Liquid-tissue behavior and differential cohesiveness during chick limb budding

Transcription

1 /. Embryol. exp. Morpli. Vol. 47, pp. -5, 978 Printed in Great Company of Biologists Limited 978 Liquid-tissue behavior and differential cohesiveness during chick limb budding By K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS From the Department of Biology, University of Virginia, and the Department of Anatomy, Wayne State University School of Medicine SUMMARY Emerging chick limb-buds at first grow only in length, not width. The growth parameters of limb mesoderm - cell shapes, distributions, division patterns and cleavage orientations - are incompatible with representations of this tissue as an elongating solid composed of proliferating but immobile cells. We observe that samples of both early limb mesoderm and also surrounding flank mesoderm round up like liquid droplets in organ culture. Therefore, liquid-like tissue rearrangments, including cell shuffling movements and neighbor exchanges, may occur in limb and flank mesoderm during in vivo limb budding. If so, differences in limb-flank surface tension properties would have to be present to keep these two fluid cell populations segregated into distinct tissues and properly positioned underneath limb and flank ectoderm. Previous studies have shown that tissue surface tensions are reflected in the spreading behavior of fused pairs of cell aggregates. To determine whether or not they possess differing surface tension properties, we pair excised pieces of early leg-bud, wing-bud or intervening flank mesoderm with pieces of 5f-day heart or liver in hanging drop cultures. For more rapid determinations of relative liquid-tissue cohesiveness than can be obtained in conventional, long-term experiments, aggregate pairs are fixed shortly after fusion. Since partial-envelopment configurations depend upon relative aggregate sizes as well as their tissue surface tensions, new procedures are used to deduce relative aggregate cohesiveness from crosssections of these briefly fused aggregate pairs. The envelopment tendencies of aggregates fixed 6-9 h after fusion are similar to those fixed 5-9 h after fusion: heart tends to surround leg; heart and wing surround each other with similar frequencies, but flank tends to surround heart. Also, liver tends to surround leg and wing, but flank tends to surround liver. When the effects of relative aggregate size are taken into account, these non-random, tissue-specific patterns of aggregate envelopment indicate that the relative cohesiveness of these tissues falls into the sequence: leg > heart ~ wing > liver > flank. The in vitro behavior of early limb-bud and neighboring flank mesoderm in these studies suggests that they are not simply mechanically identical portions of a single liquid tissue. We have previously proposed that early limb-bud mesoderm may act like a non-dispersing, cohesive liquid droplet which is embedded within a less cohesive fluid layer of flank tissue (and which is molded distally into paddle-shaped conformations by solid-like limb ectoderm and/or subjacent extracellular matrix). This proposal is not only compatible with the growth parameters of limb-bud mesoderm in vivo, but is also consistent with our observation that flank mesoderm surrounds tissues which surround limb mesoderm in these aggregate-fusion Authors' address: Department of Biology, University of Virginia, Charlottesville, VA 9, U.S.A. Author's address for reprints: Department of Anatomy, Wayne State University School of Medicine, 54 E. Canfield Ave., Detroit, MI 48, U.S.A.

