ABSTRACT
Neural crest-derived pigment cells form species-specific patterns of pigmentation in amphibian embryos. We have characterized the appearance and changes in pigment cell distribution in the embryos of the California newt, Taricha torosa. Black melanophores first appear scattered over the surface of the somites intermingled with yellow xanthophores in stage 34/35 embryos. The melanophores then migrate either dorsally to form a dorsal stripe at the apex of the somites or ventrally along the intersomitic furrows to form a midbody stripe at the somite-lateral plate mesoderm border. Xanthophores remain between the two melanophore stripes and are also found in the dorsal fin and head. The formation of the dorsal stripe coincides with a change in melanophore tissue affinity from the surface of the somites to the subectodermal extracellular matrix (ECM). The latter substratum is the location of the cue used to organize the dorsal stripe. In addition, melanophores become elongate and highly arborized, which would allow them to extend to the region where the dorsal stripe forms. In contrast, xanthophores do not form long processes in vitro. This suggests that the ability of melanophores but not xanthophores to search for a cue at the apex of the somites may account in part for the segregation of these cells types. Melanophores and xanthophores are trapped to form the midbody stripe by the pronephric duct, which is located just beneath the ectoderm at the bases of the intersomitic furrows. Ablation of the duct prevents formation of the midbody stripe, although melanophores and xanthophores still fail to migrate ventrally over the lateral plate mesoderm. Melanophores grafted to the ventral midline fail to leave the confines of the donor tissue. This suggests that a factor in the lateral plate mesoderm in addition to the pronephric duct is inhibiting further ventral migration. There is no gross morphological difference in the organization of the subectodermal ECM dorsal and ventral to the pronephric duct as revealed by alcian blue, ruthenium red and staining with antibodies to fibronectin. We also conclude that the directed dispersal of the neural crest into the space between the somites and ectoderm is due to contact inhibition of cell movement, since T. torosa neural crest cells demonstrate contact inhibition in vitro and there are enough cells in the lateral migratory spaces to make contact events likely during dispersal.
INTRODUCTION
The trunk neural crest is a transient population of mesenchymal cells that appears near the dorsal surface of the neural tube soon after neurulation is complete. In amphibian embryos, these cells migrate from their site of origin along three major pathways: dorsally, into the expanding dorsal fin; ventrally, between the somites and the neural tube; and laterally, between the ectoderm and the somites (for reviews see Hörstadius, 1950; Weston, 1980; Le Douarin, 1982; Erickson, 1986).
Among the derivatives of the amphibian neural crest are three types of pigment cells (DuShane, 1935; Stevens, 1954): black or brown melanophores; yellow xanthophores; and iridophores, which contain reflective organelles (see Bagnara et al. 1979; Frost, Epp & Robinson, 1984a,b). These cells become localized in the embryonic dermis to form species-specific patterns of pigmentation.
There is a substantial literature on the development of pigment cell patterns in the three species of newts belonging to the genus Taricha (formerly Trituras). Taricha torosa, the California newt, has two melanophore stripes. A dorsal stripe runs along the base of the dorsal fin and a second, less distinct, stripe is found at the somite-lateral plate mesoderm border. T. granulosa also has a black dorsal stripe, but, unlike T. torosa, melanophores are found ventrally over the surface of the lateral plate mesoderm. This pattern resembles the distribution of melanophores in the European newt Trituras alpestris (Epperlein, 1982; Epperlein & Claviez, 1982a,b). Finally, Taricha rivularis lacks melanophore stripes altogether. Instead, the melanophores remain as individual cells scattered over the surface of the somites and the dorsal portion of the yolk mass.
Twitty (1936,1944, 1945, 1953) and his colleagues (Twitty & Bodenstein, 1939, 1944; Twitty & Niu, 1948, 1954) attempted to determine what aspects of the species-specific melanophore patterns in T. torosa, T. rivularis and T. granulosa are controlled by environmental influences (e.g. regions of differential adhesion or regional control of melanophore differentiation) and to what extent differences in the melanophores themselves can account for pattern formation in a series of hybridization, grafting and tissue culture experiments. In brief, Twitty and his associates concluded that formation of the dorsal stripe of melanophores found in T. torosa and T. granulosa but not T. rivularis is the result of differences in the adhesive properties of the melanophores in these three species and not their environment. For example, melanophores from T. rivularis do not form stripes when orthotopically grafted into T. torosa, whereas T. torosa melanophores form a dorsal stripe when T. rivularis is the host. These results were supported with tissue culture experiments conducted by Twitty & Bodenstein (1939), who observed that T. torosa melanophores form clusters in culture, whereas melanophores from T. rivularis do not. In contrast, the midbody melanophore stripe found only in T. torosa is due to an environmental cue, since neither T. rivularis nor T. granulosa melanophores migrate ventrally over the yolk in T. torosa, and T. torosa melanophores migrate over the yolk mass in the other two species. Twitty (1936) hypothesized that the cue responsible for the formation of the midbody stripe in T. torosa is the abrupt angle taken by the ectoderm and lateral plate mesoderm as it begins to surround the bulging yolk mass. Finally, Twitty & Niu (1948, 1954) concluded from a series of in vitro experiments that the dispersal of pigment cells from the neural tube is the result of negative chemotaxis.
