The relationship between cell proliferation and the transcription of the nuclear oncogenes c-myc, c-myb and c-ets-1 during feather morphogenesis in the chick embryo

c-myc, c-myb and c-ets-1 belong to the class of nuclear proto-oncogenes (Alitalo et al. 1983; Klempnauer et al. 1984; Pognonec et al. 1989). They encode proteins that bind to DNA-specific sequences in vitro (Biedenkapp et al. 1988; Wasylyk et al. 1990). The c-myc protein may act as a positive transcriptional regulator (Onclerq et al. 1989) and a role in DNA replication has been postulated, although not demonstrated conclusively (Gutierrez et al. 1988; Studzinki et al. 1988). Several laboratories have reported that c-myb and v-myb proteins can transactivate various reporter genes (Klempnauer et al. 1989; Weston and Bishop, 1989). Furthermore a cellular gene, miml, which contains three closely spaced binding sites for the v-myb protein in its promoter region and which is strongly activated by v-myb in a cotransfection assay has been recently identified (Ness et al. 1989). The products of the c-ets-1 proto-oncogene are also transcription factors (Wasylyk et al. 1990; Gunther et al. 1990). They complement, or cooperate with, the c-fos and c-jun proteins for activation of transcription from the oncogene responsive domain of the polyoma virus enhancer (Wasylyk et al. 1990).

The role of c-myc, c-myb and c-ets-1 in cellular proliferation and differentiation has been extensively studied in vitro. The expression of c-myc is transiently induced within a few minutes after the mitogenic stimulation of T lymphocytes or quiescent fibroblasts (Kelly et al. 1983). Unlike c-myc, c-myb is not involved in the immediate response to mitogenic stimuli, c-myb mRNA reaches a maximum 4-8 h after serum stimulation of quiescent fibroblasts (Thompson et al. 1986).

In normal human lymphocytes stimulated with the mitogenic lectin phytohemagglutinin the levels of c-myb mRNA increase within 14 h after stimulation (Reed et al. 1986). In contrast, in a T cell hybridoma the induction of c-ets-1 transcription upon antigenic stimulation is associated with the functional activation of the hybridoma rather than proliferation (Kaufmann et al. 1987). In some cell types, it appears that the increased expression of c-myc and c-myb as well is not linked to the control of growth (Dotto et al. 1986; Coreos et al. 1987). Thus c-myc mRNA levels increase during the differentiation of lens cells as they withdraw from the cell cycle (Nath et al. 1987). In neuroblastoma cells, c-myb expression decreases only during induced maturation and not after growth arrest treatments (Thiele et al. 1988). The transcription of these genes can also be induced by antimitogenic agents (Bravo et al. 1985; Dixit et al. 1989).

During embryonic development the linkage between the expression of c-myc and cellular proliferation holds for only a restricted set of cells. The extraembryonic tissues (Pfeifer-Ohlsson et al. 1984) exhibit a very high level of c-myc expression. C-myc is actively expressed in the liver of both mouse and chick embryos (Schmid et al. 1989; Jaffredo et al. 1989) and in proliferating epithelial cell layers of human embryos (Pfeifer-Ohlsson et al. 1985). However, c-myc is also expressed in the postmitotic neurones of the dorsal root ganglia (Jaffredo et al. 1989). We have recently described the expression of c-myb and c-ets-1 transcripts in the chick embryo (Vandenbunder et al. 1989). c-ets-1 transcription is transiently observed in groups of mesodermal cells engaged in morphogenetic processes and appears excluded from all epithelia. Throughout the development c-ets-1 is highly expressed in endothelial cells at the onset of the formation of the blood vessels. Before E8 the expression of c-myb is most tightly linked to hemopoietic cells.

Although the overall expression patterns of these genes in the embryo have been described, our understanding of their functions remains limited. As a first step towards deciphering a functional role of these nuclear oncogenes in the control of cell proliferation, a direct comparison between their expression patterns and the distribution of cycling cells is necessary. Several features of skin morphogenesis (see Fig. 1 for a detailed description) make this system attractive for such studies. It is one of the best studied cases of secondary induction involving the interaction of the epidermis with the subjacent dermis (Sengel, 1986); the dermis induces the formation of epidermal placodes and specifies their shape and arrangement while the epidermis induces the dermal cells to colonize underlying areas. The organized dermis induces the transformation of the placodes into the appendage, the cephalocaudal orientation of which is epidermis-dependent (Sengel, 1958; Dhouailly, 1984).