2 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS experiments. Our model suggests that differences in the surface tension properties of limb vs. flank mesoderm may combine with differential cell proliferation, and possibly with active limb ectoderm expansion, to generate initial proximodistal limb outgrowth. INTRODUCTION Widely studied tissue interactions in the early chick limb-bud determine later limb development, but the initial formation of the limb-bud itself continues to pose mechanically puzzling morphogenetic problems. Searls & Janners (97) found that, during early limb outgrowth [Hamburger-Hamilton (95) stages 7-9], cell proliferation continues at a comparatively high rate in limb mesoderm but decreases markedly in surrounding flank tissue. Throughout this period, flank mesoderm expands laterally (anteroposteriorly and dorsoventrally) while maintaining a constant proximodistal thickness. By contrast, axial (proximodistal) outgrowth of limb mesoderm proceeds at a logarithmic rate, while laterally directed limb expansion is negligible at least up to stage (paddle-shaped buds). Searls & Janners concluded that some restriction on lateral limb spreading may be channeling its rapid growth into axial elongation. Lateral expansion of proliferating limb mesoderm into neighboring flank regions could be prevented in solid-\\kt tissue masses by constraints on cell movement (e.g. paralysis of locomotory organelles and/or unbreakable crosslinking between cells). Then cells could not slip past one another, so cell neighbor exchanges and tissue flowing movements would be prohibited. In that case, limb outgrowth might be due to appropriate orientation of cell division planes, to restriction of cell division to distal limb regions, to proximodistally directed cell elongation, and/or to proximodistal increases in the spacing between limb cells. However, in early bud mesoderm, at least up to stage, cell division planes are randomly oriented (Hornbruch & Wolpert, 97); cell division is rapid in proximal as well as distal limb mesoderm (Hornbruch & Wolpert, 97; Lewis, 975); axial cell elongation is not observed (Jurand, 965; Searls, Hilfer & Mirow, 97); and cell packing is relatively uniform along the proximodistal axis of the early bud (Summerbell & Wolpert, 97; Lewis, 975). In light of this negative evidence, it is by no means obvious how limb-buds could elongate without cell slippage. This dilemma has led some investigators (Hornbruch & Wolpert, 97, and see Discussion) to speculate that limb mesoderm may be a fluid tissue mass, its outgrowth somehow being shaped by surrounding tissues. We present here positive evidence that excised pieces of limb and flank mesoderm behave in organ culture like miniature liquid droplets. However, if both limb and flank mesoderm can flow like liquids in vivo, what could be suppressing lateral spreading (and thus enhance axial expansion) of the more rapidly growing limb mesoderm in this fluid-tissue system? If limb and flank mesoderm were merely adjacent, mechanically identical portions of a single liquid phase, nothing would oppose random lateral mi-

3 Chick limb budding 3 grations and/or intermixing of limb and flank tissue. Alternatively, tissuespecific differences in their surface tension properties might constrain early limb-bud mesoderm to act like a non-spreading, more cohesive liquid droplet within a less cohesive fluid layer of flank mesoderm (Phillips, Heintzelman, Daggy & Davis, \9a; Phillips, in preparation). In this study, tissue-fusion experiments provide evidence for differences in limb-flank mesoderm cohesiveness during limb-bud formation. MATERIALS AND METHODS Tissue dissection Leg-buds, wing-buds and intervening flank tissue were dissected from stages 7-8 (3 -day) White Leghorn chicken embryos in 4 C modified Hanks' Balanced Salt Solution, ph 7-4, with units/ml of penicillin and /*g/ml of streptomycin (' % pen-strep'). Heart ventricles and livers were dissected from stage 8 (5f-day) embryos. The excised limb-buds and flanks were incubated in % trypsin (Difco :5) in Hanks' modified salt solution at 4 C for 45-6 min to loosen the epidermis (Zwilling, 955). They were then transferred to cold Hanks' Balanced Salt Solution plus % horse serum and % pen-strep (HBSS + HS), where the epidermis was peeled off and discarded. Also in HBSS + HS, surface cell layers were cut away from the excised livers and heart ventricles and then, discarded, because pieces of tissue from subsurface regions of these organs, tend to round up more frequently and more completely than do pieces from surface regions. Next, all tissues were minced in HBSS + HS with fine needles into approximately cube-shaped pieces ranging from - to -4 mm on a side. These tissue fragments, each containing 3 5 cells, were then washed in cold Earle's-based Eagle's Minimal Essential Medium to which % horse serum and % pen-strep had been added (EMEM + HS). They were then cultured in fresh EMEM + HS on agar (which had previously been incubated overnight with EMEM + HS) in a 5% CO, water-saturated atmosphere at 37 C. Limb-bud and flank tissue fragments were cultured for 4 h, heart and liver fragments for 48 h, at the end of which time most fragments had rounded up into approximately spherical shapes. (Liver and especially heart fragments rounded up more slowly, and were therefore cut out and cultured a day earlier, than limb-bud and flank fragments.) Fusion experiments Rounded fragments of each tissue type were fused together in hanging-drop cultures (Phillips, Wiseman & Steinberg, 977c) in six different combinations: heart-leg, heart-wing, heart-flank, liver-leg, liver-wing and liver-flank. These combinations were used because limb-bud and flank can with ease be histologically distinguished by differential staining from heart and liver. In order to

