Initial limb budding is independent of apical ectodermal ridge activity; evidence from a limbless mutant

Limbs develop through a series of epithelial-mesenchymal interactions. The mesoderm induces the apical ectodermal ridge (Kieny, 1960, 1968; Saunders & Reuss, 1974) which appears shortly after bud formation, at stage 18 (Todt & Fallon, 1984, 1986). It has been postulated that contact between the ridge and limb bud mesoderm is necessary to maintain ridge morphology (Zwilling, 1955) and that an early interaction with normal ectoderm is necessary for subsequent maintenance ability of limb bud mesoderm (Carrington & Fallon, 1984a). In addition, the presence of the ridge is necessary for outgrowth of mesoderm and development of distal limb structures (Saunders, 1948; Summerbell, 1974; Rowe & Fallon, 1982).

Limbless is a simple Mendelian autosomal recessive mutation in the chick, first described by Prahlad et al. (1979). Limb buds of chickens homozygous for this gene never develop an apical ectodermal ridge during limb bud formation (Prahlad et al. 1979; Fallon et al. 1983). Limb buds do form, but at stage 19 cell death begins in the mesoderm and the limb buds are subsequently eliminated (Fallon et al. 1983). Stage 19 is the earliest stage at which the limbless phenotype can be detected. The absence of ridge development and the degeneration of the forming limb bud result in total amelia in homozygotes. Also, in these homozygotes the coracoid and scapula do not separate and the sternum is poorly formed (Fallon et al. unpub-fished data). In the chick, the coracoid and scapula normally develop an articulation with each other and with the humerus. Care should be taken not to confuse limbless with wingless, another mutant we have studied. The limbless mutant is quite different from wingless, in which a ridge and partial legs can form (Zwilling, 1956; Carrington & Fallon, 1984a) and wing-derived musculature can develop (Lanser & Fallon, 1987).

The failure of limbless limb buds to grow out may be due to a change in one of the normal ectodermmesoderm interactions. In an initial test, recombinants were made in this laboratory with stages 19 to 22 wild-type ectoderm (which included a ridge) and limbless mesoderm, or with limbless ectoderm (which has no ridge) and wild-type mesoderm. These recombinants showed that the limbless mesoderm is capable of maintaining a normal ridge and forming a normal wing. However, the limbless ectoderm has no ridge or ridge function in that the mutant limb bud ectoderm is incapable of supporting outgrowth of wild-type mesoderm (Fallon et al. 1983). It remains possible that, although limbless mesoderm has a normal maintenance function, it cannot induce a ridge. Without a ridge the ectoderm would not be able to support outgrowth of normal mesoderm.

To determine whether or not limbless mesoderm can induce a ridge, a recombinant technique for prelimb-bud-stage embryos (Carrington & Fallon, 19846) was used. We grafted quail flank ectoderm to presumptive right wing mesoderm of stage-15 possible limbless chick embryos. In these recombinants the limbless mutant wing bud mesoderm was capable of inducing an apical ectodermal ridge in quail flank ectoderm and forming a normal limb. In addition, the lack of separation of the coracoid and scapula seen in unoperated limbless homozygotes was corrected on the right side when the normal ectoderm was applied at stage 15 and a right wing developed. When stage-19 limbless flank ectoderm was placed on the presumptive wing bud mesoderm of normal hosts, the limbless flank ectoderm did not respond to induction of a ridge. Therefore, the limbless mutation affects the ectoderm while limbless wing bud mesoderm, in association with normal ectoderm, has all the properties of normal wing bud mesoderm. This characterization of limbless as an ectodermal mutation which precludes the formation of an apical ectodermal ridge not only adds to our knowledge of tissue interactions during limb development, but makes limbless a potentially valuable tool for biochemical and molecular studies of apical ectodermal ridge formation and function.

Flocks maintained at the University of Wisconsin were the source of fertile quail (Coturnix coturnix japonica) eggs and fertile chicken eggs from limbless stock. For studies grafting mutant and normal ectoderm to normal hosts, host white Leghorn embryos were from Sunnyside Poultry Farm, Inc. in Oregon, WI.

