Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Epicardium-derived cells organize through tight junctions to replenish cardiac muscle in salamanders

Abstract

The contribution of the epicardium, the outermost layer of the heart, to cardiac regeneration has remained controversial due to a lack of suitable analytical tools. By combining genetic marker-independent lineage-tracing strategies with transcriptional profiling and loss-of-function methods, we report here that the epicardium of the highly regenerative salamander species Pleurodeles waltl has an intrinsic capacity to differentiate into cardiomyocytes. Following cryoinjury, CLDN6+ epicardium-derived cells appear at the lesion site, organize into honeycomb-like structures connected via focal tight junctions and undergo transcriptional reprogramming that results in concomitant differentiation into de novo cardiomyocytes. Ablation of CLDN6+ differentiation intermediates as well as disruption of their tight junctions impairs cardiac regeneration. Salamanders constitute the evolutionarily closest species to mammals with an extensive ability to regenerate heart muscle and our results highlight the epicardium and tight junctions as key targets in efforts to promote cardiac regeneration.

Main

Regeneration of cardiac muscle in adult salamanders and teleost fish has been attributed to dedifferentiation and proliferation of pre-existing cardiomyocytes1,2,3,4,5,6,7,8,9. Cardiomyocytes can re-enter the cell cycle in the adult mammalian heart but this does not lead to functional regeneration after injury due to the low frequency of cardiomyocytes that complete the cell cycle and proliferate10,11,12. Hence, most efforts aim to identify paracrine cues that can stimulate cardiomyocyte proliferation13,14,15. This is however challenging given that cardiomyocytes are multinucleated and polyploid in adult mammals, in contrast to naturally regenerative non-mammalian vertebrates13,16.

An alternative approach involves resident progenitor cells that become activated following injury and differentiate into cardiomyocytes. Epicardium, the epithelial layer enclosing the heart, is a prime candidate for such an approach as it is a source of multipotent progenitors during development17,18. Studies examining the contribution of epicardium to the regeneration of the zebrafish heart have highlighted its role as a source of paracrine signalling and extracellular-matrix molecules as well as a modulator of inflammation19. Genetic lineage tracing and transplantation studies in zebrafish did not show epicardial cell differentiation into cardiomyocytes7,20,21,22. The innate epicardial fate in mammals following injury has remained unresolved due to the paucity of tools for cell type-specific tracing, as current genetic markers also label epicardial derivatives and/or non-epicardial cell types23,24,25,26. Nevertheless, it has been reported that following stimulation with factors such as thymosin-ß4 and VEGF, epicardial cells differentiate into cardiomyocytes at a low frequency, indicating the regenerative potential of the epicardium. However, infrequent conversion rates and diminished lineage plasticity after injury necessitate further studies to identify ways to enhance the regenerative response of the epicardium27,28,29.

Earlier studies of salamander heart regeneration established the ability of salamander cardiomyocytes to proliferate3,30,31,32. However, the role of the epicardium has not been investigated further than its upregulation of regeneration-specific matrix proteins and genetic markers of the salamander epicardium have not been identified33. Here we established a genetic marker-independent lineage tracing strategy in the salamander Pleurodeles waltl and show low-level conversion of epicardial cells into myocytes during homeostasis, a process that is greatly expanded in response to cryoinjury. Using single-cell RNA sequencing (scRNA-seq), we identify the tight junction protein CLDN6 as a specific marker of the homeostatic epicardium. Following cryoinjury, epicardium-derived cells (EPDCs) migrate to the injury site, form honeycomb-like structures decorated by CLDN6+ focal tight junctions and differentiate into cardiomyocytes, which engraft into the myocardium. Transcriptional profiling and trajectory analyses reveal the expression of the key cardiac transcription factors Gata4, Gata6, Foxc1 and Foxc2 during this cell-fate transition. Finally, we show that both ablation of CLDN6+ differentiation intermediates as well as disruption of tight junctions impairs cardiac regeneration in salamanders.

Results

Epicardium gives rise to cardiomyocytes under homeostasis

To study the potency of the post-metamorphic salamander epicardium, we developed a lineage tracing strategy that selectively labels the epicardium in an unbiased genetic marker-independent manner. As the epicardium forms a barrier between the underlying myocardium and the surrounding pericardial fluid (Fig. 1a,b)34, we microinjected a cell-permeant Cre recombinase (TAT-Cre) into the pericardial cavity of tgTol2(CAG:loxP-Cherry-loxP-H2B::YFP)Simon (hereafter, Cherry-loxP-H2B::YFP) reporter animals (Fig. 1c,d)35,36. Thirty hours post microinjection (h.p.i.), nuclear Cre expression was confined to the epicardium and no labelled cells were found in the underlying myocardium (Extended Data Fig. 1a,b). Accordingly, we detected the emergence of yellow fluorescent protein (YFP) signal in the epicardial cells (Extended Data Fig. 1a). At 40 h.p.i., TAT-Cre-induced recombination yielded strong YFP expression in the outermost pan-cytokeratin (pan-CK)-expressing epicardial layer, detected with a polyclonal antibody recognizing a broad spectrum of keratins (Extended Data Fig. 1a,b). A labelling efficiency of 38% was observed and spontaneous recombination did not occur in the vehicle-injected animals (Extended Data Fig. 2a–c). Reflecting previous reports of TAT fusion proteins binding to the extracellular matrix surrounding muscle tissue, we observed some extracellular matrix-associated Cre fluorescence in the myocardium (Extended Data Figs. 1a and 2d)37. However, this did not transduce cardiomyocytes, as nuclear Cre signal was absent in the myocardium (Extended Data Fig. 1a,b).

Fig. 1: Salamander epicardium gives rise to cardiomyocytes under homeostasis.
figure 1

a, TEM image showing the organization of epicardium, subepicardium and myocardium. The epicardial layer is pseudocoloured. The dashed line marks the myocardium–subepicardium border. Scale bar, 25 µm. Data represent three animals. b, Salamander epicardium forms a barrier, as shown in a biotin permeability assay. Representative immunostainings for 4,6-diamidino-2-phenylindole (DAPI), MHC (cardiomyocyte marker), pan-CK (epicardial cell marker) and biotin. Biotin staining is confined to the epicardium. Data represent four animals. Scale bars, 200 µm (top) and 10 µm (bottom). c, Cell-permeable TAT-Cre recombinase fusion protein. d, Experimental design for permanent labelling of epicardial cells in Cherry-loxP-H2B::YFP reporter salamanders. e, TAT-Cre recombinase microinjection into the pericardial cavity of Cherry-loxP-H2B::YFP allows selective labelling of epicardial cells. Representative immunostainings for pan-CK, MHC and YFP at 40 h.p.i. Labelled cells occupy the epicardial layer. Data represent six animals. f, Epicardial-to-myocardial cell conversion during homeostasis. Representative immunostainings for pan-CK, MHC and YFP on TAT-Cre recombinase-injected Cherry-loxP-H2B::YFP ventricle sections at 96 h.p.i. Labelled cells are found in the epicardial and myocardial layer (arrows). Data represent five animals. g, Experimental design for permanent epicardial-cell labelling in GFP-loxP-Cherry animals. h, EPDCs in the myocardium co-stained for cytoplasmic CHERRY and MHC indicating epicardial cell-to-cardiomyocyte conversion. Representative immunostainings for DAPI, CHERRY and MHC at 11 d.p.i. Data represent six animals. CHERRY+MHC+ cells are outlined. Magnified views of the numbered regions in the overview images (left) are shown (middle and right). e,f,h, Scale bars, 20 µm (overview), 10 µm (magnified view). The myocardium–epicardium border is marked by dashed lines. i, The number of labelled cardiomyocytes in the ventricular area were quantified at 40 h.p.i., 96 h.p.i. and 11 d.p.i. The number of epicardium-derived cardiomyocytes increased with time. Data are the mean ± s.e.m. of n = 6 (40 h.p.i. and 11 d.p.i.) and 5 (96 h.p.i.) animals. One-way analysis of variance (ANOVA) with Tukey’s multiple comparisons test; NS, not significant (P = 0.9099); ***P = 0.0001 and ****P < 0.0001. Epi, epicardium; sub-epi, subepicardium; and myo, myocardium.

Source data

From 40 h.p.i. we found rare YFP+ cells embedded within the myosin heavy chain (MHC)-expressing myocardium, which increased in number by 96 h.p.i. (Fig. 1e,f,i and Extended Data Fig. 2e). To assess the identity of these EPDCs, we performed lineage tracing in salamanders carrying the conditional reporter tgTol2(CAG:loxP-GFP-loxP-Cherry)Simon (hereafter GFP-loxP-Cherry), in which CHERRY expression on recombination is cytoplasmic and facilitates the assessment of cellular morphology (Fig. 1g). At 96 h.p.i., CHERRY+ cells coexpressing ɑ-actinin lacked myofibrillar structures and had the appearance of immature cardiomyocytes (Extended Data Fig. 2f). In contrast, we observed CHERRY+ cells coexpressing MHC and ɑ-actinin at 11 days post injection (d.p.i.; Fig. 1h and Extended Data Fig. 2g) that by virtue of their size, morphology and myofibrillar structure resembled mature cardiomyocytes. We found an increase in the number of labelled epicardium-derived cardiomyocytes over the course of 11 d (Fig. 1i), suggesting that ongoing low-level conversion of epicardial cells to cardiomyocytes contributes to cardiac homeostasis in salamanders.

Injury induces epicardial cell-to-cardiomyocyte conversion

To investigate whether epicardium contributes to cardiac regeneration in salamanders, we established a cryoinjury model. Using a liquid nitrogen-cooled probe, we injured the ventricular apex and analysed the extent of regeneration at 7, 14, 28, 64 and 210 days post cryoinjury (d.p.ci.; Fig. 2a,b and Extended Data Fig. 2h). The procedure consistently damaged approximately 25% of the ventricle, as measured at 7 d.p.ci. (Fig. 2b). Cryoinjured ventricles showed loss of myocardium as well as the deposition of collagen and fibrin (Fig. 2a). The lesion was reduced to approximately 12% and the fibrin clot started to resorb by 14 d.p.ci. (Fig. 2a,b). At 28 d.p.ci., the scar was reduced and composed of more prominent collagen networks (Fig. 2a,b). At 64 d.p.ci., the small remaining lesion—detectable by remnants of collagen—was interspersed with myocardial cells, indicating the replacement of the injury site by regenerating myocardium (Fig. 2a,b). There were no signs of injury and tissue organization was restored 210 d.p.ci. (Fig. 2a,b). We also monitored the regeneration process using echocardiography to measure the injury size for each animal and observed a gradual recovery, reflecting the histological analyses (Extended Data Fig. 2i). These results show that the salamander heart can regenerate damaged muscle tissue in response to cryoinjury.

Fig. 2: Epicardial cell-to-myocyte conversion increases in response to cryoinjury.
figure 2

a, Time course of regeneration following cryoinjury. Acid fuchsin orange G (AFOG) staining of the ventricular apex post cryoinjury. b, The size of the ventricular wound decreased with time. Data are the mean ± s.e.m. of n = 6 (7 d.p.ci.), 4 (14 and 210 d.p.ci.), 3 (21 and 64 d.p.ci.) and 5 (28 d.p.ci.) animals. c, Schematic of the permanent epicardial cell labelling combined with cryoinjury experiment. d, Epicardium-derived cardiomyocytes occupy the regenerating area. Representative images of CHERRY and MHC immunostaining at 21 d.p.ci. Ten of 258 representative frontal plane sections collected from an animal are shown, representing five animals. a,d, Scale bars, 200 µm. e, Magnified view of the box in frontal plane no. 8 in d. Immunostaining for CHERRY, MHC and sarcomeric ɑ-actinin (SɑA). The dashed line shows the myocardium (myo)–epicardium (epi) border. Scale bars, 10 µm. f, Tenascin-C marks the regenerating area. Representative images of DAPI, Tenascin-C, CHERRY and MHC immunostaining at 21 d.p.ci. on the section preceding frontal plane no. 9 in d. Note that the injured muscle at this section depth shows cardiac muscle regeneration, assessed by MHC staining, still displaying Tenascin-C positivity. Scale bars, 200 µm. g, Injury induces expansion of the epicardium-derived cardiomyocyte pool. The CHERRY+MHC+ cells were enumerated in sham-operated (n = 4 animals) and cryoinjured hearts (n = 5 animals) at 21 d.p.ci. Unpaired Student’s t-test with Welch’s two-tailed correction; **P = 0.0033. Data are the mean ± s.d. h, Schematic of the experiment to assess epicardium-derived cardiomyocyte engraftment. i, Epicardium-derived cardiomyocytes show long-term engraftment to the regenerated myocardium. High-resolution immunofluorescence images demonstrate sarcomeric organization in epicardium-derived cardiomyocytes, with transverse orientation of ɑ-actinin at z-lines overlapping with CHERRY signal. Overview (left) and magnified view of the region in the white box (right). Data represent eight animals. Scale bars, 100 µm (left) and 10 µm (right).

Source data

To evaluate the cellular contribution of epicardium to the regenerating myocardium, we performed lineage tracing as described earlier, followed by cryoinjuries at 40 h.p.i. (Fig. 2c). Absence of nuclear Cre signal in the myocardium following injury was confirmed (Extended Data Fig. 3a–c). Cryoinjury decreased the number of labelled epicardial cells, resulting in approximately 29% of the remaining epicardium being labelled, as assessed at 48 h post cryoinjury, a time point before the initiation of epicardial proliferation (Extended Data Fig. 3d–g). At 21 d.p.ci., we found CHERRY+ cells coexpressing MHC (Fig. 2d) and ɑ-actinin (Fig. 2e) within the Tenascin-C+ apical region of the myocardium (Fig. 2f). We found an increase in the number of labelled cardiomyocytes of approximately 14-fold compared with the corresponding regions in the sham-operated hearts (Fig. 2g and Extended Data Fig. 3h), indicating a substantial expansion of epicardial cell conversion into cardiomyocytes after injury. Importantly, epicardium-derived cardiomyocytes analysed at 60 d.p.ci. showed elongated morphology and sarcomere formation, indicating long-term engraftment (Fig. 2h,i). Together, these results show that epicardium-derived cardiomyocytes not only contribute to the myocardium during homeostasis but also regenerate the myocardium after injury. We also observed occasional CHERRY+vimentin+ mesenchymal cells and CHERRY+ɑ-smooth muscle actin+ smooth muscle cells/myofibroblasts, but not endothelial cells, indicating that epicardial cells may also give rise to non-myocyte lineages (Extended Data Fig. 3i–l).

To address whether epicardium-derived cardiomyocytes expand—that is, one epicardial cell gives rise to several cardiomyocytes—we utilized tgTol2(CAG:Nucbow)Simon reporter animals38. Here recombination results in random combinations of different fluorescent proteins to facilitate clonal analysis (Extended Data Fig. 4a). At 21 d.p.ci., approximately 54% of same-colour clones were comprised of two or more clonally related cells, indicating cell division (Extended Data Fig. 4b–f). Furthermore, we found that 22% of epicardium-derived cardiomyocytes were PCNA+ at 21 d.p.ci., with occasional expression of phospho-histone H3 (Extended Data Fig. 4g–j), which suggested that clonal expansion could result from the proliferation of epicardium-derived cardiomyocytes.

