Halved H. pulcherrimus embryos undergoes unique developmental process

Initially, we examined the morphology of H. pulcherrimus embryos bisected at the 2-cell stage (Fig. 1). During early cleavage, each isolated blastomere followed a cleavage pattern similar to that of intact embryos (Fig. 1b-i). For example, at the 16-cell stage, when intact embryos formed 8 mesomeres, 4 macromeres and 4 micromeres, halved embryos generated 4 mesomeres, 2 macromeres and 2 micromeres. However, after the 16-cell stage, while intact embryos continued cleaving to form a blastula (Fig. 1p-s), the halved embryos unexpectedly developed a flat, plate-like structure instead (Fig. 1j). The edges of this flattened structure then lifted, forming a cup-like shape (Fig. 1k), which gradually closed at the opening (Fig. 1l, m), eventually creating a sphere (Fig. 1n). At this stage, the blastocoel was difficult to discern in the spherical embryos. Ultimately, the halved embryos developed a clear blastocoel (Fig. 1o), becoming miniature versions of an intact blastula (Fig. 1t).

Given the essential role of the extracellular matrix (ECM), hyaline layer, which we removed during blastomere isolation, in normal sea urchin development, we considered whether its absence caused the irregular morphology. However, this possibility was ruled out, as isolated blastomeres with an intact hyaline layer exhibited the same morphological changes (Supplementary Fig. 1). This indicates that the observed developmental process is an intrinsic feature of bisected embryos. We named these distinct stages of halved embryo morphology as flat, cup and sphere (Fig. 1j-n). Although the development of halved embryos progressed more slowly than intact embryos, once the blastula was formed, the subsequent developmental processes -- such as mesenchyme cell ingression, gastrulation, and pluteus formation -- proceeded in a manner similar to that of intact embryos (Supplementary Fig. 2).

During our observations, we identified two key biological questions regarding the developmental re-organization of halved embryos. First, how does the flat morphology transition into a sphere? Second, how is a normal body axis established in the resulting sphere, given that the original axis could be disrupted during this shape transformation?

To investigate the mechanism driving the flat-to-sphere transition, we examined whether this morphological change was accompanied by additional cell proliferation. Our analysis showed no significant difference in cell number between intact embryos and double-counted halved embryos, indicating that the shape change is not due to increased cell proliferation but rather a result of cell shape changes (Fig. 2a, b). To further understand these cell shape dynamics, we closely observed the transition from cup to sphere using the membrane probe FM1-43 (Fig. 2c-f). Similar to intact blastulae, both the cup-shaped and sphere-shaped halved embryos were composed of a single layer of cells. However, the cells in halved embryos were notably more elongated along the apical-basal axis compared to those in intact embryos, forming a distinct cone-like shape (indicated by magenta double-headed arrows in Fig. 2c-f, k). The cell length in cup-shaped halved embryos was significantly greater than that in control embryos at 8 hours post fertilization (hpf) (corresponding to the flat stage of halved embryos), 10 hpf (cup stage equivalent), and 14 hpf (blastula stage equivalent) (Supplementary Fig. 3). Therefore, cell elongation along the apicobasal axis appears to be a specific feature of halved embryos. As development progressed, the apices of these cone-shaped cells converged at the center of the spherical embryo, while the edges of the cup gradually sealed laterally, leading to complete closure. During the subsequent transition from sphere to blastula, the elongated cells shortened to adopt a columnar shape, similar to those found in intact blastulae (Fig. 2k). This series of coordinated cell shape changes underlies the morphological re-organization from a flat structure to a spherical blastula-like form in halved embryos.

