A comparison of the normal and regenerated retinotectal pathways of goldfish

This is a light and electron microscopic study of the retinotectal pathway: intact and after regeneration of the optic nerve. The spatiotemporal pattern of axonal outgrowth and termination was studied with the methods of proline autoradiography, horseradish peroxidase (HRP) labeling, and fiber degeneration. The spatial order of optic fibers in the normal and regenerated pathways was assessed by labeling small groups intraretinally with HRP and then tracing them to the tectum. The labeled fibers occupied a greater fraction of the cross section of the regenerated than the normal optic tract. At the brachial bifurcation, roughly 20% of the regenerated fibers chose the incorrect brachium vs. less than 1% of the normals. In tectum, the regenerated optic fibers reestablished fascicles in stratum opticum, but they were less orderly than in the normals. The retinal origins of the fibers in the fascicles were established by labeling individual fascicles with HRP and then, following retrograde transport, finding labeled ganglion cells in whole-mounted retinas. Labeled cells were more widely scattered over the previously axotomized retinas than over the normal ones. A similar result was obtained when HRP was applied in the tectal synaptic layer. All of these results indicate that the pathway of the regenerated optic fibers is less well ordered than the intact pathway. Both autoradiography and HRP showed that the regenerating optic fibers invaded the tectum from the rostral end, and advanced from rostral to caudal and from peripheral to central tectum, along a front roughly perpendicular to the tectal fascicles. Synapses of retinal origin were noted electron microscopically in the tectum at the same sites where autoradiography indicated that the fibers had arrived. No retinal terminals were seen where grain densities were at background levels. Fiber ingrowth and synaptogenesis apparently occurred simultaneously. The synapses were initially smaller and sparser than in normals, but were in the normal tectal strata and contacted the same classes of post synaptic elements as in normals.     

Evidence for centripetally shifting terminals on the tectum of postmetamorphic Rana pipiens

In larval frogs the retina and tectum grow in topologically dissimilar patterns: new cells are added as peripheral annuli in the retina and as caudal crescents in the tectum. Retinotopy is maintained by the continual caudalward shifting of the terminals of the optic axons. After metamorphosis the pattern of growth changes. The retina continues to add new ganglion cells peripherally, but there is no neurogenesis in the tectum. To maintain retinotopy in postmetamorphic frogs, the terminals of the optic axons must continually shift toward the central tectum. We tested the proposal of centripetally shifting axons by making punctate injections of horseradish peroxidase (HRP) in the tectum of adult Rana pipiens and observing the patterns of filled cells in the contralateral retina, as was done in the goldfish (Easter and Stuermer, '84). Punctate applications of HRP in the tectum should be taken up: 1) by fascicles, and label a partial anulus of cells, 2) by terminals, and label a cluster of cells in the corresponding retinotopic site, and 3) by the extrafascicular axonal segments, and label a band of cells connecting the partial annulus to the cluster. If the terminals have shifted centripetally, the band of cells labeled through their extrafascicular segments should have a spoke-like orientation, with the center of the retina as the hub. As the tectal site moves from rostral to caudal, this band of cells should move, pendulum-like, from temporal to nasal retina. In general, the patterns of HRP-filled retinal cells we observed were consistent with our predictions. In addition, HRP taken up by the oldest (rostral) tectal axons produced more complex patterns of filled cells that indicated that these axons had shifted both caudally before metamorphosis and centripetally after.     

