ABSTRACT
We have used in situ hybridization to study the spatial and temporal distribution of the transcription of three cellular oncogenes encoding DNA-binding proteins, c-ets 1, c-myb and c-myc during the development of the chick embryo.
c-ets 1 mRNA expression appears linked to the mesodermal lineage and is strongly expressed in early endothelia; it subsequently becomes restricted to small vessel endothelia. Hemopoietic cells in extraembryonic blood islands at E2 express c-ets 1, while intraembryonic hemopoietic cells in aortic clusters (E3) and paraaortic foci (E6) express c-myb. c-myc transcripts are detected in cells undergoing hemopoiesis in both these extraembryonic and intraembryonic sites. Outside the blood forming system, c-myc is transcribed in a large variety of cells; c-ets 1 displays tissue-specific expression in groups of mesodermal cells engaged in morphogenetic processes and appears excluded from all epithelia; finally the expression of c-myb is the most tightly linked to hemopoietic cells. In any case, it is clear that these three oncogenes display complementary expression in endothelial and hemopoietic cells where their patterns are modulated in relationship to multiplication and differentiation.
Introduction
Proliferation and differentiation of a somatic cell result from the integration by this cell of multiple signals. It becomes increasingly clear that the role of a given growth factor depends on the context of the other signal molecules present (Sporn and Roberts, 1988). Similarly the role of a proto-oncogene product in the elaborate circuitry that governs the proliferation and differentiation will depend on the activity of the various pathways that carry signals from the cell surface to the nucleus.
During the past few years, the in situ hybridization technique has become a powerful method to describe the pattern of expression of various proto-oncogenes (for a review see Adamson, 1987). Information about the role of these proto-oncogenes could be gained if the relationship between the expression of various protooncogenes was established at the cell level by this method. For this purpose, we have undertaken to describe the expression of three different proto-oncogenes, c-ets 7, c-myb and c-myc by in situ hybridization in the chick embryo.
c-ets 1 and c-myb are the cellular progenitors of the two oncogenes present in the genome of the avian erythroblastosis virus E26 (Leprince et al. 1983; Nunn et al. 1983; Leprince et al. 1988). While this retrovirus induces the proliferation of both erythroid and myeloid cells (Radke et al. 1982), the ets sequence is probably essential for erythroid cell transformation by E26 (Beúg et al. 1984; Golay et al. 1988; Nunn and Hunter, 1989). Two members of the ets family, c-ets 1 and c-ets 2, have been recently described in chicken (Chen, 1985; Ghys-dael et al. 1986; Boulukos et al. 1988). c-ets 1 mRNA is abundant in adult lymphoid organs, whereas a 10-fold lower signal is detected in testes, kidney and heart (Bhat et al. 1987). In contrast, c-ets 2 is transcribed in a wide variety of tissues (Bhat et al. 1987; Boulukos et al. 1988). In adults, a very high level of c-myb mRNA is characteristic of immature hemopoietic cells (Gonda et al. 1982; Sheiness and Gardinier, 1984). The MC29 retrovirus, which contains the v-myc sequence, transforms macrophages. The expression of c-myb and c-myc increases after serum or growth factor stimulation of quiescent cells (Kelly et al. 1983; Thompson et al. 1986) ; however, unlike c-myc, c-myb is not involved in the immediate response to serum growth factors. Three members of the myc family, c-myc, N-myc and L-myc, have been characterized so far in mammals. Whereas the expression of N-myc and L-myc is restricted with respect to tissue and stage in the developing mouse (Zimmerman et al. 1986; Mugrauer et al. 1988), the expression of c-myc, which has been described in human, in chicken and in mouse embryos (Pfeifer-Ohlsson et al. 1985; Jaffredo et al. 1989; Schmid et al. 1989), is more widespread.
The products of c-ets 1, c-myb and c-myc are located in the nucleus and bind to DNA (Alitalo et al. 1983; Watt et al. 1985; Klempnauer et al. 1984; Pognonec et al. 1989). It has been shown recently that the protein encoded by v-myb recognizes a specific nucleotide sequence (Biedenkapp et al. 1988). However, the significance of the DNA binding of the proteins encoded by c-myc, c-myb and c-ets 1 remains unknown.
In early vertebrate embryos, the development of blood cells is closely related to that of vascular endothelia. In the avian yolk sac, mesodermal cells aggregate into clusters, the central cells of which become erythroblasts, while the peripheral ones flatten into endothelial cells (Sabin, 1920; Kessel and Fabian, 1987; Péault et al. 1988). These observations have led several authors to propose that both cell types are derived from a common precursor, the ‘hemangioblast’ (Murray, 1932). Close association between endothelia and blood cell precursors is also frequently observed within the embryo proper, where the latter may be seen as ‘aortic cell clusters’ (Jordan, 1915; Dieterlen-Lièvre and Martin, 1981; Olah et al. 1988). Furthermore, diffuse foci of hemopoiesis are abundant within the dorsal mesentery of the chick embryo in the vicinity of the aorta. Both the aortic clusters and the paraaortic foci are signs of the amplification and differentiation of blood stem cells formed within the embryo proper rather than in the yolk sac. The existence of these intraembryonic stem cells has been discovered through the study of chimeras consisting of a quail embryo grafted on a chick yolk sac (Dieterlen-Lièvre, 1975, 1984).
