A Transcription Factor Is Wound-Induced at the Planarian Midline and Required for Anterior Pole Regeneration

Summary

Planarian regeneration requires positional information to specify the identity of tissues to be replaced as well as pluripotent neoblasts capable of differentiating into new cell types. We found that wounding elicits rapid expression of a gene encoding a Forkhead-family transcription factor, FoxD. Wound-induced FoxD expression is specific to the ventral midline, is regulated by Hedgehog signaling, and is neoblast-independent. FoxD is subsequently expressed within a medial subpopulation of neoblasts at wounds involving head regeneration. Ultimately, FoxD is co-expressed with multiple anterior markers at the anterior pole. Inhibition of FoxD with RNA interference (RNAi) results in the failure to specify neoblasts expressing anterior markers (notum and prep) and in anterior pole formation defects. FoxD(RNAi) animals fail to regenerate a new midline and to properly pattern the anterior blastema, consistent with a role for the anterior pole in organizing pattern of the regenerating head. Our results suggest that wound signaling activates a forkhead transcription factor at the midline and, if the head is absent, FoxD promotes specification of neoblasts at the prior midline for anterior pole regeneration.

Introduction

Planarians can regenerate from nearly any injury, but how missing tissues are recognized and replaced is poorly understood. The adult population of proliferating cells (neoblasts) in Schmidtea mediterranea includes pluripotent stem cells [1] and is responsible for new tissue production in regeneration. New cell production at wounds produces an outgrowth called a blastema, which will replace some of the missing tissues [2]. At the molecular level, injuries trigger a rapid wound response program that includes conserved immediate early genes and patterning factors required for normal regeneration [3]. Wnt and Hedgehog (Hh) signaling pathways instruct the regeneration of the anterior-posterior (AP) axis, whereas the Bmp signaling pathway controls the regeneration of the dorsal-ventral (DV) axis [4]–[13]. Multiple genes required for positional identity control during embryonic development of other organisms, such as Wnt and Bmp signaling ligands, display constitutive regionalized expression in adult planarians and also guide pattern maintenance during tissue turnover [14].

Two distinct regions composed of a small cluster of cells located at the anterior and posterior animal extremities are referred to here as the anterior and posterior planarian poles. The poles are found at the midline of the animal and are subjects of current intense study. The anterior pole expresses notum, a Wnt inhibitor required for head regeneration [8], whereas the posterior pole expresses wnt1 [7], [14]. A number of genes encoding proteins predicted to be involved in signaling pathways that regulate planarian body plan patterning, and that display regional expression in planarian muscle cells have been described [15]. The genes that display these unique attributes will be referred to as position control genes (PCGs) for simplicity of discussion, but it is important to note that patterning phenotypes have not yet been described for many of these genes. PCGs expressed broadly in the planarian head include the candidate Wnt inhibitor, sFRP-1; candidate FGF inhibitors nou darake (ndk) and ndl-4; and a homeodomain transcription factor, prep [6], [7], [14]–[17]. PCGs expressed broadly in the posterior region of the animal include genes encoding additional Wnt ligands, wnt11-1, wnt11-2, and wntP-2/wnt11-5 and the Wnt receptor frizzled-4 [6], [7], [14], [15], [18]; Hox genes are also expressed in the posterior [14]. Several PCGs are expressed at the planarian poles, but have broader expression domains that extend beyond the cluster of notum⁺ or wnt1⁺ cells at the animal head and tail tips.

How the poles are formed and the role they have in organizing regeneration of an animal with a properly patterned body plan is poorly understood. Two genes encoding transcription factors of the TALE-class homeodomain family, prep and pbx, are required for regeneration of the expression domains of anterior PCGs and the anterior pole marker notum [17], [19], [20]. pbx [20] and a LIM-homeobox gene (djislet, [21]) are required for regeneration of expression domains of posterior PCGs and the posterior pole marker wnt1. follistatin, which encodes a secreted regulator of TGF-β proteins, is also expressed at the anterior pole and is required for normal head regeneration [22], but strong inhibition of this gene can also result in the absence of blastema formation indicating a broader role of this gene during regeneration [23].

