THE JOURNAL, OF COMPARATIVE NEUROLOGY 294431-442 (1990)
Development of Retinopetal Projections in the Cichlid Fish, Herotilapia d t i s p i m a ANNE C. RUSOFF AND SHARON J. HAPNER Department of Biology, Montana State University, Bozeman, Montana 59717
ABSTRACT This paper reports a study of the development of cells that project to the retina from the telencephalic nucleus olfactoretinalis and the diencephalon. Stell e t al. (Proc. Natl. Acad. Sci. USA 81r940-944, '84) have shown that the FMRFamide-immunoreactive (FMRFamide-ir) cells in the nucleus olfactoretinalis project to the retina. Therefore, we used immunocytochemistry to study the development of these cells in the nucleus olfactoretinalis. Twenty hours after fry hatched, FMRFamide-ir cells were unambiguously seen in the nucleus olfactoretinalis. At this time the axons of these cells could be traced into the optic nerve. A few hours later the axons were visible in the retina and soon attained their adult position in the inner plexiform layer near the amacrine cells. In older fry, tracers were used to fill retinopetal cells in both the nucleus olfactoretinalis and the diencephalon. Counts of these cells demonstrated that over one-third of the adult number of retinopetal cells in the nucleus olfactoretinalis are present and have axons in the retina when the fry is 9 days old, and the percentage grows to one-half by the time the fry is 1 month old. Development of the retinopetal cells in the diencephalon lags behind that of the cells in the nucleus olfactoretinalis. However, about one-third of the adult number extend their axons into the optic nerve by 1month of age. These results support our suggestion that the retinopetal cells have axons in the old part of the optic nerve because these cells were born and extended axons early in the life of the fish. K e y words: retinal efferents, centrifugal fibers, optic nerve, nucleus olfactoretinalis, FMRPamide-ir
In the previous paper (Rusoff and Hapner, '90) we demonstrated that retinopetal axons are found in the oldest part of the optic nerve and that all the retinopetal axons from the nucleus olfactoretinalis are present in the optic nerve three months after birth. These results suggested that the retinopetal axons grow into the optic nerve early in the life of the fish. This paper reports an extension of these studies to much younger fish. We used the tracers horseradish peroxidase (HRP) and cobaltous lysine to determine which cells project into the eye in fry of various ages. Since this technique was limited theoretically by the time of appearance of the retinopetal axons in the eye and practically by the extreme fragility of young fry and the tendency for tracers to diffuse out of the eye extracellularly in very young fry, we used immunocytochemistry to study the early development of the cells in the nucleus olfactoretinalis that are FMRFamide immunoreactive and project to the retina. Preliminary results from these experiments have been presented in abstract form (Hapner and Rusoff, '87). o 1990 WILEY-LISS, INC.
MATERIALS AND METHODS Rainbow cichlids, Herotilapia multispinosa (Order Perciformes, Family Cichlidae) raised in our fish-breeding facility at 27.8"C were used for all experiments. Development of the brain in these fish is extremely temperature dependent-a difference of 1°C changes its time course by several hours. Therefore, we monitored the temperature of our breeding tanks several times daily to be sure that the temperature remained constant a t 273°C.
Tracer studies Prior to application of the tracer, each fish was anesthetized with MS222 (tricaine methanesulfonate). A concentration of 0.01 5% MS222 was used for young fry; concentrations up to 0.1% MS222 were used for older fry. Very young fry were immobilized by placing them in a drop of 10% gelatin as it was beginning to solidify at 30'C. Then the Accepted November 28,1989.
A.C.RUSOFF A N D S.J.HAPNER
TABLE 2. Age and Number of Fry Processed to Reveal FMRFamide
TABLE 1. Age and Number of Fry Successfully Injected With Tracers
No. of fry
3 1 3 6 7 4 5 1 3 2
Cobalt HRP cobalt HRP cobalt Cobalt cobalt cobalt cobalt cobalt
7 9 10 14 16
24 26 'Age is calculatedfrom the time of %-laying.
