/. Embryol. exp. Morph. 73, 17-38, 1983
Printed in Great Britain © The Company of Biologists Limited 1983
Pathways of Xenopus optic fibres regenerating from
normal and compound eyes under various
conditions
By GAZE, R. M. 1 AND FAWCETT, J. W. 2
From the National Institute for Medical Research, London
SUMMARY
We have used Horseradish peroxidase to investigate the pathways taken by Xenopus optic
fibres regenerating from normal and electrophysiologically-confirmed compound eyes to the
optic tectum. Optic fibres, when sectioned near the chiasma, regenerate up both sides of the
diencephalon to both tecta. We have therefore been able, by using animals in which one eye
had or had not been removed at early embryonic stages, to look at the behaviour of regenerating axons in three different situations: (1) regeneration to the contralateral tectum, previously
innervated by the sectioned fibres; (2) regeneration to a 'virgin' ipsilateral tectum, never
before innervated by optic fibres; and (3) regeneration to an ipsilateral tectum already innervated by fibres from a normal eye.
From the chiasma to the tectodiencephalic junction regenerating fibres behave similarly in
all three situations, following roughly the course of the normal optic tract, but running in a
rather disorganised way, with frequent crossing over of fibres. Howeverfibresof nasal retinal
origin (from an NN eye) spread to occupy a much larger area of the side of the diencephalon
than those of temporal origin (from a TT eye). From the tectodiencephalic junction to the
tectal termination of thefibresthere are differences between the three situations investigated;
fibres regenerating to a 'virgin' ipsilateral, or to a denervated contralateral tectum, tend to
grow straight onto the tectum, instead of being channelled into lateral or medial brachium as
uncut fibres tend to be. There is, however, the remains of a brachial organisation, and of
differential selection of these brachia byfibresfrom the diffferent types of compound eye, this
being well seen on 'virgin' tecta.
Fibres regenerating to an ipsilateral innervated tectum behave very differently. As they
reach the tectodiencephalic junction they suddenly start to grow in a less disorganised way,
and are channelled into well defined brachia. If from a compound eye, these fibres terminate
on only that part of the tectum innervated byfibresfrom the corresponding part of the normal
eye. Thusfibresfrom a W eye and those from the ventral half of the normal eye all terminate
on medial tectum; fibres from an NN eye, and those from the nasal half of the normal eye all
terminate in caudal tectum; and temporal fibres from both normal and TT eyes terminate in
rostral tectum.
INTRODUCTION
In an earlier paper (Fawcett & Gaze, 1982) we described the pathways taken
by optic nerve fibres to reach their tectal destinations in normal Xenopus and in
1
Author's address: National Institute for Medical Research, The Ridgeway, Mill Hill,
London NW7 1AA, U.K.
2
Author's address: The Salk Institute, P.O. Box 85800, San Diego, California 92138,
U.S.A.
18
R. M. GAZE AND J. W. FAWCETT
Xenopus with one compound double-nasal (NN), double-temporal (TT) or
double-ventral (VV) eye. The optic fibres in such animals will regenerate when
cut and may then restore the orderly retinotectal projection that existed
previously. In the present paper we investigate the pathways taken by these
fibres regenerating to the tectum, and discuss how they may reform this orderly
projection.
Quite apart from the very great difference in size and shape between the
developing embryonic visual system and the regenerating adult structure,
evidence is accumulating to suggest that fibres use rather different mechanisms
to establish a retinotectal map in development and during regeneration. In
regeneration optic fibres appear to recognise tectal position, at any rate in goldfish, since if one transplants tectal grafts from rostral to caudal or vice versa,
regenerating fibres make a correspondingly translocated distribution (Hope,
Hammond & Gaze, 1976; Gaze & Hope, in preparation). We do not know
whether the tectal positional markers are themselves cellular modifications in
the tectum, fibre terminals from other projections to the tectum, or whether they
are some more direct manifestation of the original fibre distribution, for instance
fibre debris. Certainly their position can be defined by the optic fibre input to the
tectum (Schmidt, 1978; Straznicky & Tay, 1981). During the initial development
of the retinotectal projection, however, there is virtually no evidence that placespecific tectal markers function, or even exist.
In the present paper we discuss regeneration in the Xenopus visual system and
in particular the factors which may influence the pathways down which
regenerating fibres will grow to the Xenopus tectum. We have investigated
mostly the regeneration of fibres from compound eyes because, as we have
shown in the previous paper (Fawcett & Gaze, 1982), in unregenerated (uncut)
projections the course of the optic fibres through the brain is highly characteristic
for each type of compound eye and is clearly distinguishable from normal. Thus
by comparing the path taken by fibres regenerating from a compound eye with
that taken by fibres regenerating from a normal eye, we have attempted to assess
to what extent fibre pathway is influenced by the retinal positional labels carried
by the ingrowing fibres, and to what extent by simple mechanical factors.
