Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
AMER. ZOOL., 21:447-458 (1981) Growth of Fish Retinas 1 PAMELA RAYMOND JOHNS Department of Neurobiology, Harvard Medical School, Boston, Massachusetts 02115 SYNOPSIS. This review discusses development and growth of the retina. A geometric model of retinal differentiation is proposed in which four phases are recognized; the first three are common to all vertebrate embryos. The last, post-embryonic growth phase has two alternate routes, one followed by birds and mammals and the other by fish and amphibians. All retinas grow by expansion, and retinal cells are spread apart as the retina enlarges. In fish (and larval amphibians), the retina not only expands but also adds cells. In these retinas a marginal germinal zone persists and continues to produce neurons, which are added appositionally in concentric annuh around the perimeter. The genesis of one class of photoreceptor cell, the retinal rods, is different from all other retinal neurons in that the proportion of rods increases with growth in central retinal regions far from the germinal zone. The source of these centrally-added rods is not yet established; several hypotheses are discussed. Alterations of synaptic connectivity within the retina and between retina and brain are suggested by the pattern of growth and cell addition. The capacity of adult fishes to generate new neurons and to form new synapses is a remarkable property, one which most animals abandon much earlier in life. INTRODUCTION A beginning student of embryology when watching a developing embryo for the first time is immediately impressed by the eyes—they appear so early and quickly grow so large. Perhaps because they are such a prominent feature of the developing embryo, but certainly also because the eyes are such complex and exquisitely specialized organs, the study of their development has always fascinated and challenged embryologists. Surprisingly little attention has been paid to the growth and development of eyes beyond embryonic stages. Eyes do grow post-embryonically, and in some animals, such as fish, the eyes increase enormously in size during juvenile and adult life. My work and that of others has shown that in the retinas of fish, growth results in part from the addition of new neurons. A mature, functioning piece of nervous tissue that adds neurons as it grows is remarkable because neural function is to a large extent determined by the precision of connections between neurons. How then are new neuronal elements inserted into the circuit without interfering with ongoing activity? I cannot yet answer that question, but I do have some clues as to how and where new cells are added. The work reviewed here mainly concerns the morphology of growth, but the underlying motive for looking at structure is to understand function. At the end, I will mention what little we do know about how vision is affected by retinal growth. Before considering growth I must first briefly describe the structure of the adult retina. More details specific to fish retinas and the problems of growth will be added in subsequent sections. STRUCTURE OF THE RETINA The vertebrate retina is a thin sheet of neural tissue, approximately hemispherical in shape. Figure 1 is a photomicrograph of a fish retina; all retinas have a similar laminar organization (Polyak, 1957; Walls, 1967; Ramon y Cajal, 1973). At the outer (convex) surface are the photoreceptor cells, specialized sensory neurons that respond to light. Most retinas have two kinds: rods, which function at low levels of illumination, and cones, which are active in brighter light (Rodieck, 1973). The nuclei of photoreceptor cells are in the outer nuclear layer (ONL); those of the cones 1 From the Symposium on Developmental Biology of are at the level of the outer limiting memFishes presented at the Annual Meeting of the American Society of Zoologists, 27—30 December 1979, at brane and those of the rods are closer to the internal retinal layers. Tampa, Florida. 447 448 PAMELA RAYMOND JOHNS OFL FIG. 1. This photomicrograph is of the retina of a goldfish. The inside of the eyeball is downwards. Photoreceptor cells (PC), surrounded by finger-like cytoplasmic processes of pigmented epithelial cells, are at the top. The cones are large, pale and ovoid; the rods are long and slender and closer to the pigmented epithelium. Their nuclei in the outer nuclear layer (OXL) are at different levels; the cones are indicated by the small open arrow, and the rods by the filled arrow. The inner nuclear layer (INL) contains several types of cells. The ganglion cells (GC) and their axons in the optic fiber layer (OFL) are at the inner surface. The internal and external limiting membranes are indicated by the arrows on the right. The calibration bar is 50 pm. The next layer of cells is the inner nuclear layer (INL), which contains the nuclei of several types of neurons (bipolar, araacrine and horizontal cells) as well as those of the primary retinal glia (Miiller cells). In the cellular layer nearest the inner surface of the retina are the nuclei of ganglion cells (GC). Separating these three nuclear layers are two plexiform layers in which cytoplasmic processes of retinal neurons GROWTH OF FISH RETINAS make synaptic connections with one another. The axons of ganglion cells, the optic fibers, form a discrete layer at the inner retinal surface (OFL). A