Reactive astroglia-neuron relationships in the human cerebellar cortex: a quantitative, morphological and immunocytochemical study in Creutzfeldt-Jakob disease

In order to investigate the role of neuron-glia interactions in the response of astroglial to a non-invasive cerebellar cortex injury, we have used two cases of the ataxic form of Creutzfeldt-Jakob disease (CJD) with distinct neuronal loss and diffuse astrogliosis. The quantitative study showed no changes in cell density of either Purkinje or Bergmann glial cells in CJ-1, whereas in the more affected CJ-2 a loss of Purkinje cells and an increase of Bergmann glial cells was found. The granular layer in both CJD cases showed a similar loss of granule cells (about 60%) in parallel with the significant increase in GFAP+ reactive astrocytes. GFAP immunostaining revealed greater reactivity of Bergmann glia in CJ-2 than in CJ-1, as indicated by the thicker glial processes and the higher optical density. Granular layer reactive astrocytes were regularly spaced. In both CJD cases there was strict preservation of the spatial arrangement of all astroglial subtypes--Fa?anas cells, Bergmann glia and granular layer astrocytes. Reactive Fa?anas and Bergmann glial cells and microglia/macrophages expressed vimentin, while only a few vimentin+ reactive astrocytes were detected in the granular layer. Karyometric analysis showed that the increase in nuclear volume in reactive astroglia was directly related with the level of glial hypertrophy. The number of nucleoli per nuclear section was constant in astroglial cells of human controls and CJD, suggesting an absence of polyploidy in reactive astroglia. Ultrastructural analysis revealed junctional complexes formed by the association of macula adherens and gap junctions. In the molecular layer numerous vacant dendritic spines were ensheathed by lamellar processes of reactive Bergmann glia. Our results suggest that quantitative (neuron/astroglia ratio) and qualitative changes in the interaction of neurons with their region-specific astroglial partners play a central role in the astroglial response pattern to the pathogenic agent of CJD.     

Localization of neuronal and glial glutamate transporters

The cellular and subcellular distributions of the glutamate transporter subtypes EAAC1, GLT-1, and GLAST in the rat CNS were demonstrated using anti-peptide antibodies that recognize the C-terminal domains of each transporter. On immunoblots, the antibodies specifically recognize proteins of 65-73 kDa in total brain homogenates. Immunocytochemistry shows that glutamate transporter subtypes are distributed differentially within neurons and astroglia. EAAC1 is specific for certain neurons, such as large pyramidal cortical neurons and Purkinje cells, but does not appear to be selective for glutamatergic neurons. GLT-1 is localized only to astroglia. GLAST is found in both neurons and astroglia. The regional localizations are unique to each transporter subtype. EAAC1 is highly enriched in the cortex, hippocampus, and caudate-putamen and is confined to pre- and postsynaptic elements. GLT-1 is distributed in astrocytes throughout the brain and spinal cord. GLAST is most abundant in Bergmann glia in the cerebellar molecular layer brain, but is also present in the cortex, hippocampus, and deep cerebellar nuclei.     

The role of ascorbic acid in oral cancer and carcinogenesis

Severe transient focal cerebral ischemia causes brain infarction with a strong glial reaction. We have studied whether postischemic reactive glial cells express epidermal growth factor receptor (EGFR) following middle cerebral artery occlusion in the rat. We have also looked for signs of proliferating activity, as EGFR is known to be involved in cell growth and proliferation in certain non-neural cells. EGFR was studied using three different antibodies which were found to stain for a tyrosine-phosphorylated protein (p170) corresponding to the membrane-anchored EGFR. Neurons of the control brain were strongly immunoreactive to EGFR, but a decrease of EGFR-immunoreactivity was seen in the ipsilateral brain side from 24 h postischemia due to neuronal loss. However, the presence of abundant glial cells strongly immunoreactive to EGFR became apparent in this area from 4 days postischemia onward. The use of microglial (lectin or OX-42) and astroglial (GFAP) markers showed that these postischemic EGFR-stained cells were reactive microglia/macrophages and astroglia. The subcellular localization of EGFR in reactive microglia/macrophages was compatible with the network of the Golgi apparatus, as revealed with an antibody against a peripheral membrane-bound protein of the Golgi. The presence of abundant proliferating cells in the ischemic brain was detected from 4 days postischemia with an antibody against proliferating cell nuclear antigen. Proliferating reactive microglia/macrophages were abundant within the infarcted brain side, whereas proliferating astrocytes were found mainly in the immediate periphery of the infarct limiting the necrotic area from the undamaged tissue. These proliferating cells were immunoreactive to EGFR. The results show the presence of EGFR in postischemic reactive glial cells and suggest that EGFR-dependent pathways mediate signal transduction in reactive glia following transient focal cerebral ischemia.     