We have reexamined these conclusions using methods that were not available to earlier researchers in order to determine the mechanisms underlying pigment cell pattern formation in the California newt, T. torosa. These techniques, which include the identification of weakly pigmented xanthophores in situ by fluorescence microscopy, histological and immunocytochemical localization of extracellular matrix (ECM) components, time-lapse videomicroscopy and electron microscopy, have permitted us to elucidate the mechanisms of dorsal and midbody melanophore stripe formation, the segregation of different types of pigment cells and the dispersal of pigment cell precursors from their site of origin.
MATERIALS AND METHODS
Taricha torosa torosa (Twitty, 1942) egg clusters were collected from ponds in the San Francisco Bay area. Embryos were removed from their gelatinous integuments using a razor blade, and the fertilization envelopes were removed with tungsten needles. Embryos were raised in either 20 % Steinberg’s solution or tap water at 16°C. The animals were staged using the development tables of Schreckenberg & Jacobson (1975).
Videomicrography of melanophore pattern development
The appearance and changes in the distribution of melanophores were observed using a Nikon Diaphot inverted microscope with illumination provided by a substage fibre optics light source (Dyonics). Embryos were rinsed in 20% Steinberg’s solution and anaesthetized with a few crystals of ethyl m-aminobenzoate (MS-222; Sigma). For videomicrography, it was necessary to restrain the animals to prevent ciliary gliding movement. This was accomplished by gently placing the head and tail of the embryo into appropriately sized fire-polished capillary tubes. During the video recording, the embryos were kept at room temperature (22°C) to increase the rate of development. Time-lapse video recordings were made with an RCA or MTI videocamera and a Panasonic 8050 VHS videocassette player. Rates of cell movement and patterns of migration were analysed by tracing cell positions on acetate sheets placed over the display of a black and white monitor (Audiotronics).
Patterns of migration were also recorded by 35 mm photography. Photographs were taken at 3 h intervals over a 72 h period with Tri-X (Kodak) black and white film using a Nikon FE camera. Still photography was in some respects superior to videomicrography for recording cell behaviour in vivo over extended periods, because constant illumination ‘bleached’ the pigmentation of melanophores, making it difficult to resolve cell morphology.
Identification of pigment cells
Melanophores were identified as black or grey cells in living embryos, whole mounts and sections for light microscopy. In the transmission electron microscope (TEM) melanophores were identified by the presence of characteristic electron-dense melanosomes (see below).
The distribution of xanthophores was determined by NH4OH-induced autofluorescence using techniques described previously (Epperlein & Claviez, 1982a; Tucker, 1986). In brief, embryos were anaesthetized in MS-222 and immersed in 10 % NH4OH. Specimens were then mounted on a glass slide and viewed with a Zeiss epifluorescence microscope. Immediate observation was necessary since the embryos disintegrated within 15 min of NH4OH treatment. The autofluorescence of pteridines results in xanthophores appearing bluish white when viewed with appropriate filters (365 nm excitation, 395 nm barrier). This technique is specific for xanthophores since in older embryos only externally visible yellow cells fluoresced.
Iridophores were identified in plastic-embedded whole mounts (see below) as lustrous silver cells visible when viewed under bright oblique illumination.
Light and electron microscopy
Embryos (stages 16 to 40+) were embedded in plastic for whole-mount preparations as well as thick and thin sectioning. For whole mounts and light microscopy, specimens were fixed in 2 % glutaraldehyde in 0·1 M-sodium cacodylate buffer for 2 to 4 h at room temperature or overnight at 4°C. Specimens were then rinsed in buffer, dehydrated in ethanol, and infiltrated and embedded in Epon-Araldite. For electron microscopy, specimens were postfixed in 1 % OsO4 in sodium cacodylate buffer for 30 min at room temperature following the buffer rinse. Specimens were either sectioned at 5 μm on a Sorvall MT-2 microtome and stained in drops of 0·05% toluidine blue, or thin sectioned on a Reichert Om U3 ultramicrotome, stained with uranyl acetate and lead citrate, and observed at 60kV or 80 kV in a Philips 410 TEM.
Some embryos (n = 12, stages 36 to 40) were prepared for scanning electron microscopy (SEM). Specimens were fixed in 2% glutaraldehyde in sodium cacodylate buffer overnight at 4°C, rinsed in buffer, and photographed. Embryos were then transferred to wax dishes, pinned, and a rectangular piece of ectoderm was removed with tungsten needles from the side of the embryo. Both the ectoderm and the remainder of the embryo were photographed, dehydrated in ethanol and critical-point dried with CO2 in a Samdri-780A (Tousimis) critical-point drying apparatus. Specimens were then mounted on aluminium stubs with silver paint, sputter coated (Denton) with gold and observed at 20 kV in a Philips 501 SEM.
