The developing optic nerve and tract have received considerable attention in recent years, but the cellular and subcellular microenvironment of the growing axons has not been described. In the belief that such a description is essential (though certainly not sufficient) for an understanding of pathway formation, we have examined the normal development of the retinofugal projection of Xenopus laevis.

Optic fibers were labeled anterogradely at the retina with horseradish peroxidase (HRP) or the carbocyanine dye, Dil, at stages 32 to postmetamorphosis. The brains were examined both as whole mounts and in sections, light- and electron-microscopically, with the emphasis on tracts associated with the route of the optic fibers.

At stage 32, two ventral commissures were present, the anterior and postoptic. They were immediately subjacent to the pia. All tracts and even isolated axons were in similarly superficial locations. The first deep pathway (separated from the pia by cell nuclei) was seen at stage 46; it was a dorsal commissure, probably the posterior.

The first retinal axons passed from the optic stalk into the ventral part of the diencephalon, where they coursed along the rostral edge of the postoptic commissure, and maintained this position, relative to the other fibers in the tract of the commissure, throughout the remainder of their contralateral trajectory. They reached the presumptive thalamic and tectal termination sites and arborized. Subsequent optic axons followed this same route, thus enlarging the optic pathway relative to the more slowly growing nonoptic part of the commissure and its tract. Electron microscopy revealed, as early as stage 35, specialized contacts between cellular processes in the neuropil. These contacts had the form of symmetric membranous thickenings; some were associated with vesicles and were presumed to be synapses.

We conclude that the early forebrain and midbrain have only two ventral commissural pathways, and most axons that grow out after these pathways have formed add to them rather than establish new tracts. The optic axons travel a stereotyped pathway alongside a preexisting tract associated with the postoptic commissure. The possibility that optic fiber outgrowth is normally influenced by pre-existing tracts is discussed in relation to recent experimental investigations of fiber growth from ectopic eyes.

The visual system of the frog Xenopus laevis has been used in many studies of nerve connectivity. The first studies dealt with regeneration, but as techniques have improved, the early development of the retinotectal projection has been emphasized (Holt and Harris, 1983; Holt, 1984; Sakaguchi and Murphey, 1985; O’Rourke and Fraser, 1986; Harris, 1986; Harris et al. 1987.) The path taken by growing retinal ganglion cell axons and the time course of this advance have been described in both normal and in experimentally perturbed brains. Even so, little is known about the control mechanisms that guide the developing optic axons. Several studies have taken advantage of the ease of embryonic manipulation of the developing frog nervous system to transplant eye primordia to novel positions (Sharma, 1972; Constantine-Paton and Capranica, 1975; Constantine-Paton, 1978; Giorgi and Van Der Loos, 1978; Katz and Lasek, 1978; 1979; Harris, 1986), and from such studies various ideas concerning retinal axon guidance have been proposed. These include long-range attraction by the target (Giorgi and Van Der Loos, 1978; Harris, 1986) and guidance by local cues (Constantine-Paton and Capranica, 1975; Harris, 1986), including those that define a ‘substrate pathway’ (Constantine-Paton, 1978; Katz and Lasek, 1978; 1979; Katz et al. 1980.) More recently, Harris (1989) has altered the part of the neuroepithelium through which the axons grow, and produced more evidence for the role of local cues.

In an attempt to provide more specific information about the guidance of axons in the visual system, we have examined the normal development of the Xenopus retinotectal projection, using both whole-mounted and sectioned brains. The whole mounts are particularly useful because they reveal, in high contrast, the labeled optic fibers on a featureless background. But this is also a weakness of the whole mounts; they reveal nothing of the labeled fibers’ microenvironment. By embedding and sectioning the brains, and examining them with light and electron microscopy, we can visualize this microenvironment, which undoubtedly plays a role in axonal guidance.

We find that the early amphibian brain is structurally quite simple, as the early neuroembryologists had written (e.g. Herrick, 1938). The archencephalon (the presumptive fore- and midbrains: Jarvik, 1980) in the prefeeding stage of life has only very few pathways: two ventral commissures, two dorsal ones, and several closely associated longitudinal tracts on each side. Our studies have confirmed these observations and improved on them. Using better methods of fixation and visualization, we have achieved greater resolution than was possible several decades ago. In addition, we have examined earlier stages of developing pathways than the classical neuroembryologists could do, because we did not depend on reduced silver stains, which label neurofilaments, expressed relatively late during axonal maturation.

We find that the optic axons add on, in a stereotyped way, to one of the pre-existing ventral commissures and the tract associated with it, and accompany that pathway toward the presumptive tectum. We find also that all the early pathways are subjacent to the pia; indeed, electron microscopic examination of thin sections of these tiny brains reveals no axons among the cell bodies. The significance of our observations are discussed as they relate to axonal guidance.

A preliminary account of this work has appeared in abstract form (Easter and Taylor, 1988.)

Embryonic Xenopus laevis were obtained from induced matings of adults. Viable eggs were sorted and reared in dilute saline at either 16 or 23 °C, which provided embryos of a range of developmental stages over a period of several days. All embryos and larvae were staged according to the tables of Nieuwkoop and Faber (1967). We noted no differences between embryos of the same stage reared at different temperatures.

For labeling the retinotectal projection, embryos were anesthetized in 1:4500 MS222 (ethyl-m-aminobenzoate, Sandoz) in 66% Niu-Twitty saline, and placed on their sides on damp tissue. Fine tungsten needles were used to cut the sclera and then to remove the lens. Either a fewnl of a 10% HRP solution (Boehringer Type II) in 1 % Nonidet (BDH) was injected into the vitreous, or a tiny crystal of the same solution, which had been allowed to dry, was placed in contact with the retina. Following a 15 minute survival in 50% Niu-Twitty saline, the embryos were reanesthetized and then fixed in 2% glutaraldehyde in 0.1 M-phosphate buffer, pH7.4, where they remained for between 1 and 24h. After stage 46, the retinotectal projection was labeled either by injection of Nonidet with HRP directly into the vitreous, or by application of recrystallized HRP without Nonidet to the optic nerve. In older animals, fixation was by perfusion with buffered 0.25 M-sucrose, followed by buffered 2% glutaraldehyde.

After washing in buffer, the brains were dissected free and reacted as whole mounts using a modified cobalt/ diaminobenzidene procedure (Taylor and Gaze, 1985). Uncleared larval, or cleared tadpole and postmetamorphic brains were examined using a compound microscope and either drawn with the aid of a camera lucida, or photographed, or both.

Selected preparations were embedded in Araldite, in most cases after postfixation in 1 % osmium tetroxide. Brains were reconstructed three-dimensionally from serial 1 or 2 μm horizontal sections, which were obtained as follows. The block was trimmed to include the forebrain, midbrain, and part of the hindbrain. All of the sections collected in a 20 μm advance were placed in a drop of water on the first well of a multiwell slide (Hendley), a further set were then cut and placed in the next well, and so on. A full slide of sections was then dried, stained with toluidine blue, and coverslipped. A single section from each well (or every other well) was drawn in a camera lucida to give a sample section from each 20 μm (or 40 μm) through the brain. These drawings were aligned by eye, usually using the photograph of the whole mount as a guide, entered into a bit pad, and serially reconstructed using SSRCON, a program written by J. Green, of the National Institute of Medical Research, Mill Hill, London, and run on a DEC PDP 11/23 computer. A total of eleven brains were reconstructed, three each of stages 32, 35, and 39, and two of stage 46.

For electron microscopy, brains were either prepared as described above, or, since the HRP procedure tended to give bad preservation, fixed in half strength Karnovsky fixative, postfixed in osmium tetroxide, and then embedded in Araldite.

For labeling with Dil (l,l′-dioctadecyl-3,3,3′,3′-tetrameth-ylindocarbocyanine perchlorate: Honig and Hume, 1986), the eye was prepared as described for HRP labeling and a small crystal of Dil (Molecular Probes) was placed against the retina. After 5 min, the embryo was fixed in 10% buffered neutral formalin and placed in the dark at 4°C for a minimum of 5 days. The brains were dissected free, washed in buffer, and observed using fluorescence optics.

Observations of whole-mounted brains: growth of the optic fibers

During the period of outgrowth by the first optic axons, the brain is relatively undifferentiated, but certain morphological features serve as useful landmarks. These are illustrated in Fig. 1, drawings of wholemounted brains of an embryo and a late tadpole. (See Fig. 2D and K for photographs.) The embryonic brain of Xenopus, like that of other vertebrates (see Bergquist, 1952) is shaped roughly like a pistol, with the spinal cord and hindbrain corresponding to the barrel, and the mid- and forebrains to the handle. The rostral part of the handle will develop into the olfactory bulb; the caudal part develops into the hypothalamus. The presumptive midbrain is the wedge-shaped part, broad anterodorsally, narrow ventrocaudally. The epiphysis (part of the forebrain) is the earliest epaxial structure to develop, and thus provides a useful landmark for the boundary between the presumptive fore- and midbrain. A dorsal constriction, the isthmus, marks the boundary between the presumptive hind- and midbrain. The optic stalks attach to the brain near the ventral surface.

