Analysis of upper beak defects in chicken embryos following with retinoic acid

Recently we have found that local application of retinoic acid to chick wing buds leads to the development of additional digits. Associated with this effect of retinoic acid on pattern formation in the treated limb, the embryos frequently had upper beak defects (Tickle, Alberts, Wolpert & Lee, 1982). Using a carrier that releases retinoic acid more effectively to apply the chemical to the developing limb buds (Eichele, Tickle & Alberts, 1984), almost 100% of the embryos exposed to the highest concentrations of retinoic acid also have a beak effect. This provides a good system for the analysis of such beak defects since their abnormal morphogenesis can be followed during development.

Analysis of these upper beak defects can provide insights into mechanisms of orofacial morphogenesis and how such abnormalities may arise, since, at early stages, the development of the face of birds is similar to that of mammals.

The teratogenic effect of vitamin A derivatives on facial morphogenesis of mammalian embryos is well known (Morriss, 1973). Embryos of pregnant rats given excess vitamin A during day 8 of gestation show abnormal orofacial morphogenesis including upper jaw defects such as cleft palate. At the time of treatment, cranial neural crest cells are migrating into the tissues that will form the upper jaw and it has been suggested that the way in which vitamin A causes the abnormalities is by interfering with this migration (Thorogood et al. 1982). A recent report (Jelenik & Kistler, 1981) describes the teratogenic effect of retinoic acid applied systemically on orofacial morphogenesis of chick embryos on days 3 and 4, i.e. the same age as those treated in our experiments. The orofacial defect produced was bilateral cleft palate but a description of the defect was not given and its incidence was quite low. The development of the limbs was also affected.

In view of the association between orofacial and limb defects following retinoic acid treatment, the chicken mutant discovered by Abbott & MacCabe (1966) and subsequently named ‘cleft primary palate’ cpp (Yee & Abbott, 1978) is of great interest. This mutation also affects both the upper jaw and limb development, although the lower limbs only are affected. In the mutant, embryos the upper beak is absent. A study of the morphogenesis of the face of these embryos showed that the defect appears to result because fusion of the median nasal process with the lateral nasal processes does not occur (Yee & Abbott, 1978).

Here, we describe the morphology of the upper beak defects that develop following exposure of chick embryos to retinoic acid applied on a controlled release carrier to the wing bud. To analyse how the defects have arisen we have studied how the normal beak develops. Recent studies dealing with upper beak development in the chick (Yee & Abbott, 1978; Shah & Crawford, 1980; Koch & Smiley, 1981; Will & Meller, 1981; Greene et al 1983) have emphasized cytological aspects and/or the topography of early developmental stages. Since these reports did not follow normal development beyond stage 28 (about 6 days incubation: Hamburger & Hamilton, 1951) we have studied later stages up to the formation of the definitive beak. As a further aid in the analysis of the defects, we have studied the morphogenesis of the abnormal beaks.

Chicken embryos (stages 20–21, Hamilton-Hamburger stages) were treated with a\ï-trans retinoic acid (Sigma, lot Nos. 41F-0440 or 12F-0598) by implanting AG1-X2 beads (200 pm in diameter) loaded with this chemical to anterior positions in the right wing buds. The standard procedure for loading the beads was a 20 min soak in retinoic acid solutions in DMSO, followed by two rinses in tissue culture medium and a final 20 min wash in tissue culture medium at 38 °C. The beads were implanted in slits cut between the apical ectodermal ridge and the mesenchyme over the anterior part of the bud. Fig. 1 shows an embryo fixed 2h after the operation and illustrates the relative positions of the retinoic acid impregnated bead in the right wing bud, and the developing face.

For determining the incidence of beak defects the embryos were reincubated at 38 °C after implanting the bead, and the gross effects of exposure to retinoic acid were observed in 10-day embryos (stage 35–36). The heads were fixed in 5% TCA and stored in 70% alcohol. The associated effect of retinoic acid treatment on the development of the right wing was assessed by fixing the limb in 5% TCA and staining with Alcian Green to show the skeletal pattern.