4 4 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS evaluate the influence of relative aggregate size on the outcome of these fusions, we paired aggregates of differing as well as similar diameters together. Fused pairs of aggregates were incubated in hanging-drop cultures until most had formed dumbbell shapes (6-9 h), or for longer periods (5-9 h) to observe more extensive envelopment. The fused aggregates were then fixed in modified Zenker's fixative for -4 h, washed in distilled water and temporarily stored in 7% ethanol. Histology Aggregates were dehydrated further in a series of ethanols, embedded in Paraplast and serially sectioned into 5 /tm thick slices. By carefully orienting each pair of fused fragments during embedding, we were able to cut sections parallel to the axis of rotational symmetry, thus simplifying the examination of envelopment of one aggregate by the other. The sections were differentially stained with hematoxylin, eosin Y, and alcian blue 8GN. The eosin was taken up mostly by the heart and liver and the alcian blue by the limb and flank. [Occasional large, darkly stained heart cells (e.g. Figs. -4 below) are also seen in unfused heart aggregates, and so do not represent invading limb or flank cells in these fusions.] RESULTS The majority of tissue fragments rounded up and appeared healthy under the dissecting microscope and in serial sections. Larger tissue fragments were sometimes found to be necrotic and in such cases were discarded. Rounding-up of sample flank, leg and wing aggregates is shown in Fig.. Sectioned aggregate pairs were first classified into the following categories according to which tissue began to envelop the other tissue: Case I. A [leg(lg), wing(w) or flank (F)] surrounded B [heart (H) or liver (Lv)]. Case If. Neither tissue appeared to surround the other ('fusion without envelopment', Wiseman, Steinberg & Phillips, 97). Case HI. Heart or liver (B) surrounded limb or flank (A). Each of the above three cases were further subdivided according to relative size: either A was distinctly larger than B, or A and B were of approximately equal size, or B was distinctly larger than A. Estimates of direction and degree of envelopment, and of relative aggregate sizes, were made by eye. Six aggregate pairs were judged to be borderline cases; the remaining 3 pairs that clearly fit into the above categories are presented here. Tables and summarize and Figs. -7 illustrate the results obtained in these experiments. Heart usually surrounds leg (Fig. A), although exceptions occur (Fig. B, C). Heart and wing surround each other with similar frequencies (Fig. 3A, B). However, flank usually surrounds heart (Fig. 4). Liver usually surrounds leg (Fig. 5) and wing (Fig. 6), but flank usually surrounds liver (Fig. 7).

5 Chick limb budding.mm Fig. Fragments of flank (F), wing (W) and leg (Lg) mesoderm were photographed (A, C and E, respectively) immediately after excision from stage-7-8 chick embryos. The same tissue fragments were photographed again (B, D and F, respectively) after overnight culturing at 37 C. Table. Aggregates fused for 6-9 h Tissue A is leg, wing or flank; Tissue B is heart or liver. A Case I surrounded B Case II Fusioni, no envelopment Case III B surrounded A Relative size... A > B A = B A < B A > B A = B, A < B A > B A = B A < B Tissue pair A Lg W F Lg W F B H H H Lv Lv Lv I 5 8 6

6 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS Table. Aggregates fused for 5-9 h A and B as in Table. A Casel surrounded B Fusion, Case no envelopment B Case III surrounded A Relative size... Tissue pair A Lg W F Lg W F B H H H Lv Lv Lv A > B A = B A < 4 3 B A > B A = B A < B A > B A = B A 7 4 < B 4 3 C H>Lg 3A HcW 3B H)W H)F LvcLg 6 LvcW 7 Lv)F Figs. -7. Fusions of stage-8 heart (H) or liver (Lv) aggregates with stage-7-8 leg-bud (Lg), wing-bud (W) or flank (F) mesoderm aggregates fixed after 5 9 h (Fig. A-C) or 6-9 h (Figs. 3-7) in hanging-drop cultures. Symbols between letters indicate direction of envelopment.

7 B surrounds A Chick limb budding Fusion, no envelopment A surrounds B A smaller than B y BO A equal to/* ybo A larger than B Fig. 8. Diagrammatic summary of partial-spreading configurations for liquid-like aggregates A and B, with differing size ratios, fusing in medium O. Beneath each configuration is the relationship between aggregate-medium surface tensions that would yield that configuration. (Each question mark indicates that an unambiguous conclusion about relative aggregate-medium surface tensions cannot be made from that configuration without quantitative assessments in each case of the particular radii of curvature of the A-O, B- and A-B interfaces.) When the interfaces of two fusing liquid droplets, A and B, in medium O - with positive surface tensions JAO, JBO and JAB and radii RAO, RBO and R A B - are approximately spherical (Torza & Mason, 969), then 7AII _ JAO JBO. RAB RAO RBO where RAB is positive when A protrudes into B, negative when B protrudes into A, and infinite when the A-B interface isflat (Phillips, 969, and in preparation). Therefore, when B surrounds A, when A surrounds B, and when fusion without JAO JBO RAO RBO JAO,7co RAO* * RBO' envelopment is observed, 7/io JBO RAO JAO and JBO are inverse measures of liquid droplet deformability and thus direct physical measures of macroscopic droplet cohesiveness (Phillips, in preparation, and see Discussion). RBO (3)

