A dynamic transcriptional landscape of posterior regeneration
To establish a first unbiased, in-depth analysis of the transcriptomic landscape of individual cell populations during posterior regeneration in annelids, we devised a suitable sampling scheme. For this, we induced posterior regeneration by removing approximately 1/3 of the animals’ posterior tissue, including the SAZ and its rapidly proliferating progeny, amputating between segments 30 and 31. For sampling, we then isolated the posteriormost segment along with any newly regenerated tissue at distinct time points after amputation (Supplementary Fig. 1a).
We reasoned that inclusion of the last non-amputated segment in these analyses would not only provide us with data on differentiated cell types, but also allow us to detect any molecular signatures associated with the response of this segment to the adjacent wound. To assess whether this sampling scheme captured relevant molecular events in the early phases of blastema formation, we first performed a bulk RNA sequencing experiment, in which the total mRNA of each sampling time point was sequenced from biological triplicates. By using an unbiased gene-clustering approach, we determined seven major categories of gene expression dynamics over the first three days of regeneration, including four categories in which gene expression increased after amputation, with differences in the point of onset and kinetics (Supplementary Fig. 1b). Genes in these categories include known markers for stem cells and the SAZ, as well as proliferation-related transcripts (Supplementary Fig. 1b). These findings are consistent with previous observations21,24,25 and confirmed that our sampling strategy could be used to capture relevant molecular processes.
Based on these results, we devised a similar sampling scheme to build a comprehensive single cell atlas of posterior regeneration (Fig. 1a). We obtained single cells from multiple dissociated samples, representing five distinct stages of regeneration. These spanned from freshly amputated individuals (0 hpa, equivalent to uninjured trunk segments, but not the posterior-most tissues such as the SAZ and its immediate progeny) to 72 hpa, corresponding to the onset of rapid proliferation in the regenerated SAZ21, increasing the temporal resolution in early regenerative stages by adding a 12 hpa timepoint (Fig. 1a). After removing outlier and low-quality cells, we obtained a total of 80,298 transcriptomes of individual cells, sampled in two independent biological replicates of 4 and 5 timepoints, respectively (Supplementary Data 1). Even though the sampling timepoints of this single cell experiment slightly differed from those sampled in bulk (see above), we compared the two datasets using a correlation analysis. Despite the use of different sampling, sequencing and processing techniques, all replicates correspond most strongly with those in the respective other dataset sampled at the closest time point (Supplementary Fig. 1c).
a Sampling scheme illustrating posterior amputation and sampling timepoints, ranging from 0 hours post amputation (0 hpa, equivalent to a regular trunk segment) to 72 hpa, matching morphologically defined stages (st. 0 to 3). b UMAP visualization of cells, annotated by tentative cell type / population identity. c UMAP visualization showing the regenerative timepoint at which cells were sampled. d, e UMAP visualizations showing the expression of posterior identity markers (cdx, foxa) on the merged dataset, contrasting the freshly amputated sample (left) with the post-amputation time points (right) (f–i) similar UMAP visualizations, highlighting the changes in expression of stem-cell related genes (hox3, piwi, myc) (f–h) as well as the emergence of hypertranscriptomic cells (UMIs/cell) (i). j UMAP visualization of CytoTRACE values (calculated per cluster; high level indicates high differentiation potential); data for analysis provided as a Source Data file.
Single-cell data comprising multiple replicates or biological samples might suffer from batch effects, where technical differences between sampling rounds could overshadow biologically meaningful differences between samples or cell types26. To counter this effect and minimize technical variations, we took advantage of the recent establishment of a combined cell fixation and storage protocol (ACetic-MEthanol/ACME) that is compatible with single-cell sequencing27. The adaptation of this protocol for our Platynereis regenerate paradigm allowed us to sort, process and sequence cells from all sampled stages in parallel. We subsequently used standard single-cell RNA sequencing (scRNA-seq) analysis methods to process the joined data-set (see Methods).