Tight junction genes specifically mark epicardium

To follow transcriptional changes accompanying regeneration and discover specific markers of the salamander epicardium, we performed scRNA-seq on 2,386 live cells collected from sham-operated and regenerating hearts at 7, 14 and 28 d.p.ci. (Fig. 3a, Extended Data Fig. 5a–d and Supplementary Table 1). Seventeen distinct cell clusters were identified through unbiased clustering and marker-gene expression (Fig. 3b and Supplementary Table 1). As expected, immune cells represented the majority of cells recovered at 7 and 14 d.p.ci., whereas clusters identified to be endothelial or endocardial cells (Clusters 0 and 5) and myocyte-like cells (Cluster 6) were less abundant at these time points (Fig. 3b and Extended Data Fig. 5e). A small cluster (Cluster 16) was annotated as transitioning cells based on their transient appearance at 7 and 14 d.p.ci. as well as expression of the epithelial-to-mesenchymal transition (EMT) markers (Fig. 3b,d, Extended Data Fig. 5e,f and Supplementary Table 1).

Fig. 3: Cldn family genes identified via scRNA-seq as specific markers of the salamander epicardium.
figure 3

a, Schematic overview of the scRNA-seq experiments performed on metabolically active, live and nucleated cells of sham-operated and regenerating ventricles. b, Visualization of the scRNA-seq data from 2,386 individual cells using a uniform manifold approximation and projection (UMAP) plot. Endo/EC, endocardial/endothelial cell; and subepicardial ECs, subepicardial endothelial cells. c, UMAP plots showing the expression of the embryonic epicardial marker genes Wt1, Tcf21 and Tbx15 (top) as well as the marker genes identified in this study—that is, Dkk2, Cd248 and Cldn6 (bottom). d, Heatmap showing the expression levels of specific marker genes for the epicardium and EPDC cluster candidates and transitioning cells. See Extended Data Fig. 6b for the gene expression levels in all cell clusters. e, Distribution of expression for a subset of differentially expressed genes in cluster 11. f, Cldn6, Cldn7 and Cldn15 are coexpressed in epicardial cells. Representative images of fluorescent in situ hybridization of heart sections of sham-operated (top) and cryoinjured (bottom) animals at 7 d.p.ci. Note that the RNAscope fluorescent assay on fixed frozen salamander tissue causes background staining in blood cells (asterisks) that is easily distinguishable from real signal due to its non-dotty appearance. Insets: magnified views of the regions in the white boxes. Scale bars, 10 µm. g, Expression of Cldn6 messenger RNA marks the epicardium. Representative ventricle overview (left) and magnified views (middle and right) of sham-operated (top) and 7-d.p.ci. (right) heart sections. The injury area (IA) is outlined and the boxes mark the areas selected for the magnified images. Scale bars, 500 µm (left), 200 µm (middle) and 10 µm (right). f,g, Dashed lines indicate the epicardium–myocardium border.

Three clusters (Clusters 9, 11 and 12) expressed the embryonic epicardial genes Wt1, Tcf21 and Tbx15 (Tbx18 homologue in P. waltl)39,40,41 (Fig. 3c). Cluster 9 cells expressed Dkk2 (Fig. 3c,d), a Wnt-pathway inhibitor previously indicated in the specification of pro-epicardial cells, and Hoxa5, a marker of the axolotl epicardium (Fig. 3d)42,43. These cells also expressed regulators of angiogenesis such as the endothelial orphan G protein-coupled receptor Adgrl4, the transcription factor Sox18 and the endothelial Rho guanine exchange factor Fgd5 (Fig. 3d)44,45,46,47, suggesting that they might represent subepicardial endothelial cells. Providing support for this, Gene Ontology characteristics related to angiogenesis and vasculature development were enriched in Cluster 9 (Extended Data Fig. 6a).

Cluster 11 cells expressed the mesothelial marker Lrrn4 (ref. 48); the epithelial markers Cdh1, Epcam and Alcam49,50,51 and the cell-adhesion molecule Flrt2, a known epicardial marker52 (Fig. 3c–e and Extended Data Fig. 6b,c), indicating this as the main epicardial sheet enveloping the heart. In addition, cells in this cluster expressed the Claudin (Cldn) gene family members Cldn6, Cldn7 and Cldn15, important components of tight junction formation53 (Fig. 3c–e). Accordingly, Gene Ontology term and reactome over-representation analyses highlighted genes associated with cell-junction organization and tight junction interactions as over-represented among the genes with the highest expression in this cluster (Extended Data Fig. 6a).

Cluster 12 cells expressed the stromal marker Cd248 and showed enrichment for the expression of extracellular-matrix genes (Dpt, Dcn, Ogn and Lum; Fig. 3c,d and Extended Data Fig. 6b), suggesting that these cells are more mesenchymal in nature and represent fibroblast-like cells. Gene Ontology term analysis showed enrichment of genes related to extracellular-matrix assembly and connective-tissue development (Extended Data Fig. 6a).

In situ hybridizations subsequently showed that Dkk2+ cells were localized to the subepicardial region (Extended Data Fig. 6d). Cd248+ cells were found in the subepicardium and myocardial interstitial space (Extended Data Fig. 6e). Cldn6, Cldn7 and Cldn15 expression overlapped and marked the outermost layer of the heart specifically (Fig. 3f,g). Cldn6 was chosen as an epicardial cell marker for subsequent studies as it showed the highest level of expression among the tight junction genes (Fig. 3e). CHERRY+ epicardial cells marked after Cre-induced recombination in GFP-loxP-Cherry were Cldn6+ (Extended Data Fig. 6f), confirming that Cluster 11 cells represent the epicardium proper. Despite increased numbers of epicardial/subepicardial cells due to epicardial thickening, both the number of cells expressing Cldn6 and the levels of Cldn6 expression decreased at 7 d.p.ci. (Fig. 3f,g and Extended Data Fig. 6g–i). Accordingly, no cells were recovered from 7- and 14-d.p.ci. samples contributing to the Cldn6+ epicardial cell cluster, suggesting a dynamic response to injury, which we decided to explore further.

CLDN6 localizes to cell clusters in the injury

CLDN6 has a key role in the formation of embryonic epithelium and the development of endodermal tissues54,55. It is an oncofetal tight junction molecule that is expressed at high levels in stem cells and developing tissues but transcriptionally silenced in healthy adult tissues of mammals56,57. Immunohistochemical analyses in the post-metamorphic salamander heart showed that CLDN6 is expressed in the outermost epicardial layer and marks cell–cell junctions (Fig. 4a,b). Transmission electron microscopy (TEM) imaging confirmed that the outermost epicardial cells of the homeostatic epicardium are sealed via tight junctions at the epicardium–pericardial fluid interface (Fig. 4b). As expected, adherens junctions and desmosomes were located beneath the tight junctions (Fig. 4b). After cryoinjury (7 d.p.ci.), CLDN6 was present at reduced levels in the epicardial cells basal to the injury site and absent at the cell–cell borders (Fig. 4c,d). Providing further support for these data, TEM imaging revealed loss of tight junctions (Fig. 4d). Cells located subepicardially instead showed cytoplasmic CLDN6 expression, suggesting that EPDCs retain CLDN6 protein as they migrate (Fig. 4e). Subsequently, we found clusters of CLDN6+ cells in the injury forming honeycomb-like structures (Fig. 4f) connected via focal tight junctions rather than mature tight junction strands, suggesting a dynamic junctional remodelling accompanying migration of CLDN6+ EPDCs into the injury area (Fig. 4f). We did not detect expression of CLDN6 in the myocardium (Extended Data Fig. 7a).

Fig. 4: CLDN6+ cells organize in a honeycomb-like pattern in the injury area.
figure 4

a, CLDN6 marks salamander epicardium. Overview images (top) and magnified views of the regions in the white boxes (bottom) of uninjured hearts stained for DAPI, CLDN6 and sarcomeric ɑ-actinin (SɑA). The arrowhead marks an epicardial cell. Scale bars, 50 µm (top) and 10 µm (bottom). b, Schematic depicting the region of interest (box) in an uninjured heart probed for CLDN6 expression and TEM imaging (top). Representative image showing DAPI, CLDN6 and pan-CK staining (middle); the arrows mark CLDN6+ cell–cell borders. Overview and close-up TEM images of homeostatic epicardium (bottom); desmosomes (arrow), adherens junctions (arrowhead) and tight junctions (asterisks) are shown. Scale bars, 20 µm (middle), 1 µm (bottom left) and 250 nm (bottom right). c, Epicardial CLDN6 expression is reduced following cryoinjury. Overview images (top) and magnified views of the region in the white box (i) (bottom) showing DAPI, CLDN6, SɑA and pan-CK basal to the injury site at 7 d.p.ci. Scale bars, 50 µm (top) and 10 µm (bottom). d, Schematic depicting the region of interest in injured hearts probed for CLDN6 expression and analysed by TEM for tight junctions at 7 d.p.ci. (top). Representative images of CLDN6, MHC and pan-CK immunostaining (middle); the arrows mark the CLDN6 cell–cell border. Overview (left) and magnified (right) TEM images of activated epicardium (bottom); desmosomes (arrows) and adherens junctions (arrowheads) are shown. Scale bars, 20 µm (middle), 1 µm (bottom left) and 250 nm (bottom right). a,b,d, The epicardium–myocardium border is marked by dashed lines. e, Subepicardial EPDCs express cytoplasmic CLDN6 7 d.p.ci. Representative images of DAPI, CLDN6, SɑA and pan-CK immunostaining corresponding to the boxed area (ii) in panel c. The arrowheads mark CLDN6+ cells in the subepicardium. Scale bars, 10 µm. f, CLDN6+ cells form a honeycomb-like structure in the injury area at 7 d.p.ci. Representative images of CLDN6 and MHC immunostaining; low-magnification image showing the injured apex (top). High-magnification images of the outlined areas (i) and (ii) (middle). Overview (left) and magnified (right) TEM images showing focal tight junctions (marked by asterisks; bottom). Scale bars, 200 µm (top), 20 µm (middle), 1 µm (bottom left) and 250 nm (bottom right). af, The confocal images are representative of four animals and the TEM images are representative of three animals. IA, injury area; and peri, pericardial space.

Targeting CLDN6+ epicardium and EPDCs

We next wanted to ablate CLDN6+ cells to assess their relevance to regeneration. Certain members of the CLDN family proteins act as specific receptors of the Clostridium perfringens enterotoxin (CPE)58,59,60,61. Binding of CPE leads to the formation of an active pore, which subsequently enhances calcium influx and results in cell death58. Bulk and scRNA-seq data obtained from sham-operated and injured hearts showed expression of Cldn6 and Cldn7 (Extended Data Fig. 7b,c). Other CPE-sensitive Cldn receptors were not expressed (Extended Data Fig. 7b,c), allowing us to target CLDN6+ epicardium and EPDCs in the injured heart.

Although salamander CLDN6 is 68% identical to human CLDN6, with a high level of conservation of the key amino acids required for CPE binding (Extended Data Fig. 7d,e)59, we first benchmarked the use of CPE in targeting salamander CLDN6+ cells. Transfection of HEK293T cells, which are normally non-responsive to CPE62,63, with a P. waltl Cldn6 expression construct sensitized the cells to CPE and resulted in cell death following treatment (Extended Data Fig. 7f). In contrast, transfection of a mutant form of the P. waltl Cldn6 harbouring a deletion of the CPE binding domain did not sensitize the cells (Extended Data Fig. 7f), indicating that P. waltl CLDN6 is a specific receptor of CPE. To rule out off-target effects and distinguish cell type-specific effects of CPE from potential systemic toxicity, we generated previously well-characterized variants of the toxin: the binding-deficient CPE-Y306A/L315A and c-CPE that lacks the cytotoxicity domain (Fig. 5a)64,65,66. As expected, these variants did not show an effect on cell viability (Extended Data Fig. 7f). Next, we treated animals with wild-type CPE (wt-CPE), CPE-Y306A/L315A or c-CPE for 6 h at 7 d.p.ci. and performed a terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay to assess apoptosis across the injury area, epicardium, subepicardium and myocardium (Fig. 5b). Treatment with a low dose of wt-CPE (1.8 µg g−1) resulted in a substantial increase in TUNEL+ cells, which were distinctly concentrated to the injury area, with little effect observed across the epicardium and subepicardium (Fig. 5c and Extended Data Fig. 8a,b). This suggested that CLDN6+ EPDCs in the injury area, rather than the epicardium itself, were preferentially targeted by wt-CPE at this dose. In comparison, treatment with a high dose of wt-CPE (10 µg g−1) also caused cell death in the epicardium, confirming CLDN6 as a pan-epicardium marker (Extended Data Fig. 8c). Dying cells in the injury area were also confirmed to express CLDN6 (Extended Data Fig. 8d). No effect was observed across the myocardium, regardless of the dose injected, confirming the specificity of the toxin (Fig. 5c and Extended Data Fig. 8a–c). We observed a loss of CLDN6+ EPDCs in the injury area 24 h after treatment with wt-CPE (Extended Data Fig. 8e). In contrast to wt-CPE, the binding-deficient negative control CPE-Y306A/L315A did not display toxicity, thereby confirming the CLDN dependence of the CPE effects (Fig. 5c). Similarly, c-CPE treatment also did not induce apoptosis (Fig. 5c).

Fig. 5: CLDN6+ EPDCs are specifically ablated by wt-CPE.
figure 5

a, Schematic showing the interaction between CPE variants and the CLDN receptor. TJs, tight junctions. b, Schematic of the experimental design for the TUNEL assay. c, Number of TUNEL+ cells following treatment with vehicle, wt-CPE, CPE-Y306A/L315A and c-CPE for 6 h. Data are the mean ± s.d. of n = 7 (vehicle), 4 (wt-CPE and c-CPE) and 5 (Y306A/315A) animals. Two-way ANOVA with Tukey’s test; NS, not significant (P > 0.05) and ****P < 0.0001. d, Experimental design for assessing the effect of CPE treatment on lineage-traced EPDCs. e, CHERRY+ EPDCs are depleted by wt-CPE treatment. Representative images of DAPI, CHERRY and MHC immunostaining. Scale bars, 200 µm. f, Number of CHERRY+ EPDCs in the injury area at 7 d.p.ci. Data are the mean ± s.d. of n = 4 (vehicle) and 5 (wt-CPE) animals. Unpaired two-tailed Student’s t-test; **P = 0.0056. IA, injury area.

Source data

To further confirm that CLDN6+ cells are specifically ablated by wt-CPE, we combined lineage tracing of epicardial cells with wt-CPE treatment (Fig. 5d and Extended Data Fig. 8f). Under homeostatic conditions, wt-CPE treatment reduced the number of CHERRY+ epicardial cells by approximately 5.4-fold, which led to a decrease in the number of epicardium-derived cardiomyocytes of approximately fourfold (Extended Data Fig. 8g–j). The number of CHERRY+ EPDCs in the injury area of cryoinjured hearts at 7 d.p.ci. was reduced approximately fivefold following wt-CPE treatment (Fig. 5e,f). Together, these experiments validate the use of wt-CPE to ablate CLDN6+ cells for downstream functional experiments.

Ablation of CLDN6+ EPDCs impairs cardiac regeneration

To ensure sustained depletion of CLDN6+ EPDCs, we established an optimized wt-CPE treatment regimen with minimal systemic toxicity to the animals (Fig. 6a and Extended Data Fig. 9a) and confirmed that the wt-CPE treatment did not impact the integrity of the myocardium in control hearts (Extended Data Fig. 9b). We next performed cryoinjuries and monitored the regeneration process in each animal up to 21 d.p.ci. using echocardiography (Fig. 6b, Supplementary Videos 14 and Extended Data Fig. 9c). Both vehicle- and CPE-Y306A/L315A-treated animals showed progressive regeneration over the course of 21 d, whereas the regeneration of wt-CPE-treated hearts was diminished (Fig. 6b,c). Histological assessment showed impaired regeneration of the wt-CPE-treated animals compared with the CPE-Y306A/L315A- and vehicle-treated animals at 21 d.p.ci. (Fig. 6d), highlighting the requirement of CLDN6+ EPDCs for efficient regeneration.

