DEVELOPMENTAL

BIOLOGY

141,173-182(1990)

Substratum Effects on Cell Dispersal, Morphology, and Differentiation in Cultures of Avian Neural Crest Cells SHERRY L. ROGERS,*LUCYBERNARD,-~ANDJAMES

A. WESTON?

*Department of Anatomy, University of New Mexico, Albuquerque, New Mexico 87131;and ~Institute of Ne-uroscience, University of Oregon, Eugene, Oregm 97403 Accepted May 14, 1990 Adhesive extracellular matrix (ECM) molecules appear to play roles in the migration of neural crest cells, and may also provide cues for differentiation of these cells into a variety of phenotypes. We are studying the influences of specific ECM components on crest differentiation at the levels of both individual cells and cell populations. We report here that the glycoproteins fibronectin and laminin differentially affect melanogenesis in cultures of avian neural crest-derived cells. Clusters of neural crest cells were allowed to form on explanted neural tubes for 24 and 48 hr, and then subcultured on uncoated glass coverslips or coverslips coated with fibronectin or laminin. The morphology of cells varied on the three substrata, as did patterns of cell dispersal. Crest cells dispersed most rapidly and extensively on fibronectin. In contrast, cells on laminin dispersed initially, but then assumed a stellate morphology and rapidly formed small aggregates. Cell dispersal was minimal on glass substrata, resulting in a uniformly dense distribution. These patterns of dispersal were similar in subcultures of both 24- and 4%hr clusters, although dispersal of cells from older clusters was less extensive. The rate and extent of melanogenesis correlated with patterns of cell dispersal. Cells from 24-hr clusters underwent melanogenesis significantly more slowly on fibronectin than on the other two substrata. Pigment cells began to differentiate by 2 days of subculture in the cell aggregates on laminin and in the dense centers of cultures on untreated glass. By 5 days, there was significantly more melanogenesis in cultures on laminin and glass than on fibronectin substrata. Melanogenesis in cultures of 4%hr clusters was more rapid and extensive on control (glass) substrata than on fibronectin or laminin, correlating with reduced cell dispersal. We conclude that fibronectin and laminin, which are found along neural crest migratory pathways in Go, can affect melanogenesis in vitro by regulating patterns of cell dispersal. 0 19SO Academic

Press, Inc.

ences of specific ECM molecules are exerted remain to be defined. Migration of neural crest cells to diverse locations in Fibronectin and laminin are adhesive ECM glycoprovertebrate embryos has been well documented (Weston, teins that are found along neural crest migratory path1970; LeDouarin, 1982), but it is not yet clear where, ways (Newgreen and Thiery, 1980; Sternberg and during the course of this migration, developmental com- Kimber, 1986), and support neural crest cell locomotion mitment to particular lineages occurs. A critical prob- in vitro (Rovasio et ah, 1983; Runyan et uL, 1986). Allem lies in defining the roles that extrinsic cues along though crest cells can interact with both molecules, crest migratory pathways play in differentiation of available evidence suggests that the consequences of these cells. A number of extracellular matrix (ECM) these interactions differ. First, although fibronectin components are located in regions traversed by neural and laminin both may be present in embryonic basal crest cells (e.g., Newgreen and Thiery, 1980; Weston et laminae, they are not usually codistributed in other relea& 1984; Rogers et ah, 1986; Sternberg and Kimber, 1986; vant locations (Sternberg and Kimber, 1986; Rogers et Duband and Thiery, 1987; Epperlein et a& 1988), and uL, 1986; Duband and Thiery, 1987). This suggests that both in vitro and in viva studies strongly suggest that precursor populations at different sites could be exposed these components are involved in at least some phases of selectively to one or the other molecule. Second, there crest migration (e.g., Greenberg et a& 1981; Rovasio et are differences in the cell-binding domains of fibronecuZ., 1983; Runyan et ah, 1986; Bilozur and Hay, 1988; tin and laminin, and at least some cell surface receptors Bronner-Fraser and Lallier, 1988; Erickson, 1988; are specific for each molecule (Kleinman et cd., 1988; Payette et al, 1988). There is also evidence that ECM can Ruoslahti, 1988). Differences in cell-surface receptors promote or permit differentiation of neural crest cells can result in cell type-specific interactions with fibrointo certain phenotypes (Sieber-Blum et uL, 1981; Loring nectin and laminin, and with the multiple cell binding et uL, 1982; Maxwell and Forbes, 1987; Perris et uL, 1988; domains within each molecule (Rogers et uL, 1983; Edgar Campbell, 1989), but the mechanisms by which influet al, 1984; Humphries et uL, 1986; Rogers et uL, 1987, INTRODUCTION

173

0012-1606/90 $3.00 Copyright All rights

0 1990 by Academic Press, Inc. of reproduction in any form reserved.

