/. Embryo/, exp. Morph. Vol. 51, pp. 227-243, 1979 227 Printed in Great Britain © Company of Biologists Limited 1979 The mechanism of somite segmentation in the chick embryo By RUTH BELLAIRS1 From the Department of Anatomy and Embryology, University College London SUMMARY The segmentation of somites in the chick embryo has been studied by transmission and scanning electron microscopy (stages 8—14). The segmental plate mesoderm consists of loosely arranged mesenchymal cells, whereas the newly formed somites are composed of elongated, spindle-shaped cells arranged radially around a lumen, the myocoele. The diameter of each somite is thus two cells plus the myocoele. Two major factors appear to be responsible for the change in cell shape at segmentation: (1) Each prospective somite cell becomes anchored at one end to the adjacent epithelia (i.e. the neural tube, the notochord, the ectoderm, the endoderm or the aorta) by means of collagen fibrils. These fibrils are already present in the segmental plate before the somites begin to form. (2) A change in cell-to-cell adhesiveness causes the free ends of these cells to adhere to one another. (Bellairs, Curtis & Sanders, .1978). This adhesion is then supplemented by the development of tight junctions proximally in the somite. Because it is anchored at both ends, each somite cell is under tension in much the same way as a fibroblast cell in tissue culture is under tension. Each somite cell therefore becomes elongated and the somite as a whole accommodates its general shape to that of the space available between the adjacent tissues. The arrangement of the cells in the more differentiated somites (stages 17-18) has also been examined and it has been found that the chick resembles Xenopus in that the myotome cells undergo rotation and become orientated in an antero- posterior direction. INTRODUCTION One of the most lively topics in developmental biology is that of how the mesoderm becomes segmented into somites. This problem has been tackled in four main ways. The first aims at disturbing the relationship between the prospective somite mesoderm and the adjacent tissues to determine which of these are necessary for segmentation. This type of experiment was originally based on the idea that the somites formed as the result of a specific embryonic induction by another tissue (e.g. by the notochord or neural tissue), but recently that idea has been modified and a more modern concept is that there is a pro- gramming of the mesoderm at an earlier stage and that the role of the notochord may be to help in 'stabilizing' the somites once they are formed (Lipton & Jacobson, 1974 a, b; Menkes & Sandor, 1977). 1 Author's address: Department of Anatomy and Embryology, University College, Gower Street, London WC1 6BT, U.K. 228 R. BELLAIRS The second approach is essentially a theoretical one and aims to solve such problems as that of how the number of somites is determined by the embryo. The most recent of these is the application of the mathematical' Catastrophe Theory' to amphibian embryos by Cooke & Zeeman (1976), which has received support from the experiments of Elsdale, Pearson & Whitehead (1976). Catastrophe Theory has also been applied to chick somitogenesis by Zeeman (1976). The third approach has been to compare the properties of unsegmented and segmented mesoderm to gain some idea of the changes that occur in the cells at segmentation. Bellairs et ah (1978), showed that as the mesoderm became segmented, its cells became more adhesive to one another. Similarly it was found that when pieces of unsegmented mesoderm were explanted in tissue culture, they behaved differently from segmented mesoderm (Bellairs & Portch, 1977; Bellairs, Sanders & Portch, 1980). The fourth approach has been to study the morphological changes that take place in somite segmentation. Although these were described by many of the earlier authors (e.g. Duval, 1889; Williams, 1910), the only recent major analysis was by Lipton & Jacobson (1974a,b) who used both light and transmission electron microscopy, but these authors were concerned with the first six pairs of somites only. No detailed account of segmentation by scanning electron microscopy has been given previously. The present paper is therefore concerned with an SEM study of somite segmentation. Particular attention will be paid to the role of the extracellular materials in somitogenesis. MATERIALS AND METHODS Hens' eggs were incubated for periods between 30 and 72 h so that the embryos were between about stages 8 and 18 of Hamburger & Hamilton (1951). Because the process of segmentation begins at the anterior end and spreads posteriorly, there is a gradient of developmental stages along the body axis, the anterior regions being in advance of the posterior ones. Four different developmental stages of somitic mesoderm (Fig. 1) were therefore