The growth of the retina has been studied in Xenopus by use of autoradiography with tritiated thymidine. At the time when retinal polarization first occurs (around stage 30) there are only some 20 ganglion cells across the retinal equator and the rest of the retina develops later, by annular addition of cells at the ciliary margin. This process continues beyond metamorphosis.

The regeneration of retinotectal connections has been extensively studied in amphibians and it is well established that a fibre projection of approximately normal retinotopic arrangement may be re-established across the tectum when the optic nerve fibres have been cut and allowed to regenerate. The most widely accepted hypothesis which has been proposed to account for these findings is the hypothesis of neuronal specificity (Sperry, 1943, 1944, 1945, 1951, 1965). According to this hypothesis the retinal ganglion cells each acquire a specific cytochemical individuality during neurogenesis, the tectal neurons also become cytochemically specified in a comparable and matching fashion, and the eventual outgrowth of axons from the developing retina allows similarly specified retinal and tectal neurons to link synaptically.

Such a hypothesis obviously tends to focus our interest on the events taking place in retina and tectum during the critical stages of embryonic and larval life. We need to know more about the mode of development or retina, tectum and the optic nerve joining them. The nature of the retinotopic fibre projection in the adult amphibian is well known from electrophysiological analysis (Gaze, 1958; Gaze & Jacobson, 1963; Gaze, Jacobson & Székely, 1963; Cronly-Dillon, 1968). But whereas in the adult animal we are dealing with a fibre projection that is, if not static in terms of its connectivity pattern, at least quiescent, this is by no means the case during larval development. During larval life the eye and the tectum both grow in size and acquire more cells. And it seems likely that a study of the mode of this growth and the temporally ordered sequence of extra connexions that must form between the expanding retina and the expanding tectum, could help us to understand the mechanisms that exist to control the development of the ordered connexions found in the adult. As a first step in this direction this paper examines the mode of growth of the retina, studied by means of autoradiography with tritiated thymidine (3H-T).

Eggs were produced by appropriate treatment of Xenopus with chorionic gonadotrophin and the resulting embryos were reared in the laboratory, in some cases until after metamorphosis. Single injections of 3H-T were made into the belly region of young larvae and intraperitoneally in older larvae and juveniles. Larvae were staged according to the normal tables of Nieuwkoop & Faber (1956). At stage 30 or younger, 1 μCi was injected per animal; at stages 33 and 35, 2μCi; at stages 40 and 45, 5 μCi; at stages 53, 58, 61 and after metamor-phosis, 10μCi. The 3H-T had a specific activity of 5 Ci/m mole. Injections into older larvae and juveniles were made using an ordinary 1 ml tuberculin syringe with a 25 G hypodermic needle, while injections into young larvae were made using a glass micropipette with a plunger driven by a micrometer screw. Animals were injected at stages 30, 33, 35, 40, 45, 53, 58, 61, 66 (metamorphosis) and 3 weeks after metamorphosis (juvenile). In the Results section we consider animals injected at stages 30, 35, 45, 58 and 3 weeks after metamorphosis, as these were representative of the complete series. Several animals were injected at each stage and these were then killed at selected intervals after injection. The stages at which animals were killed for autoradiography corresponded to those given above. In some animals cumulative labelling with two to three injections between stages 27 and 29 was performed. The animals were reared at approximately 20° C.

Tissues were fixed in Carnoy’s fixative for 3-24 h, then rapidly processed, cleared in CHC13, embedded in paraffin wax, cut at 3–6μm and mounted on slides. Deparaffinized sections were coated with Ilford Nuclear Research Emulsion G 5 and were exposed at 5°C for 3–9 weeks before being developed. The sections were then stained with cresyl fast violet.

The mode of growth of the eye has been studied mainly from sections cut vertically so as to include the ciliary margins of the retina above and below the lens and at the same time to include the maximal diameter of the eye posteriorly. In such sections the number of ganglion cells can be counted across the equator of the eye, from the ciliary margin on one side to that on the other.

