The formation of the area centralis of the retinal ganglion cell layer in the chick

Transcription

1 Development 100, (1987) Printed in Great Britain The Company of Biologists Limited The formation of the area centralis of the retinal ganglion cell layer in the chick CHARLES STRAZNICKY and MARK CHEHADE Department of Anatomy and Histology, School of Medicine, The Flinders University of South Australia, Bedford Park, SA 5052 Adelaide, Australia Summary In adult domestic chickens, the neurones in the retinal ganglion cell layer are very unevenly disposed such that there is a sixfold increase in neurone density from the retinal edge to the retinal centre. The formation of the high ganglion-cell-density area centralis was studied on chick retinal wholemounts from the 8th day of incubation (E8) to 4 weeks after hatching (4WAH). The density of viable neurones and the number and the distribution of pyknotic neurones in the ganglion cell layer were estimated across the whole retina. Between E8 and E10, the distribution of neurones in the ganglion cell layer was anisodensitic with mm" 2 in the centre compared to mm" 2 in the periphery of the retina. Thereafter, a progressively steeper gradient of neurone density developed, which decreased from mm" 2 in the retinal centre to 6000 mm~ 2 at the retinal periphery by 4WAH. Neuronal pyknosis in the ganglion cell layer was observed between E9 and E17. From Ell onwards, consistently more pyknotic neurones were found in the peripheral than in the central retina. It was estimated that over the period of cell death approximately twice as many neurones died per unit area in the retinal periphery than in the centre. Retinal area measurements and estimation of neurone densities in the ganglion cell layer after the period of neurone generation and neurone death indicated differential retinal expansion, with more expansion in the peripheral than in the central retina. These observations allow us to conclude that the formation of the area centralis of the chick retina involves (1) slightly higher cell generation in the retinal centre, (2) higher rate of cell loss in the retinal periphery and (3) differential retinal expansion. Key words: retinal ganglion cells, neurone death, area centralis, morphometry, chick embryo, pyknosis. Introduction Most vertebrate species with good visual acuity have a specialized high cell-density region of the ganglion cell layer (GCL) in the form of an area centralis or an elongated visual streak. In the adult chick, the area centralis extends from the retinal centre well into the temporosuperior quadrant of the retina (Ehrlich, 1981). The neurone density in the area centralis of the GCL is mm~ 2 and decreases to below 4000 mm" 2 at the edge of the retina. Such regional differences are not present during early retinal development (Rager, 1980). How the specialized, highly anisodensitic retinal GCL develops from an initial isodensitic, undifferentiated sheath of cells has not yet been adequately studied in the chick. Neurones in the retinal GCL have been shown to be generated between embryonic days 5 and 12 (E5 and E12) in chick embryos (Kahn, 1974). The number of fibres in the optic nerve has been estimated during development, and peaks on E12 at 4 million followed by a steady reduction to about 2-5 million by E17 (Rager & Rager, 1978). The decrease of optic fibre number is accompanied by a ganglion cell loss of similar magnitude between E9 and E16 (Hughes & McLoon, 1979). Besides retinal ganglion cells, displaced amacrine cells are also present in the retinal GCL amounting to about % of the total neurone population according to previous estimates (Ehrlich, 1981; Layer & Vollman, 1982). Neurone generation and neurone death in the chick follow a radial pattern (Rager, 1976), both occurring first in the retinal centre and spreading peripherally at later

