Intra-epithelial spicules in a homosclerophorid sponge.

Attempts to understand the intricacies of biosilicification in sponges are hampered by difficulties in isolating and culturing their sclerocytes, which are specialized cells that wander at low density within the sponge body, and which are considered as being solely responsible for the secretion of siliceous skeletal structures (spicules). By investigating the homosclerophorid Corticium candelabrum, traditionally included in the class Demospongiae, we show that two abundant cell types of the epithelia (pinacocytes), in addition to sclerocytes, contain spicules intracellularly. The small size of these intracellular spicules, together with the ultrastructure of their silica layers, indicates that their silicification is unfinished and supports the idea that they are produced "in situ" by the epithelial cells rather than being incorporated from the intercellular mesohyl. The origin of small spicules that also occur (though rarely) within the nucleus of sclerocytes and the cytoplasm of choanocytes is more uncertain. Not only the location, but also the structure of spicules are unconventional in this sponge. Cross-sectioned spicules show a subcircular axial filament externally enveloped by a silica layer, followed by two concentric extra-axial organic layers, each being in turn surrounded by a silica ring. We interpret this structural pattern as the result of a distinctive three-step process, consisting of an initial (axial) silicification wave around the axial filament and two subsequent (extra-axial) silicification waves. These findings indicate that the cellular mechanisms of spicule production vary across sponges and reveal the need for a careful re-examination of the hitherto monophyletic state attributed to biosilicification within the phylum Porifera.