For the detailed study of the morphology of the abnormal beaks and the development of both the normal and abnormal faces, heads from stage 28 to stage 35 were prepared for scanning electron microscopy. (The bodies of these embryos were fixed in 5% TCA to aid staging of the normal series of embryos and to check right limb development in the series of treated embryos). The heads were fixed in 1% glutaraldehyde made up in a modified Tyrode’s solution (Litze & Low, 1977) at 4 °C for at least 24h. The heads were then washed several times in distilled water and then left in distilled water for at least 12h. Following several changes of absolute alcohol over a period of 24h, the heads were placed in amyl acetate as the transfer liquid (2 changes: 20min each) for critical-point drying in liquid CO₂. The dried specimens were mounted on aluminium stubs with silver conducting paint and coated with gold-palladium alloy. The specimens were viewed using either a Cambridge 410 Stereoscan or a Jeol JSM-35 scanning electron microscope. The SEM data are based on the examination of 25 normal and 16 abnormal specimens representing the relevant stages of development.

The incidence of beak defects noted following implantation of beads loaded with retinoic acid to the anterior margins of right wing buds of stage 20 hosts is shown in Table 1 for our first series of experiments. With beads soaked in solutions of 10 mg/ml and 1 mg/ml retinoic acid (Sigma lot: 41F-0440) there is virtually 100% production of beak defects. The defect is immediately recognizable in the 10-day embryos - the upper beak is completely missing, no nostrils form and the front of the face ends in a scalloped rim of tissue bordering the mouth (Fig. 2). In contrast, the lower beak appears relatively unaffected although a shortening cannot be ruled out. In a few cases the lower beak arched downwards suggesting that the upper beak may influence its direction of growth.

The symmetrically truncated upper beak is typical of embryos to which beads soaked in 10 mg/ml retinoic acid have been implanted. With beads soaked in 1 mg/ml the faces of the treated embryos are also truncated but in some cases it is now possible to detect an asymmetry in the effect; the right side being more affected than the left. Such an asymmetry in the defect and a reduction in its severity is marked in the few affected embryos treated with beads soaked in 0-1 mg/ml retinoic acid. In two of these embryos, the nostril and the beak margin on the left side are present, while on the right, the nostril and the beak margin are rudimentary. In the other two cases of affected embryos at 0-1 mg/ml, nostrils are formed on both sides but the upper beak is very short. One of these short beaks had a well-developed egg tooth.

In the majority of the cases of the first series of experiments (Table 1), the beak defect is associated with truncation of the right wing. The other effects on the treated limbs are a reduction in the number of digits, or symmetrical hand plates made up of three digits in the pattern 434 (see Tickle, Lee & Eichele, in preparation, for more details). It should be noted that the only case in which the face of an embryo was normal following implantation of a bead soaked in 10 mg/ml retinoic acid, the right limb of that embryo was also normal.

Later, experiments were carried out using different batches of retinoic acid either lot NO.12F-0598 from Sigma or Ro 1:5488 from Hoffman la-Roche. The results are qualitatively similar. Thus 100% of embryos receiving beads soaked in 10 mg/ml (11 cases) have a missing upper beak and with beads soaked in 1 mg/ml, the incidence of beak defects in treated embryos (9 cases) is 90%. However, these batches of retinoic acid appear to be less potent (Tickle et al., in preparation). Thus, the faces of the treated embryos resemble those of embryos treated with a ten times lower concentration of the batch of retinoic acid used in the first series of experiments. Thus with beads soaked in 10 mg/ml of these less ‘potent’ batches of retinoic acid the upper beak is completely absent but in some cases a difference in the effect on the right and left sides of the face can be detected. With beads soaked in 1 mg/ml the beak defects are less severe and in about 50% of the cases, nostrils are present but the upper beak is very short. At 10 mg/ml, the beak defect is still mostly associated with limb truncations but at 1 mg/ml is associated with symmetrical pattern of up to four digits (digit pattern 4334). .

The detailed study of the morphology and morphogenesis of the defective beaks was carried out on embryos into which beads soaked in 10 mg/ml retinoic acid had been implanted. In most cases, retinoic acid from Sigma, lot No. 12F-0598, was used. The effect of the retinoic-acid-impregnated implant on the face. appears to be due to exposure to the chemical that diffuses across the extraembryonic space (see Fig. 1). An asymmetrical effect on the face has already been noted, with the right side nearer the bead being more affected than the left. Furthermore, when beads soaked in 10 mg/ml retinoic acid are placed farther away from the face, for example, in posterior positions in the developing wing, the defects are not so severe. Typically in these cases, the nostrils form but the upper beak is short. These effects are consistent with the idea that proximity of the developing face to the implant site leads to the facial defect. We can estimate, from the amount of retinoic acid released from these beads into tissue culture medium (Eichele et al. 1984), that the effective concentration range of retinoic acid to which the face is exposed is rising to about 10⁻⁷M over 24h. Preliminary experiments show that at most a 24 h exposure is sufficient to bring about the facial deformity.