Evidence for the stability of positional markers in the goldfish tectum

Positional markers in the tectum, which are thought to guide growing axons to their target sites, have been proposed to be induced by axons, to be only transiently associated with the tectal cells, and then lost after long-term denervation periods (Schmidt: J. Comp. Neurol. 177:279-300, '78). To further investigate this concept, retinal axons were induced to regenerate into ipsilateral tecta which had been deprived of their retinal afferents for shorter (0-4 months) and longer periods (4-8 months). The paths of HRP-labeled regenerating axons of known retinal origin were traced and used as an operational test to decide whether the axons might navigate under the influence of positional markers. Two different kinds of experiments were performed: 1. The axons from a subpopulation of all ganglion cells in the retina were labeled by applying a small crystal of HRP at defined retinal regions. Independent of the denervation period of the tectum, the labeled regenerating axons traveled in abnormal but nonrandom routes. In early regeneration stages, axons exhibited signs of exploratory growth. They extended branches equipped with growth cones and filopodia into various regions of the tectum. In late regeneration stages, the axons lost these branches, exhibited U-turns and bends, and ended in terminal arbors in the retinotopic target region. These findings suggest that the axons travel under the influence of tectal positional markers and that these markers are not transient. 2. Axons from a surgically created temporal hemiretina were labeled by application of HRP to the optic nerve to test whether the temporal axons might expand into the caudal tectum in long-term-denervated tecta. The HRP-labeled axons coursed over rostral and midtectal regions. Instead of invading the caudal tectum they bent and terminated in the rostral tectal half. These results add further support for the conclusion that the path of regenerating retinal axons is governed by long-lasting positional markers.     

Development of the retinotectal system in normal quail embryos: cytoarchitectonic development and optic fiber innervation

The development of the optic tectum and the establishment of retinotectal projections were investigated in the quail embryo from day E2 to hatching day (E16) with Cresyl violet-thionine, silver staining and anterograde axonal tracing methods. Both tectal cytodifferentiation and retinotectal innervation occur according to a rostroventral-caudodorsal gradient. Radial migration of postmitotic neurons starts on day E4. At E14, the tectum is fully laminated. Optic fibers reach the tectum on day E5 and cover its surface on day E10. 'Golgi-like' staining of optic fibers with HRP injected in vitro on the surface of the tectum reveals that: growing fronts are formed exclusively by axons extending over the tectal surface; fibers penetrating the outer tectal layers are always observed behind the growing fronts; the penetrating fibers are either the tip of the optic axons or collateral branches; as they penetrate the tectum, optic fibers give off branches which may extend for long distances within their terminal domains; the optic fiber terminal arbors acquire their mature morphology by day E14. The temporal sequence of retinotectal development in the quail was compared to that already established for the chick, thus providing a basis for further investigation of the development of the retinotectal system in chimeric avian embryos obtained after xenoplastic transplantation of quail tectal primordia into the chick neural tube.     

Localization of optic tectal input to the ventral dendrite of the goldfish Mauthner cell

Although visually evoked Mauthner cell (M-cell) startle responses occur in the goldfish, the afferent projections underlying these reactions have not been previously studied. We have recorded from the M-cell while stimulating the left optic nerve and/or right optic tectum and have traced projections of the optic nerve and restricted areas of the optic tectum using HRP histochemistry and autoradiography. Tectal stimulation elicits similar postsynaptic potentials (PSPs) in both M-cells. The responses recorded in the right (ipsilateral) cell were localized to its ventral dendrite. The existence of uncrossed tectal projections to the ventral dendrite was confirmed morphologically following application of horseradish peroxidase (HRP) to the optic tectum. The PSPs contained both inhibitory and excitatory components, but with adequate stimulus strength, excitation of either M-cell dominated. Thus, this pathway is probably sufficient to trigger visually evoked startle responses mediated by the M-cell. Stimulation of the left optic nerve also evoked PSPs capable of bringing both M-cells to threshold. The blockage of this response by conditioning stimulation of the right tectum suggests that the visual information is relayed to the M-cells through this structure. In support of these findings, no label was found near any portion of the M-cell after either intraocular injection of tritiated proline or application of HRP to the cut end of the optic nerve. In summary, visual input to the M-cell is mediated via projections from the tectum, is segregated onto the ventral dendrite, and is capable of bringing this neuron to threshold. This pathway presumably accounts for the demonstrated behavioral efficacy of visual stimuli in evoking a startle response.     

