Regenerative responses following DNA damage – β-catenin mediates head regrowth in the planarian Schmidtea mediterranea

Stem cells drive the development and tissue renewal of all multicellular organisms. Characterizing their key properties is crucial for advancing regenerative medicine, and for a better understanding of misregulation that underpins diseases, such as tumorigenesis. Although stem cells are believed to have enhanced stress responses, ranging from upregulated DNA repair to rapid apoptosis (Dannenmann et al., 2015; Liu et al., 2013a; Maynard et al., 2008; Vahidi Ferdousi et al., 2014; Vitale et al., 2017), their molecular behaviour also depends strongly on the tissue environment in which they reside (Blanpain et al., 2004; Chacón-Martínez et al., 2018; DeGregori, 2017; Rompolas et al., 2013). Changes in the tissue landscape, for example during ageing, provide a permissive environment for malignant growth, while, in contrast, embryonic niches or specific micro-environmental cues are able to remodel and exert control over teratomas to produce normal tissue (Bianchi-Frias et al., 2019; Booth et al., 2011; Hendrix et al., 2007; Lan et al., 2019; McCullough et al., 1994; Mintz and Illmensee, 1975; Stoker et al., 1990).

The molecular pathways that regulate cell growth, morphogenesis and differentiation are key components of regeneration and tissue homeostasis but can also lead to tumorigenesis. During regeneration, strict control of these mechanisms leads to an orchestrated tissue regrowth rather than malignant transformation. Although there is a contrasting outcome for both processes, gene expression patterns during, for example, renal regeneration versus renal carcinoma are regulated concordantly for 77% of the studied genes (Riss et al., 2006). Another example of the commonality between both processes is the Wnt pathway, which is intimately linked with stem cell fate throughout development and regeneration but is also misregulated in multiple cancer types (Beachy et al., 2004; Clevers, 2006; Haegel et al., 1995; Heasman et al., 1994; Morin et al., 1997). A deeper understanding of the underlying mechanisms that protect stem cells against various types of cellular stress and the parameters that influence the switch between a healthy (regeneration) and malignant (cancer) cell fate is crucial for a myriad of stem cell applications.

Not coincidentally, tissues and species with a high regenerative ability, such as planarians, seem to have a lower predisposition towards chemically induced tumorigenesis (Enomoto and Farber, 1982; Farber, 1984; Okamoto, 1997; Oviedo and Beane, 2009; Pomerantz and Blau, 2013; Zilakos et al., 1996). While tumours have been reported in planarians, their incidence is low (Foster, 1963; Hall et al., 1986a,b; Plusquin et al., 2012; Schaeffer et al., 1991; Stevens et al., 2018; Van Roten et al., 2018). Planarians are known for their full-body regeneration, enabled by a large pool of adult pluripotent stem cells, called neoblasts. Neoblasts can provoke whole-body regeneration from a small fragment by altering proliferation, migration, differentiation and apoptosis patterns (Abnave et al., 2017; Pellettieri et al., 2010; Wenemoser and Reddien, 2010). Despite the high division rate of these stem cells, which they experience during regeneration as well as homeostasis, the incidence of malignancies is low and coincides with the absence of ageing (Oviedo and Beane, 2009; Plusquin et al., 2012; Sahu et al., 2017; Stevens et al., 2018; Tan et al., 2012). Whether this can be solely attributed to the abundantly present stem cells remains to be elucidated, as is the question of how signals from the surrounding niche are involved. With proven conservation of many signalling pathways, planarians are well-suited to the unravelling of stem cell responses within the complex multicellular entity of a regenerating tissue (Labbé et al., 2012; Onal et al., 2012; Resch et al., 2012; Swapna et al., 2018). Several studies in planarians have reported the formation of outgrowths or hyperproliferation by interfering with homologues of known mammalian tumour suppressors, such as the damage transducer SMG-1, the epigenetic regulator MLL3 and MLL4, the tumour suppressors PTEN and p53, and the Hippo pathway (de Sousa et al., 2018; González-Estévez et al., 2012; Mihaylova et al., 2018; Oviedo et al., 2008; Pearson and Sánchez Alvarado, 2010). As in other systems, neoblasts receive guiding instructions from their surroundings, such as the musculature, extracellular matrix and the nervous system (Cebria and Newmark, 2007; Fraguas et al., 2014; Isolani et al., 2013; Oviedo et al., 2010; Witchley et al., 2013). Recent studies have indicated that neoblasts respond differently to DNA damage depending on their location along the body axis (Peiris et al., 2016) and that interfering with the extracellular matrix can influence the development of outgrowths (Van Roten et al., 2018; Voura et al., 2017).

We have previously found that the planarian stem cell system reacts to genotoxic exposure by ceasing proliferation and activates different defence mechanisms depending on the developmental stage of the animal (Stevens et al., 2017, 2018). Here, we aim to further characterize how regenerative tissues respond to a genotoxic insult and whether tissue-associated factors influence the outcome. Adult worms were exposed to the genotoxic agent methyl methanesulfonate (MMS), subjected to amputation and monitored during regeneration. Our data indicate that DNA-damage-induced stem cell responses and the regenerative success are tissue-dependent and are affected by interfering with the body-axis-associated Wnt gradient.