tracer was placed inside the eye by piercing the cornea and inserting an insect pin coated with dried tracer into the retina. Either H R P or cobaltous lysine, made as described by Springer and Prokosch ('82), was used as the tracer. As the tracer-coated pin pierced the eye, some of the tracer began to dissolve in the released fluids. After the tracer was placed in the eye, the gelatin was quickly removed from around the fry, taking with it tracer that was outside the eye. Each fry was then placed in 1/4 ~Holtfreter'ssolution (Rugh, '48). In older fry either the optic nerve was cut and the tracer placed directly on the cut, or the cornea was opened, and a cut made in midperipheral retina, and the tracer placed in this cut; then the cornea was sutured closed. Young fry survived 15 minutes to 1hour; older fry survived 1day. Fry in which HRP was used as the tracer were prepared as 6 glutaraldescribed below: Fry were fixed by immersion in 2 7 dehyde in phosphate buffer for at least 2 hours; they were then rinsed in phosphate buffer and cryoprotected by immersion in 30% sucrose in phosphate buffer until they sank. (If a fry did not sink after 24 hours, its head was removed from its body to facilitate infiltration with sucrose.) Fry were then sectioned at 10-50 pm in a cryostat, and sections were placed directly on slides. The slides were then processed with o-dianisidine to reveal the H R P (Easter et al., '81), dehydrated in graded alcohols, and coverslipped with Permount (Fisher). Fry in which cobaltous lysine was used a5 the tracer were prepared basically as described by Bazer and Ebbesson ('84). Fry were immersed whole (Presson and Fernald, '86) in 2.5% ammonium sulfide in phosphate buffer for 10 minutes, rinsed in phosphate buffer, and fixed in 2 % glutaraldehyde in phosphate buffer. They were then cryoprotected with 30% sucrose in phosphate buffer, sectioned at 16 fim, and the sections were placed directly on slides. The precipitated cobalt was intensified by using the procedure of Bazer and Ebbesson ('84), and the sections were counterstained with cresyl violet, dehydrated, and coverslipped. Fifty-seven fry were used to work out the appropriate injection procedures. 'An additional 91 fry were injected to obtain 35 brains in which cell bodies that send axons into the retina were labeled. Successful injections utilized fry ranging in age from 6 to 26 days. (The day the egg was laid is considered day 0 since fertilization follows egg laying. Hatching occurs about 48 hours later. Since the eggs are often laid at night and the fry in a brood usually hatch within one hour of each other, we staged early development from hatching and added 48 hours to determine the approximate time of egg-laying or total age of the fry.) Table 1 shows the ages and numbers of successfully injected fry.
Immunocytochemistry Since some of the cells in the nucleus olfactoretinalis (NOR) that project to the retina in the adult and juvenile
No. of fry
Hatching (2 days) 8 hours p.h. 12 hours p.h. 16 hours p.h 20 hours p . h 24 hours p.h. (3days) 40 hours p.h 43 hours p.h. 49 horn p.h. (4 days) 65 hours p.h. 72hoursp.h.(5days) 96 hours p.h. (6days) 5 daysp.h. ('id& 7 days p.h. (9 day4 14dapp.h. (16dap)
4 4 4 7 6 3 5 2
3 8 1 1 2
'Age is shown both in time past hatching b h . ) and, in parenth-,
indays from egg-laying.
rainbow cichlid fish are immunoreactive for the molluscan cardioexcitatory peptide FMRFamide (Rusoff and Hapner, 'go), we used antibodies to FMRFamide to study the development of these cells and their projection to the retina. Fry were fixed by immersion in 4% paraformaldehyde in 0.05 M phosphate buffer plus 3% sucrose. (Initially, fry were fixed by immersion in 2% glutaraldehyde in phosphate buffer; however, none of the tissue from these fry displayed any antigenicity toward the FMRFamide antibodies.) After a minimum of 2 hours in the fixative, the fry were rinsed in phosphate buffer, cryoprotected in 30% sucrose in phosphate buffer, and then sectioned at 10-20 pm in a cryostat. Many fry were sectioned transversely; some were sectioned horizontally and others sagittally. Except for those sectioned sagittally, the definition of the plane of sectioning is arbitrary as the fry, especially the young ones, are curled around their yolk sacs and a horizontally sectioned fry yielded transverse sections of the telencephalon. The sections were placed directly on freshly prepared coated slides and stored in a refrigerator for a maximum of two days. The slides were bathed in each of the solutions required to reveal FMRFamide immunoreactivity as described in the accompanying paper (Rusoff and Hapner, '90). Antibodies to FMRFamide were obtained from either Immunonuclear or Cambridge Research Biologicals. The antibodies from the two sources were used either alone at a dilution of 1:1,000 or mixed with that from Immunonuclear used a t a dilution of 1:1,000 and that from Cambridge used at 1:5,000. Control slides were incubated with normal rabbit serum a t a dilution of 1:1,000. Fifty-six fry ranging in age from 2 days old (the time of hatching) to 16 days old were used for this study. Table 2 shows the ages and numbers of fry used.
