Neuronal Potentialities of Cells in the Optic Nerve of the Chicken Embryo

Proc. Nadl. Acad. Sci. USA Vol. 87, pp. 1643-1647, March 1990 Developmental Biology Neuronal potentialities of cells in the optic nerve of the chicken embryo are revealed in culture (optic stalk/differentiation/neuroepithelium/neuronal precursor cells/neurofilaments) MARIE-CLAUDE GIESS, PHILIPPE COCHARD*, AND ANNE-MARIE DUPRAT Centre de Biologie du D6veloppement, Centre National de la Recherche Scientifique, Unite de Recherche Associde 675 affilide A l'Institut National de la Sante et de la Recherche Mddicale, Universitd Paul Sabatier, Toulouse, France Communicated by J. B. Gurdon, November 13, 1989 (receivedfor review September 29, 1989)

ABSTRACT Neuronal potentialities in neuroepithelial elegant and thorough work of Raff and his colleagues (8-13) cells of the chicken embryonic optic nerve were studied in has led to the characterization of a glial precursor cell in the culture by using neurofilament antibodies as neuronal mark- optic nerve, called an O-2A cell, generating both oligoden- ers. Embryonic day4 and -5 (E4 and ES) optic stalks were drocytes and fibrous, or type 2, astrocytes. This 0-2A explanted in vitro. Within the first few days of culture, numer- progenitor cell is not recognizable in the optic stalk before ous morphologically identifiable neurons extending long neu- day 16 of gestation [embryonic day 16 (E16)] and appears to rites developed. These neurons and their processes were spe- migrate into the optic nerve from external, possibly cerebral, cifically labeled with neurofilament antibodies. Similar results sources (14). From these results, it has been inferred that were obtained by explanting only the medial portion ofE7 optic intrinsic neuroepithelial cells of the optic stalk are unable to stalks away from possibly contaminating cerebral or retinal generate neurons and that their development may be re- tissue. To determine whether neuronal potentialities persisted stricted only to the protoplasmic (type 1) astrocytic pheno- at later embryonic stages, cultures of dissociated optic stalks describes neuronal were established at Ell, E1S, and E18. Neurons labeled with type. On the other hand, a report (15) the various neurofilament antibodies appeared in all cultures of differentiation in vitro from optic stalk cells of the mouse Eli and E15 optic stalks. However, typical neurons could not embryo for a short period between E10 and E11.5 during be recognized in cultures of E18 optic nerves. These results gestation. Thus, cells with neuronal potentialities may occur indicate that cells with neuronal potentialities are present in the in this region ofthe CNS but only at very early developmental embryonic optic nerve from early stages of development and stages. persist until at least E1S. Since the adult optic nerve is devoid We have been studying the question of the occurrence of of nerve cell bodies, our observations are consistent with the cells with neuronal potentialities in the chicken embryonic hypothesis that axons of retinal ganglion cells, which course optic nerve. Our' strategy has been to isolate the optic stalk through the optic stalk, repress neuronal potentialities within from the embryonic environment at various stages of devel- a subpopulation ofprecursor cells during normal development. opment and culture it in vitro. Using morphological and immunocytochemical criteria, we demonstrate that, at least During the development of the vertebrate central nervous between 4 and 15 days of'incubation, neuronal cells develop system (CNS), it remains to be determined how undifferen- from these cultured optic stalks. tiated cells of the primitive CNS anlage are committed to a specific lineage, particularly, how neuronal and glial cell MATERIAL AND METHODS types are segregated. Whether precursor cells of the neural plate or neural tube are capable of giving rise indifferently to Optic Peduncle Disetion and Culture. Eggs (White Leghorn neurons and macroglial cells or are irreversibly committed to chicken) were