Neurono-glial changes in the cerebral cortex of animals exposed to white noise

The rat acoustic cortex has been studied at light optic and ultrastructural levels under the white noise stimulation. After the noise stimulation for 7 days, micropunctate hemorrhages, proliferation and hypertrophy in cells of microglia and astrocytic glia are noted. After the noise stimulation for 21 days, the neuroglial reaction becomes less pronounced, there are no hemorrhages, a great amount of neurons with peripheral and total chromatolysis appear. In other neurons, as well as in all types of neuroglia the number of primary lysosomes increases, their structure changes. In lysosomes lipofuscin and lipid drops are accumulating, many of lysosomes turning into multivesicular bodies. The destructive changes observed in the neurons and neuroglia underlie prolonged disturbances in the higher neural activity after the noise stimulation is stopped.     

Differential expression of superoxide dismutase isoforms in neuronal and glial compartments in the course of excitotoxically mediated neurodegeneration: relation to oxidative and nitrergic stress

To examine the cellular distribution of radical scavenging enzymes in glia, in comparison to that in neurons and their behaviour during excitotoxically induced neurodegenerative processes, protein levels and the cellular localization of cytosolic and mitochondrial superoxide dismutase (Cu/Zn- and Mn-SOD) were investigated in the rat brain undergoing quinolinic acid (Quin)-induced neurodegeneration. Evidence for the specificity of the applied antibodies to detect immunocytochemically these SOD isoforms was obtained from electron microscopy and Western blotting. In control striatum Mn-SOD was clearly confined to neurons, whereas Cu/Zn-SOD was found, rather delicately, only in astrocytes. Microglia failed to stain with antibodies to both SOD isoforms. Quin application resulted in an initial formation of oxygen and nitrogen radicals as determined by the decline in the ratio of ascorbic to dehydroascorbic acid and by increased levels of nitrated proteins, an indicator for elevated peroxynitrite formation. Morphologically, massive neuronal damage was seen in parallel. Astroglia remained intact but showed initially decreased glutamine synthetase activities. The levels of Mn-SOD protein increased 2-fold 24 h after Quin injection (Western blotting) and declined only slowly over the time period considered (10 days). Cu/Zn-SOD levels increased only 1.3-fold. Immunocytochemical studies revealed that the increase in Mn-SOD is confined to neurons, whereas that of Cu/Zn-SOD was observed only in astroglial cells. Quiescent microglial cells were, as a rule, free of immunocytochemically detectable SOD, whereas in activated microglia a few Mn-SOD immunolabeled mitochondria occurred. Our results suggest a differential protective response in the Quin lesioned striatum in that Mn-SOD is upregulated in neurons and Cu/Zn-SOD in astroglia. Both SOD-isoforms are assumed to be induced to prevent oxidative and nitric oxide/peroxynitrite-mediated damage. In the border zone of the lesion core this strategy may contribute to resist the noxious stimulus.     

Glial fibrillary acidic protein expression but no glial demarcation follows the lesion in the molecular layer of cerebellum

The present study investigates the reactive gliosis following a simple stab wound lesion to a brain area in which a characteristic astroglial architecture exists, i.e., the Bergmann-glia in the molecular layer of cerebellum. While in mammalian brain the Bergmann-glia contains glial fibrillary acidic protein (GFAP), in the avian Bergmann-glia, the cytoskeletal protein is vimentin, which is characteristic for immature astroglia in mammals. The operations were performed on chickens and rats under deep anaesthesia, using a sterile disposable needle. After a 1-week survival period, the animals were overdosed with ether and perfused transcardially with 4% buffered paraformaldehyde. Free-floating sections cut with a vibration microtome were processed for immunohistochemistry against GFAP and vimentin. GFAP immunopositivity of Bergmann-glia appeared in chicken and increased in rat in the lesioned area but the lesion was not surrounded by typical astrocytes and no demarcation was formed in the molecular layer, in contrast to the usual appearance of reactive gliosis, which was observed in the granular layer and in the white matter in both species. Vimentin immunopositivity of the Bergmann-glia also increased around the lesion in both species. The results suggest that a highly developed glial architecture fails to re-arrange into a demarcating scar, which offers an interesting model system to study the importance of glial demarcation. The observations also support that the resident glia is the main component of the glial reaction, and prove the capability of avian Bergmann-glia to express GFAP.     

