A flexible repertoire of wound closure strategies precedes whole-body regeneration

Abstract

Wound closure is an essential aspect of regeneration, often serving as the first morphogenetic step preceding downstream events. Animals capable of whole-body regeneration can heal virtually any wound, but how the process of re-epithelialization compares across injury types remains understudied. Here, we investigate wound closure in the acoel Hofstenia miamia. We find that H. miamia deploy two distinct modes of closure, depending on which epithelial layers are injured. Regardless of geometry, wounds that injure the outer epidermis close by constriction. In contrast, damage to both the outer epidermis and pharyngeal epithelium induces long, actin-rich protrusions that form transient bridges across the wound prior to re-epithelialization. Muscle contraction is differentially required across conditions, and pharmacological inhibition of actomyosin contractility is sufficient to shift closure dynamics across epithelia. This work compares wound closure within the same animal, identifying a wide toolkit of repair strategies that appears independent of subsequent regenerative processes.

Animals capable of whole-body regeneration can heal virtually any wound. Here they show that Hofstenia miamia accomplishes this feat through variable morphogenetic strategies that depend on which epithelial layers are damaged, revealing a versatile toolkit for repair.

Introduction

The ability of animals to regenerate varies substantially across phyla, ranging from more limited, tissue-specific or structural regeneration to whole-body regeneration (WBR), where any missing cell or tissue type can be recovered after amputation^1,2. Highly regenerative animals undergoing WBR face unique healing challenges, yet they scarlessly repair virtually any injury prior to downstream regeneration. Most mechanistic studies of wound closure have focused on relatively simple injuries (i.e., damaging only the epidermis) in animals that are not capable of substantial regeneration, generating foundational knowledge about cellular behaviors but lacking a strong connection to downstream regeneration^3–8. Work on wound closure in regenerative animals has primarily focused on either the formation and function of the ‘wound epidermis’, a specialized epidermis that covers the injury site and acts as an organizing center to induce downstream regeneration^9–12, or on large-scale transcriptional atlases detailing the molecular wound response^13–22. Despite some characterization of wound repair in regenerative animals, it remains unknown whether highly regenerative animals employ a uniform healing process regardless of injury context (i.e., major vs. minor wounds, varying geometries and tissue organizations, downstream regenerative outcomes) or if they employ a range of mechanisms tailored to different injury conditions.

Previous work has shown similar strategies for wound repair (e.g., cell crawling, actomyosin purse string formation) across the metazoan tree of life, including in early-branching lineages^14,16,23,24, protostomes^3,4,25–28, and deuterostomes^{15,18,29–31}. This suggests a deep conservation of cellular behaviors, evolving prior to the emergence of bilateral symmetry. These cellular behaviors work in tandem to drive re-epithelialization, in which a damaged epithelial layer undergoes cellular and morphological changes to re-establish an intact tissue⁶. One conserved behavior, cell crawling, is facilitated through the formation of actin-based membrane protrusions (such as lamellipodia or filopodia) in cells at the periphery of a wound that drive migration into the gap^7,32,33. Actomyosin contractility is essential in this crawling mechanism for both retraction at the rear of migrating cells and the formation of tension between adhesion points^23,34,35. Purse string closure, another process observed in multiple species, is a distinct behavior that is also driven by actomyosin contractility; cells around the wound form a supracellular actin cable, contracting and bringing the edges of cells inwards to pinch off the gap^32,36,37. These cellular mechanisms are not mutually exclusive and often work together, bringing the edges of epithelial cells together across the open wound and restoring tissue integrity through the reestablishment of cell junctions⁷. Although this toolkit is well characterized in small injuries, much less is understood about whether animals utilize the same closure strategies after extreme injuries, such as large-scale amputations.

Animals capable of whole-body regeneration quickly and routinely heal extreme injuries in the lab, recovering from amputations that remove half or even most of their bodies. It remains unclear, however: (1) if these animals use conserved or novel cellular behaviors to accomplish this feat, and (2) if there is any correlation between wound closure and extent or type of downstream regeneration within a fragment. Studies both in vivo and in vitro in established model organisms, which don't regenerate extensively, have revealed that re-epithelialization mechanisms vary substantially between wounding conditions, with factors such as wound size, geometry, tissue type, extracellular matrix composition, and life stage of the animal all influencing the mechanism of closure. Wound size and geometry are perhaps the most well-studied variables; work in these lab models has shown that small epithelial injuries typically rely on fast contractile responses (such as purse string contraction), whereas large injuries require cell migration^5,23. A handful of studies on regenerative animals have shown that different amputations lead to different dynamics of wound closure, but these were largely focused on nematic order instead of epithelial cell morphologies^38,39. Assessing how epithelial cells behave during closure of major wounds such as during whole-body regeneration, therefore offers a new approach to a fundamental topic in biology, capitalizing on animals’ regenerative properties to ask questions that are infeasible in other, less regenerative, species.

Hofstenia miamia is a marine invertebrate worm belonging to the enigmatic and understudied phylum Xenacoelomorpha, which likely represents an early-diverging animal lineage that is a sister group to all other bilaterians (Fig. 1a, b)^40–42. H. miamia has emerged as a new laboratory model system, largely studied for its ability to undergo whole-body regeneration. Like some animals capable of WBR, such as planarians, H. miamia relies on a population of adult pluripotent stem cells (sometimes called neoblasts) to drive regeneration^43,44. However, even in conditions that block regeneration (e.g., functional knockdown of required genes or irradiation of proliferative cells^40,43), damaged fragments are still able to visibly close their wounds—suggesting that healing and regeneration are distinct processes. In addition to their remarkable regenerative ability, H. miamia possess two epithelial tissues: an outer epidermis and a centrally located pharynx lined with an epithelium (Fig. 1c, d)⁴⁵. This dual-epithelial organization prompts both questions of how complex injuries can heal and how tissues maintain their identity during repair and regeneration.

Fig. 1. Hofstenia miamia has two distinct, atypical epithelial tissues.

a A juvenile Hofstenia miamia. b A simplified phylogenetic tree showing the position of Hofstenia miamia (red, Xenoacoelomorpha) relative to other metazoan clades. mya = million years ago. c Fluorescent in situ hybridization (FISH) images of the gene cuzd1-4 (outer epidermis label) and phar (pharyngeal epithelium label). d A schematic of major cell types in H. miamia, showing the organization of the epidermis, pharyngeal epithelium, muscle, adult pluripotent stem cells, and syncytial gut. e EpiA:RFP transgenic worms, labeling the outer epidermis and the pharyngeal epithelium in a mosaic manner. f Morphology of individual epidermal and pharyngeal (Ph.) epithelial cells, visualized by EpiA:RFP transgenic labeling. A: apical, B: basal sides of cells. Asterisks indicate basal protrusions. SiR-actin staining showing the apical organization of H. miamia epithelial tissues, particularly the presence of actin within epidermal basal protrusions (g), the polygonal shape of each cell (h), and the presence of actin at cell-cell junctions (h). i Confocal images showing the proximity of muscle (visualized through immunofluorescence staining for tropomyosin) to both epithelial layers (visualized with EpiA:RFP transgenic labeling). j Quantification of basal protrusions within epidermal and pharyngeal epithelial cells. Data represent n = 45 cells per cell type, pooled from 15 worms. All images are representative of at least 3 independent experiments. Comparison by unpaired two-tailed t-test. Scale bars: 50 µm (a, c), 100 µm (e), 5 µm (f), 20 µm (h, i), 25 µm (g), 10 µm (i, zoomed-in box). Raw values are provided within Source Data file.

