Introduction

The earliest animals are thought to have been multicellular assemblages such as the “choanoblastaea”, which evolved from a unicellular ancestor (Brunet and King, 2017; Maldonado, 2004; Nielsen, 2008; Sebé-Pedrós et al., 2017). Subsequent evolutionary events that introduced a morphologically distinct primary body axis, i.e. breaking of tissue polarity symmetry, remain largely unaddressed (Fig.1A). In the animal lineage, Wnt/β-catenin-mediated specification of the primary body axis was likely an early innovation since it is found today across many species of both bilaterians (all of the animals’ groups showing bilateral symmetry) and cnidarians (their sister group, comprising corals, sea anemones, jellyfish etc) (Holstein et al., 2011; Loh et al., 2016; Petersen and Reddien, 2009). Current evidence is insufficient to assess whether this innovation predated the branching of the more distantly related ctenophore and sponge clades (Jager et al., 2013; Reid et al., 2018). Importantly, however, transcriptional activation by β-catenin and its partner TCF at one pole cannot alone explain the morphological component of axial tissue patterning. Rather, polarity information capable of directing axis development can be distributed globally across the whole embryo, as clearly illustrated by classic experimental manipulations in the hydrozoan Clytia gregarium, which led to the concept of global polarity (Freeman, 1981).

Figure 1

Freeman’s global polarity concept retrospectively has much in common with planar cell polarity (PCP), the mechanism coordinating the cell polarity in the plane of epithelial tissues, later identified and unravelled genetically in Drosophila and vertebrates. PCP involves interactions between neighbouring cells, with each cell in a sheet, thus carrying polarity information. It is mediated by interactions of core PCP proteins: the Wnt receptor Frizzled (Fz), intracellular signal transducer Dishevelled (Dsh), transmembrane protein Strabismus/Van Gogh (Stbm), LIM-domain protein Prickle (Pk) and G-protein coupled receptor cadherin Flamingo (Fmi), to ensure tissue polarity and enable morphogenesis (Butler and Wallingford, 2017; Gray et al., 2011; May-Simera and Kelley, 2012; Seifert and Mlodzik, 2007; Vladar et al., 2009). A conserved feature of PCP is the polarised localisation of the Stbm-Pk-Fmi complex on one side of the apical cortex of each cell and Fz-Dsh-Fmi complex on the other side. These two complexes interact between adjacent cells to coordinate PCP orientation of neighbouring cells. The PCP proteins are conserved across bilaterians and cnidarians, as well as in some sponge groups (Hale and Strutt, 2015; Lapébie et al., 2011; Schenkelaars et al., 2016). Fz and Dsh are both a core PCP component and the receptor for Wnt ligands in initiating Wnt/β-catenin signalling, so Wnt, Fz and Dsh are potentially key elements in coupling the pathways.

Two crucial issues both for the developing embryo, and for the earliest animal ancestors, are i) to orient PCP and thus morphogenesis consistently and globally across the whole embryo, and ii) to coordinate PCP orientation with regionalised cell fate specification initiated by Wnt/β-catenin signalling. These issues have not been addressed in the context of the primary body axis specification, while current examples of PCP orienting mechanisms from other tissues and animals show considerable variability (Butler and Wallingford, 2017).

We address here the coupling of Wnt signalling and PCP during body axis specification in a cnidarian in order to shed light on the roles of PCP in the evolutionary emergence of the metazoan body axis, exploiting the jellyfish model species Clytia hemisphaerica (Houliston et al., 2022). Previously, we showed that maternal mRNAs for the ligand CheWnt3 (orthologue of bilaterian Wnt3, hereafter called Wnt3) and its receptors are localised with respect to the animal-vegetal polarity of the unfertilised Clytia egg (Fig. 1B), and activate Wnt/β-catenin signalling in a future oral (posterior) domain of the developing embryo to control gene expression along the developing primary body axis (oral-aboral: OA) (Momose et al., 2008; Momose and Houliston, 2007). In parallel, conserved core PCP proteins, including Stbm orthologue CheStbm (Stbm), are responsible for establishing tissue polarity (PCP) in the epidermis to support directed larval swimming via cilia beating along the oral-aboral axis, and are also required for elongation along the primary body axis during gastrulation (Momose et al., 2012). Stbm protein is localised on the aboral side of each cell (Momose et al., 2012) as expected from Drosophila and vertebrate PCP deployment mechanisms.