Unbiased clustering of the cells resulted in 38 transcriptionally distinct clusters. The comparison between biological replicates and timepoints did not suggest any batch effect affecting cluster formation (Supplementary Data 1, Supplementary Fig. 2a–e). The resulting clusters, as illustrated on a uniform manifold approximation and projection (UMAP) visualization28 (Fig. 1b), correspond to cell populations of similar transcriptomic profiles. Algorithmic prediction29 identifies one cluster (cluster 26) as the possible product of doublet formation, so this cluster was not investigated further (Supplementary Data 1). We annotated these populations based on the identities of cluster-specific marker genes, and their expression levels of known annelid cell-type markers. In total, we annotated 35 of the clusters, either as known cell populations, or based on their most diagnostic marker gene (Fig. 1b, see details in Supplementary Data 2 and 3).
As each sampled tissue contains the segment adjacent to the injury site, we were able to identify a variety of cell types in our dataset. For example, an investigation of genes previously used for assigning different Platynereis muscle cell types30, allowed us to distinguish several populations of smooth (clusters 3, 6, 8, 12 and 14) and striated (clusters 2, 10, and 17) muscle. Even less abundant cell types, such as chaetal sac cells (cluster 24) which form the bristle worm’s chitinous bristles31,32 and extracellular globin-secreting cells (cluster 15)33, were identified as distinct populations. This shows that our approach yielded a high-quality cell atlas containing biologically meaningful clusters of cell populations and with sufficient sensitivity to resolve rare and poorly understood cell types.
Molecular repatterning and emerging stem cell-like properties
As outlined above, deconvolving the dynamic injury response to individual cell populations in an annelid is expected to advance our understanding of regeneration in an evolutionary context. By capitalizing on the temporal information embedded in each transcriptome of our dataset (Fig. 1c), we were able to perform comparisons of gene expression within cell populations across time.
A common challenge in complex tissue regeneration is the re-establishment of appropriate positional information, such as the position along the antero-posterior axis. To test whether our dataset could be used to identify the individual cell types involved in repatterning, we analyzed the expression of several transcription factors involved in posterior identity. Bulk RNA sequencing of posterior regeneration and unbiased clustering of genes with similar expression dynamics using mfuzz (Supplementary Fig. 1b) revealed the presence of genes encoding posteriorly expressed transcription factors such as caudal (cdx), distalless (dlx) and foxA, in gene sets upregulated after injury. This is consistent with previous suggestions that early steps in annelid regeneration include a morphallactic adjustment of positional values34.
Using the single cell atlas, we were able to add cellular resolution to this process. For example, cells of midgut identity (cluster 16) are only found in the freshly amputated sample (0 h post amputation, hpa), subsequently yielding to a population (cluster 4) demarcated by foxA and cdx as hindgut after injury (Fig. 1d, e). This morphallactic process of gut posteriorization indicated by foxA has previously been proposed in Platynereis35, demonstrating the validity of our in silico approach. In addition, a subset of neuronal populations (clusters 11, 20) expresses cdx and foxA shortly after injury (Fig. 1d, e), while two other populations (clusters 0, epithelium; cluster 9, gcm+ neurons) started to express dlx (Supplementary Fig. 3a). Similarly, we observed a molecular shift in presumptive smooth muscle cells from cluster 14 (pre-injury) to clusters 8 (post-injury), involving genes like thrombospondin, rho kinase36 and octopamine receptor 2, which play a role in muscle attachment, function and regeneration in other species37,38,39,40 (Supplementary Data 3).
As these data supported our approach to reconstructing temporal dynamics, we next investigated the expression of stem cell related genes after posterior injury. We reasoned that if stem cells are, at least in part, regenerated by dedifferentiation or activation of wound-adjacent cells, we should detect cell populations that are already present at 0 hpa, but start to express stem cell and proliferation-related markers only after injury.
