DEVELOPMENTAL DYNAMICS 210:274–287 (1997) CHOXC-8 and CHOXD-13 Expression in Embryonic Chick Skin and Cutaneous Appendage Specification BENOÎT KANZLER, FABRICE PRIN, JACQUES THELU, AND DANIELLE DHOUAILLY* Biologie de la Différenciation Epithéliale-UMR CNRS 5538, Etude de la Différenciation et de l’Adhérence Cellulaires, Institut Albert Bonniot, Université Joseph Fourier, Grenoble, France ABSTRACT We studied the expression of two distantly clustered Hox genes which could, respectively, be involved in specification of dorsal feather- and foot scale-forming skin in the chick embryo: cHoxc-8, a median paralog, and cHoxd-13, located at the 58 extremity of the HoxD cluster. The cHoxc-8 transcripts are present at embryonic day 3.5 (E3.5)in the somitic cells, which give rise to the dorsal dermis by E5, and at E6.5–8.5 in the dorsal dermal and epidermal cells during the first stages of feather morphogenesis. The cHoxd-13 transcripts are present at E4.5–9.5 in the autopodial mesenchyme and at E10.5–12.5 in the plantar dermis during the initiation of reticulate scale morphogenesis. Both the cHoxc-8 and cHoxd-13 transcripts are no longer detectable after the anlagen stage of cutaneous appendage morphogenesis. Furthermore, heterotopic dermal–epidermal recombinations of dorsal, plantar, and apteric tissues revealed that the epidermal ability or inability to form feathers is already established by the time of skin formation. Retinoic acid (RA) treatment at E11 induces after 12 hr an inhibition of cHoxd-13 expression in the plantar dermis, followed by the formation of feather filaments on the reticulate scales. When E7.5 dorsal explants are treated with RA for 6 days, they form scale-like structures where the Hox transcripts are no more detectable. Protein analysis revealed that the plantar filaments, made up of feather b-keratins, corresponded to a homeotic transformation, whereas the scale-like structures, composed also of feather b-keratins, were teratoid. These results strengthen the hypothesis that different homeobox genes play a significant role in specifying the regional identity of the different epidermal territories. Dev. Dyn. 1997;210: 274–287. r 1997 Wiley-Liss, Inc. Key words: chick; epidermis; feather; homeobox genes; keratin; retinoic acid; scale mal keratinocytes synthetize specific keratin polypeptides according to their regional origin, thus allowing the molecular identification of the protein content inherent to a given morphology. The a-keratins range from 40 to 73 kD in molecular weight and are expressed in all epithelial cells from fish to human (Moll et al., 1982). The expression of b-keratin polypeptides, which range from 10 to 25 kD, is restricted to reptiles and birds and characterizes such terminally differentiated tissues as scales, claws, beak, and feathers (Gregg et al., 1984). Feathers, which characterize avian skin, present some differences in their morphology and distribution pattern according to the body regions. Furthermore, in the chick, the type of appendage changes from zeugopodial feathers to autopodial scales in the hindlimb. The scales themselves display two main subtypes: the scuta, which are large overlapping scales from the anterior tarsometatarsal region, and the reticula, which are small and roundish scales covering the plantar surface of the foot. Their structural protein content characterizes these three types of appendages: Feathers express lighter b-keratin polypeptides than scutate scales, while the reticulate epidermis is composed mostly of a-keratins (Kemp and Rogers, 1972; Dhouailly et al., 1978; Haake et al., 1984; Sawyer, 1983). These cutaneous appendages develop as a result of successive interactions between the two tissues comprising the skin, an epithelium, the epidermis, and an underlying mesenchyme, the dermis; their regional diversity depends both on the inductive properties of the dermis and on the competence of the epidermis (Sengel et al., 1969; Sengel, 1976; Dhouailly and Sengel, 1983, Cadi et al.; 1983; Dhouailly, 1977, 1984; Sawyer, 1983; Viallet and Dhouailly, 1994). The skin regionalization specification could involve transcription factors belonging to the homeoprotein family. Vertebrate homeobox genes, which are homologous to the first identified genes of the Drosophila antennapedia and bithorax complexes are organized into four clusters, HoxA, B, C, and D, located on different chromosomes (for reviews, see MacGinnis and INTRODUCTION The vertebrate skin provides a powerful model system to study cellular interactions which govern organ morphogenesis because of its distinct developmental pathways and the heterogeneity of its appendages. Furthermore, during integument development, epiderr 1997 WILEY-LISS, INC. Grant sponsor: ARC; Grant number: 6233; Grant sponsor: Fondation de la Recherche Médacale. *Correspondence to: D. Dhouailly, Institut Albert Bonniot, LEDAC UMR CNRS 5538-Biologie de la Différenciation Epithéliale, Domaine de la Merci, 38706 La Tronche Cedex, France. E-mail: danielle. firstname.lastname@example.org Received 18 February 1997; Accepted 1 August 1997 cHoxc-8, cHoxd-13 AND EPIDERMAL SPECIFICATION Krumlauf, 1992; Krumlauf, 1994). Within each cluster, genes most similar in sequence to a particular Drosophila gene, or paralogous genes, occupy the same relative positions within their respective cluster. In addition, the rostro-caudal boundary of one given Hox expression domain correlates with the position of the gene within its cluster, a property termed colinearity (Duboule and Dollé, 1989; Duboule and Morata, 1994). The same Hox sets have been shown to regulate the developmental processes within antero-posterior and proximo-distal specification in patterning of the vertebrae and limb segments (among others: Kessel and Gruss, 1991; Duboule, 1992). Previous results have shown that the cHoxc-6 and cHoxd-4 homeoproteins are differentially expressed during dorsal chick skin formation (Chuong et al., 1990, 1993). Likewise, in mouse, two different homeobox gene families, namely Otx and Hox, identify the facial and body skin territories, respectively, (Kanzler et al., 1994). In order to investigate the putative involvement of Hox genes in providing positional information for skin cells, and consequently in specification of chick featherversus scale-forming skin, we chose to study the expression of a median paralog, of order 8, which may be expressed at its highest level in the middorsal skin and that of the most distal one, of order 13, which similarly could be expressed in foot skin. A chick Hoxd-13 probe was provided to us by D. Duboule (Izpisùa-Belmonte