In this study, we investigate the healing capacity of H. miamia, examining whether wound repair mechanisms vary across different injury conditions. In relatively simple injuries (e.g., transverse amputations that only damage one epithelial layer, or scratch wounds), animals close their wounds through contraction of the outer epidermis. In more complex wounds (e.g., sagittal and transverse amputations that damage both epithelial layers), H. miamia form heterotypic bridges between epithelial layers prior to homotypic re-epithelialization of each layer. Interestingly, we found that the mode of closure is dependent on whether one or both epithelial layers are damaged, not on the size of the wound or the downstream regenerative response. The heterotypic bridges seen in dual-epithelial amputations are dependent on both global actomyosin contractility and on contraction of neighboring muscle, but regenerating fragments can shift their cellular processes in response to chemical perturbation. H. miamia therefore possesses an adaptable toolkit of wound closure mechanisms, with flexible strategies to ensure successful repair and downstream regeneration regardless of injury type.

Results

Hofstenia miamia has two distinct, atypical epithelial tissues

The acoel Hofstenia miamia is a marine worm that can recover from virtually any injury throughout its lifespan, scarlessly healing wounds and rebuilding lost tissues (Fig. 1a, b)^40,41. Despite the closing of wounds being an essential prerequisite to this extensive regeneration, the mechanisms of wound closure have not yet been studied in H. miamia. To fully understand the dynamics of re-epithelization, we first needed to characterize the epithelial layers of H. miamia.

H. miamia has two major epithelial tissues, distinct in molecular profiles and anatomical location: (1) an outer epidermis that covers the whole animal (Fig. 1c, left), and (2) the inner lining of the pharynx, a tube-shaped organ that extends posteriorly from the anterior mouth, spanning approximately half the length of the animal (Fig. 1c, right)⁴⁵. These epithelial tissues meet at the mouth, but are otherwise separated by internal tissues and cell types—including (but not limited to) an elaborate musculature, pluripotent stem cell population, and syncytial gut (Fig. 1d)⁴⁵. Although the dynamics of these epithelial cells have not yet been studied, previous transgenic labeling of epithelial layers has revealed their cellular morphologies⁴⁵. This transgenic labeling (using EpiA:RFP animals⁴⁵) enables the visualization of detailed cell morphologies (Fig. 1e, f), but mosaicism prevents consistent imaging across animals. We found that the cytoskeletal dye SiR-actin robustly labels both layers of epithelial cells in fixed animals and can therefore be used to visualize these tissues (Fig. 1g, h). The apical side of H. miamia epithelial cells resembles epithelia in other systems⁴⁶, with the cells organized in a polygonal pattern on the surface of the animal (Fig. 1h). As previously reported, the basal sides of both epidermal and pharyngeal epithelial cells form cytoplasmic protrusions that extend internally into the body of the worm (Fig. 1f, g, i). The protrusions are irregularly spaced and heterogeneous in number between cells but extend approximately 10–25 µm in length, with the epidermal protrusions being significantly longer than those in the pharyngeal epithelium (Fig. 1j). The basal protrusions in both cell types end in close proximity to the underlying muscle (Fig. 1i), but it is unclear whether cell-cell junctions form at the tips of these projections or whether they adhere to a substrate.

H. miamia can recover from virtually any injury configuration. Different injuries have distinct starting points and thus have different downstream regenerative outcomes; for example, upon transverse amputation, head fragments rebuild tails, whereas tail fragments will rebuild heads. Do these wounds in head and tail fragments close in the same way prior to their distinct regenerative outcomes? The central location of the pharynx also enables transverse amputations that can damage both epithelial layers, raising questions of how H. miamia coordinates epithelialization of multiple layers. Re-epithelialization is typically studied in wounds that only damage one type of epithelium—do the two epithelia interact with each other (in a heterotypic manner) or only with cells of their own layer (through homotypic interactions)? We sought to answer these questions and probe the dynamics of re-epithelialization after various types of injuries.

Amputations that damage the outer epidermal layer close through actin-based constriction

Transverse amputations through the middle of the animal, separating heads from tails, are a commonly used assay for studying whole-body regeneration^{14,40,41,43,44}. In H. miamia, a typical transverse amputation aims to cut at the middle pigmentation stripe observable on the dorsal surface of the worm, leaving behind a large gap with open epidermal edges around the wound. This amputation halves some tissues (e.g., gut, nerve net, and musculature) and completely removes anterior organs (e.g., pharynx and brain) from the posterior fragment. As a result, the two resulting fragments (heads vs tails) have different topologies and behaviors (e.g., head fragments with an intact brain swim robustly, whereas tail fragments, which now lack brains, glide only through ciliary movement⁴⁷). Additionally, these fragments face different regenerative challenges: the head fragment needs to regenerate posterior tissues, whereas the tail fragment must regenerate anterior-specific tissues and organs (Fig. 2a).

Fig. 2. Transverse amputations close through gradual constriction of the epidermis across the wound edge, regardless of downstream regenerative outcomes.

Schematic (a) and representative brightfield time course (b) showing regeneration of H. miamia fragments following transverse amputation. White arrowheads indicate the formation of a blastema (unpigmented outgrowth). c Representative images of adult pluripotent stem cells (fluorescent in situ hybridization (piwi-1, left)) and mitotic cells (immunofluorescence (pH3, right)) in uncut animals. Representative images (d) and quantification (e) of pH3+ cells during regeneration following transverse amputation. f Relative density of mitotic cells following transverse amputation, showing the accumulation of mitotic cells at the anterior end of tail fragments but not head fragments. Uncut animals are indicated with a black line, head fragments are shown in red, and tail fragments are shown in blue. g Representative images of wound closure following transverse amputation between 0 and 20 h post-amputation. Red asterisk indicates the base of the pharynx, visible in open head wounds. Yellow arrowheads indicate the actin-dense ring around the wound site, red arrowheads indicate sites of actin density at epithelial “folds”. h Quantification of wound closure following transverse amputation, comparing head and tail fragments from 3 to 10 h post-amputation. i High-magnification images showing continued dynamics of re-epithelialization following initial closure, specifically the smoothing of actin-dense folds between 10 and 20 h post-amputation in both head and tail fragments. hpa = hours post-amputation. Data represent n = 8 worms per time point, and comparisons were performed by 1-way (e) or 2-way (h) ANOVA with Šidák’s multiple comparisons test. All images are representative of at least 3 independent experiments. Error bars represent Mean ± SD in all graphs. Scale bars: 50 µm (b, i), 100 µm (c, d, g). Raw values are provided within Source Data file.

We first amputated juvenile worms transversely below the pharynx, so that only the outer epidermis (but not the pharyngeal epithelium) would be impacted by the injury. As has been previously reported^40,43, each fragment showed a rounded wound edge within 24 h, suggesting successful closure. Head fragments formed a pointed tail within 3 days and regained their original shape within 5 days. Tail fragments took longer to regenerate but formed a blastema (here defined as an unpigmented outgrowth) within 5 days (white arrowhead) and largely regained their original form within 10 days (Fig. 2b). We reasoned that this temporal difference in regeneration may be due in part to the distribution of stem cells; the pluripotent stem cell population that drives regeneration is scattered through the parenchyma of the animal but is most concentrated in the midsection of the animal (Fig. 2c)^40,43. To compare the growth dynamics of regenerating fragments, we performed immunofluorescence for pH3, a known marker of mitotic cells, at multiple time points post-amputation. aPSCs are the only mitotic cells within H. miamia, allowing an indirect readout of aPSC activation⁴¹. We observed distinct differences in proliferation between head and tail fragments. Head fragments showed peak levels of mitotic cells at 1 day post-amputation (dpa) before returning to homeostatic levels by 3 dpa (Fig. 2d, e). In contrast, tail fragments showed a consistent increase in proliferative cells across the first 7 days, only returning to homeostatic levels at 10 dpa (Fig. 2d, e).

