ABSTRACT
Scanning electron microscope studies, supported by transmission electron microscope and light microscope observations, have been made of the wing-bud apex in the region of the apical ectodermal ridge (AER) in both normal and talpid3 mutant embryos.
Ln normal embryos: the ectoderm consists of two layers, the periderm and the basal layer, resting upon a basal lamina. The external surface of the periderm cells bears numerous villi, especially numerous at the cell boundaries. In the AER the cells of the basal layer are compacted into a fan-shaped form, remaining as a single layer, though displacement of their nuclei gives the appearance of a stratified epithelium; the periderm cells tend to round up and many are necrotic. Microfilaments and microtubules are present in greater numbers in the AER than in other ectoderm cells and are orientated along the long axis in the very narrow elongated cells which comprise the middle of the ridge. The mesenchymal cells of the mesoderm tend to be flattened, with extensive flattened surface areas separated from each other by long edges of lamellar cytoplasm, from which many long filopodial extensions arise. Often the cells are elongated, with the filopodia arising predominantly at the anterior and posterior ends.
In talpid3 embryos : when these were compared with normals no differences were detected in the general ectoderm or in the AER, but in the mesoderm a statistical analysis revealed significant differences in the occurrence of sections through very fine cytoplasmic processes in transmission electron microscope micrographs. Taken in conjunction with visual examination of scanning electron microscope pictures the differences suggested that talpid3 cells have filopodia distributed more extensively around the cell, but that these extensions do not extend so far through the intercellular spaces or produce such fine terminal arborizations as in normal mesenchyme.
At the ectodermal/mesodermal boundary there is a web of extracellular material under the basal lamina, separated from the underlying mesoderm by a very narrow gap in normal embryos. Fine cytoplasmic extensions from the mesenchyme cells extend across the gap towards the basal lamina and many are in contact with it. The gap is wider in talpid3 wing-buds and a statistical analysis confirmed that far fewer filopodia approach closely to the basal lamina or make contact with it.
The significance of these findings in relation to problems of AER formation and activity and motility of mesenchymal cells in the wing-bud is discussed.
INTRODUCTION
Classical studies on limb morphogenesis were mainly concerned with the relations between tissues, in particular with the interaction between mesoderm and ectoderm (reviewed by Zwilling, 1961), but more recently the emphasis has shifted towards concentration upon the activity of individual cells and to the relations between them (e.g. Ede, 1971, 1972). Hitherto, our knowledge of the structure and arrangements of these cells in the limb-bud has been based largely upon the two-dimensional views furnished by light and transmission electron microscopy (e.g. Jurand, 1965; Gould, Day & Wolpert, 1972) and no threedimensional study using the scanning electron microscope has been published.
The activity and appearance of living mesenchyme cells derived from primary explants of limb-bud fragments in vitro have been investigated by Ede & Flint (1974a), who noted differences between the structure and behaviour of cells from normal chick embryos and those from talpid3 a mutant in which the pattern of development of the limb-bud is strikingly abnormal (Ede & Kelly, 1964; Hinchliffe & Ede, 1967; Hinchliffe & Thorogood, 1974). In order to decide whether these differences are reflected in the cells of whole embryosand whether they are related to the differences in morphogenesis between normal and mutant limb-buds, it is necessary to compare the normal and talpid3 cells both in vivo and in vitro.
In this paper we report a scanning electron microscopic study of the wing-buds of normal and talpid3 embryos examined after fixation in vivo. This is supplemented by studies using both transmission electron microscopy and light microscopy, and also by a statistical analysis of the electron micrographs. A report on the structure of normal and talpic3 cells cultured in vitro is in preparation.
MATERIALS AND METHODS
Ten normal ( + /+ or ta3/ + ) control and ten talpid3 (ta3/ta3) embryos were used in this investigation, obtained either from the flock maintained in Glasgow by D. A.E. or from the related flock kept by Dr J. R. Hinchliffe at Aberystwyth. The control embryos were the phenotypically normal siblings of the talpid3mutants. Eggs were incubated for 5 days, by which time they had reached stages 24 – 26 (Hamburger & Hamilton, 1951) when the mutant embryos could be easily distinguished by the expanded fan shape of the wing-buds (Ede & Kelly, 1964).
Wing-buds of both normal and talpid3 embryos were fixed for scanning electron microscopy (SEM), for transmission electron microscopy (TEM) and for thin-section light microscopy.
Preparation for SEM
Two methods were used:
(Seven normal and seven talpid3 embryos.) Wing-buds were fixed for 1 h in Karnovsky’s fixative (Karnovsky, 1965), washed for 15 min in cacodylate buffer and treated with 1 % osmium tetroxide. They were rinsed in 70 % ethanol and stored in 70 % ethanol. Each limb was cut into two parts vertically along the proximo-distal axis with a razor blade or a new scalpel blade and dehydrated through an ethanol series followed by an ethanol-freon series. Embryos were dried in a critical point ‘bomb’ from liquid CO2 before being mounted on aluminium stubs with UHU glue. Specimens were coated with carbon and gold and then examined in a Cambridge ‘Stereoscan’ MKI operated at 10 kV.
