Adult P. pectinifera were collected from Tokyo Bay and Mutsu Bay in Japan and kept in a glass aquarium with artificial seawater (ASW; MarineArt SF-1, Tomita Pharmaceutical, Japan) at 15°C. The embryos of P. pectinifera were obtained as previously described (25). In brief, mature eggs were prepared by treatment with 1-methyladenine (Sigma-Aldrich, US) (26) and fertilized with diluted sperm. Embryos and larvae were reared in ASW at 20°C with constant rotation at 30 rpm. From two days post-fertilization, the larvae were fed Chaetoceros calcitrans (Sun-culture, Marine Tec, Japan) at a concentration of 1.0 × 10⁶ cells/mL and reared to brachiolaria larvae with daily rearing water changes.

Coral sands (approximately 2–5 mm in diameter) from aquarium-reared adult animals were used to induce metamorphosis at the brachiolaria larval stage (27). Metamorphosed larvae were maintained until they reached the juvenile stage.

Coelomic fluid (CF) containing coelomocytes was collected using a needle (26 G × 1/2”: Terumo, Japan) and syringe (medium mouth, 1 mL: Terumo, Japan). The concentration of coelomocytes in the CF used in this study was 3.47 ± 2.06 × 10⁵ cell/mL, as measured by a hemocytometer (C-Chip, NanoEntek, Korea).

To prepare coelomocyte smears, 1 mL of 5 × 10⁵ cells/mL CF containing 0.25 M EDTA were centrifuged at 300 × g for 5 min (SC-2, TOMY, Japan) at room temperature (20–25°C). Then, May-Grünwald-Giemsa (MG) staining was performed. The MG method uses May-Grünwald (Merck, US) and Giemsa stain solution (Nacalai Tesque, Japan) containing Azure B and Eosin Y to stain the cell nuclei purple, cytoplasm blue, or pink, and the cytoplasmic granules their specific color. Staining was performed according to the manufacturer’s instructions.

Allogeneic transplantation was performed by injecting 1 mL of coelomocyte-containing CF into five recipients, immediately after collection from a single donor. Autologous transplantation was performed by injecting 1 mL of self-CF (n = 4). In the case of transplanting fluorescently labeled coelomocytes, 1 mL of CF was incubated with 2 µL of 5-(and -6)-carboxyfluorescein diacetate succinimidyl ester (CFSE; 1 mg/mL; DOJINDO, Japan) at 4°C for 45 min. Then, coelomocytes were washed once with ASW by centrifugation at 1,000 × g for 5 min at 4°C, and resuspended in 1 mL of sterilized ASW.

To observe the immune response after coelomocyte transplantation, 200 µL of CF was collected in a Petri dish with 2 mL of sterilized ASW and allowed to stand for 30 min at 20°C. When fixing the coelomocytes and their aggregate form, the collected CF was immediately mixed with an equal volume of 4% w/w paraformaldehyde-containing ASW (PFA-ASW) for bright-field or confocal microscopy or 5% w/w glutaraldehyde-containing ASW for electron microscopy.

Based on the microscopic images, the area and number of coelomocyte aggregates were measured using ImageJ software (version:1.53j). Aggregates were defined as having an area of more than 2,000 µm², which corresponds to an area of approximately 10 coelomocytes of the major cell type arranged on a flat surface.

Blastulae (approximately 16 h post-fertilization) obtained from the two parent pairs were fluorescently labeled with chloromethylfluorescein-diacetate (CMFDA, 100 µM; AAT Bioquest, US) or carboxyfluorescein diacetate (CFDA, 100 µM; AAT Bioquest, US) for 30 min at 20 °C. These embryos, having a packed volume of 0.2 mL, were dissociated by a previously described method (21), then reconstructed independently or in a mixture. The definition of the chimeras produced in this study is shown in Supplementary Figure 1 . The reconstruction was performed in a glass Petri dish at 15°C. After specimens formed a mouth, they were fed with C. calcitrans (Section 2.1), until they reached the brachiolaria stage. The fluorescently labeled specimens were observed with a laser confocal microscope (Fluoview, Olympus, Japan). For observations under the electron microscope, juveniles that transformed from the larvae stage were fixed with 2.5% glutaraldehyde in ASW.

