Neural crest cells in the trunk of vertebrate embryos have a choice of pathways after emigrating from the neural tube: they can migrate in either the medial pathway between somites and neural tube, or the lateral pathway between somites and epidermis. In zebrafish embryos, the first cells to migrate all choose the medial pathway. High resolution imaging of cells in living embryos suggests that neural crest cells do so because of repulsion by somites: cells take the medial pathway because the lateral somite surface triggers a paralysis and retraction of protrusions (contact inhibi-tion or collapse) when the medial surface does not. Partial deletion of somites, using the spadetail mutation allows pre-cocious entry into the lateral pathway, but only where somites are absent, supporting the notion that an inhibitory cue on somites delays entry. Growth cones of Rohon-Beard cells enter the lateral pathway before neural crest cells, demonstrating that there is no absolute barrier to migration. These data, in addition to providing a detailed picture of neural crest cells migrating in vivo, suggest that neural crest cells, like neuronal growth cones, are guided by a specific cue that triggers ‘collapse’ of active protru-sions.

Neural crest cells are a population of migratory cells that arise from the dorsal region of the neural tube of vertebrate embryos. As shown by a large number of cell-labelling studies in chick (Weston, 1963; Le Douarin, 1982; Bronner-Fraser, 1986; Serbedzija et al., 1989), mouse (Serbedzija et al., 1990), Xenopus (Collazo et al., 1993) and zebrafish (Raible and Eisen, 1994; Schilling and Kimmel, 1994), these cells migrate far and wide within the embryo, and differentiate into many different cell types including neurons and glial cells of the peripheral nervous system, pigment cells and facial skeleton. When migrating from the neural tube to their target sites, neural crest cells use highly stereotypical routes. In the trunk of the chick (Erickson et al., 1992) and zebrafish embryo (Raible et al., 1992), for example, the first cohort of neural crest cells all enter the pathway between the neural tube and somites (referred to here as the medial pathway) and avoid the pathway that lies between somite and ectoderm (the lateral pathway). Only after a delay of 5-6 hours in the case of the zebrafish, or 24 hours in the case of the chick, is the lateral pathway invaded by neural crest cells that give rise to melanocytes.

How are early neural crest cells directed into the medial pathway in the zebrafish embryo? According to one theory, these cells could be guided by attractant molecules in the medial pathway. Alternatively, the lateral pathway could contain molecules that prevent entry of early neural crest cells.

In either case, the guidance could depend on diffusible sig-nalling molecules, on influences from extracellular matrix, or on contact interactions between the neural crest cells and surfaces of other cells that they encounter along the way.

Previous studies have tested the involvement of specific molecules, such as chondroitin sulphate proteoglycan or fibronectin, in neural crest guidance (Thiery et al., 1982; Newgreen et al., 1986), but there has been very little direct observation of how neural crest cells move in the living embryo. This is largely because the most popular organism for neural crest studies has been the chick embryo, which is semi-opaque.

The zebrafish embryo, however, is transparent, allowing neural crest cells to be observed at high resolution as they migrate in their natural environment. Careful analysis of time-lapse films can reveal which guidance strategies are actually used, thereby identifying the phenomena for which a molecular explanation must be found. The present study establishes that early crest cell protrusions become paralysed and retract upon contact with the surface of young somites. When somites are absent, this response does not occur, and crest cells enter the lateral pathway prematurely. Neuronal growth cones, unlike crest cells, enter the lateral pathway at early stages even when somites are present. These findings indicate that neural crest cell entry into the lateral pathway in the zebrafish embryo is specifically delayed by an inhibitory cue, which triggers a ‘collapse’ of active protrusions.

High resolution microscopy of living embryos

Zebrafish embryos were dechorionated with watchmakers forceps, mounted on a glass slide and observed with differential interference contrast optics on a Leitz Diaplan, using a 100× oil-immersion objective. High resolution images were obtained by ensuring that the cells being observed were closest to the cover slip, which was supported by four layers of electrical tape (giving a thickness of 0.5 mm). This squeezed the yolk of the embryo and held the embryo securely. Devel-opment was unaffected even if the embryo was held in this way for several hours. Embryos continued to develop somites while under observation, and went on to give swimming juveniles. The number of somites at the beginning and end of the observation period was recorded to gave an indication of the stage of development during observation.

Images were collected with a CCD camera (Hamamatsu 2400-77) and fed into the video input of a BioRad MRC600 confocal microscope. Noise was minimised by averaging several images together, using the exponential averaging filter of the BioRad software. Images were also recorded simultaneously on a time-lapse video-recorder (Panasonic 6010), at a rate of 1 frame/sec. This is 1/25 the normal (real-time) rate.

The results presented in this paper are based on over 100 hours of time-lapse filming of embryos from the 6-somite until the 24-somite stage.

Labelling neural crest cells

Cells were labelled with lysinated tetramethylrhodamine-dextran (TRITC-dextran; Molecular Probes), using the method of Raible et al. (1992). Live embryos were mounted in 1.2% agarose in 10% Hanks saline, and imaged with a confocal microscope.

Quantitation of crest cell motility

For the purposes of quantitation, extension was defined as the amount of new area covered by a protrusion between two time points, while retraction was the amount of area that is lost. A protrusion was defined as a region of a cell which extends from the cell body and has filopo-dial and/or lamellipodial branches, while ‘area’ was the area of the protrusion, as seen in the image plane of the microscope. To measure extension and retraction, a series of single frames were digitized from the time-lapse record with a Macintosh Centris 660 AV. The images were then transferred to a drawing program (Canvas; ver 3.5), where the protrusion being analyzed was outlined and filled in. Next, drawings from two consecutive time-points were superimposed. The area of the protrusion in the later image that lay outside that of the protrusion in the earlier image was measured using NIH Image (public domain software) to get an indication of extension; retraction was measured by the amount of area in the earlier image that lay outside that of the protrusion in the later image. Fig. 1 illustrates the principle behind this quantitation.

