Spatial arrangement of radial glia and ingrowing retinal axons in the chick optic tectum during development

Neuroanatomical tracing of retinal axons and axonal terminals with the fluorescent dye, DiI, was combined with immunohistochemical characterization of radial glial cells in the developing chick retinotectal system. Emphasis was placed on the mode of the tectal innervation by individual retinal axons and on the distribution and fate of the tectal radial glial cells and their spatial relation to retinal axons. It was obvious from fluorescent images obtained from anterogradely filled axons that these axons deserted the superficial stratum opticum (SO) to penetrate the stratum griseum et fibrosum superficiale (SGFS) by making right-angled turns within the SO. Frequently, axons which had invaded the SGFS were bifurcated and had a superficial branch which remained within the SO. Terminal axonal arborization occurred at various depths within the SGFS. Characterization of the tectal glial cells and their radial fibers by means of the anti-filament antibody, R5, and post-mortem staining with the fluorescent dye, DiI, revealed the following. (a) At least from day E8 to P1, tectal glial fibers traversed all tectal layers from the periventricular location of their somata to the superficial interface between SO and pia mater. In this interface they enlarged and formed characteristic endfeet. (b) Glial endfeet covered the whole tectal surface. They showed at early ages anterior-posterior differences having a higher density in the posterior tectum. These differences disappeared at embryonic day E13. (c) After innervation, glial endfeet of the anterior tectal third were arranged in rows parallel to the retinal fibers within the SO. This arrangement was not observed in eyeless embryos. (d) Radial glial fibers could be stained with R5 from day E8 to late embryonic stages throughout their entire length. (e) At the first posthatching days, only the segments of the radial glial fibers restricted to the thickness of the SO were R5-positive, although the fibers still traversed throughout the depth of the tectum. The results are discussed in context to the genesis of the retinotectal projection.     

Astroglial differentiation in the opossum superior colliculus

Glial markers, namely, vimentin, glial fibrillary acidic protein (GFAP), and glycogen, as well as accumulation of axon-borne horseradish peroxidase (HRP), were used to visualize radial glial cells in the developing opossum superior colliculus (SC) and to follow changes in young astrocytes of the superficial layers. Vimentin, GFAP, and glycogen are relatively abundant in elements of the median ventricular formation (MVF), which persists at least as late as weaning time, i.e., postconception day 103, postnatal day 90 (PND90). Radial profiles and end-feet in the remaining collicular sectors (main radial system, MRS) are also vimentin-positive but show little or no glycogen or anti-GFAP staining. The numeric density of MRS profiles is very high at the final stages of neuronal migration (PND12) but falls to vestigial numbers by PND 56-60. Antivimentin staining and filling of MRS profiles by axon-borne HRP disappear in parallel. Before total regression of MRS profiles, young astrocytes of the superficial gray layer exhibit a transiently high GFAP expression that is not found in those of the subjacent layers. The results suggest that 1) radial glia at or near the collicular midline are well equipped for a mechanical supportive role, and their abundant glycogen accumulation may reflect their eventual transformation in cells with high glycolytic metabolism, including tanycytes; 2) in most collicular sectors, some radial glia cells persist for long periods after cessation of neuronal migration and may interact with afferent fibers coursing through the superficial neuropil; 3) radially oriented astrocytes of the superficial gray layer exhibit a transiently high GFAP expression that is temporally correlated with late transformations of the retinocollicular projections.     

Cross-species collapse activity of polarized radial glia on retinal ganglion cell axons

Inhibition of incorrect axonal outgrowth has been shown to be a crucial guidance mechanism during the development of the nervous system. Within the visual system of chick and rat, extension of retinal ganglion cell axons is essentially restricted to distinct layers of the retina and distinct brain regions such as the tectum opticum. In addition, populations of ganglion cells from defined retina locations project topographically to defined tectal areas, their growth possibly being inhibited by radial glia in incorrect tectal regions. In the current study, we aimed to analyse potential inhibitory activity of retinal glia during outgrowth of ganglion cell axons of embryonic chick and rat. The response of ganglion cell axons originating from different retina locations when exposed to purified retinal radial glia cell membranes were monitored in collapse assays by time lapse video recording. The interaction of axons growing on purified glial somata or glial endfeet was analysed in outgrowth assays. Our results indicate that (1) nasal and temporal chick growth cones are equally induced to collapse by cell membranes from retinal radial glia: 75% nasal and 72% temporal. (2) The collapse inducing component of radial glia can be inactivated by defined heat treatment, reducing collapsing activity to 6% nasal and 5% temporal. (3) Rat growth cones respond in a similar way to chick radial glia. (4) Rat axons grow perfectly on endfeet but not on somata of radial glia of the chick. In summary, the data suggest that radial glia are functionally polarized with permissive endfeet and inhibitory somata based on heat-labile proteins. Glia polarization is likely to inhibit aberrant growth of ganglion cell axons into outer retina layers. However, retinal radial glia are unlikely to participate in preordering axons within the retina and therefore do not affect the topographic projection. Finally, the inhibitory function of radial glia is conserved between birds and mammals and represents possibly a fundamental mechanism for structuring the central nervous system. GLIA 25:143–153, 1999. © 1999 Wiley-Liss, Inc.     

