The earliest generated cells of the cat's telencephalon that may play a role in the formation of the primary visual cortex are the subject of this study. Using [3H]thymidine autoradiography, we have found that these cells are generated between embryonic day 24 (E24) and E30 (gestation is 65 days) and that they are present in very low numbers in the white matter of the adult brain. These cells are rarely labeled by injections made after E30, when the cells destined for the cortical layers are generated. Examination of the labeling pattern in the fetal brain 10 days or more after administration of [3H]thymidine between E24 and E30 revealed a bistratified distribution of these early generated cells. Labeled cells were found in large numbers in two embryonic zones flanking the developing cortical plate: above in the marginal zone and below in the subplate. (Some if not all of the marginal zone cells constitute the population of Cajal-Retzius cells of the cat's telencephalon.). These experiments indicate that cells of the subplate and marginal zones are cogenerated in time during the days just preceding the genesis of the cortical plate. We also examined the distribution of the early generated cells shortly after their genesis--on E30, a time when cells of the cortical plate are just being generated at the ventricular zone. In this case, the labeling pattern at the occipital pole was not bistratified. Rather, labeled cells were situated within a single zone extending from the pial surface inward to the border of the ventricular zone. This finding indicates that the cells of the subplate and marginal zones are generated as a contiguous population that is subsequently split apart by the insertion of cells forming the cortical plate. A comparison between the number of early generated cells found in fetal and newborn brains with that found in adult brains suggests that these cells are generated initially in substantial numbers but then largely disappear during early postnatal life, since injections of [3H]thymidine between E24 and E30 yielded large numbers of labeled cells in the white matter and layer 1 at birth, but very few at 2 months postnatal. This significant loss contrasted with the results from injections made just a few days later (E33) that resulted in large numbers of labeled cells in cortical layer 6 not only at birth but also in adulthood.(ABSTRACT TRUNCATED AT 400 WORDS)

The prenatal development of connections between the retina and the lateral geniculate nucleus (LGN) was studied by means of the anterograde axonal transport of 3H-amino acids or horseradish peroxidase injected intraocularly in fetal cats older than embryonic day 27 (E27) and in newborn cats. (Gestation is 65 days.) A retinothalamic pathway exists as early as E28, when label can be seen in both ipsilateral and contralateral optic tracts. Afferents from the contralateral eye are the first to invade the anlage of the LGN by E32 with those from the ipsilateral eye following about 3 days later. Initially, the pattern of labeling within the nucleus is uniform, suggesting that the two sets of afferents must share a good deal of territory at early ages. By E47, however, gaps appear in the labeling pattern contralaterally, indicating that afferents from the two eyes are beginning to segregate from each other. Segregation continues so that by E54 it is possible to identify unambiguously regions of the LGN destined to comprise ipsilateral and contralateral eye layers. By birth, afferent input appears adult-like in organization, with the two sets of afferents almost completely segregated from each other into their appropriate layers. Cellular lamination of the nucleus has just commenced, however, thereby lagging the onset of afferent segregation by about 2 weeks. Prenatal development could be followed much more easily in the horizontal than in the coronal plane of section due to the finding here that the LGN is displaced approximately 90 degrees in the horizontal plane between E40 and E60. Measurements of the area occupied by the ipsilateral and contralateral afferents within the LGN indicated that even prior to segregation, the two sets of afferents are not completely intermixed within the LGN. On the contrary, those from the contralateral eye retain almost exclusive control of some territory throughout development. This detail contrasts with development in primates, in which intermixing of afferents from the two eyes is thought to be complete early on (Rakic, P. (1976) Nature 261: 467-471). Nevertheless, in the cat, as in other mammals, development of the retinogeniculate pathway is broadly characterized by an initial period of overlap followed by a period of segregation that gives rise to the adult pattern of afferent input.

