The phenotypic response of cultured quail trunk neural crest cells to a reconstituted basement membrane-like matrix is specific

Previous work has demonstrated that a reconstituted basement membrane (RBM)-like matrix stimulates the development of catecholamine (CA)-containing cells in neural crest cultures. In the present work, we found that the proportion of tyrosine hydroxylase and somatostatin immunoreactive cells was increased substantially by an overlay of the RBM matrix. In contrast, there was little or no stimulation of the development of cells possessing several other phenotypic markers including A2B5, E/C8, vasoactive intestinal polypeptide, and the low and middle molecular weight avian neurofilament proteins. These results demonstrate that the response of neural crest cells to the RBM matrix is specific to a small set of phenotypes. In addition, we demonstrate that the phenotype of the adrenergic cells which develop in the presence of the RBM gel overlay is very similar, if not identical, to that of the adrenergic cells which differentiate in the absence of the RBM gel.     

Analysis of the development of cellular subsets present in the neural crest using cell sorting and cell culture

We have tested the hypothesis that developmentally significant cellular subsets are present in the early stages of neural crest ontogenesis. Cultured quail trunk neural crest cells probed with the monoclonal antibodies HNK-1 and R24 exhibited heterogeneous staining patterns. Fluorescence-activated cell sorting was used to isolate the HNK-1+ and HNK-1- cell populations at 2 days in vitro. When these cell populations were cultured, the HNK-1+ sorted cells differentiated into melanocytes, unpigmented cells, and numerous catecholamine-positive (CA+) cells. In contrast, the HNK-1- sorted cells gave rise to melanocytes and unpigmented cells, but few, if any, CA+ cells. When neural crest cells at 2 days in vitro were labeled with R24 and sorted, both the R24+ the R24- sorted cell populations produced numerous CA+ cell, melanocytes, and unpigmented cells. These results provide evidence for the existence of developmental preferences in some subsets of neural crest cells early in embryogenesis.     

Sertoli cell glycosylation patterns as affected by culture age and extracellular matrix

This study evaluated the responsiveness of Sertoli cell glycosylation in vitro to changes in culture age and to the presence of a reconstituted basement membrane (Matrigel) or collagen IV/laminin substrata. Primary Sertoli cell cultures were prepared from 20-day-old rats and incubated with [3H]mannose, a monosaccharide specific for asparagine-linked oligosaccharides. The cells were harvested on Days 4, 6, or 10 of culture life. A supernatant enriched in cell-surface glycopeptides (the trypsinate) and a cell pellet stripped of surface glycoconjugates were evaluated separately. Glycopeptides derived from a Pronase digest of the two samples were fractionated using concanavalin-A lectin affinity chromatography into three major classes: multiantennary complex-type, biantennary complex-type, and high-mannose-type oligosaccharide structures. The proportion of radiolabeled glycopeptides appearing in each of the three classes did not differ between Days 4 and 6 of culture. In contrast, a significant increase in the percentage of radiolabeled glycopeptides containing multiantennary complex-type oligosaccharides was observed in cells harvested from the 10-day-old cultures. In other experiments, Sertoli cells were grown on various substrata: plastic; collagen IV/laminin; or Matrigel, a reconstituted basement membrane (RBM) composed of laminin, collagen IV, proteoglycan sulfate, entactin, and nidogen. Growth on RBM significantly increased multiantennary complex-type oligosaccharide formation compared to plastic, whereas the high-mannose-type glycopeptides increased in cells grown on collagen IV/laminin. These studies suggest that environmental and physiological conditions such as culture age and the presence of extracellular matrix significantly affect glycosylation patterns in Sertoli cell cultures.     

Clonal analysis of quail neural crest cells: They are pluripotent and differentiate in vitro in the absence of noncrest cells

To determine if neural crest cells are pluripotent and establish whether differentiation occurs in the absence of noncrest cells, a cell culture method was devised in which differentiation could be examined in clones derived from single, isolated neural crest cells. Single neural crest cells, which were isolated before the onset of in vivo migration, gave rise to three types of clones: pigmented, unpigmented, and mixed. Pigmented clones consisted of melanocytes only, whereas some unpigmented cells in mixed and unpigmented clones contained catecholamines, identifying them as adrenergic cells. Extracellular matrix derived from quail somite or chick skin fibroblast cultures stimulated adrenergic differentiation and axon formation. These results demonstrate for the first time the existence of pluripotent quail neural crest cells that give rise to at least two progeny, melanocytes and neuronal cells. They also suggest that continuous direct interactions with noncrest cells are not required for the differentiation of these two cell types. However, components of the extracellular matrix derived from noncrest cells may play an important role in expression of the adrenergic phenotype.     