In the current study, we aimed to determine stem cell stress responses and the associated physiological outcomes of a developing tissue after inducing DNA damage. Adult (i.e. uninjured) planarians were exposed to the DNA-alkylating compound MMS to induce DNA damage. After this period, damage responses were evaluated, and a regenerative response was provoked by amputation. Within this regenerative tissue, the progression of DNA damage, stem cell responses and regeneration efficiency were monitored over time (Fig. 1). By comparing responses in anterior and posterior body parts, and by interfering with the anterior–posterior Wnt gradient, the involvement of tissue-related factors in DNA damage responses was further investigated.

Adult worms were exposed for 7 days to 50 µM MMS, a sublethal concentration that evokes DNA damage in our model (Stevens et al., 2017, 2018). In line with previous measurements, we observed an increased amount of DNA damage (P<0.001) together with an increased amount of apoptotic cells (P<0.001), decreased stem cell proliferation [as determined by staining for phosphorylated histone H3 (H3P⁺ cells), P<0.001], a decrease in the expression of the general stem cell marker smedwi-1 (smedwi-1 intensity, P<0.05) and a decrease in the expression of the early-progeny marker NB.21.11e (P<0.01) (Fig. 2; Fig. S1) (Stevens et al., 2018). All parameters were evaluated in different parts of the animals to screen for tissue-dependent effects; a comparison between anterior (prepharyngeal) and posterior (postpharyngeal) regions did not reveal significant differences (P>0.05).

The regenerative capacity of the exposed animals was evaluated by monitoring early tissue formation (blastema size) and regenerative success at 14 days post amputation (dpa) (Fig. 3A–C). After cutting the animal in front of the pharynx, creating an anterior (head) and posterior (tail) piece, blastema growth was most affected in the tail (Fig. 3B). While both fragment types had a significantly smaller blastema at 4 dpa (P<0.001), tail fragments failed to regenerate and a protruding blastema was never observed. The blastemas of the head fragments kept growing, yet remained proportionally smaller than those in the non-exposed animals at each time point (P<0.001). The difference in blastema development between head and tail fragments was reflected in their final regenerative success at 14 dpa (Fig. 3C). This was scored as complete, aberrant or absent, based on the reappearance of missing body structures. Most heads regenerated all missing body parts, bearing a pharynx and tail, completely (63%, 26/41 animals). Manifested aberrations were the reappearance of a pharynx but not a tail (7/41) or a ‘notched’ tail (7/41). One animal did not regenerate. In contrast, only 19% of the tails regenerated a complete head with two eyes (8/43), 7% regenerated incompletely [a head with one centrally located eye (2/43) or no eyes (1/43)] and the majority of the tails did not regenerate at all (74%, 32/43). The body edges of the latter were smoothly aligned, indicating that wound closure took place but did not evolve further to reconstruct missing body parts. In accordance with the phenotypes, labelling the nervous system showed either connected nerve cords in the tail, a notched tail or no further development of nerve cords behind the pharynx (Fig. 3D). In tail fragments, nerve cords looped around the pharynx without the formation of cephalic ganglia (18/20 animals). The anterior marker notum (Petersen and Reddien, 2011) was not expressed at 3 dpa in recovering tail fragments (6/7 animals) (Fig. 3E). The posterior marker smed-wnt-1 (Adell et al., 2009; Gurley et al., 2010; Petersen and Reddien, 2009) was present in the regenerating head fragments, but the signal was diminished as compared to that in control animals (Fig. 3E). An additional experiment suggested that the location of the cut along the anterior­–posterior axis influences the final outcome, both for anterior and posterior fragments. The ability to regenerate a head increased when the cut was made more anteriorly. Likewise, the ability to regenerate a tail also increased the more posteriorly the cut was made (Fig. S2).

Anterior–posterior differences were again confirmed when excising small fragments of anterior, central (trunk) or posterior parts of the planarian body, with a decreasing regenerative success rate towards the posterior part of the animal (Fig. 3F). In addition, anterior-facing wounds had a lower success rate in all fragment types as compared to posterior-facing wounds. The highest regenerative success was observed in anterior fragments, which regenerated a complete head in 46% of the samples and a complete tail in 69% of the samples. The remaining samples regenerated aberrantly at both the anterior wound side (a head with one centrally located eye for 4/13 animals or an eyeless head for 1/13), and at the posterior-facing wound (notched-tail, 2/13). The posterior-facing wound of the trunks either regenerated completely (73%, 19/26 animals) or with a notched-tail (23%, 6/26); one fragment did not regenerate. Anterior regeneration was drastically lowered in these trunks (8% complete success, 2/26 animals) and in most cases was even absent (85%, 22/26). The overall success rate decreased further in the small posteriorly located fragments. While anterior regeneration was only successful in 1/12 of these posterior fragments, 6/12 were able to regenerate a tail. The mortality rate was higher in the smaller fragments (2/13 anterior fragments; 3/12 posterior), which is possibly related to their higher fragility during handling. Exposure to another alkylating compound (4-nitroquinoline 1-oxide) impaired regeneration in both head and tail fragments (Fig. S3).