2 412 C. Straznicky and M. Chehade developmental stages. The observed pattern of cell generation appears to be similar to those reported in fish (Johns & Easter, 1977), in amphibia (Straznicky & Gaze, 1971) and in mammals (Rapaport & Stone, 1983). One of the most promising approaches to follow retinal development is that of retinal wholemount studies (Stone, 1981). The neural retina is flatmounted on slides with the GCL lying uppermost. Since the GCL represents a thin sheath of cells in such preparations, the number of neurones, their topographic distribution and regional characteristics within the GCL can be accurately established. Analytical studies on the formation of the mammalian retinal area centralis using retinal wholemounts, suggest the involvement of a number of mechanisms, likely to be acting in combination (Stone, Maslim & Rapaport, 1984). It is possible t,hat (1) retinal ganglion cells are generated in high numbers at the area centralis, (2) ganglion cell precursors migrate intraretinally to the retinal centre, (3) preferential cell death occurs among ganglion cells in the retinal periphery, (4) a high proportion' of precursor neurones in the retinal periphery differentiate into amacrine cells and migrate out from the GCL and (5) differential retinal growth, whereby the retina expands more at its ciliary margin than at the centre. The present study in the domestic chick was undertaken to examine the distribution of neurones in the GCL during development on retinal wholemount preparations. It is also attempted to quantify neuronal loss and to examine its topographic distribution across the developing retina. In this report, we show that the high ganglion-cell-density area centralis of the chick GCL evolves from a more or less uniform ganglion cell density by differential cell loss and differential area expansion of the retinal periphery. An abstract of these observations has already been published elsewhere (Straznicky & Chehade, 1986). Materials and methods Experiments were carried out in White Leghorn and Austral Leghorn chick embryos and posthatched chicks up to 4 weeks after hatching (4WAH). Fertilized eggs and hatchlings were obtained from a local supplier (Parafield Poultry, Adelaide). Eggs were incubated at 37-8 C in a forced-draft incubator at 70 % relative humidity and embryos harvested between E8 and E21. The exact developmental stage of the embryos was determined by Hamburger & Hamilton's (1951) criteria. Preparation of retinal wholemounts Neural retinae were prepared for wholemounting according to Stone's (1981) description. In short, eyes were removed from the orbital cavity and placed in saline. The cornea, sclera, choroid and the retinal pigmentous layer were gently dissected away from the neural retina. The rest of the eye was fixed in 10% buffered formalin at ph7-4 for periods varying between 20 min and 2 h according to the size of the tissue. The eye was rinsed in saline then the pecten and the optic nerve were cut along the retinal attachment and the neural retina detached from the vitreous and the ciliary body. Radial cuts were applied on the neural retina and flatmounted on 4% gelatinized slides, the GCL lying uppermost. Retinal wholemount preparations were air dried at room temperature for 4-6 h. Wholemounts were soaked in an 8:1:1 mixture of chloroform: absolute alcohol: ether for 20min, rehydrated and stained with a 1 % solution of cresyl violet (ph4-5) for 60s. Slides were dehydrated through an alcohol series, cleared in xylene and mounted in DPX. Morphometry Retinal neurone population in the GCL was counted in areas, delineated by an ocular square grid. Sampling was carried out at regular intervals, mm apart, across the dorsoventral and nasotemporal axes of the retina at x In all, about 100 areas of mm" 2 were counted in retinae from E8 to E21 embryos and about 250 areas in retinae of posthatched chicks. The similar appearance of neurones in the GCL did not allow a distinction to be made between ganglion cells and displaced amacrine cells, therefore the latter were unavoidably included in the counting. Blood cells and other non-neuronal (glial) cells were usually clearly identifiable under high-power magnification and they were not counted. Glial cells were identified by their irregular shape, lack of Nissl substance and a rodshaped nucleus with intense staining. In retinae of E9 to E17 embryos, the number of pyknotic neurones was also counted and their retinal distribution recorded at X630, the size of the sample areas being mm 2 and 1 mm apart. Isodensity points in the retina were joined to obtain isodensity maps representative of the distribution of viable and pyknotic neurones across the retina. Retinal wholemounts were traced through a drawing tube at known magnification and their areas measured using a North Star Z80 image analyser. In some cases, retinal area measurements were carried out in wholemounts before and after fixation and processing, which revealed a shrinkage of about 5-10% in area. Total neurone number and pyknotic neurone number in the GCL were estimated on the basis of the average numbers of the sample areas and correlated with the whole retinal area. No correction for shrinkage was made. Results This report is based on the morphometric analysis of 46 retinal wholemounts obtained from 24 embryos and 7 posthatched chicks. Dorsal, ventral, temporal and nasal poles of the retinal wholemounts were identified by the presence of the pecten and the optic nerve head in the temporoinferior quadrant. The constituent cells of the retinal GCL, ganglion cells,

