The retinotectal projections from surgically rounded-up half-eyes in Xenopus

In two recent papers (Straznicky & Gaze, 1980; Gaze & Straznicky, 1980) we have shown that when various types of ‘compound eyes’ are formed surgically in Xenopus embryos, the retinal fragments that form the compound eye project to the tectum with the orientations that they would have shown, had they been normally positioned as parts of normal eyes. The partial visuo-topic map from each fragment is thus oriented, in relation to a normal map, as are the retinal fragments in relation to a normal eye. Thus the developmental programme among cells in the developing eye, that relates to the orientation of the retinotectal map, is stable after these types of operations.

If the nasal or temporal half of an eye in a Xenopus embryo of stage 32 (Nieuwkoop & Faber, 1956) is removed, the residual half-eye eventually produces, in most cases, what appears to be a full-size normal eye with a normal retinotectal projection (Berman & Hunt, 1975; Feldman & Gaze, 1975). Although this has not been investigated in detail, it seems likely that the restitution of a normal eye from a half-eye, with the eventual development of a normal retinotectal map, is related to extra mitosis following the operation, probably occurring at the unapposed cut edge (Horder & Spitzer, 1973), whereby the missing positional values of the half-eye are replaced.

We have noticed an interesting relationship between growth and pattern formation in half-eyes. If the nasal, temporal or dorsal half of an eye is removed in a Xenopus embryo of stage 32, and the residual half-eye is then surgically rounded-up, by partially freeing it from the surrounding mesenchyme and folding the fragment so that the cut edges are apposed, the half-eye heals up to form an eye which is at all stages smaller than normal and which may show distortion of the retinotectal map which reflects the distortion imposed on the developing eye at operation.

We describe in this paper the nature of these distortions and discuss the results in the light of current ideas on the question of stability versus modifiability of the programme for retinal development. Some of the results presented come from the original experiments performed in 1969 while others are recent. The findings have been consistent throughout the entire series of experiments.

Xenopus embryos of stage 32 were operated in full-strength Niu-Twitty solution containing MS 222 (tricaine methane sulphonate, Sandoz) at a concentration of 1:5000.

The right eye of an embryo was exposed and the nasal, temporal or dorsal half of the eye anlage, together with the lens, was freed by use of glass needles and then removed by suction with a Spemann pipette. The residual half-eye anlage was partially separated from the surrounding mesenchymal tissue and the edges of the eye rudiment were surgically apposed. After the rounding-up of the eye the line of fusion was directed rostrally in temporal halves, caudally in nasal halves and dorsally in ventral halves (Fig. 1). The approximated edges of the eye were held together for about 30 minutes by using splinters of coverslip glass as weights, after which the cut edges remained fused permanently. Animals were then transferred to fresh Niu-Twitty solution and 24 h after the operation each of the operated eyes was checked under a stereo microscope to determine the success of the operation. Animals which revealed a 10° or wider gap between the approximated edges of the operated eye were discarded. The remaining animals were reared to metamorphosis and beyond, and visuotectal projections from the rounded-up half-eyes were mapped electrophysiologically. The technique of mapping and the histology used have been described in a previous paper (Straznicky & Gaze, 1980).

Visuotectal maps through the operated eye to the contralateral tectum were recorded from 44 animals (Table 1). In 11 of these animals the operated eye was a nasal half (⁠ NR), in 20 a temporal half-eye (⁠ TR) and in 13 a ventral half-eye (⁠ VR). The largest class of results from each type of operation was that in which the orientation of the visuotectal map paralleled the deformation imposed on the eye at operation. Figure 2 shows the nature of the map to be expected from rounded-up half-eyes, on the basis that the programme relating to the orientation of the projection is stable and that the projection spreads to fill the available tectal space, as it does with compound eyes.

