Detailed SEM observations of the changes in cellular morphology, arrangements, and contacts that occur during the process of somite formation were made in two species of urodele amphibians, Ambystoma mexicanum and Pleurodeles waltlii, and one species of anuran amphibian, Rana sphenocephala. After fixation, embryos were fractured transversely, horizontally, and parasagittally, and the intrasomitic cellular arrangement pattern was examined with the SEM. It was found that Ambystoma and Pleurodeles embryos followed exactly the same development sequence in rosette formation and myoblast fusion. Rana somites did not, however, appear to form rosettes. Those myotomal cells underwent fusion immediately after a few segmentations occurred.

Patterns of cellular rearrangement were also described during urodele rosette formation at the time of somite segmentation and during myoblast fusion. Extensive changes in cell shape and orientation appeared to occur during those processes. When cells changed their orientation, they often exhibited a triangular configuration. Probable roles of these triangular-shaped cells in rosette formation and myoblast fusion are discussed.

During the initial period of myoblast or myotomal cell fusion, cells first send out specialized cell processes and then establish their cell–cell contacts. The establishment of such contacts eventually leads to tight membrane appositions and fusion. Since myoblast fusion appeared to occur between two cells which were tandemly arranged in a rosette, the origin of multinuclearity in the fused cells is discussed.

Finally, comparative analyses of the pattern of somite formation and subsequent muscle development were made between different species of amphibians. The possibility is discussed that patterns of somitogenesis may provide useful indicators for determining how different families of amphibians evolved.

Somite formation is a major morphogenetic event of axial structure formation during amphibian embryogenesis. Consequently, various aspects of somitogenesis have recently been studied. These include control mechanisms which regulate both the number and size of somites (Hamilton, 1969; Cooke, 1975; Elsdale, Pearson & Whitehead, 1976; Cooke, 1978; Pearson & Elsdale, 1979; Elsdale & Pearson, 1979); the kinematic nature of somite segmentation (Deuchar & Burgess, 1967; Pearson & Elsdale, 1979); comparative analyses of somite formation among different amphibian species (Hamilton, 1969); and the nature of myoblast fusion in various amphibian species (Loeffler, 1968, 1970; Muntz, 1975). It has emerged from those analyses that in most amphibian species somites form in a manner which is comparable to somitogenesis in many other vertebrates (e.g. chick). That is, each somite forms initially as a cluster of cells organized in a ‘rosette’ configuration. Initially, individual somites form by the pinching off of a group of cells from the paraxial mesoderm. Later, the majority of the somite cells, which are myotomal in character and mononucleate, elongate anteroposteriorly and fuse end to end with other cells of the same somite to give multinucleate muscle fibres. Somite formation in the anuran, Xenopus laevis does not, however, follow this typical vertebrate pattern. In Xenopus the presumptive somite cells elongate in the mediolateral direction prior to segmentation. Somite segmentation then occurs as transverse fissures isolate successive blocks of spindle-shaped cells. Bundles of the spindle-shaped cells in newly segmented somites then rotate through 90°. The cells come to lie parallel to the notochord and stretch from end to end of the length of the somite (Youn & Malacinski, 1981). Several hours later, these myotomal cells differentiate as uninucleate muscle fibres.

With recent advances in preparative techniques for scanning electron microscopy, it is now possible to observe the details of cellular morphology, arrangements, and contacts during the process of somite formation in various vertebrate embryos. In the chick embryo, for example, the surface mesoderm which lies on either side of the anteriormost end of the primitive streak (just posterior to the Hensen’s node) has been examined in stereo with the scanning electron microscope (SEM). Meier (1979) has found that this region is organized into tandemly aligned, repeating circular domains (about 180 μm in diameter). As these structures (‘somitomeres’) are added to the embryonic axis during neurulation, they condense considerably and undergo morphogenesis to become mature somites. Chick somites appear, therefore, to emerge from pre-existing somitomeres. That observation suggests that the paraxial mesoderm is programmed relatively early during axial structure development. The pattern of rosette formation at the time of somite segmentation has been studied by Bellairs (1979) with ‘fractured’ chick embryos. Those studies have led to the proposal that two factors may be responsible for changes in cell shape during somite segmentation: (1) collagen fibrils; and (2) a change in cell-to-cell adhesiveness. The arrangement of the cells in more highly differentiated somites has also been examined by that author. The results have led to the suggestion that the chick resembles Xenopus in that the myotomal cells undergo rotation and become oriented in an anteroposterior direction.

Detailed SEM investigations of amphibian somite segmentation have been restricted almost exclusively to the anuran amphibian, Xenopus laevis. In previous studies changes in cellular morphology, arrangement, and contacts among the paraxial mesodermal cells of Xenopus have been examined (Youn, Keller & Malacinski, 1980). The role individual cell shape changes play during the 90° rotation of Xenopus myotomal cells has also been studied (Youn & Malacinski, 1981). In a few other anuran and urodele amphibian species the dorsal surface of the somite and mesoderm and the temporal sequence of somite segmentation have also recently been described (Youn et al. 1980). No detailed SEM observation of the changes in cellular morphology, arrangement and contacts that occur during the process of somite formation has, however, been made in the various species of amphibians which display the typical vertebrate rosette pattern. The present paper employs two species of the urodele amphibians, Ambystoma mexicanum and Pleurodeles waltlii, and one species of the anuran amphibian, Rana sphenocephala for a comparative study. The developmental timing and sequence of rosette formation and myoblast fusion were observed. The patterns of cellular rearrangements were described during rosette formation at the time of somite segmentation and during myoblast fusion. The results of these studies provided no evidence for the presence of the chick type of somitomere-like structures in the paraxial mesoderm. It was also observed that the pattern of rosette formation and myoblast fusion in Rana is quite different from the urodele species. Special attention was, therefore, devoted to the comparative aspects of somite formation in these various species of amphibia.

Ambystoma mexicanum and Pleurodeles waltlii embryos were obtained from natural spawnings. Rana sphenocephala eggs were artificially inseminated according to the method outlined by Rugh (1962). Jelly layers were removed with fine watchmaker’s forceps in dechlorinated tap water. The stagings of the urodele embryos followed Bordzilovskaya & Dettlaff (1979) for Ambystoma, and Gallien & Durocher (1957) for Pleurodeles. Embryos of Rana sphenocephala were staged according to Shumway (1940) for Rana pipiens. Embryos at appropriate stages were fixed overnight at 4 °C in a cacodylate-buffered (0·1 M, pH 7·2) 2·5 % glutaraldehyde solution. The fixed embryos were rinsed in 0·1 M cacodylate buffer. The epidermis of each embryo was then peeled off with a fine steel knife and watchmaker’s forceps. The neural tube, and sometimes also the notochord, were removed from the ‘peeled’ embryos. Dissected embryos were then post-fixed in cacodylate-buffered (0·1 M, pH 7-5) osmium tetroxide for 3 h at 4 °C. In order to expose the cells which reside in the interior of the somitic mesoderm, a horizontal or parasagittal fracture was made through the somite file and the unsegmented mesoderm at (or above) the level of the notochord. Occasionally, a transverse fracture was made through the unsegmented plate to reveal a cross-sectional view of the morphology, arrangement, and contacts of the cells. Fine tungsten needles and forceps were employed for preparing those fractures.

Following dehydration in a graded alcohol series, samples were critical-point dried from liquid CO2 in a Peleo critical-point drier (Model H, Ted Pella Co.). Specimens were then mounted on aluminium stubs with conductive silver paint and,coated with gold–palladium (60:40) in a sputter coater E5100 (Polaron Equipment Ltd). These were examined at 20 kV in an Etec U-l.scanning electron microscope and photographed on Polaroid type 55 positive–negative film. More than 20 embryos of each stage were examined under the SEM. Five developmental stages were chosen to describe the timing and sequence of rosette formation and myoblast fusion.