To identify the factor(s) regulating the drastically morphological transition from flat to spherical shape, we first perturbed several signaling pathways known to play key roles in embryonic development. However, none of the treatments affected the shape transition (Supplementary Fig. 4). This suggested that the morphological change in halved embryos might be driven by cell-autonomous mechanisms. We therefore focused on the cytoskeleton and performed live imaging of actin dynamics in halved embryos by injecting lifeact-mCherry mRNA (Fig. 2g-j, Supplementary Movie 1). To visualize the nuclei and determine the apicobasal axis, we co-injected histone2B-Venus mRNA, as nuclei localize to the apical side of cells in normal sea urchin blastulae. In control blastula, actin accumulation was observed both apical and basal side and the accumulation at apical side is stronger than that at the basal side (Fig. 2g). Unexpectedly, in halved embryos, we observed strong actin polymerization on the basal side of the cells, opposite the nuclei, during the cup-to-sphere transition (Fig. 2h, i, white arrowheads). To determine if this basal actin polymerization is essential for the morphological change, we treated halved embryos with cytochalasin D, an inhibitor of actin polymerization, and (-)-blebbistatin, an inhibitor of myosin-II ATPase, from the flat stage (Fig. 2l-o). While over 90% of control halves successfully formed a spherical shape, 100% of cytochalasin D-treated halves were arrested at the cup stage and never formed spheres. Additionally, 94% of blebbistatin-treated embryos exhibited significant delays in sphere formation, resulting in incomplete spherical blastulae (Supplementary Fig. 5). In both treatments, cell elongation along the apicobasal axis was not observed. These results suggest that actomyosin-generated forces on the basal side of cells are crucial for cell elongation along the apicobasal axis, and this cell shape change is essential for the proper formation of the blastula.

To identify additional factors involved in the shape transition of halved embryos, we examined the role of septate junctions, which are septa-like structures between cells that function as occluding junctions in invertebrates. In Strongylocentrotus purpuratus, septate junctions are formed at the 8th cleavage stage. Notably, the timing of the flat-to-cup shape transition in halved embryos coincided with the initiation of septate junction formation, prompting us to investigate the role of septate junction-related genes, particularly Tetraspanin. in situ hybridization revealed that tetraspanin was ubiquitously expressed across the entire embryo in the flat stage, and this expression persisted through the sphere stage (Fig. 2p-s, Supplementary Fig. 6a, b). This expression pattern was identical to that in control embryos. To assess whether septate junctions are required for the shape transition in halved embryos, we attenuated tetraspanin function using a morpholino anti-sense oligo (MO) and analyzed the resulting phenotypes (Fig. 2t, u, Supplementary Fig. 6c, d). While 82% of control halved embryos successfully transitioned to the sphere stage, only 8% of tetraspanin-deficient halved embryos achieved this transition. Most tetraspanin-deficient halved embryos exhibited a "bumpy" flat shape, where cells were misaligned and failed to form a single cell layer. By 24 hpf, a small percentage of tetraspanin-deficient embryos eventually adopted a spherical shape. A similar phenotype was observed in halved embryos where ZO-1, another component of septate junctions, was knocked down using a specific morpholino (Supplementary Fig. 6d). These results strongly suggest that the mechanical forces provided by septate junctions contribute significantly to the shape transition from flat to spherical in halved embryos.

Based on these observations, we identified two molecular mechanisms regulating the shape transition in halved embryos: actomyosin constriction and the adhesive force of septate junctions. Surprisingly, neither mechanism relies on the signaling pathways typically associated with body axes formation. Instead, shape transition depends on an intrinsic system. The cells themselves actively adhere to each other via septate junctions, and this strong cellular adhesion facilitates the morphological change from a flat to a spherical structure as the cells reorganize to form a cohesive, unified shape.

Next, to characterize the gene expression feature of halved embryos, we performed bulk RNA-seq and compared the gene expression profile between intact embryos and halved embryos by principal component analysis (PCA) (Fig. 3a). Halved embryos were sampled at 4 stages: cup, sphere, 6-8 hours after sphere formation and early gastrula. Intact embryos were sampled at 2-cell stage and corresponding timing with each stage of halved embryos. At cup shape stage, the gene expression profile of control and halved embryos were similar. At the sphere stage, a large difference was detected between control and the halves. Surprisingly, the difference between control and halves was rapidly compensated 6-8 hours after the sphere shape formed. This result implies that some critical event for accomplishing self-organization occurred after sphere formation. We especially focused on genes which is involved in body axes formation and compared gene expression amount of 5 anterior specification genes (foxQ2, homeobrain, six3, sFRP1/5, dkk-1), 4 posterior specification genes (wnt8, wnt1, frizzled5/8, blimp1) and 4 lateral ectoderm specification gene (nodal, lefty, chordin, bmp2/4) between control and halves, and the only detectable difference was foxQ2 expression amount (at cup stage: cup/control = 4.63). Therefore, we focused on foxQ2 and traced the anterior-posterior (A-P) axis formation in halved embryos. We analyzed gene expression patterns in halved embryos using in situ hybridization chain reactions with foxQ2 (an anterior marker) and alx1 or foxA (posterior markers) (Fig. 3b-f). In intact embryos, foxQ2 and foxA were expressed at opposite poles, and halved embryos initially exhibited similar patterns with control (Fig. 3b-d). However, immediately after the halved embryos formed a spherical shape, foxQ2 and foxA were expressed in close proximity, almost adjacent to each other (Fig. 3e), indicating a complete disruption of the A-P axis. Remarkably, six hours later, the expression sites of foxQ2 and foxA had returned to opposite poles, resembling normal A-P axis (Fig. 3f). These observations show that the A-P axis temporarily collapses when the spherical shape was formed but is subsequently reorganized to restore normal axis patterning.

Since no aberrant cell proliferation was observed during the formation of the spherical shape (Fig. 2a, b), we hypothesized that A-P axis re-organization occurs through one of three possible mechanisms: (1) shifting the anterior end specification site from its original to its final position, (2) shifting the posterior end specification site from its original to its final position, or (3) shifting both the anterior and posterior end specification sites to their final positions. To determine the most likely scenario, we labeled the animal-pole cells of halved flat-stage embryos with DiI (Fig. 3g) and tracked the labeled cells at the gastrula stage to determine their location in the body. In intact sea urchin embryos, animal-pole cells typically contribute to the anterior end region. However, in 78.8% of halved embryos, the labeled cells were found in the lateral ectoderm, suggesting that the anterior end was specified in a different position from the original. Remarkably, in 30% of the halved embryos, the labeled cells were located in the posterior lateral ectoderm (Fig. 3h, Supplementary Fig. 7). We also performed the same labeling experiment on posterior cells of halved embryos (Veg2 lineage, which give rise to secondary mesenchyme cells and archenteron cells). In 75% of halved embryos, the labeled cells contributed to the archenteron or became secondary mesenchyme cells at the gastrula stage (Supplementary Fig. 7), indicating that the posterior region was specified at its original site. These results suggest that A-P axis re-organization is primarily accomplished through the shift of the anterior end specification site from its original to its final position, while the posterior end maintains its original position.