To study the prenatal development of connections between the lateral geniculate nucleus (LGN) and the primary visual cortex in the cat, we have examined the relationship between the position of ingrowing afferents from the LGN and their target cells in cortical layers 4 and 6 at various times during the cat's 65 d gestation period and during the first 3 weeks of postnatal life. In 1 series of experiments, the method of transneuronal transport of intraocularly injected tritiated proline (3H-proline), followed by autoradiography, was used to label the developing geniculocortical pathway. In another series, the tritiated thymidine (3H-thymidine) method was employed to keep track of the cells destined for layers 4 and 6 by labeling them on their birthdates (layer 4: embryonic day (E) 37-43; layer 6: E31-36) (Luskin and Shatz, 1985b) and then charting their locations at subsequent times during development. The results of the 2 sets of experiments were compared at corresponding ages. By E39, many of the cells of cortical layer 6 have completed their migrations and are situated within the cortical plate immediately above the subplate. However, the transneuronal labeling pattern indicates that the geniculocotical afferents have not yet arrived within the vicinity of the future visual cortex, but rather are still en route and confined within the optic radiations of the telencephalon. By E42, a week after the first afferents can be detected in the radiations, substantial transneuronal label is found in the subplate immediately below future visual cortex. However, the overlying cortical plate is free of label. Over the next 2 weeks, geniculocortical axons continue to accumulate in the subplate zone, and, in addition, transneuronal label can be found in the marginal zone. By E55 a faint geniculocortical projection can be detected within the cortical plate, but only within its deeper half (future layers 5 and 6), and even then the major portion of the projection is still confined to the subplate. The absence of a projection to cortical layer 4 at these ages is remarkable in view of the results from our 3H-thymidine experiments, which indicate that by E57 the majority of cells destined to belong to layer 4 have already completed their migrations and assumed positions superficial to the cells of layers 5 and 6. By birth, a substantial geniculocortical projection to cortical layer 4 can be detected in the transneuronal autoradiographs.(ABSTRACT TRUNCATED AT 400 WORDS)

We have studied the development of retinal ganglion cell morphology in the cat's visual system from early fetal to postnatal times. In particular, we have examined the contribution of growth and remodeling to the establishment of mature retinal ganglion cell form. Ganglion cells were identified by retrograde labeling with rhodamine latex microspheres deposited in the superior colliculus and lateral geniculate nucleus between embryonic day 34 (E34; birth = E65) and adulthood. To reveal the fine morphological details of retrogradely labeled ganglion cells, 48 hr later Lucifer yellow was injected intracellularly in living retinae that had been dissected and maintained in vitro. Our results show that at E35-37 the majority of ganglion cells are very simple in morphology, with a few dendritic processes that are generally aligned in a radial direction towards or away from the optic disc. During the ensuing 2 week period, there is a progressive growth and elaboration of dendrites. By E50, some ganglion cells resembling the adult alpha, beta, and gamma classes can be identified based on comparisons of the appearance and dimensions of their dendritic trees and somata with neighboring filled cells. However, ganglion cell dendrites and axons at this age express several transient morphological features. The axons of ganglion cells give rise to delicate processes originating from the intraretinal portion of the axon, including side branches, present in about half of the cells, and occasionally bifurcations that give rise to axon collaterals. These transient axonal features are present throughout development, including the neonatal period; no axon collaterals were observed after postnatal day 15, while axonal side branches persisted even at P31 but were gone by adulthood. Ganglion cell dendrites exhibit excessive branches and exuberant somatic and dendritic spines. Quantitative analysis of these processes shows that after E45 dendritic trees increase dramatically in complexity, reaching the peak number of spines and branch points by the first week of postnatal life. The number of dendritic processes then falls abruptly to reach near-adult levels by the end of the first postnatal month. Even though dendritic morphology closely resembles that seen in the adult at this age, ganglion cell bodies and dendrites must continue to grow to reach their adult size.(ABSTRACT TRUNCATED AT 400 WORDS)