To be able to interpret the upper beak defects, one needs to understand how the normal beak develops. The early stages in face development have been well documented (Yee & Abbott, 1978; Will & Meller, 1981). To summarize; at the time when retinoic acid is applied (stage-20 embryos), the tissue masses, that will form the face, border the stomodeum in a square-shaped configuration. The masses of tissue that will form the upper beak and the primary palate consist of a median frontonasal mass separated from the lateral nasal processes by the invaginating nasal placodes at′each top corner of the square. The lateral nasal processes have merged with the maxillary processes which form the lateral sides of the square.

By stage 28 the combined frontonasal and median nasal processes form a relatively broad, flat rectangular mass (frontonasal mass) (Fig. 3). The globular processes of the frontonasal mass are closely apposed to, and partially fused with the corners of the maxillary processes. The medial nasal processes bulge slightly at the rims of the forming nostrils. The lateral nasal processes fill the triangular area bordered by the oronasal fissures and the nasolacrymal grooves.

Inside the stomodeum (Fig. 4), the primordia of the secondary palate (palatal processes) have begun to form as medial enlargements of the maxillary processes. The primordium of the primary palate appears as a slightly bilobed extension of the frontonasal mass. Two small tubercles at the corners of the stomodeal roof are the primordia of the rostral conchae.

At stage 30 (Fig. 5), the midsaggital part of the frontonasal mass is elevated and the egg-tooth rudiment is visible. The frontonasal mass protrudes into the stomodeal antrum towards the lower beak. The edges of the lateral nasal processes bordering the developing nostril slits have a sigmoid outline thus establishing two swellings. The lower parts of the oronasal fissures are almost obliterated by fusion of the globular processes with the maxillary processes.

By stage 31 (Fig. 6), the frontonasal mass extends in the midline over the oral antrum and tip of the lower beak. The globular processes are now quite large and flare laterally between the maxillary processes and the inferior borders of the lateral nasal processes. At this stage, it is quite clear that the main protruding part of the upper beak - that region eventually encased by the keratinized epidermis, the ramphotheca - forms primarily from components originating in the frontonasal mass. In profile view (Fig. 7), the upper beak can be seen to be just beginning to protrude from the face surface while the lower beak is depressed into the stomodeal space. The palatal processes (Fig. 8) are well developed, forming cantilevered shelves that separate the nasal from the oral parts of the stomodeum laterally. The shape of the primary palate is fan-like. The rostral part of the primary palate is demarcated from the facial aspect of the developing beak by a shallow groove which is the border of the ‘lip ridge’ - the tomium. At stage 32 (Fig. 9) the entire bill begins to protrude rostrally although the tip of the upper beak still overlaps the lower one.

At stage 33 (Fig. 10), the upper beak no longer overlaps the lower one and the entire bill extends well forward from the face. Contributions of the various embryonic primordia to the upper beak are easily recognized: the globular processes form the tomium - the lateral borders of the beak, while the midsaggital part of the frontonasal mass forms the beak tip, the egg tooth and the central edge - the culmen. The opercula of the nostrils correspond to the medial nasal processes while the triangular areas between the nostrils and the tomium are formed by the lateral nasal processes.

The palate (Fig. 11) is now further developed. The primary palate forms the rostral portion of the beak. The shelves of the secondary palate now appear as enlarged ridges which bulge into the stomodeum, especially at their rostral ends where they come very close together at the mid line.

Fig. 12 shows the essentially completed upper beak at stage 35. The rostral-medial margins of the shelves of the secondary palate have fused while caudally the shelves remain unjoined to form the choanal slit which connects the oral and nasal chambers (This arrangement is typical of galliform birds). The intra-oral aspect of the tomium is delineated by a sharp groove. It should be noted that the maxillary processes form the border of the upper jaw caudal to the ram photheca as well as the shelves of the secondary palate.

Since nearly every embryo in which a bead soaked in 10 mg/ml retinoic acid has been implanted develops with an upper beak deformity, we can follow the way in which such deformities arise by examining embryos at a sequence of stages following bead implantation. In this section we describe the morphogenesis of the face in treated embryos.