Fig. 6: Ablation of CLDN6+ EPDCs impairs cardiac regeneration.
figure 6

a, Schematic of the treatment regimen to assess the impact of wt-CPE and CPE-Y306A/L315A on regeneration. b, Cardiac regeneration is impaired by wt-CPE treatment. Recovery from injury was monitored. Data are the mean ± s.d. of n = 10 (vehicle), 8 (wt-CPE) and 5 (CPE-Y306A/L315A) animals. Two-way ANOVA with Tukey’s test; 10-d.p.ci. vehicle versus wt-CPE, P = 0.52; 14-d.p.ci. vehicle versus wt-CPE, *P = 0.0167; 21-d.p.ci. vehicle versus wt-CPE, **P = 0.0034; vehicle versus CPE-Y306A/L315A, P > 0.9999; NS, not significant. c, Levels of regeneration at 21 d.p.ci. Score: mild, <45%; moderate, 45–75%; and severe, >75%, based on the remaining injury size. χ2 test; ****P < 0.0001. d, Hearts treated with wt-CPE have larger injuries at 21 d.p.ci. compared with those treated with vehicle or CPE-Y306A/L315A. Representative images of DAPI, MHC and Tenascin-C immunostaining (top). AFOG staining of sections of vehicle-, wt-CPE- and CPE-Y306A/L315A-treated hearts (bottom). Scale bars, 100 µm. IA, injury area. c,d, n = 7 (vehicle and wt-CPE) and 5 (CPE-Y306A/L315A) animals. e, Experimental design for assessing the impact of wt-CPE treatment on the epicardium-derived cardiomyocyte population. f, The epicardial contribution to cardiomyocytes is reduced by wt-CPE treatment. Representative images of DAPI, YFP, MHC and Tenascin-C immunostaining of sections of vehicle- and wt-CPE-treated hearts at 21 d.p.ci. The epicardium–myocardium border is marked by dashed lines. Insets: magnified views of the regions in the white boxes. The arrows show YFP+MHC+ cells in the regenerate. Scale bars, 200 µm (main images) and 10 µm (insets). g, Percentage of epicardium-derived cardiomyocytes in the regenerate at 21 d.p.ci. Unpaired two-tailed Student’s t-test, **P = 0.006. Data are the mean ± s.d. of n = 5 (vehicle-treated) and 4 (wt-CPE-treated) animals.

Source data

Next, we determined whether the wt-CPE treatment reduced the number of epicardium-derived cardiomyocytes. We injected TAT-Cre into Cherry-loxP-H2B::YFP reporter animals, treated them with either vehicle or wt-CPE and used echocardiography to measure the injury size and estimate the number of cardiomyocytes in the regenerate (Fig. 6e). We found that approximately 4% (percentage ± s.e.m., 4.27 ± 0.62; n = 5 animals) of the cardiomyocytes in the regenerate were derived from the epicardium in the vehicle-treated hearts at 21 d.p.ci. Given that the labelling efficiency of epicardial cells was about 29%, we estimate that epicardium-derived cardiomyocytes account for approximately 15% of new cardiomyocytes at 21 d.p.ci. (Fig. 6f,g). In comparison, wt-CPE-treated hearts showed a substantial reduction in the number of epicardium-derived cardiomyocytes (percentage ± s.e.m., 1.45 ± 0.20; n = 4 animals), confirming the outcome of wt-CPE treatment (Fig. 6f,g). Together, these data show that the impaired regeneration, which at least in part is caused by the reduction of epicardium-derived cardiomyocytes, is not compensated by the contribution of new cardiomyocytes from elsewhere.

Cardiogenic transition state revealed by scRNA-seq analyses

To molecularly profile the conversion of epicardial cells to cardiomyocytes and determine the identity of the honeycomb-forming cells, we performed scRNA-seq on live cells isolated using fluorescence-activated cell sorting (FACS) from the injury area of vehicle- and wt-CPE-treated hearts at 7 d.p.ci. (Fig. 7a). To supplement the live-cell sort and enrich for EPDCs, we also sorted cells using an antibody to CLDN6 (Fig. 7a). Seven distinct cell clusters were identified through unbiased clustering and marker-gene expression (Fig. 7b, Extended Data Fig. 10a and Supplementary Table 1). By comparing clusters across each cell-isolation strategy, we identified Cluster 0 as an intermediate cell cluster that both disappeared following wt-CPE treatment and was enriched when cells were sorted with anti-CLDN6 (Fig. 7c). Notably, Cluster 0 showed expression of well-established cardiac transcription factors such as Gata4, Gata6, Foxc1 and Foxc2, and expressed extracellular-matrix markers such as Tenascin-X, Fibulin5 and Collagen6 (Fig. 7d and Extended Data Fig. 10b). Co-immunostaining of CLDN6+ and isolectin-B4 excluded an endothelial cell identity for these cells (Extended Data Fig. 10c). Trajectory analysis of Cluster 0 suggested a differentiation path, where cells initially expressed genes related to EMT, such as Twist1, Klf8 and Fgfr2, and subsequently upregulated the expression of cardiac muscle genes such as Myl3, Myl4 and Tnnc1 (Fig. 7e,f and Extended Data Fig. 10d). Based on the expression of the early EMT marker Snail1 (ref. 67) and the absence of cardiomyocyte genes in the transitioning cells (Fig. 3d and Extended Data Table 1), we infer that they precede the intermediate cell state. These data suggest that injury induces transcriptional reprogramming of the epicardial cells into a cardiomyocyte fate via an intermediate CLDN6+ state. To further test this model, we combined lineage tracing of EPDCs in GFP-loxP-Cherry reporter animals with in situ hybridization against Twist1, Gata4, Gata6 and Myl3 at 7 d.p.ci. (Fig. 7g,h). We found that the Cherry+ EPDCs expressed Twist1 and Gata4 closer to the epicardium (Fig. 7g) and activated the expression of Myl3 after entering the injury area (Fig. 7h). These observations corroborate the proposed differentiation trajectory inferred from the scRNA-seq analysis. Collectively, the results provide molecular evidence for injury-induced activation of epicardial cell-to-myocyte conversion.

Fig. 7: scRNA-seq of CLDN6+ EPDCs reveal transcriptional reprogramming towards cardiomyocyte fate.
figure 7

a, Experimental design for obtaining single cells from vehicle- and wt-CPE-treated hearts at 7 d.p.ci. The dashed lines show apex isolation. b, UMAP visualizing the scRNA-seq data from 681 single cells isolated at 7 d.p.ci. c, Cluster 0 represents CLDN6+ differentiation intermediates. UMAP showing the distribution of cells in relation to the treatment and FACS isolation strategy (left). UMAP showing live cells isolated from hearts treated with vehicle (i) or wt-CPE (ii) as well as CLDN6+ cells isolated from the vehicle-treated hearts (iii) (right). d, Heatmap showing specific marker-gene expression for the different cell clusters. e, Pseudotime trajectory of CLDN6+ differentiation intermediates. The arrow indicates the cardiomyocyte differentiation trajectory. f, Expression, and cell density plots of genes related to EMT (top), cardiogenesis (cardiac transcription factors, TFs; middle) and muscle cell function (bottom) across pseudotime. g, TAT-Cre recombinase-mediated lineage tracing in GFP-loxP-Cherry reporter salamanders combined with in situ hybridization shows that EPDCs express Twist1 and Gata4 as they migrate into the injury area at 7 d.p.ci. Overview image showing the injury area and the epicardial layer covering it (left). Magnified views (middle and right). Scale bars, 20 µm (overview; left) and 10 µm (magnified views; middle and right). The dashed lines separate the epicardium. The arrows mark cells expressing Twist1 and Gata4. The bright autofluorescence in the injury area is unspecific staining of blood cells. h, TAT-Cre recombinase-mediated lineage tracing in tandem with in situ hybridization and immunostaining shows that EPDCs express Myl3 after entering the injury area at 7 d.p.ci. Overview image showing a section of the injury area (top); the dashed line marks epicardium-IA border. Magnified views of (i) showing Gata6+Myl3 cells in the epicardial layer surrounding the injury area (middle). Magnified views of (ii) showing Gata6+Myl3+ cells in the injury area (bottom). CHERRY+ cells are outlined by dashed lines. Scale bars, 10 µm. g,h, Data represent three animals. IA, injury area.

Disruption of tight junctions impedes cardiac regeneration

To assess the importance of focal tight junctions connecting CLDN6+ EPDCs, we took advantage of c-CPE, the non-toxic binding domain of CPE that disrupts tight junctions by temporarily displacing CLDNs (Fig. 5c)68. Fusion of recombinant c-CPE to Strep-Tag II enabled visualization of c-CPE binding to CLDN6+ cells by immunostaining (Fig. 8a). Labelled cell clusters with clear cell geometry features were observed in the injury area of animals treated with c-CPE at 7 d.p.ci. for 30 min (Fig. 8a,b). This cellular architecture was lost by 12 h post injection, as evidenced by dispersion of the clusters and the labelled cells adopting a rounded shape (Fig. 8b). This indicates that c-CPE treatment is sufficient to disturb established cell contacts between EPDCs, thus affecting cell morphology and tissue organization.

Fig. 8: Disruption of tight junctions inhibits regeneration.
figure 8

a, c-CPE binds to CLDN6+ cells in the injury area at 7 d.p.ci. Representative images of DAPI, CLDN6 and c-CPE immunostaining 30 min after c-CPE injection. The arrowheads mark a CLDN6+c-CPE+ cell. Data represent three animals. b, Treatment with c-CPE disrupts the honeycomb-like structure. Representative images of DAPI, F-actin and c-CPE immunostaining at 7 d.p.ci. The arrowheads mark cell–cell contacts. Data represent three animals per time point. a,b, Scale bars, 10 µm. c, Schematic of the in vivo c-CPE treatment regimen. d, Recovery from injury was monitored. The injury area was calculated for each animal by echocardiography at the indicated time points and normalized to the pre-treatment injury size at 3 d.p.ci. Data are the mean ± s.d. Two-way ANOVA with Tukey’s test; 10 d.p.ci., *P = 0.0130; 14 d.p.ci., **P = 0.0024; and 21 d.p.ci., **P = 0.0022. e,f, Levels of regeneration at 14 (e) and 21 d.p.ci. (f). Score: mild, <45%; moderate, 45–75%; and severe, >75%, based on the remaining injury size at 14 and 21 d.p.ci., respectively. χ2 test; ****P < 0.0001. g, The c-CPE-treated hearts have larger injuries at 21 d.p.ci. Representative images of DAPI, MHC and pan-CK immunostaining (left). AFOG staining (right). Scale bars, 200 µm. IA, injury area. dg, n = 6 animals per treatment group. h, Border-zone cardiomyocyte proliferation is not hindered by c-CPE treatment. Representative images of DAPI, MHC and PCNA immunostaining at 10 d.p.ci. The dashed lines show the border-zone area. The boxed areas in the main images are magnified (right). The arrowheads mark PCNA+ cardiomyocytes. Scale bars, 200 µm (overview; left) and 10 µm (magnified view; right). i, Percentage of PCNA+ cardiomyocytes in the injury border zone from h at 10 d.p.ci. The box-and-whiskers plots show the mean (+), median (horizontal line), quartiles (boxes) and range (whiskers). Two-tailed Mann–Whitney test: NS, not significant (P = 0.4206); n = 5 animals per treatment group. j, Model showing the conversion of EPDCs into cardiomyocytes in response to injury.

Source data

To determine how the disruption of tight junctions and subsequent dispersion of CLDN6+ EPDC clusters may affect regeneration, cryoinjured salamanders were injected daily with c-CPE (3–10 d.p.ci.) and regeneration was monitored until 21 d.p.ci. by echocardiography (Fig. 8c and Supplementary Video 5). Hearts treated with c-CPE displayed impaired regeneration across the 21-d.p.ci. period, with the effect detectable as soon as 10 d.p.ci. Notably, the inhibitory effect persisted up to 21 d.p.ci., despite the treatment being terminated at 10 d.p.ci. (Fig. 8d–f), which was also confirmed at the histological level (Fig. 8g). These observations suggest a critical role for the focal tight junctions early in regeneration. Importantly, the c-CPE treatment had no effect on the regeneration of the epicardial cell layer, as assessed by pan-CK staining (Fig. 8g), and did not affect border-zone cardiomyocyte proliferation, as assessed by PCNA expression (Fig. 8h,i). Together, these data indicate that disruption of focal tight junctions between EPDCs is sufficient to delay cardiac regeneration without impinging on the epicardial integrity or cardiomyocyte proliferation.

Discussion

Here we show that epicardium-derived intermediates migrate into the injured myocardium, create CLDN6+ honeycomb-like structures connected via focal tight junctions and become cardiomyocytes in salamanders (Fig. 8j).

The epicardium plays multiple essential roles in vertebrate heart regeneration. Epicardial cells secrete paracrine factors to neighbouring cells, including cardiomyocytes, that in the regenerative zebrafish re-enter the cell cycle and proliferate to replace lost cardiomyocytes19. Epicardial cells also produce extracellular-matrix components required for re-vascularization and muscle regeneration69. To what extent epicardial cells convert into other cell types during regeneration has remained an open question, mostly because of the lack of suitable lineage-tracing tools25. Although their contribution to fibroblasts, perivascular cells, smooth muscle cells and adipocytes is generally accepted, conversion to cardiomyocytes remains controversial25. Here we used a genetic marker-independent lineage-tracing strategy that demonstrates the conversion of epicardial cells to cardiomyocytes in salamanders. This occurs at a low rate during homeostasis and is accentuated in response to cryoinjury to regenerate cardiac muscle. Due to the relatively low labelling efficiency, it is not possible to fully quantify the contribution of epicardial cells to new cardiomyocytes but we estimate that at least 15% of the regenerated myocardium is derived from epicardium. This contribution is substantial, as indicated by observations that its prevention by ablation of the EPDCs impairs heart regeneration in salamanders. In addition to cardiomyocytes, we observed EPDCs coexpressing the mesenchymal marker vimentin or the smooth muscle cell/myofibroblast marker ɑ-smooth muscle actin, indicating that epicardial cells also give rise to non-myocyte lineages. It will be important in the future to discern, both in quantitative and qualitative terms, the different roles the epicardium has in salamanders during homeostasis as well as after injury.

The specificity of TAT-Cre-mediated recombination is a key element of our study, for which we present two lines of evidence. First, nuclear Cre is detected only in epicardium. Small amounts of Cre protein diffusing into the myocardium are sequestered in the extracellular matrix, which precludes transduction of non-epicardial cells (Extended Data Figs. 1b and 2d)37. Second, we show that ablation of lineage-labelled epicardial cells reduces the number of labelled cardiomyocytes during both homeostasis and regeneration (Fig. 6f,g and Extended Data Fig. 8e–j). It has been reported in other species that transient homotypic or heterotypic cell fusion could trigger cell-cycle re-entry of cardiomyocytes70,71. Theoretically, contribution from such a mechanism cannot be fully discounted in salamanders either. However, observations that regeneration is inhibited by the non-cytotoxic c-CPE in the absence of an effect on cardiomyocyte proliferation further supports the model that cellular contribution by the epicardium, rather than cell fusion between epicardial cells and cardiomyocytes, is essential for cardiac regeneration in salamanders (Fig. 8h,i).