Fig. 1.

Diagrammatic lateral views showing the general morphological features of the brains of a stage 39 embryo (A) and a postmetamorphic juvenile (B). Rostral to the right, dorsal up. Fig. 2D and 2K are photographs of these same ages. Hatched regions give the termination sites of the retinal projections, a small thalamic one rostrally, and a larger tectal one caudally. The lines show the optic axons. The dark line in the juvenile’s brain shows the accessory optic pathway to the basal optic nucleus. Abbreviations: BON, basal optic nucleus; C, cerebellum; E, epiphysis; Fb, forebrain; Hb, hindbrain; OT, tectum; Th, thalamic visual center. Scale bar: A=200μm; B = 1mm.

Fig. 1.

Diagrammatic lateral views showing the general morphological features of the brains of a stage 39 embryo (A) and a postmetamorphic juvenile (B). Rostral to the right, dorsal up. Fig. 2D and 2K are photographs of these same ages. Hatched regions give the termination sites of the retinal projections, a small thalamic one rostrally, and a larger tectal one caudally. The lines show the optic axons. The dark line in the juvenile’s brain shows the accessory optic pathway to the basal optic nucleus. Abbreviations: BON, basal optic nucleus; C, cerebellum; E, epiphysis; Fb, forebrain; Hb, hindbrain; OT, tectum; Th, thalamic visual center. Scale bar: A=200μm; B = 1mm.

Fig. 2.

Photographs of whole-mounted brains, from embryos of various stages, with the optic axons from the left eye labeled with HRP. Rostral is to the right. (A) Stage 34, ventral view. The labeled axons (arrowheads) emerge from the left optic stalk and the leading growth cones extend just beyond the midline (dashed line). (B) Stage 35, lateral view. The labeled axons have begun the ascent up the lateral diencephalic wall. (C) Stage 36/37, lateral view. The labeled axons have changed course from dorsad to dorsocaudad, and the leading growth cones (arrowheads) approach the tectal precursor. A single axon, tipped by a growth cone (*) veers away from the main body of the tract. (D) Stage 37/38, lateral view. Many more axons have now joined the tract, which extends toward the tectal precursor. The labeled axons form a coherent group as far as the tectum, where they splay out over a wider area. (E) Stage 39, dorsolateral view. The labeled axons have now reached the position of the developing tectum and have begun to arborize over its lateral surface. (F) Stage 42, lateral view. The tectal arbors have become more dense and expanded over the tectum. The optic tract is larger. At the midlevel of the diencephalon, a distinct region of arborization is visible (Th). The forebrain has begun to expand rostrally. (G and H) Stage 49, lateral and dorsal views of the same brain. The brain now has the more adult shape; the forebrain is as large as the tectum. The retinal projections to the basal optic nucleus and the thalamus are evident in G. (I and J) Stage 52, lateral and dorsal views of the same brain. It is noticeably larger than at stage 49. In I, the retinal projections to the pretectal neuropil, the basal optic nucleus, and the thalamus are more pronounced. In J, the extent of the tectal innervation is seen to have increased relative to H. (K and L) Postmetamorphic juvenile, lateral and dorsal views of the same brain. The forebrain has enlarged in both the rostrocaudal and mediolateral directions, resulting in a partial concealment of the midbrain and optic tracts. The retinal projection has enlarged considerably, with the optic tract covering much of the diencephalon, and the distribution of the terminal arbors covering all but the far caudal part of the tectum. In dorsal view, the ipsilateral projection can be seen (arrowhead). Abbreviations: BON, basal optic nucleus; C, cerebellum; E, epiphysis; Fb, forebrain; Hb, hindbrain; OT, optic tectum; Th, thalamic visual center; PT, pretectal neuropil. Scale bars: A,C,E=50μm; B,D,F=100 μm; G-J=400 μm; K,L=800 μm.

Fig. 2.

Photographs of whole-mounted brains, from embryos of various stages, with the optic axons from the left eye labeled with HRP. Rostral is to the right. (A) Stage 34, ventral view. The labeled axons (arrowheads) emerge from the left optic stalk and the leading growth cones extend just beyond the midline (dashed line). (B) Stage 35, lateral view. The labeled axons have begun the ascent up the lateral diencephalic wall. (C) Stage 36/37, lateral view. The labeled axons have changed course from dorsad to dorsocaudad, and the leading growth cones (arrowheads) approach the tectal precursor. A single axon, tipped by a growth cone (*) veers away from the main body of the tract. (D) Stage 37/38, lateral view. Many more axons have now joined the tract, which extends toward the tectal precursor. The labeled axons form a coherent group as far as the tectum, where they splay out over a wider area. (E) Stage 39, dorsolateral view. The labeled axons have now reached the position of the developing tectum and have begun to arborize over its lateral surface. (F) Stage 42, lateral view. The tectal arbors have become more dense and expanded over the tectum. The optic tract is larger. At the midlevel of the diencephalon, a distinct region of arborization is visible (Th). The forebrain has begun to expand rostrally. (G and H) Stage 49, lateral and dorsal views of the same brain. The brain now has the more adult shape; the forebrain is as large as the tectum. The retinal projections to the basal optic nucleus and the thalamus are evident in G. (I and J) Stage 52, lateral and dorsal views of the same brain. It is noticeably larger than at stage 49. In I, the retinal projections to the pretectal neuropil, the basal optic nucleus, and the thalamus are more pronounced. In J, the extent of the tectal innervation is seen to have increased relative to H. (K and L) Postmetamorphic juvenile, lateral and dorsal views of the same brain. The forebrain has enlarged in both the rostrocaudal and mediolateral directions, resulting in a partial concealment of the midbrain and optic tracts. The retinal projection has enlarged considerably, with the optic tract covering much of the diencephalon, and the distribution of the terminal arbors covering all but the far caudal part of the tectum. In dorsal view, the ipsilateral projection can be seen (arrowhead). Abbreviations: BON, basal optic nucleus; C, cerebellum; E, epiphysis; Fb, forebrain; Hb, hindbrain; OT, optic tectum; Th, thalamic visual center; PT, pretectal neuropil. Scale bars: A,C,E=50μm; B,D,F=100 μm; G-J=400 μm; K,L=800 μm.

Photographs of whole-mounted brains illustrating the development of the retinotectal projection are shown in Fig. 2. Note that the overall shape and size of the brain change very little between stages 34 (Fig. 2B) and 42 (Fig. 2F). From stage 42, the brain begins to straighten, and assumes the more linear form that is characteristic of the adult (Fig. 21, K, and M).

The whole-mounted brains labeled with HRP confirmed the time course of retinal axon outgrowth that had been reported previously (Harris, 1986; Holt, 1984). Labeled axons crossed the chiasm and formed the contralateral optic tract by stage 34/35 (Fig. 2A and B). By stage 37/38, the axons, frequently tipped by growth cones, had reached the presumptive tectum (Fig. 2C and D), and had begun to elaborate terminal arbors by stage 39/40 (Fig. 2E).

The formation of the optic tract also conformed to earlier descriptions (Sakaguchi and Murphey, 1985; Harris, 1986; Harris et al. 1987), except in those cases described below. A well-defined bundle of axons passed from the optic stalk into the ventral part of the brain and crossed to the contralateral side, where it passed immediately caudal to the other optic stalk. In a very few cases, labeled axons were found in the ipsilateral diencephalon, but these were always few in number (one or two axons or fine fascicles). On the contralateral side the retinal axons traveled together, with their growth cones directed dorsally towards the region of the presumptive tectum (Fig. 2C and D). The majority of axons clustered together, but isolated labeled axons were found both rostral and caudal to the cluster. The rostral boundary of the optic tract tended to be better defined than the caudal edge, from which some axons left the body of the tract and grew toward the spinal cord (Fig. 2E). Once established, the optic tract followed a course that passed dorsally to the position of the developing thalamic visual centers and then veered dorsocaudally to run directly towards the tectal primordium. From stage 42 on, as the brain straightened out, the optic tract coursed more caudally. The tract enlarged continuously, throughout development, as the continued growth of the retina (Straznicky and Gaze, 1971) produced new generations of axons. From the onset of metamorphosis (stage 54) an ipsilateral projection developed to the diencephalic visual centers (Fig. 2M and N).

At the position where, at later stages, we find the retinothalamic terminations, some cells were fluorescently labeled in preparations in which Dil had been applied to the retina (Fig. 3). We have never observed, nor has anyone else reported, HRP-labeled cells in this region following retinal application of HRP, at any stage. Therefore it seems unlikely that these fluorescent cell bodies were labeled retrogradely, and we suggest that they were labeled transneuronally following anterograde transport of the dye (Godement et al. 1987). Since Dil is believed to pass along cell membranes, this transneuronal labeling suggests that the retinal axons must have made close contacts with these cells.

Fig. 3.

Fluorescently labeled cells (arrowheads) in the region of the developing thalamic visual center. Dil has diffused from the retinal axons (open arrowheads). Scale bar=50 μm.