Relationship between isthmotectal fibers and other tectopetal systems in the leopard frog

We studied the relationship of isthmotectal input to other tectal afferent fiber systems in three ways. 1) Using horseradish peroxidase (HRP) histochemistry, we determined the nonretinal inputs to the superficial tectum. In different sets of animals we a) applied HRP to the tectal surface; b) inserted HRP crystals into the tectum; c) injected small volumes of HRP solutions into the superficial tectum. N. isthmi accounts for more than 65% of the nonretinal extrinsic input in the superficial tectal layers. One set of fibers from the contralateral n. isthmi projects to the most superficial layer. Fibers from posterior thalamus and tegmentum project to both superficial and deeper layers in the tectum, but not to the most superficial layer. The ipsilaterally projecting isthmotectal fibers terminate in the deeper superficial layers. 2) We investigated the relationship between retinofugal and contralaterally projecting isthmotectal pathways. We orthogradely labelled n. isthmi fibers by unilateral HRP injections into n. isthmi, and we also labelled retinal fibers by injecting tritiated l-proline into both eyes. In such animals contralaterally projecting isthmotectal fibers cross in the dorsal posterior region of the optic chiasm. From the chiasm to the tectum isthmotectal fibers and retinofugal fibers are admixed. 3) We determined whether other fiber systems cross with contralaterally projecting isthmotectal fibers. We cut the posterior part of the optic chiasm and applied HRP crystals to the cut. Only n. isthmi and retina are retrogradely labelled.     

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.     

Retinal projection to a rotated tectal reimplant following long-term tectal denervation in adult goldfish

Following optic nerve crush, the precise termination sites of regenerating goldfish optic axons may be influenced by the presence or abscence of degenerating axonal debris from the previous projection. We investigated whether tectal polarity reversal can be induced in the absence of axonal debris The right optic tectum was denervated by contralateral eye removal. One year later, when no debris was present, a piece of the right tectum was rotated and innervation by the right eye was induced by removal of the left tectum. The new ipsilateral projection to the rotated region was correspondingly rotated. It is concluded that retention of tectal polarity is not dependent upon degenerating axonal debris.     

Peripheral nerve grafts to the frog optic tectum: a morphological study of foreign axon regeneration in the central nervous system

The proximal stump of a transected mandibular nerve was grafted onto the rostrodorsal surface of the optic tectum in adult Rana pipiens to investigate the morphologic characteristics of nonspecific axonal regeneration in a highly organized region of central nervous system (CNS). Within the first 3 weeks postgraft surgery (WPS), the nerve-tectum interface became firmly established. Concomitant with this was an invasion of the host tectum by a small number of fine "pioneerlike" axons from the nerve. By 6 WPS there developed a concerted instreaming of a large number of peripheral fibers. Once within the CNS, the foreign axons distributed themselves throughout the rostrocaudal extent of the tectum, but primarily its dorsal aspect within superficial layers 8 and 9. Presence of intact optic fibers at the time of mandibular fiber invasion served somewhat to restrict the regenerating aberrant axons in their course through layer 9. This restriction could be avoided by removal of the optic input either before or during peripheral ingrowth. However, once peripheral fibers had entered and established themselves in the host environment, no subsequent manipulation of the retinotectal projection had any effect. The aberrant growth pattern, which appeared remarkably stable after 6 WPS, consisted of a plexus of medium- and fine-caliber peripheral axons. Many of these fibers had numerous branches and "en passant" varicosities, the latter encompassing a variety of shapes and sizes. Terminal swellings and arborizations were also found. When comparing the regeneration of optic and mandibular nerve fibers in the tectum, two distinctions were made. Whereas optic axons revealed a fascicular and layered organization, mandibular axons showed a highly segregated and disordered growth pattern. These characteristic differences were maintained even when the two fiber systems were allowed to coregenerate into the same target tectum. Thus, each of the two groups of axons interacts with the tectal substrate in a distinct manner, apparently independent of the other.     

