Animals capable of whole-body regeneration must determine the identity of lost tissues in order to execute appropriate growth programs. Planarians accomplish whole-body regeneration through control of pluripotent neoblast stem cells, which can differentiate into all adult cell types, conserved transcriptional injury responses and robust positional information (Reddien, 2018). Standing Wnt and BMP gradients are expressed regionally within muscle and determine the regional identities necessary to regenerate (Molina et al., 2007; Reddien et al., 2007; Petersen and Reddien, 2008; Gurley et al., 2010; Witchley et al., 2013; Hill and Petersen, 2015; Lander and Petersen, 2016; Scimone et al., 2016, 2018; Sureda-Gomez et al., 2016; Stuckemann et al., 2017; Clark and Petersen, 2023). Muscle is also injury responsive and expresses factors regulating blastema identity (Scimone et al., 2017). Planarian muscle cells are mononucleated and possess actinomyosin-rich contractile fibers projecting either circularly, diagonally or longitudinally within the body wall, or dorsoventrally and surrounding the gut (Cebria et al., 1997; Cebria, 2016). However, the detailed subcellular composition of planarian muscle and how this contributes to their injury responsiveness and patterning abilities remains unknown. Planarians undergo a head-versus-tail decision early in regeneration, which is a model for understanding mechanisms of blastema fating. In Schmidtea mediterranea, injury triggers muscle expression of wnt1 at any wound and expression of the Wnt inhibitor notum preferentially at anterior-facing wound sites (Petersen and Reddien, 2009, 2011). wnt1 acts through beta-catenin-1 to suppress head regeneration and promote tail formation (Gurley et al., 2008; Iglesias et al., 2008; Petersen and Reddien, 2008, 2009; Adell et al., 2009; Rink et al., 2009), while notum acts oppositely to promote head regeneration (Petersen and Reddien, 2011). Wound-induced wnt1 and notum are expressed from 6 to 24 h in muscle near the injury, followed by expression at posterior and anterior blastema poles, respectively, from 48 to 96 h. The late expression phase requires stem cells and specific differentiation programs generating anterior and posterior blastema signaling centers (Petersen and Reddien, 2009; Gurley et al., 2010; Scimone et al., 2014; Vasquez-Doorman and Petersen, 2014; Vogg et al., 2014; Akheralie et al., 2023), while the early expression phase is stem-cell independent, takes place in pre-existing muscle and has different functional requirements. Early-phase wnt1 activates generically in muscle at the injury site, depends on foxG (Pascual-Carreras et al., 2023), arx-3 (Akheralie et al., 2023) and Hedgehog signaling (Rink et al., 2009; Yazawa et al., 2009), but not Wnt/β-catenin signaling (Petersen and Reddien, 2009), and is negatively regulated by follistatin (Tewari et al., 2018) and ddx24 (Sarkar et al., 2022). Early injury-induced notum is expressed from longitudinal muscle (Scimone et al., 2017), and is positively regulated by beta-catenin-1, wnt1, wntP-2 and arx-3, and negatively regulated by Dishevelled, wnt11-1/-2 and activin-2 (Petersen and Reddien, 2011; Cloutier et al., 2021; Gittin and Petersen, 2022; Akheralie et al., 2023). Therefore, notum is a Wnt feedback inhibitor at anterior-facing wounds, while wnt1 is prevented from activating notum expression at posterior-facing wounds through a polarity mechanism contributing to the blastema head-versus-tail decision process. Inhibition of notum or wnt1 only after amputation caused polarity reversal phenotypes (Petersen and Reddien, 2009, 2011), so their wound-induced expression is essential for regeneration, but it is still unresolved what relative contributions are made by their early versus late expression programs.
Muscle is pivotal for planarian regeneration, but it is unknown how muscle structure or physiology supports activation programs for factors like wnt1 and notum. The length of fiber contributed by each muscle cell has been difficult to assess by whole-mount immunostaining, but measurements of dissociated cells suggest projections are ∼200 μm, while animals range from 1 to 20 mm (Cebria, 2016). Early activation of wnt1 and notum occurs in muscle cell bodies within a similar 200 μm distance from the wound site, suggesting injured fibers could directly control this process. Alternatively, nearby tissues such as epidermis could indirectly signal to muscle, e.g. equinox is injury-induced in epidermis and promotes wnt1, notum and follistatin expression (Scimone et al., 2022). Muscle could also communicate signals distantly, and is required for transducing a body-wide wave of phospho-Erk (Fan et al., 2023). However, the lack of reagents to detect core processes in planarian muscle has limited an understanding of how muscle structure relates to its pro-regenerative function.
Classic studies of Dugesia dorotocephala found low doses of the microtubule depolymerizer colchicine resulted in regeneration patterning defects, including cyclopia, forked blastemas and posterior-facing heads, while high doses were lethal (McWhinnie, 1955). However, these observations have not been linked to specific pathways now known to control blastema fating. Microtubules are intrinsically polarized and have numerous functions in transport, force generation, cytoskeletal support, tissue migration and cell proliferation (Etienne-Manneville, 2013; Barlan and Gelfand, 2017; Brouhard and Rice, 2018; Goodson and Jonasson, 2018; Akhmanova and Kapitein, 2022), so they could contribute to regeneration in many possible ways. Although a major physiological consequence of microtubule inhibition across organisms is mitotic arrest (Pernice, 1889; Dixon and Malden, 1908), including in planarians (van Wolfswinkel et al., 2014), early wnt1 and notum activation is independent of cell proliferation (Petersen and Reddien, 2009; Vasquez-Doorman and Petersen, 2014). Planarian microtubules are involved in ciliogenesis in epidermal and excretory cells, and have roles in the nervous system, germline and neoblasts (Thi-Kim Vu et al., 2015; Basquin et al., 2019; Magley and Rouhana, 2019; Vu et al., 2019; Lesko and Rouhana, 2020; Christman et al., 2021; Ge et al., 2021; Gambino et al., 2022; Rouhana et al., 2022). However, the relationships between microtubules and specific responses in planarian muscle are unknown. Previous efforts to detect planarian microtubules using existing reagents detect staining in other tissues, such as ciliated epidermal and excretory cells and neurons (Sánchez Alvarado and Newmark, 1999; Robb and Sánchez Alvarado, 2002; Rink et al., 2011), and the presence and/or organization of microtubules in planarian muscle is unknown. Therefore, it is unclear whether the patterning effects of colchicine originally described in planarians affect muscle.
