Neurogenesis in the retinal ganglion cell layer of the rat.

The present study has examined the birthdates of neurons in the retinal ganglion cell layer of the adult rat. Rat fetuses were exposed to tritiated thymidine in utero to label neurons departing the mitotic cycle at different gestational stages from embryonic days 12 through to 22. Upon reaching adulthood, rats were either given unilateral injections of horseradish peroxidase into target visual nuclei in order to discriminate (1) ganglion cells from displaced amacrine cells, (2) decussating from non-decussating ganglion cells, and (3) alpha cells from other ganglion cell types; or, their retinae were immunohistochemically processed to reveal the choline acetyltransferase-immunoreactive amacrine cells in the ganglion cell layer. Retinae were embedded flat in resin and cut en face to enable reconstruction of the distribution of labelled cells. Retinal sections were autoradiographically processed and then examined for neurons that were both tritium-positive and either horseradish peroxidase-positive or choline acetyltransferase-positive. Tritium-positive neurons in the ganglion cell layer were present in rats that had been exposed to tritiated thymidine on embryonic days E14-E22. Retinal ganglion cells were generated between E14 and E20, the ipsilaterally projecting ganglion cells ceasing their neurogenesis a full day before the contralaterally projecting ganglion cells. Alpha cells were generated from the very outset of retinal ganglion cell genesis, at E14, but completed their neurogenesis before the other cell types, by E17. Tritium-positive, horseradish peroxidase-negative neurons in the ganglion cell layer were present from E14 through to E22, and are interpreted as displaced amacrine cells. Choline acetyltransferase-positive displaced amacrine cells were generated between E16 and E20. Individual cell types showed a rough centroperipheral neurogenetic gradient, with the dorsal half of the retina slightly preceding the ventral half. These results demonstrate, first, that retinal ganglion cell genesis and displaced amacrine cell genesis overlap substantially in time. They do not occur sequentially, as has been commonly assumed. Second, they demonstrate that the alpha cell population of retinal ganglion cells and the choline acetyltransferase-immunoreactive population of displaced amacrine cells are each generated over a limited time during the periods of overall ganglion cell and displaced amacrine cell genesis, respectively. Third, they show that the very earliest ganglion cells to be generated in the temporal retina have exclusively uncrossed optic axons, while the later cells to be generated therein have an increasing propensity to navigate a crossed chiasmatic course.     

Direct and indirect retinal input into degenerated dorsal lateral geniculate nucleus after striate cortical removal in monkey: implications for residual vision

Summary We removed the striate cortex of one cerebral hemisphere in a macaque monkey, causing almost total retrograde degeneration of the corresponding dorsal lateral geniculate nucleus (dLGN) and extensive transneuronal degeneration of ganglion cells in the corresponding hemi-retina of each eye. The rare surviving geniculate projection neurons were retrogradely labelled by horseradish peroxidase (HRP) from extra-striate cortex and retinogeniculate terminals were labelled by an intraocular injection of HRP. Retinal terminals in the degenerated dLGN made synaptic contact exclusively with the dendrites of interneurons immunopositive for -aminobutyric acid (GABA) in both parvocellular and magnocellular regions of dLGN. As well as being postsynaptic to retinal terminals these vescicle-containing dendrites were pre- and postsynaptic to other similar dendrites, and presynaptic to relay cells. Surviving labelled projection neurons received retinal input indirectly, via both the GABA-immunopositive interneurons and GABA-immunonegative terminals characteristic of those from the superior colliculus. In the degenerated, as opposed to the normal dLGN, about 20% of retinal terminals were GABA-immunopositive and GABA-immunoreactivity was prominently elevated in the ganglion and amacrine cell layers of the degenerated half of the retina. The optic nerve also contained numerous GABA-immunopositive axons but very few such axons were found in a normal optic nerve processed in identical manner. The surviving pathways from the retina must underlie the visual abilities that survive striate cortical removal in monkeys and human patients and may involve the degenerated dLGN as well as the mid-brain.     