When the optic nerve of a newly metamorphosed Xenopus is cut close to the
chiasma, regenerating optic fibres grow up both sides of the diencephalon to
reach both optic tecta (Glastonbury & Straznicky, 1978). In regeneration, however, unlike normal development, all the fibres are found just under the pia
(Gaze & Grant, 1978). We have induced optic fibres to grow through three types
of environment: though the contralateral optic tract, which was previously filled
with fibres from the eye in question; through the ipsilateral optic tract, which
contains retinotectal fibres from the normal eye contralateral to it, and which
previously contained also ipsilateral retinodiencephalic fibres from the experimental eye; and, by excising the other eye at an early embryonic stage, through
an optic tract which has never previously had optic fibres in it (except for
Regeneratingfibresfrom normal and compound eyes
19
ipsilateral retinodiencephalic projections) to a tectum which is virgin, in that it
has never received a direct input of optic fibres. We have therefore essentially
three different experiments to consider and these will be discussed separately.
METHODS
Embryonic operations
The surgical formation of compound eyes was described in a previous paper
(Fawcett & Gaze, 1982). In brief, the left eyes of Xenopus embryos of stages
29-31 (Nieuwkoop & Faber, 1956) were operated on in full-strength Niu Twitty
solution containing 0-005 % MS 222 (Sandoz, tricaine methane sulphonate). To
form an NN eye the temporal half of the host left eye was substituted by a donor
right nasal half. VV and TT eyes were made in comparable fashion. Animals that
were to have virgin tecta had their right eye removed at the same time. Animals
were then reared to metamorphosis.
Optic nerve section
Shortly after metamorphosis, in experimental animals and in control animals
with normal eyes, the left optic nerve was exposed within the cranium, the
operation being done through the mouth. The nerve was cut near the chiasma,
the chiasma itself not being damaged. The optic nerve was then allowed to
regenerate for 4-8 weeks.
Electrophysiology
Animals of the experimental series were anaesthetised with MS 222, and their
optic tecta exposed. They were then mounted in a Perspex globe filled with
oxygenated Niu Twitty solution containing 1:10000 MS 222. This was placed at
the centre of an Aimark perimeter, with the compound eye centred, and the
other eye, if any, covered.
Visually evoked potentials, elicited by movement of a black disk in the visual
field, were recorded through a glass-insulated tungsten microelectrode. Usually
only nine tectal positions were sampled on each tectum, to avoid tissue damage,
but this was usually sufficient to show whether the map was of a typical
compound-eye type. Those animals that did not have the expected type of
visuotectal map from the compound eye were discarded.
Histology
The compound eye was removed under MS 222 anaesthesia and crystalline
Horseradish peroxidase (HRP, Sigma type 6) applied to the optic nerve stump.
Animals were left with their heads out of water until they recovered from the
anaesthetic, usually 20min, and then left overnight. They were then
anaesthetised, killed by perfusion with 0-25 M-sucrose followed by 2-5%
glutaraldehyde in 0-1 M-phosphate buffer, pH7-2. The brains were removed and
transferred to cold fixative for a further hour, during which time their pias were
20
R. M. GAZE AND J. W. FAWCETT
03
Fig. 1
Regenerating fibres from normal and compound eyes
21
stripped away. The brains were then reacted to show the HRP using the method
of Adams (1977) modified in this laboratory by R. V. Stirling.
After fixation the brains were washed for 10 min in PO 4 buffer, then for 1 h in
0-1 M-Tris buffer at pH 7-2. They were then transferred to 1 % cobaltous chloride
in Tris for 1 h, then washed in Tris for 10min, and in two changes of PO 4 buffer
for 10min each, then to 50mg/100ml di-amino benzidine and 1 % DMSO in
PO 4 buffer for 11 h. Meanwhile some more of the incubating solution was put on
ice, and, just before the brains were put in it, 4 ml/100 ml of 0-3 % H 2 O 2 was
added. Incubation continued on ice until the HRP reaction was complete, usually 20 min. The preparations were then washed in PO 4 buffer, dehydrated, and
cleared in methyl salicylate.
In two to six animals of each category the brain was prepared in this way. The
brains were then examined by transmitted light microscopy. We were unable to
find an entirely satisfactory method of reproducing our results in two dimensions, in a way suitable for publication. Due to the three-dimensional nature of
the specimens and the limited depth of focus of the microscope optics it was not
possible to take sufficiently detailed photographs. We therefore decided to make
camera-lucida drawings. These can never be an exact representation of the
specimen - it is obviously impossible to draw in every nerve fibre seen - but are
necessarily more in the nature of an interpretation by an experienced observer.
Fig. 1A. Camera-lucida drawing of a dorsolateral view of a whole-mount preparation of a normal brain in which the optic nerve had not been cut. Dorsal is upwards
and rostral is to the right. Fibres from the left eye, which was normal, are seen
coursing up the right side of the diencephalon to the right optic tectum. At the
tectodiencephalic junction many fibres enter either the medial or the lateral
brachium while some carry straight on to the tectum. The stippling represents
neuropil. The area of neuropil caudal to the chiasma is the basal optic neuropil and
that rostral to the optic tract is the neuropil of Bellonci. The bar represents 1 mm for
this and all the other drawings.