(Three normal and three talpid3 embryos.) Wing-buds were fixed in 3 % glutaraldehyde in 0·15 M cacodylate buffer at 37 °C for 2 h, then allowed to cool at room temperature and continue fixing overnight, giving a total fixation time of about 24 h. They were washed for 30 min in distilled water and then dehy-drated and treated as in the first series.
Preparation for TEM and light microscopy
After each egg was opened the embryo was removed quickly, the wing-buds snipped rapidly from the body and plunged into the fixative at pH 7·2–7·4 at room temperature. Specimens were usually fixed in Karnovsky’s fixative with either phosphate or cacodylate buffers but three normal and three talpid3 limb-buds were fixed instead by the technique used by Burnside (1971). Similar results were produced by the two fixatives, except that the microtubules appeared to be preserved more readily by the Burnside fixative. In either case, fixation was for 1 h, followed by a number of washes in the buffer used previously as a component of the fixative. After immersion in 1 % osmium tetroxide at pH 7·2–7·4 specimens were given several brief washes in maleic acid buffer at pH 5·15 before treatment for 1 h in 1 % uranyl acetate, made up in 6% maleic acid. The specimens were then dehydrated with graded ethanols and embedded in either Araldite or Spurr epoxy resin. Sections were cut on a Porter-Blum Mark II ultramicrotome and examined in a Siemens Elmiskop 1 electron microscope. Sections about 1 μm thick were also cut and, after staining with a 1 % solution of toluidine blue, were examined by light microscopy.
Statistical investigation
This was carried out entirely on material fixed with Karnovsky’s fixative. Large montages at magnifications of about 6000–9000 × each usually consisting of 10–15 consecutive photographs, were prepared of areas of ectoderm and mesoderm adjacent or nearly adjacent to, but not including, the apical ectodermal ridge (AER). This region was selected as being readily located so that valid comparisons could be made between areas from normal and talpic3 embryos.
These montages were used in making a statistical analysis, applying the sampling technique described by Bellairs (1959). In this technique, large transparent sheets of polythene were divided by fine intersecting ink lines into in squares. The polythene sheets were pinned over the montages and the structure which lay at each intersection was noted (see Fig. 1). The aim was to compare the incidence of a number of cellular structures or of intercellular spaces in the normal and talpic3 wing-buds. Each intersection was scored for one of the following categories:
Diagram to illustrate the sampling technique used in the analysis of the electron micrographs. A polythene sheet, marked with a grid of intersecting lines, is placed over the photograph. A count is made of the number of intersections falling respectively on intercellular spaces, on nuclei, and on cytoplasm. The cytoplasm is itself treated as being of three types: the mitochondria, processes less than 0·5 μm in diameter, and the remainder of the cytoplasm.
Diagram to illustrate the sampling technique used in the analysis of the electron micrographs. A polythene sheet, marked with a grid of intersecting lines, is placed over the photograph. A count is made of the number of intersections falling respectively on intercellular spaces, on nuclei, and on cytoplasm. The cytoplasm is itself treated as being of three types: the mitochondria, processes less than 0·5 μm in diameter, and the remainder of the cytoplasm.
(i)nuclei (including nucleoli), (ii) mitochondria, (iii) cytoplasmic processes less than 0·5 μm in diameter, (iv) the remaining cytoplasmic components, and (v) intercellular spaces. Separate scores were made for ectoderm (basal layer only) and mesenchyme. A further examination of the montages was made in order to investigate the relationship between the ectoderm and the underlying mesenchyme in normal and talpid3 wing-buds in this region. A count was made of the number of fine mesenchymal processes touching the ectodermal basal lamina or lying within 5–6 μm of it. The length of the strip of ectoderm in each montage was obtained by means of a map measurer graded in mm.
The statistical analysis is set out in Tables 1 and 2. In Table 1 the data for the grid counts is set out as a table of averages with frequency measured per 100 intersections of the grid. The F value is calculated and the probability (P) of that value having occurred by chance is obtained.
Variance ratio tests on grid sampling data from montages of ectoderm (basal layer) and underlying mesoderm

RESULTS
General appearance
The characteristic shapes of the normal and talpid3 wing-buds are shown in Figs. 2, 3, 12 and 13. Only half of each limb is present; in both the normal and the talpid3 the AER is a conspicuous feature extending around the perimeter of the distal part of the bud.
Figs. 2,3. Low magnification SEM views of forelimb-buds of normal and talpid3embryos, respectively. Only half of each limb-bud is shown, the cut surface passing proximo-distally at right angles to the apical ectodermal ridge, aer, Apical ectodermal ridge; d, distal region of limb-bud; p, proximal region of limb-bud. x 140.
Figs. 4, 5. SEM views of the surface of the ectoderm and apical ectodermal ridge of normal and talpid3 embryos respectively. The individual cells can be seen as raised areas, × 800.