1. Developmental timing and sequence of rosette formation and myoblast fusion

Figures 1 and 2 display a stage series of horizontally or parasagittally fractured embryos of Ambystoma and Rana. In the case of Ambystoma (Fig. 1), each somite is formed by the pinching off of a group of cells from the paraxial mesoderm. Those cells then become organized into a rosette configuration. Details of how the myotomal cells rearrange themselves into a rosette will be described in the next section of this report. Each of the segmented somites remain as rosettes until stage 27 (12–13 somites, see Fig. 1c). This observation confirms an earlier report by Loeffler (1968). Analyses of histological preparations had suggested that the critical period for myoblast fusion occurs between stages 28 and 34. Our SEM observations indicate that that is indeed the case. The embryo shown in Fig. 1d has formed about 15 somites (stages 28–29), and the myoblasts in the anterior six somites have either completed or are in the process of undergoing fusion. Later, at about stage 32 (20 somites, see Fig. 1 e), the anterior twelve somites show signs of myoblast fusion. The fusion of myoblasts, therefore, takes place in an anteroposterior direction. It also appears that the fusion process proceeds in a mediolateral direction. This can be observed in the sixth somite of a stage-28 to -29 embryo (Fig. 1 d), and the twelfth somite of a stage-32 embryo (Fig. 1 e). In those examples, fusion of myoblasts is just about to begin in those particular planes of the horizontal fracture. Only a group of cells in the medial side can be seen to be stretched from end to end of the somites. Myoblast fusion does not, therefore, occur synchronously throughout individual somites.

Fig. 1.

An atlas of somite development in Ambystoma mexicanum. Embryos were fractured longitudinally except in (b) where the embryo was parasagittally fractured in order to reveal the pattern of intrasomitic cellular arrangement. Whole embryo views are shown on the left, and higher-magnification pictures ofthe fractured somite mesoderm are on the right side, (a) Stages 19–20; (b) stages 21–22; (c) stages 27–28; (c) stages 28–29; (e) stage 32. Bars in the lower-magnification pictures represent 0·5 mm. Bar at the upper left comer of the higher-magnification micrograph in (a) represents 01 mm for all the higher-magnification micrographs in Fig. 1. Ant, anterior; Post, posterior, for all the embryos shown here.

Fig. 1.

An atlas of somite development in Ambystoma mexicanum. Embryos were fractured longitudinally except in (b) where the embryo was parasagittally fractured in order to reveal the pattern of intrasomitic cellular arrangement. Whole embryo views are shown on the left, and higher-magnification pictures ofthe fractured somite mesoderm are on the right side, (a) Stages 19–20; (b) stages 21–22; (c) stages 27–28; (c) stages 28–29; (e) stage 32. Bars in the lower-magnification pictures represent 0·5 mm. Bar at the upper left comer of the higher-magnification micrograph in (a) represents 01 mm for all the higher-magnification micrographs in Fig. 1. Ant, anterior; Post, posterior, for all the embryos shown here.

Fig. 2.

An atlas of somite development in Rana sphenocephala. Embryos were fractured longitudinally to show the dorsal aspects of the intrasomitic cellular arrangement pattern. Whole embryo views are seen on the left side, and higher-magnification micrographs taken from the fractured somite mesoderm are on the right side, (a) Stage 15; (b) stage 15+; (c) stage 16; (d) stage 18; (e) stage 20. Marked areas in (c), (d) and (e) indicate a few posteriormost somites (about six) in which the myotomal cells are undergoing fusion. Cells in the somites anterior to those seem to have completed the fusion process at these particular fraction planes. Bars in the lower-magnification micrographs represent 0-3 mm. Bar at the upper left corner of the higher-magnification micrograph in (a) represents 0·1 mm for all the higher-magnification pictures in Fig. 2. Ant. anterior; Post, posterior, for all the embryos pictured here.

Fig. 2.

An atlas of somite development in Rana sphenocephala. Embryos were fractured longitudinally to show the dorsal aspects of the intrasomitic cellular arrangement pattern. Whole embryo views are seen on the left side, and higher-magnification micrographs taken from the fractured somite mesoderm are on the right side, (a) Stage 15; (b) stage 15+; (c) stage 16; (d) stage 18; (e) stage 20. Marked areas in (c), (d) and (e) indicate a few posteriormost somites (about six) in which the myotomal cells are undergoing fusion. Cells in the somites anterior to those seem to have completed the fusion process at these particular fraction planes. Bars in the lower-magnification micrographs represent 0-3 mm. Bar at the upper left corner of the higher-magnification micrograph in (a) represents 0·1 mm for all the higher-magnification pictures in Fig. 2. Ant. anterior; Post, posterior, for all the embryos pictured here.

It is difficult to estimate accurately how long it takes for the cells of a single somite to complete the process of fusion by simple examination of the morphology of cells in these particular fracture planes. The fractures were made at slightly different dorsoventral levels in different embryos. Nevertheless, a rough approximation of the time required for rosette formation and myoblast fusion can be made. For example, the 6th somite at about stage 20 (approximately 70 h after fertilization at 20°, Fig. 1 d) is a newly formed somite which is beginning to assume the rosette form. It remains rosette-shaped until stages 27–28 (about 90 h, Fig. 1 c). At about stages 28–29 (approximately 95 h, Fig. 1 d), the cells of the 6th somite start to fuse from the medial side. Consequently, the rosette configuration lasts for approximately 25 h at 20°. At stages 32 (about 113 h, Fig. 1 e), the 6th somite appears to have completed the process of myoblast fusion. Thus it takes about 18 h at 20° for completion of the fusion process.

In Pleurodeles, also a urodele amphibian, the pattern of rosette formation is exactly the same as that in Ambystoma. Moreover, the process of myoblast fusion follows exactly the same sequence as seen in Ambystoma embryos (results not shown). When the Pleurodeles embryo reaches the 20-somite stage, the anterior eleven somites appear either to be undergoing fusion or to have already completed the fusion process. It should be pointed out here that unlike the rosettes of the chick embryo, those in the urodele amphibians are much more closely defined, with long cells radiating from the centres (see Discussion).

Scanning electron micrographs in Fig. 2 indicate that the pattern of rosette formation and myoblast fusion in Rana is completely different from that of the urodele species studied here. A group of cells first becomes separated from the paraxial mesoderm. The cells do not then exhibit any detectable sign of the formation of the urodele-type rosettes (e.g. Fig. 2). In some cases, cells appear to be arranged radially around much of the circumference of several somites (e.g. see somite no. 5 in Fig. 2b and somite no. 14 in Fig. 2e). The apparent difference in the mode of somite cell organization between Rana and the urodeles may derive in part from the shorter length and more rounded shape of Rana cells compared with urodele cglls (see below, Figs 3 and 4). However, differences in the degree of cell elongation between species may only give a superficial and variable indication of the functional polarization of the cells caused by the cytoskeletal organization or the arrangement of organelles. In a few somites anterior to the just separated paraxial mesoderm cells, fusion is already under way (Figs. 2,c, d, and e). It is interesting to note here that the cells of approximately six caudal somites are always observed in the process of fusion (see indicated areas in Figs. 2 c, d, and e), whereas the rest of the anterior somites appear to have completed the process. As in the case of Ambystoma, the fusion process proceeds in the mediolateral direction. The cellular rearrangement pattern at the time of somite segmentation and during myotomal cell fusions will be dealt with later in this paper.