Quantifying the spatial distribution of proliferative cells along the anterior-posterior axis, we saw that head fragments maintain a distribution of proliferative cells across their body axis similar to that seen in uncut animals, but tail fragments show an accumulation of mitotic cells at the anterior end (where the new head will be formed) between 3–7 dpa before returning to a near-homeostatic distribution around 10 dpa (Fig. 2f). These trajectories reflect the regenerative challenge each fragment faces; tail fragments must replace more cell types than head fragments, and the visible blastema seen in brightfield images (Fig. 2b) presumably arises from the concentration of mitotic cells and their progeny at the wound site. Studies in other whole-body regeneration model systems have shown that the downstream regenerative response is dictated by actin organization and topological asymmetries formed during wound closure^38,39, and we therefore asked whether head and tail fragments in H. miamia showed differences in re-epithelialization.

To track the dynamics of the outer epidermis during wound closure, we stained fragments with the cytoskeletal dye SiR-actin at intermediate time points during the first 20 h post-amputation (hpa) (Fig. 2g). Most wounds from both head and tail fragments closed within 10 hpa, with both showing a strong anisotropic enrichment of actin at the wound periphery (Fig. 2g, h). These observations are consistent with a process of radial constriction, where actin polarizes towards the wound edge and cells collectively pull towards the center of the wound to minimize the size of the gap. Some wound edges displayed a smooth, continuous actin cable (yellow arrowheads), whereas others appeared puckered (red arrowhead), suggesting the formation of contractile actomyosin purse-strings^32,37. Tail wounds showed faster closure in the early stages of wound closure (3 hpa), but by 6 hpa there was no difference between head and tail fragments (Fig. 2g, h). Despite closure of wounds by 10 hpa (i.e., no open space between epithelial cells), we observed visible differences in epidermal topology between 10 and 20 hpa. Closed wounds at 10 hpa had visible seams and concentration of actin where the epidermis joined, but by 20 hpa there was no detectable seam or wound edge (Fig. 2i). This suggests a sustained re-epithelialization process even after initial gap closure, where the epidermal cells continue to remodel their junctions to re-establish a smooth epithelium.

To assess whether minor wounds utilize similar mechanisms of contractility during re-epithelialization as seen in transversely amputated wounds, we performed “pinch” wounds, where fine-tipped forceps were used to pinch through the outer epidermis (Supplementary Fig 1a). These wounds closed faster than wounds in transverse amputations; we observed the formation of polarized protrusions within 5 min post-amputation, with the wound edges making contact and closing the open wound within 1 h (Supplementary Fig. 1a). This is likely due to the much smaller distance between wounded epidermal edges; in a pinch wound, the edges remain adjacent to each other, facilitating quick re-epithelialization, but a transverse amputation causes a comparatively large wound gap that requires more time to close.

Together, these results show that wounds that disrupt only one epithelial layer achieve closure through actin-based contraction. Minor wounds with small gaps (e.g., pinch wounds) can close quickly and without the formation of supracellular structures, whereas large amputations require the constriction of epithelial cells to minimize the wound gap. Importantly, we did not observe a correlation between wound closure and downstream regenerative response when comparing head and tail fragments. Both head and tail fragments were able to re-establish an intact epidermis over the amputation site within 20 h post-amputation, utilizing a similar mechanism despite significant differences in downstream regenerative outcomes.

Heterotypic epithelial bridges mediate the first step in multi-epithelial wound closure

To investigate wound closure involving multiple epithelial layers, we performed injuries that damage both the outer epidermis and the pharyngeal epithelium. We first cut animals in the sagittal plane, separating the left half from the right, and imaged fragments over a 10-day time course of regeneration (Fig. 3a, b). Interestingly, these animals were able to largely restore their original shape within 5 days, closing up their wounds without the formation of a visible blastema (Fig. 3b). The level of mitotic cells in sagittal fragments peaked at 1 dpa before significantly decreasing by 3 dpa (Fig. 3c, d), similar to the proliferation dynamics seen in head fragments (Fig. 2d, e). The spatial distribution of pH3+ cells along the anterior-posterior axis also stayed consistent despite changes in cell density (Fig. 3e), further suggesting that sagittal fragments regenerate without forming a blastema.

Fig. 3. Sagittal wounds that damage both epithelial tissues form heterotypic interactions across the wound edge.

Schematic (a) and representative brightfield time course (b) showing regeneration of H. miamia following sagittal amputation. Representative images (c) and quantification (d) of pH3+ cells during sagittal regeneration. Data represent n = 8 animals per time point. Error bars represent Mean ± SD. e Relative density of mitotic cells following sagittal amputation, showing consistent distribution along the anterior-posterior axis regardless of time point and cell density. f Representative images of wound closure following sagittal amputation between 0 and 10 h post-amputation. g High-magnification images showing polarization, protrusion, and formation of heterotypic bridges during the first 10 h of wound closure. h EpiA:RFP animals showing fluorescent signal in cells forming protrusions at 6 hpa. Quantifications of heterotypic bridge formation (i), origin of protrusions (j), and combined bridge length (k) over the first 10 hpa. Data represent n = 8 worms per condition and time point (h–i) and 30–40 bridges per time point (j); comparison by unpaired t-test. l Schematic showing the stages of heterotypic bridge formation. In the schematic, the epidermis is labeled with blue and the pharyngeal epithelium is labeled with red. In all high-magnification images, the outer epidermis is pseudocolored in blue and the pharyngeal epithelium is pseudocolored in red for clarity. All images are representative of at least 3 independent experiments. Scale bars: 50 µm (b, f, h (left)), 100 µm (d), 10 µm (f (zoomed-in boxes), h (right)), 5 µm (g). Raw values are provided within Source Data file.

We next asked how re-epithelialization occurs in sagittal-cut fragments. Following amputation in the sagittal plane, SiR-actin labeling revealed the formation of striking, long cytoplasmic protrusions that extend across the wound edge in the first 10 hpa (Fig. 3f). These protrusions originated from both the outer epidermis and the pharyngeal epithelium, connecting across the middle of the wound (Fig. 3f, g). To confirm that these cells were indeed epithelial cells of the epidermis and pharynx, we utilized the EpiA:RFP transgenic line⁴⁵ and detected protrusions in labeled cells within both the outer epidermis and the pharyngeal epithelium (Fig. 3h). We were unable to visualize the basal side of cells forming these bridges, but the extensions protruded from the apical surface of the cells—suggesting that these were new structures and not movement of the homeostatic basal protrusions.

Epidermal cells showed actin polarization within the first 3 h post-amputation (hpa), with small filopodial protrusions (5–10 µm) appearing in the outer epidermis oriented towards the wound edge (Fig. 3f, g). Long, thin extensions (>15 µm) emerged around 6 hpa, increasing in number by 10 hpa and connecting to projections originating from the pharyngeal epithelium (Fig. 3g–i). This temporal pattern was consistent across animals, and 67% of fragments displayed at least one connected bridge within 10 hpa (Fig. 3i). Notably, protrusions were always detected first in the outer epidermis, and protrusions from the pharyngeal epithelium were only visible after protrusions formed in the epidermis (Fig. 3j). Corresponding with the number of bridges increasing between 6 and 10 hpa, the average bridge length increased significantly between time points (Fig. 3k). This stepwise process (polarization, extension, connection) is consistent with a model of collective cell migration^48,49 (Fig. 3l), but the lengths of these bridges (~30–100 µm) are reminiscent of specialized projections such as cytonemes or tunneling nanotubes (which have been reported to reach up to 200 µm in vivo^50–52). To the best of our knowledge, structures like these heterotypic bridges have not been reported in wound closure literature, and we therefore sought to further investigate this phenomenon.

