20. Most of the smaller and more superficial peripheral tufted cells arise later than the deeper and larger middle and internal tufted cells. All three types of granule cells have a time of origin extending well into postnatal life, with internal granule cells arising over a longer and later period than periglomerular cells or granule cells of the mitral cell layer. Neuroglial precursors undergo final cell division chiefly between E17 and P10. In the phylogenetically less evolved accessory olfactory formation, mitral cells originate earlier than their homologues in the olfactory formation; mitral cells principally from E10 to E12 and granule cells chiefly from E12 to E18. The results support the concept that some germinal layers of the central nervous system are programmed to produce a succession of cell types, larger cells before smaller ones.     

In vitro transdifferentiation of embryonic rat retinal pigment epithelium to neural retina

Divergence of neural retinal and retinal pigment epithelial (RPE) lineages from the optic vesicle neuroepithelium starts at a very early stage of eye development. Partially or even fully differentiated RPEs of some vertebrate species are capable of transforming into neural retina. In the present study, we have shown that mammalian RPE possesses the ability to transdifferentiate into neural retina at early embryonic stages. If cultured in serum-free medium, presumptive rat RPE became pigmented and expressed a molecular marker of mature RPE. In the presence of basic fibroblast growth factor (bFGF), cultured early embryonic rat RPE did not acquire pigment and grew to form retina-like multilayer structure containing neuronal cells and cells that express markers of retinal ganglion, amacrine and rod photoreceptor cells. The effects of bFGF occurred independently of effects on cell division and became irreversible after periods that varied with tissue age. This study has demonstrated that already differentiated embryonic rat RPE still retain the ability to become neural retina up to certain stage.     

Growth of the adult goldfish eye. III. Source of the new retinal cells

The manner in which new cells are added to the growing adult goldfish retina was examined using ³H-thymidine radioutography. Cell proliferation leading to the formation of neurons is restricted to the retinal margin at the ora terminalis. New retina is added in concentric rings, with slightly more growth dorsonasally. The rate of cell addition is variable, averaging 12,000 cells/ day. These new cells account for about 20% of the total increase in retinal area; the remaining 80% is due to hypertrophy, or expansin, of the retina. In contrast to all of the other retinal cells, the rods do not appear to participate in the retinal expansion. This hypothesized immobility of the rods would create a shearing strain between the retinal layers resulting in a shift in their position relative to the other cells. Were they to maintain synaptic contacts with the same horizontal and bipolar cells, the rod axons would have to be elongated peripherally or the post-synaptic cell dendrites displaced centrally. Since neurons with this morphology have not been found in the goldfish retina, these observations suggest that the rods must be changing their synaptic connections as the retina grows.     

Morphology and birth dates of horizontal cells in the retina of a marsupial

Most eutherian (placental) mammals have two horizontal cell types; however, one type only has been seen in rodents. In order to assess whether one type of horizontal cell or two is a basic mammalian feature, we have examined the morphology of horizontal cells in a marsupial, the quokka wallaby, by Golgi staining or horseradish peroxidase labelling. The birth dates of horizontal cells have also been determined by ³H-thymidine/autoradiography. There are two types of horizontal cell in the wallaby retina. One type has no axon and corresponds to the axonless cell in eutherian species; the other has shorter dendrites, an axon, and an axonal arbor, corresponding to the eutherian short-axon cell. As in eutherian mammals, the dendrites of each horizontal cell type lie in the outer plexiform layer (OPL) and contact cones and the axonal arbor of the short-axon cell contacts rods. The dendrites of the axonless cells are long, with an average length of 250 μm, and each cell has one, sometimes two, short, stubby processes, which branch off a dendrite, traverse the inner nuclear layer, and reach the inner plexiform layer. The dendritic field of these cells is elongated, and dendrites show a preferential orientation at right angles to the trajectory of overlying ganglion cell axons. Short-axon cells have a morphology similar to that seen in other species, although the axonal arbor is relatively small. Both types of horizontal cell are generated in the first phase of retinal cell generation. © Wiley-Liss, Inc.     