Evidence for shifting connections during development of the chick retinotectal projection

The pattern in which optic axons invade the tectum and begin synaptogenesis was studied in the chick. The anterogradely transported marker, horseradish peroxidase, was injected into one eye of embryos between 5 and 16 days of development (E5 to E16). This labeled the optic axons in the brain. The first retinal axons arrived in the most superficial lamina of the tectum on E6. They entered the tectum at the rostroventral margin. During the next 6 days of development the axons grew over the tectal surface. First they filled the rostral tectum, the oldest portion of the tectum, and then they spread to the caudal pole. Shortly after the first axons entered the tectum on E6, labeled retinal axons were found penetrating from the surface into deeper tectal layers. In any given area of the tectum, optic axons were seen penetrating deeper layers shortly after arriving in that area. Electron microscopic examination showed that at least some of the labeled axons in rostral tectum formed synapses with tectal cells by E7. These results show two things which contrast with results from previous studies. First, there is no delay between the time the retinal axons enter the tectum and the time they penetrate into synaptic layers of the tectum. Second, the first retinotectal connections are formed in rostral tectum and not central tectum. Retrograde tracing showed the first optic axons that arrived in the tectum were from ganglion cells in central retina. Previous studies have shown that the ganglion cells of central retina project to the central tectum in the mature chick. This opens the possibility that the optic axons from central retina, which connect to rostral tectum in the young embryo, shift their connections to central tectum during subsequent development. As they enter the tectum the growth cones of retinal axons appear to be associated with the external limiting membrane. During the time that connections would begin to shift in the tectum a second population of axons appears at the bottom of stratum opticum, some with characteristics of growth cones. This late-appearing population may represent axons shifting their connections. These results have implications for theories on how the retinotopic pattern of retinotectal connections develops.     

Localization of Nogo and its receptor in the optic pathway of mouse embryos

We have investigated the localization of Nogo, an inhibitory protein acting on regenerating axons in the adult central nervous system, in the embryonic mouse retinofugal pathway during the major period of axon growth into the optic chiasm. In the retina, Nogo protein was localized on the neuroepithelial cells at E12 and at later stages (E13-E17) on radial glial cells. Colocalization studies showed expression of Nogo on vimentin-positive glia in the retina and at the optic nerve head but not on most of the TuJ1- and islet-1-immunoreactive neurons. Only a few immature neurons in the ventricular and peripheral regions of the E13 retina were immunoreactive to Nogo. In the ventral diencephalon, Nogo was expressed on radial glia, most strongly on the dense radial glial midline raphe within the chiasm where uncrossed axons turn and in the initial segment of the optic tract. In vitro studies showed that the Nogo receptor (NgR) was expressed on the neurites and growth cones from both the ventral temporal and dorsal nasal quadrant of the retina. In the optic pathway, NgR staining was obvious in the vitreal regions of the retina and on axons in the optic stalk and the optic tract, but not in the chiasm. These expression patterns suggest an interaction of Nogo with its receptor in the mouse retinofugal pathway, which may be involved in guiding axons into the optic pathway and in governing the routing of axons in the optic chiasm.     