Figs. 2,3. Low magnification SEM views of forelimb-buds of normal and talpid3embryos, respectively. Only half of each limb-bud is shown, the cut surface passing proximo-distally at right angles to the apical ectodermal ridge, aer, Apical ectodermal ridge; d, distal region of limb-bud; p, proximal region of limb-bud. x 140.
Figs. 4, 5. SEM views of the surface of the ectoderm and apical ectodermal ridge of normal and talpid3 embryos respectively. The individual cells can be seen as raised areas, × 800.
Higher magnification SEM views of the ectoderm near the apical ridge of normal and talpicF embryos, respectively. Villi can be seen as white spots on the surface of the cells and at the junctions between the cells, × 4000.
Figs. 8, 9. SEM views of ectoderm and underlying mesenchyme taken from regions proximal to the apical ridge of normal and talpid3 embryos, respectively, ×16000.
Higher magnification SEM views of the ectoderm near the apical ridge of normal and talpicF embryos, respectively. Villi can be seen as white spots on the surface of the cells and at the junctions between the cells, × 4000.
Figs. 8, 9. SEM views of ectoderm and underlying mesenchyme taken from regions proximal to the apical ridge of normal and talpid3 embryos, respectively, ×16000.
TEM views of ectoderm and underlying mesenchyme just proximal to apical ectodermal ridge in normal and talpic3 limb-buds, respectively. Note that the gap between the mesenchyme cells and the ectoderm is less in the normal than in the talpid3 b, cell of basal layer; p, periderm; s, intercellular space, × 8000.
TEM views of ectoderm and underlying mesenchyme just proximal to apical ectodermal ridge in normal and talpic3 limb-buds, respectively. Note that the gap between the mesenchyme cells and the ectoderm is less in the normal than in the talpid3 b, cell of basal layer; p, periderm; s, intercellular space, × 8000.
Figs. 12, 13. SEM views of the cut surface of normal and talpic3 limb-buds, respectively. Compare with Figs. 1 and 2 in which similar limb-buds are seen from a different orientation, × 370.
Figs. 14,15. Photomicrographs taken by light microscopy of sections through the tips of the limb-buds of a normal and a talpid’A embryo, respectively. Stained with toluidine blue. The arrow indicates the gap between ectoderm and mesoderm in the talpicP. × 225.
Figs. 16, 17. SEM views of apical ectodermal ridge and underlying mesenchyme in a normal and a talpid3 limb-bud, respectively. Note the fine processes forming a web between the ectoderm and the mesenchyme, × 1 350.
Figs. 12, 13. SEM views of the cut surface of normal and talpic3 limb-buds, respectively. Compare with Figs. 1 and 2 in which similar limb-buds are seen from a different orientation, × 370.
Figs. 14,15. Photomicrographs taken by light microscopy of sections through the tips of the limb-buds of a normal and a talpid’A embryo, respectively. Stained with toluidine blue. The arrow indicates the gap between ectoderm and mesoderm in the talpicP. × 225.
Figs. 16, 17. SEM views of apical ectodermal ridge and underlying mesenchyme in a normal and a talpid3 limb-bud, respectively. Note the fine processes forming a web between the ectoderm and the mesenchyme, × 1 350.
Ectoderm
Figures 4 and 5 show surface views of the AER and of the ectoderm adjacent to it. In both normal and talpid3 the cells vary in size and shape. Individually, they bear many more villi than would be expected from a casual inspection of transmission electron micrographs, and the junctions between cells are particularly well defined by the large number of villi which appear at the cell margins (Figs. 6, 7). Jurand (1965), who carried out a careful examination of the normal limb-bud from transmission electron micrographs, produced a graphic reconstruction of the outer layer of ectoderm cells which these observations show to be strikingly accurate.
Light microscope studies (Figs. 14, 15), SEM pictures (Figs. 8, 9) and TEM micrographs (Figs. 10, 11) show that the general ectoderm (i.e. adjacent to the ridge in our studies) consists of two layers -the outer periderm, whose appearance in surface view is described above, and the deeper basal layer, which rests upon a basal lamina. The cells of the basal layer are columnar and arranged in a loose epithelium with large intercellular spaces separating them, except at the basal lamina where they are in contact with their neighbours all round; there are some other points of contact randomly distributed over the main body of each cell.
The much thinner periderm of the ectoderm is composed of a single layer of flattened cells, with correspondingly flattened nuclei, orientated at right angles to the cells of the basal layer and in close contact with them. The nuclei make bulges which are clearly visible in the SEM pictures. Each cell makes close contact with its neighbouring periderm cells around the whole of its periphery and the junction between periderm cells usually shows a number of interlocking protuberances and indentations.
The specialized contacts between ectoderm cells have already been described by Jurand (1965) and consist of tight junctions, intermediate junctions and desmosomes. Microfilaments and microtubules are present in both basal and periderm layers of the general ectoderm cells but they occur less frequently here than in the cells of the AER.
No differences are apparent between normal and talpid3 ectoderm in either TEM or SEM pictures and none emerge from the quantitative data presented in Table 1.