vascular layer is interposed between them and the vitr e o u s humor, the gel-like substance that Tills the eyeball. Next I will describe the ways in which retinas grow. It is instructive to look at earlier, embryonic stages, before considering post-embryonic growth, because later growth of fish retinas is in many ways merely an extension of patterns and activities begun in the embryo. EMBRYONIC DEVELOPMENT OF RETINAS Steps in the formation of eye and retina are similar in all vertebrates (O'Rahilly and Meyer, 1959; Mann, 1969; O'Rahilly, 1975). The basic pattern of retinal histogenesis was deduced from early studies, particularly those of Ramon y Cajal (1959) who used classical histological methods. Recent electron microscopic studies (Hattori and Fujita, 1974; Hinds and Hinds, 1974; Smelser et al., 1974, and many others) have provided additional information, as have autoradiographical studies that use [3H]thymidine, a specific precursor of DNA, as a marker for dividing cells (Sidman, 1960, 1970; Fujita and Horii, 1963; Hollyfield, 1972; Kahn, 1974; and again many others). At the earliest stages the presumptive neural retina is composed of a hemispherical sheet of undifferentiated neuroepithelial germinal cells. These are elongated, spindle-shaped cells, which span the width of the neuroepithelium. The germinal cells have two critical roles: i) they generate the retinal cells and ii) they provide a vertical scaffolding around which neuronal connections are organized. Differentiation of retinal neurons does not occur simultaneously throughout the embryonic retina, but begins in the center. Figure 2 is a general model of the spatiotemporal pattern of retinal differentiation. Phase I represents the undifferentiated neuroepithelium; dividing germinal cells are indicated by "x's." Mitotic activity ceases and differentiation begins in the center (the "o's" in Fig. 2, phase II). Retina is con- 449 structed concentrically as more cells differentiate within an ever-widening circle (phase III) until finally the perimeter of the neuroepithelium is reached. Up to this point, which occurs before or shortly after the end of the embryonic period, all vertebrate retinas develop similarly. In the last phase (IV) there are two paths, and a choice must be made. The choice is to stop or to continue producing retinal cells. Birds and mammals choose the former, and all of the neuroepithelial germinal cells are consumed when the circular wave of differentiation reaches the retinal margin (the upper diagram in Fig. 2, phase IV). Cell production ceases and the retina then contains its full complement of neurons (Johns et al., 1979). In fish and amphibians, by contrast, the germinal cells are not exhausted; a few remain to constitute a circumferential germinal zone whose cells continue to proliferate throughout larval stages into adult life (the lower diagram in Fig. 2, phase IV). In these retinas, new neurons are added and new retina is formed for a prolonged period as the eye continues to grow (Straznicky and Gaze, 1971; Jacobson, 1976; Johns and Easter, 1977; Johns, 1977; Beach, 1979). POSTEMBRYONIC GROWTH OF FlSH RETINAS Unlike mammals, fish continue to grow throughout the life of the animal (Brown, 1957). The rate of growth is extremely variable and difficult to predict, however, as it depends upon a complex interplay of many factors, including the availablity and quality of food, the temperature of the water, and the density of fish in the population. Growth rate also changes with age; older fish tend to grow more slowly. The eyes grow as the body does, but not quite as rapidly, so that larger fish have relatively smaller eyes (Miiller, 1952; Lyall, 1957a; Ali, 1964; Johns and Easter, 1977). Inside the eye the retina also enlarges. In goldfish (Carassius auratus) between 1 and 4 yr of age, for example, the retinal surface area increases from 20 to 120 mm2 (Johns and Easter, 1977). In young animals this dramatic increase in retinal size is due in part to cell addition. 450 PAMELA RAYMOND J O H N S EMBRYONIC POST-EMBRYONIC FIG. 2. Four phases of vertebrate retinal development are shown. The retina is represented as a circle, which in phase I contains only undifferentiated neuroepithelial germinal cells (x). Cells that have ceased dividing (o) appear in phase II and increase in number in phase III. The vertical dashed line approximately divides embryonic from post-embryonic developmental periods, though the time of birth or hatch relative to the state of retinal development varies in different species. In phase IV the upper diagram represents retinas in which cell proliferation has ceased; further enlargement of the retina results from growth of individual cells (the "o's" are larger). The lower diagram shows how other retinas grow by adding new cells ("o" with a slash) from a circumferential germinal zone of dividing cells (x) In these retinas, the cells already present enlarge as well, and contribute to the overall growth of the retina. The germinal zone Cells are added to growing fish retinas at the margin. Here mitotic figures are seen (Muller, 1952; Lyall, 1957a; Blaxter and Jones, 1967). The dividing cells are also selectively labeled with [3H]thymidine, and examples are shown in Figure 3A, B. The labeled cells are distinctive—notice in particular their spindle-shaped nuclei. In this and in other cytological features they are identical to the neuroepithelial