Plasticity in the chick ventral lateral geniculate nucleus

The chick ventral lateral geniculate nucleus (GLv) receives topographically corresponding projections from the retina and optic tectum. Tectal lesions produced on the day of hatching removed the tectogeniculate input to the GLv region corresponding to the tectal lesion and also severed some retinotectal axons. Following a survival period of 3 to 10 weeks, a patch of augmented retinogeniculate projection was noted in the GLv segment that corresponds topographically to the damaged area of the tectum. Changing the site of the tectal lesion led to changes in the locus of heavy retinal projection to the GLv predictable from topographic maps. Nuclei which received retinal but not tectal projections did not appear to have regions of augmented retinal termination nor did nuclei which received tectal but not retinal innervation. It is unlikely that the increased retinogeniculate termination is due to rerouting of growing retinotectal axons since the chick retinofugal pathway is well established by the time of hatching. Furthermore, there was no evidence of a projection from the ipsilateral eye to the affected GLv. On the basis of these light microscopic studies, it would appear that retinogeniculate terminals have sprouted in the GLv and that competition for terminal space, conservation of terminal space, proximity, and perhaps other factors are necessary for the augmented projection to occur.     

Persistence of graded EphA/Ephrin-A expression in the adult frog visual system

Many studies have demonstrated the involvement of the EphA family of receptor tyrosine kinases and their ligands, ephrin-A2 and -A5, in the development of the temporonasal axis of the retinotectal/collicular map, but the role of these molecules in optic nerve regeneration has not been well studied. Noting that the characteristic gradients of the EphA/ephrin-A family that are expressed topographically in the retina and tectum of embryonic chicks and mice tend to disappear after birth, we took as our starting point an analysis of EphA and ephrin-A expression in leopard frogs (Rana pipiens and utricularia), species capable of regenerating the retinotectal map as adults. For the EphA family to be involved in the regeneration, one would expect these topographic gradients to persist in the adult or, if downregulated after metamorphosis, to be reexpressed after optic nerve injury. Using EphA3 receptor and ephrin-A5 ligand alkaline phosphatase in situ affinity probes (RAP and LAP, respectively) in whole-mount applications, we report that reciprocally complementary gradients of RAP and LAP binding persist in the optic tract and optic tectum of postmetamorphic frogs, including mature adults. EphA expression in temporal retinal axons in the optic tract was significantly reduced after nerve section but returned during regeneration. However, ephrin-A expression in the tectal parenchyma was not significantly elevated by either eye removal, with degeneration of optic axons, or during regeneration of the retinotectal projection. Thus, the present study has demonstrated a persisting expression of EphA/ephrin-A family members in the retinal axons and tectal parenchyma that may help guide regenerating fibers, but we can offer no evidence for an upregulation of ephrin-A expression in conjunction with optic nerve injury.     

Conduction velocity distribution of the retinal input to the hamster's superior colliculus and a correlation with receptive field characteristics

Cells driven reliably by shocks delivered to the optic nerve or optic chiasm were encountered throughout the depth of the colliculus. The incidence of such cells, however, decreased markedly in the laminae ventral to the stratum opticum. The distribution of conduction velocities for the retinal afferents to the tectum was quite broad (range: 1.7-25.5 m/sec) and clearly biomodal with peaks at about 6 and 12 m/sec. A small number of cells were innervated by rapidly (> 15 m/sec) conducting axons. No evidence of an indirect-fast pathway from the retina to the colliculus via the lateral geniculate nucleus and visual cortex was obtained. Afferent conduction velocity was not correlated with retinal eccentricity, collicular depth or speed selectivity. It was, however, clearly related to directional selectivity. Ninety percent of the tectal neurons receiving inputs from axons having conduction velocities of less than 5 m/sec were directionally selective while only 41% of those neurons innervated by more rapidly conducting fibers (> 5 m/sec) exhibited selectivity. One hundred and sixteen cells in the anterior portion of the colliculus were tested with shocks delivered to the ipsilateral optic nerve and photic stimulation of the ipsilateral eye. Of these, 11% exhibited some degree of binocularity and only 6% were responsive to optic nerve shocks. These electrophysiological findings were correlated with the limited nature of the retinal input to the ipsilateral superior colliculus.     

