New roles for Nanos in neural cell fate determination revealed by studies in a cnidarian

Nanos is a Zn-finger domain, RNA binding protein acting as a translational repressor. It is known to act in concert with its canonical partner Pumilio. Together, these proteins repress translation of their target genes. Nanos was first described as a posterior marker in Drosophila (Lehmann and Nüsslein-Volhard, 1991) and subsequent studies established three main roles for Nanos: first, it is necessary for fly posterior patterning by inhibiting hunchback translation. Second, it is crucial for germ cell development, where it mediates self-renewal of germ stem cells, their migration to the gonads, and suppresses somatic fate and premature differentiation (Hayashi et al., 2004). Finally, Nanos fulfils some functions in the nervous system of Drosophila (Ye et al., 2004; Brechbiel and Gavis, 2008). Nanos expression patterns in various animals are consistent with these functions.

The first identified role of Nanos, posterior patterning, has only been demonstrated experimentally in insects (Schmitt-Engel et al., 2012). The role of Nanos in germ cells, in contrast, seems to be evolutionarily conserved (Mochizuki et al., 2000; Shen and Xie, 2010) and is well documented in a variety of animal species. Using Nanos as a germline marker has been common practice even in the absence of functional studies in non-model organisms (Ewen-Campen et al., 2010; Juliano et al., 2010; Wu et al., 2011). Lastly, the function of Nanos in the nervous system is by far the least well studied. Work done in Drosophila has shown that Nanos is required for dendrite morphogenesis (Ye et al., 2004; Brechbiel and Gavis, 2008) through an unclear mechanism. In the mouse, Nanos1 is expressed in the central nervous system, but its function is unidentified (Haraguchi et al., 2003). A role for Nanos in other animals' nervous systems is unknown.

We aimed at studying the role of Nanos in neural development in a member of a basal metazoan phylum, the Cnidaria. Our rationale for conducting this study was twofold: first, cnidarian morphological simplicity enables investigating the role of a potentially important gene in neural development at a cellular resolution in vivo. Second, the pivotal phylogenetic position of cnidarians at the base of the Metazoa could provide insight into the early evolution and ancient role of Nanos.

The phylum Cnidaria includes sea anemones, corals, jellyfishes and hydroids. Members of this group display a relatively simple body plan with few cell types that include epithelial, muscle, gland/secretory, neural and stem cells. Some cnidarians are quite well established as model organisms (Putnam et al., 2007; Chapman et al., 2010; Plickert et al., 2012). Hydractinia echinata (Fig. 1) is a common cnidarian representative in the European North Atlantic. Throughout the animal's life cycle, a population of stem cells, called interstitial cells, or i-cells for short, continuously provides progenitors to all somatic and germ cell lineages. Hydractinia reproduces sexually on a daily basis, has a short and accessible embryonic development, and is amenable to gene expression analysis and manipulation.

The cnidarian nervous system is of particular interest because it is continuously being renewed in the adult animals and because the molecular mechanisms that specify neurons are, at least in part, well conserved with higher animals (Galliot et al., 2009). The main defining feature of cnidarians is a specific sensory/effector cell type, the stinging cell, or nematocyte (also known as cnidocyte). These chemo/mechanosensory cells include a specialized post-Golgi capsule, called nematocyst (cnida). Nematocysts can explosively discharge upon mechanical/chemical stimulation, extruding a barbed tubule and delivering a venom cocktail into prey and predator tissues.

Using gain and loss of function experiments we show that Hydractinia Nanos2 controls nematogenesis at different stages from specification through proliferation to maturation.

Searching EST databases (Soza-Ried et al., 2010) (and an unpublished 454 library) we identified two cDNA fragments with high sequence similarity to cnidarian and bilaterian Nanos genes. The complete coding sequences were obtained by RACE PCR. The genes were named Hydractinia echinata Nanos1 and Nanos2, hereafter Nanos1 and Nanos2, respectively. Two Nanos genes were also identified in other cnidarians by others. A Nanos phylogeny has been generated previously (Extavour et al., 2005), suggesting that the two cnidarian Nanos genes are a result of a lineage specific duplication. We have repeated this analysis, including also the Hydractinia sequences, and our phylogeny is consistent with the data of Extavour et al. (supplementary material Fig. S1).