incubated at 380C for 4-18 days. Developmental a given cell type is a matter of debate, despite various stages were verified using the series ofHamburger and Hamil- attempts to solve this problem (e.g., see refs. 1-7). ton (16). Embryos were placed in sterile Tyrode solution. The A model system that can be used to address this question optic nerve primordium was carefully dissected as follows is the optic nerve. The optic nerve arises from the dienceph- (Fig. 1): a semicircular dorsal incision was made around the alon, which, soon after neural tube closure, bulges bilaterally eyeball and the eyeball was then retracted to expose its to form the optic vesicles. As the optic vesicle extends to posterior surface where the optic nerve emerges from the contact the lateral ectoderm, the part of the neuroepithelium retina. Muscle and connective tissues were dissected away to connecting the optic vesicle to the diencephalon narrows and expose the entire length of optic stalk until it penetrates the forms the optic stalk, presumptive territory of the optic brain. Two transverse sections performed in the middle por- nerve. Thus, the optic stalk is merely a neuroepithelial tion of the nerve-i.e., at some distance from the retina and extension of the diencephalon. However, the adult optic diencephalon-allowed us to remove the optic stalk without nerve, although totally devoid of nerve cell bodies, contains contamination by other CNS tissue (Fig. 1 Inset). axons ofretinal ganglion cells surrounded by several glial cell At stages ranging from 4 to 7 days of incubation, optic types, including fibrous astrocytes, protoplasmic astrocytes, stalks were cultured as explants. At later stages, optic stalks and oligodendrocytes. Therefore, the question arises whether were dissociated at 370C in 0.04% collagenase (Boehringer) in optic stalk neuroepithelial cells are committed from the onset Ca2l- and Mg2"-free Tyrode solution. After 20 min, the to glial lineages or, at least for some time, have the potential explants were washed with Tyrode solution and gently trit- to develop along the neuronal lineages. urated with a blunt Pasteur pipette. The cell suspension was Until now, developmental potentialities ofoptic nerve cells centrifuged, washed in complete medium, and plated in have been documented mostly in the perinatal rat. The Abbreviations: CNS, central nervous system; NF, neurofilaments; The publication costs of this article were defrayed in part by page charge FITC, fluorescein isothiocyanate; TRITC, tetramethylrhodamine payment. This article must therefore be hereby marked "advertisement" isothiocyanate; E, embryonic day. in accordance with 18 U.S.C. §1734 solely to indicate this fact. *To whom reprint requests should be addressed.

1643 Downloaded by guest on September 24, 2021 1644 Developmental Biology: Giess et al. Proc. Natl. Acad. Sci. USA 87 (1990) tional de la Recherche Scientifique, Strasbourg, France), used at 1:100 dilution. The specificity of all antibodies was verified on Western blots containing chicken nervous system intermediate filaments, separated on polyacrylamide gels (18) and transferred to nitrocellulose (19). Results for the three anti-NF monoclonal antibodies are shown in Fig. 2. Immunofluorescence Staining. Cultures were fixed at various stages (usually 5, 7, or 15 days after plating) with 3.5% (vol/vol) formaldehyde in phosphate-buffered saline (Pi/ NaCl = 0.01 M sodium phosphate, pH 7.4/0.9% NaCl). After extensive rinsing in P,/NaCl, coverslips were preincubated for 15 min with Pi/NaCl containing 2% (wt/vol) lyophilized skimmed milk and 0.25% Triton X-100. All antibodies were diluted in P1/NaCl/milk. Incubation with specific antibodies was at room temperature for 30 min. After three 5-min rinses in P1/NaCl/milk, coverslips were incubated with the appropriate fluorescein isothiocyanate (FITC)- or tetramethylrhodamine isothiocyanate (TRITC)-conjugated anti-immunoglobulin antibody, under the same conditions. The conjugated reagents and their working dilutions were as follows: goat anti-mouse