174

DEVELOPMENTALBIOLOGY

1988). Neural crest cells are among the cell types that exhibit such different patterns of adhesive interactions (Dufour et al, 1988). Therefore, it seems reasonable to postulate that specific interactions between diverging subpopulations of crest cells and ECM components could influence crest differentiation, especially when ECM has been shown to affect cell differentiation in other systems (e.g., Blum and Wicha, 1988; Ocalan et al., 1988). It is clear that a complex variety of interactions between differentiating neural crest cells and ECM components is possible, and in vitro studies will be crucial to understand how individual matrix components influence crest cell behavior. The studies reported in this paper are part of a detailed analysis of this problem, using identified ECM macromolecules and their fragments, and cell populations with known developmental potential. We focus here upon the influences of intact fibronectin and laminin on melanogenesis in cultures of neural crest-derived cells. We report that crest cells disperse readily on fibronectin but cluster on laminin, and that these patterns of cell dispersal have consequences for differentiation of at least one neural crest phenotype.

vOLUME141,1990

neural crest cells to form. Glimelius and Weston (1981) and Vogel and Weston (1988) report that the length of time that cells in these clusters remain cohesive, independent of proximity to the neural tube, is critical to the course of differentiation once the cells are allowed to attach to a substratum and disperse. Specifically, there is a dramatic increase in melanogenic potential of the cell populations between 24 and 48 hr of cluster formation. In the experiments reported here, clusters were dissected off neural tubes at 22-24 or 48 hr after tube isolation and subcultured on the prepared substrata. The exact number of cells plated could not be determined without disrupting the clusters, but cluster size is directly related to cell number (Glimelius and Weston, 1981) and clusters of equivalent size were used in all cultures. Culture medium consisted of Ham’s F12 supplemented with 10% fetal calf serum (Hyclone), 5% chick embryo extract prepared from lo-day embryos, penicillin-streptomycin-fungizone, and 2 mM glutamine. For some experiments, serum was depleted of fibronectin by passage over a gelatin affinity column. Quantitative Estimates of the Onset and Extent of Melanogenesis

Cultures were observed daily with bright-field optics and the onset of melanogenesis in each was defined as the time when melanin granules were visable in at least Preparation of Substrata one cell. To assess the time course and extent of melanoPurified human plasma fibronectin and mouse EHS genesis, cultures were fixed with 4% paraformaldehyde tumor laminin (generous gifts of Dr. James McCarthy, on Days 3,5, and 7 of subculture, stained with a fluoresUniversity of Minnesota) were prepared as described cent nuclear dye (Hoechst 33258, Sigma), mounted on previously (Rogers et al, 1983). The proteins were di- slides and observed with a Zeiss microscope equipped luted to 25 pg/ml in Voller’s carbonate buffer, pH 9.6, with epifluorescence optics. The percentage of melanoand applied overnight to glass coverslips placed in 35- cytes in each culture was estimated by counting random mm petri dishes. Coverslips were rinsed three times nuclei in a total of 10 fields that, together, formed two with PBS prior to plating of cells. The protein concen- different transects through the middle of the culture. trations in the coating solutions correspond to approxiLevels of transmitted/uv light were adjusted to allow mately lo-’ M fibronectin and 2 X lo-* M laminin. Al- pigment granules to be seen. Counts were performed at though the actual amount of protein that bound to the a magnification of 40x, using a grid to aid in random coverslips was not determined, micro-ELISA assays sampling of nuclei (Vogel and Weston, 1988). One-way showed that the amount bound did not increase with analysis of variance and the Newman-Kuels multiple coating concentrations higher than those applied. comparisons test (Zar, 1984) were used to compare percentages of melanocytes on different substrata. Cell Culture Statistical Analyses of Cell Distribution and Den.&@ Neural tubes were isolated from stage 11-12 quail emSubjective assessments of dispersal and/or clustering bryos (Zacchei, 1961; equivalent to stage 13-14 chick embryos, Hamburger and Hamilton, 1952), as described of cells in vitro are frequently made, by us and others. In order to establish objective criteria for such subjective previously (Glimelius and Weston, 1981; Loring et d, 1982). Briefly, blocks of tissue containing neural tubes assessments, we employed an image analysis system and somites were dissected with tungsten needles, then (AIC, Irvine, CA) to determine the relative distribution incubated in pancreatin (GIBCO) and triturated to free and density of neural crest cells cultured on different the neural tubes from surrounding tissue. The isolated substrata. The cultures used for this study were not neural tubes were incubated on a nonadhesive substra- those used for the melanocyte counts, but exhibited the tum (Falcon plastic petri dishes) to allow clusters of extremely reproducible distributions of cells that were MATERIALS

AND

METHODS

ROGERS, BERNARD, AND WESTON

typical of our subjective observations. The area occupied by nuclei in 12 randomly chosen regions within each field (fields were selected along two transects as described above) was determined and their mean area and the variance around this mean were computed. We reasoned that the variance around the mean of different regions would be directly related to the degree of clustering (i.e., nonuniform distribution) of nuclei. Differences in cell aggregation among treatments (substrata) were tested by comparing variances in the area occupied by cells in each region. For each treatment, the variance among readings was estimated by nested analysis of variance (readings nested within fields nested within cultures), resulting in a “dispersal index.” These estimates were then used to conduct F tests for significant differences among treatments. Prior to analysis, the data were transformed, using a generalized log transform to remove the dependence of the variance on the mean (Zar, 1984). While variance ratio testing can distinguish uniform versus nonuniform distributions of cells, it cannot resolve differences in cell density in relatively evenly distributed cell populations. In order to compare cell density in cultures on FN and uncoated glass, numbers of nuclei were computed in consecutive fields through the center of each culture, such that the most dispersed cells at each edge were included in the analysis. To control for different total numbers of cells in different cultures, the counts were normalized by dividing the number in each field by the total number of cells counted in that culture, resulting in a “density index.” Paired t tests were used to analyze differences in the normalized values between treatments. RESULTS

