Descriptive studies of the talpid2 chick embryonic lethal mutant (ta3//a3) have suggested that the multiple effects produced by this gene are of mesodermal origin, and that they arise from defective mesenchymal cell movement and condensation (Ede & Kelly, 1964a, b). It may be argued that condensation in vivo is comparable to cell reaggregation of dissociated cells in vitro, and that defects in the former are likely to be reflected in the latter. In this case it should be possible to obtain experimental verification of this effect of the gene at the cellular level, using the quantitative methods for assessing aggregation developed by Moscona (1961a, b) and Curtis & Greaves (1965). The experiments reported here show a clear genetic effect upon cell adhesion in the wing-bud mesenchyme of the talpid3 mutant.

The wing-bud was chosen because it was hoped to establish a connexion between the effect of the gene at the cellular level and its dramatic effect on limb morphogenesis. A theoretical model of its development is proposed, which demonstrates how changes in cell adhesiveness and motility may account for the distorted shape of the mutant wing-bud, and also the role that these factors may play in the production of the normal limb-bud pattern.

Four- and five-day embryos (stages 24 and 26) were obtained from matings between known talpid3 heterozygotes, giving approximately one talpid3: three normal phenotypes. Mutant and normal embryos were easily distinguishable at these ages (Plate 1), and any talpids which were not clearly healthy (as judged by size, vigorous heart-beat and absence of capillary haemorrhages) were discarded.

The methods used for obtaining single cell suspensions of wing-bud meso-derm cells and for producing rotation-mediated reaggregation follow those of Moscona (1961 a) with slight variations. Wing-buds were dissected off in Tyrode’s solution, removed to 15 ml centrifuge tubes, rinsed in Ca2+- and Mg2+-free Tyrode’s solution (CMF) and incubated for 10 min at pH 8-0. Incubation for this and all other operations was at 37·5 °C. They were then rinsed and incubated for 10 min in trypsin solution (1% Difeo 1:250 in CMF at pH 8·0), then trans-ferred into CMF in cavity slides where the ectoderm was carefully removed and the mesoderm divided into small fragments by means of tungsten needles. The fragments were placed in centrifuge tubes with fresh trypsin solution and in-cubated for 20 min, then rinsed in CMF and the CMF finally replaced by 1 ml of culture medium. A cell suspension was obtained from the fragments by flushing them 15–20 times through the tip of a fine Pasteur pipette (0·75 mm tip diameter), transferred to a 25 ml conical flask and made up to 2 ml with culture medium. Samples of this stock suspension were taken for cell counting in a haemocyto-meter, for checking for complete dispersion of cells and for viability testing. Suspensions obtained in this way contained not more than 3–4 % ofclusters (two or more adhering cells) among cells counted, and viability, tested by eosin staining, was between 94 and 95 % for both talpid3 and normal cells.

(a) Aggregation over 3-day periods

The culture medium consisted of Eagle’s basal medium based on Earle’s BSS (Flow Laboratories) with 0·02 % glutamine, 10 % horse serum (Oxoid), 2 % embryo extract (50:50in Tyrode’s solution ; 9-day chicks embryos) and penicillin-streptomycin at a concentration of 100 units/ml. added. Media were sterilized by Millipore filtration.

The stock suspension was diluted to give 5 × 105 cells in 3 ml culture medium in 25 ml conical flasks. The flasks were gassed with 5 % CO2–95 % air mixture jto give pH 7·2, tightly stoppered and placed on a gyratory shaker inside a water-acketed incubator. The shaker (to be described in a separate publication) was constructed in the laboratory work-shop and was designed to give a very smooth swirling action. For these experiments it was run with a in. radius rotary motion, at a speed of 50 rev/min. In each experiment about six flasks each of taipid3 and normal cell suspensions were prepared from embryos of the same age, either 4- or 5-day ; eight experiments were done with 4-day and six with 5-day embryos.

At 7, 25, and 70 h after beginning rotation each flask was removed from the shaker, its stopper replaced by a glass cover, and the aggregates photographed; then it was gassed again, re-stoppered and put back on the shaker. Measure-ments and counts were made on the photographs.

After photographing at 70 h the aggregates were fixed in Bouin’s fluid, sectioned at 5 μ and stained variously with haematoxylin and eosin, iron haema-toxylin, Masson’s saffron and alcian blue with chlorantine fast red.

In order to estimate changes in cell number some flask contents, obtained from 5-day embryos, were disaggregated again after 50 h of reaggregation, and cell counts made with a haemocytometer.

(b) Aggregation over 5 h periods

Conditions for these experiments differed from the above in the following respects: horse serum and embryo extract were omitted from the culture medium; each flask contained 2 × 106 cells in 3 ml of medium; rotation was at 70 rev/min.

In each experiment one flask each of talpid3 and normal cell suspensions was prepared, gassed and stoppered and placed on the shaker; thereafter samples were taken for haemocytometer counts of single cells remaining non-aggregated, following the method of Curtis & Greaves (1965), at hourly intervals up to 5 h. Celt suspensions from 4- and 5-day embryos were used for four experiments in each case.

Normal and talpid3 wing-buds dissected from living embryos contained approximately equal numbers of cells at 4 days (Table 1), but at 5 days there were over times as many in talpid3. However, there was no indication of more rapid multiplication of talpid3 cells in culture when cells from 5-day embryos were reaggregated and then redissociated for counting after 50 h rotation at 50 rev/min. Only about one-third of the original cell number was recovered in each case, indicating that any increase by cell multiplication had been offset by cell loss during the processes of reaggregation and redissociation.

Table 1.

