Corallium rubrum, a member of the precious corals (family Coralliidae), is characterized by red-pigmented mineralized structures that give the coenenchyma its distinctive coloration, contrasting with the expanded white polyps (Figure 1A). Its anatomy was first described in detail by Lacaze-Duthiers (1864), and subsequent studies on the Pacific precious coral Corallium japonicum have confirmed a high degree of anatomical similarity (Nonaka et al., 2012). Figure 8 presents an artistic rendering of the C. rubrum coenenchyma tissues summarizing our observations. Endodermal canals occur in two subtypes: large primary canals, running parallel to the axial skeleton, and thinner secondary canals forming a network above them (see Loentgen et al., 2025 for photographs). The canals bear long tubulin-based flagella on the luminal side and an actin-striated outer wall. Both subtypes are interconnected and connect to the polyps via discrete junctions (polyps omitted for clarity).

A single type of scleroblastic network was observed: scleroblasts form a continuous, tortuous tubular system entwined with the canals in the mesoglea. Canals and scleroblasts sometimes form large intercellular bridges containing granular structures, suggesting possible exchange of molecules and organelles across the mesoglea. As previously noted by Grillo et al. (1993), sclerites are more abundant at the periphery than in the center of the coenenchyma. The early intracellular stages of sclerite formation within individual scleroblasts (Le Goff et al., 2017) were not investigated here and are not depicted. Developing sclerites were located extracellularly within the lumen of the sleeve formed by the scleroblastic network. Multiple nuclei were observed surrounding a mature sclerite, indicating that several scleroblasts contribute to the formation of a single sclerite. This suggests that the number of participating cells increases from vacuolar initiation to the extracellular calcification of a sclerite with eight protuberances. As sclerites grow, they may exceed the lumen diameter, stretching surrounding scleroblasts so that their nuclei lie in the depressions between the protuberances. Notably, naked sclerites were never observed in the mesoglea of Corallium rubrum. The cellular processes underlying the progression from the round vacuolar calcified nucleus (Le Goff et al., 2017) to the fully formed extracellular sclerite remain to be elucidated.

Figure 9 presents an artistic rendering of the Sarcophyton sp. coenenchyma, summarizing our observations. Large vertical canals, originating beneath the peripheral polyps (polyps omitted for clarity), connect to thinner transverse canals (nomenclature after Moseley, 1881), forming a network that spans the entire organism. Adjacent to the vertical canals, we observed thick ladder-like structures, here interpreted as “primary scleroblasts”.

Our terminology for primary and secondary scleroblasts follows established literature on sclerite formation: primary scleroblasts are considered responsible for the early intracellular stages of calcification, whereas secondary scleroblasts are thought to participate in the formation of mature sclerites, typically bearing microprotuberances, whether located intra- or extracellularly. For descriptive purposes, we assumed that the coenenchyma contained only two main cell types: canal cells and scleroblasts, with scleroblasts being further differentiated into primary and secondary scleroblasts in some species. Secondary scleroblasts are described as being derived from primary scleroblasts. In Sarcophyton sp., no growing sclerites were detected, at our resolution, in the tissue that we interpreted as a “primary scleroblastic network”. On the other hand, this tissue lacked the endodermal markers characteristic of canals (external striated actin fibers, flagella within the lumen, and endosymbionts), supporting our designation as a primary scleroblastic network. A possible third cell type, however, cannot be excluded. The “cell cords” described in the sea whip Pseudoplexaura flagellosa consist of cellular networks containing multiple cell types, potentially serving as precursors to various differentiated cells (Chester, 1913). In P. flagellosa, scleroblasts “in early stages of crystal formation are found clustered near the edge of the cell cord” (Goldberg and Benayahu, 1987), although no large-scale tissue views were provided. By analogy, the structures we identified in Sarcophyton sp. as primary scleroblasts may alternatively represent cell cords, a possibility that warrants further investigation.

The remainder of the mesoglea was populated by dendritic-shaped cells forming a continuous meshwork. These cells correspond to scleroblasts in the strict sense, as they contained growing sclerites, and are referred to here as secondary scleroblasts. Between their nucleated cell bodies, maturing sclerites of various sizes and with distinct protuberances were enclosed by an extended cytoplasmic envelope (Supplementary Videos 3–5). Structurally, this arrangement is reminiscent of a neuronal network, with discrete cell bodies connected only through their dendritic extensions. Similar organization has been described in P. flagellosa, where scleroblasts appear interconnected by “interdigitating pseudopodia” (Goldberg and Benayahu, 1987). However, the nature of the connectivity between the canal system and the scleroblastic networks, specifically the developmental link between primary and secondary scleroblasts, remains unresolved.

In summary, Sarcophyton sp. exhibits two distinct scleroblastic systems: (1) a primary scleroblastic (or cell cord) network running alongside the vertical canals, and (2) a secondary scleroblastic network forming a dendritic meshwork within the mesoglea, where sclerite growth occurs.