et al., 1991). In order to obtain a chick Hox gene belonging to the 8th paralog, we screened a E8.5 dorsal skin cDNA library (Michaille et al., 1994) using a mouse Hoxc-8 probe obtained from H. Le Mouellic (Le Mouellic et al., 1992) and consequently isolated and characterized the chick Hoxc-8 cDNA. The developmental expression pattern of these two genes, analyzed by in situ hybridization, suggests that the cHoxc-8 and cHoxd-13 proteins play a role in the specification of dorsal and plantar skin morphogenesis. To further define the involvement of both genes, we carried out two complementary types of experiments. First, dermal–epidermal recombinants were performed between chick dorsal and plantar skin, at stages during which Hoxc-8 and Hoxd-13 are expressed in skin and which correspond to the initiation of appendage morphogenesis. To further test the morphogenetic abilities of the plantar dermis, some of the recombinants involved apteric midventral epidermis. Second, as it is well known that retinoic acid (RA) is both able to modulate homeobox gene expression (Simeone et al., 1991; Boncinelli et al., 1991) and to promote the ectopic formation of feathers on chick feet (Dhouailly et al., 1980) and scale-like structures in dorsal skin (Chuong et al., 1992), we repeated these two types of RA treatment in order to analyze the possible correlative changes in cHoxc-8 and cHoxd-13 expression. RESULTS Overview of Chick Tegument Morphogenesis In the chick, while most of the body is covered with feathers, the feet bear scales, and some areas, referred 275 to as apteria, remain bare. All the epidermis, with the exception of the median facial epidermis (Couly and Le Douarin, 1988), originates from the ectoderm, whereas the origin of the dermis varies according to the different regions. The back dermis is formed by E5 by the migration of somitic dermatomal cells (Mauger, 1972), whereas the limb dermis is formed at E8.5 by mesodermal cells which originate from the somatopleure. The first feather primordia (Fig. 1; stage f1 as referred by Michaille et al., 1994) appear at E7 along the midline of the lumbar region and consist of a circular epidermal thickening, or placode, which then becomes associated with an underlying dermal condensation (Sengel, 1976; Dhouailly, 1984). The epidermal and dermal cells then proliferate to form the feather bud (stage f3) by E8.5, which slants backward. The feather bud then elongates and invaginates at its base to form the feather filament (stage f7) by E14.5. Feather keratinization (a-keratins in the outer epidermal sheath and b-keratins in the feather itself) is first detectable at E12.5 at the apex of the filament and then proceeds downward (Haake et al., 1984). For a better comparison with the corresponding stages of feather morphogenesis, the three main stages of scuta and reticula formation were called stages 1, 3, and 7, respectively (Fig. 1). The first indication of scutate scale formation in the anterior tarsometatarsal region is the appearance of oval placodes (stage s1) during E9. The scuta rudiments are composed of an epidermal thickening covering a lightly marked dermal condensation. By E12, the placodes give rise to the scale buds (stage s3), then to the scutate scales (stage s7), which express a- and b-keratins by E16. Scutate scale b-keratins (Fig. 8, lane 5) involve three major polypeptides (21 to 18 kD), numbered b1–b3. The polypeptides 1 to 3 are also present but barely detectable in feather, whereas feather specific b-keratins (Fig. 8, lane 6) include a minor (b4) and four major (b5–8) 16- to 10-kD polypeptides. The first step of reticula formation (stage r1) occurs in the center of the plantar region by E11.5 and is characterized by an epidermal dome without any associated dermal condensation. Reticulate scale differentiation reaches stage r7 by E16. These structures express a-keratins (Fig. 8, lane 1) with the exception of the embryonic peridermal and subperidermal shedding layers which express a 25-kD b-keratin (Sawyer and Borg, 1979; Knapp et al., 1993). cHoxc-8 cDNA Cloning Three distinct positive clones were obtained by screening the chick skin cDNA library under high-stringency hybridization conditions with the mouse Hoxc-8 cDNA probe, and their sequences were determined. The longest cDNA clone (Fig. 2A) is 2,800 bp in length and encompasses the complete open reading frame of cHoxc-8, as revealed by sequence comparison with the known Hox genes belonging to the 8th paralog. The partial sequence of this cDNA (Fig. 2B) has been submitted to the EMBL/GenBank Data Libraries and 276 KANZLER ET AL. Fig. 1. Main stages of chick feather (A), anterior tarsometatarsal scale (scuta) (B), and plantar scale (reticula) morphogenesis (C) (see text). br, barb ridge; d, dermis; dc, dermal condensation; dp, dermal pulp; e, epidermis; ec epidermal collar; ep, epidermal placode; fb, feather bud; ff, feather filament; ies, inner epithelial sheath; oes, outer epithelial sheath; p, papilla; r, reticula; rb, reticulate scale bud; s, scuta; sb, scutate scale bud. received the accession number X94179. The predicted corresponding protein contains 242 amino acids and exhibits a Mr of 28,200 kD. Sequence comparison with the human (Boncinelli et al., 1989) and mouse (Le Mouellic et al., 1988) Hoxc-8 amino acid sequences reveal a complete interspecies conservation of the homeodomain; the chick sequence (Fig. 2B) is 100% identical with its human and mouse homologues. Two other regions of the predicted cHoxc-8 protein are strongly conserved when compared with the consensus sequence, namely a hexapeptide (MYPWMK) located five amino acids upstream of the homeodomain and a peptide located at the N-terminal end. The YPWM motif has been shown to be essential for establishing cooperative interactions between a subset of Hox proteins and Pbx proteins (Chang et al., 1995). The Nterminal region contains a high number of hydrophilic residues, particularly serine and threonine. As reported for the mouse amino-acid sequence, the carboxyterminal region comprises an acidic region that is particularly rich in glutamate residues (15 out of 21 amino acids). Such a region has been previously identified in the Hoxa-7 gene, where 15 glutamate codons preceed a stop codon (Colberg-Poley et al., 1985). Expression of the cHoxc-8 and cHoxd-13 Genes During Chick Skin Morphogenesis Using in situ hybridization, we studied the spatiotemporal distribution of cHoxc-8 and cHoxd-13 transcripts in 3.5-, 4.5-, and 6.5-day whole embryos, as well as in 8.5- to 10.5-day whole hindlimbs. The tissue distribution of the cHoc-8 and cHoxd-13 