Little is known about the role of cell proliferation in feather germ morphogenesis. Colchicine treatments do not prevent dermal condensation beneath the placodes (Stuart et al. 1972) suggesting that cell condensation arises from migration rather than from a local increase of cell proliferation. According to Wessells (1965), [³H]thymidine incorporation is randomly distributed in both the ectoderm and the dermis prior to germ formation. As local dermal condensations form, DNA replicative activities cease in the aggregated cells for a period of about 20–30 h. Epidermal placodes at this stage also fail to incorporate DNA precursors. When feather outgrowth begins, large numbers of incorporating or dividing nuclei are seen in both tissues and the rate of profiferation remains high during further outgrowth. However, altogether these results do not provide a complete description of the spatial pattern and the temporal sequence in which cells proliferate during feather development; this prompted us to reinvestigate this point.

We present here the first detailed comparison between the expression patterns of c-myc, c-myb and c-ets-1 and the distribution of S-phase cells during the development of a tissue in the chick embryo. We show that the expression patterns of these nuclear oncogenes are not strictly associated with proliferating cells throughout the developmental program of the skin.

Dorsolumbar feather tracts were selected according to their peculiar program of development. The first mediodorsal row of feather rudiments appears at E6. Longitudinal rows arise sequentially in a continuous process on either sides of this initial row. Thus development proceeds in a dorsal to ventral gradient on the back of the embryo and all the early steps of feather morphogenesis can be seen on a single embryo. Two symmetrical rectangular pieces of skin were peeled off the back of embryos on either side of the cephalocaudal axis from late E5 to E10 (stages 28 to 36 according to Hamburger and Hamilton, 1951).

Freshly excised pieces of skin were cultured according to Trowell (1954), with some modifications. Briefly, expiants were transferred to filtration membranes (1.22 μm pore size, Millipore) carried by grids of stainless steel placed in 35 mm Petri dishes. The level of the medium was adjusted so as to soak the filters. The medium used was Ham’s F12 supplemented with 10% fetal calf serum (Gibco), penicillin (50i.u. ml⁻¹), streptomycin (50 μg ml⁻¹). Culture dishes were incubated at 37°C in a humidified atmosphere of 5% CO₂. Under these conditions normal feather development can proceed in vitro within a few days. In order to detect S-phase cells, expiants were exposed for 3h to 50 μM 5-bromo-2-deoxyuridine (BrdU) before being fixed.

Whole-mount specimens of skin epidermis were prepared according to Tanaka and Kato (1983). Strips of skin at E7 or E8 were spread over the Millipore filter with the epidermal side facing down, cultured for 3h and fixed for 15 min in Bouin’s fluid. After extensive washes in PBS, expiants were immersed in a 0.2 % solution of EDTA in Ca²⁺- and Mg²⁺-free PBS for 4 to 16 h at 37°C. The dermis was peeled off with forceps and discarded. The epidermal layer was then freed from the filter, fixed again for 2 h and processed in toto for the detection of BrdU-labelled nuclei as described in the following section. In all other cases, cultured strips of skin were fixed in Bouin’s fluid, washed, embedded in paraffin and serially sectioned at 7 μm either longitudinally, parallel to the cephalocaudal and dorsoventral axis of the embryo or tangentially to the skin plane.

Peeled epidermis or deparaffinized sections were rehydrated and incubated in 1.5 N HC1 for 20 min at room temperature for denaturing the DNA. After 2 washes in PBS, epidermal layers or sections were incubated for 1 h at 37°C in a 1/100 dilution of an anti-BrdU-DNA monoclonal antibody (Partec) in a PBS (pH 7.4) buffer containing 0.5% Tween 20 and 0.5 % bovine serum albumine (Sigma). Specimens were washed in PBS, bound antiserum was revealed by incubation for one hour with the fluorescein-conjugated rabbit anti-mouse IgG diluted 1/100. Epidermis or sections were then washed in PBS, stained with Evans blue (1/10000) and mounted in glycerol/ PBS. Controls were performed where the primary antibody was replaced by PBS. Epidermal whole mounts or midsagittal sections of feather rudiments were examined with an Olympus BH2 epifluorescent photomicroscope.