Alpha 1 beta 1 integrin on neural crest cells recognizes some laminin substrata in a Ca(2+)-independent manner

Neural crest cells migrate along pathways containing laminin and other extracellular matrix molecules. In the present study, we functionally and biochemically identify an alpha 1 beta 1 integrin heterodimer which bears the HNK-1 epitope on neural crest cells. Using a quantitative cell adhesion assay, we find that this heterodimer mediates attachment to laminin substrata prepared in the presence of Ca2+. Interestingly, neural crest cells bind to laminin-Ca2+ substrata in the presence or absence of divalent cations in the cell attachment medium. In contrast, the attachment of neural crest cells to laminin substrata prepared in the presence of EDTA, heparin, Mg2+, or Mn2+ requires divalent cations. Interactions with these laminin substrata are mediated by a different integrin heterodimer, since antibodies against beta 1 but not alpha 1 integrins inhibit neural crest cell attachment. Thus, the type of laminin substratum appears to dictate the choice of laminin receptor used by neural crest cells. The laminin conformation is determined by the ratio of laminin to Ca2+, though incorporation of heparin during substratum polymerization alters the conformation even in the presence of Ca2+. Once polymerized, the substratum appears stable, not being altered by soaking in either EDTA or divalent cations. Our findings demonstrate: (a) that the alpha 1 beta 1 integrin can bind to some forms of laminin in the absence of soluble divalent cations; (b) that substratum preparation conditions alter the conformation of laminin such that plating laminin in the presence of Ca2+ and/or heparin modulates its configuration; and (c) that neural crest cells utilize different integrins to recognize different laminin conformations.     

Polarity reversal of inside-out thyroid follicles cultured on the surface of a reconstituted basement membrane matrix.

When cultured in suspension, epithelial thyroid cells organized into inside-out follicles. We studied the behavior of these structures after seeding on polystyrene, type I collagen, and reconstituted basement membrane (RBM) gel. When seeded on plastic, type I collagen or mixed type I collagen-RBM gel, inside-out follicles attached and spread, forming polarized cell monolayers. In contrast, on thick RBM gel, inside-out follicles attached penetrated into the gel, and reorganized into properly oriented follicular structures. Polarity of the cell layer was progressively inverted while, after adhesion, cells penetrated the soft RBM gel. In the process of reorientation, cells with hybrid polarity were observed. The fraction of the apical pole which was not yet in the gel showed an inside-out orientation, while a modified orientation was observed in contact with the RBM gel. Cells which had penetrated completely in the matrix formed a new apical pole and displayed an opposite orientation of their polarity. A continuous basement membrane was observed, lining the basal cell surface when native RBM gel was present in the substratum.     

Lectin binding to neural crest cells. Changes of the cell surface during differentiation in vitro

To examine possible changes in cell surface carbohydrates, fluorescent lectins were applied at various times during differentiation of neural crest cells in vitro. The pattern and intensity of binding of several lectins changed as the crest cells developed into melanocytes and adrenergic cells. Considerable amounts of concanavalin A (Con A) and wheat germ agglutinin (WGA) bound to all unpigmented cells throughout the culture period. Melanocytes, however, bound much less of these lectins. Soy bean agglutinin (SBA), unlike Con A and WGA, only bound later in development to unpigmented cells at about the time when catecholamines were detected histochemically. Binding of SBA could be induced in younger cultures by pretreating the cells with neuraminidase. Melanocytes, however, did not bind detectable amounts of SBA even if treated with neuraminidase. The SBA-binding sites were often concentrated on cytoplasmic extensions and on contact points between neighboring cells, even when receptor mobility was restricted by prefixation of the cells or adsorption of lectin at 0 degrees C. All three lectins bound to cell processes resembling nerve fibers in particularly high amounts.     