transcripts were studied on serial sagittal sections of 6.5- to 14.5-day embryos and of 9.5- to 16-day hindlimb, that is during dorsal and plantar skin morphogenesis. Antisense RNA probes, labeled either with digoxygenin-UTP or 35SCTP, were synthesized from the 38 cDNA flanking region of the cHoxc-8 homeobox, as well as from the similar region of the cHoxd-13 cDNA. Whole-mount in situ hybridizations performed at E3.5 and at E4.5 (Fig. 3A) show that cHoxc-8 transcripts are present in the neural tube, in the mesodermaly derived cells of somitic origin from the midthoracic level to the caudal extremity, as well as in the antero-proximal region of the wing bud, and more lightly of that of the hindlimb bud. This gene displays an anterior expression limit in the somites 23 and 24, a level which is posterior to that observed in the neural tube. These two somites later give rise to the fifth thoracic vertebra and to the overlying thoracic dermis. By 6.5 days of incubation (Figs. 3B, 4A), the chick skin (stage f0) has formed in the dorsal region and the cHoxc-8 transcripts are detectable at the same level in the vertebrae and the overlying mesenchymal, dermal, and epidermal cells. This expression reaches anteriorly to the fifth thoracic vertebra, thus appearing colinear with the limit observed earlier in the somites. The cHoxc-8 signal is significantly more intense in the dorsal skin region corresponding to its anterior expression boundary and decreases toward the caudal region. This dorsal antero-posterior pattern is conserved at E7.5 (stage f1), where transcripts are present in the dorsal feather primordia (Fig. 4B,C), both in the dermal condensation and the epidermal placodes. By E8.5, the epidermal and dermal signal remain uniformly distributed in the feather bud (stage f3) (Fig. 4E). This signal is also present in the mid-thoracic and at a less degree in the abdominal ventral skin, comprising the midventral apterium. At stage f7 (E14.5), the cHoxc-8 transcripts are no longer present in the interfollicular skin or in the feather follicle (Fig. 3G). No specific signal was detected with the corresponding sense probes on consecutive sections (Fig. 4D,F,H). No expression of cHoxc-8 was found with the antisense probe in the head and anterior thoracic skin regions at any of the studied stages. At E3.5, the cHoxd-13 transcripts were present in somites 39–40 at the distal end of the trunk (data not shown). They were never detected in either the dermis or the epidermis during trunk skin morphogenesis, a result which is confirmed by their presence in the lombar spinal cord and caudal vertebrae in the same sections. At E4.5 and E6.5 (Fig. 3C,D) the cHoxd-13 transcripts were detected in both the wing and hindlimb autopodes and then at E9.5 (Fig. 5A) in the distal region of the feet, including the digits. On sagittal sections, it appears that these transcripts display an antero-posterior to proximo-distal gradient, being par- Fig. 2. A: Organization of the cHoxc-8 cDNA clone. The isolated cDNA contains the complete cHoxc-8 coding region (hatched box; nt 439–1168). The black box demarcates the position of the homeodomain (HD; nt 883–1063). Nucleotide positions of several restriction sites are indicated. E, EcoRI; P, PstI; K, KpnI. B: Partial nucleotide sequence of the chick cHoxc-8 cDNA and predicted amino acid sequence of the coding region. The arrow points to the 38 end of the cDNA clone. Amino acid and nucleotide positions are indicated on the left and right sides, respectively. The sequence of the putative cHoxc-8 protein is indicated below the open reading frame. Amino acid changes between the chick and mouse protein sequences are underlined. Three consensus regions, i.e., the N-terminus, the conserved hexapeptide, as well as the homeodomain (HD) are boxed. The position of the intron as deduced from the mouse Hoxc-8 cDNA sequence is indicated by a black arrowhead. The splicing site (ACG/CT) fits well with the consensus sequence (Cech, 1983). These sequence data are available from EMBL/GenBank/DDBJ sequence database under the accession number X94179. 278 KANZLER ET AL. Fig. 3. Comparison of the distribution of cHoxc-8 and cHoxd-13 transcripts in 4.5- and 6.5-day chick embryos. Whole-mount in situ hybridization. In profile (A,C) and dorsal views (B,D) with rostral at the top. At E4.5, cHoxc-8 transcripts (A) are present in the somites from the caudal extremity to the mid-thoracic level (arrowhead) as well as in the anterior proximal part of the forelimb. Labeling is more intense in the somites preceding the anterior expression limit. At E6.5, cHoxc-8 transcripts (B) form in the skin a thoracic shield and a circumference at the insertion of the wing. The cHoxd-13 transcripts are present in the distal autopodial part of both limbs, shown here at E4.5 (C) and E6.5, as well as in the genitalia (not shown). Bars: 1 mm. ticularly abundant in the mesenchymal cells surrounding the bones and in the chondrocytes located at the extremities of the bones. From E10.5, cHoxd-13 expression is present in the mesenchymal cells which give rise to the plantar dermis. This signal reaches its highest expression level by E11.5 (Fig. 5B), the transcripts being significantly more abundant in the central area of the plantar region, precisely at the place where the first reticulae appear (stage r1). At E12.5 (Fig. 5C), the cHoxd-13 signal observed in the plantar dermis decreases markedly and cannot be detected afterward (Fig. 5D). No significant signal was detected in the epidermal cell layer at any of the stages examined. Because of a high death rate in 7-day RA-treated embryos, and in order to study the effect of RA treatment on cHoxc-8 expression dorsal skin morphogenesis, we turned to in vitro organotypic culture and repeated the experiments previously performed by Chuong and coworkers (1992). Dorsal skin explants from 7.5-day chick embryo were cultured in vitro for either 2 or 6 days with or without added RA and then grafted onto the chick chorioallantoic membrane (CAM) for 6 additional days. Control explants differentiated feathers normally. Stable morphogenetic alterations were obtained only with the longer period of RA treatment (6 days); thus the results reported below will concern only this experimental series. In 12% of cases, the explants exhibited no apparent modification of feather morphogenesis (Fig. 7A,B). In 20% of cases, abnormal feather filaments developed, characterized by a dilated spherical extremity (Fig. 7C,D), and in 68% of cases structures morphologically resembling scutate scales were obtained (Fig. 7E,F). The latter were