Cellular and Molecular Responses Unique to Major Injury Are Dispensable for Planarian Regeneration

SUMMARY

The fundamental requirements for regeneration are poorly understood. Planarians can robustly regenerate all tissues after injury, involving stem cells, positional information, and a set of cellular and molecular responses collectively called the “missing tissue” or “regenerative” response. follistatin, which encodes an extracellular Activin inhibitor, is required for the missing tissue response after head amputation and for subsequent regeneration. We found that follistatin is required for the missing tissue response regardless of the wound context, but causes regeneration failure only after head amputation. This head regeneration failure involves follistatin-mediated regulation of Wnt signaling at wounds and is not a consequence of a diminished missing tissue response. All tested contexts of regeneration, including head regeneration, could occur with a defective missing tissue response, but at a slower pace. Our findings suggest that major cellular and molecular programs induced specifically by large injuries function to accelerate regeneration but are dispensable for regeneration itself.

Graphical Abstract

In Brief

In regenerative organisms, a large array of cellular responses are triggered at major injuries. However, which of these responses are fundamentally required for regeneration to occur remains unknown. Tewari et al. find that hallmark cellular and molecular responses induced uniquely at large injuries are dispensable for planarian regeneration.

INTRODUCTION

Regeneration requires the ability to respond to injury and to replace missing body parts. How animals, such as planarians, are able to regenerate after large injuries that remove multiple tissue types is a question that has fascinated biologists for centuries. Some injuries, such as incisions, wound the animal but do not remove substantial tissue. Other injuries, such as amputations, remove significant tissue that must be replaced to return the animal to normal anatomical proportions.

Planarian regeneration requires a population of dividing stem cells called neoblasts and positional information (Reddien, 2018). Position control genes (PCGs) are regionally expressed, primarily in muscle, and are proposed to guide patterning along the body axes (Witchley et al., 2013). Substantial work has shown that a major component of regeneration after significant tissue loss is the “missing tissue,” or “regenerative,” response (Pellettieri et al., 2010; Wenemoser and Reddien, 2010; Wenemoser et al., 2012; Wurtzel et al., 2015; Owlarn et al., 2017).

When challenged with an injury, planarians launch a generic wound response. This response occurs rapidly, between 1 and 6 hr post wounding, and includes induction of an estimated 200+ genes, increased cell division broadly, and increased cell death near the wound (Bagunñà, 1976; Saló and Baguñà, 1984; Pellettieri et al., 2010; Wenemoser and Reddien, 2010; Wenemoser et al., 2012; Wurtzel et al., 2015). These responses occur regardless of whether or not the injury is associated with substantial missing tissue. Injuries associated with substantial tissue loss induce a second set of responses collectively called the missing tissue response. Hallmarks of the missing tissue response include a second and sustained peak in mitoses near the wound, sustained expression of wound-induced genes until ~24 hr post injury, and a body-wide increase in apoptosis (Pellettieri et al., 2010; Wenemoser and Reddien, 2010; Wenemoser et al., 2012).

The gene follistatin, which encodes a secreted inhibitor of Activin signaling proteins, is required for the missing tissue response at anterior-facing wounds, but has no effect on the generic wound response (Gaviño et al., 2013). In addition, follistatin is required for anterior regeneration, with tail fragments unable to form any head structures after amputation (Gaviño et al., 2013; Roberts-Galbraith and Newmark, 2013). These observations suggested that the missing tissue response might be required for regeneration.

Although it is plausible that prominent cellular and molecular responses induced specifically by large injuries are required for regeneration, there are some indications that the missing tissue response is not required in all instances of regeneration. Planarians constantly replace adult tissues during homeostatic tissue turnover, even in the absence of injury. Specific removal of the planarian eye by surgical resection does not induce a missing tissue response, and yet eye regeneration occurs as an emergent property of constant progenitor production (LoCascio et al., 2017). In addition, although head regeneration is consistently affected by follistatin inhibition, tail regeneration does occur to some extent by day 6 post amputation (Roberts-Galbraith and Newmark, 2013). Here we utilized inhibition of follistatin as a tool to disrupt induction of the missing tissue response after a wide variety of injuries. This allowed us to study the fundamental requirements of regeneration after significant tissue loss in multiple wound contexts. Our data indicate that wound-induced re-establishment of positional information can lead to regeneration, even without major cellular and molecular responses that define the missing tissue response.

RESULTS

follistatin Is Required for Regeneration Only after Complete Head Amputation

To test the requirement for follistatin in planarian regeneration broadly, we inflicted a series of injuries that removed varying amounts of tissue after inhibition of follistatin by RNAi. Eye resection does not induce a missing tissue response (LoCascio et al., 2017); therefore, eye regeneration following resection was anticipated to occur in follistatin RNAi animals. Indeed, all follistatin RNAi animals regenerated missing eyes by day 20 after surgery (Figure 1A). Next, we tested larger injuries that removed different types of tissue and that were anticipated to elicit a missing tissue response (Wenemoser and Reddien, 2010; LoCascio et al., 2017). Unexpectedly, almost all follistatin RNAi animals subjected to pharynx resections, removal of tissue wedges from the head, sagittal amputations, or removal of posterior tissue regenerated by day 20 after surgery (Figure 1A). A variety of differentiated cell types removed by each of these injury types were regenerated in proper patterns by day 20 post amputation (Figure 1A). In fact, the only tested condition under which follistatin RNAi animals failed to regenerate was at anterior-facing wounds (Figure 1A).

Figure 1. follistatin Is Required for Regeneration Only after Complete Head Amputation in the Posterior.

(A) follistatin is only required for head regeneration. For each surgery: top: live images 20 days after surgery; bottom: fluorescence in situ hybridization (FISH) for differentiated tissue markers 20 days after surgery. opsin, photoreceptor neurons (green, yellow); grh-1, pharynx neurons (magenta); NB.22.1e, dorsal-ventral (DV) boundary, mouth and esophagus (yellow); cintillo, sensory neurons (magenta); gad, GABAergic neurons (magenta); notum, anterior pole (magenta); wnt1, posterior pole (magenta). Arrowheads indicate lack of regeneration. Two independent experiments. dpa, days post amputation.