In the current work, we explore the hypothesis that Wnt3, as well as activating Wnt/β-catenin signalling (Momose et al., 2008), provides the initial cue to direct global orientation of PCP and thereby define the Clytia embryonic body axis both molecularly and morphologically. A first hint of this from our previous work was that the characteristic and complete absence of embryonic axis elongation observed following Wnt3 knockdown by Morpholino (Wnt3-MO) injection could be restored by injection of synthetic Wnt3 mRNA into single blastomeres during early development (Momose et al., 2008). Importantly, Wnt3 is the sole Wnt mRNA detected before the early gastrula (EG) stage, the period when axial morphology and epidermal PCP are established, with other Wnt genes being expressed downstream of Wnt3 or at subsequent stages of larval development (Lapébie et al., 2014; Momose et al., 2008) (Fig.1B Fig. S1). This greatly facilitates the assessment of Wnt ligand function in directing PCP, in contrast to traditional model species where multiple Wnt ligands have parallel roles during embryogenesis. The studies reported here reveal how local Wnt3 cues orient the global axial PCP and PCP-driven oriented morphogenesis in the Clytia embryo. Two discrete steps define the PCP orientation: a local cueing step with orally localised Wnt3 signal and a global propagation and coordination step mediated by core PCP proteins. Interestingly, while the role of Wnt3 is the instructive cue for PCP orientation and thus morphological axis development during Clytia normal development, other cues, including mechanical strains, can orient PCP. A similar variety of PCP orientation cues has been reported in Drosophila and vertebrates. Our findings allow us to propose a ‘PCP-first’ model for body axis symmetry-breaking in the metazoan ancestors, in which the self-organising activity of the PCP pathway allowed body axis innovation in early animal ancestors.

Results

A local Wnt3 cue orients global PCP

We first analysed in detail PCP in Wnt3-depleted embryos at the EG stage, the earliest stage that shows morphological OA polarity (Fig.1B, S1). PCP manifests as structural and positional asymmetries of the ciliary basal bodies, which are consistently positioned towards the oral side of the apical cortex. As previously (Momose et al., 2012), we quantified epidermal PCP orientation by assessing the basal body positioning (Fig.1D). In unmanipulated or control-MO-injected embryos, cells showed coordinated PCP, with 85% of cells orienting within ±45° (s.d. 33 and 34° respectively) of the average orientation. In embryos injected with Morpholino antisense oligonucleotides (Wnt3-MO) to block Wnt3 translation, PCP was oriented nearly completely randomly (Fig.1E, F) (s.d. 84°, comparable to 95° for Stbm-MO embryos (Momose et al., 2012)), and individual cells were less polarised (Fig.1G), reflecting the complete lack of axial morphology. The apical cell boundary was wrinkled in Wnt3-MO-injected cells (Fig. 1E), as observed following Fz1-MO, Dsh-MO or Stbm-MO-injections (Momose et al., 2012), indicative of markedly reduced cortical tension. These analyses demonstrate that Wnt3 depletion causes PCP defects equivalent to the knockdown of the core PCP proteins (Momose et al., 2012). This supports the hypothesis that Wnt3 provides a local cue to orient global PCP, which is responsible for the development of cilia polarity and axially elongated morphogenesis in Clytia.

To test this hypothesis, we injected Wnt3 mRNA and fluorescently labelled Dextran into one blastomere at the four-cell stage in Wnt3-MO embryos, which we call the “Wnt3-rescue” experiment (Fig.1H, J). At the EG stage, we quantified PCP in a broad area (246 x 246 µm) and represented it by bars with a circular colour code (Fig.1I). Control mock injections did not affect the PCP defects caused by Wnt3-MO. In contrast, coordinated PCP was restored by Wnt3 rescue in a non-autonomous manner (Fig. 1J). Restored ciliary alignment was observed at least 200 µm away from the Wnt3 mRNA-positive area. The Wnt3 source likely influences PCP at even further distances as a virtually complete morphological axis was restored by the late gastrula stage, with the Wnt3-mRNA lineage forming a new oral pole (Fig.1K) as shown previously (Momose et al., 2008). Locally expressed Wnt3 signal is thus necessary and sufficient to orient the PCP along the whole body axis.

