Evidence for a positional memory in the development of the chick wing bud

Pattern formation in the developing chick limb bud can be viewed in terms of positional information. Cells are assigned positional values in a three-dimensional coordinate system, and interpretation of these positional values leads to the appropriate cell behaviour (Wolpert, 1969; Wolpert, Lewis & Summerbell, 1975). For the antero-posterior axis of the chick limb, it appears that specification of positional information is due to a group of cells at the posterior margin of the limb - the zone of polarizing activity (ZPA). If an additional ZPA is grafted to the anterior margin of a wing bud, the wing that develops has mirror-image symmetry about its long axis (Saunders & Gasseling, 1968; MacCabe, Gasseling & Saunders, 1973; Summerbell, 1974a; Tickle, Summerbell & Wolpert, 1975; Fallon & Crosby, 1977; Summerbell & Tickle, 1977). By grafting an additional ZPA to different positions along the antero-posterior axis, Tickle et al. (1975) found that structures form according to their distance from the ZPA. They suggested that positional information is specified by the concentration of a diffusible morphogen produced by the ZPA. In support of this Smith, Tickle & Wolpert (1978) have been able progressively to attenuate the signal from the ZPA by treating it with increasing doses of γ-radiation, and evidence for a gradient of a morphogen comes from the in vitro bioassay of MacCabe & Parker (1975, 1976).

However, some doubt has been expressed about the role of the ZPA in limb development (see Saunders, 1977, for a discussion). One difficulty is that removal of the ZPA from a limb does not affect the sequence of digits along the antero-posterior axis, even though the ZPA is not regenerated (Saunders, 1972; MacCabe et al. 1973; Tickle et al. 1975; Fallon & Crosby, 1975). This has led to the suggestion that the ZPA is effective only very early in limb development and that thereafter its influence can be remembered by the responding cells (Tickle et al. 1975; Fallon & Crosby, 1975, 1977). In the experiments described here I support this conclusion by showing that the effect of a grafted ZPA may also be remembered.

The method takes advantage of the discovery that a graft of a quail ZPA treated with 10000 rad γ-radiation is capable of producing a complete re-duplication (Smith et al. 1978). Because the irradiated cells cannot divide normally the graft tends to remain at the base of the limb, too far from the tip to exert its polarizing activity; Summerbell (1974a) found that a graft of a ZPA to a proximal level of a late limb does not produce a reduplication, presumably because the distance over which it can signal through proximal tissues is limited. It is shown that the signalling ability of the graft is not passed on to more distal tissue, and this demonstrates that a reduplication can be obtained without the continual presence of an additional ZPA.

Fertilized White Leghorn eggs were incubated at 37·5 °C and windowed on the third day of development. The embryos were staged according to Hamburger & Hamilton (1951), sealed with Sellotape, and returned to the incubator. The eggs were examined at intervals, and those at stage 17/18 were used as hosts. A graft site was prepared by excising a small piece of tissue, about 200 μm cubed, from the right wing bud opposite somite 16.

Quail embryos at stages 21–24 were irradiated as described previously (Smith et al. 1978) with 10000 rad γ-radiation, and pieces of ZPA tissue, also about 200 μm cubed, were transfixed with a platinum pin and positioned in the host embryos. Camera-lucida drawings of the host limbs were made immediately after the operation, and approximately 12, 24 and 36 h later. The grafted tissue could be recognized for at least 24 h after the operation because it appeared white against the host limb. Only embryos in which the graft could clearly be seen to be left behind at the base of the limb, and in which the operated limb bud had widened symmetrically, were used in next stage of the experiment. The wings of the remainder were fixed and stained by the method below at 10 days of incubation, and the results are not considered here.

After 36 h each suitable embryo was treated in one of three ways:

In the first group embryos were allowed to develop to 10 days of incubation, and then both wings were fixed in 5 % trichloroacetic acid, stained in 0·1% Alcian green 2GX in 1 % concentrated hydrochloric acid in 70 % alcohol, differentiated in acid alcohol, dehydrated and cleared in methyl salicylate.