The recombinant method used for this study has been described elsewhere (Carrington & Fallon, 19846) and is shown in Fig. 1. To prepare pieces of donor flank ectoderm the flank regions from both right and left sides of quail embryos of stages equivalent to chick stage 15 or 16 were treated with 1 % di-sodium salt of ethylenediaminetetraacetate (EDTA) (Errick & Saunders, 1976) in doublestrength calcium-/magnesium-free Tyrode’s solution (Kato, 1969). Each piece was then placed in cold 10 % horse serum in Tyrode’s solution where the ectoderm was dissected free and stained lightly with Nile Blue A.

To prepare hosts, a window was made in chicken eggs from limbless heterozygote matings and stage-15 embryos of unknown phenotype were selected for use. After removal of overlying membranes and light staining with Nile Blue A, the presumptive right wing bud ectoderm was mechanically peeled away using forceps and glass needles. This process of in ovo removal of ectoderm destroys the ectoderm as it is peeled off so that it cannot be used as a graft. Two pieces of quail flank ectoderm were then pipetted to each host egg, positioned and allowed to heal over the presumptive right wing mesoderm of the host. No attempt was made to orient the donor ectoderm with respect to the host body axes or with respect to internal and external surfaces of the ectoderm. At the stages used, the lack of specific orientation of the ectoderm graft on the host mesoderm does not affect the subsequent outgrowth of the recombinant bud (Carrington & Fallon, 1984b). Eggs were then sealed and allowed to develop to stages suitable for sectioning or to 10–12 days for examination of cartilage structures. The phenotype of each host embryo was determined at two days after grafting.

For experiments in which limbless ectoderm was tested for a response to inductive mesoderm, the flank ectoderm was isolated from stage-19 limbless embryos and grafted to the presumptive right wing bud mesoderm of a normal stage-15 chick host. In each case, the limb buds of the donor embryo were fixed for later histological analysis to confirm the limbless phenotype. Stage-19 limbless limb buds have no apical ectodermal ridge and there is widespread cell death in the mesoderm.

For paraffin sections, embryos were fixed overnight in a mixture of 70% ethanol, 40% formaldehyde and glacial acetic acid in a ratio of 17:2:1, dehydrated and embedded in paraffin. Sections of the limb bud, 5 μm thick, were cut so that the ridge was seen in cross-section. After staining with the Feulgen reagent for DNA, quail cells were easily distinguished from chick cells in each recombinant wing bud or wing due to darkly staining clumps of heterochromatin in the quail cell nuclei (Le Douarin & Barq, 1969). In some cases, the recombinant limb buds were stained en bloc with the Feulgen reagent (Carrington & Fallon, 1984b) and then embedded in plastic for sectioning with steel knives. Embryos to be embedded in plastic were fixed in 0·02% picric acid, 2·0% formalin and 2·5% glutaraldehyde in 0·075 M-phosphate buffer (Fallon & Kelley, 1977), rinsed and processed according to standard techniques. For examination of whole cartilage structures, embryos were fixed in 10% formalin, stained with Victoria blue, then cleared.

Grafted pieces of ectoderm generally adhered to the host mesoderm almost immediately. By 40 h after the operation, a stage-22 to -24 right wing bud had formed in most cases. Twelve recombinant buds were fixed for sectioning at stages 22–24, and 90 hosts with recombinant wings were allowed to develop to 10 days or more of incubation. Quail cells were never seen in the mesoderm of any recombinant limb bud. Recombinant wings with cartilage elements representative of all proximodistal levels were counted as distally complete. Approximately 23% of the hosts turned out to be limbless homozygotes. The remaining siblings were normal.

69 embryos with normal phenotype were allowed to develop to 10 days or more. The majority of these developed distally complete recombinant right wings (Table 1). Five buds from normal embryos were sectioned and examined for the presence of quail ectoderm. The distal tip of the bud was lost from one recombinant so this case will not be discussed further. In the other four recombinants of quail ectoderm and normal wing bud mesoderm, quail cells formed the ridge and most or all of the non-ridge ectoderm on the bud.