The difference between transverse, contraction-mediated closure and sagittal, bridge-mediated closure could be attributed to a variety of differences between the amputations other than the dual epithelial layers, including (but not limited to): angle or orientation of amputation, size of the wound, or the extent of regeneration required downstream of closure. We next asked whether heterotypic bridges occurred in transverse amputations further up the anterior-posterior axis of the animal, after injuries that impact both epithelial layers by cutting through the pharynx (Supplementary Fig. 2a). These wounds require closure in the same plane as the transverse injuries previously reported (Fig. 2), but now include damage to both the epidermis and the pharyngeal epithelium. Single transverse amputations through the pharynx led to the formation of unpigmented outgrowths in both head and tail fragments (Supplementary Fig. 2a), a phenotype consistent with spatiotemporal dynamics of proliferative cells across time points (Supplementary Fig. 2b–d). Animals amputated twice through the pharynx in the transverse plane (“donut” amputations) showed similar behaviors in both brightfield images and pH3 immunofluorescence (Supplementary Fig. 2e–h). We stained 10 hpa fragments from both types of amputations with SiR-actin and observed robust heterotypic bridges between the epidermis and the pharyngeal epithelium, similar to those seen in sagittal amputations, in all fragments (Supplementary Fig. 2i, j) (Fig. 3f, g). This bridge-based mechanism of closure is therefore not specific to the size, geometry, or regenerative response associated with the sagittal amputation, but appears instead to be a mechanism by which to close wounds that involve multiple epithelial layers.

Next, we reasoned that there must be a substrate below these protrusions acting as a physical track for their extension. The extracellular matrix (ECM) is a key regulator of wound healing, acting as a dynamic substrate on which cells can crawl and extend to complete re-epithelialization^53–55. Given the known role of collagen IV in wound closure^55–58, we identified a putative ortholog in the H. miamia transcriptome and made a custom antibody (see Methods). Using this antibody, we performed immunofluorescence staining to assess the localization of collagen IV and observed its expression solely within the pharyngeal epithelium (Supplementary Fig. 3a, b). These experiments revealed punctate expression overlapping with pharyngeal cells, but it is unclear whether the antibody is binding to pre-processed collagen within cells or if nascent, extracellular collagen simply appears punctate prior to organization. Although we cannot make firm conclusions about the localization of collagen IV, our data suggest that the pharyngeal epithelium is likely a source of new collagen.

We next asked whether collagen IV expression changed during wound closure, as heterotypic bridges are forming. We performed immunofluorescence on 10 hpa sagittal fragments and observed strong collagen IV signal within pharyngeal cells, including along wound-facing extensions (Supplementary Fig. 3c). Collagen puncta were never detected in areas without protrusions or within the outer epidermis. The same pattern was observed after transverse amputation through the pharynx (Supplementary Fig. 3d). These data implicate collagen IV as a potential substrate for epithelial bridges, possibly providing physical support for protrusions as they extend across the wound site. Although our antibody does not show localization of collagen IV underneath or within epidermal cells, we speculate that a different ECM protein may be present underneath those cells (see Discussion).

Taken together, these findings reveal the presence of heterotypic cellular bridges between epithelial layers during wound closure, which emerge within the first 10 h post-amputation. These bridges are observed across different amputation geometries that injure multiple epithelial layers, and tissue-specific collagen IV is detected within pharyngeal protrusions as they extend across the wound gap. We therefore hypothesize that these heterotypic interactions act as an intermediate step in large-scale multi-epithelial wound closure, possibly minimizing tissue loss and/or maintaining tissue identity prior to full re-epithelialization. We next asked how these heterotypic interactions relate to the rest of the wound healing process.

Wound closure involving multiple epithelial tissues utilizes distinct but temporally overlapping cellular processes

The heterotypic bridges described above take place in the early phase of wound closure, emerging within the first 10 h post-amputation (hpa). For the animal to heal fully, sagittal fragments must undergo continued remodeling to re-establish their original shape (Fig. 3b). Our regeneration time courses demonstrated that sagittal amputations in H. miamia do not form an obvious unpigmented outgrowth or localized accumulation of mitotic cells, suggesting that cells may be rearranging differently than we observed in transverse injuries (Fig. 3b–e). To gain further insight into this morphogenetic process, we stained sagittal fragments with SiR-actin over the course of 70 h post-amputation (hpa) (Fig. 4a). We observed that these animals healed through a two-step process: after the outer epidermis connects to the pharyngeal epithelium through heterotypic bridges (as shown in Fig. 3), the posterior end re-epithelializes by reconnecting the outer epidermis to itself (Fig. 4a).

Fig. 4. Sagittal wounds heal through a two-step process, combining epithelialization with whole-body rounding to restore the fragment’s original morphology.

a Schematic and time course showing epithelial tissue dynamics in sagittal wound closure during the first 70 h post-amputation (hpa), using SiR-actin labeling. Red and blue dashed outlines indicate the pharyngeal epithelium and epidermis, respectively. b Time course showing re-epithelialization of the pharyngeal epithelium, using fluorescent in situ hybridization (FISH) for the gene phar. Red dashed outline represents the wound edge of the pharynx. c Higher-magnification images demonstrating the overlapping mechanisms of epithelialization in anterior and posterior parts of the fragment over 70 hpa. Anterior tissues form heterotypic protrusions described in Fig. 3, and the posterior wound edges gradually close through homotypic re-epithelialization. Tissues are pseudocolored in red (pharyngeal epithelium) and blue (epidermis) for clarity. Quantifications of closure mechanisms (d) and fragment rounding (e). f Schematic showing temporal events in sagittal wound closure. In the schematic, the epidermis is labeled with blue and the pharyngeal epithelium is labeled with red. Data represent n = 12 animals per time point. All images are representative of at least 3 independent experiments. hpa = hours post-amputation. Scale bars: 100 µm (a), 20 µm (c). Raw values are provided within Source Data file.

As the posterior end is re-epithelializing, sagittal-cut animals round their bodies up to re-form their cylindrical shapes (and presumably re-establish their disrupted dorsoventral and mediolateral axes). To more clearly visualize the morphology of the pharynx during this process, we performed fluorescent in situ hybridization for the gene phar, a known marker of the pharyngeal epithelium⁵⁹, during sagittal wound closure. We saw rounding of the pharynx concurrent with the rounding of the outer epidermis (Fig. 4a), with the two open pharyngeal edges contacting each other to re-form a tube by approximately 70 hpa (Fig. 4b). The two main closure processes we observed (heterotypic bridges and homotypic re-epithelialization) occur in a stepwise but overlapping sequence, where all fragments first form visible bridges, then display both bridges and posterior closure, and finally end with posterior re-epithelialization alone (Fig. 4c, d). All animals had an open body between 0 and 10 hpa, characterized by a splayed pharynx and separated dorsal and ventral halves. By 20 hpa we began to see rounding of the fragments, with the outer epidermal edges curling over the pharynx towards each other. While some animals closed as early as 30 hpa, virtually all animals completed this closure by 70 hpa (Fig. 4c–f). Here, closure is defined by the original dorsal and ventral edges completely meeting at the seam of the wound, forming a cylinder with the pharynx on the inside and the epidermis on the outside. We speculate that this rounding process is aiding the posterior re-epithelialization by physically bringing the wound edges closer together.