The ablation studies using CPE rely on the expression of Claudins by epicardial cells and their progeny. The use of two different versions of the toxin, one ablating EPDCs and the other disrupting tight junctions, revealed that tight junctions per se are necessary for cardiac regeneration. These junctions are present both in the epicardial layer during homeostasis and in the EPDCs that invade the injury area. Tight junctions act as paracellular barriers for the passage of ions and solutes. They also function as mechanosensors bridging mechanical cues to the signalling platforms that regulate cytoskeletal organization and gene expression72. We observed that EPDCs form an intriguing honeycomb-like shape. Cells organized in a honeycomb-like pattern are widely found in natural contexts, ranging from retinal pigment epithelium73 to endothelial cells during angiogenesis74,75. This is thought to give mechanical support to tissues76. Muscle cells in the heart are surrounded by collagen sheaths that are organized in a honeycomb-like network77,78, which inspired the bioengineering of bilaminar scaffolds yielding electrically excitable grafts with multi-layered heart cells of neonatal rats79. It is not inconceivable that EPDCs migrating to the injury area provide support to regenerating cardiomyocytes regardless of their origin. In addition, the Hippo–YAP pathway has been shown to both translate mechanical forces into biochemical signals80,81 and regulate differentiation of EPDCs in the developing mouse heart, potentially in response to mechanical tension82. Intriguingly, CLDN6 regulates the nuclear localization of YAP1 (refs. 83,84) and renders lineage plasticity to hepatocellular carcinoma cells84, indicating a link between fate switching and CLDN6.

Through lineage tracing, scRNA-seq and toxin-mediated ablation we reveal an epicardial cell-to-cardiomyocyte conversion trajectory, characterized by the expression of EMT markers, followed by signature genes expressed in cardiac muscle. Further studies will be needed for two principal goals. First, to identify putative epicardial subpopulations with cardiomyocyte potential. Second, to understand how to stimulate such a transition in species where this does not naturally occur, such as mammals. The present data indicate that epicardial plasticity serves as a basis for naturally occurring regeneration in a vertebrate, thereby highlighting the relevance of targeting the epicardium in mammals as a strategy complementary to stimulating cardiomyocyte proliferation.

Methods

Animals

All of the procedures in this study related to animal handling, care and treatment were performed according to the guidelines approved by the Jordbruksverket/Sweden under the ethical permit numbers 18190-18 and 5723-2019. P. waltl were bred and maintained in our aquatic animal facility85. The transgenic lines tgTol2(CAG:loxP-GFP-loxP-Cherry)Simon and tgTol2(CAG:Nucbow)Simon were described previously38.

Transgenesis

Tol2-CAG:-loxP-mCherry-stop-loxP-h2bYFP plasmid and Tol2 transposase were injected into single-cell eggs to generate transgenic salamanders86,87. Only F1 and F2 CHERRY+ animals, in which the transgene is ubiquitously expressed, were used for experiments.

Tissue processing and histological analysis

Animals were anaesthetized in 0.1% MS222 (Sigma, A5040). Their hearts were excised, washed in 70% PBS with heparin (100 units ml−1; Sigma, H4784) and fixed with 4% paraformaldehyde for 2 h at room temperature (RT) or overnight at 4 °C (Santa Cruz, CAS 30525-89-4) for immunostaining and rinsed three times before being exposed to 10, 20 and 30% sucrose (Sigma, S0389) solutions at 4 °C. The tissues were then equilibrated to O.C.T. compound (Tissue-Tek, 4583) by immersing them in a 1:1 mixture of 30% sucrose and O.C.T., followed by 100% O.C.T. and finally frozen. Sections (10–14 µm) were prepared on a cryostat at −14 °C (Thermo, Cryostar NX70) in the frontal plane and stored at −80 °C.

To obtain fresh frozen tissue sections, hearts were mounted in 7% tragacanth (Sigma, G1128), dipped in pre-cooled 2-metyhlbutane (Sigma, 277258) and frozen in liquid nitrogen.

Staining procedures

For immunofluorescence staining, tissue sections were fixed in 4% formaldehyde (Millipore, 1.00496.5000) for 5 min at RT and washed three times (5 min each) with PBS. The tissue was permeabilized with PBS containing 0.25% Triton X-100 (Sigma, T8787) for 10 min, followed by three washes of 5 min with PBS. The sections were blocked in 1% bovine serum albumin (Sigma, A7906), 10% normal goat serum (Invitrogen, 31873) and 0.1% Triton X-100 in PBS for 1 h at RT. The samples were then incubated overnight with primary antibodies in blocking solution at 4 °C. The following day, the sections were washed three times (10 min each) with PBS containing 0.1% Triton X-100 (PBS-T) and rinsed with PBS. The samples were incubated with secondary antibodies diluted in blocking solution for 2 h at RT. For nuclear staining, the samples were incubated with DAPI (2 µg ml−1; Sigma, D9542) for 10 min, followed by five washes with PBS-T and one wash with PBS (5 min each). The sections were incubated with 0.1% Sudan black (Sigma, 199664) in 70% ethanol for 3 min and rinsed under water. The slides were mounted with DAKO fluorescent mounting medium (S3023, Agilent).

For the CLDN6 staining shown in Fig. 4b,d,f, fresh frozen tissue sections were fixed with pre-chilled methanol for 2 min and rehydrated in 75, 50 and 25% methanol in PBS (1 min each), followed by three washes in PBS (5 min each). The sections were blocked (1% bovine serum albumin and 10% normal goat serum in PBS) for 1 h at RT and incubated overnight with the primary antibodies at 4 °C. The following day, the slides were washed three times (10 min each) with PBS containing 0.1% Tween-20 and rinsed with PBS. Subsequently, the sections were incubated with the secondary antibodies for 2 h at RT, followed by incubation with DAPI for 10 min at RT. The slides were washed five times (5 min each) with PBS containing 0.1% Tween-20, rinsed with PBS and mounted. For the CLDN6 staining shown in Fig. 4a,c,e, the regular immunofluorescence protocol on fixed frozen tissue was followed.

For PCNA staining, the sections were incubated in citrate buffer (Sigma, C9999-100ML) at 90–95 °C for 15 min, allowed to cool at RT for 20 min and permeabilized in 0.2 % Triton X-100 for 15 min.

To detect proliferating cells, animals were injected with 0.1 mg g−1 5-ethynyl-2′-deoxyuridine (900584, Sigma), which was detected using a Click-IT EdU cell proliferation kit for imaging (C10340, Thermo Fisher).

For the AFOG staining, a dye solution was prepared by dissolving dye powder in distilled water to a final concentration of 0.5% aniline blue (Acros Organics, 401180250), 1% orange G (Sigma, O7252) and 1.5 % acid fuchsin (Sigma, F8129). The pH was set to 1.1 with hydrochloric acid (Sigma, H1758). Tissue sections were fixed in 10% neutral buffered formalin (Sigma, HT501128) for 10 min at RT and rinsed with PBS. The sections were incubated in Bouin’s solution (Sigma, HT10132) for 2 h at 60 °C, followed by 1 h at RT. The slides were washed under water for 30 min and incubated for 5 min in 1% phosphomolybdic acid (Sigma, HT153), followed by a wash in distilled water for 5 min. Tissue was stained in AFOG solution for 8 min and rinsed in water for 5 min. The sections were rinsed twice in 95% ethanol and twice in 100% ethanol (5 min each). Finally, the slides were rinsed twice with xylene (Sigma, 214736) for 2 min, air dried and mounted in Organo/Limonene mounting medium (Sigma, O8015).

The following antibodies were used: guinea pig anti-CLDN6 (1:500; custom-made against the peptides TASQPRSDYPSKNYV and CPKKDDHYSARYTATA), rabbit anti-CLDN6 (1:50; Abcam, 107059), mouse anti-CLDN6 (1:30; Thermo Fisher, MA5-24076), mouse anti-MYH-1 (1:200; DSHB, MF-20), rabbit anti-cytokeratin (1:250; Abcam, 9377), rabbit anti-GFP (1:500; Life Technologies, A-6455), chicken anti-GFP (1:1,000; Abcam, ab13970), mouse anti-Cre (1:500; US Biological, C7920), rabbit anti-Cre (1:500; Abcam, ab216262), rabbit anti-RFP (1:200; Rockland, 600-401-379), rat anti-RFP (1:200; Life Technologies, M11217), mouse anti-ɑ-actinin (1:800; Sigma, A7811), rabbit anti-ɑ-smooth muscle actin (1:100; Abcam, ab5694), isolectin GS-IB4 (1:200; Thermo Fisher, I21411 or I32450), chicken anti-vimentin (1:200; Millipore, AB5733) and Alexa Fluor 488 phalloidin (1:500; Thermo Fisher, A12379). Highly cross-adsorbed Alexa Fluor-conjugated secondary antibodies raised in goats were used at a 1:1,000 dilution. Specifically, anti-chicken 488 (A11039), anti-chicken 647 (A21449), anti-guinea pig 555 (A21435), anti-mouse IgG1 488 (A21121), anti-mouse IgG1 647 (A21240), anti-mouse IgG2b 647 (A21241), anti-mouse IgG2b 488 (A21141), anti-mouse IgG2b 568 (A21144), anti-mouse IgG2b 647 (A21242), anti-rabbit 488 (A11034), anti-rabbit 568 (A11011), anti-rabbit 647 (A21245) and anti-rat 568 (A11077). All secondary antibodies were obtained from Thermo Fisher.

Nuclear Cre signal quantification

High-magnification images were obtained using a Zeiss LSM900 confocal microscope. For each animal, at least 20 randomly selected areas (2,02 mm × 1,78 mm; scaled) capturing the epicardium–myocardium border were imaged. Macros were generated to quantify the mean intensity and raw integrated density of the nuclear Cre signal using ImageJ. Briefly, the pan-cytokeratin signal was used to define the epicardium. The DAPI signal was used to define cell nuclei. The nuclear Cre signal in a negative control area corresponding to the middle of the tissue was measured to determine the mean intensity threshold as 25 (average mean intensity + 3 × s.d.). False positives with a mean intensity of >25 and raw integrated density of <18,000 were excluded.

Biotin permeability assay

EZ-Link Sulfo-NHS-LC-Biotin (21335, Thermo Fisher) solution was prepared at a concentration of 1 mg ml−1 in amphibian PBS (aPBS). Biotin (10 μl) was microinjected (FemtoJet 4i, Eppendorf) into the pericardial cavity of the animals and allowed to perfuse for 10 min. The biotin was quenched with 100 mM glycine in aPBS and hearts were fixed overnight. Sections were stained with the addition of Rhodamine Red-X-conjugated streptavidin (1:500 of 1 mg ml−1 stock; Thermo Fisher, S6366) for 30 min at RT.

Pericardial injections of the TAT-Cre recombinase

TAT-Cre recombinase (70–100 µM; Millipore, SCR508) diluted in 10 µl amphibian HBSS (aHBSS) was microinjected (FemtoJet 4i, Eppendorf) into the pericardial sac of the reporter salamanders. The animals were kept in water at 25 °C. For labelling combined with the injury, the animals were subjected to cryoinjury at 40 h.p.i. After the injury the animals were kept in water at 18 °C and the temperature was raised to 25 °C at 10 d.p.ci.

Cryoinjury

Post-metamorphic salamanders (7–10 cm) were anesthetized by immersion in 0.1% MS222 (Sigma) and placed in a supine position. A skin incision was made and the pericardium was cut open. The heart was manoeuvred out of the pericardial cavity. Excess moisture was removed with a tissue. A liquid nitrogen-cooled copper filament (1.2 mm) was applied to the ventricle apex for 10 s. The heart was then placed back into the pericardial cavity. The pericardium was sealed with ethilon 8-0 black monofilament sutures (Ethicon) and the outer skin was sealed with Vicryl Rapid 6-0 (Ethicon). The animals were placed in 0.5% sulfamerazine solution (Sigma, S0800) on ice overnight and transferred to regular water the next day. The water temperature was raised to 25 °C at 10 d.p.ci.

Heart dissociation and FACS

We adopted a dissociation protocol biased towards non-myocyte cell populations of the heart88. Hearts were rinsed with ice-cold 70% aHBSS (Sigma, 55037C). The ventricles were minced into smaller pieces and collected in cold aHBSS. The tissue pieces were allowed to settle, rinsed once with aHBSS and incubated with 2 mg ml−1 Collagenase/Dispase (Sigma, 10269638001) in aHBSS for 2 h at 27 °C. The tissue pieces were rinsed with aHBSS, resuspended in aPBS with 10% fetal bovine serum (FBS) and mechanically broken with the help of a pipette. Cells were passed through a 100-µm filter and spun down at 300g for 5 min at 4 °C. The pellets were resuspended in 1 ml aPBS with 1% FBS. The cells were stained with Sytox Blue dead-cell stain (1:1,000; Thermo Fisher, S34857), calcein AM (1:250; Thermo Fisher, C1430) and Vybrant DyeCycle ruby (1:1,000; Thermo Fisher, V10273) and then sorted on a FACS Aria III system (BD Biosciences) using a 130-µm nozzle.

To establish a milder dissociation protocol to capture the cells in the injury area at 7 d.p.ci., the apical regions of the hearts were removed with a scalpel and minced into smaller pieces that were collected in aHBSS. The tissue pieces were rinsed once with cold aHBSS and incubated in digestion buffer containing 1.5 mg ml−1 bovine serum albumin (Sigma, A7906), 3 mg ml−1 glucose (Sigma, G6152) and 2 mg ml−1 Collagenase/Dispase (Sigma, 10269638001) in aHBSS. The tubes were placed in a 27 °C waterbath for 30 min with shaking (90 r.p.m.). After 30 min, solution containing tissue pieces and dissociated cells was gently pipetted up and down without disturbing the larger tissue pieces and collected in a separate tube with FBS. Fresh digestion buffer was added to the remaining tissue pieces and the procedure was repeated.

To perform CLDN6 staining on isolated cells, 5 µg anti-CLDN6 (Abcam, 107059) or rabbit isotype control (Abcam, ab171870) was conjugated to Dylight 650 (Abcam, 201803). The cells were incubated with the conjugated antibodies for 1 h at 4 °C and washed twice with FACS buffer to remove the unbound antibodies. Sytox blue and vybrant orange staining was performed. The FACS data were analysed using the FACSDiva and FlowJo software.

scRNA-seq

The single-cell transcriptome data were generated at the Eukaryotic Single-cell Genomics facility of the Science for Life Laboratory in Stockholm, Sweden.

Single-cell libraries were sequenced on a HiSeq2500 system by Illumina. Single-end 43-base-pair reads were mapped to the P. waltl reference coding sequences consolidated in orthology groups using STAR (v.2.5.3a)89. Reads mapping uniquely to one orthology group were assigned to that group and a final matrix of unique read counts per orthology group was used for downstream analysis.

