Neural crest (NC) cells from the periorbital mesenchyme (POM) invade the acellular stroma of the chick cornea at stage 27 of development (∼6 days). The invading cells become collagen-producing fibroblasts while the NC cells remaining in the POM differentiate into a wide range of cell types, the most easily recognizable of which is the pigment-producing melanocyte. In this paper, we report observations on the differentiation in vitro of cells within and migrating from explants of corneal stroma and compare their behaviour with that of cells within and migrating from explants of the POM. In ∼70 % of cases, POM explants produced black, eumelanin pigmentation within 2–3 days in culture and gave rise to a mixed outgrowth of fibroblasts and melanoblasts that produced brown pigment. In no case, however, did a corneal explant produce black pigment (so demonstrating that any POM contamination was negligible). However, in 28 % of cultures from stage-27 and -28 corneas, some of the cells in the outgrowth contained brown pigment indistinguishable from that produced by the POM control, although the majority of the cells in each case were fibroblasts. Two lines of investigation demonstrated that this pigment was melanin: first, transmission electron microscopy showed that the pigment organelles were incompletely melanized, granular melanosomes; second, tests designed to demonstrate the presence of lipofuscin, an alternative pigment, proved negative. Migrating cells from older corneas, in contrast, showed no evidence of even the first stages of melanogenesis. These results show, first, that some of the NC cells that invade the cornea are at least bipotent and hence representative of the POM population rather than being a unique subgroup and, second, that the acellular stroma of the cornea determines the state of differentiation of the NC cells that colonize it. The results thus provide an unequivocal demonstration that extracellular matrix can induce postmigratory NC cells to differentiate into fibroblasts.
The problem of what controls the differentiation of postmigratory neural crest (NC) cells cannot in general be investigated experimentally because it is usually only possible to isolate pure populations of these cells before they leave the crest (e.g. Sieber-Blum & Cohen, 1980). The NC cells that invade the acellular chick corneal stroma (Fig. 1; for review, see Hay, 1980) provide an exception to this general rule as they all become fibroblasts. These cells come from the group of postmigratory NC cells in the periorbital mesenchyme (POM), a NC-derived transitional ectomesenchyme (Horstadius, 1950; Le Douarin, 1980) that surrounds the optic cup in early development. Transplants of quail neural crest into chick embryos (for review, see Le Douarin, 1980) have shown that those cells remaining in the POM later differentiate into melanocytes, muscle cells, nerve cells, fibroblasts, chondroblasts and osteoclasts (Noden, 1978a,b;Le Lièvre, 1978), while the cells from the POM that colonize the 4-day cornea become an endothelium (Bard, Hay & Meller, 1975). In contrast, the group of POM cells that colonizes the swollen primary stroma of the cornea (Toole & Trelstad, 1971) on the 6th day of development all become fibroblasts which deposit collagen on the existing stroma in vivo (Bard & Higginson, 1977) and have typical fibroblast morphology both in vivo and in vitro (Hay & Revel, 1969; Bard & Hay, 1975). As it is possible to culture fragments of corneal stroma immediately after the secondary migration and compare their differentiation with that of cultured POM fragments, the developing anterior eye is a particularly good system for investigating fibroblast differentiation in postmigratory NC cells.
The possible factors that can control the differentiation of the neural crest have recently been the subject of extensive investigation. It has now been shown that the fate of NC cells can be influenced by inducer tissues along the route of migration (Cohen, 1972; Norr, 1973; for review, see Bronner-Fraser & Cohen, 1981) and by direct cellular interactions (Nichols, Kaplan & Western, 1977; Bee & Thorogood, 1980; Hall, 1981). In addition, there is some evidence that the extracellular matrix can affect the differentiation of isolated premigratory NC cells. Thus Sieber-Blum, Sieber & Yamada (1981) showed that fibronectin and collagen could cause cultured NC cells to become neurones and induce adrenergic differentiation, while Newsome (1976) has shown that the matrix deposited by pigmented retinal cells in vitro will induce chondroblastic differentiation. The system on which we report here provides an opportunity to examine whether extracellular matrix in situ can direct the differentiation of postmigratory NC cells.