Average number of mesoderm cells in 4- and 5-day wing-buds, and in aggregates from 5 × 105 5-day cells after 50 h rotation culture. Total number of wing-buds of flasks in parentheses

Average number of mesoderm cells in 4- and 5-day wing-buds, and in aggregates from 5 × 105 5-day cells after 50 h rotation culture. Total number of wing-buds of flasks in parentheses
Average number of mesoderm cells in 4- and 5-day wing-buds, and in aggregates from 5 × 105 5-day cells after 50 h rotation culture. Total number of wing-buds of flasks in parentheses

During the 3-day period reaggregation proceeded in the way described by Moscona (1961a), the large number of very small aggregates appearing after a few hours reducing to smaller numbers of larger aggregates as rotation con-tinued. Direct observation suggested a difference between normal and talpid3 in the pattern of reaggregation as regards size and number of aggregates, and also in their form, stabilizing at about 50 h (Plate 2, figs. G-L). The data regarding size and number in cultures from 4-day embryos are summarized in the histo-gram (Text-fig. 1), which omits aggregates of 0·1 mm or less since these are difficult to count and measure accurately. It shows little difference between normal and talpid2 at 7 h, but thereafter normal aggregates are generally larger and less numerous than talpid2 aggregates, this tendency increasing up to 50 h, when the pattern becomes stabilized,

Text-fig. 1.

Number and size distribution of 4-day normal and talpid3 aggregates > 0·1 mm at 7, 25, 50 and 70 h. semilog.

Text-fig. 1.

Number and size distribution of 4-day normal and talpid3 aggregates > 0·1 mm at 7, 25, 50 and 70 h. semilog.

Changes in total aggregate number (not including aggregates of 0·1 mm and under) and aggregate size are shown in Table 2 and in Text-fig. 2. In order that the significance of the larger aggregates should not be obscured by measurements of the smaller ones (which would include at the extreme limit any group of two or more cells) size is expressed as an index , where , d is the maxi-mum diameter of an aggregate and summation is over aggregates. In this way, the aggregate to which a given cell belongs is effectively averaged over cells rather than over aggregates. Put in another way, the contribution which any aggregate makes to the average is weighted in proportion to its approximate volume (d3).

Table 2.

Average size (all aggregates, 100 log ƌ index) and average number (per flask, aggregates >01 mm) at 7, 25, 50 and 70 h, with analysis of variance (* = p < 0·05, ** = P < 0·01, n.s. = not significant)

Average size (all aggregates, 100 log ƌ index) and average number (per flask, aggregates >01 mm) at 7, 25, 50 and 70 h, with analysis of variance (* = p < 0·05, ** = P < 0·01, n.s. = not significant)
Average size (all aggregates, 100 log ƌ index) and average number (per flask, aggregates >01 mm) at 7, 25, 50 and 70 h, with analysis of variance (* = p < 0·05, ** = P < 0·01, n.s. = not significant)
Text-fig. 2.

Aggregates from 4-day normal (○, solid line) and talpid3 (•, dotted line) at 7, 25, 50 and 70 h. A, Average size (100 log index ƌ) of all aggregates. B, Average number of aggregates > 0·1 mm per flask.

Text-fig. 2.

Aggregates from 4-day normal (○, solid line) and talpid3 (•, dotted line) at 7, 25, 50 and 70 h. A, Average size (100 log index ƌ) of all aggregates. B, Average number of aggregates > 0·1 mm per flask.

For statistical analysis it was found to be more convenient to use log .

The graphs indicate that there is indeed only a small difference in aggregate size at 7 h, but that this difference increases up to 50 h. Aggregate number is greater in normal cultures at 7 h, but from 25 h it is smaller than in talpid3.

The same picture emerges from the experiments using cells from 5-day embryos. Table 2 suggests that there is also a difference between cultures from 4- and 5-day embryos, showing a consistently higher average number of aggre-gates in the latter, but an analysis of variance shows that this apparent difference is not significant at the 5 % level. On the other hand the differences between normal and talpid2 are significant in all cases, except for the difference in aggre-gate number at 25 h, where an inexplicably wide divergence from the general pattern in one out of the six experiments reduced the statistical significance of the result.

In both 4-day and 5-day experiments normal aggregates are generally not only larger but also irregular in shape, whereas talpid3aggregates are nearly spherical. This is very marked at 7 h (Plate 2, figs. E, F, G and J), when normal aggregates are extremely irregular, with cells adhering loosely at their surfaces, and tangling with neighbouring aggregates. Talpid3 aggregates at this stage are rounded, have few loose cells at their surfaces, and aggregates are clearly separated from their neighbours.

This difference in the form of normal and talpid3 aggregates is evident in histological sections made from aggregates fixed after 3 or 4 days or rotation (Plate 3, figs. M-P). Cells at the boundary of normal aggregates usually remain rounded and project irregularly to give a rough surface, whereas those at the boundary of talpid3 aggregates are usually flattened and closely packed together, forming a virtual epithelium around each aggregate and giving it a smooth surface.

The cells of the chondrogenic interior of the aggregates at the end of the culture period are stellate in normal aggregates, and are in contact with each other chiefly at the tips of their cytoplasmic projections, whereas in talpid aggregates these cells are more rounded, and adhere to each other along more extensive regions of the cell boundaries (Plate 3, figs. Q, R).

The results of the experiments in which cells were rotated at 70 rev/min over a 5 h period in order to observe the decline in number of single (non-aggregated) cells are given in Text-fig. 3, and in Table 3 together with the results of an analysis of variance of the differences in steepness of decline at the beginning, middle and end of this period. They show that talpid3 declines more steeply than normal during the first hour, after which there is no significant difference between them in this respect. They also show that 5-day cultures decline more steeply than 4-day cultures; the difference in this case is not significant during the first hour but becomes significant during the middle and end periods.