Another experimental approach, involving interspecies quail-chick grafting of embryonic rudiments and their analysis with a monoclonal antibody that recognizes the hemangioblastic lineage in the quail, has recently revealed two different modes of development of endothelia during ontogeny (Pardanaud et al. 1989). In internal organ rudiments, composed of endoderm and mesoderm, endothelia arise in situ. In contrast, external rudiments composed of ectoderm and mesoderm, in particular limb buds and bone marrow, are colonized by extrinsic endothelial precursors. The first process is designated as ‘vasculogenesis’ and the second as ‘angiogenesis’. Concerning the ontogénie relationship between endothelia and hemopoietic cells, these experiments have also shown that hemopoietic cells from an extrinsic source colonize internal organs, thus that their origin is independent from that of endothelial cells.
The present analysis revealed that c-ets 1 is prominently expressed in a greater variety of cells than previously realized. All these cells are of mesodermal origin and the expression of c-ets 1 is transient during their evolution. The most conspicuous expression of c-ets 1 occurs in endothelial cells during the early stages of blood vessel formation, independently of their mode of differentiation, c-myc mRNA is detected in extra-embryonic as well as in intraembryonic hematopoietic cells, whereas c-myb mRNA is first detected in the hematopoietic cells of paraaortic foci.
Materials and methods
Preparation of RNA probes
A 750bp Bg/11-H/TidIII fragment containing the 3′ half of vets was obtained from lambda E26Q1, a molecularly cloned E26 provirus (Leprince et al. 1983), and subcloned into the polylinker region of pSP64 and 65 (Promega Biotec) by standard methods. Thus this fragment was inserted in both orientations relative to the promoter for Salmonella bacteriophage SP6 RNA polymerase. The same procedure was followed with a 350 bp EcoRl-Sall fragment containing the major part of v-myb within the same E26 provirus and with a 600 bp EcoRI fragment of the 3′ part of the chicken c-myc gene (Saule et al. 1984).
The RNA probes were transcribed from truncated plasmids in 20 μl reaction mixtures by the SP6 RNA polymerase (Promega Biotec) using 100 μCi [32P]UTP (400 Ci mmole-1, NEN) for blot analysis or 20 μCi [3H]UTP and 20μCi [3H]GTP (∼40 Ci mmole-1, NEN) for in situ analysis. We checked by electrophoresis on 2·2M-formaldehyde/2·2 % agarose gels that the sizes of the sense- and of the antisense-labelled transcripts were identical. The probes for in situ hybridization were subjected to a limited alkaline hydrolysis according to Cox et al. (1984). After ethanol precipitation, the probes were dissolved in 2 mM-EDTA containing 2 mg ml-1E. Coli tRNA and stored at —80°C for a maximum of 9 months.
RNA blot analysis
Tissues were placed immediately after dissection in guanidinium isothiocyanate and homogenized using a Polytron apparatus. Total cellular RNA was isolated from chicken tissues and from embryos by the method of Chirgwin et al. (1979). Total RNA (20 μg/lane) was subjected to electrophoresis in agarose gels containing T2% formaldehyde and transferred onto nitrocellulose membranes (Hybond c-extra, Amersham). The RNA probe activity in the hybridization buffer was 4×106ctsmin-1 ml-1. After hybridization at 60°C, the blots were washed in 2xSSC, 0·1% SDS at room temperature and in 0·1×SSC, 0·1 % SDS at 75°C. In addition, blots were incubated for 15 min at room temperature with If/gml-1 RNAseA (Sigma) depending upon the degree of background signals (Heuman et al. 1983).
Tissue preparation
The eggs of brown Leghorn chicken were incubated at 37 °C in a humidified air chamber. The age of embryos is indicated as El, E2 etc…. E0 being the first day of incubation. Two or three embryos were examined on each of the following days, E2, E3, E6 and E8. They were fixed at 4°C for 16 h in 4% paraformaldehyde in PBS containing 5mM-MgC12, dehydrated and embedded in paraffin. 5 to 6 μm thick transverse sections were transferred to polylysine-coated slides and heated at 42°C for two days. Slides were stored at 4°C for up to 6 months.
In situ hybridization
In situ hybridization was performed essentially by the method of Cox et al. (1984). After deparaffinization and hydration, slides were incubated in 0·1 M-glycine, 0·2M-THS-HC1, pH7·4 for 10min at room temperature, treated with proteinase K (1μgml-1 Boehringer-Mannheim) for 15min at 37°C and further fixed in 4% paraformaldehyde in PBS. The slides were subsequently washed in PBS, acetylated and dehydrated as described by Cox et al. (1984).
Immediately prior to hybridization, the probes were denatured at 80°C for 3 min and diluted in the hybridization buffer (Cox et al. 1984). A low level of background and a highly specific signal were obtained when the hybridization was performed at 60°C for 16 h with a probe concentration of 50pg μl−1 in the hybridization buffer. Thereafter, the slides were washed in 4xSSC at room temperature and the sections treated with RNAse A (10ggml-1, type III A, Sigma) for 30 min at 37°C. The final washes were in 0-lxSSC at 60–70°C for 10 min. The slides were dehydrated and dipped in a nuclear track emulsion (Kodak NTB2 diluted 1:1 with 0·6 M-ammonium acetate). The slides were exposed at 4 °C for three months and developed according to the recommendations of Cox et al. (1984) and coverslips were mounted. In some cases, labelling with Hoechst dye 33258 (bisbenzimidine) was performed to visualize the nuclei (Hilwig and Grapp, 1972). The sections were examined under phase-contrast and dark-field illumination with a Zeiss microscope (IM35).