3 Formation of area centralis in chick retinal ganglion displaced amacrine cells and glia were identified on morphological criteria. In sample area, counting approximately 250 cells, the glial cells amounted to about 1-3% of the total population. These figures correspond well with published data on the proportion of glial cells in the retinae of several species: 1 % in frogs (Dunlop & Beazley, 1984), about 5 % in the chick (Ehrlich, 1981) and about 1 % in the cat (Stone, 1978). The majority of neurones (both displaced amacrine cells and small ganglion cells) at any stage of development was of similar appearance: ovoid nuclei, surrounded by a narrow, faintly stained cytoplasmic collar (Fig. 1A-C). Some of the larger neurones (mostly ganglion cells), predominantly in the peripheral retina, had abundant cytoplasm with 413 Nissl substance and an occasionally visible axon hillock (Fig. ID). Area measurements on the retinal wholemounts showed an increase of approximately threefold from an average of 75 mm"2 on E8 to 220 mm"2 on 4WAH. The number and distribution of viable neurones in the ganglion cell layer The number of neurones in the retinal GCL was counted in sample areas and the total neurone number estimated. Table 1 shows that maximal neurone number of 4-25 million was found on E10; this decreased to 2-9 million by E17 and thereafter it remained stable. Neurone densities in the retinal GCL were established at selected stages. On E9, Fig. 1. Histophotographs of neurones in the retinal GCL at various stages of development. (A) Undifferentiated neurones in the central and (B) in the peripheral retina of an E10 embryo. (C) Small neurones in the central and (D) large neurones in the peripheral retina of a posthatched (4DAH) animal. (E) Pyknotic neurones with vacuolated nuclei (arrowheads) in the GCL of the peripheral retina of an E13 embryo. Scale bars, 20^m. Bar in D applies also to A-C.

4 414 C. Straznicky and M. Chehade highest densities of cells mm were found in the central retina decreasing to cells mm~ 2 in the periphery (Fig. 2A). Neurones in the GCL at this stage revealed similar size and ovoid shape both in the central retina and periphery (Fig. 1A,B). Neurone densities decreased considerably across the retina by E17 where values were cell mm~ 2 in the centre and cells mm" 2 in the periphery (Fig. 2B). At this stage, cells tended to be larger in the peripheral retina than those in the central retina and they contained abundant Nissl substance. At the time of hatching, there were cells mm" 2 centrally and cells mm" 2 peripherally. By 4 weeks after hatching, the density of neurones in the GCL decreased further, approaching the adult levels (Ehrlich, 1981), of cells mm" 2 centrally and 6000 cells mm" 2 peripherally (Fig. 2C). D D The number and distribution of pyknotic neurones in the ganglion cell layer Dying neurones could be distinguished clearly from normal viable neurones at any stage during the time of naturally occurring cell loss in the retinal GCL. Degenerating neurones appeared with fragmented, shrunken and vacuolated nuclei with intense basophilic spheres (Fig. IF). No degenerating neurones in the GCL were found on or before E8. A few pyknotic neurones could be seen for the first time at the Table 1. Summary of retinal area measurements (in mm 2 ) and counts of viable neurones (in millions) in the retinal ganglion cell layer during development Age of embryos Neurone number Retinal area B T E8R*1 E8Lfl E10R1 E10L1 E13R1 E13L1 E17R1 E17R2 E18R1 E18L2 E20R1 E20R2 HJ Rl 4DAH Rl 2WAHH Rl 2WAH LI 4WAH Rl 4WAH LI * Right eye. t Left eye. JE21 hatchling. 4 days after hatching. \2 weeks after hatching Fig. 2. Isodensity maps of neurones in the retinal GCL of E9 (A) and E17 (B) embryonic and 4WAH posthatched (C) retinae. Numbers on the contour lines refer to thousand neurones per 1 mm 2 area estimated from counts of mm 2 sample areas. In these and in each of following Figs the symbols are as follows D, dorsal; V, ventral; T, temporal; N, nasal. Shaded areas in the retinal maps denote the position of the pecten. Scale bars, 10 mm.