(B) The requirement for follistatin in head regeneration diminishes at amputations made at serially anterior locations along the anterior-posterier (AP) axis. “n” describes the number of animals that look like the representative image. % head regeneration is provided in yellow. Bottom: FISH for differentiated tissue markers notum, cintillo, gad pool (anterior pole, sensory neurons, GABAergic neurons; magenta), and opsin (photoreceptor neurons; green) at 20 dpa (dorsal view). Two independent experiments.

Scale bars, 200 μm. See also Figure S1.

The Requirement for follistatin in Head Regeneration Is Specific to the Injury Location on the Anterior-Posterior Axis

The requirement of follistatin specifically for head regeneration was puzzling. Given that follistatin RNAi animals were able to replace anterior cell types after removal of tissue wedges from the head (Figure 1A), we reasoned that there was not a defect in the capacity to make anterior tissue but rather an inability to initiate head regeneration at certain wounds. We therefore assessed whether the position of amputation along the anterior-posterior (AP) axis affects the head regeneration outcome after follistatin RNAi. At amputation planes immediately anterior to the base of the pharynx, 0% of follistatin RNAi animals regenerated a head (n = 0/30), as expected (Gaviño et al., 2013; Roberts-Galbraith and Newmark, 2013). Following amputation performed slightly anterior to this location, however, 16.1% of follistatin RNAi animals regenerated heads (n = 5/31) (Figure 1B). In fact, with each progressively anterior amputation, a larger fraction of animals successfully regenerated a head. At an amputation plane immediately posterior to the auricles (AP1), a location posterior to the brain demarcating the head, 100% of animals regenerated heads (n = 36/36) (Figures 1B and S1A). Amputation at AP1 removes most specialized structures in the head, including the anterior pole, a structure known to act as an organizer for head tissue (Scimone et al., 2014; Vásquez-Doorman and Petersen, 2014; Vogg et al., 2014; Oderberg et al., 2017) (Figures S1A and S1B). Nonetheless, these follistatin RNAi fragments regenerated heads without detectable abnormalities, as evidenced by the presence of eyes and a variety of neuronal cell types by day 20 post amputation and successful formation of the anterior pole (Figure 1B). We determined that regeneration at AP1 was not a consequence of RNAi efficacy declining over time. follistatin inhibition levels remained comparable from day 0 to day 14 post amputation at AP1 without further delivery of double-stranded (ds)RNA (Figure S1C). In an additional experiment, regeneration at AP1 occurred despite continual injection of follistatin dsRNA up to 14 days post injury, which also showed a consistent follistatin inhibition level throughout regeneration (Figure S1D). Finally, 100% of follistatin RNAi animals that successfully regenerated after amputation at AP1 failed to regenerate heads when subsequently amputated in the tail without further delivery of follistatin dsRNA (Figure S1E). These results suggest that the requirement of follistatin in head regeneration is dependent upon the location of injury along the AP axis.

follistatin Is Required for the Missing Tissue Response at Multiple Injuries

Prior work has shown that follistatin is required for the missing tissue response at anterior-facing amputations in tail fragments (Gaviño et al., 2013). The results described above raised the question of whether follistatin is required for the missing tissue response specifically at anterior-facing wounds made in the posterior, with this requirement explaining the head regeneration failure phenotype after this injury. To address this question, we examined the missing tissue response at diverse wound types after follistatin RNAi by assessing its three hallmarks: the second peak in proliferation at wounds, sustained expression of wound-induced genes, and the animal-wide increase in apoptosis 72 hr post injury. We found that at posterior-facing wounds and at wounds associated with the removal of a wedge of tissue from the head, the second peak in mitotic cell numbers at the wound was significantly reduced after follistatin RNAi (Figures 2A–2C). After tissue wedge removal from the head, proliferation near the wound in follistatin RNAi animals was comparable to day 0 from day 2 to day 14 post injury despite eye regeneration having occurred by this time point, indicating regeneration could occur with no detectable proliferative response (Figures 2B and S2A). Mitotic numbers were comparable in uninjured follistatin and control RNAi animals, suggesting normal levels of proliferation exist during homeostatic tissue turnover (Figures S2B and S2C). Second, the expression levels of inhibin-1 and runt-1 (Wenemoser et al., 2012) were reduced at posterior-facing wounds 24–48 hr post tail amputation in follistatin RNAi animals (Figures 2D and S2D). Finally, TUNEL⁺ cell numbers were significantly reduced in follistatin RNAi head fragments 72 hr post injury, whereas TUNEL⁺ cell numbers at 0 hr post injury were comparable to control RNAi animals (Figures 2E and 2F). These data indicate that follistatin is required for multiple components of the missing tissue response at diverse wounds, rather than just at anterior-facing wounds.

Figure 2. follistatin Is Required for the Missing Tissue Response at Multiple Injury Types.

(A) follistatin is required for the second mitotic peak after tail amputation and removal of tissue wedges from the head. Phospho-histone H3 (H3P) antibody labeling to mark mitotic cells (green) after the indicated injury at the specified time point (ventral view). Two independent experiments.

(B) Graph displays the number of H3P⁺ cells counted in the region marked by the blue box at the indicated times after head wedge removal. ****p < 0.0001. n > 8 per time point; n > 30, 2 dpa; n > 20, 0 hours post amputation (hpa).

(C) Graph displays the number of H3P⁺ cells counted in the region marked by the blue box at 0 hpa and 36 hpa. ****p < 0.001, ***p < 0.001, *p < 0.05.

(D) follistatin is required for perduring wound-induced gene expression at posterior-facing wounds. FISH for the wound-induced gene inhibin-1 (yellow) at 24 hpa in regenerating head fragments (dorsal view). Black box indicates the region shown. Two independent experiments.

(E) follistatin is required for the 72-hr wave of apoptosis after tail amputation. TUNEL marked cells undergoing apoptosis (orange) at 72 hpa in regenerating head fragments (ventral view). Four independent experiments.

(F) Graph displays the number of TUNEL⁺ cells counted in regenerating head fragments at the indicated time points. ****p < 0.0001.