Regulation of retinal ganglion cell axon arbor size by target availability: Mechanisms of compression and expansion of the retinotectal projection

The ability of pre- and postsynaptic populations to achieve the proper convergence ratios during development is especially critical in topographically mapped systems such as the retinotectal system. The ratio of retinal ganglion cells to their target cells in the optic tectum can be altered experimentally either by early partial tectal ablation, which results in an orderly compression of near-normal numbers of retinal projections into a smaller tectal area, or by early monocular enucleation, which results in the expansion of a reduced number of axons in a near-normal tectal volume. Our previous studies showed that changes in cell death and synaptic density consequent to these manipulations can account for only a minor component of this compensation for the population mismatch. In this study, we examine other mechanisms of population matching in the hamster retinotectal system. We used an in vitro horseradish peroxidase labeling method to trace individual retinal ganglion cell axons in superior colliculi partially ablated on the day of birth, as well as in colliculi contralateral to a monocular enucleation. We found that individual axon arbors within the partially lesioned tectum occupy a smaller area, with fewer branches and fewer terminal boutons, but preserve a normal bouton denstiy. In contrst, ipsilaterally projecting axon arbors in monoculary enucleated animals occupy a greater area than in the normal condition, with a much larger arbor length and greater number of boutons and branches compared with normal ipsilaterally projecting cells. Alteration of axonal arborization of retinalganglion cells is the main factor responsible for matching the retinal and tectal cell populations within the tectum. This process conserves normal electrophysiological function over a wide range of convergence ratios and may occur through strict selectivity of tectal cells for their normal number of inputs. © 1994 Wiley-Liss, Inc.     

The response of optic tract glia during regeneration of the goldfish visual system. II. Tectal factors stimulate optic tract glia

After transection, retinal ganglion cell axons of the goldfish will regenerate by growing into a primary target tissue, the optic tectum. To determine what role the target tissue may play in regulating glial cell growth, we measured biosynthetic activity of optic tract glia following excision of the optic tectum and compared it to activity of glia found in the regenerating visual system. Ablation of the tectum reduced glial incorporation of both [³H]thymidine and [³⁵S]methionine. Tectal ablation also led to nearly 80% reduction of amino acids incorporated by oligodendroglia as well as a decrease in the amount of newly synthetized protein found within multipotential glia and within cytoplasmic projections of astroglia. Since the tectal influence upon optic tract glia was detected at a time when tract and tectum are physically separated, we sought to determine if the optic tectum contained soluble glia-promoting factors. A soluble fraction recovered from tecta of the regenerating visual system increased amino acid incorporation within optic tract glia at 2–3-fold above preparations incubated with fractions from control, intact tecta. Comparisons of radiolabeled proteins separated by sodium dodecyl polyacrylamide gel electrophoresis from regenerating and factor-stimulated optic tract were similar and indicated that a soluble tectal fraction promoted biosynthesis of specific glial proteins. Our findings suggest that during regeneration of the goldfish visual system glia are influenced by humoral factor(s) released from the synaptic target site.     

Goldfish retina and tectum influence each other's growth activity during regrowth of the retinotectal projection

Variations in retinal and tectal growth activity, during regrowth of the goldfish retinotectal projection, were monitored by measuring the rates of incorporation of [14C]leucine into soluble protein and tubulin-enriched fractions at different times after crushing the optic nerves. Other experiments tested for growth-modulating interactions between tectum and retina. Here we studied how the absence of one of these structures (i.e. tectal ablation or eye removal) affected the profile of biosynthetic activity in the other. Experiments were also conducted on groups of fish in which the tectum was reinnervated by a half-retina (either half-nasal (1/2 N) or half-temporal (1/2 T) retina). This was done to ascertain if growth interactions between retina and tectum display any position-dependent differences that may be relevant to retinotopic ordering during regeneration. Our studies have revealed that: the retina and tectum of 1/2 T and 1/2 N groups differ in their growth responses during regeneration of the visual pathway: the tectum may exert a stimulatory and at other times an inhibitory influence on retinal protein synthesis; and retina and tectum display a bimodal profile of biosynthetic activity during regeneration that coincides with two stages of increased cell division (primarily glia) which other workers have found occurs in the tectum and tract during regeneration of the retinotectal projection. Indeed it seems there may be a link between this glial proliferation and the neurotrophic and guiding influences which tectum and retina exert upon one another during regeneration.     