To assess this point, we first investigated the expression of the homeobox gene hox3, whose transcripts are rapidly upregulated in posterior regeneration of Platynereis dumerilii41 and mostly restricted to a population of PSCs that are generally referred to as ectodermal PSCs in accordance with their presumed developmental origin18,21. Whereas homeostatic trunk cells (0 hpa) are almost entirely devoid of hox3 expression, we could detect a strong and mostly cluster-specific upregulation of this gene in post-injury time points of cluster 0 (Fig. 1f). Likewise, we find that this cluster expresses Platynereis piwi (Fig. 1g), a key member of the GMP19, and myc (Fig. 1h), both of which are expressed in Platynereis PSCs18. These data suggest that cluster 0 is a source of ectodermal PSCs.
As outlined above, hox3 is preferentially expressed in ectoderm-derived PSCs. However, additional populations of stem cells contributing to Platynereis growth and regeneration have previously been hypothesized, including mesoderm-derived PSCs21,42. We therefore systematically queried our single-cell atlas with a combined signature of stem cells (piwi, vasa, nanos), proliferation (proliferating cell nuclear antigen/pcna) and chromatin remodeling (dnmt1, chd1) These genes are expressed in cells of post-injury timepoints within several clusters, hinting at additional sources of PSCs (Fig. 1g, h, Supplementary Fig. 3b–f).
To identify the most stem-like cells in each cluster in an unbiased, systematic way, we used CytoTRACE, a computational method which assigns cells a score representative of their “developmental potential”, a proxy for stemness43. Cells were ranked by their CytoTRACE score (within each cluster), and genes correlated with this score were calculated. This analysis provides an unbiased, systematic overview of transcriptional changes within each cell population as cells acquire a higher degree of developmental potential (Source Data). We further determined gene ontology (GO) terms associated with the transcriptional changes within each cluster, providing a more comprehensive resource for the involved biological processes (Fig. 1i; Supplementary Data 4 and below).
As an additional approach to identify potential stem cells, we took advantage of the observation that Platynereis PSCs exhibit larger nuclei and nucleoli18,23, a feature usually associated with increased transcriptional (and translational) activity44. Increased, broad transcription, referred to as “hypertranscription”, is frequently observed in active stem cells and progenitors, closely associated with proliferation, and plays a role in stem cell activation and function during growth and regeneration. Recently, absolute scaling of single cell transcriptomes using Unique Molecular Identifier (UMI)-based sequencing data has been shown to identify hypertranscriptomic stem cells and progenitors45. We therefore investigated the dynamic changes in transcriptional activity upon injury in our dataset and found evidence for hypertranscription (increased numbers of total UMIs detected per cell) (Fig. 1j). Our analysis shows that there is a progressive increase in high-UMI cells during regeneration (Supplementary Fig. 3g). Hypertranscriptomic cells are located within hotspots of GMP-related gene expression and high CytoTRACE values on the UMAP (Fig. 1i) and can be found in several clusters. Our analysis also revealed a sub-population of smooth muscle cells showing high CytoTRACE values and piwi expression even before injury (cluster 12, Fig. 1b, g,ii), which could imply the existence of a dedicated progenitor state within this specific tissue.
Taken together, we found several features associated with stem cells and the SAZ in injury-adjacent cells. These features are predominantly detected after injury, increasing as posterior regeneration proceeds. We found cells that strongly display these features distributed among multiple, but not all clusters in this dataset, indicating multiple different sources of regenerating PSCs.
To understand these putative sub-populations of the regenerated SAZ, and their cellular origins, we focused our analysis on two major cell populations (clusters 0 and 1), which show a strong activation of stem-cell-related features as described above.