Forkhead-box (Fox) genes are an evolutionary ancient family of winged-helix transcription factors involved in a wide variety of biological processes such as regulation of cell proliferation, growth, and differentiation [24]. During embryogenesis, Fox genes are involved in organogenesis and patterning of several tissues from all three germ layers [25]. Mutations in Fox genes have a profound impact in human disease causing a variety of phenotypes, from eye abnormalities to speech impediments [25]–[27]. Some members of the Fox gene family are expressed in restricted regions of embryos. In Drosophila, genes encoding Forkhead-family proteins, fkh, sloppy paired 1 and 2, and crocodile, are all expressed in the anterior region of the embryo, and are required for midline establishment as well as head patterning [28]–[31]. In amphioxus, FoxQ is expressed at the anterior pole during embryogenesis [32]. In Xenopus, the forkhead family gene, XDF1 is expressed in Spemann's organizer and at later stages in the anterior neural region [33]. In planarians, few Fox genes have been described. In particular, DjFoxD is expressed in few cells at the anterior pole region of the planarian D. japonica [34] and FoxD influences expression of follistatin in planarian heads [22], [34]. Given the potential importance of the planarian anterior pole in organizing head regeneration, we investigated the role of Schmidtea mediterranea FoxD in regeneration.

Results

Co-expression of FoxD with a number of anterior-expressed genes defines the anterior pole

A number of genes have been identified that are expressed in different domains of planarian heads. To provide a molecular definition of the anterior-most end of the planarian head, the anterior pole, we investigated the expression of a number of genes expressed near the planarian head tip. FoxD is expressed in a very small number of cells at the head tip (Fig. 1 and [22], [34]), but the pole and its role(s) are poorly defined; we focused our investigation of the anterior pole on the Schmidtea mediterranea ortholog of DjFoxD, Smed-FoxD, or FoxD in short (Fig. S1). FoxD expression in intact animals was dorsal-biased and most FoxD⁺ cells also expressed the gene notum. notum is required for the head-versus-tail regeneration decision known as AP regeneration polarity, and encodes a secreted inhibitor of Wnt signaling [8]. Like FoxD, notum expression in uninjured animals is largely restricted to a very small number of cells at the head tip (Fig. 1 and [8]). FoxD⁺ cells at the head tip also expressed multiple anterior markers (PCGs) that are expressed in AP transcriptional domains extending beyond the FoxD⁺ cells, including sFRP-1, ndk, ndl-4, and prep, but not sFRP-2 (Fig. 1). We also assessed whether FoxD is co-expressed with two planarian genes expressed at the DV boundary (a lateral domain surrounding the animal at the midpoint of dorsal and ventral surfaces) and/or at the midline (the median plane about which bilateral symmetry is organized): admp, encoding a BMP-family signaling ligand expressed in the ventral midline and at the DV boundary [10], [11] and slit, a conserved midline cue with a prominent role in regulation of axon guidance [35] expressed in the ventral and dorsal planarian midline [35]. admp and slit expression did not substantially coincide with pole cells expressing FoxD; midline slit expression did not reach the anterior-most region where FoxD-expressing cells are found (Fig. 1). We propose a definition for the planarian anterior pole in the mature head as the few cells restricted to the head tip and that co-express the highly restricted FoxD and notum genes together with a set of anterior PCGs, but displaying little expression of DV boundary and midline genes. The cellular basis for formation of these cells and the roles of these cells in regeneration are investigated below.

FoxD expression is wound-induced in the ventral midline

FoxD expression was highly induced by three hours following wounding in subepidermal cells, with expression peaking at approximately six hours after wounding (Fig. 2A). FoxD expression after amputation occurred at both anterior- and posterior-facing wounds, raising the possibility that FoxD is a generically wound-induced gene (Fig. 2A). FoxD expression after amputation was greatly diminished by 18 to 24 hours following injury (Fig. 2B), but increased again between 24 and 48 hours after amputation. At this later time FoxD expression was only observed at anterior-facing wounds that required the formation of a new anterior pole (Fig. 2B and Fig. S2A). Irradiation eliminates neoblasts [36], which comprise the entire population of dividing adult planarian cells; therefore, amputation experiments in irradiated animals can determine whether new gene expression in regeneration occurs in pre-existing cells at wound sites, or requires new cell production. Early wound-induced expression of FoxD was irradiation-insensitive, indicating that it is a transcriptional response in cells present at the time of wounding (Fig. 2B).