A single Hydractinia Pumilio gene sequence was obtained from the draft assembly of the Hydractinia genome (The Hydractinia Genome Consortium) and named Hydractinia echinata Pumilio, hereafter Pumilio. Additional Nanos or Pumilio sequences were neither identified in the Hydractinia EST database or genome draft, nor in other cnidarians genomic and transcriptomic databases, suggesting that cnidarian genomes encode 2 Nanos and 1 Pumilio genes.

We initially performed in situ hybridization to study the normal expression patterns of Nanos2, Nanos1 and Pumilio. Zygotic transcription in Hydractinia commences at the morula/gastrula stage (Plickert et al., 2006). Nanos2 was a maternal transcript, asymmetrically deposited in the prospective oral pole of the embryo and in a Nuage-like deposition around the nuclei of early cleavage stages (Fig. 2A,B). A similar expression has been observed in the hydrozoan Clytia (Leclère et al., 2012). Polar bodies are given off at this pole of cnidarian oocytes (Freeman, 1981). This is also the pole where β-catenin is stabilized and the blastopore is formed in species with polar gastrulation (Wikramanayake et al., 2003). Later on this pole develops into the larval posterior end (Plickert et al., 2006). However, following metamorphosis the larval posterior end develops into the adult oral pole of the polyp (also referred to as head; Fig. 1).

Following gastrulation, Nanos2 was expressed in developing stinging cells (nematocytes; Fig. 2C). Hydrozoan nematocytes and their progenitors, nematoblasts, can be identified morphologically by two ways: first, as nematoblasts proliferate after becoming committed they form characteristic nests or clusters (Lindgens et al., 2004); second, postmitotic nematoblasts or early maturing nematocytes are migratory and can be identified by the developing nematocyst capsule in their cytoplasm.

In the gastrula, at 24 hours post fertilization (hpf) and in the planula larva (48 hpf), Nanos2 was expressed in two stages of nematocyte differentiation: first, in clusters of dividing nematoblasts in the endodermal cell mass; second, in maturing nematocytes in both the endoderm and in the ectoderm (Fig. 2C). Note that in cnidarian literature ‘ectoderm’ and ‘endoderm’ usually refer to the outer and inner cell layers (also known as epidermis and gastrodermis), respectively, regardless of their embryonic origin. In hydroids like Hydractinia, for example, i-cells, neurons and nematocytes are predominantly found in the ‘ectoderm’ of the adult, but are of embryonic endodermal origin and migrate to the outer cell layer (‘ectoderm’ or epidermis) in late embryonic development and metamorphosis. Nanos2 was also expressed in metamorphosed animals in mitotic nematoblasts clusters at the base of the polyp, in single migrating nematocytes in the upper part of the body column, and in basally positioned, maturing nematocytes in the tentacles (Fig. 2D; supplementary material Fig. S2C). Fully mature nematocytes, recognized by their capsule with coiled tubule and by their apical position, being mounted in battery cells, were devoid of Nanos2 transcripts; only basally positioned nematocytes were Nanos2+ (supplementary material Fig. S2C). In sexual polyps, the Nanos2 probe marked early developing germ cells in both male and female individuals, but was undetectable in late developing and mature gametes (Fig. 2E,F; supplementary material Fig. S2D). In the stolons, strong Nanos2 expression was observed in nematoblasts and maturing nematocytes (Fig. 2G), but it was also expressed in cells resembling uncommitted i-cells, based on their very large nucleus and little cytoplasm (Fig. 2G; supplementary material Fig. S2A,B). No Nanos2 expression was observed in epithelial cells, in contrast to Nanos2 in Hydra (Mochizuki et al., 2000).

Nanos1 was strongly expressed in early developing female gametes (supplementary material Fig. S2E), but never observed in males. Furthermore, its mRNA levels in all other cell types were undetectable by in situ hybridization. Noteworthy, in a related cnidarian, Clytia hemisphaerica, CheNanos1 has a similar expression as CheNanos2 (Leclère et al., 2012). In Hydra, Nanos1 is expressed in germ cells and in multipotent stem cells (Mochizuki et al., 2000). Hydractinia Nanos1 has a similar expression pattern as mammalian Nanos2-3. However, while mammalian Nanos2 is male specific, Hydractinia Nanos1 is female specific. Hydractinia Nanos2 expression in the nervous system resembles mammalian Nanos1. Yet, unlike Hydractinia Nanos2, mammalian Nanos1 is not expressed in germ cells. The Drosophila single nanos gene is expressed in both germ cells and in the nervous system, similar to Hydractinia Nanos2.