distinguishable: (a) Unsegmented mesoderm, which is a thick band of tissue, the segmental or paraxial plate, which runs longitudinally down either side of the neural tube and notochord in the trunk region. (b) Transitional mesoderm, which is partly segmented; for example, two pairs of partly formed somites are present at stage 12. (c) Newly formed somites, which lie just anterior to the transitional mesoderm, and which have not yet begun to form dermo-myotomes and sclerotomes. (d) Mature somites, which are the most anteriorly situated somites and which have begun to form dermo-myotomes and sclerotomes. At stage 12, 16 pairs of somites are present but by stage 18 a further 20 or more pairs have differentiated and there is little unsegmented mesoderm Somite segmentation in the chick embryo 229 Differentiated somites Newly segmented somites Transitional region Lateral plate Segmental plate Fig. 1. Diagram to show the different regions of somitic mesoderm and lateral plate in a stage-14 embryo. remaining. The differentiation of most of the somites into dermo-myotomes and sclerotomes is well advanced in these older embryos. Fourteen specimens were prepared for transmission electron microscopy (TEM) and sixteen for scanning electron microscopy (SEM). Specimens for TEM were fixed in 2-5 % glutaraldehyde in 0-1M sodium cacodylate at a pH of 7-2 for 1-4 h and then washed three times in 0-1 M sodium cacodylate contain- ing 0-333 g CaCl2 for a total of l|h. They were treated with 1 % osmium tetroxide in phosphate buffer (pH 7-2 for 1 h at 4 °C then rinsed in phosphate buffer. After dehydration in graded ethanols, followed by two changes of propylene oxide, they were embedded in Araldite. Sections were stained with 2 % uranyl acetate at 38 °C for 2 min then counterstained with lead citrate. Twenty-seven specimens were fixed for periods of 4-24 h in 3 % glutaraldehyde in cacodylate buffer, at pH 7-2. After washing in cacodylate buffer, the specimens were immersed in 1 % osmium tetroxide for 30 min, washed again in buffer and dehydrated in graded ethanols. They were dried in a Polaron critical point drying apparatus from liquid CO2, mounted on stubs with UHU glue (Fishmar, Ltd, Waterford, Eire) and coated with gold. 230 R. BELLAIRS Somite segmentation in the chick embryo 231 RESULTS Figure 2 shows a recently formed somite cut transversely and viewed by SEM. It is triangular in section and comparable to the specimen illustrated with a light micrograph by Lipton & Jacobson (1974a; their fig. 8). Those cells which lie on the dorso-medial side of the somite and beneath the neural plate are arranged in a columnar manner, but the cells of the more ventral region are less well organized. There are many contacts between the somite and the neural plate and these consist mainly of extracellular materials (Fig. 3). Similar materials are present between the somite and the endoderm. A large part of these extracellular materials consists of fibrils which are probably collagenous (see Discussion). Similar fibrils are also present between the lateral plate mesoderm and the underlying endoderm. It is well known that soon after the first somites have formed, the neural folds rise up towards one another, and the dorso-medial wall of each somite rises with it (Lipton & Jacobson, 1974a) eventually becoming the vertically orientated medial wall of each somite (Williams, 1910). In this way these first formed somites change shape and acquire the rosette shape which is generally considered to be a characteristic of all somites. When seen in longitudinal section a somite is typically rosette-shaped (Fig. 4) though in transverse section it is box or wedge-shaped (Fig. 5). Lipton & Jacobson (1974a) have already pointed out that only the first three pairs of somites are formed in association with a flat neural plate, and that these are FIGURES 2-7 Fig. 2. SEM micrograph of part of a stage-8 embryo. The specimen has been fractured across the 3rd somite. Note the collagen fibrils (/) which lie between the dorsal side of the somite (s) and the neural plate (/?). e, Endoderm; Ip, lateral plate, x 650. Fig. 3. Enlargement of part of Fig. 2 to show the collagen fibrils. Fig. 4. SEM micrograph of a longitudinal section through a somite of a stage-1.1 embryo. Spindle-shaped cells are arranged around a lumen (/) which contains other cells. Extracellular materials cover the surface of the somite, n, Neural plate; no, Notochord. x 585. Fig. 5. SEM micrograph of a transverse section through a somite and the associated lateral plate mesoderm of a stage-] 2 embryo. The aorta (a) and endoderm (e) are visible but other tissues have been removed, x 780. Fig. 6. SEM micrograph of the centre of a transversely broken somite from a stage-14 embryo. The myocoele is packed loosely with mesenchymal cells (arrowed), which make contact with the spindle-shaped cells of the somite (sp).