Evidence for sexual differentiation of glia in rat brain

It is well established that gonadal steroids mediate sexual differentiation of the brain via direct effects on neurons during a restricted critical period. In addition, estrogen can influence glial morphology in the adult brain, and in vitro studies suggest estrogen induces glial differentiation. However, there is a lack of in vivo evidence for steroid effects on glia during the critical period. We report here a hormone-mediated sexual differentiation of arcuate glia as early as Postnatal Day 1. Using glial fibrillary acidic protein immunoreactivity (GFAP-ir), we compared the responsiveness of astroglia in the rat arcuate nucleus among five hormonally different groups. The results indicate increased GFAP-ir cell surface area 24 hr after hormonal manipulation in castrate males compared to intact males, intact females (ANOVA; P < 0.01), and females injected with testosterone propionate (50 microg; ANOVA; P < 0.05). However, astroglia in intact males extended their processes significantly greater distances from the cell body compared to all other treatment groups (ANOVA; P < 0.01). The GFAP-ir cells were categorized into four distinct classes ranging from a simple bipolar to a fully stellate morphology. The frequency distribution of classes varied between groups with more stellate cells found in intact males. Finally, these sex differences in arcuate glia persisted into adulthood. We hypothesize that during the critical period, testosterone, or its metabolite estrogen, induce sexual differentiation of glia. We further hypothesize that in females glial cells remain partially undifferentiated and this may be important to glial plasticity seen in adult female arcuate.     

Application of infrared spectroscopy to development of stable lyophilized protein formulations

Global cerebral ischemia selectively damages neurons, but its contribution to glial cell death is uncertain. Accordingly, adult male rats were sacrificed by perfusion fixation at 1, 2, 3, 5, and 14 days following 10 minutes of global ischemia. This insult produces CA1 hippocampal neuronal death at post-ischemic (PI) day 3, but minor or no damage to neurons in other regions. In situ end labeling (ISEL) and immunohistochemistry identified fragmented DNA of dead or dying glia and distinguished glial subtypes. Rare ISEL-positive oligodendroglia, astrocytes, and microglia were present in control brain. Apoptotic bodies and ISEL-positive glia significantly increased at PI day 1 in cortex and thalamus (p < 0.05), but were similar to controls in other regions and at other PI intervals. Most were oligodendroglia, although ISEL-positive microglia and astrocytes were also observed. These results show that oligodendroglia die rapidly after brief global ischemia and are more sensitive than neurons in certain brain regions. Their selective vulnerability to ischemia may be responsible for the delayed white matter damage following anoxia or CO poisoning or that associated with white matter arteriopathies. Glial apoptosis could contribute to the DNA ladders of apoptotic oligonucleosomes that have been found in post-ischemic brain.     

Astrocytes and extracellular matrix following intracerebral transplantation of embryonic ventral mesencephalon or lateral ganglionic eminence