Pumilio transcripts were found in germ cells (supplementary material Fig. S2F) and in dividing nematoblasts clusters at the base of polyps (supplementary material Fig. S2G). The cell type in the latter could not be identified definitely, but at least some of the cells had nematoblast morphologies. Hence, Pumilio expression overlapped with Nanos2 in germ cells but only partly in somatic cells.

Because Nanos2 expression was associated with nematogenesis, we also studied the expression pattern of the known nematocyte lineage marker Ncol1, which has not been studied in Hydractinia previously. Ncol1 and other minicollagens are classical cnidarian nematocyte differentiation markers (Kurz et al., 1991). These proteins form part of the nematocyst capsule and are not expressed by any other cell type. Minicollagen mRNA is detectable only in nematogenesis but not in mature nematocytes. Hydractinia Ncol1 mRNA was first expressed in single, nematoblasts in the endodermal cell mass of 24 hpf embryos and in larvae (Fig. 2H,I), and almost never in clusters of dividing nematoblasts. Young metamorphosed animals had Ncol1+ cells primarily in their stolons (Fig. 2J). In mature polyps, Ncol1 was expressed in a belt-like fashion, primarily in single, post-mitotic nematoblasts and only rarely in proliferating clusters (Fig. 2K). Note, however, that nematogenesis occurs in two different anatomical locations: in the polyps and in the stolons. There is no evidence for migration of mature nematocytes from polyps to the stolons or vice versa, but i-cells, their progenitors, generally migrate from the stolons to developing polyps.

Nematocytes are thought to belong to the cnidarian nervous system (Marlow et al., 2009). We therefore sought a definitive neuronal marker and have chosen the RFamide precursor gene (hereafter Rfamide). Although not expressed in all Hydractinia neurons (Katsukura et al., 2003), RFamide is a definitive and most commonly used neuronal marker in cnidarians, and is not expressed in any other cell type (Fig. 2L). A pan-neuronal marker is difficult to identify: Elav is also not expressed in all neurons in Nematostella (Nakanishi et al., 2012), and with the use of anti-acetylated tubulin antibodies it is difficult to quantify cells by counting because they stain the neurites. The proneural gene Ash (known in Hydra as Cnash) is expressed in both nematoblasts (Grens et al., 1995) and in neurons (Hayakawa et al., 2004).

Given the expression pattern of Nanos2 in proliferating nematoblasts clusters and in maturing nematocytes, we performed two-color double in situ hybridization of Ncol1 and Nanos2 to better understand gene expression along nematogenesis (Fig. 3). We found that Ncol1 and Nanos2 were largely expressed in distinct domains along the nematocyte differentiation pathway with only little overlap at the interface (Fig. 3A). Hence, Nanos2 is expressed in mitotic nematoblast clusters, but its mRNA dissipates soon after Ncol1 expression commences. Nanos2 expression resumes as Ncol1 transcription ceases and is maintained until late nematocyte maturation, absent from mature nematocytes.

The role of Nanos in germ cells is well established in Drosophila and in mammals. However, its neuronal role is not well understood. Because Hydractinia Nanos1 was exclusively expressed in germ cells, whereas Nanos2 was expressed also in proliferating nematoblasts (which are sensory cells), we decided to study the function of Nanos2. Based on its expression pattern, as described above, we thought that Nanos2 might be required in early nematogenesis. To test this, we used a morpholino antisense oligonucleotide to lower Nanos2 protein levels in embryos. Fertilized eggs were microinjected with a translation block morpholino at 0.25 mM. A scrambled morpholino at the same concentration was used as control (supplementary material Table S1). We found that morphant embryos had a significantly reduced number of Ncol1 positive nematoblasts (32.5% decrease comparing to control; P = 0.01; 10 control animals counted and 12 morphants; Fig. 4A) and mature nematocytes in both larvae (37.5% decrease, P<0.05; 49 control animals counted and 49 morphants; Fig. 4B) and metamorphosed polyps (50% decrease, P<0.0001; 29 control animals and 29 morphants counted; Fig. 4C).