(G) follistatin is required for the missing tissue response after AP1 amputation. FISH for the wound-induced gene inhibin-1 (yellow) at 24 hpa (dorsal view). H3P antibody labeling to mark mitotic cells (green) at 48 hpa (ventral view). TUNEL marked apoptotic cells (orange) in pharynges at 72 hpa. Black box indicates the region shown. Two independent experiments.

(H) Graph displays the number of H3P⁺ cells counted near the wound, indicated by the blue box. Anterior regeneration at AP1 displayed a higher second mitotic peak than posterior regeneration in control RNAi animals. n ≥ 8 for each time point. n > 20, 0 hpa; n > 25, 2 dpa. ****p < 0.0001.

(I) Graph displays the number of TUNEL⁺ cells counted in the pharynges of regenerating trunks indicated by the blue box. n ≥ 6 for each time point. ***p < 0.001, *p < 0.05.

(J) Live images of head regeneration outcome 10 dpa at AP1 after follistatin RNAi.

Scale bars, 200 μm. Error bars represent mean ± SD. NS indicates no significant difference. See also Figure S2.

As described above, follistatin RNAi animals can regenerate heads after amputation at AP1 (Figure 1B). We examined the missing tissue response at this injury and found that follistatin RNAi animals displayed reduced wound-induced gene expression at 24 hr post injury, no detectable second mitotic peak, and no detectable elevation in apoptosis levels 72 hr after wounding (Figures 2G–2I). In fact, levels of mitosis and apoptosis in regenerating follistatin RNAi animals at AP1 were comparable to levels at 0 hr after injury from day 2 through day 10 post amputation, despite successful regeneration of eyes by 10 days post injury. Therefore, regeneration occurred despite no detectable missing tissue response during the course of regeneration from this injury (Figures 2H–2J). These findings together indicate that the missing tissue response is not required for regeneration following a large array of injury types, including small injuries such as eye resection, large injuries such as removal of the posterior half (or more) of the body, and injuries that remove the head.

Regeneration Occurs with a Defective Missing Tissue Response, but at a Slower Rate

Although follistatin RNAi animals were able to fully regenerate from multiple injury types, they formed very small blastemas and appeared to replace tissue slower than did controls (Figure 3A). To test this possibility, we used multiple differentiated tissue markers to assess the rate of tissue formation after follistatin inhibition and various amputations. In the case of tail amputation, we found that by day 7 post amputation, follistatin RNAi animals, unlike controls, had not regenerated the posterior zone of marginal adhesive gland cells (mag-1⁺) and had accumulated mouth cells at the wound face but not yet formed a defined mouth (NB.22.1e⁺) (Figure 3B). In addition, when we tracked the appearance of pharyngeal tissue over time, using a marker for neurons located at the pharynx tip (Collins et al., 2010), we found that follistatin RNAi animals regenerating the posterior half of their bodies had significantly smaller pharynges compared to controls between day 4 and day 10 post amputation (Figures 3B and 3C). However, by day 20 after injury, the posterior zone of marginal adhesive gland cells, the mouth, and the pharynx were completely formed in follistatin RNAi animals (Figure S3A).

Figure 3. Posterior Regeneration Occurs with a Defective Missing Tissue Response, but at a Slower Rate.

(A) follistatin RNAi head fragments regenerate slowly with small blastemas. Live images at the indicated time points post amputation. Asterisks mark a newly formed pharynx. Arrowheads indicate a blastema.

(B) Multiple differentiated tissue structures appear small or absent at 7 days post tail amputation after follistatin RNAi. FISH to mark marginal adhesive gland cells (mag-1⁺, green), the mouth (NB.22.1e⁺, yellow), and pharyngeal neurons (grh-1, magenta) (ventral view). Black box indicates the region shown. Two independent experiments.

(C) follistatin RNAi animals regenerate pharynges slowly and are indistinguishable from controls by 14 days post tail amputation. Graph indicates the number of grh-1⁺ cells counted per animal at the specified time points post tail amputation. Data are plotted as mean ± SD. n > 8 for each time point. ****p < 0.0001, *p < 0.05.

(D) The posterior pole forms slowly in regenerating head fragments after follistatin RNAi. FISH for wnt1 to mark the posterior pole (magenta). Dorsal view. Black box indicates the region shown. Two independent experiments.

(E) Posterior patterning occurs slowly in regenerating head fragments after follistatin RNAi. FISH for posterior patterning genes wntP-2 (magenta) and fz-4, wnt11-1, and wnt11-2 pool (green) at the indicated time points after tail amputation (ventral view). Black box indicates the region shown. Two independent experiments.

Scale bars, 200 μm. See also Figure S3.

Regeneration also involves the replacement of lost positional information. Specifically, expression domains of patterning genes (PCGs) return. Some of these expression changes occur in existing muscle cells at wounds and some occur in newly produced muscle cells (Witchley et al., 2013). At 72 hr post injury, posterior-facing wounds begin to form a posterior pole, a group of muscle cells that accumulate near the midline at the posterior tail tip and express the gene wnt1 (Petersen and Reddien, 2009). In follistatin RNAi head fragments, wnt1⁺ cells near the midline of the animal were present at 72 hr but had not yet coalesced into a pole, whereas coalesced poles were present in control animals (Figure 3D). The gene wntP-2 (sometimes referred to as wnt11-5; Gurley et al., 2010) is expressed in a broad posterior-to-anterior transcriptional gradient in muscle and is detectably upregulated at posterior-facing wounds ~30 hr following amputation (Petersen and Reddien, 2009). wntP-2 was expressed at posterior-facing wounds of follistatin RNAi animals, but its expression at 3–5 days post amputation was lower than in controls (Figure 3E) (Gaviño et al., 2013). Regeneration of the expression domains of several posteriorly expressed PCGs is dependent on new cell formation (neoblast proliferation) (Gurley et al., 2010). Regeneration of these expression domains was delayed in follistatin RNAi animals (Figure 3E). These data suggest that re-establishment of posterior patterning information can occur with a defective missing tissue response; however, it proceeds slowly. Because regenerative patterning involves both changes in pre-existing muscle cells and new muscle cell production, the slower pace of PCG expression re-establishment is consistent with the lower rate of proliferation in follistatin RNAi animals during regeneration.