Results
Specificity of the hybridization signals
Previous studies have shown that a PstI fragment derived from v-ets (nt 1504-1927, Nunn et al. 1983) used as a probe was able to detect mRNAs of 7·5, 2·0 and 1·5 kb, which are transcribed from c-ets 1 (Leprince et al. 1983), and a 4·0 kb mRNA, which is transcribed from c-ets 2 (Boulukos et al. 1988). The 705 bp Bg/II-HindIII fragment of v-ets (nt 1065-1770), used as template for the synthesis of single-stranded RNA probes, contains c-ets 1 -specific sequences along most of its length. The antisense 32P-labelled transcript hybridizes with a major band of 7·5 kb on a Northern blot of total RNA from newborn chicken spleen, thymus, lungs and from 3-day whole chicken embryo (Figure 1A); no 4·0 kb RNA species is detected in the MC29-trans-formed macrophage cell line HD11, which expresses the 4·0 kb c-ets 2 RNA (Boulukos et al. 1988). On this basis we conclude that the 705 bp BglII-HindIII fragment of v-ets is a specific probe for c-ets 7.
(A) RNA blot analysis of c-ets 1 expression in chicken tissues. 20 μg of total RNA were extracted from whole 3-day embryos (lane 1) and from the liver (lane 2), gut (lane 3), lung (lane 4), kidney (lane 5), spleen (lane 6) and thymus (lane 7) of chickens 10 days post hatching. Hybridization with the c-ets 1 antisense probe yields a main band at 7·5 kb; under the same conditions, the control c-ets 1 sense probe gives no signal (not shown). (B) Detection of c-ets 1 transcripts by in situ hybridization. Serial sections were prepared from a paraffin block containing fragments of muscle (M), gut (G), the spleen (S) and lobes of the thymus (T) of a chick 10 days post hatching. The sections were hybridized with 32P;-labelled c-ets 1 antisense and sense probes according to Material and methods, and exposed to Amersham Hyperfilm β3 max for 10 days at 4 °C.
(A) RNA blot analysis of c-ets 1 expression in chicken tissues. 20 μg of total RNA were extracted from whole 3-day embryos (lane 1) and from the liver (lane 2), gut (lane 3), lung (lane 4), kidney (lane 5), spleen (lane 6) and thymus (lane 7) of chickens 10 days post hatching. Hybridization with the c-ets 1 antisense probe yields a main band at 7·5 kb; under the same conditions, the control c-ets 1 sense probe gives no signal (not shown). (B) Detection of c-ets 1 transcripts by in situ hybridization. Serial sections were prepared from a paraffin block containing fragments of muscle (M), gut (G), the spleen (S) and lobes of the thymus (T) of a chick 10 days post hatching. The sections were hybridized with 32P;-labelled c-ets 1 antisense and sense probes according to Material and methods, and exposed to Amersham Hyperfilm β3 max for 10 days at 4 °C.
On Northern blots, the antisense probes for c-myb and c-myc hybridize, respectively, with 4·0 kb mRNA species and with 2·6 kb mRNA species, which is the expected size for the corresponding mRNA in the chicken (data not shown). The 0·6kb chicken c-myc probe used in these studies contains sequences at the 3′ end of the third exon. In both human and mouse, these sequences encode for domains with amino acid homology between the three members of the myc family, c-myc, N-myc and L-myc (Alt et al. 1986). However, when Southern blots of chicken DNA digested with various restriction enzymes were hybridized with this probe (Saule et al. 1984), one single band was observed on each lane, indicating that the 0·6 kb c-myc probe was specific for one single locus under the conditions prevailing during the hybridization. Since the thermal stabilities of RNA-RNA duplexes in a cell and of DNA-DNA duplexes on a blot are different, one cannot deduce directly from Southern blot analysis what is the specificity of the c-myc probe when used for in situ hybridization analysis. The experiments discussed in this paper have been done under stringent conditions (hybridization at 60°C and washes at 70°C in a 0·l ×SSC). Previous in situ hybridization experiments with the 0·6 kb c-myc probe (Jaffredo et al. 1989) revealed in the chick embryo a pattern of expression similar to the pattern obtained in the mouse embryo with probes specific for exon 1, 2 and 3 of the mouse c-myc gene, with a high signal in the endoderm, in particular in the liver (Schmid et al. 1989).
The sense 32P-labelled transcripts derive from the vets, v-myb and c-myc fragments give no visible signal on Northern blots under the same conditions. Thus these sense transcripts have been used as negative controls for in situ hybridization. Following the suggestion of Lawrence and Singer (1985), we took advantage of the high specific radioactivity of the 32P-labelled probes to adjust the conditions of the in situ hybridization so as to match the results of these experiments with the results obtained on Northern blots. With the appropriate temperatures for hybridization and washing, sections of chicken thymus and spleen give a strong positive signal, whereas sections of muscle or gut are weakly stained with the 32P antisense v-ets probe (Figure 1B). Under the same conditions, the sense probes give no visible signal. Subsequently, tritiated probes, which achieve a good spatial resolution, were routinely used. In all the tissues described in these studies, no subpopulation of cells gave a signal above the background with the sense probes.