Transplantation of embryonic neurons to the adult mammalian central nervous system (CNS) offers the possibility of re-establishing neural functions lost after traumatic injuries or neurodegenerative disease. In the adult CNS, however, transplanted neurons and their growing neurites can become confined to the graft region, and there may also be a relative paucity of afferents innervating grafted neurons. Because glia may influence the development and regeneration of CNS neurons, the present study has characterized the distribution of astrocytes and developmentally regulated glycoconjugates (chondroitin-6-sulfate proteoglycan and tenascin) within regions of the embryonic mouse CNS used as donor tissues, and in and around these grafts to the adult striatum and substantia nigra. Both chondroitin-6-sulfate proteoglycan and tenascin are present in the embryonic ventral mesencephalon (in association with radial glia and their endfeet, and glial boundaries that cordon off the ventral mesencephalon dopamine neuron migratory zone) and lateral ganglionic eminence before transplantation, and they are conserved within grafts of these tissues to the adult mouse. Neostriatal grafts exhibit a heterogeneous pattern of astrocyte and extracellular matrix molecule distribution, unlike ventral mesencephalon grafts, which are rather homogeneous. There is evidence to suggest that, in addition to variation in astroglial/extracellular matrix immunostaining within different compartments in striatal grafts to either adult striatum or substantia nigra, there are also boundaries between these compartments that are rich in glial fibrillary acidic protein/extracellular matrix components. Substantia nigra grafts, with cells immunoreactive for tyrosine hydroxylase, are also rich in immature astroglia (RC-2-immunopositive), and as the astroglia mature (to glial fibrillary acidic protein-positive) over time the expression of chondroitin-6-sulfate proteoglycan and tenascin is also reduced. These same extracellular matrix constituents, however, are only slightly up-regulated in an area of the adult host which surrounds the grafted tissue. Glial scar components exhibit no obvious differences between grafts from different sources to homotopic (e.g., striatum to striatum) or heterotopic (e.g., substantia nigra to striatum) sites, and likewise grafts of non-synaptically associated structures (e.g., cerebellum to striatum), needle lesions or vehicle injections all yield astroglial/extracellular matrix scars in the host that are indistinguishable. Studies utilizing the ROSA-26 transgenic (beta-galactosidase-positive) mouse as a host for non-5-bromo-4-chloro-3-indolyl-beta-d-galactopyranoside-labeled grafts indicate that the early astroglial/extracellular matrix response to the graft is derived from the surrounding host structures. Furthermore, biochemical analysis of one of the "boundary molecules", tenascin, from the developing ventral mesencephalon versus adult striatal lesions, suggests that different forms of the molecule predominate in the embryonic versus lesioned adult brain. Such differences in the nature and distribution of astroglia and developmentally regulated extracellular matrix molecules between donor and host regions may affect the growth and differentiation of transplanted neurons. The present study suggests that transplanted neurons and their processes may flourish within graft versus host regions, in part due to a confining glial scar, but also because the extracellular milieu within the graft site remains more representative of the developmental environment from which the donor neurons were obtained [Gates M. A., et al. (1994) Soc. Neurosci. Abstr. 20, 471].     

Neurturin and GDNF promote proliferation and survival of enteric neuron and glial progenitors in vitro

Signaling through the c-Ret tyrosine kinase and the endothelin B receptor pathways is known to be critical for development of the enteric nervous system. To clarify the role of these receptors in enteric nervous system development, the effect of ligands for these receptors was examined on rat enteric neuron precursors in fully defined medium in primary culture. In this culture system, dividing Ret-positive cells differentiate, cluster into ganglia containing neurons and enteric glia, and create extensive networks reminiscent of the enteric plexus established in vivo. Glial cell-line-derived neurotrophic factor (GDNF) and neurturin both potently support survival and proliferation of enteric neuron precursors in this system. Addition of either neurturin or GDNF to these cultures increased the number of both neurons and enteric glia. Persephin, a third GDNF family member, shares many properties with neurturin and GDNF in the central nervous system and in kidney development. By contrast, persephin does not promote enteric neuron precursor proliferation or survival in these cultures. Endothelin-3 also does not increase the number of enteric neurons or glia in these cultures.     

Novel defensin subfamily from spinach (Spinacia oleracea)

Bcl-2 has a role in suppressing the production of reactive oxygen species and lipid peroxidation. To explore the in situ localization of 4-hydroxy-2-nonenal (HNE)-modified proteins and the Bcl-2 oncoprotein, we used double immunofluorescence labeling and confocal imaging in the rat brain after 3 h of middle cerebral artery (MCA) occlusion followed by reperfusion. Immunoreactivity for HNE or Bcl-2 was not detected at 1 h, but appeared in some intact neurons in the boundary between the infarcted and non-infarcted zones at 12 h. At 48 h, HNE-positive microglia were colocalized with Bcl-2 in the infarcted area and the boundary zone. Bcl-2 may play an important role in the antioxidant system promoting survival of the neurons and activated microglia following reperfusion injury.     