We generated an antibody recognizing an alpha-tubulin protein expressed in planarian muscle and used it to identify a network of microtubules running in parallel to the contractile fiber that can connect to muscle cell bodies. This reagent enabled detection of muscle fiber growth early in the process of wound repair. Using microtubule depolymerizing drugs, we determine that wnt1 and notum are among a small number of muscle-expressed, wound-induced genes strongly affected by microtubule disruption. Furthermore, we integrate these findings into known pathways for blastema identity determination to reveal how microtubules participate in regeneration specificity.
The planarian genome encodes 8 β-tubulins and 62 α-tubulins (Table S1). To examine the distribution and potential functions of microtubules in planarian muscle, we identified an α-tubulin with highly enriched expression across planarian muscle cell types (Fig. S1A), tuba-2 (dd508). We raised a rabbit polyclonal antibody predicted to uniquely target TUBA-2 protein (Fig. S1B) and found it labeled a dense filament network associated with muscle fibers (Fig. 1A). RNAi to target tuba-2 depleted anti-TUBA-2 staining, indicating appropriate targeting (Fig. 1C). However, we did not detect any defects in regeneration or homeostasis after tuba-2 RNAi (10/10 animals), suggesting potentially redundant functions for this protein, consistent with the large number of planarian alpha-tubulins. Co-labeling with planarian 6G10 antibody (Ross et al., 2015), which labels actinomycin-rich muscle fibers (Cebria et al., 1997), revealed that multiple TUBA-2 filaments ran in parallel to each 6G10 fiber (Fig. 1B). Higher-resolution imaging of individual z-slices detected TUBA-2 staining in circular, diagonal and longitudinal body-wall muscle fibers distributed across the body axis (Fig. 1C, Movies 1 and 2). Carnoy's fixative enabled detection of TUBA-2 protein, but 4% formaldehyde fixation did not. Carnoy's fixation also resulted in suboptimal Hoechst staining of nuclei, but we were able to identify z-slices that revealed the presence of TUBA-2 protein surrounding rare nuclei that we suggest represents the muscle cell body (Fig. 1D). Using a recently developed fixation (nitric/formic acids and paraformaldehyde, NAFA) (Guerrero-Hernandez et al., 2024), we detected collagen muscle cell bodies by FISH followed by anti-TUBA-2 immunostaining to find close association between muscle microtubules within fibers and associated cell bodies (Fig. 1E, Movies 3 and 4). Together, TUBA-2 protein is strongly expressed throughout planarian body-wall muscle.
To determine whether the filaments labeled by anti-TUBA-2 antibody are microtubules, we treated animals with chemical microtubule depolymerizers. Based on historical studies finding a role for microtubules in blastema patterning (McWhinnie, 1955), we chose a dose schedule in which animals were treated with colchicine or nocodazole for 24 h prior to fixation and immunostaining followed by recovery in drug-free media (Fig. 1F). By staining animals for the mitotic marker H3P over a dosage series of colchicine treatment, we determined that mitotic arrest reached maximal levels at 400 µg/ml (Fig. S2). In animals stained using anti-TUBA-2, doses that caused sub-maximal mitotic arrest (125 and 200 µg/ml colchicine) were sufficient to disrupt muscle fiber staining and cause accumulation of signal around cell bodies located nearby (Fig. 1F). Nocodazole administered at concentrations previously used to arrest mitosis maximally (400 ng/ml) (van Wolfswinkel et al., 2014) and also a lower dose (100 ng/ml) resulted in similar disruptions to TUBA-2 staining that reduced filament density and accumulated signal in cell bodies, and under these conditions without loss of 6G10 fibers (Fig. 1F,G). The disruptions to TUBA-2 from colchicine or nocodazole did not appear regionalized across the body axis (20/20 animals). Nocodazole caused stronger effects on TUBA-2 staining than colchicine but also greater lethality (8/95 survived 24 h of 100 ng/ml nocodazole but 28/30 survived 125 µg/ml colchicine), so we used colchicine for the majority of subsequent experiments. Together, these results indicate TUBA-2 marks muscle microtubules, and that the muscle cell body could be a site of unpolymerized tubulin heterodimer accumulation. The sensitivity of muscle microtubules to inhibitors of microtubule polymerization indicates they are in a state of dynamic maintenance.
We examined how the muscle microtubules change dynamically after injury (Fig. 2A,B). To enable the most straightforward imaging, we focused on flank incisions, which heal rapidly within 1 day. In the first 2 h following incision injury, anti-TUBA-2 antibody detected strong signal at the wound site, but the staining had a disorganized and blebbed appearance compared to intact tissue. By 4 h, TUBA-2 muscle fibers projected outward from each side of the wound, and by 6 h some fibers had crossed the injury site. By 18-24 h, the muscle fiber network had fully traversed the wound, but TUBA-2 staining remained stronger at this location (Fig. 2A). These overall behaviors were mimicked by 6G10 staining (Fig. 2B), suggesting that the growth of microtubules is likely concurrent with growth of muscle fibers across the wound. Colchicine treatment for 24 h prior to injury (at 125 µg/ml), and followed by recovery in control media after surgery, prevented any formation of TUBA-2 muscle fibers projecting across the wound by 4-6 h (Fig. 2A). By 18 h after injury, wound sites from colchicine-treated animals appeared closed, but they tended to accumulate muscle cell body expression of TUBA-2 (Fig. 2A). 6G10 expression at the wound site was also disrupted by colchicine treatment (4-24 h, Fig. 2B), suggesting microtubules are likely important for the process of muscle fiber growth. We suggest that, under these conditions, the overall process of wound healing is likely be driven by other tissues, such as the overlying epidermis, but that microtubule-dependent muscle fiber growth contributes to restoration of the muscle system at wounds. Given the rapid responses to regrow muscle fibers across incisions, we predicted this process would likely be driven by growth of pre-existing muscle rather than new differentiation. To examine this possibility, we tested muscle fiber growth after incisions in animals lethally irradiated with 6000 Rads of X-rays, a treatment known to eliminate planarian stem cells and any new differentiation (Reddien et al., 2005). Irradiated animals could restore TUBA-2 muscle across an incision site with the same kinetics as uninjured animals, indicating repair likely occurs through growth of pre-existing muscle (Fig. 2C). We also stained regenerating trunk fragments with the anti-TUBA-2 antibody during the first 18 h after amputation (Fig. S3), although it was difficult to unambiguously assign TUBA-2 fibers at the injury site as either body-wall muscle projecting across the amputation or DV muscle fibers parallel to the wound plane. However, we did not notice substantial differences between TUBA-2 at anterior or posterior wounds. We conclude microtubules have a likely involvement in muscle fiber growth early, within the first day following incision injuries, and detection of muscle microtubules can track the behavior of muscle during injury repair and regeneration.
