INTRODUCTION

Planarians show a striking morphological plasticity that becomes evident during regeneration and normal tissue homeostasis. They are able to regenerate a whole organism from a piece of almost any part of their body, and, furthermore, they have the ability to grow and degrow according to culture conditions (Morgan, 1898; Saló, 2006). These properties rely on the neoblasts, multipotent stem cells present in adult organisms that are able to differentiate into any planarian cell type (Saló, 2006).

The canonical Wnt signalling pathway has a common role in establishing the anteroposterior (AP) axis during development in several species, including mouse (Marikawa, 2006), chick (Nordstrom et al., 2002), zebrafish (Schier and Talbot, 2005), Xenopus (Kiecker and Niehrs, 2001), amphioxus (Holland, 2002), C. elegans (Huang et al., 2007) and Platynereis (Schneider and Bowerman, 2007). In early vertebrate embryos it is also required for dorsoventral (DV) polarity (De Robertis and Kuroda, 2004). In cnidarians, it specifies the oral-aboral embryonic axis, and it also has a reported role in axial patterning during regeneration (Lee et al., 2006). However, in classical models of regeneration, such as fish or amphibians, canonical Wnt signalling has only been demonstrated to be involved in the regenerative capacity (Kawakami et al., 2006; Yokoyama et al., 2007).β -Catenin is the key intracellular effector of the canonical Wnt signalling pathway, although it is a bi-functional protein that also regulates cell adhesion as a component of adherens junctions (Schneider et al., 2003).

Here, we report the characterization of two β-catenin homologs in the planarian species Schmidtea mediterranea (Smed-βcatenin1 and Smed-βcatenin2). Silencing of Smed-βcatenin1 in regenerating and intact planarians induces a gradual anteriorization of the animals that finally leads to a radial-like hypercephalized phenotype, demonstrating the requirement for Smed-βcatenin1 in AP axis re-establishment and maintenance. Analysis of the protein domains of the S. mediterranea β-catenins, and functional assays using Xenopus embryos, demonstrate the involvement of Smed-βcatenin1 but not Smed-βcatenin2 in Wnt signalling, suggesting a functional specialization of S. mediterranea β-catenins.

MATERIALS AND METHODS

Organisms

The planarians used belong to an asexual race of S. mediterranea collected from Montjuïc, Barcelona, Spain, and maintained as described elsewhere (Molina et al., 2007).

Identification and cloning of S. mediterranea genes

Fragments of Smed-βcatenin1 and Smed-βcatenin2 were identified from the S. mediterranea genomic database through a BLAST search. The corresponding full-length transcripts were amplified by rapid amplification of cDNA ends (RACE) using the Invitrogen GeneRacer Kit (Invitrogen). Smed-HoxD, Smed-AbdBa and Smed-TCEN49 were identified from the S. mediterranea genomic database using homologs from other planarian species (Orii et al., 1999; Garcia-Fernandez et al., 1993; Nogi and Watanabe, 2001; Bueno et al., 1996). Specific primers were designed to isolate the corresponding full-length cDNA sequences.

Accession numbers

Smed-TCEN49, EU082822; Smed-AbdBa, EU082823; Smed-HoxD, EU082824; Smed-βcatenin1, EU082826; Smed-βcatenin2, EU082825.

RNAi silencing

RNAi analyses were performed by feeding planarians with bacteria expressing double-stranded RNA (dsRNA) or by dsRNA microinjection, as described by Newmark et al. (Newmark et al., 2003) and Sánchez Alvarado and Newmark (Sánchez Alvarado and Newmark, 1999), respectively. The corresponding cDNA for Smed-βcatenin1 and Smed-βcatenin2 subcloned into the pPR242 vector was used for feeding. Control animals were fed bacteria containing the vector alone. When dsRNA microinjection was used, Smed-βcatenin1 and Smed-βcatenin2 dsRNA was synthesized by in vitro transcription (Roche). Control animals were injected with water. For regeneration experiments, treated planarians were amputated pre- and postpharyngeally, and the trunk pieces allowed to regenerate. Planarians were processed for whole-mount in situ hybridization or whole-mount immunostaining between 3 and 30 days after amputation or last injection (intact animals).

Whole-mount in situ hybridization

Whole-mount in situ hybridization was carried out essentially as described previously (Nogi and Levin, 2005; Umesono et al., 1999). Digoxigenin-labelled riboprobes for Smed-βcatenin1, Smed-βcatenin2, Smed-HoxD, Smed-AbdBa, Smed-TCEN49, Smed-Opsin (K. Eckelt), H.10.2f (Sánchez Alvarado et al., 2002), cintillo (Oviedo et al., 2003), Smed-GluR (F. Cebrià and P. Newmark), septin (Zayas et al., 2005) and eye53 (Zayas et al., 2005; Molina et al., 2007) were synthesized using an in vitro transcription kit (Roche).