Label given at stage 30

At stage 30 the eye has only some 20 ganglion cells across the equator. If the animal is given 3H-T at stage 30 and killed for autoradiography 2 h later the eye is found to be extensively labelled (Fig. 1). If an animal is labelled at stage 30 and killed at stage 35, autoradiography reveals massive labelling near the ciliary margins, the relative distribution of the label indicating that the cells farthest away from the margin have undergone fewer divisions than those at the margin (Fig. 2). Most of the ganglion cells to be seen across the equator are unlabelled and thus were probably formed prior to the administration of the 3H-T. Animals labelled at stage 30 and autoradiographed at or 3 months after metamorphosis show that the retinal label is then confined to a small region around the exit of the optic nerve at the back of the eye (Figs. 3, 4). Thus the entire extent of the retina of the stage 30 eye eventually comes to comprise only this small disk of retina around the optic nerve head in the postmetamorphic animal, and all the rest of the juvenile retina has developed later than stage 30. There are some 300-400 ganglion cells across the equator of the retina in a 6-month juvenile.

Fig. 1.

In all the illustrations the bar represents 100 μm

Fig. 1. Eye from an animal labelled at stage 30 and killed 2 h later.

Fig. 1.

In all the illustrations the bar represents 100 μm

Fig. 1. Eye from an animal labelled at stage 30 and killed 2 h later.

Fig. 2.

Eye from an animal labelled at stage 30 and killed at stage 35.

Fig. 2.

Eye from an animal labelled at stage 30 and killed at stage 35.

Fig. 3.

Fundus of the eye from an animal labelled at stage 30 and killed at stage 66 (metamorphosis). The arrows indicate the regions of labelled cells adjacent to the optic nerve head (ON). The rest of the retina was unlabelled.

Fig. 3.

Fundus of the eye from an animal labelled at stage 30 and killed at stage 66 (metamorphosis). The arrows indicate the regions of labelled cells adjacent to the optic nerve head (ON). The rest of the retina was unlabelled.

Fig. 4.

Optic nerve head (ON) from an animal labelled at stage 30 and killed 3 months after metamorphosis. Arrows point to groups of labelled cells. The rest of the retina was unlabelled.

Fig. 4.

Optic nerve head (ON) from an animal labelled at stage 30 and killed 3 months after metamorphosis. Arrows point to groups of labelled cells. The rest of the retina was unlabelled.

If an animal is labelled at stage 35 (when there are 20–30 ganglion cells across the equator) and killed 2 h later, autoradiography shows heavy labelling at the ciliary margins while the central 20 or so ganglion cells in the section are unlabelled (Fig. 5). By stage 45 the continuing cell division at the ciliary margin has diluted the label somewhat in this region and the mass of labelled cells appears farther towards the fundus (Fig. 6). In the animal illustrated there were some 40 ganglion cells across the retinal equator, of which the central 20 were mostly unlabelled and had thus been formed before administration of the label. The labelled edges of the retina had been formed between stages 35 and 45. The retina grows very rapidly at this period in the animal’s development and Fig. 7 shows the distribution of retinal label at stage 48. Here there are some 66 ganglion cells across the retinal equator. The central 20 cells are mostly unlabelled, having undergone their final DNA synthesis before administration of the thymidine at stage 35. The two fringes of labelled ganglion cells now appear to be displaced considerably towards the fundus, due to the extensive mitosis which has occurred at the ciliary margin and has added on new, unlabelled (and dilutely labelled) cells at the edge of the retina. By 3 months after metamorphosis the retina is completely free of label except for a small group of cells gathered round the optic nerve head (Figs. 8, 9).

Fig. 5.

Eye from an animal labelled at stage 35 and killed 2 h later.

Fig. 5.

Eye from an animal labelled at stage 35 and killed 2 h later.

Fig. 6.

Eye from an animal labelled at stage 35 and killed at stage 45.

Fig. 6.

Eye from an animal labelled at stage 35 and killed at stage 45.

Fig. 7.

Eye from an animal labelled at stage 35 and killed at stage 48.

Fig. 7.

Eye from an animal labelled at stage 35 and killed at stage 48.

Fig. 8.

Optic nerve head (ON) from an animal labelled at stage 35 and killed three months after metamorphosis. Arrows indicate groups of labelled cells. The rest of the retina was unlabelled.