Cell Mmphology and Dispersal Twenty-j&r-hour clusters of neural crest celLs.Neural crest cell clusters attach to fibronectin (FN), laminin (LM), and uncoated glass substrata, and these cells rapidly (within an hour after plating) begin to disperse onto all three substrata. However, time-lapse video analysis shows that following initial migration away from the center of the clusters, cell morphology and patterns of dispersal become markedly different on each substratum (Fig. 1). The following observations were not affected by culturing the cells in medium containing serum that had been depleted of FN. In response to FN, crest cells flatten and often form large lamellae, exhibiting the polarized appearance typical of motile cells (Fig. 1B). Cells at the periphery of clusters on FN tend to move away from one another, and usually continue to disperse throughout the culture period. In combination with cell proliferation, this behavior results in cells being relatively evenly distributed

Eff&ts of ECM on Crest D$%rentiatim

175

over a large area, with cell density at the center of the culture becoming much greater than at its periphery (Fig. 1A). Although crest cells disperse initially on laminin substrata, after several hours they lose their lamellipodia and become stellate or spindle-shaped (Fig. 1D). As this change in morphology takes place, the cells form small aggregates and appear to become immobile. The majority of cells in cultures on LM become incorporated into the aggregates, which increase in size due to recruitment of motile cells and probably also to cell proliferation. The behavior of crest cells plated on untreated glass coverslips differs from that on both FN and LM. They do not disperse as extensively as they do in response to FN, and this results in relatively high cell density. Despite the difference in density, the distribution of cells on both FN and glass is relatively uniform compared with those on LM. Individual cells at the edges of clusters on glass do not form the large lamellae typical of those on FN, or become stellate like those on LM, but their morphology ranges from flat to slightly rounded (Fig. 1F). In order to analyze quantitatively the observed differences in cell aggregation on FN, LM, and glass, statistical tests were performed on data collected from 3day cultures of 24-hr clusters. The patterns of cell distribution in these cultures were consistent with those observed at this time point during analysis of melanogenesis (see below). To address differences in patterns of cell dispersal/aggregation, the variance in numbers of cell nuclei among regions within each culture (obtained from nested analyses of variance, see Methods) provided an “index of dispersal” (Table 1). Larger values for this index indicate less uniformity of cell distribution. Comparison of pairs of these indices were performed by variance ratio testing, and the significance of differences assessed by F-statistic. Table 1 shows that this nonuniformity of cell distribution in cultures on LM is significantly greater than in cultures on either FN or glass. This reflects the aggregation of cells into discrete clumps on LM substrata, compared with the relatively even distribution of cells on FN and glass. Although variance ratio testing detects differences in patterns of cell dispersal with respect to uniform vs nonuniform distribution, it does not address differences in cell density in relatively evenly distributed populations. In order to compare densities of cells in cultures on FN and glass, nuclei were counted in consecutive fields through the center of each culture. A “density index” (Table 1) was derived by dividing the average number of cells per field by the total number of cells counted in each culture. Thus, the index represents the mean percentage of cells in a culture found in an “average” field, and indicates how closely cells are packed. This index takes into consideration variations in (1) the

1’76

DEVELOPMENTAL BIOLOGY

VOLUME 141,1990

FIG. 1. Neural crest cells from 24-hr clusters subcultured on fibronectin (A and B), laminin (C and D) and uncoated glass coverslips (E and F) for 5 days. Cells plated on fibronectin disperse extensively, resulting in a large area of the substratum being covered with cells (A). Melanogenesis begins at the center of the cluster (arrows) where cells are most dense. At the edge of these cultures, individual cells tend to be flat, with broad lamellae (B, arrows) typical of motile cells. On laminin, crest cells begin to disperse, but shortly thereafter tend to become stellate or spindle-shaped and form small aggregates (C). Melanogenesis proceeds rapidly within aggregates (C and D, large arrows), and isolated cells also differentiate (D, small arrows). Neural crest clusters plated on uncoated glass do not readily disperse, and the dense centers of the cell