Genesis of resting microglia in the gray matter of mouse hippocampus

The genesis of resting microglia in the gray matter of mouse hippocampus was studied by ³H-thymidine autoradiography in combination with electron microscopy. Newborn mice were injected with ³H-thymidine singly or repeatedly at different postnatal stages, and killed shortly after the injection or after various intervals. Tissue specimens of the hippocampus at CA1 and CA2 were processed for light and electron microscopic autoradiography. The results showed that at least 91% of glial cells in the stratum radiatum of the hippocampus are produced after birth. About three-fourths of astroglia in this area are produced before the sixth postnatal day, and a larger part of resting microglia are formed after the ninth postnatal day. Morphological transition can be traced from either proliferating cells in the stratum radiatum at late postnatal days to resting microglia or from those in early postnatal days to astroglia. A continuous morphological transition was observed between the proliferating cells at the late postnatal days (microglial production period) and those at the early postnatal days (astroglial production period). The latter retain some fine structural characteristics similar to small glioblasts in the subependymal layer. These findings strongly suggest that resting microglia, as well as astroglia, are derived from glioblasts, and are of neurectodermal origin.     

Growth of the adult goldfish eye. II. Increase in retinal cell number

The retinas of adult goldfish, one to four years of age, 4–23 cm in length, were examined with standard paraffin histology to determine if new cells were being added with growth. Retinal cell nuclei were counted and the area of the retina was measured. An analysis of cell densities in various regions throughout the retina showed that the cells are distributed nearly homogeneously. The density (No./mm² of retinal surface) of ganglion cells, inner nuclear layer cells and cones decreases with growth, but the density of rods remains constant. Thus the rods account for a larger proportion of the cells in larger retinas. The total number of cells per retina increases: the ganglion cells from 60,000 to 350,000; the inner nuclear layer cells from 1,500,000 to 4,000,000; the cones from 250,000 to 1,400,000; the rods from 1,500,000 to 15,000,000. This increase in the number of retinal neurons implies the formation of even more new synapses, and suggests the adult goldfish retina as a model for both neuro- and synaptogenesis.     

Postnatal development of 3H-GABA-accumulating cells in rabbit retina

Light and electron microscopic autoradiography demonstrates that 3H-GABA is accumulated by horizontal cells in neonatal rabbit retina but not in the adult. A specific population of horizontal cells appears to be mature at birth and they avidly accumulate 3H-GABA during a 15-minute incubation period in vitro. Uptake into horizontal cells is not observed after the fifth postnatal day; 3H-GABA-accumulating horizontal cell bodies and their processes are the first identifiable components that clearly mark the future location of the outer plexiform layer at birth and as such, may be considered pioneering elements. Our observations raise the interesting possibility that the pioneering horizontal cell may provide structural and/or chemical factors necessary for the subsequent development of the outer plexiform layer of the retina. Labeling patterns of other retinal cells also show varying degrees of change during development. A population of amacrine cells accumulate 3H-GABA at birth. These cells show little change in their morphological or 3H-GABA uptake properties from birth to adulthood. Müller cells show weak accumulation of 3H-GABA at birth. Subsequent to this time, labeling of Müller cells is significantly more robust, resulting in Müller cell domination of retinal autoradiographic patterns in more mature retinas. Every cell body in the ganglion cell layer accumulates 3H-GABA at birth. The number of labeled cells declines during postnatal development, resulting in a very limited adult population. We conclude that the ability of retinal cells to accumulate 3H-GABA does not remain constant during postnatal development; rather each cell population displays a unique maturation sequence that results in a dramatic developmental shift in the number and types of GABA-accumulating cells present in the retina.     

Differentiation-dependent sensitivity to cell death induced in the developing retina by inhibitors of the ubiquitin-proteasome proteolytic pathway

The effects of inhibitors of proteasome function were studied in the retina of developing rats. Explants from the retina of neonatal rats at postnatal day (P) 3 or P6 were incubated with various combinations of the proteasome inhibitor carbobenzoxyl-leucinyl-leucinyl-leucinal (MG132), the protein synthesis inhibitor anisomycin, or the adenylyl cyclase activator forskolin. MG132 induced cell death in a subset of cells within the neuroblastic (proliferative) layer of the retinal tissue. The cells sensitive to degeneration induced by either MG132 or anisomycin, were birthdated by bromodeoxyuridine injections. This showed that the MG132-sensitive population includes both proliferating cells most likely in their last round of cell division, and postmitotic undifferentiated cells, at a slightly earlier stage than the population, sensitive to anisomycin-induced cell death. The results show that sensitivity to cell death induced by proteasome inhibitors defines a window of development in the transition from the cell cycle to the differentiated state in retinal cells.     