Fig. 3.

Fluorescently labeled cells (arrowheads) in the region of the developing thalamic visual center. Dil has diffused from the retinal axons (open arrowheads). Scale bar=50 μm.

In a few cases the optic axons followed abnormal routes. Fig. 4A shows an example in which they failed to form a defined tract, but splayed out across the contralateral diencephalon. The axons in this projection appeared to be heading dorsally, but many with large growth cones at their tips pointed rostrodorsally rather than caudodorsally. In a second case, labeled optic axons were misrouted where they left the optic stalk (Fig. 4B). A small group of axons headed rostrally away from the optic tract into the region of the developing telencephalon. Some of these axons bent sharply and re-entered the optic pathway at the chiasm, while others passed across the midline and continued towards the contralateral side of the brain.

Fig. 4.

Aberrant retinal axons in the brains of two stage 35 embryos. Camera lucida drawings. Rostral to the right. (A) Lateral view. The labeled axons crossed the midline together, but the aberrant group separated from the main bundle in the ventral diencephalon to deploy over the forebrain. (B) Ventral view. In this case, the aberrant axons separated from the main bundle early, and entered a more anterior pathway (probably the anterior commissure) in which they decussated. The loop of labeled axons near the base of the optic stalk suggests that some axons left the main route but then regained it. Abbreviations: E, epiphysis; Fb, forebrain; OS, optic stalk. Scale bar=100 μm.

Fig. 4.

Aberrant retinal axons in the brains of two stage 35 embryos. Camera lucida drawings. Rostral to the right. (A) Lateral view. The labeled axons crossed the midline together, but the aberrant group separated from the main bundle in the ventral diencephalon to deploy over the forebrain. (B) Ventral view. In this case, the aberrant axons separated from the main bundle early, and entered a more anterior pathway (probably the anterior commissure) in which they decussated. The loop of labeled axons near the base of the optic stalk suggests that some axons left the main route but then regained it. Abbreviations: E, epiphysis; Fb, forebrain; OS, optic stalk. Scale bar=100 μm.

Observations of sections: the microenvironment of the axons

At stage 32, nearly the entire cross section of the brain was occupied by a uniform population of neuroepithelial cells, with nuclei packed tightly together at all levels. Fig. 5 shows horizontal semithin sections at three levels in the same brain. The low magnification views (Fig. 5A, C, and E) show the brain and its surrounding structures. The more highly magnified details (Fig. 5B, D, and F) illustrate small regions, immediately deep to the pia, from which the nuclei are excluded. At the light-microscopic level, the texture of these nucleus-free zones was quite granular, suggestive of a neuropil or a tract cut transversely. (This interpretation is confirmed, electron microscopically, below.) The nucleus-free region was quite substantial in the medulla and spinal cord, and its texture suggested a longitudinally cut tract.

Fig. 5.

Horizontal semithin sections from a brain of a stage 32 embryo. Rostral is up. A, C, and E are low magnification views at progressively more dorsal levels (see Fig. 6). B, D, and F are more highly magnified views of the regions indicated by the boxes in A, C, and E, respectively. (A and B) The tract of the anterior commissure. (C-F) The tract of the postoptic commissure. Abbreviations: OS, optic stalk; P, presumptive pia; R, retina; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure; V, ventricle; YG, yolk granule. Scale bars: A,C,E=100 μm; B,D,F=20 μm.

Fig. 5.

Horizontal semithin sections from a brain of a stage 32 embryo. Rostral is up. A, C, and E are low magnification views at progressively more dorsal levels (see Fig. 6). B, D, and F are more highly magnified views of the regions indicated by the boxes in A, C, and E, respectively. (A and B) The tract of the anterior commissure. (C-F) The tract of the postoptic commissure. Abbreviations: OS, optic stalk; P, presumptive pia; R, retina; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure; V, ventricle; YG, yolk granule. Scale bars: A,C,E=100 μm; B,D,F=20 μm.

The three-dimensional reconstruction shown in Fig. 6 reveals that the nucleus-free regions shown in Fig. 5C-F aligned as a vertical tract just beneath the surface of the brain, and caudal to the optic stalk (which is not shown in the reconstruction, but can be seen in the micrographs of Fig. 5). In other stage-32 brains, a more anterior tract could also be discerned, just rostral to the optic stalk. The nucleus-free region shown in Fig. 5A is probably an early stage of this tract, which will be seen in later figures. Both of the pathways are stirrup-shaped, suggestive of commissures across the ventral midline. The more caudal longitudinally oriented tract is also evident in Fig. 6. All three of these are in locations where Herrick (1938) described tracts. The more anterior of the two commissures was named the anterior commissure, and thought to be the primordium of the adult structure of the same name. The more posterior was named, with reference to the optic stalk, the postoptic commissure. The longitudinally oriented strip is probably what Herrick called the forerunner of the medial longitudinal fasciculus. We accept Herrick’s description, and hereafter we refer to these elongated nucleus-free regions as tracts, using his nomenclature.

Fig. 6.

Computer-assisted reconstruction of a brain from a stage 32 embryo. Rostral to the left, dorsal up. The outer boundary of the brain and the nucleus-free regions are shown. The latter line up to form the tract of the postoptic commissure, on the surface. One nucleus-free region anteriorly probably presages the tract of the anterior commissure (arrow). A longitudinal nucleus-free region, more caudally, is presumed to be the forerunner of the medial longitudinal fasciculus. The arrowheads indicate the sections shown in Fig. 5A, C, and E. See text for details. Abbreviations: C, cerebellum; E, epiphysis; MLF, medial longitudinal fasciculus; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure. Scale bar=100 μm.

Fig. 6.

Computer-assisted reconstruction of a brain from a stage 32 embryo. Rostral to the left, dorsal up. The outer boundary of the brain and the nucleus-free regions are shown. The latter line up to form the tract of the postoptic commissure, on the surface. One nucleus-free region anteriorly probably presages the tract of the anterior commissure (arrow). A longitudinal nucleus-free region, more caudally, is presumed to be the forerunner of the medial longitudinal fasciculus. The arrowheads indicate the sections shown in Fig. 5A, C, and E. See text for details. Abbreviations: C, cerebellum; E, epiphysis; MLF, medial longitudinal fasciculus; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure. Scale bar=100 μm.

Electron microscopic evidence supported the interpretation of the nucleus-free strips as tracts. Ultrathin sections at several levels were examined, and Fig. 7A and B illustrate the appearance of the dorsal extension of the postoptic commissure (hereafter referred to as the tract of the commissure), roughly at the level of Fig. 5C. There were numerous transversely cut processes, many with microtubules. They resembled young axons described by others (e.g. Cima and Grant, 1980) and we assume that they are axons. Longitudinally cut processes, oriented perpendicularly to the surface of the brain, were also abundant. Many of them widened into the shape typical of the end feet of neuroepithelial cells at the pial surface, and so we assign them that identity. The end feet generally sealed off the parenchyma of the brain from the pial basal lamina. The tract was also characterized by more extracellular space than adjacent regions (compare Fig. 7B and C).

Fig. 7.

Electron micrographs of a stage 32 brain. (A) A low magnification view of the surface of the brain, including the tract of the postoptic commissure, inside the box. (B) A more highly magnified view of the region in the box. The basal lamina surrounding the brain is underlain by neuroepithelial cell processes. Nuclei are excluded (compare to C) Many processes are cut transversely, and are probably axons. (C) A region near the one in B, but different in that the cell nuclei extend right to the edge of the brain, the extracellular space is much reduced, and there are no transversely cut processes. This is typical of most of the outer edge of the brain. Abbreviations: BL, basal lamina; N, nucleus; NEF, neuroepithelial end foot; *, transversely cut process. Scale bar; A=10 μm; B,C=l μm.

Fig. 7.

Electron micrographs of a stage 32 brain. (A) A low magnification view of the surface of the brain, including the tract of the postoptic commissure, inside the box. (B) A more highly magnified view of the region in the box. The basal lamina surrounding the brain is underlain by neuroepithelial cell processes. Nuclei are excluded (compare to C) Many processes are cut transversely, and are probably axons. (C) A region near the one in B, but different in that the cell nuclei extend right to the edge of the brain, the extracellular space is much reduced, and there are no transversely cut processes. This is typical of most of the outer edge of the brain. Abbreviations: BL, basal lamina; N, nucleus; NEF, neuroepithelial end foot; *, transversely cut process. Scale bar; A=10 μm; B,C=l μm.

By stage 35, the tract of the anterior commissure was more evident, the tract of the postoptic commissure was enlarged and continuous with the ventral marginal zone of the midbrain, and the posterior commissure appeared. These developments are illustrated in Fig. 8–10. (The fourth commissure identified by Herrick is the habenular commissure, a dorsal structure just rostral to the posterior commissure. We do not question its existence, but in our material, we could not distinguish it from the posterior commissure, so our description of the latter may actually include the habenular commissure and tract as well.)

Fig. 8.