During development of the mammalian cerebral cortex, thalamic axons must grow into the telencephalon and select appropriate cortical targets. In order to begin to understand the cellular interactions that are important in cortical target selection by thalamic axons, we have examined the morphology of axons from the lateral geniculate nucleus (LGN) as they navigate their way to the primary visual cortex. The morphology of geniculocortical axons was revealed by placing the lipophilic tracer Dil into the LGN of paraformaldehyde-fixed brains from fetal and neonatal cats between embryonic day 26 (E26; gestation is 65 d) and postnatal day 7 (P7). This morphological approach has led to three major observations. (1) As LGN axons grow within the intermediate zone of the telencephalon toward future visual cortex (E30-40), many give off distinct interstitial axon collaterals that penetrate the subplate of nonvisual cortical areas. These collaterals are transient and are not seen postnatally. (2) There is a prolonged period during which LGN axons are restricted to the visual subplate prior to their ingrowth into the cortical plate; the first LGN axons arrive within visual subplate by E36 but are not detected in layer 6 of visual cortex until about E50. (3) Within the visual subplate, LGN axons extend widespread terminal branches. This represents a marked change in their morphology from the simple growth cones present earlier as LGN axons navigate en route to visual cortex. The presence of interstitial collaterals suggests that there may be ongoing interactions between LGN axons and subplate neurons along the entire intracortical route traversed by the axons. From the extensive branching of LGN axons within the visual subplate during the waiting period, it appears that they are not simply "waiting." Rather, LGN axons may participate in dynamic cellular interactions within the subplate long before they contact their ultimate target neurons in layer 4. These observations confirm the existence of a prolonged waiting period in the development of thalamocortical connections and provide important morphological evidence in support of the previous suggestion that interactions between thalamic axons and subplate neurons are necessary for cortical target selection.

The morphological changes in individual retinal ganglion cell axons associated with the formation of the eye-specific layers in the dorsal lateral geniculate nucleus (LGN) were studied during the prenatal development of the cat's visual system. Previous work has shown that the pattern of segregated eye inputs found in the adult arises from an immature state in which inputs from the two eyes are intermixed within the nucleus (Shatz, 1983). Here, this developmental process is examined at its fundamental unit of connectivity--the individual retinal ganglion cell axon. To do so, an in vitro method was used to label fetal cat optic tract axons with HRP at various times during development between embryonic day 38 (E38) and postnatal day 2 (P2) (gestation = 65 d). The results presented here are based on reconstructions of 172 axons. During the initial period of intermixing (E38-43), axons are relatively simple in morphology. Many axons studied at the earliest ages (E38) end in growth cones and have very few branches along the main axon trunk as they traverse the nucleus. By E43, the number of side branches given off along the main axon trunk has increased and most axons also have a simple terminal arbor. Over the next 2 weeks (E43-55), the majority of axons are studded with side branches and the terminal arbor is well defined. Then, between E55 and birth, axons lose their side branches and the eye-specific layers appear. By birth, nearly all axons have a smooth trunk and an elaborate terminal arbor restricted to the LGN layer appropriate to the eye of axon origin. When the number of side branches per axon was quantified, the time course of appearance and subsequent loss of side branches was found to parallel the time course of the initial intermixing of inputs and subsequent reduction in territory shared by the two eyes as determined from previous intraocular injection experiments. Our results also showed that the side branches along each axon were located primarily within LGN territory destined to be occupied by the other eye. Thus, the side branches are likely to represent a morphological substrate for the intermixing of inputs from the two eyes. These observations suggest that the segregation of eye input to the LGN involves two fundamental and simultaneous events. One event is the remodeling of the branching pattern along the length of the main axon trunk so that the side branches present early on are eliminated and the main axon trunk becomes smooth.(ABSTRACT TRUNCATED AT 400 WORDS)

with the onset of cell death and maturation of axonal connections. This study demonstrates that members of the trk family, previously identified in the CNS on the basis of mRNA transcripts, are present as receptors with specific binding affinities for BDNF and NT-3. Moreover, the correspondence between the developmental shift from full-length to truncated trkB and the critical periods for cell fate determination, cell death, and axonal remodeling suggests an important role for neurotrophic factors in the development of the visual system.

The adult cerebral cortex extends axons to a variety of subcortical targets, including the thalamus and superior colliculus. These descending projections are pioneered during development by the axons of a transient population of subplate neurons (McConnell et al., 1989). We show here that the descending axons of cortical plate neurons appear to be delayed significantly in their outgrowth, compared with those of subplate neurons. To assess the possible role of subplate neurons in the formation of these pathways, subplate neurons were ablated during the embryonic period. In all cases, an axon pathway formed from visual cortex through the internal capsule and into the thalamus. In half of all cases, however, cortical axons failed to invade their normal subcortical targets. In the other half, targets were innervated normally. Subplate neurons are therefore likely to provide important cues that aid the process by which cortical axons grow toward, select, and invade their subcortical targets.