To assess whether this dual closure process (i.e., heterotypic bridges followed by homotypic re-epithelialization) (Fig. 4f) was unique to the sagittal amputation, we next asked whether similar processes occurred in transverse injuries that form heterotypic bridges. We amputated animals both once (single transverse) and twice (donut) through the pharynx and stained fragments with SiR-actin at intermediate time points (Supplementary Fig. 4a, b). Unlike the sagittal fragments, the transverse fragments did not show extensive body rounding after the formation of bridges, likely due to the fact that all main axes (anterior-posterior, dorsal-ventral, and medio-lateral) remain intact after transverse amputation. However, both amputations showed a similar two-step process: heterotypic bridges form between epithelial layers by 10 hpa, followed by homotypic re-epithelialization of the outer epidermis over the wound site. Interestingly, the re-epithelialization of the outer epidermis appeared to utilize purse string-like radial constriction, a process more similar to the transverse amputations in Fig. 2 than the sagittal fragments in Fig. 4, suggesting that fragment or wound geometry might influence the directionality of homotypic closure (Supplementary Fig. 4a, b).

Given the extent of cellular movements taking place during this healing process, we asked whether the animals’ pluripotent stem cells (aPSCs) were necessary. Previous work has shown that, although necessary for regeneration, aPSCs are dispensable for visible wound healing in transverse amputations⁴⁰. We irradiated animals to eliminate the aPSC population (see Methods) and amputated animals in the sagittal plane, collecting fragments between 10–70 hpa (Supplementary Fig. 5). Irradiated animals formed heterotypic bridges at the same rate as control animals but were unable to complete comparable homotypic epithelialization or body rounding by 70 hpa (Supplementary Fig. 5a–d). We observed increased spreading of the pharyngeal epithelial cells at 70 hpa in 67% of animals (Supplementary Fig. 5e), although the exact reason for this phenotype is unclear. Stem cells are therefore necessary for fragment rounding and homotypic re-epithelialization, but are not required for the formation of heterotypic bridges. We next asked how cell and tissue contractility was involved in these mechanisms of wound closure, seeking to combine the characterization of these processes with functional perturbations.

Muscle contraction is necessary for both heterotypic wound closure and downstream morphogenetic movements

Contraction of muscle at the wound site is an essential component of wound closure in both vertebrates and invertebrates, and drives physical and biochemical events through the generation of mechanical force^27,30,60,61. Because of the proximity of epithelial cells to the underlying muscle in H. miamia (Fig. 1j), we next asked whether muscle contraction was involved in wound closure. H. miamia have an extensive network of muscle (Fig. 5a), and we utilized Tropomyosin immunofluorescence⁶² to visualize how muscle was reorganized during single and dual epithelial wound closure. In transversely-amputated fragments (both head and tail), muscle was largely absent from the wound site itself but was instead present at the periphery of the wound edge as it closed (Fig. 5b). In contrast, sagittal wound sites contained parenchymal muscle fibers that link the outer epidermis to the pharyngeal epithelium, coming together as the animal rounds up (Fig. 5c). We hypothesized that this difference in muscle organization might influence the closure mechanisms used by each fragment, and asked whether muscle contraction was required for each type of wound closure.

Fig. 5. Muscle contraction is dispensable for traditional transverse amputations but required for the formation of heterotypic bridges during sagittal wound closure.

Images showing the organization of muscle in uncut worms (a), transverse wound closure (b), and sagittal wound closure (c), visualized by immunofluorescence staining for tropomyosin. Representative images (d) and quantification (e) of wound closure in control and tricaine-treated animals after transverse amputation, labeled with SiR-actin. Error bars represent Mean ± SD. f Representative images of wound closure in control and tricaine-treated animals after sagittal amputation, labeled with SiR-actin. The outer epidermis is pseudocolored in blue, and the pharyngeal epithelium is pseudocolored in red for clarity. Quantification of closure mechanism (g) and fragment rounding (h) in control and tricaine-treated animals after sagittal amputation. Data represent n = 12 animals per condition, per time point; quantification in (e) by unpaired two-tailed t-test with Welch’s correction. ns not significant, hpa hours post amputation. All images are representative of at least 3 independent experiments. Scale bars: 100 µm (a, c), 50 µm (b, d, f). Raw values are provided within Source Data file.

To test the functional role of muscle contraction, we anesthetized freshly-amputated animals in 0.1% tricaine, which blocks neural action potentials⁶³ and effectively inhibits muscle contraction in H. miamia⁶⁴. When kept in tricaine, single epithelial wound closure in transverse amputations occurs at the same rate as controls (Fig. 5d–e), indicating that the epidermis is capable of constricting and re-epithelializing without the aid of muscle contraction. Animals with multi-epithelial wounds, on the other hand, did not form heterotypic bridges when kept in tricaine (Fig. 5f–h). We observed actin polarization and formation of small protrusions from the epidermis and pharyngeal epithelium, but no bridges ever formed in tricaine-treated animals (Fig. 5f). The requirement of muscle contraction for the formation of heterotypic bridges was consistent across all multi-epithelial injuries (Supplementary Fig. 6a, b).

Upon closer analysis, however, we found that muscle contraction is required for the participation of the outer epidermis in closure—but not the pharyngeal epithelium. In the presence of tricaine, pharyngeal epithelial cells still formed extensions, spreading to independently reach across the wound gap (Supplementary Fig. 6a–d). This extension was more obvious in transverse wounds than in sagittal wounds, with protrusions often seeming to reach the adjacent epidermis but never succeeding in covering the wound entirely (Supplementary Fig. 6a, b). While the exact mechanism is not yet clear, these data implicate muscle contraction as a necessary driver of multi-epithelial wound closure. This contraction may be involved in generating tension, inducing mechanical or biochemical signals, and/or physically bringing the wound edges together to allow contact between the epithelia. Regardless, without muscle-generated mechanical force, the outer epidermis cannot generate or maintain long protrusions to form heterotypic bridges, and the wound closure process stalls and collapses (Supplementary Fig. 6e).

We next performed time courses with tricaine treatment to determine whether muscle contraction was only necessary for the initiation of the heterotypic bridges or whether it was needed for both their formation and maintenance (Supplementary Fig. 6f). Animals treated with tricaine for the first 10 h following amputation, then placed in seawater for the next 20 h, closed wounds at a similar rate to controls. This indicates that there is no critical window for protrusions to form after injury. Injured animals can fully recover after tricaine-based anesthetization, re-starting the wound closure process even after a 10-h delay in initiation. In contrast, animals that were in seawater for the first 10 h post-amputation and then placed in tricaine for the next 20 h were indistinguishable from animals treated with tricaine for all 30 h (Supplementary Fig. 6g, h). This indicates that persistent muscle contraction is necessary for the maintenance of heterotypic bridges.

Together, these findings highlight the importance of muscle contractility in mediating cellular events during multi-epithelial wound closure. To complement our data regarding the role of tissue-scale mechanical forces generated by muscle, we next assessed the role of actomyosin contractility in protrusive behaviors and overall wound closure.

Types of wound closure are differentially regulated by actomyosin contractility

Actomyosin contractility is generated through the interactions of non-muscle myosin with actin, producing mechanical forces within cells that enable cytoskeletal rearrangements and cellular movements^65–67. While actomyosin contractility is broadly implicated in wound closure across species, we wanted to know how its role varied across injury types in H. miamia. We treated amputated animals with the inhibitor blebbistatin, which inhibits the ATPase activity of myosin and blocks actomyosin contractility^67,68. In the presence of blebbistatin, transverse wounds cut below the pharynx (which only disrupt the outer epidermis) closed at comparable speeds to control conditions but appeared to show different morphologies (Fig. 6a–d). Instead of the formation of contractile purse strings seen in control fragments, we saw short protrusions emerging from cells around the wound edge (Fig. 6c, d). These protrusions more closely resembled filopodia than the long bridges seen in control conditions. Despite this switch in morphology, all fragments successfully closed their wounds, indicating that protrusion-based mechanisms can accomplish wound closure in blebbistatin-treated conditions.

