The spatial and temporal dynamics of Sax1(CHox3) homeobox gene expression in the chick’s spinal cord

In the chick, gastrulation via the primitive streak starts at stage 3 H&H (Hamburger and Hamilton, 1951). The first cells to ingress into the lower layer are the endodermal cells which are followed by the ingression of mesodermal cells that form the middle layer between the epiblast and hypoblast (Bellairs, 1986; Nicolet, 1971; Stern, 1991). At stage 4 H&H, the primitive streak acquires its full length with Hensen’s node at its anterior end. The mediorostral triangle of the node is the anlage of the notochord, and the epiblast rostral to it is the future neural plate (Selleck and Stern, 1991). On both sides of the anlage of the notochord are the cells that after ingression into the middle layer will form the somites on each side of the ingressed notochord (Schoenwolf, 1979). The somites segregate from the mesodermal plate in an orderly manner, rostral to caudal (Ooi et al., 1986). As development proceeds along the anteroposterior axis, the notochord induces the ventral part of the already formed neural tube to form the floor plate and together with the floor plate, induces motor neurons (Jessel and Dodd, 1992; Ruizi i Altaba and Jessel, 1993), while other, presumably dorsal signals induce the dorsal part of the neural tube (Goulding et al., 1993).

One of the characteristics of the above directional morphogenetic processes is a series of induction interactions between neighboring tissues. A major subject in contemporary research are the molecules that are involved in these inductions, which include secreted molecules and their receptors (reviewed in Jessel and Melton, 1992; Kingsley, 1994; see also Niswander et al., 1993; Riddle et al., 1993; Thaller and Eichele, 1987) and transcription factors that respond to the above factors and transmit the induction signal (reviewed in Gruss and Walther, 1992; McGinnis and Krumlauf, 1992; see also Dolle et al., 1993; Izpisua-Belmonte et al., 1993). A major group of genes which encode transcription factors that are involved in early embryonic development are the homeobox genes, which contain a conserved 183 bp long coding sequence, termed the homeobox (McGinnis and Krumlauf, 1992). Much of the specificity of the homeobox gene products, including their binding to specific DNA sequences, lies within the region encoded by the homeobox, termed the homeodomain (Scott et al., 1989). Classical and molecular genetic studies of the homeobox genes showed that they function coordinately to initiate and maintain developmental decisions within defined spatial and temporal boundaries of the embryo (reviewed in Kessel and Gruss, 1990; Krumlauf, 1993; McGinnis and Krumlauf, 1992).

Numerous homeobox genes are expressed in the central nervous system (CNS) of vertebrates during its development (Krumlauf, 1993; Wilkinson, 1993). Most of the genes in the four HOX complexes are expressed in the hindbrain and in the spinal cord. The relative position of the genes within the HOX complexes correlates with their anteroposterior pattern of expression in the embryo (reviewed in McGinnis and Krumlauf, 1992). The anterior border of the expression of HOX genes is usually sharp and, in the hindbrain, the expression is colocalized with the borders of rhombomeres (Krumlauf, 1993). It was also shown that even before the boundaries of the rhombomeres become visible, the expression of HoxB-1 and rhombomeres identity are already autonomous in the hindbrain (Guthrie et al., 1992).

The homeobox gene, HbxE (previously CHoxE; Rangini et al., 1991) and genes of the Pax gene family (reviewed in Gruss and Walther, 1992) are transcription factors that are probably involved in the differentiation of the hindbrain and spinal cord along the dorsoventral axis at somewhat later stages of development. In the case of the Pax genes it was shown that they respond to signals that come from the notochord (Goulding et al., 1993).

Several additional chicken homeobox genes that are expressed in the neural tube have been cloned (reviewed in Fainsod and Gruenbaum, 1994). One such gene, which is not included in the chicken HOX complexes is Sax1 (previously CHox3; Rangini et al., 1989). Sax1 is conserved in evolution, since the Drosophila NK1 (Kim and Nirenberg, 1989) and the honeybee HHO genes (Waldorf et al., 1989) contain a highly homologous homeodomain.

