Stem cells (SCs) are undifferentiated cells capable of self-renewal and with the ability to differentiate into specialised cells. Given their pivotal role in animal development and regeneration, understanding the molecular mechanisms controlling their potency and evolution is of critical importance. Studies in SC biology are mainly based on mammals or other vertebrates, despite their limited regenerative capacities (Zhao et al., 2016). Tunicates are the sister group of vertebrates (Delsuc et al., 2006), and colonial tunicates are the only chordates able to regenerate entire organisms from circulating SCs (Rinkevich et al., 1995; Sunanaga et al., 2006; Voskoboynik et al., 2007; Brown et al., 2009; Kassmer et al., 2019; Manni et al., 2019; Scelzo et al., 2019). As such, this group offers unique opportunities to investigate the similarities and differences in SC biology between vertebrate and invertebrate chordates.

Among colonial tunicates, the ascidian Botryllus schlosseri is recognized as a model species for the study of SCs and asexual reproduction (Manni et al., 2007; Voskoboynik and Weissman, 2015; Manni et al., 2019). This species reproduces both sexually and asexually. In a colony, each individual (zooid) is hermaphroditic and produces swimming larvae via sexual reproduction. Larvae metamorphose into sessile individuals that reproduce asexually by budding (blastogenesis), forming clonal zooids grouped in star-shaped systems and embedded in a common tunic (Figures 1A,B). In each colony, three generations coexist: adult zooids, their primary buds (1Bs), and secondary buds (2Bs). In laboratory conditions (18°C), each week all adult zooids are resorbed during a process called takeover and replaced by their 1Bs; simultaneously, 2Bs become 1Bs and produce a new generation of 2Bs. Each new 2B originates from a thickening of the peribranchial epithelium of a 1B, which further develops into a vesicle undergoing organogenesis. Organogenesis nears completion when the 2B becomes a 1B where tissue differentiation occurs. B. schlosseri SCs are able to freely circulate in the colony haemocoel where, over generations, they contribute to tissue differentiation in buds (Kowarsky et al., 2021). Candidate SCs (cSCs) were identified in B. schlosseri as hemoblast, i.e., round cells, with a diameter of about 5 μm and a high nucleus-cytoplasm ratio (Rosental et al., 2018; Kowarsky et al., 2021). Two cSC transient niches have also been proposed in adult zooids (Figures 1E–H): the haemocoel, ventral to the anterior endostyle (the gland on the pharynx floor which produces the mucus net for filtration) (Voskoboynik et al., 2008; Rosental et al., 2018), and the “cell islands'', i.e., groups of haemocytes in the ventral body wall (Rinkevich et al., 2013). Cell islands are numerous in adult individuals belonging to early-cycle colonies and decrease in number toward the late-cycle. cSCs reside and proliferate in these transient niches and migrate before takeover through the haemocoel to relocate in newly formed niches of the following generations (Rinkevich et al., 2013).

The four transcription factors c-Myc, Sox2, Pou5f1 (also named Oct3/4) and Klf4 are of interest given their role in tissue differentiation and in both regulating and inducing pluripotency in vertebrates. Sox2 and Pou5f1 are involved in maintaining pluripotency of embryonic stem cells (Niwa et al., 2000; Avilion et al., 2003; Boyer et al., 2005), while c-Myc and Klf4 play a key role in self-renewal and maintenance of the undifferentiated state (Cartwright et al., 2005; Li et al., 2005). Remarkably, the forced co-expression of these four transcription factors, also named Yamanaka Factors (YFs), is sufficient to reprogram murine and human fibroblasts into induced pluripotent stem (iPS) cells (Takahashi and Yamanaka, 2006; Takahashi et al., 2007).

c-Myc belongs to the MYC gene family, characterized by a C-terminal basic-helix-loop-helix-leucine zipper (bHLHZ) domain mediating protein dimerization and DNA binding, and two N-terminal trans-activation motifs (Atchley and Fitch, 1995; James and Eisenman, 2002). Vertebrates possess multiple Myc genes (c-Myc, N-Myc, L-Myc), with c-Myc involved in cell division and growth control, and considered a central hub for pluripotent SC regulatory networks (Chappell and Dalton, 2013; Fagnocchi and Zippo, 2017). In contrast, two genes are found in a cnidarian with high regenerative ability (Hartl et al., 2010), and only a single homolog is present in Drosophila and most other invertebrates (Gallant, 2006; McFerrin and Atchley, 2011). Despite differences in gene number, the expression of the single Myc gene in the undifferentiated cells of invertebrates suggests that its function is similar to that of c-Myc in vertebrates (Kawamura et al., 2008; Hartl et al., 2010; Fujiwara et al., 2011).