成体金黄地鼠神经节细胞轴突在外周神经移植到视网膜后的再生

在成体金黄地鼠的视网膜上移植一段自体坐骨神经1～4个月后,视网膜神经节细胞纤维长入移植的坐骨神经中长达2 cm。具有再生纤维的神经节细胞分布于外周神经插入处与视网膜边缘之间的扇形或带状区域内。细胞数目与插入处的位置有关,接近视神经乳头处较远离视神经乳头处标记细胞数量多。移植后的视网膜标记神经元细胞体面积分布直方图表明:具有再生纤维的神经元胞体面积范围除包括正常大小的神经节细胞外,还包括相当多的胞体增大的神经元。用荧光染料核黄施于视束、真蓝施于移植的坐骨神经后的逆行荧光双标记法实验表明,再生的轴突起源于神经节细胞的损伤轴突,而不是完整神经节细胞轴突的侧芽。     

An immunocytochemical marker for hamster retinal ganglion cells

Summary We examined the specificity and developmental time course of the labelling of retinal ganglion cells in Syrian hamsters by a monoclonal antibody AB5. In adult hamsters, AB5 selectively labelled somata in the ganglion cell layer, dendrites in the inner plexiform layer and axons in the nerve fibre layer. When retinal ganglion cells were retrogradely labelled with Dil prior to AB5 immunocytochemistry, all of the retrogradely labelled retinal ganglion cells in the ganglion cell layer were AB5 immunoreactive, indicating that AB5 labels all classes of ganglion cell in that layer. In retinae depleted of retinal ganglion cells by neonatal optic nerve transections, AB5 did not label any somata or processes, indicating that AB5 specifically labels retinal ganglion cells. During development, AB5 labelling first appeared as a weak staining of cell bodies in the ganglion cell layer on postnatal day 12 (P12; PO=first 24 h following birth) and acquired the staining pattern seen in the adult by postnatal day 14. From the onset of AB5 immunoreactivity, AB5-labelled somata of varying sizes were present across the entire retinal surface. Although AB5 labelled retinal ganglion cell axons in the nerve fibre layer of the retina it did not label the optic nerve or retinal ganglion cell axons in the brain at any age examined. AB5 labelling was also found to be compatible with bromodeoxyuridine immunocytochemistry and, therefore, useful for determining the time of generation of hamster retinal ganglion cells.     

Specific transcellular carbocyanine-labelling of rat retinal microglia during injury-induced neuronal degeneration

The present work employed a new technique for labelling phagocytizing microglia in the axotomized retinal of adult rats. Transection axotomy was performed within the intraorbital segment of the optic nerve, and the fast-transported, vital fluorescent carbocyanine dyes DiI and 4Di-10ASP were deposited at the ocular stump of the nerve in order to retrogradely prelabel the ganglion cells which were destined to die. Optic nerve transection resulted in progressive degradation of ganglion cell axons, perikarya and dendrites within the retina and in release of fluorescent material which was then incorporated into cells identified as microglia but not into other cells of the retina. Incorporation of labelled material into microglia occurred only when the ganglion cells degenerated and not when the non-lesioned ganglion cells were labelled from the superior colliculus. Double-staining of microglia with both dyes helped to compare the pattern of labelling for each dye. After progression of ganglion cell degeneration, microglia displayed a staggered, bilaminated distribution within the ganglion cell layer and within the inner plexiform layer. Fluorescent microglia were not found within the deeper layers of the retina indicating that transneuronal degeneration and subsequent labelling of microglial cells do not occur. The results show that one major function of microglia within the ganglion cell and inner plexiform layers of the lesioned retina is to remove debris produced after degradation of neurons.     

Cell populations of the ganglion cell layer: displaced amacrine and matching amacrine cells in the pigeon retina

Summary The proportion and size distribution of ganglion and non-ganglion cells in the ganglion cell layer of different areas of the pigeon retina was examined in whole-mounts of the retina by retrograde axonal transport of horseradish peroxidase (HRP) from large brain injections. A maximum of 98% of cells were labelled in the red field and a maximum of 77% in the peripheral yellow field. Unlabelled cell bodies were 30% smaller than labelled ganglion cells and had a mean diameter of 6.2 m and a size range of 4 to 9 m. The morphology of cells in the ganglion cell layer was examined by Golgi staining of retinal whole-mounts. Small glia, displaced amacrine and ganglion cells were found. Displaced amacrine cell bodies were about 30% smaller than ganglion cells and their size distribution was similar to the unlabelled cells in HRP preparations. Displaced amacrine cells had small rounded cell bodies (mean diameter 6.2 m) increasing in size with eccentricity, and a unistratified dendritic tree of fine, nearly radial, varicose dendrites in sublamina 4 of the inner plexiform layer. They had elliptical dendritic fields (mean diameter 66 m) aligned parallel to the retina's horizontal meridian. A population of amacrine cells was found with somas at the inner margin of the inner nuclear layer and soma and dendritic morphology matching those of displaced amacrines. These amacrine cells had unistratified dendritic trees at the junction of sublaminae 1 and 2 of the inner plexiform layer. Pigeon displaced amacrine cells and their matching amacrines are similar to starburst cells of the rabbit retina. They may participate in on and off pathways to ganglion cells and their lamination suggests that they are cholinergic.     