Apical ectodermal ridge
SEM pictures show that the cells of the AER are more irregular and tend to appear larger at the surface than those of the adjacent ectoderm (Figs. 4, 5). There appear to be fewer villi.
Figures 14 and 15 are typical thick sections cut through the apical region of wing-buds of a normal and talpic3 embryo respectively, stained with toluidine blue and viewed with the light microscope. They do not appear to differ greatly from each other; in each the AER is a thickened region measuring about 25–35 //m at the apex, and though in some talpid3 specimens the ridge is flatter and the central groove less conspicuous it is likely that this difference is due merely to variations in the angle of sectioning.
In TEM pictures the basal layer usually gives the appearance of being several cells deep because the nuclei lie at various levels. It has been possible, however, to trace some cells through from the basal to the distal limits and this is confirmed in the SEM pictures which show that individual cells extend from the basal lamina to the periderm, just as they do in the adjacent ectoderm. The nuclei become displaced at different levels, giving a false appearance of a stratified epithelium.
The medial part of the AER is the thickest and here the individual cells of the basal layer are extremely narrow and elongated. They are characterized by having many microfilaments, each about 5 run in diameter, which are orientated mainly in the longitudinal direction (Fig. 23). By contrast, there were no bundles of microfilaments running transversely along the base of the cells, whose orientation might have supported a ‘purse-string’ theory of AER formation similar to that which is now widely held to explain the rolling-up of the neural tube (see discussion by Bellairs, 1974). Microtubules are also present in these cells of the apical ectoderm (Fig. 22). Each measures about 20 nm in diameter and, like the microfilaments, the tubules run predominantly in a longitudinal direction. There appear to be considerably more microfilaments and microtubules present in the ridge cells than in the cells of the general ectoderm. The mitochondria, Golgi bodies and endoplasmic reticulum are all similarly extended along the long axis in these middle cells. A further characteristic is that in many specimens the basal lamina is interrupted in this region and is continuous with extracellular material lying between the cells (Figs. 18, 19).
Figs. 18, 19TEM views of the base of the apical ectodermal ridge of normal and talpic3 embryos, respectively. Note the whisps of granular material, collagen fibrils and small cytoplasmic processes of the mesoderm (m). Not all sections through normal limb-buds show the mesenchyme cells so closely pressed to the ectoderm (e) in this region, but in talpic3 limb-buds such a close apposition has never been seen, × 15000.
Figs. 20, 21. TEM micrographs of the mesenchyme from normal and talpic3 limb-buds, respectively. × 3000.
Figs. 18, 19TEM views of the base of the apical ectodermal ridge of normal and talpic3 embryos, respectively. Note the whisps of granular material, collagen fibrils and small cytoplasmic processes of the mesoderm (m). Not all sections through normal limb-buds show the mesenchyme cells so closely pressed to the ectoderm (e) in this region, but in talpic3 limb-buds such a close apposition has never been seen, × 15000.
Figs. 20, 21. TEM micrographs of the mesenchyme from normal and talpic3 limb-buds, respectively. × 3000.
Fig. 22. TEM micrograph to show microtubules in the middle cells of the apical ectodermal ridge of a normal limb-bud. Similar microtubules are present in talpid3cells, × 45000.
Fig. 23. TEM micrograph to show microfilaments in the middle cells of the apical ectodermal ridge of a talpid3 limb-bud. Similar microfilaments are present in the cells of normal limb-buds, × 30000.
Figs. 24, 25. SEM views of distal mesenchyme from normal and talpid3 limb-buds, respectively. × 3400.
Figs. 26, 27. SEM views of part of proximal region of limb-bud of normal and talpid3 embryos, respectively. The mesenchyme (m) is pressed more closely beneath the ectoderm (e) in the normal than in the talpid3. × 3400.
Fig. 22. TEM micrograph to show microtubules in the middle cells of the apical ectodermal ridge of a normal limb-bud. Similar microtubules are present in talpid3cells, × 45000.
Fig. 23. TEM micrograph to show microfilaments in the middle cells of the apical ectodermal ridge of a talpid3 limb-bud. Similar microfilaments are present in the cells of normal limb-buds, × 30000.
Figs. 24, 25. SEM views of distal mesenchyme from normal and talpid3 limb-buds, respectively. × 3400.
Figs. 26, 27. SEM views of part of proximal region of limb-bud of normal and talpid3 embryos, respectively. The mesenchyme (m) is pressed more closely beneath the ectoderm (e) in the normal than in the talpid3. × 3400.
The basal layer cells on either side of the middle region of the AER radiate fanwise (Figs. 16-17). Individual cells are broader than the middle cells in section and their organelles do not show the same longitudinal orientation as in the middle cells. The microfilaments are less easy to see; it is possible that fewer of them are present but their random orientation renders them less conspicuous and they are further obscured by the presence of many more ribosomes in these cells.
SEM pictures of the AER show that the periderm cells become much less flattened than in the general ectoderm (Figs. 4, 5). Many of them, equally in both normal and talpid’3 limb-buds, become necrotic, with very electron-dense cytoplasm and large lysosomes. These cells have been described by Jurand (1965) in the normal limb-bud.