Fig. IB. Preparation from an animal in which thefibresfrom a normal eye have been
cut and allowed to regenerate. Dorsolateral view. Compared to the previous illustration it can be seen that the optic tract up the side of the diencephalon is here wider,
and that more of thefibrespass straight onto the front of the tectum. Very few fibres
are seen passing lateral to the tectum, but there is a poorly defined medial brachium.
Dorsal is upwards, rostral to the right. Magnification as in 1A. The ipsilateral side
of this brain is shown in Fig. ID. This brain is also shown in Fig. 2A.
Fig. 1C. A dorsolateral view of the pathway taken by fibres regenerating from a
normal left eye to the ipsilateral tectum. Dorsal is upwards, rostral to the left. This
tectum is also innervated by fibres from the normal right eye. On the side of the
diencephalon the tract is wide, and the course of thefibrestortuous. However, at the
tectodiencephalic junction the fibres suddenly become 'tramlined' into well-defined
brachia. The terminal optic neuropil in this preparation is sparse and confined to the
rostral part of the tectum. Magnification as in 1A.
Fig. ID. Dorsolateral view of fibres regenerating from a normal left eye up the
ipsilateral diencephalon. Dorsal is upwards, rostral to the left. The tectum also
carries the projection from normal right eye. The regenerated tract is extensively
disorganised until the fibres reach the tectodiencephalic junction, when marked
'tramlining' is seen. Magnification as in 1A. The ipsilateral side of this brain is shown
in Fig. 2C and the contralateral side in Fig. IB.
22
R. M. GAZE AND J. W. FAWCETT
RESULTS
The experiments on which this report is based are summarised in Table 1. All
the animals with compound eyes used in these experiments were mapped
electrophysiologically and had recognisable compound eye maps regenerated
onto at least one tectum. The few operated animals that did not satisfy this
criterion were discarded and are not included in the table.
Table 1. Regenerated Optic Nerve
Type of preparation
Normal eye
NN eye
TT eye
Weye
NN eye (virgin tectum)
TT eye (virgin tectum)
W eye (virgin tectum)
No.
Visuotectal maps
(+, mapped; - , not mapped)
contra
ipsi
3
6
4
4
3
4
2
Regeneration of fibres back to the contralateral tectum
Regeneration from a normal eye
Between the optic chiasma and the tectodiencephalic junction, fibres
regenerating back to the contralateral tectum from a normal eye form an optic
tract which is wider and less well organised than a normal tract (Fig. 1A, B). The
regenerated fibres interweave more than normal fibres. In addition to the formation of loosely organised brachia (mainly medial) there is a marked tendency for
many fibres to charge straight onto the front of the tectum (Figs IB and 2A). The
coverage of the tectum with neuropil (meaning here the densely distributed
terminal arborizations of the optic fibres) is normal.
Regeneration from a compound eye
Contralateral NN
As we have shown in the previous paper, the non-regenerated contralateral
tract from an NN eye tends to be wide, with strongly formed medial and lateral
brachia. Almost all the fibres travel in either the medial or the lateral brachium;
few enter the tectum at its rostral pole.
Regenerating fibres from NN eyes behave in a variable fashion. Between the
chiasma and the tectodiencephalic junction the tracts tend to be wide and disorganised (Figs 3A, 4A). There is strong brachial formation and in addition many
fibres charge straight onto the tectum. In this respect, fibre regeneration from
an NN eye resembles that from a normal eye. Only one of our nine preparations
produced a regenerated tract which closely resembled an unregenerated tract but
2A
\
Fig. 2A. Fibres regenerating from a normal eye onto the
contralateral tectum. Dorsal is upwards and rostral to the right.
Many fibres may be seen charging straight onto the tectum. This is
the same brain as that shown in Fig. IB. Bar = 500fjm.
Fig. 2B. Fibres regenerating from an NN eye to an already innervated ipsilateral tectum. Dorsal is upwards and rostral to the left. The
fibres mostly form conspicuous brachia, rather than passing straight
onto the tectum. This is the same brain as that shown in Fig. 10.
Bar = 500 /an.
Fig. 2C. Fibres regenerating from a normal eye to an innervated
ipsilateral tectum. Thefibresmainly travel in either the medial or the
lateral brachium. Few enter straight onto the front of the tectum.
This is the same brain as that shown in Fig. ID. Bar = 500 fjm.
to
s
a
Io
24
R. M. GAZE AND J. W. FAWCETT
in this case there were many fibres which entered directly onto rostral tectum
(Fig. 5A). Tectal coverage by neuropil tended to be complete.
Contralateral TT
Uncut fibres from a TT eye form a rather tightly grouped tract and this is also
true, to a lesser degree, for regenerated fibres from a TT eye (Figs 6A, 7 A). There
is much crossing over of fibres in the tract but fibres enter straight onto the rostral
part of the tectum. Terminal neuropil is restricted mainly to rostrolateral tectum
(in these recently metamorphosed animals, 4-8 weeks after nerve section).