germinal cells of the embryonic retina. From their appearance and position at the retinal perimeter I conclude that these cells are the descendents of embryonic neuroepithelial germinal cells that never relinquished their capacity to proliferate. The germinal zone in larval and adult fish surrounds the entire retina and adds new cells at the margin (Johns, 1977; Meyer, 1978). Figure 2, phase IV illustrates how concentric annuli of new retina are produced by the germinal zone. Some of the autoradiographical results from which this scheme was developed are illustrated in Figure 3C. This is a histological section from the retina of a goldfish killed months after injection of [3H]thymidine Labeled cells in this section are distributed in a vertical band, which includes all three nuclear layers. These labeled neurons are the progeny of germinal cells that incorporated the label and differentiated soon after. The unlabeled segment of retina now interposed between the labeled cells and the margin was produced in the subsequent 6 month interval by germinal cells whose label was diluted by multiple divisions in the absence of radioactive precursor. Further evidence that the germinal zone produces neurons is the immature appearance of the retina adjacent to it (Muller, 1952; Johns, 1977). The most obvious signs of immaturity are the stubby, undeveloped cones in the photoreceptor layer near the margin; toward the center they grow longer and larger (Fig. 4). Addition of cells How many cells is the germinal zone producing? To answer that question we counted retinal cells in goldfish from 5 to 20 cm in length. In the goldfish retina, regional differences in cell density are minor and inconsistent (Stell and Harosi, 1976; Johns and Easter, 1977), so total number of cells was calculated from average cell density. Cell number, excluding vascular cells and glia in the optic fiber layer, is plotted in Figure 5; it increases from 3,000,000 to 8,000,000 cells as the retina grows from 4 to 8 mm in length. Muller (1952) counted cells in guppy retinas and found an analogous increase in total cell number from 525,000 to 1,900,000 as the fish grew during the first year. In both guppies and goldfish, this increase in total cell number results from the addition of all types of retinal cells. Others have counted neurons in growing fish retinas but have confined their counts to either cones or ganglion cells. The cones in most teleost fish are ar- GROWTH OF FISH RETINAS 431 A Fie. 3. A and B. This autoradiograph shown in both darkfield (A) and brightfield (B) illumination was prepared from the retina of a cichlid fish (Haplochromis burtoni) injected with [3H]thymidine. Dividing cells at the margin of the retina are labeled with silver grains. The retina extends toward the lower left corner; pigmented epithelium is across the top. The calibration bar is 20 pirn. C. This darkfield autoradiograph is the retina from a goldfish injected with [3H]thymidine six months previously. Labeled retinal cells are in a vertical band (larger arrow), now displaced about 150 /urn from the retinal margin, which is at the right. The smaller arrow indicates a labeled cell in the outer nuclear layer displaced centrally from the band of other labeled cells (see text for further discussion). The calibration bar is 100 /xm. D. This is the retinal margin of a large goldfish, 23 cm in length. No germinal zone is present. The calibration bar is 100 //.m. ranged in a regular mosaic pattern (Engstrom, 1960, 1963), which makes them easy to count in tangential sections. The size of the mosaic unit and the size of individual cones increases as the retina grows, but not enough to account for the entire increase in retinal area. The conclusion is that mosaic units must be added with growth (Lyall, 1957a; O'Connell, 1963; Ahlbert, 1968). Ganglion cells can be conveniently counted in intact flat mounts of the whole retina, though they must be distinguished from surrounding glia and vascular cells. In crucian carp (Carassius carassius), a species closely related to goldfish, Kock and Reuter (1978) found that ganglion cells increased steadily in number as eyes grew from 4 to 7 mm' in diameter, but with further increase in eye diameter up to 10 mm the number of ganglion cells remained constant. Their results imply that the germinal zone finally ceased producing ganglion cells in these large fish, which were up to 7 yr old. Recently we have examined retinas from goldfish larger than those used in our previous study and found that in these older fish retinal cells are no longer being added (Johns and Easter, unpublished observations). In sup- 452 PAMELA RAYMOND J O H N S Germinal Zone Zone of Differentiation Differentiated Retina FIG. 4. This is the peripheral retina of a medium-sized goldfish (12 cm in length). The large short arrow points to germinal cells, one of which is in mitosis (M). The most peripheral retina (in the zone of differentiation) contains undifferentiated cells; immature cones (C) are especially distinctive. In more central, differentiated retina, the cones are larger and more mature. The calibration bar is 50 jim. port of this apparent cessation of ganglion cell addition, the retinal margin from a very large goldfish in Figure 3D shows no evidence of a germinal zone. (We have not yet injected such large fish with [3H]thymidine.) Counts of optic nerve fibers in electron micrographic