Identification of glycosaminoglycans
The distribution of glycosaminoglycans (GAG) was determined using alcian blue staining of paraffin sections and ruthenium red staining of embryos prepared for TEM using methods described in detail elsewhere (Turley, Erickson & Tucker, 1985; Tucker, 1986; Tucker & Erickson, 1986b). For alcian blue staining, embryos (n = 7, stages 32 to 40) were fixed overnight in 10 % formaldehyde with or without 0·5 % cetylpyridinium chloride (CPC; Polysciences; see Derby & Pintar, 1978) at room temperature, dehydrated in ethanol, cleared in xylene, and infiltrated and embedded in Paraplast (Lancer). Embryos were serially sectioned at 7μm, rehydrated, treated with 5M-HC1 for 30 s and stained in 1 % alcian blue 8G-X (MCB) in 0·025 M-MgCl2 (pH2·6) overnight. Stained sections were then rinsed, dehydrated, cleared in xylene, and mounted in Permount (Fisher). For ruthenium red staining, embryos (n = 5, stages 38 to 40) were fixed in 2 % glutaraldehyde with 0·1 % ruthenium red (Polysciences, see Luft, 1971) and 0·1% tannic acid (Fisher) in PBS at 4°C. After 30 min, specimens were cut transversely, approximately in half, with a razor blade to facilitate penetration of the fixative and stains, and the specimens were kept in the fixative overnight at 4°C. Embryos were then rinsed in 0·1 M-sodium cacodylate buffer, postfixed in 1 % OsO4 for 30 min at room temperature (if stain was present in the fixative, ruthenium red was also present in the postfixative), rinsed in double-distilled H2O and further prepared for thin sectioning as outlined above. Thin sections were made following thick sectioning of the specimen to a point approximately 50 μm from the cut surface. The types of GAG stained were determined by digestion of the matrix with chondroitinase ABC (Seikagaku Kogyo) or Streptomyces hyaluronidase (Seikagaku Kogyo) using methods modified from Hay & Meier (1974) and reported in detail elsewhere (Tucker, 1986).
Immunocytochemistry
Stage 39/40 T. torosa embryos were fixed in Carnoy fixative for 4h at room temperature, embedded in OCT compound (Miles) and frozen in 2-methyl butane cooled with dry ice. 10 pm sections were cut using a Bright cryostat, air dried, and rinsed three times in phosphate-buffered saline (PBS) and once in PBS with 0-5 % bovine serum albumin (BSA, Sigma). Sections were then incubated in anti-Xmopus plasma fibronectin (FN) anti-sera (1:40; a gift from C. C. Wylie and R. O. Hynes, see Heasman et al. 1981) for Ih at room temperature. Following incubation with the primary antibody, sections were rinsed in PBS and incubated in goat anti-rabbit secondary antibody conjugated with FITC for 30 min at room temperature, rinsed and mounted in glycerol. Some sections were incubated in PBS-BSA for Ih instead of primary anti-sera before incubation in secondary antibody to control for nonspecific background staining. Stained preparations were observed with a Leitz Dialux 20 epifluorescence microscope and photographed using a Leitz Vario-Orthomat camera.
Surgical manipulations
Ablation and grafting operations were carried out under sterile conditions on wax dishes in 100 % Steinberg’s solution. To determine the effects of the ventral environment on melanophore differentiation and migratory behaviour, fragments of neural folds were grafted ventral to the pronephric duct and the resulting ipelanophore distribution was observed. Neural folds and the underlying neurectoderm were excised from the future anterior trunk region of stage 19 neurulae (n = 3) using tungsten needles. The prospective neural tissue was then grafted into a slit made in the ectoderm overlying the lateral plate mesoderm of an anaesthetized stage 23/24 host embryo. The grafted tissue was held in place until healed (approximately 30 min) using a fragment of a glass coverslip. The host embryo was then transferred to 20 % Steinberg’s solution and monitored daily.
Pronephric duct primordia were removed from the right side of stage 22/23 embryos (n = 5) following procedures modified from Poole & Steinberg (1982). It was possible to remove completely the ovoid primordium without noticeably disturbing the underlying somites or removing the overlying ectoderm. After recording the development of melanophore and xanthophore patterns, presumably ductless larvae were embedded in plastic and sectioned at 5 i/m (see above) to confirm the complete removal of the pronephric duct.
In vitro analysis of pigment cell behaviour and morphology
Detailed descriptions of T. torosa neural crest tissue culture methods have been presented elsewhere (Tucker & Erickson, 1986A,B). In brief, fragments of neural folds were excised with tungsten needles from stage 16/17 embryos and explanted onto tissue culture plastic dishes (Corning) containing half-strength Leibovitz’s L-15 medium (GIBCO) with antibiotics (gentamicin sulphate, Sigma; fungizone, GIBCO). To promote the differentiation of melanophores and xanthophores in vitro, 10 % foetal calf serum (FCS; GIBCO) was added to the medium. For other experiments, neural crest cells were cultured in saline alone or with 25 μgml-1 plasma FN (BRL) added to the medium.
Some neural folds were cultured on top of collagen gels (see Elsdale & Bard, 1972; Tucker & Erickson, 1984). Eight parts collagen (Vitrogen 100; Collagen Corporation), 1 part 2× L-15 medium and 1 part 0·142 N-NaOH were combined and diluted with with antibiotics (see above) and FCS. The final concentration of the collagen was approximately 6 %.
Contact behaviour was recorded using videomicroscopy (see above) with a Nikon Diaphot inverted phase-contrast microscope with an ice-chilled stage. Speed of movement was determined using methods described previously (Tucker & Erickson, 1984). In brief, nuclear displacement was traced on acetate sheets at 20min intervals over a 2h period. A Hipad Digitizer was used to enter the positional information into a computer to determine the rates of translocation on different substrata.
RESULTS
The development of melanophore patterns in T. torosa larvae can be separated into three phases: ‘appearance’ (stage 33/34), when the melanophores first become visible beneath the ectoderm in the lateral pathway; ‘segregation’ (stages 35 to 38), when the melanophores form the dorsal and midbody stripes; and the ‘primary pattern’ (stage 39/40), when melanophore motility wains and the dorsal stripe is accentuated by extensive branching of cell processes. The results of timelapse analysis of melanophore pattern formation are described below.