In octocorals, sclerite formation is mediated by scleroblasts within a controlled microenvironment. Earlier reports proposed that scleroblast clustering around forming sclerites could be arranged as a syncytium in some species scleroblasts (Woodland, 1905a) (Goldberg and Benayahu, 1987). However, in Corallium rubrum, electron microscopy imaging showed that two individual scleroblasts surrounding a developing sclerite are separated by a continuous plasma membrane and linked by a septate junction (Le Goff et al., 2017). Septate junctions, a hallmark of individualized cells in invertebrate tissues, are intercellular junctions characterized by an intermembrane spacing of ~20 nm between adjacent plasma membranes (Ganot et al., 2015; Tambutté et al., 2007). Thus, although the resolution of our present confocal analyses is insufficient to distinguish membrane boundaries between paired scleroblasts, previous ultrastructural analyses in C. rubrum demonstrated that the scleroblastic network is composed of individual cells.

In Sarcophyton sp., the absence of electron micrographs prevents determination of whether distinct boundaries exist between the numerous scleroblasts revealed by DAPI staining of their nuclei, or of the precise nature of the cellular compartment enclosing the growing sclerites. In particular, it remains unclear whether sclerites develop entirely within a vacuole-like compartment of a single cell or within an extracellular space formed by multiple scleroblasts. However, we labeled growing sclerites within the scleroblastic network using calcein, a cell-impermeant dye. As such, calcein cannot label mineralization occurring within a closed intracellular compartment, which would require passage across both the plasma membrane and the vacuolar membrane. By contrast, calcein diffuses through extracellular spaces in anthozoan tissues, as the permselective barrier formed by septate junctions allows calcein to pass between adjacent cells (Ganot et al., 2015; Venn et al., 2011). Consistently, in other calcein-labeling experiments in which sclerites were isolated from Sarcophyton tissues following bleaching and analyzed for calcein incorporation as a function of size, labelling was observed in sclerites ranging from 50 to 300 µm in length (Forin et al., 2024, Forin et al., 2026). This indicates that the vacuolar phase of calcification is restricted to sclerites smaller than ~50 µm in Sarcophyton. Together, these observations suggest that, during most of their growth, sclerites calcify within an extracellular compartment formed by multiple scleroblasts. Ultrastructural analyses of Sarcophyton sp. will be essential to definitively resolve the histological organization of the scleroblastic dendritic meshwork.

Another significant observation in C. rubrum was the presence of intercellular bridges between canal cells and the scleroblastic network, as well as between cells of the same subtype. Grillo et al. (1993) previously reported such cytoplasmic extensions, although without detailed characterization. These intercellular bridges are actin-based and resemble the filopodia, cytonemes, or tunneling nanotubes described in other metazoans. Each of these terms, however, carries specific functional implications, for example, cytonemes are specialized for the exchange of signaling proteins between cells in bilaterians (Belian et al., 2023). As the molecular nature of the material exchanged between C. rubrum cells remains unknown, we retain the neutral term “bridges”.

In animals, intercellular communication and exchange rely on multiple mechanisms that depend on both the distance between cells and the nature of the information being transmitted. These mechanisms include gap junctions, cellular intercellular bridges (or analogous structures), and the diffusion of molecules through the extracellular matrix (González-Méndez et al., 2019; Heinke, 2025; Kornberg and Roy, 2014). Intercellular bridges are actin-based plasma membrane protrusions that establish cytoplasmic contacts between neighboring cells, whereas gap junctions are built around core channel-forming proteins (connexins or their homologs) that directly connect the cytosol of adjacent cells and mediate the exchange of small cytosolic molecules. Interestingly, although gap junctions have been reported in hydrozoans (a clade of Medusozoa, the sister group to Anthozoa) (Magie and Martindale, 2008), no homologs of connexins, innexins, or pannexins are encoded in octocorallian genomes (P.G., unpublished results). This absence suggests that canonical gap junctions are absent in octocorallians, thereby restricting intercellular signaling to molecular diffusion through the extra cellular matrix, such as the mesoglea, and direct intercellular bridges.

Based on our observations, these bridges appear to originate as cytoplasmic extensions from a cell of either subtype (Figure 3; Supplementary Video 2, Table 1), which then meet and connect to a cell of another subtype, forming a continuous link. In C. rubrum, these bridges can reach diameters of up to 2 µm, and the presence of granular structures within suggests the potential transfer of large cytosolic components, possibly even organelles. Such structures could therefore serve as conduits for exchange and communication between tissues otherwise separated by tens of micrometers of mesoglea.

Our histological observations indicate that bridge formation is transient, or at least unevenly distributed, within the coenenchyma: in some regions of a given preparation they are abundant, while in others they are absent. This variation suggests that bridge-based communication may be labile to endogenous stimuli or regulated in a spatiotemporal manner. Clusters of scleroblasts within the continuous scleroblastic network throughout the colony would thus be independently subjected to additional levels of developmental regulation, based on their local microenvironment, such as sclerite formation and their distance from canal cells. Such regulation would be required for the spatially coordinated production of sclerites, as well as for the supply of nutrients from the gastrodermal canals. Although the physiological conditions that promote or inhibit bridge formation remain unknown, this spatial variability may reflect a level of regulation in cell–cell communication not previously documented in octocorals. Importantly, the intercellular bridges of C. rubrum should not be confused with the dendritic processes of the secondary scleroblastic meshwork in Sarcophyton.