Given the involvement of muscle microtubules in early injury responses, we hypothesized that microtubules might contribute to transcriptional responses to injury and used RNAseq to analyze microtubule-dependent transcriptome changes after amputation. Animals were treated with colchicine for 24 h with 0 or 200 µg/ml colchicine, then amputated and allowed to recover in colchicine-free media, followed by isolation of total RNA from wound-proximal tissue at 0, 4 and 18 h after amputation. 118 genes were identified as wound-induced in normal animals because they were upregulated at either 4 or 18 h after amputation under control treatments. A large number of genes were perturbed by colchicine at the 0-h timepoint (2691 downregulated and 2348 upregulated), reflecting broad roles for microtubules in the animal (Fig. S4A, Table S2). However, a more limited number of genes were differentially expressed by colchicine treatment in regeneration (375 genes were downregulated at either 4 h or 18 h, and 278 genes were upregulated at either 4 or 18 h), likely due to allowing time for recovery in colchicine-free media following amputation (Fig. S4A, Table S2). Genes downregulated by colchicine at 0 h included tuba-2 and many other tubulins, consistent with a conserved feedback mechanism that downregulates tubulin mRNA and translation when tubulin monomers are in excess (Cleveland et al., 1981; Batiuk et al., 2024) (Fig. S4C). Using this dataset, we assessed the behavior of wound-induced genes (Wenemoser et al., 2012; Wurtzel et al., 2015) (Fig. 3A,B, Fig. S5). Only a subset of injury-induced genes had significantly different expression at 4 or 18 h due to colchicine treatment. These included reduced expression for wnt1, notum, runt-1 and h2b, while dd1039, dd9204, HSP20L, TNFAF1 and hadrian had increased expression (Fig. 3A,B, Fig. S5). wnt1 and notum are injury induced exclusively in muscle, while runt-1 and h2b are induced in neoblasts (Wurtzel et al., 2015). Colchicine treatment still enabled the expression of many other injury-induced genes known to be expressed from muscle, including inhibin-1, wntless, follistatin, nlg-1, CALM1, 35exo and fascin. We confirmed these effects on runt-1, inhibin-1, wntless, nlg-1 and follistatin by FISH (Fig. 3C, Fig. S6A,B). Therefore, muscle injury-induced genes displayed a variation in their sensitivity to colchicine treatment. Likewise, colchicine inhibited some but not all injury-induced genes expressed from neoblasts and epidermis. Among the genes wound-induced in neoblasts, runt-1 and h2b were sensitive to colchicine but not inx-13 and HSP20L, and colchicine-treated animals still activated the epidermally expressed genes equinox and TNFAF. Together, although colchicine caused a complex transcriptional response, its effects on injury-induced genes were surprisingly specific and prominently affected wnt1 and notum.
We further evaluated whether the effect on wnt1 and notum could be due to wound healing failure or loss of muscle. Wounds close to stop the outflow of debris within ∼1-2 h and acquire a smoother appearance, and treatment with 125 and 200 µg/ml colchicine still enabled this process, as monitored in live animals (Fig. S7), consistent with ultrastructural studies finding planarian epidermal wound closure is impaired by inhibition of actin (Pascolini et al., 1984) and not microtubules (Hori, 1978). Therefore, the sensitivity of some wound-induced factors like notum to colchicine is unlikely to arise from a failure of wound closure. Likewise, a trivial explanation for the elimination of notum or wnt1 expression could be failure to produce or sustain muscle cells capable of expressing these genes at the wound site. However, animals treated with colchicine still had abundant collagen body-wall muscle, including myoD longitudinal muscle cells, which are the sole source of notum (Fig. S8). We conclude that microtubules have a specific role in the capacity for muscle cells to express a subset of injury-induced genes.
Given the relatively specific effect of microtubule inhibition on notum and wnt1, we carried out experiments to better understand how their expression changes spatially in response to various doses of colchicine. We began with notum as it is the only gene expressed asymmetrically at anterior versus posterior wound sites (Wurtzel et al., 2015). A colchicine concentration series testing notum expression at 18 h confirmed the RNAseq data that doses of at least 200 µg/ml led to strong reductions or failures of notum expression from anterior-facing wounds (Fig. S9A). Surprisingly, however, 125 µg/ml colchicine caused elevation of notum expression at posterior-facing wounds at ∼20-40% penetrance (Fig. S9A). In separate cohorts measured across 4 independent experiments, we confirmed that 125 µg/ml colchicine treatment caused 25/59 animals to have >10 notum cells at posterior-facing wounds, while all control animals had fewer than 10 notum cells (36/36 animals) (Fig. 4A). This excess expression phenotype at posterior-facing wounds occurred between 75-125 µg/ml colchicine, but no concentration tested could fully separate from doses leading to less notum at anterior-facing wounds (Fig. S9A). We also verified nocodazole-treated animals also displayed similar phenotypes, with a lower dose (100 ng/ml) elevating notum at posterior-facing wounds, while higher doses (400 ng/ml) caused loss of notum from anterior-facing wounds (Fig. S9B). Therefore, microtubules likely participate in early wound-induced notum expression in two different ways: one that promotes notum expression and is only inhibited by relatively higher amounts of colchicine, and another that polarizes notum expression and is inhibited by relatively lower amounts of colchicine.