mosaics of cross sections through whole optic nerves of these large goldfish also suggest that the number of ganglion cell axons reaches a plateau (Easter et ai, 1979). Despite the waning of cell production, the eye and retina in old fish continue to enlarge, though very slowly. Without cell addition the only mechanism for further growth is expansion. Expansion of the retina By "expansion" I mean that the retinal surface area enlarges and the cells are spread apart. As an aside, it is worth pointing out that all retinas expand as the eye grows to adult size. Histological studies of postnatal development of the retina in several mammals indicate that the cellular layers become thinner as if the retina were being stretched (Mann, 1969; Vogel, 1969; Braeckevelt and Hollenberg, 1970). The mechanism of expansion is unclear. Cou- lombre and co-workers (Coulombre, 1955; Coulombre et al, 1973) suggest that increases in intraocular pressure serve to inflate the eyeball and so stretch the retina. In 1-yr-old goldfish injected with [3H]thymidine, expansion of the retina was demonstrated directly: The retina present at the time of injection, and enclosed by the labeled annulus, doubled in area after one year (Johns, 1977). This increase in the size of the existing retina accounted for 80% of the total increase in retinal area; new retina added at the margin contributed the remaining 20%. In large goldfish, which are no longer adding retinal cells, the retina becomes thinner as it expands, as illustrated in Figure 6. These are photomicrographs of sections from mid-dorsal retina of three sizes of goldfish, small (A), medium (B) and large (C). In the retina from the mediumsized fish (B), the density of cells is less than in the small fish (A), as it must be since the retinal area is increased. (The rods are an exception and are discussed below.) The overall thickness of the retina is slightly increased in (B), presumably because the cells are bigger and have more processes in the larger retina. In the larg- 453 GROWTH OF FISH RETINAS est fish (C) growth is due entirely to expansion, which now stretches the retina (it is thinner by about 50%) as well as spreading apart the cells. When all three retinas in Figure 6 are compared, the increased •>ize and decreased density of the cones from A to B to C are obvious, as is the depletion of cells in the inner nuclear and ganglion cell layers. It is equally obvious, however, that the density of rod nuclei in the outer nuclear layer does not decrease at all; if anything it is slightly increased in the medium-sized fish. Miiller (1952) saw the same pattern in growing guppy retinas—the density of rods remains constant as the eye enlarges. 10 <o o 8 S 6 . ^ < •o • o 4 The dilemma of the rods The simple story of how these retinas grow, add cells at the edges and spread apart those in the middle, has now become 10 4 6 8 more complicated. The dilemma to be exRETINAL LENGTH (mm) plained is: How do the rods maintain a constant density while the retina expands? Fie. 5. This shows the increase in retinal cell numWe might imagine that new rods are in- ber with growth in goldfish. Each point is one retina. serted into the outer nuclear layer (The different symbols represent different experithroughout the retina, but if so, where do mental groups which are not important for the presdiscussion.) On the abscissa is plotted an index of they come from? The germinal cells at the ent retinal size—the linear distance along the retina from retinal margin were implicated as the one margin to the other in a meridional section source of new retinal cells in the above dis- through the center of the eye. cussion, but I also showed that the cells produced by the germinal zone differentiate where they were born, at the perimeter. Simple apposition of new cells at the already present in the inner nuclear layer, margin, as shown in Figure 2, phase IV, which then migrate when needed into the cannot, however, explain how the ratio of outer nuclear layer and there differenrods increases in central regions far from tiate. This is not a new idea, nor is it the germinal zone. The problem is even unique to fish: Bernard (1900) and Gliicksmore perplexing if we consider the retinas mann (1940) reached similar conclusions of larval fish. Blaxter and co-workers from histological studies of developing (Blaxter and Jones, 1967; Blaxter and amphibian retinas. If movement of cells Staines, 1970; Blaxter, 1975) have shown from inner to outer layers continues in the that in ten different species of teleost fish, growing adult retina, it would account for in which the adults have both rods and the observed increase in the ratio of rods. cones, the larvae have only cones in an oth- A depletion of cells in the inner nuclear erwise differentiated retina. At metamor- layer would be required, and there is some phosis rods suddenly appear interspersed evidence that such a loss occurs with among the cones. Similar observations growth in herring (Blaxter and Jones, have been made by Ali (1959) and Lyall 1967) and in goldfish (Johns and Easter, (1957a) in two other species. Because they 1977). In goldfish during the first four saw no obvious evidence of mitotic activity years of life, however, the number of cells in central retina, Blaxter and Jones (1967) lost from the inner nuclear layer is only suggested that rods may develop from cells about one half the number