Fig. 6. Actomyosin contractility is differentially utilized across wound closure mechanisms.

Representative images (a) and quantification (b) of wound closure in control and blebbistatin-treated animals after transverse amputation, labeled with SiR-actin. Error bars represent Mean ± SD. c High-magnification images showing the difference between control and blebbistatin-treated wound edges in transverse amputations. Asterisks indicate blebbistatin-induced protrusions within the epidermis. d Schematic showing transverse epidermal closure dynamics in control and blebbistatin-treated conditions. e Representative images of wound closure in control and blebbistatin-treated animals after sagittal amputation. f High-magnification images showing heterotypic bridges in control animals and the lack thereof in blebbistatin-treated animals. Quantification of closure mechanism (g) and fragment rounding (h) in control and blebbistatin-treated animals after sagittal amputation. i Representative images of control and blebbistatin-treated animals after 10, 30, and 50 hpa. j Schematic showing sagittal closure dynamics in control and blebbistatin-treated conditions. In high-magnification images, the outer epidermis is pseudocolored in blue, and the pharyngeal epithelium is pseudocolored in red for clarity. Data represent n = 16 (b) or 12 (g, h) animals per condition, per time point; comparison in (b) by unpaired two-tailed t-test with Welch’s correction. ns not significant, hpa hours post amputation. Scale bars: 50 µm (a, e), 10 µm (c, f), 100 µm (i), 20 µm (i, zoomed-in box). All images are representative of at least 3 independent experiments. Raw values are provided within Source Data file.

We next asked how actomyosin contractility was involved in multi-epithelial wound closure. In contrast to the mild phenotype in single epithelial wounds, in which epithelial cells were able to utilize a different process to accomplish closure, we observed significant closure defects in the blebbistatin-treated sagittal fragments. Blebbistatin treatment eliminated the formation of stable heterotypic bridges in sagittal fragments at 10 hpa (Fig. 6e). We observed short protrusions emerging from both the outer epidermis and pharyngeal epithelium, forming multiple protrusions rather than connected bridges (Fig. 6f and Supplementary Fig. 7a). These data indicate that actomyosin contractility is essential for heterotypic bridges, although we cannot currently distinguish whether this phenotype is due to an inability to initially extend protrusions or due to a failure to maintain connections. The formation of bridges must require the stabilization of protrusions as they extend, the application of pulling forces to bring the epithelia together, and/or the stabilization of nascent adhesions at heterotypic interfaces—all behaviors that have been linked to actomyosin contractility^{35,65,69–72}. Without this contractility, small filopodia-like protrusions can form, but they are ultimately unable to form bridges.

Contrasting with the inhibition of heterotypic bridges (which act as the first stage of sagittal wound closure), we observed an acceleration of the second stage of closure (the homotypic re-epithelialization) in blebbistatin-treated sagittal fragments. In control conditions, re-epithelialization of the posterior end began around ~20–30 h after amputation, after the formation of heterotypic bridges at the anterior end (Fig. 6g). In contrast, animals treated with blebbistatin showed posterior closure as early as 10 hpa, even in the absence of heterotypic bridges (Fig. 6e–g). By 30 hpa, blebbistatin-treated animals showed robust closure of the outer epidermis, but no heterotypic bridges and no rounding of the pharyngeal epithelium were observed. Prolonged treatment with blebbistatin for 50 h, however, led to reduced tissue integrity in the epidermis, manifesting first as holes between cells and leading to eventual epidermal ruptures (Fig. 6i). This phenotype is consistent with that seen in other systems, in which the reduction of junctional tension loosens the epithelial sheet and allows epidermal cells to move more freely relative to each other^30,69,73. The hypothesized decrease in junctional tension that enables accelerated cell spreading may also prevent the maintenance of strong junctions between cells, resulting in a weakened epidermal layer that cannot withstand the mechanical demands of healing and regeneration.

Lastly, we assessed whether this phenotype was reversible and if the epidermis could re-establish its integrity after accelerated re-epithelialization. We placed blebbistatin-treated animals in fresh sea water after a 30-hour incubation for either 20 (50 hpa time point) or 40 (70 hpa time point) hours, comparing directly to control animals and animals left in blebbistatin for the full time course (Supplementary Fig. 7b). Animals that were placed in fresh ASW for 20 h were rounder than their counterparts that remained in blebbistatin, suggesting a re-entry into the morphogenetic trajectory for full wound closure (Supplementary Fig. 5c, d). Animals that were placed in fresh ASW for 40 h were able to completely recover, resembling the control animals at 70 hpa (Supplementary Fig. 7c, d). This suggests that while prolonged blebbistatin treatment leads to a loss of epidermal integrity, animals can restore epidermal function after removal of the drug and continue the wound closure process. This wash-out data, paired with the similar tricaine wash-out experiments (Supplementary Fig. 6c–e) suggest that injured fragments do not possess critical windows for wound closure. Once animals have recovered from the contractile perturbation, the cells are able to sense the injury and restore their normal behaviors to fully close the wound.

Discussion

In this study, we demonstrate that the acoel Hofstenia miamia uses context-dependent wound healing, employing complementary and overlapping mechanisms of re-epithelialization to repair wounds prior to whole-body regeneration. H. miamia have two distinct epithelial layers (an outer epidermis and a pharyngeal epithelium), and we found that the type of wound closure largely depended on which epithelial layers were damaged. Wounds disrupting the outer epidermis (but not the pharyngeal epithelium) use rapid and contractile closure processes, similar to wound closure in other species. In normal conditions, these wounds primarily rely on radial constriction of the epidermis, but in conditions of reduced actomyosin contractility, the epidermal cells shift morphologies and close through protrusions reminiscent of those seen in cell crawling⁵. Multi-epithelial wounds close through a combination of heterotypic bridges and homotypic re-epithelialization, requiring both actomyosin contractility and muscle contraction to achieve successful closure. In conditions of reduced actomyosin contractility, other closure mechanisms are transiently accelerated, but ultimately fail to restore long-term tissue integrity unless contractility is restored. Mechanisms of wound closure in H. miamia are therefore not uniform or discrete processes, but instead seem to be dynamic and cooperative, shifting in response to both intrinsic and extrinsic factors.

The heterotypic bridges described here are, to the best of our knowledge, an atypical wound-induced response. Collective migration of epithelial sheets is a well-characterized phenomenon in homotypic wound closure; leader cells at the wound edge typically polarize and extend protrusions towards the wound, “tugging” on neighboring cells to collectively migrate and close the gap⁴⁸. The cellular phenotypes we see in H. miamia are consistent with this mechanism. The transient heterotypic bridges we observe, however, present a striking variation of epithelial contact. Protrusions from both epithelial layers extend across the gap and seem to contact each other, an interface that does not occur in an uninjured context aside from the mouth. Unlike homotypic wound closure, in which cells integrate into a continuous tissue made up of the same cell type, these bridges are temporary—suggesting a different functional role rather than true re-epithelialization. Heterotypic contacts have been described in other systems as mechanisms by which tissue boundaries are maintained during development^73–77, but the filopodial matching that has been observed⁷⁷ appears distinct from these bridges both in size and duration. Instead of fast filopodial probing between proximal cells to form the proper adhesive complexes^3,78, the bridges observed in H. miamia are long (>50 µm) and visible for at least 10 h. At present, we cannot rule out the possibility that these bridges are dynamic, but their length, paired with the clear progression of formation (polarization of actin, followed by short protrusions, then connected bridges), suggests some temporal stability.