We examined the impact of colchicine on the pole-specific phase of notum expression from 48-96 h and head/tail regeneration outcomes. At both the 125 and 200 µg/ml doses, colchicine delayed expression of a notum anterior pole that began at 24 h in control animals but was not detected in any colchicine-treated fragments until 96 h (Fig. 4B). Likewise, while transverse fragments treated with 0 or 50 µg/ml colchicine produced anterior pole sfrp-1 by 5 days, 125 µg/ml colchicine treatment delayed sfrp-1 anterior pole formation until between days 5 and 16 (Fig. S10). This delay could be due to a combination of modified injury signaling and also recovery from mitotic arrest, which through H3P staining described above was more severe in the anterior half of the body (Fig. S2). Next, given the classic studies of microtubule inhibition on planarian regeneration (McWhinnie, 1955), we analyzed a series of experiments to determine whether colchicine could cause posterior head regeneration. Although posterior blastemas underwent a reproducible but incompletely penetrant elevation of notum expression during the early polarity decision after microtubule inhibition, we did not detect regeneration of posterior heads from animals treated over a range of colchicine concentrations (0 to 200 µg/ml) (total of n=80 animals as assessed by live scoring). Likewise, posterior blastemas of colchicine-treated animals did not form notum anterior pole tissue within 96 h of regeneration (Fig. 4B), and no colchicine-treated animals formed sfrp1 anterior pole tissue within their tail blastemas at either 5 or 16 days post-amputation (61/61 animals stained, Fig. S10). We attempted to examine head/tail regeneration in animals treated with higher doses (200 or 500 µg/ml colchicine 24 h prior to amputation) or after dosing for longer periods of time (72-120 h of 125 or 200 µg/ml colchicine prior to amputation), but these caused lethality within 4-5 days (50 animals total). We conclude that, despite modifying the early expression of injury-induced head/tail regulatory factors, microtubule inhibition alone was not able to lead to blastema head/tail fate transformations. notum expression polarity correlates with head/tail regeneration outcomes in normal animals, but earlier work found that disruptions of notum polarity do not always correlate with head/tail regeneration outcomes. For example, a greater fraction of activin-2(RNAi) animals express excess notum than go on to form posterior-facing heads (Cloutier et al., 2021). A possible explanation is that early notum polarity acts redundantly with other factors to regulate blastema fating.
However, notum expression represents the earliest known symmetry-breaking step in early planarian regeneration, and so we sought to place microtubule regulation within this pathway. wnt1 and beta-catenin-1 promote notum expression from anterior-facing wounds (Petersen and Reddien, 2011; Gittin and Petersen, 2022), so we tested whether these factors were required for ectopic notum produced at posterior-facing wounds under the 125 µg/ml colchicine (Fig. 4C). Indeed, inhibition of beta-catenin-1 or wnt1 prevented expression of notum at posterior-facing wounds in animals treated with 125 µg/ml colchicine (Fig. 4C). By contrast, inhibition of wnt11-2 increases notum expression from posterior-facing wounds at 18 h (Gittin and Petersen, 2022), and 125 µg/ml colchicine did not enhance this effect (Fig. S11). One possibility is that wnt11-2 acts through a microtubule-dependent step to control notum polarization, consistent with the failure to observe phenotypic enhancement in this experiment. However, because both the RNAi and also microtubule inhibitions are unlikely to eliminate function, these factors could act in parallel but in some way preventing the detection of enhancement. We conclude that, at a minimum, wnt1 and beta-catenin-1 are important for the overactivation of notum at posterior-facing wounds after 125 µg/ml colchicine.
We next examined how colchicine affected wnt1 expression behavior (Fig. 5A). In normal animals, wnt1 wound-induced expression normally peaks at 12 h and is declining by 18 h, followed by expression selective to the posterior pole starting at 48-72 h. By contrast, 125 or 200 µg/ml colchicine strongly reduced or eliminated expression of wnt1 at 12 or 18 h in anterior-facing wounds (Fig. 5A). Posterior-facing wounds had a more complex response, with 200 µg/ml colchicine preventing wound-induced expression at both timepoints, while 125 µg/ml colchicine delayed the onset of wnt1 until 18 h. In addition, posterior wound sites had expression of wnt1 along the dorsal posterior midline, reminiscent of its expression in uninjured animals (Fig. 5A). Similarly, wnt1 expression remained expressed at the posterior pole of colchicine-treated animals from 48 to 96 h and occupied a larger domain within the posterior domain (Fig. 5B). Homeostatic animals treated with 125 or 200 µg/ml colchicine for 3 days also underwent an expansion of the wnt1 posterior domain (Fig. 5C). However, domain sizes of other Wnt-related factors marking the AP axis (wnt11-2, wntP-2 and notum) were not significantly different (Fig. S12A). Further demonstrating that specificity of microtubule inhibition effects on wnt1, midline expression of slit expression was normal at 18 h in colchicine-treated animals (Fig. S12B). Because animals all died within 4-5 days of continuous colchicine treatment, it is possible that any effects of microtubule inhibition on AP patterning downstream of wnt1 could occur over a longer timescale that cannot be observed because of animal death. However, these results indicate that microtubules can limit the posterior domain of pre-existing and tail blastema-specific wnt1 at the posterior pole. Together, these results indicate a complex relationship between microtubule function and both notum and wnt1 expression in planarians. High colchicine doses (200 µg/ml) prevented wound-induced wnt1, lower doses (125 µg/ml) delayed the onset of injury-induced wnt1 at posterior-facing wounds, and both doses still enabled regeneration of a wnt1 posterior pole after amputation whose domain was expanded.
Given the role of wnt1 in promoting notum expression, the dose-dependent expression behavior of wnt1 after microtubule inhibition offers potential explanations for why lower colchicine doses result in ectopic notum expression but high doses lead to no notum expression. We suggest a single microtubule-dependent process may control the extent and/or timing of wnt1 activation, and this process may either be under A-P control or differentially sensitive to microtubule inhibitions at distinct axis positions. High colchicine doses strongly reduce this response, leading to lack of injury-induced wnt1 and consequently lack of notum expression. Lower colchicine doses still enable delayed wnt1 expression that, at posterior-facing wounds, can activate notum and reveal a simultaneous disruption to the polarity ordinarily restricting notum to the anterior. A potentially separate microtubule-dependent process additionally regulates wnt1 at the posterior pole. The net effect of modifying wnt1 and notum after microtubule inhibition still enables appropriate tail blastema specification to occur, because the dual reductions to injury-induced wnt1, and/or increases in notum expression at posterior-facing wounds, could be counterbalanced by posterior-pole wnt1 expression. Consistent with a model in which early injury-induced and posterior-pole wnt1 expression phases act in parallel, a recent study found that S. polychroa embryos transit through a stage where they can regenerate tails and the wnt1 posterior pole without the earlier wound-induced expression phase of wnt1 (Booth et al., 2025).