Denervation of non-optic brain areas along the course of the optic tract does not affect the success of optic nerve regeneration in frogs

Unsuccessful axonal regeneration in frog or goldfish spinal cord is thought to be due in part to inappropriate denervated synaptic sites attracting regenerating axons as they pass by the lesion zone (Bernstein and Bernstein, '69; Bernstein and Gelderd, '73). The reported success of optic nerve regeneration in these same animals may result because there are no inappropriate synaptic sites available to the regenerating axons. To test for this we denervated areas within thalamus and mesencephalon which lie along the path of regenerating optic axons to determine whether or not abnormal connections would form in these areas and affect the success of optic nerve regeneration. After transection of the brainstem through the isthmal region, terminal degeneration was found in several zones adjacent to optic nerve targets or the optic tract (i.e., nucleus rotundus, corpus geniculatum laterale, and torus semicircularis). In adult frogs receiving left isthmal transection, we also crushed the right optic nerve and then examined the regenerated optic projection at intervals up to six months after nerve crush using anterograde transport of ³H-proline. In no instance did the regenerating optic axons alter their distribution within visual neuropil zones or invade those areas deafferented by the isthmal lesion. Histological study showed that the axons cut by the isthmal lesion did not regenerate back to their sites to prevent the invasion of optic axons into these zones. We then attempted to force optic axons into foreign territory by removing a major projection zone, the optic tectum. With tectal ablation and isthmal transection, regenerating optic axons were offered synaptic space made available by both lesions. However, we found abnormal optic projections only in the middle and posterior thalamic neuropil and in the remaining tectal hemisphere. Optic axons did not expand into any of the areas deafferented by the isthmal transection, even though some of them were further denervated by the tectal ablation. We conclude that optic axons will not invade non-optic areas deafferented by an isthmal lesion even if a large number of optic axons have no normal target to innervate.     

Intraocular colchicine inhibits competition between resident and foreign optic axons for functional connections in the doubly innervated goldfish optic tectum

After unilateral optic tectum ablation in the goldfish, regenerating optic axons grow into the optic layers of the remaining ipsilateral tectal lobe and regain visual function. The terminal arbors of the foreign fibers are initially diffusely distributed among the resident optic axons, but within two months the axon terminals from each retina are seen to segregate into irregular ocular dominance patches. Visual recovery is delayed until after segregation. This suggests that the foreign fibers compete with the residents for tectal targets and that the segregation of axon terminations is an anatomical characteristic of the process. Here we investigate whether inhibiting axonal transport in the resident fibers inhibits competition with foreign fibers. The eye contralateral to the intact tectal lobe received a single injection of 0.1 μg colchicine, which does not block vision with the intact eye. We measured visual function using a classical conditioning technique. Segregation of axon terminations was examined shortly following visual recovery by autoradiography. The no-drug control fish showed reappearance of vision with the experimental eye at 9 weeks postoperatively and ocular dominance patches were well developed. Colchicine administered to the intact eye (resident fibers) several weeks postsurgery decreased the time to reappearance of vision with the experimental eye by several weeks. Autoradiography revealed some signs of axonal segregation but the labeled foreign axons were mainly continuously distributed. Administration of colchicine at the time of tectum ablation, or of lumicolchicine at two weeks postoperatively produced normal visual recovery times. Fast axonal transport of³H-labeled protein was inhibited by 1.0 and 0.5 μg but not by 0.1 μg of colchicine or by 1.0 μg of lumicolchicine. Previous studies showed that while 0.1 μg of colchicine does not block vision it is sufficient to inhibit axonal regeneration following optic nerve crush. We conclude that two retinas can functionally innervate one tectum without forming conspicuous ocular dominance columns, and that the ability of residents to compete with the in-growing foreign axons is very sensitive to inhibition of axoplasmic transport or other processes that are inhibited by intraocular colchicine.     