A smaller class of results from NR and TR experiments was one where the maps showed complete or partial mirror-reduplication, resembling that seen in maps from compound eyes. A few results, from each type of operation, gave normal maps. In all cases the projection extended across the whole of the mappable surface of the tectum.

Normal maps, even from normal eyes, differ from one another, and the decision as to what can be called a normal map is subjective. Figure 3 illustrates what we have called a ‘normal map’, in this case coming from a rounded-up temporal half-eye. Essentially, in a normal map, latero-medial rows of tectal recording positions give corresponding ventro-dorsal rows of stimulus positions in the visual field. To lessen the selective and subjective elements inherent in the classification of the experimental results, we have chosen to illustrate the findings with several examples from each class of map so that the reader can decide whether or not the descriptions are valid.

Five out of 11 maps were oriented according to the nature of the operation (Fig. 4). Rows of stimulus positions in the temporal field (corresponding to the least affected part of the retina) are normally oriented. In the nasal field, corresponding to the most distorted part of the retina, the rows of stimulus positions curl towards the nasal pole of the field. These eyes were smaller than the normal eyes.

In four animals the operated eye gave a mirror-reduplicated projection resembling that from a surgically formed compound double-nasal eye (Fig. 5). In one of these the dorso-ventral alignment of the projection was normal and in the other the map was rotated 90° clockwise, in accord with the position of the fissure, which pointed caudally.

One animal gave an approximately normal visuotectal map; and the map from the remaining animal was unclassifiable, in that the tectum was covered with a chaotic projection from a small region of central visual field.

Eleven out of 20 maps were oriented according to the nature of the operation (Fig. 6). Rows of stimulus positions in the nasal field, corresponding to the least affected part of the retina, are normally oriented. In the temporal field, corresponding to the most distorted part of the retina, the rows of stimulus positions curl towards the temporal pole of the field. The eyes were small.

Four animals gave maps showing mirror-reduplication resembling that seen with surgically produced double-temporal compound eyes (Fig. 7). In both the cases shown the reduplication is seen in dorsal field, while temporoventral field (naso-dorsal retina) is not reduplicated but shows instead the distortion typical of this operation.

Four maps were classified as approximately normal and one showed a chaotic projection from a small region of central field to the whole tectum.

Eleven out of 13 eyes gave maps which were oriented according to the nature of the operation (Fig. 8). These maps show ‘barrelling’ of the visual field contour lines, from dorsal to ventral, in both nasal and temporal parts of the visual field. The field projection also tends to extend much further ventrally than it does in a normal animal. The operated eyes were smaller than normal and all showed gold pigmentation dorsally with a normally positioned ventral fissure. Two animals gave an approximately normal projection.

Previous observations on compound eyes have shown that the retinal fragments retain their original programming for (1) generating within the eye the particular cell characteristics responsible for the orientation of the visuo-tectal map (Straznicky & Gaze, 1980; Gaze & Straznicky, 1980); (2) carrying out the characteristically timed pattern of retinal histogenesis (Straznicky & Tay, 1977); and (3) forming selectively nerve fibre tracts which are positioned appropriately according to the nature of the retinal fragment (Straznicky, Gaze & Horder, 1979). Further evidence for the existence of stable programming in retinal fragments has come from study of the regeneration of optic nerve fibres from compound eyes, where it has been shown that, when such fibres regenerate to the ipsilateral tectum, they initially innervate only part of it; and the part innervated corresponds to the nature of the compound eye (Gaze & Straznicky 1979).

If we assume that the rounded-up half-eyes in the present experiments also maintain intact their developmental programmes for map orientation, an assumption which implies that retinal cell properties relating to map orientation are, in this situation, related to cell lineage, we would expect the maps to reflect the distortions imposed on the retina at operation (Fig. 2) and the results presented show that this is so. Since each operated eye will initially contain only half the positional values of a complete eye, we might expect the initial tectal projection to be restricted to the corresponding half of the tectum, as occurred in the study on regeneration quoted above (Gaze & Straznicky, 1979). We know, however, that each (similar) part of a conventional compound eye will eventually spread across the whole of the contralateral tectum during development (Straznicky, Gaze & Keating, 1971), and the present results indicate that such a spreading also occurs with the projections from rounded-up half-eyes.