immunoglobulin/FITC (Nordic, Tilburg, FIG. 1. Schematic drawing of the head of an E7 chicken embryo immu- summarizing the dissection procedure of the embryonic optic nerve. The Netherlands) at a 1:100 dilution; goat anti-mouse The optic stalk (OS), well-individualized at this stage, is exposed noglobulin/TRITC (Nordic) at a 1:20 dilution; goat anti- from a dorsal view. CG, ciliary ganglion; E, eyeball. (Inset) Enlarge- rabbit IgG/FITC (Nordic) at a 1:40 dilution; and goat anti- ment of the dissection field. Two transverse sections (double arrows) rabbit IgG/TRITC (Immunotech, Marseille, France) at a 1:50 allow the medial part of the optic stalk to be removed. dilution. In double-labeling experiments, cells were incubated se- 4-well plastic dishes (Nunc) onto 16-mm glass coverslips quentially with the specific antibodies, washed, and stained, coated with rat-tail collagen (Sigma). again sequentially, with the appropriate FITC- or TRITC- The culture medium was Dulbecco's modified Eagle's conjugated antibodies. None of these fluorescent antisera medium supplemented with 10% (vol/vol) fetal calf serum cross-reacted with inappropriate immunoglobulins. (Flow Laboratories) and gentamicin (Sigma) (50 gg/ml). Control coverslips, for which the primary antibody or, for Cultures were maintained at 370C in a humidified atmosphere double-labeling experiments, either the primary or secondary of'5% C02/95% air. Cultures were fed every 3 days. antibody was omitted, were always included. Antibodies. Antibodies used were as follows: NF70 (Sig- After incubation with the fluorescent antibodies, cover- ma), a mouse monoclonal antibody specific for the light slips were washed for three 5-min periods in P1/NaCl/milk, neurofilament (NF) subunit (Fig. 2, lane a) used at a 1:20 rinsed briefly in distilled water, mounted in Mowiol 4-88 a dilution; NFT, a mouse monoclonal antibody recognizing the (Hoechst) on glass slides, and examined with Leitz epi- three NF subunits (Fig. 2, lane b), kindly provided by D. fluorescence microscope equipped with rhodamine and fluorescein filter sets. Photographs were taken on Kodak T-Max Paulin (Institut Pasteur, Paris), used at a 1:400 dilution; RT97, 400 films. a mouse monoclonal antibody recognizing predominantly the heavy NF subunit (17) (Fig. 2, lane c), kindly provided by J. Wood (Sandoz Institute for Medical Research, London), RESULTS used at 1:500 dilution; NF200, a rabbit antiserum also rec- In initial experiments carried out with E4 and E5 optic stalks ognizing predominantly the heavy NF subunit (data not cultivated for 7 days, an extensive neuritic outgrowth was shown), generously provided by D. Paulin, used at a 1:500 visualized by phase-contrast microscopy, extending radially dilution; anti-glial fibrillary acidic protein, a rabbit antiserum, from the explant, in all 31 cultures examined. These neurites, purchased from Dakopatts, used a 1:100 dilution; and anti- as well as cell bodies and cell processes within the explant galactocerebroside, a rabbit antiserum, kindly provided by itself, were intensely decorated with RT97 antibody (Fig. 3). M. Sensenbrenner (Centre de Neurochimie du Centre Na- In addition, numerous nerve cell bodies extending long processes, specifically labeled with the anti-NF antibodies, a b c were found alongside the explant lying on a dense monolayer of cells that migrated out of the explant. They presented the morphology of bipolar or multipolar neurons and were either isolated or grouped in small aggregates. Their long thin 216- FIG. 2. Immunoblots using anti-NF neurites frequently displayed varicosities and extended over monoclonal antibodies. Triton-insoluble acellular regions to contact fibroblast-like cells. The neuritic were halo from the was transient and could not 150- extracts of E18 chicken brains sub- emerging explant mitted to SDS/PAGE in reducing condi- be observed over long culture periods, due to the degener- tions and blotted onto nitrocellulose. In- ation of explant neurons. However, some of the neurons dividual lanes were treated with mono- localized away