H3-thymidine autoradiographic studies on the cell proliferation and differentiation in the external and the internal granular layers of the mouse cerebellum

Cell proliferation and migration in the external granular layer of the mouse cerebellum were studied with autoradiography after cumulative labeling with H³-thymidine. The germinative cells in the external granular layer were considered as externally dislocated matrix cells. Their generation time, presynthetic time, duration of DNA synthesis, postsynthetic time and mitotic time were determined in one-, three-, seven- and ten-day-old mice. The entire sequence of the ontogeny of the external granular cell-system was separated into three consecutive stages; stage 1 or stage of pure external matrix cell proliferation, stage 2 or stage of neuroblast production, and stage 3 or stage of neuroglia differentiation. Production of neuroblasts in the external granular layer at seven and ten days of life and their migration into the internal granular layer were demonstrated by means of autoradiography. Transit times of the neuroblasts migrating across the external mantle layer and the molecular layer of ten-day-old mice were estimated at 21 and four hours, respectively. More than 50% of the inner granular cells migrated from the external granular layer later than ten days of life and almost 81 to 92% were produced later than seven days of postnatal life. In conclusion, on the basis of the matrix cell concept, the authors tried to unify observations of previous and present investigators and presented a scheme of pre- and postnatal histogenesis of the mouse cerebellum.     

Pedf (Pigment epithelium-derived Factor) promotes increase and maturation of pigment granules in pigment epithelial cells in neonatal albino rat retinal cultures

Pigment Epithelium-Derived Factor (PEDF), purified from human retinal pigment epithelial (RPE) cell culture medium, is a neurotrophic factor which potentiates the differentiation of human Y-79 retinoblastoma cells and increases the survival of cerebellar granule cells. To investigate the effects of PEDF on non-transformed retinal cells, we used primary cultures of neonatal albino rat retinas, where the three principal cell types of the retinal layers (neuronal, glial and epithelial) were all present and focussed our attention on RPE cells, which are of special relevance for retinal pathophysiology. PEDF had a dramatic effect on these cells. They showed a modified phenotype, with larger dimensions, higher cytoplasmic spreading, presence of phagocytic vacuoles, development of wide intercellular contacts, and increase and maturation of pigment granules. These results suggest that PEDF may have a role in regulating RPE cell differentiation.     

Freeze-fracture study of filipin binding in photoreceptor outer segments and pigment epithelium of dystrophic and normal retinas

We have studied sterol distribution in the retinal pigment epithelial (RPE) microvillous and outer segment disc membranes of rats with inherited retinal degeneration (RCS; RCS-p/+) and of normal genetic controls (RCS-rdy⁺, RCS-rdy⁺-p/+) by using the polyene antibiotic filipin, which binds specifically to 3-B-hydroxy-sterols, and freeze-fracture techniques. Retinas were perfusion-fixed, incubated with filipin in the same fixative, and prepared routinely for freeze-fracture electron microscopy. In the normal retina, the distribution of filipin binding sites on both RPE microvillous and outer segment disc membranes changes during development. Prior to outer segment elongation and the onset of phagocytosis (10 days postnatal), filipin sterol complexes are homogeneously distributed in both microvillous and outer segment membranes. With the onset of phagocytosis (2 weeks postnatal and later) filipin binding in both tissues forms a proximal-to-distal gradient, and binding sites decrease as distance from the cell body increases. In the normal RPE microvillous membranes, binding sites are numerous proximally and sparse on the distal tips. In the normal outer segment disc membranes, binding sites are often present on the basal discs, but are sparse on the intact apical discs prior to shedding. As the discs are cast off and engulfed by the RPE, however, filipin binding increases on both disc and phagosome membranes. In the dystrophic retina, the distribution of filipin binding sites differs from the normal. First, in the microvillous membranes, the proximal-to-distal gradient in filipin binding is rarely present at 2 weeks postnatal and becomes prominent only after the buildup of membranous debris has begun (3–6 weeks postnatal). Second, as the photoreceptors degenerate and the membrane debris disappears (4 months postnatal), filipin binding on the microvillous membranes becomes relatively sparse and homogeneous. Third, filipin binding on the intact disc membranes does not change with outer segment elongation, and numerous filipin binding sites are present on both apical and basal outer segment disc membranes. Fourth, large aggregates of filipin binding sites occupy the vast expanses of particlefree areas of debris membranes which accumulate between the photoreceptors and the RPE. These changes in the amount and distribution of filipin binding sites in the dystrophic retina add to the evidence that the disease process involves outer segment as well as RPE membranes and suggest that alterations in cholesterol distribution could contribute to the phagocytic defect.