ABSTRACT
We have examined the neural tube in Xenopus laevis tadpoles to investigate the anatomical guidance elements which may be present in the presumptive marginal zone. With appropriate fixation protocols the neuroepithelial cells appeared in contact; electron microscopic observations failed to show any specialized intercellular spaces preceding the growing axons. The first fibres were found in the intercellular clefts between the neuroepithelial cells near the surface of the neural tube. Reconstructions of the neural tube from examination of serial 1 μm sections showed that the intercellular clefts are non-aligned at this stage and branching. Scanning electron microscopy of the surface of the neural tube confirmed that the intercellular spaces are non-aligned and often branch caudal to the growing front of descending axons. Thus to grow in a consistent direction the developing axons may have to make consistent and selective (specific) selections of pathway at numerous branch points if their growth is restricted to these intercellular clefts. As more iixons grow along the neural tube, the intercellular clefts become wider, and the neuroepithelial cells bounding the clefts become indented. At later stages many fibres were observed with both scanning and transmission electron microscopy to grow along the surface of the neural tube. These changes in neuroepithelial cell morphology and fibre pathway allow axons to form bundles which take a fairly straight course in contrast to the winding path which must be taken by the first axons to grow through the intercellular clefts.
INTRODUCTION
It has been claimed that physical (Weiss, 1934), or chemical (Sperry, 1963), guidance systems exist in the developing nervous system, and that they are responsible and necessary for normal pathway selection by growing fibres. There is, however, controversy with respect to the part played by the various factors implicated in fibre guidance, and to the distribution of these factors in the nervous system. Physical and chemical guidance systems have been the subject of recent reviews (Letourneau, 1982; Purves & Lichtman, 1983), but only physical or morphological systems are considered here.
Some workers (Katz & Lasek, 1978, 1981; Singer, Nordlander & Egar, 1979; Kevetter & Lasek, 1982), claim that stereotyped axon tracts between specific neuronal populations are a critical requirement in the central nervous system, and that ‘aligned substrate pathways’ exist to guide growing axons. The idea that these substrate pathways may consist of some form of morphological structure which can guide the fibres appears to have been accepted by some (Silver & Robb, 1979; Silver & Sidman, 1980; Silver, Lorenz, Wahlsten & Coughlin, 1982), while others deny that such physical guidance exists (Valentino & Jones, 1982), or is necessary (Fujisawa, Tani, Watanabe & Ibata, 1982) for fibres to follow their normal pathways.
Most of the evidence for anatomically identifiable structures which could play a role in the guidance of central nervous system fibre tracts has come from studies of the optic pathway or the neural tube. Some workers have observed aligned spaces in the embryonic retina and optic stalk which precede, and could therefore play a role in the guidance of, later growing axons (Silver & Robb, 1979; Silver & Sidman, 1980 in mice; Silver & Sapiro, 1981 in mice, rat, chick and Xenopus;Krayanek & Goldberg, 1981, in the chick; Halfter & Deiss, 1984, in the chick and quail), while others have been unable to find any alignment of such spaces (Suburo, Carri & Adler, 1979).
The first suggestion that glial channels or spaces could guide the outgrowth of CNS axon tracts came from studies of the regenerating spinal cord (Egar & Singer, 1972; Nordlander & Singer, 1978). Later studies using transmission electron microscopy have found such specialized structures in the developing Xenopus spinal cord (Nordlander & Singer, 1982a,h) but have not investigated whether these channels are aligned in any special way which would suggest that they could play a role in the predominantly rostrocaudal growth of the main developing fibre tracts of the neural tube. Other workers using scanning and transmission electron microscopy have shown that the first identifiable fibre tracts grow on the surface of the neural tube and therefore suggest that the intercellular clefts can play little role in the guidance of these fibres (Taylor & Roberts, 1983).