Similar to the case of posterior amputation, regeneration after amputation of follistatin RNAi animals at AP1 occurred despite appreciably reduced blastema size (Figure 4A). Formation of both tissue-specific progenitors and differentiated structures occurred slowly at this injury. Between 2 and 5 days post amputation at AP1, ovo⁺ eye progenitors were fewer in follistatin RNAi animals than in controls (Figure S4A) and, at day 7 post amputation, follistatin RNAi animals had significantly fewer cintillo⁺ sensory neurons (Figures 4B and 4C), opsin⁺ photoreceptor neurons (Figures 4B and 4D), and gad⁺ GABAergic neurons (Figure S4B).

Figure 4. Head Regeneration at AP1 Occurs in the Absence of a Detectable Missing Tissue Response, but at a Slower Rate.

(A) follistatin RNAi animals regenerate heads slowly after amputation at AP1. Live images at the indicated time points post amputation. Arrowheads indicate the first appearance of eyes.

(B) Differentiated anterior tissues are less developed than in controls 7 days post AP1 amputation in follistatin RNAi animals. FISH for differentiated tissue markers opsin (photoreceptor neurons, green; dorsal view) and cintillo (sensory neurons, magenta; ventral view). Black box indicates the region shown. Two independent experiments.

(C) Number of cintillo⁺ sensory neurons at 7 dpa. Blue box indicates the region quantified. ****p < 0.0001.

(D) Number of opsin⁺ photoreceptor neurons at 7 dpa. Blue box indicates the region quantified. ****p < 0.0001.

(E) Anterior patterning and pole formation do not occur after amputation below the pharynx in follistatin RNAi animals. Top: FISH for notum (magenta) and sFRP-1 (green) at 5 dpa (ventral view). Bottom: FISH for ndl-2 (magenta) and ndl-5 (green) at 5 dpa (ventral view). Black box indicates the region shown. Two independent experiments.

(F) Anterior patterning and pole formation occur slowly after amputation at AP1 in follistatin RNAi animals. Top: FISH for notum (magenta) and sFRP-1 (green) at 72 hpa (ventral view). Bottom: FISH for ndl-2 (magenta) and ndl-5 (green) at 5 dpa (ventral view). Black box indicates the region shown. Two independent experiments.

Scale bars, 200 μm (A, E, and F) and 100 μm (B). Error bars represent mean ± SD. See also Figure S4.

We also examined the re-setting of positional information after amputation at different AP locations in follistatin RNAi animals. follistatin RNAi tail fragments obtained by amputation below the pharynx did not form an anterior pole or express anterior PCGs that depend on new muscle cell formation for their expression during regeneration (Figures 4E and S4C) (Roberts-Galbraith and Newmark, 2013). Tail fragments also did not perform patterning steps that occur in existing muscle cells, such as rescaling the expression domain of the posterior PCG wntP-2 (Gaviño et al., 2013) and expressing the anterior PCG wnt-2 (Figure S4C). Posterior fragments obtained by amputation at an intermediate location, AP2, expressed notum and sFRP-1 and formed photoreceptor neurons in ~50% of fragments (Figure S4D). By contrast, all follistatin RNAi animals amputated at AP1 re-established anterior patterning information, but slowly, with reduced expression of the anterior PCG ndl-5 and delayed coalescence of the anterior pole between days 3 and 5 post injury (Figures 4F and S4E).

Taken together, these findings support a model in which the missing tissue response accelerates patterning and tissue replacement but is not required to bring about regeneration of any missing body part, including a head, following a large diversity of injury classes.

follistatin Inhibits Early Wound-Induced wnt1 Expression at All Injuries

Why is follistatin required for head regeneration at amputations in the trunk and posterior of animals, but not required for head regeneration after amputation in the anterior? Wnt signaling is known to be important for re-setting positional information and for replacing missing body parts after injury to the AP axis (Gurley et al., 2008; Iglesias et al., 2008; Petersen and Reddien, 2008, 2009, 2011; Adell et al., 2009). Planarians maintain a gradient of Wnt signaling activity along their AP axis during homeostasis and re-establish this gradient during regeneration (Gurley et al., 2008, 2010; Petersen and Reddien, 2008, 2009; Adell et al., 2009; Sureda-Gómez et al., 2016; Stückemann et al., 2017). Soon after wounding, both anterior-facing and posterior-facing wounds (indeed, essentially all wounds) express the wnt1 gene (Petersen and Reddien, 2009). Another wound-induced gene, notum, is preferentially induced at anterior-facing wounds over posterior-facing wounds (Petersen and Reddien, 2011; Wurtzel et al., 2015). notum encodes a broadly conserved deacylase that inhibits the action of Wnt ligands (Kakugawa et al., 2015; Zhang et al., 2015). Both wnt1 and notum are wound induced in muscle cells (Witchley et al., 2013), and within 24 hr post amputation a low-Wnt signaling environment is established at anterior-facing amputations (head generating) and a high-Wnt environment is established at posterior-facing amputations (tail generating) (Stückemann et al., 2017). notum RNAi leads to regeneration of tails instead of heads at anterior-facing wounds (Petersen and Reddien, 2011), and wnt1 RNAi leads to either failed tail regeneration or the regeneration of heads instead of tails at posterior-facing wounds (Adell et al., 2009; Petersen and Reddien, 2009).