At stage 28 (Fig. 13), the maxillary processes and the globular processes have failed to approximate to each other. Hence the oronasal fissures are wide: the lateral nasal processes form the upper boundary and their medial margins are slightly indented.

At stage 29 (Fig. 14), the oronasal fissures are much wider and the lateral nasal processes fill much of the space between the frontonasal mass and the maxillary processes. At this stage, the indentations of the margins of the lateral nasal processes are very marked, thus dividing them into two tubercles (labelled 2 and 3 in Fig. 14). The medial ones (2) are the smaller ones and they slope towards the frontonasal mass (1) but are separated from it by slits which are homologous to the nostrils in normal embryos. The lateral tubercles β) of the lateral nasal processes extend downwards almost to the margin of the maxillary processes (4) but are separated from them by well-developed depressions. Inside the stomodeum, the palatal shelves have a fairly normal relationship with the maxillary processes.

An asymmetrical specimen was deliberately chosen to illustrate abnormal stage 30 (Figs. 15–16) because the homologies to normal primordia are more easily appreciated in this embryo. Here, the frontonasal mass slopes towards the left side of the embryo and the left oronasal fissure is extremely deep. Tubercles of the left lateral nasal process (2 and 5) and the medial part of the maxillary process (4) form the lateral side of the fissure. The left medial nasal process is a small extension of the frontonasal mass lying above the nostril slit. On the right side of the embryo, the oronasal fissure is relatively shallow and corresponds to the wide space lying between the maxillary process (4) and the right edge of the frontonasal mass. Thus the tubercles of the right lateral nasal process actually border the rim of the stomodeal antrum between the maxillary process and the globular part of the frontonasal mass. The right side shows the morphology most commonly found in the experimental embryos and demonstrates why the nostrils are absent in the abnormal specimens. Fig. 16 shows this specimen in left profile view to emphasize the retarded growth of the frontonasal mass, i.e. it curves into the stomodeum and does not overlap the lower beak.

The facial aspect of stage 31 (Fig. 17) shows the more common, bilaterally symmetrical pattern of abnormal beak development. The upper rim of the stomodeal antrum is formed by seven tubercles: the frontonasal mass with its globular processes projecting laterally to the vestigial nostril slits, the lateral nasal processes in the form of double tubercles and lastly the maxillary processes extending to the corners of the stomodeum on either side. The intrastomodeal aspect of this specimen is shown in Fig. 18. Here the caudal part of the frontonasal mass is seen to form a separate, unpaired tubercle (5) which lies between a pair of smaller tubercles (6) arising from the anterior part of the stomodeal vault. The latter are homologous to the primordia of the rostral conchae. Laterally, the palatal shelves (7) extend part of the way into the stomodeum but their medial edges remain widely separated.

From the foregoing account, it is seen that the nasomaxillary complex in experimental embryos, undergoes arrested growth and distorted morphogenesis wherein the upper ‘beak’ is formed by twelve tubercles homologous to normal primordia, i.e. two unpaired tubercles (7 and 5) which form from the frontonasal mass and five sets of paired tubercles representing the lateral nasal processes (2 and 3), the maxillary processes (4), the palatal processes (7) and the primordia of the rostral conchae (6). Figs 19 and 20 (stage 35) show some of the variations that these elements can assume in different specimens. However, the basic number and arrangement of tubercles in all the observed abnormal embryos followed essentially the same pattern.

Implanting beads that have been soaked in high concentrations of retinoic acid to anterior positions in wing buds leads to abnormal morphogenesis of the face. Typically the upper beak is completely missing and the front of the face forms a scalloped rim of tissue above the mouth. No nostrils are formed. In contrast, the lower beak appears to be essentially normal.

Our analysis of the morphogenesis of normal beaks and the beaks of experimental embryos shows that the scalloped border of the mouth is derived from the tissue masses that would normally fuse to form the upper beak. Thus, seven bulges can be distinguished at the upper margin of the mouth: these are a central tubercle derived from the frontonasal mass, flanked by two tubercles on either side derived from the lateral nasal processes, which in turn are flanked laterally, at the corners of the mouth, by tubercles derived from the maxillary processes. The indentations, that represent the sites where the nostril slits would have formed, lie between the central mass and the tubercles immediately flanking it, i.e. tubercles derived from the upper part of the lateral nasal process. Initially, it was difficult to visualize the derivation of the pair of tubercles lying on either side of the frontonasal mass, but their origin can be understood by reference to the normal development of the lateral nasal process. Normally, this tissue mass borders the lateral edge of the developing nostril and, soon after obliteration of the oronasal fissure, assumes a sigmoid shape with two swellings, one below the other, along the rim of the nostril. In the abnormal embryos these two swellings give rise to the pair of tubercles on either side of the frontonasal mass. They occupy the space created by the lack of fusion across the oronasal fissures, and their position is radically modified since they now come to lie side by side at the upper rim of the mouth.