Fig. 8.

Optic nerve head (ON) from an animal labelled at stage 35 and killed three months after metamorphosis. Arrows indicate groups of labelled cells. The rest of the retina was unlabelled.

Fig. 9.

Section through fundus, close to optic nerve head, from a different animal, labelled at stage 35 and killed 3 months after metamorphosis. Arrows indicate labelled regions of retina.

Fig. 9.

Section through fundus, close to optic nerve head, from a different animal, labelled at stage 35 and killed 3 months after metamorphosis. Arrows indicate labelled regions of retina.

Label given at stage 45

When 3H-T is given at stage 45 and the animal is killed 2 h later, the distribution of retinal label is as shown in Fig. 10. Essentially the labelled cells are confined to the ciliary margin, where mitosis and DNA synthesis are occurring. Occasionally labelled cells can be found in the fundus of the retina; these may represent glial elements developing later than neural cells (Jacobson, 1968b). At this stage there are some 40 ganglion cells across the retinal equator and they are all unlabelled. By stage 61 extensive further growth of the retina gives a result such as that shown in Figs. 11, 12 and 13. The two pools of labelled cells indicate what part of the retina was actively proliferating when the label was given at stage 45; the central, unlabelled, part of the retina (37 cells across) had been formed prior to administration of the label and the unlabelled edges, comprising the greater part of the retina, have developed since the thymidine was given. In Fig. 11 the dorsal unlabelled edge of the retina contains 80 ganglion cells while the ventral unlabelled edge contains 65. In this section there were approximately 200 ganglion cells across the whole equator of the retina.

Fig. 10.

Eye from an animal labelled at stage 45 and killed 2 h later. Arrows indicate groups of labelled cells at the ciliary margins of the retina.

Fig. 10.

Eye from an animal labelled at stage 45 and killed 2 h later. Arrows indicate groups of labelled cells at the ciliary margins of the retina.

Fig. 11.

Eye from an animal labelled at stage 45 and killed at stage 61. Arrows indicate bands of labelled cells in the fundus.

Fig. 11.

Eye from an animal labelled at stage 45 and killed at stage 61. Arrows indicate bands of labelled cells in the fundus.

Fig. 12.

Higher magnification of the fundus shown in Fig. 11.

Fig. 12.

Higher magnification of the fundus shown in Fig. 11.

Fig. 13.

Fundus and optic nerve head (ON) from an animal labelled at stage 45 and killed at stage 6.1. Arrows indicate narrow bands of labelled cells, one on each side of the optic disk.

Fig. 13.

Fundus and optic nerve head (ON) from an animal labelled at stage 45 and killed at stage 6.1. Arrows indicate narrow bands of labelled cells, one on each side of the optic disk.

Label given at stage 58

Growth of the eye in Xenopus continues up to and after metamorphosis, although the rate of addition of new retinal cells slows down in the later phases of growth. Fig. 14 is taken from an animal to which 3H-T had been given at stage 58, shortly before metamorphic climax, and which had been killed 24 h later. The retinal margin is seen to be actively proliferating. If the animal has been labelled at stage 58 and allowed to survive until 3 months after metamorphosis, the retinal label is then found to occupy a position some short way in from the ciliary margin (Fig. 15), thus indicating again that further proliferation has occurred at the edge of the retina. Fig. 15 also illustrates a common finding, which is that the labelled cells in the bipolar and receptor layers extend farther towards the fundus than do the labelled cells in the ganglion cell layer.

Fig. 14.

Ciliary margin of the retina from an animal labelled at stage 58 and killed 24 h later. A group of labelled cells is shown at the ciliary margin. The rest of the retina was unlabelled.

Fig. 14.

Ciliary margin of the retina from an animal labelled at stage 58 and killed 24 h later. A group of labelled cells is shown at the ciliary margin. The rest of the retina was unlabelled.

Fig. 15.

Segment of retina from an animal labelled at stage 58 and killed 3 months after metamorphosis. The part of retina shown is some distance in from the ciliary margin, which is beyond the top of the photomicrograph. The arrows indicate the positions of the labelled cells farthest from the ciliary margin in each of the three layers of the retina.