predominant in the posterior region of the skin explants. In situ hybridization performed on longitudinal sections from explants after 6 days of in vitro culture with both antisense and sense cHoxc-8 probes revealed that the cHoxc-8 transcripts were no longer specifically detectable (data not shown). In order to discriminate between a single shape convergence and a real transformation from feather to scale or the reverse, the keratin content of both the RA-induced feather filaments and scale-like structures was subsequently analyzed and compared with that of hatchling feather, scutate, and reticulate scales (Fig. 8). The control keratin composition of hatchling scale (Fig. 8, lane 5) and feather (Fig. 8, lane 6) revealed characteristic, previously reported b-keratin profiles (Dhouailly et al., 1978; Sawyer, 1983), while reticulate scale epidermis contained mostly a-keratins about 65 to 67 kD, and a 25-kD (peridermal) b-keratin (Fig. 8, lane 1). The Effect of Retinoic Acid (RA) Treatment on cHoxc-8 and cHoxd-13 Expression and the Resulting Skin Phenotypes and Keratin Expression Chick embryos treated in ovo with RA at E11 showed by E17 enlarged and short feather filaments which look like scales in the cephalic, cervical, alar, and femoral tracts, and abnormal feather location on the reticulate scales from the center of the plantar region of the feet (Fig. 6A), as previously observed (Dhouailly et al., 1980). In situ hybridization was consequently performed with the cHoxd-13 probe on longitudinal frozen sections of feet of treated embryos at 11.5 and 12 days of incubation, thus 12 and 24 hr after the RA injection. This analysis revealed a lack of cHoxd-13 transcripts in the plantar dermal cells at both stages (Fig. 6B), whereas the controls showed a normal reticula morphogenesis (Fig. 6C) and cHoxd-13 plantar dermal expression (Fig. 6D). Conversely, chick embryos treated with RA at E10 never formed feathered reticulae (ectopic feathers only formed on the anterior face of the tarsometatarsus). In this case, the appearance of cHoxd-13 transcripts was just delayed by about 12 hr, and thus the transcripts were present by the time of reticulate morphogenesis. cHoxc-8, cHoxd-13 AND EPIDERMAL SPECIFICATION 279 Fig. 4. Distribution of cHoxc-8 transcripts during dorsal chick skin morphogenesis. In situ hybridization of sagittal sections with the antisense probe (A–C,E,G) and the sense probe (D,F,H). At E6.5 (stage fo) (A), transcripts are present in the vertebrae (vt) as well as in the overlying skin, both in the epidermis (e) and the dermis (d), in the midthoracic region. The arrowhead marks the anterior level. At E7.5 (stage f1) (B), the newly formed dorsal feather primordia (fp) express the cHoxc-8 gene. The transcripts are located in the epidermis (e), the placode (p), and the dermal condensation (dc). They display the same anterior expression boundary (arrowhead) (C), both in the skin (s) and the vertebrae, while the section hybridized with the sense probe (D) does not show any specific labelling. At E8.5 (stage f3), transcripts are present (E) in the feather bud (fb) epidermis, in the dermal condensation, and in the interappendage epidermis and superficial dermis. Compare with the sense control (F). At E14.5 (stage f7), the cHoxc-8 transcripts are no longer detectable in the skin (G). The labeling which is confined to the epidermal cells (arrowhead) surrounding the dermal papilla at the base of the feather filament (ff) is an artifact. Compare with the sense control (H). Darkfield illumination. Bars 5 210 µm (A,B), 3 mm (C,D), 150 µm (E–H). analysis of the protein content of scutate scale-like structures (Fig. 8, lane 4), obtained by treating E7 dorsal skin explants with RA, as well as of the ectopic feather filaments (Fig. 8, lane 3) formed on plantar skin of treated E11 embryos, showed a similar protein content: specific feather b4- to b8-keratins together with nonspecific b1- and b2-keratins as well as akeratins, making an embryonic feather-type keratin profile. It should be noted that, contrary to the feather keratin profile, the scutate scale keratin profile does not comprise any bands in the lower part of the gel. The presence of a minor b4-keratin and that of a-keratins, 280 KANZLER ET AL. the latter originating either from the interfollicular epidermis or the epidermal sheath of feather filaments, is usual when analyzing embryonic feathers or feathered skin explants. Analysis of Epidermal and Dermal Abilities Using Heterotopic Skin Recombinants (Table 1) When E7 dorsal epidermis was associated with either an E7 dorsal dermis (Fig. 9A) or an E10–E11 plantar dermis (Fig. 9B), it developed numerous long feather filaments, whereas it remained bare when it was recombined with a midventral apterium dermis (Fig. 9C; Sengel et al., 1969). Three types of differentiation were obtained with the E10–E11 plantar epidermis, depending on its associated dermis: reticulate scales with a plantar dermis (Fig. 9D), arrested buds distributed according to the feather hexagonal motif, plus a few hypomorphic feather on the edges of the explant, with a dorsal dermis (Fig. 9E), glabrous skin when recombined with a midventral apterium dermis (Fig. 9F). The reverse association of E10–11 plantar dermis and E10 midventral apterium epidermis resulted in the formation of numerous reticula-like dome-shaped buds, diplaying a tight distribution pattern with 2–3 days after grafting. However, after 8 days of culture, a few cases showed abnormal but short feather filaments, recognizable by their barb ridges (Fig. 9G). In the other cases (Fig. 9H), the epidermal top of the reticula-like structures contained b-keratins within 8 days as shown by immunofluorescent analysis (Fig. 9I,J). These were feather-type b-keratins, added to the 25-kD b peridermal and the a-keratins of the reticulate scales, as determined by electrophoretic analysis (Fig. 8, lane 2). In situ hybridization with both the cHoxc-8 and cHoxd-13 probes was performed 1, 12, 24, 36, and 48 hr after the recombination. No transcripts were detected for any of the dermal–epidermal recombinants. DISCUSSION We report the isolation and characterization of the chick homeobox gene cHoxc-8. The predicted sequence of the cHoxc-8 coding region analyzed is strikingly similar in sequence to the corresponding region of the mouse Hoxc-8 paralog (Le Mouellic et al., 1988). Particularly, the homeodomain appears identical with its mouse and human (Boncinelli et al., 1989) cognates. This evolutionary conservation probably extends to the developmental role of this gene during mouse and chick embryonic development, as