Analysis of the scRNA-seq was performed using Seurat package version 4.0.1 (ref. 90). Genes expressed in fewer than three cells and cells with less than 200 expressed genes were omitted. Cells with total read counts lower than 5,000 were also discarded. After quality control, data from 2,908 cells were analysed. The average counts of individual cells were 155,888. Total read counts and cell-cycle regression were performed and normalized with the CellCycleScoring() and ScaleData() functions in Seurat. The FindVariableGene() function was used to identify 2,000 variable genes, which were subjected to subsequent analysis. Principal component analysis was performed, and the first 12 (Fig. 3) and 24 (Fig. 7) principal components were used for UMAP dimension reduction analysis. The UMAP was used for graphic-based clustering using the FindNeighbors and FindClusters functions.

Top markers of individual clusters were identified using the FindAllMarkers function. The top 300 marker genes with the highest log-transformed fold change and P < 0.05 were subjected to over-representation analysis using the protein analysis through evolutionary relationships (PANTHER) tool91. The analysis was conducted with the PANTHER GO-slim Molecular Function, PANTHER GO-slim Biological Process and Reactome pathway databases, using an over-representation test by Fisher’s exact test with a false discovery rate cutoff of 5%. Features in the scRNA-seq data were used as the background.

Monocle 3 was used for single-cell trajectories and pseudotime analysis based on the UMAP generated in the Seurat analysis92,93,94,95. Genes that were differentially expressed across the pseudotime axis were identified using the graph_test() function. The corresponding heatmaps were plotted using the plot_pseudotime_heatmap function in Monocle 2.

Scatter, bar, dot and violin plots, and other data representation graphs were generated using the ggplot2 R package.

RNAscope in situ hybridization

Custom RNAscope probes against Cldn6, Cldn7, Cldn15, Dkk2, Cd248, Twist1, Gata4, Gata6 and Myl3 were designed and manufactured by Advanced Cell Diagnostics (ACD). Catalogue probe against Cherry was used to detect the CHERRY signal (ACD, 431201-C2 and 431201-C3). RNAscope fluorescent multiplex assay (ACD, 320850), RNAscope 2.5 HD duplex assay (ACD, 322430) and RNAscope 2.5 HD assay-RED (ACD, 322350) were performed according to the manufacturer’s recommendations, with minor modifications: fixed frozen tissues were treated with protease III for 15 min. High-magnification images were acquired with a Zeiss LSM700 confocal microscope or Zeiss AxioScan Z1. The Zen Blue, CaseViewer and HALO software were used for visualization and quantification.

TEM

Transmission electron microscopy was performed at the TEM unit (EMil) of the Karolinska Institute. Salamander hearts were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4, and stored at 4 °C. Following the primary fixation, the hearts were rinsed with 0.1 M phosphate buffer and post-fixed in 2% osmium tetroxide in 0.1 M phosphate buffer, pH 7.4, at 4 °C for 2 h. The hearts were then stepwise ethanol dehydrated, followed by acetone and embedded in LX-112. Semi- and ultrathin sections were prepared using a Leica EM UC7 ultramicrotome. The ultrathin sections were contrasted with uranyl acetate, followed by lead citrate and examined in a Tecnai 12 Spirit Bio TWIN transmission electron microscope operated at 100 kV. Digital images were acquired using a 2kx2k Veleta CCD camera.

Phylogenetic tree and sequence alignment

Multiple sequence alignment of CLDN6 was performed using the T-Coffee programme with default settings using the EMBL-EBI96,97. A phylogenic tree was generated using the ClusterW2 package in the EMBL-EBI API98. The alignment and tree were visualized using Jalview and TreeDyn99,100,101.

Bulk RNA sequencing

Sequencing libraries were prepared using the TruSeq stranded total RNA sample prep LS protocol. Reads were mapped and annotated as described earlier. Trimmed mean of M values normalization was performed using the edgeR package102.

Molecular cloning

The P. waltl Cldn6 sequence was retrieved from the P. waltl genome. Amino acids at positions 140–157 were removed to create the PwCldn6Δ sequence, where the ECL2 domain was removed. The T2A-H2B-EBFP2 sequence was inserted to the 5′ end of the stop codon in wild-type P. waltl Cldn6 and the PwCldn6Δ sequence, as a selection marker. The recombinant sequences were synthesized as double-stranded DNA fragments (IDT gblock) and inserted into a piggyBac-CAG expression plasmid using an infusion kit (Takara). Plasmids were transformed into One Shot Stbl3 chemically competent Escherichia coli (Thermo Fisher, C737303).

Cell culture and transfection

The HEK293T cell line was purchased from the American Type Culture Collection (CRL-3216). The cells were maintained in Dulbecco’s modified eagle medium with 10% FBS (Life Technologies). The cells were transfected using Lipofectamine 2000 (Invitrogen, 11668-027).

Cell viability assay

HEK293T cells (10,000) were plated into each well of a 96-well plate coated with poly-l-lysine (Sigma, P5899). The cells were exposed to wt-CPE, Y306A/L315A or c-CPE diluted in culture medium (3 mg ml−1) the following day. A Pierce LDH cytotoxicity assay kit was used to determine the levels of cytotoxicity (Thermo Scientific, 88953).

Expression and purification of NH2-His-tagged CPE

The pET16b-10×HIS-CPE plasmid was generated by GenScript103,104. An overnight culture (10 ml) of BL21-Codon Plus (DE3)-RIL competent cells (Agilent Technologies, 230245) transformed with 60 ng pET16b-10×HIS-CPE was inoculated into 1 l of LB medium containing 100 mg ml−1 carbenicillin and 25 mg ml−1 chloramphenicol. The culture was cultured at 37 °C with vigorous shaking to an optical density at 600 nm of 0.5–0.6 and induced with 1 mM isopropyl-1-thio-β-d-galactopyranoside (Thermo Fisher, R1171) for 3 h. The cells were harvested by centrifugation at 4,000g for 20 min at 4 °C. The cell pellets were resuspended in native lysis buffer containing 50 mM NaH2PO4, 0.5 M NaCl, 1 mM imidazole, rLysozyme (60 KU g−1 cell paste; Millipore, 71110), benzonase nuclease (250 units ml−1; Sigma, E1014-25KU) and EDTA-free protease inhibitor (Thermo Fisher, A32965), pH 8.0, and kept on ice for 30 min, followed by sonication with Bandelin Sonopuls (Cycle 2, 6 min, 50% power output). The suspension was centrifuged at 26,500g for 30 min at 4 °C. The supernatant was applied to Ni-NTA agarose affinity resin for 1 h at 4 °C. Unbound proteins were washed away with a buffer containing 20 mM imidazole and the His-tagged wt-CPE was eluted with 250 mM imidazole. Elutions containing wt-CPE were pooled and concentrated using a Pierce protein concentrator (10 kDa molecular-weight cutoff; Pierce, 88517). Triton X-114-mediated endotoxin removal was performed105 and high-affinity Triton-binding beads (Bio-Rad, 152-8920) were used to remove the residual TX-114. Zeba spin desalting columns (Thermo Fisher, 89890) were used to remove the residual salts. Purified wt-CPE protein was kept at −80 °C.

Expression and purification of Strep-Tag II-tagged Y306A/L315A, CPE and c-CPE

BL21-Codon Plus (DE3)-RIL competent cells (Agilent Technologies, 230245) were transformed with 50 ng Strep-Tag II–CPE_pET22b(+), Strep-Tag II–cCPE_pET22b(+) or Strep-Tag II–Y306A/L315A_pET22b(+), generated by GenScript. An overnight culture (10 ml) was used to inoculate 1 l of LB medium containing 100 mg ml−1 carbenicillin and 25 mg ml−1 chloramphenicol. When the absorbance at 600 nm reached 0.5–0.7 after incubation at 37 °C with vigorous shaking, the culture was induced with 1 mM isopropyl-1-thio-β-d-galactopyranoside (Thermo Fisher, R1171). After 3 h, the cells were harvested by centrifugation at 4,000g for 20 min at 4 °C. The pellets were resuspended in lysis buffer containing 100 mM Tris–HCl, 150 mM NaCl, 1 mM EDTA (iba, 2-1003-100), rLysozyme (60 KU g−1 cell paste; Millipore, 71110), benzonase nuclease (250 units ml−1; Sigma, E1014-25KU) and EDTA-free protease inhibitor (Thermo Fisher, A32955), pH 8.0, and kept on ice for 30 min, followed by sonication with Bandelin Sonopuls (Cycle 2, 6 min, 50% power output). After centrifugation of the lysed cells at 26,500g for 30 min at 4 °C, the supernatant was passed through a Strep-Tactin Sepharose column (iba, 2-1202-051). Unbound proteins were washed away with 100 mM Tris–HCl, 150 mM NaCl and 1 mM EDTA, pH 8.0 (IBA, 2-1003-100), and the Strep-Tag II-tagged protein was eluted with 2.5 mM desthiobiotin (IBA, 2-1000-025). The eluate was concentrated using a Pierce protein concentrator (10 kDa molecular-weight cutoff; Pierce, 88517) and endotoxins were removed by incubation with Triton X-114 (Sigma, X114-1L). Triton X-114 residuals were removed using high-affinity Triton-binding beads (Bio-Rad, 152-8920) and salt removal was performed with Zeba spin desalting columns (Thermo Fisher, 89890). Purified Strep-Tag II-tagged CPE, Y306A/L315A and c-CPE protein were kept at −80 °C.

Treatment of salamanders with wt-CPE and its variants

For experiments where epicardium was ablated following TAT-Cre labelling (Extended Data Fig. 8f–j), 200 ng Strep-Tag II-tagged wt-CPE in a volume of 5 µl was microinjected into the pericardial cavity of the animals.

Echocardiography

Serial echocardiography was performed using a Vevo 3100 imaging system (VisualSonics) equipped with a high-frequency transducer (MX700, 29-71 MHz). Animals were anesthetized with pH-adjusted MS222 (0.05%; pH 7.0–7.5; tricaine methanesulfonate, Sigma) and placed in the supine position (VisualSonics). B-mode parasternal long axis (PLAX) images were acquired by placing the transducer directly above the ventral side of the animal, parallel to the midline of the chest. For each animal and time point, we first obtained B-mode images corresponding to the PLAX plane containing both the ventricle and outflow tract. Within the context of regeneration, additional planes containing the largest cross-sectional area of the injury were obtained for injury-size calculations when necessary. This was achieved by scanning the transducer along the left–right axis using the in-built micromanipulator. The focus depth, two-dimensional gain and image dimensions were optimized according to the manufacturer’s recommendations and at least four cardiac cycles were recorded per imaging sequence. After the procedure, the animals were returned to water to resuscitate. Injury size was calculated by measuring the area of the whole ventricle (excluding outflow tract but including injury area) and the injury itself in Vevo Lab 3.2.0 using the 2D area function. The following formula was used: (injury size ÷ ventricle size) × 100.

Quantification

In each type of experiment, hearts from different animals were analysed in multiple independent experiments performed on different days. Six series with frontal cryosections, each containing representative regions of the heart were generated. At minimum, one of these series was immunostained and all sections were documented via high-throughput imaging to ensure fair sampling for quantification. Subsequently, at least three representative sections were selected for further quantification.

Quantification of epicardium-derived cardiomyocytes in Fig. 1i was performed by counting the number of recombined cells that were MHC+ and normalizing to the area of the section.

Quantification of nuclear Cre signal in Extended Data Figs. 1b and 3c was performed using ImageJ as described above; 698, 308, 800 epicardial cell nuclei, and 1,569, 1,719 and 3,846 myocardial cell nuclei were analysed at 30, 40 and 96 h.p.i., respectively (Extended Data Fig. 1b). Cell nuclei (3,105) in the injury area and border myocardium were analysed at 5 h.p.ci. (Extended Data Fig. 3c).

In Fig. 2g, the number of epicardium-derived cardiomyocytes was normalized to the injury area on day 3 as a proxy to the starting injury size. Serial echocardiography was used to measure the injury sizes at 3 and 21 d.p.ci. for each animal. This led to the calculation of a regeneration ratio by dividing the injury size at 3 d.p.ci. by the injury size at 21 d.p.ci. Tenascin-C staining on cryosections obtained on day 21 closely matched the injury sizes calculated via echocardiography at this time point. Therefore, measuring the Tenascin-C+ area and multiplying it by the calculated regeneration ratio allowed us to estimate the starting injury size for each animal. For the sham-operated hearts, the area corresponding to 25% of the ventricle size was measured.

To analyse the clone size for the experiments in Extended Data Fig. 4, an overview image of the apex was taken, followed by a z-stack image of the clone. Cells were considered clonally related only if they were nearby and showed identical colours.

Quantification of RNAscope data in Extended Data Fig. 6g–i was performed using the HALO software employing custom parameters for cell detection. Analysis was extended to the epicardium–subepicardium, defined as the region surrounding the myocardium.

Quantification of TUNEL+DAPI+ nuclei in Fig. 5c and Extended Data Fig. 8c was stratified to specific regions of the heart. Epicardium was defined as the pan-CK+ layer, whereas subepicardium was defined as the area between the pan-CK+ layer and myocardium. All counts were normalized to the area of each respective compartment. In the case of epicardium and subepicardium, the counts were normalized to the area of the myocardium as a proxy.

Quantification of labelled EPDCs per mm2 in Fig. 5f was done by counting the number of CHERRY+ cells in the injury area and normalizing the count to the size of the injury. Labelled cells on the epicardial layer covering the injury area were not included.

Quantification of the percentage of epicardium-derived cardiomyocytes in the regenerate in Fig. 6g was done by calculating the starting injury sizes as described in the previous section for Fig. 2f and counting the total number of cardiomyocytes in this area.

Quantification of PCNA+ cardiomyocytes within the border-zone myocardium in Fig. 8i was calculated by first counting the number of MHC+DAPI+ nuclei within a distance of 250 µm from the injury border and then counting the number of PCNA+ cells in this population. All counts were normalized to the size of the counting area to account for variations in animal-heart size.

To calculate the injury-recovery percentages in Figs. 6b and 8d, the injury sizes were measured based on serial echocardiography images and normalized to the pre-treatment injury size at 5 and 3 d.p.ci., respectively, for each animal.

For experiments involving CPE-variant treatments, salamanders were pre-screened for normal heart function via echocardiography. The injury sizes at 5 (wt-CPE-related experiments) or 3 d.p.ci. (c-CPE-related experiments) were measured to ensure consistent tissue damage and animals were allocated randomly into treatment groups. Sample sizes were not pre-determined based on statistical power calculations but were based on our experience with this type of in vivo experiments. For assays in which variability is high, we typically used n ≥ 8 salamanders and repeated the assay multiple times to ensure reproducibility.

Quantification of the percentage of vimentin+ or MHC+ EPDCs in Extended Data Fig. 3l was done by counting the number of YFP+ EPDCs in the regenerating area that coexpressed the respective markers. Five sections were randomly selected for each animal to perform the quantification.

Quantification of labelled epicardial cells in Extended Data Fig. 8i was performed by counting the number of CHERRY+pan-CK+ cells in the epicardial layer and normalizing the count to the area of the ventricle.

Quantification of labelled myocytes in Extended Data Fig. 8j was performed by counting the number of CHERRY+MHC+ cells in the myocardial layer and normalizing the count to the area of the ventricle. Ten sections were randomly selected for each animal to perform the quantification.

Statistics and reproducibility

No statistical methods were used to pre-determine the sample sizes. The sample sizes were empirically estimated on the basis of pilot experiments and previously performed experiments with a similar set-up to provide sufficient sample sizes for statistical analyses. Experiments were repeated multiple times. The number of biological and technical replicates are reported in the legend of each figure. Statistical analyses were performed using GraphPad Prism 9. Two-parameter comparisons of samples from in vivo studies were performed using a two-tailed unpaired Student’s t-test. Analyses of in vivo studies with multiple parameters were performed using a two-way ANOVA with Tukey’s post-hoc test. Statistical significance was defined as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001 for all figures.