Table 3.

Number of single cells (% of original suspension of 2×10e cells) at 0–5 h, with analysis of variance of steepness of decline between 0–1, 2–3 and 4–5 h (* = P < 0·05, n.s. = not significant)

Number of single cells (% of original suspension of 2×10e cells) at 0–5 h, with analysis of variance of steepness of decline between 0–1, 2–3 and 4–5 h (* = P < 0·05, n.s. = not significant)
Number of single cells (% of original suspension of 2×10e cells) at 0–5 h, with analysis of variance of steepness of decline between 0–1, 2–3 and 4–5 h (* = P < 0·05, n.s. = not significant)
Text-fig. 3.

Decline in number of single cells over the first 5 hours of reaggregation. A.Cells from 4-day normal (○, solid line) and talpid3 (•, broken line) embryos. B.Cells from 5-day normal (▵, solid line) and talpid3 (▴, broken line) embryos.

Text-fig. 3.

Decline in number of single cells over the first 5 hours of reaggregation. A.Cells from 4-day normal (○, solid line) and talpid3 (•, broken line) embryos. B.Cells from 5-day normal (▵, solid line) and talpid3 (▴, broken line) embryos.

1. Aggregation patterns in normal and talpid3, and their relation to cell adhesion

From these experiments it is clear that a difference in aggregation pattern exists between normal and talpid3 wing-bud mesenchyme cells, visible at 7 h and stabilizing at about 50 h in rotation-mediated cultures; in general, normal aggregates are larger and irregular in shape while talpid3 aggregates are smaller and more nearly spherical. The number of cells in pooled aggregates after 50 h is approximately the same in each, so that the difference between them is a matter of cell distribution, uncomplicated by differences of cell number.

Moscona (1961a) showed that, under conditions similar to these, different cell types (e.g. neural retina, liver, mesonephros) produced aggregates of charac-teristic size and shape, and, on the basis of experiments in which speed of rotation and temperature were varied, concluded that there was a direct relation between size and cell cohesiveness: the stronger the cohesiveness, the larger the aggregates produced from a given initial cell suspension. If this rule is applicable to our case it would indicate that talpid3 cells are less adhesive to each other than normal cells, since they form smaller aggregates.

However, estimates of the decline in number of single cells, which gives a direct measure of cell adhesiveness during the first stage of aggregation, show a steeper fall in talpid3 than in normal, indicating that on the contrary talpid3 cells are the more adhesive. The histological appearance of the aggregates is consistent with this conclusion and at the same time suggests why it is that normal aggregates should, nevertheless, be the larger. The appearance in sections of the cells in the chondrogenic interior of aggregates at the end of the culture period strongly suggests that talpid3 chondroblasts are more closely adherent to each other than normal chondroblasts. More important for our present argument is the contrast between the rather rough surface, produced by loosely adhering cells, of normal aggregates and the smooth quasi-epithelial surface of talpid3 aggregates, which is evident as early as 7 h.

These observations may all be resolved if in this instance aggregate size is indirectly related to cell adhesiveness, not directly as Moscona’s rule requires —‘The size distribution of aggregates thus formed reflects an equilibrium between cell cohesiveness under the given conditions, and the shearing forces of the system: the lesser the capacity of cells to cohere (or the fewer the number of cohesive cells, or the stronger the shearing forces), the smaller the resulting aggregates.’ (Moscona, 1961 b). It is therefore important to see whether this generalization, which was proposed for its heuristic value, does not require modification.

The production of aggregates in this system is due chiefly, after the first hour or so, not to collisions of single cells but of smaller aggregates, which may unite to give larger aggregates, and if ‘aggregate’ were substituted throughout for ‘cell’ the rule would be indisputable. However, it is not permissible to equate aggregate cohesiveness with cell adhesiveness ; that is to say, the mutual adhesiveness of individual cells may not be reflected in the mutual adhesiveness of the aggregates they make up, since whether two colliding bodies stick together or not depends not only upon the adhesiveness of their surfaces but also upon how much of these surfaces come into mutual contact, i.e. upon their shape and deformability.

The characteristic shapes of aggregates are not accounted for by Moscona, except as depending on a process of sorting out within aggregates of cells of different degrees of adhesiveness. Thus, liver cells produce one or a few large irregular aggregates under standard conditions while limb-bud cells produce many small subspherical ones. According to Moscona (1965), ‘One assumes that in 24 h the aggregates acquire a surface layer of cells which are relatively non-adhesive at the side exposed to the medium. Hence a cell population com-prising many such cells may be expected to yield an aggregation pattern consisting of many small aggregates, and vice versa. Cells from dissociated whole limb-buds produce aggregates which, perhaps because they are rapidly covered by epidermal cells, become non-adhesive and remain quite small; liver cells, on the other hand, cohere into large masses’. However, in earlier work, Moscona (1961b) obtained similar results using only the central core (skeleto-myogenic mesoblast) of 4-day chick limb rudiments; in fact the character of the limb-bud aggregates does not depend upon the presence of epidermal cells, and there is no reason to believe that at this stage the central mesoblast cells are heterogeneous with respect to adhesiveness. In our experiments differences in aggregate shape are apparent at 7 h, when no sorting will have occurred, and we suppose them to be due not to any heterogeneity within the cell populations, but to differences in the way in which the cells become packed together.

