Germ layers are defined during embryogenesis and identify groups of interacting cells that give rise to distinct tissues and structures. Developmental biology textbooks contrast the diploblastic Cnidaria, composed of ectodermal and endodermal germ layers, with triploblastic bilaterians, possessing an additional mesodermal layer. The search for a unifying theory explaining the evolution of animal germ layers, their link to gastrulation and their derivatives has been a source of controversy for over a century. The release of three preprints (Haillot et al., 2024 preprint; Lebedeva et al., 2022 preprint; Sun et al., 2024 preprint) has provided new elements to the debate, offering insights into the mechanisms that govern germ layer formation in the sea anemone Nematostella vectensis.
The traditionally accepted correspondence between cnidarian and bilaterian germ layers has been called into question, based on molecular and fate-mapping data (Steinmetz et al., 2017). Building essentially on findings regarding the anthozoan model system Nematostella vectensis, which gastrulates by invagination and develops a defined pharynx, Steinmetz and colleagues proposed that the territories called ‘endoderm’ in Cnidaria and Bilateria are, in fact, not homologous (Steinmetz et al., 2017). Instead, the cnidarian innermost layer would correspond to the mesoderm of bilaterians, while the bilaterian endoderm would be homologous to the portion of the ectoderm entering the gastric cavity and forming the pharynx. Several key observations support this hypothesis: the inner layer of Nematostella embryos displays a molecular profile similar to that of the bilaterian mesoderm and gives rise to typical mesodermal derivatives such as muscles, gonads and nutrient-storage organs. In parallel, the ectodermal territory of the pharynx (pharyngeal ectoderm) expresses several conserved endodermal markers, and its derivatives include digestive gland cells and insulin-secreting cells, both of which are characteristic of endoderm in Bilateria.
Several studies have shown that canonical Wnt signalling (cWnt), notably mediated by nuclear localisation of β-catenin, is essential for patterning and gastrulation, and that several players contribute to the patterning of the Nematostella embryo, including ERK/MAPK signalling (reviewed by Technau, 2020). In a recent preprint, Haillot and colleagues further investigate the early regulatory cascade leading to the definition of germ layers in Nematostella (Haillot et al., 2024 preprint). Through functional manipulations of cWnt and MAPK signalling components, Haillot and colleagues show that these signalling pathways act antagonistically at the blastula stage: nuclear localisation of β-catenin leads to ectodermal cell fates, while ERK/MAPK signalling is responsible for the specification of the future innermost layer and for the morphogenetic processes leading to gastrulation. Simultaneous inhibition of cWnt and MAPK signalling generates aboralised embryos, composed only of ectodermal tissue. These experiments support the idea that the default cell fate in Nematostella is ectoderm, which aligns with previous findings from blastomere isolation studies (Leclère et al., 2016) and parallels what had been observed in bilaterians (e.g. Darras et al., 2011).
Haillot and colleagues further show that cell fate decisions are precocious and, already at 6 h from fertilisation, a domain of cells can be distinguished transcriptionally (e.g. pitx1-like, duxABC) at the animal-most pole (future oral pole); the same cells will shortly after express ERK signalling components (e.g. erg, fgfa1) (Haillot et al., 2024 preprint). The initial specification of this erg territory, devoid of nuclear β-catenin, likely depends on a still unknown signal. After gastrulation, these cells make up the innermost layer. The blastoporal ring of cells expressing foxA and brachyury, which are markers of endoderm and gut opening in bilaterians, emerge later as a one-cell-wide domain (10-12 h of development) and undergo an expansion shortly before gastrulation. Functional manipulations of Notch/Delta signalling, which in bilaterians participates in the definitions of germ layers, show that this foxA domain is derived from ectodermal cells in contact with the animal-most domain of cells, expressing paralogues of the delta ligands. Altogether, these findings provide a mechanistic understanding of the new germ layer hypothesis proposed by Steinmetz et al. (2017).
When exploring homologies between the germ layers of cnidarians and bilaterians, a longstanding challenge has been the observation that gastrulation occurs on opposite sides, relative to egg polarity (Martindale and Hejnol, 2009): from the animal-derived side (side of the oocyte where the female nucleus is localised, marked by the extrusion of polar bodies) in cnidarians and from the vegetal-derived side in bilaterians. It has been assumed that β-catenin is active at opposite poles in cnidarians and bilaterians, with the inner, orally derived layer of Nematostella being controlled by cWnt signalling (e.g. Lee et al., 2007; Wikramanayake et al., 2003). Supporting this hypothesis, the restriction of nuclear β-catenin along the animal-vegetal axis of early Nematostella embryos has been shown to be regulated by differential β-catenin degradation along the primary body axis (Wikramanayake et al., 2003) and by Dishevelled, a signalling protein that acts downstream of Frizzled receptors. In fact, Dishevelled is maternally deposited around the female pronucleus, and remains localised at the animal pole during cleavage and embryogenesis (Lee et al., 2007).