Next, we sought to characterize underlying stem cell responses leading to the observed regenerative success or failure. We assessed how the induced genotoxic effects in adult animals (Fig. 2) affected regenerative stem cell dynamics over time (Figs 4 and 5). During normal regeneration, an initial body-wide proliferation peak is followed by a local peak near the wound site at 2–3 dpa (Wenemoser and Reddien, 2010). The latter was chosen as a first time point to assess stem cell dynamics (Fig. 4). In both heads and tails, the exposure-induced DNA damage was still present at 3 dpa (P<0.001 and P<0.05, respectively) and was concomitant with a significant decrease in the expression of a general stem cell marker (smedwi-1 intensity) at 2 dpa as compared to control animals, which was stronger in exposed tails than heads (P<0.001 versus P<0.05, Fig. 4A,B). The number of proliferative cells (H3P⁺) in exposed head fragments was similar to that in controls (no significant difference), while their early progeny (NB.21.11e⁺) was significantly lower (P<0.001) (Fig. 4C,D). In exposed tail fragments, both proliferation (P<0.001) and early differentiation (P<0.001) were significantly decreased as compared to non-exposed animals. Both proliferation and differentiation were significantly higher in exposed heads than in exposed tails (P<0.01 and P<0.05, respectively) (Fig. 4C,D).

At 7 dpa, no significant differences in the measured parameters were observed for exposed heads (Fig. 5A–D). In exposed tails, only the number of early progeny cells was significantly decreased (P<0.05). Since proliferation and differentiation did increase in impaired tail fragments over time, although slower than in head fragments (Fig. S5F), we assessed whether these increases could support a regenerative response at a later time point. Re-cutting impaired tail fragments at 7 dpa rescued regeneration completely (Fig. S4).

When animals were continuously exposed to MMS (i.e. before and after amputation), regenerative success again decreased in tails but not in heads (Fig. S5C). Stem cell proliferation in continuously exposed heads remained significantly higher than in continuously exposed tails both in the short term (2 dpa, P<0.05) and long term (7 dpa, P<0.01) (Fig. S5D,E). Both continuously exposed heads and tails had a significantly lower amount of proliferation and differentiation compared to their respective controls 2 dpa (P<0.001, Fig. S5D). While differentiation remained significantly decreased in both fragment types at 7 dpa (P<0.001 and P<0.01, respectively), only continuously exposed heads showed a similar proliferation level as controls at that time point (no significant difference).

Independently of the treatment (before±after amputation), head and tail fragments showed increasing patterns in proliferation and differentiation over time, but both parameters increased faster in heads than tails (Fig. S5F). While proliferative responses in treated animals eventually equal the levels measured in non-exposed animals, this effect became apparent in heads at an earlier time point. Similarly, proliferation increased faster in fragments that were not continuously exposed. Control levels of early differentiation, on the other hand, were never achieved in continuously exposed animals.

A long-term follow-up experiment (up to 21 dpa) was performed to estimate the occurrence of malignant development. Abnormalities were only observed in animals that were continuously exposed to MMS before and after amputation (3% for all conditions combined; n=156) (Fig. S5B). Their incidence in regenerating anterior and posterior fragments was equal. Aberrancies manifested themselves as blister-like structures on the dorsal side of the animal but were not further characterized because of their low occurrence and fragility.

In search for factors responsible for blocking or activating regeneration specifically in posterior fragments, we interfered with the Wnt gradient, which is known to determine and maintain the anteroposterior axis during homeostasis and regeneration (Gurley et al., 2010; Petersen and Reddien, 2011, 2009). An increasing gradient of β-catenin-1 activity towards the posterior end determines posterior identity, while knockdown of β-catenin-1 induces the formation of a head structure, also in posteriorly facing wounds (Gurley et al., 2008; Iglesias et al., 2008; Petersen and Reddien, 2008). The failure to regenerate a head structure in exposed tails was rescued by β-catenin-1 knockdown (β-cat-1 KD), both phenotypically (54% developed completely, 37% aberrantly) and at the level of brain development (16/17 animals developed cephalic ganglia) (Fig. 6A,B). Eyes were nonetheless under-developed as compared to those in non-exposed animals, and their regenerated brains were significantly smaller (P<0.001). Also under continuous MMS exposure, β-cat-1-KD tails had an increased regenerative capacity (Fig. S6A). A similar rescue effect was observed after knockdown of teashirt, a known downstream target of β-catenin-1 (Reuter et al., 2015), confirming the involvement of Wnt signalling (Fig. 6C). When damage was induced via irradiation, a known and well-studied phenomenon in planarians (Wagner et al., 2011, 2012, Lei et al., 2016, Shiroor et al., 2019 preprint), β-cat-1 KD also ameliorated the effects of irradiation (2/3 animals), although regeneration was not rescued (Fig. S6B).

