In the S&Z gastrulation model, the organizer and the prospective neuroectoderm are originally established in the blastocoel roof of blastula embryos, and these embryonic regions move downward to get into contact with each other by their inner surfaces at the dorsal marginal zone in the ACE embryo (Fig. 1). In this study, we found that the dorsal one-third of the marginal zone at ACE is sufficient for the construction of the complete dorsal structure, indicating that this small embryonic region should contain not only the organizer and the prospective neuroectoderm but also the precursors of somitic mesoderm, neural crest, and other dorsal components. Since organizer marker genes such as goosecoid and chordin have been shown to be expressed in around the dorsal one-third of the marginal zone at early gastrula (Cho et al. 1991; Sasai et al. 1994), the entire organizer tissue may be necessary for the formation of the complete dorsal structure. In addition, when the mGFP-labeled dorsal one-third of the marginal zone at ACE was transplanted into the ventral side of a non-labeled ACE embryo, almost all the structures of the secondary body axis were labeled (Fig. 4). According to the conventional gastrulation model, the organizer migrates on the inner surface of the blastocoel roof and induces neural fate, which should result in the construction of neural structures by non-labeled host cells in the transplantation assay. However, the current result showed that not only the organizer tissues but also the neural and neural crest derivatives are constructed by the transplants, indicating that this secondary axial structure is not the result of induction by the Spemann organizer (Spemann and Mangold 1924) but rather is already determined within the transplanted dorsal one-third of the marginal zone at ACE as described in the S&Z model (Fig. 1; Yanagi et al. 2015). In addition, the head and tail of the secondary axis were constructed from the transplanted dorsal one-third of the marginal zone, but the trunk region was formed by both transplant and host cells (Fig. 4, Suppl. Fig. 1). Therefore, it is suggested that the head and tail structures are constructed from the tissues within the dorsal one-third of the marginal zone, and the elongated trunk is constructed from both the dorsal one-third of the marginal zone and the neighboring marginal tissues by convergent extension movement.

It is noteworthy that the Keller sandwich explant constructed from ACE embryos developed the organized neural tissues expressing otx2 and krox20 in order (Koide et al. 2002). Since the Keller sandwich is the back-to-back conjugate of dorsal ectodermal tissues (Keller and Danilchik 1988), it contains the prospective neuroectoderm but not the underlying organizer or other mesodermal/endodermal tissues, in principle. Therefore, it is suggested that the neural domains such as forebrain, midbrain, and hindbrain have already been specified within the prospective neuroectoderm by the time of ACE. However, the Keller sandwich can never form more complicated and detailed neural structures, indicating that the construction of the complete dorsal structures requires the organizer even after ACE for purposes other than the induction and the initial patterning of neural tissues. For example, the formation of two separate eyes requires the underlying axial mesoderm (Li et al. 1997), and the prechordal mesoderm is known to be necessary for hypothalamus formation (Yamaguti et al. 2005; Xie and Dorsky 2017).