Generation of retinal cells in the wallaby, Setonix brachyurus (quokka)

We have examined the generation of retinal cells in the wallaby, Setonix brachyurus (quokka). Animals received a single injection of tritiated thymidine between postnatal days 1-85 and retinae were examined at postnatal day 100. Retinae were sectioned, processed for autoradiography and stained with Cresyl Violet. Ganglion cells were labelled by injection of horseradish peroxidase into the optic tracts and primary visual centres. Other cells were classified according to their morphology and location. Retinal cell generation takes place in two phases. During the first phase, which concludes by postnatal day 30, cells destined to lie in all three cellular layers of the retina are produced. In the second phase, which starts by postnatal day 50, cell generation is almost entirely restricted to the inner and outer nuclear layers. Cells produced in the first phase are orthotopic and displaced ganglion cells, displaced and orthotopic amacrine cells, horizontal cells and cones. Glia in the ganglion cell layer, orthotopic amacrine cells, bipolar and horizontal cells. Muller glia, and rods are generated in the second phase. Cells became heavily labelled with tritiated thymidine in the central retina before postnatal day 7, over the entire retina (panretinal) by postnatal day 7 and from postnatal day 18, only in the periphery. The second phase of cell generation is initiated at P50, in a region extending from the optic nerve head to mid-temporal retina. Subsequently, cells are generated in annuli, centred on mid-temporal retina, which are seen at progressively more peripheral locations. Therefore, cell addition to the inner and outer nuclear layers continues for longer in peripheral than in mid-temporal retina. We suggest that such later differential cell addition to the inner and outer nuclear layers contributes to an asymmetric increase in retinal area. This non-uniform growth presumably results in more expansion of the ganglion cell layer peripherally than in mid-temporal retina and may play a role in establishing density gradients of ganglion cells.     

Quantitative analysis of GABA-immunoreactive synapses in the inner plexiform layer of theBufo marinus retina: identification of direct output to ganglion cells and contacts with dopaminergic amacrine cells

Summary We have recently reported that about 50% of amacrine cells and some of the bipolar and ganglion cells are GABA-immunoreactive in the retina ofBufo marinus. Synapses formed by these elements in the inner plexiform layer were studied. GABA-immunoreactive amacrine cell processes were found most frequently in synaptic contact with non-immunoreactive amacrine cells. Double-label experiments showed that some of these non-GABA-immunoreactive elements contain tyrosine hydroxylase immunoreactivity. Another source of input to the GABA-immunoreactive amacrine cells were the bipolar cells; some of which were GABA-immunoreactive. GABA-immunoreactive amacrine cells synapsed also onto bipolar cell terminals, and ganglion cell dendrites that were identified by the retrograde transport of horseradish peroxidase from the optic nerve. Synapses between GABA-immunoreactive amacrine cells and bipolar and ganglion cells were non-uniformly distributed in the inner plexiform layer. Synaptic contacts with bipolar cells were more frequent in the OFF-sublamina, and those with ganglion cell dendrites in the ON-sublamina. These results demonstrate that GABA-immunoreactive amacrine cells (1) preferentially synapse with OFF-responding bipolar and ON-centre ganglion cells in the through-pathway, (2) synapse with tyrosine hydroxylase-immunoreactive amacrine cells in both the OFF- and ON-sublaminae, and (3) synapse directly with GABA-immunoreactive ganglion cells. The synapses between GABA-immunoreactive amacrine and GABA-immunoreactive ganglion cells may inhibit the centrally projecting inhibitory ganglion cells, causing disinhibition in the visual centres.On leave from Department of Zoology, Attila József University, Szeged, Hungary.     