Rapid global PCP development precedes basal body alignment

To understand the initial process by which PCP becomes coordinated by Wnt3, we examined stages preceding EG, again using basal body localisation as the read-out. We reasoned that if PCP propagation occurs through successive cell interactions, cell alignment might spread as a wave across the embryo. PCP was detectable from the mid-blastula stage (8-9 hpf Fig. 2A-D), slightly after the onset of the zygotic Wnt3 expression (Momose et al., 2008) and epithelialisation of the blastula cells (Kraus et al., 2020). We did not, however, detect any wavefront of local cell alignment; rather, the PCP gradually increased in coherence across the embryo over time (Fig.2A, B). This was confirmed by live imaging, injecting synthetic mRNAs in the egg to visualise cell contours with PH-Venus and basal bodies with Poc1-mCherry (Fig. 2E, Movie S1). The movies further revealed active mitosis in the blastodermal cells. A weak bias in cleavage orientation was observed along the OA axis (Fig. 2F). During mitosis, duplicated centrioles from the ciliary basal bodies are deployed in the mitotic spindles (Fig. 2E, Movie S1) such that basal body positioning becomes scrambled immediately after mitosis. Subsequently, the displaced basal bodies migrate unidirectionally toward the oral side of each cell, where the cilium is re-established, a clear indication that PCP was already in place across the tissue (Fig 2E, G, H). Together these observations indicate that PCP is established at the mid-blastula stage, presumably via canonical localisation of core PCP proteins to opposite cortices of each cell through continuous Fz-Stbm interaction. The protein partitioning then manifests as asymmetric basal body positioning. Localised PCP proteins may attract basal bodies to the oral cortex of the cell as described in ascidian embryos (Negishi et al., 2016). During subsequent stages, PCP is either maintained (Bellaiche et al., 2004) in each cell during mitosis or rapidly restored through interactions with neighbouring cells (Devenport et al., 2011).

Figure 2

Wnt3 orients PCP in a β-catenin-independent manner

We previously showed that local Wnt3 expression is necessary and sufficient to activate the Wnt/β-catenin pathway in the oral pole of the Clytia embryo (Momose et al., 2008). We next examined whether the action of Wnt3 for PCP orientation and morphogenesis is mediated by the Wnt/β-catenin pathway. First, we showed that PCP orientation by Wnt3 is independent of β-catenin: ectopic Wnt3 could orient PCP in the Wnt3-rescue experiment even in the presence of a dominant-negative form of CheTCF (dnTCF) (Fig.3A), which blocks downstream transcription of the β-catenin pathway. Gastrulation was completely prevented, confirming that the Wnt/β-catenin pathway was inactive (Fig. 3C, D, Fig. S2). Furthermore, body axis elongation was partially restored at the early gastrula stage in the absence of an active β-catenin pathway (Fig. 3C), although they did not develop pointed morphology of the oral pole nor elongate further (Fig. 3D, Fig.S2). These results support a model in which axial elongation requires both PCP coordination and gastrulation (migration and intercalation of ingressed endodermal cells), as suggested previously (Kraus et al., 2020). Conversely, we found that constitutively active β-catenin (CA-β-cat) mRNA was not able to restore PCP coordination nor elongation of Wnt3-depleted embryos (Fig.3B) but did restore β-catenin-dependent gastrulation (Fig. S2). Taken together, these experiments demonstrate that global PCP orientation along the Clytia OA axis by Wnt3 is not mediated by the Wnt/β-catenin pathway. Wnt3 thus has dual independent roles: PCP orientation and Wnt/β-catenin activation.

Figure 3

Global PCP is established in two distinct steps

We next tested the contribution of the core PCP proteins CheStbm and CheFmi (hereby called Stbm and Fmi) to linking PCP orientation to the local Wnt3 source. As expected, locally confined expression of Wnt3 mRNA did not restore global PCP coordination in Stbm-MO or Fmi-MO embryos (Fig.3E, F). Importantly, however, polarity was restored in the area immediately adjacent to the Wnt3 mRNA-positive cell patch. The effect covered a distance of a few cells, potentially derived from cells directly in touch with the Wnt source at earlier stages. This observation raised the possibility that global PCP orientation in Clytia embryos is established in two steps: Stbm/Fmi-independent PCP orientation by local Wnt3 at the oral end, followed by global PCP propagation through the intercellular interaction of core PCP proteins (two-step model), as predicted by the domineering non-autonomy property of mosaic PCP mutations (Amonlirdviman et al., 2005). Alternatively, Wnt3 may diffuse from the oral Wnt3-expressing cells (Fig.1A) to form a long-range gradient along the OA-axis and orient the PCP (Wnt-gradient model). Wnt activity gradients have been shown to control or orient PCP in mice (Gao et al., 2011; Minegishi et al., 2017). To distinguish these two models, we created mosaic embryos with a small patch of core-PCP negative cells within Wnt3-rescued embryos.