Fig. 3.

(a) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Ambystoma embryo, right caudal to the last segmented somite. Cells of various sizes are seen to be arranged in various directions. No specialized structure such as somitomeres, can be detected in this figure. This micrograph was taken from the embryo shown in Fig. 1c. Med, medial; Lat, lateral; Ant, anterior; Post, posterior. Bar represents 25 μm.

(b) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Rana embryo, showing that cells appear to be oriented somewhat mediolaterally. The width of the paraxial mesoderm at this particular fracture plane appears to consist of approximately four cells. The number of cells does not seem to change even after somite segmentation (see arrow for the segmentation line). No somitomere-like structure can be observed in this micrograph. This picture was taken from the embryo shown in Fig. 2 c. Med, medial; Lat, lateral; Ant, anterior; Post, posterior. Bar represents 25 μm.

Fig. 3.

(a) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Ambystoma embryo, right caudal to the last segmented somite. Cells of various sizes are seen to be arranged in various directions. No specialized structure such as somitomeres, can be detected in this figure. This micrograph was taken from the embryo shown in Fig. 1c. Med, medial; Lat, lateral; Ant, anterior; Post, posterior. Bar represents 25 μm.

(b) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Rana embryo, showing that cells appear to be oriented somewhat mediolaterally. The width of the paraxial mesoderm at this particular fracture plane appears to consist of approximately four cells. The number of cells does not seem to change even after somite segmentation (see arrow for the segmentation line). No somitomere-like structure can be observed in this micrograph. This picture was taken from the embryo shown in Fig. 2 c. Med, medial; Lat, lateral; Ant, anterior; Post, posterior. Bar represents 25 μm.

Fig. 4.

Transverse view of the unsegmented mesoderm of Ambystoma (a) and Rana (b) embryos. Arrow indicates approximate parasagittal plane of fractures which will be shown in Fig. 5. Med. medial; Lat, lateral. Bars represent 0·1 mm.

Fig. 4.

Transverse view of the unsegmented mesoderm of Ambystoma (a) and Rana (b) embryos. Arrow indicates approximate parasagittal plane of fractures which will be shown in Fig. 5. Med. medial; Lat, lateral. Bars represent 0·1 mm.

An accurate estimation of the developmental timing of myotomal cell fusions could not be made in this case, since a normal staging series of Rana sphenocephala has not yet been established. The developmental rate at room temperature is, however, known to be about 20 % as fast again as Rana pipiens (J. Frost, personal communication).

In order to approximate the timing of fusion, the 5th somite was chosen. It is newly formed at the neural-fold stage, and the cells within it remain unfused until the neural folds begin to fuse. Once the folds have fused, cells in the medial side of the somite begin to fuse. By the muscular movement stage the cells in the 5th somite appear to be completely fused. Substantially less than 24 h is required, therefore, for completion of the fusion process.

2. Cellular rearrangement during rosette formation

In the chick embryo, somites have been shown to develop from pre-existing somitomeres (Meier, 1979). That observation suggests an early programming of the paraxial mesoderm which later develops into somites. In the case of amphibian embryos, examination of the surface paraxial mesoderm has, however, failed to provide any evidence for such somitomere-like structures (Youn et al. 1980). In this study a further attempt was made to search for the presence of such structures in the rosette-forming species of amphibians (urodeles). The pattern of intrasomitic cellular arrangement was examined in the unsegmented mesoderm which resides immediately caudal to a newly segmented somite. First, Ambystoma embryos were horizontally fractured through the unsegmented plate. Dorsal aspects of the pattern of cells which lay deep (ventrally) in the plate were examined with the SEM (Fig. 3a). No specialized structure which resembled the somitomere of the chick embryo could befound. Cells in the medial side appeared, however, to be elongated mediolaterally (see pointers in Fig. 3a).

Second, the cellular arrangement pattern of the paraxial mesoderm was examined in transverse sections. The photograph in Fig. 4 a demonstrates that cells which lie on the lateral side are arranged in a columnar manner, but cells of the medial region are less well organized. Some of those in the medial region near the notochord and the neural tube appear to be elongated in the mediolateral direction. This apparently explains why the cells of the medial side shown in Fig. 3 a were seen to be oriented in a mediolateral fashion. Fig. 4 a also shows that the paraxial mesoderm consists of two layers of cells in the lateral side and approximately three layers in the region where the mesoderm is thickest. These cells are, no doubt, elongated dorsoventrally. Therefore, the cellular arrangement pattern displayed in Fig. 3 a is obtained when a fracture is made between those cell layers. Only the upper (dorsal) portions of those dorsoventrally oriented cells which reside in the lower layer were exposed in the horizontal section (Fig. 3a).

In Rana embryos, no somitomere-like cellular arrangement could be observed in horizontal sections (Fig. 3 b). The paraxial mesoderm consists of about four cell layers in width at this particular fracture plane. The number of layers does not appear to change even after the last somite is formed. As in the case of Ambystoma embryos, the medial cells of the paraxial mesoderm are mediolaterally oriented. When viewed in transverse section (Fig. 4 b), it can be observed that the paraxial mesoderm consists of many cell layers dorsoventrally. The highest number is found in the most medial side near the notochord and decreases further laterally.

In order to gain insight into the cellular mechanisms involved in rosette formation in Ambystoma and Pleurodeles embryos, it is essential to understand the three-dimensional aspects of the cellular arrangements in the paraxial mesoderm. The horizontal section shown in Fig. 3a reveals the morphology of mediolaterally elongated cells but fails to display any structural details of dorsoventrally oriented cells. The transverse section in Fig. 4a provides a substantial amount of information regarding the cell morphology and arrangement pattern in the paraxial mesoderm. Yet that view fails to reveal any temporospatial sequence of cell shape changes that occur in the posteroanterior direction as the myotomal cells are progressively organized into a rosette. Nevertheless, transverse sections can provide a general indication of how the cells in the paraxial mesoderm might behave at the time of somite segmentation. The mediolaterally elongated cells seen in the transverse section (Fig. 4 a) and the horizontal section (Fig. 3a) could provide the mediolaterally arrayed cells of a prospective rosette. Cells in the middle layer of the three-layered region of the paraxial mesoderm might serve as a source for anteroposteriorly arranged cells in either the anterior or posterior half of a rosette. In that case they would have to undergo several degrees of rotation, since they appear to be elongated dorsoventrally. In the lateral region of the paraxial mesoderm, the cells are arranged in two layers. Those cells must also rotate in order to provide the lateromedially arranged cells of a prospective rosette.

Both the pattern of cellular arrangements in the unsegmented mesoderm and the changes which accompany rosette formation in a newly formed somite can be best displayed in parasagittal sections. Parasagittal fractures were made with Pleurodeles (Figs. 5,a, b) and Ambystoma (Fig. 5 c) embryos at various stages. The approximate location of the fracture plane is indicated by the arrow in Fig. 4 a. Since the fractures were made somewhat laterally, two layers of dorsoventrally elongated cells were usually observed in the paraxial mesoderm. It was predicted that cells in the prospective cranial and caudal segmentation lines should undergo extensive rearrangement in order to provide the anteroposteriorly oriented cells in the complete rosette. In Figs 5 a and c, in which the posterior segmentation line is shown to be fully established (thick arrows), cells in the most anterior portion of the as yet unsegmented mesoderm assume an oblique orientation (short arrows in Figs. 5 a and c). Generally, those cells are conical. The apex of the cells at the outside margin of the prospective rosette is wide. The other ends are sharply pointed toward the prospective centre of the rosette. Where segmentation is still in progress (thick arrow, Fig. 5 b), however, the prospective anterior cells of the emerging somite do not show any detectable sign of rotation. In such cases it was observed that in the caudal part of the newly segmented somite, cells are often triangular shaped (long arrow, Fig. 5 b). Such triangular-shaped cells are also frequently observed in other parts of rosettes (long arrows, Fig. 5 a). It can be speculated that those cells are actively engaged in rotating 90°. Perhaps they are triangular because one of their apices actively moves towards the prospective centre of the rosette (see Discussion). As rosette formation continues, the other apices probably retract toward the thickest portions of the cells. In this fashion dorsoventrally arranged cells could change to an anteroposterior orientation.