When the blastocoel roof of the blastula was explanted, the wound healed quickly and the explant underwent gastrulation movement and eventually developed an embryo lacking only a yolk mass (Fig. 5). Consistent with the current results, it was reported that Xenopus fertilized eggs in which 40% of the vegetal cytoplasm was removed after the cortical rotation could develop into embryos lacking a yolk mass (Sakai 1996). Collectively, all of these findings suggest that the development of the complete dorsal structure does not require the vegetal part of the embryo before gastrulation in Xenopus. The wound-healed blastula blastocoel roof explant may show a hollow globular structure which is similar to the blastula of a basal chordate (amphioxus). The amphioxus blastula forms the blastopore, and the presumptive mesoderm which expresses the organizer genes such as goosecoid and chordin undergoes simple invagination with little involution to make a physical contact with the ectodermal region between their respective inner surfaces (Fig. 6a; Zhang et al. 1997; Yu et al. 2007). At this standpoint, the gastrulation movement of amphioxus appears to be equivalent to the amphibian S&Z movement. Thus, the wound-healed explant of the blastocoel roof of Xenopus blastula might mimic the situation of the amphioxus blastula, and if so, it could be thought that the gastrulation movement is conserved from amphioxus to amphibians. One of the differences between blastula embryos of amphioxus and amphibians is thought to be the amount of yolk. The vegetal side of the Xenopus oocyte has numerous yolk platelets, which are highly conserved membrane-bound organelles storing yolk proteins such as the derivatives of Vitellogenin (Danilchik and Gerhart 1987), and the yolk platelets segregate to the cleaved blastomeres resulting in the larger amount of yolk platelets in the vegetal blastomeres. In addition, the developing gut of amphibians, which is derived from the vegetal blastomere, consumes the numerous yolk platelets to feed the embryo like a yolk sac in later developmental stages (Jorgensen et al. 2009). Therefore, it could be thought that the vegetal half of the amphibian blastula corresponds to amniote yolk (Fig. 6c). On the other hand, the amphioxus embryo contains only a small amount of yolk (Fig. 6a). Therefore, it could be thought that the vertebrate precursors increase the amount of yolk on the vegetal side of the embryo to store more nutrients for development (Fig. 6b). From this point of view, the Xenopus embryo might be regarded as the embryo of a basal chordate with a certain amount of yolk in its vegetal side but the S&Z gastrulation movement is basically conserved. Based on this idea, the endoderm-less ACE embryo and the blastocoel roof explant from the blastula in this study could be regarded as essentially the same: the difference between the two operated embryos is whether the yolk was removed before or after ACE, and both contain the regions required for the formation of the complete dorsal structure. Therefore, these operated embryos developed similar tailbud embryos lacking a yolk mass (Figs. 3 and 5).

S&Z movement is conserved, whereas the amount and the location of yolk are varied, among chordates. a Amphioxus shows a hollow globular embryo which possesses only a small amount of yolk, but ACE has occurred by the simple invagination with little involution, which corresponds to S&Z movement. Thereafter, the A-P axis is elongated posteriorly. b Though vertebrate precursor might acquire the yolk on the vegetal side of the embryo, the gastrulation movement could be conserved. c Amphibians such as Xenopus laevis also store a certain amount of yolk within their dividing vegetal hemisphere. d Odorrana supranarina has a larger amount of yolk, which impedes cleavage, but the gastrulation movement, in which ACE occurs at the dorsal marginal zone and elongates the A-P axis posteriorly, is indeed confirmed. e In avian development such as that of Gallus gallus, the anterior end of the primitive streak regresses after the streak reaches its full length. The regression might lead to physical contact between the organizer and the prospective neuroectoderm, which could be referred to as the head process. The morphogenetic movement is similar to S&Z movement. Red region, the organizer. Blue region, the prospective neuroectoderm. Orange, yolk