Development of melanophore patterns
Melanophores become visible beneath the transparent ectoderm in T. torosa at stage 33/34 (Fig. 1A). The grey cells appear scattered over the surface of the somites in the lateral pathway, where they have migrated from their site of origin on the neural tube.
As development proceeds, the scattered melanophores segregate into two populations: most of the cells move a short distance (<200 μm) dorsally to form the dorsal stripe and others move ventrally to form the midbody stripe. The cells that move ventrally display rapid (up to 0·30μm min-1), highly directed motility until they reach the somite-lateral plate mesoderm border (Figs 1B,C, 2A). These cells are invariably following intersomitic furrows. 48 h after their initial appearance (Fig. 1D), the melanophores are found evenly spaced in the midbody stripe at the base of the furrows. This orderly arrangement is short-lived.
During the following 24 h, melanophores migrate rostrally and caudally within the stripe along the border of the yolk mass at rates of up to 0·18 μm min-1 (Fig. 2B). Once a cell joins the midbody stripe, it does not venture more ventrally over the yolk mass, nor does it migrate dorsally to join the dorsal stripe. The number of melanophores that are found in the ventral stripe at stage 39 is highly variable in different clutches of embryos. This variability appears to be related to the total number of melanophores that are found in the embryo, since animals with broad, intensely pigmented dorsal stripes tend to have the most cells in the midbody stripe as well. Occasionally an embryo will be found that totally lacks a midbody stripe. These animals invariably have thin dorsal stripes with interspersed gaps and no melanophores over the surface of the somites (see below). A sample of 20 embryos (stage 39/40) from four different clutches had an average of 13·0 ±4·4 melanophores in the midbody stripe between the forelimb bud and the posterior end of the yolk mass.
Frequently individual melanophores are encountered over the surface of the somites between the midbody and dorsal melanophore stripes. These cells are usually, but not always, found in intersomitic furrows, and they tend to be seen in the posterior half of the trunk more frequently than in anterior regions. On average, 2·9 ± 2·3 (n = 17) individual melanophores are found between the stripes (stage 39/40).
A typical primary pattern (the secondary pattern forms as the larva approaches metamorphosis; see Lehman, 1950) of melanophores is illustrated in Fig. 2C.
Distribution of xanthophores and iridophores
Using NH4OH-induced pteridine fluorescence, the distribution of xanthophores has been determined during melanophore pattern formation. Xanthophores are first seen soon after melanophores appear at stage 35. Although there is considerable variability in their number and precise distribution, these cells are generally scattered among the melanophores in the anterior half of the trunk and in the head. As the melanophores become segregated into the dorsal and midbody stripes, xanthophores are seen in the head, dorsal fin, between the melanophore stripes over the surface of the somites and among the melanophores in the midbody stripe (Fig. 3A–C). This codistribution of yellow and black pigment cells in the midbody stripe is in marked contrast to the dorsal stripe, which is completely void of xanthophores. Like melanophores, xanthophores are not seen ventral to the midbody stripe over the surface of the yolk mass.
Iridophores are found in fixed specimens viewed with oblique illumination as early as stage 35. Like xanthophores, there is considerable variation in the number and distribution of these cells from embryo to embryo. Iridophores are found aligned along the dorsal surface of the neural tube in some younger (stage 35) specimens (Fig. 4A), and in slightly older embryos (stage 35/36) iridophores are found deep within the embryo (as determined by optical sectioning) along what appears to be the ventral surface of the notochord or the dorsal aorta. Iridophores are found in the trunk of older embryos (stage 36 to 40) ventral to the midbody melanophore stripe, especially along the ventral midline near the heart (Fig. 4B), but also scattered over (or within) the lateral plate mesoderm.
Pigment cell–tissue affinities and cell morphology in situ and in vitro
Melanophores are initially observed scattered over the somites, although they are restricted to more dorsal portions of the somites in the anterior half of the trunk. When a roughly rectangular piece of ectoderm is peeled from the side of a fixed or anaesthetized embryo with tungsten needles before stage 35, nearly all of the melanophores remain associated with the surface of the somites. During the segregation phase (stages 35 to 38), melanophores remain attached to both the overlying ectoderm and the surface of the somites (Fig. 5A,B). By the time the primary melanophore pattern has developed (stage 40), all of the melanophores remain attached to the overlying ectoderm, even those cells that are in the midbody stripe (Fig. 6A,B).
Cell morphology and cell–substratum relationships were also investigated using SEM early in the ‘segregation’ phase of melanophore pattern formation (stage 35) when the migratory substratum of the melanophores is shifting from the surface of the somites to the ectoderm. Individual melanophores were identified in the SEM from light micrographs of the specimens taken before processing for electron microscopy (e.g. Fig. 5A,B). Melanophores are highly branched cells at this stage, with their processes frequently extending 75 μm to 100 μm from the elongate cell body (Figs7D, 8B). Melanophores are restricted to the region dorsal to the pronephric duct. The only cells that are seen in the space between the lateral plate mesoderm and the overlying ectoderm are large, rounded cells that are identical in size, shape, number and distribution to leukocytes that have been identified by the dopa-reaction and TEM in earlier studies (Tucker & Erickson, 1986c). Cells that remain on the surface of the somites are frequently aligned along the intersomitic furrows; other cells are found associated with the pronephric duct (Fig. 7B), which is located superficially beneath the ectoderm (Fig. 7A). At stage 35, the midlateral line placode primordium is passing between the ectoderm and the ectoderm’s basal lamina in the anterior part of the trunk approximately midway between the dorsal fin and the pronephric duct. The placode primordium displaces cells of the underlying somites to form a groove that is visible in the SEM (Fig. 7B). Melanophores frequently align along the placode primordium at this stage.