Mesoderm
The characteristic appearance of the mesoderm in light microscope sections in normal and talpid3 wing-buds is shown in Figs. 14 and 15. There is a tendency for the cells immediately beneath the epidermis to be elongated and orientated at right angles to it in the normal, but not so evidently in taipid3
TEM micrographs show the mesenchyme cells in both to be of characteristically irregular shape (Figs. 20, 21) in section. The nuclei are about 4–8 μm in diameter and usually of a relatively regular ovoid shape; they are smaller than those of the ectoderm and usually contain several patches of nucleolar material. The cytoplasm is rich in ribosomes but, like the ectoderm, contains little endo-plasmic reticulum. The mitochondria are about 0·5–1 μm in diameter and are usually round or oval in section. The nuclei are overlain over much of their surface by a very thin layer of cytoplasm with large cytoplasmic extensions at two or three points in the section, with much finer tapering extensions from the surface in these regions, chiefly at the tips. It is of course impossible to say from the sections whether these are thread-like or leaf-like, but thread-like cytoplasmic extensions (filopodia), cut in cross-section are obvious in the intercellular spaces. In the TEM micrographs it is impossible to visualize or estimate how long these filopodia are, but the SEM micrographs (Figs. 24, 25) show that these very fine cytoplasmic extensions are of considerable length, spanning the gaps between the cells and travelling in some cases over the surface of cells (though this particular appearance may be an artefact of the preparative treatment) with fine arborizations at their tips. When stereopairs of mesenchyme cells are examined with a stereo-viewer, it can be seen that each cell is in contact with many others. Furthermore, the contact is not confined to immediately adjacent cells, but may, by means of these long thread-like processes, extend over several cells.
It is not always easy to decide from these pictures whether any individual process extended from a cell is a true cytoplasmic filopodium or is composed of extracellular strands of material; TEM micrographs enable us to see that the processes are almost all cytoplasmic in the interior of the mesoderm but that both types are present at its junction with the ectoderm. Both microfilaments and microtubules are present in the cell cytoplasm and can be seen extending along the cytoplasmic processes.
From inspection of the TEM micrographs we gained the impression that the spaces between talpid3 mesenchyme cells were slightly larger than those between normal cells and also that they were much emptier, in the sense of their being occupied by fewer sectioned areas of the filopodial extensions, especially the very fine ones of less than 0·5 μm diameter. This impression is confirmed in the quantative data presented in Table 1. No significant difference is obtained for nuclei and mitochondria, suggesting that these cover an equal area and are found equally frequently in normal and talpic3. On the other hand, the figures for cytoplasmic processes less than 0·5 μm across, remaining cytoplasm and inter-cellular spaces are all significantly different, the greatest difference being in the relative numbers of the very fine cytoplasmic processes (2 × greater in normal than in talpic3 montages). These figures suggest that the nuclei and main cell bodies are not much, if any, more widely spaced in talpic3 than in normal mesenchyme but that the difference lies in the distribution of the cytoplasm, i.e. that where in talpic3 the space between cells is relatively empty and more likely actually to be scored as intercellular space, in normals it is more often traversed at grid points by a cytoplasmic extension, especially by a fine filopodium. This would suggest that either talpic3 cells are less endowed with fine filopodial processes than normal cells, or that the filopodia are shorter and therefore occur in fewer sections. It may also be significant that the origins of filopodia in talpic3 cells often appear to be ‘bunched’ at the ends of the larger cell extensions, giving the appearance of a sea-anemone with tentacles (Fig. 21); this would mean (see Fig. 1) that where one cytoplasmic extension was scored the chances of scoring its close neighbours would be reduced. From this analysis, therefore, it appears that talpic3 wing-bud mesenchyme cells differ from normals in having fewer or less extensive fine filopodial cytoplasmic processes.
More light is thrown on these differences by a detailed examination of the SEM pictures. These show, especially in the normal embryos (Figs. 8, 16, 24, 26), that the mesenchyme cells are flatter than might be expected. In the mass, they give somewhat the appearance of a heap of cornflakes, each mesenchyme cell having extensive flattened surface areas, bounded by long edges of lamellar cytoplasm, from which most of the filopodia arise either directly or from larger cytoplasmic extensions. Our impression is that normal cells tend to be elongated, with more filopodia arising from the lamellar edges at each end. There appear to be usually 3–4 flattened surfaces in most cells, reflected in the characteristic preponderance of cells appearing triangular or quadrilateral in section (Figs. 14, 20).