Fig. 15.

Segment of retina from an animal labelled at stage 58 and killed 3 months after metamorphosis. The part of retina shown is some distance in from the ciliary margin, which is beyond the top of the photomicrograph. The arrows indicate the positions of the labelled cells farthest from the ciliary margin in each of the three layers of the retina.

Throughout its development the eye grows in a fashion which is asymmetrical about the optic nerve head. This is well shown in Figs. 7, 10 and 11 and is further illustrated in Fig. 16. In this case the label was given at stage 58 and the animal killed 3 months after metamorphosis. The labelled regions on each side of the retina are shown in Figs. 17 and 18. On the dorsal limb of the retina (Fig. 17) the labelled cells are some way from the ciliary margin, while on the ventral limb (Fig. 18) they are much closer to the edge of the retina. More cells have thus been added to the dorsal edge than to the ventral edge of the retina.

Fig. 16.

Eye from an animal labelled at stage 58 and killed 3 months after metamorphosis. The labelled regions of retina are indicated by boxes A and B and these regions are shown, at higher magnification, in Figs. 17 and 18.

Fig. 16.

Eye from an animal labelled at stage 58 and killed 3 months after metamorphosis. The labelled regions of retina are indicated by boxes A and B and these regions are shown, at higher magnification, in Figs. 17 and 18.

Fig. 17.

Enlargement of box A from Fig 16. Arrows indicate the site of labelled cells in the ganglion cell layer and in the bipolar cell layer.

Fig. 17.

Enlargement of box A from Fig 16. Arrows indicate the site of labelled cells in the ganglion cell layer and in the bipolar cell layer.

Fig. 18.

Enlargement of box B from Fig 16. Arrow shows the site of labelled cells.

Fig. 18.

Enlargement of box B from Fig 16. Arrow shows the site of labelled cells.

It is perhaps worth mentioning that cells labelled at any particular time form more or less distinct groups in the retina (Figs. 3, 12 and 15). This may indicate that cells which had undergone their final mitosis at a given time kept their relative positions and did not move farther away from each other or spread over a large sector of the retina.

Further observations

Cumulative labelling performed between stages 27 and 29 resulted in those cells developing before stage 30 being massively labelled. Autoradiography of such animals at various stages of larval life suggested to us that the numbers of labelled cells did not decrease significantly over a period of up to 12 weeks. This persistence of labelling was also found consistently in the other groups of animals (Figs. 2–3, 7–9) and indicates that no considerable proportion of these cells later degenerate and disappear.

The growth of Xenopus eye continues after metamorphosis for an undetermined period. Animals given 3H-T 3 weeks after metamorphosis (at stage 66 + ) and killed 48 h later, show labelled cells present at the ciliary margin of the retina.

The Xenopus sensory retina grows throughout larval and into postmeta-morphic life by the addition of cells to all three layers at the ciliary margin. The development of the retina is shown diagrammatically in Fig. 19, which represents a vertical section through the eye, lens and optic nerve.

Fig. 19.

Synoptic diagram indicating the temporal mode of growth of the retina in Xenopus. The stage of the retina and the size of the subdivisions are only approximately correct. S, stage of development (Nieuwkoop & Faber, 1956); M, metamorphosis; M + 3, 3 months after metamorphosis; D, days; W, weeks.

Fig. 19.

Synoptic diagram indicating the temporal mode of growth of the retina in Xenopus. The stage of the retina and the size of the subdivisions are only approximately correct. S, stage of development (Nieuwkoop & Faber, 1956); M, metamorphosis; M + 3, 3 months after metamorphosis; D, days; W, weeks.

It can be seen that the ontogenetically oldest cells are grouped around the exit of the optic nerve. This group of cells is followed by progressively younger cells as we go towards the retinal margin. These results are in accord with previous autoradiographic studies on developing amphibian (Jacobson, 1968 b), chick (Fujita & Horii, 1963), and mouse (Sidman, 1961) eye, which showed that mitosis first ceases in the central zone of the retina. The present results show that after stage 35 in Xenopus, newly formed retinal cells originate entirely from the retinal margin. The situation thus resembles that found in the regenerating adult newt retina (Gaze & Watson, 1968) and in the developing retina of Rana (Hollyfield, 1968). This latter author has shown labelled cells moving from the retinal margin into the inner nuclear towards the fundus of the eye. Apart from the sort of situation shown in Figs. 15 and 17, we have found no evidence of intraretinal cell movement of Xenopus.