In view of this controversy we have investigated the first growth of ascending and descending fibre tracts in the Xenopus spinal cord using both scanning and transmission electron microscopy. The sources and targets of the fibres present at these early stages have been well characterized (Forehand & Farel, 1982; Nordlander, 1984; van Mier & Ten Donkelaar, 1984; Nordlander, Baden & Ryba, 1985). The earliest fibres come from intrinsic ‘primary’ sensory and motor neurones within the spinal cord. The first descending fibres from the brainstem come from the interstitial nucleus of the fasciculus longitudinalis medialis in the mesencephalon and arrive in the rostral spinal cord by stage 28. These fibres are soon joined by a more extensive reticulo and vestibulospinal input.
Because of the accumulating evidence for fibre-fibre recognition (Katz & Lasek, 1981; Bonhoffer & Huf, 1985) and the known tendency of fibres to fasciculate (Letourneau, 1982; Taylor & Roberts, 1983), we have concentrated on the guidance of this initial fibre outgrowth on the assumption that later added fibres will follow the fibre tracts laid down in early development. We have concentrated on looking for one possible sort of pathway guidance, a physical structure which can be identified at the light or electron microscopic level and shows some structural specialization such as rostrocaudal alignment which would suggest that it could play a role in the guidance of growing axons. Our methodology does not allow us to say anything about the more subtle aspects of the pathways along which the fibres grow. It is quite possible (for example) that a ‘pathway’ made of a chemical such as laminin, fibronectin or the neural cell adhesion molecule can act as a guiding factor and has been shown to do so in tissue culture (Rutishauser, 1984) and in vivo (Silver & Rutishauser, 1984).
We then ask the following questions, if such substrate pathways exist, what are their components and how are they fitted to perform the function of guiding axons? It has been proposed that aligned channels and spaces precede the growth of axons in the embryonic spinal cord (Nordlander & Singer, 1982a,b) and that these could guide the growth of the later descending axons. Does the substrate pathway take the form of such a gutter or aligned channel which guides axonal growth? Are the neuroepithelial end-feet of the developing nervous system morphologically adapted to axonal guidance?
MATERIALS AND METHODS
Twelve Xenopus laevis embryos of stages 22, 23 and 24 (Nieuwkoop & Faber, 1956), were fixed by immersion in a solution containing 4% glutaraldehyde, 2% paraformaldehyde, 1 % acrolein and 2·5 % dimethylsulphoxide in 0·05 M-sodium cacodylate with enough CaCl2 to make the solution 0-001 M for Ca2+. A further five embryos from each of stages 19 to 25 and a further thirty embryos of stages 26 to 54 were fixed in half-strength Karnovsky’s fixative (Nordlander & Singer, 1982a,b). From each stage between 19 and 25 a minimum of four embryos were processed for fight microscopy and transmission and scanning electron microscopy.
Light microscopy
Embryos fixed as above were rinsed in 0·05 M-sodium cacodylate buffer, postfixed for 1 h in 1 % osmium tetroxide in phosphate buffer, then dehydrated through graded ethanols and embedded in Araldite resin. The caudal half of each embryo was then sectioned serially at 1 μm either in the transverse or parasagittal plane. Sections prepared in this way were stained with toluidine blue. Photographs were taken of the neural tube using ×40 objectives at 10 μm intervals through the sets of serial sections obtained from each embryo. The negatives were projected to give a final magnification of ×1000, then drawings were made of the intercellular spaces at the lateral surface of the neural tube. From these drawings the intercellular spaces were reconstructed to determine the degree of alignment present.
Transmission electron microscopy
From blocks prepared for light microscopy, thin sections were cut for electron microscopy. Sections were stained with 2% aqueous uranyl acetate and lead citrate (Reynolds, 1963). Photographs were taken of the intercellular spaces close to the lateral surface of the neural tube.
Scanning electron microscopy
Embryos fixed as for light microscopy were pinned to Plasticine and the neural tube exposed using an electrolytically sharpened tungsten wire. Various orientations were prepared. The dissected embryos were dehydrated through ethanol, substituted in freon then dried using a critical-point dryer. The dried specimens were mounted on stubs and coated for examination in a scanning electron microscope.
RESULTS
Light microscopy
The intercellular spaces at the lateral margin of the neural tube were examined in sets of serial 1 μm sections taken from stage-23 embryos. The clefts were examined to see whether or not they were continuous along the tube, and whether or not they formed a tubular bed which could at later stages of development receive growing axons. In confirmation of earlier studies (Hayes & Roberts, 1973, 1974; Van Mier & Ten Donkelaar, 1984), examination of the sets of serial sections produced from the caudal half of the neural tube showed that development progressed in a rostrocaudal direction. While rostral-most sections examined in stage-20 embryos showed unspecialized intercellular clefts without axons, by stage 25 axons were present in rostral sections, and in progressively more caudal sections in later stages. Between stages 20 and 25, the intercellular clefts were narrow, except close to the surface, where they changed shape. Close to the surface many of the superficial cells were observed to expand to .form a tongue which passes sideways and partly overlaps the neighbouring cells. This type of surface formation was also seen in adjacent developing somites. At later stages the intercellular clefts were seen to contain a few, then many fibres. The space was observed to become wider at these later stages and the tongue processes of the surface cells become longer and thinner as the marginal zone is formed.
Reconstructions of serial sections taken from the neural tube caudal to axonal invasion demonstrate (Fig. 1) that although intercellular spaces are continuous, they are not necessarily well aligned along the neural tube. Axons following an intercellular cleft would be required to weave between successive cells. As more axons appear, the intercellular space widens in such a way that the spaces are in better alignment, and the axons are able to run straight.
A line drawing representing a set of eight serial sections taken 10 μm apart from the neural tube of a stage-22 embryo, caudal to axon invasion. I, lumen of neural tube; r, rostral; c, caudal. The intercellular clefts have been exaggerated for emphasis. At the surface the clefts can be seen to form a network, without good alignment in the long axis of the tube. Bar equals 20 μm.
A line drawing representing a set of eight serial sections taken 10 μm apart from the neural tube of a stage-22 embryo, caudal to axon invasion. I, lumen of neural tube; r, rostral; c, caudal. The intercellular clefts have been exaggerated for emphasis. At the surface the clefts can be seen to form a network, without good alignment in the long axis of the tube. Bar equals 20 μm.
Transmission electron microscopy
During the first few days of life the Xenopus embryo increases dramatically in size without any consumption of food. This implies that there is a great intake of water and a possible gradual reduction in the osmolality of the embryo’s tissues. The relationship between the osmolality of a tissue and the osmolality of the fixative required for good fixation is a complex one as the fixation itself changes the semipermeable properties of the tissue (Karnovsky, 1965). Most fixatives commonly employed for electron microscopy (including Karnovsky’s) are hyperosmotic when compared with the osmolality of the tissue to be fixed, we found that the half-strength Karnovsky’s fixative caused considerable shrinkage of the neuroepithelial cells and an artifactual widening of the intercellular clefts in the younger animals (up to stage 25). The opposing cell surfaces separated by the intercellular clefts were often of a complementary shape in spite of the large intervening gap. This picture, reminiscent of illustrations of ‘continental drift’ suggests that before fixation the two surfaces may have been in contact (Fig. 2). When different fixation protocols were investigated it was found that the acrolein fix described in the Materials and Methods section produced minimal shrinkage of the tissue (Figs 3, 4). All descriptions and illustrations of embryos younger than stage 25 that follow are drawn from the acrolein-fixed material.
An electron micrograph of the lateral wall of the neural tube of a stage-22 embryo fixed with half-strength Karnovsky’s fixative. The unspecialized clefts (arrows) which will eventually contain axons are covered over by tongue-like processes of the neuroepithelial cells. Note the large spaces between the cells and contrast with Figs 3,4 and 5 which show similar specimens fixed with the fixative containing acrolein. Bar equals 3 μm.
An electron micrograph of the lateral wall of the neural tube of a stage-22 embryo fixed with half-strength Karnovsky’s fixative. The unspecialized clefts (arrows) which will eventually contain axons are covered over by tongue-like processes of the neuroepithelial cells. Note the large spaces between the cells and contrast with Figs 3,4 and 5 which show similar specimens fixed with the fixative containing acrolein. Bar equals 3 μm.
An electron micrograph of the lateral wall of the neural tube of a stage-23 embryo, caudal to fibre development. The neuroepithelial cells of the neural tube are at the top of this micrograph. The lower cells are mesenchymal cells which surround the neural tube. The large amorphous black and grey structures are yolk (y) and lipid (/) droplets. With acrolein fixation (see Materials and Methods) the intercellular clefts (arrows) which will eventually contain axons are narrow and appear unspecialized. Note the varying amounts of glycogen (dark granules) in the cells. One cell in particular (g) is packed with glycogen. Bar equals 2μm.
An electron micrograph of the lateral wall of the neural tube of a stage-23 embryo, caudal to fibre development. The neuroepithelial cells of the neural tube are at the top of this micrograph. The lower cells are mesenchymal cells which surround the neural tube. The large amorphous black and grey structures are yolk (y) and lipid (/) droplets. With acrolein fixation (see Materials and Methods) the intercellular clefts (arrows) which will eventually contain axons are narrow and appear unspecialized. Note the varying amounts of glycogen (dark granules) in the cells. One cell in particular (g) is packed with glycogen. Bar equals 2μm.
A high-powered view of the intercellular cleft before the arrival of any axons taken from a stage-23 embryo fixed with acrolein-containing fixative. The margin of the neural tube is to the left. The cell membranes appear to interdigitate and there does not appear (with this fixation protocol) to be any pre-existing space to accommodate any later growing axons (arrow). As in Fig. 3 note the variation in glycogen content of the cells illustrated. Bar equals 1 μm.
A high-powered view of the intercellular cleft before the arrival of any axons taken from a stage-23 embryo fixed with acrolein-containing fixative. The margin of the neural tube is to the left. The cell membranes appear to interdigitate and there does not appear (with this fixation protocol) to be any pre-existing space to accommodate any later growing axons (arrow). As in Fig. 3 note the variation in glycogen content of the cells illustrated. Bar equals 1 μm.
By electron microscopy the intercellular spaces appear to be unspecialized (Figs 3, 4). The acrolein fix, but not the half-strength Karnovsky’s, preserved the glycogen well. The cells surrounding the first fibres often contained large quantities of glycogen and this could be used as a marker for the rapid localization of the first fibres in a large section (Fig. 5A). As seen by light microscopy, the cells forming the surface of the neural tube often have tongue-like expansions (Fig. 5B) which form a roof over the intercellular clefts. No specialized gutters or channels could be identified. By stage 22 single fibre-like structures which could be traced through several sections could be seen lying in the clefts (Fig. 5A,B). As more axons are added, the intercellular cleft becomes wider and the cells forming the boundaries become indented. Small processes from the bounding cells partly encircle the fibres, forming a bundle (Fig. 6). At no stage were similar changes seen in the bounding cells without the presence of axons. As the cleft expands to accommodate more axons the cells become narrow at this point, leaving only a thin cell layer forming the lateral surface (Fig. 7). On occasions the axons in the intercellular cleft are uncovered, except for a layer of basement membrane (Fig. 7).
(A) Electron micrograph showing the typical appearance of the first fibre-like structures observed in Xenopus neural tube. The margin of the neural tube runs from top right to bottom left. Note the absence of spaces between intercellular clefts around the border of the neural tube. This structure (arrowed and shown in greater detail in Fig. 5B) could be identified in several sections taken at points 10 pm apart along the neural tube. Stage-23 embryo fixed with acrolein-containing fixative. Bar equals 5 pm. (B) Detail from Fig. 5A showing the position of a presumed fibre (arrow) in an intercellular cleft. The displacement of the yolk droplet (y) shows that there may be some stretching of the section under the electron beam. This may contribute to an artifactual opening of the intercellular cleft around the ‘fibre’. Bar equals 0·5 μm.