ROGERS,BERNARD,AND WESTON

Effects of ECM 012Crest @%erentiatim

177

than in cultures on LM (P < .005) and glass (P < .05). There are differences in the proportions of melanocytes on all three substrata at 5 days of subculture, with the most dramatic increase in the control cultures and a relatively small increase in the cultures on FN. After an extended time in culture, these distinctions disappear. Thus, by 7 days the proportion of melanocytes in cultures on all three substrata are nearly identical, due in part to a rapid increase in older cultures on FN. Forty-eight-hour clusters. Melanocytes usually are not present in subcultures of 4%hr clusters after 1 day, but most cultures have significant numbers by 2 days (approximately 20-35s; Fig. 4). By 3 days of subculture, melanocyte differentiation in control cultures has reached a peak level, while the process is significantly slower in cultures on both FN and LM. This difference is maintained until 7 days, when proportions of melanocytes appear to be relatively equal on all three substrata. As in cultures of 24-hr clusters, melanogenesis begins in areas of highest cell density, but more cells in cultures from older clusters differentiate into melanocytes, irrespective of their location in the culture. Figure 5 compares proportions of melanocytes in 5day subcultures of 24 and 4%hr clusters at 5 days of subculture, expressed as percentages of control (cultures on untreated glass) values. Three salient conclusions are emphasized by representing the data in this manner. First, proportions of melanocytes are lower when cells are plated on FN or LM rather than uncoated Melanogenesis glass substrata. This is true of cells from both 24- and Twenty-four-hour clusters. The presence of pigment 4%hr clusters. Second, while melanogenesis is least exgranules in one or more cells in a culture was defined as tensive in all cultures exposed to FN substrata, this the beginning of melanogenesis, and each culture was treatment has its most dramatic effect on cells from scored on the second and third days of subculture for the “young” clusters. Third, in contrast, the reduction in presence or absence of melanocytes (Table 2). By 2 days, melanogenesis in cultures on LM substrata is approxipigmentation had begun in an average of half, or more, mately the same with respect to cluster age. of the cultures on LM (47%) and untreated glass (62%). DISCUSSION By 3 days, melanogenesis had begun in almost all culECM, present in all neural crest migratory pathways, tures on these two substrata. In contrast, only 21% of contains a number of molecular components. Many of the cultures plated on FN contained melanocytes at Day these components can influence crest cell behavior, by 2, and 16% had not yet begun to undergo pigmentation virtue of their particular adhesive characteristics and/ on Day 3. In all cultures, the onset of melanogenesis or by more indirect effects upon processes of differenoccurs where cell density is highest, in the center of cul- tiation. In this study, we have examined how two ECMtures on FN and glass, and in the aggregates that form adhesive glycoproteins, fibronectin and laminin, affect on LM (Figs. lA, lC, 1E). Isolated cells on LM and glass aspects of neural crest cell behavior that appear to be also frequently differentiated into melanocytes, while involved in differentiation of melanocytes, important the flat cells with lamellipodia on FN rarely did so (Figs. neural crest derivatives. lB, lD, and 1F). Figure 3 compares the extent of melanogenesis on the Fibronectin and Lam&in Can Regulate Patterns of Neural Crest Cell Dispersal three substrata at 3, 5, and 7 days of subculture. At 3 days, an average of 7% of cells in cultures on FN have Cell morphology that is conductive to motility is an differentiated into melanocytes, significantly fewer obvious prerequisite for cell dispersal. Fibronectin-

initial number of cells plated, and (2) possible differences in proliferation rates in different cultures. A paired t test was performed on the density indices for cultures on FN and glass, and reveals a statistically significant difference in these values, confirming our subjective impression that cell dispersal is most extensive in cultures on FN. In addition, there was a significant difference in the number of consecutive fields required to perform the counts of nuclei described above, such that cultures on FN covered a larger area than those on glass. Although this assessment supports the conclusion drawn from the density indices, it should be interpreted cautiously, since variations in initial cluster size could affect the area of dispersal. Forty-eight-hour clusters. Clusters that are allowed to remain attached to neural tubes for 48 hr are more cohesive than are those that are removed a day earlier. Isolated 4%hr clusters adhere to FN, LM, and untreated coverslips, but the cells disperse less on all three substrata compared to cells from the younger clusters. The patterns of cell distribution are similar to those described for 24-hr clusters, with cells dispersing evenly on FN (Fig. 2A), forming small aggregates on LM (Fig. 2B), and remaining relatively nondispersed on glass (Fig. 2C). The morphology of cells on each substatum resembles that of cells from 24-hr clusters, although in all cases lamellipodia are fewer and smaller.

populations become heavily pigmented (E). At the edge of these clusters, the morphology of individual cells ranges from fairly flat (F, large arrow), to relatively rounded. As on laminin, isolated cells often differentiate into melanocytes (F, small arrows). Bars = 50 pm.

178

DEVELOPMENTALBIOLOGY

VOLUME141,1%0

TABLE 1 PA!ITERNSOFCELLDISPERSAL~ Substratum

Dispersal index b

LM FN Glass

.034 .033

.048

Signif.”

Density indexd

P < .Ol

&2

NS

.167

Signif.

P < .Ol

’ An image analysis system was employed to assess quantitatively (1) patterns of cell distribution on different substrata, and (2) differences in cell density on FN and glass substrata. See Materials and Methods for details. bThe index of dispersal represents the variance around the mean number of nuclei per field. A relatively large value for this index (i.e., for cultures on LM) indicates a nonuniform distribution of nuclei. See Materials and Methods for derivation of this index. ’ Statistical significance of differences between dispersal indices was evaluated by variance ratio tests on each pair of treatments (LM:FN, FN:Glass; LM:Glass), with differences in ratios assessed by F-statistics with 440 degrees of freedom. For each treatment, seven cultures from two separate experiments were analyzed. The difference in dispersal indices for cultures on LM and glass was also significant (P < .Ol), but is not shown here. NS = not significant. d Difference in density of cells on FN and glass was tested by comparing the average percentage of total cells found in any given field (i.e., the density index = mean number of cells per field divided by the total number of cells counted), which is directly related to density. See Methods. eDifferences in density indices were assessed with a paired t test.

treated substrata support morphological changes that are characteristic of motile cells for at least 5 days of subculture of cells from 24-hr neural crest clusters, and to a lesser extent in cultures of cells from 4%hr clusters. Cells plated on uncoated substrata exhibit such morphology less frequently and, accordingly, do not disperse as extensively. In contrast, crest cells begin to migrate onto laminin-treated substrata, but their morphology then changes and most cells reaggregate, forming small clumps that are scattered throughout these cultures. In all cultures, the patterns of cell dispersal on different substrata have clear consequences for interactions among cells, so that cell-cell contacts predominate over cell-substratum interactions on glass and laminin. However, it may be important that the timing and overall extent of cell-cell contacts differs on these two surfaces. Patterns of Cell Llispemal Can Influence the Course of Melanogenesis Differences in the morphology of neural crest cell cultures on different substrata correlate predictably with variations in the rate and extent of melanogenesis. Melanocytes begin to differentiate first in areas of high cell density, and their proportions relative to nonmelanocytes become greatest in these areas. This is true in the dense centers of cell populations on both fibronectin and

FIG. 2. Neural crest cells from 4%hr clusters subcultured on fibronectin (A), laminin (B), and untreated coverslips (C) for 5 days. Patterns of cell dispersal resemble those of 24-hr clusters, but melanogenesis is more extensive, especially in control cultures on uncoated glass (C) where cells are most dense. Shapes of individual cells resemble those in cultures of 24-hr clusters, although most are considerably less flat. Bar = 50 pm.