Fig. 1.

The principle behind the algorithm used to quantitate motility of neural crest cell protrusions. The amount of extension was obtained by subtracting the image at t0 from that at t1, while the amount of retraction was obtained by subtracting the image at t1 from that at t0.

Fig. 1.

The principle behind the algorithm used to quantitate motility of neural crest cell protrusions. The amount of extension was obtained by subtracting the image at t0 from that at t1, while the amount of retraction was obtained by subtracting the image at t1 from that at t0.

spadetail analysis

Homozygous mutants were obtained by crossing heterozygotes, which were provided by Charles Kimmel. Since mutant embryos lack proper somites, they were staged both by the shape of the yolk and by comparison with wild-type siblings that were grown at the same temperature.

To quantitate the position of neural crest cells in the trunk of embryos, the distance of their nuclei from the edge of the neural tube was measured. The cells were then grouped in terms of number of cell diameters. The value of one cell diameter was measured to be 6.1 µm (average of 28 measurements). Positions were measured for 10 mutant embryos and 5 wild-type siblings at the 13-to 15-somite stage in the region corresponding to the first four somites.

HNK-1 labelling

Embryos were fixed and permeabilized for antibody staining (Westerfield, 1993), incubated in 1:1000 anti-HNK-1 (Sigma) in 1% BSA/PBS overnight at 4°C, given two 1-hour washes in PBS, then incubated in 1:100 FITC goat anti-mouse IgM (µ-chain specific; Sigma) in PBS, again overnight at 4°C. After several rinses in PBS, embryos were mounted in Citifluor (UKC Chem. Lab.) and viewed with a confocal microscope.

Neural crest cells in the living zebrafish embryo can be imaged with high resolution

The zebrafish embryo is transparent. With careful mounting and contrast-enhanced video microscopy, neural crest cells within the intact embryo can be visualized in detail, so that fine protrusions such as filopodia can be detected (Fig. 2). In the anterior trunk, neural crest cells are first visible at the 10-to 12-somite stages, when they emerge from the neural tube and occupy a position dorsal to somites. That these are indeed neural crest cells was confirmed by labelling them with TRITC-dextran and tracking them for at least 24 hours. As shown in Fig. 3, they moved as neural crest cells have been reported to do (Raible et al., 1992). Specifically, all 17 cells (in 13 embryos) that were labelled at the level of somites 1-6, in embryos up to the 14-somite stage, migrated ventrally in the medial pathway (Fig. 4).

Fig. 2.

Neural crest cells in the living zebrafish embryo can be visualized in detail. This image, obtained with differential interference contrast optics, shows neural crest cells that have just emerged from the neural tube near the 2nd and 3rd somites of a 11-somite embryo. The white arrowheads indicate their filopodia. Dorsolateral view; nc, neural crest cell. Bar,10 µm.

Fig. 2.

Neural crest cells in the living zebrafish embryo can be visualized in detail. This image, obtained with differential interference contrast optics, shows neural crest cells that have just emerged from the neural tube near the 2nd and 3rd somites of a 11-somite embryo. The white arrowheads indicate their filopodia. Dorsolateral view; nc, neural crest cell. Bar,10 µm.

Fig. 3.

Tracking cell movement in the zebrafish embryo. (A) Lateral view of a 14-somite embryo, with cells (arrowhead) that have just been labelled with TRITC-dextran. (B,C) The cells move ventrally in the medial pathway (note the proximity of the ventral cell to the notochord), consistent with their being neural crest cells. not, notochord. Bar, 100 µm.

Fig. 3.

Tracking cell movement in the zebrafish embryo. (A) Lateral view of a 14-somite embryo, with cells (arrowhead) that have just been labelled with TRITC-dextran. (B,C) The cells move ventrally in the medial pathway (note the proximity of the ventral cell to the notochord), consistent with their being neural crest cells. not, notochord. Bar, 100 µm.

Fig. 4.

Schematic cross-section of a zebrafish trunk, showing the medial and lateral pathways. The first cohort of neural crest cells all migrate in the medial pathway.

Fig. 4.

Schematic cross-section of a zebrafish trunk, showing the medial and lateral pathways. The first cohort of neural crest cells all migrate in the medial pathway.

Neural crest cell protrusions become paralyzed and retract after contacting somites

Early neural crest cells are highly active after emigrating from the neural tube. They extend a large number of filopodia; in one focal plane, an average of 6.6±2.6 (average ± s.d., n=23) were detected per cell, at one time point. These slender struc-tures, measuring 0.3±0.1 µm in diameter (n=17) can expand into thick protrusions measuring 2.0±1.1 μm (n=15) within 1 to 2 minutes. Branches are formed and retracted at a relatively high rate (Fig. 5), and neural crest cells move away from the neural tube by extending protrusions that constantly change direction.

Fig. 5.