To achieve this, we transplanted single 32- or 64-cell stage blastomeres from Stbm-MO/Wnt3-MO or Fmi-MO/Wnt3-MO injected embryos to Wnt3-rescued embryos with Dextran lineage markers (Fig. 4A-E). The two-step model predicts that the PCP orientation would make a detour around the Stbm-MO/Fmi-MO mosaic area; in contrast, under the Wnt-gradient model, PCP would be affected only in the mosaic area. We found that in the Stbm-MO or Fmi-MO mosaic Wnt3-rescued embryos, PCP was correctly oriented or slightly disturbed within the Stbm-MO patch while often de-polarised in the Fmi-MO patch. Outside the Sbtm-MO or Fmi-MO patches, PCP orientation was inverted on the aboral side of the patches unless these were in close proximity to the Wnt3-mRNA patches. Stbm- or Fmi-defected cells thus exhibited domineering non-autonomy effects. This is consistent with the intercellular Fz-Fmi and Sbm-Fmi interaction model shown in Drosophila and vertebrates and the localisation of Stbm protein on the aboral side of each epidermal cell in Clytia (Momose et al., 2012). PCP was oriented normally in the cells on the oral or lateral sides of the patches (Fig. 4B, D). PCP orientation from oral to aboral detouring around the Stbm-negative patches was similarly observed when mosaics were generated by injection of Stbm-MO into a single blastomere in a Wnt3-rescue embryo at the 16-cell stage (Fig.4F, G). These observations strongly support the two-step model in which Wnt3 acts locally to orient PCP, and PCP coordination is then transmitted from oral (i.e. Wnt3-positive lineage) to aboral through core PCP interactions. They are not compatible with the alternative model that a long-range Wnt3-gradient orients PCP.

Figure 4.

Mechanical cues can orient PCP

Transplantation of Wnt3-MO blastomeres into Wnt3-MO host embryos at the 32- or 64-cell stage did not cause any PCP defects in most cases. However, in a few cases (N=3 out of 8), reorientation of PCP toward the transplanted donor lineage was observed (Fig.5A-E). In these cases, the donor cells were incompletely incorporated into the host ectodermal epithelium (Fig. 5D), such that the donor cells remained as a protrusion with a “rosette” (radial) arrangement. In these embryos, host and donor-derived epithelial cells acquired an elongated morphology around the circumference of the smooth host-donor boundary (Fig.5B). Although the grafted tissue influenced the orientation of PCP in the surrounding host cells, the direction was opposite to that driven by Wnt3-expressing sources. The rosette-like transplants thus operate as artificial “aboral” cues, in contrast with Wnt3 sources, which provide an oral cue (Fig. 5C). Unlike Stbm-MO or Fmi-MO mosaic embryos, cell polarity was continuous across the host-transplant border in these Wnt3-MO to Wnt3-MO transplantations (Fig.5A), suggesting the core PCP protein interaction was not disturbed at the boundary and thus not likely due to the nonautonomous effect of PCP defects. Epithelial ‘scarring’ can thus act to orient PCP, perhaps via discontinuities in cell adhesion or mechanical strain. PCP re-orientation by mechanical strain has been reported in wing PCP repatterning in Drosophila and for the orientation of Xenopus epidermal motile cilia (Aigouy et al., 2010; Chien et al., 2022; Guirao et al., 2010; Mitchell et al., 2007; Olguin et al., 2011). We also performed transplantations into Wnt3-MO-only hosts but did not obtain any such rosettes, so we could not determine whether Wnt has a permissive role for PCP in this context. Nevertheless, we can conclude that any local PCP asymmetry can potentially act as a PCP orientation cue via core-PCP proteins, and thus again is consistent with the two-step orientation model.

Figure 5.