ROGERS,BERNARD,AND WESTON TABLE 2 MELANOGENESISAT Two DAYS OFSUEXXLTURE” % Cultures pigmented

Substratum

6070--

-i 50-5 -s 30-60

Signif.

21

FN Glass LM

P < .Ol P < .Ol

62 47

40

O--O

x

a The onset of melanogenesis was recognized by the appearance of melanin granules in at least one cell in a culture. Data from 11 experiments were pooled, and were analyzed by partitioning of the xz statistic (Maxwell, 1961).

glass, and in the cell aggregates scattered throughout cultures on laminin. Where cells remain dispersed, for example at the edges of cultures on fibronectin substrata and at the outgrowth boundary of other cultures, the rate and extent of pigmentation are lowest. Studies by Perris and co-workers (1988) also demonstrate a relationship in vitro between ECM-mediated amphibian crest cell aggregation/dispersal and differentiation, although in this case cell dispersal is associated with enhanced melanogenesis. The discrepency between these observations and our own might be explained by (1) species differences, (2) culture conditions and complexity of the substrata, and/or (3) cell populations with different developmental potential. In other experimental systems, mesenchyme also supports melanocyte differentiation (Derby, 1982; Campbell, 1989), and the importance of a mesenchymal component in this process is underscored by studies with the Steel mutant mouse, in which the defect underlying faulty melano-

70

0 x :

50 40

l --0FN l . . . l LM a--@ Control

20 --

l -•Control 10 -0-l

I 1

3

!

:,/

/

1

3

Length of Subculture

5

.’

:‘,

5

7

(days)

FIG. 4. The time course of melanogenesis in subcultures of 48-hr neural crest clusters. Melanogenesis occurs rapidly in all cultures, but, until approximately 7 days of subculture, proportions of melanocytes are fewer in cultures on fibronectin and laminin than in controls. Error bars = standard errors of means.

genesis appears to reside in the ECM produced by skin cells (Morrison-Graham et aL, 1990). Here too, melanogenesis and cell dispersal seem to be inversely related. Mechanisms by which Adhesive Substrata May Regulate Neural Crest Cell LXspersal and l&%rentiatim D$fe-rential adhesiveness. Regulation of cell dispersal by extracellular matrix components is likely to be highly dependent upon the particular adhesive characteristics of these components. Neural crest cells probably respond to a hierarchy of adhesive surfaces (Rovasio et ah, 1983; Dufuor et a& 1988) as do neuronal growth cones (Letourneau, 19X5), and in this case, fibronectin appears to provide an attractive adhesive alternative to the surfaces of adjacent cells. Cell aggregation on laminin substrata could then be interpreted as a preference

:

t---t

FN

l . . . l LM

Length of Subculture

60

179

Effect.s of ECM on Crest Lh&rentiaticm

c P z

80 70

T EJ

T

0

24 hr clusters 48 hr clusters

7

(days)

FIG. 3. Percentages of melanocytes on fibronectin (FN), laminin (LM), and untreated glass (control) substrata at 3, 5, and 7 days of subculture of 24 hr neural crest clusters. Melanogenesis proceeds more slowly in cultures on fibronectin than in cultures on laminin or untreated glass. By 6-7 days, the proportions of melanocytes are equivalent in all cultures. For the 3- and S-day time points, data were from five separate experiments; for ‘I-day cultures, four experiments were analyzed. Error bars = standard errors of means.

FN

LM

FIG. 5. Comparison of proportions of melanocytes in 5-day subcultures of 24- and 48-hr neural crest clusters, expressed as percentages of control (cultures on uncoated glass) values. Note that fibronectin has a selective effect on melanogenesis in cultures of cells from “young” clusters.