To assess the rescue effects of β-cat-1 KD on the stem cell population, stem cell-related parameters were assessed in regenerating tail fragments at 1 dpa (Fig. 7). The expression of the general stem cell marker smedwi-1 decreased following exposure and visually increased (6/7 animals) due to β-cat-1 KD (Fig. 7A). In addition, stem cell proliferation significantly increased following β-cat-1 KD in control (P<0.05) as well as exposed (P<0.001) animals (Fig. 7B). This stimulating effect on the cell cycle was confirmed by an increased expression of cdc23 (P<0.05 in exposed animals), a positive regulator of the G2/M progression, while the expression of cdc73, a negative regulator of the cell cycle, was not affected by exposure or by β-cat-1 KD (Fig. 7C).

To identify how MMS exposure and β-cat-1 KD affect specific stem cell subpopulations, the expression of neoblast subclass-associated genes was measured (Fig. 7D) (van Wolfswinkel et al., 2014). The expression of the general stem cell marker smedwi-1 significantly decreased in response to MMS treatment (P<0.001) and increased following β-cat-1 KD (P<0.001), confirming the whole-mount observations (Fig. 7A,D). Similarly, the expression of the putative pluripotency marker tgs-1 (Zeng et al., 2018) significantly decreased following MMS exposure and increased following β-cat-1 KD, both in control and MMS-exposed animals (P<0.001). Similar decreases following exposure and increases following β-cat-1 KD were observed for the genes associated with the sigma neoblast subclass, soxP-1 (P<0.001) and soxP-2 (non-significant increasing trend following KD). Also the expression of zeta subclass-associated genes (soxP-3 and zfp-1) decreased following exposure (P<0.001 and P<0.05, respectively), but was not influenced by β-cat-1 KD. The expression of the gamma subclass-associated genes gata4/5/6 and prox1, on the other hand, was unaffected by MMS exposure alone, but decreased in combination with β-cat-1 KD (P<0.05 and non-significant trend, respectively).

By 2 dpa, β-cat-1 KD did not affect stem cell proliferation or the number of early progeny cells (Fig. S7A,B). The amount of DNA damage was unaffected by β-cat-1 KD 3 dpa (Fig. S7C).

The inherent plasticity of stem cells confers a tissue with regenerative potential. However, an uncontrolled execution of this plasticity can cause the onset of diseases, failed wound healing or regenerative impairment (Beachy et al., 2004; Knoepfler, 2009; Ma et al., 2010; Sundaram et al., 2018). Increased lifetime, frequent replication and micro-environmental signals make stem cells vulnerable to both exogenously and endogenously induced DNA damage (Espada and Ermolaeva, 2016; Tao et al., 2015; Tomasetti and Vogelstein, 2015). Regenerative tissues and tissues in early development have the ability to restrict and tightly control cellular functioning, but alternatively can also pave the way for malignancies to arise because of the increased replication and division rates (Enomoto and Farber, 1982; Hendrix et al., 2007; Li and Ye, 2017; Ma et al., 2010; Oviedo and Beane, 2009; Pomerantz and Blau, 2013). In order to use stem cells at their full potential, it is crucial to gain more insight into how newly regenerating tissues respond to inflicted damages, preserve genomic integrity, and guard their cellular and physiological functions. Studying these processes in a regenerative model can reveal key determinants that tip the scale towards successful tissue restoration or lead to abnormal growth. Planarians have the advantage of linking underlying molecular mechanisms and stem cell behaviour to outcomes at the tissue level, such as successful development or impaired regeneration. We used the genotoxic compound MMS to challenge the planarian stem cell system and characterized (1) the regeneration potential upon induced DNA damage, (2) the underlying stem cell dynamics, (3) whether these responses led to adverse outcomes and (4) which factors influenced these outcomes. To achieve this, the regeneration process was induced via amputation after MMS exposure; stem cell responses were then monitored in the regenerating fragments over time (Fig. 1).

The alkylating agent MMS irreversibly binds to DNA strands, stalls the replication fork and disturbs normal DNA replication (Lundin et al., 2005). With stem cells being the only dividing cells in planarians, stem cell cycling was affected by the increased amount of DNA damage in our set-up (Fig. 2A), probably by halting them in the S-phase (Stevens et al., 2018). As a consequence, the amount of proliferating and descending progeny cells decreased (Fig. 2C,D). The overall decrease in the expression of the general stem cell marker smedwi-1 could be due to damage-induced cell death, although the detected increases in apoptosis can also take place in differentiated cells (Fig. 2B; Fig. S1). The observed alterations in stem cell dynamics did not result in severe phenotypic effects in adult animals. As far as could be determined, the amount of DNA damage was equal in both the anterior and posterior regions of the adult animal (Fig. 2A), as was previously also observed after rad51 and ubc9 knockdown (Peiris et al., 2016; Thiruvalluvan et al., 2018). Following amputation, the amount of DNA breaks decreased in both heads and tails after 3 days, and DNA damage was completely absent at 7 dpa (Figs 2A, 4A and 5A). This corresponds to a previous study in Dugesia schubarti describing a similar decrease in, but not the absence of, DNA breaks 24 h after an acute MMS exposure in adult animals (Guecheva et al., 2001). Also after sublethal irradiation, smedwi-1-positive cells repopulate and regain their mitotic activity within a similar timespan (Lei et al., 2016; Salvetti et al., 2009; Wagner et al., 2012, 2011).