In a second group the operated wing bud was fixed in half-strength Karnov-sky’s fixative (Karnovsky, 1965), dehydrated, and embedded in Araldite. μm serial sections were cut in a plane containing the proximo-distal and antero-posterior axes of the limb, and they were stained alternately with toluidine blue and by the Feulgen technique. These sections were to confirm that the grafted tissue had remained at the base of the limb.

In the last group the operated limb bud was removed from the embryo and placed in a dish of Hanks Balanced Salts Solution. Pieces of tissue from the anterior and posterior margins were assayed for polarizing activity by grafting them to the anterior margins of wing buds of host embryos at stages 18−21. In some cases the tip of the operated bud was allowed to continue its development by grafting it, in its normal orientation, to the stump of a stage-18 or stage-19 wing bud. These embryos were allowed to develop for a further 6 days before the limbs were fixed, stained and whole-mounted by the method above. The unoperated wing bud was fixed, stained and whole-mounted. It was staged by measuring its proximo-distal length and referring to Summerbell (1976).

Sixty-two grafts of irradiated quail ZPA were made to stage-17/-18 hosts. In 25 of these the ZPA was clearly left at the base of the limb, and the limb had widened symmetrically. A typical series of camera-lucida drawings is shown in Fig. 1. The widening of the limb is one of the earliest responses to a ZPA graft, and indicates that a reduplicated limb will be formed. In the 37 other cases either the grafted tissue was carried forward as the limb grew out, or the limb bud failed to widen sufficiently, or both. The last camera-lucida drawing was made about 36 h after the graft, when the embryo was at stage 24 or 24/25, and the limbs were then treated as described in the Methods.

Nine embryos were allowed to continue development, of which seven survived to 10 days of incubation. All produced good reduplications, five with digit 4 posterior to the graft, and two with digit 3 posterior to the graft (Fig. 2).

Three limbs were examined histologically 36 h after the operation to confirm the position of the graft. The grafted tissue was easily recognizable even in sections stained with toluidine blue (Fig. 3). The cells were larger than the surrounding host mesoderm cells, and there were macrophages present. The cell density in the graft was lower than in the host, and there was a well-defined border between the two. The nuclei of the grafted cells stained weakly with Feulgen, so that the quail nucleolar marker (Le Douarin, 1973) was indistinct. In all three cases the main body of the graft was positioned at the base of the limb, and a thin layer of cells extended distally up to almost half the length of the limb, just under the ectoderm. These cells could have been dragged by the ectoderm, which slides distalwards over the mesoderm (Searls & Zwilling, 1964; Amprino, 1965).

Tissue from 13 operated limbs was assayed for polarizing activity, as shown in Fig. 1D. Twelve grafts of the more distal piece of tissue on the anterior margin (piece a, Fig. 1D) survived, and none of these grafts produced reduplications in the host wing (Fig. 4A). Quite frequently, however, there were abnormalities in the cartilage pattern of the host limb. In two limbs the humerus was short and thick, and in another two the radius was thick. A further two limbs had a spur of cartilage coming from the elbow. Twelve grafts of the proximal piece of tissue on the anterior margin (piece b, Fig. 1D) survived. Of these nine produced no reduplication and three produced an extra digit 2 in the host limb (Fig. 4B). It is possible that the pieces of tissue grafted in these three cases contained some of the original graft. This would produce the weak polarizing activity that was found. As a control, ZPA tissue from the posterior margin of the limb (pieces c and d, Fig. 1D) was grafted. Eighteen operations survived, and 17 gave good reduplications, 13 with digit 4 (Fig. 4C) and four with digit 3 posterior to the graft. One gave only an extra digit 2. Finally, six grafts of the tip of the limb survived. All gave a reduplicated hand on the host humerus or forearm. Three of these had digit 4 anteriorly, in two cases rather reduced (Fig. 4D), and three had digit 3 anteriorly. The lack of a digit 4 in these cases might be because presumptive digit 4 was removed in the assay for polarizing activity.