Reporting Summary

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

Data availability

Sequencing data that support the findings of this study have been deposited at the Gene Expression Omnibus under the accession code GSE180914. The PANTHER GO-SLIM database and REACTOME pathway database are publicly available (www.pantherdb.org/panther/goSlim.jsp, https://reactome.org/). All other data supporting the findings of this study are available from the corresponding authors on reasonable request. Source data are provided with this paper.

References

  1. Cano-Martínez, A. et al. Functional and structural regeneration in the axolotl heart (Ambystoma mexicanum) after partial ventricular amputation. Arch. Cardiol. Mex. 80, 79–86 (2010).

  2. Flink, I. L. Cell cycle reentry of ventricular and atrial cardiomyocytes and cells within the epicardium following amputation of the ventricular apex in the axolotl, Amblystoma mexicanum: confocal microscopic immunofluorescent image analysis of bromodeoxyuridine-label. Anat. Embryol. 205, 235–244 (2002).

    Article  Google Scholar 

  3. Becker, R. O., Chapin, S. & Sherry, R. Regeneration of the ventricular myocardium in amphibians. Nature 248, 145–147 (1974).

    Article  CAS  PubMed  Google Scholar 

  4. Oberpriller, J. O. & Oberpriller, J. C. Response of the adult newt ventricle to injury. J. Exp. Zool. 187, 249–259 (1974).

    Article  CAS  PubMed  Google Scholar 

  5. Bader, D. & Oberpriller, J. Autoradiographic and electron microscopic studies of minced cardiac muscle regeneration in the adult newt, Notophthalmus viridescens. J. Exp. Zool. 208, 177–193 (1979).

    Article  CAS  PubMed  Google Scholar 

  6. Kikuchi, K. et al. Primary contribution to zebrafish heart regeneration by gata4+ cardiomyocytes. Nature 464, 601–605 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Jopling, C. et al. Zebrafish heart regeneration occurs by cardiomyocyte dedifferentiation and proliferation. Nature 464, 606–609 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Poss, K. D., Wilson, L. G. & Keating, M. T. Heart regeneration in zebrafish. Science 298, 2188–2190 (2002).

    Article  CAS  PubMed  Google Scholar 

  9. Godwin, J. W., Debuque, R., Salimova, E. & Rosenthal, N. A. Heart regeneration in the salamander relies on macrophage-mediated control of fibroblast activation and the extracellular landscape. npj Regen. Med. 2, 22 (2017).

    Article  PubMed  PubMed Central  Google Scholar 

  10. Bergmann, O. et al. Evidence for cardiomyocyte renewal in humans. Science 324, 98–102 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Bergmann, O. et al. Dynamics of cell generation and turnover in the human heart. Cell 161, 1566–1575 (2015).

    Article  CAS  PubMed  Google Scholar 

  12. Senyo, S. E. et al. Mammalian heart renewal by pre-existing cardiomyocytes. Nature 493, 433–436 (2013).

    Article  CAS  PubMed  Google Scholar 

  13. Sadek, H. & Olson, E. N. Toward the goal of human heart regeneration. Cell Stem Cell 26, 7–16 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Tzahor, E. & Poss, K. D. Cardiac regeneration strategies: staying young at heart. Science 356, 1035–1039 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Yuan, X. & Braun, T. Multimodal regulation of cardiac myocyte proliferation. Circ. Res. 121, 293–309 (2017).

    Article  CAS  PubMed  Google Scholar 

  16. Herget, G., Neuburger, M., Plagwitz, R. & Adler, C. DNA content, ploidy level and number of nuclei in the human heart after myocardial infarction. Cardiovasc. Res. 36, 45–51 (1997).

    Article  CAS  PubMed  Google Scholar 

  17. Simões, F. C. & Riley, P. R. The ontogeny, activation and function of the epicardium during heart development and regeneration. Development 145, dev155994 (2018).

  18. Smits, A. M., Dronkers, E. & Goumans, M. J. The epicardium as a source of multipotent adult cardiac progenitor cells: their origin, role and fate. Pharmacol. Res. 127, 129–140 (2018).

    Article  CAS  PubMed  Google Scholar 

  19. Cao, J. & Poss, K. D. The epicardium as a hub for heart regeneration. Nat. Rev. Cardiol. 15, 631–647 (2018).

    Article  PubMed  PubMed Central  Google Scholar 

  20. Kikuchi, K. et al. Tcf21+ epicardial cells adopt non-myocardial fates during zebrafish heart development and regeneration. Development 138, 2895–2902 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Wills, A. A., Holdway, J. E., Major, R. J. & Poss, K. D. Regulated addition of new myocardial and epicardial cells fosters homeostatic cardiac growth and maintenance in adult zebrafish. Development 135, 183–192 (2008).

    Article  CAS  PubMed  Google Scholar 

  22. González-Rosa, J. M., Peralta, M. & Mercader, N. Pan-epicardial lineage tracing reveals that epicardium derived cells give rise to myofibroblasts and perivascular cells during zebrafish heart regeneration. Dev. Biol. 370, 173–186 (2012).

    Article  CAS  PubMed  Google Scholar 

  23. van Wijk, B., Gunst, Q. D., Moorman, A. F. M. & van den Hoff, M. J. B. Cardiac regeneration from activated epicardium. PLoS ONE 7, e44692 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Zhou, B. et al. Adult mouse epicardium modulates myocardial injury by secreting paracrine factors. J. Clin. Invest. 121, 1894–1904 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  25. Redpath, A. N. & Smart, N. Recapturing embryonic potential in the adult epicardium: prospects for cardiac repair. Stem Cells Transl. Med. 10, 511–521 (2021).

    Article  PubMed  Google Scholar 

  26. Cai, W. et al. Limited regeneration potential with minimal epicardial progenitor conversions in the neonatal mouse heart after injury. Cell Rep. 28, 190–201 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Smart, N. et al. De novo cardiomyocytes from within the activated adult heart after injury. Nature 474, 640–644 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Zangi, L. et al. Modified mRNA directs the fate of heart progenitor cells and induces vascular regeneration after myocardial infarction. Nat. Biotechnol. 31, 898–907 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Zhou, B. et al. Thymosin beta 4 treatment after myocardial infarction does not reprogram epicardial cells into cardiomyocytes. J. Mol. Cell. Cardiol. 52, 43–47 (2012).

    Article  CAS  PubMed  Google Scholar 

  30. Bettencourt-Dias, M., Mittnacht, S. & Brockes, J. P. Heterogeneous proliferative potential in regenerative adult newt cardiomyocytes. J. Cell Sci. 116, 4001–4009 (2003).

    Article  CAS  PubMed  Google Scholar 

  31. Laube, F., Heister, M., Scholz, C., Borchardt, T. & Braun, T. Re-programming of newt cardiomyocytes is induced by tissue regeneration. J. Cell Sci. 119, 4719–4729 (2006).

    Article  CAS  PubMed  Google Scholar 

  32. Uemasu, H. et al. Cryo-injury procedure-induced cardiac regeneration shows unique gene expression profiles in the newt Pleurodeles waltl. Dev. Dyn. (2021); https://doi.org/10.1002/dvdy.450

  33. Mercer, S. E., Odelberg, S. J. & Simon, H.-G. A dynamic spatiotemporal extracellular matrix facilitates epicardial-mediated vertebrate heart regeneration. Dev. Biol. 382, 457–469 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Velecela, V. et al. Epicardial cell shape and maturation are regulated by Wt1 via transcriptional control of Bmp4. Development 146, dev178723 (2019).

  35. Vivès, E., Brodin, P. & Lebleu, B. A truncated HIV-1 Tat protein basic domain rapidly translocates through the plasma membrane and accumulates in the cell nucleus. J. Biol. Chem. 272, 16010–16017 (1997).

    Article  PubMed  Google Scholar 

  36. Wadia, J. S., Stan, R. V. & Dowdy, S. F. Transducible TAT-HA fusogenic peptide enhances escape of TAT-fusion proteins after lipid raft macropinocytosis. Nat. Med. 10, 310–315 (2004).

    Article  CAS  PubMed  Google Scholar 

  37. Caron, N. J. et al. Intracellular delivery of a Tat–eGFP fusion protein into muscle cells. Mol. Ther. 3, 310–318 (2001).

    Article  CAS  PubMed  Google Scholar 

  38. Joven, A. et al. Cellular basis of brain maturation and acquisition of complex behaviors in salamanders. Dev. 145, dev160051 (2018).

  39. Robb, L. et al. epicardin: A novel basic helix-loop-helix transcription factor gene expressed in epicardium, branchial arch myoblasts, and mesenchyme of developing lung, gut, kidney, and gonads. Dev. Dyn. 213, 105–113 (1998).

    Article  CAS  PubMed  Google Scholar 

  40. Moore, A. W., McInnes, L., Kreidberg, J., Hastie, N. D. & Schedl, A. YAC complementation shows a requirement for Wt1 in the development of epicardium, adrenal gland and throughout nephrogenesis. Development 126, 1845–1857 (1999).

    Article  CAS  PubMed  Google Scholar 

  41. Kraus, F., Haenig, B. & Kispert, A. Cloning and expression analysis of the mouse T-box gene Tbx18. Mech. Dev. 100, 83–86 (2001).

    Article  CAS  PubMed  Google Scholar 

  42. Phillips, M. D., Mukhopadhyay, M., Poscablo, C. & Westphal, H. Dkk1 and Dkk2 regulate epicardial specification during mouse heart development. Int. J. Cardiol. 150, 186–192 (2011).

    Article  PubMed  Google Scholar 

  43. Gaur, A. et al. Expression of HoxA5 in the heart is upregulated during thyroxin-induced metamorphosis of the Mexican axolotl (Ambystoma mexicanum). Cardiovasc. Toxicol. 1, 225–235 (2001).

  44. Favara, D. M., Banham, A. H. & Harris, A. L. A review of ELTD1, a pro-angiogenic adhesion GPCR. Biochem. Soc. Trans. 42, 1658–1664 (2014).

  45. Lilly, A. J., Lacaud, G. & Kouskoff, V. SOXF transcription factors in cardiovascular development. Semin. Cell Developmental Biol. 63, 50–57 (2017).

    Article  CAS  Google Scholar 

  46. Cheng, C. et al. Endothelial cell-specific FGD5 involvement in vascular pruning defines neovessel fate in mice. Circulation 125, 3142–3158 (2012).

    Article  PubMed  Google Scholar 

  47. Hernández-García, R., Iruela-Arispe, M. L., Reyes-Cruz, G. & Vázquez-Prado, J. Endothelial RhoGEFs: a systematic analysis of their expression profiles in VEGF-stimulated and tumor endothelial cells. Vasc. Pharmacol. 74, 60–72 (2015).

    Article  CAS  Google Scholar 

  48. Kanamori-Katayama, M. et al. LRRN4 and UPK3B are markers of primary mesothelial cells. PLoS ONE 6, e25391 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Lecuit, T. & Yap, A. S. E-cadherin junctions as active mechanical integrators in tissue dynamics. Nat. Cell Biol. 17, 533–539 (2015).

    Article  CAS  PubMed  Google Scholar 

  50. Fujiwara, H. et al. Human blastocysts and endometrial epithelial cells express activated leukocyte cell adhesion molecule (ALCAM/CD166). J. Clin. Endocrinol. Metab. 88, 3437–3443 (2003).

    Article  CAS  PubMed  Google Scholar 

  51. Trzpis, M., McLaughlin, P. M. J., De Leij, L. M. F. H. & Harmsen, M. C. Epithelial cell adhesion molecule: more than a carcinoma marker and adhesion molecule. Am. J. Pathol. 171, 386–395 (2007).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Müller, P. S. et al. The fibronectin leucine-rich repeat transmembrane protein Flrt2 is required in the epicardium to promote heart morphogenesis. Development 138, 1297–1308 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Tsukita, S., Tanaka, H. & Tamura, A. The claudins: from tight junctions to biological systems. Trends Biochem. Sci. 44, 141–152 (2019).

  54. Turksen, K. & Troy, T. C. Claudin-6: a novel tight junction molecule is developmentally regulated in mouse embryonic epithelium. Dev. Dyn. 222, 292–300 (2001).

    Article  CAS  PubMed  Google Scholar 

  55. Anderson, W. J. et al. Genetic targeting of the endoderm with claudin–6CreER. Dev. Dyn. 237, 504–512 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. Wang, L. et al. Claudin 6: a novel surface marker for characterizing mouse pluripotent stem cells. Cell Res. 22, 1082–1085 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  57. Zavala-Zendejas, V. E. et al. Claudin-6, 7, or 9 overexpression in the human gastric adenocarcinoma cell line AGS increases its invasiveness, migration, and proliferation rate. Cancer Invest. 29, 1–11 (2011).

    Article  PubMed  Google Scholar 

  58. Shrestha, A., Uzal, F. A. & McClane, B. A. The interaction of Clostridium perfringens enterotoxin with receptor claudins. Anaerobe 41, 18–26 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  59. Mitchell, L. & Koval, M. Specificity of Interaction between Clostridium perfringens enterotoxin and claudin-family tight junction proteins. Toxins 2, 1595–1611 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  60. Vecchio, A. J., Rathnayake, S. S. & Stroud, R. M. Structural basis for Clostridium perfringens enterotoxin targeting of claudins at tight junctions in mammalian gut. Proc. Natl. Acad. Sci. USA 118, e2024651118 (2021).

  61. Saitoh, Y. et al. Structural insight into tight junction disassembly by Clostridium perfringens enterotoxin. Science 347, 775–778 (2015).

    Article  CAS  PubMed  Google Scholar 

  62. Becker, A. et al. Functionalization of gold-nanoparticles by the Clostridium perfringens enterotoxin C-terminus for tumor cell ablation using the gold nanoparticle-mediated laser perforation technique. Sci. Rep. 8, 14963 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. Piontek, J. et al. Formation of tight junction: determinants of homophilic interaction between classic claudins. FASEB J. 22, 146–158 (2008).

    Article  CAS  PubMed  Google Scholar 

  64. Takahashi, A. et al. Domain mapping of a claudin-4 modulator, the C-terminal region of C-terminal fragment of Clostridium perfringens enterotoxin, by site-directed mutagenesis. Biochem. Pharmacol. 75, 1639–1648 (2008).

    Article  CAS  PubMed  Google Scholar 

  65. Eichner, M., Protze, J., Piontek, A., Krause, G. & Piontek, J. Targeting and alteration of tight junctions by bacteria and their virulence factors such as Clostridium perfringens enterotoxin. Pflug. Arch. 469, 77–90 (2017).

    Article  CAS  Google Scholar 

  66. Protze, J. et al. Directed structural modification of Clostridium perfringens enterotoxin to enhance binding to claudin-5. Cell. Mol. Life Sci. 72, 1417–1432 (2015).

    Article  CAS  PubMed  Google Scholar 

  67. Tran, D. D., Corsa, C. A. S., Biswas, H., Aft, R. L. & Longmore, G. D. Temporal and spatial cooperation of Snail1 and Twist1 during epithelial–mesenchymal transition predicts for human breast cancer recurrence. Mol. Cancer Res. 9, 1644–1657 (2011).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  68. Sonoda, N. et al. Clostridium perfringens enterotoxin fragment removes specific claudins from tight junction strands. J. Cell Biol. 147, 195–204 (1999).