One possible contributor to the effects observed on expression of these genes could be the particular strategy of microtubule inhibition used in these experiments. In the experiments above, we used a scheme for microtubule inhibition for 24 h prior to amputation, followed by recovery in microtubule-free media, in order to follow the design of classic studies that found microtubule inhibition could cause head/tail regeneration polarity defects (McWhinnie, 1955) and avoid eventual lethality. The inclusion of a wash-out recovery step enables animals to survive but could complicate the interpretation that microtubules are necessary for wnt1 and notum expression behavior, because recovering animals may have unique responses as they eventually re-establish their microtubules during recovery. To examine these possibilities, we tested the importance of the wash-out step in the ability of colchicine to reduce wnt1 and notum expression at 200 µg/ml and elevate notum expression at 125 µg/ml. Continuously treating animals with 200 µg/ml colchicine for 24 h pre-amputation and 18 h post-amputation caused a similar reduction to notum and wnt1 expression at anterior-facing wounds as in treatments that included a washout step. Likewise, animals treated with 125 µg/ml colchicine continuously also caused increased notum expression at posterior-facing wound sites (Fig. S13). Therefore, microtubule inhibition is the cause of wnt1 and notum expression effects and not as a consequence of recovering from inhibition.
Our model suggested that microtubules may control wnt1 and notum expression phases through parallel pathways. To examine for any participation of microtubules in head/tail decisions that might occur in conjunction with other regulators, we tested whether colchicine could modify the appearance of regeneration phenotypes after Wnt RNAi. RNAi of wnt1 under these conditions caused a 22% penetrant phenotype of sfrp-1 posterior head formation (5/23 animals), but simultaneous treatment with 125 µg/ml colchicine increased the penetrance to 88% (15/17 animals, P<0.0001 by 2-tailed Fisher's exact test) (Fig. 6A). Similarly, wnt11-2 RNAi animals all failed to regenerate tails without regenerating posterior heads (16/16 animals by live scoring and 28/28 animals lacked posterior sfrp1 expression), but simultaneous treatment with 125 µg/ml colchicine caused a fraction of these animals to form posterior heads (4/21 animals as measured by posterior sfrp-1 expression, P=0.0432 by two-tailed Fisher's exact test) (Fig. 6A). Therefore, microtubules act with Wnt signaling to regulate the specificity of regeneration. Together, our results indicate that muscle contains a network of microtubules, that microtubules are required for repair of pre-existing muscle fibers across injury sites, that specific microtubule regulatory processes regulate injury-induced expression of wnt1 and notum from muscle, and that microtubules functionally contribute to the wnt1-dependent process of head/tail blastema identity determination (Fig. 6B,C).
Our analysis reveals that planarian body-wall muscle harbor a network of microtubules along their actinomyosin-rich contractile fibers and identifies functional roles for microtubules in the process of muscle fiber growth and expression of key patterning genes activated within muscle. By imaging of TUBA-2 muscle microtubules after incision injuries, we show that body-wall muscle regrows following wounding (Fig. 6B). Because this process occurs in animals irradiated at doses that eliminate stem cells, pre-existing muscle cells are likely the source of repair, rather than differentiation of new muscle cells. Colchicine prevented fiber regrowth, indicating microtubules are crucial for this process. Although microtubules are present broadly in many planarian cell types with a diverse set of functions, we find that microtubule inhibition affects expression strongly for a subset of injury-induced genes. Regeneration polarity factors wnt1 and notum were among the most strongly downregulated genes following microtubule inhibition, indicating that some microtubule-dependent process selectively promotes their expression after injury. At lower colchicine doses, animals underwent a selective upregulation of notum expression at posterior-facing wound sites in a step requiring wnt1 and beta-catenin, similar to normal expression of notum at anterior-facing wounds. Under conditions of strong microtubule inhibition, no wnt1 expression occurs and therefore wnt1-dependent expression of notum at anterior-facing wounds does not occur. At lower levels of microtubule inhibition (125 µg/ml), wnt1 wound-induced expression still activates at posterior-facing wounds, though in a delayed fashion, and the process repressing notum at posterior-facing wounds is partially disrupted. This polarity mechanism could be driven by wnt11-2, activin-2 or other as yet unidentified factors. In addition, microtubule inhibition also expanded the wnt1 posterior pole in uninjured animals and also animals regenerating a new tail. In normal animals, the dual sources of wnt1 from the posterior pole and also early injury activation phases could act in parallel to promote tail versus head blastema fate. This model would explain why dual inhibition of microtubules along with either wnt1 or wnt11-2 resulted in increases to the fraction of animals undergoing regeneration with inverted polarity (Fig. 6C).
Our study reveals that microtubules regulate injury-induced genes and control regeneration decision making, and also raises several questions about how microtubules contribute to these processes on a mechanistic level in planarians. Microtubules are known to mutually interact with Wnt signaling through a wide variety of mechanisms across proliferative and post-mitotic cells across many organisms (Kikuchi and Arata, 2024), suggesting myriad opportunities for planarian microtubules to participate in Wnt-mediated control of head versus tail regeneration. In canonical Wnt signaling, Wnt binding to Frizzled receptors activates Dishevelled, which inactivates the GSK3 kinase and thereby prevents a destruction complex from constitutively degrading β-catenin protein, which can then undergo nuclear translocation and transcriptional regulation in conjunction with TCF transcription factors. β-Catenin can be transported to the nucleus via microtubules and is dependent on Kinesin-2 and the IFT-A complex (Vuong et al., 2014, 2018; Balmer et al., 2015). APC, a component of the β-catenin destruction complex, is also a microtubule plus-end binding protein and can be transported along microtubules (Jimbo et al., 2002). Wnt signaling components also regulate microtubule dynamics. GSK3, in some cases acting through the Wnt pathway, can regulate microtubule stability and dynamics through phosphorylation and modification of the microtubule plus-ended binding ability of Tau, MAP1B and APC (Lucas et al., 1998; Zumbrunn et al., 2001; Johnson and Stoothoff, 2004; Noble et al., 2005; Caccamo et al., 2007). Axin, a scaffolding protein that functions in the β-catenin destruction complex, binds to γ-tubulin and can be involved in microtubule organizing center formation (Ruan et al., 2012). Dishevelled regulates both canonical and non-canonical Wnt pathways, can participate in the regulation of MAP1B to control microtubule stability (Ciani et al., 2004), and can regulate the orientation of microtubules linked with the cell cortex in migration and cell division (Matsumoto et al., 2010; Yang et al., 2014; Kikuchi et al., 2018). Furthermore, planar cell polarity pathways that involve regulation through Dishevelled and the asymmetric localization of Frizzled and Vangl cell-surface proteins can rely on microtubules for polarized transport of these proteins within the cell (Shimada et al., 2006; Matis et al., 2014), and downstream outcomes of planar cell polarity can also be regulated through control of microtubule dynamics (Vladar et al., 2012). Therefore, the function of microtubules to support wound-induced expression of wnt1, the regulation of notum polarity and the control of head-versus-tail blastema specification could, in principle, arise from these or other potential relationships. The analysis of microtubule regulatory factors and also the generation of new tools to allow assessment of Wnt pathway component localization in planarians will be an important future step to further understand these mechanistic links.