8 K. F. HENTZELMAN, H. M. PHILLIPS AND G. S. DAVIS (A) Heart<leg (B) Heart wing (C) Heart > flank (9,4) (5/4) (9/9) (8/9) (6/7) (/9), VM H>W H = W H<W (/7)^ (/7) IZp H>F H = F H<F (D) Liver< leg (E) Liver < wing (4/) (F) Liver > flank (9 ) (/7) (7/) (4/7) Lv>Lg Lv = Lg Lv<Lg JO/) Lv>W Lv = W Lv<W ( ) 7 ( /) 'A \A ^r~* Lv > F Lv = F Lv < F Fig. 9. Histograms of the number of cases of relative cohesiveness of one tissue fragment compared to its partner in fusions of 6-9 h (hatched) or 6-9 h (solid) or both (fractions above brackets) calculated from Tables and as described in text. In any tissue combination, a larger aggregate may start to envelop (Fig. A) or to be enveloped by (Fig. C) a smaller aggregate; or aggregates of different sizes may begin to fuse without mutual envelopment (Fig. B). However, since aggregate-fusion times were intentionally minimized in these experiments (see Discussion), partial-envelopment configurations were always obtained. When envelopment is only partial, ordinary liquid-droplet configurations are determined not only by relative droplet cohesiveness (see Discussion) but also by relative droplet size (Fig. 8). For example, when two equally cohesive, immiscible droplets coalesce, the larger will tend to envelop the smaller. Thus, if fusing aggregates are acting here like coalescing immiscible liquid droplets (see Discussion), then Case I behavior with A larger than B (column in Tables and ) and also Case behavior with B larger than A (column 9) may be due

9 Chick limb budding 9 to differences in relative aggregate size and/or cohesiveness. Therefore, when expressing our results as evaluations of relative aggregate cohesiveness (Fig. 9), we have omitted these ambiguous cases. In addition (Fig. 8), the configurations represented by columns, 3 and 6 in Tables and indicate that B is more cohesive than A; the configurations represented by columns 4, 7 and 8 indicate that A is more cohesive than B; and the column 5 configuration indicates that A and B are equally cohesive. Hence, for comparisons of aggregate cohesiveness, our observations have been recategorized accordingly in Fig. 9. With the effects of relative aggregate size thus taken into account, the observed patterns of aggregate envelopment in our 6-9 h fusions are virtually identical with those in our 5-9 h fusions (see hatched vs. solid bars in Fig. 9); and these data have been pooled to obtain the fractions in Fig. 9. This analysis demonstrates that the non-random tissue-specific patterns of aggregate envelopment observed here cannot be attributed merely to variations in relative aggregate size, but instead reflect differences in aggregate cohesiveness (see Discussion). According to Fig. 9, on the average, the relative cohesiveness of these tissue samples falls into the sequence leg > heart ~ wing > liver > flank. DISCUSSION Tissue liquidity By definition (Symon, 97) subunitsin a 'liquid' must cohere to one another (unlike subunits in a gas), yet must be able to slide past one another to relax internal shear forces (unlike subunits in a solid). Thus, when 'liquid' tissues 'flow', cohering cells in these tissues must be capable of slipping past one another. Despite the conventional view of coherent tissues as solid-like masses of interlocking, immobile cells, several indirect lines of evidence suggest that limb-bud mesoderm may be capable of in vivo liquid-tissue flow. Amprino & Ambrosi (973) demonstrated that, when cavities are created in intact developing limb-buds, adjacent mesenchyme cells migrate into those wounds, not individually as separate cells, but rather en masse as 'coherent cell groups continuous with the surrounding mesenchyme'. Ede, Bellairs & Bancroft (974) noted that limb mesenchyme cells, unlike contact inhibited cells in tissue culture, possess ruffled membranes in vivo, and were persuaded that 'cell movement is definite; the question is whether it is passive or active'. Since major tissue migrations evidently do not occur during normal limb morphogenesis (Searls, 967; Stark & Searls, 973), early limb-bud mesoderm may be a relatively quiescent liquid tissue in which passive and/or active cell slippage movements function, not to accommodate extensive tissue streaming, but merely to relax shear stresses by permitting local cell shuffling. Four-day limb-bud mesoderm is in fact one of the embryonic cell populations which in organ culture exhibit a whole syndrome of liquid-like behavior - rounding-up, cell sorting and aggregate envelopment (Steinberg, 963, 964,