We then analyzed the effect of Nanos2 downregulation on neurons. This was done by in situ hybridization using an Rfamide probe. We found that morphants had higher numbers of Rfamide positive neurons compared to animals injected with control morpholino (36% increase, P<0.03; 7 control larvae counted and 16 morphants; Fig. 4E,F). Note also, that because we work with wild-type, genetically heterogeneous animals the number of neurons varies between individuals. To minimize this effect we performed each experiment on a single batch of sibling animals.

Knocking down of Nanos2 also impacted the morphology of metamorphosed polyps; they had significantly reduced tentacle numbers, from an average of 5.9 tentacles per polyp (n = 32) in controls, to 1.4 tentacles per polyp (n = 25) (77% decrease, P<0.0001; Fig. 4C,D) in morphants, but appeared normal otherwise.

Hence, downregulation of Nanos2 resulted in decreased number of nematoblasts and nematocytes, but an increase in the number of neurons. It was also associated with a decrease of tentacle numbers post metamorphosis.

The specificity of the morpholino was demonstrated by in vitro expression of Nanos2-His-tag with the Nanos2 morpholino and the control morpholino. Western blot using anti-His-tag antibody detected a band at the approximate expected size (37 kDa) in the lane containing the product of the control morpholino but not in the one with the Nanos2 morpholino (Fig. 4G).

We then went on to study the effect of excessive Nanos2 expression. For this, we generated transgenic animals expressing Nanos2 ectopically. We used a construct consisting of the Hydractinia Actin1 promoter upstream of the full Nanos2 coding sequence, fused in-frame with enhanced green fluorescent protein (eGFP) at the 3′ end (supplementary material Fig. S3). The Hydractinia Actin1 promoter is active ubiquitously in embryogenesis and in the larval stage. However, following metamorphosis its activity switches to become epithelial specific, being quiescent in all other cell types of the animal (Künzel et al., 2010; Millane et al., 2011; Duffy et al., 2012). Hence, using this construct enabled us to study not only the consequences of Nanos2 ubiquitous overexpression in embryogenesis, but also its long-term effect in epithelial cells that do not express the gene naturally in Hydractinia.

Fertilized eggs were microinjected with the construct solution. Stable GFP expression was evident 24–48 hours later (supplementary material Fig. S4A). The integration of the construct into the embryo's genome is a stochastic process and the injected embryos were therefore mosaics (supplementary material Fig. S4A). Animals with a high Nanos2–GFP expression mostly looked distorted, having severe developmental defects, and most of them died before reaching the larval stage (supplementary material Fig. S4B). Those with fewer Nanos2–GFP expressing cells appeared normal. Control animals injected with a construct consisting of the same promoter driving only GFP expression appeared normal independent of the level of their GFP expression (not shown).

Because Nanos2 downregulation compromised nematogenesis, we counted nematoblasts in transgenic embryos, which overexpressed Nanos2. Anti-NCol1 antibodies showed, as expected, that the numbers of nematoblasts were significantly higher in Nanos2 transgenic larvae comparing to wild-type larvae (Fig. 5A; average increase of 110%; P<0.005; 33 wild-type larvae counted, 34 transgenic ones). Following metamorphosis, when the transgene's expression shifted to epithelial cells and could no longer affect nematogenesis, the numbers of NCol1+ cells was still elevated in 24 hours primary polyps (Fig. 5B; 25% average increase) but was no longer statistically significant (P = 0.1). The expression domain of NCol1 was also still broadened in transgenic primary polyps as compared to wild-type animals in which NCol1+ nematoblasts were always restricted to the lower part of the polyp body column and never reached the tentacle line. In Nanos2 transgenic animals, in contrast, we observed NCol1⁺ nematoblasts well above the tentacle line (Fig. 5B). Surprisingly, when we counted the numbers of mature nematocytes in transgenic larvae we discovered that, despite the elevated numbers of nematoblasts, these transgenic embryos and larvae contained significantly fewer mature nematocytes (average decrease of 45%; 11 wild-type and 5 transgenic animals analyzed; P<0.03) (Fig. 5C). This trend was reversed in post-metamorphic transgenic animals in which the number of mature nematocytes was higher than in wild-types as expected. Nematocytes were not directly counted in post-metamorphic animals due to their excessive numbers. Instead, we measured their average density and found no significant differences between transgenic and wild-type polyps (Fig. 5D). As transgenic polyps had more tentacles (see below), we concluded that nematocyte numbers were proportionately higher in transgenic animals.