Fig. 5.

Mediolateral view of the parasagittally fractured Ambystoma (c) and Pleurodeles (a and b) embryos, showing how changes in cell shape occur during rosette formation. Short, thick arrows indicate the caudal segmentation line. Short arrows point to conically shaped cells which appear to be undergoing rotation. In these cases, rotation takes place in such a way that dorsoventrally elongated cells become oriented anteroposteriorly. Triangular-shaped cells (long arrows) are often observed in the anterior or the posterior corners of rosettes. Ant, anterior; Post, posterior, for all the micrographs in Fig. 5. Bars represent 50/un. Same scale of bar in (a) can be used in (b).

Fig. 5.

Mediolateral view of the parasagittally fractured Ambystoma (c) and Pleurodeles (a and b) embryos, showing how changes in cell shape occur during rosette formation. Short, thick arrows indicate the caudal segmentation line. Short arrows point to conically shaped cells which appear to be undergoing rotation. In these cases, rotation takes place in such a way that dorsoventrally elongated cells become oriented anteroposteriorly. Triangular-shaped cells (long arrows) are often observed in the anterior or the posterior corners of rosettes. Ant, anterior; Post, posterior, for all the micrographs in Fig. 5. Bars represent 50/un. Same scale of bar in (a) can be used in (b).

Details of the changes in cell-to-cell contacts which accompany the formation of a new somite were also examined. When observed in horizontal sections, cells of the paraxial mesoderm appear to make contacts with each other by many thin, thread-like projections (Fig. 6 a). In the newly segmented somite (Fig. 6 b), cells appear to move toward the prospective centre of the rosette (small circle) by sending out long, thin, and sharply pointed projections (arrow). When viewed under higher magnification (Fig. 6 c), these cell processes appear to bifurcate into short, finger-like projections which connect with adjacent cells (arrows). Moreover, it was observed that cells in the newly formed somite are attached to each other by thin, broad processes which, in turn, send out a few finger-like projections (pointers, Fig. 6 c). Even after rosette formation is completed, cells near the centre of the rosette continue to display specialized cell processes (arrows, Fig. 6d). Those processes are thin and broad and have relatively smooth edges.

Fig. 6.

(a) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Ambystoma, showing thin, thread-like projections between the cells. Bar represents 20 μm.

(b) Dorsal view of the longitudinally fractured, newly formed somite of Abmystoma, showing that cells appear to move towards the prospective centre of the rosette (circled area) by sending out long projections (arrow). The caudal line of segmentation is indicated by pointers. Bar represents 20 μm.

(c) Higher-magnification micrograph taken from the region near the long projection indicated by arrow in Fig. 6b. The long projection is seen to have tiny, finger-like processes around its margin. Cells appear to make contacts with each other by thin but broad projections which, in turn, send out a few finger-like processes. Bar represents 20 μm.

(d) Lateral view of the parasagittally fractured mature somite of Ambystoma, showing thin, broad cell processes around the centre of the rosette. Bar represents 20 μm.

(e) Dorsal view of the longitudinally fractured mature somite of Ambystoma, showing the myocoel. The lateromedial direction of the somite is indicated by long arrow. Bar represents 20 μm.

(f) Higher-magnification micrograph of the myocoel taken from Fig. 6e, showing specialized endings of myoblasts in the myocoel. They seem to be interconnected by a network of fibrils. Bar represents 20 μm.

Fig. 6.

(a) Dorsal view of the longitudinally fractured, unsegmented mesoderm of Ambystoma, showing thin, thread-like projections between the cells. Bar represents 20 μm.

(b) Dorsal view of the longitudinally fractured, newly formed somite of Abmystoma, showing that cells appear to move towards the prospective centre of the rosette (circled area) by sending out long projections (arrow). The caudal line of segmentation is indicated by pointers. Bar represents 20 μm.

(c) Higher-magnification micrograph taken from the region near the long projection indicated by arrow in Fig. 6b. The long projection is seen to have tiny, finger-like processes around its margin. Cells appear to make contacts with each other by thin but broad projections which, in turn, send out a few finger-like processes. Bar represents 20 μm.

(d) Lateral view of the parasagittally fractured mature somite of Ambystoma, showing thin, broad cell processes around the centre of the rosette. Bar represents 20 μm.

(e) Dorsal view of the longitudinally fractured mature somite of Ambystoma, showing the myocoel. The lateromedial direction of the somite is indicated by long arrow. Bar represents 20 μm.

(f) Higher-magnification micrograph of the myocoel taken from Fig. 6e, showing specialized endings of myoblasts in the myocoel. They seem to be interconnected by a network of fibrils. Bar represents 20 μm.

It appears that myocoel formation and the emergence of rosette centres are separate phenomena. The micrographs in Fig. 1 display the centres of the rosettes in each somite but not the myocoels. That may be due to the fact that horizontal fractures were not made deep enough to uncover them. The myocoel can be observed when a fracture is made more ventrally through the somite file. Figure 6e shows both the centre of a rosette and the myocoel. Cells in the medial region which are adjacent to the myocoel (opposite to the direction of the arrow in Fig. 6e) appear to be shorter than those in the more lateral portions of the somite. Therefore, the presence of the myocoel may not be required for cellular rearrangements during rosette formation. Actually, the myocoel exists prior to the rosette formation (Hamilton, 1969). Within the myocoel itself, cells appear to be interconnected by a network of fibrils (Fig. 6f). Unlike chick embryos (Bellairs, 1979), no mesenchymal cells were detected within the myocoel.

3. Cellular rearrangement during myoblast fusion

The process of myoblast fusion was examined in horizontally fractured embryos of Ambystoma (Fig. 7) and Rana (Fig. 8). In the mature rosette configuration cells appear to radiate from the centre of the rosette in all directions. In the medial and lateral sides, cells which are radiating from the centre of the rosette are conically or triangularly shaped. In the middle region, however, cells are somewhat rod-like or spindle-shaped, and are arranged directly in the anteroposterior direction. Before they begin to fuse, two main events apparently occur. First, cells in the medial side undergo extensive changes in shape. Figure 7a shows that they are in the process of retracting their conical side away from the centre and towards the medial region of the rosette (short arrows). While that retraction process is under way, it appears that cells begin to establish end-to-end contacts with each other (pointer). Those kinds of contact may be the first indication that myoblast fusion is about to begin. It was speculated in the previous section of this report that cell retraction is required for elongated cells to change direction of orientation. As was shown in Fig. 5, dorsoventrally elongated cells became arranged anteroposteriorly. Likewise, mediolaterally arranged cells become oriented in the anteroposterior direction. Cell retraction may, therefore, be a common mechanism used for changing direction of cell orientation during rosette formation and myoblast fusion (see Discussion).

Fig. 7.

(a) Dorsal view of the longitudinally fractured somite of Ambystoma, showing that the myoblasts start to fuse from the medial side. Short arrows indicate the region of probably the retraction trails of two cells in the most medial side. They begin to show an initial sign of fusion. One of them is seen to send out a long process (pointer) on to another. Long arrows point to the prospective dermatomal cell, which appears to have rather smooth lateral surfaces. Bar represents 20 μm.