    Article  PubMed  PubMed Central  Google Scholar 

  69. Wang, J., Karra, R., Dickson, A. L. & Poss, K. D. Fibronectin is deposited by injury-activated epicardial cells and is necessary for zebrafish heart regeneration. Dev. Biol. 382, 427–435 (2013).

    Article  CAS  PubMed  Google Scholar 

  70. Sawamiphak, S., Kontarakis, Z., Filosa, A., Reischauer, S. & Stainier, D. Y. R. Transient cardiomyocyte fusion regulates cardiac development in zebrafish. Nat. Commun. 8, 1525 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  71. Yellamilli, A. et al. Abcg2-expressing side population cells contribute to cardiomyocyte renewal through fusion. FASEB J. 34, 5642–5657 (2020).

    Article  CAS  PubMed  Google Scholar 

  72. Citi, S. The mechanobiology of tight junctions. Biophys. Rev. 11, 783–793 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  73. Valiente-Soriano, F. J. et al. Tracing the retina to analyze the integrity and phagocytic capacity of the retinal pigment epithelium. Sci. Rep. 10, 1–18 (2020).

    Article  CAS  Google Scholar 

  74. Nagata, D., Mogi, M. & Walsh, K. AMP-activated protein kinase (AMPK) signaling in endothelial cells is essential for angiogenesis in response to hypoxic stress. J. Biol. Chem. 278, 31000–31006 (2003).

    Article  CAS  PubMed  Google Scholar 

  75. Park, H. J. et al. Human umbilical vein endothelial cells and human dermal microvascular endothelial cells offer new insights into the relationship between lipid metabolism and angiogenesis. Stem Cell Rev. 2, 93–102 (2006).

    Article  CAS  PubMed  Google Scholar 

  76. Wang, S. et al. Honeycomb structure is promising for the repair of human bone defects. Mater. Des. 207, 109832 (2021).

    Article  CAS  Google Scholar 

  77. Hanley, P. J., Young, A. A., LeGrice, I. J., Edgar, S. G. & Loiselle, D. S. 5-Dimensional configuration of perimysial collagen fibres in rat cardiac muscle at resting and extended sarcomere lengths. J. Physiol. 517, 831–837 (1999).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  78. Macchiarelli, G. et al. A micro-anatomical model of the distribution of myocardial endomysial collagen. Histol. Histopathol. 17, 699–706 (2002).

    CAS  PubMed  Google Scholar 

  79. Engelmayr, G. C. et al. Accordion-like honeycombs for tissue engineering of cardiac anisotropy. Nat. Mater. 7, 1003–1010 (2008).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  80. Dupont, S. et al. Role of YAP/TAZ in mechanotransduction. Nature 474, 179–184 (2011).

    Article  CAS  PubMed  Google Scholar 

  81. Cai, X., Wang, K.-C. & Meng, Z. Mechanoregulation of YAP and TAZ in cellular homeostasis and disease progression. Front. Cell Dev. Biol. 9, 673599 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

  82. Xiao, Y. et al. Hippo signaling plays an essential role in cell state transitions during cardiac fibroblast development. Dev. Cell 45, 153–169 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  83. Yu, S. et al. CLDN6 promotes tumor progression through the YAP1–snail1 axis in gastric cancer. Cell Death Dis. 10, 949 (2019).

  84. Kong, F. E. et al. Targeting tumor lineage plasticity in hepatocellular carcinoma using an anti-CLDN6 antibody-drug conjugate. Sci. Transl. Med. 13, eabb6282 (2021).

  85. Joven, A., Kirkham, M. & Simon, A. in Salamanders in Regeneration Research (eds. Kumar, A. & Simon, A.) 47–70 (Humana Press, 2015); https://doi.org/10.1007/978-1-4939-2495-0_4

  86. Sandoval-Guzmán, T. et al. Fundamental differences in dedifferentiation and stem cell recruitment during skeletal muscle regeneration in two salamander species. Cell Stem Cell 14, 174–187 (2014).

    Article  CAS  PubMed  Google Scholar 

  87. Khattak, S. et al. Optimized axolotl (Ambystoma mexicanum) husbandry, breeding, metamorphosis, transgenesis and tamoxifen-mediated recombination. Nat. Protoc. 9, 529–540 (2014).

    Article  CAS  PubMed  Google Scholar 

  88. Pinto, A. R., Chandran, A., Rosenthal, N. A. & Godwin, J. W. Isolation and analysis of single cells from the mouse heart. J. Immunol. Methods 393, 74–80 (2013).

    Article  CAS  PubMed  Google Scholar 

  89. Elewa, A. et al. Reading and editing the Pleurodeles waltl genome reveals novel features of tetrapod regeneration. Nat. Commun. 8, 2286 (2017).

  90. Butler, A., Hoffman, P., Smibert, P., Papalexi, E. & Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411–420 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  91. Mi, H. et al. PANTHER version 7: improved phylogenetic trees, orthologs and collaboration with the Gene Ontology Consortium. Nucleic Acids Res. 38, 204–210 (2009).

    Article  CAS  Google Scholar 

  92. Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by pseudotemporal ordering of single cells. Nat. Biotechnol. 32, 381–386 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  93. Qiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. Nat. Methods 14, 979–982 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  94. Cao, J. et al. The single-cell transcriptional landscape of mammalian organogenesis. Nature 566, 496–502 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  95. Levine, J. H. et al. Data-driven phenotypic dissection of AML reveals progenitor-like cells that correlate with prognosis. Cell 162, 184–197 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  96. Notredame, C., Higgins, D. G. & Heringa, J. T-coffee: a novel method for fast and accurate multiple sequence alignment. J. Mol. Biol. 302, 205–217 (2000).

    Article  CAS  PubMed  Google Scholar 

  97. Madeira, F. et al. The EMBL-EBI search and sequence analysis tools APIs in 2019. Nucleic Acids Res. 47, W636–W641 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  98. Larkin, M. A. et al. Clustal W and Clustal X version 2.0. Bioinformatics 23, 2947–2948 (2007).

    Article  CAS  PubMed  Google Scholar 

  99. Waterhouse, A. M., Procter, J. B., Martin, D. M. A., Clamp, M. & Barton, G. J. Jalview version 2-A multiple sequence alignment editor and analysis workbench. Bioinformatics 25, 1189–1191 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  100. Dereeper, A. et al. Phylogeny.fr: robust phylogenetic analysis for the non-specialist. Nucleic Acids Res. 36, 465–469 (2008).

    Article  CAS  Google Scholar 

  101. Chevenet, F., Brun, C., Bañuls, A. L., Jacq, B. & Christen, R. TreeDyn: towards dynamic graphics and annotations for analyses of trees. BMC Bioinform. 7, 439 (2006).

    Article  CAS  Google Scholar 

  102. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140 (2010).

    Article  CAS  PubMed  Google Scholar 

  103. Kitadokoro, K. et al. Crystal structure of Clostridium perfringens enterotoxin displays features of β-pore-forming toxins. Biol. Chem. 286, 19549–19555 (2011).

    Article  CAS  Google Scholar 

  104. Santin, A. D. et al. Treatment of chemotherapy-resistant human ovarian cancer xenografts in C.B-17/SCID mice by intraperitoneal administration of Clostridium perfringens enterotoxin. Cancer Res. 65, 4334–4342 (2005).

    Article  CAS  PubMed  Google Scholar 

  105. Teodorowicz, M. et al. Optimized Triton X-114 assisted lipopolysaccharide (LPS) removal method reveals the immunomodulatory effect of food proteins. PLoS ONE 12, e0173778 (2017).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  106. Loulier, K. et al. Multiplex cell and lineage tracking with combinatorial labels. Neuron 81, 505–520 (2014).

    Article  CAS  PubMed  Google Scholar 

  107. Veshnyakova, A. et al. On the interaction of Clostridium perfringens enterotoxin with claudins. Toxins 2, 1336–1356 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  108. Krause, G. et al. Structure and function of claudins. Biochim. Biophys. Acta 1778, 631–645 (2008).

    Article  CAS  PubMed  Google Scholar 

  109. Winkler, L. et al. Molecular determinants of the interaction between Clostridium perfringens enterotoxin fragments and claudin-3. J. Biol. Chem. 284, 18863–18872 (2009).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We thank S. Kagioglou for technical help; E. Balado, C. Donaldson, E. Arvidsson, A. Karakatsanis and A. Giannakakis for their help with raising animals; T. Davit-Béal for the donation of adult P. waltl; A. Becker and A. Ngezahayo for sharing the Strep-Tag II–c-CPE plasmid and protein purification protocol; L. Haag and K. Hultenby for their help with the interpretation of the TEM images; G. Palano and E. Hansson for helping with the establishment of the cell-dissociation protocol; E. Tanaka for scientific discussions leading to the use of TAT-Cre for lineage tracing; F. Salomons for helping with microscopy and automated image analysis training; and the members of the Chien and Simon laboratories for rigorous scientific discussions. Analysis of P. waltl gene expression was enabled by resources provided by the Swedish National Infrastructure for Computing at the Uppsala Multidisciplinary Center for Advanced Computational Science, partially funded by the Swedish Research Council through grant agreement no. 2018-05973. Some illustrations (Figs. 2c, 5a and 7a) were created with BioRender.com. This work was supported by a Swedish Research Council Distinguished Professor Grant (Dnr 541-2013-8351; K.R.C.), the Swedish Heart Lung Foundation (grant no. 20140623; K.R.C.), the Wallenberg Foundation (KAW Dnr 2013.0028; K.R.C.), AstraZeneca Pharmaceuticals (ICMC) (K.R.C.), the Karolinska Institutet (K.R.C), the Swedish Research Council (grant no. 2018-02443; A.S.), Cancerfonden (grant no. 19 0417 Pj 01 Hx; A.S), Stiftelsen Olle Engkvist (A.S.), SFO Stem Cells and Regenerative Medicine (A.S.), a EMBO long-term fellowship (grant no. ALTF-729-2015; E.E), a NIH Ruth Kirschstein postdoctoral fellowship (grant no. F32GM117806; A.E.) and the German Research Foundation (grant no. GO3220/1-1; A.G.).

Funding

Open access funding provided by Karolinska Institute.

Author information

Authors and Affiliations

Authors

Contributions

Conceptualization: N.W. and K.R.C. initiated the salamander heart regeneration project, with input from A.S. E.E., A.S. and K.R.C. conceived and co-designed the current CLDN6+ EPDC study. Methodology: E.E., N.W., A.J.A. and H.W. Formal analysis: E.E., C.Y.T.Y., Y.-L.T., A.E. and A.G. Investigation: E.E., C.Y.T.Y., T.S., C.H.U. and J.S. Resources: A.J.A. and H.W. Writing of the original draft: E.E., C.Y.T.Y. and A.S. Writing—review and editing: E.E., C.Y.T.Y, A.S. and K.R.C. Visualization: E.E., C.Y.T.Y. and Y.-L.T. Funding acquisition: A.S. and K.R.C.

Corresponding authors

Correspondence to Elif Eroglu, András Simon or Kenneth R. Chien.

Ethics declarations

Competing interests

K.R.C. is a scientific founder and equity holder in Moderna Therapeutics and Procella Therapeutics, and chair of the External Science Panel for AstraZeneca. The remaining authors declare no competing interests.

Peer review

Peer review information

Nature Cell Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data

Extended Data Fig. 1 TAT-Cre recombinase transduces epicardial cells.

a, Nuclear Cre expression is detectable 30-h post injection in the epicardial layer. Representative immunostaining showing DAPI, Cherry, YFP and Cre at 30-, 40- and 96-h post TAT-Cre recombinase injection (h.p.i.) into the pericardial sac of Cherry-loxP-H2B::YFP. Arrowheads mark nuclear Cre+ epicardial cells. Dashed line marks epi-myo border. Scale bars, 20 µm. Data shown represent 9 animals at 30 h.p.i., 6 animals at 40 h.p.i., 5 animals at 96 h.p.i. b, TAT-Cre recombinase does not transduce myocardial cell nuclei. Quantification of the number of Cre+ nuclei in the epicardium and myocardium at 30-, 40- and 96-h.p.i. Dashed line at mean intensity of 25 depicts the threshold separating negative versus positive nuclei. Note that scale is adjusted based on nuclear Cre signal in the epicardium at 30-h.p.i. Insets show data plotted with different scale to provide magnified view. See methods for details of the quantification.

Source data

Extended Data Fig. 2 TAT-Cre recombinase selectively labels epicardial cells.

a, TAT-Cre recombinase induces epicardium specific recombination. Overview image of a ventricle immunostained for YFP and pan-CK 40-h post TAT-Cre injection (h.p.i.). Scale bars, 500 µm. b, Table showing the percent labelled epicardial cells at 40-h.p.i. Pan-CK+ cells were counted to determine the number of epicardial cells. c, Vehicle injection does not result in recombination. Representative immunostaining for DAPI, CHERRY, YFP and pan-CK 96-h post vehicle injection. Dashed line marks epi-myo border. Boxed area is magnified. Arrowheads mark epicardial cells. Scale bars, 20 µm (left), 10 µm (right). d, Cre is localized to the myocardial extracellular matrix at 96-h.p.i. Representative immunostaining showing DAPI, SɑA, wheat-germ agglutinin (WGA) and Cre at 96-h post TAT-Cre recombinase injection into the pericardial sac of Cherry-loxP-H2B::YFP. Scale bars, 10 µm. e, Representative immunostaining for MHC, YFP and pan-CK on TAT-Cre recombinase injected Cherry-loxP-H2B::YFP heart sections at 40- and 96-h.p.i. Orthogonal views of YFP+ cells in the epicardium (at 40-h.p.i. and 96-h.p.i.) and myocardium (at 96-h.p.i.). Dashed line shows epi-myo border. Scale bars, 10 µm. f, At 96-h.p.i., epicardium-derived cardiomyocytes are morphologically immature. Representative immunostaining for DAPI, CHERRY and SɑA on TAT-Cre recombinase injected GFP-loxP-Cherry heart section at 96-h.p.i. CHERRY+ epicardium-derived cardiomyocyte is outlined. Scale bars, 10 µm. g, At 11-dpi, epicardium-derived cardiomyocytes show sarcomeric organization. Representative immunostaining for DAPI, CHERRY and SɑA on TAT-Cre recombinase injected GFP-loxP-Cherry heart section at 11-dpi. CHERRY+ epicardium-derived cardiomyocyte is outlined. Boxed area shown in the overview image is magnified. Scale bars, 20 µm (overview), 10 µm (close-up). h, Representative images of the ventricle before and immediately after cryoinjury, showing the damage to the apex. Scale bars, 1 mm. i, Representative still images of longitudinal axis views of the regenerating salamander heart at 3-, 10-, 14- and 21-d.p.ci. obtained by ultrasound imaging. Ventricle is outlined by dashed lines and injury area is pseudocolored in the bottom panels. Asterisk marks the lumen. OFT, outflow tract. a, b, c, d, e, f, g, i, Data shown represent 6 animals (a, b, d), 3 animals (c, e), 4 animals (f, g) and 16 animals (i).