Horizontal semithin sections from a brain of a stage 35 embryo. Rostral is up. A, C, and E are low magnification views at progressively more dorsal levels (see Fig. 9). B, D, and F are more highly magnified views of the regions inside the boxes of the associated low magnification views. (A, B) A tangential section through the postoptic commissure, showing the labeled optic fibers (arrowheads) in the right optic stalk and along the rostral border of the postoptic commissure. (C, D) Slightly more dorsal section through the tracts of the postoptic commissure, and on the left side, the labeled optic fibers (arrowheads) are cut transversely. (E, F) At the level of the epiphysis, the tract of the postoptic commissure is confluent with a larger nucleus-free region, the ventral marginal zone, but the labeled fibers remain in the same region as before, between pia and nuclei. Abbreviations: E, epiphysis; OS, optic stalk; POC, postoptic commissure; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure; V, ventricle; VMZ, ventral marginal zone. Scale bar: A,C,E=100 μm; B,D,F=20 μm.

Fig. 8.

Horizontal semithin sections from a brain of a stage 35 embryo. Rostral is up. A, C, and E are low magnification views at progressively more dorsal levels (see Fig. 9). B, D, and F are more highly magnified views of the regions inside the boxes of the associated low magnification views. (A, B) A tangential section through the postoptic commissure, showing the labeled optic fibers (arrowheads) in the right optic stalk and along the rostral border of the postoptic commissure. (C, D) Slightly more dorsal section through the tracts of the postoptic commissure, and on the left side, the labeled optic fibers (arrowheads) are cut transversely. (E, F) At the level of the epiphysis, the tract of the postoptic commissure is confluent with a larger nucleus-free region, the ventral marginal zone, but the labeled fibers remain in the same region as before, between pia and nuclei. Abbreviations: E, epiphysis; OS, optic stalk; POC, postoptic commissure; tAC, tract of the anterior commissure; tPOC, tract of the postoptic commissure; V, ventricle; VMZ, ventral marginal zone. Scale bar: A,C,E=100 μm; B,D,F=20 μm.

At this stage, the optic tract was first seen. HRP-labeled optic axons coursed along the rostral border of the postoptic commissure (Fig. 8A and B) and its tract (Fig. 8C-F, and 9). At the leading edge of the optic tract, the axons were found in the space between the pia superficially and the perikarya deep (Fig. 8F; see also Fig. 13). Closer to the chiasm in the optic tract (Fig. 8B and D), the axons were wedged into the same location, suggesting that the pioneering axons have all occupied this strip contiguous with the rostral edge of the postoptic commissure and its tract. In the presumptive chiasm (Fig. 8A and B), the optic axons apparently intermingled with the unlabeled axons from the other eye and perhaps with the nonoptic axons that constituted the commissure before the optic axons arrived.

Electron microscopic examination of the tract of the postoptic commissure (Fig. 10) showed that it resembled that of stage 32. Extracellular space was still abundant. Neuroepithelial end feet sealed the parenchyma of the brain from the basement membrane on the outside. Probable growth cones made extensive contacts with other processes, probably slightly older axons (Fig. 10B and C; also compare Easter, 1987). A few specialized intercellular contacts (membranous thickenings), not seen in stage 32 brains, were evident (Fig. 10C). Axons were more numerous than at stage 32, both in the tracts and in other locations around the perimeter of the brain (not shown). As at stage 32, all of the axons seen at stage 35 were located immediately beneath the subpial neuroepithelial end feet or in a major tract that was in direct contact with the neuroepithelial end feet. No axons or bundles of axons were separated from the pia by cell nuclei. Specialized intercellular contacts were found only in the regions with many axons.

Fig. 9.

Computer-assisted reconstruction of a brain from a stage 35 embryo. Rostral to the left, dorsal up. The outer boundary of the brain and the nucleus-free regions are outlined, and the regions occupied by labeled optic fibers are blackened. Arrowheads show the four sections in Fig. 8A, C, and E. Abbreviations: C, cerebellum; E, epiphysis; MLF, medial longitudinal fasciculus; PC, posterior commissure; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Fig. 9.

Computer-assisted reconstruction of a brain from a stage 35 embryo. Rostral to the left, dorsal up. The outer boundary of the brain and the nucleus-free regions are outlined, and the regions occupied by labeled optic fibers are blackened. Arrowheads show the four sections in Fig. 8A, C, and E. Abbreviations: C, cerebellum; E, epiphysis; MLF, medial longitudinal fasciculus; PC, posterior commissure; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Fig. 10.

Electron micrographs from horizontal sections of the brain of a stage 35 embryo. No HRP labeling. (A) A very low magnification field, illustrating that the entire cross section fits in the open space of a one-hole grid. Rostral is up. The box indicates the region shown in B. (B) The tract of the postoptic commissure; compare with Fig. 7A, the comparable region in a younger brain. The boxed regions are shown in C and D. (C) The superficial region of the tract. Neuroepithelial end feet seal the stroma from the outside, specialized intercellular contacts (arrows), transversely sectioned axons (*), and probable growth cones are identified. (D) Deeper in the tract, the same structures, apart from the endfeet, are evident. Abbreviations: BL, basal lamina; GC, growth cone; N, nucleus; NEF, neuroepithelial end foot; tPOC, tract of the postoptic commissure; V, ventricle; YG, yolk granule. Scale bar: A=100 μm; B = 10 μm; C,D=1 μm.

Fig. 10.

Electron micrographs from horizontal sections of the brain of a stage 35 embryo. No HRP labeling. (A) A very low magnification field, illustrating that the entire cross section fits in the open space of a one-hole grid. Rostral is up. The box indicates the region shown in B. (B) The tract of the postoptic commissure; compare with Fig. 7A, the comparable region in a younger brain. The boxed regions are shown in C and D. (C) The superficial region of the tract. Neuroepithelial end feet seal the stroma from the outside, specialized intercellular contacts (arrows), transversely sectioned axons (*), and probable growth cones are identified. (D) Deeper in the tract, the same structures, apart from the endfeet, are evident. Abbreviations: BL, basal lamina; GC, growth cone; N, nucleus; NEF, neuroepithelial end foot; tPOC, tract of the postoptic commissure; V, ventricle; YG, yolk granule. Scale bar: A=100 μm; B = 10 μm; C,D=1 μm.

In the stage 39 brain, the pre-existing nucleus-free regions have enlarged and new ones have appeared, all in the immediately subpial region (Fig. 11–13). Intercellular contacts were more numerous than at stage 35, and included contacts between presumed neural processes deep in the tract (Fig. 14A and B) and between neural processes and neuroepithelial processes at the pia (Fig. 14C).

Fig. 11.

Horizontal semithin sections from a brain of a stage 39 embryo. Rostral is up. A, B, and C are from progressively more dorsal locations (see Fig. 12), and D is a more highly magnified view of the left box in C. (A) The optic axons (arrow) are still restricted to the space between pia and nuclei on the anterior edge of the tract of the postoptic commissure. (B) More dorsally, as the tract becomes continuous with a more extended nucleus-free region, the axons remain in place.(C) Still more dorsally, the optic fibers begin to spread out (see D). Abbreviations: tAC, tract of the anterior commissure; E, epiphysis; Hb, hindbrain; Mb, midbrain; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; V, ventricle. Scale bar: A-C=100 μm; D=20 μm.

Fig. 11.

Horizontal semithin sections from a brain of a stage 39 embryo. Rostral is up. A, B, and C are from progressively more dorsal locations (see Fig. 12), and D is a more highly magnified view of the left box in C. (A) The optic axons (arrow) are still restricted to the space between pia and nuclei on the anterior edge of the tract of the postoptic commissure. (B) More dorsally, as the tract becomes continuous with a more extended nucleus-free region, the axons remain in place.(C) Still more dorsally, the optic fibers begin to spread out (see D). Abbreviations: tAC, tract of the anterior commissure; E, epiphysis; Hb, hindbrain; Mb, midbrain; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; V, ventricle. Scale bar: A-C=100 μm; D=20 μm.

Fig. 12.

Computer-assisted reconstruction of a brain from a stage 39 embryo. Rostral to the left, dorsal up. Most of the hindbrain is not shown. Arrowheads indicate the three sections in Fig. 11A, B, and C. Abbreviations: C, cerebellum; E, epiphysis; PC, posterior commissure; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Fig. 12.

Computer-assisted reconstruction of a brain from a stage 39 embryo. Rostral to the left, dorsal up. Most of the hindbrain is not shown. Arrowheads indicate the three sections in Fig. 11A, B, and C. Abbreviations: C, cerebellum; E, epiphysis; PC, posterior commissure; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Fig. 13.

Electron micrographs of HRP-labeled axons and growth cones in a brain of a stage 39 embryo. (A) Low magnification field of the anterior boundary of the postoptic commissure, showing a nucleus, the pial boundary, and several labeled processes. (B and C) More highly magnified views of several of the labeled processes in A. The one in B, deep in the tract, is in intimate and extended contact with several processes, perhaps axons. It is assumed (from the presence of the several processes) to be advancing. Four separate elements are evident in C. Numbers 2 and 4 are presumably established axons; number 3 is probably a growth cone with three lamellipodia making intimate contact with the glial endfoot on the left and other unidentified processes; number 1 is an isolated filopodium or lamellipodium, perhaps attached to 3. Abbreviations: BL, basal lamina; N, nucleus; NEF, neuroepithelial end foot; P, pia. Scale bars: A=5 μm; B,C=0.5 μm.