acquired by the 454 PAMELA RAYMOND J O H N S FIG. 6. These three photomicrographs compare equivalent retinal regions in three goldfish, 6 cm (A), 12 cm (B) and 23 cm (C) in length. The photomicrographs are aligned at the level of the external limiting membrane. Bending of the photoreceptor cell processes is an artifact of the histological preparation. The calibration bar is 100 /xm. outer nuclear layer, so it is unlikely that this is the complete explanation (Johns, 1977). Another potential source of rods is suggested by the results from autoradiography of the growing retinas of juvenile and adult fish. Following injections of [3H]thymidine, labeled nuclei, presumably those of rods, have been observed in the outer nuclear layer displaced from the annular region that contains other labeled nuclei. These cells spread centrally into the outer nuclear layer; an example is shown in Figure 3C. That they are rods is suggested, but not proven, by their location in the outer nuclear layer, and by their persistence for months or years (Johns, 1977; Meyer, 1978). Their origin is unclear. Scholes (1976) suggested that proliferating cells which generate rods in the Black Molly retina are not confined to the germinal zone, but are found in the outer nuclear layer adjacent to the margin. He saw occasional mitotic figures and sometimes labeled cells there. Meyer (1978) also saw labeled cells in the outer nuclear layer displaced 200 /u,m or more from the labeled annulus, but not until eleven or more days following injections in goldfish. I saw a similar distribution of labeled cells, again in goldfish retinas, at 80-336 days (Johns, 1977). The apparent delay in the appearance of these displaced labeled cells suggests the possibility that they may have been generated at the margin and then migrated centrally. It is not easy to distinguish between these two possibilities—a scattered population of proliferating rod precursors or a migration of those progeny of the circumferential germinal cells that are destined to become rods. The interpretation of retinal autoradiographs at GROWTH OF FISH RETINAS 455 short times after injection is complicated In summary, the mystery of where new by the presence of dividing glia and rods come from is still unsolved. Several phagocytes. Both of these cells are labeled suggestions have been made, and evidence with [3H]thymidine, and they are scat- given to support them. It is certainly postered in and among the layers of retinal sible and even likely that more than one is eurons, throughout the retina. They dis- the correct explanation. No matter what appear after longer survival times when the answer is, it is certain that as the retina further cycles of mitotic division have di- grows, the rods account for an increasing luted their label. The histological tech- proportion of the neuronal population. So niques used in the previous studies were even if there is no shear between layers, inadequate to allow undifferentiated rod new synaptic relations must be formed in precursors, if they existed, to be distin- central, differentiated retinal regions. guished from these other kinds of prolif- From counts of synapses in the inner plexierating cells. To do so requires electron form layer in retinas from goldfish of difmicroscopy or specific histochemical mefh- ferent sizes, Fisher and Easter (1979) conods; these studies are in progress. clude that new synapses are formed with There is yet another way to account for growth. Though the synapses they countthe displacement of labeled cells in the out- ed did not include those of the rods, their er nuclear layer and to explain the con- results provide quantitative evidence for stant density of rods. This idea was first synaptogenesis in mature fish retinas. suggested by Miiller (1952) in his compreUp to now I have only asked "how" and hensive study of the guppy retina; the re- not "why" rod density is maintained with sults of my autoradiographical study in growth. All I can offer is the speculation goldfish were consistent with this theory that it may be important for visual sensi(Johns, 1977). Imagine that the rods resist tivity. At low levels of illumination where the forces of expansion, and instead re- the rods function the goal is to catch all main closely-packed while the ganglion available quanta of light, and that may recells, cones and inner nuclear layer cells quire a tightly-packed sheet of photorespread apart. This leads to a progressive ceptors no matter how large the eye. lateral displacement of these cells away Growth of the eye has other predictable from their contemporaries in the lay- effects on visual function, which are diser of rods. The observed displacement of cussed next. labeled cells in the outer nuclear layer in this scheme results not from rods migrat- Effects of retinal growth on vision ing centrally but from other cells moving Does a bigger eye see more of the world? peripherally. If all new cells, including We (Easter et ai, 1977) measured the size rods, are produced at the margin and and shape of the visual field in goldfish and none penetrate into already differentiated found that it is 185° and spherically symretina, then a central circular region metric in eyes of all sizes. So a big eye sees should remain free of labeled cells. This is the same world as does a small eye. what I observed in the goldfish retina Does a bigger eye see better? How