To explore further whether mechanical strain can orient PCP, we disrupted the morphology of the embryo by expressing dominant negative and constitutively active forms of Clytia RhoABC, the single cnidarian orthologue of the vertebrate Rho GTPases including RhoA. Late gastrula embryos expressing dominant negative RhoABC showed defects in axial elongation, consistent with a role in morphogenesis. Conversely, embryos injected with constitutively active RhoABC showed a scrambled axial morphology with ectopic protrusions induced along the major axis (Fig.5F). The induced protrusions appeared prior to gastrulation. PCP orientation was locally coordinated along the micro-axes of each protrusion so that they became extra aboral poles with respect to PCP (Fig.5G, H), while PCP orientation along the major body axis was globally maintained. Enhanced cortical actin fibres were arranged around the circumference of the induced protrusion (Fig. 5I). This method to provoke mechanical deformation of the embryo thus generated protrusions that mimicked the morphological extension and circumferential cell arrangement seen in the transplantation-induced rosettes, with in both cases PCP orientation aligning along the experimentally formed ectopic mini-axes.

Discussion

In this study, we have shown that during embryogenesis in the cnidarian Clytia, the ligand Wnt3, known to initiate oral gene expression via Wnt/β-catenin pathway activation, also provides β-catenin-independent cues to orient PCP globally in the embryo and thereby direct morphogenesis along the oral-aboral axis. PCP orientation occurs in two distinct steps: local orientation by Wnt3 and global propagation via conserved core PCP machinery. These findings demonstrate deep conservation of PCP orientation and tissue morphogenesis mechanisms in eumetazoans, and allow us to novel scenarios for PCP-driven axis symmetry-breaking underlying the emergence of the animal body plan.

A single Wnt ligand coordinates molecular and morphological axis determination

Wnt/β-catenin signalling (Holstein et al., 2011; Loh et al., 2016; Petersen and Reddien, 2009) and PCP pathway components (Gray et al., 2011; Kumburegama et al., 2011; Momose et al., 2012; Wallingford, 2012; Zallen, 2007) are involved respectively in embryonic body axis specification and morphogenesis across many metazoan species, but how these aspects are linked has been complicated to address. For instance, in vertebrates, Wnt5a and Wnt11 play crucial roles during body axis morphogenesis, (Heisenberg et al., 2000; Kilian et al., 2003; Tada and Smith, 2000; Wallingford and Harland, 2001) but it is not straightforward to demonstrate whether they provide positional cues to orient PCP along the body axis, while they do in other contexts (eg. orientation of ciliated node cells and muscle fibre) (Gros et al., 2009; Minegishi et al., 2017). During Clytia embryogenesis, a single Wnt ligand directs both molecular and morphological aspects of axis formation. Wnt3 translated from maternally localised mRNA controls axial gene expression along the body axis via β-catenin regulation (Momose et al., 2008). This dual role of the Wnt3 elegantly ensures the coherence of gene expression along the developing OA axis with morphogenetic aspects such as embryo elongation during gastrulation and beating of ectodermal cilia for directed locomotion. Maternal XWnt11 in Xenopus activates the Wnt/β-catenin pathway (Tao et al., 2005), and it is also implicated in the β-catenin-independent role to control convergent extension (Tada and Smith, 2000). The dual roles of a Wnt ligand for PCP and Wnt/tβ-catenin pathway may be a common strategy in vertebrates and cnidarians.

Two-step PCP orientation, initiated by CheWnt3 and propagated by core PCP interactions

The experiments presented here reveal two distinct steps of Wnt3-driven PCP and axial morphology in Clytia embryos (Fig.6A): A Wnt3-dependent local PCP orientation (cueing) step is followed by a propagation step mediated by core PCP proteins. This resembles the “domineering non-autonomy” phenomenon described in Drosophila mosaic mutants for core PCP genes (Amonlirdviman et al., 2005). How Wnt3 locally orients PCP in the cueing step in Clytia embryos remains to be understood. A likely candidate is CheFz1 (a single orthologue to vertebrate Fzd1/2/7/3/6), whose predicted localisation is on the oral side of each cell, opposite to the aboral accumulation of Stbm. An attractive hypothesis is that Wnt3 from the oral pole anchors CheFz1 at the oral side of the cells adjacent to the Wnt3 source in a mechanism similar to zebrafish Wnt11 that locally accumulates Fz7 in the plasma membrane (Witzel et al., 2006). A further implication of the two-step model and the domineering non-autonomy character of PCP proteins is that PCP orientation may act as a morphogen-independent body axis maintenance mechanism in certain regeneration contexts in cnidarians (Freeman, 1981; Livshits et al., 2017; Sinigaglia et al., 2020). Established local PCP coordination may be propagated to restore the body axis in regeneration. In this context, it would be interesting to investigate the mechanism by which core PCP modulates axial gene expression. In Clytia, PCP defects resulting from Stbm knockdown led to changes in gene expression at the early gastrula stage, notably affecting cells in the oral pole that ingress during gastrulation (Lapébie et al., 2014).