Axonal pathfinding during the regeneration of the goldfish optic pathway

Retinal ganglion cells in fish and amphibians regenerate their axons after transection of the optic nerve. Fiber tracing studies during the third month of regeneration show that the axons have reestablished a basically normal fiber order in the two brachia of the optic tract; axons originating in the ventral hemiretina are concentrated in the dorsal brachium, axons from the dorsal hemiretina in the ventral brachium. Attardi and Sperry (Exp. Neurol. 7:46-64, 1963) first suggested that the reestablishment of the fiber order reflects path-finding by the regenerating axons. Recently, however, Becker and Cook (Development 101:323-337, 1987) have claimed that the fiber order observed at later stages of regeneration is due to secondary axonal rearrangements and that the initial brachial choice is random. In order to evaluate whether regenerating axons are capable of navigating in the optic tract and brachia and on the tectum, the present study examined the pathway choices and the morphology of regenerating axons en route to their tectal targets in goldfish. Subsets of axons were labeled at various time intervals (2 to 30 days) following an optic nerve crush, by intraretinal application of the lipophilic fluorescent tracer 1,1-dioctadecyl-3-3-3'-3'-tetramethylcarbocyanine (DiI). After a survival time of 18 to 72 hours (to allow for diffusion of DiI along the axons), the experimental animals were perfused with fixative and their right and left optic pathways (nerve, tract, and tectum) were dissected free and separated at the chiasm. Fluorescently labeled axons were traced in whole-mounted pathways. Pathway choices were examined at the brachial bifurcation where axons from ventral and dorsal hemiretinae normally segregate. DiI was found to label axons reliably up to their growth cones, even at the earliest stages of regrowth. The pathway choices of the axons were nonrandom. The majority of the ventral axons reached the appropriate, dorsal hemitectum through the appropriate dorsal brachium of the tract. Dorsal axons reached the ventral hemitectum mainly through the ventral brachium. This suggests the presence of specific guidance cues, accessible to the regenerating axons. Differences in the complexity of the growth cones of the regenerating axons (simple in the nerve and tectal fiber layer, complex in the tract and the synaptic layer of the tectum) provide further evidence for specific interactions between the regenerating axons and their substrates along the pathway. These results argue that regenerating retinal axons in fish are capable of axonal path-finding.     

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.     

Anomalous retrograde labeling of brain cells from the eye in the goldfish: evidence for long distance growth of sprouted neurites

We have used retrograde labeling with horseradish peroxidase (HRP) and a wheat germ agglutinin conjugate of HRP (WGA:HRP) to investigate the projections of the nucleus postglomerulosus (nPg) both in normal goldfish and in animals which had undergone retinal removal. In normal animals, our evidence indicates that nPg projects only to the optic tectum. Using small HRP and WGA:HRP application sites in the tectum, we have shown that nPg cells have broadly spread terminals in the tectal neuropil and that there is no obvious correspondence between the rostrocaudal axis of the nPg and the deployment of the terminal arbors of its cells along the rostrocaudal axis of the tectum. In addition, we found no evidence for an nPg projection to the eye in normal animals. After retinal removal we found that nPg cells were more readily backfilled from small tectal applications of HRP. However, our most interesting observation was that at 4-6 weeks and more after ocular surgery, we could retrogradely label the cells of the nPg with intraocular or retroocular injections of WGA:HRP. At the same postoperative times, we were also able to label neurites in the atrophied optic nerve by microinjecting WGA:HRP into the contralateral midbrain tegmentum. Finally, we found that the cells of the nPg undergo a hypertrophic response, similar to that seen in other neurons after axotomy, following retinal removal or section of the dorsomedial brachium of the optic tract. Thus, these cells respond to retinal denervation of the tectum with a response characteristic of axotomized cells although their axons have not been cut. Similar changes were also seen in the nucleus isthmi on both sides of the brain following retinal removal. We interpret our data to indicate that cells of the nPg can respond to optic (and thus heterotypic) denervation of their terminal field by sprouting processes which grow away from the terminal field, through denervated optic pathways, to the retinaless eye. This interpretation requires that the sprouted processes grow for several millimeters.     