5 Formation of area centralis in chick retinal ganglion 415 Table 2. The total number of pyknotic neurones in the ganglion cell layer during development Age of embryos Total number Average number Average density per mm 2 E9R'l E9L1 E9iRl E9JR2 ElORl E10 Lfl E10L2 E10L3 E11R1 E11L2 E11R3 Ell L E12L1 E12R2 E12L2 E12L3 E13R1 E13L1 E13R2 E14R1 E14L1 E14L2 E14R E15 Rl E15 LI E15L2 E15R E16R1 E16R2 E16L3 E16iRl E16JL1 E17R1 E17 LI Fig. 3. Isodensity maps of pyknotic neurones in the retinal GCL of E% (A) and E10 (B) embryos. Numbers on the contour lines refer to pyknotic neurones per 1 mm 2 area. Total between E9 and E17 * R = right retinji. tl = left retina beginning of E9 in the central retinal area, followed by a rapid increase by the end of E9. The last dying neurones in the retinal periphery disappeared by E17. Neurone loss across the retina peaked on E12. Over the 8-day period, the aggregate average daily loss was cells in the retinal GCL (Table 2). The distribution of pyknotic neurones revealed a clear regional pattern (Table 3). The retinal area was subdivided into central, intermediate and peripheral thirds by two concentric circles. The density of pyknotic neurones was highest in the central and intermediate zones in E9i and E10 embryos (Fig. 3). From Ell onwards, consistently higher pyknotic density was observed in the intermediate and peripheral retinal zones. The detailed results of the sampling of pyknotic neurones across the retina are given in an E12 embyro. Fig. 4 shows higher pyknotic counts in the peripheral dorsotemporal and dorsonasal retina than in the central retina. Despite the abrupt decrease in the number of pyknotic neurones in the central retina, high pyknotic counts were still present in the peripheral retina in E13 (Fig. 5) and in E14 (Fig. 6) embryos. Naturally occurring neurone loss virtually ceased in the central retina by E15 (Fig. 6); however, it continued at a lower rate in the peripheral retina up to E16i (Fig. 7). By the end of E17, no dying neurones were observed in the GCL. A comparison of the isodensity maps also shows that the period of nuclear pyknosis in the retinal GCL was maintained longest in the periphery of the dorsonasal retina. Total neurone loss between E9 and E17 was calculated for each of the central, intermediate and peripheral retinal zones. Fig. 8 and Table 3 show that both the average density of pyknotic neurones per 1 mm 2 unit area, as well as the total neurone loss, are

6 416 C. Straznicky and M. Chehade about twice as high in the peripheral as in the central retinal zones. It should be noted, however, that the aggregate neurone loss in the peripheral retina might have been slightly underestimated, because the peripheral retinal sector in E9 and ElO embryos does not quite correspond to the same sector in E13 or older embryos. The reason for this is the later generation of neurones in the retinal periphery on ElO to E12 by continuous cell addition at the ciliary margin (Kahn, 1974). E12 Fig. 4. Isodensity map of pyknotic neurones in the retinal GCL of an E12 embryo. Small numbers correspond to readings in mm 2 sample areas. Contour lines with larger numbers refer to pyknotic neurones per 1 mm 2 area. Estimation of differential retinal expansion The central-to-peripheral gradient in neurone density in the GCL is changing until the cessation of retinal growth by expansion in young adult birds. According to our analysis, the neurone density ratio between the central and peripheral retina was 1-5:1 in E9, 2:1 in E17, 2-3:1 in E21, 4:1 in 4VVAH, in contrast to the 6:1 ratio in young adult (Ehrlich, 1981). Retinal area increases, for example, from 174 mm 2 on E17 to mm 2 by E21 (Table 1), with the accompanying increase of the retinal radius from 7-1 to 8-2 mm. Taking into account that neither further neurone loss nor additional cell generation took place in the retinal GCL during the last 4 days of incubation, the increase of retinal area was exclusively due to expansion through interstitial growth. Since, according to our estimation, the neurone