Fig. 13.

Electron micrographs of HRP-labeled axons and growth cones in a brain of a stage 39 embryo. (A) Low magnification field of the anterior boundary of the postoptic commissure, showing a nucleus, the pial boundary, and several labeled processes. (B and C) More highly magnified views of several of the labeled processes in A. The one in B, deep in the tract, is in intimate and extended contact with several processes, perhaps axons. It is assumed (from the presence of the several processes) to be advancing. Four separate elements are evident in C. Numbers 2 and 4 are presumably established axons; number 3 is probably a growth cone with three lamellipodia making intimate contact with the glial endfoot on the left and other unidentified processes; number 1 is an isolated filopodium or lamellipodium, perhaps attached to 3. Abbreviations: BL, basal lamina; N, nucleus; NEF, neuroepithelial end foot; P, pia. Scale bars: A=5 μm; B,C=0.5 μm.

Fig. 14.

Electron micrographs from the transversely sectioned tract of the postoptic commissure in a stage 38 embryo, about midway between chiasm and tectum. All show intercellular contacts (arrowheads) with membranous thickenings. (A and B) Presumed axo-axonal contacts without (A) and with (B) associated vesicles. (C) Contact between an unidentified process and an end foot, just beneath the basal lamina. Abbreviations: BL, basal lamina; NEF, neuroepithelial end foot; Ve, vesicles. Scale bar=0.5 μm.

Fig. 14.

Electron micrographs from the transversely sectioned tract of the postoptic commissure in a stage 38 embryo, about midway between chiasm and tectum. All show intercellular contacts (arrowheads) with membranous thickenings. (A and B) Presumed axo-axonal contacts without (A) and with (B) associated vesicles. (C) Contact between an unidentified process and an end foot, just beneath the basal lamina. Abbreviations: BL, basal lamina; NEF, neuroepithelial end foot; Ve, vesicles. Scale bar=0.5 μm.

The number of labeled optic axons has also increased, as a comparison of Fig. 11B and 8D shows. The cross-section of the presumptive optic tract has become both broader and deeper. The labeled axons and some growth cones (Fig. 13) continued to occupy a superficial position, just beneath the pia, along the rostral edge of the postoptic commissure and tract. They were not found in the anterior commissure or its tract, even dorsal to the region where the tracts of the two commissures met (Fig. 12). Many of the labeled axons had arrived at the presumptive tectum, where they spread out and arborized (Fig. 2E, 11C, and D). In this study, we did not investigate whether the new axons followed retinotopic rules in the formation of the optic tract, as they do at later stages of development (Taylor, 1987; Wilson et al. 1988), but such a study is currently in progress (Taylor, unpublished).

In the stage 46 brain, the same trends continued, as Fig. 15 illustrates. The nucleus-free regions enlarged, the labeled optic axons remained clustered along the rostral border of the postoptic commissure and tract, the cross-sections of all tracts enlarged, and the labeled axons arborized further on the tectum. These developments are evident in the reconstruction of Fig. 16. A comparison of the four reconstructions (Fig. 6, 9, 12, and 16), shows that the optic tract has grown relatively more than the nonoptic part of the postoptic commissure and tract.

Fig. 15.

Horizontal semithin sections from the brain of a stage 46 embryo. See Fig. 16 for their levels. (A) Low magnification view including part of the forebrain, midbrain, and hindbrain. The labeled optic fibers (arrowheads) now occupy a broad cross-section, still restricted to the anterior edge of the enlarged nucleus-free zone. (B) Detail of the box in A, showing the different cytoarchitectonies at the boundary between mid- and hindbrain. (C) Low magnification view, dorsal to A. The labeled fibers (arrowhead) have spread out. (D) Detail of the boxed region in C, showing the cytoarchitectonie difference between the periventricular and superficial cells. Abbreviations: E, epiphysis; tPC, tract of the posterior commissure; V, ventricle. Scale bars: A,C=100 μm; B,D=20 μm.

Fig. 15.

Horizontal semithin sections from the brain of a stage 46 embryo. See Fig. 16 for their levels. (A) Low magnification view including part of the forebrain, midbrain, and hindbrain. The labeled optic fibers (arrowheads) now occupy a broad cross-section, still restricted to the anterior edge of the enlarged nucleus-free zone. (B) Detail of the box in A, showing the different cytoarchitectonies at the boundary between mid- and hindbrain. (C) Low magnification view, dorsal to A. The labeled fibers (arrowhead) have spread out. (D) Detail of the boxed region in C, showing the cytoarchitectonie difference between the periventricular and superficial cells. Abbreviations: E, epiphysis; tPC, tract of the posterior commissure; V, ventricle. Scale bars: A,C=100 μm; B,D=20 μm.

Fig. 16.

Computer-assisted reconstruction of the brain of a stage 46 embryo. Rostral to the left, dorsal up. Most of the hindbrain is excluded. Arrowheads indicate the levels of the sections shown in Fig. 15A and B. Abbreviations: E, epiphysis; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Fig. 16.

Computer-assisted reconstruction of the brain of a stage 46 embryo. Rostral to the left, dorsal up. Most of the hindbrain is excluded. Arrowheads indicate the levels of the sections shown in Fig. 15A and B. Abbreviations: E, epiphysis; tAC, tract of the anterior commissure; tPC, tract of the posterior commissure; tPOC, tract of the postoptic commissure; VMZ, ventral marginal zone. Scale bar=100 μm.

Cytoarchitectonies

The relatively homogeneous cellular core of the brain at stage 32 has differentiated slightly by stage 35. The cells near the center of the presumptive tectum were round, but at the isthmus, they were elongate (not shown). These regional differences became quite exaggerated by stage 46, as Fig. 15A-B illustrate. The differences in shape correlated with differences in staining, too. The cells of the stage 35 brain stained nearly uniformly, irrespective of their positions relative to either the ventricle or the boundaries of a subfield like the tectum. In the stage 46 brain, in contrast, cells near the ventricle or at a boundary had rather sparse cytoplasm and small nuclei, both of which were stained deeply by the toluidine blue. They were quite different from the more superfiçial cells which were larger, rounder and paler.

The regional variations in the numbers of these cells presumably contributed to make the brain less tubular than at earlier stages. In the tectum, now identified by the presence of extensive optic arbors, the large and pale cells were quite numerous. At the boundaries of the tectal field they were absent, and the wall of the brain was spanned by radial processes of the darker periventricular cells (Fig. 15C and D). The thin boundary zones and the thick tectum combined to produce an inward bulge of the wall, thereby constricting the ventricle. Two similar classes of cells have been noted in the goldfish tectum (Raymond and Easter, 1983) and, by analogy, we suggest that the dark elongate cells are proliferative, the round ones, postmitotic.

Superficial and deep pathways

The embryonic CNS, according to the classic description, is a nuclear core and a fibrous exterior. But all vertebrate brains eventually contain tracts that are deep; that is, separated from the pia by cell nuclei. How soon do the deep tracts appear, and how do they originate? We have confirmed that all of the early tracts in the archencephalon are immediately subjacent to the pia at stages 32, 35, and 39. The first exception to this generalization was seen at stage 46. Fig. 15C illustrates this tract, just caudal to the epiphysis, and tentatively identified as the posterior commissure; it was separated from the pia by cell nuclei. A superficial tract was evident in this same region at stage 35 (Fig. 9), and by stage 39 (Fig. 11C) it was still superficial. We suggest that the deep tract of Fig. 15C began as the superficial one of Fig. 9 and 11C, and was secondarily enveloped by nuclei that migrated from the periventricular zone to the pia.

Electron microscopy permits the detection of small numbers of axons that might be missed light microscopically. We have examined individual sections for axons, to check the validity of the classical notion that they are found exclusively in a superficial location. A horizontal section through the archencephalon is small enough to fit onto the open space of a one-hole grid, with no overlap onto the metal part of the grid. Fig. 10A shows one such section, supported by a transparent formvar film. It can therefore be examined, in its entirety, at a magnification high enough to reveal axons. We have searched for them in several sections of two stage 32 brains. The great majority were found in one of the tracts described at the beginning of Results. A few others were found relatively isolated - either alone, or in groups of 2 –4. All of the relatively isolated axons were found immediately subjacent to the pia, among the end feet of the neuroepithelial cells. None were found in the cellular core. We assume that these relatively isolated axons on the perimeter of the stage 32 brain were the pioneers of other pathways to be formed later. These isolated axons were much less numerous than the ones added to the pre-existing tracts.

In summary, pioneering axons, recognizable by their relative isolation, always grew in the subpial lamina, never in the cellular core. Most later axons joined preexisting tracts, at least up to stage 46, the latest stage that we examined in this context.