(b) Dorsal view of the Ambystoma myoblasts in the lateral region of the somite. Cells in the most lateral side seem to lose their triangular or conical configuration and become oriented anteroposteriorly. Bar represents 20 μm.

(c) Dorsal view of the longitudinally fractured somite of Ambystoma which is in a more advanced stage of fusion than the somite shown in Fig. la. Fracture was made more deeply (ventrally) in the lateral side than in the medial side. Arrow indicates sclerotomal cell mass. Bar represents 20 μm.

(d) Higher-magnification micrograph taken from the lateral region of the somite shown in Fig. 7 c. Arrow indicates thin, broad cell processes which appear to cover the end portions of neighbouring cell surfaces which are to be fused. Bar represents 20 μm.

(e) Dorsal view of the longitudinally fractured somite of Ambystoma in the posterior portions. Cells in the anterior half of the somite are shown to develop narrow and long processes (arrows), and begin to surround the shorter-looking cells. Bar represents 20 μm.

(f) Dorsal view of the longitudinally fractured somite of Ambystoma, in which cells appear to have already completed the fusion process. Pointers indicate dermatomal cells. The mediolateral direction of all the somites shown in Fig. 7 is indicated by a long arrow here. Bar represents 20 μm.

Fig. 7.

(a) Dorsal view of the longitudinally fractured somite of Ambystoma, showing that the myoblasts start to fuse from the medial side. Short arrows indicate the region of probably the retraction trails of two cells in the most medial side. They begin to show an initial sign of fusion. One of them is seen to send out a long process (pointer) on to another. Long arrows point to the prospective dermatomal cell, which appears to have rather smooth lateral surfaces. Bar represents 20 μm.

(b) Dorsal view of the Ambystoma myoblasts in the lateral region of the somite. Cells in the most lateral side seem to lose their triangular or conical configuration and become oriented anteroposteriorly. Bar represents 20 μm.

(c) Dorsal view of the longitudinally fractured somite of Ambystoma which is in a more advanced stage of fusion than the somite shown in Fig. la. Fracture was made more deeply (ventrally) in the lateral side than in the medial side. Arrow indicates sclerotomal cell mass. Bar represents 20 μm.

(d) Higher-magnification micrograph taken from the lateral region of the somite shown in Fig. 7 c. Arrow indicates thin, broad cell processes which appear to cover the end portions of neighbouring cell surfaces which are to be fused. Bar represents 20 μm.

(e) Dorsal view of the longitudinally fractured somite of Ambystoma in the posterior portions. Cells in the anterior half of the somite are shown to develop narrow and long processes (arrows), and begin to surround the shorter-looking cells. Bar represents 20 μm.

(f) Dorsal view of the longitudinally fractured somite of Ambystoma, in which cells appear to have already completed the fusion process. Pointers indicate dermatomal cells. The mediolateral direction of all the somites shown in Fig. 7 is indicated by a long arrow here. Bar represents 20 μm.

Fig. 8.

Dorsal views of the longitudinally fractured somites of Rana embryos. (a), (b), (c) and (d) are the higher-magnification micrographs taken from the 8th, 7th, 5th and 4th somites of the stage-16 embryo shown in Fig. 2c, respectively. Long arrows in (a) and (b) indicate probable fusion lines along which the myotomal cells are aligned and fused later. In (a), arrow indicates long projection which may play an active role in the fusion process. In (b), arrows indicate the cells which appear to be elongated along the prospective fusion line. Arrow in (c) points to widely spread cell process. In (d), most of the myotomal cells at this particular fracture plane have completed fusion. Bar in (a) represents 20 μm for (b), (c), (d) and (f). Arrow in (e) also indicates long projections which are usually observed during the initial period of fusion. Bar in (e) represents 20 μm. In (f), arrow points to the cell which is near completion of fusion. The mediolateral direction of all the somite is shown by long arrow in(f).

Fig. 8.

Dorsal views of the longitudinally fractured somites of Rana embryos. (a), (b), (c) and (d) are the higher-magnification micrographs taken from the 8th, 7th, 5th and 4th somites of the stage-16 embryo shown in Fig. 2c, respectively. Long arrows in (a) and (b) indicate probable fusion lines along which the myotomal cells are aligned and fused later. In (a), arrow indicates long projection which may play an active role in the fusion process. In (b), arrows indicate the cells which appear to be elongated along the prospective fusion line. Arrow in (c) points to widely spread cell process. In (d), most of the myotomal cells at this particular fracture plane have completed fusion. Bar in (a) represents 20 μm for (b), (c), (d) and (f). Arrow in (e) also indicates long projections which are usually observed during the initial period of fusion. Bar in (e) represents 20 μm. In (f), arrow points to the cell which is near completion of fusion. The mediolateral direction of all the somite is shown by long arrow in(f).

Second, extensive changes in cellular rearrangement occur in the lateral side. Conically or triangularly shaped cells seem to rearrange themselves to become elongated directly in the anteroposterior direction (Fig. 7b). Cell retraction may also be involved in this process. As a result, radially oriented cells in the lateral side change their orientation. Some of those cells which are located on the extreme lateral side of the somite remain unfused and, at later stages, become dermatomal cells.

As mentioned before, the fusion of myoblasts proceeds in the mediolateral direction. The observations included in Figs. 7 c and d indicate that the establishment of end-to-end contacts between cells in the anterior and posterior halves of the rosette may lead to fusion. The higher-magnification micrograph in Fig. 7d shows that cells send out wide and broad processes (e.g. see arrow) surrounding the partner cell with which they will fuse. Moreover, cells in the anterior half of the rosette may play a more active role in the fusion process than those in the posterior half. Those in the anterior half possess long and narrow processes stretching all the way to the caudal end of the somite (see arrows in Fig. 7e). Conversely, cells in the posterior half are somewhat shorter in length in this particular fracture plane. Later, these processes may actually broaden and begin to surround the shorter cells. Membrane fusion probably then follows.

The same kind of cell processes seen in Ambystoma at the time of myoblast fusion can also be observed during myotomal cell fusion in Rana. Cells appear to send out long processes toward neighbouring cells (arrows in Figs. 8a, b). It is also observed that cells frequently develop broad, wide processes, probably at the end of the fusion process (arrow, Fig. 8c). Figure 8e shows an example of cells which are probably toward the completion of fusion (arrow). The cells are almost completely fused, except in the posterior region where a smaller cell was detached during the fracture leaving behind an artifactual hole. We suspect that the fused cell was actively surrounding the detached cell underneath it in order to complete fusion and to stretch from one end of the somite to the other. With the aid of SEM alone, however, it is difficult to interpret cell morphologies and contacts in terms of the fusion process itself. In order to determine precisely where and when cell fusion is occurring, further evidence provided by the transmission electron microscope is needed.

As mentioned before, the myotomal cells in Rana undergo fusion immediately after a few segmentations have occurred. During the fusion period they position themselves in lines which are not exactly parallel to the long axis of the embryo. In fact, it appears that the fusion lines are approximately 45° off the axis (long arrows in Figs. 8a and b). Cells in that formation which are preparing for fusion (short arrows, Fig. 8b) appear to be somewhat elongated in the direction of the long arrow. However, Figs. 8c and d reveal that cells which have already fused, or are in the process of fusion, become arranged parallel to the notochord. This observation indicates that Rana somites rotate in a fashion which resembles somitogenesis in Xenopus (Hamilton, 1969; Youn & Malacinski, 1981). The degree of rotation, however, is approximately 45°. That amount of rotation is substantially less than the 90° rotation known to occur in Xenopus somites. Like Ambystoma, the fusion process proceeds in the mediolateral direction.