Given the importance of Wnt signaling in the decision to make a head or tail, we asked whether follistatin might impact wound-regulated Wnt signaling to mediate its role in head regeneration. We assessed the expression of notum and wnt1 early after wounding in regenerating tail fragments, amputated below the pharynx. Wound-induced notum expression was normal between 6 and 24 hr post amputation (Figures 5A, S5A, and S5B); however, the formation of a notum⁺ anterior pole at later time points did not occur in follistatin RNAi tail fragments, as previously reported (Roberts-Galbraith and Newmark, 2013) (Figure S5A). Wound-induced wnt1 expression, however, was robustly higher compared to controls in follistatin RNAi tail fragments between 6 and 12 hr post amputation, before returning to normal levels by 24 hr after injury (Figures 5B, 5C, and S5A). Elevated wound-induced expression of wnt1 after follistatin RNAi was observed at all wound types tested, including wounds that did not remove substantial tissue, and occurred in muscle cells, which is the normal site of wound-induced wnt1 expression (Figures 5B and S5C). This effect on wound-induced gene expression was specific to wnt1, with multiple other wound-induced genes, including those induced in muscle, not affected by follistatin RNAi at 6 hr post injury (Scimone et al., 2017) (Figures 5A and 5C). We also did not observe any changes in the expression of β-catenin-1-sensitive genes during homeostasis in follistatin RNAi animals, suggesting that homeostatic Wnt signaling levels were unaffected (Figure S5D). RNAi of activin-1, a gene encoding an Activin-like TGF-β-family signaling ligand, is known to suppress the head regeneration defect of follistatin RNAi tail fragments (Gaviño et al., 2013; Roberts-Galbraith and Newmark, 2013). We found that increased wnt1 expression at wounds after follistatin RNAi was dependent on activin-1, (Figures 5D, S5E, and S5F), but activin-1 RNAi had no effect on notum expression at wounds (Figure S5G). This suggests that follistatin is required to inhibit the expression levels of wound-induced wnt1 in an Activin-dependent manner.

Figure 5. follistatin Inhibits Wound-Induced Expression Levels of wnt1 at Diverse Injuries.

(A) Expression levels of wound-induced notum and inhibin-1 are normal after follistatin RNAi. FISH for notum (green) and inhibin-1 (yellow) at 6 hpa in regenerating tail fragments (ventral view). Black box indicates the region shown. Two independent experiments.

(B) follistatin is required to inhibit wnt1 expression early after wounding at many injuries. FISH for wnt1 (magenta) at 6 hr after the indicated surgeries (ventral view). Black box indicates the region shown. Two independent experiments.

(C) follistatin is not required to inhibit expression of other wound-induced genes at 6 hpa. Heatmap depicts RNA sequencing (RNA-seq) data of anterior-facing wounds collected from follistatin RNAi and control RNAi tail fragments (Scimone et al., 2017). Data are presented as log₂ fold change in gene expression between follistatin RNAi and control RNAi at the indicated time points post amputation. *p_adj < 0.05, **p_adj < 0.01, ***p_adj < 0.001, ****p_adj < 0.0001. ^†Best human BLAST hits.

(D) follistatin-mediated inhibition of wnt1 is dependent on activin-1. Top: feeding regimen for RNAi. Bottom: FISH for wnt1 (magenta) at 6 hpa in each RNAi condition. Black box indicates the region shown. Two independent experiments.

Scale bars, 200 μm. See also Figure S5.

Inhibition of wnt1 Suppresses the Head Regeneration Defect of follistatin RNAi Animals

Defective follistatin-mediated inhibition of the Wnt-ligand-encoding wnt1 gene could explain why head amputations made at posterior locations fail to regenerate. The planarian posterior has inherently high Wnt signaling during homeostasis (Petersen and Reddien, 2008; Adell et al., 2009; Gurley et al., 2010; Sureda-Gómez et al., 2016; Stückemann et al., 2017), and anterior-facing wounds in the posterior must generate a low-Wnt environment for head regeneration to occur. Increased wound-induced wnt1, in what is naturally a high-Wnt signaling environment, could therefore lead to head regeneration failure in follistatin RNAi animals. Conversely, many genes encoding Wnt inhibitors, including notum and sFRPs, are expressed in the low-Wnt anterior region of the animal during homeostasis (Gurley et al., 2008, 2010; Petersen and Reddien, 2008, 2011). This low-Wnt anterior environment might favor head regeneration even in the presence of elevated levels of wnt1 at the wound face.

To determine whether the elevated level of wnt1 expression in follistatin RNAi animals is required for the head regeneration defect observed, we simultaneously inhibited follistatin and wnt1 and assessed regeneration in tail fragments. Whereas almost all tails treated with follistatin; control dsRNA (36/37) did not regenerate a head, 32/49 tails treated with wnt1; follistatin dsRNA successfully regenerated, despite similar follistatin expression-level reduction in both conditions (Figures 6A and S6A). The regenerated heads of wnt1; follistatin double-RNAi animals had no observed morphological abnormalities (Figures 6A and S6B). We performed a similar experiment by simultaneous inhibition of follistatin and β-catenin-1, the intracellular effector of canonical Wnt signaling. β-catenin-1 RNAi also rescued the head regeneration failure phenotype of follistatin RNAi animals (Figures S6C and S6D). In addition, we performed double RNAi of myoD and β-catenin-1. myoD is required for wound-induced expression of follistatin but does not affect wound-induced wnt1 expression, and myoD RNAi leads to regeneration failure (Scimone et al., 2017). We found that myoD; β-catenin-1 RNAi tails were also able to regenerate despite reduced expression of follistatin at wounds after 1 week of myoD RNAi (Figures S6E and S6F). Previous work has shown that myoD; β-catenin-1 RNAi tails do not regenerate heads after 3 weeks of myoD RNAi (Scimone et al., 2017). myoD is required for the specification of longitudinal muscle fibers, suggesting that after 3 weeks of dsRNA treatment a larger reduction in longitudinal fibers, and further loss of any attendant additional roles of these fibers, results in the failed regeneration of myoD; β-catenin-1 double-RNAi animals. These data indicate that inhibition of Wnt signaling can rescue the head regeneration defect caused by RNAi of follistatin.

Figure 6. wnt1 Inhibition Restores Head Regeneration after follistatin RNAi Despite a Defective Missing Tissue Response.

(A) wnt1 RNAi suppresses the head regeneration defect after follistatin RNAi. Top: feeding regimen for RNAi. Bottom: live images of tail fragments at 14 dpa. Two independent experiments.

(B) Simultaneous inhibition of wnt1 and follistatin does not restore the secondary mitotic peak in regenerating tail fragments. Graph displays the number of H3P⁺ cells counted in the region indicated by the blue box at 48 hpa for each RNAi condition. ***p < 0.001. Three independent experiments.