In 55 cases, the shoulder girdle was examined. In other cases, the body of the embryo was used for a separate experiment and was not available for examination of the shoulder girdle. 53 of the hosts that were examined had an outgrowth on the operated side; 51 of these had normal shoulder girdles. In two of the recombinants with an outgrowth the coracoid and scapula had not separated. Therefore, it appears that the grafting operation itself can result in lack of separation of the coracoid and scapula in a small percentage (4%) of cases. In the remaining two embryos that were examined, no right wings formed and the coracoid and scapula separated in one embryo but not the other. In all cases, the left coracoid and scapula articulated normally and the sternum was normal.

21 limbless hosts with grafts of quail ectoderm were allowed to develop to 10 days or longer (Table 1). The left wing bud and leg buds of these embryos failed to form limbs in every case and served as controls, having had no graft of normal ectoderm. The right wing bud, with the graft of quail ectoderm, developed as a normal limb bud in the majority of embryos (Fig. 2). The proportion of outgrowths (90%) of limbless wing bud mesoderm and normal flank ectoderm recombinants was comparable to the proportion of outgrowths (96%) of recombinants of normal mesoderm and ectoderm.

Two 10-day-old limbless hosts produced no wing at all. One of these was sectioned through the wing level of the body. Examination of approximately every tenth section showed only a few scattered groupings of quail cells in the skin, indicating poor survival of the graft in this case.

Seven recombinant buds from limbless hosts were sectioned at stages 22–24. All of these were morphologically normal right wing buds. Ln every case, the apical ectodermal ridge was composed of quail cells and the mesenchyme was composed of chick cells from the host (Fig. 3). No abnormal cell death was detected in any of the recombinant buds. In each case, the left, unoperated, wing bud showed widespread, abnormal cell death, was regressing and had no ridge.

The bodies of eight limbless recombinant hosts were available for examination of the right and left shoulder girdles. Seven recombinants had outgrowths consisting of at least the humerus and ulna. In all seven of these embryos, the coracoid and scapula articulated normally on the operated side. In the remaining case, where no limb developed, the coracoid and scapula did not separate. In every limbless homozygote, the left coracoid and scapula failed to separate. In addition, the sternum in each embryo examined was slightly asymmetrical; the right side, with the graft of normal ectoderm, being slightly longer than the left side. This was true, as well, of the embryo that developed no wing.

The foregoing experiments imply that the limbless mesoderm is normal and limbless ectoderm cannot respond to the normal ridge inductive signals in the mesoderm. A direct test of this would be to graft stage-15 limbless ectoderm to normal hosts according to our procedure. However, in these experiments the ectoderm donor is killed to obtain the graft ectoderm. Because the limbless phenotype cannot be distinguished at stage 15, this would make it impossible to distinguish the phenotype of the graft ectoderm. In the past, early recombination studies with mutant ectoderm have been done by grafting presumptive limb bud mesoderm to a host flank and allowing the host ectoderm to heal over the mesoderm. The host (i.e. ectoderm) phenotype was determined later (Fraser & Abbott, 1971). However, this method gives a bud with mesoderm from two sources, host and donor, mixed together (Dhouailly & Kieny, 1972). In addition, the host mesoderm that is in the bud is from a non-limb source. Simply peeling off ectoderm in ovo from a possible limbless donor at stage 15 or 16 does not yield a viable graft of ectoderm, making impossible a simple reversal of our procedure using limbless embryos as donors.

We have tried several methods of obtaining a graftable ectoderm from in ovo stage-15 embryos without success. Experiments using embryos from limbless heterozygote matings randomly selected as ectoderm donors would require a prohibitively large number of operations to show that the possible limbless ectoderm results in the expected 25 % proportion of failed outgrowths. Taking into account a 95 % expected success rate for the operation and a = 0·01, demonstrating that the ideal proportion of successful operations (71 % successful and 29 % without right wings), was a true representative proportion of the population of donors used would require over 104 operations. Our experience with this procedure at early stages makes it clear that this would be a minimum number and that, in fact, several times that number of operations could be necessary; a practical impossibility considering the limited size of the limbless carrier flock.