Whole-mount immunostaining

Immunostaining was carried out essentially as described previously (Cebrià and Newmark, 2005; Sánchez Alvarado and Newmark, 1999). The following monoclonal antibodies were used: anti-arrestin (VC-1) (Sakai et al., 2000) at a 1:15,000 dilution; anti-synapsin (anti-SYNORF1, Developmental Studies Hybridoma Bank) at 1:25; and P-Tyr-100 (Cell Signalling Technology) at 1:500. Highly cross-absorbed Alexa Fluor 488-conjugated goat anti-mouse IgG secondary antibody (Molecular Probes) was used at a 1:400 dilution.

Xenopus microinjection of mRNA and in situ hybridization

The entire coding regions from S. mediterranea β-catenins genes were amplified by PCR and inserted into pCS2+ (Turner and Weintraub, 1994). To generate GFP-tagged constructs, a DNA fragment from the 5′ region of each cDNA, which includes unique sites within the open reading frame, was PCR-amplified. The 5′ primers contained an EcoRI site to clone the fragments in frame within the pCS2-GFP plasmid. The PCR fragments were cloned in pGEM-T Easy vector and sequenced prior to fusion with their corresponding cDNA. For mRNA preparation, the DNAs were linearized and transcribed with SP6 RNA polymerases as described (Harland and Weintraub, 1985), with GTP cap analog (New England Biolabs). Xenopus embryos were injected at the two-cell stage, into one blastomere at the prospective ventral marginal region, with 500-1000 pg of each mRNA, and fixed at the tailbud stage. Antisense RNA probes were prepared from Otx2, Krox20 and Cad3 cDNAs and labelled with digoxigenin (Roche). Xenopus specimens were hybridized as described (Harland, 1991). Antibody staining was performed as described (Gómez-Skarmeta et al., 2001), using the monoclonal antibody 12/101 (Developmental Studies Hybridoma Bank, developed by J. P. Brockes) and rabbit anti-GFP (Molecular Probes).

RESULTS AND DISCUSSION

Smed-βcatenin1 inhibition induces a gradual anteriorization of regenerating planarians

A search for β-catenin homologs in the S. mediterranea genome database yielded two genes, which we called Smed-βcatenin1 and Smed-βcatenin2. In situ hybridization experiments revealed different expression patterns for the S. mediterraneaβ -catenins: Smed-βcatenin1 was expressed ubiquitously but more strongly in the central nervous system (CNS), whereas Smed-βcatenin2 mRNA was mostly located in the digestive system (see Fig. S1 in the supplementary material). During the process of regeneration, expression of both S. mediterranea β-catenin genes was detected in anterior and posterior blastemas (see Fig. S2 in the supplementary material). To assess their potential role in axis re-establishment during regeneration, we carried out RNAi experiments. In situ hybridization for each gene demonstrated that RNAi effectively silenced its expression (see Fig. S3 in the supplementary material). Following Smed-βcatenin1 silencing, 100% of the regenerating trunk pieces exhibited AP polarity defects, although with varying degrees of transformation: tailless planarians; two-headed planarians (Fig. 1A); two-headed planarians with ectopic eyes next to the normal anterior eyes (Fig. 1B); two-headed planarians with ectopic eyes next to the anterior and posterior ones (Fig. 1C); and, the most severe phenotype, animals displaying apparently radial symmetry with eyes all around the periphery of their body (Fig. 1D; see also Movie 1 in the supplementary material). This last phenotype was referred to as radial-like hypercephalized planarians. The different phenotypes obtained correspond to the degree of severity of the transformation, which was dependent on the time of regeneration and also on the dose of inhibition (Fig. 1I′). After Smed-βcatenin2 silencing, regenerating planarians did not show morphological defects or altered expression of tissue markers (data not shown).