The temporal pattern of expression of Sax1 was studied by developmental RNA blot analysis (Rangini et al., 1989). Sax1 codes for five transcripts, which are probably the result of alternative splicing, since different genomic probes identify different subsets of transcripts. These Sax1 transcripts were detected from the time the egg is laid until 5 days of development. In the present report, we show that Sax1 is exclusively expressed in the spinal part of the neural plate at a relatively early stages. Transcripts of Sax1 were found to become spatially localized at stage 8 ⁻ and their anterior limit marks the border between the spinal part of the neural plate and the future hindbrain. At later stages, the anterior border of expression gradually shifts continuously in a caudal direction at the same pace as the segregation of the somites from the mesodermal plate. However, once the spinal part of the neural plate is induced, Sax1 transcription is autonomously regulated within it.

Commercial Yarkon Tint hens’ eggs were incubated at 38°C in a humidified incubator. Embryos were isolated in Ringer’s solution, staged according to Hamburger and Hamilton (1951) and transferred onto fresh vitelline membranes which were isolated from unincubated eggs and stretched around glass rings (New, 1955). Each glass ring encompassing the embryos was put in a small Petri dish containing some solid egg albumen. The various microsurgical manipulations were performed in these dishes after a 1-2 minutes exposure of the embryo to 0.15% trypsin in PBS. After the operation, the medium was aspirated and the dishes were incubated at 38°C for the desired length of time.

Embryos of stages 8-10 H&H were placed ventral side up. An endodermal flap, covering the most posterior three somites of one side, was carefully peeled off and folded sidewards and the exposed somites were cut out and discarded. The endodermal flap was then put back into its place and the embryos were incubated at 38°C for additional 6 hours.

Embryos of stages 8-11 H&H were placed ventral side up. The endoderm caudal to the last somite was carefully peeled from the segmental plate on one side and pushed sidewards. A section of the segmental plate either corresponding in length to three already formed posterior somites (approximately 300 μm) or including all the non-segmental tissue until Hensen’s node was isolated and discarded. After the operation, the endoderm was put back on the operated area and the embryos were incubated at 38°C for 5.5-6.75 hours.

Pairs of embryos, one of stage 7-9 H&H and the other of stage 9-11 H&H were placed on the same vitelline membrane ventral side up. The younger embryo served as the donor of the notochord, while the older embryo was the recipient. The posterior portion of the notochord from behind the youngest somite down to Hensen’s node, was isolated and transferred to the older embryo. In the recipient embryo, a longtitudal incision was made in the endoderm and then deepened to separate the notochord from the segmental plate. The younger notochord fragment was inserted into the slit, while care was taken to keep its original orientation when transplanted adjacent to the host’s notochord and neural palte. The embryos were incubated at 38°C for additional 5-6 hours.

Embryos at stage 4⁺ H&H were used. The embryos were placed ventral side up and Hensen’s node was cut out and discarded. The operated embryos were incubated at 38°C for 24 hours. About 20% of the embryos did not regenerate their normal Hensen’s node and were recognized by having only one row of large median somites underneath most or all the neural plate. A similar phenomena was described in spontaneously occurring notochord-less embryos (Stern and Bellairs, 1984). These embryos were taken for further analyses.

Embryos of stages 8-12 H&H were placed dorsal side upwards. The right lateral half of the neural plate, caudal to the last somite, was isolated with the aid of a micro-knife (15° angle). The isolated fragment corresponded in length either to three somites (approximately 300 μm) or to the entire section between the last somite and Hensen’s node. The isolated neural plate fragment was transferred to the anterior extraembryonic area, lateral to the forebrain and inserted into the middle layer underneath the ectoderm.