Contralateral W
Uncut fibres from a VV eye reach the tectum via the medial brachium of the
tract only (Straznicky, Gaze & Horder, 1979; Steedman, 1981; Fawcett & Gaze,
1982). Regenerated fibres from a VV eye form a contralateral tract which is wide
and disorganised (Fig. 8A) but there is a strong tendency for many fibres to enter
the tectum medially. Some fibres charge straight on to the tectum and there is
little sign of any lateral brachial formation. Terminal neuropil is restricted mainly to rostromedial tectum.
Regeneration of fibres to an ipsilateral, virgin, tectum
Ipsilateral NN to virgin tectum
Fibres regenerating to an uninnervated ipsilateral tectum run up the side of the
diencephalon in an even more disorganised manner than on the contralateral
side. Fibres mostly enter the tectum either by charging straight onto the rostral
pole or through a poorly organised medial brachium. The three animals in this
category show a trend where the most sparsely innervated and most disorganised
tract produced a tectal projection that was completely disorganised (Fig. 9C, D)
and the most densely innervated tract, which resembled a regenerated
contralateral tract rather closely, gave rise to a map that was fairly well ordered
(Fig. 5C, D). All three animals gave densely innervated contralateral tracts,
associated with fair NN maps.
Ipsilateral TT to virgin tectum
Fibres regenerating from a TT eye along the ipsilateral tract to a tectum that has
not previously received retinotectal fibres, form tracts that are considerably
disorganised in comparison with normal TT tracts, but which still show a tendency
for the fibres to form a rather compact bundle passing up the side of the
diencephalon. The fibres do not form brachia but pass straight onto the rostral
part of the tectum, where they terminate. The four animals in this category all
behaved in the same way in respect to the tract. Two of them gave ipsilateral maps
that resembled the contralateral projections (Fig. 6C, D). In one animal the ipsilateral projection was poorly organised in comparison with the contralateral
projection. The fourth animal could not be assessed in relation to its ipsilateral
Regenerating fibres from normal and compound eyes
Fig. 3. These three drawings show the results from an animal with a left NN eye and
a right normal eye. The optic fibres from the NN eye were cut. (A) The right
(contralateral) tectum in dorsolateral view. Fibres travel up the side of the
diencephalon in a wide band (the gap is a mechanical tear) and then mostly pass
straight onto the rostral margin of the tectum. There is no lateral brachium and only
a poorly defined medial one. Neuropil covers the entire tectum. Magnification as in
Fig. 1A. (B) The visuotectal map from the right tectum shown in 3A. In this and the
other visuotectal maps shown, the upper diagram represents the optic tectum (right
or left) seen in dorsal view, with the open arrow pointing rostrally along the midline.
Numbers and filled circles represent electrode recording positions. The lower
diagram represents a perimeter chart of the left visual field with numbers and filled
circles indicating optimal stimulus positions corresponding to the tectal recording
positions. The eye is to be thought of as being behind the perimeter chart, looking
out through its centre. The chart extends for 100 ° outwards from the centre. N, nasal;
S, superior; T, temporal; I, inferior. In this map response positions in the visual field
tend to be reduplicated about the dorsoventral axis in the manner (although poorly
ordered) characteristic of NN projections. (C) The ipsilateral (left) tectum, which is
also innervated by fibres from the normal (right) eye. Fibres travel in a wide band
up the side of the diencephalon. At the tectodiencephalic junction, several fibres
pass straight onto the tectum, a few travel in a well-defined lateral brachium, but the
majority travel in the medial brachium. The neuropil is restricted to the caudomedial
tectum. Magnification as in 1A.
25
R. M. GAZE AND J. W. FAWCETT
Fig. 4. A series of illustrations from an animal similar to that in Fig. 3. The observations are also similar. (A) The right, contralateral, tectum. The tract is wide, and the
majority of the fibres pass straight onto the tectum. Magnification as in 1 A. (B) The
visuotectal map from this right tectum. This is of the classical NN type, and quite well
ordered. The map suggests that the eye was rotated 15° anticlockwise. (C) The
ipsilateral (left) tectum. The tract is wide, with a tear dorsally. At the tectodiencephalic junction most fibres enter the medial brachium, a few enter the
lateral brachium and a few pass straight onto the tectum. Magnification as in 1A.
Fig. 5. Results from an animal with an NN eye on the left and no eye on the right.
The optic nerve from the compound eye was cut just after metamorphosis. (A) The
right (contralateral) tectum in dorsolateral view. The tract is broad, and most of the
fibres pass straight onto the tectum. Neuropil covers the entire tectum. Magnification as in 1A. (B) The visuotectal map from this contralateral right tectum. This is
of the typical NN type and quite well ordered. In this and other maps, open circles
on the tectal diagram represent positions from which no tectal response was obtained. (C) The left (ipsilateral) tectum in dorsolateral view. The tract is very wide
(the gaps are mechanical tears), and mostfibrespass straight onto the tectum. There
is an ill-defined medial brachium. Neuropil covers the entire tectum. Magnification
as in 1A. (D) The visuotectal map from the virgin ipsilateral tectum shown in 5C.
This is of a typical NN type and is reasonably well ordered.
Regeneratingfibresfrom normal and compound eyes
Fig. 5
27
28
R. M. GAZE AND J. W. FAWCETT
Fig. 6.