Fortunately, there is a way to solve this difficulty. We have shown that flank ectoderm from normal embryos at stages 15 through 19, when grafted over stage-15 presumptive wing bud mesoderm, can form a ridge and promote outgrowth of a wing (Carrington & Fallon, 1984b). Therefore, because the limbless phenotype can first be identified at stage 19, we tested limbless flank ectoderm at this stage for its capacity to respond to inductive mesoderm by promoting outgrowth of a wing.

In 13 control operations, stage-19 flank ectoderm from normal siblings was grafted to the wing region of stage-15 normal hosts. Right wing outgrowths developed in ten cases (77%). Of these, seven were distally complete wings. This result is comparable to previous results of grafting stage-19 flank ectoderm to normal stage-15 wing bud mesoderm (Carrington & Fallon, 1984b). One recombinant bud resulting from a graft of stage-19 normal ectoderm on a normal host was sectioned at stage 22 and had a normal ridge. In ten operations, stage-19 flank ectoderm from limbless donors was grafted to the presumptive right wing region of normal stage-15 hosts. None of these recombinants formed any wing elements on the operated side. A bud formed at the graft site by 24–30h after the operation, but soon after that time the bud began to regress and no limb elements developed (Fig. 4). In addition, the coracoid and scapula failed to separate on the right side in each of the hosts. In every case the unoperated limbs developed normally and the left shoulder girdle was normal. Two additional recombinant buds made from limbless flank ectoderm and normal chick mesoderm were sectioned at stage 22. Neither of these had a ridge and extensive mesodermal cell death was observed in both cases (Fig. 5).

The results from our experiments show that limbless ectoderm is affected by the mutation and does not participate fully in limb development, while the wing bud mesoderm in the limbless mutant behaves normally. Specifically, the ectoderm does not form a ridge on the limbless embryo in response to the phenotypically normal limbless mesoderm: the limb mesoderm buds, yet no ridge forms on limbless limb buds at any time. Further, stage-19 limbless flank ectoderm does not respond to ridge inductive signals from normal presumptive limb bud mesoderm when normal flank ectoderm does form a ridge in similar recombinants. Lack of ridge activity results in total elimination of the bud mesoderm by cell death beginning at stage 19. This indicates that ectoderm on the limbless embryo at early-limb-bud-forming stages and through stage 19 is affected by the mutation and cannot form a ridge. Comparisons of limbless tissues with normal tissues could be valuable in further understanding ridge induction and function. In addition, because limb buds do form in the limbless embryo, we now know budding must be independent of apical ectodermal ridge formation. Further, ridge function becomes necessary for mesoderm stabilization and bud elongation by stage 19.

In general, the pseudostratified columnar morphology of the ridge epithelium is correlated with the ability of limb bud ectoderm to support outgrowth of limb bud mesoderm (Saunders, 1949; Rubin & Saunders, 1972). However, under certain conditions, ridge ectoderm can retain potential function after loss of its typical morphology. For example, when a stage-19 apical ectodermal ridge is removed from contact with stage-19 mesoderm and placed, instead, over stage-28 mesoderm for two days the ridge cells lose their normal high Columnar ridge configuration and form a flattened, thick ectoderm. When this same ectoderm is then placed in contact with stage-19 limb bud mesoderm, it again forms a morphologically and functionally normal ridge (Rubin & Saunders, 1972). Further, evidence for the possible separation of ridge structure and function has been presented by Errick & Saunders (1974). Limbless ectoderm does not form a ridge structure and it has no potential ridge function, as can be seen by its lack of ability to promote outgrowth of both stage-15 presumptive wing bud mesoderm and the apparently normal limbless mesoderm.