Neural crest cell protrusions collapse upon contact with somites. This sequence shows the behaviour of a neural crest cell (arrow) near the 4th somite of a 13-somite embryo. Prior to contact with the somite, the crest cell protrusion branched 5 times (A-F), either by extending filopodia (arrowhead in C, for example) or lamellipodia (F, arrowhead). Upon contact with the somite (H), however, filopodial branches (G, arrowheads) were not thickened, and no further branching occurred; the only apparent activity was a broadening at the tip (I, arrowhead). (J) 1.5 minutes after contact, the protrusion began to thin while the tip remained attached to the somite. As thinning continued, a new region of the cell became motile (K,L, arrowhead). Dorsolateral view, with dorsal at the top and anterior to the left; s, somite. Bar, 10 µm.

Fig. 5.

Neural crest cell protrusions collapse upon contact with somites. This sequence shows the behaviour of a neural crest cell (arrow) near the 4th somite of a 13-somite embryo. Prior to contact with the somite, the crest cell protrusion branched 5 times (A-F), either by extending filopodia (arrowhead in C, for example) or lamellipodia (F, arrowhead). Upon contact with the somite (H), however, filopodial branches (G, arrowheads) were not thickened, and no further branching occurred; the only apparent activity was a broadening at the tip (I, arrowhead). (J) 1.5 minutes after contact, the protrusion began to thin while the tip remained attached to the somite. As thinning continued, a new region of the cell became motile (K,L, arrowhead). Dorsolateral view, with dorsal at the top and anterior to the left; s, somite. Bar, 10 µm.

Strikingly, this active behaviour ceases when the tip of a pro-trusion contacts the surface of a somite (Fig. 5G-I). Very soon after contact, filopodial or lamellipodial branches are no longer produced by the protrusion, and pre-existing filopodial branches do not grow larger. The tip does not extend any further, although it does broaden slightly. The constant cen-tripetal movement of ruffles and inhomogeneities within the protrusion, which is visible during extension, also ceases. After a period of quiescence, the protrusion retracts. In some cells, the tip does not remain attached to the somite during retrac-tion, and only a slight narrowing occurs; in other cells, such as the one shown in Fig. 5, the tip continues to adhere to the somite, causing the protrusion to thin considerably as cytoplasm retracts (Fig. 5J,K). During the period of paralysis and retraction, another region of the cell becomes motile: it extends filopodia and grows into a thick protrusion which actively branches (Fig. 5J-L).

Quantitation of the amount of extension and retraction in protrusions confirms the impression that contact with somites is followed by a rapid loss of motility. These parameters were quantified by measuring the change in the shape and area occupied by protrusions, which is here defined as a region of a cell which extends from the cell body and has filopodial and/or lamellipodial branches. As shown in Fig. 6, the amount of extension and retraction prior to contact is high. Soon after contact, extension decreases to zero; the small value seen immediately after contact reflects the broadening at the tip. Prolonged loss of extension is only seen after contact with a somite; this distinguishes the behaviour of protrusions that retract after contacting somites from the behaviour of those where individual branches retract without contact.

Fig. 6.

Quantitation of the amount of extension (■) and retraction (○) in neural crest cell protrusions indicates that contact with a somite is followed by prolonged loss of extension. These graphs depict activity in a crest cell protrusion near the 4th somite of a 12-somite embryo (A), and near the 5th somite of a 13-somite embryo (B). Rapid extension and retraction occurred prior to contact with the somite, which took place at the time-point indicated by the left arrow. Immediately after contact, the protrusion ceased branching, but its tip thickened; this is manifested here by the lower extension values. At the time-point indicated by the arrow on the right, the entire protrusion began withdrawing into the cell.

Fig. 6.

Quantitation of the amount of extension (■) and retraction (○) in neural crest cell protrusions indicates that contact with a somite is followed by prolonged loss of extension. These graphs depict activity in a crest cell protrusion near the 4th somite of a 12-somite embryo (A), and near the 5th somite of a 13-somite embryo (B). Rapid extension and retraction occurred prior to contact with the somite, which took place at the time-point indicated by the left arrow. Immediately after contact, the protrusion ceased branching, but its tip thickened; this is manifested here by the lower extension values. At the time-point indicated by the arrow on the right, the entire protrusion began withdrawing into the cell.

The response to contact with somites is specific

The paralysis and retraction seen in neural crest cell protru-sions that contacted somites did not occur when neural crest cells contacted one another. In 19 cases of interactions between one neural crest cell and another observed in embryos at the 12-to 15-somite stages, for example, no retraction was seen. Fig. 7 shows an example of such an interaction: instead of retracting, the protrusion of one neural crest cell thickened after contacting another neural crest cell.

Fig. 7.

Neural crest cell protrusions do not become paralyzed nor do they retract after contacting another neural crest cell in vivo. In this sequence, a filopodium from one neural crest cell (black arrowhead) was seen to contact the protrusion of another crest cell (A). The tip of the filopodium broadened at the point of contact (white arrowhead). (B-D) The protrusion thickened and spread along the contacted protrusion, while continuing to extend more filopodial branches (white arrowheads). (E,F) Eventually, the entire cell moved and became aligned with the contacted protrusion. This behaviour is in striking contrast to that seen when neural crest cells contact somites. Bar, 10 µm.

Fig. 7.

Neural crest cell protrusions do not become paralyzed nor do they retract after contacting another neural crest cell in vivo. In this sequence, a filopodium from one neural crest cell (black arrowhead) was seen to contact the protrusion of another crest cell (A). The tip of the filopodium broadened at the point of contact (white arrowhead). (B-D) The protrusion thickened and spread along the contacted protrusion, while continuing to extend more filopodial branches (white arrowheads). (E,F) Eventually, the entire cell moved and became aligned with the contacted protrusion. This behaviour is in striking contrast to that seen when neural crest cells contact somites. Bar, 10 µm.