Several markers were used to characterize Smed-βcatenin1-silenced animals. The planarian CNS is composed of two anterior cephalic ganglia (CG) located at one end of two ventral nerve cords (VNCs), which extend along the body and converge in the tail (Fig. 1Q) (Agata et al., 1998). From the two dorsally located eyes, visual projections extend to the CG (Okamoto et al., 2005). During the first regeneration stages (3-7 days) after Smed-βcatenin1 silencing, in situ hybridization for opsin and cintillo revealed the ectopic differentiation of photoreceptors and mechanoreceptors, respectively, in the posterior blastema (Fig. 1F,J). Analysis of the pan-neuronal marker H.10.2f revealed differentiation of ectopic CG in the posterior blastema (Fig. 1N). Analysis of radial-like hypercephalized planarians (after 20-30 days of regeneration) showed that ectopic photoreceptors and mechanoreceptors expanded from both ends of the regenerating animal to surround the planarian body (Fig. 1H,L). Synapsin immunostaining, which labels neuronal synapses, and in situ hybridization for H.10.2f and Smed-GluR (which is specifically expressed in the lateral branches of the brain), revealed that the CNS of radial-like hypercephalized planarians appears as a thick ring from which cephalic branches ectopically differentiate all around (Fig. 1P,T,V). Note that all photoreceptor cells appeared to be connected through their visual axonal projections to the circular brain, as seen with an anti-arrestin antibody (Fig. 1R). The digestive system of planarians is composed of a pharynx located in the middle of the trunk, from which one anterior and two posterior branches extend (Saló, 2006). The pharynx evaginates through the mouth, which is located ventrally in the middle part of the planarian body. Staining with an anti-P-Tyr antibody revealed that the two posterior branches converge to form a single branch in Smed-βcatenin1-silenced planarians (Fig. 1E′). Use of the same marker also showed that secretory cells located in the prepharyngeal mesenchyme of the wild-type planarian were ectopically differentiated in the posterior part of Smed-βcatenin1-silenced planarians (Fig. 1G′).

Altogether, these results show that both neural and non-neural structures acquire anterior identity throughout the body of Smed-βcatenin1-silenced animals, demonstrating a role for Smed-βcatenin1 in AP axis re-establishment during planarian regeneration.

Trunk and tail identities are lost in Smed-βcatenin1-silenced planarians

To assess whether the differentiation of anterior tissues in central and posterior regions was accompanied by the loss of trunk and tail identities, we analyzed central and posterior markers in Smed-βcatenin1-silenced planarians. Smed-HoxD and Smed-AbdbA, central-posterior and posterior Hox genes, respectively, were not expressed at 3 days of regeneration, the earliest stage at which they are detected in wild-type animals (Fig. 2A,E). Smed-HoxD continued not to be expressed throughout regeneration (Fig. 2C). Note that the absence of this marker indicated the disappearance of the mouth. Expression of Smed-TCEN49, a central marker associated with the pharynx, had almost disappeared at the latest stage (Fig. 2G), and disappearance of the pharynx was also evident when analyzed with other markers (Fig. 1T,V and Fig. 2I,M). These results demonstrate that the anteriorization of Smed-βcatenin1-silenced planarians is accompanied by the disappearance of trunk and tail identities. A possible role for Hox genes as targets of Smed-βcatenin1 needs further investigation.

The DV axis is unaffected in Smed-βcatenin1-silenced planarians

As the AP axis is extensively transformed, we analyzed whether the DV axis was affected after Smed-βcatenin1 inhibition. Analysis of septin and eye53 expression (dorsal and ventral markers, respectively) revealed that the DV axis of radial-like hypercephalized planarians was not affected (Fig. 2I-P).

Taken together, these results demonstrate that in regenerating Smed-βcatenin1-silenced animals the posterior blastema acquires anterior identity and the trunk region is anteriorized, ultimately generating radial-like hypercephalized animals in which the AP axis is lost but the DV axis remains unaffected. Although the canonical Wnt signalling pathway has a common role in establishing the AP axis during development in several species, from cnidarians to vertebrates (Holland, 2002; Kiecker and Niehrs, 2001; Nordstrom et al., 2002; Schneider and Bowerman, 2007), in vertebrates, at earlier developmental stages, nuclear accumulation ofβ -catenin leads to dorsalization of the embryo (De Robertis and Kuroda, 2004). Our results support a conserved role for the canonical Wnt signal in AP axis specification throughout evolution, and provide further confirmation that its role in DV axis establishment is a vertebrate innovation.

In vertebrate models of regeneration, the Wnt pathway has a reported role in cell proliferation and regenerative outgrowth, but not in axis re-establishment (Kawakami et al., 2006; Yokoyama et al., 2007). Cnidarians are the only species in which this pathway has a demonstrated axial patterning role during regeneration (Lee et al., 2006), although a direct functional study of cnidarian β-catenin has not been reported. Our results, together with two recent reports also in planarians (Gurley et al., 2008; Petersen and Reddien, 2008), are the first direct demonstration of the requirement of β-catenin in AP patterning during the process of regeneration. In contrast to other models, such as Xenopus and mouse, in which the inhibition of β-catenin during development produces complex defects and non-viable embryos due to the inhibition of organizer formation and the impairment of gastrulation (Heasman et al., 2000; Huelsken et al., 2000), the planarian model allows in vivo analysis of mutant phenotypes for essential developmental genes.