To characterize these populations, we identified strongly expressed marker genes and performed in situ Hybridisation Chain Reactions (HCR, see Methods and Supplementary Data 5) to detect their expression in the tissue.16,46. Co-labeling of both genes in posterior parts of uninjured, posteriorly growing animals revealed that they demarcate two spatially distinct tissues, corresponding to the ectodermal epidermis (cluster 0) and a sub-epidermal mesodermal cell type (cluster 1) (Fig. 2a–e). The latter population covers the sub-epidermal region, but does not include muscle. We therefore refer to this population as coelomic mesoderm.
a, c UMAP visualization of markers specific to cluster 0 (epig1) and 1 (ccdc134). b, d, e in situ HCR (uninjured animal, posterior end), showing mutually exclusive expression of epig1 and ccdc134 in the epidermis and coelomic mesoderm, respectively. Scale bar = 50 µm. (n = 3–5). f Nuclear staining and in situ HCR of col6a6 expression in wound-adjacent epidermal tissue at 3 timepoints after posterior amputation. Scale bar upper panels = 250 µm, lower panels = 25 µm. g Quantification of nuclear and nucleolar size change in regenerating tissue at 0, 12 and 48 hpa; each timepoint represents 3 individuals with 50 nuclei or corresponding nucleoli per individual. Statistical significance was calculated using a one-way ANOVA test with multiple comparisons. P-values for nucleus quantifications: 0pa vs 12hpa < 0.0001, 0hpa vs 48hpa < 0.0001, 12hpa vs 48hpa = 0.8862. P-values for nucleolus quantifications: 0pa vs 12hpa < 0.0001, 0hpa vs 48hpa < 0.0001, 12hpa vs 48hpa = 0.6475. Data for analysis provided as a Source Data file. h–i In situ HCR of regenerating tissue at 0 and 72 hpa (stage 3), showing the emergence of the expression of ectodermal stem cell marker hox3, combined with a ubiquitous stem cell marker (piwi) and a proliferation label (EdU, 30 min pulse before fixation); Scalebar = 50 µm. Zoom-in on hox3 positive population. j, k In situ HCR of regenerating tissue at 0 and 72 hpa (stage 3), showing the emergence of the expression of mesodermal stem cell marker prrx, combined with a ubiquitous stem cell marker (piwi) and a proliferation label (EdU, 30 min pulse before fixation); Scalebar = 50 µm. Zoom-ins on hox3– or prrx-positive population.
Distinct signatures for PSCs of ecto- and mesodermal origin
Having found evidence for distinct sources of regenerated PSCs, we next aimed to molecularly characterize them and profile them in situ. If indeed multiple populations of wound-adjacent cells acquire stem cell properties and repopulate the regenerating SAZ, our in silico data allows us to make certain testable predictions:
First, we examined whether cells of somatic origin change towards a teloblast-like morphology. As described above, Platynereis PSCs display a unique morphology with notably increased nuclear and nucleolar sizes. To test whether cells of this morphology emerge in wound-adjacent tissue, we stained tissue of posteriorly amputated Platynereis worms for the expression of collagen alpha 6(VI) chain (col6a6). Based on our CytoTRACE calculation, col6a6 is strongly expressed in epidermal cells (cluster 0) and progressively lost as they acquire PSC-like properties (Supplementary Fig. 4a, Source Data). Quantifying the surface area of nuclei and nucleoli in this population during regeneration showed a strong increase in both metrics after injury (Fig. 2f, g; Supplementary Fig. 4a–c, Source Data), along with a gradual reduction of col6a6 levels. These data are consistent with the gradual acquisition of a teloblast-like morphology.
Next, we reasoned that if these PSC-like sub-populations are distinct from each other, we should be able to find genes specifically enriched in either of them and should find their expression in distinct groups of cells in situ. As mentioned above, hox3 has previously been described as a marker predominantly expressed in ectoderm-derived PSCs, and accordingly is mostly restricted to the PSCs we identified among epidermal cells (cluster 0). Based on this observation, we sub-clustered cells of both the epidermal (cluster 0) and the coelomic mesodermal (cluster 1) populations to define their respective PSC-like sub-populations. We used CytoTRACE-scores, the total number of UMIs and the expression of GMP, SAZ, proliferation and epigenetic remodeling-related genes to identify the respective subclusters (Supplementary Fig. 4) and discovered novel molecular markers unique to these cells (Supplementary data 3; Supplementary Figs. 4d–f, 5a–s). These new markers include genes encoding putative receptors, as well as proteins with DNA binding motifs such as transcription factors, thus establishing a set of molecules with possible regulatory functions (Supplementary Data 3).