In situ hybridizations with ³⁵S-labeled probes were performed exactly as described in Rangini et al. (1991). Whole-mount in situ hybridization was performed according to the method of Wilkinson (1992) with some modifications. Chick embryos were dissected out in PBS and fixed for 2-20 hours in 4% paraformaldehyde in PBS, washed twice in PBT (PBS; 0.1% Tween 20), dehydrated for 5 minutes each with 25%, 50%, 75% methanol in PBT and twice with 100% methanol. Embryos were then stored at –20°C. Rehydration of the embryos was done by transferring them through the same methanol series in reverse order followed by 2 washes with PBT. Bleaching was performed with 6% hydrogen peroxide in PBT for 1 hour, followed by three washes with PBT. The embryos were digested with 10 μg/ml of Proteinase K in PBT containing 0.05% SDS for 6-15 minutes, depending on the developmental stage of the embryo. After washing once with freshly prepared 2 mg/ml glycine in PBT and twice with PBT, the embryos were refixed in fresh 0.2% glutaraldehyde and 4% paraformaldehyde in PBT for 20 minutes and washed again with PBT for two more times. Prehybridization was in 1 ml of 50% formamide, 5× SSC pH 4.5, 1% SDS, 50 μg/ml yeast RNA and 50 μg/ml heparin at 70°C for 1 hour. Digoxigenin-labeled RNA probes were transcribed from Sax1 templates (Fig. 1A) in Bluescript vectors using T7 or T3 RNA polymerases. The labeled probes were added to fresh prehybridization buffer at a final concentration of ∼1 μg/ml and incubated with the embryos overnight at 70°C. Following hybridization, the embryos were twice washed for 30 minutes at 70°C in 50% formamide, 5× SSC pH 4.5, 1% SDS, once in a 1:1 mixture of the former solution with 0.5 M NaCl, 10 mM Tris-HCl pH 7.5, 0.1% Tween 20 and three times in the later solution. Embryos were then incubated twice, 15 minutes each, with 100 μg/ml of RNase A in 0.5 M NaCl, 10 mM Tris-HCl pH 7.5, 0.1% Tween 20 at 37°C, followed by a wash with the same buffer without the enzyme, then three washes with 50% formamide, 2× SSC pH 4.5, the first at ∼22°C and the other two at 65°C, which were followed by three washes with TBST (0.14 M NaCl, 2.7 mM KCl, 25 mM Tris-HCl pH 7.5, 0.1% Tween 20 and 2 mM Levamisole). Blocking of non-specific binding of antibodies was in 10% lamb serum in TBST for 90 minutes. Anti-digoxigenin alkaline phosphatase-conjugated antibodies (Fab’) were preabsorbed for 1 hour with powder of chick embryos (precipitated with acetone, 3 mg/μl of 150 units/μl antibody) in TBST containing 1% lamb serum. Embryos were then incubated overnight in 1:2,000 dilution of the antibody in TBST containing 1% lamb serum at 4°C. After three washes, 5 minutes each, and five washes, 10 minutes each, with TBST they were washed three times for 10 minutes with NTMT (0.1 M NaCl, 0.1 M Tris-HCl pH 9.5, 50 mM MgCl₂ 0.1% Tween 20, 2 mM Levamisole). The solution was then changed into NTMT containing 0.33 mg/ml nitro-blue tetrazolium salt (NBT) and 0.17 mg/ml 5bromo-4-chloro-3-indolyl-phosphate (BCIP) and the embryo was then incubated for 90-120 minutes in the dark. The reaction was monitored microscopically and stopped by two washes with PBS, fixation in 4% paraformaldehyde for 1 hour, three more washes in PBS, one wash in PBS containing 50% glycerol and one with PBS containing 80% glycerol. The embryos were mounted in 80% glycerol in PBT and photographed using a Zeiss Stemi SVII stereoscope and a contax 167 MT camera.

Some of the embryos were then washed three times, 10 minutes each, with PBS, embedded in 7% gelatin and fixed overnight in 4% paraformaldehyde in PBS. The embryos were then sectioned, using a vibratome (series 1,000). The 70 μm-thick sections were placed on gelatin-coated slides, mounted with entellan and photographed using a Zeiss Axiovert 135TV microscope.

Developmental northern analyses revealed that Sax1 transcripts are already present at stage 2 H&H of embryonic development (Rangini et al., 1989). In order to localize Sax1 transcripts spatially during the stages of gastrulation and early neurulation, in situ hybridization analysis was performed on whole-mounted as well as on serially sectioned embryos of stages 4-7 H&H, using Sax1 probes A or B (Fig. 1A). The results of these experiments revealed that at the above stages there was no localized signal of hybridization (data not shown).

In situ hybridization of embryos at stages 8^– to 15 H&H (Fig. 2) revealed an anteroposteriorly regressing localization of Sax1 transcripts along the spinal section of the neural plate. The most anterior border of the hybridization signal at stage 8^– was approximately at the level of the 4th somite, which marks the border between the future brain and the spinal cord (Fig. 2A). Since that stage onwards, the anterior border of Sax1 transcripts in the neural plate continuously regressed in a posterior direction in a spatial and temporal correlation with the segregation of somites from the mesodermal plate, and was always at the same plane as the border line between the youngest somite and the mesodermal plate (Figs 1B, 2). The posterior border of Sax1 transcripts continuously shifted backwards with the posterior elongation of the neural plate and always coincided with the posterior end of the neural plate. No hybridization signal was observed when the in situ hybridization was performed on control embryos using the sense strand of either probes A or B (data not shown).

In order to verify the spatial expression of Sax1, in situ hybridization was performed on frontal sections of stages 1011 H&H embryos, using a [³⁵S]-labeled probe (Fig. 3A,B). The results of these experiments revealed, again, that the anterior border of Sax1 expression is roughly in the same transverse plane as the posterior border of the youngest somite. A control probe of ³⁵S-labeled Sax1-sense RNA did not show hybridization signal to neighboring sections (data not shown).

The in situ hybridization to both whole-mounted embryos (Figs 1B, 2) and frontal sections (Fig. 3A,B) also revealed that there is probably a quantitative posteroanterior gradient of the amounts of Sax1 transcripts in the neural system. The strongest signal was observed posteriorly while its intensity declined toward the anterior border of expression. The spatial localization of the various Sax1 transcripts seems to be similar, since probe B (Fig. 1A), which lacks the homeobox and recognizes the 1.9 kb transcript, and probe A (Fig. 1A), which contains the homeobox and recognizes all five Sax1 transcripts (Rangini et al., 1989), both showed a similar spatial pattern of hybridization. However, in most experiments, the intensity of the in situ hybridization signal was higher when probe A was used.

In order to study the pattern of Sax1 expression along the dorsoventral axis of the embryo, the whole-mounted hybridized embryos were serially cross-sectioned at 70 μm. Examination of these sections revealed that the intensity of hybridization of the probes to Sax1 transcripts was roughly uniform in the neural cells along the dorsoventral axis. No hybridization signal was detected in the neural folds, which are to form the future neural crest (Fig. 4). In some cases, the signal seemed to be stronger either at the dorsolateral or the ventral section of the neural tube, but it might have been a result of oblique sectioning of the embryos, with a weaker expression representing a slightly more anterior part of the tube. To support these observations further, stage 12 H&H embryos were first serially sectioned and then in situ hybridized with a ³⁵S-labeled RNA probe. The results of these experiments again demonstrated that Sax1 transcripts were restricted to the neural plate cells with a more or less uniform hybridization signal along the dorsoventral axis, which was restricted from the neural crest region (Fig. 3C,D). Similar experiments with stage 8 H&H or stage 15 H&H embryos gave identical results (data not shown). Control probes A or B of ³⁵S-labeled Sax1-sense RNA did not give a signal when hybridized to neighboring sections (data not shown).

Sax1 pattern of expression in the neural plate correlates very nicely with the dynamics of the segregation of somites from the mesodermal plates. We therefore decided to check experimentally a possible correlation between the differentiation processes in the neural system and its neighboring mesodermal structures, such as the posterior somites, the non-segmented mesodermal plate and the notochord. We have therefore changed the spatial relations between the neural tube and each of the above structures and checked if and how this would affect Sax1 pattern of expression.

In the first set of experiments (n=5), we wanted to determine whether the unilateral removal of somites will cause the neighboring region of the neural plate to maintain the expression the Sax1 gene. Three newly formed somites were unilaterally removed from the vicinity of the neural plate and the embryos were further incubated for additional 3 to 6 hours, during which 2-3 new more posterior somites were added on both sides of the neural tube (Fig. 5A). The results of these experiments revealed that Sax1 expression remained equal on both the operated and the control non-operated sides of the nervous system (Fig. 5B). Thus, the removal of somites did not affect the disappearance of Sax1 expression in the operated side. In a second set of experiments (n=10), we tried to avoid any possible contact between the neural plate and defined somites. Thus, we removed unilaterally sections of the mesodermal plate posterior to the already formed somites corresponding in length either to future 3-4 somites (Fig. 5C,D) or comprising most of the mesodermal plate (Fig. 5C,E). After 6 hours of incubation, the distribution of Sax1 transcripts proved again to be identical on both sides of the neural system. Thus, the removal of the mesodermal plate and the lack of somite formation on the operated side did not affect the disappearance of Sax1 expression in the corresponding region of the neural tube.

Another tissue that we assumed may be involved in the regulation of Sax1 expression in the spinal cord is the notochord, which is known as an inducer of the neural tube (Hara, 1961; Izpisua-Belmonte, 1993; Yamada et al., 1993).