MATERIALS AND METHODS
Fertile eggs of randomly-mated, J-line Brown Leghorn chicks were obtained from the Poultry Research Centre, Roslin, and incubated in a 38°C humidified incubator. Embryos were staged according to the criteria of Hamilton & Hamburger (1957), supplemented by the observations that the retinal corona around the cornea darkens at stage 28 and the first folds of the ciliary body are visible at stage 29 (Bard & Ross, 1982).
The anterior parts of eyes from stage-27 to -36 embryos were carefully stripped of lens, vitreous and pigmented retinal epithelium (PRE), care being taken to ensure that there was no contamination by any pigmented material. Isolated anterior eyes were placed in a dissociation mixture for 6–8 min at 20°C (stock solution: 250mg pancreatin (Sigma, U.K.) and 150mg trypsin (Difco, U.K.) were dissolved in 10ml distilled water, 10 ml of calcium- and magnesium-free Tyrodes solution were then added and the solution sterilized by filtration, see Bee & Thorogood, 1980). The corneal and conjunctival epithelium was then removed with a bent microscalpel made from a disposable 21-gauge hypodermic needle ground with a hand-held rotary polishing wheel (Identoflex, Switzerland).
Squares of either POM or cornea were then excised with a knife made from fragments of a brittle steel razor blade (Gillette Blue, courtesy of Gillette Ltd., London). Great care was taken to ensure that the cornea was not contaminated with POM or vice versa (Fig. 2). The corneal fragments thus contained undifferentiated NC cells from the POM, endothelial cells, collagen, hyaluronic acid, chondroitin sulphate and perhaps other glycoproteins (see Hay, 1980). POM fragments contained NC-derived mesenchyme in an extracellular matrix. To check that corneal sample squares were contaminated by neither anterior epithelium nor pigmented retinal epithelium, two squares were examined in the scanning electron microscope (SEM) and two cultures in the transmission electron microscope (TEM): there was no evidence of either epithelial morphology in the sample squares or junctional complexes between cultured, melanincontaining cells as might have been expected had PRE been present.
Heart fibroblasts were obtained by removing the heart from 5- or 6-day-old embryos, tearing away the pericardium and mincing the remaining tissue with two scalpel blades to give fine fragments for culturing. Fibroblasts from 18-day corneas were obtained from collagenase-treated fragments.
Expiants were transferred in fine capillary tubes to a 30 mm non-vented Petri dish (Nunclon) containing a thin film of culture medium and allowed to adhere for 1–4 h at 37 °C, 1–2 ml of medium were then added to the dish. Dishes were sealed with sterilized silicone high vacuum grease (Edwards, U.K.) to maintain a humid atmosphere within the culture dish and so prevent evaporation of the medium. Expiants were cultured in Ham’s F10 medium supplemented with 1 mM-glutamine, lOmM-sodium bicarbonate, 20mM-HEPES, buffer, 100 μg/ml penicillin, 100 μg / ml streptromycin, 10 % foetal calf serum (Gibco, U.K.) and 2 % chick-embryo extract prepared according to the method of Sieber-Blum & Cohen (1980).
Cultures were examined and photographed on a Wild M 40 inverted phase microscope. In order to visualize the pale brown melanosomes, specimens were obliquely illuminated by placing a mask across half of the condenser; this illumination gave an interference-contrast-like effect which allowed unpigmented cells to be distinguished from those with pale pigmentation. Cultures for TEM were routinely fixed and processed (Bard, 1979), sections were cut on an LKB Ultratome V microtome, stained with uranyl acetate and lead citrate and viewed on Philips 300 TEM.