Displaced retinal ganglion cells in normal frogs and those with regenerated optic nerves

Summary We have analysed the number and spatial distribution of displaced retinal ganglion cells in the frog Litoria (Hyla) moorei. A series of normal animals was compared with one in which the optic nerve was crushed and allowed to regenerate. Ganglion cells were labelled with horseradish peroxidase (HRP) applied to the optic nerve, and retinae were examined as sections or whole mounts. We analysed separately ganglion cells with somata displaced to the inner nuclear (Dogiel cells, DGCs) and to the inner plexiform layer (IPLGCs). These findings were related to data for the orthotopic ganglion cells (OGCs). The mean number of DGCs in the normal series was 2,550 (±281) and fell to 1,630 (±321) after regeneration, representing a mean loss of 36%. This reduction was not significantly different from the mean loss of 43% from the OGC population in which mean values fell from 474,700 (±47,136) to 268,700 (±54,395). In both the normal and the regenerate series, DGCs were estimated to represent means of only 0.6% of the OGC population. Densities of DGCs were highest in the nasoventral and temporo-dorsal peripheries; densities of both DGCs and OGCs were lower after optic nerve regeneration. We conclude that the factors which affect ganglion cell death during optic nerve regeneration, do so to similar extents amongst the DGC and the OGC populations. The IPLGCs were very rare in normal animals with a mean of 420 (±95). However, their numbers increased after regeneration to a mean of 3,350 (±690), estimated to be 1.2% of the OGC population. These cells normally favoured peripheral retina but became pan-retinal after regeneration. The primary dendrites of the majority of IPLGCs were oriented in the same direction as those of OGCs. We conclude that most IPLGCs were OGCs which had relocated their somata to the inner plexiform layer.     

Is Thy-1 expressed only by ganglion cells and their axons in the retina and optic nerve?

Summary The distribution of Thy-1 in the retina and optic nerve has been examined immunohistochemically, and compared to that of the astrocytic marker glial fibrillary acidic protein. The axons and cell bodies of ganglion cells were found to be Thy-1 positive as were processes within the inner plexiform layer. Transection of the optic nerve in the neonatal rat results in the rapid degeneration of the ganglion cells but some Thy-1 staining remains in the inner plexiform layer. We have estimated using an immunoassay of normal and optic nerve transected retinae that about 70% of the Thy-1 in the retina is on ganglion cells and their axons and the remainder is on cells which contribute processes to the inner plexiform layer, presumably amacrine, bipolar or Müller cells.In the optic nerve the Thy-1 was found to be limited to the fascicles of optic nerve fibres and the intrafascicular spaces, containing astrocytes and their processes, were not stained. Axotomy of the adult nerve, which produced axonal degeneration and astrocytic proliferation, led to a loss of over 95% of the Thy-1 from the nerve. We found no evidence that the astrocytes of the retina or optic nerve were Thy-1 positive in normal animals or during degeneration.     

Binocular interaction in the fetal cat regulates the size of the ganglion cell population

During fetal development of the cat's visual system there is a marked overproliferation of optic nerve axons. In utero binocular interaction contributes to the severity of fiber loss since removal of an eye during gestation attenuates axon loss in the remaining optic nerve. The purpose of the present study was to determine whether this reduced loss of optic nerve fibers is due to a failure of retraction by supernumerary axon branches or to a reduction in ganglion cell death. To resolve this issue, we compared the number of ganglion cells and optic nerve fibers in adult cats which had one eye removed at known gestational ages. Retinal ganglion cells were backfilled with horseradish peroxidase and counts were made from retinal wholemounts. The axon complement was assessed with an electron microscopic assay. In the retinas of a normal cat we estimated 151,000 and 152,000 ganglion cells. The optic nerves of two other normal cats contained approximately 158,000 and 159,000 axons. In comparison, an animal enucleated on embryonic day 42 had 180,000 ganglion cells and 178,000 optic nerve fibers, while in an animal enucleated on embryonic day 51 the corresponding estimates were 182,000 and 190,000. The close agreement between cell and fiber counts indicates that axonal bifurcation does not contribute appreciably to the axon surplus in the optic nerve of prenatally enucleated cats. These results demonstrate that prenatal binocular interaction regulates the size of the mature retinal ganglion cell population.     

Identification of amacrine and ganglion cells in the carp retina.