Regeneratingfibresfrom normal and compound eyes
29
projection, since there was only one reduplicated point; in this case, however, the
contralateral map also was poor and showed only three reduplicated positions.
The contralateral projections in these animals were recognisably TT but, as usual
with TT projections, their ordering was less good than with NN maps.
Ipsilateral VV to virgin tectum
In the two animals in which fibres regenerated up the ipsilateral tract to a virgin
tectum, the tracts were disorganised and wide (Fig. 8C) but allfibrespassed medially. Most of the contralateral fibres also passed medially. In one animal both
maps were recognisably VV (Fig. 8B, D) and in the other the contralateral tectum
only gave three responses but the ipsilateral projection gave a fair VV map.
Regeneration of fibres to an ipsilateral innervated tectum
Normal eye to innervated tectum
The arrangement of regenerated optic fibres in the ipsilateral tract from chiasma to tectodiencephalic junction is very variable from one animal to another, but
the organisation is usually worse than in the contralateral tract. The distribution
can be very wide with greatly spread out straggly edges (Fig. 1C). The tract is
extensively disorganised and an extreme case is shown in Fig. ID.
Both contralateral and ipsilateral tracts may thus be much disorganized when
regenerated, but a major difference between the two is that, at the tectodiencephalic junction, ipsilateral fibres do not, usually, charge straight onto
the tectum. They are diverted from their courses and turn, mainly medially, to
run round the margin of the tectum, as a well-defined medial brachium, before
entering onto its surface. There is also a well-organised lateral brachium, but
always with very few fibres in it (Fig. 2C). Tectal coverage may be normal.
Ipsilateral NN to innervated tectum
Fibres from an NN eye regenerating along the ipsilateral tract, which also carries the projection from the normal eye, tend to show extensive disorganisation
Fig. 6. A brain from an animal with a TT eye in the left orbit and no right eye. The left
optic nerve was cut. (A) The right (contralateral) tectum in dorsolateral view. The
optic tract up the side of the diencephalon is of approximately normal width; narrower
than that formed by fibres regenerating from a normal eye and much narrower than
that formed byfibresfrom an NN eye. Fibres enter the tectum from the rostral margin
and the neuropil is restricted to the rostral part of the tectum. Magnification as in 1A.
(B) Visuotectal map from this contralateral tectum. This is typical of maps from TT
eyes, with the maximum separation between reduplicated visualfieldpositions being
found when recording from rostral tectum. The order is quite good. (C) The ipsilateral
(virgin) tectum in dorsolateral view. The optic tract is of greater width than the
contralateral one, but narrower than that from an NN eye regenerating to a virgin
tectum. Fibres enter the tectum by its rostral margin, and neuropil is restricted to the
rostral half-tectum. Magnification as in 1A. (D) The visuotectal map from the virgin
tectum. This is recognisably reduplicated about the dorsoventral axis, but is less well
ordered than that from the contralateral tectum.
EMB73
30
R. M. GAZE AND J. W. FAWCETT
Fig. 7. Results from an animal with a TT eye in the left orbit and a normal eye in the
right orbit. The left optic nerve was cut. (A) The right (contralateral) tectum in
dorsolateral view. The optic tract is narrow and neuropil is restricted to the rostral
tectum. Magnification as in 1A. (B) The visuotectal map from this contralateral right
tectum. This is recognisably reduplicated about the dorsoventral axis of the visual
field, and moderately well ordered.
between chiasma and tectodiencephalic junction. Despite the tract disorganisation, as fibres reach the tectal margin they form groups rather suddenly and
follow the presumed positions of the brachia of the underlying normal fibres
(Figs 2B, 4C and 10), particularly the medial brachium. The fibres tend to leave
the rostral and lateral tectum empty and to reach the region of termination,
which is in caudomedial tectum, by running round the tectal margins. Some few
fibres enter straight onto rostral tectum and cross it to their terminal zone (Fig.
3C). However the great majority of fibres travel in the brachia until a point
adjacent to their eventual termination sites, when they turn onto the tectum.
Ipsilateral TT to innervated tectum
Fibres regenerating from a TT eye up the ipsilateral tract when the projection
from the normal eye is present, tend to concentrate in the middle of the tract
region (Fig. 11), forming a narrow tract. Neuropil distribution is confined to
rostrolateral tectum, and the fibres all enter the tectum by its rostral pole, there
being little or no sign of fibres entering either brachium.
Ipsilateral VV to innervated tectum
Fibres form a wide and straggly tract up the side of the diencephalon. At the
tectodiencephalic junction most of the fibres enter a fairly well-defined medial
Regeneratingfibresfrom normal and compound eyes
Fig. 8. Results from an animal with a VV eye in the left orbit and no right eye. The
left optic nerve was cut. (A) The contralateral (right) tectum in dorsolateral view.
The optic tract is quite wide. At the tectodiencephalic junction the majority of fibres
form an ill-defined medial brachium, and a few pass straight onto the tectum.