The general impression given by the talpid3 mesenchyme cells in SEM pictures is clearly different and may be described by saying that they appear, en masse (Fig. 13) and individually (Figs. 9, 17, 25), more ‘ragged’ than normal cells. There are fewer of the very fine filopodial extensions lying over their surface, but there are more filopodia arising from the lamellar cytoplasmic boundaries and their extensions all around the cell instead of predominantly at leading or trailing lamellae, with rather long uninterrupted lamellar regions between, as in the normal cells (compare Figs. 24 and 25). The ‘bunching’ of filopodial origins, which would account for the appearance mentioned in TEM micrographs, can be clearly seen in Fig. 25. Our impression is that the larger cytoplasmic extensions in normal cells are more finely drawn out than in talpic3 cells, with the filopodia branching off in a tree-like manner, whereas these extensions in talpid3are shorter, with the filopodia arising closer together. These differences are summarized in diagrammatic form in Fig. 28. Taken in conjunction with the quantitative evidence obtained from the TEM micrographs it appears that talpid3 cells have a larger number of filopodial extensions than normal cells and that these arise more generally around the edges of the cell, but that they do not extend so far through the intercellular spaces or produce such fine terminal arborizations as do the filopodia in the normal cells.
Diagrammatic comparison of normal and talpic3 wing-bud mesenchyme cells, partly in section (see text).
Junctions between mesenchymal cells are sometimes seen in both normals and talpid3, similar to those in the ectoderm, but no desmosomes have been seen connecting them in this region of the limb-bud.
Ectodermal/mesenchymal boundary
SEM micrographs (Figs. 12, 13, 16, 17), TEM micrographs (Figs. 10, 11) and light microscope sections (Figs. 14,15) all reveal that there is a slight gap between the ectoderm basal lamina and the underlying mesoderm in both normal and talpid3 embryos, but that this gap is much more extensive in the mutant, in which it frequently measures as much as 2 μm. In both normals and talpids the mesenchyme cells make contact with the basal lamina by means of many fine processes (Figs. 16, 17); some of these are as thin as 10 nm and they form a web between the mesenchyme and the ectoderm. Many of these processes are obviously cytoplasmic but the finest of them show a periodic structure in TEM micrographs at high magnification and almost certainly consist of collagen. Thus TEM micrographs show that the mesenchymal cells are separated from the ectoderm basal layer cells by the basal lamina, but also, beneath that, by strands of extracellular material, including collagen fibres. In Fig. 27 this extracellular material occupies much of the large gap between ectoderm and mesoderm in a talpid3 wing-bud. In the region immediately beneath the AER (Figs. 18, 19), the space contains wisps of granular material resembling the substance of the basal lamina (which, as already noted, is not continuous in this region) as well as fine collagen fibrils and small cytoplasmic processes.
The visual appearance of a much more extensive gap between ectoderm and mesoderm in talpid3 than in normal limb-buds is confirmed by the data presented in Table 2, where both the number of mesodermal cell processes approaching close to (within 5–6 μm) the basal lamina and the number actually touching it is much greater in normal than in talpid3 embryos. SEM micrographs suggest that in normal embryos contact with the basal lamina (Figs. 16, 26) is often made by an extensive region of one of the lamellar cytoplasmic cell edges mentioned above, and this is confirmed in TEM micrographs (Fig. 10). Such extended connexions do not appear to occur in the talpid3 wing-buds. In both normal and talpid3 embryos, though much more numerous in normal limb-buds (see Table 2, and compare Figs. 26 a.nd 27), are found the fine cytoplasmic processes which connect the basal lamina with the underlying mesoderm cells and which appear to be drawn taut, like guy ropes. Though the cell processes are in close contact with the basal lamina, none have been seen to extend through it.
DISCUSSION
Structure of the ectoderm and formation of the AER
Amprino (1965) and Amprino & Ambrosi (1973) have presented clear evidence that the ectoderm slides in a proximo-distal direction over the underlying mesoderm towards the apex of the limb-bud; they believe that the formation and maintenance of the AER depends, at least in part, on packing of the cells at the distal tip where cells from the dorsal and ventral sides meet, after the manner outlined, explicitly by Ede (1971).
The structural features of the non-ridge ectoderm and of the AER described here support this interpretation. The basal layer of the non-ridge ectoderm, with its columnar cells separated by wide gaps except at their base, forms in effect an extended caterpillar track which is well adapted to flow smoothly over the limb contours. The thin periderm layer, with its closely interdigitating pavement-like cells, forms a flexible cover over this. The deformation of the basal cells at the collision line would give the appearance we find in transverse sections of the AER: tall wedge-shaped cells, arranged fan-wise, which retain their connexion with the basal lamina, i.e. pseudo-stratified rather than stratified as suggested by Jurand (1965). The elimination of the large spaces between the cells of the basal layer and the compression of the cells would lead to the rounding off of the peridermal cells which we observed. The necrosis of many of these cells may be connected with their being forced out of their previous cell contacts. The gaps which occur in the basal lamina beneath the AER might also be due to compression of the ectodermal basal layer cells in this region.