Figure 6.

Conservation of the PCP orientation and morphogenesis mechanisms in metazoans

In the Drosophila wing disc, local membrane-tethered Wg/Wnt4 in the wing margin can globally orient the PCP (Wu et al., 2013; Yu et al., 2020). In vertebrate tissues, Wnt ligands either act as global cues for PCP orientation (Chu and Sokol, 2016; Gros et al., 2009; Minegishi et al., 2017), or triggers involved in the body elongation (Heisenberg et al., 2000; Kilian et al., 2003; Tada and Smith, 2000; Wallingford and Harland, 2001). The role of Wnt for PCP orientation is thus common across cnidarians and bilaterians. The Wnt subtypes employed in PCP orientation are, however, not conserved. The Wnt gene family gained its complexity prior to the bilaterian-cnidarian bifurcation (Kusserow et al., 2005). Then, different Wnt subtypes were likely co-opted in different taxa and developmental contexts. In zebrafish, PCP orientation by wnt11 and core PCP regulation are distinctive in mechanosensory hair cell orientation (Acedo et al., 2019). The modulable two-step mechanism of Wnt-driven PCP orientation is likely common in metazoans.

Other cues than Wnt also function as PCP orientation cues in the animal kingdom. In the developing Drosophila eye and abdomen, another signalling module comprising the atypical cadherins Fat(Ft) and Dachsous(Ds) and kinase Four-jointed(Fj) regulate the localisation of core PCP complexes via gradients of Ds and Fj (Strutt and Strutt, 2021; Yang et al., 2002). Cnidarians have conserved Ft and Ds ortholog genes. In hydra Fat-like protein (HyFat1) is implicated in epithelial cell alignment and adhesion (Brooun et al., 2020). In Clytia, inhibition of Fat-like (CheFat1) and Ds (CheDs) disrupted global PCP coordination, and the Ds disruption was restored by Wnt3 expression (Fig.S3), suggesting that these molecules participate in global PCP orientation in a manner yet to be studied. Besides molecular signalling cues, mechanical constraints or fluid flow are also known to direct PCP (Aigouy et al., 2010; Chien et al., 2022; Guirao et al., 2010; Mitchell et al., 2007; Olguin et al., 2011) and can potentially act as the PCP symmetry-breaking factor. The availability of alternative PCP orientation cues might have allowed evolutionary flexibility to link PCP to available structures, such as the cell polarity scaffold of eggs. While tissue polarity orientation by PCP is strongly conserved, the cellular mechanisms underlying the morphogenetic movement are highly variable. Oriented cell division or convergent extension are likely major driving forces of blastula elongation prior to gastrulation in Podocoryne carnea (Momose and Schmid, 2006), another hydrozoan jellyfish. In contrast, in Clytia, elongation likely involves PCP-dependent intercalation of both ectoderm (Byrum, 2001) and ingressing endoderm (Sande et al., 2020) cells during gastrulation. In Nematostella, hydraulic pressure created by body wall muscle, organised along the body axis, drives the elongation (Stokkermans et al., 2022) (Fig. S4)