from the explant could still be observed after clonal antibodies and subsequently with longer culture periods-e.g., 16 days (Fig. 4)-and appeared a peroxidase-conjugated goat anti-mouse more differentiated than at 7 days in vitro. immunoglobulin antiserum. Reactions At these there was some 73- were visualized using 4-chloro-1-naph- early embryonic stages, however, neurons. thol and H202. Lanes: a, NF70; b, NFT; doubt regarding the origin of the observed Indeed, c, RT97. Numbers indicate apparent mo- at 4 or 5 days of incubation, the optic stalk was not anatom- lecular masses (kDa) of the three NF ically well defined and could contain cells that later in protein subunits in chicken. development would be integral parts ofthe diencephalon, due Downloaded by guest on September 24, 2021 Developmental Biology: Giess et al. Proc. NatL. Acad. Sci. USA 87 (1990) 1645

FIG. 3. E4 optic stalk explant cultured for 7 days. Immunofluorescence staining by RT97. Staining is found in the explant from which large bundles of fluorescent neurites emerge radially. (Bar = 50 Aim.) to possible morphogenetic movements. In contrast, at 7 days extensive as that at previous stages. In addition, fewer of incubation, the optic peduncle formed a well-defined neurons could be observed outside the explant. Again, a large anatomical entity and could be easily dissected. Thus, ex- number ofneurons started degenerating after about a week in plants as small as possible were removed at this stage, taking culture (see Fig. 5). care to dissect tissue only from the medial part of the optic It was then of interest to determine the stage to which stalk. After a week in vitro, all 65 cultures stained with the neurons could develop in vitro. Medial parts of optic stalks various NF antibodies had NF-positive neurons (Fig. 5) were dissected from Eli, E15, and E18 embryos, dissoci- inside and outside the explant, indicating that neurons indeed ated, and cultured. After 7 days in vitro, double-labeling arise from the optic stalk and not from contaminating pre- experiments using the polyclonal antibody NF200 plus one of sumptive neural tissue. However, it must be pointed out that the monoclonal antibodies, RT97, NFT, or NF70, unambig- the radial neurite outgrowth observed was not nearly as uously demonstrated the presence of neuronal cells in all 31

FIG. 4. ES optic stalk explant cultured for 16 days. Immunofluorescence staining by NFT. Cell bodies (arrowheads) and their processes are stained. These neurons are found at some distance of the explant itself, lying on a dense layer of nonneuronal cells. Note that some processes present varicosities (arrows). (Bar = 50 Am.) Downloaded by guest on September 24, 2021 1646 Developmental Biology: Giess et al. Proc. Natl. Acad. Sci. USA 87 (1990) contrast and short ramifying processes that were not stained by any ofthe other NF markers. The nature ofthese cells has not been yet identified. In fact, these cells were present in small numbers in cultures of Eli optic stalks and became progressively more abundant at later developmental stages. To determine whether our culture conditions supported differentiation of other cell types, anti-glial fibrillary acidic protein and anti-galactocerebroside antibodies were used to label specifically astrocytes (20, 21) and oligodendrocytes (22). Glial fibrillary acidic protein immunoreactivity was present in E15 optic stalks cultured for 5 days in cells with a typical astrocytic morphology. Similarly, galactocerebroside immunoreactive cells with small rounded cell bodies and highly anastomosed processes were visualized in 7-day cultures of Eli optic stalks, indicating that oligodendrocytes also differentiated in our cultures. None of these antibodies stained neuronal cells. FIG. 5. E7 optic stalk explant cultured for 7 days. Immunofluo- rescence staining with NF200. Several labeled cell bodies are isolated (arrowheads) or grouped in a small aggregate (arrow) from DISCUSSION which several long thin neurites emerge. A small aggregate of degenerating neurons (double arrow) is