density decreased less in the central (from cells mm" 2 to cells mm" 2 ; a 6-7 % decrease) than in the peripheral retina over this period (from cells mm" 2 to cells mm" 2 ; a 20 % decrease), the retinal expansion had to be uneven, with greater area increase in the peripheral than in the central retina. A simple model indicating the assumed pattern of differential retinal growth is given in Fig. 9. Indeed, such differential retinal expansion, predominantly in the peripheral retina, continues beyond hatching (E21) which further enhances the central-to-peripheral gradient in neurone density. Table 3. The distribution of pyknotic neurones in the central, intermediate and peripheral third of the retinal ganglion cell layer, mean values and ±s.d. for 1 nun 2 unit areas Age Central Intermediate Peripheral Fig. 5. Isodensity map of pyknotic neurones in the retinal GCL in an E13 embryo. Note the increased pyknotic neurone density towards the retinal periphery. E9&9} ElO Ell E12 E13 E14 E15 E16 E16i & 17 Total between E9 and E ± ± ± ±8 136 ±31 65 ±33 24 ±11 12 ± ±4 187 ±8 300 ± ±7 205 ± ±16 54±30 43 ± ±7 153 ± ± ± ± ± ±23 45 ±

7 Formation of area centralis in chick retinal ganglion 417 D Fig. 6. Isodensity maps of pyknotic neurones in the retinal GCL in E14 (A) and E15 (B) embryos. Nuclear pyknosis virtually ceased in the retinal centre (hatched area) in the latter case. D E16V5 Fig. 7. Isodensity map of pyknotic neurones in the retinal GCL of an E16i embryo. A few pyknotic neurones are still present in the dorsonasal margin. The retinal centre is devoid of pyknotic neurones (hatched area). Discussion We have studied the development of area centralis of the chick retinal GCL from the 8th day of incubation to 4 weeks after hatching. Our quantitative analysis yielded the following main observations. (1) Neurone density difference of the retinal GCL between central and peripheral retina increased from 1-5:1 to 4:1. (2) Approximately twice as many neurones per unit area were lost in the peripheral retina than in the central Fig. 8. Sketch of an idealized retinal wholemount. The retina is divided into thirds by two concentric circles, showing mean pyknotic densities in the central, intermediate and peripheral sectors. AC, area centralis; ONH, optic nerve head with the pecten. retina. (3) Area expansion of the peripheral retina was more than of the central retina. The results of our study confirm previous observations that a significant proportion of initially generated neurones in the retinal GCL die and that the time course of neurone death is between E9 and E17 (Rager, 1976, 1980; Hughes & McLoon,.1979). The number of fibres in the optic nerve has been shown to reach a peak of about 4 million on E10-E11 followed by an abrupt decrease to about 2-5 million

8 418 C. Straznicky and M. Chehade D T Fig. 9. Model of differential retinal expansion, drawn from retinal wholemounts of an E17 embryo (broken line) and an E21 hatchling (solid line), shown in the right part of the panel. Pie slices of the two retinae, containing the same number of neurones are given in the left part of the panel. Note that differential expansion of the eye enhances the neurone density gradient from the retinal centre to the retinal periphery. by E16-E17 (Rager & Rager, 1978). Axon loss in the optic nerve is a reliable reflection of ganglion cell death and this correlation suggests that the observed neurone death in the GCL occurred mostly among ganglion cells and not among amacrine or nonneuronal cells. Displaced amacrine cells in the GCL have been estimated to be about % of the total population (Ehrlich, 1981; Layer & Vollmer, 1982). Our peak neurone number of 4-25 million, which includes both ganglion cells and displaced amacrine cells, appears to represent a significantly lower estimate. However, it should be noted that previous estimates of neurone number in the GCL during development were based on cell counts on retinal serial sections which may not give figures as accurate as can be obtained.on retinal wholemounts. Our preliminary