In vitro clonal analysis of quail cardiac neural crest development

The developmental potentials of cardiac neural crest cells were investigated by in vitro clonal analysis. Five morphologically distinct types of clones were observed: (1) "pigmented" clones contained melanocytes only; (2) "mixed" clones consisted of pigmented and unpigmented cells; (3) "unpigmented dense" clones consisted of flattened, closely aligned unpigmented cells; (4) "unpigmented loose" clones consisted of a few loosely arranged, flattened cells; and (5) "unpigmented large" clones included a large number of small, stellate cells that were highly proliferative. The binding patterns of antibodies against lineage-specific markers showed that cells in the different clones expressed characteristic phenotypes. The following phenotypes were expressed in addition to pigment cells: smooth muscle cells, connective tissue cells, chondrocytes, and cells in the sensory neuron lineage. Mixed clones expressed all five phenotypes. Unpigmented dense clones contained smooth muscle cells, connective tissue cells, chondrocytes, and sensory neurons. Unpigmented loose clones exclusively consisted of smooth muscle cells, whereas unpigmented large clones contained chondrocytes and sensory neuron precursors. Based on these results, the following conclusions can be drawn: (1) Pigmented and unpigmented loose clones are most likely formed by precursors that are committed to the melanogenic and myogenic cell lineages, respectively. (2) Mixed and unpigmented dense clones are derived from pluripotent cells with the capacity to give rise to four or five phenotypes. (3) Unpigmented large clones originate from progenitor cells that appear to have a partially restricted developmental potential, that is, these cells are capable of generating two phenotypes in clonal cultures. Thus, the data indicate that the early migratory cardiac neural crest is a heterogeneous population of cells, consisting of pluripotent cells, cells with a partially restricted developmental potential, and cells committed to a particular cell lineage.     

Cellular fibronectin promotes adrenergic differentiation of quail neural crest cells in vitro

We have investigated the interaction of cellular fibronectin (CFN) with cultured quail neural crest cells and its possible role in crest cell migration and differentiation. In vitro, quail neural crest cells from the trunk region differentiate into at least two morphologically recognizable cell types, melanocytes and adrenergic nerve cells. The latter often aggregate spontaneously into ganglia-like structures. We found that neither melanocytes nor adrenergic nerve cells synthesize CFN. However, both cell types readily interacted with exogenous CFN: Melanocytes removed CFN from the substratum and accumulated it in an aggegated form on their upper cell surface, whereas unpigmented cells migrated on the CFN substratum, often rearranging it into a fibrillar network. The adsorption of CFN by melanocytes was apparently without further consequences. However, catecholamine-positive cells were substantially increased after treatment with exogeneous fibronectin. The stimulation of adrenergic differentiation of neural crest cells is the first evidence for a positive regulatory role of fibronectin in differentiation.     

Distribution of laminin and collagens during avian neural crest development

The distribution of type I, III and IV collagens and laminin during neural crest development was studied by immunofluorescence labelling of early avian embryos. These components, except type III collagen, were present prior to both cephalic and trunk neural crest appearance. Type I collagen was widely distributed throughout the embryo in the basement membranes of epithelia as well as in the extracellular spaces associated with mesenchymes. Type IV collagen and laminin shared a common distribution primarily in the basal surfaces of epithelia and in close association with developing nerves and muscle. In striking contrast with the other collagens and laminin, type III collagen appeared secondarily during embryogenesis in a restricted pattern in connective tissues. The distribution and fate of laminin and type I and IV collagens could be correlated spatially and temporally with morphogenetic events during neural crest development. Type IV collagen and lamin disappeared from the basal surface of the neural tube at sites where neural crest cells were emerging. During the course of neural crest cell migration, type I collagen was particularly abundant along migratory pathways whereas type IV collagen and laminin were distributed in the basal surfaces of the epithelia lining these pathways but were rarely seen in large amounts among neural crest cells. In contrast, termination of neural crest cell migration and aggregation into ganglia were correlated in many cases with the loss of type I collagen and with the appearance of type IV collagen and laminin among the neural crest population. Type III collagen was not observed associated with neural crest cells during their development. These observations suggest that laminin and both type I and IV collagens may be involved with different functional specificities during neural crest ontogeny. (i) Type I collagen associated with fibronectins is a major component of the extracellular spaces of the young embryo. Together with other components, it may contribute to the three-dimensional organization and functions of the matrix during neural crest cell migration. (ii) Type III collagen is apparently not required for tissue remodelling and cell migration during early embryogenesis. (iii) Type IV collagen and laminin are important components of the basal surface of epithelia and their distribution is consistent with tissue remodelling that occurs during neural crest cell emigration and aggregation into ganglia.     