Repulsion by somites appears to delay entry into the lateral pathway

The finding that crest cell protrusions undergo a specific paralysis and retraction upon contact with somites raises the possibility that the first cohort of neural crest cells in the zebrafish trunk is guided into the medial pathway because of repulsion by somites. For this to be the case, all protrusions that contact the lateral surface of somites, from the start of emi-gration until the beginning of lateral migration several hours later, should retract. Observations in 56 embryos at different stages and axial levels show that this is indeed what happens. In the region of somites 1 to 5, when observed up to the 19-somite stage (prior to lateral migration), the protrusions of all 43 cells that contacted the lateral surface of somites became paralyzed and then retracted. For somites 6-10, retraction was seen in all 14 cases of contact recorded up to the 24-somite stage. Further posteriorly, specifically at somites 14, 19 and 27, 6 interactions were observed and all resulted in retraction. At the stages when lateral migration begins, which is at the 20-somite stage at the anterior somite region, not all protrusions retracted and cells were able to crawl on the lateral surface of somites. In 5 embryos imaged at the 21-to 23-somite stages at the 3rd and 4th somites, for example, 14 interactions with somites were seen and only 4 retractions occurred.

As somites develop, their ability to repel neural crest cells, or the sensitivity of crest cells to the repulsive signal, is gradually lost. One indicator of this is the gradual increase in the time between contact and retraction. In embryos at the 10-to 13-somite stage, for example, the average time for retrac-tion of crest cell protrusions contacting the lateral surface of somites 1-5 was 2.6±1.1 minutes (n=14), while the average time measured for embryos at the 14-to 19-somite stage was 4.1±2.8 minutes (n=10). A further indicator of a gradual loss in repulsiveness is that only partial retraction occurs at later stages. In one 20-somite embryo imaged in the region of the 4th somite, for example, only the branch of the protrusion that contacted the somite retracted. Three other branches remained active, judging by their ability to continue extending.

If neural crest migration is prevented by an inhibitory cue on somites, then early entry into the medial pathway will occur if the medial face of somites loses its inhibitory property before the lateral face. During the early stages of migration, which is at the 10-to 12-somite stages, protrusions that contacted the medial surface of somites 2-5 retracted quickly, in less than 3 minutes, in a similar manner to protrusions con-tacting the lateral surface (5 interactions observed). But, 2 hours after the start of emigration (at the 13-to 14-somite stages), retraction was seen in only 4 of the 8 interactions observed; in these, the time from contact to retraction was longer than that for younger embryos (average of 8.4±4.5 minutes, n=4). In the other 4 cases of contact, the protru-sions did not retract, but continued to extend branches and to thicken for over 1 hour (which was the period of recording), indicating that the medial surface of the somite was no longer inhibitory to migration. This was confirmed by observations of another 5 interactions at the 15-to 17-somite stages, where none of the protrusions withdrew. Protrusions that contacted the lateral face at this time, however, still became paralyzed and retracted. These observations support the proposal that the entry of neural crest cells into the lateral pathway is delayed because the lateral faces of somites retain their ability to trigger retraction longer than the medial faces.

Neural crest cells enter the lateral pathway precociously in the absence of somites

Neural crest cell protrusions were seen to become paralyzed and withdraw only when their tips contacted somites, and not when they were visibly separated from the somites, implying that retraction does not occur in the absence of somites. Hence, one way to test whether repulsion by somites prevents entry into lateral pathway would be to observe the migration of neural crest cells in embryos where somites have been partially deleted. Here, somites were removed using the spadetail (spt-1) mutation, which affects the movement of a subset of mesoderm cells during gastrulation and thereby deletes entire or parts of somites in the anterior trunk (Kimmel et al., 1989).

Sixteen such mutant embryos, from the 10-somite to 24-hour stages, were examined with differential interference contrast optics to determine the position of neural crest cells. In embryos at the 10-somite stage, crest cells had not yet emerged from the neural tube and the only cells visible in the region adjacent to the neural tube and notochord were somite cells, which were tightly clumped into groups of fragmentary somites. At the 13-to 15-somite stage, migrating crest cells were visible in the anterior trunk. These cells could be easily distinguished from somite cells by their mesenchymal mor-phology and by their origin in the neural tube, as seen in time-lapse videos. They are also morphologically distinct from other mesodermal mesenchymal cells, which are located in more ventral and lateral positions. In the ten mutants examined, only a subset of crest cells were located immediately next to the neural tube. The remaining cells were aberrantly located in the gaps between somite fragments and also in the spaces between somite fragments and the ectoderm (Fig. 8A, arrowheads).

Fig. 8.

Partial deletion of somites, caused by the spt-1 mutation, leads to precocious migration in the lateral pathway. (A) Neural crest cells (arrowheads) stream laterally between somite fragments and the ectoderm in this mutant, in the region where somites 1-3 should be. (B) A diagram of the same mutant, showing positions of all crest cells from several different focal planes projected onto a single plane. (C) Neural crest cells in wild-type embryos, in contrast, were always found beside the neural tube, as shown in this 14-somite sibling. (D) A diagram of the wild-type embryo, showing positions of crest cells. (E) A plot of the average number of neural crest cells as a function of the distance from the neural tube, for both spt-1 and wild-type embryos. A and C are dorsolateral views. nt, neural tube; s, somite. Bar, 10 µm.

Fig. 8.