Development of the transient ipsilateral retinotectal projection in the chick embryo: A numerical flourescence-microscopic analysis

The ipsilateral retinotectal projection in the developing chick was examined by using rhodamine-B-isothiocyanate (RITC)as an anterograde and retrograde vital marker for the retinal ganglion cells and their axons. Staining of the entire retina following intravitreal RITC injection between incubation days 3 and 16 revealed a small number of anterogradely labeled fibers in the optic tract and the anterior half of the optic tectum ipsilateral to the injection site. The total number of ipsilaterally projecting fibers was estimated to be about 2,000 on developmental day 9. The ipsilateral projection totally disappeared after day 15. The arrangement of fibers within the ipsilateral projection was examined by local anterograde RITC staining of localized retinal regions between days 9 and 10. The projection was retinotopically organized along the dorsoventral axis such that fibers of dorsal retinal origin projected on the ventral tectal half, whereas fibers of ventral retinal orgin projected on the dorsal tectal half. The localization of ipsilaterally projecting ganglion cell bodies was examined by retrograde RITC staining during days 9 and 15. Ganglion cells of all four quadrants of the central retina contributed to the production of the ipsilateral projection. The ipsilaterally growing retinotectal fibers did not represent collaterals of contralaterally projecting retinotectal axons. We assume that the tendency of early growing retinotectal axons to grow straight, as well as the ability of axonal growth cones to “sample” the environment, lead to a crossing of axons to the contralateral side. Ipsilateral projections would therefore represent “pathfinding errors.” Explanations for the elimination of the ipsilateral retinotectal projection are discussed.     

Topographic projections between the nucleus isthmi and the tectum of the frog Rana pipiens.

The connections between the nucleus isthmi and the tectum in the frog have been determined by several anatomical techniques: iontophoresis of horseradish peroxidase into the tectum, iontophoresis of 3H-porline into the nucleus isthmi and the tectum, and Fink-Heimer degeneration staining after lesions of the nucleus isthmi. The results show that the nucleus isthmi projects bilaterally to the tectal lobes. The ipsilateral isthmio-tectal fibers are distributed in the superficial layers of the tectum, coincident with the retionotectal terminals. The contralateral isthmio-tectal fibers travel anteriorly adjacent to the lateral optic tract and cross the midline in the supraoptic ventral decussation, where they turn dorsally and caudally; upon reaching the tectum, the fibers end in two discrete layers, layers 8 and A of Potter. The tectum projects to the ipsilateral nucleus isthmi and there is a reciprocal topographic relationship between the two structures. Thus, a retino-tecto-isthmio-tectal route exists which may contribute to the indirect ipsilateral retinotectal projection which is observed electrophysiologically. The connections between the nucleus isthmi and the tectum in the frog are strinkingly similar to the connections between the parabigeminal nucleus and the superior colliculus of mammals.     

Isoform‐specific localization of Nogo protein in the optic pathway of mouse embryos

Expression of Nogo protein was investigated in the optic pathway of embryonic mice by using isoform‐specific antibodies Bianca and 11C7, which recognize Nogo‐A/B and Nogo‐A, respectively. Our previous reports from using antibody N18 have shown that Nogo is localized on the radial glia in the retina and at the midline of the ventral diencephalon in mouse embryos during the ingrowth of retinal ganglion cells (RGCs) axons. This glial‐specific localization is markedly different from findings in other studies. This study showed Nogo‐A/B primarily on radial glia in the retina at E13 and then later on retinal ganglion cells and axons at E14 and E15, whereas Nogo‐A was expressed preferentially by RGCs and their axons. In the ventral diencephalon, Nogo‐A/B was expressed strongly on radial glia, particularly in those located in the midline region of the chiasm but also on RGC axons. In Nogo‐A knockout embryos, the isoform Nogo‐B (revealed by Bianca) was observed on radial glia in the ventral diencephalon and on RGCs and their axons. We concluded that Nogo‐A is localized on the ganglion cells and retinal axons, whereas Nogo‐B is expressed by the radial glia in the optic pathway. Nogo‐B may play an important role in guiding axon growth in decisive regions of the visual pathway, which include the optic disc and the optic chiasm. J. Comp. Neurol. 524:2322–2334, 2016. © 2016 Wiley Periodicals, Inc.     

