Smed-Evi/Wntless is required for β -catenin-dependent and -independent processes during planarian regeneration

Whole-mount in hybridization hybridization as Digoxigenin-labelled riboprobes (Roche) Planarians can regenerate a whole animal from only a small piece of their body, and have become an important model for stem cell biology. To identify regenerative processes dependent on Wnt growth factors in the planarian Schmidtea mediterranea ( Smed ), we analyzed RNAi phenotypes of Evi, a transmembrane protein specifically required for the secretion of Wnt ligands. We show that, during regeneration, Smed-evi loss-of-function prevents posterior identity, leading to two-headed planarians that resemble Smed- β -catenin1 RNAi animals. In addition, we observe regeneration defects of the nervous system that are not found after Smed- β -catenin1 RNAi. By systematic knockdown of all putative Smed Wnts in regenerating planarians, we identify Smed-WntP-1 and Smed-Wnt11-2 as the putative posterior organizers, and demonstrate that Smed-Wnt5 is a regulator of neuronal organization and growth. Thus, our study provides evidence that planarian Wnts are major regulators of regeneration, and that they signal through β -catenin-dependent and -independent pathways.


INTRODUCTION
Planarians have a nearly unlimited capability of renewing lost body structures (reviewed by Saló, 2006). Candidates for providing positional information during planarian regeneration are gradient-forming growth factors, such as Wnts. Wnts are secreted glycoproteins that behave as major organizers during embryonic development in various metazoan organisms (reviewed by Logan and Nusse, 2004). Canonical Wnt signaling is transduced through β-catenin, and mediates developmental processes, including axis specification (reviewed by Grigoryan et al., 2008). Some Wnts, such as mammalian Wnt5 (Slusarski et al., 1997), can signal through β-catenin-independent mechanisms, and control processes, such as cell polarity and directional movement (Witze et al., 2008).
Recently, β-catenin has been demonstrated to be essential in the establishment of anteroposterior (AP) identity during regeneration of the planarian species Schmidtea mediterranea (Gurley et al., 2008;Iglesias et al., 2008;Petersen and Reddien, 2008). Smed-β-catenin1 silencing leads to a loss of posterior identity and complete anteriorization ('radial-like hypercephalyzed' planarians) (Iglesias et al., 2008). However, although these studies suggest that Wnt proteins might be essential organizers of regeneration in planarians, direct evidence is missing.
Wnt secretion requires the transmembrane protein Evenness interrupted [Evi; also known as Wntless (Wls) and Sprinter] (Bänziger et al., 2006;Bartscherer et al., 2006;Goodman et al., 2006). In Drosophila and Caenorhabditis elegans, Wnts are not secreted in the absence of Evi, leading to a loss of Wnt signaling in the surrounding tissue. Owing to the lack of non-canonical phenotypes in evi mutant flies (Bartscherer et al., 2006), it is currently not known whether Evi also controls β-cateninindependent Wnt signaling.
Here, we report the effect of RNAi-mediated silencing of Smedevi and all putative S. mediterranea Wnt genes during planarian regeneration. We show that, like Smed-β-catenin1, Smed-evi, as well as Smed-wnt11-2 and Smed-wntP-1, are required for posterior identity in regenerating animals. In addition, we demonstrate that Smed-evi RNAi causes the same β-catenin-independent defects as does Smed-wnt5 loss-of-function, suggesting that Smed-Wnt5 is a non-canonical Wnt that requires Smed-Evi for its secretion, and which acts to control neuronal growth during regeneration of the planarian nervous system.

Animals
The planarians used belong to an asexual race of S. mediterranea, and were maintained as described elsewhere (Molina et al., 2007).

RNAi silencing
dsRNA microinjection was performed as described elsewhere (Sánchez-Alvarado and Newmark, 1999). dsRNAs were synthesized as described (Boutros et al., 2004). Primer details are available upon request. Control animals were injected with water. Injected planarians were amputated preand post-pharyngeally, and the head-, trunk-, and tail-pieces were allowed to regenerate for the times indicated.

Whole-mount in situ hybridization
Whole-mount in situ hybridization was carried out as described previously (Nogi and Levin, 2005;Umesono et al., 1999). Digoxigenin-labelled riboprobes were synthesized using an in vitro transcription kit (Roche) (Iglesias et al., 2008). Primer details are available upon request.