Partial deletion of somites, caused by the spt-1 mutation, leads to precocious migration in the lateral pathway. (A) Neural crest cells (arrowheads) stream laterally between somite fragments and the ectoderm in this mutant, in the region where somites 1-3 should be. (B) A diagram of the same mutant, showing positions of all crest cells from several different focal planes projected onto a single plane. (C) Neural crest cells in wild-type embryos, in contrast, were always found beside the neural tube, as shown in this 14-somite sibling. (D) A diagram of the wild-type embryo, showing positions of crest cells. (E) A plot of the average number of neural crest cells as a function of the distance from the neural tube, for both spt-1 and wild-type embryos. A and C are dorsolateral views. nt, neural tube; s, somite. Bar, 10 µm.

The presence of cells in the region lateral to somite fragments suggests that early entry into the lateral pathway is possible in the absence of contact with somites. In wild-type embryos at these stages, in contrast, all crest cells were located adjacent to the neural tube (Fig. 8C). To quantitate the distri-bution of crest cells in mutant embryos, and to assess the dif-ference in migration between mutant and wild-type embryos, the position of crest cells in various different planes of focus was projected onto one plane, as shown in Fig. 8B and D, and their distance from the neural tube measured. This analysis, which is shown in Fig. 8E, indicates that approximately half of the crest cells (62 out of 149) are ectopically located in mutant embryos.

The ectopically located crest cells were always found around but not on the surface of somite fragments themselves, as though there is a localized inhibition on the somite fragments. Time-lapse observation of neural crest migration in the anterior trunk of two mutant embryos at the 13-somite stage confirms this: neural crest protrusions that contacted somite fragments became paralyzed and retracted.

Since the spadetail mutation is known to affect the direction of cell movement during gastrulation (Ho and Kane, 1990), it is conceivable that the aberrant migration of neural crest cells is caused directly by the spt-1 mutation acting in the neural crest cells, and not by the defect in somites, which is a secondary effect of the mutation. To examine this possibility, the position of early neural crest cells was examined in the posterior trunk of mutant embryos, where somites are normal in size. In 3 mutant embryos examined at the 24-hour stage, crest cells at the level of somites 22-25 were all located adjacent to the neural tube, just as in wild-type embryos. This suggests that the aberrant migration in mutant embryos is due to the defect in somites, and not to a defect in the neural crest cells themselves.

Neuronal growth cones enter the lateral pathway before neural crest cells

The preceding observations suggest that neural crest entry into the lateral pathway is delayed by a collapse-inducing signal on the surface of somites, rather than a complete inability of the environment to support migration, because neural crest cells that did not contact somites in mutants were able to migrate pre-cociously in the gap between somite and ectoderm. If the inhibition is specific to neural crest cells, then other cells which do not detect the signal should be able to migrate in the pathway before neural crest cells do so. In the course of observing neural crest migration, it was found that some growth cones, presumably of Rohon-Beard cells – sensory neurons with cell bodies in the dorsal part of the neural tube – do exactly this. In 4 embryos, at approximately the 17-somite stage, growth cones were seen migrating past the dorsolateral surface of somites without retracting, leaving their axons on the region of the somite that neural crest cells are unable to migrate over (Fig. 9A). The identity of these lateral migrating growth cones was confirmed by examining embryos labelled with the HNK-1 antibody, which binds to the entire surface of Rohon-Beard cells (Metcalfe et al., 1990), including their rostro-caudal processes within the neural tube. In the 9 embryos examined at the 18-somite stage, five Rohon-Beard axons were seen extending approximately 10 μm into the lateral pathway, while one extended more than 40 μm (Fig. 9B). Rohon-Beard growth cones are thus able to migrate in the lateral pathway (Fig. 9C-E) before neural crest cells.

Fig. 9.

Early inhibition within the lateral pathway is specific to neural crest cells, as growth cones can migrate in this pathway before neural crest cells. (A) An axon (arrow) extends laterally to the 4th somite in a 17-somite embryo, while neural crest cells (arrowheads) are all unable to enter. (B) A HNK-1 labelled axon (arrow) and growth cone in another 17-somite embryo, extending laterally to the 3rd somite. (C-E) Different planes of focus at the 17th somite of a 22-somite embryo, showing that the growth cones migrate in the lateral pathway, i.e. between somites and the ectoderm. At a deep plane (C), a Rohon-Beard axon is visible (arrowhead). At a slightly shallower plane (D), ectodermal cell nuclei are visible (arrowhead), while at an even shallower plane (E), enveloping layer cells are visible. All embryos are shown in dorsolateral view. s, somite; nt, neural tube. Bar, 20 µm for A and C-E; 25 µm for B.

Fig. 9.

Early inhibition within the lateral pathway is specific to neural crest cells, as growth cones can migrate in this pathway before neural crest cells. (A) An axon (arrow) extends laterally to the 4th somite in a 17-somite embryo, while neural crest cells (arrowheads) are all unable to enter. (B) A HNK-1 labelled axon (arrow) and growth cone in another 17-somite embryo, extending laterally to the 3rd somite. (C-E) Different planes of focus at the 17th somite of a 22-somite embryo, showing that the growth cones migrate in the lateral pathway, i.e. between somites and the ectoderm. At a deep plane (C), a Rohon-Beard axon is visible (arrowhead). At a slightly shallower plane (D), ectodermal cell nuclei are visible (arrowhead), while at an even shallower plane (E), enveloping layer cells are visible. All embryos are shown in dorsolateral view. s, somite; nt, neural tube. Bar, 20 µm for A and C-E; 25 µm for B.