Neural crest migration in 3D extracellular matrix utilizes laminin, fibronectin, or collagen

The trunk neural crest originates by transformation of dorsal neuroepithelial cells into mesenchymal cells that migrate into embryonic interstices. Fibronectin (FN) is thought to be essential for the process, although other extracellular matrix (ECM) molecules are potentially important. We have examined the ability of three dimensional (3D) ECM to promote crest formation in vitro. Neural tubes from stage 12 chick embryos were suspended within gelling solutions of either basement membrane (BM) components or rat tail collagen, and the extent of crest outgrowth was measured after 22 hr. Fetal calf serum inhibits outgrowth in both gels and was not used unless specified. Neither BM gel nor collagen gel contains fibronectin. Extensive crest migration occurs into the BM gel, whereas outgrowth is less in rat tail collagen. Addition of fibronectin or embryo extract (EE), which is rich in fibronectin, does not increase the extent of neural crest outgrowth in BM, which is already maximal, but does stimulate migration into collagen gel. Removal of FN from EE with gelatin-Sepharose does not remove the ability of EE to stimulate migration. Endogenous FN is localized by immunofluorescence to the basal surface of cultured neural tubes, but is not seen in the proximity of migrating neural crest cells. Addition of the FN cell-binding hexapeptide GRGDSP does not affect migration into either the BM gel or the collagen gel with EE, although it does block spreading on FN-coated plastic. Thus, although crest cells appear to use exogenous fibronectin to migrate on planar substrata in vitro, they can interact with 3D collagenous matrices in the absence of exogenous or endogenous fibronectin. In BM gels, the laminin cell-binding peptide, YIGSR, completely inhibits migration of crest away from the neural tube, suggesting that laminin is the migratory substratum. Indeed, laminin as well as collagen and fibronectin is present in the embryonic ECM. Thus, it is possible that ECM molecules in addition to or instead of fibronectin may serve as migratory substrata for neural crest in vivo.     

Follicle-like structure and polarized monolayer: Role of the extracellular matrix on thyroid cell organization in primary culture

The thyroid follicle, the morphofunctional unit of thyroid gland, is a spheroidal structure formed by a monolayer of polarized cells surrounding a closed cavity in which thyroglobulin accumulates. Newly isolated porcine thyroid cells reorganize into two types of structures which differ by the orientation of cell polarity: in follicle-like structures, obtained in the presence of TSH, the apical pole delineates a closed cavity and cells express most parameters characteristic of thyroid function; in inside-out follicles the apical pole is oriented towards the culture medium and cells do not express properly the thyroid function. The organization of newly formed follicles can be modified by stimulation of cell migration or by interaction of their apical poles with a new cell environment. Seeded on a hard surface (glass, plastic), cells of follicle-like structures or inside-out follicles formed in suspension migrate giving a monolayer. On the contrary, cells organized into a monolayer treated with hexamethylene bisacetamide, reorganize into follicle-like structures. Inside-out structures reoganize upon interaction of their apical poles with collagen I gel, a coherent matrix, or with a reconstituted basement membrane (RBM), a soft matrix. Overlaid with collagen I, monolayers reorganize into follicles. Embedded in collagen I or in RBM, inside-out follicles reorient their polarity giving functional follicles. On the RBM surface, cells pull on the gel and embed themselves in the soft matrix gel, finally reorienting their polarity to inside-in polarity. When comparing thyroid cells with other epithelial cell types (mammary cells, Sertoli cells), it appears that the obtention in culture of follicle-like structures, ie closed inside-in polarized cell organization, is the best way to express in culture both morphology and function of any specific epithelial tissue, the polarized monolayer in porous bottom culture chamber coming just behind.     

Xenopus neural crest cell migration in an applied electrical field

Xenopus neural crest cells migrated toward the cathode in an applied electrical field of 10 mV/mm or greater. This behavior was observed in relatively isolated cells, as well as in groups of neural crest cells; however, the velocity of directed migration usually declined when a cell made close contact with other cells. Melanocytes with a full complement of evenly distributed melanosomes did not migrate of their own accord, but could be distorted and pulled by unpigmented neural crest cells. Incompletely differentiated melanocytes and melanocytes with aggregated melanosomes displayed the same behavior as undifferentiated neural crest cells, that is, migration toward the cathode. An electrical field of 10 mV/mm corresponded to a voltage drop of less than 1 mV across the diameter of each cell; the outer epithelium of Xenopus embryos drives an endogenous transembryonic current that may produce voltage gradients of nearly this magnitude within high-resistance regions of the embryo. We, therefore, propose that electrical current produced by the skin battery present in these embryos may act as a vector to guide neural crest migration.     