10 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS 97). Tn this study, we have confirmed in vitro aggregate rounding-up and envelopment behavior for limb mesoderm excised at the earliest stages of bud formation. Direct physical tests for in vivo tissue liquidity in intact limbbuds, comparable with the in vitro tests for cell slippage in embryonic chick liver aggregates (Phillips, Steinberg & Lipton, 9776), are in progress (Phillips & Daggy, in preparation). Even if limb-bud mesoderm per se is liquid-like, neighboring flank mesoderm might nevertheless be composed of cross-linked, immobilized cells, and thus might provide solid-tissue constraints on lateral limb spreading. However, we find that flank mesoderm between wing- and leg-buds also displays liquid-like rounding-up and envelopment behavior in vitro. Therefore, less obvious, liquid-tissue constraints on limb-flank spreading may instead be operating during in vivo limb-bud formation. Tissue surface tension in liquid limb budding We have proposed (Phillips et al. 977a; Phillips, in preparation) that limbbud mesoderm may behave like a more cohesive liquid droplet embedded within a less cohesive fluid layer of flank mesoderm. Laterally, surface tension differences could serve to keep these two contiguous fluid populations segregated into separate tissues. Distally, surface tension differences could keep limb mesoderm from spreading between flank mesoderm and ectoderm and also prevent flank mesoderm from spreading between limb mesoderm and ectoderm. (Too little is yet known about proximal limb-flank boundaries and their precise relationship to the underlying coelom to attempt here to predict what role, if any, limb-flank surface tension differences might play in the control of basal limbbud mesoderm morphogenesis.) Limb-bud outgrowth could be propelled by active expansion of limb ectoderm (Amprino, 965; Amprino & Ambrosi, 973) and/or by underlying limb mesoderm growth pressure. Tn any case, cell slippage would permit axially directed outgrowth of limb-bud mesoderm, despite uniform, non-oriented cell shapes, spacing and division patterns. This liquid-tissue model presupposes that, sometime during the differentiation of limb and flank mesoderm before limb budding, these two cell populations acquire different surface tension properties affecting tissue spreading and immiscibility. However, until now there has been no experimental evidence to eliminate the simpler alternative possibility that limb and flank mesoderm have identical mechanical properties and thus form a single liquid phase. The aggregate-fusion experiments described here were designed to test whether or not any differences in the surface tension properties of limb and flank mesoderm are in fact present at the time that limb-buds first begin to bulge out from the sides of the chick embryo. Whether it actively propels or merely passively permits the axial outgrowth of limb mesoderm, some solid-like portion of the limb-bud ectoderm (and/or subjacent extracellular matrix) would in any case have to bs responsible for shaping the liquid-like bud mesoderm into its paddle-like (rather than spherical-droplet-like) conformation.