In this paper, a technique for high resolution time-lapse filming of neural crest cells inside the intact zebrafish embryo is described. This technique is used to identify a guidance mechanism that acts on trunk neural crest cells, delaying entry into the lateral pathway.

By observing neural crest cells with time-lapse video-microscopy, active protrusions were found to become paralyzed and retract after contacting somites. This response occurred dramatically in all interactions between neural crest cells and young somites. It involves the formation of a tentative adhesion, rather than the mere absence of adhesion, since pro-trusions can be seen to thin even while their tips remains attached to the somite. Only a limited part of the cell is paralyzed, in that protrusions can extend and branch in other regions of the cell. A similar sequence of contact, adhesion, paralysis, retraction and motility elsewhere in the cell has been previously described for fibroblasts (Abercrombie, 1970) and neuronal growth cones (Kapfhammer and Raper, 1987) in vitro. This behaviour, which is termed contact inhibition of locomotion or collapse, is a response to specific signals. Growth cones of dorsal root ganglia, for example, collapse only upon contact with axons of retinal ganglion cells, and not when they contact other dorsal root ganglia (Kapfhammer and Raper, 1987); retinal ganglion axons from the posterior retina collapse upon contact with membranes from the posterior, but not anterior, tectum (Cox et al., 1990). The response in zebrafish trunk neural crest cells also appears specific, as it is seen only when they contact young somites and not when they contact one another. The term ‘collapse’ is used here inter-changeably with ‘contact inhibition’; collapse describes the phenomenon – i.e. rapid retraction of entire protrusions – more accurately, but is usually associated with neuronal growth cones, whereas contact inhibition has been extensively used for fibroblastic cells such as neural crest cells. Since both involve a similar loss of activity in the leading edge, which is func-tionally and structurally homologous in motile cells and neuronal growth cones (Trinkaus, 1984), it is highly likely that the underlying cytoskeletal responses are similar (Keynes and Cook, 1990).

The period in which collapse/contact inhibition occurs in neural crest cells suggests that inhibition by somites provides at least one mechanism by which neural crest entry into the lateral pathway is delayed. Initially, all protrusions that contact somites become paralyzed and retract, and cells remain in the space dorsal to somites. As the embryo develops, only partial collapse occurs. Eventually, protrusions no longer retract and the cells move ventrally. Since the lateral surface of somites continues to be inhibitory when the medial surface no longer is, the first cells to move all enter the medial pathway.

If entry into the lateral pathway is indeed delayed by a specific inhibitory signal, then removal of the signal or absence of sensitivity to the signal should lead to early entry into the lateral pathway. The spadetail mutation effectively removes the signal in restricted regions of the embryo by partially deleting somites. This causes neural crest cells to enter the lateral pathway precociously, except where somite fragments are present, confirming that a cue on somites delays entry. Rohon-Beard cells provide a way of assessing the behaviour of cells lacking sensitivity to the inhibitory signal. Growth cones from these cells do not retract upon encountering the dorsolateral somite surface, at stages when neural crest cells do, and are able to migrate in the lateral pathway. This indicates that there is no general barrier to migration, such as an extracellular matrix incapable of supporting migration or an absence of physical space.

In the avian embryos, neural crest cells are thought to be guided by inhibitory signals which act by preventing adhesion to the extracellular matrix. Chondroitin sulphate proteoglycan, for example, which interferes with neural crest cell attachment to fibronectin and thereby inhibits their migration (Erickson and Turley, 1983; Perris and Johansson, 1987), appears respon-sible for the exclusion of neural crest cells from the perinoto-chordal region (Newgreen et al., 1986). This molecule is also present in the lateral pathway (Oakley et al., 1994), at the period when neural crest cells do not enter, so it is possible that an ‘anti-adhesion’ mechanism operates here. Another molecule, which binds peanut agglutinin (PNA), is also tran-siently present in this pathway (Oakley et al., 1994). This molecule has been proposed to guide neural crest cells by repulsion, because a PNA-binding molecule that can induce collapse of neuronal growth cones has been found in the regions of the sclerotome that are devoid of motor axons (Davies et al., 1990). However, it is not known if the PNA-binding molecule in the lateral pathway is able to induce a similar collapse response in neural crest cells. Hence, up to now, there has only been relatively weak evidence that neural crest cells are guided by signals that trigger paralysis and retraction.

The observations reported here suggest that the pattern of neural crest migration in the zebrafish trunk may be dependent on the medial surface of somites losing their inhibitory property before the lateral surface. How could this be possible? In the zebrafish, it is known that the notochord induces differ-entiation in somites. In particular, muscle pioneers are first induced in the medial cells adjacent to the notochord (van Raamsdonk et al., 1978; Halpern et al., 1993). A wave of differentiation then spreads from the medial region, so that cells on the lateral surface differentiate later than those on the medial surface (Waterman, 1969). If loss of inhibition to neural crest migration is one consequence of this differentiation, then this medial to lateral wave can explain the early difference between the two surfaces as well as the eventual loss of inhi-bition on the lateral surface. This hypothesis can be tested by examining the pattern of neural crest migration in mutants lacking muscle pioneers, such as floating head and no tail (Halpern et al., 1993).

Changes in production of the signal by somite cells may be accompanied by change in responsiveness to the signal by crest cells, reinforcing the effect. Indeed, in avian embryos, at least, late-migrating neural crest cells appear insensitive to the signal that inhibits migration of early crest cells, as indicated by a recent experiment: late migrating quail neural crest cells that were transplanted into young chick embryos were found to pre-cociously enter the lateral pathway (Erickson and Goins, 1995).