To study the source of the excessive nematoblast numbers in Nanos2 transgenic larvae, we compared the numbers of S-phase cells in transgenic versus wild-type embryos by BrdU labeling. We found that transgenic animals overexpressing Nanos2 had significantly higher numbers of BrdU positive cells than wild-type animals (Fig. 5E; 90.5% increase; P<0.0001; 19 wild-type embryos counted; 18 transgenic ones).

We then looked at the numbers of neurons. We found that the numbers of Rfamide positive neurons were significantly reduced in transgenic embryos, larvae and metamorphosed polyps (average of 83% decrease in larvae, 22 wild-types counted, 25 transgenic animals; P<0.0001; Fig. 5F,G).

In addition to the changes in nematoblasts and neurons numbers, metamorphosed animals also displayed abnormal morphologies. The mildest phenotypes was supernumerary, but correctly positioned, tentacles (Fig. 6A,B). Their numbers increased significantly from an average of 9.32 tentacles per wild-type polyp (n = 54) to 14.28 in transgenic ones (n = 34) (increase of 53% in transgenic animals; P<0.0001). This abnormal phenotype was not evident immediately post metamorphosis, but required at least one week to develop. A more severe phenotype was observed in polyps which had not only supernumerary tentacles but also grew them ectopically down the body column (Fig. 6C,D). All ectopic tentacles had normal morphologies, contained numerous stinging cells (nematocytes) mounted in epithelial battery cells, and were fully functional as attested by their ability to kill brine shrimp nauplii. Many transgenic animals also developed hyperplastic stolons (Fig. 6B,D). In normal animals, these organs form as temporarily modified stolons during allogeneic encounters between histoincompatible animals (Ivker, 1972; McFadden et al., 1984; Lange et al., 1989; Lange, et al., 1992; Müller, 2002). Hyperplastic stolons are swollen by excessive nematocytes that are used to inflict damage on allogeneic tissues. However, transgenic animals were not exposed to a direct contact with other colonies. Finally, three animals developed as a ball of tentacles without any further structure (Fig. 6E).

Following metamorphosis, the transgene's expression became restricted to epithelial tissue, and expression in other transgenic cells had shut down. Concomitantly, the phenotypes of the animals gradually dissipated over the next weeks and all newly budded clonal polyps after the primary polyp (i.e. the first polyp that emerges during metamorphosis) appeared completely normal despite strong Nanos2–GFP expression in their epithelia (Fig. 6F,G). Nevertheless, up to several weeks later, the first polyp in these colonies was still recognizable due to its excessive tentacle number (Fig. 6F), comparing to its genetically identical clone mates that had normal tentacle numbers.

In cnidarians, anterior–posterior polarity is controlled by canonical Wnt signaling. Ectopic activation of the canonical Wnt pathway induces ectopic tentacles (Lee et al., 2006; Plickert et al., 2006; Duffy et al., 2010), reminiscent of the Nanos2 mediated excessive tentacles shown here. To investigate a possible involvement of Nanos2 in Wnt mediated axial patterning we studied the expression of Wnt3 in transgenic animals. This gene is expressed exclusively in the oral pole of Hydractinia throughout its life cycle and is a known target of its own protein through canonical Wnt signaling. Consequently, global activation of canonical Wnt signaling deregulates Wnt3, rendering it ubiquitous (Plickert et al., 2006; Duffy et al., 2010). We thought that if Nanos2 were upstream of Wnt, then Nanos2 transgenic animals should also have a depolarized, more ubiquitous Wnt3 expression. However, the expression of Wnt3 in transgenic animals remained restricted to the very oral end as in wild-type polyps and no Wnt3 mRNA was detected anywhere else in the animal (Fig. 7A,B). This shows that Nanos2 does not affect canonical Wnt signaling, and that the ectopic tentacle formation in Nanos2 transgenic animals was likely Wnt independent.

Conversely, ectopic Wnt activation by treatment with azakenpaullone, a specific inhibitor of Gsk3 (Kunick et al., 2004; Müller et al., 2004b</emph>; Teo et al., 2006), drove ubiquitous Nanos2 expression (Fig. 7C,D), suggesting that Nanos2 is a putative downstream target of canonical Wnt signaling.