Based on our phenotypic analysis, our work rules out several different candidate models for how microtubules promote injury-induced expression of wnt1 and notum. Muscle from injured animals treated with 200 µg/ml colchicine could still activate expression of several wound-induced genes, such as nlg1, wntless and inhibin, and these animals still had abundant collagen and myoD muscles, so the lack of wnt1 activation in these animals is unlikely to be due to a nonspecific requirement for microtubules to maintain muscle itself or its general wound responsiveness. Our RNAseq analysis to identify genes differentially expressed following colchicine treatment in planarians also did not detect modifications to expression of the wnt1-inducing transcription factors foxG or arx-3 (Akheralie et al., 2023; Pascual-Carreras et al., 2023). Additionally, wnt1 expression does not require beta-catenin-1 (Petersen and Reddien, 2009), so microtubule-mediated regulation of wnt1 occurs likely through another pathway. Planarian hedgehog is necessary for wnt1 expression (Rink et al., 2009) and does not control notum expression polarization (Petersen and Reddien, 2011). Hedgehog signaling involves Kif7-mediated microtubule transport within the primary cilium in vertebrate cells (He et al., 2014), although earlier analyses have not found support for Cos2/Kif27/Kif7 kinesins being involved in the transduction of Hedgehog signaling effects relating to head/tail determination in planarians, but rather in Hedgehog-independent ciliogenesis (Rink et al., 2009). However, our experiments do not rule out the possibility that a variant Hedgehog pathway could be relevant for microtubule activity and wnt1 expression in muscle. In addition, microtubules could regulate wnt1 through follistatin, which is induced from muscle cells at injury sites and negatively regulates injury-induced wnt1 expression independent of its role to enable the activation of proliferation and cell death in an animal following large injuries (Tewari et al., 2018). However, wound-induced follistatin expression still occurred but at reduced levels following colchicine treatments that, nonetheless, eliminated wnt1 and notum expression. These observations are inconsistent with a hypothetical model in which microtubules suppress follistatin activation in order to limit wnt1 activation. Microtubule action in other types of muscle could also be important for wnt1 expression. The expression of wnt1 on the posterior dorsal midline is negatively regulated by STRIPAK components mob4 and striatin (Schad and Petersen, 2020), bmp (Clark and Petersen, 2023), and the kinase pak1 acting upstream of warts and yorkie in the Hippo pathway (Lin and Pearson, 2014, 2017; Doddihal et al., 2024). Microtubules similarly restrict this wnt1 domain posteriorly within the tail, suggesting possible uses for microtubules upstream or downstream of these factors. Future work will be necessary to resolve how microtubules regulate the wnt1-related pathways that together impinge on head/tail determination.
Likewise, several candidate models could potentially explain how microtubules are involved in the polarization of notum expression. One possibility is that microtubules are responsible for asymmetric transport or sequestration of mRNA/protein polarity factors along muscle fibers, such that these cells are primed for either activation or suppression of notum, depending on their orientation to a wound site. Because notum is a feedback inhibitor of β-catenin signaling, it is also possible that known mechanisms relating microtubule regulation to canonical or noncanonical Wnt signaling could be responsible for the role of microtubules in suppressing notum expression specifically at posterior-facing wounds (described above and reviewed by Kikuchi and Arata, 2024). Alternatively, colchicine could disrupt planar cell polarity signals that emerge from polarized tissues, such as the planarian epidermis (Almuedo-Castillo et al., 2011; Vu et al., 2019), that generate polarized notum expression outcomes within nearby muscle at injury sites; however it unknown whether epidermis is necessary for notum polarization. Microtubules are intrinsically polarized polymers whose organization within muscle could itself imbues these cells with polarity. Microtubules could either be oriented in a consistent direction within each type of muscle aligned with the body axis (e.g. plus-ends anterior in all longitudinal muscle) or such alignments could occur in each cell but not uniformly for all muscle cells (e.g. all longitudinal muscle has polarized microtubules but they adopt random orientations). Alternatively, they could adopt a distribution of orientations within each fiber (i.e. plus ends of individual tubules aligned in either direction along the fiber) or have a common orientation with respect to the cell body (e.g. minus ends all located at the cell body and plus ends distally in the fibers). Although no A/P differences in individual muscle fibers are known, light-sheet imaging of 6G10-labeled muscle found a distinct geometry of fiber alignment at the anterior versus posterior termini, and the fiber network morphology was overall modified by beta-catenin-1 RNAi, suggesting that the muscle network structure is responsive to head-tail patterning (Lu et al., 2025). The establishment of additional reagents for detection of microtubule end proteins and microtubule-associated factors will be important for resolving how microtubules are oriented in planarian muscle to further understand how they may contribute to polarized blastema choices.
Our study demonstrates that microtubules functionally participate in conjunction with Wnt pathway components to control regeneration decisions in planarians. This participation could include both promoting injury-induced wnt1 expression and suppressing notum at posterior-facing wounds, as both would bias blastema specification toward tail fates. In addition, we find microtubule inhibition does not eliminate, but actually expands, wnt1 expression at the posterior pole. Our findings that colchicine treatment dramatically enhanced the wnt1(RNAi) and wnt11-2(RNAi) phenotypes could be consistent with these three pathway branches acting in series, given likely incomplete inhibition from RNAi and drug treatments, or acting in parallel. Consistent with this idea, while notum expression asymmetry is prominent in some planarians, i.e. Schmidtea mediterranea (Petersen and Reddien, 2011) and Dugesia japonica (Durant et al., 2019), its expression is symmetric in Girardia sinesis, which undergoes whole-body regeneration but with naturally frequent head/tail polarity reversals (Cleland et al., 2025 preprint). Therefore, while notum asymmetry is not a universal feature of planarian regeneration, it may facilitate robust blastema decision making. Our analysis suggests the head/tail decision process involves multiple redundant events that could vary across species and help ensure regeneration robustness in S. mediterranea.