The nature of the operation suggests that the distortions found in the visuotopic projections from rounded-up half-eyes should be symmetrical about the vertical meridian for ventral half-eyes and symmetrical about the horizontal meridian for nasal and temporal half-eyes (Fig. 2). This prediction is fulfilled for ventral half-eyes but not for the other varieties, where the distortion seen is confined mainly to the dorsal part of the field. This is perhaps because more ventral parts of the field would project too far laterally on the tectum to be recorded. It is also possible that the injury to the region of the ventral fissure, which must occur in operations for nasal or temporal half-eyes but which does not occur with ventral half-eyes, may disturb the retinotopic arrangement of fibres as they form the optic nerve, and this may be reflected in the nature of the map.

Four out of 11 NR eyes, and 4 out of 20 TR eyes, gave maps showing mirror-reduplication resembling that seen in maps from conventional NN and TT compound eyes respectively. The occasional recurrence of mirror-redupli-cation in unrounded-up half-eyes has been described previously (Gaze, 1970; Berman & Hunt, 1975; Feldman & Gaze, 1975); and in the previous paper of this series (Gaze & Straznicky, 1980) we have shown that mirror-reduplication may be induced when compound eyes are formed from retinal fragments that would not be expected to give this type of pattern, provided that the operations are performed in a medium of low ionic strength. We have suggested, therefore, that mirror-reduplication should be regarded as a form of tissue response to injury, particularly when the injury is aggravated by the use of operating solutions that damage the cells and slow down healing.

Perhaps the most interesting aspect of the reduplications seen in the present work is that several of them are partial. Figure la and b illustrate maps from TR eyes; in each case the main projection is reduplicated in TT style and the small region of ventro-temporal field shows an additional, non-reduplicated, partial projection which has the orientation characteristic of maps from a nonreduplicated TR eye. Figure 6 c shows a projection from a TR eye in which two points are reduplicated. This combination of reduplicated projection with a distorted, partial, non-reduplicated projection, was not seen in the maps from NR eyes. Moreover, no map from a VR eye showed reduplication. The cellular events underlying reduplication of this sort are unknown, and the phenomenon will repay extensive study.

Surgically rounded-up half-eyes are smaller than half-eyes that have not been rounded-up in this way. The difference in size persists into adult life. We have previously shown that the Xenopus eye grows by mitosis at the ciliary margin (Straznicky & Gaze, 1971). It is probable that when a half-eye is left unrounded-up, a new complete ciliary margin is rapidly formed by mitosis at the cut edge of the residual half-eye. Furthermore, regeneration of a new halfeye from tissue in the optic stalk is a likely possibility (Gaze, Feldman, Cooke & Chung, 1979) and in this case the eye eventually formed might be expected to be normal in all respects.

It seems that the process of surgically rounding-up the half-eye, whereby the cut ciliary margins are apposed, prevents in some way the re-formation of the missing ciliary margin; and the eye so treated is thereafter stunted in its growth, never regains its full cell complement, and remains a half-eye in terms of its cellular positional values. Further evidence for this last point comes from the work of Steedman, who has shown, by the method of cobalt impregnation, that the optic pathway from a rounded-up ventral half-eye forms only a medial tract; the lateral tract, which normally carries retinal fibres of dorsal origin, is missing (Steedman, in preparation).

Further work is needed to reveal the nature of the cellular mechanisms responsible for the different behaviour of half-eyes that are rounded-up and those that are not. The present experiments, however, provide further evidence for the stability of the retinal developmental programme in various abnormal situations.

These experiments were partly performed during the tenure by C.S. of a Wellcome Research Fellowship.