The description of gene expression by in situ hybridization requires passing through a minefield of potential artifacts and one may argue that some cells give a negative signal because their mRNAs have been specifically washed out during the procedure or because the diffusion of the probe throughout the cytoplasmic matrix is hindered. Proteolytic treatment after fixation has been introduced to improve the accessibility of mRNA for hybridization, but its extent must be adjusted to avoid the loss of the mRNAs from the sections. In the absence or in the presence of proteinase K (1 or 4 μml−1), we obtain the same qualitative results, although when the proteinase K treatment is omitted, the overall efficiency of hybridization is reduced. Thus the patterns of expression that have been obtained do not result from a differential accessibility of the mRNA in the cells. The different patterns obtained with different probes are a good indication of the specificity of the procedure and serve as positive controls.
General pattern of expression of the three oncogenes in the early embryo
It has long been assumed from the results of Northern blot analysis that the expression of these three oncogenes was not restricted to a single cell lineage. The particulars of the in situ hybridization technique allows a detailed description of this expression. Both c-ets 1 and c-myb probes detect numerous mRNA copies in E10-E15 thymuses (data not shown). In younger embryos (E2-E6), we found that the c-ets 1 probe detects transcripts in a greater variety of cells than previously realized, cells of mesodermal origin displaying a signal clearly higher than the background during the whole period studied. The signal is particularly strong on endothelial cells; the pattern of expression in these latter cells will be detailed below. However, of all c-ets 1 labelling observed over the period studied, by far the most striking is found over the mesencephalic neural crest (Figure 2). The signal over this migrating group of cells appears much stronger than over any other rudiment. Truncal neural crest is also labelled. Remarkably, c-ets 1 transcripts are never detected in epithelia, whether they are derived from endoderm, mesoderm or ectoderm (Figures 4A, 5A, 6A). c-myb expression may be observed in a few cell types outside the bloodforming system but this expression is so limited that c-myb may be considered as restricted to the hemopoietic system.
c-ets 1 in the neural crest, 8-somite embryo. Midbrain level. Double labelling for c-ets 7 (A) and for nuclei with Hoechst stain (B). Bar: 100 μm. Migrating mesencephalic neural crest cells are heavily labelled (arrows). Other c-ets 1 signals are located over endothelia. One of the two dorsal branches of the aorta appears especially labelled, due to a favorable angle of the section, d, dorsal branch of aorta; nt, neural tube; ph, pharynx; v, ventral branches of aorta
c-ets 1 in the neural crest, 8-somite embryo. Midbrain level. Double labelling for c-ets 7 (A) and for nuclei with Hoechst stain (B). Bar: 100 μm. Migrating mesencephalic neural crest cells are heavily labelled (arrows). Other c-ets 1 signals are located over endothelia. One of the two dorsal branches of the aorta appears especially labelled, due to a favorable angle of the section, d, dorsal branch of aorta; nt, neural tube; ph, pharynx; v, ventral branches of aorta
c-ets 1 and c-myc in extraembryonic blood islands. Sections from a 7-somite-embryo were double-labelled with one of the oncogene probes and with Hoechst nuclear stain. Bar: 200 μm. (A, C) c-ets 1 preobe; (E) c-myc probe; (B, D, F) Hoechst stain, e, ectoderm; en, endoderm; m, mesoderm; nt, neural tube; v, vessel. (A, B) Proximal early extraembryonic blood islands strongly express c-ets 1 mRNA in most or all their cells. Theese two blood islands (stippled circles in B) are so immature that their cells are loosely arranged and cannot be distinguished from surrounding mesoderm in (B). (C, D) Distal, more mature, extraembryonic blobd island (arrows). The c-ets 1 signal is restricted to endothelia of the blood island and of vessels. (E, F) The c-myc probe yields a signal over these three distal blood islands (arrows). Endothelia of vessels are not labelled. Vitelline granules display non specific light scattering in C and E.
c-ets 1 and c-myc in extraembryonic blood islands. Sections from a 7-somite-embryo were double-labelled with one of the oncogene probes and with Hoechst nuclear stain. Bar: 200 μm. (A, C) c-ets 1 preobe; (E) c-myc probe; (B, D, F) Hoechst stain, e, ectoderm; en, endoderm; m, mesoderm; nt, neural tube; v, vessel. (A, B) Proximal early extraembryonic blood islands strongly express c-ets 1 mRNA in most or all their cells. Theese two blood islands (stippled circles in B) are so immature that their cells are loosely arranged and cannot be distinguished from surrounding mesoderm in (B). (C, D) Distal, more mature, extraembryonic blobd island (arrows). The c-ets 1 signal is restricted to endothelia of the blood island and of vessels. (E, F) The c-myc probe yields a signal over these three distal blood islands (arrows). Endothelia of vessels are not labelled. Vitelline granules display non specific light scattering in C and E.