Associated with the failure of formation of nostrils and upper beak, the development of the palate of treated embryos is also abnormal. The roof of the abnormal oral cavity has five tissue masses - a central bulge and two lateral bulges, flanked by the flat palatal shelves. By reference to the stages of normal development, the central bulge of tissue is derived from the frontonasal mass. In normal embryos, this bulge becomes the part of the primary palate which forms much of the intraoral surface of the beak. The two inner lateral bulges at the roof of the mouth of abnormal embryos are derived from the primordia of the rostral concjiae but they are now distorted and lie level with the palatal surface. The secondary palate in the treated embryos is represented by the maxillary palatal shelves which project medially and are separated by a wide space along their length. In normal embryos, these shelves meet each other in the midline behind the primary palate and extend caudally for a short distance as a united surface until interrupted by the choanal slit. This is the, so-called, ‘schitzognathic’ condition characteristic of galliform birds. (In anseriforms, the shelves of the secondary palate are essentially fused along their entire length) (Bellairs & Jenkin, 1960)).

From the foregoing, the primary facial defect in abnormal embryos can be classified as severe bilateral clefting of the primary palate. The accompanying complete separation of the secondary palate appears to be the consequence of the abnormal development of the primary palate.

Clefting of the primary palate could, in principle, arise because of defective morphogenesis at a number of different steps during development of the face (Trasler & Fraser, 1977). Thus, for example, it has been suggested that clefting could result from failure of the fusion process itself - involving adhesion of the facial primordia and dissolution of the adjacent epithelia - or from only a weak seam being established with subsequent rupture (Poswillo, 1974). However, the severe clefting of the primary palate in our experiments does not result from interference with this step in face development. In our embryos, the origin of the beak defect can be traced back to the failure of the frontonasal mass to approximate to the maxillary processes. The gap between the facial processes appears to arise because the frontonasal mass does not enlarge whereas the maxillary processes appear to be less affected. Indeed, this effect is permanent and results in the later inhibited extension of the beak.

Failure of the facial primordia to enlarge has been recognized as a possible cause of cleft palate (Trasler & Fraser, 1977) and has been ascribed to a reduction in the number of cells within the developing facial processes. Since cells that populate the facial processes are derived from cranial neural crest, it has been suggested that interference with cell migration could lead to fewer cells reaching these primordia. This hypothesis is consistent with the effects of retinoic acid on the facial development of mammals. Thus rat embryos treated at the time of neural crest emigration have reduced cell numbers in the facial primordia and go on to develop cleft palate (Morriss, 1973). Embryonic chicks treated with the Vitamin A derivative, retinol, during neural crest cell emigration into the face have a reduced number of cells in the face primordia (Hassell, Greenberg & Johnston, 1977; Keith, 1977). Tissue culture tests of the effect of retinol on the motility of quail cranial neural crest cells show that cell motility is lessened because their adhesiveness to the substratum is greatly reduced (Thorogood et al., 1982). However, the face primordia are already populated by crest cells at the time of treatment. Indeed, in the experiments in which chicks were cultured in the presence of retinol (Hassell et al., 1977), the end point for assessing the effect is the stage at which we begin the application of retinoic acid.

In our experiments, retinoic acid appears to act on cells that have already migrated into the face primordia, by killing them and/or affecting their proliferation. This would diminish the number of cells and inhibit the enlargement of the frontonasal mass. This interpretation is consistent with experiments in which chick embryos treated with a vitamin A analog, 13 cis retinoic acid, that is known in other systems to be less cytotoxic (reviewed, Pawson, Ehmann, Itri & Sherman, 1982) did not have facial abnormalities (our unpublished observations).