The c-myc protein was revealed with the steptavidin–biotin system as described previously (Jaffredo et al. 1989).

The c-myc probe derives from a 520 bp EcoRI fragment from genomic DNA (Saule et al. 1984), which contains 205 bp from the third exon of c-myc and the 3′ non coding sequences. Since, in both human and mouse, the three members of the myc family share regions with amino acid homology in this third exon (Alt et al. 1986), it could be argued that the 520 bp EcoRI probe also detects N-myc or L-myc transcripts. On northern blots with total mRNA from the brain, liver, intestine and heart of a 6-day embryo, this probe hybridizes with single 2.6 kb RNA species. Thus the 520bp EcoRI probe does not detect L-myc transcripts whose size is 4 kb in mouse (Zimmerman et al. 1986) and 3.8 kb in human (Alt et al. 1986). According to Sawai et al. (1990) a probe containing most of the chicken N-myc third exon does not hybridize under stringent conditions with the corresponding part of the c-myc gene. Thus the amino acid homologies between the different members of the myc family are not sufficient to produce cross hybridization of either gene to the other. Probes containing the third exon of the N-myc gene or most of the coding sequences of c-myc have been found to detect respectively N-myc and c-myc transcripts specifically in mouse embryos (Mugrauer et al. 1988; Downs et al. 1989).

The c-myb probe derives from a 350 bp EcoRI-Sa/I fragment of the E26 provirus (Leprince et al. 1983). We have also used a nearly full-length (3.5 kb) chicken c-myb cDNA in the plasmid pSG-3 (Gerondakis and Bishop, 1986) obtained from Joseph Lipsick (Stony Brook, NY).

The probe specific for c-ets-1 transcripts is a 750 bp Bg/H-Windlll fragment obtained from the same molecularly cloned E26 provirus (Leprince et al. 1983).

In situ hybridization was performed essentially by the method of Cox et al. (1984) with RNA probes labelled with tritium or ³⁵S for the 3.5 kb c-myb probe. Details of the protocol are given elsewhere (Vandenbunder et al. 1989). Control hybridizations with sense probes gave no signal above background. The slides were exposed at 4°C prior to development. The slides were stained using the Hoechst dye 33258 (bisbenzimi-dine, Img in 250ml PBS) and mounted with DAKO-glycergel (Sebia). The sections were observed and photographed under double illumination on an Olympus BH2 photomicroscope with epifluorescence for Hoechst staining and with a dark-field condenser for silver grain detection. Identical results were obtained from independent experiments performed on skin specimens peeled off the back of several embryos.

Prior to feather rudiment formation, the chick embryo epidermis comprises flattened cells forming the peridermal layer and tightly associated columnar cells forming the basal layer. The densification phase of the initially loose dermal stem cells leads to the formation of a uniform thick layer of dermal cells clearly individualized between the epidermis and the subjacent loose and vascularized hypoderm. At the onset of the densification phase, some BrdU-incorporating nuclei were scattered both in the epidermal cells and in the loose dermis (Fig. 2A). During the process of accumulation, the number of BrdU-incorporating nuclei greatly increased. But the labelled cells were randomly distributed in the dermis and the whole epidermis including both the peridermal and basal layers (Fig. 3A). Tangential sections of dermis and epidermal whole mounts corroborated this observation (Fig. 4A,B).

The c-myc and the c-myb probe used in this study detected no transcripts over the period of this first increase in dermal cell density. Neither the epidermis nor the dermis displayed a signal higher than the background (Figs 2B, 3B, 2C, 3C). Similarly no peculiar pattern of c-myc protein expression was detected with a specific antiserum until the beginning of feather formation (Fig. 12A,B). By contrast the c-ets-1 probe detected a strong signal both on loose dermal cells underlying the epidermis and more deeply on clusters of endothelial cells in the hypoderm. The epidermis remained unlabelled (Fig. 2D). As the dermis had completed its reorganization, a strong signal persisted within the whole dermis and especially around blood vessels at the level of the hypoderm (Fig. 3D).