11 Chick limb budding Aggregate-fusion assays for tissue cohesiveness Surface tension is the force in ordinary liquid droplets which promotes droplet rounding-up and opposes droplet deformation (Adamson, 96). When pairs of similarly-sized liquid droplets coalesce, the surrounding droplets tend to have lower surface tensions with respect to the medium than do the surrounded droplets (Phillips, 969, and see Fig. 8). Similarly, for a variety of embryonic chick tissues exhibiting liquid-like behavior, less cohesive (more deformable) aggregates tend to envelop more cohesive (less deformable) aggregates (Phillips & Steinberg, 969; Phillips et al. 977c). Final tissue positioning in cell-sorting and aggregate-fusion experiments can therefore be used as a qualitative indicator of relative liquid-tissue cohesiveness. Until now, however, aggregate configurations have been observed only several days after fusion; and, during prolonged culture, tissue cohesiveness may change (Phillips et al. 977c) due to normal or abnormal processes of in vitro tissue differentiation. In this study, we fixed aggregate pairs shortly after fusion in order to minimize time in culture after aggregate rounding-up. Our longer, 5- to 9-hour fusions demonstrate that relative aggregate cohesiveness assessed after the aggregate pairs had proceeded further toward their final envelopment configurations is similar to that deduced here from briefer (6-9 h) fusions. One potential problem with partial-spreading configurations is that, unlike complete-envelopment configurations (Phillips, 969, and in preparation), they are dependent upon relative aggregate size as well as upon relative aggregate-medium surface tensions (Fig. 8). Therefore, whenever possible, it is Internal shear stresses in fluids dissipate spontaneously, so surface tension is the sole force in a liquid substance that offers permanent resistance to increases in its surface area (Adamson, 96; Symon, 97). In the context of this study, 'aggregate cohesiveness' - i.e. the net macroscopic morphogenetic force opposing permanent deformations of cell aggregates in culture medium-is synonomous with 'aggiegate-medium surface tension' (Phillips, 969, and in preparation). The simplest and in our opinion most likely factors influencing macroscopic tissue surface tensions are tissue-specific differences in cell-cell adhesiveness (possibly modified by extracellular substances). Constructive criticism of the more complex and in our opinion less probable (Phillips, in preparation) alternative speculations by Harris (976) are beyond the scope of this publication (but see Steinberg, 978). Flow rates of liquids vary inversely with viscosity as well as directly with surface tension. However, final droplet configurations are independent of viscosity, so the effects of viscosity should decrease as droplet spreading progresses toward mechanical equilibrium. No such progressively decreasing effects of viscosity are detectable here. Evidently, any viscosity effects are already trivial after 6-9 h of fusion, since coalescence patterns at this time are similar to those after 5-9 h of fusion (Fig. 9). In ordinary liquid systems, tenfold differences in droplet viscosity have negligible effect upon pre-equilibrium droplet configurations (Torza & Mason, 969). In the present study, limb and flank aggregates rounded up at similar rates, so if limb has the higher aggregate-medium surface tension, it must be less viscous than flank. However, heart and liver displayed the most dramatic difference in rates of aggregate rounding up. Heart, with the apparently higher aggregate-medium surface tension, took approximately twice as long to round up as liver, and thus should be more than twice as viscous. Despite these various possible differences in relative tissue viscosity, limb and flank displayed the same differences in spreading tendencies against heart as they did against liver.

12 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS most efficient to fuse aggregates of equal size, since in that case relative aggregate cohesiveness in early aggregate fusions can be read directly from the direction of aggregate envelopment. However, it is sometimes more convenient, or even necessary, to fuse aggregates of different sizes. This study illustrates (Fig. 8) how relative aggregate cohesiveness can be deduced from partial-envelopment configurations involving differently sized aggregates in early fusion stages. Relative cohesiveness of limb vs. flank mesoderm For technical simplicity (see Materials and Methods), flank and limb mesoderm were fused with tissues from which they can easily be distinguished by standard differential staining techniques. Tested against both heart and liver aggregates, flank and limb aggregates displayed different tendencies consistent with the hypothesis that flank mesoderm is less cohesive than both wing and leg mesoderm in early limb-bud stages. (Since leg and wing tissues do not normally contact one another during limb budding, it is not clear what morphogenetic significance, if any, should be ascribed to the difference in cohesiveness of leg vs. wing mesoderm observed here.) Previous aggregate-fusion experiments with six different embryonic chick cell populations (including 4-day limb-bud mesoderm and 5-day heart and liver) have shown that, for all 5 tissue-pairs tested, if tissue A surrounds B, and if B surrounds C, then A will surround C (Steinberg, 97). Given our deduction from heart-wing and liver-wing fusions that heart should be more cohesive than liver, it is interesting to note that, in direct fusions of undissociated tissue fragments, 5-day liver does surround heart (Steinberg, 963, 964, 97). Our results with heart-wing fusions differ from those of Steinberg, since he found that heart always surrounds wing. This could be due to his use of wing mesoderm from older embryos and/or heart ventricle from younger embryos than ours. The latter is less likely, however, since Gershman (97) and Wiseman et al. (97) found that 5-day and 7-day heart aggregates fuse without envelopment, indicating that they are of equal cohesiveness. Also, the cohesiveness of our heart fragments may have increased during the days of culturing prior to fusion (e.g. see Phillips et al. 977 c) compared with Steinberg's heart fragments, which were fused immediately after excision. The interesting but exceptional and equivocal results of limb-flank cell-sorting experiments by Crosby (967) need repeating under our culture conditions before they can be compared with our observations. The tissue-specific differences in cohesiveness in the combinations tested here are apparently not absolute, since exceptions to the general trend of envelopment were often observed. Evidently, there is sufficient variability in the aggregate-medium surface tensions within any one group of aggregates so that, while one tissue may on the average be more cohesive than another, a less cohesive aggregate of the former may envelop a more cohesive aggregate of the latter. In principle, variability in cohesiveness might reflect local in vivo