Tyrosinase (dopa oxidase)
Cultures were fixed for 4 h in 5 % gluteraldehyde in 0·1 M-sodium cacodylate buffer of pH 7·3, and then washed six times in cacodylate-buffered 10 % sucrose. Where appropriate, specimens were incubated in 1 mg/ml L-dopa (Sigma) in cacodylate buffer for 4 h at 37 °C. Tyrosinase activity resulted in the formation of dark-coloured, electron-dense dopa melanin at the sites where the enzyme was present (Hunter, Patterson & Fairley, 1978). Control cultures were incubated in the same way with D-dopa (Sigma), in order to distinguish autoxidation of dopa from that catalysed by the enzyme. Cultures were then rewashed in 10 % buffered sucrose, and if required, postfixed in osmium tetroxide for transmission electron microscopy.
Melanin detection methods
Two methods were used to establish that melanin, rather than a neuronal lipofuscin (Pearse, 1960), was being made by cultured cells.
Lack of autofluorescence
Cells were examined live or after fixation in 4 % formaldehydecacodylate buffer with a Leitz Ortholux II microscope with epifluorescent illumination. Excitation was achieved with a Leitz KP500 or a BG12 filter in conjunction with a KP470 filter, and suppression with a TK 510/K 515 barrier filter. This configuration was used in order to encompass a broad spectrum of emitted light.
Bleaching and Nile blue sulphate staining
Hydrogen peroxide and potassium permanganate were used to bleach cultures air dried from water or ethanol (Pearse, 1960). In addition, some cultures were stained with 0·5 % Nile blue sulphate in distilled water and differentiated by incubation in ethanol for 15 min. This caused pigmented cells to stain dark green and the blue background of the fibroblasts to be removed. The specimen was then bleached with 10 % hydrogen peroxide for 24 h. Melanin, unlike lipofuscin, does not retain stain after such bleaching.
Periorbital mesenchyme cultures
Periorbital mesenchyme fragments taken from stage-25 to -32 embryos usually adhered to the culture dish and cells migrated from them to give radial outgrowths; the explant remained discrete in all but a few cases where it dissociated after a few days.
Close examination of the initial explants confirmed that they contained no pigment. Within 2–3 days, however, ∼70 % of them contained black melanocytes (Table 1) which, in section, could be seen throughout the explant. Cells started to migrate from the explants within 12 h. Two days later, the outgrowths comprised mainly cells with typically fibroblastic morphology (Fig. 3), but, in more than 90% of cultures where outgrowths formed, there were a minority of stellate cells with brown pigmentation which had differentiated after migration from the explant. Moreover, in all but one of the 121 explants which developed pigmentation, pigmented cells were present in the corresponding outgrowths (Table 1). The states of pigmentation seemed stable for, after 2 weeks of culture, the explant remained black, while peripheral cells had pale pigmentation (Fig. 4).
Corneal stroma cultures
Corneal explants from stage-27 to -28 eyes produced outgrowths within 24 h in vitro in about 30 % of cases. The remainder had few or no outgrowing cells around them even after 6 days of culture; this was probably because of the known preference for cells to adhere to collagen rather than plastic (Elsdale & Bard, 1972). The explants differed markedly from those of the POM for in none of the 126 explants examined was black pigmentation present (Table 1, Fig. 5), even after 2 weeks of culture. One explant did appear brown after one week in vitro, but no discrete melanocytes could be seen and TEM showed the explant to be necrotic. These observations confirm that the corneal explants had not been significantly contaminated by POM even though they had been grown in an environment which had been shown to support the development of pigmentation.
The type of cells that migrated from the explants depended on the age of the embryo from which the explant had been taken (Table 1). In all stage-29 to -36 fragments which gave explants, the cells were totally lacking in pigmentation and had typical fibroblastic morphology. Outgrowths from stage-27 to -28 explants were, however, of two types: in ∼70 % of cases, the cells were indistinguishable from the fibroblast-like cells of the stage-29 to -36 cultures. In the remaining ∼30 % of cultures, the outgrowths comprised a mixture of fibroblasts and cells with brown pigmentation (Fig. 6). In one case, there was, in addition to pigmented and fibroblastic cells, a small patch of unpigmented, epithelial cells which migrated from the explant after 2 days in vitro and then rolled up after 4 days; this was probably an endothelial outgrowth.