Based on observations of avian neural crest cells in vitro, it has been proposed that contact inhibition occurs when neural crest cells contact one another and that mutual repulsion is what drives neural crest dispersal from the neural tube, by causing cells to move away from one another (Erickson, 1985). As described in this paper, however, there is no retrac-tion when neural crest cells contact one another in vivo (although there is cessation of extension, or contact paralysis, when lamellipodia contact another crest cell in vitro; Jesuthasan, unpublished data). Contact inhibition, which is defined as paralysis followed by retraction (Abercrombie, 1970), appears to occur specifically upon contact with somites. As in neuronal growth cones (Cox et al., 1990; Davies et al., 1990), this response seems to be involved in guidance, rather than in dispersal.

In neuronal growth cones, collapse-inducing molecules act by ultimately triggering depolymerization of the actin and microtubule cytoskeleton, as indicated by findings that collapse can be mimicked by depolymerizing drugs such as cytochal-asin (Fan et al., 1993) and viniblastine (Tanaka et al., 1995). Several collapse-inducing molecule, such as collapsin (Luo et al., 1993), have already been identified, and there is now evidence that some collapse-inducing signals are transduced by Ca2+ (Bandtlow et al., 1993) and/or G-proteins (Igarashi et al., 1993). Indeed, the list of molecular players involved in growth cone repulsion is steadily increasing, with a recent finding that a member of the Eph family of receptor tyrosine kinases is involved in collapse of retinal ganglion cell growth cones (Drescher et al., 1995).

Despite our extensive knowledge of neural crest migration pathways and of their derivatives, there is still very little under-standing of how these cells are guided, either at the level of cell behaviour or at the level of molecules. This study, by pre-senting evidence that crest cells are guided by inhibitory cues that trigger contact inhibition or collapse, provides a foun-dation and a starting point for molecular analysis of neural crest pathfinding. It seems, moreover, that neural crest cells and neuronal growth cones are all susceptible to the same type of guidance, encouraging the view that they may all be steered by fundamentally similar mechanisms.

I thank Julian Lewis for encouraging this work; Phil Ingham for the fish facility; Uwe Strähle and Patrick Blader for the embryos; Richard Gardner for the CCD; Catherine Haddon for transmitting the photographs; the Embryology Course at Woods Hole for the intro-duction to video microscopy, and the Rhodes Trust, ICRF and Balliol College, Oxford, for financial support.