Retinotopic organization of the developing retinotectal projection in the zebrafish embryo

Developing retinal axons in the zebrafish embryo were stained with HRP or with the fluorescent dyes dil and diO to study the formation of the retinotectal projection. Retinal axons leave the eye at 34-36 hr postfertilization (PF), invade the tectum at 46-48 hr PF, and innervate the tectal neuropil at 70-72 hr PF. Dorsal and ventral axons occupy separate aspects of the optic nerve and tract and pass into their retinotopically appropriate ventral and dorsal hemitectum, respectively. Nasal and temporal axons are segregated in the nerve, mixed in the tract, and are coextensive over the rostral half of tectum until 56 hr PF. They then segregate again, due to the progression of nasal axons into the open caudal tectum. Thus, at 70-72 hr PF, dorsal and ventral as well as temporal and nasal axons occupy their retinotopically appropriate tectal quadrants. After ablation of the temporal retina prior to the time of axonal outgrowth, the nasal axons bypass the vacant rostral tectum to terminate in the caudal tectal half. Temporal axons in the absence of nasal axons remain restricted to their appropriate rostral tectal half, suggesting that nasal and temporal axons possess a preference for their retinotopically appropriate tectal domains. Measurements of individual terminal arbors and the tectal areas in embryos and in adult zebrafish showed that individual arbors are large with respect to the embryonic tectum but are about 14-15 times smaller than in the adult. However, the proportion of tectum covered by embryonic arbors is about 7 times larger than in the adult, suggesting that a higher precision of the adult projection is achieved as a result of a greater enlargement of the tectum than of the arbors.     

Axon terminals from the nucleus isthmi pars parvocellularis control the ascending retinotectofugal output through direct synaptic contact with tectal ganglion cell dendrites

The optic tectum in birds and its homologue the superior colliculus in mammals both send major bilateral, nontopographic projections to the nucleus rotundus and caudal pulvinar, respectively. These projections originate from widefield tectal ganglion cells (TGCs) located in layer 13 in the avian tectum and in the lower superficial layers in the mammalian colliculus. The TGCs characteristically have monostratified arrays of brush‐like dendritic terminations and respond mostly to bidimensional motion or looming features. In birds, this TGC‐mediated tectofugal output is controlled by feedback signals from the nucleus isthmi pars parvocellularis (Ipc). The Ipc neurons display topographically organized axons that densely ramify in restricted columnar terminal fields overlapping various neural elements that could mediate this tectofugal control, including the retinal terminals and the TGC dendrites themselves. Whether the Ipc axons make synaptic contact with these or other tectal neural elements remains undetermined. We double labeled Ipc axons and their presumptive postsynaptic targets in the tectum of chickens (Gallus gallus) with neural tracers and performed an ultrastructural analysis. We found that the Ipc terminal boutons form glomerulus‐like structures in the superficial and intermediate tectal layers, establishing asymmetric synapses with several dendritic profiles. In these glomeruli, at least two of the postsynaptic dendrites originated from TGCs. We also found synaptic contacts between retinal terminals and TGC dendrites. These findings suggest that, in birds, Ipc axons control the ascending tectal outflow of retinal signals through direct synaptic contacts with the TGCs. J. Comp. Neurol. 524:362–379, 2016. © 2015 Wiley Periodicals, Inc.     

Mode of growth of retinal axons within the tectum of Xenopus tadpoles, and implications in the ordered neuronal connection between the retina and the tectum

Retinal axons of Xenopus tadpoles at various stages of larval development were filled with horseradish peroxidase (HRP), and their trajectories and the patterns of branching within the tectum were analyzed in wholemount preparations. To clarify temporal and spatial modes of growth of retinal axons during larval development, special attention was directed to labeling a restricted regional population of retinal axons with HRP, following reported procedures (H. Fujisawa, K. Watanabe, N. Tani, and Y. Ibata, Brain Res. 206:9-20, 1981; 206:21-26, 1981; H. Fujisawa, Dev. Growth Differ 26:545-553, 1984). In developing tadpoles, individual retinal axons arrived at the tectum, without clear sprouting. Axonal sprouting first began when growing tips of each retinal axon had arrived at the vicinity of its site of normal innervation within the tectum. Thus, the terminals of the newly added retinal axons were retinotopically aligned within the tectum. The retinotopic alignment of the terminals may be due to an active choice of topographically appropriate tectal regions by growth cones of individual retinal axons. The stereotyped alignment of the newly added retinal axons was followed by widespread axonal branching and preferential selection of those branches. Each retinal axon was sequentially bifurcated within the tectum, and old branches that had inevitably been left at ectopic parts of the tectum (owing to tectal growth) were retracted or degenerated in the following larval development. The above mode of axonal growth provides an adequate explanation of cellular mechanisms of terminal shifting of retinal axons within the tectum during development of retinotectal projection. Selection of appropriate branches may also lead to a reduction in the size of terminal arborization of retinal axons, resulting in a refinement in targeting.