c-ets 1, c-myb and c-myc in intraaortic clusters. Transverse sections from E3 embryos at trunk level, a, aorta; g, gut; 1, liver; nt, neural tube. Bar: 100urn. The c-ets 1 signal is present in the mesenchyme, usually associated with endothelia. This association is particularly clear in the sinus venosus inside the liver. The gut endoderm (arrow) is unlabelled, c-myb and c-myc are detected only in the aortic clusters (arrows). While c-myb is restricted to this group of cells, the background is more important for the c-myc signal and the somite displays some signal above it. Erythrocytes in large vessels display non-specific light scattering (in A and C).
c-ets 1, c-myb and c-myc in intraaortic clusters. Transverse sections from E3 embryos at trunk level, a, aorta; g, gut; 1, liver; nt, neural tube. Bar: 100urn. The c-ets 1 signal is present in the mesenchyme, usually associated with endothelia. This association is particularly clear in the sinus venosus inside the liver. The gut endoderm (arrow) is unlabelled, c-myb and c-myc are detected only in the aortic clusters (arrows). While c-myb is restricted to this group of cells, the background is more important for the c-myc signal and the somite displays some signal above it. Erythrocytes in large vessels display non-specific light scattering (in A and C).
c-ets 1, c-myb and c-myc in paraarortic hemopoietic foci. Neighbouring transverse sections through an E6 embryo. (A) c-ets 1; (B) c-myb; (C) c-myc; (D) phase contrast, a, aorta; er, erythrocytes; s, segmental arteries; k, kidney. Bar: 100 μm. C-ets 1 is strongly expressed in the endothelium of segmental arteries, between the tubules of the mesonephros and on a glomerulus (arrow). By contrast, the tubules are completely negative. The c-ets 1 signal has disappeared from the aortic endothelium, c-myb and c-myc transcripts are expressed only by the hemopoietic paraaortic foci (arrowheads). The same group of maturing erythrocytes (er) displays non-specific light scattering in the four pictures.
c-ets 1, c-myb and c-myc in paraarortic hemopoietic foci. Neighbouring transverse sections through an E6 embryo. (A) c-ets 1; (B) c-myb; (C) c-myc; (D) phase contrast, a, aorta; er, erythrocytes; s, segmental arteries; k, kidney. Bar: 100 μm. C-ets 1 is strongly expressed in the endothelium of segmental arteries, between the tubules of the mesonephros and on a glomerulus (arrow). By contrast, the tubules are completely negative. The c-ets 1 signal has disappeared from the aortic endothelium, c-myb and c-myc transcripts are expressed only by the hemopoietic paraaortic foci (arrowheads). The same group of maturing erythrocytes (er) displays non-specific light scattering in the four pictures.
c-ets 1 in the mesoderm lineage. E6 embryo, dm, dorsal mesentery; lu, lung; nt, neural tube. Bar: 100 μm. (A) Internal organs. Mesenchyme of the lung, the gonad and the mesonephros displays fairly uniform labelling. The dorsal mesentery is more weakly labelled. Note absence of detectable expression within the bronchial epithelium (arrow), the germinal epithelium of the gonad (arrowhead) and in the mesonephric tubules (the latter structure is better illustrated in Figure 5A). (B) Axial structures in the trunk. Most labelling is associated with endothelia around the neural tube and in the hypoderm. The dermis (arrows) in this dorsal region is uniformly labelled.
c-ets 1 in the mesoderm lineage. E6 embryo, dm, dorsal mesentery; lu, lung; nt, neural tube. Bar: 100 μm. (A) Internal organs. Mesenchyme of the lung, the gonad and the mesonephros displays fairly uniform labelling. The dorsal mesentery is more weakly labelled. Note absence of detectable expression within the bronchial epithelium (arrow), the germinal epithelium of the gonad (arrowhead) and in the mesonephric tubules (the latter structure is better illustrated in Figure 5A). (B) Axial structures in the trunk. Most labelling is associated with endothelia around the neural tube and in the hypoderm. The dermis (arrows) in this dorsal region is uniformly labelled.
On the other hand, high levels of expression of c-myc are observed in the hematopoietic cells of El to E6 embryos, but the c-myc signal is not restricted to these cells, numerous cell types lighting up at different periods of embryogenesis according to a time- and tissue-specific schedule. The present paper will deal only with c-myc expression related to the hemopoietic system. Since the early steps of erythropoiesis and of vasculogenesis in the chick embryo appear tightly linked, the patterns of expression of c-ets 1, c-myb and c-myc will be analyzed in detail in relationship to one another during these events.
c-ets 1 and c-myc are transcribed in extraembryonic blood islands
After formation of the mesoderm by ingress of cells through the primitive streak, tight clusters of cells appear in the mesoderm of the area opaca in contact with the endoderm. At about the 6-to 8-somite stage, erythroblasts containing hemoglobin are detectable in these clusters, which have increased in size (Wilt, 1974). At the same time, the cells on the surface of these blood islands flatten to become endothelial. In the 7-somite embryo, the c-ets 1 probe raises a strong signal in small groups of cells in close contact with the endoderm, which otherwise cannot yet be recognized as blood islands (Figure 3A, B). The c-ets 1 signal disappears from the central cells of more mature blood islands but it remains on endothelia (Figure 3C, D). In contrast, the c-myc probe hybridizes strongly to the central cells of these blood islands (Figure 3E, F). At this stage, the c-myb probe used in our studies detects no mRNA anywhere within the embryo.