As the observation that pigmented cells could come from stage-27 to -28 (6-day) corneal explants was novel, we examined two further sets of cultures. The first was an age-matched culture of heart fibroblasts (non-NC), the second was cultures and subcultures of 18-day embryonic corneal fibroblasts; in no case was a single pigmented cell seen within either explants or outgrowths. Thus non-NC fibroblasts do not become pigmented in our culture conditions and differentiated NC cells do not transdifferentiate.
The morphology of the corneal cells with brown pigment was similar to the corresponding cells in POM cultures, but differed from the fibroblasts which lacked pigment. While the latter were flattened, had ruffling membranes and were discrete, many of the former had blunt-ended processes and over-lapped; furthermore, they tended to associate in small clumps (Fig. 6). Because of this clumping, it was not possible to count the number of pigmented cells in the cultures. Moreover, had such counts been possible, their significance would have been negligible as pigmented and fibroblastic NC cells have different growth rates (Maxwell, 1976).
Transmission electron microscopy
The cells that had migrated from the cornea and become pigmented contained a large number of melanosomes (Fig. 7). At higher power (Fig. 8), these melanosomes were seen to be incompletely melanized: they were round, often granular and sometimes contained poorly formed filamentous arrays on which melanin was deposited. The morphology of the melanosomes, even after three weeks in vitro, was thus similar to stages I and II of melanogenesis (Jimbow, Oikawa, Sugiyama & Takeuchi, 1979). TEM confirmed that the ultrastructure of the melanosomes in the light-pigmented POM cells was identical to that of the pigmented corneal cells. POM explants, in contrast, contained elongated mature melanosomes (Fig. 9): these were either filamentous arrays on which was deposited melanin or more melanized organelles with totally electron-opaque deposits (corresponding to stages III and IV of melanogenesis; Jimbow et al., 1979). There was, however, no evidence of melanosomes within the corneal explants. (A more detailed description of the melanogenesis of POM cells and their cultured NC precursors will be reported elsewhere).
The ultrastructural evidence that the pigmented corneal cells contained melanin was supplemented by histochemical tests to eliminate the possibility of the presence of an alternative pigment, lipofuscin, that can be confused with melanin (Crichton, Bassuttil & Price, 1980; Pearse, 1960). Pigmented corneal cultures were therefore tested for staining with Nile blue sulphate or L-dopa, autofluorescence and bleachability. In all cases, the cultures gave the positive result for melanin and the negative result for lipofuscin.
The melanin-forming potential (dopa oxidase activity) of migrating unpigmented cells from corneal explants was examined after 6 days in vitro for evidence of an alternative to fibroblast differentiation in cells which had spent only a few hours within the cornea. For this, the cells were tested for the presence of tyrosinase, a marker for a melanin-synthesizing capacity (Hunter, et al., 1978). Six stage-28 to -29 cultures were therefore incubated with L-dopa and four controls with D-dopa; if there were cells containing tyrosinase, the former would have stained dark while the latter would not. Careful examination of the cultures after treatment revealed no dark staining in any cells of fibroblastic morphology. These cells do not therefore display a mixed phenotype.
The observations presented here show that some of the NC cells that invade the chick cornea at stage 27 of development have the potential to become melanocytes, but that they soon lose this ability. The possibility that these melanocytes result from contamination of the corneal cultures by fragments of POM is ruled out by the observation that migrating cells become pigmented, while those that remain in the corneal explant do not; a result that contrasts with the observation that POM explants become pigmented in vitro. Moreover, the fact that the corneal NC cells can differentiate into both melanocytes and fibroblasts shows that the invading cells are pluripotent and do not comprise a unique group of predetermined cells.