Although our investigation focused primarily on scleroblasts, it also revealed processes involved in the overall irrigation of an octocoral colony, specifically fluid exchange within the canal system. Octocorals possess networks of canals that connect the gastrodermal cavities of the polyps. Seawater entering through autozooids or siphonozooids forms part of this internal fluidic system. In our previous study (Loentgen et al., 2025), long-term time-lapse videos of C. rubrum demonstrated that the opening and closing of polyp mouths were synchronized across the colony (see supplementary videos in Loentgen et al., 2025). Moreover, polyp closure and contraction, which induce changes in hydrostatic pressure, were accompanied by pronounced dilation of the canals; conversely, polyp opening was associated with canal contraction. Such polyp contractions likely act as one motor driving fluid circulation within the colony, although intervals between contraction events may extend over days or even weeks.

In the present work, we further show that octocoral canals possess an external sheath of striated actin, a feature typically associated with contractile mechanisms. In addition, the canal walls are covered by tubulin-based flagella, closely resembling the motile flagella described in the coelenteric cavities of hexacorallian corals (Tambutté et al., 2021). Taken together, these observations suggest that fluid movement within the colony is driven by three coordinated mechanisms: (1) polyp contraction, (2) canal contraction, and (3) flagellar beating within the canal lumen.

All aquaria were supplied with an open-circuit flow of Mediterranean seawater and coral cultures were fed daily. Corallium rubrum colonies were maintained in separate aquaria at 15°C and cultured as described in Loentgen et al. (2025). Sarcophyton sp. mother colonies were maintained in large aquaria at 25°C, following Forin et al. (2024).

Calcein labeling: Sarcophyton sp. nubbins of approximatively 3cm long were cut from the main colony and allowed to regenerate for two weeks. They were then incubated in a 1L beaker for 24 h in seawater supplemented with calcein, followed by fixation and immunofluorescence without a demineralization step.

The protocol from fixation to immunolabeling was adapted from Ganot et al. (2015). Colony fragments of interest from C. rubrum or Sarcophyton sp. were cut with a scalpel and fixed for 24 h at 4°C in 50 ml of chilled artificial seawater/paraformaldehyde (PAF) fixation buffer [425 mM NaCl, 9 mM KCl, 9.3 mM CaCl₂, 25.5 mM MgSO₄, 23 mM MgCl₂, 2 mM NaHCO₃, 100 mM HEPES pH 7.9, 4.5% PAF]. Samples were then demineralized for 24 h in 50 ml of buffer [100 mM HEPES pH 7.8, 500 mM NaCl, 250 mM EDTA pH 8.0, 0.4% PAF], rinsed three times in PBS (10 min each), and manually sliced into thin sections.

Sections were incubated in blocking buffer [1× PBS, 0.05% Tween-20 (PBST), 2% BSA] for 1 h, followed by staining with phalloidin in blocking buffer (15 µl phalloidin/ml blocking buffer) for 1 h. Samples were then incubated with primary rabbit anti-AcTubK40 antibody (1:500) for 48 h at 4°C, rinsed three times in PBST (5 min each), and incubated with Alexa Fluor 568–conjugated anti-rabbit secondary antibody (1:1000) in blocking buffer for 24 h. Nuclei were counterstained with DAPI during the second PBST rinse. Finally, samples were mounted on coverslips in ProLong™ Gold Antifade Mountant and allowed to polymerize in the dark for 48 h before mounting on slides for confocal microscopy.

For preparations without demineralization, samples were manually cut after fixation and processed directly for phalloidin and DAPI staining. This approach is technically challenging due to the hardness of the sclerites, which necessitate larger coral sections and can hinder laser penetration during imaging.

Confocal imaging was performed using Leica SP8 confocal microscope and associated LAS-X software. Large panels were acquired using motorized stage and las-X Navigator software. 3D reconstruction videos were performed using Las-X 3D sofware, Table 1. Each channel was acquired sequentially to ascertain lack of cross-emission. Laser and PMT setups are given in the Supplementary Table 1.

For paraffin-embedded sections, a branch of C. rubrum was fixed and decalcified as described above. The sample was then dehydrated through a graded ethanol series and embedded in paraffin (Sigma). Seven-micrometer sections were cut on a Leica RM 2155 microtome and mounted on silane-coated glass slides (Sigma S4651-72EA). Paraffin was removed using baths of xylene substitute, followed by rehydration through a graded ethanol series ending with a Milli-Q water rinse. Sections were imaged using a Leica microscope equipped with a 40× objective and a CCD camera.