Extended Data Fig. 3 Salamander epicardial cells are multipotent.

a, Scheme showing the experimental design. b, Representative overview and close up images showing immunostaining for DAPI, YFP and Cre. Injury area (IA) is delineated by dashed line. Boxed areas are magnified. Scale bars, 100 µm (overview), 50 µm (i), 10 µm (ii). c, Quantification of Cre+ nuclei in the injury area and myocardium 5-hpci. 3105 cells were analysed, n = 3 animals. Dashed line shows the threshold. d, Representative immunostainings of DAPI, pan-CK, MHC and EdU at 5-d.p.ci. Dashed lines separate epicardium and myocardium. Scale bars, 100 µm. e, Schematic showing the experimental design. f, Arrows mark YFP+ epicardial cells. Scale bars, 100 µm (overview), 10 µm (close-up). Data shown represent 6 animals (40 h.p.i.) and 5 animals (48 hpci). g, Quantification of labelled epicardial cell frequency (percentage of YFP+, pan-CK+ cells) before (n = 6 animals) and after cryoinjury (n = 5 animals). Box and whiskers plot shows mean (+), median, quartiles (boxes), and range (whiskers). h, Representative immunostaining on TAT-Cre recombinase injected GFP-loxP-Cherry heart sections, showing DAPI, CHERRY and MHC at 21-days post sham operation. Boxed area is shown in the inset, arrow marking a labelled cardiomyocyte. Scale bars, 200 µm (overview), 20 µm (inset). i-l, Epicardium-derived fibroblasts/myofibroblasts are found in the injured area at 21-d.p.ci. Experimental scheme as shown in Fig. 2c. i, Representative immunostainings showing DAPI, MHC, Vimentin (Vim) and CHERRY at 21-d.p.ci. Boxed area is magnified. Arrow marks CHERRY+, Vim+ cell. Scale bars, 20 µm (overview), 10 µm (close up). j, Representative immunostainings showing DAPI, MHC, ɑ-Smooth muscle actin (ɑ-SMA) and CHERRY at 21-d.p.ci. Boxed area is magnified. Arrow marks CHERRY+, ɑ-SMA+ cell. Scale bars, 50 µm (overview), 10 µm (close up). k, Representative immunostainings showing DAPI, MHC, Isolectin-B4 and CHERRY at 21-d.p.ci. Boxed area is magnified. Arrows show CHERRY, Isolectin-B4+ cells. Scale bars, 20 µm (overview), 10 µm (close up). l, Percentage of epicardium-derived cells at 21-d.p.ci. b, d, h, i, j, k,l, Data shown represent 3 animals (b, d, l) and 6 animals (h, i, j, k).

Source data

Extended Data Fig. 4 EPDCs expand clonally.

a, Schematic drawing showing the Nucbow reporter construct106. b, Pie chart showing the abundance of 1- to 6-cell epicardium-derived cardiomyocyte clones at 21-d.p.ci. n = 4 animals, 80 clones. c, Representative image of heart apex showing multi-colour labelling induced by TAT-Cre recombinase in Nucbow reporter. EBFP2 (white), mEYFP (green), mCherry (orange), mCerulean (blue) and MHC (magenta). Arrows mark clones of various colours and sizes. IA, injury area. Boxed area is shown in panel d. Scale bar, 200 µm. d, Representative image showing a two-cell clone (arrows) expressing EBFP2, mCerulean and mCherry. Note that adjacent labelled cells (white and yellow arrowheads) do not show same colour combinations and therefore do not belong to the same clone. Scale bar, 10 µm. e, Representative images showing two different Z-layers of a five-cell clone expressing mEYFP and mCeru. Arrows mark mEYFP+, mCeru+ cells belonging to the clone. See Extended Data Fig. 4f for orthogonal projection. Scale bars, 10 µm. f, Orthogonal projection of the image shown in panel e. Scale bars, 10 µm. g, Scheme depicting the experimental design to assess the proliferative potential of epicardium-derived cardiomyocytes. h, Epicardium-derived cardiomyocytes are proliferative. Representative immunostaining for DAPI, YFP, MHC and PCNA at 21-d.p.ci. Arrowheads mark epicardium-derived YFP+ cardiomyocytes that are PCNA (white) and PCNA+ (red). Scale bar, 10 µm. i, Quantification of PCNA+ epicardium-derived cardiomyocytes percentage. j, Representative immunostaining for DAPI, YFP, MHC and Phospho-Histone H3 (pH3) at 21-d.p.ci. Arrowheads mark epicardium-derived YFP+ cardiomyocytes that are pH3 (white) and pH3+ (yellow). Scale bars, 10 µm. (c, d, e, f, h, i, j) Data shown represent 4 animals.

Source data

Extended Data Fig. 5 Quality control of scRNA-seq data.

a, FACS plots showing the gating strategy to obtain live, metabolically active and nucleated cells. Percentage of cells after each gate in comparison to all cells is displayed. b, Violin plot of log10 of total read counts of each individual cell in different plates obtained from single-cell sequencing. See Supplementary Table S1 for information on plate identity. c, Violin plot of log10 of the number of unique genes of each individual cell in different plates obtained from single-cell sequencing. See Supplementary Table S1 for information on plate identity. d, Table showing the number of cells recovered in each cell cluster. e, UMAP plots showing the distribution of recovered single-cells according to time point. Epicardial cells and transitioning cells are circled to highlight dynamic recovery. f, Dot plot showing the relative expression of marker genes for the 17 clusters identified in the scRNA-seq data as presented in Fig. 3b.

Extended Data Fig. 6 Cluster 11 represents the epicardium-proper.

a, Representative GO terms upregulated in candidate epicardium/epicardium-derived cell clusters. b, Heatmap showing select marker gene expression presented in Fig. 3d across all cell clusters. c, UMAP showing the expression of epithelial/mesothelial markers expressed in cluster 11. d, Dkk2 expressing cells are located to the subepicardium (marked by arrows). Representative images showing the expression of Cldn6 (red) and Dkk2 (cyan) in the sham operated hearts. Scale bars, overview; 50 µm close-up; 20 µm. Data shown represent 3 animals. e, Cd248 expressing cells are in the subepicardium and myocardial interstitium (marked by arrows). Representative images showing the expression of Cd248 (red) and Cldn6 (cyan) in the sham operated hearts. Scale bars, overview; 100 µm close-up; 20 µm. Data shown represent 3 animals. f, Epicardial cells labelled by the TAT-Cre recombinase are Cldn6+. Representative images showing in situ hybridization combined with immunostaining for Cldn6 mRNA, DAPI, CHERRY and SɑA on sections from TAT-Cre recombinase injected GFP-loxP-Cherry. Arrows mark Cldn6+ epicardial cells labelled by TAT-Cre induced recombination. Scale bars, 100 µm (i), 20 µm (ii), 10 µm (iii). Data shown represent 4 animals. g, Epicardium/sub-epicardium thickens in response to the injury. Quantification of epicardial/subepicardial cell numbers of the sham operated (n = 3 animals) and injured hearts (7-d.p.ci.) (n = 3 animals). Student’s t-test, unpaired, two-tailed. ns p = 0.1377. Data are represented as mean values ± sem. h, Cryoinjury reduces the number of cells expressing Cldn6 mRNA in the epicardium. Quantification of cells expressing Cldn6 mRNA in the epicardial and sub-epicardial space of the sham operated and injured hearts (n = 3 animals per group). Student’s t-test, unpaired, two-tailed, ***p = 0.0002, Data are represented as mean values ± SD. i, Cldn6 mRNA expression levels decrease in response to the injury. Quantification of Cldn6 mRNA expression levels in the epicardium/subepicardium cells of the sham operated (n = 3 animals) and injured hearts (7 d.p.ci.) (n = 3 animals).

Source data

Extended Data Fig. 7 CPE-sensitive Cldn expression is restricted to the epicardial cells.

a, CLDN6 is absent in the myocardium. Representative immunostaining showing DAPI, CLDN6, SɑA and pan-CK at 7-d.p.ci. See Fig. 4c for overview image. Scale bar, 10 µm. Data shown represent 4 animals. b, Cldn6 and Cldn7 are the only CPE sensitive Cldn gene family members expressed in the salamander heart. Expression of CPE sensitive Cldn genes in the salamander heart as determined by bulk RNA sequencing of the sham operated and injured hearts (4-d.p.ci. and 10-d.p.ci., n = 3 hearts per time point). Box and whiskers plot shows median, quartiles (boxes), and range (whiskers). TMM, Trimmed mean of M values. Note that Cldn9 gene is not annotated in the Pw genome and therefore not shown despite being a CPE-sensitive CLDN. c, Violin plots showing the distribution of expression for the Cldn gene family members. Cldn6, Cldn7 and Cldn15 are expressed in the epicardial cluster 11. Cldn11, a nonclassical CLDN family member that is not a CPE receptor, is expressed in cluster 9107. d, CLDN6 protein is well conserved. Amino acid sequence alignment of Pw CLDN6 with sequences from other vertebrates, H. sapiens, M. musculus, R. norvegicus, A. mexicanum, X. leavis and D. rerio. Extracellular loop 1 (ECL1) and ECL2 domains are highlighted108. Asterisk shows the key motif NP(V/L)(V/L)(P/A) within the ECL2 domain required for the CLDN-CPE interaction109. e, A phylogenetic tree of vertebrate CLDN6 protein. The scale bar indicates evolutionary distance. f, Salamander Cldn6 sensitizes HEK293T cells to wt-CPE. Quantification of cytotoxicity % in response to treatment of sensitized HEK293T cells with wt-CPE, CPE-Y306A/L315A or c-CPE for 3 hrs. Two-way ANOVA, Tukey’s multiple comparisons test. ****p < 0.0001, error bars represent sem. n = 10,000 HEK293T cells examined in 3 independent experiments for each condition.

Source data

Extended Data Fig. 8 wt-CPE treatment ablates CLDN6+ epicardium and/or EPDCs in a dose-dependent manner.

a, wt-CPE treatment causes cell death. Representative immunostainings showing TUNEL+ cells following treatment for 6 h. Shown are overview images of apex. Healthy myocardium is separated by dashed lines. Boxed areas are magnified in panel b. IA, injury area. Scale bars, 200 µm. b, Representative close-up images. Scale bars, 10 µm. c, (Left) Quantification of TUNEL+ cells upon high dose wt-CPE treatment for 6 hrs. Error bars represent ± SD. (Right) Representative images of a ventricle section immunostained for DAPI, MHC, TUNEL and pan-CK. Scale bars, 200 µm. d, Representative immunostainings showing CLDN6+, TUNEL+ EPDCs in the injury area, following wt-CPE treatment for 6 hrs. Scale bars, 50 µm (overview), 20 µm (close up). e, wt-CPE treatment ablates CLDN6+ cells in the injured area. Scheme showing the experimental design. Representative immunostaining of DAPI, CLDN6 and MHC. Scale bars, 100 µm (overview), 20 µm (close-up). f, Scheme showing the experimental design to assess the impact of epicardial cell ablation under homeostasis. g, wt-CPE treatment disrupts the epicardial layer. Representative immunostainings for DAPI, SɑA and pan-CK. Scale bars, 500 µm. h and i, wt-CPE treatment reduces the number of epicardial cells. h, Overview and close up images showing DAPI, SɑA and CHERRY immunostaining 4 days and 8 h post vehicle or wt-CPE injection. Scale bars, 500 µm (overview), 250 µm (close up), 10 µm (insets). i, Quantification of CHERRY+ epicardial cells 104 h post wt-CPE treatment. vehicle; n = 4 animals, wt-CPE; n = 5 animals. Student’s t-test, unpaired, two-tailed. *p = 0.0275. Error bars represent ± sem. j, Epicardial cell ablation reduces the number of CHERRY+ cardiomyocytes. Quantification of CHERRY+, SɑA+ cardiomyocytes 104 h post wt-CPE treatment. 10 sections for each animal are plotted. Vehicle; n = 4 animals, wt-CPE; n = 5 animals. Student’s t-test, unpaired, two-tailed. ****p < 0.0001. (a, b, c, d, e, g, h) Data shown represent vehicle n = 7 animals, wt-CPE n = 4 animals, Y306A/315 A n = 5 animals, c-CPE n = 4 animals (a), 4 animals (b, c), 3 animals (d), 3 animals for each group (e), 4 animals for the vehicle group and 5 animals for the wt-CPE injected group (g, h).

Source data

Extended Data Fig. 9 wt-CPE treatment does not affect healthy myocardium.

a, Survival curve showing the effect of wt-CPE treatment over the course of 21 days for the following treatment groups: sham + vehicle (n = 4 animals), sham + wt-CPE (n = 8 animals), injured + vehicle (n = 9 animals) and injured + wt-CPE (n = 9 animals). Vehicle or wt-CPE injection scheme as depicted in Fig. 6a. b, Immunofluorescence staining for MHC shows unaffected muscle morphology on sham operated, wt-CPE-treated hearts after 21 days. Overview and close-up images of vehicle or wt-CPE-treated heart sections stained for DAPI, pan-CK and MHC. wt-CPE treatment had a minor long-term effect on epicardial integrity as shown by pan-CK staining, while muscle tissue was unaffected. Scale bars, 200 µm (overview), 10 µm (close-up). Data shown represent 6 animals for each group. c, Representative still images of the Supplementary Movies. Ventricles are outlined by dashed lines. The injury area is pseudocolored in the bottom panels. Data shown represent 8 animals for each group.

Source data

Extended Data Fig. 10 Cardiogenic transcription programme is activated in the epicardium-derived intermediate cells.

a, Table showing the number of cells recovered in each cluster. b, UMAP showing the expression of selected genes expressed in the intermediate cells. c, Epicardium-derived CLDN6+ intermediates do not express endothelial cell marker Isolectin-B4. Representative immunostaining showing DAPI, CLDN6 and Isolectin-B4. Arrows mark Isolectin-B4+ CLDN6 cells. Scale bars, 10 µm. Data shown represent 3 animals. d, Heatmap showing differentially expressed genes along the pseudotime. Note that “cluster” refers to the hierarchical clustering of differentially expressed genes and not the UMAP clusters.

Supplementary information

Reporting Summary

Peer Review File

Supplementary Video 1

Video showing ultrasound recording of healthy ventricle previous to cryoinjury. Long axis view, half-speed.

Supplementary Video 2

Video showing ultrasound recording of vehicle-treated ventricle at 21 d.p.ci. Long axis view, half-speed.

Supplementary Video 3

Video showing ultrasound recording of wt-CPE-treated ventricle at 21 d.p.ci. Long axis view, half-speed.

Supplementary Video 4

Video showing ultrasound recording of CPE-Y306A/L315A-treated ventricle at 21 d.p.ci. Long axis view, half-speed.

Supplementary Video 5

Video showing ultrasound recording of c-CPE-treated ventricle at 21 d.p.ci. Long axis view, half-speed.

Supplementary Table

Table summarizing the details of the scRNA-seq experiments, metadata of scRNA-seq data, top 120 enriched genes of each cluster identified with the FindAllMarkers() method in Seurat, with a threshold of adjusted P < 0.05 and ranked according to –log(fold change). nCount_RNA, total number of reads; nFeature_RNA, total number of unique genes detected.

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Statistical source data.

Source Data Fig. 5

Statistical source data.

Source Data Fig. 6

Statistical source data.

Source Data Fig. 8

Statistical source data.

Source Data Extended Data Fig. 1

Statistical source data.

Source Data Extended Data Fig. 3

Statistical source data.

Source Data Extended Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 8

Statistical source data.

Source Data Extended Data Fig. 9

Statistical source data.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Eroglu, E., Yen, C.Y.T., Tsoi, YL. et al. Epicardium-derived cells organize through tight junctions to replenish cardiac muscle in salamanders. Nat Cell Biol 24, 645–658 (2022). https://doi.org/10.1038/s41556-022-00902-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/s41556-022-00902-2

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing