The growth of optic axons towards experimentally rotated tecta has been studied. In stage 24/25 embryos, a piece of the dorsal neural tube, containing the dorsal midbrain rudiment, was rotated through 180°. At later stages of development, the pathways of growing optic axons were investigated by labelling with either horseradish peroxidase or fluorescent dye. It is shown that retinal ganglion cell axons followed well-defined pathways, in spite of the abnormal structure of the brain, and were able to locate displaced tecta. This directed outgrowth of retinal axons in the optic tracts appears to be related either to the tectum or to some other component included in the graft operations. In tadpoles in which the midbrain rudiment was removed, optic axons still followed the normal course of the optic tract. This observation argues against long-range target attraction as being essential in guiding growing retinal axons towards the tectum. An alternative axon guidance mechanism, selective fasciculation, is discussed as a possible alternative to exaplain the directed axon outgrowth which occurs in both the normal and in these experimentally manipulated tadpoles.

Studies concerning directed nerve outgrowth and the formation of specific nerve connections have suggested numerous control mechanisms (Sperry, 1963; Macagno, 1978; Goodman et al. 1984; Lumsden and Davies, 1983). In general these suggestions form partial answers to the question of axonal guidance; it is increasingly evident that in most systems a multiplicity of factors is acting or able to act to direct growing nerve fibres. The dependence of a nerve fibre on any such cues at any time in its outgrowth will relate to the state of development of the animal and hence the immediate environment of the growing nerve fibre. Further, it may also depend on intrinsic factors within that nerve cell. Perhaps of more importance to experimental studies, the use by an axon of any particular guidance system may be altered by experimental interference, thus requiring the axon to place an increased reliance on what might ordinarily be considered a secondary system. It appears that there normally exists a range of strategies that can be utilized in the ‘wiring up’ of a nervous system.

Recently the retinotectal system of the frog Xenopus has been favoured for investigation of both the guidance of axons and the initial stages of the formation of specific connections (Holt and Harris, 1983; Holt, 1984; Sakaguchi and Murphey, 1985; O’Rourke and Fraser, 1986; Harris, 1986; 1989; Harris et al. 1987). In a series of experiments, Harris showed that optic axons from grafted eyes, which entered the brain at a novel site, still located the optic tectum. One mechanism proposed by Harris for this directed outgrowth was target attraction. An alternative was that cues for directing axons towards their correct targets were based upon positional information.

Preliminary findings from this study (Taylor, 1987) argued that the target was not necessary for directed growth of optic axons and that some form of local cue must reside within the pathway. Recently, Harris (1989) has provided experimental evidence for the existence of local cues and we have suggested a possible basis for such cues in retinal axon guidance (Easter and Taylor, 1989).

In order to examine further hypotheses of axon guidance, and as a part of a series of investigations into the initial development of the Xenopus retinotectal projection, I have performed embryonic tectal rotation and translocation operations in which the target has been displaced to a novel site. At later stages of development horseradish peroxidase (HRP) or the carbocyanine dye 1,1’ dioctadecyl 3,3,3’,3’-tetra-methylindocarbocyanine perchlorate (Dil) have been used to label the retinal ganglion cell axons. These observations confirm that retinal axons have a remarkable ability to locate the tectum, even in such grossly disturbed brains. The axons grow along specific routes within the midbrain forming a range of abnormally positioned, yet well defined, optic tracts. In the absence of the tectum, optic axons still follow the normal course of the optic pathway. These results are discussed with reference to ideas of target attraction and local pathway cues as axon guidance mechanisms.

Dil labelling

Embryos were prepared as described for HRP labelling with the lens removed. A minute crystal of Dil (Molecular Probes) was inserted into the vitreous cavity and allowed to ‘heal’ into place for approximately 10 min. The animals were then fixed in 10% buffered neutral formalin and kept for 5−7 days at 4°C. After washing in buffer the brains were dissected free and viewed using either air or water immersion optics with incident fluorescent light (540 nm). Photographs were taken at 1600 ASA using either Ilford XP1 monochrome or Koda-chrome 400 ASA slide film. Brains were temporarily stored in phosphate buffer at 4°C in the dark, to minimize the progressive loss of specific labelling.

Embryonic operations

Pairs of adult frogs were induced to lay eggs by injection with human chorionic gonadotrophin (Intervet). Viable eggs were sorted and reared in 10% Niu-Twitty saline at either 16°C or 23°C in an incubator on a 12h light/dark cycle. Embryos at stages 24/25 (Nieuwkoop and Faber, 1967) were mechanically removed from their jelly coats and anaesthetized in a 1:4500 ethyl-w-aminobenzoate solution in 66 % Niu-Twitty solution containing twice the normal levels of Ca2+ and Mg24”. The embryos were then inserted into small depressions in a wax dish containing the same anaesthetic solution. All operations were performed using sharpened tungsten needles. The following operations were carried out (Fig. 1):

Fig. 1.

The four types of embryonic operation performed in these experiments. (A) Rotation of the anterior part of the dorsal neural tube at stage 24/25, resulting in the rostrocaudal inversion and anterior displacement of the tectum and dorsal midbrain structures (TROT). (B) Sham operation in which the anterior part of the dorsal neural tube was excised and then replaced without rotation. This acts as a control for the rotation operations. (C) Removal of the anterior part of the dorsal neural tube at stage 24/25 resulting in deletion of dorsal midbrain structures. (D) Removal of the tectal primordium at stage 34/35 leaving other midbrain structures relatively undisturbed. Bar 100μm.

Fig. 1.

The four types of embryonic operation performed in these experiments. (A) Rotation of the anterior part of the dorsal neural tube at stage 24/25, resulting in the rostrocaudal inversion and anterior displacement of the tectum and dorsal midbrain structures (TROT). (B) Sham operation in which the anterior part of the dorsal neural tube was excised and then replaced without rotation. This acts as a control for the rotation operations. (C) Removal of the anterior part of the dorsal neural tube at stage 24/25 resulting in deletion of dorsal midbrain structures. (D) Removal of the tectal primordium at stage 34/35 leaving other midbrain structures relatively undisturbed. Bar 100μm.

i) Rotation; in which a piece of tissue containing the tectal rudiment was rotated, displacing the subsequently developing tectum to a more anterior position within the brain (TROT).

ii) Sham operations; in which a similar portion of tissue was excised but reimplanted without rotation.

iii) Tectal removal; in which the same piece of tissue as was used in the graft operations was removed.

The grafts were held in place with small glass fragments for a few minutes, then the embryos were left for approximately 30 min to allow for healing. Operated embryos were then returned to 10 % saline for rearing to the appropriate stage for labelling.

HRP labelling

Embryos were staged, anaesthetized in 1:4500 ethyl-m-aminobenzoate in 66% Niu-Twitty saline and placed left side up on a piece of damp tissue. Tungsten needles were used to first cut through the cornea and then to lift out the lens. A fine micropipette filled with a solution of 10 % HRP (Boehringer, Grade I) with 1 % Nonidet P40 (BDH) in 50 % Niu-Twitty saline was lowered into the vitreous cavity and the HRP was allowed to passively diffuse into the eye cup for a few seconds. The pipette was removed and the animal left in air for a further 5 min, before being returned to 10% saline. As an alternative labelling technique, which proved as successful, the eye was prepared as described and a small piece of recrystallized HRP/nonidet solution was inserted into the vitreous. After 10min, the animals were reanaesthetized, then fixed in 2% glutaraldehyde in 0.1M-phosphate buffer pH 7.2, for between 1 and 12 h. Following a wash in buffer, the brains were dissected free and cleared of surface adhering cells including melanocytes. The HRP was visualized by reacting according to the Cobalt-DAB method of Adams (1977). Brains with labelled optic projections were dehydrated in ethanol, cleared using methyl salicylate, drawn with a camera lucida and, in some cases, photographed.

The time course and normal pattern of retinotectal development

The results from the control preparations made in this study are in general agreement with previous reports (Fig. 2). For further description of the normal development of the optic pathways, see Easter and Taylor (1989). Briefly, axons were seen passing from the optic stalk to the ventral part of the diencephalon at stage 34/35. At the midline, the growing axons crossed the projection from the other eye, forming the chiasma, and passed caudal to the other optic stalk. In the optic tracts, axons travelled as a coherent group, following a predictable dorsocaudal trajectory. From stage 39, terminal arbors were seen extending over the region of the developing tectum. After application of Dil, axons, their growth cones bearing fine filopodia and, at later stages, their terminal arbors all showed intense labelling (Fig. 3). The results obtained using this technique are qualitatively similar to those obtained using the HRP method.

Fig. 2.

Stage 39 control preparation in which the optic axons have been labelled with HRP. The optic tract passes from the ventral part of the midbrain, caudal to the other optic stalk, then dorsal to the midregion of the diencephalon where it veers slightly caudal to head directly to the tectum (t). Branches of the retinal axons are seen in the midregion of the diencephalon (arrowhead) and over the lateral part of the tectum. The tectum lies within the dorsolateral bulge of the midbrain rostral to the constriction which demarcates the midbrain/hindbrain junction. Rostral is to the right. Bar 100μm.

Fig. 2.

Stage 39 control preparation in which the optic axons have been labelled with HRP. The optic tract passes from the ventral part of the midbrain, caudal to the other optic stalk, then dorsal to the midregion of the diencephalon where it veers slightly caudal to head directly to the tectum (t). Branches of the retinal axons are seen in the midregion of the diencephalon (arrowhead) and over the lateral part of the tectum. The tectum lies within the dorsolateral bulge of the midbrain rostral to the constriction which demarcates the midbrain/hindbrain junction. Rostral is to the right. Bar 100μm.

Fig. 3.

The normal retinotectal projection of a stage 41 tadpole whose right eye has been labelled with Dil. Axons in the left optic tract have been heavily labelled and can be seen extending to the tectal primordium (t), where they have formed terminal arbors. Arborisation can also be seen in the midregion of the diencephalon (arrowhead) in the developing thalamic visual centres. Individual axons, growth cones and the fine branches of the arborizing axons are all identifiable. Bar 100μm.

Fig. 3.

The normal retinotectal projection of a stage 41 tadpole whose right eye has been labelled with Dil. Axons in the left optic tract have been heavily labelled and can be seen extending to the tectal primordium (t), where they have formed terminal arbors. Arborisation can also be seen in the midregion of the diencephalon (arrowhead) in the developing thalamic visual centres. Individual axons, growth cones and the fine branches of the arborizing axons are all identifiable. Bar 100μm.

General anatomy of the brains of rotated preparations

In the most easily interpreted cases, the position of the pineal gland proved to be a convenient marker. If the intended anterior rotation was successful then the tissue that would have formed at the junction between the midbrain and the hindbrain should have been moved to a more anterior position; whilst the pineal body, which marked the rostral edge of the grafted tissue, should have come to occupy a position adjacent to the hindbrain. In many cases, this displacement of the pineal was obvious, whilst the transposition of the tectal precursor tended to be recognized by an abnormal bulge in the most rostral neural tissue (Fig. 4). At the edges of the graft, the wound healing appeared to be excellent with no obvious gross discontinuities. However, at the graft borders, the neural tissue tended to be constricted forming a ‘seam’, delineating the graft’s position. In no case was there evidence of derotation of the graft or of degeneration of the operated tissue and, as far as could be discerned, there was no large-scale regeneration of the normally adjacent structures at the graft borders. Naturally, without extensive cell marking experiments these findings are not conclusive, but the evidence suggests that the grafts healed into place and formed a continuum with the surrounding neural tissue.

Fig. 4.

Drawing to illustrate the effect of rotation on the general morphology of the brain. In the rotated brain, the pineal body (p) serves as a useful marker, lying caudal, adjacent to the hindbrain. The hatched region indicates the normal and rotated position of the tectal precursor tissue.

Fig. 4.

Drawing to illustrate the effect of rotation on the general morphology of the brain. In the rotated brain, the pineal body (p) serves as a useful marker, lying caudal, adjacent to the hindbrain. The hatched region indicates the normal and rotated position of the tectal precursor tissue.

Pathways of optic axons

The results are based upon 193 operations of which 128 were sufficiently well labelled to allow the distribution of optic axons to be described. The results from the rotations (90 cases), can be placed in three main categories (Table 1); optic pathways that were directed in an abnormal anterior direction; pathways that were split, with one group of optic axons heading in an anterior direction, but with a second more caudodorsally directed group; and finally those that appeared to follow a grossly normal pattern. In addition, a minority of pathways were more abnormal and have been categorized as ‘other’.

Table 1.

The numbers of preparations in each category of results

The numbers of preparations in each category of results
The numbers of preparations in each category of results

Anterior pathways

In animals with an anterior rotated graft labelled at stages 38−41, the optic axons were found to have entered the ventral part of the brain and passed through the optic chiasma in a normal fashion (Fig. 5A). In the ventral part of the contralateral optic tract, the axons exhibited a similar tendency to form a well-defined, close-knit grouping, and grew dorsally, as was found for axons in control preparations. Beyond the midpart of the optic tract, the axons turned in an abnormal anterior direction and headed straight for the rostral region of the graft tissue, the position to which the tectal precursor tissue would have been expected to have been moved. Although axons generally remained as a coherent grouping, there were greater numbers of ‘stray’ axons than in normal preparations.

Fig. 5.

(A) An anterior tract shown in micrograph and camera-lucida drawing. The course of the optic axons can be seen to be strongly directed in an abnormal anterior direction. The morphology of the brain suggests that the axons are invading the transposed tectal precursor tissue which now lies rostral at the forebrain/midbrain border. The pineal body has been displaced caudally to lie adjacent to the hindbrain. (B) A preparation in which the optic axons form a split tract. The axons course in a normal trajectory in the ventral part of the tract but then split into two groups travelling in opposite directions. The anterior group invades the predicted position of the translocated tectal tissue, whilst the caudal group grows towards the position at which a normal tectum would be located. In this preparation, the morphology of the brain suggests that, in addition to the rotated tectal tissue, a normal tectum is developing rostral to the hindbrain. (C) A preparation in which the majority of axons head anteriorly towards the rotated tectum, but a minor group courses caudally. It is unclear whether the caudally growing axons would have turned dorsal towards the position of the normal tectum, or passed caudal into the hindbrain. (D) A ‘normal’ projection in a rotated brain. The optic axons course caudal to the graft (dashed lines) which has left the tectal precursor tissue undisturbed. In all cases rostral is to the right. Abbreviations: t, tectal precursor; t’ rotated tectal precursor; p, pineal body; fb, forebrain; hb, hindbrain. Bar 100 μm.

Fig. 5.

(A) An anterior tract shown in micrograph and camera-lucida drawing. The course of the optic axons can be seen to be strongly directed in an abnormal anterior direction. The morphology of the brain suggests that the axons are invading the transposed tectal precursor tissue which now lies rostral at the forebrain/midbrain border. The pineal body has been displaced caudally to lie adjacent to the hindbrain. (B) A preparation in which the optic axons form a split tract. The axons course in a normal trajectory in the ventral part of the tract but then split into two groups travelling in opposite directions. The anterior group invades the predicted position of the translocated tectal tissue, whilst the caudal group grows towards the position at which a normal tectum would be located. In this preparation, the morphology of the brain suggests that, in addition to the rotated tectal tissue, a normal tectum is developing rostral to the hindbrain. (C) A preparation in which the majority of axons head anteriorly towards the rotated tectum, but a minor group courses caudally. It is unclear whether the caudally growing axons would have turned dorsal towards the position of the normal tectum, or passed caudal into the hindbrain. (D) A ‘normal’ projection in a rotated brain. The optic axons course caudal to the graft (dashed lines) which has left the tectal precursor tissue undisturbed. In all cases rostral is to the right. Abbreviations: t, tectal precursor; t’ rotated tectal precursor; p, pineal body; fb, forebrain; hb, hindbrain. Bar 100 μm.

Split tracts

In the 19 cases where split tracts occurred (Fig. 5A,C), it was again found that the axons in the chiasma region and in the ventral part of the optic tract followed normal pathways. From the midregion of the tract, a subgroup of axons headed in an abnormal anterior direction into the rostral part of the graft, towards the predicted position of the tectal rudiment. A second group of axons also departed from the main body of the tract, heading in a more caudal direction towards the normal position of the optic tectum in control preparations. In some cases, this caudal branch of the optic tract occupied a similar position to the normal tract, whilst in others it appeared to follow the graft border. In terms of the variability of the result, minor and major branches were commonly found, with either the majority of axons heading anteriorly whilst a diminished population passed caudally, or vice versa (Fig. 5C).

Normal tracts

In fourteen cases, a ‘normal’ optic tract was formed. In the majority of these preparations, the axons followed the normal position of the optic pathway but, in some cases, the axons deviated around the graft tissue (Fig. 5D). It is assumed that in these normal cases the operation failed to include the tectal rudiment, being too rostral, and thereby leaving the normal target undisturbed. This assumption is supported by those cases where the axons obviously deviated around a more rostral graft area.

Sham operations

In these experiments, the tissue including the tectal primordium was excised and then reimplanted in normal orientation. This control should test the effects of the act of excision and the possible role of the wound at the graft border in interference with the normal course of optic axons. In all 19 animals in which the retinal ganglion cell axons were successfully labelled, the optic pathway appeared to be normal (Fig. 6A,B). The only detectable difference was a tendency for the axons to be less tightly grouped in the optic tract and for there to be increased numbers of ‘stray’ axons, especially those projecting caudal into the spinal tracts.

Fig. 6.

(A) A Dil-labelled preparation at stage 40, which had received a sham operation in embryo, viewed using fluorescent and bright-field optics. There is a greater tendency for axons to diverge from the main body of the tract, but the majority of optic axons follow a relatively normal course to the tectum. In the mid-diencephalon, arbors can be seen at the thalamic visual centres. Bar 100/rm. (B) A stage 42 HRP labelled projection after a sham operation, showing the virtually normal appearance of the projection to the tectum. Rostral to the right. Bar 100μm.

Fig. 6.

(A) A Dil-labelled preparation at stage 40, which had received a sham operation in embryo, viewed using fluorescent and bright-field optics. There is a greater tendency for axons to diverge from the main body of the tract, but the majority of optic axons follow a relatively normal course to the tectum. In the mid-diencephalon, arbors can be seen at the thalamic visual centres. Bar 100/rm. (B) A stage 42 HRP labelled projection after a sham operation, showing the virtually normal appearance of the projection to the tectum. Rostral to the right. Bar 100μm.

Tectal removals

Two types of tectal removal operation were performed; in the first, at stage 24/25, a piece of tissue, similar to that described for the above graft operations, was removed and discarded; in the second approach, the operation was performed at stage 34/35, when the region of the presumptive tectal tissue itself could be identified and ablated. Information from control preparations suggests that, at stage 34/35, growing optic axons are in the region of the optic chiasma and would not have been directly affected by the surgery.

Early ablations

Normal optic tracts were found with axons passing dorsocaudally around a deletion. In these cases, it would appear that the initial operation was too rostral, preserving the tectal rudiment and deleting tissue anterior to the optic pathway (Fig. 7A). In the second class of results, the optic axons passed dorsally up the optic tract to the deletion border where they abruptly turned caudally, entered the hindbrain and then passed into the spinal tracts. The third and most remarkable class of results showed optic axons looping around the diencephalon (Fig. 8). The optic axons entered the contralateral optic tract as normal, passed dorsally towards the deletion border, then continued dorsally recrossing the midline. Once over the dorsal midline, they entered the ipsilateral optic tract passing ventrally back down into the chiasma. In this manner, a closed loop was formed around the diencephalon.

Fig. 7.

(A) A brain showing the obvious gap (dashed lines) left after dorsal neural tube removal at stage 24/25. The operation has not included the tectum or the region of the diencephalon through which the optic tract normally forms. In this case, the labelled optic axons follow a relatively normal pathway to the tectum. Bar 100pm. (B) A caudally sweeping optic projection in a stage 40 brain which had the tectal primordium removed at stage 34/35. The optic axons follow a relatively normal pathway in the diencephalon, but at the deletion border (dashed lines) they turn caudal, entering the hindbrain and continuing into the spinal cord. At the deletion border a few axons have passed dorsally towards the midline. Bar 100 μm.

Fig. 7.

(A) A brain showing the obvious gap (dashed lines) left after dorsal neural tube removal at stage 24/25. The operation has not included the tectum or the region of the diencephalon through which the optic tract normally forms. In this case, the labelled optic axons follow a relatively normal pathway to the tectum. Bar 100pm. (B) A caudally sweeping optic projection in a stage 40 brain which had the tectal primordium removed at stage 34/35. The optic axons follow a relatively normal pathway in the diencephalon, but at the deletion border (dashed lines) they turn caudal, entering the hindbrain and continuing into the spinal cord. At the deletion border a few axons have passed dorsally towards the midline. Bar 100 μm.

Fig. 8.

Two brains in which deletion of the dorsal midbrain has resulted in a looping projection, one shown in micrographs and the other in camera lucida drawings. Rostral is to the right. (A and D) The contralateral, right, optic tracts. Axons form a coherent group, which is directed dorsocaudally, following the normal route of the optic tract. At the border of the deletion they continue dorsally over the midline. (B and E) Dorsal views, showing the axons passing a coherent group over the midline. The caudal border of both of these deletions appears to be cerebellum (c), indicating that the operation has removed the entire tectal precursor tissue. (C and F) Ipsilateral optic tracts. The axons follow a well defined, ventral trajectory through the ipsilateral diencephalon returning to their entry point adjacent to the optic stalk (os). Bar 100μm.

Fig. 8.

Two brains in which deletion of the dorsal midbrain has resulted in a looping projection, one shown in micrographs and the other in camera lucida drawings. Rostral is to the right. (A and D) The contralateral, right, optic tracts. Axons form a coherent group, which is directed dorsocaudally, following the normal route of the optic tract. At the border of the deletion they continue dorsally over the midline. (B and E) Dorsal views, showing the axons passing a coherent group over the midline. The caudal border of both of these deletions appears to be cerebellum (c), indicating that the operation has removed the entire tectal precursor tissue. (C and F) Ipsilateral optic tracts. The axons follow a well defined, ventral trajectory through the ipsilateral diencephalon returning to their entry point adjacent to the optic stalk (os). Bar 100μm.

Stage 34/35 ablations

Axons followed a normal route through the ventral part of the optic pathway, growing dorsocaudally towards the deletion border. At this position, the axons formed caudally oriented tracts which swept through the hindbrain and into the spinal cord (Fig. 7B). Unfortunately these axons could only be followed for 500 μm. These axons did not appear to be labelled to their tips, so it is assumed that they continued caudally for an even greater distance. This limitation was unavoidable using a labelling protocol with a short survival time. In a few cases, a minor contingent of axons left the main caudally sweeping tract and passed dorsally over the midline in the region of the deletion. These axons formed a similar pathway to the looping retinal axons described above, but were fewer in number. As in the early deletions, some cases of normal pathways were found, and again it was assumed that these represented misjudged operations, performed too far rostrally.

Developmental considerations

In order to make certain that the patterns of optic tract observed in the stage 38−41 preparations did not result from an initial random outgrowth, followed by the retraction of certain axons, we have examined similar anterior rotation preparations at stages 35−36. In these cases, axons tipped with large growth cones were seen at mid-diencephalic levels. The axons appeared to be more widespread than those in controls at this stage of development. However, the majority of axons appeared to be heading straight towards the target region, suggesting that the outgrowth was directed from its earliest stages (Fig. 9).

Fig. 9.

Micrograph of a whole mounted brain and camera lucida drawing to show the anterior directed outgrowth of the earliest arriving optic axons in the stage 35/36 brain. The axons, tipped by growth cones are all heading anteriorly towards the predicted position of the rotated tectal tissue (t’). Rostral is to the right. Bar 100 μm.

Fig. 9.

Micrograph of a whole mounted brain and camera lucida drawing to show the anterior directed outgrowth of the earliest arriving optic axons in the stage 35/36 brain. The axons, tipped by growth cones are all heading anteriorly towards the predicted position of the rotated tectal tissue (t’). Rostral is to the right. Bar 100 μm.

Target identity

In all the above experiments, I have made the assumption that the grafted tissue remains in position and differentiates normally to form the same tissue that it would have formed if left in situ. Indeed, the identity of the graft can be inferred from the morphology of the brain (the position of the pineal organ and a more rostral enlargement of the neural tissue comparable to that normally known to form the tectum), and from the presence of constrictions delineating the graft borders. In most cases, the axons can be seen to have crossed such borders and to have entered the graft tissue. However, it has also been noted that in some cases subgroups of axons appear to track the graft borders, suggesting that this might form some physical constraint on certain of the outgrowing axons. Whilst there is no evidence in any of the preparations to suggest that the assumption of normal development of the graft is invalid, it was decided that more positive evidence was required.

To confirm that the development of anterior grafts was as expected, embryos were raised to stage 46, when the tectum was clearly identifiable in whole-mount preparations. HRP labelling of control animals revealed the details of the normal optic pathway at this stage. In operated animals, a similar clear identification of the tectum could be made but, in these cases, the tectum was found to be in reversed orientation. Fig. 10A,B shows both control and rotated preparations in dorsal orientation, before HRP labelling. In control preparations with HRP-labelled tracts, a clear laterally positioned region of axon termination could be seen, which extended caudally away from the entry point of the optic axons (Fig. 10C). In the rotated tecta, there was a similar region of optic fibre arborization, but this was found to extend rostrally away from the optic tract (Fig. 10D). In the preparation shown, the reversed polarity and rostrally directed optic innervation are clear. The more dorsal direction taken by the optic tract probably reflects the caudal displacement of midbrain structures which occurs as the forebrain enlarges during development (Easter and Taylor, 1989). From the eight preparations showing this result, we can conclude that the rotation operation does include the tectal precursor tissue and that this subsequently develops into a ‘normal’ tectum.

Fig. 10.

(A) A normal stage 46 brain viewed dorsally. The forebrain is to the right, the hindbrain to the left with the cerebellum, marking its rostral border. On each side the tectum bulges laterally, with the ventricle forming a triangular, bowing, expansion within the brain, at the apex of which lies the pineal body. (B) A similar view to that in A of a rotated brain. The characteristic shape of the tecta and the ventricular expansion can be seen to have been reversed by the operation. (C) Lateral view of a normal brain at stage 46 showing the course of the HRP-labelled optic axons. The optic tract passes dorsocaudal through the diencephalon towards the rostral pole of the tectum. The terminal arbors of the retinal axons are distributed from rostral to caudal over the surface of the tectum. (D) Lateral view of a stage 46 brain with a rotation. The optic tract passes dorsal rather than caudal and innervates the caudal pole of the tectum. The terminal arbors of the optic axons extend from caudal to rostral corresponding with the reversed polarity of the tectum. The more dorsal trajectory of the optic tract (compared to the anterior direction seen at earlier stages), reflects the passive caudal displacement of the optic tract, which occurs as the forebrain enlarges. Rostral is to the right. Abbreviations: fb, forebrain; hb, hindbrain; p, pineal body; t, tectum, t’, rotated tectum; c, cerebellum; v, ventricle. Bars 100 μm.

Fig. 10.

(A) A normal stage 46 brain viewed dorsally. The forebrain is to the right, the hindbrain to the left with the cerebellum, marking its rostral border. On each side the tectum bulges laterally, with the ventricle forming a triangular, bowing, expansion within the brain, at the apex of which lies the pineal body. (B) A similar view to that in A of a rotated brain. The characteristic shape of the tecta and the ventricular expansion can be seen to have been reversed by the operation. (C) Lateral view of a normal brain at stage 46 showing the course of the HRP-labelled optic axons. The optic tract passes dorsocaudal through the diencephalon towards the rostral pole of the tectum. The terminal arbors of the retinal axons are distributed from rostral to caudal over the surface of the tectum. (D) Lateral view of a stage 46 brain with a rotation. The optic tract passes dorsal rather than caudal and innervates the caudal pole of the tectum. The terminal arbors of the optic axons extend from caudal to rostral corresponding with the reversed polarity of the tectum. The more dorsal trajectory of the optic tract (compared to the anterior direction seen at earlier stages), reflects the passive caudal displacement of the optic tract, which occurs as the forebrain enlarges. Rostral is to the right. Abbreviations: fb, forebrain; hb, hindbrain; p, pineal body; t, tectum, t’, rotated tectum; c, cerebellum; v, ventricle. Bars 100 μm.

In the present study, I have shown that growing retinal axons locate the tectum when it has been displaced to an abnormal position within the brain. Before considering what these results might tell us about axon guidance, the assumptions made regarding the tissue included in the operations and its subsequent development will be discussed. When the tectal rudiment was rotated and translocated rostrally, the growing optic axons formed an abnormal anteriorly directed tract. In cases where ‘split’ optic tracts were formed, it was assumed that the operation had left part of the tectal precursor intact, thereby forming two targets for the growing retinal axons. In cases where the pathway was ‘normal’, the operation was interpreted as being too far rostral and not including the tectal primordium. Given these differing results it is apparent that in spite of every attempt to make the operations reproducible in each batch of operations there were variations. This variation reflects the minute size of the embryonic Xenopus brain and the small tissue masses used in the grafts (30-0 cell diameters, see Chung and Cooke, 1978). Unfortunately, the precise nature of the graft tissue can only be determined when tissue differentiation has taken place and, during the period of initial outgrowth of retinal ganglion cell axons, the tectum has not differentiated. At these stages of development the identity of the tectum can only be inferred from the gross morphology of the brain and by its innervation by optic fibres. The tectal precursor cannot be identified by any morphological or histological means. In support of these assumptions concerning the identity of the tectum, operated embryos were reared to a stage at which positive identification of the tectum was possible.

The question as to what guides growing nerve fibres to their targets has been asked on numerous occasions and has produced a range of partial answers. On current evidence it would appear that there are no simple answers; in almost every case multiple factors are likely to be involved.

One suggested mechanism involved in axon guidance, which is pertinent to the results presented here, is target-derived chemoattraction. This idea was suggested almost a century ago to explain axon guidance in development (Ramon y Cajal, 1892) and in specific axon regeneration (Langley, 1895; 1897). Although there are several examples of directed nerve outgrowth that have been explained by chemoattraction, there is only one molecule that has been proposed as having chemoattractant properties, nerve growth factor (NGF), (Menesini-Chen et al. 1978; Letourneau, 1978; Gunderson and Barrett, 1980). Even for NGF, some of the evidence from which its chemoattractant properties have been suggested has been questioned (Davies, 1987). Recently, Lumsden and Davies, in a series of elegant in vitro experiments, have shown that sensory axons innervating the maxillary whiskers of the mouse are guided by a target-derived factor which was not NGF (Lumsden and Davies, 1983; 1986; Davies et al. 1987). Similarly, the fibres of cultured commissural neurones from the rat spinal cord appear to respond to a chemoattractant signal emanating from ventral floor plate cells (Tessier-Lavigne et al. 1988).

In the visual system, recent in vivo studies have also suggested chemoattraction as a mechanism to explain target-directed outgrowth of retinal ganglion cell axons. Harris (1986) grafted the eye primordium to a position just caudal to the orbit and showed that optic axons arising from these displaced eyes possessed a remarkable ability to locate the optic tectum. Retinal axons entering the brain in the near vicinity of the tectum showed highly directed outgrowth towards the tectum and, in many cases, did so independently of the normal optic pathway. In contrast, axons that entered the more caudal regions of the medulla grew both towards and away from the tectum. One explanation that was proposed to account for this directed outgrowth was that a target-derived signal could attract axons, providing they were within a certain distance of the target. In the mammalian system, retinal expiants from mice, grafted to the midbrain parenchyma of rats, produced axons that grew directly towards the superior colliculus (Hankin and Lund, 1987). In these studies, long-range guidance, by some ‘tropic’ factor emanating from the tectum was also evoked as a possible mechanism responsible for the directed outgrowth. The suggested chemoattraction was accentuated by the apparent lack of any directed outgrowth from expiants placed deep within the midbrain, which were assumed to be ‘out of range’ of the tropic signal. In vitro studies of mammalian retinotectal cocultures have also shown directed outgrowth of retinal axons and have suggested target attraction may occur (Smallheiser et al. 1981). Unfortunately, in Xenopus comparable retinotectal coculture experiments have not provided any positive evidence for such target attraction (Harris et al. 1985; Jack and Taylor; unpublished observations). In the chick, grafting the tectal precursor tissue into the diencephalon led to an abnormally positioned supernumerary tectum that was innervated by retinal ganglion cell axons (Alvarado-Mallart and Sotelo, 1984). Further, the optic fibre innervation of the supernumerary tecta was greater than that of the host tectum. These results were interpreted as evidence for target-attraction.

The results in this study, which show directed outgrowth of retinal axons towards displaced targets, may be regarded as substantive of the idea of target-derived tropism (Chung and Cooke, 1978; Giorgi and Van Der Loos, 1978; Harris, 1986). However, experiments in which the tectal primordium was removed, in all cases, still showed directed growth of optic axons. This observation suggests that the target is not essential for directed axon growth within the optic tracts (see also Reh et al. 1983).

What alternative explanations could account for the anteriorly directed growth of retinal axons in animals with midbrain rotations? Recently, Harris (1989) has provided experimental evidence for the existence of local pathway cues for retinal axon guidance, a suggestion made in a preliminary report of these results (Taylor, 1987). In brief, Harris rotated that part of the presumptive diencephalon through which optic axons would later grow and showed that the retinal axons grew with a corresponding rotation in the optic tracts. These compelling findings suggested that there were cues for retinal axon guidance located within the pathway. In our earlier study, the nature of the tissue through which the growing retinal ganglion cell axons travel was described (Easter and Taylor, 1989). We have shown that the optic axons join the rostral part of a common commissural tract which develops before retinal ganglion cell fibre outgrowth (Easter and Taylor, 1989). This fasciculation leads to the formation of the optic tract. One explanation that could be offered for the abnormal pathways found after rotation of parts of the optic pathway’s precursor tissue, is that the operation transposes cells that will give rise to the commissural pathway. For example, the anterior displacement of these cells could lead to an abnormally rostrally positioned commissure, which optic axons may follow giving rise to a rostrally directed optic tract.

The main questions to be answered are: what are the constituent axons of the commissural pathway?; how are the commissural tracts affected by the rotation operations? and, do the optic axons follow abnormally directed commissural fibres in operated embryos?

The nature of the cells that give rise to axons in the commissures is uncertain. Roberts and co-workers have shown GABA-expressing cells giving rise to axons, which course in the commissure at stage 29/30 (Roberts et al. 1987). It is unclear whether these cells are the pioneers of the commissure or, like optic axons, a secondary component. Ongoing studies should allow the resolution of this question. The question as to what happens to the commissure in rotated preparations is more complex, since at these early developmental stages clear definition of the anatomy of the brain, especially in operated embryos, is difficult (Taylor, unpublished data). Double-labelling experiments using fluorescent markers and immunohistochemical staining are being undertaken in an attempt to address this issue.

It should be stressed that the results of tectal removal do not rule out a recognition phenomenon between the growing optic axons and their target cells in the tectum. Indeed, some form of recognition must occur, otherwise axons would not stop at the tectal precursor. In normal embryos, it has been shown that axons leave the optic tract, slow down and begin to arborize as they reach the region of the tectum (Harris et al. 1987). There is evidence for target-specific labelling in the Xenopus visual centres. In a recent study by Fujisawa and colleagues, a monoclonal antibody was raised that recognizes cell surface antigens in the retinorecipient centres of Xenopus and is expressed at stage 40, the stage of development just after the arrival of the first optic axons (Takagi et al. 1987). Indeed, if retinal ganglion cell axons did not recognise their targets, then one might expect them to follow underlying axon tracts, past the tectal region, into the spinal cord (Easter and Taylor, 1989). This may be what happened to those axons that are growing caudally beyond the tectum in control preparations (Fig. 3, see Harris et al. 1987).

In summary, we have shown that retinal ganglion cell axons will grow directly towards tectal precursor tissue grafted to rostral locations in the brain. We do not believe that the growth is directed solely by a long-range target-attractant mechanism, since after deletion of the tectal precursor the outgrowth of optic axons through the diencephalon is directed as normal. We suggest an alternative explanation for retinal axon guidance in the optic tracts is fasciculation with axons of the post optic commissural system. However, since axons begin to arborize when they reach the tectum, whether it is in its normal position or has been displaced by operation, some form of local target recognition is thought likely to occur.

I would like to extend my thanks to Mike Gaze and Steve Easter for the many helpful discussions we had during the course of these experiments. I also thank them and Ray Guillery for their comments on this manuscript. This work was supported by the MRC and by the Wellcome Trust.