We have described the development of the Xenopus retinofugal pathway and the relation of the optic axons to the changing morphology of the brain and to their microenvironment. Our discussion will focus first on several issues relating to the early development of the brain and its tracts, and close with special attention to the factors that may guide the optic axons.

Brain differentiation

The pistol shape of the brain changes relatively little from stage 32 to 40. The only distinctive features are the optic stalks, the epiphysis and the isthmus. After stage 40, the most noticeable changes are the rapid expansion of the forebrain and the unbending of the primary flexure ventral to the isthmus. The elongation of the forebrain has the effect of displacing dorsal midbrain structures caudally, so the orientation of the distal optic tract changes from dorsad to caudad.

As development proceeds, the midbrain acquires a more distinctive regional morphology. The presumptive tectum, initially just a slight bulge, increases in size and protrudes both medially and laterally by stage 35. Bulges of this sort provided the basis for dividing the brain into ‘neuromeres,’ an exercise that commanded considerable attention by the neuroembryologists of the 1940s and 50s (e.g. Bergquist, 1952), and which has recently undergone a revival of interest (Hanneman et al. 1988; Lumsden and Keynes, 1989).

At the cellular level, morphological differentiation is also underway. At stage 35, the cells at the center of the tectum are morphologically distinct from those on the edge. By stage 38/39, the different shapes are accompanied by different staining, and these differences become very pronounced by stage 46. Since the optic axons do not arrive until stages 39/40 (Gaze et al. 1974; Holt and Harris, 1983), but tectal morphogenesis and cellular differentiation are already evident at stage 35, we conclude that these two events do not depend on the arrival of the retinal axons.

At the subcellular level, we saw evidence for specialized intercellular contacts, symmetric membranous thickenings, sometimes associated with vesicles, as early as stage 35. Where the axons bend, in the midregion of the diencephalon, wisps of terminal arbors have been seen. It is in this area that we have found fluorescently labeled cells after Dil application to the retina. Since Dil is known to diffuse along cell membranes, this suggests that the labeled cells were in intimate contact with the optic axons (Godement et al. 1987). These early arbors and contacts may be the beginnings of the retinothalamic projection.

Initial scaffold

Our use of horizontal sections, which intercept most reliably those tracts oriented dorsoventrally, has directed our attention to the anterior and postoptic commissures and the tracts that extend dorsally from them. These two stirrup-shaped pathways are evident at stage 32, and they enlarge steadily thereafter. We anticipated seeing new tracts in later stages, but by stage 46, these two were still the only substantial dorsoventral pathways in the fore- and midbrain. With the electron microscope, we saw occasional axons elsewhere, but the vast majority of dorsoventrally oriented axons that appeared between stages 32 and 46 were added to these two pathways. Herrick (1938) believed that the initial axons originated from cells lining the pathways and, although we have not evaluated that claim in Xenopus, it does appear to be true for the tract of the postoptic commissure in the zebrafish, Brachydanto rerio (Wilson and Easter, 1989). Many of the later axons in, or alongside, the postoptic commissure and its tract originated in the contralateral eye. The selective enlargement of pre-existing tracts, rather than the formation of new ones, is very striking. The mature brain has more than two dorsoventral tracts, so new pathways must develop at some time, but it remains to be shown if they originate as new ones, separate from the pre-existing ones, or by subdivision of those already in place. Irrespective of how that later process occurs, we suggest that a general rule of the early development of pathways may be the following: first create a relatively few tracts (the initial scaffold) and add later axons to these few rather than form new ones.

The initial pathways are superficial, immediately subjacent to the pia. This confirms classical descriptions of the embryonic brain as a cellular core and a fibrous exterior. These observations on the fore- and midbrain are consistent with others made on Xenopus, both in the spinal cord (Singer et al. 1979; Nordlander and Singer, 1982a; 1982b; Nordlander, 1984) and the hindbrain (Kevetter and Lasek, 1982). Our observations are also consistent with recent immunocytochemical investigations of avian and mammalian embryos (Dodd and Jessell, 1988; Dodd et al. 1988; Lumsden and Keynes, 1989). We have sought axons at deeper locations, and failed to find them, even with the electron microscope. To be sure, they must emerge from the cellular core, oriented toward the pia, (see Lumsden and Keynes, 1989), but our results suggest that axonal growth in the direction parallel to the surface of the brain must be very rare at any depth other than the most superficial. Once a pathway has been established, growth cones can apparently advance in association with other axons deep to the pia (see Fig. 13), but when the choice is either the underside of the pia or the cellular core, the axons choose the former. It appears that the early CNS contains two zones that are hospitable to axonal growth - the subpial zone and pre-existing tracts - and one zone that is inhospitable - the cellular core.

Apparently all regions of the subpial zone are not equally conducive to axonal growth, though, since tracts appear in particular places and quite well spaced from one another. The axons that form them must respond to cues in the subpial region, cues that may be associated with neuroepithelial endfeet or with the extracellular matrix. We use the word ‘cue’ guardedly, fully aware that the choice of one particular route may be a result either of favorable conditions there, such as the presence of certain substrate molecules (Silver and Rutishauser, 1984; Neugebauer et al. 1988; Matsunaga et al. 1988) or of unfavorable and even hostile conditions in nearby regions (Kapfhammer et al. 1986), or both. Growth cones arriving later - such as those of the retinal ganglion cells - are probably influenced similarly, with the additional complexity that the axons preceding them may also carry molecular cues.

Optic pathway

The optic axons are easily labeled at the eye, and this permits us to distinguish them from their nonoptic predecessors. We have exploited this to examine the microenvironment through which they advance. Earlier studies have traced the retinal fibers through the optic stalk to the chiasm. Our observations begin at the chiasm, which is formed along the rostral border of the postoptic commissure. Within the chiasm, the axons of the two eyes apparently course side-by-side, in opposite directions, with relatively few mistakes. Rare ipsilateral axons have been reported by others (Sakaguchi and Murphey, 1985), but no one has described retinoretinal trajectories, as have been seen in rodents (Bunt and Lund, 1981; Godement et al. 1987). The rarity of ipsilateral axons can perhaps be explained by the fact that the two optic tracts approach one another nearly head on, so that an ipsilateral deflection would require a 180 degree turn. We can provide no such explanation for the absence of retinoretinal axons, however.

As the axons emerge from the chiasm, they retain the same position on the rostral border of the tract as it turns dorsally along the diencephalon, just caudal to the optic stalk contralateral to the fibers’ origins. They keep this same position until the boundary of the tract of the postoptic commissure becomes indistinct, and the tectal neuropil is approached.

Guidance of optic axons

Our original interest in this subject arose from the question of what guides retinal axons to their target. Several studies had addressed this issue by describing the trajectories of optic axons that had been induced to enter the brain at an abnormal location (Sharma, 1972; Constantine-Paton and Capranica, 1975; Constantine-Paton, 1978; Giorgi and Van Der Loos, 1978; Katz and Lasek, 1978; 1979; Harris, 1986). These findings prompted a range of interpretations, including attraction by the target (Giorgi and Van Der Loos, 1978; Harris, 1986), pathfinding on the basis of directional (Constantine-Paton and Capranica, 1975) or positional cues (Harris, 1986), and selective axonal fasciculation (Constantine-Paton, 1978; Katz and Lasek, 1978; 1979). In all, the results were quite variable, as is usually the case in experimental embryology, and the interpretations were suitably guarded. We believe that our description of the normal case can help to interpret the results of experimental approaches, both past and future (e.g. Taylor, 1989).

There are two salient points to be emphasized. First, the retinal axons did not pioneer a pathway, but joined a pre-existing one. Second, they did not intermix significantly with the pathway, but carved out a parallel and contiguous strip alongside it, in the most superficial part of the brain.

The fact that the optic fibers joined a pre-existing pathway suggests immediately that they may have been guided by other axons. This idea has recently received strong support from the work on insect central nervous systems by Ghysen (1978) and Goodman (Goodman et al. 1984). The initial growth of CNS pioneer axons appears to be directed by surface cues on glial cells that overlie the developing neuroepithelium (Bastiani and Goodman, 1986). The basic scaffold of longitudinal and commissural axon tracts is established by these pioneer axons. Subsequent axonal outgrowth occurs in a highly organised fashion by axons selectively fasciculating with specific axons within the fascicles (Raper et al. 1983; Bastiani et al. 1984). At choice points, the growing axons appear to be able to discern differences between axons and to make specific choices. The mechanism underlying this selective fasciculation involves cell surface molecules, differentially distributed on specific axons and on different regions of an individual axon (Bastiani et al. 1987; Patel et al. 1987; Snow et al. 1988). In this fashion, a highly complex system is thought to be gradually constructed by relatively simple choices, involving such cell surface markers (Harrelson and Goodman, 1988).