The large numbers of planarian wound-induced genes described so far are expressed broadly at the wound site [3], [8], [18], [37]. By contrast, wound-induced FoxD expression following amputation was unique: restricted to cells found in the ventral midline (Fig. 2A,B). To better understand the regulation of FoxD expression by wounding, we examined the impacts of several injury types on FoxD expression (Fig. 2C). First, we observed that expression of FoxD was not exclusive to wounds requiring pole regeneration; both parasagittal and sagittal amputations induced FoxD expression in the midline broadly along the wound site (first two cuts from the left, Fig. 2C). Incision into the planarian side with a scalpel, a wound not requiring blastema formation for repair, was sufficient to induce FoxD expression. Strikingly, FoxD was expressed at anterior and posterior areas of the incised wound site, but only within the midline on the ventral side (third cut from left, Fig. 2C). This wound-induced FoxD expression occurred even in the absence of the anterior pole, indicating that local cues rather than signals from the pole control wound-induced expression of FoxD. An even more minor injury, a dorsal puncture with a needle (third panel from right, Fig. 2C), induced FoxD expression in the vicinity of the dorsal puncture, but only ventrally in the midline. All of these wound types impinged upon the midline of the animal, raising the possibility that any midline injury is sufficient to trigger FoxD expression, regardless of whether blastema formation would be required for repair or not. Supporting this hypothesis, two wound types that do not damage the midline (lateral puncture and lateral edge removal; first and second panels from right, Fig. 2C) did not trigger expression of FoxD. These two wounds did, by contrast, elicit normal expression of other wound-induced genes (Fig. 2D).

The wound-induced expression of FoxD occurred within ventral midline cells expressing the midline markers admp and slit (Fig. 2E). Furthermore, FoxD was also co-expressed at wounds with other defined wound-induced genes, such as noggin-like1 (nlg1) [3], [38], wntless [3], [12], [38], notum [8], and wnt1 [37] (Fig. 2E and Fig. S2B). Wound-induced genes that peak in expression around six hours following injury and that are mostly expressed subepidermally at wound sites are known as W2 genes [3]. Although most W2 genes are predicted to be secreted proteins, such as signaling and matrix remodeling factors, our results indicate that the transcription factor FoxD itself belongs in this category. W2 gene expression is induced in muscle cells expressing collagen [15], and most wound-induced FoxD expressing cells (92.5±4.6%) also co-expressed the muscle gene collagen (Fig. 2E). We conclude that FoxD is a wound-induced gene in planarian muscle, but unique among known planarian wound-induced genes with expression occurring only in ventral midline cells and only following injury that impinges on the midline. FoxD also defines a fourth gene (together with wnt1, notum, and follistatin) that is wound-induced and subsequently expressed in either the anterior or posterior pole.

FoxD is expressed in a subset of smedwi-1⁺ cells during anterior pole regeneration

The second phase of FoxD expression during regeneration, initiating around 24 hours post-amputation, occurred in cells that were coalesced at the anterior pole. This expression phase presented the opportunity to define the cellular steps of anterior pole formation. Irradiated animals did not form an anterior pole and did not express FoxD at 24 to 72 hours following wounding (Fig. 2B), indicating the requirement of neoblasts for this process. FoxD-expressing cells at the regenerating anterior pole at this time also co-expressed multiple anterior PCGs as well as the anterior pole-expressing gene notum (Fig. 3A). follistatin, which is required for anterior regeneration [22], [23], was also co-expressed with FoxD at the regenerating anterior pole at this regeneration stage (Fig. 3A). FoxD expression gradually became restricted with time to fewer cells at the regenerating anterior pole, adopting the same appearance as in intact animals (Figs. 1 and 3A).

The planarian smedwi-1 gene encodes a PIWI family protein and is expressed in all dividing adult planarian cells, marking the neoblast population [1]. We found that some FoxD-expressing cells located at the forming anterior pole co-expressed smedwi-1 in regenerating blastemas 72 hours following amputation (Fig. 3B). In addition to FoxD, expression of the anterior pole gene notum and the anterior-patterning gene prep was also found in smedwi-1-expressing neoblasts at the regenerating anterior pole (Fig. 3B). These data suggest that by three days of regeneration some neoblasts have been specified to produce the new pole cells of the regenerating head, potentially marking the first cellular step in formation of a new anterior pole.