The morphology of pigment cells and their precursors changes dramatically between stages 33/34, when they are initially entering the lateral migratory pathway, and stage 36/37, when melanophores are segregating into dorsal and midbody stripes. At the onset of migration (Fig. 8A), these cells have broad lamellae and tend to be roughly rectangular in shape. In contrast, at later stages of development, melanophores tend to be elongate with many long processes (Figs7C, 8B). Numerous filopodia are observed associated with the lateral surfaces of the long processes as well as at the tips of the processes. These tips frequently flare into a fan shape that resembles a neuronal growth cone (Fig. 8B).
Similar changes in melanophore morphology are observed when neural crest cells are maintained in vitro. On tissue culture plastic, neural crest cells lack long processes (Fig. 11 A,B). Soon after melanization, however, melanophores develop long processes extending up to 250μm. Xanthophores do not develop similar processes. Instead these cells remain ovoid (Fig. 9A). When cultured on collagen gels, melanophores show a similar propensity for arborization, unlike xanthophores, which remain round or fusiform (Fig. 9B).
The control of midbody stripe formation
The position of the midbody stripe coincides precisely with the location of the pronephric duct. When viewed in the SEM (Fig. 7A,B) or in thick sections (Fig. 12E), the pronephric duct is found just beneath the surface of the ectoderm from stage 34/35, when melanophores are appearing, to stage 38, when the primary pattern of melanophores is nearing completion. Furthermore, melanophores are frequently observed closely associated with the pronephric duct, especially during the early stages of pattern formation.
At stage 36/37, when melanophores are migrating caudally and rostrally along the midbody stripe, it is possible to cut a slit in the ectoderm just ventral to the pronephric duct and carefully extract the duct with tungsten needles. The melanophores that were part of the midbody stripe remain attached to the excised duct (Fig. 10A,B). This intimates that the duct (or its associated vasculature) is acting as a migratory substratum for melanophores in the midbody stripe, perhaps functioning as a trap for the melanophores dispersing ventrally along the intersomitic furrows.
The pronephric duct is eventually internalized in an anterior–posterior wave, bringing it nearer to the dorsal aorta. Time-lapse analysis of the midbody stripe during this internalization (stages 38 to 40) shows that many melanophores in the midbody stripe remain associated with the pronephric duct and are gradually lost from external view (Fig. 11A,B). This would explain why there tend to be fewer cells in the midbody stripe of older embryos than are found in the stripe during the earlier segregation phase. Melanophores that remain in the midbody stripe are firmly attached to the ectoderm (Fig. 6B).
To test directly the role of the pronephric duct in midbody stripe formation, the pronephric primordium was removed from one side of the embryo before the appearance of melanophores, and the development of melanophore patterns was observed. At stage 37, some melanophores are found ventrally over the lateral plate mesoderm on the flank without a duct (Fig. 12A,B), although most of the melanophores are distributed over the surface of the somites. At stage 38/39, there is no midbody stripe on the side of the embryo without a duct, in contrast to the unoperated side of the embryo (Fig. 12C,D). Instead, melanophores are scattered over the surface of the somites. Melanophores are no longer seen on the dorsal part of the lateral plate mesoderm as they are at earlier stages. The distribution of xanthophores was determined in one duct-ablated embryo at stage 38/39 using NH4OH-induced fluorescence. Xanthophores, like melanophores, are found over the surface of the somites in these animals, and not over the lateral plate mesoderm.
When the experimental animals are serially sectioned, it is clear that the duct is completely removed from the right side of the embryo (Fig. 12E). The remaining duct, which has melanophores associated with it, is unusually large, perhaps to compensate for the loss of function of the missing pronephros.
Grafts to the ventral regions
Melanophore precursors were introduced into the regions ventral to the pronephric duct by grafting neural folds to a site near the ventral midline. Melanophores appear in the graft at the same time as in siblings of the donor. In one of three experiments, melanophores were observed spread along a large blood vessel that had formed at the ventral midline, but in no other cases were melanophores observed to disperse from the donor tissue (Fig. 13A). The melanophores that were associated with blood vessels were no longer visible less than 24 h after they were first observed. It was not determined if these cells died, lost their pigmentation, or migrated out of view into the opaque endoderm. Serial sections through the graft (Fig. 13B) show structures resembling a neural tube, dorsal fin and portions of somite-like mesoderm. Melanophores in the graft, like those in the normal situation, are associated with the neuronal tissue and the ‘dorsal’ margin of the somite-like mesoderm. There are no apparent heterogeneities at the boundaries of the grafted tissue (i.e. scar tissue) that could account for the failure of most of the pigment cells to leave the area of the graft.
Extracellular matrix in the ventral regions of the embryo
Alcian blue staining of paraffin sections reveals GAG in the dorsal fin and around the notochord, ventral neural tube, dorsal aorta and heart. However, there is relatively little staining in the spaces between the somites and the ectoderm as well as ventrally between the lateral plate mesoderm and the ectoderm (Fig. 14). This staining pattern is confirmed by observations of thin sections stained with tannic acid and ruthenium red. Striated collagen fibres coated with proteoglycan aggregates are seen to the same extent in the subectodermal ECM both dorsal and ventral to the pronephric duct (Fig. 15A,B). There is considerably less ruthenium red-positive material in the subectodermal matrix than in the tail fin ECM or near the notochord (Tucker & Erickson, 1986b). Indirect immunofluorescence reveals intense anti-FN staining around blood vessels ventral to the pronephric duct, in the ECM beneath the ectoderm dorsal and ventral to the duct, as well as around the duct itself (Fig. 16A,B).
The effect of fibronectin in vitro on cell morphology and translocation
Since FN is found along the neural crest migratory pathways during pigment cell pattern formation, we wished to determine the response of Taricha neural crest cells to substrata coated with FN in vitro. Our analysis of the effect of FN was limited to unpigmented cells, because FN is found in the serum that must be present for T. torosa neural crest cells to differentiate into pigment cells in culture. Neural crest cells were cultured on tissue culture plastic in saline alone or saline with 25 μg ml-1 plasma FN. The presence of FN in the saline affects both cell morphology and the distance that cells spread from the neural fold explant. In the presence of FN, neural crest cells are more flattened on the substratum, with more extensive lamellipodia than cells in cultures with saline alone (Fig. 17A,B). After 13 days in vitro, neural crest cells migrate significantly farther (P< 0·001) from the edge of the explant in the presence of FN (747 ± 365 μm, n = 103) than cells cultured in saline alone (394 ± 152 μm, n = 57). When FN is present in the medium, the average rate of neural crest cell translocation (0·37 ± 0·14μm min-1, n = 10) is not significantly different from the rate of translocation in saline alone (0·39 ± 0·22μm min-1, n = 7). These rates are similar to the speed of movement of melanophores along the intersomitic furrows in vivo (Fig. 2A).
Contact inhibition of cell movement in vitro
Contact inhibition of cell movement was frequently observed in cultures of neural crest cells. Contact events were especially common and dramatic in cultures containing FN. This is apparently related to the extensive lamellipodia that form when neural crest cells are cultured on FN-coated substrata (Fig. 17A,B). When lamellipodia from two cells touch one another, ruffling of the lamellae ceases (contact paralysis). Soon after, the lamellipodia pull away from one another, and protrusive activity commences elsewhere (Fig. 18A–F). This usually results in the cells that have just collided moving away from the site of contact.
DISCUSSION
A number of mechanisms probably contribute to the development of pigment cell patterns in the embryos and larvae of Taricha torosa. Potential regulatory mechanisms include: contact inhibition of cell movement, contact guidance, mechanical obstruction and differential adhesion. Control of pathways of cell migration, but not regional control of cell differentiation, can account for nearly all aspects of pigment cell pattern formation. However, at least one aspect of this pattern formation apparently involves the control of pigment cell differentiation. Hyaluronate inhibits the differentiation of T. torosa melanophores in vitro (Tucker & Erickson, 1986a,b). Since the dorsal fin, which at stage 40 contains melanoblasts but not melanophores (Tucker & Erickson, 1986b,c), is rich with hyaluronate (Tucker & Erickson, 1986b), the fin ECM may affect pigment cell patterning in part by inhibiting differentiation. It is important to remember, however, that this control may be acting by delaying the expression of a phenotype and not by altering the phenotypic fate of a pigment cell precursor. In urodeles, the phenotype of a pigment cell appears to be determined before the cell migrates from the dorsal surface of the neural tube (Epperlein & Löfberg, 1984).
Control of neural crest cell dispersal
The amphibian neural crest initially appears as a compact cord of cells in a groove in the dorsal surface of the neural tube. These cells disperse from their site of origin by migrating laterally over the surface of the somites, dorsally into the expanding dorsal fin and ventrally into the space between the somites and the neural tube (MacMillan, 1976; Löfberg, Ahlfors & Fällström, 1980; Tucker, 1986). What controls the directed dispersal of the neural crest into the surrounding ECM? Twitty (1944,1953) and Twitty & Niu (1948,1954) approached this question in T. torosa by culturing melanophores in capillary tubes and beneath glass coverslips. In the capillary tubes, two melanophores tended to move away from one another; when translocating beneath a coverslip, melanophores would migrate farther than surrounding cells that were not covered. They concluded that this behaviour was the result of negative chemotaxis, which is the directed migration of cells away from high concentrations of a diffusible factor produced by the pigment cells themselves. Thus, the restricted spaces of a capillary tube or beneath a coverslip are analogous to the neural crest migratory spaces in vivo, where pigment cells would move away from their site of origin (i.e. the site where the factor is at its highest concentration). Recent attempts to repeat these experiments with avian melanophores and freshly isolated neural crest cells have been unsuccessful (Erickson & Olivier, 1983). Since we did not observe the formation of a ‘no man’s land’ between colliding outgrowths of T. torosa unpigmented neural crest or pigment cells as would be predicted by negative chemotaxis (Oldfield, 1963), another mechanism may be responsible for the migration of neural crest cells away from the neural tube over the surface of the somites.
A second possible mechanism for directing the outgrowth of neural crest cells is contact guidance (Weiss, 1945). Löfberg et al. (1980) have reported that collagen fibres tend to be aligned parallel to the direction of neural crest cell movement in the lateral pathway and that these fibres may direct cell migration in this space. Spieth & Keller (1984), however, report evidence that collagen fibres may be aligned by traction produced by the advancing cells themselves. In fact, our results show ECM fibres aligned at a right angle to the direction of dispersal in T. torosa (Fig. 8A) when the neural crest is beginning to disperse and in the chick embryo ECM fibrils are frequently arranged normal to the direction of neural crest migration (Tosney, 1982).