RESULTS The rainbow cichlid is a tropical fish that prefers to lay its eggs at about 27.8OC. At this temperature the eggs hatch in about 48 hours; the fry have huge yolk sacs so they settle to the bottom of the tank and are unable to swim for several days (Sigall, '87).At the time of hatching the nervous system is very immature (Fig. 1).Although the eye is obvious and makes up a large part of the nervous system, there is no obvious differentiation of cells within the retina, no separation of cells to form the ganglion cell layer, and no indication of axons in the optic stalk at the light microscopic level. The telencephalon is extremely immature; no separate nuclei can be discerned although the marginal zone contains a neuropil (Fig. 1, arrow) and the cells farthest from the
DEVELOPMENT OF RETINOPETAL PROJECTIONS
Figs. 1-9. Unless otherwise stated, all figures are photographs of portions of 10 pm-thick sections and were processed to reveal FMRFamide immunoreactivity. Fig. 1. A: A 1pm-thick section cut transversely through the head of a fry killed at hatching. Although the eyes and various other regions of the brain can be distinguished, the cells in the nervous system are mostly undifferentiated. The cells of the pigmented epithelium are just beginning to produce pigment at this time so there is no pigmented zone around the retina. The solid arrow indicates one of the few regions of
ventricle have begun to differentiate. They are round and stain more palely than the immature cells a t the ventricular margin (Fig. 1B). None of these cells can be identified as belonging to the future nucleus olfactoretinalis as no cells in this region are clearly immunoreactive. Presumably the nucleus olfactoretinalis will be near the region where the olfactory placode abuts the neural tube (Fig. 1A). Eight hours later the appearance of the brain has only changed slightly. The central part of the retina contains a layer of differentiated ganglion cells, but the inner plexiform layer is not yet visible. In the region of the telencephalon closest to the olfactory placode a few large, pale-staining cells are close to the margin of the neural tube. Again, these cells are not clearly immunoreactive. By 12 hours after hatching differentiation is beginning to be more obvious in the retina and telencephalon. The inner plexiform layer is visible in the central portion of the retina (Fig. 2A,B) and the optic fiber layer is present a t the optic fissure (Fig. 2C), suggesting that ganglion cell axons have begun to grow out the optic stalk toward their targets. There are more differentiated cells in the telencephalon, and some of them have started to separate from the main mass of cells (Fig. 3A,B). A few of the cells at the margin of the telencephalon are very close to the olfactory placode and may be FMRFamide immunoreactive neurons (Fig. 3B, arrow); however, a t this stage the cells are small and their
neuropil in the telencephalon visible at this early time. The scale line indicates 100 pm. AO, adhesive organ; L, lens; OP, olfactory placode; R, retina; TEC, tectlm; TEL, telencephalon;V, ventricle. B: A more highly magnified view of part of the telencephalon from an adjacent section. The majority of the cells still appear neuroblastic with highly elongated shapes and dark cytoplasm. However, a few cells located near the neuropil have begun to differentiate; these cells are much rounder, and their nuclei are round and pale with a prominent nucleolus (arrow). The scale line indicates 20 pm. N, neuropil; V, ventricle.
features are obscured by the immunoreaction product so that it is not possible to positively identify the cells as immunoreactive neurons rather than vascular elements. However, absence of staining in this region in the controls is evidence for the neuronal nature of the stained cells. Sixteen hours after hatching the retina is much more mature. In the central retina cells in the ganglion cell layer are rounded and clearly separated from the rest of the retina by the inner plexiform layer; the outer plexiform layer is visible. The optic fiber layer is clear and continuous with neuropil in the optic stalk, and it is continuous with the neuropil of the ventral diencephalon. No FMRFamide-ir fibers were seen in the optic stalk or retina a t this stage, and the immunoreactivity of the region of the telencephalon that will become the nucleus olfactoretinalis is still questionable. Labeled cells in this region are still not clearly neurons. However, by 20 hours after hatching the incipient nucleus olfactoretinalis can be recognized both by the position of a few cells in the brain, at the margin adjacent to the connection between the olfactory placode and the telencephalon, and by the presence of FMRFamide-ir cells (Fig. 3C,D). The identity of these cells as FMRFamide-ir neurons rather than vascular elements is confirmed by the presence of immunoreactive fibers both in the neuropil of the telencephalon approaching the diencephalon and in the optic
A.C. RUSOFF AND S.J. HAPNER
Fig. 2. A: A section through the most differentiated part of the retina of a 12 hour post-hatching fry. The arrow indicates the inner plexiform layer that is present in only a small part of the retina a t this time. The scale line indicates 50 pm. B A more highly magnified view of' the region of the retina indicated by the arrow in A. The inner plexiform layer clearly separates the ganglion cell layer from the neuroblastic layer. T h e scale line indicates 20 pm. GCL, ganglion cell layer; IPL, inner plexiform layer; MZ, mitotic zone. C A section through the other eye of this same fry a t the level of the entrance of the optic stalk into the eye. This section is not through the most differentiated part of the retina so the inner plexiform layer is not visible. However, the optic fiber layer is visible near the optic nerve head (arrow). T h e scale line indicates 50 pm.