180

DEVELOPMENTALBIOLOGY

of the cells for each others’ surfaces when presented with laminin as an alternative substratum (see Weston, 1970). Cells cultured on uncoated glass disperse even less than they do initially in response to laminin. The hypothesis that neural crest cell dispersal in vivo can be explained, at least in part, by contact inhibition of movement (reviewed in Erickson, 1988) might be reassessed to include the idea that these cells “search” for substrata of appropriate adhesiveness in complex extracellular environments. When considering how hierarchies of adhesive responses might influence crest cell behavior, it is important to emphasize that such hierarchies probably involve multiple cell binding domains within individual ECM molecules (Dufuor et aL, 1988) as well as the intact molecules, as has been reported for nerve fiber elongation (Humphries et al., 1986; Rogers et al, 1987,1988; Kleinman et al, 1988). The association of dispersing cells with laminin substrata undergoes a change during the first few hours after plating. The mechanism(s) responsible for the progressive morphological alterations that we have observed are intriquing, but unknown at present. Possibilities include (1) changes in cell surface’receptors for laminin, and (2) specific modifications of the substratum by the cells. Further investigation of this phenomenon may enhance understanding of how cell interactions with an adhesive molecule, or with specific domains within it, can change with time. Regulation of cell-cell associations. The different developmental potentials of cell populations in 24 vs 48 hr clusters were demonstrated in previous studies (Glimelius and Weston, 1981, Vogel and Weston, 1988), and it was suggested that some aspect(s) of the environment within the older clusters causes either loss of neurogenic precursors or enhanced differentiation of uncommitted cells along a nonneurogenic pathway (Vogel and Weston, 1988; Weston et al., 1984). This hypothesis would also apply to the aggregates of cells in the present study, in cultures of clusters of both ages. Certainly, the microenvironments created within aggregates of cells would be different from those encountered by cells that are actively moving away from each other by interacting with an adhesive substratum. It is also likely that the timing and extent of cell aggregation, which can be regulated by the composition of the substratum, can affect the composition of these microenvironments. In addition, cellular differentiation in response to extracellular matrices might also involve access to growth factors that bind to particular substrata from the extracellular milieu (Folkman et al, 1988; Bashkin et a,?.,1989), and that access to growth factors could in turn be affected by patterns of cell-cell and cell-substratum associations. In any of the microenvironments that evolve in vivo, an underlying regulatory mechanism may be con-

vOLUME141.1990

trol of cell proliferation, leading to expansion of select cell populations. Eflects of eetl morphology. Adhesive substrata might also influence neural crest cell differentiation by mechanisms that operate separately from, or in addition to regulation of cell dispersal/aggregation. In the experiments reported here, we consistently observed that the flat, motile cells at the periphery of neural crest cultures on fibronectin rarely differentiate into melanocytes, whereas isolated cells with stellate or spindle-like morphology on laminin and untreated glass often do so (Fig. 1; also see Maxwell, 1976). In other systems, the morphological changes that accompany cell interaction with particular substrata have been shown to have consequences for cell proliferation and/or gene expression (e.g., Folkman and Moscona, 1978; Blum and Wicha, 1988; Ocalan et aL, 1988). In our studies, it will be critical to determine whether cell morphology can directly influence differentiation along the melanocyte pathway, or whether shape changes are a consequence of that developmental decision. Similarly, an intriguing question is whether the morphological alterations associated with cell motility, and/or motility itself, prevent melanocyte differentiation. Potential Relevance to Crest Cell Behavior in Vivo While it seems clear that adhesive glycoproteins can have profound influences upon neural crest cell behavior, it must be emphasized that these molecules comprise only one aspect of complex extracellular environments (Weston et al, 1984; Tucker and Erickson, 1984; Newgreen, 1989). Further, until we know more about what substrata cells actually use in vivo (see Boucaut et al, 1984; Bronner-Fraser and Lallier, 1988), we can only speculate about how our in vitro observations might apply to crest migratory and differentiative behavior in the embryo. One site at which such speculation seems warranted is along the ventral crest migratory pathway, where the presence of abundant fibronectin (Newgreen and Thiery, 1980) might promote rapid initial cell dispersal. An absence of melanogenesis and the predominance of neuronal and glial derivatives in this region might be a consequence of this rapid dispersal. Indeed, in vitro studies do suggest a reciprocal relationship between differentiation of neurons and glia and differentiation of melanocytes (Vogel and Weston, 1988; our unpublished observations). Also, reports of enhanced neurogenesis in vitro in response to fibronectin (SieberBlum, 1981; Loring et al, 1982) complement our observations. It is important to consider that environmental factors such as fibronectin may exert effects on migrating cells before they reach their final destinations. In terms of neuronal and glial differentiation, fibronectin

ROGERS,BERNARD,AND WESTON

is conspicuously absent as cells coalesce to form spinal ganglia (Rogers et aL, 1986). Interestingly, laminin appears at these sites, where it might be involved in halting cell migration, similar to the effects that we have observed in witro. In contrast to the ventral wave of crest migration, cells that migrate along the dorsal, subepidermal pathway appear to do so at a slightly later time (Derby, 1978; Loring and Erickson, 1987; Serbedzija et al, 1989). Our observations of increased proportions of melanocytes in cell aggregates in subcultures of either 24- or 48-hr neural crest clusters, certainly suggest that a delay in dispersal in viva might favor melanocyte differentiation. Although laminin is present in the region containing premigratory crest cells (Rogers et UC,1986; Erickson, 1988), it is not known whether it plays a role in delaying the dispersal of melanogenic precursors. These precursors may also interact with laminin in the developing skin, where it might affect cell positioning and morphology. To address these questions adequately, more information is needed concerning (1) timing and mechanisms of migration of melanocyte precursors into the developing skin, (2) sites of precursor proliferation and cell-cell interactions, and (3) the precise composition of the microenvironments in which these activities take place.