Although they differ in presentation from previously studied heterotypic contacts, it is possible that these bridges are functioning similarly to maintain tissue borders. During regeneration, injuries that damage both epithelial layers must preserve both tissue identity and orientation, ultimately restoring the original anatomy. Because we do not yet have clear tools to differentiate between epithelial membranes, we don’t know how the bridges resolve during secondary closure. However, the framework established here sets the stage for future studies, including better visualization of cellular interactions and tissue coordination, identification of adhesion and migration machinery, the roles of signaling pathways, and more.

When assessing the dynamics of these heterotypic interactions, we observed that the outer epidermis requires muscle contraction to form stable protrusions, but the pharyngeal epithelium can spread across the wound even in the absence of muscle contraction. We also see pharyngeal epithelial cells, but not epidermal cells, spreading aberrantly in later stages of closure after stem cell irradiation. This difference in cell behaviors could be due to intrinsic differences in the epithelia; for example, perhaps the pharyngeal epithelium has intrinsic cell motility machinery that is primed for migration regardless of external forces. Intrinsic differences in adhesive and migratory properties of epithelial layers have been previously observed in the cnidarian Hydra vulgaris, in which endodermal and ectodermal epithelia behave differently in the context of self-assembly and local wounding⁷⁹.

The differences between the responses of the two epithelia could also be due to their extrinsic interactions. The epithelial cells in H. miamia are anatomically unique, lacking a flat basal surface as is typically observed in most animal epithelia and instead forming long projections that end in close proximity to muscle. Electron microscopy in acoels has failed to identify a visible basement membrane underneath the outer epidermis^47,80, but the composition and structure of the extracellular matrix (ECM) in H. miamia have not been studied. Using a custom antibody in this study, we demonstrate that there is tissue-specific collagen underneath the pharyngeal epithelial cells, and that collagen is present along wound-induced protrusions. We did not observe this collagen underneath the outer epidermis, but a similar mechanism could be occurring with a different extracellular matrix protein, perhaps one whose remodeling is mechanosensitive. Differences in ECM stiffness and composition could affect the mechanical response of each tissue, as well as differences in regular tensile loads. It has also been shown in other systems that wound-induced muscle contraction triggers the release of matrix remodeling proteins^53,54, which could be necessary for epidermal cells to extend far enough to reach their pharyngeal epithelial counterparts. Future studies will need to further characterize the H. miamia ECM and its dynamics to distinguish between these hypotheses.

The difference in wound closure processes across injury types raises several questions about the connection between healing and regeneration. Extensive work in the whole-body regeneration field has shown that fragments begin to re-establish their polarity within hours of amputation^40,41,44,81, and we have shown here that the temporal dynamics of regeneration are variable across different amputations. The amount of proliferation and the distribution of mitotic cells differ depending on the type of injury, but we did not see a clear correlation between the type of regeneration (i.e., anterior, posterior, sagittal) and the type of wound closure. Instead, the major determinants of cellular behavior during closure were which types of epithelia were damaged and the geometry of the fragment. Although not seen in this study, it seems likely that different strategies of wound closure correlate with different cellular and molecular environments in regenerating fragments, whether through biochemical and/or mechanical events associated with re-epithelialization. Future work will be required to test whether the different closure strategies we describe here have direct consequences for downstream tissue regeneration, whether healing and regeneration are truly nonoverlapping biological processes, and whether other species capable of whole-body regeneration utilize distinct mechanisms to heal different types of wounds.

Methods

Hofstenia miamia strains and husbandry

Adult worms were housed in artificial seawater (ASW) (37 ppt, pH 7.70–8.00) in plastic boxes within 21 °C incubators. Embryos were collected from adult boxes and placed into petri dishes for development. Once hatched, juvenile worms were transferred to tanks for post-embryonic growth. All animals were fed twice a week; adult worms were fed brine shrimp (Artemia, Brine Shrimp Direct), and juvenile hatchlings were fed rotifers (Brachionus plicatilis, Reed Mariculture). EpiA:RFP worms were generated as previously described⁴⁵ and reared identically to wild-type worms.

Irradiation of Hofstenia miamia

Juvenile worms were isolated and irradiated as previously described⁴¹. Briefly, animals were isolated and placed in a dual Gammacell-40 cesium source to deliver 10,000 rads (65 min at 153 rad/min).

Amputations and sample preparation

Juvenile worms were isolated and starved for 1–2 weeks prior to use to minimize autofluorescence of ingested food. For regeneration experiments, animals were placed in 1 mg/mL ethyl 3-aminobenzoate, methanesulfonic acid (“tricaine”, stock solution 5 mg/mL of ASW) (Thermo Fisher, AC118000100) in ASW in a petri dish for 3–5 min until they stopped swimming, then amputated using a scalpel. After amputation, fragments were transferred to a dish containing fresh ASW until their designated time point. For fixation, animals were placed in 4% paraformaldehyde (PFA) (Thomas Scientific, C993M31) in ASW for 1 h at room temperature. The PFA was removed, and animals were washed with PBST (PBS + 0.1% Triton-X-100) three times prior to immunofluorescence staining. For pinch wounds, animals were grasped with fine-tip forceps, then allowed to recover in ASW prior to fixation. For transverse and donut cuts, animals were cut once or twice transversely, then transferred to fresh ASW until their designated time point. For single transverse wounds, animals were cut transversely once again, above the previous cut, to remove the wound site for imaging. These fragments were then fixed in 4% PFA in ASW at their designated time points.

Small molecule inhibitors and treatment with anesthetics

For small molecule inhibitor experiments, animals were first amputated as described above, then flushed with ASW to remove tricaine prior to incubation in designated inhibitors. Animals were left in an inhibitor for 3–70 h, depending on the experiment. Blebbistatin (Sigma-Aldrich, B0560) was dissolved in DMSO per manufacturer instructions, then diluted in ASW for use at 25 µM. DMSO control treatments used an equal volume of DMSO as Blebbistatin treatments. For tricaine time point experiments, animals were left in 0.5 mg/mL tricaine in ASW for 3–50 h.

Immunostaining

For antibody staining, all steps were carried out using an orbital shaker. Fixed animals were washed in PBST for 3 × 5 min, then placed in blocking solution (PBST + 10% goat serum) (Thermo Fisher, 16210072) for one hour at room temperature. Animals were then placed in primary antibody in block solution for two nights at 4 °C. Primary antibody was washed out with PBST, 6 × 20 min. Animals were then blocked once again for one hour at room temperature, then incubated with secondary antibody (1:800) (Thermo Fisher, A-11011) in blocking solution overnight at 4 °C. The next day, the secondary antibody was washed out using PBST, 6 × 20 min. Finally, animals were incubated in SiR-actin (1:1000) in PBST (Cytoskeleton, CY-SC001) for 90 min at room temperature prior to mounting. All samples were mounted on double cover slips using Prolong Glass Antifade Mountant (Thermo Fisher, P36980). If animals were only stained with SiR-actin, they were placed in SiR-actin in PBST for 90 min at room temperature, then mounted immediately. The custom H. miamia Tropomyosin antibody⁶² (Rabbit IgG) and the custom Collagen IV antibody were used at 1 µg/mL. To label mitotic cells, we used the anti-phospho-Histone H3 (Ser10) antibody (1:1000) (Sigma-Aldrich, 06-570). Where relevant, Hoechst (1:500) (Invitrogen, H3570) was used to counter-stain nuclei.