Target regulation of protein biosynthesis in retinal ganglion cells during regeneration of the goldfish visual system

Transected ganglion cells of the goldfish retina demonstrate a marked increase in protein biosynthesis as their axons regrow into a primary target tissue, the optic tectum. In order to examine what role the target tissue may play in regulating the pattern of neuronal protein biosynthesis, the tectum is removed at the time of axotomy. By 10 days after surgery, the production of specific polypeptides including the structural proteins, tubulin and actin, are not affected by tectal ablation. However, during a later phase of regeneration when axons would normally begin to re-connect with the tectum, the appearance of a radiolabeled polypeptide of 300 kDaltons is blocked by tectal ablation. These data suggest that ganglion cells are regulated by target tissues late in the regenerative process perhaps by contact between axons and cells of the tectum.     

Frog tectal efferent axons fail to regenerate within the CNS but grow within peripheral nerve implants

Tectal efferent axons, located adjacent to the optic tract, fail to regenerate past diencephalic lesions in Rana pipiens even though optic axons regenerate after the same injury (M. J. Lyon and D. J. Stelzner, J. Comp. Neurol. 255: 511-525). We tested the possibility that tectal efferent axons can regenerate within peripheral nerve implants. A 6- to 8-mm segment of autologous sciatic nerve was implanted into the anterolateral (N = 23) or centrolateral (N = 22) portion of the dorsal surface of the tectum. Frogs survived for 6 (N = 16) or 12 weeks (N = 29) before the free end of the nerve was recut and HRP applied. A control group had the nerve crushed prior to the HRP application. Neurons within the tectum, near and medial to the implant site, were retrogradely labeled from the nerve graft in most experimental operates but no neurons were labeled in controls. In addition, neurons were also labeled in nuclei which projected to the tectum in a number of cases. Three times as many neurons were labeled in 12-week operates (42 +/- 46) as in 6-week operates (15 +/- 12). The morphology and location of labeled neurons in the tectum was similar to tectal efferent neurons except that the somal area of neurons labeled from the graft was significantly larger (41%) than normal tectal efferent neurons. The basic finding is similar to experiments using the same paradigm in the mammalian central nervous system (CNS). One difference is the minimal glial reaction at the graft insertion site.(ABSTRACT TRUNCATED AT 250 WORDS)     

Rules of order in the retinotectal fascicles of goldfish

Individual fascicles of retinal axons were labeled in the goldfish tectum with horseradish peroxidase (HRP). The contralateral retina was later processed for HRP histochemistry to mark the cells that had axons in the fascicles. Labeled cells were found in a partial half anulus in ventral hemiretina, centered on the optic disk. The distance of the partial anulus from the disk depended on which tectal fascicle had been labeled; the more rostrocentral the fascicle, the smaller was the annular radius. The angular subtense of the partial anulus with respect to the disk depended on where (along its tectal course) the fascicle had been labeled; the more rostral the label site, the longer was the angular subtense. These results were interpreted in the context of retinotectal growth, and it was inferred that the axons followed two rules: (1) grow in along the edge of the tectum and (2) exit and terminate in order, axons from temporal retina first, nasal retina last. These rules would produce a retinotopic projection in peripheral tectum, but they require that some of the terminals already in place must shift as the tectum grows.     