In contrast to what was observed after rad51 KD, we did not find regional differences in proliferation or differentiation responses upon MMS-induced DNA damage in adult animals (Fig. 2) (Peiris et al., 2016). However, when challenged by tissue loss, the regenerative outcome differed along the body axis as only heads were able to regenerate (Fig. 3C). We did notice a delay in the regenerative responses of these heads, exemplified by the restricted blastema size, decreased wnt-1 expression and an initially decreased number of early progeny cells 2 dpa (Figs 3B,E and 4D). In line with the delayed, but successful, regenerative capacities, proliferative patterns were compensated at 2 dpa, followed by an increase in early differentiation at 7 dpa (Figs 4C and 5D). In tails, on the other hand, the increasing numbers of proliferating and differentiating stem cells during regeneration were insufficient to induce the formation of a blastema (Figs 3B,C, 4C,D and 5C,D; Fig. S5F). The absence of the anterior marker notum (Petersen and Reddien, 2011) indicates that the anterior–posterior decision does not take place in impaired tails and regenerative responses are already affected at an early time point (Fig. 3E). In general, the numbers of proliferating and differentiating cells were significantly higher in exposed heads than tails, at early and later stages of regeneration, and this was also the case when exposure continued after amputation (Figs 4C,D and 5C,D; Fig. S5D,E). This indicates that, although amputation-associated stem cell responses remain active following genotoxic stress in both fragment types, their outcome depends on the tissue location (Fig. 3C; Fig. S5F).

Anterior-located amputation sites have a faster rate of head regeneration in a number of planarian species and differences in mitotic activity between body regions were described previously (Baguñà, 1976; Oviedo and Levin, 2007; Reddien and Sánchez Alvarado, 2004). Accordingly, in S. mediterranea, higher proliferative capacities and a higher resistance to cell death in the anterior body region have been found following rad51 KD-induced DNA damage (Peiris et al., 2016), which is in line with our observed differences between head and tail fragments, specifically affecting anterior regeneration of posterior body parts (Figs 3–5; Fig. S5). We did not observe regional differences in stem cell responses in adult animals before amputation (Fig. 2), suggesting that regeneration-associated signals can differentially influence stem cell fate following MMS exposure (Stevens et al., 2018). To date, several (positional) cues that influence stem cell fate have been identified in planarians, with muscle cells acting as an important signalling centre for positional information (Fraguas et al., 2014; Miller and Newmark, 2012; Pirotte et al., 2015; Rossi et al., 2012; Scimone et al., 2017; Seebeck et al., 2017; Witchley et al., 2013). Neuronal signalling and the presence of the brain are often associated with a stimulating effect on stem cells (Cebria and Newmark, 2007; Peiris et al., 2016; Rossi et al., 2012), but we found no significant differences in regenerative success rates in smaller anterior fragments without a brain (Fig. 3F).