Neuropil covers the entire tectum. Magnification as in 1A. (B) The visuotectal map
from this contralateral right tectum. This is of the typical VV type, the visual field
positions being reduplicated about the naso-temporal axis. The maximum distance
apart of reduplicated points is found when recording from medial tectum. The order
is quite good. (C) The ipsilateral (virgin) tectum. The fibres form a poorly defined
medial brachium. None go laterally and none pass straight onto the front of the
tectum. Neuropil is restricted to rostromedial tectum. Magnification as in 1A. (D)
The visuotectal map from the virgin tectum. This is reduplicated about the nasotemporal axis, but the order is almost non-existent.
31
R. M. GAZE AND J. W. FAWCETT
Fig. 9. Results from an animal with an NN left eye and no right eye. The left optic
nerve was cut. (A) Dorsolateral view of the right (contralateral) tract and tectum.
The tract is very wide with massive and poorly organised brachial formation.
Numerous fibres pass straight onto the tectum. There is a tear in the rostral tract.
Tectal coverage is complete. Magnification as in 1A. (B) The visuotectal map from
the contralateral right tectum. The map is reduplicated in the manner of NN maps
and the order is good. (C) Dorsolateral view of the ipsilateral (virgin) tectum and
tract. The tract is disorganised. Magnification as in 1A. (D) The visuotectal map
from the virgin tectum. The map is completely disordered.
Regeneratingfibresfrom normal and compound eyes
33
Fig. 10. The left, ipsilateral, tectum in an animal with an NN left eye and a normal
right eye. The fibres from the NN eye were cut and have regenerated back to the
already innervated ipsilateral tectum. The fibres are seen to run in brachia round the
tectum to reach their caudomedial destinations. No fibres are seen crossing the
tectum from the front. This brain is also shown in Fig. 2B.
brachium, a few enter a well-defined lateral brachium, while some enter directly
onto the tectum (Fig. 12). The sizes of the medial and lateral brachia are similar
to what one sees in the case of both normal and NN eyes regenerating to an
innervated tectum. Neuropil is restricted to medial or rostromedial tectum.
Accuracy of regeneration: visuotectal maps
Maps recorded from the contralateral tecta showed that the regenerating optic
fibres made the same types of projection as uncut fibres from such eyes. However, even from the restricted nine-point recordings made here, it could be seen
that the maps were less well ordered than those from non-regenerated control
animals. In general, NN maps were better ordered than those of the other two
types investigated.
Maps from the ipsilateral virgin tectum were also, in most cases, recognisably
of the type characteristic of the compound eye being investigated, but they were
in general even less well ordered than contralateral regenerated maps. Ipsilateral
projections to already innervated tecta were not mapped.
DISCUSSION
In this series of experiments we have tried to identify some of the factors which
guide regenerating fibres in the Xenopus visual system. We have looked at the
pathways taken by fibres from normal and compound eyes in three situations:
regenerating back to their own tectum; regenerating to a tectum that already
34
R. M. GAZE AND J. W. FAWCETT
carries a projection from a normal eye, and regenerating to a previously uninnervated virgin tectum. This has allowed us to identify some of the factors which
affect the pathways followed by regenerating fibres.
The behaviour of fibres between the chiasma and the tectodiencephalic junction
Between the chiasma and the tectodiencephalic junction, the presence or
absence of fibres or fibre debris has little influence on the pathway taken by
regenerating fibres; in all three experimental situations the fibres grow over the
pial surface of the brain (Gaze & Grant, 1978), following roughly the course that
would be taken by a normal optic tract. The fibres run tortuously, with frequent
corssings over, a feature not seen in normal brains. The only consistent difference we see in our preparations depends not on which of the three different
experimental environments the fibres are growing in, but on the embryonic
position of origin of the retina from which they came; thus fibres of temporal
retinal origin, from TT eyes, grow to form a narrower optic tract than those of
nasal origin, from NN eyes, which form a very wide tract. The tract from normal
and VV eyes is of intermediate width.
The behaviour of fibres as they grow onto the tectum
Between the tectodiencephalic junction and the termination of the fibres on
the tectum, there are considerable differences in the behaviour of the fibres,
depending on the environment through which they are growing.
The pathways of fibres regenerating back to the contralateral tectum
Fibres regenerating back to their original, contralateral, tectum, have a tendency to grow directly onto the tectum via its rostral pole, rather than running
Fig. 11. The left (ipsilateral) tectum from an animal with a left TT eye and a right
normal eye. The fibres from the TT eye have been cut. The fibres mostly run in the
centre of the tract, though a few wander. They enter the tectum by its rostral margin
and terminate over its rostral half. Magnification as in 1A.
Regeneratingfibresfrom normal and compound eyes
35
to near their eventual sites of termination in either medial or lateral brachium,
as they do in normal brains. The fibres do, however, form some brachial organization. Thus fibres regenerating from VV eyes show a greater tendency to
run in a rather poorly defined medial brachium than fibres from either normal
or NN eyes. In the latter two cases it is possible to see an ill defined medial
brachium, little lateral brachium, and many fibres which charge straight on to the
rostral pole of the tectum.