One day later (not shown), only the endothelial cells in the extraembryonic area display the c-ets 1 signal. The c-myc signal is detected in the circulating blood cells of the E3 embryo but it is weaker than that observed earlier in the blood islands.
c-ets 1, c-myb and c-myc in intraaortic clusters and paraaortic foci
The first intraembryonic vessels to form are the dorsal aortae, the endothelial cells of which express c-ets 1 (Figure 2A) in the 7-somite embryo. Thereafter, the dorsal aortae come into apposition and fuse from the aortic arches caudalwards. The ventrolateral aspects of the aortic endothelium of 3-day embryos display two thickened ridges, composed of budding intraembryonic blood cells (Dieterlen-Lièvre and Martin, 1981). The expression of c-ets 1 is diffuse in the mesenchyme surrounding the aorta and is relatively faint in the endothelial cells of the aorta at E3 (Figure 4A). The cell clusters that protrude both in the lumen of the aorta and in the mesentery express c-myb and c-myc (Figure 4B, C). Since neighbouring sections of the aorta were hybridized to the different probes, it was not possible to determine whether some cells in these clusters express both c-ets I and c-myb or whether the cells that express c-myb are distinct from the cells expressing c-myc. Diffuse foci of basophilic cells are found throughout the dorsal mesentery three days later. Cells within these foci are engaged in hemopoiesis (Dieterlen-Lièvre and Martin, 1981). In the E6 embryo, a conspicuous hybridization signal was seen on these foci with the myb probe and the myc probe, whereas the c-ets 1 did not hybridize (Figure 5). Interestingly, Figure 5 shows that the endothelial cells of the aorta at E6 no longer expressed c-ets 1, whereas in the segmental arteries flowing out from the aorta the endothelial cells expressed c-ets 1.
c-ets 1 in the endocardium
The avian heart is composed of three tissue layers, the epicardium, the myocardium (or muscular layer) and the endocardium which lines the heart’s internal surface. Whereas the myc probe produces a weak signal in the developing heart from the 7-somite stage to E6, no myb signal is found in the heart. In the 7-somite embryo, the endocardium expresses c-ets 1 (data not shown). In the E3 heart, which is pumping blood, the expression of c-ets 1 is similarly restricted to the endocardium. Later in development, this expression decreases; in the E6 endocardium, the c-ets 1 probe no longer hybridizes. Thus the features of c-ets 1 expression in the heart appear similar to that in the major blood vessels, a strong signal in the endothelium being the feature of early stages; when several mesenchymal layers become organized around the primitive endothelium, the expression of c-ets 1 disappears from the endothelium of major blood vessels.
The establishment of the early vascular tree occurs by local differentiation. The formation of blood vessels also proceeds through the growth and extension of preexisting ones. We have subsequently investigated the influence of the development process of the blood vessels on the expression of c-ets 1, c-myb and c-myc by describing their expression in various organs.
c-ets 1 in angiogenesis in the kidney and the neural tube
The brain is colonized by extrinsic endothelial cells and the vascularization of the nervous system occurs by extension and branching of existing capillaries (Stewart & Wiley, 1981). The same is true in the kidney (Ekblom et al. 1982). The early stages of the development of the kidney are characterized by the expression of c-ets 1 in the intermediate mesoderm which gives rise to the nephric blastema. However, when kidney tubules have differentiated, they are negative with the c-ets 1 probe. In the E3 embryo, these negative mesonephric tubules are surrounded by cells that express c-ets 1. In the E6 embryo, it becomes clear that c-ets 1 is expressed in endothelial cells of the sinusoids between the tubules of the mesonephros and in the glomeruli (Figure 5A, 6A). After hatching, the expression of c-ets 1 in the definitive kidney is restricted to the glomeruli (presumably the capillaries).
We have detected the expression of c-ets 1 in the vessels that sprout from the aorta in the intersomitic space of the E3 embryo, in the vessels that envelop and penetrate the spinal cord and the brain and in the capillaries around the spinal ganglia. This signal is well above background between E3 and E8. Figure 6B shows the expression of c-ets 1 in the endothelium of the vessels surrounding the neural tube of an E6 embryo.
Neither c-myc nor c-myb probes produce a signal in the blood vessels irrigating the kidney and the brain.
c-myc and c-ets 1 during liver vasculogenesis
The formation of the vessels in the liver involves emergence of endothelial cells from in situ located mesoderm, rather than colonization by extrinsic buds as it does in the nervous system and the kidney. Angio-blastic cells differentiating from local mesenchyme proliferate and hepatocytes grow closely apposed to them. In the E3 embryo, c-ets 1 expression is detected in the endothelia of the ductus venosus and of sinusoids, whereas a high expression of c-myc is detected in the hepatocytes (Figure 7A); the c-myc signal decreases drastically by E6 (Figure 7B). The signal produced with the myc probe in the hepatocytes is higher in the peripheral portions of the liver than in the more central portions. The endothelial wall of the ductus venosus no longer expresses c-ets 1 in the E6 embryo. By contrast, the endothelium of the sinusoids still expresses a high level of c-ets 1 mRNA (Figure 7C). During the whole period, no signal is observed with the myb probe, a finding well correlated with the fact that hemopoiesis in the liver of the avian embryo is modest or absent, in contrast to the prominent hemopoietic role of the mammalian fetal liver. Ten days after hatching, we observe no signal in the liver with the probes used in these studies.