(A) Electron micrograph showing the typical appearance of the first fibre-like structures observed in Xenopus neural tube. The margin of the neural tube runs from top right to bottom left. Note the absence of spaces between intercellular clefts around the border of the neural tube. This structure (arrowed and shown in greater detail in Fig. 5B) could be identified in several sections taken at points 10 pm apart along the neural tube. Stage-23 embryo fixed with acrolein-containing fixative. Bar equals 5 pm. (B) Detail from Fig. 5A showing the position of a presumed fibre (arrow) in an intercellular cleft. The displacement of the yolk droplet (y) shows that there may be some stretching of the section under the electron beam. This may contribute to an artifactual opening of the intercellular cleft around the ‘fibre’. Bar equals 0·5 μm.
Electron micrograph showing a small fibre bundle containing approximately 50 fibres, growth cones and glial processes. The bundle still follows the intercellular cleft closely. The surrounding neuroepithelial cells are undercut and the fibre bundle is covered by a tongue-like process from one of the cells (compare with Fig. 9A,B). This micrograph shows a section taken from the caudal part of a stage-33 embryo fixed with acrolein. Bar equals 1 μm.
Electron micrograph showing a small fibre bundle containing approximately 50 fibres, growth cones and glial processes. The bundle still follows the intercellular cleft closely. The surrounding neuroepithelial cells are undercut and the fibre bundle is covered by a tongue-like process from one of the cells (compare with Fig. 9A,B). This micrograph shows a section taken from the caudal part of a stage-33 embryo fixed with acrolein. Bar equals 1 μm.
Electron micrograph of the neural tube of a stage-32 Xenopus tadpole fixed with half-strength Karnovsky’s fixative. By this stage, in the more rostral part of the neural tube, a true ‘marginal zone’ has formed with numerous fibre bundles becoming contiguous and growing over the surface of the neuroepithelial cells in superficial galleries whose ‘roof is made from numerous thin and discontinuous processes of the neuroepithelial cells. Note that in places the fibres appear to be open to the surface of the neural tube, only separated from the surrounding mesenchymal cell (ni) by a thin covering of basement membrane (arrows). Bar equals 5 μm.
Electron micrograph of the neural tube of a stage-32 Xenopus tadpole fixed with half-strength Karnovsky’s fixative. By this stage, in the more rostral part of the neural tube, a true ‘marginal zone’ has formed with numerous fibre bundles becoming contiguous and growing over the surface of the neuroepithelial cells in superficial galleries whose ‘roof is made from numerous thin and discontinuous processes of the neuroepithelial cells. Note that in places the fibres appear to be open to the surface of the neural tube, only separated from the surrounding mesenchymal cell (ni) by a thin covering of basement membrane (arrows). Bar equals 5 μm.
Scanning electron microscopy
The cells making up the wall of the stage-22 neural tube are in general columnar with clear intercellular clefts between them. The clefts are not formed into gutters or tubes and are not well aligned. Examination of the lateral wall of the neural tube (Fig. 8) shows that the individual cell boundaries do not line up along the tube. In contrast, in the older specimens, the intercellular clefts contain more fibres and undercut the edges of the cells making up the lateral surface of the neural tube. Parallel fascicles of fibres can be seen running in the intercellular clefts just below the surface (Fig. 9A,B) and partly exposed (Fig. 10).
A scanning electron micrograph of the lateral surface of the neural tube of a stage-24 embryo. The rostrocaudal axis runs from right to left. The intercellular clefts can be seen to be non-aligned (arrows). Dorsally (towards the top of the micrograph) a bundle of fibres can be seen crossing over the surface of the neural tube (hollow arrows), ignoring the irregular cell boundaries. Bar equals 10 μm.