also observed. (Bar = 50 Our aim was to investigate the developmental potential of Am.) cells in the embryonic chicken optic stalk and in particular to determine whether there were cells with the potential to cultures of Eli and E15 optic stalks. In these cultures of develop into neurons in this CNS tissue. We demonstrate that dissociated cells, neurons appeared isolated (Fig. 6) or in a during almost all stages of embryonic development, the few (two or three) small aggregates from which neuritic primitive anlage of the optic nerve can give rise in culture to networks emerged. Neurons were generally fewer than at subpopulations of nerve cells, in addition to the well- previous stages, and their numbers varied largely from one characterized glial cell populations, astrocytes and oligoden- culture to another (from two or three to a few dozen neurons). drocytes. In marked contrast, cultures of E18 optic stalks did not have The identification of neurons was based on immunocyto- readily identifiable neurons, irrespective of the time in cul- chemical criteria, using the NF triplet proteins as the now ture (up to 15 days). However, NFT antibodies stained a classical and highly specific neuronal markers (for reviews, subpopulation of cells with very dark cell bodies in phase- see refs. 23-25 and see also ref. 26). As demonstrated by Western-blot analysis, the antibodies used specifically rec- ognize either all three NF subunits (NFT) or only one subunit-the low molecular weight subunit, NFL (NF70), or the high molecular weight subunit, NFH (RT97 and NF200). In double-labeling experiments, there was a perfect correla- tion in the visualization of the various NFs within the same cells, leaving no doubt about the presence of true neuronal- specific intermediate filaments. Neurons appeared early, within the first days of culture and acquired with time a certain degree of morphological maturity. The neuronal population did not appear morphologically homogeneous, and neuronal cell bodies and neuritic extensions varied in shape and size, suggesting that various neuronal subpopulations emerged. No attempts were made to further characterize neuronal types. Extensive neuronal death was observed within the first 7-10 days of culture, possibly indicating that the culture environment lacked proper survival factors. To further define the nature of the neurons formed in our cultures, culture conditions should be devised that allow long-term neuronal survival. It is unclear whether neuronal potentialities in optic stalks are present over the entire period ofembryonic development. Cultures of E18 optic stalks contained only a subpopulation of atypical cells that extended short processes and were labeled with only one NF antibody (NFT). Further experiments are required to ascertain the nature ofthese cells. Thus we cannot determine the persistence or loss of neuronal potentialities at late embryonic stages. The potential of the optic stalk for neuron generation does not appear to be a particularity of chickens. Juurlink and Fedoroff (15) showed neuronal differentiation in explant cultures of E10 to E11.5 mouse optic stalk. This phenomenon FIG. 6. E15 dissociated optic stalk cells cultured for 7 days. was transient and appeared to be precisely restricted to the Double labeling with NF200 (Upper) and NF70 (Lower). Two period of initial extension of neural retinal axons along the rounded cell bodies and their processes are labeled in a similar optic stalk. However, the validity of these results can be fashion by the two antibodies, despite the less-intense reaction in cell questioned, since at these early stages, the boundary between processes given by NF70. (Bar = 50 ,um.) the optic stalk and eye cup or diencephalon is not clear-cut Downloaded by guest on September 24, 2021 Developmental Biology: Giess et al. Proc. Natl. Acad. Sci. USA 87 (1990) 1647