Correlates of this idea have been found in the development of the spinal cords of fish (Kuwada, 1986) and mammal, where axonal choices appear to be made in response to changes in expression of cell surface markers (Dodd et al. 1988; Dodd and Jessell, 1988). We suggest that the founders of the postoptic commissure and its associated tract may be analogous to the pathway pioneers in grasshoppers, and the first optic axons are probably fasciculating selectively with this group. This must be considered speculative, as we have not shown that the first optic axons grow on other axons already in place, only that the optic axons are contiguous to the pre-existing tract.

At later stages of life, optic axons are known to fasciculate with their optic predecessors during normal development (Bodick and Levinthal, 1980; Cima and Grant, 1982; Easter et al. 1984; Bork et al. 1987). Following optic nerve section, late-regenerating axons behave similarly (Easter, 1987). Apparently, optic axons provide an excellent substrate for the growth of other optic axons.

More extreme surgical intervention has produced additional support for the role of axons as substrate. In adult goldfish, when one tectal lobe is removed, the severed optic axons grow across the midline to innervate the remaining tectal lobe, but they do not grow directly toward it. Instead, they join a small number of established pathways, and one of them, the horizontal commissure, leads initially away from the midline and follows a very circuitous route to the remaining tectal lobe (Easter et al. 1978; Lo and Levine, 1980). If the tectum were providing some long-range attractant, or if local positional or directional cues indicated the midline, one would expect the axons to grow medially. The fact that they did not supports the role of axons as a very favorable, and nondirectional, substrate for axonal growth.

If optic axons grow best, and perhaps exclusively, in contact with other axons, then the earlier experimental work on frog embryos can be interpreted more readily. Most of this work produced variable results, and interpretation necessarily entailed exclusion of some of the data. In particular, the report by Harris (1986) included cases in which the optic axons grew down the spinal cord, directly away from the presumptive tectum, rather than toward it. This cannot be easily reconciled with ideas of local directional or positional cues or long-range attraction. But such a result would be expected if the axons from the ectopic eye grew best along preexisting tracts, with the direction of growth dictated more by the angle from which the axons approached the tract than by positional or directional cues associated with it.

We are therefore inclined to interpret ‘homing behavior’ (Harris, 1986) as due, at least in part, to the ectopic axons growing into and then following one of the relatively few pre-existing tracts that make up the initial scaffold. Several of these tracts pass near the presumptive tectum and might plausibly lead axons there. Specifically, the tract of the postoptic commissure coincides with the normal pathway to the tectum, so any ectopic axons on the archencephalon that found their way into this tract would already be on the designated path to the tectum. The two other tracts most likely to be encountered by axons in the caudal forebrain or midbrain are the medial longitudinal fasciculus and the posterior commissure; both pass within a few tens of/rm of the presumptive tectum (Herrick, 1938; Roberts et al. 1987) and could therefore channel axons into it.

Our results also provide an interpretation of the more recent report by Harris (1989). In stage 26 Xenopus embryos, he rotated a patch of neuroepithelium through which the optic axons would grow. Later, he observed that their trajectories turned away from the normal upon entering the rotated patch; specifically, when the patch had been rotated clockwise, the optic axons were deflected clockwise. He interpreted this as evidence against long-range attraction and in favor of local cues. We agree, and propose that the salient cue may not have been in the neuroepithelial cells, but in the axons of the tract of the postoptic commissure, which would probably have formed in an abnormal orientation inside the rotated patch.

We do not claim to have ruled out the possibility of axonal guidance by long-range attraction or by local directional or positional cues in the neuroepithelium. We have provided evidence for a third mechanism - association with pre-existing axons - that can account for the behavior of normal optic axons, for much of the previously inexplicable behavior of ectopic ones, and for the novel routes taken through experimentally altered brains.

This work was carried out with the support of the Royal Society, the Medical Research Council, the Wellcome Trust, and research grant EY-00168 from the National Institutes of Health. We thank Dr R. M. Gaze and Dr P. Grant for useful discussions of the work in progress, Mr John Burrill, Ms Riva Marcus, Dr Ray Guillery, Dr Pamela Raymond, Dr Linda Ross, and Dr Steve Wilson for useful comments on the paper, and Ms Celeste Malinoski and Mr Terry Richards for help with the illustrations.