A third possible mechanism is contact inhibition of cell movement. Keller & Spieth (1984) observed contact inhibition between axolotl neural crest cells in vitro, but discounted this as a possible dispersal mechanism in vivo (except at the earliest phases of migration) since there were not enough cells in the lateral pathway to make contact events likely. As in the axolotl and Trituras alpestris (Epperlein, 1974), T. torosa neural crest cells display contact inhibition of cell movement in vitro on plastic substrata. In contrast to the situation reported by Keller & Spieth (1984), however, there are numerous cells in the lateral spaces of T. torosa and these cells are usually close enough to another cell for contact to occur (Fig. 7B–D).
Dorsal stripe formation
Pigment cell precursors initially disperse ventrally in the lateral pathway. Soon after melanization, however, many of these cells reverse their direction of migration and translocate dorsally toward the apex of the somites to form the dorsal melanophore stripe. Thus, we are confronted with three questions. What is responsible for this change of orientation? Why are xanthophores excluded from the dorsal stripe? What is the nature of the cue that signals the cells to stop at the apex of the somites?
Twitty (1936) determined by a series of grafting experiments that the cue for dorsal stripe formation was associated with the somites themselves. When pieces from the dorsal portion of the somites were grafted ventrally with the neural tube, neural crest and overlying ectoderm, melanophores still aggregated along the somites. If somites were missing from the graft, the melanophores did not form a distinct stripe in the donor tissue. To explain the reversal of melanophore orientation in the lateral pathway, Twitty (1945) likened the situation to the clumping of T. torosa melanophores in vitro. He concluded that stripe formation was the result of melanophores pulling themselves together via connecting (possibly anastomozing) processes. There are only passing references to xanthophores in these early reports, and their possible interactions with melanophores were not discussed. Our observations of melanophores in situ and in vitro do not support the notion that melanophores aggregate to form the dorsal stripe by dumping. In fact, clumping is observed only after several weeks in culture, long after the development of the dorsal stripe in the larva that provided the neural folds for culture.
There are two conspicuous changes that take place in situ at the time when the direction of melanophore migration changes: the migratory substratum shifts from the surface of the somites to the ECM underlying the ectoderm (as evidenced from the ectoderm-peel experiments) and recently differentiated melanophores become dendroid. The shift of migratory substratum is probably a response to increasing adhesivity of the subectodermal ECM. Throughout early development, collagen fibres are being incorporated into the dermal ECM. An increase in the number of collagen fibres or an ECM molecule associated with the fibres (such as FN, to which neural crest cells adhere tenaciously) could account for the change in substratum affinity. Unlike the somite cells in Xenopus, which acquire an enveloping basal lamina as the melanophores are migrating (Tucker, 1986), there is no basal lamina around the somite cells of T. torosa until well after the development of the primary pattern. Thus, it would be unlikely that the somite surface would be as adhesive as the dermal ECM. If the dorsal stripe cue is associated with the dermal ECM, as is suggested by ectoderm grafting experiments (Tucker & Erickson, 1986a), then this change in substratum affinity could contribute to a change in the direction of cell migration by bringing the cells in contact with a haptotactic cue. Directional migration would also be facilitated by changes in cell shape. The elongate processes that extend from melanophores in vitro and in situ frequently resemble neurite growth cones, which may find appropriate targets by haptotaxis (Nardi, 1983). It is possible that the development of long processes could lead to the change of direction of locomotion by permitting direct contact with the more adhesive substrata near the apex of the somites.
This latter hypothesis is also a possible explanation for the mechanism of xanthophore-melanophore segregation over the surface of the somites. In vitro, xanthophores lack long processes. If xanthophores also tend to lack long, ‘searching’ processes in vivo, these cells would be less likely to recognize the cue used by melanophores to form the dorsal stripe. The morphology of xanthophores in situ was not examined in detail in this study. Occasionally elongate xanthophore processes were visible using NH4OH fluorescence, but most cells examined in this way were rounded. The destructive nature of this technique (see Materials and Methods) did not permit further ultrastructural analysis of xanthophores.
We have proposed that contact inhibition of cell movement may be responsible for the initial directed migration of neural crest cells away from the neural tube. Nevertheless, neural crest-derived melanophores eventually reverse their direction of migration and aggregate near their point of origin, the neural tube. Melanophores, unlike unpigmented neural crest cells and xanthophores observed in culture, have greatly reduced lamellipodia. Their stellate morphology and reduced lamellipodia undoubtedly contribute to a reduction in contact inhibition amongst the melanophores. Changes in the cell surface that accompany differentiation may also lead to reduced contact inhibition. Xanthophores in culture do have lamellipodia. If these cells continue to display contact inhibition of cell movement after differentiation, this may inhibit a similar aggregation of the yellow pigment cells in situ.
The cue used by the melanophores to form the dorsal stripe is possibly produced by the somites (Twitty, 1936), but the cue is located in the subectodermal ECM since melanophores will form a stripe under ectoderm grafted ventrally over the lateral plate mesoderm (Tucker & Erickson, 1986b). The GAG identified in the lateral pathway using alcian blue staining and TEM appears to be homogeneous over the surface of the somites. More research is needed to identify the molecular nature of the dorsal stripe cue.