(C) Inhibition of wnt1 does not suppress the defect in the 72-hpa apoptotic wave caused by follistatin RNAi. Graph: TUNEL⁺ cell numbers per pharynx (indicated by the blue box) at 72 hpa, in regenerating trunks, for each RNAi condition. ****p < 0.0001. Three independent experiments.

(D) wnt1; follistatin double-RNAi animals regenerate eyes slowly compared to controls. FISH for opsin (photoreceptor neurons, green) at 7 dpa (dorsal view). Black box indicates the region shown.

(E) Graph displays the number of opsin⁺ photoreceptor neurons counted per animal for each RNAi condition at the specified time points.

Scale bars, 200 μm (A) and 100 μm (D). Error bars represent mean ± SD. NS indicates no significant difference. See also Figures S6 and S7.

Head Regeneration from the Posterior in wnt1; follistatin Double-RNAi Animals Occurs Despite a Diminished Missing Tissue Response

We considered the possibilities that inhibition of wound-induced wnt1 by follistatin initiates the missing tissue response or that follistatin separately regulates wnt1 expression and the missing tissue response. To distinguish between these two possibilities, we tested whether wnt1 inhibition suppresses the missing tissue response defect in follistatin RNAi tail fragments. wnt1; follistatin double-RNAi tail fragments still showed a diminished second mitotic peak at 48 hr post amputation (Figure 6B) and reduced cell death at 72 hr after injury (Figure 6C), suggesting that the missing tissue response defect of follistatin RNAi animals had not been suppressed by wnt1 RNAi. wnt1; follistatin double-RNAi tail fragments initiated anterior pole formation at 72 hr post injury, suggesting these animals could re-set positional information (Figure S7A). Furthermore, regeneration of multiple differentiated tissues, including the eyes, brain, and pharynx, occurred in these fragments, but slowly compared to wnt1; control RNAi and control RNAi tails (Figures 6D, 6E, and S7B–S7D). These findings indicate that head regeneration in the posterior does not require a normal missing tissue response, and suggest that follistatin separately regulates wound-induced Wnt signaling and the missing tissue response (Figure 7).

Figure 7. Model for the Roles of the Missing Tissue Response and follistatin in Regeneration.

(A) Model for the role of follistatin in planarian regeneration.

(B) Schematic depicting head regeneration after follistatin RNAi in high- (posterior) and low- (anterior) Wnt signaling environments. Head regeneration can occur in a low-Wnt signaling environment even in the absence of a detectable missing tissue response, but at reduced speed. Amputation in a high-Wnt signaling environment requires appropriate regulation of wound-induced Wnt signaling. Elevated wound-induced wnt1 expression after follistatin RNAi causes regeneration of anterior patterning information and anterior tissue formation to fail.

DISCUSSION

Identifying the cellular and molecular processes that are required for regeneration is a fundamental problem in understanding adult wound repair. In many regenerative organisms a large array of cellular responses are triggered at major injuries, prominently including elevated cell proliferation (Poleo et al., 2001; Nechiporukand Keating, 2002; Chera et al., 2009; Poss, 2010; Tanaka and Reddien, 2011; Srivastava et al., 2014). For instance, similar to the case of planarian regeneration, limb regeneration in axolotls involves phases of gene expression and sustained cell proliferation specifically associated with blastema formation after amputation and absent in lateral wounds that do not remove substantial tissue (Knapp et al., 2013). However, which cellular processes are required for tissue regeneration to occur across animals remains an open question. Our findings suggest that prominent cellular and molecular responses unique to substantial tissue loss that are hallmarks of regeneration in planarians, collectively called the missing tissue or regenerative response, are not required for regeneration. We determined that regeneration from all injury types tested can occur with a defective missing tissue response, although at a significantly reduced speed. We propose that the missing tissue response functions to accelerate, rather than to bring about, regeneration (Figure 7).

This leads to the question of what is fundamentally required to bring about regeneration. All new cell production in planarian regeneration requires neoblasts, and neoblasts constantly divide and produce fate-specified progenitors during tissue turnover (Reddien, 2013; Zhu and Pearson, 2016). Neoblasts are therefore required for regeneration (Bardeen and Baetjer, 1904; Dubois, 1949; Reddien et al., 2005; Baguñà, 2012). Data suggest that re-setting of positional information is a critical component of regeneration. Regeneration of the AP axis after amputation begins with the regulation of Wnt signaling by generic wound signaling, followed by anterior pole formation, re-establishment of positional information, and tissue replacement. We found that follistatin affects the earliest of these steps through inhibition of the level of wound-induced wnt1 expression. This increase in wnt1 expression at wounds was associated with failure to re-set positional information and an inability to regenerate anterior tissue at amputations in the posterior, where homeostatic Wnt signaling levels are high. Tail fragments that fail to regenerate a head continue to turn over existing tissue, similar to the phenotype of myoD RNAi animals, suggesting follistatin RNAi amputated tail fragments have an inability to generate any tissue that is missing (Gavinño et al., 2013; Scimone et al., 2017). Blocking generic wound-induced gene expression with an inhibitor to Erk signaling also blocks regeneration (Owlarn et al., 2017). These data suggest generic wound signaling, prominently involving wnt1 and notum, is required for re-setting positional information and that this can be required for regeneration. By contrast, all other wound types from which follistatin RNAi animals could regenerate were able to re-pattern tissue after injury but did so slowly. In addition, inhibition of wnt1 was sufficient to rescue patterning and tissue regeneration in follistatin RNAi tail fragments. These data suggest that the re-setting of positional information by wound regulation of Wnt signaling could mediate regeneration even without a detectable missing tissue response.

In some conditions, such as following RNAi of patched, notum, or APC, increased Wnt signaling at anterior-facing wounds can result in ectopic tail regeneration rather than failed regeneration (Gurley et al., 2008; Rink et al., 2009; Yazawa et al., 2009; Petersen and Reddien, 2011). In the case of follistatin RNAi, the significantly reduced missing tissue response at wounds is associated with a slow rate of positional information re-setting. This might not allow enough time for induction of an ectopic tail program while wnt1 is overexpressed. follistatin RNAi also differs from these other conditions in that increased wnt1 expression occurs only between 6 and 12 hours post amputation, whereas it persists up to at least 24 hours post amputation after patched RNAi (Rink et al., 2009). Because these follistatin RNAi anterior-facing wounds have normal expression levels of wound-induced notum up to 24 hours post amputation, this could make ectopic tail formation unlikely at these injuries.