well (Johns, 1977). The consequence of shear an animal sees depends on the optical rebetween retinal layers is dramatic—syn- solving power of its eye as well as the efaptic connections must be altered. As the ficiency of its retinal detectors and procesretina expands, the post-synaptic partners sors. Acuity is said to be limited by the size of each rod would move away and other and packing density of cones (Tamura, cells, formerly in more central positions, 1957). The optical "grain" of the retina in would assume their place. This theory pos- a big fish is finer than in a small fish betulates, then, a fundamental instability in cause there are more cones to divide up the retinal circuity, requiring whole-scale the same amount of visual space, so acuity shifts in synaptic connections, as a result of could potentially improve with growth a normal growth process. Such an idea is (Muller, 1952; Tamura, 1957; Hester, bizarre, but not impossible. 1968; Johns and Easter, 1977). A few be- 456 PAMELA RAYMOND JOHNS havioral studies have shown that bigger fish do have better acuity than smaller conspecifics (Baerends et ai, 1960; Hester, 1968; Northmore and Dvorak, 1979). A bigger eye produces a bigger retinal image, and this may affect the activity of retinal neurons. Each neuron has a receptive field, an area on the retina and its corresponding region in visual space over which photic stimuli will evoke a response. The size of the receptive field of an individual cell, measured either in degrees of visual angle or in retinal area, must change as the retina enlarges. Electrophysiological analysis of goldfish retinal ganglion cells suggests that receptive fields change in size by an amount predicted if the overlap between adjacent fields is held constant as the eye grows (Macy and Easter, 1979). Thus the number of retinal cells responsible for a given point in visual space is the same in small and large fish. Finally, the geometry of cell addition has a profound effect on the spatial properties of the retinal neurons. A ganglion cell at the peripheral margin in a small retina is displaced centrally as new cells are added at the perimeter. Likewise, the position of its receptive field shifts centrally, and the spatial information contained in the message it carries to the brain is altered. The central neural link between visual input and motor output must be continually reinterpreted and readjusted. How this is accomplished is unknown. CONCLUSION Structures related to the eye also grow. Bigger eye muscles are required to move a bigger eye, and new muscle fibers are added to the extraocular muscles (Easter, 1979). The brain of the fish also enlarges, and like the retina, it contains germinal zones which persistently generate new neurons (Kirsche, 1967; Segaar, 1965; Richter and Kranz, 1977). In the optic tectum, which receives most of the retinal input, new cells are added at the edges, but not all the way around. The tectal germinal zone is not circular but U-shaped, with a gap at the anterior pole (Kirsche, 1967; Meyer, 1978; Johns and Easter, 1979). Since the optic fibers synapse in topo- graphic order across the tectum, and since new ganglion cells are added at all points along the retinal perimeter, some new optic fibers (those from the temporal margin) innervate the anterior tectal margin where no new cells are being generated. A simil mismatch in the pattern of cell producti in retina and tectum in larval amphibians led Gaze and colleagues to propose that retinotectal connections shift with growth (Straznicky and Gaze, 1972; Gaze et al., 1979). The synaptic remodeling implied here is analogous to that proposed for the retinal rods by the hypothesis of laminar shear. Progressive and systematic movement of synapses in functional neural tissue as a normal consequence of growth is not an established concept in neurobiology. Unfortunately we as yet have no direct techniques to demonstrate the growth and movement of neuronal processes in the living animal, but the fish visual system is a good place to search for such a process. What is astonishing about the retina of fishes, and indeed about their entire nervous system, is that generation of neurons and production of synaptic connections continues into adult life. These embryoniclike activities perhaps compete with or even hinder neural function, but certainly must influence it. ACKNOWLEDGMENTS Dr. S. S. Easter, Dr. W. A. Harris, Dr. D. H. Hubel and Dr. A. C. Rusoff commented on the manuscript, and my thanks to them. Mr. M. Peloquin assisted with the photography and Ms. O. Brum helped with the typing. The autoradiograph in Figure 3A and B was prepared with the assistance of Dr. R. D. Fernald. P.R.J. is supported by PHS grant EY-03301. REFERENCES Ahlbert, I.-B. 1968. The organization of the cone cells in the retinae of four teleosts with different feeding habits (Perca jiuviatilts L., Lucioperca luctoperca L., Acerina cernua L., and Coregonus albula L.). Arkiv. Zool. 22:445^481. Ali, M. A. 1959. The ocular structure, retinomotor and photo-behavioral responses of juvenile Pacific salmon. Can. J. Zool. 37:965-995. Ali, M. A. 1964. Stretching of the retina during growth of salmon (Salmo salar). Growth 28:8389. GROWTH OF F I S H RETINAS Baerends, G. P., B. E. Bennema, and A. A. Vogelzang. 