Midbody stripe formation
Why do melanophores and xanthophores stop and align at the somite-lateral plate mesoderm border in T. torosa?Twitty (1936) speculated that melanophores halted to form a midbody stripe due to the extreme angle that the lateral pathway takes as the ectoderm begins to surround the bulging yolk mass. From grafting experiments, Twitty (1936) also believed that some factor ventral to the somites inhibited either melanophore differentiation or migration. Our results suggest that melanophores are trapped in the lateral pathway by the pronephric duct. Melanophores reach the site of the midbody stripe by migrating through the intersomitic furrows, which apparently direct the cells ventrally as well as provide a pathway of least resistance beneath the bulging lateral line placode primordium. At the base of the intersomitic furrow, melanophores (and probably xanthophores) turn and migrate either anteriorly or posteriorly along the midbody stripe. These cells do not leave the midbody stripe to continue migrating ventrally or to return dorsally, suggesting that they are migrating on a particularly adhesive substratum (Figs 1B–D, 2B). This substratum is the pronephric duct. When the pronephric duct is surgically removed at this stage, melanophores in the midbody stripe are attached (Fig. 10A,B), and in serial sections, these cells are closely associated with the duct (Fig. 12E). The duct is located just beneath the surface of the ectoderm when the cells arrive at the base of the intersomitic furrows (Fig. 7A,B) and when it is internalized at a later stage in development many of the melanophores in the midbody stripe remain attached (Fig. 11 A,B). The pronephric duct in prehatching embryos is surrounded by a basal lamina (Tucker, unpublished observations) and it stains brightly with anti-FN antibodies (Fig. 16B). Since Taricha neural crest cells find FN to be an adhesive substratum in vitro, the presence of FN in the basal lamina surrounding the duct may account, in part, for the trapping of melanophores migrating ventrally. In fact, most of the tissues surrounded by a basal lamina in the trunk of Taricha torosa during melanophore patterning (e.g. the dorsal neural tube, the pronephric duct and blood vessels [see below]) have melanophores associated with them. The only exceptions are the ventral neural tube and portions of the subectodermal basal lamina. Melanophores are separated from the subectodermal basal lamina by a thick layer of collagen and GAG (Fig. 15A,B), and the ventral neural tube is also surrounded with GAG (Fig. 14). GAG can interact with FN and inhibit its adhesive properties (Rich, Pearlstein, Weissmann & Hoffstein, 1981; Erickson & Turley, 1983; Tucker & Erickson, 1984).
Although the pronephric duct halts the ventral migration of the melanophores and xanthophores in T. torosa, these pigment cells do not migrate ventrally between the lateral plate mesoderm and ectoderm when the duct has been ablated. This may be due, in part, to trapping of pigment cells by the posterior cardinal vein, which develops normally at the somite-lateral plate mesoderm border in spite of the absence of the pronephric duct (Fig. 12E). Nevertheless, there does appear to be some factor ventral to the duct and posterior cardinal vein that prevents the invasion of melanophores and xanthophores, because melanophores do not leave the boundaries of tissues grafted from dorsal regions onto the lateral plate mesoderm (Twitty, 1936; Tucker & Erickson, 1986b; Fig. 13A,B), except rarely along vasculature. It is interesting to note that blood vessels in the ventral tissues, like the pronephric duct, stain intensely with antibodies to FN (Fig. 16A,B) and are surrounded by a basal lamina.
Histological examination of the quantity and organization of the GAG in the lateral pathway dorsal and ventral to the pronephric duct does not reveal noteable differences (Fig. 15A,B). Our methods, however, provide only crude estimates of the composition and concentration of GAG. This aspect of the study deserves reexamination following the development of more quantitative techniques.
Distribution of iridophores
Observations of Taricha torosa whole mounts show iridophores along the dorsal surface of the neural tube and, in older embryos, in the eye and over the lateral plate mesoderm (Fig. 4A,B). The latter observation confirms the results of Finnegan (1955), who grafted lateral plate mesoderm with its overlying ectoderm to the dorsal surface of T. torosa embryos and observed numerous iridophores in the grafted tissue just after larval metamorphosis. How do iridophores reach the ventral tissues from which melanophores and xanthophores are excluded? One possibility is that iridophores migrate primarily through the ventral pathway, between the somites and neural tube, instead of through the lateral pathway over the apex of the somites. Although iridophores were not identified in the TEM in the ventral pathway, this may be due to the relative scarcity of these cells, as well as to the irregularity of their numbers and distribution from animal to animal. In one whole mount, lustrous cells were present in the ventral pathway near the notochord. Perhaps T. torosa iridophores, like the pigment cell precursors in Xenopus laevis (MacMillan, 1976; Tucker, 1986), migrate through the ventral pathway and continue ventrally along the mesoderm-endoderm border or along the coelomic face of the lateral plate mesoderm instead of between the ectoderm and mesoderm. These cells would not be trapped by the superficial pronephric duct and they would not be inhibited from migrating by a factor in the dermal ECM. Iridophores are seen dorsally near the neural tube before they begin to migrate. This suggests that these cells select the ventral pathway after they have differentiated and that their phenotype is not determined by a factor in the ventral pathway. It would be interesting to observe iridophore behaviour in vitro and compare interactions between iridophores and other pigment cells with substrata composed of isolated ECM macromolecules.
ACKNOWLEDGEMENTS
We would like to thank T. J. Poole for his assistance with ablation experiments, as well as P. B. Armstrong and R. E. Keller for useful comments and advice during the course of this research. We are also grateful for the cooperation of East Bay Regional Park District. This research was supported by NIH grant PHS-DE 05630 to C. A.E. and DHHS National Research Service Award GM 07377 to R.P.T.