Planarians can regenerate from small injuries, such as eye resection, as an emergent property of homeostatic tissue turnover without elevated cell proliferation (LoCascio et al., 2017). However, major injuries pose additional challenges–in addition to the loss of organs, organ-specific progenitors and the positional information for specifying those progenitors can be lost. A reasonable assumption has been that responses such as increased cell proliferation during blastema formation play a key role in overcoming these challenges. Our findings suggest that this is not the case in planarians. We propose a model in which regenerative re-patterning on the AP axis, induced by regulation of generic wound-induced Wnt signaling, coupled with continuous production of progenitors associated with tissue turnover, can bring about regeneration from diverse wound scenarios. The generic wound response (the follistatin-independent response occurring 1 to 6 hr post injury) has been implicated in initiating re-establishment of positional information and regeneration (Petersen and Reddien, 2009, 2011; Owlarn et al., 2017; Scimone et al., 2017). Aspects of the generic wound response could prove to have additional contributions to regeneration. Furthermore, there could prove to exist specific responses to missing tissue injuries that are follistatin independent and that contribute to the capacity for regeneration. These will be important targets for continued investigation. Taken together, our results suggest that in the absence of detectable cellular and molecular responses specific to major injuries, regulation of wound-induced Wnt signaling for regenerative re-patterning together with continuous tissue turnover can mediate successful regeneration in essentially any wound context (Figure 7).

STAR★METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Dr. Peter Reddien (reddien@wi.mit.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Asexual Schmidtea mediterranea strain animals (CIW4) were cultured in 1x Montjuic planarian water at 20° (Sánchez Alvarado et al., 2002). Animals were starved 1-2 weeks prior to experiments.

METHOD DETAILS

Double-stranded RNA synthesis for RNAi experiments

dsRNA was prepared from in vitro transcription reactions (Promega) using PCR-generated forward and reverse templates with flanking T7 promoters (TAATACGACTCACTATAGGG). Each template (16 μl) was mixed with 1.6 μl of 100 mM rNTPs (Promega); 0.6 μl of 1M dithiothreitol (DTT; Promega); 4 μl of T7 polymerase; and 24 μl of 5x Transcription optimized buffer (Promega). Reactions were incubated for 4h at 37°C. RNA was purified by ethanol precipitation, and re-suspended in a final volume of 30 μl milliQ H2O. Forward and reverse strands were combined and annealed by heating at 56°C followed by cooling to 37°C. Animals were starved for 1-2 weeks prior to first RNAi feeding and were fed twice a week. RNAi food mixture was prepared using 12 μl dsRNA for 30 μl planarian food (homogenized beef liver) (Rouhana et al., 2013). For RNAi of two or more genes, dsRNA for each gene was diluted in half (Scimone et al., 2016). For β-catenin-1 double-RNAi experiments, animals were fed once with β-catenin-1 or control dsRNA at the end of the feeding regimen. C. elegans unc-22 was used as the control condition (Benian et al., 1989).

Double-stranded RNA injections for RNAi

dsRNA injections were performed using a Drummond Nanoject II Auto-nanoliter injector. Needles were pulled from Borosilicate capillaries (#BF100-78-15) on a Sutter Model P-2000 micropipette puller. Animals were injected 2-3 times with 32.9nl of dsRNA per injection.

Microsurgery

Head wedges were made by performing cuts from the medial point of the eye to the base of the auricle and the top of the head. Sagittal fragments were made with a single sagittal cut, immediately lateral to the pharynx. Pharynx resection was performed by creating a longitudinal dorsal incision followed by pharynx extraction. For serial cuts along the AP axis, AP cut 1 was made immediately posterior to the auricles, AP4 was made immediately anterior to the pharynx.

Fixation

Animals were killed in 5% NAC in PBS for 5 minutes before fixation in 4% formaldehyde for 15 minutes. Fixative was removed and worms were rinsed 2X with PBSTx (PBS + 0.1% Triton X-100). Animals were dehydrated and stored in methanol at −20°C. For Anti-phospho-Histone H3 labeling and TUNEL labeling, animals were not dehydrated after fixation (King and Newmark, 2013).

Whole-mount in situ hybridizations

RNA probes were synthesized as described previously (Pearson et al., 2009). Fluorescence in situ hybridizations (FISH) were performed as previously described (King and Newmark, 2013) with minor modifications. Briefly, fixed animals were bleached, rehydrated and treated with proteinase K (2 μg/ml) in 1xPBSTx. Following overnight hybridizations, samples were washed twice in pre-hyb solution, 1:1 pre-hyb-2X SSC, 2X SSC, 0.2X SSC, PBSTx. Subsequently, blocking was performed in 0.5% Roche Western Blocking reagent and 5% inactivated horse serum in 1xPBSTx. Animals were incubated in antibody overnight at 4°C. Post-antibody washes and tyramide development were performed as described (King and Newmark, 2013). Peroxidase inactivation was done in 1% sodium azide for 90 minutes at RT. Specimens were counterstained with DAPI overnight (Sigma, 1 μg/ml in PBSTx).

phospho-Histone H3 labeling

Fixed animals were bleached overnight at room temperature in H₂O₂ (Sigma, 6% in 1xPBSTx). Bleached animals were permeabilized in Proteinase K solution (2 μg/ml in 1xPBSTx with 0.1% SDS) and post-fixed in formaldehyde (4% in 1xPBSTx). Animals were placed in anti-phospho-Histone H3 antibody (Millipore 05-817R-I, clone 63-1C-8) overnight at room temperature at a concentration of 1:300 in 5% inactivated horse serum. Samples were washed with PBSTx, then placed in goat anti-rabbit antibody (ThermoFisher 65-6120) overnight at room temperature at 1:300 in 5% inactivated horse serum. After PBSTx washes, samples were developed in fluorescein tyramide (1:3000 in PBSTx, with 0.003% H₂O₂) for 10 minutes at room temperature. Samples were washed in PBSTx and labeled with DAPI (Sigma, 1 μg/ml in PBSTx) before mounting.