A single cell in rotating medium will assume a spherical shape, which will be deformed if it collides with another cell and adheres to it. The degree of adhesion will be related to the degree of deformation, and will be strongest when the cells are deformed into hemispheres (i.e. adhere over a maximum area), producing a spherical aggregate (i.e. one with minimal surface) conjointly. Looser adhesions will occur when there is less deformation of each cell, producing a dumb-bell shaped aggregate, with a smaller area of adherence and a larger surface. Other cells colliding with these aggregates and adhering to them will do so with a maximum of contact and produce nearly spherical aggregates if they are very adhesive, and with less contact and produce irregular aggregates if they are less so.

After some time aggregation by collision of single cells with other single ceils and with small aggregates will be superseded by collision of aggregates with aggregates. Spherical aggregates will come into contact over an area which becomes relatively smaller as their bulk becomes greater, and though they will still tend to stick together more frequently because their cells are more adhesive there will be an increasing opposing tendency, eventually dominant, for them to bounce off each other. Irregular aggregates, on the other hand, will make contact at a number of points and will tend increasingly to tangle with each other, so that although their cells are less adhesive the chances of fusion will be greater. More adhesive cells will, in fact, tend to produce numerous small spherical aggregates, less adhesive cells a few large irregular ones.

How this may be related to our data, on the hypothesis that talpid3 cells are more adhesive than normal cells, may be explained in conjunction with the accompanying diagram (Text-fig. 4), in which aggregation is divided into four phases, as follows.

Phase 1. The initial single cell suspension, before aggregation has begun.

Phase 2. Single cells are colliding together, and since they adhere more fre-quently the number of single cells declines more steeply and the number of small aggregates is larger in talpid3 (Text-fig. 3 A, B, 0–1 h).

Phase 3. Single cells are colliding with small aggregates, which are also colliding with each other; the irregular normal aggregates are already fusing more often than the subspherical talpid3 aggregates, and consequently becoming larger and less numerous (Text-fig, 2 A, 7 h; Plate 2, figs. G, J). In the number graph (Text-fig. 2 B, 7 h) normal aggregates appear as more numerous, since aggregates of 0·1 mm and less are not included, and a greater proportion of talpid3 aggregates fall into this category.

Phase 4. The tendency for irregular aggregates to fuse more readily than spherical ones continues; all aggregates have reached countable size and they are clearly fewer and larger in normal than in talpid3 cultures (Text-fig. 2A, B, 25 and 50 h; Plate 2, figs. H, I, K, L).

Text-fig. 4.

Diagram illustrating proposed relationship between cell adhesiveness and aggregate size and number in normal and talpid3. The number of cells is 150 in each case and remains constant through all phases. The top left half of each frame repre-sents single cells and aggregates < 0·1 mm, the bottom right half aggregates > 0·1 mm.

Text-fig. 4.

Diagram illustrating proposed relationship between cell adhesiveness and aggregate size and number in normal and talpid3. The number of cells is 150 in each case and remains constant through all phases. The top left half of each frame repre-sents single cells and aggregates < 0·1 mm, the bottom right half aggregates > 0·1 mm.

If this interpretation is accepted, all data indicate that the mutual adhesive-ness of wing-bud mesenchyme cells is greater in talpid3 mutant than in normal embryos. Whether the talpid3 gene produces this effect by altering some property of the cell membrane or by affecting the production of some extracellular substance will be a matter of further investigation.

2. Adhesion and motility in talpid3 cells

The adhesiveness of cells may play an important part in determining their motility; sarcoma cells, fibroblasts and epithelial cells appear to lie on a scale of increasing adhesiveness and decreasing motility. If cells are very adhesive their motility might be much reduced, and the following evidence suggests that on this scale talpid3 cells are towards the epithelial end:

  1. The appearance of cells at the surface is more epithelioid in talpid3 than in normal aggregates (Plate 3, figs. N, P).

  2. The cells of the chondrogenic interior appear to be united by more exten-sive regions of cytoplasm in talpid3 than in normal aggregates (Plate 3, figs. Q, R).

  3. Hinchliffe & Ede (1968) found that whereas in normal development the chondroblasts of the shoulder girdle elements become orientated at right angles to the long axis of the rudiments, no such orientation occurs in talpid3. This absence of chondroblast orientation is also found in the cartilage rudiments of the limb in 6-day talpid3 embryos, when it is clearly established in normals(Plate 3, figs. S,T). The emergence of this order within a rudiment in which the chondro-blasts are at first arranged randomly must entail some movement of these cells upon one another, and inhibition of this movement because of their abnormally high adhesiveness would account for its absence in talpid3.

We conclude that talpid3 cells are not only more adhesive to each other than normal cells, but that as a consequence their motility is reduced.

3. Defective mesenchymal condensation andpolydactyly in talpid3 mutants

Defective segregation of mesenchyme cells in talpid3 was first described in the head (Ede & Kelly, 1964a), where the distortions of the face—e.g. eyes drawn together in the mid-line, absence of upper beak and fusion of the halves of the lower beak—were attributed to the prechordal mesoderm failing to separate to form the lateral maxillary rudiments, remaining instead as a mesial block of mesenchyme. Widespread failure of separate precartilage condensations to appear during skeletal development was also found (Ede & Kelly, 1964b), and in particular extensive fusion between the elements of the limb skeleton. These fusions are well shown in histochemical studies on the development of the latter, especially as regards mucopolysaccharide distribution (Hinchliffe & Ede, 1967).

We believe that the failure of these mesenchymal cells to segregate properly to form normal clearly separated condensations is due to their abnormal adhesive-ness and reduced motility.