Inflammation and glial responses in ischemic brain lesions

Focal cerebral ischemia elicits a strong inflammatory response involving early recruitment of granulocytes and delayed infiltration of ischemic areas and the boundary zones by T cells and macrophages. Infiltration of hematogenous leukocytes is facilitated by an upregulation of the cellular adhesion molecules P-selectin, intercellular adhesion molecule-1 and vascular adhesion molecule-1 on endothelial cells. Blocking of the leukocyte/endothelial cell adhesion process significantly reduces stroke volume after transient, but not permanent middle cerebral artery occlusion. In the infarct region microglia are activated within hours and within days transform into phagocytes. Astrocytes upregulate intermediate filaments, synthesize neurotrophins and form glial scars. Local microglia and infiltrating macrophages demarcate infarcts and rapidly remove debris. Remote from the lesion no cellular infiltration occurs, but astroglia and microglia are transiently activated. Astrocytic activation is induced by spreading depression. In focal ischemia neurons die acutely by necrosis and in a delayed fashion by programmed cell death, apoptosis. Proinflammatory cytokines such as tumor necrosis factor-alpha and interleukin-1 beta are upregulated within hours in ischemic brain lesions. Either directly or via induction of neurotoxic mediators such as nitric oxide, cytokines may contribute to infarct progression in the post-ischemic period. On the other hand, inflammation is tightly linked with rapid removal of debris and repair processes. At present it is unclear whether detrimental effects of inflammation outweigh neuroprotective mechanisms or vice versa. In global ischemia inflammatory responses are limited, but micro- and astroglia are also strongly activated. Glial responses significantly differ between brain regions with selective neuronal death and neighbouring areas that are more resistent to ischemic damage.     

Tetracyclines inhibit microglial activation and are neuroprotective in global brain ischemia

Ischemic stroke is the most common life-threatening neurological disease and has limited therapeutic options. One component of ischemic neuronal death is inflammation. Here we show that doxycycline and minocycline, which are broad-spectrum antibiotics and have antiinflammatory effects independent of their antimicrobial activity, protect hippocampal neurons against global ischemia in gerbils. Minocycline increased the survival of CA1 pyramidal neurons from 10.5% to 77% when the treatment was started 12 h before ischemia and to 71% when the treatment was started 30 min after ischemia. The survival with corresponding pre- and posttreatment with doxycycline was 57% and 47%, respectively. Minocycline prevented completely the ischemia-induced activation of microglia and the appearance of NADPH-diaphorase reactive cells, but did not affect induction of glial acidic fibrillary protein, a marker of astrogliosis. Minocycline treatment for 4 days resulted in a 70% reduction in mRNA induction of interleukin-1beta-converting enzyme, a caspase that is induced in microglia after ischemia. Likewise, expression of inducible nitric oxide synthase mRNA was attenuated by 30% in minocycline-treated animals. Our results suggest that lipid-soluble tetracyclines, doxycycline and minocycline, inhibit inflammation and are neuroprotective against ischemic stroke, even when administered after the insult. Tetracycline derivatives may have a potential use also as antiischemic compounds in humans.     

Selective induction of c-Jun and NGF in reactive astrocytes after cholinergic degenerations in rat basal forebrain

Cholinergic basal forebrain neurons are the major source of cortical cholinergic innervation. The number of these neurons is regulated by the availability of nerve growth factor (NGF) during development while in adulthood their cholinergic activity is modulated by NGF. In previous studies we have shown that cholinergic immunolesions of basal forebrain neurons increase local immediate early gene expression and NGF synthesis in the regions of degeneration. In this study we identify the cellular source of c-Jun and NGF expression using dual immunolabeling of c-Jun and NGF in combination with neuronal and glial markers. We demonstrate that both c-Jun and NGF are exclusively expressed in reactive astrocytes but not in microglia or in GABAergic basal forebrain neurons. These observations support the hypothesis that reactive astrocytes synthesize neurotrophic substances in vivo in response to neuronal degeneration in the basal forebrain.     