The Wnt pathway is one of the most important positional cues and is present in a gradient along the planarian anterior–posterior axis, decreasing its activity from tail to head and determining posterior identity (Adell et al., 2009; Gurley et al., 2008; Iglesias et al., 2008; Petersen and Reddien, 2008, 2009; Sureda-Gómez et al., 2016). The dual role of β-catenin is segregated in two distinct genes in planarians, and β-catenin-1 solely functions as a positive mediator of Wnt signalling (Chai et al., 2010; Iglesias et al., 2008). By downregulating β-catenin-1, or its downstream target teashirt, we were able to induce regeneration in the impaired tail fragments (Fig. 6). This is in line with the previously reported rescue of head regeneration by β-catenin-1 KD in Dendrocoelum lacteum, Procotyla fluviatilis and Phagocata kawakatsui, all species that are normally unable to regenerate anteriorly in posterior body fragments (Liu et al., 2013b; Sikes and Newmark, 2013; Umesono et al., 2013). Similar to what was found in D. lacteum and P. fluviatilis, we did not observe the presence of the anterior marker smed-notum in impaired (non-KD) tail fragments (Fig. 3E). In contrast to a normal proliferative response in impaired D. lacteum and P. fluviatilis wild-type tails, we did detect a diminished proliferative response and the absence of blastema growth, indicating that the MMS exposure also affects other processes than establishing anterior or posterior fates (Figs 3B and 4C). Wnt signalling is correlated with stem cell control and with the DNA damage response pathway in other organisms (Clevers, 2006; Karimaian et al., 2017; Reya et al., 2003; Zhang et al., 2011; Zhao et al., 2018). In murine embryonic stem cells, for example, Wnt signalling was upregulated following genotoxic stress in a p53-dependent manner, thereby inhibiting differentiation (Lee et al., 2010). Murine intestinal stem cells are more sensitive to DNA damage when localized in a region with high Wnt activity and as a result apoptotic responses increased (Tao et al., 2015). A transcriptomic study in S. mediterranea found stem cell gene expression to be affected by β-catenin knockdown, suggesting that stem cells respond to differences in β-catenin signalling (Reuter et al., 2015). In accordance with this, we found that the expression of stem cell-associated genes increased in rescued (β-cat-1 KD) tail fragments at an early time point (1 dpa), together with an increased proliferation (Fig. 7A–D). The increased expression of a pluripotency marker (tgs-1) and the sigma subclass-associated SoxP-1, which is necessary for stem cell maintenance, suggest that β-catenin is a regulator of proliferation in tails (Wagner et al., 2012; Zeng et al., 2018). The specific increase of these genes' expression at the expense of the expression of less potent gamma subclass-associated genes is in line with this hypothesis, as it is the sigma subclass that is suggested to be the only self-renewing neoblast subclass (Lai et al., 2017). As we did not detect significant effects on the amount of DNA damage following knockdown and we found similar (although less pronounced) responses of β-cat-1 KD without MMS exposure, it is possible that β-catenin counteracts the effects of DNA damage by increasing proliferation, without having a direct role in DNA damage responses (Fig. S7C). Future research is needed to determine the specific function of β-catenin in response to various types of DNA damage and to reveal how DNA damage affects Wnt signalling. The fact that a downstream target of β-catenin also elicits a similar rescue effect could imply a broader role for Wnt signalling rather than β-catenin being a primary regulator. In addition, other types of DNA damage need to be assessed to fully understand the role of Wnt signalling in these processes. Inducing DNA damage via irradiation suggests that β-cat-1 KD also ameliorates irradiation effects in tail fragments (Fig. S6B). Exposure to another alkylating compound (4NQO), on the other hand, impaired regeneration in both head and tail fragments, possibly related to its stronger proliferation inhibition as compared to MMS (Fig. S3) (Stevens et al., 2017). Further research will thus be needed to determine if and how Wnt signalling functions in response to (specific types of) DNA damage.

While Wnt signalling is also active during homeostasis (Iglesias et al., 2008), we did not observe differential effects in the anterior versus posterior regions of adult (non-KD) animals (Fig. 2). β-catenin might function differently in a regenerative versus adult tissue state, in accordance with the variable role of Wnt signalling in different cell types or tissues (Clevers, 2006).

In conclusion, MMS exposure provokes equal stem cell responses in adult animals along the body axis, while stem cell responses to previously inflicted DNA damage in a regenerative context depend on their body-axis location. Anteriorly localized fragments quickly regain normal stem cell functioning and are capable of restoring missing tissue. Posterior fragments activate stem cell proliferation and differentiation processes to a lesser extent and fail to regenerate. The latter can be improved by β-catenin-1 KD, which activates regeneration in otherwise impaired tail fragments by stimulating stem cell proliferation at an early time point. Future research will focus on further elucidating its role in planarians in a context of DNA damage in planarians.

An asexual line of the planarian Schmidtea mediterranea was cultivated at 20°C in the dark in freshwater medium as described previously by Pirotte et al. (2015). Worms were fed once a week with veal liver, but were not fed during the experiments. As such, a fasting period of at least 7 days prior to measurements was taken into account to avoid food-related effects.

Worms were exposed to 50 µM methyl methanesulphonate (MMS, Sigma-Aldrich, catalogue number 129925) in various experimental set-ups in freshwater medium (1.6 mM NaCl, 1 mM CaCl₂, 1 mM MgSO₄, 0.1 mM MgCl₂, 0.1 mM KCl, 1.2 mM NaHCO₃.). In each case, medium was refreshed twice a week (every 2–3 days) with a freshly prepared MMS working solution, prepared in freshwater medium. Control animals were kept in freshwater medium and were refreshed concomitantly. Exposure was achieved in six-well plates with three or four animals per well in 3 or 4 ml of medium, respectively. Phenotypic follow-up was performed with individual worms in 1 ml of medium in a 24-well plate.

To examine how regenerative tissue recovers after a genotoxic insult, a set-up was used in which adult animals were exposed prior to the induction of regeneration (Fig. 1). Adult worms were exposed to 50 µM MMS for 7 days. The genotoxic effect of MMS was established after this initial exposure period. Regeneration was then induced with a pre-pharyngeal cut to obtain a head and tail piece. The animals were kept in freshwater medium during the regenerative phase; medium was refreshed twice a week (every 2–3 days). Samples for various measurements, as indicated below, were taken just before cutting the animals on day 0 and at 2, 3, 7 and 14 days post amputation (dpa). Alternative cutting strategies, exposures or combination experiments with RNA interference are indicated in the relevant figures and were exposed and refreshed as explained above.