In a set of experiments (n=6), posterior pieces of a newly formed notochord from stages 7-9 H&H embryos were implanted into embryos at stages 9-11 H&H (Fig. 6A,B). In order to check whether the young notochord will prolong the expression period of Sax1 on the operated side, the notochord fragment was placed unilaterally adjacent to the neural plate at the transition area between the last somite and the mesodermal plate. The embryos were then incubated for about 6 hours and processed for whole-mount in situ hybridization, using a Sax1 probe. The results of the in situ analysis revealed that the spatial distribution of Sax1 transcripts was identical on both sides of the neural tube (Fig. 6C). Thus, we conclude that at the time of the operation Sax1 pattern of expression was already determined and the interaction with a young notochord did not affect it.

In the above experiments with the implanted young notochord, the neural system continued to be under the influence of the original notochord or its precursor cells. Therefore, we decided to test whether the picture would be altered by avoiding any contact between the induced neural tissue and the notochord during its entire differentiation.

In a complementary set of manipulations (n=3), the region of the Hensen’s node, from which the notochord develops was removed from stage 4⁺ H&H embryos (Fig. 7A), in which the induction of the neural tissue is known to already have taken place (Selleck and Stern, 1991, 1993). The embryos were further incubated until they reached stages 9-11 H&H. Notochord-less embryos were formed which contained only a single row of median somites underneath the neural plate (Fig. 7B). The pattern of Sax1 expression was found not to be influenced by the total lack of notochord (Fig. 7C-F). Thus we conclude that neither the absence nor the presence of a notochord affect Sax1 pattern of expression.

All the above experiments point to the possibility that Sax1 expression is autonomously regulated within the spinal section of the neural plate. To verify this possibility, longtitudal stripes of one half of the neural plate from the region that expresses Sax1 were cut out from embryos at stages 8-12 H&H and transplanted into the anterior extraembryonic region, far from conventional neighboring structures, while the other half of the neural plate was left in place (Fig. 8A,D). Some of the neural plate corresponded in length to future three somites while others included the entire length of the neural plate. In some of the above experiments the transplant was inserted into the extraembryonic region in an anteroposterior orientation conforming with the original orientation of the embryo (Fig. 8A), while in others it was inserted in the reverse orientation (Fig. 8D). After incubation periods of 2.5-6 hours, the spatial distribution of Sax1 transcripts in the transplant was compared to that of the identical intact part of the neural plate.

In all our experiments (n=12), both kinds of transplants proved to have the same pattern of Sax1 expression as the non-manipulated side of the neural plate (Fig. 8B-C, E-F). After a longer incubation period of 16-20 hours, the Sax1 signal disappeared altogether from the transplants as well as from both the corresponding non-manipulated region (data not shown).

Most vertebrate homeobox genes are expressed in more then one germ layer (reviewed in Kessel and Gruss, 1990; McGinnis and Krumlauf, 1992; Krumlauf, 1993; Fainsod and Gruenbaum, 1994). Sax1 is unique among those genes since it is currently the only known homeobox gene which during stages 8^–-15 H&H is exclusively expressed in a defined ectodermal section — the spinal part of the neural plate. Another gene that is specifically expressed in the neural tube is HbxE (Rangini et al., 1991). However, unlike Sax1, HbxE is expressed in the intermediate plate of the neural tube at later stages of development and is probably involved in differentiation events along the dorsoventral axis of the neural tube.

At stage 8^–,Sax1 is first localized and its anterior border is behind the 4th visible somite. The neural plate section above the 4th somite, which will become a part of the hindbrain (Jacobson, 1992), does not express Sax1. The timing, as well as the pattern of expression of Sax1 in the spinal region, resembles those of Hox B5, which is expressed in mice at a similar developmental stage (3-4 somites), starting at the border between head and trunk and spreading backwards (Wilkinson et al., 1989). Wilkinson et al. (1989) suggested that this limit of expression may correlate with a transition in CNS organization, which may apply also to the limits of expression of Sax1. However, in contrast to Hox B5, Sax1 is the only known gene with an anterior border that continuously regresses in a posterior direction and roughly coincides with the plane where the youngest somite has just segregated from the mesodermal plate (Figs 1, 2). The posterior border of Sax1 transcripts always coincides with the posterior border of the neural plate merging into the primitive streak.