Glücksmann described cell degeneration in the early stages of the developing frog eye (Glücksmann, 1965) and assumed that new cells would form to replace the degenerated ones. Our observations do not support the idea of any considerable degeneration of differentiated cells in the retina of Xenopus’, if this were to happen, noticeable numbers of cells labelled at an early larval stage should have disappeared by the time of or after metamorphosis. Indeed, it would be surprising if any significant amount of late cell degeneration, or of intraretinal cell movement were to occur, since this would upset the topological relationship between retina and tectum.

From the present analysis of retinal growth two problems become obvious. The first has to do with the ‘specification’ of retinal ganglion cells and the second with the way in which the developing retina connects with the developing brain.

During development the eye connects with the brain in an orderly fashion. In Xenopus, this orderly fibre projection is such that the nasal part of the retina connects with the caudal part of the optic tectum; temporal retina connects with rostral tectum; inferior retina connects with dorsal tectum and superior retina connects round the lateral edge of the tectum. This orderly projection may be restored when an adult optic nerve is cut and allowed to regenerate and there is much evidence (reviewed in Gaze, 1970) to suggest that each ganglion cell ‘knows’ where it is in the retina and ‘knows’ when its axon has got to the right place on the tectum. The system behaves, in other words, as though the ganglion cells of the retina are specified in a positional sense. Furthermore, this positional labelling of the retinal ganglion cells occurs at a known stage of development. Jacobson (1968 a) has shown that the eye becomes polarized across the nasotemporal axis at stage 30 and across the dorsoventral axis a few hours later. If an eye from a stage 28 embryo is rotated, the animal later shows normal vision. If the eye is rotated in a stage 32 embryo, the animal later shows visual behaviour that is upside-down and back-to-front. Thus at about stage 30 polarization of the eye takes place and thereafter the ganglion cells will normally only connect with their appropriate regions of tectum; i.e. they behave as if they have been specified.

The initial formation of ganglion cells in Xenopus, which occurs at about stage 30, was studied autoradiographically by Jacobson (1968b), who showed that the ganglion cell precursors underwent their final DNA synthesis shortly before the retina became polarized across the nasotemporal axis. However, at stage 30 only a minute part of the cells that will make up the adult retina have been formed (Fig. 19). Jacobson (1968 b) claimed that cumulative labelling (of Xenopus eye) started at stage 30 and continued over several days gave no labelling of ganglion cells; and furthermore, he stated that by stage 35 DNA synthesis had ceased in all the cells which later differentiated into receptor cells. Both these statements would appear to be incorrect in view of the results described in the present paper. Simple consideration of the differences between the size of the eye at stage 35 and in the adult must indicate that all layers of the stage 35 retina receive many new cells to permit adult proportions to be achieved and this addition of cells is illustrated in Figs. 15, 17 and 18. Jacobson (19686) held that, at stage 29, DNA replication ceases in all the neuroblasts which later differentiate into ganglion cells, with the exception of a small percentage at the periphery of the retina. This ‘small percentage’ is what we are concerned with in the present paper; and it gives rise to almost the whole of the adult retina, which develops after retinal polarization has occurred. And since the rest of the eye develops with a polarity consistent with the initially polarized part (even if rotated), it would seem that the ‘specification’ of the later developing ganglion cells is transmitted to the new cells somehow as they appear. The details of this process of specification are completely unknown but may be related to the differing life histories of the two daughter cells that result from the division of a cell at the retinal margin. Thus, the autoradiographic evidence suggests that, of two such daughter cells, the one nearer the fundus heads in the direction of differentiation, while the cell nearer the margin of the retina will divide further, and so on.