nerves. Figure 4 shows a section from another fry fixed 20 hours after hatching. A FMRFamide-ir fiber is visible in the optic nerve approaching the retina. No immunoreactive fibers were seen within the retina itself a t this time. Figure 5 shows the course of the FMRFamide-ir fibers between the nucleus olfactoretinalis and the retina in one 24 hour post-hatching fry. Figure 5A shows the nucleus olfactoretinalis (arrow) in the telencephalon with many FMRFamide-ir cells; FMRFamide-ir axons from these cells pass through the telencephalic neuropil toward the diencephalon (Fig. 5B); the axons are present in the diencephalic neuropil
before entering the optic tract. They follow the tract across the chiasm (Fig. 5C,D) into the optic nerve (Fig. 5D,E) and enter the retina via the optic nerve head. (Fig. 5C,D shows axons from the left nucleus olfactoretinalis while the other figures show axons from the right nucleus olfactoretinalis in this same fry.) Consistently the FMRFamide-ir axons are seen at the margins of the optic nerve (Fig. 5E) and cross the ganglion cell fiber layer of the retina close to the ganglion cell bodies (Fig. 5F). After passing dorsally across the surface of the ganglion cell bodies for a short distance, the FMRFamide-ir axons penetrate the ganglion cell layer. We
DEVELOPMENT OF RETINOPETAL PROJECTIONS
Fig. 3. A: A section cut transversely through the telencephalon of a fry killed 12 hours after hatching. The arrow indicates the region shown in higher magnification in B. Both olfactory placodes are visible in the lower half of the section while an adhesive organ is visible on the top, left side of the section. The scale line indicates 50 pm. B A more highly magnified view of the region indicated by the arrow in A. The arrow indicates cells that may be FMRFamide-ir but may also be vascular elements. Note the close proximity of these cells to the olfactory placode and its close proximity to the marginal zone of the telencephalon. Some of the cells of the telencephalon have separated from the ventricular zone
and appear differentiated. The scale line indicates 20 Wm. C: A part of a section cut transversely through the telencephalon of a fry killed 20 hours after hatching. Cells in both the olfactory placode and telencephalon appear more differentiated than in A and B. The arrow indicates FMRFamide-ir cells separated from the ventricular zone in the presumptive nucleus olfactoretinalis. The scale line indicates 50 Fm. D: A more highly magnified view of the FMRFamide-ir cells (arrow) shown in C . The olfactory placode is still in very close proximity to the telencephalon. The scale line indicates 20 pm.
Fig. 4. A section from a fry killed 20 hours after hatching and cut horizontally through the eye at the level of the optic fissure. The arrow indicates the optic nerve as it enters the eye through the fissure. The scale line indicates 50 pm. The inset shows a more highly magnified view of the optic nerve; the arrow indicates a FMRFamide-immunoreactive fiber in the optic nerve.