Nanos is a classical germ cell marker in virtually all animals (Ewen-Campen et al.). Its general mode of action is the inhibition of translation of its target mRNAs, but it can also mediate mRNA degradation by recruiting deadenylating enzymes (Kadyrova et al., 2007). The role of Nanos in germ cells is well documented, based on work done on flies and mice. Because both Hydractinia Nanos homologs were expressed in germ cells it is reasonable to assume that they function similarly in this context.

Less understood is the role of Nanos in the nervous system. In Drosophila, Nanos has been associated with dendrite morphogenesis (Ye et al., 2004; Brechbiel and Gavis, 2008), with presynaptic terminal growth, and with postsynaptic glutamate receptor subunit composition (Kadyrova et al., 2007). In the mouse, only one out of three Nanos homologues (Nanos1) is expressed in the nervous system. The function of which, however, is unknown, as its complete absence causes no visible developmental abnormality. By contrast, Nanos' canonical partner, Pumilio, does function in the mammalian nervous system (Vessey et al., 2010; Marrero et al., 2011; Siemen et al., 2011). Because Nanos and Pumilio usually act in concert, the role of Nanos in the mammalian nervous system might have been overlooked, owing perhaps to the complexity of the mammalian nervous system. The by far simpler nervous system of Hydractinia may be more informative in studying Nanos functions.

Indeed, our results point to a novel and unexpected role for Nanos in neural cell fate determination. Nanos2 overexpression resulted in increased numbers of nematoblasts, by increasing their proliferation, but decreased numbers of mature nematocytes in pre and early post metamorphosis stages. This suggests that Nanos2 also acts in inhibiting nematocyte maturation. An additional role of Nanos2 in late nematocyte differentiation may be related to migration. Hence, while promoting nematoblast migration, Nanos2 may prevent their last maturation stages until the cells reach their final destination in the tentacles where they lose Nanos2 expression, stop migrating and fully mature. This is consistent with its expression pattern in these cells (Fig. 2C,D; supplementary material Fig. S2A–D), and explains why Hydractinia larvae overexpressing Nanos2 had only few mature nematocytes despite having an excess of Ncol1+ progenitors, until after metamorphosis when the transgene's expression became solely epithelial. Transgenic animals had, in contrast, lower numbers of Rfamide+ neurons in both pre- and post-metamorphic stages. Although we have no definite marker for early neuroblasts, it is unlikely that neuroblasts numbers were increased and merely their maturation inhibited by ectopic Nanos2, as in the case of nematoblasts. If this were the case, increased numbers of Rfamide+ neurons should have been produced post metamorphosis, similar to nematocytes, which was not the case. Therefore, Nanos2 expression was key in determining the balance between nematocytes and Rfamide+ neurons. Nanos2 may be acting on i-cells, promoting them to become committed to nematogenesis and also promotes their proliferation following commitment. This would result in a decrease in other i-cell derivatives, such as neurons. We have not counted other cell types but cannot exclude that numbers of other cell types would be affected too. Alternatively, Nanos2 could act on a common nematocyte neural progenitor, affecting only nematocytes and neurons. The existence of such a cell has been proposed previously (Miljkovic-Licina et al., 2007). The results of Nanos2 expression manipulation are summarized in Fig. 8A.