Sox2 is a member of the ancient SOX gene family that has undergone frequent gene duplications and is therefore divided into ten subgroups (A-J) based on the similarity of the DNA-binding domain (Wegner, 1999; Bowles et al., 2000). In vertebrates, Sox2, Sox1 and Sox3 comprises the SoxB1 group, while the close SoxB2 group consists of Sox14 and Sox21. Invertebrates have only a single member each of the SoxB1 and SoxB2 groups (Bowles et al., 2000; Heenan et al., 2016). The role of SoxB1 in pluripotency appears to be conserved among invertebrates (Onal et al., 2012; Niwa et al., 2016), where it is fundamental for neural development and regeneration (Mizuseki et al., 1998; Buescher et al., 2002; Bylund et al., 2003; Graham et al., 2003; Miya and Nishida, 2003; Meulemans and Bronner-Fraser, 2007; Marlow et al., 2009; Ross et al., 2018) and is expressed in multiple adult self-renewing epithelia (Arnold et al., 2011).

Pou5f1 belongs to the metazoan POU gene family, characterized by a modular DNA-binding domain (POU) consisting of a N-terminal POU-specific domain (POUs) and a C-terminal homeodomain (POUh) separated by a linker of variable length. Based upon similarity across the entire POU domain, Pou genes are grouped into six classes (Rosenfeld, 1991), that have evolved through multiple lineage-specific events of gene duplications and losses (Frankenberg and Renfree, 2013; Gold et al., 2014; Onichtchouk, 2016). Pou5f1 belongs to the vertebrate-specific POU5 class (also including Pou5f3 and the primate/rodent-specific Pou5f2), hypothesised to be arisen from a Pou3-like ancestor together with the bilaterian-specific POU2 class (Gold et al., 2014).

Klf4 is a member of the Kruppel like factors (Klf) family, characterized by the presence of three C-terminal highly conserved zinc finger motifs and a high diversification of the associated transactivation/repression domains (Vangapandu and Ai, 2009; Presnell et al., 2015). Depending on their structure and/or function, the Klf members are partitioned into different groups whose number varies across studies (McConnell and Yang, 2010; Pei and Grishin, 2015; Presnell et al., 2015; Pollak et al., 2018). Pei and Grishin (2015) found a minimum of 18 Klf genes in vertebrates and classified them into seven groups (A-G), with invertebrate deuterostomes possessing a single member of each group and the loss of KlfE in the ascidian Ciona intestinalis. Together with Klf1/2/17, the human Klf4 belongs to the group KlfA.

In this study we investigate the presence and the evolution of the YF genes in ascidians, and analyze when and where the orthologs of the vertebrate YF genes are expressed in the blastogenetic cycle of B. schlosseri by using in situ hybridization (ISH) experiments. Our results show patterns of expression consistent with the reported function of these genes in invertebrate and vertebrate development and differentiation. Moreover, the expression of Myc, SoxB1 and Pou3 is retrieved in cSCs residing in cSC niches, suggesting these transcription factors may play a role in the maintenance of undifferentiated state, and highlighting the necessity to better investigate whether similar molecular mechanisms control pluripotency in vertebrates and tunicates.

All YF gene families have a complex evolutionary history characterized by several gene duplications, some of which also occurred in the ancestor of vertebrates. Our evolutionary analyses show that a one-to-many relationship exists between ascidian and human orthologs of some YF genes, which means that the ascidian gene has more than one ortholog in humans. Indeed, the ascidian SoxB1 is the ortholog not only of the human/vertebrate YF Sox2, but also of Sox1 and Sox3. Likewise, the ascidian Myc gene is the ortholog not only of the human/vertebrate YF c-Myc, but also of N-Myc and L-Myc. This implies that the functional role and the expression pattern of these ascidian genes could have been inherited by one or many human/vertebrate orthologs. With that in mind, we have analysed the expression pattern of the YF genes in B. schlosseri, a highly regenerative model species belonging to the taxon evolutionarily closest to vertebrates.

Our results show that Myc is expressed in all developing tissues of buds and in cSCs in niches (Figures 1C,D), suggesting a crucial role in undifferentiated territories. Noteworthy, Myc expression in cSCs in their niches is a new finding, not observed in previous studies on Myc expression in two other colonial ascidians. Indeed, in Botryllus primigenus, Myc was proposed as a marker of undifferentiated, but not proliferative, somatic and germline SCs, as it was expressed by multipotent epithelial cells in buds, undifferentiated coelomic cells, and developing gonads (Kawamura et al., 2008). In Polyandrocarpa misakiensis, Myc expression was observed in the atrial epithelium in the organ-forming region of developing bud, and in fibroblast-like cells, which participate in the organogenesis together with the epithelial cells (Fujiwara et al., 2011). Moreover, the noteworthy identification of an ascidian-specific C-terminal tail, located just downstream the bHLHZ domain, incentivises further investigation on the role and the possible additional functions performed by this protein region.