TUNEL

TUNEL was performed using reagents from the ApopTag Red In Situ Apoptosis Detection Kit (Millipore, #S7165). Fixed animals were bleached overnight at room temperature in H₂O₂ (Sigma, 6% in PBSTx). Bleached animals were permeabilized in Proteinase K solution (2 μg/ml in PBSTx with 0.1% SDS) and post-fixed in formaldehyde (4% in PBSTx). Samples were transferred to 1.5mL micro centrifuge tubes. PBSTx was replaced with 20 μL reaction mix (3 parts ApopTag TdT enzyme mix, 7 parts ApopTag reaction buffer), and incubated overnight at 37°C. Animals were washed in PBSTx followed by development in 20 μL development solution (1 part blocking solution, 1 part ApopTag anti-digoxigenin rhodamine conjugate), and incubated in the dark at room temperature overnight. Samples were washed in PBSTx and counterstained with DAPI (Sigma, 1 μg/ml in PBSTx). TUNEL protocol was adapted from previous work (Pellettieri et al., 2010).

Quantitative real-time PCR (qRT-PCR)

Three to five animals were collected per biological replicate with three biological replicates per condition. Total RNA was isolated in 1mL Trizol (Life Technologies) as per manufacturer’s instructions. Samples were triturated using a P1000 tip to homogenize tissue. Following RNA purification and resuspension in MilliQ H₂O, concentrations for each sample were determined using the Qubit RNA HS Assay Kit (Life Technologies). 1 μg RNA input was used to prepare cDNA with the SuperScript III Reverse Transcriptase kit (Invitrogen). Ct values from three technical replicates were averaged and normalized by the Ct value of the housekeeping gene g6pd to generate ΔCt values. Relative expression levels were determined by the -ΔΔCt method by calculating the difference from the average ΔCt value of control RNAi replicates. Bar graphs show relative expression values as 2^−ΔΔCT with standard deviation and individual expression values. Primer pairs used are provided in Table S2.

RNA sequencing Analysis

RNAseq data for follistatin RNAi tail regeneration was analyzed from previously published experiments (Scimone et al., 2017) (GEO: GSE99067). Reads were mapped to the dd_Smed_v6 transcriptome (Brandl et al., 2016) (http://planmine.mpi-cbg.de/planmine/begin.do) using bowtie-1 (Langmead et al., 2009). Raw read counts were subjected to independent filtering with the filter criterion, overall sum of counts, to remove genes in the lowest 40% quantile. Differential expression analysis was performed using DEseq (Anders and Huber, 2010). Heatmaps were generated using pheatmap and are displayed as log₂Fold change values. Genes without a best BLAST hit are identified by transcriptome ID (dd_xxx) in heatmaps. Significance is reported as p_adj values, with p_adj < 0.05 used as a cutoff.

Microscopy and image analysis

Fluorescent images were taken with a Zeiss LSM700 Confocal Microscope. All images are Maximum intensity projections. Images of samples that did not fit the microscope field of view were obtained using the tile scan function (Zeiss ZEN) that aligns and stitches tiles to obtain a single image of a large specimen. Images were processed using ImageJ (Fiji) (Schindelin et al., 2012). Light images were taken with a Zeiss Discovery Microscope. Cell counting was performed manually after blinding control and experimental conditions.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were performed using the Prism software package (GraphPad Inc., La Jolla, CA). Comparisons between the means of two populations were done by a Student’s t test. Comparisons of means between multiple populations were done by one-way ANOVA. Comparisons between means of time-points in a time course was done by two-way ANOVA. Significance was defined as p < 0.05. Statistical tests, significance, data points, error bars and animal numbers (n) for each figure are provided in the legends.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-digoxigenin-POD, Fab fragments	Roche	Cat# 11 207 733 910; RRID: AB_514500
anti-fluorescein-POD, Fab fragments	Roche	Cat# 11 426 346 910; RRID: AB_840257
anti-DNP-HRP conjugate	Perkin-Elmer	Cat# FP1129; RRID: AB_2629439
Anti-phospho-Histone H3 (Ser10) Antibody, clone 63-1C-8	Millipore	Cat# 05-817R; RRID:AB_11215621
Goat anti-rabbit antibody	ThermoFisher	Cat# 65-6120; RRID:AB_2533967
Chemicals
TRIzol	Life Technologies	Cat#15596018
Western Blocking Reagent	Roche	Cat#11921673001
Critical Commercial Assays
ApopTag Red In Situ Apoptosis Detection Kit	Millipore	Cat# S7165
Deposited Data
follistatin RNAi RNaseq	Scimone et al., 2017	GEO: GSE99067
Smed_dd_v6 transcriptome	Brandl et al., 2016	http://planmine.mpi-cbg.de/planmine/begin.do
Experimental Models: Organisms/Strains
Schmidtea mediterranea, clonal strain CIW4, asexual	Laboratory of Peter Reddien	N/A
Oligonucleotides
Sequences used for all FISH probes and dsRNA provided in Table S1	N/A	N/A
Primers used for qRT-PCR are provided in Table S2	N/A	N/A
Software and Algorithms
ImageJ (FIJI)	Schindelin et al., 2012	https://fiji.sc
ZEN digital imaging software	Zeiss	https://www.zeiss.com/microscopy/us/products/microscope-software/zen.html
GraphPad Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
R v.3.2.3	The R Foundation	https://www.r-project.org/
bowtie v1.1.2	Langmead et al., 2009	http://bowtie-bio.sourceforge.net/index.shtml
DESeq	Anders and Huber, 2010	https://bioconductor.org/packages/release/bioc/html/DESeq.html

Highlights.

• follistatin is required for the missing tissue response at many injuries

• The missing tissue response is not required for regeneration

• Regeneration occurs slowly in the absence of a missing tissue response

• follistatin inhibits wnt1 at wounds to mediate its role in head regeneration