observations on E11-E15 chick retinal wholemounts, incubated with Lucifer Yellow which is selectively taken up by displaced amacrine cells, indicate that the amacrine cell population may be as low as 5-10 % of neurones in the GCL (Hiscock & Straznicky, 1987). It is thus likely, on the basis of the present observations and preliminary studies, that the number of displaced amacrine cells has to be revised downwards from previous estimates. The main findings of the present study relate to the estimation of the magnitude of naturally occurring cell loss and to the regional distribution of pyknotic neurones in the retinal GCL. Total neurone loss between E9 and E17 was found to be about when estimated by counting pyknotic neurones. Since displaced amacrine cells appear to represent less than 10 % of the neurone population, the bulk of pyknotic neurones correspond to ganglion cells. In contrast, neurone number in the GCL decreases by about 1-6 million over the same period when estimated by counting fibres in the optic nerve (Rager & Rager, 1978). Since degenerating neurones were easily identifiable on the basis of Pannese's (1976) criteria, it is unlikely that some pyknotic neurones escaped detection. In accord with other neurone pools (Arens & Straznicky, 1986), we propose that the time course of nuclear pyknosis is much shorter than the 24 h interval of sampling. At present, no independent measure is available to establish accurately the time it takes from the beginning of pyknosis to the disappearance of the degenerating neurone. Considering that about pyknotic neurones were accounted for between E9 and E17 and the number of neurones decreased by 1-6 million, we assume that the pyknotic cycle among neurones of the retinal GCL may have been as short as 3-4 h. Previous observations have revealed that retinal neurone generation has a clear-cut regional pattern. Ganglion cells are first generated in the central retina, followed by more peripherally located neurones (Kahn, 1974). The same generation pattern appears to hold true for displaced amacrine cells (Vollmer & Layer, 1986) and for the differentiation of cells of other retinal layers (Liu, Layer & Gierer, 1983). The present study shows a similar regional pattern of neurone death in the retinal GCL. First, most.of the pyknotic neurones are located in the central retina, the ontogenetically oldest part of the retina. Later, the wave of nuclear pyknosis moves outwards such that more neurones per unit area are lost in the

9 Formation of area centralis in chick retinal ganglion 419 peripheral than in the central retina. These observations confirm and extend Rager & Rager's (1978) previous report on the regional character of neurone death in the GCL. It is also worth noting that as the wave of nuclear pyknosis progresses to the peripheral retina, the 'hot spot' or highest level of neurone death appears first in the dorsotemporal retina (E12) and then it moves to the dorsonasal retina (E14 onwards) which is the ontogenetically least differentiated part of the retina. Therefore it is quite clear from our results that nuclear pyknosis and its regional distribution in the GCL follows the pattern of general retinal differentiation (Layer & Vollmer, 1982; Liu et al. 1983). Retinal growth has been shown to have two distinct components: first, a continuous addition of neurones at the ciliary margin throughout the whole life of fish (Johns, 1977; Johns & Easter, 1977) and amphibia (Straznicky & Gaze, 1971; Dunlop & Beazley, 1984); and second, a continuous retinal expansion, in the form of interstitial growth (Easter, Johns & Baumann, 1977; Koch, 1982). It is well known that during development the retina undergoes a large increase in area that is not accompanied by the addition of neurones to the GCL. Mammalian studies have suggested that retinal growth is not uniform, with peripheral retinal area growing more than central retinal area (Rapaport & Stone, 1983; Mastronarde, Thibeault & Dubin, 1984; Stone, Maslim & Rapaport, 1984). Differential retinal growth, after the