Smed-βcatenin1 is required for AP axis maintenance during homeostasis

To address whether Smed-βcatenin1 activity could also be required for maintenance of the AP axis, we silenced Smed-βcatenin1 in intact planarians. At 10-17 days after Smed-βcatenin1 silencing, ectopic eyes began to differentiate first in the tail region and afterwards anteriorly, adjacent to the original eyes (Fig. 3A,D). Ectopic mechanoreceptor cells appeared in the tail and lateral regions (Fig. 3G), and ectopic brain differentiated along the length of the VNCs (Fig. 3J,M). Around 30 days after Smed-βcatenin1 silencing, radial-like hypercephalized organisms that were indistinguishable from Smed-βcatenin1-silenced regenerating animals were observed. These animals showed ectopic eyes (Fig. 3B,E) that were all connected to the brain by their visual axons, ectopic mechanoreceptor cells all around the periphery of the body (Fig. 3H), and ectopic branching of the CNS all around the body (Fig. 3K,N). Analysis of central and posterior markers revealed the disappearance of trunk and tail identities (Fig. 3P,R). These data demonstrate that Smed-βcatenin1 activity is required not only during regeneration, but also during normal tissue homeostasis to maintain AP axis polarity in planarians. Interestingly, our data show that the anteriorization process in intact animals follows the same pattern as in regenerating trunks: it starts mainly in the tail region and later expands to the rest of the body (see Fig. 3T for quantification). Further studies are required to elucidate whether this is a direct consequence of the suppression of Smed-βcatenin1 activity or an indirect effect due to, for example, a higher rate of cell turnover in the tail.

The extreme radial-like phenotype obtained after Smed-βcatenin1 inhibition suggests that, in adult planarians, Smed-βcatenin1 activity is required to confer posterior identity, and its inhibition is required to acquire head fate. The differential Smed-βcatenin1 activity along the AP axis in planarians could be a response to a morphogenetic Wnt gradient, as has been described for the patterning of the CNS in Xenopus embryos (Kiecker and Niehrs, 2001). Accordingly, during Xenopus development, repression of Wnt signalling is a characteristic of the head organizer (Niehrs, 1999; Niehrs, 2004). However, the effect could also be explained by a mechanism in which β-catenin regulates sister-cell asymmetry following cell division, as occurs along the animal-vegetal axis during Platynereis and C. elegans embryogenesis (Schneider and Bowerman, 2007; Kaletta et al., 1997; Lin et al., 1998). The classical janus-headed phenotype (Morgan, 1898), in which two-headed planarians are produced after cutting extremely thin fragments, is a well-known model that supports the gradient hypothesis (Child, 1911), because the phenotype may be explained by the inability of such a small field to support a Smed-βcatenin1 activity gradient. The analysis of Smed-βcatenin1 activity, by analysis of its subcellular localization, for instance, will help to resolve this question.

Functional characterization of S. mediterraneaβ -catenins in Xenopus embryos

Sequence analysis of S. mediterranea β-catenins demonstrated that the GSK3-binding domain, required for β-catenin degradation during Wnt signalling, is present in Smed-βcatenin1 but not in Smed-βcatenin2. By contrast, residues involved in cell-cell adhesion were conserved in Smed-βcatenin2 but not in Smed-βcatenin1 (for details, see Fig. S4 in the supplementary material). To test whether S. mediterranea β-catenins could have undergone a functional specialization, we assayed their ability to induce a secondary axis in Xenopus embryos (McMahon and Moon, 1989). A secondary axis was induced after Smed-βcatenin1 injection but not after Smed-βcatenin2 injection (Fig. 4A,B,K). A GFP-tagged construct of each S. mediterranea β-catenin was injected to quantify the amount of translated protein in each embryo, demonstrating that even when twice the amount of Smed-βcatenin2 was present, it could never induce a secondary axis (Fig. 4C-H). Moreover, anti-GFP staining demonstrated that Smed-βcatenin1, but not Smed-βcatenin2, accumulates in the nucleus (Fig. 4I,J).

Altogether, these results demonstrate the functional conservation of Smed-βcatenin1 in Wnt signalling, and point to a functional specialization of S. mediterranea β-catenins, such that Smed-βcatenin1 would be involved in signalling and Smed-bcatenin2 would be involved in the membrane cell-cell contacts. All metazoans studied to date have a single β-catenin gene encoding a protein containing both kinds of functional domains, involved in signaling and in cell-cell adhesion (Schneider et al., 2003), with the exception of C. elegans, which is the only species in whichβ -catenin gene duplication and functional specialization have been reported (Korswagen et al., 2000). A phylogenetic analysis using β-catenin homologs from several species demonstrates that the duplication in S. mediterranea occurred independently from the one in C. elegans (see Fig. S5 in the supplementary material). Molecular studies of other species belonging to different phyla will clarify the significance of these duplications.

Supplementary material

Supplementary material for this article is available at http://dev.biologists.org/cgi/content/full/135/7/1215/DC1