13 Chick limb budding 3 differences in cohesiveness within each tissue. However, both aggregate-fusion and -centrifugation experiments (Wiseman et ah 97; Phillips et ah 977 c) have revealed similar variations in tissue cohesiveness even in reaggregates formed from dissociated and randomly intermixed heart cells. Clearly in that case such variations could not merely be attributed to initial differences in the relative cohesiveness of excised fragments prior to dissociation. Therefore, the apparent intra-tissue differences in cohesiveness observed here may simply reflect minor diverging responses of tissue fragments to excision and in vitro culturing. In any case, whether or not flank cohesiveness overlaps with limb cohesiveness cannot be deduced from fusions with heart or liver, but instead will have to be determined from direct limb-flank fusions (or from other, more direct methods of assessing their aggregate-medium surface tensions). We conclude that both limb and flank mesoderm aggregates exhibit liquidlike rounding-up and envelopment behavior in vitro, but that these two populations of somatic lateral-plate mesoderm cells are evidently not mechanically identical, adjacent portions of a single liquid-tissue phase. Instead, their contrasting spreading tendencies, tested against both heart and liver aggregates, indicate that their surface tension properties differ, with limb being more cohesive than flank. Thus, our results suggest a model for in vivo limb-bud mesoderm as a more cohesive droplet embedded within a less cohesive fluid layer of flank mesoderm. Unlike solid-tissue models, this model is compatible with the exclusively axial expansion of early limb-bud mesoderm. Moreover, it identifies specific new mechanical parameters that could govern (i) the segregation of neighboring populations of limb and flank mesoderm into discrete tissues, (ii) their proper positioning under limb and flank ectoderm, and (iii) their different directions of spreading during limb-bud formation. Finally, since tissue surface tensions may be generated by tissue-specific differences in intercellular adhesiveness (Steinberg, 963, 964, 97, 975, 978; Phillips, 969, and in preparation; Phillips et ah 977c), this model suggests new control mechanisms by which cell surface (and extracellular matrix) interactions may regulate early limb-bud morphogenesis. Assessments of limb-mesoderm-flank-mesoderm, limb-mesoderm-limb-ectoderm and flank-mesoderm-flank-ectoderm surface tensions are in progress We are grateful to Mr Martin Eglitis and Ms Carolyn Anderson for their technical assistance in this work. This research was supported at the University of Virginia Biology Department by N.S.F. grant GB-44 to H.M.P. G.S.D. was supported by N.I.H. traineeship HD 43. Portions of this paper are from K.F.H.'s thesis, submitted to the faculty of the University of Virginia in partial fulfillment of the requirements for the M.S. degree. REFERENCES ADAMSON, A. W. (96). Capillarity. In Physical Chemistry of Surfaces, pp. -5. New York: Interscience Publishers. AMPRINO, R. (965). Aspects of limb morphogenesis in the chicken. In Organogenetis (ed. R. L. DeHann & H. Ursprung), pp New York: Holt, Rinehart and Winston. EMB 47