While the eggs of basal chordates and amphibians undergo holoblastic cleavage, the eggs of amniotes such as birds and reptiles show meroblastic cleavage because they harbor a huge amount of condensed yolk which impedes cleavages. Therefore, it is difficult to compare the gastrulation movement between amphibians and amniotes directly. However, the egg of Odorrana supranarina, which is a species of frog inhabiting Ishigaki Island and Iriomote Island in Japan, has a condensed yolk and shows meroblastic cleavage on the animal pole side similar to avian meroblastic discoidal cleavage. Therefore, O. supranarina could be considered an intermediate species between general amphibians and amniotes. When the fertilized egg was embedded into gelatin to prevent the rotation following the shift of the center of gravity as previously described (Yanagi et al. 2015), the dorsal structures were formed on the vegetal side in O. supranarina (Supplementary movie 5), as in Xenopus embryos. When the cortical rotation was prevented by treatments such as UV irradiation or nocodazole treatment, the dorsal structure could not be constructed (Gerhart et al 1989). Therefore, the cortical rotation was not disturbed by embedding into gelatin, and the elongation of the body axis to the vegetal side of the embryo could be considered a physiological and not an artificial process, which is conserved among amphibians, even in O. supranarina. Thus, it is suggested that the amphibian embryos showing meroblastic cleavage also follow the S&Z gastrulation model even though these amphibians have an apparent condensed yolk in their eggs (Fig. 6d). From this perspective, we also think that a movement similar to the amphibian S&Z movement can be found in the chick developmental process. After the primitive streak has reached its full length, the anterior end (Hensen’s node) begins to regress, leaving in its trail a structure commonly referred to as the head process (Nicolet 1971; Wakely and England 1979). Hence, it could be thought that the prospective head neural tissue, which is localized anterior to Hensen’s node, also moves posteriorly with the regression and contributes to the formation of the head process (Fig. 6e). If so, these relative movements to form the head process look like the S&Z movement of amphibian gastrulation. Therefore, from the comparison of the gastrulation movement of an amphibian to that of a basal chordate and an amniote, it is suggested that the fundamental manners of gastrulation are conserved among chordates though individual processes are different from species to species: the prospective neuroectoderm and the organizer make a physical contact by their inner surfaces and elongate the A-P axis posteriorly (Fig. 6).

Wnt/β-catenin signaling is an evolutionarily conserved signaling pathway regulating numerous biological processes (Clevers and Nusse 2012). When the Xenopus egg is fertilized, the cortical rotation occurs and the maternal dorsal determinants including Disheveled (Dsh) and GSK3-binding protein (GBP) are actively transported to the presumptive dorsal region along the microtubule bundles by binding to kinesin (Weaver et al. 2003). The translocated dorsal determinants then lead to the nuclear accumulation of β-catenin, which induces gene expression including expression of the Nodal-related factors (McKendry et al. 1997; Takahashi et al. 2000). Of note, the maternal transmembrane protein Huluwa, which also localizes at the vegetal pole of the oocyte and dorsally translocates upon the cortical rotation, was recently reported to induce the nuclear accumulation of β-catenin at the dorsal region in a Wnt protein-independent manner (Yan et al. 2018). Furthermore, macropinocytosis and the formation of lysosomes are suggested to be required for β-catenin signaling in the dorsal region of Xenopus early embryos (Tejeda-Muñoz and De Robertis 2022). The Spemann organizer has been reported to be established by the cooperation of β-catenin and Nodal signaling, which is conserved among the early chordate embryos (Kozmikova and Kozmik 2020). Consistent with these reports, the nuclear accumulation of β-catenin at the dorsal blastocoel roof of the early blastula has been reported (Schohl and Fagotto 2002). Therefore, we suggest that Dsh, GBP, Hwa and other molecules translocate to the dorsal region of the embryo upon the cortical rotation to lead to the establishment of the organizer at the blastocoel roof of the blastula, and then, it moves downward to the dorsal marginal zone of the early gastrula.

In the early chick development, the midgut remains open and connects to the yolk sac, while the dorsal structures including head, trunk, and tail are definitely constructed from the epiblast. As described above, in Xenopus development, the vegetal half of the blastula, which contributes to the gut, corresponds to the yolk of amniotes, and the blastocoel roof of the blastula develops the complete dorsal structure. In addition, it is known that the Xenopus animal cap cells can be converted into different types of cells such as endodermal, mesodermal, epithelial, and neural cells by several growth factors (Asashima et al. 1990; Ariizumi et al. 1991; Kengaku and Okamoto 1993). Therefore, though the animal cap of the Xenopus blastula is frequently referred to as undifferentiated ectoderm, it could be equivalent to the epiblast in amniotes rather than ectoderm. Thus, we proposed that all chordate embryos could be divided into two fundamental portions: (1) the embryonic tissues, which are the blastocoel roof of the blastula in amphibians and the epiblast in amniotes, and (2) the extraembryonic tissues acquired by increasing the amount of yolk in the evolution of chordates, which are the vegetal half of the blastula in amphibians and the yolk in amniotes.