Coelomocyte aggregates and juveniles fixed with glutaraldehyde were embedded in 1.0% agarose and post-fixed with 2.0% osmium tetroxide in filtered ASW for 2 h at 4°C. After washing with distilled water, the sections were dehydrated using a graded ethanol series and embedded in Epon 812 (TAAB) at 65°C. Ultra-thin sections (70 nm thickness) of coelomocytes and coelomocyte aggregates were cut using a diamond knife on an Ultracut S ultra-microtome (Leica, Germany), stained with 2.0% uranyl acetate solution and 2.0% lead citrate solution, and observed using a transmission electron microscope (Tecnai G2 20, FEI, US) operating at 200 kV. For the juveniles, semi-thin sections (thickness, 500 nm) of juveniles were cut using a diamond knife, stained with 0.1% toluidine blue in 0.1 M phosphate buffer [pH 7.4], and observed using a BX-51 light microscope (Olympus, Japan). Juvenile sections were stained with uranyl acetate and lead citrate solutions, coated with osmium using an OPC80 osmium coater (Filgen, Japan), and observed using a field-emission scanning electron microscope (Quanta 450 FEG; FEI, US) with a backscattered electron detector operating at 5 kV.

Survival time was measured after metamorphosis of the reconstructed chimeras. Death was determined by morphological distortion and atrophy in addition to motility, including tube foot movements. Kaplan–Meier survival curves were used to visualize the data, and a log-rank test was performed with p < 0.05 considered statistically significant. R (version 4.2.0) software was used for all statistical analyses (28).

Coelomocytes are the cells present in the CF of echinoderms that mediate immune responses. The various types of coelomocyte are characterized by their morphology, and include phagocytes, spherule cells (also called spherulocytes, morula cells, and granulocytes), and progenitor cells (29–31). However, the coelomocytes in P. pectinifera remain uncharacterized.

Here, we attempted to classify coelomocytes based on morphology and MG staining ( Figure 1A ). At least five types of coelomocytes were discriminated using MG staining. The first type were cells that comprised the majority of the smear and appeared in contact with each other in a network-like manner ( Figures 1A, B ). These cells contained small granules and adhered to the substrate via widely spreading lamellipodia, with filopodia observed in some cells ( Figure 1B ). These features suggested that these cell types were the phagocytes corresponding to amoebocytes (also called petaloid phagocytes), which are most abundant in the CF of other echinoderms (31–33). The second and third cell types both presented a disc-shaped morphology similar to the discoidal phagocytes in sea urchin (34); however, differences in the stainability were clearly observed ( Figures 1C, D ). In the cells depicted in Figure 1C , the nucleus size was relatively small compared to the cells in Figures 1B, D , indicating condensed chromatin. These cells included small granules, and the cytoplasm stained light pink was localized near the nucleus. Membrane ruffling was observed at the periphery of the lamellipodia, indicating that these cells adhered to the substrate with their membranes spread to the full extent. In contrast, the third cell type had a large nucleus and evenly stained cytoplasm, with no observed granules or membrane ruffling ( Figure 1D ). Therefore, we concluded that the second cell type was discoidal phagocyte and the third cell type was agranulocytes. The fourth cell type shown in Figure 1E was small with a high nuclear-cytoplasmic ratio, which is characteristic of undifferentiated cells, corresponding to progenitor cells in other echinoderms (29–31). The fifth cell type was spherical, and the cytoplasm was filled with many basophilic granules ( Figure 1F ), consistent with the characteristics of morula cells in other echinoderms (31, 35). Although no nucleus was visible in Figure 1F , the presence of the nucleus was confirmed in the other cells ( Supplementary Figure 2 ). This cell type was present in only approximately 0.02% of the smear.