did not observe any FMRFamide-ir axons penetrating into the inner plexiform layer at this stage. The next stage that we sampled was at 40 hours after hatching. The retina yias much more differentiated at this time. The inner plexiform layer was much thicker and the cells in the inner nuclear layer were divisible into an outer and an inner group that stained differently. The outer plexiform layer was also more prominent, as was a band of horizontal cells along its inner margin. The FMRFamide-ir fibers were visible across the ganglion cell fiber layer by this time and had proces;es on both sides of the inner plexiform layer. In horizontal sections through the eye the fibers appear first in the optic fissure ventral in the eye, then they are seen crossing the surface of the ganglion cell layer as they ascend in the eye until they penetrate the ganglion cell layer and are visible a t the outer aspect of the inner plexiform layer next to the amacrine cell bodies. Thus, FMRFamide-ir fibers extend from the nucleus olfactoretinalis to the outer aspect of the inner plexiform layer of the retina before the fry begin to swim at 3 days post-hatching. Figure 6 shows a section from a 43 hour post-hatching fry that was sagittally sectioned. The entire pathway of the FMRFamide-ir fibers (solid arrows) from the nucleus olfactoretinalis (open arrows) into the optic tract is clear in this fish. The only major change that occurred in the pathway of the FMRFamide-ir fibers as the fish matured after this point was the extension of the fibers in the inner plexiform layer near the amacrine cells. As the retina grows, the fibers
A.C. RESOFF AND S.J. HAPNER extend toward the growing margin remaining near the amacrine cells as they grow (Fig. 7). Since we do not yet know any specific antigens contained by the diencephalic efferents, we were not able to use immunocytochemistry to study their development. However, by the time the fry were 6 days old (4 days posthatching), we were able to inject tracers into one eye and label the optic projection to the tectum and the diencephalon (Fig. 8A,B). We were not able to verify the presence of retrogradely labeled cells in the nucleus olfactoretinalis at this age because the tracer spread extracellularly into the telencephalon and obscured the cells there. However, the visual projections could be demonstrated a t this age. In the 6 day old fry shown in Figure 8A,B, ganglion cell axons filled with cobalt are visible in the optic tract coursing along the diencephalon on their way to the tectum as well as terminating in the diencephalon. The diencephalic nuclei are visible; the small size of their cells and the presence of terminals around them makes it difficult to determine if any of the cell bodies are really filled with tracer, although the arrow indicates one possible retinal efferent. By the time the fry are 9 days old (7 days post-hatching), obvious cells in the diencephalon filled with tracer are visible (Fig. 8C,D). By this time it is also possible to visualize tracer-filled cells in the nucleus olfactoretinalis although they become easier to see in older fry. Figure 9A,B shows cells filled with cobalt in the nucleus olfactoretinalis of an 18 day old fry, and Figure 9C,D shows similarly filled cells in the diencephalon of the same fry. The number of retinopetal cells in the nucleus olfactoretinalis is relatively easy to estimate since the tracer-filled cells may be directly counted or their axons may he counted in sections between the nucleus and the optic tract. The number of such cells increases very rapidly. Table 3 gives the results of counts of these cells in several fry between 9 and 26 days old. One 9-day-old fry already had over one-third of the adult number of retinopetally-projecting cells. (The adult rainbow cichlid has approximately 100 retinopetally-projecting cells in the nucleus olfactoretinalis [Rusoff and Hapner, '901.) Before the fry were 1 month old, they had over half of their adult complement of retinop-
Fig. 5. A series of sections cut transversely through the head of a fry killed 24 hours after hatching. The scale line indicates 30 gm and applies to each of the parts. A: A section through the telencephalon (TEL) a t the level of the nucleus olfactoretinalis. The arrow indicates FMRFamide-ir cells in that nucleus on the right side. B: A more caudal section through the junction of the telencephalon with the diencephalon. The arrow indicates a FMRFamide-ir axon that is in the neuropil ipsilateral to the nucleus olfactoretinalis (right) in which its cell body lies. R, retina. C Section through the diencephalon (D) a t the level of the optic chiasm. A FMRFamide-ir axon (arrow) is visible in the optic tract approaching the chiasm. (This axon originated from the left nucleus olfactoretinalis.) D: Section adjacent to the one shown in C. A portion of the FMRFamide-ir axon seen in C is visible across the chiasm (C) in the right optic nerve (arrow). D, diencephalon; R, retina. E Section adjacent to the one shown in D. Portions of two FMRFamide-ir axons (arrows) are visible in the left optic nerve. These axons are very close to the margins of the nerve. C, chiasm). F: Section through the ventral retina of this fry. The optic fiber layer (OFL) is present only in the ventral retina near the optic nerve head. Arrows indicate a FMRFamide-ir axon that enters the retina from the margin of the nerve head and courses dorsally across the retina a t the interface between the optic fiber layer and the ganglion cells ( G C ) .