Fluorescent in situ hybridization

Following fixation, animals were transferred to small baskets with a mesh bottom and placed in 24-well plates. Eight hundred microliters of solution was used for each well. Animals were permeabilized using proteinase K solution (0.1% SDS, 1 μL/10 mL ProK in PBST). After permeabilization, worms were post-fixed in 4% formaldehyde in PBST, washed twice in PBST, and then washed in a 1:1 PBST:PreHyb solution for 10 min. Animals were transferred to PreHyb solution (50% DI formamide, 25% 20X SSC, 0.05% 10% Tween-20, 1 mg/mL yeast tRNA, 20% water) for 2 h in a 56 °C hybridization oven, followed by transfer to hybridization solution (50% DI formamide, 25% 20X SSC, 0.05% 10% Tween-20, 1 mg/mL yeast tRNA, 10% 50% dextran sulfate, 10% water) containing riboprobe and left overnight. The following day, animals were washed at 56° C (PreHyb x2, 1:1 PreHyb:2XSSCT x2, 2XSSCT (2XSSC with 0.1% TritonX-100) x2, 0.2XSSCT (0.2X SSC with 0.1% TritonX-100) x2). Animals were transferred to room temperature, washed twice with PBST, and moved to blocking solution (5% horse serum and 5% casein in PBST) for 1 h. Animals were incubated with anti-Digoxigenin-POD (1:1500 dilution; Roche, 11633716001) in blocking buffer overnight. The following day, animals were washed with PBST x6, incubated with tyramide buffer (1.1688 g/mL NaCl, 6.18 mg/mL boric acid, filled with water, and adjusted pH to 8.5) for 10 min, then developed with rhodamine-conjugated tyramide solution for 10 min. Animals were washed again with PBST and incubated for 1 h in 1:500 Hoechst in PBST to label nuclei. Animals were then mounted on glass coverslips. Probes for the genes phar, piwi-1, and cuzd1-4 were previously generated in the lab^43,59.

Generation of custom collagen IV antibody

Rabbit polyclonal antibodies were custom-made by GenScript through their PolyExpress immunization protocol, using a peptide sequence from H. miamia Collagen IV (CPPGAKGNDGEPGEKG). This peptide was expressed in E. coli using the provided sequence and then used to immunize two rabbits. Two antibodies were generated, purified, and validated in-house by Western Blot with the protein immunogen.

Microscopy and image processing

All images were collected using 10×/0.25× (dry), 20×/0.75× (dry), or 63×/1.40 (oil) fluorescent objectives on a Leica DMI SP8 inverted confocal microscope, equipped with Leica Application Suite software (3.5.27). Images were processed using Image J/FIJI software.

Quantifications

Length of epithelial basal protrusions

To quantify the average length of basal protrusions in outer epidermal and pharyngeal epithelial cells, images were taken of labeled cells in EpiA:RFP animals. The Measure function in FIJI was used to quantify the length of each protrusion, tracing from the end of the protrusion to where it meets the body of the cell. For each cell type (pharyngeal epithelium vs. outer epidermis), n = 15 cells from n = 5 animals (total = 45 cells) were measured.

Percentage of closure in transverse amputations

For quantification of wound closure in transverse cuts, images were taken of both the wound edge and the open edge of each sample, using FIJI to measure the open area of each side. The percentage of wound closure was determined by dividing the open area of the wound edge by the total open area, then subtracting that from 1. For each time point and fragment (head vs. tail), measurements were collected from n = 8 animals.

Percentage of closure in sagittal amputations

To quantify the percentage of wound closure after sagittal amputations, images of SiR-actin-stained fragments were imported into FIJI. Because of mounting imperfections, only one side of each animal was measured, choosing the side with the clearest signal for the most accurate quantification. Six to twelve animals were quantified per time point and condition. Within FIJI, the Measure function was used to quantify: (1) the length of the pharynx and (2) the cumulative distance of connected tissue between the epidermis and the pharyngeal epithelium. The percentage of wound closure was determined by dividing the closed distance by the total pharynx length. For each time point, measurements were collected from n = 12 animals.

Bridge length

To quantify the length of heterotypic bridges, images of SiR-actin-stained fragments were taken at each time point, then imported into FIJI for analysis. Using the Measure function, bridges were quantified by drawing a line along the membrane extensions protruding between cell types. Cell bodies were not included in this measurement. Because of the technical challenge of robustly identifying where one protrusion met the other, the combined bridge length was reported instead of individual cell contributions. For each time point, measurements were collected from n = 30–40 animals.

pH3 levels and mitotic cell distribution

To quantify the number of mitotic cells per mm², z-stacks of pH3-stained animals were taken of entire fragments. Images were imported into FIJI and flatted into maximum-intensity projections for analysis. For each animal, we measured the total area of the fragment (in µm²) using the Hoechst channel. We thresholded the pH3 channel and used the Analyze Particles function to quantify the number of cells per animal, normalizing for area and converting µm² to mm² for graphing purposes. To generate the distribution plots of pH3+ cell density, each thresholded image was aligned horizontally and oriented (from L to R) along its anterior-posterior axis in FIJI. The Plot Profile function was used to generate average intensity profiles along each image’s y-axis, and each plot was saved as a spreadsheet. The spreadsheets were imported into MATLAB and plotted. Plotting code was developed with assistance from Claude (Anthropic, Sonnet 5) and completely reviewed by the authors prior to use. Profiles were smoothened using the pandas smooth_window function with a window size of 20, and each image was normalized to an axis of 200 points, with 0 = anterior end and 1 = posterior end of the animal. Eight animals were quantified per amputation and per time point.

Morphological descriptions

To report the morphological characteristics of healing fragments, images of SiR-actin-stained fragments were taken at each time point. For bridge formation, animals were binned into four categories: none (no visible polarization of the epithelial layers and no protrusions), polarization (visible polarization of cell membranes towards the wound edge), protrusions (at least one membrane extension <15 µm in length), or bridges (at least one connected heterotypic bridge). For protrusion origin, animals were assessed for visible protrusions coming from only the outer epidermis, only the pharyngeal epithelium, both, or neither. Rounding was assessed by looking at the shape of the animal, binning animals into either open (no visible rounding), rounding (partway through the rounding process), or closed (where the outer edges of the fragments have met, re-forming a cylindrical body plan). The values plotted for each morphological trait were the representative percentage of total fragments, ranging from n = 10–45 animals (exact values are reported in accompanying figure legends).

Graphing and statistical analyses

All graphs were generated using GraphPad PRISM 10 or Python. Data with two or more factors were analyzed using two-way ANOVA with Dunnett’s multiple comparisons test. Data comparing changes in one condition over different time points were analyzed by one-way ANOVA with Dunnett’s multiple comparisons test. A two-tailed unpaired t-test with Welch’s correction was used to compare data between two independent groups. All statistical tests are stated in the figure legends. P values were obtained using GraphPad PRISM 10 and used to determine the level of varying significance between experimental groups; groups were considered different when P < 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

A.P.K. conceived the project, performed experiments, quantified and analyzed data, and wrote the manuscript under the supervision of M.S. A.P.K. and M.S. discussed the results and edited the manuscript together.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

APK discloses support for publication of this work from the Jane Coffin Childs Memorial Fund for Medical Research. MS discloses support for publication of this work from the National Institute of General Medical Sciences [R35GM128817] and [R35GM153252].

Data availability

Source data are provided with this paper.

Competing interests

The authors declared no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-76555-8.