1. Amacrine and ganglion cells in the carp retina were identified from such criteria as photoresponses, intracellular dye staining, responses to optic nerve stimulation and behaviour to a synapse blocking agent. 2. Responses of ganglion cells were accompanied by spike discharges., either facilitated or suppressed by photic stimulation. The cells were also invaded by antidromic impulses, which survived after chemical synapses had been blocked by application of atomized CoCl2 solution. In subsequent histology of the Procion-stained neurones, the cell bodies were found in the ganglion cell layer and the axons were often traced. 3. Amacrine cells were subdivided into two types. The first type gave rise to transient depolarizations at both on- and offsets of spot and annulus illuminations, usually being associated with spike discharges of which the amplitudes varied in different cells. In histology, the cell bodies of this type were situated in the inner nuclear layer and dendrites ramified in two or more discrete sublayers of the inner plexiform layer (the stratified amacrine cell of Cajal). 4. The second type of amacrine cells produced sustained responses during illumination, being associated with no spike but with small oscillatory wavelets. The cell bodies were situated in the inner nuclear layer and the dendrites ramified in a single sublayer of the inner plexiform layer (the monolayered amacrine cell). 5. An attempt was made to see the effect of activation of centrifugal fibres on amacrine cells, but almost all of about 200 cells examined did not respond to optic nerve stimulation. Only two cells produced, with long latency, a small post-synaptic depolarization which disappeared after chemical synapses in the retina had been blocked. It is considered that the physiological role of the centrifugal system is insignificant in the carp retina.     

Uptake and axonal transport of horseradish peroxidase isoenzymes by different neuronal types

The uptake and transport of basic and acidic horseradish peroxidase isoenzymes was compared in the neuromuscular, visual and olfactory systems of Xenopus larvae and postmetamorphic frogs. The concentration (w/v) of the two preparations was corrected to compensate for their difference in enzymatic activity (unit/w), which was seven-fold higher in basic horseradish peroxidase. Uptake and transport of horseradish peroxidase isoenzymes could be demonstrated with 7% basic horseradish peroxidase, but not with equal amounts of 49% acidic horseradish peroxidase in all systems investigated: retrograde transport from terminals of retinal ganglion cells, isthmotectal neurons and spinal motoneurons, as well as anterograde transganglionic transport along olfactory neurons. A very weak labelling of the same neuronal pathways by acidic horseradish peroxidase was obtained only after increasing the amount injected by approximately two-fold. Basic horseradish peroxidase isoenzymes were also preferentially taken up and transported retrogradely by broken axons of the optic nerve. When tested, similar results were obtained in both larvae and frogs suggesting that preferential uptake and transport of basic horseradish peroxidase is a general feature of all neurons and of all developmental stages. Electron microscopical analysis of the outer layers of the optic tectum revealed that, in the same experimental conditions producing no retrotrade labelling of optic axons, acidic horseradish peroxidase was rarely found to enter nerve terminals. It appears that interactions between horseradish peroxidase and neuronal membranes occur during uptake and transport and that molecular charge plays an important role, beyond non-specific fluid-phase endocytosis. We suggest that differences between horseradish peroxidase isoenzymes as neuronal tracers reflect a process of adsorptive endocytosis related to general characteristics of neuronal membranes (regardless of age) and not to specific receptor-mediated interactions characteristic of neuronal specificity.     

Naturally occurring and induced ganglion cell death

Summary The retina in frogs grows continuously throughout the whole life of the animal by the addition of rings of cells at the ciliary margin. Naturally occurring neuron death cannot, consequently, be established by counting surviving neurons. A new approach, retinal wholemount autoradiography was introduced in this study to estimate cell loss occurring in the ganglion cell layer over a long period of time. ³H-thymidine injection at stage 53 (midlarval stage) labels a ring of cells, thereby marking the extent of retina formed up to the time of isotope administration. In the present study the number of neurons in the ganglion cell layer within the autoradiographically identified central retinal sector was estimated from midlarval stage to 6 months after metamorphosis in Xenopus laevis.The mean neuron number in the central retinal sector formed up to stage 53 was 17,420 and this was reduced by 20% to 13,515 by 6 months after metamorphosis. Optic nerve section at the time of isotope injection and subsequent regeneration brought about a reduction of the number of surviving neurons in the part of the retina formed up to stage 53 to 7,720, or to about 57% of the normal neuron number in an equivalent retinal area of an intact eye of the same age. A further reduction to 20% of normal neuron population was observed in retinae where the optic nerve failed to regenerate. The surviving neurons are assumed to be amacrine cells.The bulk of natural neuron loss in the retinal centre occurs during premetamorphic stages while little further loss takes place in the next 6 months suggesting that the underlying mechanism is a fine tuning of the developing retinal projections.     