Hh signaling impacts wound-induced FoxD expression

Both Wnt and Hh signaling pathways are required for the head-versus-tail regeneration decision made at planarian amputation planes [4]–[8], [39]. To test whether wound-induced FoxD expression was regulated by these signaling pathways, we inhibited Wnt (β-catenin and APC) and Hh (patched (ptc) and hedgehog (hh)) pathway genes with RNAi and examined FoxD expression six hours following amputation. As controls, we analyzed the numbers of notum- and wnt1-expressing cells in the different RNAi conditions and, as expected from prior reports [5], [8], notum expression was exclusively affected in β-catenin and APC RNAi animals and wnt1 expression was affected following perturbation of the Hh signaling pathway (Figs. 4 and S3A). Perturbation of Wnt signaling did not affect wound-induced FoxD expression (Figs. 4 and S3A). By contrast, ptc(RNAi) animals displayed fewer than normal FoxD-expressing cells at wounds and hh(RNAi) animals had increased numbers of FoxD-expressing cells (Figs. 4 and S3A). patched (ptc) encodes a receptor for Hh that antagonizes pathway output [40]. Hh signaling therefore negatively regulates FoxD induction at the midline following wounding. Because hh impacts wound-induced wnt1 expression oppositely to FoxD [4], [5], this result does not reflect a generic requirement for hh in wound-induced gene activation. FoxD expression at the anterior pole was normal in hh(RNAi) anterior blastemas (Fig. S3B), indicating a specific role for hh in regulating the FoxD wound-induced phase of expression.

ptc(RNAi) animals with a strong phenotype regenerate tails in place of heads, but we found no evidence that FoxD influences the head-versus-tail regeneration choice.

However, ptc(RNAi) animals with a weak phenotype (e.g., cyclopic heads) do resemble FoxD(RNAi) animals [4], [5]. Cyclopic or headless ptc(RNAi) animals showed decreased expression of the anterior PCG sFRP-1 ([5] and Fig. S3D) and decreased expression of the anterior pole marker notum (Fig. S3D), indicating a defect in anterior pole regeneration. Therefore, the defect in FoxD expression might contribute to the ptc(RNAi) phenotype. In vertebrates, Sonic hedgehog (a member of the Hh family) signaling is required for normal forebrain development and midline induction [40], [41]. In planarians, hh is expressed ventrally and medially [4], [5]; the impact on wound-induced expression of FoxD at the midline raises the interesting possibility that hh might also have a role in midline biology in planarians. In ptc(RNAi) animals, midline expression of slit at six hours following amputation was normal (Fig. S3C). 86.3±5.5% of FoxD-expressing cells following wounding co-express the midline gene slit in wild-type animals. Therefore, these results indicate that the reduced wound-induced expression of FoxD in ptc(RNAi) animals is not a consequence of the absence of the midline cells that normally express FoxD.

Disruption of pole formation with FoxD RNAi perturbs anterior regeneration

FoxD(RNAi) animals displayed defective regeneration, with variable blastema size and heads that regenerated either one or no eyes (Fig. 5A). FoxD(RNAi) head blastemas had abnormal anatomy, with medial collapse of cephalic ganglia (labeled with an RNA probe to choline acetyltransferase (chat) [42]), one or no eyes (detected with an anti-ARRESTIN antibody (VC1) [43]), and a slightly abnormal anterior intestine morphology (labeled with an RNA probe to mat [1]) (Fig. 5A). Tail fragments regenerated pharynges (Fig. S4B), demonstrating that some missing tissues can regenerate in FoxD(RNAi) animals. Moreover, intestinal branches were normally regenerated in FoxD(RNAi) tail blastemas (Fig. S4C), indicating that FoxD has a largely specific role in anterior regeneration. Parasagittal thin fragments of FoxD(RNAi) animals also regenerated pharynges (Fig. S6B), further demonstrating that neoblasts can replace missing tissues. However, most of these fragments regenerated only one eye, showed slightly reduced expression of the anterior PCG sFRP1, and regenerated asymmetric cephalic ganglia (Fig. S6B).

Blastema size abnormalities in FoxD(RNAi) animals did not appear to be an overt consequence of a neoblast maintenance defect, because normal mitotic cell numbers were present in FoxD(RNAi) animals at the time of the amputation (0 hours following wounding, Fig. S4A). Following FoxD RNAi, neoblast (smedwi-1-expressing cells) proliferation (6 hours post-amputation, Fig. S4A) and migration (18 hours post-amputation, Fig. 5B, upper panel) in response to wounding were normal. At 48 hours following wounding, neoblasts of regenerating FoxD(RNAi) tail, but not trunk, fragments displayed slightly reduced neoblast proliferation (Fig. 5B and Fig. S4A). Reduced proliferation in FoxD(RNAi) tail fragments persisted five days following wounding (Fig. S4A).