Within the limb-bud, development of the pentadactyl precartilage pattern in the mesoderm entails two processes whose mutual relationships are still obscure ;

  1. Condensation, i.e. its division into a central core of closely packed and a peripheral region of loosely packed mesenchyme cells, followed by subdivision of the core into condensed regions representing the skeletal elements.

  2. Differentiation, proceeding in parallel with condensation, in which the cells of the core become transformed into chondroblasts, with characteristic inter-cellular matrix, and separated by sheaths of spindle-shaped perichondrial cells from the myogenic and other tissues developing in the peripheral mesenchyme.

The aspect of differentiation will be dealt with in a further study; for the present condensation only will be considered, and it will be supposed that it is produced by aggregative cell movements. Of the two other mechanisms suggested by Trinkaus (1965), i.e. (1) local increase in cell division and (2) contraction of the whole cellular mass, there is no evidence for the first in this instance, and the experiments of Hampe (1960) suggest that individual cell movement is involved rather than the second. The condensations appear at definite locations within the developing limb-bud, and their foci are presumably the centres of production of some morphogenetic substance (morphogen) leading to aggregation in their neigh-bourhood. The process appears to be similar to the appearance of aggregation centres in dispersed populations of slime mould amoebae (Bonner, 1963), though whether the morphogen of the limb bud produces its effect chemotactically as the hormone acrasin does in slime moulds, or by some sort of trapping by cell immobilization, is an open question.

Whatever the mechanism, the tendency to aggregate will be greatest at the focal point and diminish with increasing distance from it, so that in a two-dimensional diagram regions of decreasing aggregation potential may be repre-sented by concentric lines around it (Text-fig. 5 A). Neighbouring regions over-lap, and the degree of overlapping is related to their position on the proximo-distal axis of the limb bud; those at the distal end overlap least because, being at the border of the expanded portion of the bud, they are the most widely spaced. Where lines overlap, the regions they enclose become fused.

Text-fig. 5.

Diagram showing precartilagc condensation fields in normal and talpid wing-buds.

Text-fig. 5.

Diagram showing precartilagc condensation fields in normal and talpid wing-buds.

According to this scheme, cells move away from the periphery towards the centre, and normally condense into the black regions in the diagram, forming (omitting the carpals for the sake of simplicity) four bands of elements in the wing: (1) humerus, (2) radius and ulna, (3) four metacarpals and (4) four sets of phalanges. Talpid2 cells, because of their abnormal adhesiveness and reduced motility, manage to condense only into the outer rings, so that radius and ulna elements and metacarpal elements become fused, only the phalanges appearing as separate condensations. Expressing this in another way, we may say that as it grows out from the trunk the limb bud generates a sequence of individuation fields (Waddington, 1967) of chondrogenic potential, and that each of these fields (except the most proximal) normally becomes divided into subfields, but that in talpid3 this subdivision manifests itself only in the most distal field.

However, talpid3 is characterized not only by fusion of normally separate elements, but also by polydactyly in the formation of up to eight phalangeal elements in the wing, and this is clearly related to the distinctive fan shape of the limb-bud. In the normal embryo the number as well as the spacing of elements in each field is related to its position on the proximo-distal axis: 1 (humerus) proximally, where the limb is narrowest, 2 (radius and ulna) next distally, 4 (metacarpals) and 4 (phalanges) most distally. This in itself suggests that the number of condensations is related to the breadth of the field, and this would account for the increased number of phalanges in talpid3, where the distal margin is extremely extended. The fan-shaped distortion of the limb-bud will also in-crease the number of metacarpal foci, but since condensations extend to the outer ring these elements remain fused and indistinguishable. If the effect of the gene on cell adhesion were less strong, so that condensation proceeded up to the middle ring, the increased number of metacarpals would be revealed (Text-fig. 5B). This is precisely what is shown in photographs of the cartilage limb skeleton of the talpid2 chick embryo (Goetinck & Abbott, 1964), a lethal mutant which shows the same general effects as talpid3, but which survives longer and whose genetic effects on cell properties are therefore probably less intense.

The same sort of relationship was found by Coulombre, Coulombre & Mehta (1962) to exist between the diameter of the eye and the number of scleral ossicles produced in the conjunctival mesenchyme. Both are clear examples of the type of development characterized (Smith, 1960) by the production of an integral number of structures from a homogeneous field of an extent which can vary continuously from individual to individual, with a preferred spacing between structures, the actual number formed being a compromise between the preferred spacing and the requirement that the number be integral. Whether in this case the field is strictly homogeneous, requiring explanation of the origin of subfields by reference to a Turing-type system, or whether gradients and competitive interactions alone would be sufficient mechanisms, must remain open questions for the present.

We conclude that the fusion of cartilage elements in the talpid3 wing skeleton is accounted for by the increased adhesiveness and reduced motility of the mesenchyme cells, and that this also accounts, in conjunction with the distorted shape of the limb-bud, for the characteristic phalangeal polydactyly.

4. Cell motility in relation to limb-bud shape

It may be asked whether the characteristic distortion of the talpid3 limb bud is produced by an intrinsically abnormal rate of cell division. There are about 70 % more cells in the mutant than in normal wing buds at 5 days, but there is no difference in numbers of dissociated and reaggregated cells after 2 days in culture (Table 1); although more evidence is required on this point, we suspect that the cell multiplication rate is potentially equal, but that because most proliferation occurs just behind the apical ectodermal ridge (see below), and because this ridge is up to 65 % more extensive in the mutant, more cells are in fact dividing in it than in the normal. The increased cell number is a result of the abnormal geometry of the limb-bud rather than vice versa.