c-ets 1 and c-myc in the liver, h, heart; 1, liver; m, mesenchyme of the gut. Bar: 100 μm. (A) c-myc in E3 embryo. Transverse section at the heart level. Note that while the expression is intense in the liver, few transcripts are present in the heart. (B) c-myc in E6 embryo. Transverse section through the liver. The signal has drastically decreased in the liver by comparison to E3. Gut mesenchyme adjacent to the liver is negative. (C) c-ets 1 in E6 embryo. B and C are from the same liver. The signal delineates the endothelium of the vascular tree. In adjacent mesoderm the signal is uniformly distributed.
c-ets 1 and c-myc in the liver, h, heart; 1, liver; m, mesenchyme of the gut. Bar: 100 μm. (A) c-myc in E3 embryo. Transverse section at the heart level. Note that while the expression is intense in the liver, few transcripts are present in the heart. (B) c-myc in E6 embryo. Transverse section through the liver. The signal has drastically decreased in the liver by comparison to E3. Gut mesenchyme adjacent to the liver is negative. (C) c-ets 1 in E6 embryo. B and C are from the same liver. The signal delineates the endothelium of the vascular tree. In adjacent mesoderm the signal is uniformly distributed.
Discussion
Knowing how cellular oncogene expression is modulated in specific cell types during development will help to shed light on the roles of these genes. At the present moment, however, the products of most oncogenes have been evaluated by biochemical means on tissue extracts, so the cells expressing them have not been identified. Furthermore, few of the oncogenes known to date have been investigated during ontogeny. Of the three nuclear oncogenes whose in situ expression we report here, only c-myc has been subjected to similar analyses in the human embryo (Pfeifer-Ohlsson et al. 1985), the mouse embryo (Schmid et al. 1989) and the chick embryo (Jaffredo et al. 1989). These reports showed that c-myc is expressed in a large variety of cells, where it is modulated according to time- and tissue-specific developmental patterns, and that it is not tightly linked with cell proliferation. In particular, c-myc is regularly expressed in postmitotic neurons, where it has been detected both as mRNA (Ruppert et al. 1986) and as the protein product (Jaffredo et al. 1989).
None of the two other oncogenes investigated here has been studied previously in situ in the embryo. In the adult chicken, a high level of c-ets 1 exists in lymphoid cells and a lower level in erythroblasts and bone marrow cells (Chen, 1985; Ghysdael et al. 1986). c-myb is expressed at high levels in hemopoietic cells from various lineages (Gonda et al. 1982; Coll et al. 1983; Sheiness and Gardinier, 1984; Duprey and Bdttiger, 1985; Emilia et al. 1986; Bading et al. 1988).
In the present study, in situ hybridization to c-myc, c-ets 1 and c-myb probes was carried out on sections of whole chicken embryos, making it possible to determine the time- and cell-specific transcription patterns during the periods of morphogenesis and organogenesis. While c-myc transcripts were widely distributed, the other mRNAs were more restricted and appeared strikingly related to the blood-forming system. These observations enticed us to center the present report on this system, though other sites of expression occur that will be reported in detail later. Schematizing, c-ets 1 was displayed by endothelial cells, while c-myc and c-myb were expressed by hemopoietic cells; the less mature the cells, the stronger the hybridization. In the case of c-ets 1, for instance, grain number decreased at E6 on the aortic endothelium, while it was high on the newly formed segmental arteries. At the period of development concerned (E2–E8), hemopoietic cells differentiate in two locations, extraembryonic blood islands on the one hand, intraembryonic aortic clusters and paraaortic foci on the other (Dieterlen-Lièvre, 1984; Cormier et al. 1986; Cormier and Dieterlen-Lièvre, 1988). The expression of the three nuclear oncogenes is not identical in these two sets of hemopoietic rudiments. Early extraembryonic blood islands express c-ets 1, while aortic clusters and paraaortic foci express c-myb. Both sets express c-myc intensely. It cannot be resolved from the present data whether the two sets of hemopoietic rudiments really have different oncogene patterns or whether transient expression of one or the other escaped our scrutiny, bearing in mind the stages selected. However, it is well known that the primitive hemopoietic generation has a number of special morphological and biochemical features. For instance, in mammals the primitive erythroid lineage does not depend on erythropoietin for differentiation (Cole and Paul, 1966). In the face of such different requirements, it seems plausible that oncogene expression is also different.
In any case, it is interesting to find that, within the embryo and in the adult, c-myb is expressed in hemopoietic cells. In this regard, it should be mentioned that a c-myb antisense oligodeoxynucleotide was shown to inhibit the formation of human myeloid colonies in vitro (Gewirtz and Calabretta, 1988). This recent piece of work elegantly demonstrated the critical role of c-myb in the amplification of blood cell precursors. The expression of most oncogenes is, however, not restricted to a single cell lineage, c-myb, although prominently associated with hemopoietic cells, does not escape this rule. Later in development, a signal is detected in some proliferating epithelia (Vandenbunder, unpublished results); this observation is consistent with the transient increase of c-myb mRNA observed in chicken embryo fibroblasts following serum stimulation (Thompson et al. 1986).