It is clear from these results that the signalling ability of the irradiated ZPAs is not passed on to more distal tissue. Twenty-one out of 24 pieces of tissue from the anterior margin gave no reduplication, and it seems likely that the polarizing activity found in three pieces (all of them from next to the grafted tissue) was due to contamination by the graft. The controls, grafts of ZPA tissue, gave good reduplications in 17 out of 18 cases.

The signalling ability of an irradiated ZPA left behind at the base of a limb as the limb grows out is not transmitted to more distal tissue. This has enabled me to demonstrate that the continued presence of an additional ZPA is not required to obtain a reduplicated limb. I shall discuss these two points separately.

Maps of the ZPA (for example, MacCabe et al. 1973) show that its activity is found progressively more distal as development proceeds. There are two theories suggesting how this occurs. The earlier is that of Saunders (1972) who proposed that ZPA cells induce cells distal to them to become ZPA. In contrast, Summerbell & Tickle (1977) suggested that cells which are to be ZPA are clonally propagated throughout limb development. The results in this paper show that the signalling ability of an irradiated ZPA is not transmitted distally, and this supports the view that ZPA activity is clonally propagated. One reservation is that irradiation, which attenuates positional signalling (Smith et al. 1978), might similarly abolish the ability of the ZPA to induce adjacent tissue to become ZPA (Summerbell, personal communication). However, as I discuss below, it is clear that distal propagation of ZPA activity is not required to produce a reduplication.

Grafts of a quail ZPA treated with 10000 rad γ-radiation to a stage-17/-18 wing bud produce a complete reduplication even though the graft remains at the base of the limb and the signalling ability of the ZPA is not transmitted to adjacent tissue. By stage 24, when the digits have yet to be specified (Summerbell, 1974b), the ZPA is too far from the tip to exert its polarizing activity (Summer-bell, 1974a). It follows that the cells at the tip remember their exposure to the ZPA when they go on to form a reduplication. They may be said to have a positional memory. It is not known how long cells must be exposed to the ZPA, but preliminary experiments in which an additional ZPA is removed at different times after the graft suggest that it is between 12 and 17 h (J. C. Smith, work in progress). The ability of cells to remember their positional value explains why removal of the ZPA from a limb does not affect pattern along the antero-posterior axis. Such experiments can no longer be interpreted as showing that the ZPA has no role in limb development.

This example of positional memory illustrates the more general point that positional value is a stable cell state that does not depend upon the continued presence of any positional cue (Wolpert & Lewis, 1975; Lewis, Slack & Wolpert, 1977). Such stability of positional value is well illustrated in amphibian limb regeneration, where positional values ascribed during embryonic development direct the development of the regenerate. A special case of this has recently been described by Slack & Savage (1978,a, b) who find that amputation of a reduplicated axolotl limb is followed by the regeneration of a reduplicate. Another example is seen in the cockroach leg, where grafting experiments reveal that cells in the epidermis are aware of their circumferential position, and apposition of tissues with different positional values brings about intercalary regeneration (French, Bryant & Bryant, 1976).

Positional values can, however, change, as in the ZPA grafts performed here and by many others (see Introduction). It has been suggested by Tickle et al. (1975) for the chick limb that positional value is stable to a decrease in morphogen concentration but can be raised by an increase in morphogen concentration. This rule has also been applied to the insect egg by Meinhardt (1977), and may be useful in elucidating the mechanisms by which positional information is interpreted (Lewis et al. 1977; Meinhardt, 1978).

I thank Dennis Summerbell, Cheryll Tickle, Fiona Watt and Lewis Wolpert for their encouragement and comments on the manuscript. I also thank the MRC for a research studentship.