In its outward growth the limb-bud becomes flattened and the main features of its growth can be seen in lateral view, as in Text-fig. 6, where tracings from stage 22 to stage 29 have been superimposed. On the right of the diagram the pictures have been idealized by smoothing the outlines, eliminating the elbow joint in the normal and making the pre-axial and post-axial portions symmetrical in order to emphasize the essential difference between the two. It becomes clear that this lies in the fact that whereas the normal limb bud grows outwards without much increase in breadth for some time, forming a stem, and only later begins to fan out to form a distal paddle, in talpid3 there is no stem because the limb bud begins to increase in breadth immediately.

Text-fig. 6.

A. Superimposed outlines of wing-bud stages 22–29, normal and talpid3. B. Idealized diagrams of the same.

Text-fig. 6.

A. Superimposed outlines of wing-bud stages 22–29, normal and talpid3. B. Idealized diagrams of the same.

The mechanisms of limb growth have not been fully worked out even at the descriptive level. Saunders (1948) established that growth and differentiation of new prospective regions in the mesoderm occurs at the distal end of the bud, while the more proximal regions established previously continue to develop. Mitosis is most intense just behind the apical ectodermal ridge, but it is not certain whether there is a distinct zone of proliferation here as Saunders believed, and for which Searls (1965) provides some evidence, or a gradient of mitotic activity with its high point distally as postulated by Camosso, Jacobelli & Pappalettera (Amprino, 1965).

In order to account for elongation of the stem region of the normal limb bud by cell division alone it would be necessary for the mitotic spindles to be orien-tated along the proximo-distal axis, and this has not been observed. The simplest alternative is to suppose that there is a tendency to cell movement in a distal direction; if this were necessary for normal development, inhibition of move-ment in talpid3 cells would account for the absence of a stem region and the consequent distortion of the mutant limb-bud. In order to test this hypothesis, programmes are currently being designed for the generation of models simu-lating relevant aspects of limb-bud growth on a digital computer (Ede & Law, unpublished). Pending completion of this work, when the detailed design of the programmes will be published, their basic principles have been used to produce simpler models on graph paper, examples of which are shown in Text-figure 7.

Text-fig. 7.

Models of limb-bud growth, with outlines of 6-day normal (4a) and talpid3 (5a) wing-buds. Growth starts from a single row of‘cells’, representing prospective limb region of the flank, and proceeds by generation of a series of pictures, each produced by scan-ning the previous picture and transforming it by causing some ‘cells’ to reproduce and positioning new ‘cells’ according to simple rules. The diagrams represent the 35th picture in each case. Model 1.1 in 5 of‘cells’ in each row is reproduced, and each new cell placed in the nearest available space. Model 2. A ‘gradient’ of reproduction is obtained by limiting reproduction to ceils in the 10 outer rows, giving a first approximation to talpid3 wing-bud shape. Model′S. The ‘gradient’ is retained, but new cells are moved distalwards by a small amount (three places) before being placed in the nearest available space, giving a first approximation to normal wing-bud shape. Model 4. As model 3, but after an arbitrary number of pictures the amount of distal movement is reduced, giving a closer approximation to the normal shape. Model 5. As model 4, but with more restricted distal movement, giving a closer approximation to the talpid3 shape.

Text-fig. 7.

Models of limb-bud growth, with outlines of 6-day normal (4a) and talpid3 (5a) wing-buds. Growth starts from a single row of‘cells’, representing prospective limb region of the flank, and proceeds by generation of a series of pictures, each produced by scan-ning the previous picture and transforming it by causing some ‘cells’ to reproduce and positioning new ‘cells’ according to simple rules. The diagrams represent the 35th picture in each case. Model 1.1 in 5 of‘cells’ in each row is reproduced, and each new cell placed in the nearest available space. Model 2. A ‘gradient’ of reproduction is obtained by limiting reproduction to ceils in the 10 outer rows, giving a first approximation to talpid3 wing-bud shape. Model′S. The ‘gradient’ is retained, but new cells are moved distalwards by a small amount (three places) before being placed in the nearest available space, giving a first approximation to normal wing-bud shape. Model 4. As model 3, but after an arbitrary number of pictures the amount of distal movement is reduced, giving a closer approximation to the normal shape. Model 5. As model 4, but with more restricted distal movement, giving a closer approximation to the talpid3 shape.

The diagrams show a series of models representing increasing additions of simple rules controlling cell reproduction and the position of the cells when they are produced. In spite of the simplicity of the rules it is clear that distinct patterns of growth are generated, that some of these patterns approximate to patterns found in developing limb-buds in normal embryos, and that slight changes in the rules produce dramatic changes in shape which are comparable to those found in the mutant embryos. The models suggest that in addition to cell multiplication with a higher rate of proliferation distally, cell movement towards the distal end is also a necessary factor in normal growth of the limb bud. Where this movement is restricted there is a distortion of its shape such as is found in talpid3. Models 4 and 5 incorporate a change in motility, suggested by the experimental evidence that in both normal and talpid3 cells from 5-day embryos are more adhesive than cells from 4-day embryos. Their resemblance to the outlines of 5-day or 6-day normal and talpid3 wing-buds respectively is very striking, suggesting that changes in motility of the mesenchyme cells with time may play a part in determining the basic pattern of the limb-bud shape.