Developmental northern analysis has shown that Sax1 is already expressed in blastoderms at the early primitive streak and head fold formation stages, namely stages 2-5 H&H (Rangini et al., 1989). However, Sax1 transcripts could be spatially localized by in situ hybridization from stage 8^– H&H and on, after the first 3-4 pairs of somites have segregated from the mesodermal plate. Since the in situ hybridization technique gives a clear signal only when enough transcript(s) are localized in a spatially restricted manner, it is likely that prior to stage 8^– H&H either most or all cells in the blastoderm express low amounts of Sax1 or that the levels of the transcripts in specific cells are below the limits of the sensitivity of the technique.

Looking for a connection between Sax1 expression and its contemporary morphogenetic events, we sought a correlation with the following events: the closure of the neural plate into a tube, the segregation of the somites from the mesodermal plate, and the process of formation and maturation of the notochord. The experiments in this study were planned to address each of the mentioned possibilities except for the first one, as it became clear that even in normal embryos a signal could be detected in the closed section of the neural tube in stages 14-15 H&H, while in embryos of stages 8-9 H&H, a section of an open neural plate was found to have lost its signal (data not shown).

The timing and the mechanism of the induction of the posterior neural plate in avian embryos are not yet entirely clear. Storey et al. (1992) have suggested two alternative mechanisms by which the more posterior regions of the CNS could be generated: (i) homeogenetic induction by the anterior sections of the neural plate which have already been induced, spreading posteriorly in the epiblast, and (ii) elongation of posterior neural primodia that have been induced at a very early stage, and which are probably confined into a relatively small ectodermal area anterior and lateral to Hensen’s node (Schoenwolf, 1991). Whatever the mechanism might be, it is clear that from stage 5 H&H and on, the formation of the spinal part of the neural plate, as demonstrated by Sax1 expression, does not require the presence of either normal Hensen’s node or notochord. This is why in either case the removal of the node, which results in partially notochord-less embryos, does not interfere either with the formation of the spinal section of the neural plate, or with its extension above the single row of well as those in which a young notochord was implanted near medial somites, all the way down to the defective streak. In the signal transition area, all point to the same conclusion. The other words, the expression of Sax1 seems to be already deter-gradual disappearance of the hybridization signal from the mined simultaneously with the induction of the spinal region anterior towards the posterior end of the spinal neural plate, of the neural plate, after which it is not affected by the probably exhibits the temporal gradient of differentiation notochord. The experiments in which the neural plate was which is initially reached in the anterior section of the develremoved from the influence of either notochord or somites as oping spinal cord and gradually shifts posteriorly.

Sax1 seems to be more or less expressed in all the cells of a given transverse plane of the neural plate but not in the neural folds, which are the primodia of the neural crest. In contrast, other genes that encode for transcription factors such as HbxE and the Pax genes, which are also expressed in the spinal cord of early embryos, have a restricted dorsoventral pattern of expression. Experimental evidence points to the impact that the notochord has on the expression of the Pax genes (Goulding et al., 1993). Thus, the remarkable differences between Sax1 and Pax genes is that the expression of Sax1 in the spinal cord is determined earlier, is not limited to a certain section of the spinal cord and is probably related to an initial induction or determination of the spinal part of the neural plate. This state of determination, as signaled by Sax1 expression, spreads from the anterior end of the spinal neural plate posteriorly and is probably related to the physiological age of the tissue. In contrast, the expression of the Pax genes, which is restricted along the dorsoventral axis, is somewhat later in development and is related to differentiation processes along the dorsoventral axis within the previously determined spinal cord.

Another prominent morphogenetic change in the embryo that occurs at the period studied is the gradual anteroposterior formation of somites. The dynamics of Sax1 expression follows that of somite formation. Since induction signals between neighboring tissues are common during early embryogenesis, we looked for a possible connection between the neural plate and the forming somites in the form of a signal that might coordinate the differentiation rhythms of the two. The results of our experiments, which included the removal of posterior somites, removal of mesodermal plate or transplantation of sections of the neural plate into the extraembryonic area, where it had neither contact with the somites nor with the mesodermal plate, showed that Sax1 pattern of expression is independent of the above tissues.

Although we can not rule out the possibility of just a coincidence, the results presented here suggest that in the avian embryo there is a developmental clock, yet unknown, which normally determines and coordinates the level of differentiation of all the tissues in a given transverse plane at a given time.

We are grateful to Dr C. Kalcheim for her advise and help in demonstrating some of the techniques of embryonic manipulations. We also thank Dr C. D. Stern for advise on the removal of Hensen’s node. This work was supported by a grant from the Israeli Academy for Arts and Sciences to H. E. G and Y. G.