When dermal cell aggregates were seen under epidermal placodes, the skin displayed a peculiar pattern of BrdU incorporation. Midsagittal sections of feather germs showed that cells within the core of dermal condensations failed to incorporate the thymidine analogue, whereas both peripheral cells and the sparse interplumar dermal cells displayed a conspicuous labelling (Figs 5A, 6A). It is noteworthy that basal cells within epidermal placodes were never labelled, whereas incorporating nuclei could be seen within the peridermal layer (Fig. 6A). Tangential sections of dermis (Fig. 4A), as well as epidermal whole mounts (Fig. 4B), showed unlabelled circular areas surrounded by numerous incorporating nuclei corresponding to dermal condensations and epidermal placodes, respectively. At this stage c-myc transcripts were detected in the epidermal placodes and a low level of c-myc expression appeared at the periphery of the dermal condensations (Figs5B, 6B). We did not detect c-myb transcripts at this stage (Figs 5C, 6C). As placodes and subjacent condensations formed, the ets-1 signal decreased in aggregated dermal cells (Figs 5D, 6D).

When the outgrowth of feather buds began, numerous incorporating nuclei were seen in the previously unlabelled dermal condensations beneath unlabelled placodes. Bud growth rapidly became asymmetrical and, concomitantly, the number of incorporating nuclei increased in the caudal part of each bud: the bud dermis and the edge of the epidermal placode contained more numerous labelled nuclei than did the rostral part of the growing bud and the interplumar spaces (Fig. 7A). Tangential sections in the dermis showed the accumulation of incorporating nuclei within the caudal half of condensations and the low level of labelling within the previously highly proliferative interplumar spaces (Fig. 8A). Whole mounts of epidermis showed fluorescent crescents delineating the caudal edge of each slanting bud and high levels of labelling in interplumar spaces contrasting with the unlabelled dorsal part of the bud (Fig. 8B).

At this stage, c-myc mRNA was highly expressed both in the epidermal sheath, and in the apical and the central part of the dermal condensation. The distribution of the c-myc protein is similar to the distribution of the c-myc mRNA (Fig. 12C,D). The transcription pattern of c-myc (Fig. 7B) also appeared asymmetrical since the dermal cells at the caudal edge displayed a lower hybridization signal than at the rostral edge. Thus, at this stage, the pattern of c-myc transcription was complementary to the pattern of BrdU incorporation. However, the c-myb probe, which so far had not detected any transcripts, gave a conspicuous signal on the apex of the epidermal sheath, whereas dermal cells displayed a faint signal (Fig. 7C). The transcription of c-ets-1 remained very high in the endothelial cells of the capillaries underlying the dermis (Fig. 7D).

As the bud clearly emerged from the back skin, midsagittal sections showed a general increase of BrdU incorporation both in the dermis and the epidermis (Fig. 9A). Labelled nuclei were particularly numerous in the cephalic part of the epidermal covering whereas the thick caudal edge displayed unlabelled cells. A high proportion of underlying dermal cells incorporated the thymidine analogue.

At this stage, the c-myb probe gave a strong hybridization signal on the epidermal layer, whereas in the dermis the signal was much lower (Fig. 9C). The intensity of this c-myb hybridization signal clearly decreased in the interplumar epidermis. Both the c-myc and the c-ets-1 expression patterns were similar to those described at the onset of feather bud outgrowth (Fig. 9B,D).

As the bud elongated and slanted backwards (Fig. 10A), labelled nuclei were essentially seen in the epidermis of the protruding feather germ. Interplumar epidermis failed to incorporate BrdU, whilst few labelled dermal nuclei were randomly scattered in the core of the bud. At the follicle stage, the feather filaments showed cylinders of epidermis which developed prominent barb ridges where incorporating nuclei were abundant. The inner face of papillar ectoderm entering into the dermis showed sparsely labelled nuclei (Fig. 11 A). A low level of labelling was observed within the dermal pulp of the feather filament and at the root of the feather.