No such movement of limb-bud mesenchyme cells has been observed, but as demonstrated in the models it need be only of a very small order, amounting to slight displacements of the cells on each other in an orientated way, and it is difficult to see how such small movements could be demonstrated directly. However, Thornton (1960) showed that in regenerating limbs of Amblysloma the epidermal cap exercised an orientating influence on the aggregation and outgrowth of mesodermal cells to form the new limb, and pointed out, as has Faber (1965), that the apical cap in amphibia is essentially similar to the apical ectodermal ridge in higher vertebrates. It may be that one of the functions of the apical ectodermal ridge is to initiate and direct the outward cell movement postulated here. More generally, this recalls the demonstration by Clarkson & Wolpert (1967) that bud elongation in hydra is caused not so much by cell division, as previously thought, but rather by orientated movements of cells immigrating from surrounding tissues, directed by the tip of the bud.

5. Cell contact and cell death in talpid®

If the above is accepted, the talpid3 gene affects three processes, sequential in time but overlapping and to some extent superimposed:

  1. The drift of mesodermal cells in the direction of the apical ectodermal ridge to produce the basic shape of the limb bud, (2) the aggregation of mesen-chymal cells to form the precartilage condensations of the limb skeleton, and (3) the alignment of chondroblasts within the developing cartilages. Each of these, it has been argued, may result from a genetic effect on cell adhesion and motility.

In addition, Hinchliffe & Ede (1967) showed that certain regions of normal cell death, the anterior and posterior necrotic zones, were absent in talpid3, and suggested that this might play a part in producing the abnormal shape of the limb bud: these zones might act as ‘end-stops’ to the apical ectodermal ridge, which in their absence would extend beyond its normal limits and, by stimulating mesodermal proliferation beneath it, cause much more expansive growth at the distal margin. There is no evidence at present to determine whether the findings reported in this paper render this hypothesis superfluous by showing that factors acting within the mesoderm alone account for its fan-shaped development, or whether both mechanisms are acting and reinforcing each other, but on either view the absence of these zones of cell death remains. The inhibition of cell death in highly circumscribed regions where it normally occurs seems a much more mysterious and arbitrary genetic effect than a general increase of cell adhesion within the limb bud, and it cannot be coincidental that both occur in the same mutant. It seems most likely that the first is a particular result of the second, and that programmed embryonic cell death (Saunders, 1966), like other types of differentiation, is not a property of individual cells but depends upon forms of contact between them in cell groups.

  1. Differences between aggregation patterns in reaggregating wing-bud mesenchyme cells from normal and talpid3 embryos are interpreted as indicating that the mutant cells are abnormally adhesive.

  2. The role of cell adhesion and motility is discussed in relation to the formation of the normal pattern of skeletal condensations in the embryonic wing, and to its distortion in talpid mutants.

  3. A model of normal wing-bud growth is proposed in which slight move-ments of the mesodermal cells towards the apical ectodermal ridge play an important part, accounting for the characteristic shape of the talpid3 wing-bud by inhibition of these movements owing to the abnormal mutual adhesiveness of the cells.

Adhésion et mouvement cellulaire en relation avec le développement du membre chez les embryons de Poulet normaux et mutants ‘talpid3

  1. Des différences observées dans le mode d’agrégation des cellules du mésen-chyme du bourgeon alaire d’embryons normaux et ′talpid3 indiqueraient une adhésivité anormale des cellules mutantes.

  2. La discussion porte sur le rôle de l’adhésion et de la motihié cellulaire en rapport avec la formation d’un type normal de condensations squelettiques dans l’aile embryonnaire et l’altération de ce type chez les mutants ′talpid′.

  3. On propose un modèle de croissance pour le bourgeon d’aile normal où de faibles mouvements de cellules mésodermiques vers la crête apicale ectodermique joueraient un rôle important; il est valable pour le bourgeon ′talpid3 avec sa forme caractéristique, où ces mouvements seraient inhibés à cause de l’adhésivité mutuelle anormale des cellules.

We wish to thank Mr Bernard Dugdale for constructing the gyratory shaker, Mr. R. Morley Jones for statistical advice, Mr Hamish Law for discussions on mathematical aspects of model construction, and Miss Irene Thomson and Miss Morag Sylvester for technical assistance.

PLATE 1

Figs. A and B. Normal embryos, 4- and 5-day. Figs. C and D. talpicF embryos, 4- and 5-day.

PLATE 1

Figs. A and B. Normal embryos, 4- and 5-day. Figs. C and D. talpicF embryos, 4- and 5-day.

PLATE 2

Aggregates from dissociated wing-bud mesenchyme cells from 4- and 5-day embryos at different stages of rotation culture. Fig. E. 4-day normal aggregate, 7 h. Fig. F. 4-day talpid3 aggregate, 7 h. Figs. G-I. 5-day normal aggregates; the same culture at 7, 25 and 50 h. Figs. J-L. 5-day talpid3 aggregates; the same culture at 7, 25 and 50 h.

PLATE 2

Aggregates from dissociated wing-bud mesenchyme cells from 4- and 5-day embryos at different stages of rotation culture. Fig. E. 4-day normal aggregate, 7 h. Fig. F. 4-day talpid3 aggregate, 7 h. Figs. G-I. 5-day normal aggregates; the same culture at 7, 25 and 50 h. Figs. J-L. 5-day talpid3 aggregates; the same culture at 7, 25 and 50 h.

PLATE 3

Figs. M and N. Section of a 4-day normal aggregate, fixed at 70 h. Stained haematoxylin and eosin. Figs. O and P. Section of a 4-day talpid3 aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. Q. Section of the chondrogenic interior of a 5-day normal aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. R. Section of the chondrogenic interior of a 5-day talpid3 aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. S. Longitudinal section of the wing of a 6-day normal embryo showing chondroblast orientation in the ulna region. Stained haematoxylin and eosin. Fig. T. Longitudinal section of the wing of a 6-day talpid3 embryo showing failure of chondroblast orientation in the ulna region. Stained haematoxylin and eosin.