To investigate whether the Wnt/β-catenin, which is the first and the most responsible signaling for A-P patterning, plays a role in the shifting of the anterior end specification site in halved embryos, we blocked the pathway by injecting mRNA encoding a truncated form of cadherin's cytoplasmic domain (Δcad). This truncated cadherin prevents β-catenin from entering the nucleus and functioning as a transcription factor. We labeled the animal pole cells of the halved Δcad embryos with DiI to trace the cell fate shift. As a result, the labeled cells remained at the most anterior end, indicating that the animal-pole cells retained their original fate (Fig. 3i [cf. Fig. 3g, h]). This result strongly suggests that endogenous Wnt/β-catenin signaling is necessary for the anterior end shift in halved embryos. Intriguingly, at the timing of the anterior-fate shifting in halved embryos, Wnt/β-catenin originally does not play a role in restricting the anterior end region in intact embryos. Thus, we hypothesized that Wnt/β-catenin signaling is reactivates irregularly in halved embryos at this stage. To test this, we performed live imaging of halved embryos injected with β-catenin-Venus mRNA, which allowed us to trace where Wnt/β-catenin signaling was active. In these embryos, we observed a temporal nuclear β-catenin signals in a region adjacent to the original posterior site after the halved embryos formed a sphere (white asterisks in Fig. 3j-l, Supplementary Movie 2). To confirm whether canonical Wnt/β-catenin signaling is truly reactivated after the embryo has reached the sphere stage -- at a time point when it would normally be too late to suppress foxQ2, we treated embryos with the Wnt inhibitor C59 from cup shape stage and analyzed the expression patterns of foxQ2 and foxA at blastula stage. While C59 does not achieve the same level of canonical Wnt inhibition as ∆cadherin injection in H. pulcherrimus, it has a relatively strong inhibitory effect and was thus used in this experiment. In control halved embryos, 100% (16/16) showed polar expression of foxQ2 and foxA at opposite ends (Fig. 3m, m'), consistent with successful A-P re-formation. In contrast, 44% (7/16) of C59-treated halved embryos exhibited foxQ2 and foxA expression in adjacent regions, indicating a failure of foxQ2 to shift to the opposite pole after initial proximity to foxA after being formed sphere (Fig. 3n, n'). Overall, these results indicate that a transient Wnt/β-catenin signal plays a crucial role in shifting the anterior end region from its original to the final position, facilitating the re-organization of the A-P axis in halved embryos following sphere formation. In addition, non-canonical Wnt signaling has been reported to restrict anterior neuroectodermal fate downstream of canonical Wnt in normal sea urchin embryos, we next investigated whether this pathway is involved in anterior fate shifting in halved embryos. We inhibited c-Jun N-terminal kinase (JNK), a key component of non-canonical Wnt signaling, using the JNK inhibitor SP600125, and observed the positional relationship between foxQ2-positive anterior regions and foxA-positive posterior regions in halved embryos (Supplementary Fig. 8). In 97% of control halves, the anterior and posterior regions were located at opposite poles, similar to intact embryos. However, in 27% of JNK-inhibited embryos, the anterior and posterior regions were adjacent to each other, suggesting that non-canonical Wnt signaling is partially required for anterior end shifting. These findings suggest that signaling pathways, which normally restrict the anterior-most region in intact embryos, are also utilized during A-P axis re-formation in halved embryos.

Considering the organization of body axes, we next focused on the dorsal-ventral (D-V) axis. During sea urchin embryogenesis, D-V axis formation is regulated by members of the TGF-β superfamily, including Nodal, Lefty, Chordin, and BMP2/4. To determine whether the D-V axis forms normally in halved embryos, we performed in situ hybridization with nodal and lefty (Fig. 3o-x). In intact embryos, nodal is expressed exclusively in the future ventral ectoderm. In halved embryos, although the precise expression position was challenging to completely identify in the flat stage, we detected biased nodal expression. Intriguingly, from the cup stage onward, nodal was ubiquitously expressed across the entire embryo, except at the anterior and posterior ends, this pattern that persisted until the spherical stage. Four hours after the halved embryos formed a sphere, nodal expression became re-biased toward the future ventral side, resembling that of intact embryos. We also observed nuclear phosphorylated-Smad2/3 (pSmad2/3), a downstream transcription factor of the Nodal signaling pathway, throughout the embryo at the cup stage, except at the anterior and posterior ends (Supplementary Fig. 9), supporting the notion that ectopically expressed Nodal functions as a signaling ligand. The expression pattern of lefty closely mirrored that of nodal during the cup stage, though it was less prominent immediately after the embryos reached the sphere stage; however, the precise mechanism responsible for this transient reduction remains undetermined in the current study (Fig. 3w). To investigate BMP signaling, we performed immunohistochemistry to detect phosphorylated-Smad1/5/8 (pSmad1/5/8), a transcription factor downstream of BMP, which marks the region where BMP2/4 functions (Supplementary Fig. 10). In halved embryos, pSmad1/5/8 signals were first observed at 16 hpf, coinciding with the timing in intact embryos, and were localized on one side, as in normal embryos. Notably, no pSmad1/5/8 signals were detected before the embryos reached the sphere stage (Supplementary Fig. 10g). These findings suggest that the D-V axis is initially present in halved embryos and is later reorganized after the spherical shape is established.