Body axis symmetry-breaking in the evolutionary history of metazoa

The dual role of Wnt3 and the two-step PCP orientation mechanism revealed here are particularly intriguing in the context of understanding the emergence of animal body axes during evolution. The metazoan ancestor is considered to have taken a multicellular state such as a “choanoblastaea” from a unicellular ancestor (Arendt et al., 2015; Nielsen, 2008) (Fig. 6B). Then metazoans acquired a body axis and germ layers and formed a hypothetical ancestral state such as Haeckel’s “gastraera” (Haeckel, 1873) or Hyman’s “planuloid-acoeloid” (Hyman, 1951). Early multicellular ancestors thus acquired morphological axis patterning and coupled it with regional gene expression. It is tempting to speculate that the β-catenin/TCF-dependent local gene expression and morphogenetic patterning coordinated by core PCP proteins were independently acquired and coupled by co-opting Wnt ligands and signal transducers from one pathway to the other. For example, Wnt, Fz and Dsh might have first been employed in the Wnt/β-catenin pathway, then recruited to a primordial PCP regulatory system (Fig. 6B). Genome analyses indicate that PCP effectors and a part of core PCP components such as Cdc42, RhoA, ROCK, Prickle or Inversin are evolutionarily older than metazoans (Lapébie et al., 2011) (Fig. S5, Table S1). Wnt-Fz-Dsh would then have been recruited to create morphogenesis machinery directed by PCP based on the existing local gene expression by Wnt/β-catenin signalling. Alternatively, core PCP interactions and Wnt-driven PCP orientation may be more ancestral with Wnt-Fz-Dsh used in PCP orientation prior to its involvement in “canonical” Wnt signalling, acquired later though recruitment of the cell adhesion molecule β-catenin and its interaction with the DNA-binding TCF. Comparisons of the genomic repertoires for these components between species from across the metazoan phylogeny do not allow us to conclude which modules were acquired first (Fig.S5, S6, Table S1). In the outgroups of the cnidaria-bilateria clade, a full set of core PCP proteins are present in a sponge Oscarella carmera (Lapébie et al., 2011; Nichols et al., 2006; Schenkelaars et al., 2016), and full Wnt/β-catenin pathways were identified in both Porifera and Ctenophora, while multiple cases of loss of Stbm/Fmi occurred in Porifera and potentially in Ctenophora (Fig.S5, Table S1).

PCP is a fundamental character in multicellular tissue organisation seen wide across metazoans (Fig. S5, S6) and may have been particularly important in the metazoan ancestor with lower anatomical complexity. It is known that a rotational polarity of the flagellar basal body is coordinated across the spheroid of green alga Volvox carteri, which thus independently acquired PCP-like character (Hoops, 1993) (Fig.S6), suggesting that a PCP equivalent evolved independently in Volvox. PCP-driven axis evolution is an attractive scenario and can explain how an initial “choanoblastaea” multicellular ancestor acquired coordinated cilia/flagella flow. For example, PCP regulation might initially have improved filter-feeding efficiency in a choanoblastaea ancestor (Cavalier-Smith, 2017; Nielsen, 2008). A comparative cell biology, or “Cellular Evo-Devo” approach, of the PCP/tissue polarities in diverse metazoans, in particular “basal metazoans” including Cnidaria, Ctenophora and Porifera, will be the key to addressing how symmetry-breaking to create a body axis arose during evolution.

Materials and methods

Animal culture and embryo preparation

Animal culture methods are described in (Lechable et al., 2020). Clytia hemisphaerica Laboratory strains were provided by the Service Aquariologie of CRB (IMEV-FR3761, Sorbonne Université/CNRS contact: Axel Duchene axel.duchene@imev-mer.fr) as an EMBRC-Fr service (https://www.embrc-france.fr/). Daily spawning was induced by light with a 24-hour day-night cycle. Eggs and sperm were separately collected from jellyfish and transferred into 10 cm glass dishes two hours after the light stimulation. Eggs were fertilised less than one hour after the spawning. To acquire EG stages, embryos were incubated at 16°C for 18 hours, which corresponds to the 12∼13 hpf stage at the standard culture temperature (18°C).

Microinjection

Morpholinos and mRNA were microinjected into unfertilised eggs or blastomeres at 4-cell ∼ 16-cell stages using an Eppendorf Femtojet microinjector with microcapillaries pulled with Narishige PN-30 needle puller. The tip of the needle was broken by touching it to a glass surface scratched by a diamond pen to open the needle tip to give an outer diameter of 2 to 4 µm. Eggs were aligned in a microinjection chamber made by casting and polymerising PDMS in a negative mould printed with a Form3 3D printer. 2∼3% of the egg/cell volume of solution was injected by inserting a microcapillary needle with constant back pressure (Femtojet compensation pressure). It typically takes 0.5∼1 seconds for each egg at 5∼10 kPa when the needle has an appropriate opening size. The volume of injection for each injection was visually controlled under an Olympus SZX6 stereomicroscope with SZ-ILLT oblique illumination. 3∼5 mg/ml lysine fixable 10 kDa Dextran labelled with Alexa647 or Cascade Blue were co-injected with Wnt3 mRNA and Stbm-MO/Fmi-MO for the lineage tracking in the Wnt3-Rescue experiment and PCP mosaic experiments.