its restricted expression pattern during early embryogenesis and the first stages of skin morphogenesis appear to be very similar in these two species. Indeed, the pattern of expression of the cHoxc-8 gene in the neural tube, somites, and Fig. 5. Distribution of cHoxd-13 transcripts during plantar chick skin morphogenesis. A: In situ hybridization of a longitudinal section of a E9.5 foot (stage r0). The transcripts are present in the perichondrial mesenchymal cells in the distal region of the foot and the digits, as well as in the plantar mesenchyme. B: Longitudinal foot section at the plantar level at E11.5 (stage r1): the cHoxd-13 transcripts are present in the perichondrial mesenchyme (pm) and in the plantar dermis (d). C: At E12, on a similar section, the dermal cHoxd-13 signal appears significantly decreased and is no longer detectable at E.16 (D). tmt, tarsometatarse; e, epidermis; r, reticula. Darkfield illumination, Bars 5 25 mm (A); 125 µm (B–D). cHoxc-8, cHoxd-13 AND EPIDERMAL SPECIFICATION Fig. 6. Effects of retinoic acid (RA) treatment on plantar skin phenotype and cHoxd-13 expression. A: Plantar view of a foot from a 17-day embryo, treated with RA at E11, showing ectopic feather filaments (f) that formed on the reticulate scales. Compare with the untreated control (C). B: In situ hybridization of a longitudinal section of a 11.5-day foot RA-treated at E11, showing the lack of cHoxd-13 expression in the plantar 281 dermis (d). Compare with the untreated control (D), in which the cHoxd-13 transcripts are present in the plantar dermis and in the perichondrial mesenchyme (pm), whereas the cHoxd-13 signal does not exceed the background level in the anterior skin region which forms scutate scales (ss). B,D: Darkfield illumination. Bars 5 1.3 mm (A,C); 280 µm (B,D). 282 KANZLER ET AL. Fig. 7. Effects of retinoic acid (RA) treatment on dorsal skin phenotype. Whole explants (A,C,E) and their corresponding sections (B,D,F). Dorsal RA-treated skin explant (dissected at E7.5) after 6 days of in vitro culture, followed by 6 days on the chick chorioallantoic membrane, displays either almost normal feather filaments (A,B), abnormal enlarged feather filaments in which the barb (b) and barbule cells (bl) are still recognizable (C,D), or even scutate scale-like structures (sl) in the caudal region and arrested very short feathers (af) in the anterior region of the explant (E,F). e, epidermis; d, dermis. D,E,F, Hematoxylin/Biebrich Scarlet; Bars 5 1.3 mm; (A–C) 120 µm (D–F). proximodistal region of the limb buds of the 3-day chick embryo is similar to the pattern previously reported for the mouse at a comparable stage of development (E 10.5) (Le Mouellic et al., 1988, 1992). Likewise, the pattern of expression of cHoxc8 in the skin at E7.5 corresponds with the overall expression of this gene along the cephalo-caudal and dorso-ventral axes previously reported in embryonic murine skin at a similar stage (Kanzler et al., 1994). The cHoxc-8 transcripts are present in both the epidermis and the dermis of the feather primordia and feather buds and disappear afterward. These results suggest that cHoxc-8 expression plays a role in the thoracic region during the initiation and first stages of feather embryonic morphogenesis. In contrast, no cHoxd-13 expression was found in the trunk skin over the period of morphogenesis in the different feather pterylae, including the caudal tract. The cHoxd-13 expression characterizes the ventral skin of both wing and leg autopode, and particularly the plantar dermis during skin morphogenesis. This is in concordance with the restricted pattern of this gene in the distal part of the limb, already shown by D. Duboule and coworkers (Dollé et al., 1991, 1993). In the mouse, the Hoxd-13 gene is expressed in a slightly more distal skin region, namely in the anterior and plantar dermis from the digits when skin differentiation takes place (unpublished results from Dr. J.P. Viallet in our group). It should be noted that in both species, the cHoxd-13 transcripts were only found in the dermis. Given their distinct expression pattern in skin, the cHoxc-8 and cHoxd-13 could be part of different homeoproteins sets, providing positional information specifying the dorsal thorax and autopodial ventral skin territories, respectively. Nevertheless, the difference between the ventral skin of the wing, which is feathered, and the plantar skin implies that the formation of reticulate scales must involve a further set of instructions which prevents the plantar epidermis from developing into feathers and which remains to be identified. Indeed, the results of heterotopic recombi- cHoxc-8, cHoxd-13 AND EPIDERMAL SPECIFICATION Fig. 8. Molecular identification of the protein content of chick cutaneous appendages formed in normal and experimental skin. (MW) Molecular weight; (lane 1) hatchling foot pad epidermis (reticulate scales); (lane 2) reticulate-like structures formed after 8 days of culture by recombinants composed of E11 plantar dermis and E10 apteric midventral epidermis (see Fig. 9,H,I,J); (lane 3) 18-day ectopic feathers formed on foot pad after RA-treatment at E11 (see Fig. 6A); (lane 4) scutate-like structures formed by E7 dorsal skin cultured for 6 days with added RA, and grafted for 6 more days on chick CAM (see Fig. 7,E,F); (lane 5) hatchling anterior tarsometatarsal epidermis (scutate scales); (lane 6) hatchling dorsal feathers. The a-keratins are designated as a-K, the b-keratins 1–8 as b-K; the histidine-rich proteins by an asterisk. 12% acrylamide gel stained with Coomassie blue. nants show that the plantar epidermis acquires a restricted ability to only differentiate reticulate scales on E11, which coincides with the expression of cHoxd-13 in the underlying dermis, while the dorsal epidermis, and to a lesser degree the midventral epidermis, which both express cHoxc-8, are endowed with featherforming ability, as shown by their ability to interpret inducing clues originating either from a dorsal or a plantar dermis to go over the feather differentiation program. The morphogenesis of such heterotopic dermal– epidermal recombinants is consistent with the hypothesis of an early regional specification of the epidermal abilities (Dhouailly and Sengel, 1983; Viallet and Dhouailly, 1994). The fact that we were unable to detect Hox gene expression in the dermal–epidermal recombinants, following the recombination, supports this hypothesis. The transcripts present in the dissected skin are likely to be destroyed by the enzymatic treatment of the skin, which is required to split it into its two