Bastiani
,
M. J.
and
Goodman
,
C. S.
(
1986
).
Guidance of neuronal growth cones in the grasshopper. DU. Recognition of specific glial pathways
.
J. Neurosa
.
6
,
3542
3551
.
Bastiani
,
M. J.
,
Harrelson
,
A. L.
,
Snow
,
P. M.
and
Goodman
,
C. S.
(
1987
).
Expression of fasciclin I and II glycoproteins on subsets of axon pathways during neuronal development in the grasshopper
.
Cell
48
,
745
755
.
Bastiani
,
M. J.
,
Raper
,
J. A.
and
Goodman
,
C. S.
(
1984
).
Pathfinding by neuronal growth cones in grasshopper embryo: HI. Selective affinity of the G growth cone for the P cells within the A/P fascicle
.
J. Neurosci
.
4
,
3211
3228
.
Bergquist
,
H.
(
1952
).
Studies on the cerebral tube in vertebrates. The neuromeres
.
Acta Zool
.
33
,
1
61
.
Bodick
,
N.
and
Levinthal
,
C.
(
1980
).
Growing optic nerve fibers follow neighbors during embryogenesis
.
Proc. natn. Acad. Sci. U.S.A
.
77
,
4374
4378
.
Bork
,
T.
,
Schabtach
,
E.
and
Grant
,
P.
(
1987
).
Factors guiding optic fibres in developing Xenopus retina
.
J. comp. Neurol
.
264
,
147
158
.
Bunt
,
S. M.
and
Lund
,
R. D.
(
1981
).
Development of a transient retino-retinal pathway in hooded and albino rats
.
Brain Res
.
211
,
399
04
.
Cima
,
C.
and
Grant
,
P.
(
1980
).
Ontogeny of the retina and optic nerve of Xenopus laevis. IV. Ultrastructural evidence of early ganglion cell differentiation
.
Devi Biol
.
76
,
229
237
.
Cima
,
C.
and
Grant
,
P.
(
1982
).
Development of the optic nerve in Xenopus laevis. I. Early development and organisation
.
J. Embryol. exp. Morph
.
72
,
225
249
.
Constantine-Paton
,
M.
(
1978
).
Central projections of anuran optic nerves penetrating hindbrain or spinal cord regions of the neural tube
.
Brain Res
.
158
,
31
43
.
Constantine-Paton
,
M.
and
Capranica
,
R. P.
(
1975
).
Central projections of optic tract from translocated eyes in the leopard frog (Rana pipiens)
.
Science
189
,
480
482
.
Dodd
,
J.
and
Jessell
,
T. M.
(
1988
).
Axon guidance and the patterning of neuronal projections in vertebrates
.
Science
242
,
700
708
.
Dodd
,
J.
,
Morton
,
S. B.
,
Karagogeos
,
D.
,
Yamamoto
,
M.
and
Jessell
,
T. M.
(
1988
).
Spatial segregation of axonal glycoprotein expression on subsets of embryonic spinal neurons
.
Neuron
1
,
105
116
.
Easter
,
S. S.
(
1987
).
Retinal axons and the basal lamina
.
In Mesenchymal-Epithelial Interactions in Neural Development
, (ed.
J. R.
Wolff
et al. 
)
Springer
.
Berlin
, pp.
385
396
.
Easter
,
S. S.
,
Bratton
,
B.
and
Scherer
,
S. S.
(
1984
).
Growth-related order of the retinal fiber layer in goldfish
.
J. Neurosci
.
4
,
2173
2190
.
Easter
,
S. S.
,
Schmidt
,
J. T.
and
Leber
,
S. M.
(
1978
).
The paths and destinations of the induced ipsilateral retinal projection in goldfish
.
J. Embryol. exp. Morph
.
45
,
145
159
.
Easter
,
S. S.
and
Taylor
,
J. S. H.
(
1988
).
Early development of the Xenopus visual system
.
Invest. Ophth. vis. Sa
.
29
,
378
.
Gaze
,
R. M.
,
Keating
,
M. J.
and
Chung
,
S. H.
(
1974
).
The evolution of the retinotectal map during development in Xenopus
.
Proc. R. Soc. Lond. B
185
,
301
330
.
Ghysen
,
A.
(
1978
).
Sensory neurones recognise defined pathways in the Drosophila central nervous system
.
Nature, Lond
.
IT A
,
869
873
.
Giorgi
,
P. P.
and
Van Der Loos
,
H.
(
1978
).
Axons from eyes grafted in Xenopus can grow into the spinal cord and reach the optic tectum
.
Nature, Lond
.
275
,
746
748
.
Godement
,
P.
,
Vanselow
,
J.
,
Thanos
,
S.
and
Bonhoeffer
,
F.
(
1987
).
A study of the developing visual systems with a new method of staining neurones and their processes in fixed tissue
.
Development
101
,
697
713
.
Goodman
,
C. S.
,
Bastiani
,
M. J.
,
Doe
,
C. Q.
,
Du Lac
,
S.
,
Helfand
,
S. L.
,
Kuwada
,
J. Y.
and
Thomas
,
J. B.
(
1984
).
Cell recognition during neuronal development
.
Science
225
,
1271
1279
.
Hanneman
,
E.
,
Trevarrow
,
B.
,
Metcalfe
,
W. K.
,
Kimmel
,
C. B.
and
Westerfield
,
M.
(
1988
).
Segmental pattern of development of the hindbrain and spinal cord of the zebrafish embryo
.
Development
103
,
49
58
.
Harrelson
,
A. L.
and
Goodman
,
C. S.
(
1988
).
Growth cone guidance in insects: Fasciclin n is a member of the immunoglobulin superfamily
.
Science
242
,
700
708
.
Harris
,
W. A.
(
1986
).
Homing behaviour of axons in the embryonic vertebrate brain
.
Nature, Lond
.
320
,
266
269
.
Harris
,
W. A.
(
1989
).
Local positional cues in the neuroepithelium guide retinal axons in embryonic Xenopus brain
.
Nature, Lond
.
339
,
218
221
.
Harris
,
W. A.
,
Holt
,
C. E.
and
Bonhoeffer
,
F.
(
1987
).
Retinal axons with and without their somata, growing to and arborising in the tectum of Xenopus embryos: a time-lapse video study of single fibres in vivo
.
Development
101
,
123
133
.
Herrick
,
C. J.
(
1938
).
Development of the cerebrum of Amblystoma during the early swimming stages
.
J. comp. Neurol
.
68
,
203
241
.
Holt
,
C. E.
(
1984
).
Does timing of axon outgrowth influence retinotectal topography in Xenopus?
J. Neurosci
.
4
,
1130
1152
.
Holt
,
C. E.
and
Harris
,
W. A.
(
1983
).
Order in the initial retinotectal map in Xenopus: a new technique for labelling growing nerve fibres
.
Nature, Lond
.
301
,
150
152
.
Honig
,
M. G.
and
Hume
,
R. I.
(
1986
).
Fluorescent carbocyanine dyes allow living neurons of identified origin to be studied in long-term cultures
.
J. Cell Biol
.
103
,
171
187
.
Jarvik
,
E.
(
1980
).
Basic Structure and Evolution of Vertebrates
, vol.
2
.
Academic Press
:
New York
, pp.
17
18
.
Kapfhammer
,
J. P.
,
Grünewald
,
B. E.
and
Raper
,
J. A.
(
1986
).
The selective inhibition of growth cone extension by specific neurites in culture
.
J. Neurosci
.
6
,
2527
2534
.
Katz
,
M. J.
and
Lasek
,
R. J.
(
1978
).
Eyes transplanted to tadpole tails send axons rostrally in two spinal cord tracts
.
Science
109
,
202
203
.
Katz
,
M. J.
and
Lasek
,
R. J.
(
1979
).
Substrate pathways which guide growing axons in Xenopus embryos
.
J. comp. Neurol
.
183
,
817
832
.
Katz
,
M. J.
,
Lasek
,
R. J.
and
Nauta
,
H. J. W.
(
1980
).
Ontogeny of substrate pathways and the origin of the neural circuit pattern
.
Neuroscience
5
,
821
833
.
Kevetter
,
G. A.
and
Lasek
,
R. J.
(
1982
).
Development of the marginal zone in the rhombencephalon of Xenopus laevis
.
Dev. Brain Res
.
4
,
195
208
.
Kuwada
,
J.
(
1986
).
Cell recognition by neuronal growth cones in a simple vertebrate embryo
.
Science
233
,
740
746
.
Lo
,
R. Y. S.
and
Levine
,
R. L.
(
1980
).
Time course and pattern of optic fiber regeneration following tectal lobe removal in the goldfish
.
J. comp. Neurol
.
191
,
295
314
.
Lumsden
,
A.
and
Keynes
,
R.
(
1989
).
Segmental patterns of neuronal development in the chick hindbrain
.
Nature, Lond
.
337
,
424
428
.
Matsunaga
,
M.
,
Hatta
,
K.
,
Nagafuchi
,
A.
and
Takeichi
,
M.
(
1988
).
Guidance of optic nerve fibres by N-cadherin adhesion molecules
.
Nature, Lond
.
334
,
62
64
.
Neugebauer
,
K. M.
,
Tomaselli
,
K. J.
,
Lujen
,
J.
and
Reichardt
,
L. F.
(
1988
).
N-cadherin, NCAM, and integrins promote retinal neurite outgrowth on astrocytes in vitro
.
J. Cell Biol
.
107
,
1177
1187
.
Nieuwkoop
,
P. D.
and
Faber
,
J.
(
1967
).
A Normal Table of Xenopus laevis (Daudin)
.
Amsterdam
,
North-Holland
.
Nordlander
,
R. H.
(
1984
).
Developing descending neurons of the early Xenopus tail spinal cord in the caudal spinal cord of early Xenopus J
.
comp. Neurol
.
228
,
117
228
.
Nordlander
,
R. H.
and
Singer
,
M.
(
1982a
).
Spaces precede axons in the embryonic spinal cord
.
Expl Neurol
.
75
,
221
228
.
Nordlander
,
R. H.
and
Singer
,
M.
(
1982b
).
Morphology and position of growth cones in the developing Xenopus spinal cord
.
Devi Brain Res
.
4
,
181
193
.
O’Rourke
,
N. A.
and
Fraser
,
S. E.
(
1986
).
Dynamic aspects of retinotectal map formation revealed by a vital-dye fiber-tracing technique
.
Devi Biol
.
114
,
265
276
.
Patel
,
N. H.
,
Snow
,
P. M.
and
Goodman
,
C. S.
(
1987
).
Characterization and cloning of fasciclin III: A glycoprotein expressed on a subset of neurons and axon pathways in Drosophila
.
Cell
48
,
975
988
.
Raper
,
J. A.
,
Bastiani
,
M. J.
and
Goodman
,
C. S.
(
1983
).
Pathfinding by neuronal growth cones in grasshopper embryos. II Selective fasciculation onto specific axonal pathways
.
J. Neurosci
.
3
,
31
41
.
Raymond
,
P. A.
and
Easter
,
S. S.
(
1983
).
Postembryonic growth of the optic tectum in goldfish. I. Location of germinal cells and numbers of neurons produced
.
J. Neurosci
.
3
,
1077
1091
.
Roberts
,
A.
,
Dale
,
N.
,
Ottersen
,
P.
and
Storm-Mathisen
,
J.
(
1987
).
The early development of neurones with GABA immunoreactivity in the CNS of Xenopus laevis embryos
.
J. comp. Neurol
.
261
,
435
449
.
Sakaguchi
,
D. S.
and
Murphey
,
R. K.
(
1985
).
Map formation in the developing retinotectal system: An examination of ganglion cell terminal arborisations
.
J. Neurosci
.
5
,
3228
3245
.
Sharma
,
S. C.
(
1972
).
Retinotectal connexions of a heterotopic eye
.
Nature, New Biol
.
238
,
286
287
.
Silver
,
J.
and
Rutishauser
,
U.
(
1984
).
Guidance of optic axons by a preformed adhesive pathway on neuroepithelial end feet
.
Devi Biol
.
106
,
485
499
.
Singer
,
M.
,
Nordlander
,
R. H.
and
Egar
,
M.
(
1979
).
Axonal guidance during embryogenesis and regeneration in the spinal cord of the newt: the blueprint hypothesis of neuronal pathway patterning
.
J. comp. Neurol
.
185
,
1
22
.
Snow
,
P. M.
,
Zinn
,
K.
,
Harrelson
,
A. L.
,
McAllister
,
L.
,
Schilling
,
J.
,
Bastiani
,
M. J.
,
Makk
,
G.
and
Goodman
,
C. S.
(
1988
).
Characterization and cloning of fasciclin I and II glycoproteins in the grasshopper
.
Proc. natn. Acad. Sci. U.S.A
.
85
,
5291
5295
.
Straznicky
,
K.
and
Gaze
,
R. M.
(
1971
).
The growth of the retina in Xenopus laevis: an autoradiographic study
.
J. Embryol. exp. Morph
.
26
,
67
79
.
Taylor
,
J. S. H.
(
1987
).
Fibre organisation and reorganisation in the retinotectal projection of Xenopus
.
Development
99
,
393
410
.
Taylor
,
J. S. H.
(
1989
).
The directed growth of retinal axons towards surgically transposed tecta; an examination of homing behaviour by retinal ganglion cell axons. (In preparation
.)
Taylor
,
J. S. H.
and
Gaze
,
R. M.
(
1985
).
The effects of the fibre environment on the paths taken by regenerating optic nerve fibres in Xenopus
.
J. Embryol. exp. Morph
.
89
,
383
401
.
Wilson
,
M. A.
,
Taylor
,
J. S. H.
and
Gaze
,
R. M.
(
1988
).
A developmental and ultrastructural study of the optic chiasma in Xenopus. Development
102
,
537
553
.
Wilson
,
S. W.
and
Easter
,
S. S.
(
1989
).
Early development of the postoptic commissure in zebrafish embryos
.
Abstr. Soc. Neurosci
.
(in press)
.