There are other examples of experimental production of cleft palate in which the primary effect appears to be on cells that have already migrated into the facial primordia. Administration of the drug ‘hadacidin’ to pregnant hamsters causes the embryos to develop severe clefts of the primary palate (Shah, 1977). In these experiments, the greatest production of defects was noted when treatment occurred at an analogous stage of development to that of the chick embryos in our experiments. Furthermore in another example, in which tunicamycin was locally applied to the developing primary palate of cultured rat embryos (Eto, Figueroa, Tamura & Pratt, 1981), enlargement of the frontonasal mass is specifically affected and this could result from inhibition of the proliferation of its constituent cells.

It should be noted that the effect of retinoic acid on the growth of the frontonasal process in our embryos is permanent. Another factor, in addition to the proposed diminution of cell number by killing and/or reduction in proliferation, may be involved. This concerns the contribution of mesenchyme to the frontonasal mass from the maxillary processes as a part of the normal fusion sequence, i.e. after the epithelial partitions are perforated, mesenchyme from the maxillary processes is thought to migrate medially (Patten, 1961).

Thus the failure of the upper facial primordia to fuse could be an important factor leading to abnormal morphogenesis of the frontonasal mass. Such an explanation could also shed light on why the lower beak in these treated embryos, in contrast to the upper beak, appears to be unaffected. Morphogenesis of the lower beak does not involve the contribution of cells from adjacent primordia since the paired mandibular processes are essentially in continuity throughout development. Recent experiments on the developmental potential of the frontonasal mass, however, show that small fragments can undergo morphogenesis in isolation (Wedden: work in progress). Thus the lack of contribution of cells from adjacent primordia does not appear to be relevant to the abnormal morphogenesis of the frontonasal mass. Furthermore, this leaves the puzzling question of the apparent immunity of the lower beak still open.

It is of interest to compare our analysis of the morphogenesis of the face of retinoic acid treated embryos with that of the mutant cpp. The origin of the defect in this mutant (as in our study), can be traced back to the failure of the frontonasal mass to enlarge (Yee & Abbott, 1978). However it is not clear whether the basis of abnormal morphogenesis in the mutant involves defective migration of neural crest cells into the face or whether there is a later effect on the proliferation of cells in the established primordia. A further interesting feature of the cpp embryos relevant to experiments with retinoic acid, is that the mutation affects the limbs as well as the face. Indeed the effect of retinoic acid on facial morphogenesis reported here, emerged from a series of experiments in which we were primarily interested in the effects on pattern formation in developing wing buds. At the concentrations of retinoic acid that result in abnormal facial morphogenesis, the treated wings are truncated or have incomplete sets of symmetrical digits. In the cpp mutant, the legs only are defective and have reduced numbers of digits. This mutant is thus reminiscent of embryos treated with the highest concentrations of retinoic acid. At the lower retinoic acid concentrations effective in producing severe clefting of the primary palate, the associated limbs have symmetrical digit patterns. In this context, it is of interest that mutant chickens, such as diplopodia, that have additional digits also have short upper beaks (reviewed, Romanoff, 1972).

An association between limb and facial defects following experimental treatments of embryos has been noted for other chemicals in addition to retinoic acid. Thus, in the example of hadacidin treatment of hamster embryos already mentioned (Shah, 1977), the morphogenesis of the limbs as well as that of the face was affected (the limb defects ranged from amelia to polydactyly but were not described in any detail). In another example, the peak of incidence of limb defects following treatment with pilocarpine coincides with that for facial defects in chicken embryos (Landauer, 1954), although in this case the limbs were truncated. One can speculate that the cellular mechanisms by which these teratogens bring about defects are the same for both the limbs and the upper beaks. However, at present an explanation in these terms is not possible because the mechanisms involved in the development of these two structures are incompletely understood. Nevertheless, the view that beak and limb defects result from interference with the same cellular mechanisms can provide new insights. For example, the limb truncations produced by high concentrations of retinoic acid could be interpreted as resulting from damage to the apical ectodermal ridge that is necessary for limb outgrowth (reviewed, Wolpert, 1976). One could then speculate that outgrowth of the upper beak may similarly require a special region of ectoderm, and failure of beak development results from damage to this epithelium.

This work was supported by the MRC. We are grateful to J. MacCabe for drawing our attention to the cpp mutant. We thank Professor N. Woolf, The Bland Sutton Institute of Pathology, The Middlesex Hospital Medical School for the use of the Jeol scanning electron microscope and Professor A. Boyde, Department of Anatomy and Embryology, University College, London, for the use of the Cambridge Stereoscan. We are also grateful to P. Rowles for technical advice.