DEVELOPMENT OF RETINOPETAL PROJECTIONS
A.C. RUSOFF AND S.J. HAPNER
Fig. 6. A: A 16 pm-thick section cut sagittally through a fry killed 43 hours after hatching. Rostra1 is to the right and caudal is to the left. FMRFamide-ir cells (open arrows) are visible a t the junction between the olfactory bulb (OB) and the telencephalon. Portions of FMRFamide-ir fibers (solid arrows) may be traced caudally through the telencephalon (TEL) into the optic tract (OT). The lowest solid arrow indicates FMRFamide-ir fibers in a more ventral portion of the optic tract. The scale line indicates 50 pm. B A more highly magnified view of the FMRFamide-ir cells in the nucleus olfactoretinalis (open arrows) and of axons leaving the nucleus (solid arrow). The scale line indicates 20 pm.
etally projecting cells in that nucleus. The accompanying paper shows that the remainder of the retinopetal cells in the nucleus olfactoretinalis extend axons into the retina within the next two months (Rusoff and Hapner, '90). The diencephalic efferents are harder to count than the efferents from the nucleus olfactoretinalis because their axons are indistinguishable from neighboring axons of retinal ganglion cells and their cell bodies are spread throughout more sections so that the chance of losing sections and,
therefore, missing cells is higher. The counts from those fish in which we were convinced that all the diencephalic efferents were visible are also given in Table 3. A few of the retinopetally projecting cells in the diencephalon are already demonstrable when the fry is 9 days old, and their number increases with age. The percentage of the adult number of retinopetal cells present in the diencephalon lags behind that present in the nucleus olfactoretinalis at all ages shown in Table 3. Most fish probably do not have the adult
DEVELOPMENT OF RETINOPETAL PROJECTIONS
Fig. 7. A A section cut transversely through a peripheral portion of the retina from a fry killed 48 hours after hatching. The section includes relatively mature retina a t the top of the photograph and immature retina a t the bottom. Dorsal is to the right. The solid arrow indicates FMRFamide-ir fibers a t the junction between the inner plexiform and inner nuclear layers. T h e scale line indicates 50 fim. B A more highly magnified view of the mature portion of the retina shown in A. Notice
the different staining of cells in the inner and outer parts of the inner nuclear layer. T h e different types of cells in the layer are becoming differentiated. The solid arrow indicates the same fibers marked in A. T h e open arrow indicates immunoreactive fibers at the junction between the ganglion cells and the inner plexiform layer. T h e scale line indicates 20 Wm. INL, inner nuclear layer; IPL, inner plexiform layer; PR, photoreceptors.
number of diencephalic efferents even a t 3 months of age (see Table 1 in Rusoff and Hapner, 'go), although they are approaching the low end of the adult range in numbers. Presumably, diencephalic cells that project to the retina after 3 months of age do so within a few months as they are all present by 10 months of age (see Table 1 in Rusoff and Hapner, '90). No other sources of retinopetal axons were found in any of these fry.
same stage of retinal development in this fish as in other species. In the mouse ganglion cell axons begin to grow into the optic stalk shortly after ganglion cells migrate to their final position in the retina and begin to assume a differentiated morphology (Hinds and Hinds, '74) and about the time of appearance of the inner plexiform layer. In the rainbow cichlid ganglion cells are distinguishable 8 hours after the egg hatches and the inner plexiform layer is clearly visible 12 hours post-hatching. (We did not sample a t 10 hours post-hatching.) We assume, therefore, that the first ganglion cell axons grow into the optic stalk between 8 and 1 2 hours after hatching. The retinopetal projection from the nucleus olfactoretinalis is clearly demonstrable by 20 hours posthatching and the cells may be immunoreactive as early as 12 hours post-hatching. When clearly visible at 20 hours posthatching, the FMRFamide-ir fibers already extend through the optic stalk into the optic fissure. Thus, they too must have grown into the optic stalk in the diencephalon earlier. Since the centripetally growing retinal ganglion cell axons and the centrifugally growing axons from the nucleus olfactoretinalis probably enter the optic stalk within a few hours of each other, we wondered if they interact with each other. Unfortunately here the technique fails us. We cannot separate centrifugally growing axons from centripetally growing axons until the retinopetal axons begin to express the antigen that binds antibodies to FMRFamide. If the axons are present but not yet expressing this antigen, we can not detect them with our present technique.