Mesoderm
Structure of individual cells
The descriptive term generally applied to the mesenchyme cells in the chick has been ‘stellate’; e.g. this has been used to describe the primary mesenchyme of the early embryo (Trelstad, Hay & Revel, 1967) and the undifferentiated mesenchyme of the early limb-bud (Gould et al. 1972). It implies a more or less radially symmetrical cell with pointed cytoplasmic projections radiating from it and this is indeed the impression given by the appearance of the cells in section. It is perhaps the chief contribution of the SEM technique in this study to show that it is misleading: normal wing-bud mesenchyme cells, at any rate in the region we examined, appear to be rather flattened, often longer than broad and presenting extensive flat surfaces bounded by long edges of lamellar cytoplasm from which fine filopodial extensions arise. The picture is closer to that presented by limb-bud mesenchyme cells in culture (Ede & Flint, 1974A) than has been supposed, and it is possible that valid comparisons may be drawn between the lamellar edges of the cells in vivo and the lamellipodia (ruffled membranes) of the same cells in vitro.
Connexions between cells
The filopodia form an extremely complex system, connecting each cell not only with its immediate neighbours but with others several cells away and, in the case of cells near the surface of the limb-bud, with the basal lamina of the ectoderm. It is not possible to be quite clear where the terminations of the filopodia fall on the contacted cell, but many of them appear to be on the flat surfaces of the main cell bodies.
Activity and motility of the mesoderm cells
The cells in SEM pictures give the appearance of being in a highly active state, for it is unlikely that either the filopodia, or the fine lamellar edges of the cells from which they arise, are rigid or static structures. The question as to what part cell motility plays in the developing limb-bud has been raised by Ede & Agerbak (1968), Ede & Law (1969), Ede (1971) and subsequently by Gould et al. (1972). The appearance of the cells in SEM suggests that at least the peripheral regions of the cytoplasm are in a state of dynamic activity. The absence of desmosomes in these cells lends support to this interpretation.
Ectodermal I mesodermal junction
The two layers are completely separated by the basal lamina except at the region immediately beneath the AER where there is a number of perforations. Beneath the basal lamina there is a mat of very fine fibres, of which some at least have the characteristic banded appearance of collagen. Fine filo-podial cell extensions are also present, many of them attached or at least closely juxtaposed to it, and they appear to be stretched taut between the basal lamina and their mesodermal cell bodies. In normal embryos the connexion, as seen in both TEM micrographs (Fig. 10) and SEM pictures (Fig. 26), is often along a considerable length of a lamellar edge, but in talpid3 embryos the connexion is in almost all cases only by the tips of the filopodia. In the normals particularly, it appears that the filopodia are making contact with the basal lamina, adhering, and then increasing the contact beyond the original point. We are reminded of the inference drawn by Trelstad et al. (1967) from observations on primary mesen-chyme cells in the early embryo, that ‘filopodia extending from the advancing edge of the uppermost and lowermost mesenchymal cells seem to attach to the basement epithelial lamina (of the epiblast and hypoblast) as if they were feeling their way along this substratum’, In both normal and talpid3 wing-buds there is a small gap of the order of 4–6 μm in depth immediately beneath the central cells of the AER. Elsewhere in the parts of the limb-bud which we have examined, the gap is much narrower and often almost absent in normal embryos, but a clear gap between ectoderm and mesoderm is a marked feature of ta Ip id3 and explains why, in experiments involving removal of the ectodermal cap from the under-lying mesoderm following trypsinization, separation has always been found to occur more easily in the mutant.
Differences between normal and talpid3 cells
Two differences have emerged from this study: (a) in the morphology of individual cells; (b) in the extent of the gap between ectoderm and mesoderm.
The distinct morphology of the talpid3 cells is presumably a direct genetic effect. Other differences have been discovered and have been or will be described elsewhere: (i) in cell-to-cell adhesion (Ede & Agerbak, 1968; Ede & Flint, 19746); (ii) in rate of locomotion on plastic (Ede & Flint, 1974a); (iii) in resistance to injury and death (Ede & Flint, 1972). Such genetic evidence as is available suggests that, though no fine genetic analysis has been possible, talpid3represents a mutation of a single gene; our assumption is therefore that all of these effects at the cellular level are related to each other and to the effect of the mutation on morphogenesis. The relation between the morphological differences and the differences in cell adhesion and rate of locomotion is beginning to emerge, appearing to depend upon the differences in the number, structure and distribution of fine filopodial cytoplasmic extensions, and is discussed in detail elsewhere (Ede & Flint, 1974a, b).
The relatively more prominent ectodermal/mesodermal gap in talpid3 wing-buds must be a more indirect effect of the gene and there appear to be two possible explanations. (1) Ede & Kelly (1964) described a generalized oedema in later talpid3 embryos which developed from the 8th day. No oedema was observed before this stage but it is possible that a fluid pressure, sufficient to force the ectoderm away from the underlying mesoderm, has already built up at the stage we have examined here. (2) Perhaps the more extensive gap in talpid3 is caused by differences in the nature of the connexion of the mesodermal cells with the basal lamina, depending upon the differences in cellular morphology which we have observed.
