The Sox genes, a metazoan-specific family of HMG-domain containing transcription factors, are important regulators of animal development. In mammals and in classical non-vertebrate models (Drosophila melanogaster, Caenorhabditis elegans), studies of Sox gene expression and function have highlighted their crucial involvement in a great diversity of developmental contexts, for example, in neurogenesis, cardiogenesis, angiogenesis, chondrogenesis, and endoderm development [1–3]. Sox genes are also involved in adult tissue homeostasis and in disease, notably cancer [2–4]. At the molecular level, Sox genes activate, repress or modulate transcription of target genes through physical interaction with a variety of partner proteins. The mechanisms whereby this transcriptional regulation is mediated are remarkably diverse [5, 6].
There is significant diversity within the Sox multigenic family, with, for example, 20 paralogues in the mammalian genome and 8 in the fly genome . Previous gene phylogenies have identified five major Sox groups (named B, C, D, E and F) [8–12]. With the exception of SoxD, all of them are represented in the genomes of bilaterians as well as non-bilaterian eumetazoans (cnidarians and ctenophores) [8–10]. Several more artificial "groups" (for example, group A, G, H, I, J) have been created to accommodate single genes that are difficult to position in the Sox tree. The presence of putative members of families B, C and F in sponges furthermore suggests that Sox genes started to duplicate before the last common ancestor of Metazoa [9, 11, 12].
A recurrent theme in functional studies of Sox functions at the cellular level in bilaterian models is the involvement of various members of the family in the critical balance between self-renewing stem cells/proliferating progenitors, and cells undergoing differentiation, and their pivotal role in the regulation of this equilibrium in numerous developmental contexts [2, 3, 13, 14]. For example, the vertebrate Sox2 gene is widely known as a key factor for maintenance of mammalian ES cell pluripotency [13–17]. Its forced expression (together with Oct4, Klf4 and c-Myc) in differentiated fibroblasts leads to their re-programming into ES-like pluripotent cells . In central nervous system development, the same Sox2 gene acts in synergy with other SoxB group genes (Sox1 and Sox3) to maintain neural stem cells and to repress neuronal differentiation, whereas yet other SoxB genes (Sox21 and Sox14) promote cell cycle exit and neuronal differentiation under the control of the proneural genes [14, 17]. The same Sox protein can sometimes act on one side or the other of the balance between proliferating and differentiating cells depending on the developmental context, as is the case of the vertebrate Sox2 gene, involved in the terminal differentiation of some neuronal subtypes [13, 14, 17], in addition to its earlier function in neural stem cell maintenance.
Data from non-vertebrate bilaterians such as insects, Caenorhabditis elegans (nematode), sea limpet (mollusc), Platynereis (annelid) and sea urchin (echinoderm) suggest evolutionary conservation of at least some aspects of Sox gene functions, notably in neurogenesis and in gametogenesis [19, 20]. In the annelid Platynereis dumerilii, a SoxB group gene was found to be expressed in the neurectoderm before the formation of committed neural precursors, while the expression of a SoxC group gene evoked a role in neuronal differentiation . These data are consistent with involvement of these two genes at different sides of the balance between cell proliferation and differentiation along the neuronal cell lineage. However, for most invertebrate Sox genes (including Drosophila genes), expression and function have not been precisely characterised in terms of stages and progression along cell lineages. Therefore, it remains unclear whether Sox family genes have evolutionarily conserved roles in these processes, and if it is the case, whether each particular Sox orthology group was ancestrally associated with one particular side of the balance, that is, either with stem cells/progenitors, or with differentiating cells.
Studies in animal lineages that branch outside bilaterians are expected to be informative about the early stages of animal evolution. Phylogenetic relationships between the early-diverging animal phyla remain contentious [21–24], but a critical re-analysis of data sets used in recent phylogenomic studies suggests that apparent conflicts between them disappear when errors are corrected and appropriate taxon sampling and models are used . Currently the best-supported phylogenomic estimate of basal metazoan relationships implies the monophyly of animals with nerve cells and muscle cells (Eumetazoa: cnidarians, ctenophores and bilaterians) in line with classical views, and the grouping of cnidarians and ctenophores in a coelenterate clade sister-group to the Bilateria . Previously-published data on Sox gene expression in two anthozoan cnidarians (the sea anemone Nematostella vectensis, and the coral Acropora millepora and in a ctenophore  have started to unveil conserved features of Sox gene expression at the eumetazoan level. In particular, these three studies all concluded that involvement of some Sox genes in neuro-sensory cell specification and differentiation probably dates back to the common eumetazoan ancestor. In addition, two of the ctenophore Sox genes were found expressed in the germ line as well as in several somatic territories recently characterised as reservoirs of somatic stem cells . It was therefore proposed that Sox roles in the balance between stem cells/progenitors and differentiating cells might be conserved at the eumetazoan scale . However, there is currently no data from cnidarians to fuel this hypothesis, notably because stem cells and progenitors have not been characterised in the larvae and adults of the two anthozoans in which Sox gene expression has been investigated [8, 26].
To gain insight into evolutionary conservation and divergence of Sox gene expression characteristics in relation to progression along cell lineages, we investigated the expression of Sox genes in the hydrozoan cnidarian Clytia hemisphaerica. Hydrozoan cnidarians have multipotent stem cells, called interstitial stem cells, whose progeny comprises neuro-sensory cells (including the stinging cells or nematocytes), gland cells, and germ cells [29–34]. These interstitial cells appear in the endoderm after gastrulation . The planula larva has endodermal patches of interstitial stem cells already providing larval nematoblasts, nerve cells and gland cells . Upon metamorphosis, interstitial cells migrate to the ectoderm, where they remain localised in the adults [29, 34]. The C. hemisphaerica life cycle comprises two alternating adult forms: the asexual benthic colony of polyps, and the sexual pelagic medusa . In a previous work , it was shown that the medusa contains localised populations of somatic stem cells, notably two symmetrical patches of stem cells positioned in the proximal region of each tentacle bulb. Tentacle bulbs are specialised basal swellings of the tentacles, in which tentacle nematocytes are generated all life long. There is a gradient of nematogenesis stages from the proximal to the distal pole of the tentacle bulb axis . Thanks to these features, genes expressed during nematogenesis in the medusa can be easily characterised either as stem cell/progenitor genes or as early or late differentiation genes, based on the spatial position of their expression zone along the tentacle bulb axis.
Here, we present detailed expression analyses of 10 Sox genes (five members of group B, one group C gene, 2 members of group E, one group F gene and one unclassified Sox gene) in the Clytia hemisphaerica planula larvae, medusae and eggs. The results suggest conservation at a deep evolutionary level of the general features of Sox gene expression: the SoxF orthologue has endodermal expression in Clytia like in other non-bilaterian animals investigated so far, whereas for all other orthology groups, the genes are expressed either in somatic stem cells and in the germ line, or in differentiating/differentiated cells with a neuro-sensory identity (either nematocytes or nerve cells). However, comparison with gene expression data from ctenophore and bilaterians reveals total lack of correlation between any particular Sox group and expression/function in stem cells/progenitors vs. in differentiating cells, thus indicating that the roles of individual Sox genes can easily switch from one side to the other of the balance, in different developmental and evolutionary contexts.