Regenerative success was assessed by scoring the regrowth of missing body structures at 14 dpa. The reappearance of a head with two photoreceptors, a centrally located pharynx and the presence of a tail were visually inspected. Regenerative success was classified into three categories, scoring how well the missing body structures regenerated (i.e. complete, aberrant or absent). Within the considered time period, heads that have regenerated their missing body parts completely (i.e. a pharynx and tail) are scored ‘complete’. In cases where a pharynx was present but the tail did not develop completely, samples were scored ‘aberrant’. In cases where the wound closure did not develop further into the missing body structures, this was scored as ‘absent’. For tails, the regeneration of a head structure with two visible eyes was scored as ‘complete’. In cases where a head with no eyes or only one eye developed, an ‘aberrant’ score was given. If wound closure did not lead to development of the missing body structures, this was considered as ‘absent’. For the experiments where animals were cut differently (anterior–trunk–posterior), the same categories to score the presence of missing body structures were used per wound-type (anterior-facing or posterior-facing). Blastema areas were used as indication of early tissue regeneration and were measured at 4, 7 and 14 dpa. The blastema is an unpigmented region that is formed at the wound site, within which missing distal structures differentiate. The size of the blastema was measured relative to the size of the worm, determined as an average value, which was calculated based on three pictures per worm. Pictures were taken with a Nikon Ds-Ri2 digital camera mounted on a Nikon SMZ800 stereomicroscope (Nikon Instruments Inc.). The number of replicates and experiments is indicated in the respective figures.

The expression of smedwi-1 (general stem cell marker), NB.21.11e (early progeny marker that is committed to the epidermal cell lineage), notum (anterior pole marker) and wnt-1 (posterior pole marker) were determined using whole-mount fluorescent or colorimetric in situ hybridization (ISH) (Adell et al., 2009; Eisenhoffer et al., 2008; Gurley et al., 2010; Petersen and Reddien, 2011, 2009; Reddien et al., 2005). The protocol was performed as described by King and Newmark (2013). For colorimetric ISH, development was performed using NBT/BCIP (Roche) as described by Pirotte et al. (2015). Probes were synthesized using the DIG RNA (SP6/T7) Labelling Kit (Roche) as indicated by the manufacturer, starting from a purified PCR product of the gene of interest. To obtain this, a general PCR was performed with gene-specific primers as follows: smedwi-1 primers forward, 5′-GTGACGCAGAGAAACGGAAG-3′, reverse, 5′-TTGGATTAGCCCCATCTTTG-3′; NB.21.11e primers forward, 5′-GTGATTGCGTTCGCGTATATT-3′, reverse, 5′-ATTTATCCAGCGCGTCATATTC-3′; notum primers forward, 5′-CGAGTGATTTGTGGTCTGG-3′, reverse, 5′-CGTGGAGTCGTTGATTGTTG-3′; and wnt-1 primers forward, 5′-TGAAGGAATCAGAAAGGGTA-3′, reverse: 5′-TTGTTAGGAAAGGTCGGTTG-3′ Samples were mounted with Immu-Mount (Thermo Fisher Scientific) and analysed with a Nikon i80 fluorescence microscope with a Nikon Ds-Ri2 digital camera using the Nikon NIS-Br software (Nikon Instruments Inc.). Because of the dense signal for smedwi-1, the intensity of the signal was determined using the Nikon NIS-Br software (average intensity for the region of interest). The average intensity of the entire worm or a specified region, as indicated in the figures, was used. For NB.21.11e, cells were counted in specified areas. In heads, one region covering the centre of the animal was used. For tails, a prepharyngeal, postpharyngeal and the region at the right side of the pharynx were used to calculate an average value per animal. For adult animals, one region located either pre- or post-pharyngeally was used as indicated in the figure. The number of replicates and independent experiments is indicated in the respective figures.

The mitotic activity of stem cells was determined by immunolabelling phosphorylated histone H3 (Ser10) (H3P; Merck-Millipore, catalogue number 09-797, diluted 1:600), detected by a secondary anti-rabbit-IgG Alexa Fluor 568-conjugated antibody (Thermo Fisher Scientific, catalogue number A-11036, diluted 1:500) as described by Leynen et al. (2019). The total number of positive cells was normalized against the total body size of the animal (determined post mounting) and was counted using the NIS-Br software (Nikon Instruments Inc.). The number of replicates and independent experiments is indicated in the respective figures.

Neuroregeneration was determined by immunolabelling synapsin (mouse anti-SYNORF1, C311, diluted 1:50, Developmental Studies Hybridoma Bank), detected by a secondary goat anti-mouse-IgG Alexa Fluor 488-conjugated antibody (Thermo Fisher Scientific, catalogue number A-1101, diluted 1:400) as described by Leynen et al. (2019). Samples were mounted with Immu-Mount (Thermo Fisher Scientific) with the ventral side upwards. Pictures were captured with a Nikon Ds-Ri2 camera mounted on a Nikon eclipse i80 fluorescence microscope and analysed using the Nikon NIS-Br software (Nikon Instruments Inc.). The presence of nerve cords and cephalic ganglia was scored visually. The relative brain size was measured as the width of the brain relative to the width of the entire head.