1960. Uber die Anderung der Sehscharfe 457 microscopic studies on morphology of matrix cells, and on development and migration of neumit dem Wachstum bei Aequidens portalagrensis. roblasts in human and chick embryos. J. Elect. Zool. Jb. 88:67-78. Micros. 23:269-276. Beach, D. H. 1979. Patterns of cell proliferation in Hester, F. J. 1968. Visual contrast thresholds of the goldfish (Carassius auratus). Vision Res. 8:1315the retina of the clawed frog during develop1335. ment. J. Comp. Neurol. 183:603-613. ernard, H. M. 1900. Studies in the retina: Rods Hinds, J. W. and P. L. Hinds. 1974. Early ganglion and cones in the frog and in some other amcell differentiation in the mouse retina: An elecphibia. Quart. J. Micros. Sci. 43:23-47. tron microscopic analysis utilizing serial section. Devel. Biol. 37:381-416. Blaxter, J. H. S. 1975. The eyes of larval fish. In M. A. Ali (ed.), Vision in fishes: New approaches in reHollyfield, J. G. 1972. Histogenesis of the retina in search, pp. 427-444. Plenum Press, New York. the killifish Fundulus heteroclitus. J. Comp. Neurol. 144:373-380. Blaxter, J. H. S. and M. P.Jones. 1967. The development of the retina and retinomotor responses Jacobson, J. 1976. Histogenesis of retina in the in the herring (Clupea harengus). J. Mar. Biol. clawed frog with implications for the pattern of Assoc. U.K. 47:677-697. development of retinotectal connections. Brain Res. 103:541-545. Blaxter, J. H. S. and M. Staines. 1970. Pure-cone retinae and retinomotor responses in larval te- Jacobson, J. 1978. Developmental neurobiology, 2nd ed. leosts. J. Mar. Biol. Assoc. U.K. 50:449-460. Plenum Press, New York. Braekevelt, C. R. and M. J. Hollenberg. 1970. The Johns, P. R. 1977. Growth of the adult goldfish eye. development of the retina of the albino rat. J. III. Source of the new retinal cells. J. Comp. Anat. 127:281-302. Neurol. 176:343-358. Brown, M. E. 1957. The physiology of fishes, Vol. 1, Johns, P. R. and S. S. Easter. 1977. Growth of the adult goldfish eye. II. Increase in retinal cell Metabolism. Academic Press, New York. number. J. Comp. Neurol. 176:331-342. Coulombre, A. J. 1955. Correlations of structural and biochemical changes in the developing retina Johns, P. R. and S. S. Easter. 1979. The germinal of the chick. Amer. J. Anat. 96:153-189. zone in goldfish optic tectum—where it is and what its cells look like. Neurosci. Abstr. 5:165. Coulombre, A. J., S. N. Steinberg, and J. L. Coulombre. 1973. The role of intraocular pressure Johns, P. R., A. C. Rusoff, and M. W. Dubin. 1979. in the development of the chicken eye. V. PigPostnatal neurogenesis in the kitten retina. J. mented epithelium. Invest. Ophthal. 2:83-89. Comp. Neurol. 187:545-556. Easter, S. S. 1979. The growth and development of Kahn, A. J. 1974. An autoradiographic analysis of the superior oblique muscle and trochlear nerve the time of appearance of neurons in the develin juvenile and adult goldfish. Anat. Rec. oping chick neural retina. Devel. Biol. 38:30-40. 195:683-698. Kirsche, R. 1967. Uber postembryonale MatrixzoEaster, S. S., P. R.Johns, and L. R. Baumann. 1977. nen im Gehirn verschiedener Vertebrate und Growth of the adult goldfish eye. I. Optics. Videren Beziehung zur Hirnbauplanlehre. Z. Mision Res. 17:469-477. kros. Anat. Forsch. 77:313-406. Easter, S. S., P. E. Kish, and S. S. Scherer. 1979. Kock, J.-H. and T. Reuter. 1978. Retinal ganglion Growth and development of the optic nerve in cells in the crucian carp (C. carassius). I. Size and juvenile goldfish. Neurosci. Abstr. 5:158. number of somata in eyes of different size. J. Comp. Neurol. 179:535-548. Engstrom, K. 1960. Cone types and cone arrangements in the retina of some cyprinids. Acta. Zool. Lyall, A. H. 1957«. The growth of the trout retina. Quart. J. Micros. Sci 98:101-110. 41:277-295. Engstrom, K. 1963. Cone types and cone arrange- Lyall, A. H. 19576. Cone arrangements in teleost retinae. Quart. J. Micros. Sci. 98:189-201. ments in teleost retinae. Acta Zool. 44:197—243. Fisher, L. J. and S. S. Easter. 1979. Retinal synaptic Macy, A. and S. S. Easter. 1979. Retinal dimensions of receptive fields increase during growth in the arrays: Continuing development in the adult goldfish. Neurosci. Abstr. 5:794. goldfish. J. Comp. Neurol. 185:373-379. Fujita, S. and M. Horii. 1963. Analysis of cytogenesis Mann, I. 1969. The development of the human eye. Grune and Stratton, New York. in chick retina by 3H-thymidine autoradiography. Arch. Histol. Jap. 23:359-366. Meyer, R. L. 1978. Evidence from thymidine labelGaze, R. M., M. J. Keating, A. Ostberg, and S.-H. ing for continuing growth of retina and tectum Chung. 1979. The relationship between retinal in juvenile goldfish. Exp. Neurol. 59:99-111. and tectal growth in larval Xenopus: Implications Morest, D. K. 1970. The pattern of neurogenesis in for the development of the retinotectal projecthe retina of the rat. Z. Anat. Entwick. 131:45tion. J. Embryol. Exp. Morph. 53:103-143. 67. Gliicksmann, A. 1940. Development and differen- Miiller, H. 1952. Bau und Wachstum der Netzhaut des Guppy (Lebistes reliculatus). Zool. Jb. 63:275— tiation of the tadpole eye. Brit. J. Ophthal. 324. 24:153-178. Gliicksmann, A. 1965. Cell death in normal devel- Noel, VV. K. 1958. Differentiation, metabolic organization and viability of the visual cell. Arch. opment. Arch. Biol. 76:419-437. Ophthal. N.Y. 60:702-733. Hattori, T. and S. Fujita. 1974. Scanning electron 458 PAMELA RAYMOND J O H N S Northmore, D. P. M. and C. A. Dvorak. 