components. They may just be some leftovers, without the need to be replaced. At the time of skin recombination, 7 days for the dorsal, 10 days for the midventral, and 11 days for the plantar skin, the Hox genes have already played their role in specifying epidermal abili- 283 ties. This may even occur earlier, at the time of skin formation, or even when the global pattern of hox gene expression within the body is established by E3. Phenotypic changes following RA treatment add indirect proof, which nevertheless supports our hypothesis. Retinoic acid treatment at E10 or E11 resulted in an inhibition of cHoxd-13 expression in the plantar dermis. Embryos treated on day 11 developed feather filaments on the reticulate scales. In contrast, in embryos treated on day 10, the cHoxd-13 expression reappeared by E11, followed by normal reticulate scale morphogenesis. The ectopic plantar feather filaments displayed a feather b-keratin electrophoretic profile, suggesting that this is a real homeotic transformation. A possible explanation is that retinoic acid acts by inhibiting the posterior paralogs and respecifying the plantar cells to a more proximal positional identity, thus allowing the formation of feathers. It cannot be excluded that RA may also act by enhancing the expression of more anterior paralogs. The RA-treated dorsal skin explants formed abnormal structures which can be interpreted as abnormal short and enlarged feather filaments, since the expressed set of b-keratins was of feather type. Similar malformed feathers were already obtained in the cephalic, alar, and femoral pterylae by in ovo RA treatment at E11 (Dhouailly et al., 1980). By the time of treatment the feathered specification of the dorsal skin might have already occurred. Furthermore, it should be noted that the RA effect on cHoxd-13 expression in embryonic skin is consistent with the known effects of RA on Hox genes expression in teratocarcinoma cells in vitro (Boncinelli et al., 1991; Simeone et al., 1991), during vertebrae differentiation (Kessel and Gruss, 1991), and during limb morphogenesis (Hayamizu and Bryant, 1994): The expression of Hox genes belonging to the median paralogs 4 to 8 is generally not modified by RA, whereas the 58 clustered genes are downregulated. Further experiments of mis-expression of both Hoxc8 and Hoxd13 are required to refine our hypothesis. In particular, the question still remains to know whether the thoracic skin is specified by the expression of cHoxc8 as well as other median paralog Hox genes, or by the absence of expression of posterior paralogs as cHoxd13. Likewise, is the autopodial plantar skin specified by the expression of cHoxd13 or the absence in a sufficient amount of homeoproteins of more anterior paralogs? Finally, bird scale morphogenesis must require, in addition to the proximodistal clues provided by the Hox genes, still-unknown information involved in the specification of the hindlimb versus the wing. EXPERIMENTAL PROCEDURES cDNA Cloning Approximately 1.5 3 106 recombinant phages of a LZapII (Stratagene) cDNA library prepared from E8.5 chick dorsal skin RNA (Michaille et al., 1994) were screened under high-stringency conditions (50% formamide; 53 standard saline citrate [SSC]; 1% standard 284 KANZLER ET AL. TABLE 1. Skin Morphogenesis in Chick Heterotopic Epidermal–Dermal Recombinants Cultured 6–8 Days on Chick Chorioallantoic Membrane Dermis Dorsal E7 Dorsal E7 (number of cases) Feather filaments (5) Plantar E11 Feather filaments (10) Midventral apterium E10 Glabrous skin (2) (and Sengel et al., 1969) saline sulfate [SDS]; 50 mM Tris-HCl pH 7.5; 0.1 mg · ml21 denatured salmon sperm DNA at 44°C). The probe, a gift from Dr. H. Le Mouellic, was a 840-bp SalI-EcoRI probe comprising the main part of the mouse Hoxc-8 homeodomain sequence and that extended to the 38 end (Le Mouellic et al., 1988). We followed the automatic excision protocol with helper phage and recircularization to generate the subclones containing the different chick cDNA inserts in pBluescripttII SK phagemid vector, as stated in the STRATAGENE Kit instructions. DNA Sequencing The nucleotide sequences were determined with the dideoxy chain-termination method, using (35S)-dATP and the Pharmacia T7-sequencingy Kit, according to the manufacturer’s instructions. The sequence of the cDNAs were read on both DNA strands. Embryos Warren breed fertile chick eggs were obtained from the ‘‘Centre d’aviculture de Cerveloup’’ (Moirans, France). Some Bresse breed eggs were obtained from the ‘‘Centre de Sélection de la Race Bressane’’ (Béchanne, France) as we have previously shown (unpublished data) that this scaled-feet breed is more prone than the Warren breed to develop feathers on their feet through RA treatment (70% instead of 10% of treated embryos), which facilitates the corresponding in situ hybridization analysis. The eggs were incubated at 38°C. For in situ hybridization, embryonic and postnatal samples were embedded in Tissue Teckt medium and frozen. In Situ Hybridization A 600-bp XhoI-PstI fragment of the cHoxd-13 cDNA (a gift from Dr. D. Duboule) containing the homeodomain region and the following 38 sequences and a 1,500 bp KpnI-EcoRI fragment of the isolated chick Hoxc-8 cDNA clone were used as templates for synthesis of either (35S)-CTP or digoxygenin-11-UTP–labeled riboprobes for in situ hybridization on frozen sections or whole-mount embryos, respectively. In vitro transcription reactions were performed using either T7 or T3 Epidermis Plantar E10–11 (number of cases) Arrested buds (18) Arrested buds plus hypomorphic feathers (18) Reticulate scales (5) Glabrous skin (4) Midventral apterium E10 (number of cases) Feather filaments (Sengel et al., 1969) Abnormal reticulae (12) Abnormal feathers and reticulae (11) Glabrous skin (2) RNA polymerase as directed by the manufacturer (Boehringer Mannheim Biochemicals). Whole-mount in situ hybridization protocol was based on the Conlon and Rossant (1992) procedure with minor modifications. Following fixation, bleaching, and proteinase K treatment, either 3.5-, 4.5-, or 6.5-day embryos as well as 9.5-day feet were hybridized overnight at 70°C with 1 µg · ml21 of probe. After several washes under stringent conditions and RNase A and T1 treatment, embryos and feet were incubated with antidigoxigenin Fab conjugated to alkaline phosphatase (Boehringer Mannheim). Staining was allowed to proceed at room temperature by the addition of alkaline phosphatase substrates NBT/BCIP. In situ hybridization on frozen sections was carried out essentially as follows. Serial cryostat sections of 6–10 µm thickness were