It is also interesting to note here that the sclerotome does not seem to form by partitioning from the main somite mass. Figure 7c shows the mass of sclerotomal cells lying deep in the medial wall of the somite (arrow). Those cells are probably not of neural crest origin because neural crest cell migration has not begun at this stage (approximately stage 32). Figure 7f clearly shows the dermatomal cells in the lateral side of the somite (pointers). These cells can be characterized by their smaller size and the many finger-like projections on their lateral surfaces. For comparison, the lateral side of the somite shown in Fig. 7a displays only cells which are larger in size and have fewer projections (long arrows).

1. Cellular basis of rosette formation in urodele species

The scanning electron microscopic observations of the interior of the paraxial mesoderm and somites have provided new insights into the cellular mechanisms which direct rosette formation and myoblast fusion in Ambystoma and Pleurodeles embryos. A variety of cellular mechanisms such as shape changes, rearrangements, and individual cell movements are probably involved in such processes. Our observations reveal that during somite morphogenesis some cells appear to change orientation several times. The first change in cell arrangement occurs during rosette formation at the time of appearance of new somites. The parasagittal sections in Fig. 5 indicate the manner in which the double-layered cells, which are elongated in the dorsoventral direction, become organized into a rosette form. A particularly interesting feature of somite segmentation is the appearance of triangular-shaped cells at the corners of rosettes. These cells may turn through as much as 90°. The cellular arrangement pattern seen in Fig. 5 is diagrammatically illustrated in Fig. 9. To simplify the issue, only two layers of cells elongated dorsoventrally in parasagittal section are considered. During the establishment of the segmentation line cells in the bottom portion of the upper layer and top portion of the lower layer begin to send out long processes toward the prospective centre of the rosette (see Figs. 6b, c). In order to become organized into a rosette, cells in the anterior- and posterior-most regions of the somite must move a longer distance towards the centre of the rosette than those in the middle. In the process they become triangular shaped. It appears that such cellular reshaping, followed by active cell movements, takes place only after the segmentation lines appear. Since segmentation lines are established first in the anterior region, cells in that region are the first to rearrange themselves into triangular configurations. Shortly afterwards, while the caudal line of segmentation is being established, cells in the posterior region of a newly formed somite begin to change shape.

Fig. 9.

Schematic diagram of the changes in cell shape accompanying rosette formation at the time of somite segmentation. Only two . layers of cells are considered. See text for detailed explanation of the cell shape changes in each step.

Fig. 9.

Schematic diagram of the changes in cell shape accompanying rosette formation at the time of somite segmentation. Only two . layers of cells are considered. See text for detailed explanation of the cell shape changes in each step.

In Step II (Fig. 9), those triangular-shaped cells undergo another change in shape in order to complete the process of rosette formation. In more matured somites, cells are radially arranged so that those in the anterior and the posterior halves of the rosette become oriented more or less parallel to the long axis of the embryo (Step III). In order to become arranged in that fashion, the apices of the triangular-shaped cells around the outside margin of the somite appear to move towards the future anteroposterior axis (see the direction of arrows in Step II). This phenomenon provides one definitive example of a cell rotation which occurs during somite morphogenesis in amphibian embryos. Another example can be found in the rotation through 90° of myotomal cells of Xenopus laevis somites (Hamilton, 1969). Our SEM studies have demonstrated that rotating myotomal cells of Xenopus often exhibit bent configurations (Youn & Malacinski, 1981). That shape change may indicate that they have an intrinsic capacity for bending. Perhaps specially oriented or polarized microtubules (reviewed by Karfunkel, 1974; Trinkaus, 1976) participate in the process. In the case of rosette formation, however, cell motility mechanisms might be somewhat different. Cells retract, probably because they have a tendency to compensate for previous migratory movements towards the centre of the rosette (see Step I). In contrast to Xenopus somitogenesis, therefore, active cytoplasmic flow (in the direction of arrows in step II) accompanied by membrane turnover may be directly involved in the rotational reorientation of myotomal cells in Ambystoma and Pleurodeles embryos (Harris, 1973). Transformation of microfilaments from a network formation to bundles, or vice versa, during tail elongation and retraction as seen in fibroblasts in culture (Chen, 1981) could account for myotomal cell retraction during rosette formation.

A question then arises concerning the motive force that drives cells to the prospective centre of the rosette. Morphogenetic movements of cell groups may be mediated by cell–cell interactions such as adhesion, contact inhibition, or changes in the shapes of firmly affixed individual cells (Phillips, Steinberg & Lipton, 1977). Alterations in the adhesive interactions between cells have been thought to be active factors in morphogenetic processes (Townes & Holtfreter, 1955; Gustafson & Wolpert, 1967). Bellairs, Curtis & Sanders (1978), Bellairs, Sanders & Portch (1980) have shown that when chick somites segment and the somite cells subsequently differentiate, the rate of cell-cell aggregation increases and the cells exhibit characteristic behaviour patterns. It was proposed from those observations that changes in cellular adhesiveness play an important role in somite formation. That proposal may imply that the surface of each cell is not uniformly adhesive during rosette formation. Rather, it is localized largely at one end near the prospective centre of the rosettes. Subsequently, clusters of cells in a newly formed somite aggregate together in the area of high adhesiveness. That presumably leads to the organization of cells into rosettes.

There is much evidence that the molecules which determine adhesive specificity may reside at or external to the cell surface. For example, research on aggregation factors has suggested that specific glycoproteins presumably located at the cell surface are responsible for mediating cell–cell adhesion in embryonic chick neural tissue (Thiery, Brackenbury, Rutishauser & Edelman, 1977) and cellular slime moulds (Gerisch, 1976). It is very likely that somitic cell rearrangements are also initiated to a significant degree by modifications in cell-surface adhesion determinants. In order to understand the mechanisms and forces which mediate and direct morphogenetic cell rearrangements accompanying rosette formation, it would be essential to seek further evidence for changes in the adhesion/recognition properties of the somitic cells and to relate these properties to specific cell-surface biochemical determinants.

Our SEM studies have revealed several major differences between somitogenesis in amphibian and in chick embryos. In chick embryos, the presence of mesenchymal cells in the mycoel, the appearance of somitomere-like structures in the paraxial mesoderm, the mode of sclerotome formation, and the loosely packed nature of the somitic cells all differ from the amphibian embryos studied here. Such variant features lead to the suggestion that amphibian and chick somites may employ different mechanisms for rosette formation. Extensive discussion of comparisons between amphibian and chick somitogenesis may ultimately be, therefore, misleading.

2. Cellular basis of myoblast fusion

Extensive changes in cellular arrangement were shown to occur during the initial period of fusion (Fig. 7a). Radially arranged cells in the medial side of the rosette appeared to retract their apices from the centre of the rosette and become arranged anteroposteriorly. Those in the lateral side also began to undergo rotational reorientation at about the same developmental stage (Figs. 7b, c). Medial cells of the rosette become elongated anteroposteriorly before the lateral ones do. This mediolateral directionality can also be observed during amphibian somite morphogenesis in the following two instances: (1) myoblast fusion, and (2) formation of dermatome in the lateral side and of sclerotome in the medial side. It would be interesting to determine when and how information on the directional morphogenesis is given to the somitic and/or presomitic cells during development. Somite reversal experiments (Deuchar & Burgess, 1967) are currently being undertaken to obtain such information.