DISCUSSION The early appearance of antigens that, bind antibodies to FMRFamide allowed us to study the early development of one source of retinopetal projections, the nucleus olfactoretinalis. We demonstrated the presence of the projection from the nucleus olfactoretinalis through the optic stalk to the retina within one day after hatching. One would like to compare their first appearance in the optic stalk to that of the first axons of retinal ganglion cells. However, since we examined our tissue a t the light microscopic level, we could not detect the presence of the first retinal ganglion cell axons in the optic stalk and diencephalon. Since the pattern of development of the retina of the rainbow cichlid is similar to that seen in other fish (Muller, '53; Blaxter and Jones, '67; Hollyfield, '72; Wagner, '74; Griin, '75; Sharma and Ungar, '80) as well as in mammals (Sidman, '61), we assume that the ganglion cell axons grow into the optic stalk a t about the
Fig. 8. A A section cut transversely through the diencephalon (D) and rostra1 tectum (T) of a fry that received an injection of cohaltous lysine into the left eye 4 days after hatching and was killed 10 minutes later. Cobalt filled axons are visible in the optic tract (0) and in portions of the diencephalon. Filled terminals are visihle in the tectum. A retinorecipient region of the diencephalon is indicated by the arrow. The scale line indicates 100 p m . B: A more highly magnified view of the right diencephalon. A cell that appears to be filled with cobalt is indicated by
A.C. RUSOFF AND S.J. HAPNER
the arrow. The scale line indicates 50 pm. C: A section cut transversely through the diencephalon (D) and tectum ( T )of a fry that received an injection of cohaltous lysine into the right eye 7 days after hatching. Cobalt-filled axons are visible approaching the left tectum and terminating there. The arrow indicates the diencephalic region that is retinorecipient and is shown in D. The scale line indicates 100 pm.D: A more highly magnified view of the cobalt-filled terminals and cells (arrows) in the diencephalon. The scale line indicates 20 pm.
DEVELOPMENT OF RETINOPETAL PROJECTIONS
Fig. 9. A A 16 pm-thick section cut transversely through t h e telencephalon of a fry t h a t received an injection of cobaltous lysine into the right eye 16 days after hatching. Cobalt-filled cells are indicated by the arrow. T h e scale line indicates 20 pm. B: The cobalt-filled cells are shown in higher magnification. The scale line indicates 50 pm. C: Section showing the diencephalon and rostra1 tectum (TEC) from t h e same fish. Cobalt-filled cells (arrow) and axons are visible in the diencephalon. A
cluster of terminals is also visible in the tectum. (The section was accidentally inverted on the slide so t h a t the filling appears to be on the right side of the brain. I t is actually on the left as is the filling in the nucleus olfactoretinalis.) T h e scale line indicates 20 fim. D: A more highly magnified view of the cohalt-filled cells in the diencephalon. T h e scale line indicates 10 pm.
A.C. RUSOFF AND S.J. HAPNER
442 TABLE 3. Numher of Tracer-Filled Cells in the Nucleus Olfactoretinalis and in the Diencephalon Diencephalic cells
NOR cells Fish no. 20%
20% 192k 209h 189c 190h 211a
9 9 9
38 17' 63
14 16 18 24 26
67' 74 129 114
Fritzsch ('83) in that we both report early development of the cells that project to the retina with those from the diencephalon lagging behind those from the nucleus olfactoretinalis.
51 96 80
'An Abercrombie correction designed to prevent double counting of 1 cell that appears in two sectionswas used to calculate the correctedvalues. This calculation wumes an equal number of cells in all sections which was clearly not true. Therefore, the real value lies between the counted p d correctedvalues (Ahercrombie,'46). 'Cells were relatively scattered so that double counting did not appearto he a problem.
When do the diencephalic efferents first reach the retina? The absence of a way to study them without using tracers precludes studies of their very early development. Our tracer study suggests that they develop later than the retinopetal projection from the nucleus olfactoretinalis. In 6 day old fish we found only a couple of filled bodies in the diencephalon that may have been filled cells; 9-day-old fish clearly had filled cells in the diencephalon but the numbers were small (between 10 and 30). (See Table 3.) Compare this result with the counts in the nucleus olfactoretinalis. The 14-day-old nucleus olfactoretinalis already has at least one-third of all the retinopetal cells that it will contain in the adult, while the 16-18-day diencephalon has only about one-fourth of the retinopetal cells that it will have in the adult. The work of Presson and Fernald ('86) also suggests that the diencephalic efferents develop later than the visual projections. They studied the development of the divisions of the optic tract in another cichlid fish. Their youngest fish were developmentally younger than ours, since the divisions of the optic tract were well-formed in the youngest fish that we successfully injected (6 day old fish that were beginning to swim). However, they do not mention observing any filled cell bodies in the diencephalon. The later appearance of diencephalic efferents fits well with the results of another study of the development of retinopetal cells in a fish. Crapon de Caprona and Fritzsch ('83) studied the development of the nucleus olfactoretinalis in two other cichlid fish and found that cells there could be filled with reaction product by placing H R P in the eye of fry 1 day after a projection was first present from the retina to the brain. They did not observe any diencephalic efferents until a few days later. According to their cell counts, all of the cells in the nucleus olfactoretinalis that project to the retina are present by 3 weeks of age. They did not attempt to count the diencephalic efferents. Although the exact days of appearance of all the retinopetal cells differ in the different fish, our results agree with those of Crapon de Caprona and
We thank Dr. Wm. Stell for the gift of an aliquot of antibodies to salmon LHRH and Dr. Stephen S. Easter, Jr., for critically reading the manuscript. This work was supported by NIH grant EY06495 to A.C.R.