Effects of ECM 072Crest Di&renkiation

181

against a laminin-heparan sulfate proteoglycan complex pertubs cranial neural crest migration in viva. J. CeU Bid 106,1321-1329. CAMPBELL,S. (1989). Melanogenesis of avian neural crest cells in vitro is influenced by external cues in the periorbital mesenchyme. Deuelgpment 106, ‘71’7-726. DERBY, M. (1978). Analysis of glycosaminoglycans within the extracellular environment encountered by migrating neural crest cells. De-u.Bid. 66,321-336. DERBY,M. (1982). Environmental factors affecting neural crest differentiation: Melanocyte differentiation by crest cells exposed to cellfree (deoxycholate-extracted) dermal mesenchyme matrix. Cell Tissue Res. 225,379-386. DUBAND, J.-L., and THIERY, J. P. (1987). Distribution of laminin and collagens during avian neural crest development. Deuelopment 101, 461-478. DUFOUR,S., DUBAND, J. L., HUMPHRIES,M. J., OBARA, M., YAMADA, K. M., and THIERY, J. P. (1988). Attachment, spreading and locomotion of avian neural crest cells are mediated by multiple adhesion sites on fibronectin molecules. EMBO J. 7,2661-2671. EDGAR, D., RMPL, R., and THOENEN,H. (1984). The heparin-binding domain of laminin is responsible for its effects on neurite outgrowth and neuronal survival. EMBO J. 3,1463-1468. EPPERLEIN,H.-H., HALTER, W., and TUCKER,R. P. (1988). The distribution of fibronectin and tenascin along migratory pathways of the neural crest in the trunk of amphibian embryos. Deuekqmzent 103, 743-756. ERICKSON,C. A. (1988). Control of pathfinding by the avian trunk neural crest. Develqwnwnt lOS(Suppl.), 63-80. FOLKMAN,J., KLAGSRRUN,M., SASSE,J., WADZINSKI, M., INGBER,D., and VLODAVSKY,I. (1988). A heparin-binding angiogenic proteinbasic fibroblast growth factor-is stored within basement membrane. Amer. J. Pathol 130,393-400. This work was supported by NIH Grant NS23368 and a Research FOLKMAN, J., and MOSCONA,A. (1978). Role of cell shape in growth control. Nature (London) 273,345-349. Career Development Award to S.L.R. and DE04316 to J.A.W. The GLIYJZLIUS,B., and WESTON,J. A. (1981). Analysis of developmentally authors thank Ms. Susan Alexander for excellent technical assistance, homogeneous neural crest cell populations in vitro. III. Role of culand Drs. Lorraine Heisler (Department of Biology, University of Oreture environment in cluster formation and differentiation. Cell gon) and Dorothy Pathak (Department of Family, Community, and D&-r. 10,57-67. Emergency Medicine, University of New Mexico) for performing many of the statistical analyses. We also thank Dr. Kristine Vogel GREENBERG,J. H., SEPPA, S., SEPPA, H., and HEWIIT, A. T. (1981). Role of collagen and fibronectin in neural crest cell adhesion and (Institute of Neuroscience, University of Oregon) for valuable advice migration. Den Biol 87,259-266. and discussions. HAMBURGER, V., and HAMILTON, H. L. (1951). A series of normal stages in the development of the chick embryo. J. Mcn$w1!88,49-92. HUMPHRIES, M. J., AKNAMA, S. K., KOMORIYA,A., OLDEN, K., and REFERENCES YAMADA, K. M. (1986). Identification of an alternatively spliced site in human plasma fibronectin that mediates cell type-specific adheBASHKIN, P., DOCTROW,S., KLAGSBRUN,M., SVAHN, C. M., FOLKMAN, sion. J. Cell Biol. 103,2637-2647. J., and VLODAVSKY,I. (1989). Basic fibroblast growth factor binds to KLEINMAN, H. K., OGLE,R. C., CANNON,F. B., LITTLE, C. D., SWEENEY, subendothelial extracellular matrix and is released by hepartinase T. M., and LUCKENBILL-EDDS,L. (1988). Laminin receptors for neuand heparin-like molecules. Biochxmist~ 28,1’737-1’743. rite formation. Proc. Nat. Acad Soi USA 85,1282-1286. BILOZUR,M. E., and HAY, E. D. (1988). Neural crest migration in 3D extracellular matrix utilizes laminin, fibronectin, or collagen. Deu. LEDOUARIN, N. (1982). “The Neural Crest.” Cambridge Univ. Press, Cambridge. Biol 125,19-33. LETOURNEAU,P. C. (1975). Cell-to-substratum adhesion and guidance BLUM, J. L., and WICHA, M. S. (1988). Role of the cytoskeleton in lamiof axonal elongation. Dev. Biol. 44,92-102. nin induced mammary gene expression. J. CeU Physiol 135,13-22. BOUCAUT,J. C., DARRIBERE,T., POOLE,T. J., AOYAMA, H., YAMADA, LORING, J., GLIMELIUS, B., and WESTON,J. A. (1982). Extracellular matrix materials influence quail neural crest cell differentiation in K. M., and THIERY, J.-P. (1984). Biologically active synthetic pepvitro. Dev. Biol. 90,165-174. tides as probes of embryonic development: A competitive peptide LORING,J. F., and ERICKSON,C. A. (1987). Neural crest cell migratory inhibitor of flbronectin function inhibits gastrulation in amphibian embryos and neural crest migration in avian embryos. J. Cell Biol. pathways in the trunk of the chick embryo. Dew. Biol 121,220-236. MAXWELL, A. E. (1961). “Analysing Qualitative Data.” Chapman and 99,1822-X30. BRONNER-FRASER, M. (1985). Alterations in neural crest migration by Hall, London. a monoclonal antibody that affects cell adhesion. J. CeU Biol. 101, MAXWELL, G. D. (1976). Cell cycle changes during neural crest cell 610-617. differentiation in vitro. Dev. Biol 49,66-79. BRONNER-FRASER, M., and LALLIER, T. (1988). A monoclonal antibody MAXWELL, G. D., and FORBES,M. E. (1987). Exogeneous basement-