While the fusion process is under way in the medial side, somites are composed of two layers of spindle-shaped myoblasts tandemly arranged in the anterior and the posterior halves of the rosette. Our SEM observations indicate that fusions take place between two cells from each layer. Are the resulting fused cells dinucleate? According to Holtfreter (1965), most of the myotubes of Ambystoma maculatum somites contain five to seven nuclei. Muchmore (1965) has examined mitoses in developing myotomes of Ambystoma maculatum. He found that the nuclei of myoblasts divide rapidly until the mid-tailbud stage, when mitosis ceases and the cells fuse to form the characteristic multinucleate muscle fibres. Even prior to fusion, therefore, nuclear division in the absence of cytoplasmic partitioning has probably occurred in the myoblasts. The multiple fusions of mononucleate myoblasts suggested by Loeffler (1965) do not appear to take place. Our SEM results (e.g. Fig. 7) failed to reveal signs of multiple fusions between myoblasts.

When the cells undergo fusion, they elaborate cell processes which may play an important role in the fusion process itself. Similar types of cell processes were observed in each of the amphibian species included in this study. In the present studies cells were observed to send out either long processes (see arrows in Figs. 7e, 8a and 8e) or thin, broad processes (see arrows in Figs. 7d and 8c). It can be speculated that the appearance of such long processes is characteristic of cells which are in the initial phase of fusion. Conversely, the presence of thin, broad processes may indicate that cells are near the completion of fusion. Close apposition of membranes by means of such cell processes appears to characterize the early events of membrane fusion in many experimental systems. A direct morphogenetic role of surface cell processes produced during the process of cell fusion has been suggested for neural fold fusion (Bancroft & Bellairs, 1975; Mak, 1976; Waterman, 1975) and for fusion in cultured muscle cells (Fischman, 1970; Shimada, 1971,1972). Ultrastructural alteration of membranes accompanying this process has also been the subject of several recent reports (Shimada, 1971; Lipton & Konigsberg, 1972; Rash & Fambrough, 1973). However, the detailed cellular mechanisms of fusion still remain obscure.

3. Comparative aspects of amphibian somite morphogenesis

Table 1 summarizes different aspects of somite formation in four species of amphibia. Embryos of the urodele species, Ambystoma and Pleurodeles, exhibit exactly the same pattern of somite morphogenesis. A few differences exist among the anuran species, Xenopus and Rana. However, there are many differences in somite formation between the anuran and urodele species. No comprehensive comparisons will be made for each aspect listed in the table, since many of the details have already been mentioned. Further discussion concerning the establishment of segmentation lines and the ploidy of myoblasts is, however, warranted. First, the shape of newly formed somites in Xenopus and Rana appears to be arrowhead-like. Youn & Malacinski (1981) have suggested that the formation of ‘arrowhead-like’ somites may be caused by extensive differential movements that occur between the prospective neural area, epidermis, and prospective mesodermal mantle during neurulation. As a result, the segmentation lines are made diagnonally. In the case of the urodele species, the somites appear to be formed by transverse fissures, giving the somites a ‘stump-like’ shape. This may be due to the fact that differential movements are not as prominent in Ambystoma and Pleurodeles as in Xenopus and Rana. The question of how different shaping mechanisms might affect the intrasomitic cellular arrangement pattern remains obscure.

Table 1.

Comparative aspects of amphibian somite morphogenesis

Comparative aspects of amphibian somite morphogenesis
Comparative aspects of amphibian somite morphogenesis

Second, with regard to the ploidy of myoblasts and myotubes, it is known that Xenopus rotated myotomal cells remain uninucleate until about stage 45 (Kielbowna, 1975). Nuclear division then follows, and mutinucleate myotubes are formed. Conversely, in Ambystoma and Pleurodeles embryos, multinucleate myotubes remain in an interkinetic stage until the beginning of metamorphosis. It was, therefore, concluded that the myotomal cells of Xenopus were unusual in that they developed to a fully functional state and relatively large size while remaining uninucleate (Muntz, 1975). Multinucleation and the establishment of contractibility are obviously not causally related.

The pronounced differences in early embryonic development of the anuran and urodelean amphibians have recently been summarized (Chung & Malacinski, 1980). Nieuwkoop & Sutasurya (1976) have also indicated that variations are exemplified in many ways, including differences in mesoderm formation. But, perhaps most importantly, fundamental differences in primordial germ cell (PGC) formation exist between the two groups. Comparative analyses of the origin of the PGCs have led to speculation that the two groups of amphibians are only remotely related and could have originated polyphyletically from different ancestral fishes. Table 1 also displays striking differences between the two groups in the mode of somite formation and subsequent muscle development. However, it is perhaps not appropriate to speculate further on the origin of the modern Amphibia with those data. Rather, we would like to suggest that Rana may be intermediate between Xenopus and the urodele species, and that more diversity can be found among the anuran species than the urodele species. The pattern of somitogenesis may provide a useful indicator for determining how different families of amphibians evolved. Further clues as to the origin of modern Amphibia may be obtained by examining and comparing various aspects of somitogenesis among teleosts, reptiles, and other amphibians.

We wish to express our gratitude to Dr Ray Keller for his valuable advice and continuous encouragement during the course of this investigation. We wish to thank Dr John Frost, Dr Sally Frost and Ms Fran Briggs (I.U. Axolotl Colony) for providing Rana, Pleurodeles and Ambystoma embryos, respectively. We also thank Ms Diane Malacinski for assisting in the preparation of the manuscript. This work was initiated with support from NSF PCM 77-04457 and completed with NSF PCM 80-06343.