cessation of retinal neurogenesis certainly contributes to the production of the central-to-peripheral gradient in neurone density. Our present observations on the chick retinal development confirm this notion for this species. Peripherally located ganglion cells become more widely spread than those centrally located as development progresses'; and this is due, in part, to differential peripheral retinal expansion and, in part, to the appearance of very large ganglion cells in the peripheral part of the retina (Lunam, Hiscock & Straznicky, 1987). Our present results do not allow us to calculate the extent to which differential retinal expansion contributes to the formation of the adulttype neurone-density gradient in the GCL. It is likely that a higher rate of neurone production in the central retina and a higher rate of neurone death in the peripheral retina are largely responsible for the establishment of neurone-density gradient, which is further increased by differential retinal growth. The mechanisms for the formation of the high ganglion-cell-density region of the retina in various vertebrate classes are strikingly different. It has been shown in frogs that the postmetamorphic formation of the visual streak is based mainly on differential neurone generation at the nasal and temporal ciliary margins, maintaining high ganglion cell densities across the nasotemporal retinal axis (Coleman, Dunlop & Beazley, 1984). Naturally occurring neurone death is diffuse and insufficient in magnitude to create the ganglion cell density gradient characteristic of the adult animal (Jenkins & Straznicky, 1986). In the chick, although there is a slightly higher neurone production in the central retina, this in itself, is not enough to form a neurone density gradient comparable to the adult. In contrast to the frog, the chick area centralis develops, after the period of neurone generation, as a result of differential neurone death: more neurones die in the peripheral than in the central part of the retina. Retinal neurone-density gradient is further enhanced in the chick by differential retinal expansion. In mammals, neurone generation and subsequent neurone loss in the GCL have been shown to be more or less even across the retina (Rapaport & Stone, 1983; Young, 1985). However, differential maturation and migration sculpture the area centralis as more progenitor neurones in the peripheral part of the GCL differentiate into amacrine cells and migrate out of the GCL than in the central part (Stone etal. 1984). During postnatal life, the ganglion cell density gradient becomes steeper due to differential retinal expansion. Thus the formation of the high ganglion cell density region of the retina in frogs, birds and in mammals is brought about by different mechanisms except for the common feature of retinal expansion. It is indeed tempting to speculate about the biological meaning of such diverse mechanisms. One reason might be that, in fish and amphibia, the eye and the retina grow continuously throughout the whole lifespan of the animal, in part by the addition of newly generated neurones at the ciliary margin. The slowing down of neurone generation In parts of the ciliary margin, whilst maintaining higher mitotic activity in other parts is sufficient to assure the formation of a ganglion cell density gradient in the form of the visual streak. In contrast, in the chick and in mammals, the period of active neurone generation is very sh6rt, requiring other mechanisms to form high ganglion cell density at the site of acute vision. Differential neurone death in the chick and differential neurone migration in mammals in the retinal periphery bring about a ganglion-cell-density gradient and, consequently, the formation of the area centralis from a retina which is initially isodensitic. Miss Shefia Nemer's help with the preparation of retinal wholemounts and Mr Dennis Jones's assistance with the illustrations are gratefully acknowledged. This work was supported by a grant from the Flinders University Research Budget to C.S.