FoxD(RNAi) animals have reduced numbers of follistatin-expressing cells at the anterior pole [22]. At 48 hours following wounding notum⁺ and notum⁺ follistatin⁺ cells were reduced from regenerating FoxD(RNAi) anterior blastemas (Fig. 5B and Fig. S5A). Furthermore, neoblasts expressing notum and prep were also fewer or absent in FoxD(RNAi) animals (Fig. S7), suggesting that FoxD is required for neoblast specification into anterior pole cell progenitors. These observations suggest a role for FoxD in anterior pole regeneration by specifying pole progenitors.

At 72 hours following head amputation, we observed significantly lower or complete absence in expression of the anterior PCGs prep, ndl-4, and sFRP-1, and decreased or no notum-coalesced cells in FoxD(RNAi) animals (Fig. 5C). After seven days of regeneration, FoxD(RNAi) anterior blastemas also showed severe defects in the expression of sFRP-1 and ndl-4, as well as the pole-restricted gene notum (Fig. 5D and Fig. S5B). Together, these results establish a role for FoxD and the anterior pole in head patterning during regeneration.

FoxD(RNAi) animals had normal numbers of notum- and wnt1-expressing cells at six hours following wounding (Fig. 6A), indicating that FoxD does not prevent the wound-induced phase of notum expression during the specification of head regeneration. Furthermore, wound-induced expression of follistatin [22], [23] was normal in FoxD(RNAi) animals (Fig. S5A). We also did not observe ectopic expression of posterior markers (wnt11-2 and wnt1) in regenerating anterior blastemas of FoxD(RNAi) animals (Fig. S4D), demonstrating that the choice to regenerate a head instead of a tail (AP regeneration polarity) was not detectably affected. In addition, we observed normal expression of posterior markers (wnt11-2 and wnt1) in regenerating posterior blastemas, indicating a specific role for FoxD in anterior regeneration (Fig. 6B).

pbx and prep are required for FoxD expression during anterior pole formation

The TALE-homeodomain-encoding pbx and prep genes are required for regeneration and maintenance of anterior PCG expression, including for markers of the anterior pole [17], [19], [20]. Because prep is expressed broadly at the head tip and pbx in most cells of the regenerating head [17], [19], [20], it was not previously possible to determine whether poles promoted anterior PCG expression or vice versa. However, because FoxD expression is restricted to the regenerating pole, the FoxD RNAi phenotype described above suggests the pole is required for anterior PCG expression. This model predicts that pbx and prep might be required for FoxD expression. Wound-induced FoxD expression at six hours following injury was normal in both pbx and prep RNAi animals (Fig. 7A). Similarly, other wound-induced genes (notum and wnt1) are expressed normally in pbx(RNAi) animals [19], [20], indicating that pbx and prep act downstream of wound-induced expression of these genes. By contrast, anterior pole-specific expression of FoxD and notum was completely absent in both pbx and prep RNAi animals 72 hours following head amputation (Fig. 7B). Because both FoxD and notum expression is induced in neoblasts at 72 hours following amputation of wild-type animals, the complete absence of expression of these two genes at this time point suggests a requirement for pbx and prep in the specification of neoblasts into anterior pole progenitors and that this defect might underlie the pbx and prep phenotypes.

FoxD is required for midline regeneration

The restricted, wound-induced expression of FoxD in the midline raises the possibility of a regenerative connection between the midline and the anterior pole. To explore this possibility further, we analyzed the expression pattern of several midline genes, such as slit, admp, and ephR1, in FoxD(RNAi) animals. Indeed, these midline genes were not properly expressed in regenerating FoxD(RNAi) anterior blastemas (Fig. 5D). In addition, parasagittal thin pieces showed reduced slit expression at the anterior midline (Fig. S6B). By contrast, tail blastemas of transversely amputated FoxD(RNAi) animals displayed normal extension of slit-expressing cells to the posterior pole (Fig. S4C), further demonstrating the specific role of FoxD in the biology of the anterior pole.