The second problem to be raised by this study is that of how the developing retina connects with the brain. The adult projection from retina to tectum has already been mentioned; the edges of the retina project in order round the edges of the tectum and the centre of the retina projects (approximately) to the centre of the dorsal surface of the tectum. But when optic nerve fibres first reach the tectum (shortly after stage 35), only the central regions of the retina have yet been formed. All the rest of the retina is added later, at the edges. Thus, if the initially innervated piece of tectum is the place to which the central retinal fibres project later in the animal’s life, then the tectum should also grow at the edges, to accommodate, in a proper order, the newly arriving fibres from the retinal margins. The mode of growth of the tectum in relationship to the development of the retina is presently being investigated.

We would like to thank Miss E. M. Forrest for her expert histological assistance. K. Straznicky was a Wellcome Research Fellow for 1969.

Cronly-Dillon
,
J.
(
1968
).
Pattern of retinotectal connections after retinal regeneration
.
J. Neurophysiol.
31
,
410
418
.
Fujita
,
S.
&
Horii
,
M.
(
1963
).
Analysis of cytogenesis in chick retina by tritiated thymidine autoradiography
.
Archvm. histol. jap.
23
,
359
366
.
Gaze
,
R. M.
(
1958
).
The representation of the retina on the optic lobe of the frog
.
Q. Jl exp. Physiol.
43
,
209
214
.
Gaze
,
R. M.
(
1970
).
The Formation of Nerve Connections.
London
:
Academic Press
.
Gaze
,
R. M.
&
Jacobson
,
M.
(
1963
).
A study of the retinotectal projection during regeneration of the optic nerve in the frog
.
Proc. Roy. Soc. B
157
,
420
448
.
Gaze
,
R. M.
,
Jacobson
,
M.
&
Székely
,
G.
(
1963
).
The retinotectal projection in Xenopus with compound eyes
.
J. Physiol., Lond.
165
,
484
499
.
Gaze
,
R. M.
&
Watson
,
W. E.
(
1968
).
Cell division and migration in the brain after optic nerve lesions
, in
Growth of the Nervous System. Ciba Foundation Symposium
13
(eds.
G. E. W.
Wolstenholme
and Maeve O’Connor
), pp.
53
67
.
London
:
Churchill
.
Glücksmann
,
A.
(
1965
).
Cell death in normal development
.
Archs Biol., Liège,
76
,
419
437
.
Hollyfield
,
J. G.
(
1968
).
Differential addition of cells to the retina in Rana pipiens tadpoles
.
Devi Biol.
18
,
163
179
.
Jacobson
,
M.
(
1968a
).
Development of neuronal specificity in retinal ganglion cells of Xenopus
.
Devi Biol.
17
,
202
218
.
Jacobson
,
M.
(
1968b
).
Cessation of DNA synthesis in retinal ganglion cells correlated with the time of specification of their central connections
.
Devi Biol.
17
,
219
232
.
Nieuwkoop
,
P. D.
&
Faber
,
J.
(
1956
).
Normal Table of Xenopus laevis (Daud
.).
Amsterdam
:
North Holland
.
Sidman
,
R. L.
(
1961
).
Histogenesis of mouse retina studied with thymidine -3H
. In
Structure of the Eye
(ed.
G. K.
Smelser
), pp.
487
505
.
New York
:
Academic Press
.
Sperry
,
R. W.
(
1943
).
Visuomotor co-ordination in the newt (Triturus viridescens) after regeneration of the optic nerve
.
J. comp. Neurol.
79
,
33
55
.
Sperry
,
R. W.
(
1944
).
Optic nerve regeneration with return of vision in anurans
.
J. Neurophysiol.
7
,
57
70
.
Sperry
,
R. W.
(
1945
).
Restoration of vision after crossing of optic nerves and after contralateral transplantation of eye
.
J. Neurophysiol.
8
,
15
28
.
Sperry
,
R. W.
(
1951
).
Mechanisms of neural maturation
. In
Handbook of Experimental Psychology
(ed.
S. S.
Stevens
), pp.
236
280
.
New York
:
Wiley
.
Sperry
,
R. W.
(
1965
).
Embryogenesis of behavioral nerve nets
. In
Organogenesis
(eds.
R. L.
De Haan
and
H.
Ursprung
), pp.
161
186
.
New York
:
Holt, Rinehart and Winston
.