At these stages, both c-myc and c-myb transcripts were detected on the bud epidermis (Fig. 10B,C). TTie interplumar epidermis contained no transcripts. As the feather follicles formed, an obvious signal was observed with the c-myc and the c-myb probes both on epidermal folds at the base of the feather filaments and on the barb ridge-forming epidermal layer (Fig. 11B,C). Over the dermis, the signal with the c-myc and c-myb probes was not significantly higher than the background (Figs 10B,C, 11B,C).

The pattern of c-ets-1 expression remained clearly linked to dermal cells with an obvious higher level within endothelial cells of blood vessels (Fig. 11D). However, blood vessels located in the hypoderm displayed a lower signal than young invading blood vessels within the feather filaments. We never observed a peculiar signal either in the epidermis or in dermal cells undergoing epidermal folding at the follicle stage.

We have presented here a detailed comparison of the maps showing the cells engaged in DNA synthesis with the maps showing the transcription of the nuclear proto-oncogenes c-myc, c-myb and c-ets-1 during the development of the skin. We do not detect c-myc expression during the early stages when proliferating cells are scattered in the dermis and in the epidermis. Later, during development, c-myc becomes associated with proliferation in both layers. We detect c-myb transcripts during the later stages in the proliferating epidermis. We show that c-ets-1 is transiently expressed in dermal cells during the accumulation stage, as well as in endothelial cells of the invading blood vessels.

In vitro it is a common observation that c-myc is expressed in dividing cells and shut off when these cells enter a resting state (Kelly et al. 1983; Campisi et al. 1984). Examples of cells that divide without apparent increase of c-myc expression have been essentially found in embryos (Pfeifer-Ohlsson et al. 1985; Downs et al. 1989; Jaffredo et al. 1989). Spermatogenic cells proliferate and have barely detectable c-myc transcripts (Stewart et al. 1984). During the first stages of the development of the skin when the cells that proliferate are scattered throughout the epidermis and the dermis, we did not detect c-myc mRNA and protein. One of the major differences in the control of the cell cycle between somatic cells and some early embryonic cells is the existence of fewer checkpoints in the embryonic cell cycles (Hartwell and Weinert, 1989). Thus it is tempting to speculate that loose mesenchymal cells and ectodermal cells in the skin do not require for their proliferation the various steps, including the expression of c-myc, that will be subsequently activated when they will cycle synchronously in the feather buds. The existence of a mitotic signalling pathway bypassing c-myc induction has already been suggested in human fibroblasts stimulated by the platelet-derived growth factor (Coughlin et al. 1985; Rozengurt and Sinnett-Smith, 1988).

In quiescent fibroblasts stimulated by serum, the transient increase in the transcription of c-myc precedes the entry into the S phase (Thompson et al. 1986; Panet et al. 1989). During the regeneration of the liver, c-myc transcripts are induced within two hours after hepa-tectomy and peak at 6 h whereas DNA synthesis occurs 20–24 h later (Sobczak et al. 1989). It has been suggested from these experiments that the induction of c-myc is the result of an activational event that renders the cells competent to enter the cell cycle. In the skin of the chick embryo after the condensation stage, mesodermal and epidermal basal cells enter a state of decreased proliferative activity (Fig. 5A). The incorporation of BrdU is only seen within sparse peridermal cells and at the periphery of the placodes. The features of BrdU labelling of the placodes on each row that arise every six hours and previous results obtained with [³H]thymidine (Wessells, 1965) show that this period of latency lasts 20 to 30 h. Within the cells of the epidermal placodes, the c-myc probe detects a high number of transcripts. One may hypothesize that the transcription of c-myc is part of the response of quiescent epidermal cells to an external signal from the dermis, which triggers their entry in the cell cycle. However, it should be stressed that the distribution of c-myc transcripts in the epidermal placodes is not asymmetrical, whereas the growth of the feather buds becomes overtly asymmetrical immediately after the placode stage. If the transcription of c-myc is part of the G₀→ G₁ transition in the epidermal cells of the placodes, one must postulate that other events trigger the S phase in the caudal edge (Fig. 7A) before the anterior edge of each bud. Alternatively, it is possible that the role of the expression of c-myc at this stage is not related to the control of growth.