To further investigate the possible connection of the regenerating anterior pole and the midline, we examined the midline-expressed slit and ephR1 genes in pbx and prep RNAi animals, which are unable to regenerate an anterior pole (Fig. 7B and [19], [20]). pbx(RNAi) animals have been reported to have defective slit expression in anterior and posterior blastemas [19], [20]. As predicted, expression of both slit and ephR1 was defective in anterior blastemas of pbx and prep RNAi animals seven days following head amputation (Fig. 7B). Moreover, the midline expression of ephR1 was defective in all ptc(RNAi) animals showing intermediate phenotypes (cyclopia and headless; and therefore, reduced or absent anterior pole) at seven days following head amputation (Fig. S3D); because of the broad role of ptc it is unknown whether this defect is solely explained by a pole defect in ptc(RNAi) animals. hh(RNAi) animals, which can regenerate a normal anterior pole (Fig. S3B), showed proper expression of the ephR1 gene in anterior blastemas at seven days following amputation (Fig. S3B). Intact FoxD(RNAi) animals undergoing long-term RNAi did not show an obvious phenotype (Fig. S6A); following transverse amputation of these animals, small blastemas with one or no eyes were regenerated, and the anterior pole was rarely formed or very reduced (Fig. S6C, D). In some of these amputated FoxD(RNAi) animals a small cluster of anterior pole cells were regenerated but offset from the midline (n = 5/19 offset, with 14/19 showing medial but severely reduced poles, Fig. S6D). Altogether, our results suggest a role for the anterior pole in organizing the regeneration of the new midline and patterning of the head.

Discussion

The ends of embryos that will become the poles of the primary body axis can be involved in organizing embryonic tissue pattern in many animals. For example, bicoid mRNA in Drosophila is locally translated at the prospective anterior end of the oocyte and controls establishment of AP body pattern [44], [45]. PAR-3/PAR-6/aPKC-3 proteins localize to one side of the C. elegans zygote after fertilization, establishing the anterior embryo end [46]. In neurula-stage Xenopus embryos, Wnt genes regulate AP neurectoderm patterning and are active at the posterior, with Wnt inhibitory genes active at the anterior [47]. Early steps in development, such as asymmetric deposition of maternal determinants in oocytes or events triggered at the sperm entry site can be involved in establishment of polarized regions that impact primary axis pattern formation. In animals capable of whole-body regeneration, by contrast, the anterior and posterior poles must be formed de novo without reliance on events specific for embryogenesis. This problem exemplifies a general challenge faced by all regeneration paradigms: the establishment of pattern in an adult tissue context utilizing initial steps that can differ from the cues accessible to embryos. Planarian pole regeneration presents an ideal setting to explore the mechanistic basis for this process.

FoxD encodes a Forkhead-family transcription factor that functions during planarian regeneration to specify cells at a key signaling center occurring at the intersection of the anterior extreme and the midline. Transcription factors expressed regionally in embryos can promote establishment of signaling domains that control tissue pattern. A classic example is goosecoid, which encodes a homeodomain transcription factor required for expression of many patterning factors at Spemann's organizer in frog embryos [48]. In regeneration, how transcription factors might be suitably activated to organize regeneration of tissue pattern is poorly understood. The early and restricted midline expression of FoxD following wounding, the subsequent expression of FoxD at the anterior pole (which is located at the midline), and the anterior pole and midline regeneration phenotype in FoxD, pbx, and prep RNAi animals suggest that the anterior pole organizes many aspects of head regeneration. We propose the following model for head regeneration (Fig. 7C):

(i) Anterior wound response

In the first phase, the anterior wound response (3–18 hours post-amputation), signaling mechanisms occurring in pre-existing muscle cells at wounds determine whether a new head is to be regenerated. Wound signaling triggers rapid (within ∼6 hours) expression of wnt1 at all wounds [37]. Wnt signaling is selectively inhibited at anterior-facing wounds, involving activation of the notum gene [8]. The mechanism that leads to selectivity in notum activation is unknown. This first phase of head regeneration results in inhibition of Wnt signaling at anterior-facing wounds, whereas Wnt signaling is active at posterior-facing wounds. FoxD is activated at the midline of most wounds during this initial phase of wound-induced wnt1 and notum expression, but our data do not indicate a role for this gene in the decision to regenerate a head-versus-tail.