Trajectories of regenerating retinal axons in the goldfish tectum: I. A comparison of normal and regenerated axons at late regeneration stages

To visualize and compare the intratectal path of normal and regenerated retinal axons, HRP was applied to localized sites in the dorsotemporal and dorsonasal retina in normal goldfish and in goldfish at 3-12 months after optic nerve section. The anterogradely labeled axons were traced in tectal whole mounts. In normal animals the axons were confined to the appropriate ventral hemitectum. Therein they ran in very orderly routes (Stuermer and Easter: J. Neurosci. 4:1045-1051, '84) and terminated in regions retinotopic to the labeled ganglion cells in the retina. The terminal arbors of dorsotemporal axons resided in the ventrorostral tectum and those of dorsonasal axons in the ventrocaudal tectum. In regenerating animals the terminal arbors also resided at retinotopic regions, where they sometimes formed two separate clusters. In contrast to normal axons, the regenerating ones traveled in abnormal routes through the appropriate and inappropriate hemitectum. From various ectopic positions, they underwent course corrections to redirect their routes toward the retinotopic target region. In their approach toward their target sites, dorsotemporal and dorsonasal axons behaved differently in that the vast majority of dorsotemporal axons coursed over the more rostral tectum whereas dorsonasal axons progressed into the caudal tectal half. This differential behavior of regenerating dorsonasal and dorsotemporal axons was substantiated by a quantitative evaluation of axon numbers and orientations.     

Trajectories of regenerating retinal axons in the goldfish tectum: II. Exploratory branches and growth cones on axons at early regeneration stages

HRP was applied to small sites in the dorsotemporal or dorsonasal retina in fish at 10-36 days after optic nerve section. The anterogradely labeled axons were visualized in tectal whole mounts. Axons traveled through all regions of the tectum in various abnormal routes. Misrouted axons were also seen to alter their orientation and to direct their course toward their target. At all regeneration stages the majority of dorsotemporal axons coursed and achieved target-related orientations preferentially within the rostral tectal half whereas dorsonasal axons proceeded into the caudal tectum. The growing axons exhibited various morphologies. All axons in the superficial fascicle layer stratum opticum (SO) and some in the synaptic layer stratum fibrosum et griseum superficiale (SFGS) were unbranched and tipped with a leading growth cone. Other axons in the synaptic layer carried one to several growth cones at their ends and often filopodia proximal to the growth cone, or they had sprouted numerous side branches with growth cones and filopodia on the shaft and on branches. Some axons at retinotopic or ectopic sites gave rise to several long branches of several hundred microns in length, with growth cones and filopodia. From 32 days onward axons ending in terminal arbors at retinotopic sites became apparent. Thus, numerous axons at early regeneration stages go through a phase of exploratory growth on their way toward their target sites.     

Studies of the early stages of optic axon regeneration in the goldfish

We have studied the early stages (4-14 days) of axonal regeneration following intraorbital optic nerve crush in the goldfish. We used 3H-proline autoradiography to anterogradely label and visualize the growing axons and wheat germ agglutinin-conjugated horseradish peroxidase (WGA:HRP) for retrograde labeling to determine the cells of origin of the earliest projections. The first retinal ganglion cells (RGCs) that could be retrogradely filled from the optic tract, following optic nerve crush, were in the central retina and were seen at 8 days postoperative. More peripheral cells were only labeled with longer postcrush survival periods. Thus, the first axons to regenerate past the lesion were from central RGCs. The axons of these cells extended into the cranial nerve stump between 4 and 5 days postcrush and entered the nerve as a fascicle, which travelled just beneath its surface. Studies of nerve cross sections from animals at 5-8 days postoperative demonstrated that initial outgrowth was not confined to any particular locale within the nerve although the early fibers appeared to avoid its temporal aspect. When the regenerating axons reached the optic tract they remained in fascicles but left the surface to run along the medial, deep portion of the tract, immediately adjacent to the diencephalon and pretectum. The positions occupied by the earliest-regenerating axons in the optic nerve were variable and not always appropriate for their central retinal origin. However, the abrupt change in growth trajectory as the fibers entered the optic tract brought them into the areas of the visual paths that are occupied by central axons in intact animals. We suggest that this change in position is related to both changes in the structural organization of the intracranial visual paths and to possible axon guidance signals in the region of the nerve-tract juncture.