Although Nanos seems to be involved in posterior patterning in many insects (Lall et al., 2003), evidence for a similar function in other animals is scarce, and based exclusively on polarized Nanos expression patterns in various animals such as leeches (Pilon and Weisblat, 1997), Hydractinia (this study) and other cnidarians (Mochizuki et al., 2000; Torras et al., 2004; Extavour et al., 2005; Torras and González-Crespo, 2005; Leclère et al., 2012). The presence of the ectopic tentacles of post metamorphic Nanos2 transgenic animals could support a similar role in cnidarians, as tentacles are oral structures and the cnidarian head corresponds to the larval posterior pole (Meinhardt, 2002). However, six lines of evidence dispute this hypothesis: First, although Nanos transcripts were concentrated at the prospective oral pole of the early embryo (Fig. 2A), in gastrulae and larval stages Nanos2 expression was not polarized (Fig. 2C). Second, the only evidence for polarity defects (i.e. ectopic tentacles) became evident post-metamorphosis, concomitant with the increasing numbers of mature nematocytes (Figs 5,6). In contrast, ectopic Wnt activation in Hydractinia embryos oralizes the animals already at the pre-larva stage (Plickert et al., 2006), different from ectopic Nanos2 expression. Third, with the exception of three cases (out of 25), Nanos2 ectopic expression did not cause polarity defects other than extra tentacles (Fig. 6). Fourth, Nanos2 downregulation had no effect on axis polarity except of lowering nematocyte and tentacle numbers (Fig. 4). Fifth, long-term surviving transgenic animals had normal morphologies despite ectopic Nanos2 in their epithelial cells (Fig. 6F,G). Finally, expression of Wnt3, a major regulator of cnidarian axis polarity, was unaltered in Nanos2 transgenic animals (Fig. 7A,B), showing that Nanos2 mediated ectopic tentacles are Wnt-independent. Hence, we favor the hypothesis that excessive numbers of nematocytes, rather than polarity defects, drove extra tentacle formation, but cannot exclude the alternative hypothesis. Of note, i-cell (and therefore nematocyte) free Hydra can regenerate tentacles (Marcum and Campbell, 1978; Sugiyama and Fujisawa, 1978), but recent studies showed that neurogenesis is requited for head (hence, tentacle) regeneration of normal Hydra (Miljkovic-Licina et al., 2007). In Hydractinia, elimination of i-cells results in tentacles, and eventually polyps, being resorbed (Müller et al., 2004a</emph>). Hence, nematocytes may have a role in normal cnidarian tentacle formation.

A model summarizing Nanos2 function in nematogenesis is presented in Fig. 8B. Analysis of Nanos2 expression pattern and functional studies suggest that Nanos2 is expressed in two distinct phases of nematogenesis. In the first, it promotes i-cells commitment to nematocytes and represses neuronal fate. During this phase, Nanos2 is likely to act in concert with its partner, Pumilio, as their expression patterns overlap in early nematoblasts. Later, as nematoblasts exit the cell cycle, Nanos2 transcription is silenced and replaced by Ncol1. At the end of the Ncol1 phase, Nanos2 expression resumes until stinging cells mature. In later nematocyte development Nanos2 promotes migration of maturing nematocytes, but prevents their final maturation until they reach their final destination. Nanos2 and Pumilio expression patterns, overlapping only in proliferating nematoblasts, may suggest that Nanos acts independent of Pumilio in the later phase of nematocytes development.

Taken together, Nanos was shown previously to have a conserved role in germ stem cell maintenance, self-renewal, migration, and prevention of differentiation. Our data add a similar role for Nanos in a neural lineage, and suggest it has no direct role in cnidarian axis patterning, a role that might be an insect or arthropod innovation.

Animals were collected in Galway Bay by SCUBA, and cultured in seawater. They were fed brine shrimp nauplii five times a week supplemented by ground fish once a week. Embryos were collected about an hour following light-induced spawning. Metamorphosis was induced by a pulse treatment in 100 mM CsCl (Müller and Buchal, 1973).

In situ hybridization was performed as described (Gajewski et al., 1996). DIG- and Fluorescein labeled RNA probes (Roche) were synthesized using SP6 and T7 RNA polymerases according to the manufacturer's protocol (Fermentas). In situ hybridization was performed at 50°C. For fluorescent in situ hybridization we followed the protocol by Toledano et al. (Toledano et al., 2012).

We prepared a vector consisting of the Hydractinia Actin1 promoter, driving expression of Nanos2 fused in frame with enhanced green fluorescent protein (supplementary material Fig. S3). One-two cell stage embryos were microinjected with 200 pl volume of the plasmid at a concentration of 4–5 µg/µl as previously described (Künzel et al., 2010; Duffy et al., 2012; Millane et al., 2011).

Fluorescein conjugated translation block morpholino oligonucleotides (see supplementary material Table S1 for sequence) were designed and synthesized by Gene Tools. One-two cell stage embryos were microinjected with 100 pl at 0.25–0.5 mM. As a control scrambled morpholinos were injected at the same concentration.

Western Blotting was performed to verify the specificity of the translation block morpholino oligonucleotide. For this purpose, Nanos2-His-tag protein was synthesized using the in vitro transcription-translation system TNT®T7 Coupled Reticulocyte Lysate System (Promega, cat.no. L4611). We added 1 µl of 1 mM Nanos2 morpholino to 50 µl transcription-translation reaction. As control, a transcription-translation reaction was performed with the non-coding scrambled morpholino. Synthesized, His-tagged proteins were detected using rabbit anti-His-tag antibodies (1:1000 dilution, Santa Cruz, cat no. SC-803), followed by HRP-conjugated Goat anti-rabbit IgG (1:1000 dilution, Invitrogen, cat. no. 656120) and a chemiluminescent substrate.