Exogenous glutamate enhances glutamate receptor subunit expression during selective neuronal injury in the ventral arcuate nucleus of postnatal mice

Much attention has been paid to proteinases derived from not only neurons but also microglia in relation to neuronal death. There is accumulating evidence that intra- and extracellular proteinases in these cells are part of the basic machinery of neuronal death pathways. Some members of the ced-3/interleukin-1 beta converting enzyme (ICE) (caspase) family of cysteine proteinases have been thought to play a major role in apoptosis of not only non-neuronal cells but also neurons. Calpain has also been demonstrated to be a mediator of the neurodegenerative response. Recent studies have shown that excitotoxic and ischemic neuronal injury could be attenuated by inhibitors of caspases and calpain. Several recent studies have suggested the involvement of endosomal/lysosomal proteinases, including cathepsins B, D and E, in neuronal death induced by excitotoxins and ischemia. Furthermore, it has been reported that the extracellular tissue-type plasminogen activator/plasmin proteolytic cascade is involved in excitotoxic injury of the hippocampal neurons. In addition to such neuronal proteinases, microglial proteinases are believed to be important for the modification of neuronal functions positively or negatively. Cathepsins E and S derived from microglia have been suggested to contribute to neuronal survival through degradation and removal of beta-amyloid, damaged neurons and cellular debris. On the other hand, 6-hydroxydopamine-induced microglial cell death was inhibited by inhibitors of aspartic proteinases and caspases, suggesting the involvement of cathepsins E and D and caspases in microglial cell death. Therefore, identification of which proteinases play a causative role in neuronal death execution and clarification of the regulators and substrates for such proteinases is very important for understanding the molecular basis of the neuronal death pathways and to develop novel neuroprotective agents.     

Injury-induced physiological events that may modulate gene expression in neurons and glia

Damage to the brain triggers a host of reactive responses in neurons and glia which are seen at sites of focal injury as well as at sites that are at a distance from the injury. Although many of these responses have been studied extensively, the signals that initiate the different responses have not been fully characterized, and it is still not understood how focal injury affects neurons and glia in distant sites. The present review summarizes recent findings that suggest that physiological events that occur at the time of the injury or during the early postlesion period can play an important and variable role in modulating neuronal and glial responses to injury. We focus on the events that occur in the hippocampal formation following unilateral lesions of the entorhinal cortex - a model system that has been used extensively for studies of cellular responses following focal brain injury. This lesion destroys the cells of origin of a massive excitatory projection to the dentate gyrus and hippocampus proper. Over time, the denervated neurons in the hippocampal formation are almost completely reinnervated as a result of local sprouting of systems that survive the lesion. Thus, this model system has been useful for studying cellular responses to both denervation and reinnervation. We summarize the information that this injury triggers physiological events that can strongly modulate gene expression in neurons and glia, including episodes of spreading depression that occur at the time of the injury, seizures that occur during the early postlesion period, the loss of afferent drive which leads to decreases in postsynaptic activity, and the restoration of activity that occurs in conjunction with reinnervation. We describe recent studies which suggest that some of these physiological events occur to a variable extent in different animals, especially the episodes of spreading depression and the recurrent seizures. Thus, the spatial pattern and temporal dynamics of altered gene expression following this "model" experimental injury may vary from animal to animal. The fact that physiological events strongly modulate the reactive changes in gene expression that occur following injury has important implications for understanding the sequelae of injury, and offers new opportunities for experimental and therapeutic interventions that may improve cellular repair, regeneration, and recovery of function.     

Cell-type- and developmental-stage-specific metabolism and storage of retinoids by embryonic chick retinal cells in culture

Biological functions of retinoids in the vertebrate retina include the role of 11-cis retinaldehyde as visual pigment chromophore, and possible effects of retinoic acid in histogenesis and cell survival. Qualitative and quantitative regulation of retinoid availability for these complex processes could involve several cell types, including retinal pigment epithelium, Müller glia and retinal photoreceptors and non-photoreceptor neurons; their relative contributions, however, have not been fully elucidated. Using purified cultures, we have carried out a study of cell-type-specific metabolism and storage of retinoids in chick embryo retinal photoreceptors and other neuronal cells, as compared to those of retinal glia. Retinal glia were found to synthesize both retinoic acid and retinyl esters, and to hydrolyse the latter; they also displayed retinol dehydrogenase activities. Cultured neurons and photoreceptors also synthesized and hydrolysed retinyl esters; their capacity for retinaldehyde synthesis from a retinol or retinyl ester substrate suggested the presence of retinol dehydrogenase activity. Retinoic acid was not synthesized in differentiated neuronal cultures, although some synthesis was detectable at early culture stages when the cells were still morphologically undifferentiated. These findings indicate that cell-type-specific metabolic activities are expressed during retinal cell differentiation in vitro, and that embryonic retinal photoreceptors and nonphotoreceptor neurons are active participants in the metabolism and storage of retinoids.