The behaviour of the POM cultures, in addition to acting as a control for the corneal cells, leads to two conclusions. First, four days after the NC cells reach the environs of the optic cup, only some can differentiate into melanocytes in vitro. Second, potential melanocytes produce more mature melanosomes within the POM environment than away from it. The NC cells that invade the cornea, in contrast, can produce melanin only if they escape the corneal environment soon after entry. The evidence thus shows unequivocally that the corneal environment inhibits melanogenesis, a conclusion that might be expected as the corneal environment is not, for obvious reasons, one where pigmentation is likely to occur.
It is worth noting that this apparently irreversible inhibition of melanogenesis is complete within a very few hours of the cells entering the corneal environment. This observation tallies with the observations of Hay & Revel (1969) who showed that the colonizing cells started to increase their rough endoplasmic reticulum and Golgi apparatus (and thus differentiate into collagen-producing fibroblasts) between stages 29 and 30, or within ∼6h of entering the cornea. The mechanism by which this inhibition is achieved remains unclear. It might at first thought be supposed that the endothelium induced this change through inductive messages transported to adjacent colonizing cells and then from one cell to another through fine-filopodial contacts. Such contacts are known to occur in the cornea between one NC cell and another (Hay & Revel, 1969). However, close examination of electron micrographs showed that very few, if any, of the colonizing cells seemed to make contact with the endothelium. It thus seems that non-cellular stromal constituents are responsible for the inhibition of melanogenesis.
The number of candidates for controlling the differentiation of the NC cells that invade the corneal stroma at stage-27 to -28 is limited. The environment in which the cells find themselves is a connective tissue known to contain collagen of types I and II (von der Mark, von der Mark, Timpl & Trelstad, 1977) and the proteoglycans chondroitin sulphate and heparan sulphate all of which are produced by the epithelium (Hart, 1978; Hay & Dodson, 1973), and hyaluronic acid produced by the endothelium (Toole & Trelstad, 1971); but there is no reason to suppose that this list is complete. If macromolecular matrix components made by either the corneal epithelium or endothelium are responsible for inhibiting melanogenesis, the former is clearly the more likely source and it is significant, in this context, that collagen type II has recently been implicated in neural crest chondrogenesis (Thorogood & Smith, 1984).
There is also the possibility that some control over melanogenesis in the anterior part of the cornea may be achieved by the physical constraint of cellular migration. Close examination of 13-day (stage-39) corneas showed that melanocytes appeared about then within the epithelium and subjacent stroma of the limbus (or periphery) of the cornea. Indeed, it is known that NC cells invading the epithelia of pigmented vertebrates will themselves become pigmented (Teillet, 1971). In the cornea, however, there is a region of dense, unswollen collagen stroma, Bowman’s zone, subjacent to the epithelium. Migrating cells do not, and probably cannot, invade this region (Hay & Revel, 1969). The corneal epithelium adjacent to the limbus may thus be physically protected from neural crest invasion by Bowman’s zone.
The corneal endothelium, subjacent to the stroma, is another NC tissue whose differentiation may be affected by extracellular matrix. At 4-days of development, a group of NC cells from the POM migrates between the posterior stroma of the cornea and the subjacent lens capsule (Bard, et al., 1975). Both of these are thick (∼5 μm) extracellular laminae, but with very different molecular compositions. As only those cells which come in contact with these laminae seem to differentiate into endothelial cells, it is likely that corneal endothelial differentiation is controlled by the extracellular matrix.
In the more general context of the control of NC cell differentiation, the results presented here provide clear and unequivocal evidence that a target tissue may control the final cell type in a way that cannot depend on contact with target cells. In organs other than the cornea, the initial position in the crest, the route of migration or target cell contacts may influence differentiation. The results presented here show that there can be non-cellular influences that will inhibit melanogenesis and stimulate the formation of fibroblasts.
We thank Duncan Davidson and Tom Elsdale for commenting on the manuscript, Allyson Ross for technical assistance and Sandy Bruce for help in preparing the plates. S.C. was supported by an M.R.C. studentship during the course of this work.