A scanning electron micrograph of the lateral surface of the neural tube of a stage-24 embryo. The rostrocaudal axis runs from right to left. The intercellular clefts can be seen to be non-aligned (arrows). Dorsally (towards the top of the micrograph) a bundle of fibres can be seen crossing over the surface of the neural tube (hollow arrows), ignoring the irregular cell boundaries. Bar equals 10 μm.
(A) A scanning electron micrograph of the wall of the neural tube of a stage-28 embryo rostral to the caudal-most point of fibre development. The intercellular clefts can be seen between the columnar cells. A fibre bundle can be seen at the lateral surface (arrow). The position from which Fig. 9B is taken is indicated by the box. Bar equals 10μm. (B) A magnified micrograph of the area enclosed in Fig. 9A. The fibre bundle can be seen, passing along close to the surface of a cell which is narrower at this point where the intercellular cleft is enlarged. The region of narrowing of the neuroepithelial cells (arrows) is shown diagrammatically in Fig. 11B and a similar structure is shown in the transmission electron micrograph shown in Fig. 6. Bar equals 1μm.
(A) A scanning electron micrograph of the wall of the neural tube of a stage-28 embryo rostral to the caudal-most point of fibre development. The intercellular clefts can be seen between the columnar cells. A fibre bundle can be seen at the lateral surface (arrow). The position from which Fig. 9B is taken is indicated by the box. Bar equals 10μm. (B) A magnified micrograph of the area enclosed in Fig. 9A. The fibre bundle can be seen, passing along close to the surface of a cell which is narrower at this point where the intercellular cleft is enlarged. The region of narrowing of the neuroepithelial cells (arrows) is shown diagrammatically in Fig. 11B and a similar structure is shown in the transmission electron micrograph shown in Fig. 6. Bar equals 1μm.
A scanning electron micrograph of the surface of a stage-30 embryo, rostral to the caudal-most extent of fibre development. Fibres can be seen (arrows) exposed at the surface (compare with Fig. 7).
A scanning electron micrograph of the surface of a stage-30 embryo, rostral to the caudal-most extent of fibre development. Fibres can be seen (arrows) exposed at the surface (compare with Fig. 7).
DISCUSSION
This study has not found any evidence for the presence of preformed structures in the form of aligned channels or gutters which could guide pioneering axons along the neural tube, other than unspecialized intercellular clefts. We have observed that, when the first fibres grow along the spinal cord, the intercellular clefts are narrow, tortuous and branching as they follow the boundaries of the cells forming the surface of the neural tube. Although the size of the intercellular clefts may be expected to vary with the species, stage of development, and fixation, it is unlikely that these factors could have had much effect on the observed lack of alignment of these spaces.
From these observations it appears that axons in the developing spinal cord initially must take a twisting path between cells, and that this pathway becomes aligned through addition of fibres and further development of the neuroepithelial cells (Fig. 11).
A line dravμng showing the changes which take place in the intercellular spaces and the cells bounding them at the surface of the neural tube, as axons grow along the tube. A view of the lateral surface is shown above, while its appearance in transverse section is shown below. Initially the intercellular clefts are narrow and not well aligned, so that a fibre growing along the tube in the intercellular cleft would have to wind its way between the cells as suggested by the arrows in (A). Later, as fibres are added, the cells become undercut, allowing more fibres to be added as in (B). In yet later stages (C) large numbers of fibres form bundles, and the neuroepithelial cells have been reduced to a narrow stalk, and a thin surface plaque.
A line dravμng showing the changes which take place in the intercellular spaces and the cells bounding them at the surface of the neural tube, as axons grow along the tube. A view of the lateral surface is shown above, while its appearance in transverse section is shown below. Initially the intercellular clefts are narrow and not well aligned, so that a fibre growing along the tube in the intercellular cleft would have to wind its way between the cells as suggested by the arrows in (A). Later, as fibres are added, the cells become undercut, allowing more fibres to be added as in (B). In yet later stages (C) large numbers of fibres form bundles, and the neuroepithelial cells have been reduced to a narrow stalk, and a thin surface plaque.