Fig. 12. The left (ipsilateral) tectum from an animal with a left VV eye and a right
normal eye; fibres from the W eye were cut. The tract is wide and straggly. On
reaching the tectodiencephalic junction most fibres enter a well-defined medial
brachium, a few enter a well-defined lateral brachium and a few pass straight onto
the tectum. Neuropil is restricted to medial and rostral tectum. Magnification as in
1A.
Fibres regenerating to an ipsilateral, virgin, tectum
The situation here is similar to, but rather more extreme than, that found with
fibres regenerating to the contralateral tectum. That is, most fibres do not show
much sign of being directed into brachia, but instead advance straight on to the
rostral pole of the tectum. However fibres of ventral origin do tend to run in a
poorly defined medial brachium. A lateral brachium is not seen.
Fibres regenerating to an innervated ipsilateral tectum
Regenerating fibres here have their pathways markedly influenced by the
existence of the underlying normal projection. Instead of charging on to the
rostral pole of the tectum, the fibres are almost all directed into narrow, welldefined brachia. They leave those brachia only at points adjacent to their areas
of termination. Only a few fibres, probably of nasal retinal origin, are seen to
pass straight onto the tectum. Fibres of temporal origin, of course, normally
enter the tectum by its rostral pole and terminate rostrally; and they do so also
36
R. M. GAZE AND J. W. FAWCETT
in our regenerates. As found by Gaze & Straznicky (1980), the tectal terminations of the regenerating fibres are defined by those of the incumbent normal
projection. Thus fibres from VV eyes all terminate medially, over the area where
the ventral fibres from the normal eye are found, and so on.
Interestingly, despite forming anatomically well-defined brachia, almost all
the fibres pass medially, as indeed is also the case with fibres regenerating to
contralateral or virgin tecta. Fibres, then, regenerating to an innervated tectum
mostly reach it via the medial brachium; the lateral brachium contains few fibres,
and its size does not vary depending on whether the fibres come from a normal,
VV or NN eye. This is quite different to what one finds in normal, uncut projections. Despite this, fibres continue to grow around the tectal margin until adjacent to their areas of termination - only then do they grow onto the tectum. They
may get onto the ring road going in the wrong direction, but they still take the
correct exit.
We do not understand why fibres regenerating to an innervated tectum often
select the wrong brachium. In normally developing animals of the same age,
newly arriving fibres which grow through a similar environment must be able to
select the medial or lateral brachium, depending on the retinal position from
which they come; for instance fibres from ventral retina all grow via the medial
brachium. There are several possible explanations for this difference: the large
number of regenerating fibres arriving in a short time may swamp the system; the
width and disorganization of the optic tract may make correct brachium selection
difficult, and the tendency of regenerating fibres in whatever situation, for
reasons we do not understand, to grow medially may override selection. However, once regenerating fibres are in a brachium, their pathways are clearly
related to those of the underlying fibres of the incumbent normal projection.
In our previous paper on the pathways of normal, uncut retinotectal fibres
(Fawcett & Gaze, 1982) we put forward the idea that the orientation of the
retinotectal map, and a crude degree of retinotopic order, are defined by the
pathways taken by the fibres as they grow to the tectum. There is evidence that
previously innervated tecta (Attardi & Sperry, 1963; Jacobson & Levine, 1975;
Hope et al. 1976) have become positionally labelled, and in such cases it may
therefore not be important for the correct formation of the retinotectal map,
what pathway the fibres take.
However, fibres regenerating to both contralateral and virgin tecta do exhibit,
to a degree, the pathway patterning seen in normal brains. Moreover, the degree
to which this patterning looks normal seems to correlate with the ordering of the
retinotectal map recorded from individual tecta in our experiments. Thus the
ordering of regenerated maps on contralateral tecta is consistently better than
the ordering of those on virgin tecta, and these contralateral projections consistently show the more normal looking optic tracts. Also, comparing all the NN
regenerates to contralateral tecta, we find that those with the most normal
looking tracts tend to have the most ordered visuotectal maps. It is possible,
Regeneratingfibresfrom normal and compound eyes
37
then, that the pathways taken by regenerating fibres to the tectum affect the
resultant retinotectal map, although there may well be other factors operating
as well. These arguments do not apply to fibres regenerating to an innervated
ipsilateral tectum, where both their pathway and tectal termination are determined by the incumbent normal projection.
The sudden change of behaviour ('tramlining') of fibres regenerating to an
innervated ipsilateral tectum at the tectodiencephalic junction, is most interesting. In normal visual projections, both in Xenopus (Fawcett & Gaze, 1982;
Fawcett, Gaze, Grant & Hirst, in preparation) and in Ciclid fish (Scholes, 1979),
there is an ordering or reordering of the fibres at this point; and we would like
to know what change in the environment of the growing fibres brings this about.
Our results suggest that for full expression of the phenomenon in postembryonic
life there must already be optic fibres present. This could indicate that pioneer
fibres may be important in laying down the anatomy of the developing visual
pathway.