Adams
,
J. C.
(
1977
).
Technical considerations on the use of HRP as a neuronal marker
.
Neurosci
.
2
,
141
145
.
Alvarado-Mallart
,
R-M.
and
Sotelo
,
C.
(
1984
).
Homotopic and heterotopic transplantations of quail tectal primordia in chick embryos: Organisation of the retinotectal projection in chimeric embryos
.
Devl Biol
.
103
,
378
398
.
Chung
,
S. H.
and
Cooke
,
J. C.
(
1978
).
Observations on the formation of brain and of nerve connections following embryonic manipulation of the amphibian neural tube
.
Proc. R. soc. Lond. B
.
201
,
335
373
.
Davies
,
A. M.
(
1987
).
Molecular and cellular aspects of patterning sensory neurone connections in the vertebrate nervous system
.
Development
101
,
185
208
.
Davies
,
A. M.
,
Bandtlow
,
C.
,
Heumann
,
R.
,
Korsching
,
S.
,
Rohrer
,
H.
and
Thoenen
,
H.
(
1987
).
Timing and site of nerve growth factor synthesis in developing skin in relation to innervation and expression of receptor
.
Nature, Lond
.
326
,
353
358
.
Easter
,
S. S.
, JR
. and
Taylor
,
J. S. H.
(
1989
).
The development of the Xenopus retinofugal pathway: Optic fibres join a preexisting tract
.
Development in press
.
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
.
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
.
Gunderson
,
R. W.
and
Barrett
,
J. N.
(
1980
).
Characterisation of the turning response of dorsal root neuntes toward nerve growth factor
.
J. Cell Biol
.
87
,
546
554
.
Hankin
,
M. H.
and
Lund
,
R. D.
(
1987
).
Role of the target in directing the outgrowth of retinal axons: Transplants reveal surface-related and surface-independent cues
.
J. comp. Neurol
.
263
,
455
466
.
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 the 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 fibre in vivo
.
Development
101
,
123
133
.
Harris
,
W. A.
,
Holt
,
C. E.
,
Smith
,
T.
and
Galenson
,
N.
(
1985
).
Growth cones of developing retinal cells in vivo, on collagen surfaces and in collagen matrices
.
J. Neurosci. Res
.
13
,
101
122
.
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
.
Langley
,
J. N.
(
1895
).
Note on the regeneration of pre-gangliomc fibres of the sympathetic
.
J. Physiol. Lond
.
18
,
280
284
.
Langley
,
J. N.
(
1897
).
On the regeneration of pre-ganglionic and of post-ganglionic visceral nerve fibres
.
J. Physiol. Lond
.
22
,
215
230
. .
Letourneau
,
P. C.
(
1978
).
Chemotactic response of nerve fibre elongation to nerve growth factor
.
Devi Biol
.
66
,
183
196
.
Lumsden
,
A. S. G.
and
Davies
,
A. M.
(
1983
).
Earliest sensory nerve fibres are guided to their peripheral targets by attractants other then nerve growth factor
.
Nature, Lond
.
306
,
786
788
.
Lumsden
,
A. S. G.
and
Davies
,
A. M.
(
1986
).
Chemotropic effect of specific target epithelium in the developing mammalian nervous system
.
Nature, Lond
.
323
,
538
539
.
Macagno
,
E. G.
(
1978
).
A mechanism for the formation of synaptic connections in the arthropod visual system
.
Nature, Lond
.
275
,
318
320
.
Menesini-Chen
,
M. G.
,
Chen
,
J. S.
and
Levi-Montalcini
,
R.
(
1978
).
Sympathetic nerve fibre growth in the central nervous system of neonatal rodents upon intracerebral injection of NGF
.
Archs Ital. Biol
.
116
,
53
84
.
Nieuwkoop
,
P. D.
and
Faber
,
J.
(
1967
).
A Normal Table of Xenopus laevis
(Daudin..
Amsterdam
:
North Holland
.
O’Rourke
,
N. A.
and
Fraser
,
S. E.
(
1986
).
Dynamic aspects of retinotectal map formation revealed by a vital-dye fibre-tracing technique
.
Devi Biol
.
114
,
265
276
.
Ramon Y Cajal
,
S.
(
1892
).
The Structure of the Retina. English translation
1972
.
Charles C. Thomas
,
Springfield
.
Reh
,
T. A.
,
Pitts
,
E.
and
Constantine-Paton
,
M.
(
1983
).
The organisation of fibres in normal and tectum-less Rana pipiens
.
J. comp. Neurol
.
218
,
282
296
.
Roberts
,
A.
,
Dale
,
N.
,
Ottersen
,
O. P.
and
Storm-Mathisen
,
J.
(
1987
).
The early development of neurons 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
.
Smallheiser
,
N. R.
,
Peterson
,
E. R.
and
Crain
,
S. M.
(
1981
).
Neurites from mouse retina and dorsal root ganglion expiants show specific behaviour within co-cultured tectum or spinal cord
.
Brain Res
.
208
,
499
505
.
Sperry
,
R. W.
(
1963
).
Chemoaffinity and the orderly growth of nerve fibre patterns and connections
.
Proc natn. Acad. Sci. U.S.A
.
50
,
703
710
.
Takagi
,
S.
,
Tsuji
,
T.
,
Amagai
,
T.
,
Takamatsu
,
T.
and
Fujisawa
,
H.
(
1987
).
Specific cell surface labels in the visual centers of Xenopus laevis tadpoles identified using monoclonal antibodies
.
Devi Biol
.
122
,
90
100
.
Taylor
,
J. S. H.
(
1987
).
Target recognition and pathway cues in the primary development of the retinotectal projection
.
Soc. Neurosci. Abstr
.
13
104.12
.
Tessier-Lavigne
,
M.
,
Placzek
,
M.
,
Lumsden
,
A. G. S.
,
Dodd
,
J.
and
Jessel
,
T. M.
(
1988
).
Chemotropic guidance of developing axons in the mammalian central nervous system
.
Nature, Lond
.
336
,
775
778
.