Somatostatin-like immunoreactive material in associational ganglion cells of human retina

The retinal ganglion cell is classically viewed as the output cell of the retina, sending a single axon via the optic nerve to synapse in visual relay nuclei of the brain. However, some ganglion cells, termed associational ganglion cells, have axons which do not leave the retina and presumably serve intraretinal communication. Using high-affinity and specific monoclonal antibodies to somatostatin-14 and the avidin-biotin-peroxidase immunohistochemical procedure, somatostatin-immunoreactive associational ganglion cells are specifically stained in human retinas obtained at necropsy. These cells are more numerous in the inferior than the superior retina; they have dendrites which ramify in the inner plexiform layer; and they have sparsely branching axons, many of which can be traced over 1 cm. These axons do not enter the optic nerve. They follow remarkably straight courses at the border of the inner plexiform layer and ganglion cell layer and thereby form a gridwork of fibers covering the entire retinal area. These observations verify the existence of associational ganglion cells in the human and establish somatostatin as a neurotransmitter or neuromodulator candidate for these neurons. The morphology of these cells suggests that they are involved in long-distance interactions within the retina.     

Evidence for differential effects of terminal and dendritic competition upon developmental neuronal death in the retina

Retinal ganglion cells with ipsilaterally projecting axons were labelled with horseradish peroxidase injected unilaterally along the optic pathway in adult rats. Unoperated controls were compared with three groups of animals operated at birth, given (a) contralateral enucleation, (b) contralateral lesion to the optic tract or (c) both lesions simultaneously. The numbers of ipsilaterally projecting cells were increased in all three operated groups, presumably because of a reduction in natural neuronal death following diminished terminal and dendritic competition. The pattern of increase of labelled cell density varied with the type of lesion: enucleation led to a major increase within lower temporal retina; optic tract lesion caused its major increase in upper temporal retina, centred at the location of the area centralis; and the double lesion combined both effects above. The distribution of cell-body sizes was differentially affected by the lesions: enucleation led to a shift in the distribution towards the small cell side of the spectrum, when compared with the controls; optic tract lesion shifted the distribution towards the large cell side of the spectrum, but only outside the temporal crescent; and the double lesion led to a shift towards small cells within the temporal crescent and towards large cells outside the crescent, again combining the effects of the single lesions. Large alpha-like neurones with ipsilateral axons were common in the nasal retina of both groups given optic tract lesions but they were rare in the nasal retina of unoperated and, especially, of enucleated rats. The limits of the temporal crescent were unchanged, notwithstanding the large numbers of cells outside the crescent in operated rats. It is suggested that postnatal competitive interactions at the level of terminals and of dendrites control natural neuronal death in the rat retina with different requirements regarding retinal topography and ganglion cell types. The postnatal regulation of neuronal numbers is not responsible for the generation of the nasotemporal division but may be involved in the development of differential distributions of specific ganglion cell types across the retina.     

The role of nitric oxide in the apoptosis of neurons in the retina of the human fetal eye

The locations of NADPH-diaphorase (NADPH-d), inducible NO synthase (iNOS), and TUNEL-immunoreactive neurons in the retina of human fetuses collected during the first to third trimesters of pregnancy were studied. High levels of NADPH-d activity were seen in the inner segments of light-sensitive cells, amacrine cells, and ganglion cells. The population of NADPH-d-positive amacrine cells included three types of neuron. Type 1 neurons were large and had sparse dendritic fields occupying the inner nuclear and outer retinal layers. Small type 2 neurons were located in the inner retinal layer. Ectopic amacrine cells, type 3, were located in the outer part of the ganglion layer. A high density of NADPH-d-positive neurons was seen in the central part of the retina, surrounding the central fovea and optic disk area. NADPH-d activity increased progressively during ontogenesis and correlated with the appearance of immunoreactive iNOS in neurons. iNOS labeled a subpopulation of amacrine and ganglion cells, which appeared at 20–21 weeks of development and reached a peak of immunoreactivity by the end of the third trimester. TUNEL-immunopositive neuron nuclei with signs of apoptotic destruction were seen at 30–31 weeks of pregnancy. The greatest apoptotic index was seen in the ganglion and amacrine cell populations. These data identify NO as a factor mediating apoptosis of neurons during the critical period of differentiation of interneuronal connections in the human retina. Director: Doctor of Biological Sciences M. A. Vashchenko Director: Doctor of Biological Sciences S. L. Kondrashov __________ Translated from Morfologiya, Vol. 129, No. 1, pp. 42–49, January–February, 2006.