Many authors have suggested that some form of physical guidance is present in the nervous system to guide both central and peripheral axons to their targets (Singer, Nordlander & Egar, 1979; Bentley & Keshishian, 1982). Kevetter & Lasek (1982) have claimed that stereotyped axon tracts between specific neuronal populations are a critical requirement in the nervous system. While there is some support for this proposal it is not universally accepted.
The evidence for the existence of aligned substrate pathways is that fibre tracts in the CNS are found at consistent locations. Transplanted neurones send their axons through the CNS via established routes (Katz & Lasek, 1978, 1981). This latter evidence, however, is weak since, as axons have a natural tendency to fasciculate with existing fibres and form bundles (Weiss, 1934; Young, 1942; Nakai, 1960; Letourneau, 1982; Rutishauser, 1984), it is necessary to demonstrate these substrate pathways by transplanting cells into a site where their axons will be the first to invade the neural tissue rather than by transplanting them into sites where fibre tracts are already present.
Nordlander & Singer (1982a,b) have claimed that spaces precede axons in Xenopus spinal cord, and that these spaces, actually intercellular clefts, have some special significance for growing axons. In urodeles the intercellular spaces are claimed to be aligned longitudinally and thus to offer a guidance channel to growing axons (Singer, Nordlander & Egar, 1979). This study has found no specialization of intercellular spaces and no evidence for alignment of the spaces in Xenopus. This difference in observations may depend in part on the different fixatives used. When we used the half-strength Karnovsky’s fixative employed by Nordlander & Singer (1982a,b) we observed considerable shrinkage of the neuroepithelial cells and an increase in size of the intercellular clefts (Fig. 2). Similar shrinkage was observed when chick retina was fixed with hypertonic fixatives (Krayanek & Goldberg, 1981). An increase in the width of the intercellular spaces would make it appear as though the fibres growing through them could take a fairly straight path, even if the spaces were not aligned. Krayanek & Goldberg also point out that such shrinkage may reveal underlying patterns of cell-to-cell adhesion so that clefts containing fibres may widen further than those where neuroepithelial cell can contact neuroepithelial cell directly.
While pre-existing channels have been described in situations other than newt spinal cord, such as the optic nerve (Silver & Sidman, 1980), similar preferential guidance channels were not observed in the developing rat corpus callosum (Valentino & Jones, 1982), the developing corticospinal tract in the foetal and neonatal rat (Schreyer & Jones, 1980), or in the regenerating spinal cord in Xenopus tadpoles (Michel & Reier, 1979).
There is no evidence for a universal distribution of anatomical structures which provide guidance cues in the nervous system. They appear to be available to growing axons in some situations, and absent in others. No system has been found where such morphological guidance is obvious and essential for correct fibre growth, so that, like impulse activity (Harris, 1984), morphological guidance may be only one of many factors involved in axonal guidance, but may not itself be essential. From our results we can conclude that in the neural tube of Xenopus if any preformed pathways for the guidance of later growing axons exist they must be of a more subtle nature than the gross anatomical features which we have examined.
From recent studies it appears that extracellular matrix components such as fibronectin, laminin (Rogers, Letourneau, Palm, McCarthy & Furcht, 1983), and neural cell adhesion molecule (Silver & Rutishauser, 1984), may play a role in axonal guidance as important as more gross features of neuroepithelial cell morphology. Such chemicals may be laid down in precise patterns which could act as preformed pathways to guide the later fibre growth.
ACKNOWLEDGEMENTS
We would like to thank G. J. Brown, J. L. James and A. Tracey for their technical assistance, the Nuffield Foundation, NIH grant number 11066 and the Medical Research Council for their financial support.