Reconstituted basement-membrane matrix modulates fibroblast activities in vitro

Cellular growth and collagen biosynthesis were compared in dermal calf fibroblasts cultured on plastic or on a reconstituted basement membrane gel, termed matrigel. This matrix, extracted from Engelbreth-Holm-Swarm tumors, consists mainly of laminin, entactin, type IV collagen, and heparan sulfate proteoglycan. The multiplication rate of fibroblasts grown on matrigel was stimulated compared to that of monolayered cells cultured on plastic, and these cells formed multilayers after 4 days. Protein and collagen biosynthesis was reduced in fibroblasts cultured on matrigel. A higher proportion of the newly synthesized collagen (40%) was incorporated to the extracellular matrix in cultures grown on matrigel than in those grown on plastic (14%). Type III collagen was the preferential collagen type deposited on matrigel, and the ratio of type III:type I collagens secreted in the medium was also slightly higher in cultures grown on matrigel. Partially processed collagen was more abundant in fibroblasts grown on matrigel than in cells cultured on plastic. Finally, cells grown on matrigel exhibited a higher catabolic activity than cells grown on plastic. In this experimental model, the reconstituted basement-membrane matrix seems to influence the activities of fibroblasts significantly.     

An antiglycolipid antibody inhibits Madin-Darby canine kidney cell adhesion to laminin and interferes with basolateral polarization and tight junction formation

Epithelial cells polarize not only in response to cell-cell contacts, but also to contacts with a substratum composed of extracellular matrix molecules. To probe the role of specific matrix constituents in epithelial cell polarization, we investigated the effects of an adhesion-blocking mAb, 12B12, on initial polarization of MDCK cells. The 12B12 antibody, raised against whole MDCK cells, blocks adhesion to laminin by 65% but has no effect on adhesion of cells to collagen type I. Taking advantage of this antibody's function-blocking activity, as well as the fact that MDCK cells secrete laminin, the role of endogenous laminin in polarization was examined by plating cells on collagen-coated substrata in the presence of the antibody. Under these conditions, cell spreading was reduced 1.5h after plating, and cells were flatter and had fewer microvilli after 24 h. Even though lateral cell membranes were closely apposed, transepithelial resistance in the presence of the antibody was significantly reduced relative to controls. When the polarization of specific apical and basolateral markers was examined both biochemically and immunocytochemically in the presence of the antibody, we observed that the apical marker polarized at normal rates while basolateral markers did not. Surprisingly, the 12B12 antibody was not directed against any known cell adhesion protein but reacted specifically with Forssman antigen, a glycosphingolipid. These results suggest that glycolipids may play a significant role in cell adhesion via laminin and in epithelial cell polarization.     

Tumor-Promoting Phorbol Esters Promote Melanogenesis and Prevent Expression of the Adrenergic Phenotype in Quail Neural Crest Cells

Tumor-promoting phorbol esters were used to manipulate the in vitro development of neural crest cells. When plated at clonal density in secondary culture, quail neural crest cells from the trunk region gave rise to three types of colonies, pigmented, unpigmented, and mixed. Pigmented colonies consisted exclusively of melanocytes; up to 50% of the unpigmented and mixed colonies contained adrenergic nerve cells which could be identified by a catecholamine-specific histofluorescence method. Addition of potent tumor promoters to the culture medium shortened the doubling time of neural crest cells and altered their morphologic appearance. It also delayed the onset of pigmentation, prevented the expression of the adrenergic phenotype, reduced the number of unpigmented and mixed colonies, and increased the number of pigmented colonies, most likely by directing progenitor cells preferentially to the melanogenic pathway. There was a clear correlation between the ability of phorbol esters to promote skin tumors in mice and their ability to interfere with the in vitro development of quail neural crest cells. The potent promoters 12–0–tetradecanoyl phorbol 13–acetate (TPA) and phorbol 12,13–didecanoate (PDD) were most effective, phorbol 12,13–diacetate (PDA) was considerably less effective, the nonpromoting analogues 4–0–methyl 12–0–tetradecanoyl phorbol 13–acetate (4–0–Me-TPA) and 4α-phorbol 12,13–didecanoate (4α-PDD) and the parent alcohol phorbol (PHR) had little or no effect.     

Cranial and trunk neural crest cells use different mechanisms for attachment to extracellular matrices.