PLATE 3

Figs. M and N. Section of a 4-day normal aggregate, fixed at 70 h. Stained haematoxylin and eosin. Figs. O and P. Section of a 4-day talpid3 aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. Q. Section of the chondrogenic interior of a 5-day normal aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. R. Section of the chondrogenic interior of a 5-day talpid3 aggregate, fixed at 70 h. Stained Masson’s saffron. Fig. S. Longitudinal section of the wing of a 6-day normal embryo showing chondroblast orientation in the ulna region. Stained haematoxylin and eosin. Fig. T. Longitudinal section of the wing of a 6-day talpid3 embryo showing failure of chondroblast orientation in the ulna region. Stained haematoxylin and eosin.

Amprino
,
R.
(
1965
).
Aspects of limb morphogenesis in the chicken
. In
Organogenesis
, pp.
255
81
(eds.
Delhan
,
R. L.
&
Ursprung
,
H.
).
New York
:
Holt, Rinehart and Winston
.
Bonner
,
J. T.
(
1963
).
Epigenetic development in the cellular slime moulds
. In
Cell Differentiation
, pp.
341
58
(
17th Symp. Soc. exp. Biol.
, ed.
Fogg
,
G. E.
).
Cambridge University Press
.
Clarkson
,
S. G.
&
Wolpert
,
L.
(
1967
).
Bud morphogenesis in Hydra
.
Nature, Lond
.
214
,
780
83
.
Coulombre
,
A. J.
,
Coulombre
,
J. L.
&
Mehta
,
H.
(
1962
).
The skeleton of the eye. I. Conjunctival papillae and scleral ossicles
.
Develop. Biol
.
5
,
382
401
.
Curtis
,
A. S. G.
&
Greaves
,
M. F.
(
1965
).
The inhibition of cell aggregation by a pure serum protein
.
J. Embryol. exp. Morph
.
13
,
309
26
.
Ede
,
D. A.
&
Kelly
,
W. A.
(
1964a
).
Developmental abnormalities in the head region of the talpid3 mutant of the fowl
.
J. Embryol. exp. Morph
.
12
,
161
82
.
Ede
,
D. A.
&
Kelly
,
W. A.
(
1964b
).
Developmental abnormalities in the trunk and limbs of the talpid3 mutant of the fowl
.
J. Embryol. exp. Morph
.
12
,
339
56
.
Faber
,
J.
(
1965
).
Autonomous morphogenetic activities of the amphibian regeneration blastema
.
Proc. Regeneration in Animals
, pp.
404
19
.
Amsterdam
:
North Holland Publishing Company
.
Goetinck
,
P. F.
&
Abbott
,
U. K.
(
1964
).
Studies on limb morphogenesis. I. Experiments with the polydactylous mutant, talpid3
.
J. exp. Zool
.
155
,
161
70
.
HampÉ
,
A.
(
1960
).
La compétition entre les éléments osseux du zeugopode de poulet
.
J. Embryol. exp. Morph
.
8
,
241
5
.
Hinchliffe
,
J. R.
&
Ede
,
D. A.
(
1967
).
Limb development in the polydactylous talpid3 mutant of the fowl
.
J. Embryol. exp. Morph
.
17
,
385
404
.
Hinchliffe
,
J. R.
&
Ede
,
D. A.
(
1968
).
Abnormalities in bone and cartilage development in the talpid3 mutant of the fowl
.
J. Embryol. exp. Morph
.
19
,
327
39
.
Moscona
,
A. A.
(
1961a
).
Rotation-mediated histogenetic aggregation of dissociated cells
.
Expl Cell Res
.
22
,
455
75
.
Moscona
,
A. A.
(
1961b
).
Effect of temperature on adhesion to glass and histogenetic cohesion of dissociated cells
.
Nature, Lond
.
190
,
408
9
.
Moscona
,
A. A.
(
1965
).
Recombination of dissociated cells and the development of cell aggregates
. In
CellsandTissues in Culture
. Vol.
1
, pp.
489
599
(ed.
Willmer
,
E. N.
).
London
:
New York : Academic Press
.
Saunders
,
J. W.
, Jr
. (
1948
).
The proximo-distal sequence of origin of the parts of the chick wing and the role of the ectoderm
.
J. exp. Zool
.
108
,
363
403
.
Saunders
,
J. W.
, Jr
. (
1966
).
Death in embryonic systems
.
Science
154
,
604
12
.
Searls
,
R. L.
(
1965
).
An autoradiographic study of the uptake of S36-sulfate during the differentiation of limb bud cartilage
.
Develop. Biol
.
11
,
155
68
.
Smith
,
J. Maynard
(
1960
).
Continuous, quantized and modal variation
.
Proc. Roy. Soc. B
152
,
397
409
.
Thornton
,
C. A.
(
1960
).
Influence of an eccentric epidermal cap on limb regeneration in Amblystoma larvae
.
Develop. Biol
.
2
,
551
69
.
Trinkaus
,
J. P.
(
1965
).
Mechanisms of morphogenetic movements
. In
Organogenesis
, pp.
55
-
104
(eds.
Dehaan
,
R. L.
&
Ursprung
,
H.
).
New York
:
Holt, Rinehart and Winston
.
Waddington
,
C. H.
(
1967
).
Fields and gradients
. In
Major problems in Developmental Biology
, pp.
105
124
(ed.
Locke
,
M.
).
New York
:
Academic Press
.