We believe that the second hypothesis is more likely, since no obvious oedema has been observed until a much later stage, and since a gap is as much a feature of the normal wing-bud and is only exaggerated in the mutant. Gustafson & Wolpert (1961) described a mode of cell locomotion in sea-urchin morphogenesis: the mesenchyme cells concerned threw out fine cytoplasmic processes, up to 30 μm long, which exhibited random exploratory movements; if they did not make contact with the ectodermal body wall they collapsed, but if a successful contact was made the processes shortened and pulled the cell bodies towards the point of attachment. Essentially similar forms of cell locomotion have subsequently been found to occur in a wide variety of developmental situations (reviewed by Trinkaus, 1973) and the taut appearance of the filopodial processes connecting the mesenchymal cells with the basal lamina across the ectodermal/ mesodermal gap which we have found in normal embryos, together with the orientation of many subectodermal cells towards the ectoderm, makes it seem likely that the same activity occurs in the limb-bud. If this were so, the more extensive gap found in talpid3 embryos might again be attributed to differences in the structure and distribution of the filopodia referred to above, e.g. that since they were less extended they might be less effective in making connexions with the basal lamina, as suggested diagrammatically in Fig. 29. For whatever reason, it appears that in normals the mesenchymal cells approach closely to the ectoderm through out-reaching filopodial and lamellipodial processes, whereas in talpid3 wing-buds this close connexion fails to become established.
Diagrammatic representation of the ectoderm and underlying mesoderm in the region of the AER, with a comparison of the ectoderm/mesoderm boundary in normal and talpid3 wing-buds. ‘The collagen web’ contains collagen fibres embedded in additional extracellular material.
Outgrowth of the limb-bud
These observations may throw some light on the problem of limb-bud outgrowth. Ede & Law (1969) drew attention to the parameters at cell level affecting changes in shape of a developing organ, particularly the limb-bud, in developing a computer simulation programme. These were eight in number: cell number, proliferation, position, movement, size, shape, packing density and the constraint produced by the ectodermal layer on the underlying cell mass. Amprino (1965, 1973) has suggested that the main role in shaping limb-bud outgrowth is played by the last of these factors and Hornbruch & Wolpert (1970) have also taken this view; all of them consider that the main shaping force is the relatively rigid structure of the apical ectodermal ridge.
This is probably an over-simplification ; the shaping of the limb must be due to a balance between the forces exerted by all of the factors enumerated above. The distribution of cell divisions within the developing limb-bud might obviously be important, and Amprino (1965), Ede & Law (1969) and Hornbruch & Wolpert (1970) have found evidence for a gradient of mitosis within the limb-bud and used it in their models. Summerbell & Wolpert (1972) have proposed a hypothesis relating this gradient to the outgrowth of the ectoderm, which is controlled -i.e. given a firm shape -by the AER, thus giving the AER the key role.
It is certainly the case that the ectoderm, supported by the AER, does provide a relatively rigid jacket in which, in fact, dissociated cells can be packed to produce a fairly well-formed limb. But when ectoderm and mesoderm are dissociated after trypsin treatment, the mesoderm also maintains its shape and continues to do so until cells begin to migrate away from it. We may suppose that in the course of evolution, ectodermal outgrowth and underlying mesodermal growth have come to be correlated so as to act harmoniously without one being necessarily the sole driving force. Indeed, according to Amprino’s hypothesis of ridge formation by ectoderm sliding from two directions and colliding, the sliding must take place over a rigid determined form, provided presumably by the underlying mesoderm. In the earliest stages of limb-bud formation no ridge exists and we may suppose that in this case the limb-bud form is determined principally by the mesodermal cell behaviour, particularly by mitosis and possibly by orientated movement. The possibility that movement is important in limb-bud outgrowth was suggested by Ede & Agerbak (1968) and shown by Ede & Law (1969) to produce, in conjunction with a mitotic gradient, a good approximation to the limb-bud shape in computer simulations. Mitolo (1971) and Wilby (personal communication) have since shown by computer simulation that it is possible to produce a limb-bud-like outgrowth in the absence of cell movement, providing that the mitotic distribution is very precisely controlled in a particular way. But it is certainly much simpler and more flexible to produce outgrowth if movement is included. Moreover, exclusion of movement also eliminates the necessity, by implication, of postulating any modelling role for the ectoderm, which would produce outgrowth by forcing cell movement in certain directions by the constraints its structure imposed on simple expansion through cell division. On the whole it seems likely that cell movement is a factor in limb outgrowth, and the question is really whether the movement is active or passive, i.e. whether the modelling role of the ectodermal jacket is so predominating that the cells are squeezed forward like toothpaste in a tube, or whether they move actively towards the distal tip. The contribution of the present study to this problem is that the cells do not give the appearance of either static immobilized bodies or passively flowing particles; on the contrary, they appear to be in a state of considerable activity in the production of cytoplasmic processes, and in the regions we have looked at there is evidence that cells near the surface are attaching themselves to the basal lamina and hauling themselves towards it. This movement might then be transmitted to cells lying deeper within the mesoderm, producing the slight shifting movement required in the original computer model.
ACKNOWLEDGEMENTS
One of us (D. A. E.) wishes to acknowledge support in the course of this work by grants from the Agricultural Research Council, the Science Research Council and Unilever Ltd. We are grateful to Mr Raymond Moss for technical assistance, and to O. P. Flint and O. K. Wilby for critical discussion of the manuscript.