treated successively with acetone, 4% formaldehyde at 4°C, 0.1 M triethanolamine/0.25% acetic anhydride, 50% formamide/13 SSC at 60°C, and two ethanol washes. The hybridization buffer included 50% formamide, 10 mM DTT, 500 µg/ml tRNA, 13 saltsDenhardts, 10% dextran sulfate, and heat-denatured, (35S)-labeled antisense riboprobe (specific activity of 5 3 108 c.p.m.21 · mg21; final concentration: 2 · 104 c.p.m./µl) or the corresponding opposite control probe. After overnight hybridization at 54°C, slides were washed under stringent conditions and treated with RNase A and T1. After ethanol dehydration, slides were dipped in Kodak NTB-2 nuclear track emulsion and exposed for about 3 weeks before developing. Sections were stained with propidium iodide, mounted in Surgipatht, and then analyzed with an AX70 Olympus microscope using both darkfield and fluorescence illuminations. In Vivo and In Vitro Retinoic Acid Treatment Retinoic acid (125 µg of all-trans retinoic acid, a gift from Hoffmann-La Roche, Basel, Switzerland), previously dissolved in absolute ethanol (50 µl) was injected into the amniotic cavity of 10- or 11-day chick embryo (Bresse breed). Preliminary experiments (Dhouailly et al., 1980) showed that 125 µg was the most suitable dose to obtain a high percentage of both surviving Fig. 9. Homotopic and heterotopic skin recombinants after 8 days of culture on the chick chorioallantoic membrane. A–C: Recombinants involving E7 dorsal thoracic epidermis, associated with a dermis from (A) E7 dorsal thorax, (B) E11 plantar foot pad, (C) E10 midventral apterium: formation of feather filaments when the dermis originates from an appendage-forming region. D–F: Recombinants involving E11 plantar epidermis, associated with a dermis from (D) E11 plantar foot pad, (E) E7 dorsal thorax, (F) E10 midventral apterium: formation of reticulate scales with a plantar dermis, of reticulate abnormal structures distributed according to the feather hexagonal motif, of a few hypomorphic feathers with a dorsal dermis, and of nude skin when the dermis originates from an apteric region. G–J: Recombinants involving an E10 apteric midventral epidermis associated with an E11 plantar dermis. Formation of reticulate structures (G,H) and in a few cases (G) of a few hypomorphic feather filaments recognizable by their barb ridges (arrow). Even in most of the cases which only differentiate reticulate structures as in (H), the immunofluorescent staining with specific polyclonal antibodies to b- (I) and a- (J) keratin polypeptides shows that the epidermis located at the top of the reticulate-like structures (arrows) elaborates both a- and b-type keratins. Bars 5 0.5 mm (A–H); 0.2 mm (I,J). 286 KANZLER ET AL. embryos and morphogenetic alterations. Feet from control and treated embryos were collected 12 and 24 hr after injection for in situ hybridization, as well as 6 days later to study the resulting skin phenotype and to determine the keratin composition of the appendages by SDS-polyacrylamide gel electrophoresis (PAGE). Dorsal skin tissues from 7.5-day chick embryo (Warren breed) were microdissected from the level of the wing to the caudal extremity. Cultures were then set up by placing the explants onto a grid in a Falcon dish with DMEM (Dulbecco’s modified Eagle medium) supplemented with 20% fetal calf serum and either all-trans retinoic acid previously dissolved in ethanol (final concentration: 5 µg/ml) or an equivalent volume of ethanol alone. The medium was changed every 2 days. After 2 or 6 days of in vitro culture, the explants were grafted onto the chick chorioallantoic membrane (CAM) for 6 more days. Some of the control and RA-treated grafts were conserved at 220°C for further protein content analysis, and some of them were embedded in Tissue Teckt medium for cryostat sectioning and in situ hybridization analysis. Heterotopic Dermal–Epidermal Recombinations Midthoracic dorsal, ventral, and plantar skin fragments were dissected from 7-, 10-, and 11-day Warren chick embryos, respectively, in Ca21- and Mg21-free Earle’s saline. Dermis and epidermis were separated in Tyrode’s solution after incubation in 2% trypsin and 1% pancreatin in Earle’s saline for 10 min at 4°C; protease digestion was stopped in 50% fetal calf serum. Tissues were reassociated in heterotopic recombinations (referred in Table 1) on a semisolid nutritive agar and placed for 30 min at 37°C to obtain sufficient mutual adhesion between dermis and epidermis. Recombinants were transferred to the CAM of 10-day chick embryos, cultured for 6 to 8 days, photographed, and then some of them were processed for electrophoretic keratin analysis or immunofluorescent staining with a polyclonal rabbit antibody which recognizes all bkeratins (Dhouailly and Sawyer, 1984). Keratin Analysis Control skin fragments were dissected from the tarsometatarsus, central foot pad, and feather plucked from the back of a 21-day hatchling chicken. Skin recombinants and RA-treated dorsal explants were retrieved from the CAM 7 or 8 days after grafting. Ectopic feather filaments formed after RA-treatment on the reticula were plucked at E19. For gel electrophoresis, keratin polypeptides from control and experimental specimens were isolated and S-carboxymethylated according to the procedure of Dhouailly et al. (1978). Then one-dimensional SDS-PAGE (12% acrylamide) was performed according to Laemmli (1970). ACKNOWLEDGMENTS We are grateful to Dr. D. Duboule for the gift of the chick Hoxd-13 probe and to Dr. H. Le Mouellic for the mouse Hoxc-8 probe, as well as for their advice, and to Dr. R.H. Sawyer, in the laboratory of whom D. Dhouailly became interested in chick reticulate morphogenesis. We thank Dr. J.P. Viallet for his advice in performing the whole mounts, Mr. Ray Dunn and Mrs. G. Chevalier for their help in keratin analysis, and Mrs. B. Peyrusse for the illustrations. We are indebted to Dr. Bernigaud for providing the Bresse breed eggs and to Dr. Jennifer Seed for critical reading of the manuscript. This work was supported by ARC grant 6233 and a ‘‘Fondation de la Recherche Médicale’’ grant to D. Dhouailly. REFERENCES Boncinelli E, Acampora D, Pannese M, D’Esposito M, Somma R, Gaudino G, Stornaiuolo A, Cafiero M, Faiella A, Simeone A. Organization of human class I homeobox genes. Genome 1989;31:745–756. Boncinelli E, Simeone A, Acampora D, Mavilio F. HOX gene activation by retinoic acid. TIG 1991;7:329–334. Cadi R, Dhouailly D, Sengel P. Use of retinoic acid for the analysis of dermal-epidermal interactions in the tarsometatarsal skin of the chick embryo. Dev Biol 1983;100:489–495. Cech TR. RNA splicing: three themes with variations. Cell 1983;34:713– 716. Chang C-P, Shen W-F, Rozenfeld S, Lawrence HJ, Largman C, Cleary ML. Pbx proteins display hexapeptide-dependent cooperative DNA binding with a subset of Hox proteins. Genes Dev 1995;9:663–674. Chuong CM, Olivier G, Ting SA, Jegalian BG, Chen HM, De Robertis EM. Gradients of homeoproteins in developing feather buds. Development 1990;110:1021–1030. Chuong CM, Ting SA, Widelitz RB, Lee YS. Mechanism of skin morphogenesis. II. Retinoic acid modulates axis orientation and phenotypes of skin appendages. Development 1992;115:839–852. Chuong CM, Randall B, Widelitz B, Ting-Xin J. Adhesion molecules and homeoproteins in the phenotypic determination of skin appendages. J Invest Dermatol 1993;101:10S–14S. Colberg-Poley AM, Voss SD, Chowdhury K, Gruss P. Structural analysis of murine genes containing homeobox sequences and their expression in embryonal carcinoma cells. Nature 1985;314:713–718. Conlon RA, Rossant J. Exogenous retinoic acid rapidly induces anterior ectopic expression of murine Hox-2 genes in vivo. Development 1992;116:357–368. Couly G, Le Douarin N. The fat map of the cephalic neural primordium at the presomitic to the 3-somite stage in the avian embryo. Development 1988;103s:101–113. Dhouailly D. Dermo-epidermal interactions during morphogenesis of cutaneous appendages in amniotes. Front Matrix Biol 1977;4:86– 121. Dhouailly D. Specification of feather and scale patterns. In: Malincinski GM, Bryant SV, eds. Pattern Formation. New York: Macmillan Pub. C., 1984:581–601. Dhouailly D, Sengel P. Feather forming properties of the foot integument in avian embryos. In: Sawyer RH, Fallon JF, eds. EpithelialMesenchymal Interactions in Development. New York: Praeger Press, 1983:147–161. Dhouailly D, Sawyer RH. Avian scale development: XI. Initial appearance of the dermal defect in Scaleless skin. Dev Biol 1984;105:343– 350. Dhouailly D, Rogers GE, Sengel P. The specification of feather and scale protein synthesis in dermal-epidermal recombinations. Dev Biol 1978;65:58–68. Dhouailly D, Hardy MH, Sengel P. Formation of feathers on chick foot scales: a stage-dependant morphogenetic response to retinoic acid. J Embryol Exp Morphol 1980;58:63–78. Dollé P, Izpisüa-Belmonte JC, Boncinelli E, Duboule D. The Hox4–8 gene is localised at the extremity of the Hox-4 complex and is expressed in the most posterior parts of the body during development. Mech Dev 1991;36:3–14. cHoxc-8, cHoxd-13 AND EPIDERMAL SPECIFICATION Dollé P, Dierich A, LeMeur M, Schimmang T, Schuhbaur B, Chambon P, Duboule D. Disruption of the Hoxd-13 gene induces localized heterochrony leading to mice with neotenic limbs. Cell 1993;75:431– 444. Duboule D. The vertebrate limb: a model system to study the Hox/HOM gene network during development and evolution. Bioassays 1992;14:375–383. Duboule D, Dollé P. The structural and functional organization of the murine HOX gene family resembles that of Drosophila homeotic genes. EMBO J 1989;8:1497–1505. Duboule D, Morata G. Colinearity and functional hierarchy among genes of the homeotic complexes. TIG 1994;10:358–364. Gregg K, Wilton SD, Parry DAD, Rogers GE. A comparison of genomic coding sequences for feather and scale keratins: structural and evolutionary implications. EMBO J 1984;3:175–178. Haake AR, König G, Sawyer RH. Avian feather development: relationships between morphogenesis and keratinization. Dev Biol 1984;106: 406–413. Hayamizu TF, Bryant SV. Reciprocal changes in Hoxd-13 and RAR-b2 expression in response to retinoic acid in chick limb buds. Dev Biol 1994;166:123–132. Izpisùa-Belmonte J-C, Tickle C, Dollé P, Wolpert L, Duboule D. Expression of the homeobox Hox-4 genes and the specification of position in chick wing development. Nature 1991;350:585–589. Kanzler B, Viallet JP, Le Mouellic H, Boncinelli E, Duboule D, Dhouailly D. Differential expression of two different homeobox gene families during mouse tegument morphogenesis. Int J Dev Biol 1994;38:633–640. Kemp DJ, Rogers GE. Differentiation of avian keratinocytes. Characterization and relationship of the keratin proteins of the adult and embryonic feathers and scales. Biochemestry 1972;11:969–975. Kessel M, Gruss P. Homeotic transformations of murine vertebrae and concomitant alteration of Hox codes induced by retinoic acid. Cell 1991;67:89–104. Knapp LW, Shames RB, Barnes GL, Sawyer RH. Region-specific patterns of beta keratin expression during avian skin development. Dev Dyn 1993;196:283–290. Krumlauf R. Hox genes in vertebrate development. Cell 1994;78:191– 201. 287 Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970;264:377–379. Le Mouellic H, Condamine H, Brûlet P. Pattern of transcription of the homeogene Hox-3.1 in the mouse embryo. Genes Dev 1988;2:125– 135. Le Mouellic H, Lallemand Y, Brûlet P. Homeosis in the mouse induced by a null mutation in the Hox-3.1 gene. Cell 1992;69:251–264. MacGinnis W, Krumlauf R. Homeobox genes and axial patterning. Cell 1992;68:283–302. Mauger A. Rôle du mésoderme somitique dans le développement du plumage dorsal chez l’embryon de poulet. II Régionalisation du mésoderme plumigène. J Embryol Exp Morphol 1972;28:343–366. Michaille JJ, Blanchet S, Kanzler B, Garnier JM, Dhouailly D. Characterization of cDNAs encoding the chick retinoic acid receptor g2 and preferential distribution of retinoic acid receptor g transcripts during chick skin development. Dev Dyn 1994;201:334–343. Moll R, Franke WW, Shiller DL. The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell 1982;31:11–24. Sawyer RH. The role of epithelial-mesenchymal interactions in regulating gene expression during avian scale morphogenesis. In: Sawyer RH, Fallon JF, eds. Epithelial-Mesenchymal Interactions in Development. New York: Praeger Press, 1983:115–146. Sawyer RH, Borg TK. Avian scale development. VI. Ultrastructure of the keratinizing cells of reticulate scales. J Morphol 1979;161:111– 122. Sengel P. Morphogenesis of skin. In Abercrombie M, Newth DR, Torrey JG, eds. Developmental and Cell Biology Series. Cambridge: Cambridge University Press, 1976:1–277. Sengel P, Dhouailly D, Kieny M. Aptitude des constituants cutanés de l’aptérie médio-ventrale du poulet à former des plumes. Dev Biol 1969;19:436–446. Simeone A, Acampora D, Nigro V, Faiella A, D’Esposito M, Stornaiuolo A, Mavilio F, Boncinelli E. Differential regulation by retinoic acid of the homeobox genes of the four HOX loci in human embryonal carcinoma cells. Mech Dev 1991;33:215–227. Viallet JP, Dhouailly D. Retinoic acid and mouse skin morphogenesis: II. Role of epidermal competence in hair glandular metaplasia. Dev Biol 1994;166:277–288.