How damage signals are integrated to produce specific regeneration outcomes is still not well understood at the cellular level in any organism. Our study raises the possibility that damage to muscle fibers provides a signal that activates expression of a set of genes that subsequently regulate regeneration decisions. Future work will be crucial to uncover the triggers and relay mechanisms central to this process. The use of microtubules in actinomycin-rich cell projections could have a more general purpose in mediating injury responses. Our analysis identifies a network of microtubules in planarian muscle and defines microtubules as crucial for regulating specific injury-induced programs involved in the regulation of blastema identity.
Asexual Schmidtea mediterranea (CIW4 strain) were cultured in 1× Montjuic salts at 18-20°C. Animals were fed pureed calf liver once a week and cleaned at least once every week in static cultures for maintenance. Animals were starved for at least 7 days before experiments.
RNAi treatments were performed by feeding with dsRNA (16% v/v) at a final concentration of ∼300 ng/µl, green food dye (4% v/v) and pureed liver (80% v/v). Animal of each condition were cultured in separate Petri dishes in 1× Montjuic salts. dsRNA was synthesized as previously described (Bonar et al., 2022; Clark and Petersen, 2023). RNAi negative controls used dsRNA targeting C. elegans unc-22, a gene not present in the Schmidtea mediterranea genome. Animals were fed 10 µl of RNAi food mixture per worm every 2-3 days for the indicated length of the experiment. For RNAi treatment without injury, animals were fixed 5 days after the final feeding. For a regeneration time courses following injury, surgery was performed 2 days after the final feeding, prior to fixations at the indicated times.
For preparation of colchicine-containing media, 500 µg/ml colchicine (Sigma C9754-500MG) dissolved in 1× Montjuic salts was prepared freshly and then diluted to desired concentrations in 1× Montjuic salts. Animal media were replaced with colchicine-containing media for 24 h and, if surgeries were performed, animals were allowed to recover in colchicine-free media while they regenerated until further processing. For experiments involving colchicine, planarian water (1× Montjuic salts) was used as a negative control. Nocodazole (Sigma 487929-10MG-M) was dissolved in 1% DMSO in 1× Montjuic salts at the indicated concentrations and the solution made freshly before use. Animal media were replaced with nocodazole-containing media for 24 h at the indicated concentrations, then if surgeries were performed, animals were allowed to recover in 1× Montjuic salts without nocodazole for the indicated times prior to fixation and analysis. For experiments involving nocodazole, mock treatment of animals with 1% DMSO in planarian water were used as negative controls.
Fluorescence in situ hybridization (FISH) was performed using established protocols (King and Newmark, 2013). Briefly, animals were treated with 7.5% (w/v) N-acetyl cysteine (NAC, Sigma A7250-100G) in 1×PBS, fixed in 4% formaldehyde (w/v) (Sigma F8775-500ML) and stored in methanol (ThermoFisher A4544). They were rehydrated with methanol: 1× PBSTx (PBS with 0.3% Triton X-100 Sigma, T8787-250ML), bleached in 1.2% HO (v/v) (H1009-500ML), 5% formamide and 0.5xSSC in 1× PBS, permeabilized with proteinase K (10 µg/ml; Invitrogen, 25530049), then prehybridized at 56°C in pre-hybridization hybridization buffer (50% formamide, 5×SSC, 1 µg/ml yeast RNA, 0.5% Tween). Animals were incubated for 16 h in hybridization buffer (50% formamide, 5×SSC, 1 µg/ml yeast RNA, 0.5% Tween and 5% dextran sulfate) containing 1:1000 riboprobes, followed by two washes each of prehybridization buffer, then in 2×SSC/0.1% Triton-X, and 0.2×SSC/0.1% Triton-X. Hybridization occurred with digoxigenin- or fluorescein-labeled riboprobes at a 1:1000 concentration (v/v), which were synthesized using T7 RNA-binding sites for antisense transcription. Anti-digoxigenin-POD (Roche/Sigma 11207733910) or anti-fluorescein-POD antibodies (Roche/Sigma 11426346910) were in a solution of 1× TNTx/10% (v/v) horse serum (Sigma, H1138-500ML)/10% (v/v) Western Blocking Reagent (Roche 11921673001) at a concentration of 1:2000 (v/v). Homemade fluorescein tyramide or rhodamine tyramide prepared as described previously (King and Newmark, 2013) was used in TSA buffer (2 M NaCl and 0.1 M Boric acid and pH 8.5) for 10 min to label the specimens, followed by seven washes in TNTx. For double FISH, the enzymatic activity of tyramide reactions was inhibited by sodium azide (100 mM). Nuclei were stained using 1:1000 Hoechst (v/v, ThermoFisher H3570) in 1×TNTx.