RESULTS AND DISCUSSION An Evi homolog is expressed in S. mediterranea
We found a single evi gene in the genome of S. mediterranea ( Fig.  1A; see also Fig. S1 in the supplementary material). Smed-evi mRNA mainly localized to the nervous system, the brain/cephalic ganglia (CG) and the ventral nerve cords (VNCs), as well as to the pharynx, and to the mouth. In addition, it was expressed in discrete cells of the posterior parenchyma (Fig. 1B).
To analyze the expression of Smed-evi in regenerating animals, we dissected heads and tails from the trunk parts of the animals and followed their regeneration. Smed-evi was expressed in the regenerating nervous system and upregulated in the pharynx primordia, as well as in the tips of posterior wounds (blastemas; Fig.  1C), suggesting that Smed-evi might be especially important for posterior regeneration. As Evi is required for the secretion of Wnts, the Smed-evi expression pattern might indicate sites of Wnt production and release in planarians.

Smed-evi is required for posterior identity
To reveal Wnt-dependent processes in planarian regeneration, we depleted Smed-evi by RNAi, and dissected heads and tails from the trunk parts of the animals. Within 25 days of regeneration, all head fragments developed an ectopic posterior head with eyes and a brain ( Fig. 2A,B). In addition, the expression of the central-posterior Hox gene Smed-hoxD was lost (Fig. 2C). These results indicate that, in the absence of Smed-evi, posterior fate is suppressed to the gain of anterior structures.
We obtained similar results for evi RNAi trunk fragments, as most of them developed an ectopic posterior head (Fig. 2D,E). In all animals, the anterior head showed several ectopic eyes. Both, anterior and posterior eyes differentiated laterally along the anteroposterior axis and were connected by their visual axons. In most cases, they did not cross the commissure and the optic quiasm was not formed (Fig. 2D). In ~50% of the Smed-evi RNAi trunks, the posterior head dissociated spontaneously during regeneration, a process we refer to as 'scission'. Tail pieces regenerated a head with ectopic eyes (for Smed-evi RNAi phenotypes, see Fig. S2 in the supplementary material).
The role of planarian Evi in AP polarity, but not in dorsoventral (DV) polarity (see Fig. S3 in the supplementary material), is consistent with the recently described Smed-β-catenin1 RNAi phenotype. Smedβ-catenin1 RNAi causes a loss of AP polarity and complete anteriorization of regenerating planarians (Gurley et al., 2008;Iglesias et al., 2008;Petersen and Reddien, 2008) (Fig. 2F). However, we never observed full anteriorization after Smed-evi knockdown, which might be due to inefficient knockdown of Smed-evi, or to a stronger blockage of signaling after the loss of Smed-β-catenin1.
Together our results suggest that Smed-Evi is required for the release of Wnts that act through a β-catenin-dependent pathway to organize AP axis polarity during planarian regeneration.

Smed-evi is required for neuronal growth regulation during regeneration
To analyze the brains of Smed-evi RNAi animals in more detail, we tested head fragments for Synapsin expression (Fig. 2G-J). In wildtype animals the CG were located dorsally above the VNCs and extended as a pair into the regenerating tail ( Fig. 2G; see Movie 1 in the supplementary material). However, in the posterior head of Smedevi RNAi animals, both CG grew laterally along the VNCs. Confocal sections revealed that the posterior part of the CG projected into the ventral region ( Fig. 2GЈ; see Movie 2 in the supplementary material). We refer to this phenotype as 'deflected-brain phenotype'. Even though Smed-β-catenin1 knockdown also resulted in the formation of a posterior brain, the ectopic CG never appeared deflected from the VNCs (Fig. 2H).
In bipolar regenerating trunk fragments, Synapsin expression showed the same deflected-brain phenotype. In addition, longitudinal neuronal tissue projected from posterior CG into the ventral half of the animal ( Fig. 2IЈ; see Movie 4 in the supplementary material; a control trunk fragment is shown in Movie 3). We can only speculate whether these projections were new VNCs growing from posterior to anterior. As they grew laterally to the old VNCs, it is possible that old ones might repel new neuronal tissue, causing the lateral protrusions we observed in regenerating trunks (Fig. 2D,IЈ). Consistent with the absence of any optical chiasm, Synapsin staining revealed that the commissure was lost in anterior and posterior brains. We also observed a disconnection between the old and the new VNCs. Smed-β-catenin1 RNAi animals showed neither a deflection nor any disconnections of the nervous tissue (Fig. 2J). Thus, we propose

RESEARCH REPORT
Development 136 (6) that Smed-Evi is required for the release of a Wnt protein that does not signal through β-catenin, and which restricts and defines the position of neuronal tissue during planarian regeneration.