Such a role of cWnt signalling in the formation of the innermost layer in Nematostella has, however, been called into question by several more recent studies, including Haillot et al. (2024 preprint), which suggest that this process relies instead on the likely complete downregulation of this pathway. In fact, knockdown of β-catenin results in embryos expressing only ‘mesoderm’ markers (Haillot et al., 2024 preprint; Leclère et al., 2016). A solution to this inconsistency was provided by the work of Lebedeva and colleagues (Lebedeva et al., 2022 preprint). They elegantly demonstrate through live imaging of a β-catenin:GFP reporter line that β-catenin in Nematostella is actually first active in the embryonic domain derived from the vegetal portion of the egg, contrary to earlier expectations that it would be active at the animal-derived pole (Lee et al., 2007; Wikramanayake et al., 2003). Considering that the foxA-expressing cells (‘endoderm’) derive from the ectodermal territory (Haillot et al., 2024 preprint; Steinmetz et al., 2017), and that higher levels of nuclear β-catenin are later found in this domain (Lebedeva et al., 2022 preprint), these data further suggest that, as in bilaterians, the ‘endodermal’ layer in Nematostella develops from territories that are under sustained β-catenin signalling.
The cytoplasmic levels of β-catenin are modulated by the so-called destruction complex, the main components of which – particularly Axin and APC – are broadly conserved. However, the domain composition of Axin proteins differs across species, with the β-catenin-interacting domain absent in the Axin gene of Nematostella but typically present in its bilaterian homologues. These differences suggest that, although the cWnt pathway is conserved between cnidarians and bilaterians, its components may function differently. In a recent preprint, Sun and colleagues functionally investigate the molecular interactions of the destruction complex components Axin and APC with β-catenin (Sun et al., 2024 preprint). They reveal that, despite the structural differences, both can bind β-catenin in Nematostella. Furthermore, Nematostella Axin can modulate cWnt signalling not only in the sea anemone, but also in sea urchin embryos (Sun et al., 2024 preprint). This work has also shown that overexpression of Nematostella Axin and APC (and thus downregulation of cWnt signalling) leads to downregulation of pharyngeal ectoderm markers such as foxA – in line with Haillot et al. (2024 preprint) – but also of genes expressed in the innermost layer, notably a Frizzled receptor (Fz10). At a later stage, Axin overexpression led to disorganisation of the gastrodermis, consistent with previous studies downregulating cWnt signalling components in Nematostella embryos (Lee et al., 2007; Kumburegama et al., 2011; Röttinger et al., 2012).
The current discrepancies regarding the involvement of cWnt signalling in the development of the innermost territory in Nematostella (e.g. Wijesena et al., 2022; Niedermoser et al., 2022) will need to be addressed by assessing the precise roles of the cWnt pathway components throughout gastrodermis development (Kusserow et al., 2005; Kumburegama et al., 2011; Röttinger et al., 2012). It will also be crucial to functionally investigate which maternal components contribute to the very early restriction of cWnt and MAPK signalling into mutually exclusive territories, key to the formation of Nematostella germ layers as shown by Haillot et al. (2024 preprint).
Nematostella research is rewriting our textbooks and reviving the debate about whether cnidarians should be considered triploblastic, at least in some species. Under this framework, the conventional nomenclature of Nematostella germ layers requires re-evaluation. Several studies used the term ‘endomesoderm’ to describe the inner layer of Nematostella (e.g. Wijesena et al., 2017), borrowing the term from bilaterian embryology. The findings of Haillot and colleagues suggest that this terminology is inaccurate because bilaterian endoderm and mesoderm layers most likely evolved from distinct embryonic territories in the cnidarian-bilaterian common ancestor. Haillot et al. (2024 preprint) aimed to simplify the comparisons between Nematostella and bilaterian fate maps by renaming the invaginating inner layer, traditionally known as endoderm, to ‘mesoderm’, and the foxA territory, the previous pharyngeal ectoderm, to ‘endoderm’. However, we have barely started to explore functionally the remarkable diversity of gastrulation modes found across Cnidaria, notably hydrozoans and scyphozoans (Kraus and Markov, 2017): the findings in Nematostella might turn out to be poorly applicable to other species, particularly those which do not gastrulate by invagination. Furthermore, a defined pharynx is not formed in the larvae in all species, even in some that still gastrulate through invagination, such as Cyanea or Aurelia (Jägersten, 1972; Kraus et al., 2022).
Interestingly, the terms ‘ectoderm’ and ‘endoderm’ were first coined by George J. Allman to describe the inner and outer layers of the Cordylophora and Hydra polyps (Allman, 1853a,b), before being applied to bilaterian embryos (reviewed by Oppenheimer, 1940). Given the current state of the debate, it may be more stable to maintain the classical definitions of ectoderm and endoderm for cnidarians. Whether the ectoderm-derived ‘endoderm-like’ foxA domain constitutes a true germ layer and thus qualifies cnidarians as triploblastic remains an open question. Building a comprehensive framework for understanding the evolution of metazoan germ layers will require comparing the patterns emerging from Nematostella research across cnidarians, and notably evaluating the conservation of this newly defined cnidarian embryonic territory.
Note added in proof
Lebedeva et al., 2022 has now been published as: Lebedeva, T., Boström, J., Kremnyov, S., Mörsdorf, D., Niedermoser, I., Genikhovich, E., Hejnol, A., Adameyko, I. and Genikhovich, G. (2025). β-catenin-driven endomesoderm specification is a Bilateria-specific novelty. Nature Commun. 16, 2476. doi:10.1038/s41467-025-57109-w.
Footnotes
Funding
This work was supported by an ATIP-Avenir grant (2021) from the Centre National de la Recherche Scientifique (CNRS) and the Institut National de la Santé et de la Recherche Médicale (INSERM).
References
Competing interests
The authors declare no competing or financial interests.