14 4 K. F. HEINTZELMAN, H. M. PHILLIPS AND G. S. DAVIS AMPRINO, R. & AMBROSI, G. (973). Experimental analysis of the chick embryo limb bud growth. Archs Biol. 84, CROSBY, G. M. (967). Developmental capabilities of the lateral somatic mesoderm of early chick embryos. Ph.D. dissertation, Brandeis University, Waltham, Massachusetts. EDE, D. A., BELLAIRS, R. & BANCROFT, M. (974). A scanning electron microscope study of the early limb-bud in normal and talpid 3 mutant chick embryos. /. Embryo!, exp. Morph. 3, GERSHMAN, H. (97). On the measurement of cell adhesiveness. /. exp. Zool. 74, HAMBURGER, V. & HAMILTON, H. L. (95). A series of normal stages in the development of the chick embryo. /. Morph. 88, HARRIS, A. K. (976). Is cell sorting caused by differences in the work of intercellular adhesion?: A critique of the Steinberg hypothesis. J. theor. Biol. 6, HGRNBRUCH, A. & WOLPERT, L. (97). Cell division in the early growth and morphogenesis of the chick limb. Nature, Lond. 6, JURAND, A. (965). Ultrastructural aspects of early development of the forelimb buds in the chick and the mouse. Proc. R. Soc. B 6, LEWIS, J. H. (975). Fate maps and the pattern of cell division: a calculation for the chick wing bud. /. Embryol. exp. Morph. 33, PHILLIPS, H. M. (969). Equilibrium measurements of embryonic cell adhesiveness; Physical formulation and testing of the differential adhesion hypothesis. Ph.D. dissertation, The Johns Hopkins University, Baltimore. PHILLIPS, H. M. & STEINBERG, M. S. (969). Equilibrium measurements of embryonic chick cell adhesiveness. I. Shape equilibrium in centrifugal fields. Proc. natn. Acad. Sci., U.S.A. 64, -7. PHILLIPS, H. M., HEINTZELMAN, K. F., DAGGY, B. P. & DAVIS, G. S. (977a). Liquid-tissue mechanics in chick limb bud formation. Proc. 9th Ann. Southeastern Develop. Biol. Conf. PHILLIPS, H. M., STEINBERG, M. S. & LIPTON, B. H. (9776). Embryonic tissues as elasticoviscous liquids.. Direct evidence for cell slippage in centrifuged aggregates. Devi Biol. 59, PHILLIPS, H. M., WISEMAN, L. L. & STEINBERG, M. S. (977 C). Self VS. non-self in tissue assembly: correlated changes in recognition behavior and tissue cohesiveness. Devi Biol. 57, SEARLS, R. L. (967). The role of cell migration in the development of the embryonic chick limb bud. /. exp. Zool. 66, SEARLS, R. L. & JANNERS, M. Y. (97). The initiation of limb bud outgrowth in the embryonic chick. Devi Biol. 4, SEARLS, R. L., HILFER, S. R. & MIROW, S. M. (97). An ultrastructural study of early chondrogenesis in the chick wing bud. Devi Biol. 8, STARK, R. J. & SEARLS, R. L. (973). A description of chick wing bud development and a model of limb morphogenesis. Devi Biol. 33, STEINBERG, M. S. (963). Reconstruction of tissues by dissociated cells. Science, N.Y. 4, STEINBERG, M. S. (964). The problem of adhesive selectivity in cellular interaction. In Cellular Membranes in Development (ed. M. Locke), pp New York: Academic Press. STEINBERG, M. S. (97). Does differential adhesion govern the self-assembly processes in histogenesis? Equilibrium configurations and the emergence of a hierarchy among populations of embryonic cells. /. exp. Zool. 73, STEINBERG, M. S. (975). Adhesion-guided multicellular assembly: a commentary upon the postulates, real and imagined, of the differential adhesion hypothesis, with special attention to computer simulation of cell sorting. /. theor. Biol. 55, STEINBERG, M. S. (978). Specific cell ligands and the differential adhesion hypothesis: how do they fit together? In Specificity of Embryo logical Interactions (ed. D. Garrod). London: Chapman and Hall. (In the Press.) SUMMERBELL, D. & WOLPERT, L. (97). Cell density and cell division in the early morphogenesis of the chick wing. Nature, New Biol. 39, 4-6.

15 Chick limb budding 5 SYMON, K. R. (97). Mechanics, pp. 47, 3. Reading, Massachusetts: Addison-Wesley Publ. Co. TORZA, S. & MASON, S. G. (969). Coalescence of two immiscible liquid droplets. Science, N.Y. 63, WISEMAN, L. L., STEINBERG, M. S. & PHILLIPS, H. M. (97). Experimental modulation of intercellular cohesiveness: Reversal of tissue assembly patterns. Devi Biol. 8, ZWILLING, E. (955). Ectoderm-mesoderm relationship in the development of the chick embryo limb bud. /. exp. Zool. 8, (Received 5 September 977, revised 6 April 978)

16