(see ref. 27). On the other hand, a report describes (28) the neural crest origin forming the presumptive glial cell popu- differentiation of nerve cells in long-term cultures (3-4 lation ofthe ganglion differentiate into neurons when allowed weeks) of newborn rat optic nerve. These results are in to migrate away in the peripheral structures of a younger contrast with the extensive studies of Raff and his colleagues chicken host. Accordingly, differentiated neurons and their (8, 29), who never showed neuronal differentiation in cultures axons could provide local cues repressing neuronal potenti- of perinatal rat optic nerve. More precisely, Abney et al. (29) alities in precursor cells. These undifferentiated cells could have shown that cultures of embryonic (E17-E18) optic be bipotential neuron-glia progenitors or, alternatively, de- stalks were always devoid of neurons. In the same work, termined neuronal precursor cells destined to die in situ and however, they mentioned that cultures of optic nerves be rescued, at least for some time, by the culture conditions. younger than E17 always contained NF-bearing neurons but This working hypothesis should be tested in our model by concluded to a possible contamination of their cultures by using retinal ablation experiments in vivo and cocultures of adjacent cerebral tissue. This apparent discrepancy in these optic stalk cells with various types of differentiated neurons. results could also originate from the different culture conditions used. Preliminary experiments in our laboratory indi- We thank Drs. D. Paulin, J. Wood, and M. Sensenbrenner for their cate that neurons in vitro from generous gifts of antibodies. The expert technical assistance of Mrs. develop rapidly dissociated F. Foulquier and M. J. Guinaudy is also gratefully acknowledged. cells of E17 rat optic nerves (M.-C.G. and P.C., unpublished This work was supported by the Centre National de la Recherche results). Therefore, neuronal potentialities in rat and chicken Scientifique and the Ministbre de l'Education Nationale. embryos do not seem to be restricted to early developmental stages, as noted in the mouse embryo (15). 1. Levitt, P., Cooper, M. L. & Rakic, P. (1981) J. Neurosci. 1, 27-39. the various the ofthese 2. Turner, D. L. & Cepko, C. L. (1987) Nature (London) 328, 131-136. Among hypotheses regarding origin 3. Price, J., Turner, D. & Cepko, C. (1987) Proc. Nati. Acad. Sci. USA 84, neurons, we can exclude the possibility that they arise from 156-160. other contaminating neural tissues: optic stalks were care- 4. Bartlett, P. F., Reid, H. H., Bailey, K. A. & Bernard, 0. (1988) Proc. fully dissected and always stripped ofextraconnective tissue, Natl. Acad. Sci. USA 85, 3255-3259. if there was left. At E4 and the between 5. Luskin, M. B., Pearlman, A. L. & Sanes, J. R. (1988) Neuron 1, 635- any E5, boundary 647. optic stalk and diencephalon was not well defined. Thus, as 6. Price, J. & Thurlow, L. (1988) Development 104, 473-482. discussed above, presumptive cerebral tissue or retinal pig- 7. Holt, C. E., Bertsch, T. W., Ellis, H. M. & Harris, W. A. (1988) Neuron mented epithelium, which is known to form neurons when 1, 15-26. from the retina could have been included 8. Raff, M. C., Miller, R. H. & Noble, M. (1983) Nature (London) 303, separated (30, 31), 390-395. in the explant. This was not the case at later stages, especially 9. Raff, M. C., Abney, E. R., Cohen, J., Lindsay, R. & Noble, M. (1983) since we always took care to dissect only the most medial part J. Neurosci. 3, 1289-1300. of the optic stalk. Therefore, we are confident that the sole 10. Raff, M. C., Abney, E. R. & Miller, R. H. (1984) Dev. Biol. 106, 53-60. source of neurons in our cultures is the stalk itself. 11. Miller, R. H., David, S., Patel, R., Abney, E. R. & Raff, M. C. (1985) optic Dev. Biol. 111, 35-41. Consequently the following two possibilities remain. 12. Temple, S. & Raff, M. C. (1985) Nature (London) 313, 223-225. (i) Neurons preexist in the embryonic optic nerve. It is 13. ffrench-Constant, C. & Raff, M. C. (1986) Nature (London) 319,499-502. generally agreed that the adult optic nerve is devoid of nerve 14. Small, R. K., Riddle, P. & Noble, M. (1987) Nature (London) 328, cell bodies. In several authors have described the 155-157. addition, 15. Juurlink, B. H. J. & Fedoroff, S. (1980) Dev. Biol. 78, 215-221. fine structure of the developing optic nerve (32-36), but, to 16. Hamburger, V. & Hamilton, H. L. (1951) J. Morphol. 