To verify the allorecognition ability of coelomocytes, we injected fluorescently labeled allogeneic coelomocytes into the coelom of recipient animals. Aggregates of coelomocytes developed over time in the recipient’s CF, and fluorescent signals derived from the injected cells were observed in the aggregates with a scattered pattern ( Figures 2A, B ). In contrast, in the case of injected self-coelomocytes, no aggregate was observed ( Figure 2A ). These results clearly indicated that the coelomocytes of P. pectinifera possessed allorecognition ability. The size and number of aggregates peaked at 12 h after the allograft then decreased ( Figures 2C, D ). The number of aggregates varied widely among individuals ( Figure 2D ); however, the aggregates from individuals with a small number of aggregates tended to be large, and a large variation was observed in aggregate size, especially at 12 h and 18 h.

Figure 3 demonstrates the internal morphology of the aggregates. Each coelomocyte in the aggregate was densely packed, but no multinucleated cells were observed, whereas numerous cross-sections of the pseudopodia were observed in the intercellular spaces ( Figure 3A ). Coelomocytes that independently phagocytosed other cells were scattered inside the aggregates ( Figure 3B ). Moreover, coelomocytes containing secondary lysosomes were observed ( Figure 3C ). These results corresponded to the detected pattern of fluorescent signals derived from allografted cells, as indicated in Figure 2B . To determine whether such aggregate formation without cell fusion occurred only in response to allogeneic heterologous cells, the bacteria were injected into the adults. As a result, coelomocytes were shown to aggregate without cell fusion, and each coelomocyte phagocytosed the bacteria individually ( Supplementary Figure 3 ).

The normal individual reconstruction processes performed by the dissociated cells of P. pectinifera embryos are as follows: (i) the mixture of dissociated cells reaggregates within 2 h after dissociation; (ii) 8 h after the start of reconstruction, a germ layer forms via the sorting of ectodermal and endodermal cells; and (iii) by 24–40 h, as the blastocoel gradually expands, the ectoderm invaginates and eventually fuses with the endoderm, forming a blastopore. After these reconstruction processes, the reconstructed embryos develop normally into bipinnaria larvae, forming mouth and coelomic pouches (21). This reconstruction phenomenon is also achieved by mixing dissociated cells from sibling embryos derived from a common parent (22).

In this study, we first stained blastulae derived from two pairs of parents (batches A and B) using different fluorescent dyes ( Figure 4A ; Supplementary Figure 1A ). The sibling embryos were dissociated and reconstructed individually ( Figure 4B , batch A or B) or mixed ( Figure 4B , batch A + B). The reconstructed embryos from batches A and B were sibling chimeras and developed normally into bipinnaria larvae, as previously reported (21). In contrast, although the reconstructed embryos from a cell mixture of batches A and B were chimeras comprising cell populations without kin-relationships, these allogeneic chimeras also successfully developed into bipinnaria larvae, as expected from our previous research (24), in which the larval immune system was tolerant of living allogeneic cells ( Figure 4B , allogeneic chimera). We observed no difference in morphogenesis between the sibling chimeras ( Figure 4B ). However, in the reconstructed embryos of allogeneic chimeras, we noted that the fluorescent signals derived from each batch were distributed in a patchy fashion, indicating that cells derived from each batch tended to accumulate separately.

Upon feeding, bipinnaria larvae developed into brachiolaria larvae able to metamorphose, and adult rudiment developed at the posterior end of the larvae ( Figure 4B , arrows). Regardless of the presence or absence of a kin-relationship, all chimeras developed into brachiolaria larvae. When we induced these chimeras to metamorphose, all chimeras, including allogeneic chimeras, reached the juvenile stage within two days through the typical metamorphic process of P. pectinifera ( Figure 4B , juvenile). No abnormalities were found in the external morphology of the chimeric juveniles. Moreover, we observed no statistically significant difference in body length between the sibling and allogenic chimeras throughout the developmental stages ( Supplementary Figure 4 ).