182

DEVELOPMENTALBIOLOGY

membrane-like matrix stimulates adrenergic development in avian neural crest cultures. Development 101, ‘76’7-7’76 MORRISON-GRAHAM,K., WEST-J• HNSRUD, L., and WESTON, J. A. (1990). Extracellular matrix from normal but not Steel mutant mice enhances melanogenesis in cultured mouse neural crest cells. Den BioL, 139,299-307. NEWGREEN,D. F. (1989). Physieal influences on neural crest cell migration in avian embryos: Contact guidance and spatial restriction. Lkv. Bid 131,136-148. NEWGREEN,D. F., and THIERY, J.-P. (1980). Fibronectin in early avian embryos: Synthesis and distribution along the migration pathways of neural crest cells. Cell Tissue Res. 211,269~291. OCALAN,M., GOODMAN,S. L., KUHL, U., HAUSCHKA,S. D., and VONDER MARK, K. (1988). Laminin alters cell shape and stimulates motility and proliferation of murine skeletal myoblasts. Dev. Bid 125,158167. PAYETTE, R. F., TENNYSON,V. M., POMERANZ,H. D., PHAY, T. D., ROTHMAN,T. P., and GERSHON,M.D. (1988).Accumulation of components of basal laminae: Association with the failure of neural crest cells to colonize the presumptive aganglionic bowel of Is/Is mutant mice. Dev. Biol 125,341-360. PERRIS,R., VON BOXBERG,Y., and LOFBERG,J. (1988). Local embryonic matrices determine region-specific phenotypes in neural crest cells. Science 241,86-89. ROGERS,S. L., EDSON,K. J., LETOURNEAU,P. C., and MCLOON,S. C. (1986). Distribution of laminin in the developing peripheral nervous system of the chick. Der. Bid 113,429-435. ROGERS,S. L., LETOURNEAU,P. C., PALM, S. L., MCCARTHY,J. B., and FURCHT,L. T. (1983). Neurite extension by peripheral and central nervous system neurons in response to substratum-bound fibronectin and laminin. Dev. Biol 98,212+X0. ROGERS,S. L., LETOURNXAU,P. C., PETERSON,B. A., FURCHT,L. T., and MCCARTHY,J. B. (1987). Selective interaction of peripheral and central nervous system cells with two distinct cell-binding domains of fibronectin. J. CeUBid 105,1435-1442. ROGERS, S. L., PALM, S. L., LETOURNEAU, P. C., HANLON, K., MCCARTHY,J. B., and FURCHT,L. T. (1988). Cell adhesion and neurite extension in response to two proteolytic fragments of laminin. J. Neurosci. Res. 21,315-322.

VOLUME141.1990

ROVASIO,R. A., DELOUVEE,A., YAMADA, K. M., TIMPL, R., and THIERY, J. P. (1983). Neural crest cell migration: Requirements for exogenous fibronectin and high cell density. J. Cell BioL 96,462-473. RUNYAN,R. B., MAXWELL,G. D., and SHUR,B. D. (1986). Evidence for a novel enzymatic mechanism of neural crest cell migration on extracellular glycoconjugate matrices. J. Cell Biol 102,432-441. RUOSLAHTI,E. (1988). Fibronectin and its receptors. Annu Rev. Bio&em. 57,375413. SERBEDZIJA,G. N., BRONNER-FRASER, M., and FRASER,S. E. (1989). A vital dye analysis of the timing and pathways of avian trunk neural crest cell migration. Deuelqpment 106,809-816. SIEBER-BLUM,M., SIEBER,F., and YAMADA, K. M. (1981). Cellular fibronectin promotes adrenergic differentiation of quail neural crest cells in vitro. Exp. Cell Res. 133,285-295. STERNBERG,J., and KIMBER, S. J. (1986). Distribution of fibronectin, laminin, and entactin in the environment of migrating neural crest cells in early mouse embryos. J. Emrgol Exp. Morpti 91,276~282. TUCKER,R. P., and ERICKSON,C. A. (1984). Morphology and behavior of quail neural crest cells in artificial three-dimensional matrices. Lkv. Biol 104,390~405. VOGEL, K. S., and WESTON,J. A. (1988). A subpopulation of cultured avian neural crest cells has transient neurogenic potential. Neuron 1,569-57’7.

WESTON,J. A. (1970). The migration and differentiation of neural crest cells. Adv. Morphogex 8.41-114. WES~N, J. A., CIMENT, G., and GIRDLESTONE,J. (1984). The role of extracellular matrix in neural crest development: A reevaluation. In “The Role of Extracellular Matrix in Development” (R. L. Trelstad, Ed.), pp. 433-460. A. R. Liss, New York. WESTON,J. A., VOGEL, K. S., and MARUSICH,M. F. (1988) Identification and fate of neural crest subpopulations in early embryonic development. In “From Message to Mind” (S. S. Easter, K. F. Barald, and B. M. Carlson, Eds.) pp. 224-237. Sinauer Assoc., Inc., Sunderland, MA. ZACCHEI,A. M. (1961). Lo sviluppo embrionale della quagglia giapponese (Coturnix coturnix japonica). Arc Iti An& Embr@ 66, &‘72. ZAR, J. H. (1984). “Biostatistical Analysis.” Prentice-Hall, Englewood Cliffs, NJ.