88, 49-92. our knowledge, no one has ever reported the presence of 17. Anderton, B. H., Breinburg, D., Downes, M. J., Green, P. J., Tomlin- nerve cell bodies. Nevertheless, there is a slight possibility son, B. E., Ulrich, J., Wood, J. N. & Kahn, J. (1982) Nature (London) that if neurons and in small 298, 84-86. appear transiently numbers, they 18. Laemmli, U. K. (1970) Nature (London) 227, 680-685. could have been missed in these studies. 19. Towbin, H., Staehelin, T. & Gordon, J. (1979) Proc. Natl. Acad. Sci. (ii) A more likely hypothesis is that the neurons seen in USA 76, 4350-4354. culture arise from precursor cells that would not differentiate 20. Antinatus, D. S., Choi, B. H. & Lapham, L. W. (1975) Brain Res. 89, the neuronal in situ or Conse- 363-367. along lineage degenerate. 21. Raff, M. C., Fields, K. L., Hakomori, S.-I., Mirsky, R., Pruss, R. M. & quently, our culture conditions would favor the expression of Winter, J. (1979) Brain Res. 174, 283-308. a neuronal phenotype. One of the main differences between 22. Raff, M. C., Mirsky, R., Fields, K. L., Lisak, R. P., Dorfman, S. H., the in vivo and in vitro conditions is evidently the presence or Silberberg, D. H., Gregson, N. A., Liebowitz, S. & Kennedy, M. (1978) absence of axons from retinal cells. In chicken Nature (London) 274, 813-816. ganglion 23. Lazarides, E. (1980) Nature (London) 283, 249-256. embryos, the first recognizable retinal ganglion cell axons 24. Franke, W. W., Schmid, E., Schiller, D. L., Winter, S., Jarasch, E. D., appear in the optic fiber layer at about E3 (37). At E3.5, they Moll, R., Denk, H., Jackson, B. W. & lllmensee, K. (1982) Cold Spring have left the eye and travel in the ventral-most part of the Harbor Symp. Quant. Biol. 46, 431-454. optic stalk, entering the chiasm and diencephalic optic tract 25. Holtzer, H., Bennett, G. S., Tapscott, S. J., Croop, J. M. & Toyama, Y. at (1982) Cold Spring Harbor Symp. Quant. Biol. 46, 317-330. E4 (35). Thus optic stalk neuroepithelial cells have been in 26. Cochard, P. & Paulin, D. (1984) J. Neurosci. 4, 2080-2094. contact with retinal axons from the earliest stages. It is 27. Silver, J. & Robb, R. M. (1979) Dev. Biol. 68, 175-190. possible to speculate that the disappearance of axons, which 28. Omlin, F. X. & Waldmeyer, J. (1989) Dev. Biol. 133, 247-253. occurs after the stalk from the would be 29. Abney, E. R., Williams, B. P. & Raff, M. C. (1983) Dev. Biol. 100, severing optic retina, 166-171. in some way responsible for the appearance of neurons. Such 30. Orts-Llorca, F. & Genis-Galvez, J. M. (1960) Acta Anat. 42, 31-70. a speculation can be supported by a number of experiments. 31. Coulombre, J. L. & Coulombre, A. J. (1965) Dev. Biol. 12, 79-92. In the retina of amphibian larvae, selective ablation of 32. Vaughn, J. E. (1969) Z. Zellforsch. Mikrosk. Anat. 94, 293-324. neuronal leads to an 33. Skoff, R., Price, D. & Stocks, A. (1976) J. Comp. Neurol. 169, 291-312. specific populations overproduction of 34. the same Skoff, R., Price, D. & Stocks, A. (1976) J. Comp. Neurol. 169, 313-333. neuronal types by retinal precursor cells (38, 39). 35. Silver, J. & Sapiro, J. (1981) J. Comp. Neurol. 202, 521-538. This suggests a mechanism of local feedback control of 36. Silver, J. & Rutishauser, U. (1984) Dev. Biol. 106, 485-499. neuronal proliferation and differentiation by already differ- 37. Krayanek, S. & Goldberg, S. (1981) Dev. Biol. 84, 41-50. entiated neurons. In birds, using back-transplantation exper- 38. Reh, T. A. & Tully, T. (1986) Dev. Biol. 114, 463-469. iments of 39. Reh, T. A. (1987) J. Neurosci. 7, 3317-3324. quail-chicken chimeric nodose ganglia, Ayer-Le 40. Ayer-Le Lievre, C. S. & Le Douarin, N. M. (1982) Dev. Biol. 94, Lievre and Le Douarin (40) have demonstrated that cells of 291-310. Downloaded by guest on September 24, 2021