(ii) Anterior pole progenitor specification

If an anterior-facing wound is not juxtaposed by anterior tissue, a second phase of head regeneration involving anterior pole progenitor specification ensues. This phase (24–72 hours post-amputation) involves formation of a new anterior pole in an emerging blastema. The initial medial FoxD expression presents a candidate mechanism for establishing the location of pole regeneration–at the prior midline. This location of FoxD induction highlights the connection between the midline and pole regeneration as an important area for further investigation. We found that a small cluster of cells at the midline expressing notum, follistatin, and FoxD emerges from neoblasts in this second phase. Neoblasts can be specified to form eyes [48] or protonephridia [49] in regeneration, and here we demonstrate that some neoblasts near the forming anterior pole express FoxD, notum, and prep. We propose that neoblast induction of these genes near the midline at wounds represents an initial cellular step in formation of the cells of the new anterior pole.

(iii) Anterior patterning

In a third phase of head regeneration (∼48 hours+ post-amputation), as the blastema grows substantially, pattern of the head blastema is established. FoxD RNAi severely reduced anterior pole regeneration. Whereas these FoxD(RNAi) anterior blastemas still displayed anterior character (such as the presence of brain cells), the blastemas lacked the normal tissue organization of wild-type head blastemas. FoxD(RNAi) blastemas also had defects in the expression of a number of anterior PCGs with broad anterior expression domains (e.g., prep, sFRP-1). This raises the possibility that the anterior pole is involved in establishing and maintaining more general anterior patterns of gene expression during this third phase of head regeneration. Consistent with this hypothesis, head tissue can also be regenerated in pbx or prep RNAi animals lacking poles, but establishment of anterior patterning gene expression domains is severely affected [17], [19], [20]. Finally, expression domains of genes expressed at the midline were aberrant in FoxD(RNAi) head blastemas. Together, these results suggest that the regenerating anterior pole promotes midline regeneration for proper bilateral patterning of the head and promotes establishment of gene expression domains for AP patterning of the head.

Materials and Methods

Animals and radiation treatment

Asexual Schmidtea mediterranea strain (CIW4) animals starved 7–14 days prior experiments were used. Animals were exposed to a 6,000 rads dose of radiation using a dual Gammacell-40 137 cesium source and amputated three days after irradiation. No vertebrate animals were used in this study, and usage of planarians (invertebrates) is unregulated.

RNAi

Animals were injected with control (the C. elegans gene unc-22), FoxD, or pbx dsRNA. Animals were transversely amputated and trunk pieces injected within 1 hour post-amputation with dsRNA. A booster dsRNA injection of trunk pieces was performed the following day. This procedure (amputation, injection and booster injection) was performed a total of three times, every two to three days. Following the third cycle of amputation and injections, animals were scored for phenotype at 72 hours and seven days post amputation, fixed and analyzed with in situ hybridizations [50] and for many experiments involved azide quenching as described [50].

For RNAi by feeding experiments, dsRNA-expressing bacteria cultures were mixed with 70% liver solution in a 1∶300 ratio to culture volume. β-catenin, APC, and prep RNAi animals have been fed four times (days 0, 4, 7 and 11), and amputated at d12. 72 hours or seven days following amputation, animals were fixed and in situ hybridizations performed. hh and ptc RNAi animals were fed six times every three to four days. Long-term FoxD(RNAi) homeostasis experiments were performed by feeding the animals every three to four days during a ten week-period.

Whole-mount in situ hybridizations and immunostainings

Animals were wounded and fixed at six hours following injury in 4% formaldehyde [50] and nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP) colorimetric whole-mount in situ hybridizations or fluorescence in situ hybridizations (FISH) were performed as described [50]. For immunostainings, animals were fixed as for in situ hybridizations and then treated as described [51]. A mouse anti-ARRESTIN antibody was kindly provided by Kiyokazu Agata and used in a 1∶5,000 dilution, and an anti-mouse-Alexa conjugated antibody was used in a 1∶500 dilution. For the neoblast wound response assay, RNAi animals were fed during the course of eight weeks every three to four days, then were transversely amputated and trunk or tail fragments were fixed at 0, 6, 18, 48, and 120 hours following wounding. Animals were immunostained using a rabbit anti-phospho histone 3 antibody and an anti-rabbit HRP in a 1∶100 dilution as previously described [52]. Fluorescent images were taken with a Zeiss LSM700 Confocal Microscope. Light images were taken with a Zeiss Discovery Microscope.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7