Abercrombie
,
M.
(
1970
).
Contact inhibition in tissue culture
.
In Vitro
6
,
128
142
.
Bandtlow
,
C. E.
,
Schmidt
,
M. F.
,
Hassinger
,
T. D.
,
Schwab
,
M. E.
and
Kater
,
S. B.
(
1993
).
Role of intracellular calcium in NI-35-evoked collapse of neuronal growth cones
.
Science
259
,
80
83
.
Bronner-Fraser
,
M.
(
1986
).
Analysis of early stages of trunk neural crest migration in avian embryos by monoclonal antibody HNK-1
.
Dev. Biol
.
115
,
44
55
.
Collazo
,
A.
,
Bronner-Fraser
,
M.
and
Fraser
,
S. E.
(
1993
).
Vital dye labelling of Xenopus laevis trunk neural crest reveals multipotency and novel pathways of migration
.
Development
118
,
363
376
.
Cox
,
E. C.
,
Müller
,
B.
and
Bonhoeffer
,
F.
(
1990
).
Axonal guidance in the chick visual system: posterior tectal membranes induce collapse of growth cones from the temporal retina
.
Neuron
4
,
31
37
.
Davies
,
J. A.
,
Cook
,
G. M. W.
,
Stern
,
C. D.
and
Keynes
,
R. J.
(
1990
).
Isolation from chick somites of a glycoprotein fraction that causes collapse of dorsal root ganglion growth cones
.
Neuron
4
,
11
20
.
Drescher
,
U.
,
Kremoser
,
C.
,
Handwerker
,
C.
,
Löschinger
,
J.
,
Noda
,
M.
and
Bonhoeffer
,
F.
(
1995
).
In vitro guidance of retinal ganglion cell axons by RAGS, a 25kDa tectal protein related to ligands for Eph receptor tyrosine kinases
.
Cell
82
,
359
370
.
Erickson
,
C. A.
(
1985
).
Control of neural crest cell dispersion in the trunk of the avian embryo
.
Dev. Biol
.
111
,
138
157
.
Erickson
,
C. A.
,
Duong
,
T. D.
and
Tosney
,
K. W.
(
1992
).
Descriptive and experimental analysis of the dispersion of neural crest cells along the dorsolateral path and their entry into ectoderm in the chick embryo
.
Dev. Biol
.
77
,
251
272
.
Erickson
,
C. A.
and
Goins
,
T. L.
(
1995
).
Avian neural crest cells can migrate in the dorsolateral path only if they are specified as melanocytes
.
Development
121
,
915
924
.
Erickson
,
C. A.
and
Turley
,
E. A.
(
1983
).
Substrata formed by combinations of extracellular matrix components alter neural crest cell motility in vitro
.
J. Cell Sci
.
61
,
299
323
.
Fan
,
J.
,
Mansfield
,
S. G.
,
Redmond
,
T.
,
Gordon-Weeks
,
P. R.
and
Raper
,
J.
(
1993
).
The organization of f-actin and microtubules in growth cones exposed to a brain-derived collapsing factor
.
J. Cell Biol
.
121
,
867
878
.
Halpern
,
M. E.
,
Ho
,
R. K.
,
Walker
,
C. W.
and
B., K. C
. (
1993
).
Induction of muscle pioneers and floor plate is distinguished by the zebrafish no tail mutation
.
Cell
75
,
99
111
.
Ho
,
R. K.
and
Kane
,
D. A.
(
1990
).
Cell-autonomous action of zebrafish spt-1 mutation in specific mesodermal precursors
.
Nature
348
,
728
730
.
Igarashi
,
M.
,
Strittmatter
,
S. M.
,
Vartanian
,
T.
and
Fishman
,
M. C.
(
1993
).
Mediation by G proteins of signals that cause collapse of growth cones
.
Science
259
,
77
79
.
Kapfhammer
,
J. P.
and
Raper
,
J. A.
(
1987
).
Collapse of growth cone structure on contact with specific neurites in culture
.
J. Neurosci
.
7
,
201
212
.
Keynes
,
R.
and
Cook
,
G.
(
1990
).
Cell-cell repulsion: clues from the growth cone?
Cell
62
,
609
610
.
Kimmel
,
C. B.
,
Kane
,
D. A.
,
Walker
,
C.
,
Warga
,
R. M.
and
Rothman
,
M. B.
(
1989
).
A mutation that changes cell movement and cell fate in the zebrafish embryo
.
Nature
337
,
358
362
.
Le Douarin
,
N.
(
1982
).
The Neural Crest
.
Cambridge, UK
:
Cambridge University Press
.
Luo
,
Y.
,
Raible
,
D.
and
Raper
,
J. A.
(
1993
).
Collapsin: a protein in brain that induces the collapse and paralysis of neuronal growth cones
.
Cell
75
,
217
227
.
Metcalfe
,
W. K.
,
Myers
,
P. Z.
,
Trevarrow
,
B.
,
Bass
,
M. B.
and
Kimmel
,
C. B.
(
1990
).
Primary neurons that express the L2/HNK-1 carbohydrate during early development in the zebrafish
.
Development
110
,
491
504
.
Newgreen
,
D. F.
,
Scheel
,
M.
and
Kastner
,
V.
(
1986
).
Morphogenesis of sclerotome and neural crest in avian embryos. In vivo and in vitro studies on the role of notochordal extracellular material
.
Cell Tiss. Res
.
244
,
299
313
.
Oakley
,
R. A.
,
Lasky
,
C. J.
,
Erickson
,
C. A.
and
Tosney
,
K. W.
(
1994
).
Glycoconjugates mark a transient barrier to neural crest emigration in the chicken embryo
.
Development
120
(
1
),
103
114
.
Perris
,
R.
and
Johansson
,
S.
(
1987
).
Amphibian neural crest cell migration on purified extracellular matrix components: a chondroitin sulphate proteoglycan inhibits locomotion on fibronectin substrates
.
J. Cell Biol
.
105
,
2511
2521
.
Raible
,
D. W.
and
Eisen
,
J. S.
(
1994
).
Restriction of neural crest cell fate in the trunk of the embryonic zebrafish
.
Development
120
,
495
503
.
Raible
,
D. W.
,
Wood
,
A.
,
Hodsdon
,
W.
,
Henion
,
P. D.
,
Weston
,
J. A.
and
Eisen
,
J. S.
(
1992
).
Segregation and early dispersal of neural crest cells in the embryonic zebrafish
.
Dev. Dynamics
195
,
29
42
.
Schilling
,
T. F.
and
Kimmel
,
C. B.
(
1994
).
Segment and cell type lineage restrictions during pharyngeal arch development in the zebrafish embryo
.
Development
120
,
483
494
.
Serbedzija
,
G. N.
,
Bronner-Fraser
,
M.
and
Fraser
,
S. E.
(
1989
).
A vital dye analysis of the timing and pathways of avian neural crest cell migration
.
Development
106
,
809
819
.
Serbedzija
,
G. N.
,
Fraser
,
S. E.
and
Bronner-Fraser
,
M.
(
1990
).
Pathways of trunk neural crest cell migration in the mouse embryo as revealed by vital dye labeling
.
Development
108
,
605
612
.
Tanaka
,
E.
,
Ho
,
T.
and
Kirschner
,
M. W.
(
1995
).
The role of microtubule dynamics in growth cone motility and axonal growth
.
J. Cell Biol
.
128
,
139
155
.
Thiery
,
J. P.
,
Duband
,
J. L.
and
Delouvée
,
A.
(
1982
).
Pathways and mechanisms of avian trunk neural crest cell migration and localization
.
Dev. Biol
.
93
,
324
343
.
Trinkaus
,
J. P.
(
1984
).
Cells Into Organs: The Forces That Shape The Embryo
.
Englewood Cliffs, NJ
:
Prentice-Hall
.
van Raamsdonk
,
W.
,
Pool
,
C. W.
and
te Kronnie
,
G.
(
1978
).
Differentiation of muscle fiber types in the teleost, Brachydanio rerio
.
Anat. Embryol
.
153
,
137
155
.
Waterman
,
R. E.
(
1969
).
Development of the lateral musculature in the teleost, Brachydanio rerio: A fine structural study
.
Am. J. Anat
.
125
,
457
494
.
Westerfield
,
M.
(Ed.). (
1993
).
The Zebrafish Book (2nd ed
.).
Eugene, OR
:
University of Oregon Press
.
Weston
,
J. A.
(
1963
).
A radioautographic analysis of the migration and localization of trunk neural crest cells in the chick
.
Dev. Biol
.
6
,
279
310
.