Animals were fixed in 4% paraformaldehyde in PBS for 60 minutes and then dehydrated in ethanol through four steps (25, 50, 75 and 100%). Samples were rehydrated and blocked for 30 minutes in 2% BSA in PBS (BSA/PBS), then blocked for 30 minutes in 5% goat serum in BSA/PBS (GS/BSA/PBS). The antibody was diluted 1:500 in GS/BSA/PBS and incubated for 1 hour at room temperature, followed by three washes with BSA/PBS and re-blocking as above. Pre-adsorbed secondary antibodies (Alexa Fluor 594 and 488 goat anti-rabbit IgG, cat. no. A-11012 and A-11008, Invitrogen) were diluted 1:500 in GS/BSA/PBS and incubated for 1 hour at room temperature. Animals were mounted in 90% glycerol/PBS.

Animals were incubated in seawater containing 200 mM BrdU for 1 hour and then washed with seawater. Forty minutes later they were fixed with paraformaldehyde. The specimens were washed in PBS, then with 0.4 M glycine pH7.2. Next they were treated with 2 M HCl, then washed with 0.25% Triton PBS. Unspecific binding of the antibodies was blocked with 1% BSA in 0.25% Triton PBS. Samples were then incubated for 1 hour at room temperature in 1:500 dilution of anti-BrdU (Roche), then washed with 0.25% Triton PBS and incubated in the dark at room temperature for 1 hour in pre-absorbed 1:500 dilution of secondary antibody, Alexa Fluor 488 anti-mouse antibody (Invitrogen, cat. no. A11059). Finally, they were washed with 0.25% Triton PBS and mounted in glycerol.

Stinging cell progenitors were visualized using the rabbit polyclonal anti Hydra NCol1 antibody (Adamczyk et al., 2010), a kind gift from Dr Suat Özbek (University of Heidelberg, Germany). The numbers of nematoblasts were too high to be counted in full. We therefore counted all NCol1+ cells in a single, anterior-posterior confocal plane, choosing the one with the highest number of cells in each counted animal. Statistical analyses were performed by the two-sample Student's t-test method using the Minitab software.

Mature nematocytes were counted by staining the poly-γ-glutamate contents of their nematocyst capsules with DAPI as described (Szczepanek et al., 2002) and viewing them under the FITC channel of a confocal microscope. Under these conditions only nematocyst capsules, and not nuclei, were visible. Nematocysts in embryos and larvae were counted in full confocal z-stacks of the entire animals. Counting nematocysts in animals older than a week post metamorphosis is difficult due to their high numbers and distribution among the different animal compartments (tentacles, body column and stolons). As alternative, we counted nematocysts in the most distal 80 µm of 24 wild-type and 9 Nanos2 transgenic tentacles but found no significant difference between them (9% increase in transgenic animals P = 0.119). We therefore took the number of tentacles in a polyp as a proxy for the total number of nematocysts.

All BrdU+ cells per animal were counted for single confocal layer, choosing the one with the highest number of cells in each counted animal.

Neurons expressing Rfamide were visualized by in situ hybridization. We counted all the cells in the animals using brightfield microscopy.

Nanos protein sequences from various animals were taken from GenBank and boundaries of their conserved domains were established based on a previously described phylogenetic analysis (Extavour et al., 2005). Alignment of conserved domains was prepared using ClustalW2 (http://www.ebi.ac.uk/Tools/msa/clustalw2) and adjusted manually. Bootstrapped (100 replicates) phylogenetic tree of Nanos was inferred by the MEGA 5.10 software, using the maximum likelihood method with the Jones-Taylor-Thornton (JTT) model of amino acid substitution and the Nearest-Neighbor-Interchange (NNI) method, and prepared using the NJ/BioNJ option.

We thank members of the Frank laboratory for their continuous support and discussions. Albert Lawless, Eoin MacLoughlin, John Galvin and Terry Callanan are kindly acknowledged for technical assistance. Anti-NCol1 antibody was a kind gift from Suat Özbek (University of Heidelberg).