The alkaline comet assay was used to analyse the DNA-damaging effect of MMS in variable set-ups. After exposure, a stem-cell-containing fraction was obtained (from one worm per sample) using a dissociation protocol as described previously (Stevens et al., 2018), with some modifications that are mentioned below. After chemical and physical maceration, the samples were serially filtered with a 35 µm filter and a 10 µm filter (Pluriselect) to purify the fraction from larger cell types and debris. The obtained cell pellet was dissolved in 160 µl of 0.8% low-melting-point-agarose (Thermo Fisher Scientific) in phosphate-buffered saline (PBS) and immediately gelled on an agar-coated slide (1% agar in ultrapure water, Invitrogen). Each sample was made in technical duplicates (70 µl per gel). The gels were lysed overnight in lysis buffer (2.5 M NaCl, 0.1 M Na₂EDTA, 10 mM Tris, 1% Triton X-100, 10% DMSO; pH 10) at 4°C. After washing twice with cold PBS, the samples were denatured for 20 min and electrophoresis was carried out during 20 min at 4°C in electrophoresis buffer (300 mM NaOH, 1 mM Na₂EDTA, pH 13.0) at 0.95 V/cm over the platform. The buffer was continuously recirculated. Samples were neutralized first with PBS and subsequently ultrapure water, and then stained with Sybr Gold Nuclear Acid Gel Stain (Invitrogen, catalogue number S11494) for 20 min. After two washing steps with water, samples were dried and scored with a Zeiss fluorescence microscope (AxioImager.Z2, equipped with SlideFeeder X80) using the 10× objective magnification. Images were automatically captured with Metafer 5 (Metasystems). Analysis of ∼100 comets per sample (50 per technical duplicate) was performed with the comet assay IV software (Instem, Perceptive Instruments). Tail intensity (i.e. the percentage of intensity in the tail relative to the total intensity of the comet) was used as output. Four biological replicates per condition with two technical replicates per sample were analysed. A median value per technical duplicate was determined; the average of both technical duplicates was used per sample. The total number of samples and independent experiments is indicated in the respective figures.

To knockdown the expression of β-catenin-1, worms were injected for 3 consecutive days with a gene-specific dsRNA probe. Injection was undertaken prepharyngeally with three injections of 32.2 nl of a 1000 ng/µl dsRNA probe. The day after the final injection, the exposure period of adult worms began. The dsRNA probe was made based on a gene-specific amplified PCR product that was transcribed to dsRNA with the T7 Ribomax TM Express RNAi system kit (Promega) following the manufacturer's instructions. The control group was injected with ultrapure water following the same injection scheme. Primer sequences were: β-catenin-1, forward 5′-GCTGGATTGTTGGTTGAGGT-3′, reverse 5′-TGGTTGTGCATAATCGGAGA-3′; and teashirt, forward 5′-ACAATGACACCGGGAAATGTT-3′, reverse: 5′-ACAGTCGCTATCAGGTTGGG-3′. The total number of samples and independent experiments is indicated in the respective figures.

Quantitative real-time PCR (qPCR) was performed as described previously (Stevens et al., 2018). The RNA was extracted from one animal per sample using a standard phenol:chloroform protocol. Snap-frozen samples were dissolved in 100 µl lysis buffer (Qiagen, catalogue number 79216) containing 1% β-mercaptoethanol. The extracted RNA was precipitated using sodium acetate and ethanol. The concentration and purity was determined spectrophotometrically with the Nanodrop ND-1000 (NanoDrop Technologies). DNA was removed with the Turbo DNA-free kit (Invitrogen, catalogue number AM1907) following the manufacturer's instructions. cDNA was prepared using the Superscript III first strand synthesis supermix for qRT-PCR (Thermo Fisher Scientific, catalogue number 11752250), diluted and measured under universal cycling conditions with an ABI PRISM 7500 platform (Thermo Fisher Scientific). The most stable reference genes were selected using geNorm analysis. Gene expression analysis was performed with the MIQE guidelines taken into account and details of the procedure are given in Table S1; primer sequences are listed in Table S2 (Bustin et al., 2009).

Statistical analysis was performed in Open Source RStudio version 1.0143 (RStudio Team, 2015). Depending on the research question, a t-test or two-way ANOVA with a Tukey honest significant difference (HSD) post-hoc test was used. Normality of the data was analysed with the Shapiro–Wilk test and equal variance with the Bartlett test. If normality criteria were not met, the dataset was transformed (log, sqrt, 1/x or e^x). If the normality criteria were still not met, a non-parametric Kruskal–Wallis test followed by a pairwise Wilcoxon rank sum test was used. P<0.05 was considered significant. Pictures were processed and assembled in Adobe Photoshop. The background of pictures was removed and replaced by a black background for better visualization.

The authors thank Natascha Steffanie and Ria Vanderspikken for their skilful technical assistance.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.237545.reviewer-comments.pdf