Plasmid construction

Mutant and fluorescent protein templates were constructed in pCX3 vector, which contains short 5’ and 3’UTR sequences from CheStbm mRNA. Wildtype CheWnt3 with 3 mispaired nucleotide residues at the Morpholino target and CheTCF lacking amino acid residues 2-27 (dnTCF) as a dominant negative form (lacking β-catenin binding site) of human TCF-4 (Korinek et al., 1997) were amplified by PCR and cloned into the vector. A constitutively active form of Che-β-catenin (CA-β-cat) was created by introducing S97A T101A S105A mutations corresponding to the constitutively active mice β-catenin mutant lacking a GSK3-β phosphorylation target (Baba et al., 2006). Constitutively active forms (CA1: G14V, CA2: Q63L) that gave identical phenotypes and a dominant form (DN: T19N) of CheRhoABC were designed based on the identical vertebrate RhoA mutants with the known phenotypes. These point mutations were introduced into wildtype cDNA clones in pCX3 vector using the QuickChange site-directed mutagenesis kit (Stratagene).

Poc1-mCherry and PH-domain-Venus fusions were constructed in a transgenic pMiniTol2-ACT2 vector (Weissbourd et al., 2021) using NEBuilder Assembly kit from individually amplified PCR products or synthetic genes. mRNA was synthesised in vitro using mMessage mMachine T3 kit and poly(A)-tailing kit (Thermo Fisher) from PCR products amplified from the plasmids.

Immunostaining

Immunofluorescent visualisation of PCP by gamma-tubulin antibody and phalloidin staining was described previously (Momose et al., 2012). List of the antibodies are in the Supplementary method. In brief, the embryos were fixed with 4% Paraformaldehyde in 0.1 M Hepes, 50 mM EGTA, 80 mM Maltose, 10 mM MgSO₄, 0.1% Triton X-100 and washed in PBST (1x PBS with 0.1%Triton X-100) then incubated in antibody solution in 1% BSA fraction V in PBST, followed by three times of washes for at least 5 minutes in PBST. After removing the detergent by washing in PBS, actin was stained with fluorescently labelled phalloidin in PBS (0.13 µM). After staining, embryos were cleared and mounted in Citiflour AF-1 diluted with an equal volume of PBS and scanned using Leica SP5 or Stellaris confocal microscopes with 63x objectives. The surface of the ciliated epidermis was Z-scanned (up to 10 µm) in the entire 246 x 246 µm field of view and z-stacked to visualise cell contour with apical cortical actin and basal body, typically at 0.081 µm resolution in XY-axis and 0.5∼1 µm resolution in Z-axis.

Image analysis

The x-y coordinates of the apical surface centroid (x_C, y_C) of each cell and basal body of its cilium (x_b, y_b) are identified from the z-stacked confocal images. Detailed procedures and ImageJ scripts are described in the supplementary method. The polarity angle θ for each cell was calculated as

The mean orientation (arrow in the radar plot) and circular standard deviation (circ. s.d.) are calculated (Berens, 2009). A Matlab script was used to summarise the PCP orientation and statistics (Supplementary Method). The morphology of embryos in three dimensions (Figure. 5D, H) is reconstructed from a series of cortical actin confocal images by building a surface of the volume containing the cortical actin signals using Imaris software (Oxford Instruments).

Acknowledgements

This work was supported by ANR DiploDevo (ANR-09-BLAN-0236-01), iMMEJ (ANR-17-CE13-0016-01) and the CNRS INSB Diversity of Biological Mechanisms program. We thank the “Service Aquariologie” of the CRB platform (IMEV, FR3761) supported by EMBRC-France and ANR funding (ANR-10-INBS-02) for access to facilities and resources and the PIM platform for imaging. We thank Brandon Weissbourd for the critical reading of the manuscript.

Additional information

Author Contributions

TM designed the project. TM and EH wrote the manuscript. TM, AD, JU and MLé performed the experiments. MLe, FL prepared Clytia animal setup. FL prepared plasmid vectors. TM, JU and MLé performed image data analysis. TM, EH and CV acquired funding.

Abbreviations

• PCP: planar cell polarity

• EG: early gastrula,

• LG: late gastrula,

• MO: Morpholino,

• OA: Oral-Aboral

Additional files

Supplemental Figures 1-7

Supplemental Method

Supplemental Movie