Individuals in the allogeneic chimeras with abnormal morphologies, such as those bent in half or some swollen balloons, gradually emerged two weeks after reaching the juvenile stage. ( Figure 5A ). These morphologies were not observed in sibling chimeras. Furthermore, one month after metamorphosis, the number of dead individuals increased rapidly in allogeneic chimeric juveniles only, with all allogeneic chimeras eventually dying. Many of the juveniles that died during this period did not show any morphological abnormalities, as mentioned above; however, the motility of their tube feet and their adhesive property to coral sands, which is a metamorphosis-induced substrate, were completely lost. Figure 5B depicts the survival time curve of the juvenile stage using the Kaplan–Meier method. Using a log-rank test, we observed a significant difference in the survival time between sibling and allogeneic chimeras, and the hazard risk of allogeneic chimeras was 40.901 times (95% CI: 28.654–58.38) higher than that of sibling chimeras. However, in a preliminary experiment conducted in other batches, allogeneic chimeric juveniles were completely destroyed in approximately two weeks ( Supplementary Table 1 ). A batch effect may exist for the time to death of the allogeneic chimeric juveniles.

To investigate whether the death of allogeneic chimeras was caused by genetic incompatibility, we prepared half-sibling chimeras with a different mother or father ( Supplementary Figure 1B ). These half-sibling chimeras also developed normally into brachiolaria larvae and showed no morphological abnormalities after metamorphosis to the juvenile stage, as with the allogeneic chimeras. Although we observed a significant difference in the mortality in the juvenile stage of half-sibling chimeras compared to that of the sibling chimeras (hazard risk: 2.348 fold, 95% CI: 1.354–4.07), and the survival rate was much higher than that of the allogeneic chimeras (hazard risk of allogeneic chimeras to half-sibling chimeras: 22.27 fold, 95% CI: 13.16–37.69) ( Figure 5B ).

We examined the internal morphology of allogeneic chimera juveniles in the process of death. An epidermal cell layer was present near the body surface of the healthy sibling chimera juvenile five days post-metamorphosis (dpm); beneath this cell layer, we observed condensed mesenchymal regions corresponding to larval blastocoel filled with extracellular matrix ( Figure 6A ). At this time, skeletal plate formation, which is a characteristic of echinoderms, had not yet been observed, and a hyaline layer derived from the larvae covered the dorsal surface of the individual. Furthermore, the vast coelomic cavity, a major characteristic of the adults, was still underdeveloped. These morphologies revealed that the post-metamorphic juvenile continued tissue organization for a period of time. Similar morphologies were also observed in the allogeneic chimera five days after development into the juvenile ( Figure 6A ), with no difference in morphological characteristics between sibling chimeras and allogeneic chimeras within five days after metamorphosis.

In contrast, in the allogeneic chimera juvenile, the mesenchymal region of the individuals with morphological abnormalities ( Figure 5A ) indicated abnormal expansion and was not condensed, as shown in the sibling juvenile (14 dpm) ( Figure 6A ; abnormal, MR). Furthermore, part of the epidermal cell layer was lost, and only a hyalin layer-like structure was observed ( Figure 6A , pink line). In this region, larval mesenchyme cells, which developed pseudopodia, were observed along the hyalin layer ( Figure 6A , MC). This observation was not limited to allogeneic chimeras with morphological abnormalities; the mesenchymal region was also expanded in chimeric individuals that died without showing deformation (30 dpm) ( Figure 6A , death). Conversely, development of the coelomic cavity was not observed. Furthermore, most of the epidermal cell layers of dead allogeneic chimera juveniles, with or without morphological abnormalities, disintegrated, and only larval mesenchyme cells remained along the hyalin layer (30 dpm) ( Figure 6A ; death, MC). In areas of severe tissue damage, we observed many collapsed cells and vesicles that appeared to have originated from these cells, as well as cells containing phagosome-like vesicles ( Figure 6B ).

Figure 7 indicates a magnified electron microscopic image of the digestive tract of the juveniles. In the sibling chimeras, the intestinal cells contained nuclei that exhibited a relatively high electron density and numerous lipid droplets ( Figures 7A, B ). In contrast, such cells were not observed in allogeneic chimeras, whereas numerous cells contained enlarged nuclei and cytoplasm ( Figure 7C, D ). Moreover, in some cells, we observed uniform chromatin with low electron density, nuclear membrane damage, an absence of distinct cell membrane, and loss of organelles, suggesting that these cells were dead ( Figures 7D, E ).