As summarized in Figure 1C, SoxB1 is widely expressed during B. schlosseri asexual development. It is found in the developing nervous system, as has also been reported for Sox2 and other SoxB1 group genes in many metazoans (Wood and Episkopou, 1999; Penzel et al., 1997; Rex et al., 1997; Meulemans and Bronner-Fraser, 2007; Ferrero et al., 2014; Ross et al., 2018). In the developing nervous system we also detected the expression of Pou3 (Figure 1C), known to be involved in neural system development in vertebrates and other invertebrates, (Candiani et al., 2002; Cole and Arnone, 2009; Wollesen et al., 2014; Garner et al., 2016; Wei et al., 2016; Prünster et al., 2019). The expression, in samples at the same developmental stages, of SoxB1 and Pou3 in the developing nervous system, observed here and in previous studies (Iwafuchi-Doi et al., 2011; Hui et al., 2015), suggests a co-regulation of these two factors during neurogenesis. We also retrieved SoxB1 expression in the developing oesophagus, stomach and proximal intestine, suggesting a role in the differentiation and maintenance of these epithelia. This result corresponds with the Sox2 expression in the adult stomach tissues of mammals (Arnold et al., 2011). Moreover, SoxB1 was found expressed in the endostyle of 1B, and in cSCs of the endostyle niche and cell islands, indicating potential involvement in stemness. Supporting this hypothesis, Niwa et al. (2016) demonstrated that C. robusta and other invertebrate SoxB1 proteins are able to replace the pluripotency-associated function of mouse Sox2. Thus the role of SoxB1 in the control of pluripotency in B. schlosseri is plausible.

Our evolutionary analyses demonstrate the absence of the POU5 class in ascidians, indicating that POU5 arose after the tunicate-vertebrate divergence. Since there is no true ortholog of Pou5f1 to be investigated in B. schlosseri, the ancestor of the POU5 class could instead be investigated in its place. Indeed, it remains unclear whether the role of vertebrate Pou5f1 in SC pluripotency was inherited from the ancestral paralog (see discussion in Gold et al., 2014). Our POU evolutionary tree shows a basal polytomy that prevents identifying the POU5 ancestor (Figure 4A). However, our tree is in agreement with a previous wide evolutionary analysis of the entire POU family (including POU1) in all Metazoa (Gold et al., 2014). In that study, a polytomous clade including POU5, POU3 and POU2 classes was identified, and the POU4 class was found as the paralog that diverged after the POU6+POU1 basal clade. Relying on the ancestral state reconstruction and on the taxonomic distribution of POU2 and POU5, Gold et al. (2014) hypothesized that a Pou3-like gene was the last common ancestor of the POU5/3/2 polytomous cluster, in which POU2 and POU5 classes represent bilaterian-specific and vertebrate-specific duplications, respectively. On this basis, and given the goal of our study, we investigated the expression pattern in B. schlosseri of both Pou3, as representative of the Pou3-like ancestor of Pou5f1, and Pou2, as the other closest paralog of Pou5f1.

Remarkably, while POU2 and POU4 are present in all analysed ascidians, Pou3 was found only in ascidians of two Stolidobranchia families (i.e., Styelidae and Pyuridae; Supplementary Table 1 and Figure 4A) and seems to be absent even in the two widely studied Ciona species with highly curated genomic sequence annotations. This leads us to hypothesize a Pou3 loss in most ascidian groups, although the long branch/fast evolution of the ascidian Pou3 clade (Figure 4A) could also explain the failure to identify these genes. As summarized in Figure 1D, we found that B. schlosseri Pou3 is expressed in the endostyle, even in the adult phase, especially in the anterior-most part, close to the previously identified SC niche (Rinkevich et al., 2013). Moreover, Pou3 is expressed in the few pluripotent cells of the early 2B and cSCs in their niches. This expression pattern highlights a likely role of B. schlosseri Pou3 in undifferentiated cells, as also recently evidenced in Botrylloides diegensis/leachii (Viard et al., 2019; Kassmer et al., 2020). Overall, our and previous results support a role in SC pluripotency for the Pou3-like ancestor that gave rise to the vertebrate Pou5f1, here represented by the Pou3 of ascidians. Finally, Pou3 was found expressed in the differentiating heart, gut and gonads, and in the developing nervous system (see above).

As for Pou2, it is expressed exclusively in the immature previtellogenic oocytes, both in gonads and circulating in the colonial vascular system. Remarkably, this expression pattern significantly differs from those of vertebrates and invertebrates, where Pou2 is expressed in many more territories. For example, Pou2 is involved in neurogenesis in cephalopods (together with Pou3) (Wollesen et al., 2014), in the establishment of the dorsal/ventral axis in echinoderms (Range and Lepage, 2011) and in immunity in vertebrates (Strubin et al., 1995; Karnowski et al., 2012). Notably, this is the first report of Pou2 expression in tunicates.

Overall, these results point to an expression pattern of the vertebrate Pou5f1 more similar to that of B. schlosseri Pou3 rather than Pou2, thus supporting the Gold’s and coll. (2014) hypothesis of a Pou3-like gene as ancestor of Pou5f1.

To summarize, we identified in the model species B. schlosseri the orthologs of vertebrate YFs through deep evolutionary analyses. Then, we found that Myc, SoxB1 and Pou3 are individually expressed in several differentiating territories, consistently with the reported function of these genes in chordate embryogenesis. Moreover, we found their expression in cells with features of cSCs exhibiting the same morphological and morphometric features. These cells are located close to the endostyle and in cell islands, and according to our analyses, should coexpress at least Pou3 and Myc. Further studies, such as a deep analysis of niche dynamic and double in situ hybridization experiments, are needed to understand the actual co-expression of these factors in cSCs, which could in turn suggest a shared ancestral pluripotency program in chordates.