10 420 C. Straznicky and M. Chehade References ARENS, M. & STRAZNICKY, C. (1986). The development of the trigeminal (V) motor nucleus in normal and tubocurare treated chick embryos. Anat. Embryol. 174, COLEMAN, L. A., DUNLOP, S. A. & BEAZLEY, L. D. (1984). Patterns of cell division during visual streak formation in the frog Limnodynastes dorsalis. J. Embryol. exp. Morph. 83, DUNLOP, S. A. & BEAZLEY, L. D. (1984). A morphometric study of the retinal ganglion cell layer and optic nerve from metamorphosis in Xenopus laevis. Vision Res. 24, EASTER, S. S., JOHNS, P. R. & BAUMANN, L. R. (1977). Growth of adult goldfish eye. I: Optics. Vision Res. 17, EHRLICH, D. (1981). Regional specialization of the chick retina as revealed by the size and density of neurons in the ganglion cell layer. /. comp. Neurol. 195, HAMBURGER, V. & HAMILTON, H. L. (1951). A series of normal stages in the development of the chick embryo. J. Morph. 88, HISCOCK, J. & STRAZNICKY, C. (1987). Amacrine cells in the developing chick ganglion cell layer. Neurosci. Letts Suppl. 37, p85. HUGHES, W. F. & MCLOON, S. C. (1979). Ganglion cell death during normal retinal development in the chick: Comparison with cell death induced by early target field destruction. Expl Neurol. 66, JENKINS, S. & STRAZNICKY, C. (1986). Naturally occurring and induced ganglion cell death, a retinal whole-mount autoradiographic study in Xenopus. Anat. Embryol. 174, JOHNS, P. R. (1977). Growth of the adult goldfish eye. LTI. Source of new retinal cells. J. comp. Neurol. 176, JOHNS, P. R. & EASTER, S. S. (1977). Growth of the adult goldfish eye. II. Increase in retinal cell number. /. comp. Neurol. 176, KAHN, A. J. (1974). An autoradiographic analysis of the time of appearance of neurons in the developing chick neural retina. Devi Biol. 38, KOCH, J.-H. (1982). Neuronal addition and retinal expansion during growth of the crucian carp eye. /. comp. Neurol. 209, LAYER, P. G. & VOLLMER, G. (1982). Lucifer Yellow stains displaced amacrine cells of the chicken retina during embryonic development. Neurosci. Letts. 21, Liu, L., LAYER, P. G. & GIERER, A. (1983). Binding of FITC-coupled peanut-agglutinin (TTTC-PNA) to embryonic chicken retinas reveals developmental spatio-temporal patterns. Devi Brain Res. 8, LUNAM, C. A., HISCOCK, J. & STRAZNICKY, C. (1987). Classification of ganglion cells in the chick retina based on HRP : labelling. Neurosci. Letts. Suppl. 27, p97. MASTRONARDE, D. N., THTBEAULT, M. A. & DUBIN, M. W. (1984). Non-uniform postnatal growth of the cat retina. J. comp. Neurol. 228, PANNESE, E. (1976). An electron microscopic study of cell degeneration in chick embryo spinal ganglia. Neuropathol. Appl. Neurobiol. 2, RAGER, G. H. (1976). Morphogenesis and physiogenesis of the retino-tectal connection in the chick. I. The retinal ganglion cells and their axons. Proc. R. Soc. B 192, RAGER, G. H. (1980). Development of the retinotectal projection in the chicken. Adv. Anat. Embryol. Cell Biol. 63, RAGER, G. H. & RAGER, U. (1978). Systems-matching by degeneration. I. A quantitative electronmicroscopic study of the generation and degeneration of retinal ganglion cells in the chicken. Expl Brain Res. 33, RAPAPORT, D. H. & STONE, J. (1983). The topography of cytogenesis in the developing retina of the cat. J. Neurosci. 3, STONE, J. (1978). The number and distribution of ganglion cells in the cat's retina. J. comp. Neurol. 180, STONE, J. (1981). The Wholemount Handbook. A Guide to the Preparation and Analysis of Retinal Wholemounts. Maitland. Sydney, Australia. STONE, J., MASLIM, J. & RAPAPORT, D. H. (1984). The development of the topographic organisation of the cat's retina. In Development of Visual Pathways in Mammals (ed. J. Stone, B. Dreher & D. H. Rapaport), pp New York: Alan R. Liss, Inc. STRAZNICKY, C. & CHEHADE, M. (1986). The development of the high ganglion cell density area centralis in the chick. Neurosci. Letts. Suppl. 26, p262. STRAZNICKY, K. & GAZE, R. M. (1971). The growth of the retina in Xenopus laevis: an autoradiographic study. J. Embryol. exp. Morph. 26, VOLLMER, G. & LAYER, P. G. (1986). Lucifer Yellow stains a subpopulation of chicken retinoblasts preferentially 3 h after their last round of DNA synthesis. Devi Brain Res. 27, YOUNG, R. W. (1985). Cell proliferation during postnatal development of the retina in the mouse. Devi Brain Res. 21, {Accepted 24 March 1987)