We have used a quantitative cell attachment assay to compare the interactions of cranial and trunk neural crest cells with the extracellular matrix (ECM) molecules fibronectin, laminin and collagen types I and IV. Antibodies to the beta 1 subunit of integrin inhibited attachment under all conditions tested, suggesting that integrins mediate neural crest cell interactions with these ECM molecules. The HNK-1 antibody against a surface carbohydrate epitope under certain conditions inhibited both cranial and trunk neural crest cell attachment to laminin, but not to fibronectin. An antiserum to alpha 1 intergrin inhibited attachment of trunk, but not cranial, neural crest cells to laminin and collagen type I, though interactions with fibronectin or collagen type IV were unaffected. The surface properties of trunk and cranial neural crest cells differed in several ways. First, trunk neural crest cells attached to collagen types I and IV, but cranial neural crest cells did not. Second, their divalent cation requirements for attachment to ECM molecules differed. For fibronectin substrata, trunk neural crest cells required divalent cations for attachment, whereas cranial neural crest cells bound in the absence of divalent cations. However, cranial neural crest cells lost this cation-independent attachment after a few days of culture. For laminin substrata, trunk cells used two integrins, one divalent cation-dependent and the other divalent cation-independent (Lallier, T. E. and Bronner-Fraser, M. (1991) Development 113, 1069-1081). In contrast, cranial neural crest cells attached to laminin using a single, divalent cation-dependent receptor system. Immunoprecipitations and immunoblots of surface labelled neural crest cells with HNK-1, alpha 1 integrin and beta 1 integrin antibodies suggest that cranial and trunk neural crest cells possess biochemically distinct integrins. Our results demonstrate that cranial and trunk cells differ in their mechanisms of adhesion to selected ECM components, suggesting that they are non-overlapping populations of cells with regard to their adhesive properties.     

Effect of Collagen Gel Configuration on the Cytoskeleton in Cultured Rat Hepatocytes

The cytoskeleton is important in the maintenance of cellular morphology and differentiated function in a number of cell types, including hepatocytes. In this study, adult rat hepatocytes sandwiched between two layers of collagen gel were compared to cells cultured on a single collagen gel for differences in the organization and expression of the cytoskeletal proteins actin and tubulin. Hepatocytes cultured between two layers of hydrated rat tail tendon collagen (sandwich gel) morphologically resembled cells in intact liver for several weeks. Actin filaments (F-actin) in these hepatocytes were concentrated under the plasma membrane in regions of cell-cell contact. In contrast, hepatocytes cultured on a single collagen gel were flattened and motile and had F-actin containing stress fibers. This was accompanied by a severalfold increase in actin mRNA. Microtubules formed an interwoven network in hepatocytes cultured in a sandwich gel, but in single gel cultures they formed long parallel arrays extending out to the cell periphery. Tubulin mRNA was severalfold greater in hepatocytes cultured on a single gel. Fibronectin and laminin staining were greater in single gel cultures, and these proteins were concentrated in fibrils radiating from the cell periphery. Overlaying a second collagen gel onto hepatocytes that had been cultured on a single gel (double gel rescue) reversed cell spreading and reduced stress fibers. Double gel rescue also resulted in a decrease in actin and tubulin mRNA to levels present in sandwich gel cultures and freshly isolated hepatocytes. These results show that the configuration of the external matrix has a dynamic effect on cytoskeletal proteins in cultured rat hepatocytes.     

Analysis of developmentally homogeneous neural crest cell populations in vitro: I. Formation,morphology and differentiative behavior

When early embryonic quail neural tubes are dissected free from surrounding tissues and placed in culture, small stellate neural crest cells usually migrate from the explant onto the substratum. This outgrowth has been reported to consist of a mixture of cells, some of which undergo melanogenesis, while the rest remain unpigmented. We have, in contrast to earlier observations, obtained a spatial separation of the two phenotypes. In these cultures the primary outgrowth of migrating cells remained almost free of pigment-forming cells, whereas small spherical clusters containing several hundred pigment-forming cells appeared on the explanted neural tubes. Whether the clusters remained with the tube explants or were subcultured, all cluster cells differentiated into melanocytes. Prior to melanogenesis, the appearance of the cultured cells from a cluster was indistinguishable from the cells in the outgrowth. The clusters provide a source of neural crest cells, that (1) can be easily obtained in comparatively large numbers, (2) is not contaminated with any other cell type, (3) can be isolated before the onset of differentiation, and (4) is developmentally homogeneous. Thus, the cluster population is well suited for many types of experiments, such as the identification of specific environmental factors that might control neural crest cell differentiation.