Anti-TUBA-2 antibody was generated by Genscript using a C-terminal peptide derived from the residues 440-454 of the TUBA-2 (dd508, SMEST054547001.1) protein sequence with an N-terminal cysteine appended for KLH conjugation (N-CVGYDSADIGNADQD-C). Blast searching confirmed no other exact matches to this peptide exist in the planarian proteome. For immunostaining, animals were fixed in 10 ml glass scintillation vials by replacing planaria water with 2% HCl in water for 30 s with gentle swirling, then media were gently replaced with Carnoy's solution (containing 60% ethanol, 30% chloroform and 10% glacial acetic acid) without disturbing specimens, incubated without mixing for 2 h on ice (4°C), followed by incubation in methanol at -20° for at least 1 h. Animals were bleached in 6% hydrogen peroxide (H1009-500ML) in methanol (v/v) (ThermoFisher A4544) overnight on a lightbox prior to immunostaining. For immunostaining with anti-TUBA-2 (this study) and 6G10 (Developmental Studies Hybridoma Bank 6G10-2C7) antibodies, animals were blocked with 10% horse serum (Sigma H1138-500ML) and 10% Roche Western Blocking Reagent (Roche 11921673001) in PBSTx then incubated overnight at room temperature with primary antibody (anti-TUBA-2 polyclonal rabbit antibody at 1:1000, 6G10 mouse monoclonal antibody at 1:1000) in blocking solution. For H3P staining, animals were incubated in 5% horse serum in PBSTx then incubated overnight at room temperature with rabbit anti-ser10-H3P antibody (Cell Signaling Technology, 3377S) at 1:300 diluted in blocking solution. Animals were washed with PBSTx six times over 6 h prior to incubation with secondary antibodies. Secondary antibodies and concentrations were: goat-anti-rabbit-alexa568 (ThermoFisher A11036) at 1:1000 and goat-anti-mouse-Alexa488 (ThermoFisher A32723) at 1:1000 in Fig. 1A-D,G; goat anti-rabbit-HRP (ThermoFisher G21234) at 1:1000 and goat anti-mouse-alexa568 (ThermoFisher A11031) at 1:1000 in Figs 1F, 2, Figs S1C, S3; or goat anti-rabbit-HRP (ThermoFisher G21234) at 1:1000 in Fig. S2). Tyramide development in 1× PBSTi (PBSTx with 10 mM Imidazole) was performed as described previously (Pearson et al., 2009) to label samples stained with HRP-conjugated secondary antibodies. For nuclear counterstain labeling, samples were incubated in Hoechst dye at 1:1000 and washed at least four times prior to mounting in Vectashield (Vector Laboratories, H-1000) and imaging. Experiments to perform FISH and immunostaining followed a NAFA (nitric acid and formic acid) fixation and staining protocol described previously (Guerrero-Hernandez et al., 2024). Briefly, animals were fixed in NA (nitric acid) solution containing 4% paraformaldehyde (Electron Microscopy Services 15710) in 100 mM HEPES (pH 7.5), 25 mM EGTA, 50 mM MgSO4 and 0.53% nitric acid) for 1-2 min, followed by FA (formic acid) solution [4% paraformaldehyde, 100 mM HEPES (pH 7.5), 25 mM EGTA and 4.80% formic acid (Sigma F0507-500ML)] for 45 min, washed twice in 1× PBS, once in 50% methanol/1×PBS and then transferred into 100% methanol, incubated in 100% methanol for at least 1 h at -20°C, followed by transfer in to 1×PBS and bleaching in formamide bleach solution for 2 h under a light source. Tyramide-FISH detection proceeded as above described in the FISH procedures but omitted the proteinase K step as in the published NAFA-FISH protocol (Guerrero-Hernandez et al., 2024). Samples were then detected by immunostaining, using 1:1000 anti-TUBA-2 primary antibody, followed by 1:1000 anti-rabbit-488 as described above. Samples were labeled with 1:1000 Hoechst dye and washed six times prior to mounting and imaging.
Genes were cloned by RT-PCR after cDNA synthesis (Superscript III, oligo-dT priming) from mixed stage planarian total RNA. Primers used for cloning wnt1, wntP-2, wnt11-2, runt-1, beta-catenin, bmp and slit have been described previously (Petersen and Reddien, 2008, 2009; Wenemoser et al., 2012; Lander and Petersen, 2016; Cloutier et al., 2021) or were as follows: nlg-1 (CGAGAACCGTTGATAGTTAATGC, CAGCTACATGTGCAAGATTCAT), wntless (TCGATTGGATGGAGATGAGGT, AACTCCTTCGATGATGCCGT), inhibin-1 (TGTTACAATGTAGCAGTTGCCA, TCGTCTTTGCACTTCAAGAGGA) and tuba-2 (TCCGCATGTGTCTTTTGGAA, GCCAAATCTTCACGAGCCTC). PCR-mediated addition of T7 sequences was used to generate templates for riboprobe and dsRNA synthesis.
Animals were irradiated using a Radsource RS-2000 X-ray to deliver 6000 rads to animals in 1× Montjuic salts (Fig. 2B), a treatment that is known to eliminate stem cells (Vasquez-Doorman and Petersen, 2014), and flank incisions were performed 3 days later.
For experiments involving quantification of notum, wnt1 or follistatin cells, maximum projection images from ∼200 μm regions near the wound sites were manually scored by tabulation using Fiji's cell counting plugin. For measurements of wnt1, wnt11-2 and wntP-2 expression domains, relative lengths of gene expression domains from the posterior tip were normalized to body length in Fiji. H3P cells and animal areas were counted using an automated implementation of Analyze Particles function. Boxplots with overlayed jittered dotplots were generated using shiny.chemgrid.org/boxplotr or ggplot in R. Statistical analyses are described in each figure legends and calculated in R. For comparing multiple samples across a single variable if data were normally distributed (Shapiro's test) and of equal variance (Levene's test), one-way ANOVA followed by Tukey's post-hoc test was used. For data not normally distributed and/or having unequal variance, and in which multiple samples were compared to a common control condition, Kruskal-Wallis non-parametric tests followed by Dunnett's post-hoc tests were applied. For data not normally distributed and/or having unequal variance, and in which multiple samples were compared to each other, Kruskal-Wallis non-parametric tests followed by Dunn's post-hoc tests were applied. Comparisons with P<0.05 were considered significant.
Animals were treated with control media or 200 µg/ml colchicine for 24 h, followed by head amputation and recovery in colchicine-free media for 0, 4 and 18 h prior to isolation of total RNA from wound-proximal tissue. Samples were derived from 10 animals per biological replicate over 4 biological replicates per condition. Animal fragments were placed in Trizol then homogenized using a Turrex tissue homogenizer, followed by extraction of total RNA. cDNA libraries were prepared and sequenced by Novogene using a directional eukaryotic mRNA expression profiling protocol, which involved capture of mRNA on oligo-dT magnetic beads, reverse transcription using random hexamers, followed by second-strand synthesis, adaptor ligation and paired-end Illumina sequencing to a depth of 30 M reads per sample. Reads were mapped to the Dresden ddv6 transcriptome (https://planmine.mpinat.mpg.de/planmine/begin.do) using HISAT2, followed by differential expression analysis with DESeq2 and using the Benjamini-Hochberg method to correct P-values for false-discovery. Heatmaps were created through Rstudio to determine z-scores of log2 (FPKM+1) values of each gene across the treatments. Venn diagrams were constructed in R studio the gplots and ggplot2 packages.
We thank members of the Petersen lab, Dr S. Wignall and Dr V. Gelfand for helpful advice and discussions, and B. Stevens for help with automated H3P cell counting.
The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.204669.reviewer-comments.pdf