1979. ConSchade (eds.), Degeneration patterns in the nervous trast sensitivity and acuity of the goldfish. Vision system, Prog. Brain Res. 14:143-231. Elsevier, Res. 19:255-262. New York. O'Connell, C. P. 1963. The structure of the eye of Sidman, R. L. 1960. Histogenesis of mouse retina Sardinops caeulea, Engraulis mordax and four other studied with thymidine-H3. In G. K. Smelser, The pelagic marine teleosts. J. Morph. 113:287-329. structure of the eye, pp. 487—506. Academic Press, New York. O'Rahilly, R. 1975. The prenatal development of the human eye. Exp. Eye Res. 21:93-112. Sidman, R. L. 1970. Autoradiographic methods ancl O'Rahilly, R. and D. B. Meyer. 1959. The early deprinciples for study of the nervous system with velopment of the eye in the chick. Acta. Anat. lhymidine-H3. In W. J. H. Nauta and S. O. E. 36:20-58. Ebbesson (eds.), Contemporary research methods in neuroanatomy, pp. 252-274. Springer-Verlag, Polyak, S. L. 1957. The vertebrate visual system. Univ. New York. of Chicago Press, Chicago. Rager, G. and U. Rager. 1978. Systems-matching by Silver, J. and F. W. Hughes. 1973. The role of cell degeneration. I. A quantitative electron microdeath during morphogenesis of the mammalian scopic study of the generation and degeneration eye. J. Morph. 140:159-170. of retinal ganglion cells in the chicken. Exp. Smelser, G. K., V. Ozanics, M. Rayborn, and D. SaBrain Res. 33:65-78. gun. 1974. Retinal synaptogenesis in the priRamon y Cajal, S. 1959. Studies on vertebrate neuromate. Invest. Ophthal. 13:340-361. genesis. L. Guth (trans.). Charles C Thomas, Stell, W. K. and F. I. Harosi. 1976. Cone structure Springfield, Illinois. and visual pigment content in the retina of the goldfish. Vision Res. 16:647-657. Ramon y Cajal, S. 1973. The vertebrate retina. D. Magiuire and R. W. Rodieck (trans.). In R. W. Straznicky, K. and R. M. Gaze. 1971. The growth of Rodiek (ed.), The vertebrate retina: Principles of the retina in Xenopus laevis: an autoradiographic structure and function, Appendix I, pp. 775—904. study. J. Embryol. Exp. Morph. 26:67-79. W. H. Freeman, San Francisco. Straznicky, K. and R. M. Gaze. 1972. Development Richter, W. and D. Kranz. 1977. Uber die Bedeuof the tectum in Xenopus laevis: an autoradiotung der Zellproliferation fur die Hirnregenegraphic study. J. Embryol. Exp. Morph. 26:87ration bei niederen Vertebraten. Autoradiogra115. phische Untersuchungen. Verh. Anat. Ges. Tamura, T. 1957. A study of visual perception in 71:439-445. fish, especially on resolving power and accommodation. Bull. Jap. Soc. Sci. Fish. 22:537-557. Rodieck, R. W. 1973. The vertebrate retina: Principles of structure and function. W. H. Freeman, San Vogel, M. 1978. Postnatal development of the cat's Francisco. retina. Adv. Anat. Embryol. Cell Biol. 54:6-64. Scholes, J. H. 1976. Neuronal connections and cel- Walls, G. L. 1967. The vertebrate eye and its adaptive lular arrangement in the fish retina. In F. Zettler radiation. Hafner, New York. and R. Weiler (eds.), Neural principles in vision. Weiss, P. 1949. Differential growth. In A. K. Parpart Springer-Verlag, New York. (ed.), The chemistry and physiology of growth, pp. Segaar, J. 1965. Behavioral aspects of degeneration 135-186. Princeton University Press, New Jerand regeneration in fish brain: A comparison sey. with higher vertebrates. In M. Singer and J. P. retinas of larval goldfish, rods are not present until 3 days after hatching. Production of rods is delayed until larval stages, although all of the cones in central retina Recent radioautographic studies in juvenile and adult fish, goldfish (Johns, P. R., 1980, Neurosci. Abst. are post-mitotic at hatching. This is consistent with 6:639) and the cichlid, Haplochromus burtoni (Fernald, previous observations in other teleost species. R. D. and P. R.Johns, 1981, Invest. Ophthal. Vis. Sci. Radioautography demonstrated that rods in larval 20:77), have demonstrated that new rods are pro- goldfish are generated by scattered, proliferating cells, duced by dividing precursor cells which are scattered analogous to and, indeed, the progenitors of those among mature rod nuclei in the outer nuclear layer. which continue to produce rods in the adult retina. Most of the labeled precursors for rods were in The precursors for rods derive from clusters of what peripheral (the youngest) regions of the adult retina, appear to be neuroepithelial germinal cells whose but some labeled nuclei were found even in central nuclei are sequestered in the inner nuclear layer until retina. New rods are thus continually inserted into the after about the third postembryonic day, when some of mature retina, and this probably accounts for the their progeny move into the outer nuclear layer where maintenance of a constant density of rods as the grow- they divide and generate rods. Thus another of the ing retina expands. The hypothesis of laminar shear proposed hypotheses, migration of cells from the with which my earlier study was consistent, is not sup- inner to the outer nuclear layer, is also supported by ported by these new results, but neither do they dis- these results. The dilemma of the rods is now less troublesome, but none the less intriguing as we realize prove it. I have also recently shown (Johns, P. R. and Y. that the ontogenesis of rods in the teleost retina follou s Hwang. 1981, Invest. Ophthal. Vis. Sci. 20:150; Johns, a special course. P.R.J. P. R., 1981, Neurosci. Abst. 7, in press) that in the NOTE ADDED IN PROOF