When the later stages are examined, the distribution of S-phase nuclei coincides with the distribution of c-myc transcripts. The preferential expression of c-myc in the cephalic part of the bud (Fig. 7B) precedes the pattern of BrdU incorporation that characterizes the following stage where numerous S-phase cells were observed in the rostral part of the bud (Fig. 9A). During the elongation of the feathers, both BrdU-incorporating nuclei and cells labelled with the c-myc probe are preferentially localized within the epidermal sheath of the feather buds or filaments as well as within the papillar ectoderm. The pulp displays some BrdU-labelled nuclei and correlatively the level of c-myc expression drops in the dermis. During this process, a putative delay between the proto-oncogene expression and the S-phase entry within each lineage cannot be accurately determined, as feather outgrowth is a long and continuous process.

The transcription of c-myb in the proliferating epidermis is a new finding. This transcription is detected as the bud emerges from the back skin, later than the first detection of c-myc transcripts. The distribution of c-myb transcripts is similar to the distribution of S-phase cells during the elongation of the feather buds. The expression of c-myb has been most readily detected in hemopoietic cell precursors both in vitro and in vivo (Gonda et al. 1982; Coll et al. 1983; Sheiness and Gardiner, 1984). Recently, we have detected a strong transcription of c-myb in the clusters of hemopoietic cells budding on the ventral wall of the aorta at E3 or forming foci in the dorsal mesentery at E6 (Vandenbunder et al. 1989). The intensity of the hybridization signal suggests that the level of c-myb transcripts is higher in hemopoietic cells than in epidermal cells.

The expression of c-myb has been detected in other proliferating epithelia. The mRNA levels of c-myb are increased in neoplastic human mucosa (Torelli et al. 1987). In related studies, we have also detected c-myb transcripts in the epithelium of the bursa of Fabricius or of the gut (Quéva and Vandenbunder, unpublished results). The hybridization signal is higher at the tips of these branching epithelial tissues where mitosis are more abundant. Interestingly c-ets-1 transcripts have never been detected in epithelia (Vandenbunder et al. 1989).

In this paper, we have also described the expression of c-ets-1 in the developing skin. C-ets-1 is expressed transiently in the dermis when mesenchymal cells accumulate under the uniform epidermis. This increase in cell density under the epidermis has been ascribed to an increased proliferation of dermal cells starting at E6 (Sengel, 1970). These results suggest that c-ets-1 may control the proliferation of dermal cells at a time when c-myc and c-myb are inactive.

As previously reported (Vandenbunder et al. 1989) we show that a high expression of c-ets-1 mRNA is detected in endothelial cells, from the earliest bloodvessel-forming cell clusters to the new capillaries in the pulp of feather filaments. We do not detect BrdU incorporation in these endothelial cells. Thus in the developing blood vessels the expression of c-ets-1 is associated with the differentiation or the migration of endothelial cells. The expression of c-ets-1 in migrating cells has also been reported in the neural crest during neurulation (Vandenbunder et al. 1989).

As discussed by Spom and Roberts (1988), the informational content of signalling molecules as well as nuclear transcription factors does not reside in individual peptides, but in the pattern or set of regulatory peptides molecules to which a cell is exposed. Thus, nuclear transcription factors themselves act in sets in a contextual manner. It is important in understanding their role to consider the expression of different transcription factors that cooperate in the transformation of hemopoietic cells (Golay et al. 1988) and of neuroretina cells (Amouyel et al. 1989) when they are introduced in the genome of recombinant viruses. Further information should be obtained from experiments designed to disturb normal development and/or to dissociate proto-oncogene expression from cell proliferation.

We thank P. Sengel, J. Coll and V. Laudet for critical reading of the manuscript, J. Lipsick for providing the 3.5 kb c-myb probe. We acknowledge L. Pardanaud who drew our attention to the expression of c-myc and c-ets-1 in the skin, when we were working on the blood-forming system of the chick embryo. We thank L. Meunier, E. Ferreira, M.A. Mirabel for technical assistance, F. Laloux and N. Devassine for patient typing. This work was supported by funds from the Institut Pasteur de Lille, Institut National de la Santé et de la Recherche Médicale, Centre National de la Recherche Scientifique and Association pour la Recherche sur le Cancer.