Expression of c-ets 1 mRNA in endothelial cells is a new finding. It will be interesting to define the circumstances of this expression and learn whether it becomes extinct in the quiescent state that is characteristic for most mature endothelial cells. In any event, it should be noted that the endothelial signal in the embryo was present, irrespective of the process through which the endothelia were forming, i.e. vasculogenesis or angiogenesis (Pardanaud et al. 1989).
c-ets 1 expression is not restricted to endothelial cells, any more than c-myb is restricted to the hemopoietic lineage. However, it does appear tightly linked to the mesoderm lineage. A signal of medium intensity is present over all mesenchyme in E3 embryos, a higher signal characterizes the intermediate plate and an impressive number of grains decorates the migrating mesencephalic neural crest. It will be important to determine whether this c-ets 1 expression is related to the mesectodermal potencies of this region of the neural crest. Finally, in the dermis at E6, the c-ets 1 probe lights up small collections of cells, which are going to multiply into aggregates that initiate feather bud formation. The transient expression of c-ets 1 in various types of cells suggest that it is part of the response to an external signal inducing mesodermal cells in a differentiation pathway. These interesting patterns of c-ets 1 expression will be submitted to experimental analysis in the near future. One final feature of c-ets 1 expression deserves emphasis; the signal is never observed over epithelia, irrespective of their germ layer origin.
It should be stressed that the detection of a hybridization signal with one probe in various cell types does not mean that these cells are producing the same proteins or even that they are producing a protein. Differences due to alternative splicing, translational and post-translational controls may occur which modify the functional significance of the hybridization signals. Tissue-specific splicing has been shown for c-src: neural tissues express a uniquely spliced c-src mRNA which contains different 5′ sequences encoding the amino portion of pp60c-src (Martinez et al. 1987). c-ets 1 is expressed in chicken cells as RNAs of 7·5, 2·0 and 1·5 kb (Leprince et al. 1983; Chen, 1985) and one can speculate that the various situations where c-ets 1 is detected correspond to different transcripts. On Northern blots (Figure 1), the c-ets 1 probe hybridizes only with 7·5 kb mRNA species. Recently, Leprince et al. (1988) demonstrated the existence of an alternative splicing within the chicken c-ets 1 locus. The resulting mRNAs are both 7·5 kb in size and encode two different proteins of 54 and 68×103Mr. Since the c-ets 1 probe used in these studies is complementary to the exons which are common to the two mRNAs encoding the 54 and 68×103Mr proteins, we cannot conclude whether the signals observed in the different classes of tissues (lymphoid and non-lymphoid) correspond to the same or to two distinct 7·5 kb mRNAs. Probes specific to the two 7·5 kb mRNAs have been prepared and work is currently in progress to distinguish between these two c-ets 1 transcripts in the chicken embryo.
The regulation of the expression of c-myc by a post-transcriptional mechanism has been demonstrated in Go-arrested fibroblasts stimulated by growth factors (Blanchard et al. 1985). The examination of both c-myc mRNA and protein reveals a stringent translational regulation in the chick embryo (Jaffredo et al. 1989). Whereas hepatocytes display a strong hybridization signal, no protein is detected in this tissue with the myc- specific antibodies. The striking discrepancy in endodermal c-myc RNA and protein uncovered in the chick embryo should be an incentive to systematic comparison of the two levels of expression.
In conclusion, by means of in situ study of whole embryos, it has been possible to detect expression of c-ets 1 in new cell types, all of mesodermal lineage, some of which give rise to the blood-forming system. The strongest expression of c-ets 1 in the embryo is in endothelial cells, c-myb, in the embryo and in the adult, is expressed preferentially in immature hemopoietic cells, while c-myc expression is ubiquitous, though tightly time-regulated. The preferential expression of c-myb and c-ets 1 in related cell lineages which may have a common ancestor is striking in view of the capture of these two oncogenes by retrovirus E26. This relationship suggests that the interaction of the v-myb and of the v-ets domains in the 135 ×103Mrgag-myb-ets fusion protein mimics the transient interaction of the c-myb and the c-ets proteins during the development of the blood-forming system. The functional significance of these findings may become clearer when they have been extended by a detailed study of other cell types expressing these oncogenes. In this regard, it is interesting that early hemopoietic foci express c-ets 1 and later foci c-myb. Combinations of signals are thought to be involved in cell multiplication and differentiation. If confirmed, the alternate expression of either c-ets 1 or c-myb in cells undergoing similar differentiation processes suggests that a similar function is carried out by different oncogenes at different times of ontogeny.
ACKNOWLEDGEMENTS
We thank Dr F. Dieterlen for numerous discussions and continuous support throughout the course of this work; Drs M. Brahic, R. Angerer, B. Gosselin and the technicians in his laboratory, A. Flourens and J. Vanderdonckt for their assistance during our first attempts with the in situ hybridization technique; Dr Julian Smith for critical reading of the manuscript; B. Henri and Y. Rantier for their excellent photographic contribution; N. Devassine, M. C. Bouchez and S. Redjem for patient typing. This work was supported by funds from the Institut National de la Santé et de la Recherche Médicale, Centre National de la Recherche Scientifique, Institut Pasteur de Lille and Association pour la Recherche sur le Cancer.