The large amount of cell death in the limbless limb bud mesoderm (Fallon et al. 1983) indicates that limbless ectoderm does not maintain healthy limb bud mesoderm after the initial formation of the bud. A direct effect of the limbless gene causing mesodermal cell death is possible. However, such an effect would have to be reversible because combining the mesoderm with normal ectoderm, either at stage 15 or at stages 19–22 (Fallon et al. 1983), results in survival of the mesoderm. Alternatively, the death of limbless limb bud mesoderm may be similar to the situation where removal of the ridge from stage-18 to -20 normal limb buds results in cell death in the distal 150–200μm of underlying mesoderm (Rowe et al. 1982) and deletion of distal limb parts (Saunders, 1948). In our experiments with limbless mesoderm, the normal ectoderm of recombinant wing buds formed a ridge and the buds had no abnormal cell death. Thus, the fact that no ridge forms on unoperated limbless limb buds accounts for the observed mesodermal cell death and eventual regression of the buds.

We wish to stress that, although cell death is seen in normal wing buds after ridge removal (Rowe et al. 1982), proximal wing parts do form (Saunders, 1948). Normal wing buds from which the ridge is removed when it first appears at stage 18 or 19 form a shoulder girdle and partial humerus or humerus and partial radius and ulna, respectively (Saunders, 1948; Summerbell, 1974). Removal of the ridge after it is morphologically distinguishable at stage 18 does not affect parts already stabilized in the mesoderm (Summerbell, 1974). Limbless embryos do form stage-18 wing buds but never form any part of a humerus or more distal wing cartilages. However, the limbless limb bud mesoderm cells have not been in contact with ridge activity at any time. Therefore, we conclude that in normal limb buds at the initial stages when a ridge is not yet visible, the future apical ectoderm already has ‘ridge’ activity that is necessary for the stability of elements that are determined in the mesoderm. Because the limbless mesoderm is never exposed to a ridge, the entire limb bud mesoderm is unstable and degenerates. In spite of this lack of stabilization (ridge activity), a bud can form.

During normal development, the shoulder girdle in the chick forms by division of a single precartilagi-nous mesenchymal condensation (Hinchliffe & Ede, 1968; Hall, 1986) formed from cells of somatopleural and somitic origin destined to form the coracoid and scapula, respectively (Chevallier, 1977). The coracoid and scapula of unoperated limbless embryos do not separate. From the data presented here, it appears that this is not an irreversible, direct effect of the mutant gene. Rather, the abnormality of the limbless shoulder girdle is secondary to the gene’s effect on the ectoderm. Grafting a normal ectoderm leads to normal separation of the coracoid and scapula. It is likely that the presence of the humerus is required for the separation of the coracoid and scapula, as we have found that any time the humerus is missing (for example, in limbless and wingless) the shoulder girdle elements fail to separate (personal observation). There has been no information to date indicating a direct effect of overlying ectoderm on separation of these two elements.

The limbless mutation, through an autosomal recessive gene, results in totally limbless chickens because the embryonic ectoderm does not form an apical ectodermal ridge. Study of this mutant indicates that ridge activity in normal limb buds is present during the time the buds form and before a ridge is morphologically distinguishable. The fact that limbless embryos lack ridge activity but form buds leads to the conclusion that budding of limb bud mesoderm can take place without ridge activity, adding to our knowledge of tissue interactions necessary for limb bud formation and outgrowth. In addition, our studies characterizing limbless as an ectodermal mutant that does not form an apical ectodermal ridge make limbless a useful tool for the study of cellular and molecular mechanisms inherent in ridge formation and function.

This research was supported by NIH grant no. 5T32HD07118 and NSF grant no. PCM 8406638. We thank A. Clark, L. Dvorak, R. Fuldner, L. Klein, J. Petterson, M. Savage, B. K. Simandl, D. Slautterback and W. Todt for their constructive criticism of this manuscript. We thank Lucy Taylor for the drawing. Special thanks are due to Dennis Summerbell for reading the final manuscript.

A preliminary report of some of this work appeared in Fallon et al. 1983. No data on the experiments reported here were presented.