Bancroft
,
M.
&
Bellairs
,
R.
(
1975
).
Differentiation of the neural plate and neural tube in the young chick embryo
.
Anat. Embryol
. (2. Anat. Entwbesch.)
147
,
309
335
.
Bellairs
,
R.
(
1979
).
The mechanisms of somite segmentation in the chick embryo
.
J. Embryol. exp. Morph
.
51
,
227
243
.
Bellairs
,
R.
,
Curtis
,
A. S. G.
&
Sanders
,
E. J.
(
1978
).
Cell adhesiveness and embryonic differentiation
.
J. Embryol. exp. Morph
.
46
,
207
213
.
Bellairs
,
R.
,
Sanders
,
E. J.
&
Portch
.
P. A.
(
1980
).
Behavioural properties of chick mesoderm and lateral plate when explanted in vitro
.
J. Embryol. exp. Morph
.
56
,
41
58
.
Bordzilovskaya
,
N. P.
&
Dettlaff
,
T. A.
(
1979
).
Tables of stages of the normal development of axolotl embryos and the prognostication of timing of successive developmental stages at various temperatures
.
Axolotl Newslett
.
7
,
2
22
.
Chen
,
W.T.
(
1981
).
Mechanism of retraction of the trailing edge during fibroblast movement
.
J. Cell. Biol
.
90
,
187
200
.
Chung
,
H. M.
&
Malacinski
,
G. M.
(
1981
).
A comparative study of the effects of egg rotation (gravity orientation) and UV irradiation on anuran vs. urodele amphibian eggs
.
Differentation
,
18
,
185
189
.
Cooke
,
J.
(
1975
).
Control of somite number during development of a vertebrate, Xenopus laevis
.
Nature
254
,
196
199
.
Cooke
,
J.
(
1978
).
Somite abnormalities caused by short heat shocks to pre-neurula stages of Xenopus laevis
.
J. Embryol. exp. Morph
.
45
,
283
294
.
Deuchar
,
E. M.
&
Burgess
,
A. M. C.
(
1967
).
Somite segmentation in amphibian embryos: Is there a transmitted control mechanism?
J. Embryol. exp. Morph
.
17
,
349
358
.
Elsdale
,
T.
&
Pearson
,
M.
(
1979
).
Somitogenesis in amphibia. II. Origins in early embryogenesis of two factors involved in somite specification
.
J. Embryol. exp. Morph
.
53
,
254
267
.
Elsdale
,
T. R.
,
Pearson
,
M.
&
Whitehead
,
M.
(
1976
).
Abnormalities in somite segmentation following heat shock to Xenopus embryos
.
J. Embryol. exp. Morph
.
35
,
625
635
.
Fischman
,
D. A.
(
1970
).
The synthesis and assembly of myofibrils in embryonic muscle
.
Curr. Top. devl. Biol
.
5
,
235
280
.
Gallien
,
L.
&
Durocher
,
M.
(
1957
).
Table chronologique du développement chez Pleurodeles waltlii Michah
.
Bull. Biol
.
2
,
97
114
.
Gerisch
,
G.
(
1976
).
Membrane sites implicated in cell adhesion: their developmental control in Dictyostelium discoideum
.
In International Cell Biology
(ed.
B. R.
Brinkley
&
K. R.
Porters
), pp.
36
42
.
New York
:
Rockefeller University Press
.
Gustafson
,
T.
&
Wolpert
,
L.
(
1967
).
Cellular movement and contact in sea urchin morphogenesis
.
Biol. Rev
.
42
,
442
498
.
Hamilton
,
L.
(
1969
).
The formation of somites in Xenopus
.
J. Embryol. exp. Morph
.
22
,
253
264
.
Harris
,
A. K.
(
1973
).
Cell surface movements related to cell locomotion
.
In Locomotion of Tissue Cells (Ciba Foundation Symp. 14)
.
New York
:
Elsevier/North-Holland
.
Holtfreter
,
J.
(
1965
).
Differentiation of striated muscle cells in vitro
.
Amer. Zool
.
5
,
719
.
Karfunkel
,
P.
(
1974
).
The mechanisms of neural tube formation
.
Int. Rev. Cytol
.
38
,
254
271
.
Keller
,
R. E.
(
1976
).
Vital dye mapping of the gastrula and neurula of Xenopus laevis. II. Prospective areas and morphogenetic movements of the deep layer
.
Devl Biol
.
51
,
118
137
.
Kielbowna
,
L.
(
1975
).
Utilization of yolk platelets and lipid bodies during the myogenesis of Xenopus laevis (Daudin)
.
Cell Tissue Res
.
159
,
279
286
.
Lipton
,
B. H.
&
Konigsberg
,
I. R.
(
1972
).
A fine-structural analysis of the fusion of myogenetic cells
.
J. Cell Biol
.
53
,
348
364
.
Loeffler
,
C. A.
(
1965
).
Homoplastic and xenoplastic multinucleate myotubes produced by the fusion of anuran and urodele mononucleate myoblasts
.
Amer. Zool
.
5
,
204
.
Loeffler
,
C. A.
(
1968
).
Evidence for the fusion of myoblasts in amphibian embryos. I. Homoplastic transplantations of somitic material labeled with tritiated thymidine
.
J. MorpA
.
128
,
403
426
.
Loeffler
,
C. A.
(
1970
).
Evidence for the fusion of myoblasts in amphibian embryos. II. Xenoplastic transplantations of somitic cells from anuran to urodele embryos
.
J. Morph
.
130
,
491
500
.
Mak
,
L. L.
(
1976
).
Scanning and transmission electron microscopic observations of neural fold fusion in amphibians
.
J. Cell Biol
.
70
,
224a
.
Meier
,
S.
(
1979
).
Development of the chick embryo mesoblast: formation of the embryonic axis and establishment of metameric pattern
.
Devl Biol
.
73
,
25
45
.
Muchmore
,
W. B.
(
1965
).
Mitoses in developing myotomes of Ambystoma maculatum
.
Amer. Zool
.
5
,
721
.
Muntz
,
L.
(
1975
).
Myogenesis in the trunk and leg during development of the tadpole of Xenopus laevis (Daudin 1802)
.
J. Embryol. exp. Morph
.
33
,
757
774
.
Nieuwkoop
,
P. D.
&
Sutasurya
,
L. A.
(
1976
).
Embryological evidence for a possible polyphyletic origin of the recent amphibians
.
J. Embryol. exp. Morph
.
35
,
159
167
.
Pearson
,
M.
&
Elsdale
,
T.
(
1979
).
Somitogenesis in amphibian embryos. I. Experimental evidence for an interaction between two temporal factors in the specification of somite pattern
.
J. Embryol. exp. Morph
.
51
,
27
50
.
Phillips
,
H. M.
,
Steinberg
,
M. S.
&
Lipton
,
B. H.
(
1977
).
Embryonic tissues as elasticoviscous liquids. II. Direct evidence for cell slippage in centrifuged aggregates
.
Devl Biol
.
59
,
124
134
.
Rash
,
J. E.
&
Fambrough
,
D.
(
1973
).
Ultrastructural and electrophysiological correlates of cell coupling and cytoplasmic fusion during myogenesis in vitro
.
Devl Biol
.
30
,
166
186
.
Rugh
,
R.
(
1962
).
Experimental Embryology: Techniques and Procedures
, 3rd ed.
Minneapolis
:
Burgess
.
Shimada
,
Y.
(
1971
).
Electron microscope observations on the fusion of chick myoblasts in vitro
.
J. Cell Biol
.
48
,
128
142
.
Shimada
,
Y.
(
1972
).
Scanning electron microscopy of myogenesis in monolayer culture: a preliminary study
.
Devl Biol
.
29
,
227
233
.
Shumway
,
W.
(
1940
).
Stages in the normal development of Rana pipiens. I. External form
.
Anat. Rec
.
78
,
139
146
.
Thiery
,
J.-P.
,
Brackenbury
,
R.
,
Rutishauser
,
U.
&
Edelman
,
G.
(
1977
).
Adhesion among neural cells of the chick embryo. II. Purification and characterization of a cell adhesion molecule from neural retina
.
J. biol. Chem
.
252
,
6841
-
6845
.
Townes
,
P. A.
&
Holtfreter
,
J.
(
1955
).
Directed movements and selective adhesion of embryonic amphibian cells
.
J. exp. Zool
.
128
,
53
120
.
Trinkaus
,
J. P.
(
1976
).
On the mechanism of metazoan cell movement
.
In The Cell Surface in Animal Embryogenesis and Development
(ed.
G.
Poste
&
G. L.
Nicolson
), pp.
225
239
.
Amsterdam
:
North-Holland
.
Vogt
,
W.
(
1929
).
Gestaltungsanalyse am Amphibienkeim mit ortlichter Vitalfarbung. II. Teil. Gastrulation und Mesodermbildung bei Urodelen und Anuren
.
Wilhelm Roux Arch. EntwMech. Org
.
120
,
384
706
.
Waterman
,
R. E.
(
1975
).
SEM observations of surface alterations associated with neural tube closure in the mouse and hamster
.
Anat. Rec
.
183
,
95
98
.
Youn
,
B. W.
,
Keller
,
R. E.
&
Malacinski
,
G. M.
(
1980
).
An atlas of notochord and somite morphogenesis in several anuran and urodelean amphibians
.
J. Embryol. exp. Morph
.
59
,
223
247
.
Youn
,
B. W.
&
Malacinski
,
G. M.
(
1981
).
Somitogenesis in the amphibian Xenopus laevis\ scanning electron microscopic analysis of intrasomitic cellular arrangements during somite rotation
.
J. Embryol. exp. Morph. (In the Press
.)