We see consistently in the optic pathway that fibres of temporal origin tend to
grow as a tightly coherent group, whereas those from retina of nasal origin do
not. We have noted this previously in the optic nerve (Fawcett, 1981; Fawcett &
Gaze, 1981); it has been found in the uncut optic tract (Steedman, 1981; Fawcett
& Gaze, 1982) and the same phenomenon is here shown to exist in the
regenerated optic tract. The tendency of the neuropil from a TT eye to be
restricted to a limited area of rostral tectum, while that from an NN eye covers
the whole tectum might be a manifestation of the same phenomenon. The tendency of temporal fibres to remain together could be due to their growing
through a pathway in the brain which is narrower than that for nasal fibres and
which is specific for temporal fibres. The phenomenon could also be due to
interactions between the growing fibres themselves, with temporal fibres tending
preferentially to grow close to other temporal fibres.
The observations presented in this paper suggest to us that the second of these
possibilities is correct, that temporal fibres tend to stick together. In the part of
the optic tract between chiasma and tectodiencephalic junction, regenerating
fibres are not precisely following their original pathways, they are growing up the
side of the diencephalon just under the pia (Gaze & Grant, 1978). Despite this
the tract from a TT eye is narrower than that from a normal, VV, or NN eye,
indicating that the fibres from the TT eye are more closely packed.
This ability of growing temporal fibres to remain together could give them a
competitive advantage over nasal fibres in innervating rostral tectum; in other
words it could establish the rostrocaudal polarity of the retinotectal map. A study
of possible differences in the behaviour of growth cones from nasal and temporal
fibres could perhaps help us to understand how fibres in an array might recognise
appropriate neighbours and pathways.
In conclusion, we present in this paper results which suggest that fibres
regenerate onto a non-innervated tectum with a degree of order sufficient to
38
R. M. GAZE AND J. W. FAWCETT
form a relatively disorganised, but recognisably retinotopic projection, such as
we see, without any mechanisms other than fibre interactions being involved. We
present evidence for there being a sudden change in optic fibre behaviour at the
tectodiencephalic junction when the fibres approach an innervated tectum, and
for a gradient, from temporal to nasal retina, of a property which tends to make
temporal fibres grow as a coherent group.
REFERENCES
ADAMS, J. C. (1977). Technical considerations on the use of HRP as a neuronal marker.
Neuroscience 2, 141-145.
ATTARDI, D. G. & SPERRY, R. W. (1963). Preferential selection of central pathways by
regenerating optic fibers. Exptl. Neurol. 7, 46-64.
FAWCETT, J. W. (1981). How axons grow down the Xenopus optic nerve. J. Embryol. exp.
Morph. 65, 219-233.
FAWCETT, J. W. & GAZE, R. M. (1981). Fiber order in the Xenopus optic nerve after quadrantic grafts of embryonic retina. Soc. Neurosci. Abs. 7', 404.
FAWCETT, J. W. & GAZE, R. M. (1982). The retino-tectal fibre pathways from normal and
compound eyes in Xenopus. J. Embryol. exp. Morph., in the press.
GAZE, R. M. & GRANT, P. (1978). The diencephalic course of regenerating retino-tectal fibres
in Xenopus tadpoles. /. Embryol. exp. Morph. 44, 201-216.
GAZE, R. M. & STRAZNICKY, C. (1980). Regeneration of optic nerve fibres from a compound
eye to both tecta in Xenopus: evidence relating to the state of specification of the eye and
the tectum. /. Embryol. exp. Morph. 60, 125-140.
GLASTONBURY, J. & STRAZNICKY, K. (1978). Aberrant ipsilateral retino-tectal projection
following optic nerve section in Xenopus, Neuroscience Letts. 7, 67-72.
HOPE, R. A., HAMMOND, B. J. & GAZE, R. M. (1976). The arrow model: retino-tectal specificity and map formation in the goldfish visual system. Proc. R. Soc. Lond. B 194, 447-466.
JACOBSON, M. & LEVINE, R. L. (1975). Stability of implanted duplicated tectal positional
markers serving as targets for optic axons in adult frogs. Brain Research 92, 468-471.
NIEUWKOOP, P. D. & FABER, J. (1956). Normal table of Xenopus laevis (Daudin). Amsterdam:
North Holland.
SCHMIDT, J. T. (1978). Retinal fibers alter tectal positional markers during the expansion of
the half retinal projection in goldfish. J. comp. Neurol. 177, 279-300.
SCHOLES, J. H. (1979). Nervefibretopography in the retinal projection to the tectum. Nature
278, 620-624.
STEEDMAN, J. G. (1981). Pattern formation in the visual pathways of Xenopus laevis. Ph.D.
thesis, University of London.
STRAZNICKY, C., GAZE, R. M. & HORDER, T. J. (1979). Selection of appropriate medial branch
of the optic tract byfibresof ventral retinal origin during development and in regeneration;
an autoradiographic study in Xenopus. J. Embryol. exp. Morph. 50, 253-267.
STRAZNICKY, C. & TAY, D. (1981). Spreading of hemiretinal projections in the ipsilateral
tectum following unilateral enucleation: a study of optic nerve regeneration in Xenopus with
one compound eye. /. Embryol. exp. Morph. 61, 259-276.
(Accepted 1 October 1982)