LITERATURE CITED Ahercrombie, M. (1946) Estimation of nuclear population from microtome sections. Anat. Rec. 94239-247. Bazer, G.T., and S.O.E. Ebbesson (1984) A simplified cobalt-lysine method for tracing axon trajectories in the central nervous system of vertebrates. Neurosci. Lett. 51:315-318. Blaxter, J.H.S., and M.P. Jones (1967) The development of the retina and retinomotor responses in the herring. J. Mar. Biol. Assoc. U.K. 47:677697. Crapon de Caprona, M.-D., and B. Fritzsch (1983) The development of the retinopetal nucleus olfacto-retinalis of two cichlid fish as revealed by horseradish peroxidase. Dev. Brain Res. 11:281-301. Easter, S.S., Jr., A.C. Rusoff, and P.E. Kish (1981) The growth and organization of the optic nerve and tract in juvenile and adult goldfish. J. Neurosci. 2:793-811. GrIin, G . (1975) Structural basis of the functional development of the retina in the cichlid Tilapia leucosticta (teleostei). J. Embryol. Exp. Morphol. 33243-257. Hapner, S.J., and A.C. Rusoff (1987) Development of the nucleus olfactoretiNeurosci. Abstr. 13:1692. nalis in the rainbow cichlid. SOC. Hinds, J.W., and P.L. Hinds (1974) Early ganglion cell differentiation in the mouse retina: An electron microscopic analysis utilizing serial sections. Dev. Biol. 37:381-416. Hollyfield, J.G. (1972) Histogenesis of the retina in the killifish, Fundulus heteroclitus. J. Comp. Neurol. 144~373-380. MUller, H. (1953) Bau und Wachstum der Netzhaut des Guppy (Lebistes reticulatus). Zool. Jahrb. Abt. Allg. Zool. Physiol. 63:275-324. Presson, J.C., and R.D. Fernald (1986) Development of the optic tract in the cichlid fish Haplochromis burtoni. Dev. Brain Res. 26:179-186. Rugh, R. (1948) Experimental Embryology. 2nd Ed. Minneapolis: Burgess Publishing Co., p. 10. Rusoff, A.C., and S.J. Hapner (1990) Organization of retinopetal axons in the optic nerve of the cichlid fish, Herotilapia multispinosa. J. Comp. Neurol. 294:418-430. Sharma, S.C., and F. Ungar (1980) Histogenesis of the goldfish retina. J. Comp. Neurol. 191:373-382. Sidman, R.L. (1961) Histogenesis of mouse retina studied with thymidine "H. In G.K. Smelser (ed): Structure of the Eye. New York: Academic Press, pp. 487-505. Sigall, M.W. (1987) At the end of the rainbow: T h e ideal cichlid. Trop. Fish Hobbyist 35(i I):26-37. Springer, A.D., and J.H. Prokosch (1982) Surgical and intensification procedures for defining visual pathways with cobaltous-lysine. J. Histochem. Cytochem. 30:1235-1242. Stell, W.K., S.E. Walker, K.S. Chohan, and A.K. Ball (1984) The goldfish nervus terminalis: A luteinizing hormone-releasing hormone and molluscan cardioexcitatory peptide immunoreactive olfactoretinal pathway. Proc. Natl. Acad. Sci. USA 81:940-944. Wagner, H.-J. (1974) Die Entwicklung der Netzhaut von Nannacara anomala (Regan) (Cichlidae, Teleostei) mit besonderer Berucksichtigung regionaler Differenzierungsunterschiede.Z. Morphol. Teire 79:113-131.