Posterior identity requires Smed-wntP-1 and Smed-wnt11-2
Next, we set out to identify the Wnts responsible for the specification of posterior structures during regeneration. In situ hybridization experiments showed that Smed-wnt11-2, as well as Smed-wntP-1 (Petersen and Reddien, 2008), was expressed in discrete cells along the most posterior midline of the S. mediterranea body (Fig. 3A,B). In regenerating animals, Smed-wnt11-2 and Smed-wntP-1 mRNA levels were upregulated in discrete cells of the posterior blastemas, indicating an important role in the control of posterior identity (Fig. 3C,D). We tested whether posterior regeneration was affected by Smed-wnt11-2 and Smed-wntP-1 RNAi. Silencing of either mRNA led to a 'tailless' morphology, which was characterized by a shorter and more rounded posterior end, in which VNCs terminated shortly behind the pharynx (Fig. 3F,G). Furthermore, 10% of regenerating trunks, and more than half of all regenerating head fragments of Smed-wntP-1 RNAi animals resulted in two-headed planarians (Fig. 3G), a phenotype that was 907 RESEARCH REPORT Smed-Evi in planarian regeneration DEVELOPMENT not enhanced by the simultaneous knockdown of wnt11-2 (not shown). As Smed-β-catenin1 RNAi also causes anteriorization of planarians (Fig. 2), Smed-Wnt11-2 and Smed-WntP-1 are likely to signal through β-catenin to permit posterior fate during regeneration.

Smed-Wnt5 controls neuronal growth and patterning during regeneration
To identify Wnt proteins responsible for the deflected-brain phenotype of Smed-evi RNAi animals, we silenced all remaining S. mediterranea Wnt transcripts. Smed-wnt2-1, Smed-wntP-3 and Smed-wntP-4 had no obvious phenotype, possibly owing to inefficient knockdown. Consistent with the phenotype after DjwntA silencing in the planarian species Dugesia japonica (Kobayashi et al., 2007), we detected an abnormal elongation of the new brain and visual axons in Smed-wntA RNAi animals (see Fig. S5 in the supplementary material).
In Smed-wnt5 RNAi animals, we discovered a deflection and ventral expansion of the regenerating CG ( Fig. 4D; see also Movie 5 in the supplementary material). In the regenerating tail, new Synapsin-positive tissue appeared, which was thicker than, and appeared disconnected from, the old nervous tissue (Fig. 4D). Expression analysis of Smed-glutamate receptor (gluR) mRNA showed that the new posterior neuronal tissue was not of brain identity (Fig. 4E), suggesting that it was new VNCs that grew deflected from the old ones. These phenotypes were similar to those observed after Smed-evi silencing ( Fig. 2I; see also Movie 5 in the supplementary material).
Consistent with this, we found that Smed-wnt5 mRNA localized mainly to distinct cells along the CG and VNCs, and was upregulated in the regenerating nervous system and the blastemas (Fig. 4B). Together, our data suggest that Smed-Evi restricts and coordinates the growth of regenerating neuronal tissue by regulating the secretion of Smed-Wnt5.

Concluding remarks
Our study reveals several regenerative processes that are regulated by Wnts in planarians (for a summary, see Table S1 in the supplementary material). Using Smed-evi RNAi to block Wnt secretion, we identify AP axis polarity and neuronal growth as being Wnt-regulated processes. Specifically, we identify Smed-wntP-1 and Smed-wnt11-2 as being the secreted molecules responsible for AP axis polarity. Consistent with the role of Wnts as morphogens (reviewed by Bartscherer and Boutros, 2008), Smed-WntP-1 and Smed-Wnt11-2 might activate posterior fate, from their site of production in the tail (Fig. 3), along the AP body axis. As posterior identity also depends on Smed-β-catenin1, we propose that Smed-WntP-1 and Smed-Wnt11-2 signal through β-catenin to permit posterior fate during regeneration.
Furthermore, we show that knockdown of Smed-wnt5 results in a deflected-brain phenotype. Our data are consistent with the reported role of Wnt5 in axonal growth in several organisms (Yoshikawa et al., 2003;Fradkin et al., 2004;Zhang et al., 2007). Smed-Wnt5 belongs to the Wnt5 family (see Fig. S4 in the supplementary material), which has been linked to non-canonical Wnt signal transduction (Wong et al., 1994;Olson and Papkoff, 1994;Shimizu et al., 1997) and seems to regulate cell motility rather than specification (Moon et al., 1993;Wallingford et al., 2001;Witze et al., 2008). As we did not observe any deflectedbrain phenotype after Smed-β-catenin1 RNAi, we suggest that Smed-Wnt5 signals through a mechanism that is β-catenin independent.
In planarians, Evi function is therefore required for the secretion of Wnts that signal through β-catenin-dependent and -independent pathways (Fig. 4F). This suggests that Evi is an ancient factor that had been an important facilitator of Wnt secretion before Wnts functionally diverged. Even though the same Wnts might be able to signal through both canonical and non-canonical pathways,