For ectoderm-derived PSCs, our analysis not only identifies the previously described genes hox3 and evenskipped (evx) but adds markers such as a gene encoding a fibronectin leucine-rich transmembrane protein of unclear orthology (flrtl, Supplementary Fig. 6f–j) and a gene (sp/btd) encoding the Platynereis homolog of the transcription factor Sp947. This population of cells further expresses early neuronal progenitor genes and patterning factors, such as the transcription factor gene soxb1 (Supplementary Data 6) and the gene four-jointed that is involved in planar cell polarity, and was previously demonstrated to be expressed in developing medial neuroectoderm36. These data are consistent with the concept that these cells are the source of new neurons in post-regenerative growth.
For the mesoderm-derived population of PSCs, our analysis also predicts distinct marker genes. These include the gene chd3/4/5b that encodes a chromodomain helicase DNA-binding protein, and has previously been detected in regenerating mesoderm24, as well as a gene we identify as Platynereis paired-related homeobox gene (prrx) (Fig. 2j, k, Supplementary Data 6; Supplementary Fig. 6k–o). The putative purinoreceptor gene p2x (Fig. 2i) and the Platynereis orthologue of the mesoderm related homeobox factor msx48,49 are also predicted to be expressed in mesoderm-derived PSCs, albeit less exclusively than prrx (Supplementary Fig. 6k–t; Supplementary Data 3).
If prrx and flrtl are novel markers of distinct populations of stem cells, they should be expressed in separate, injury-adjacent populations of cells and exhibit morphological and molecular properties of stem cells. To test this prediction, we designed specific in situ HCR probes (Supplementary Data 5) and used these to analyze the expression of both genes in posterior regenerates (Fig. 2j, k; Supplementary Fig. 4d, e; whole-mount in situ hybridisation in Supplementary Fig. 4e). In agreement with our digital data, flrtl transcripts were co-expressed with hox3 in cells of the epidermal layer, both at stage 1 and 3 (Supplementary Fig. 4d). By contrast, the predicted mesodermal stem cell marker prrx labeled cells at a deeper layer (Fig. 2j; Supplementary Fig. 4d). Consistent with the time-resolved atlas (Supplementary Fig. 6k–o), prrx was not yet detectable at stage 1 (Fig. 2l), but from stage 2 on (Fig. 2p). In both cases, a subset of labeled cells shows enlarged nucleoli as described for PSCs (arrowheads in Fig. 2h–k).
These sub-populations, based on our in silico data and their putative identity as stem cells, are predicted to be proliferating and expressing GMP genes. We therefore co-stained markers for ectoderm (hox3) and mesoderm (prrx) derived putative stem cells with the proliferation marker EdU and the key GMP factor piwi. We found both markers expressed in proliferating, piwi positive cells with teloblast-like morphology (Fig. 2h–j).
Taken together, these results are consistent with the notion that, during regeneration, distinct populations of PSCs of mesodermal and ectodermal origin derive from existing cells not displaying any stem cell related properties prior to injury. Our single-cell atlas allows the identification of novel markers of these cells.
Germ-layer-based lineage restriction of growth and regeneration
Whereas our data argued for PSCs of distinct ecto- and mesodermal origin in the blastema, it still remained unclear if these cells had identical potency, contributing to derivatives in all of the regenerate, or if they were more restricted in their developmental potential. We therefore turned towards a transgenic strategy that would allow us to address this question at least on the broad level of germ layers.
Several other lines of evidence from previous studies suggest the existence of lineage-restricted ectodermal and mesodermal stem cells during larval and juvenile posterior growth in Platynereis18,50,51,52. The early embryogenesis of Platynereis follows a stereotypical programme known as spiral cleavage53,54. Highly asymmetric cell divisions (unequal cleavage) in the early embryo produce blastomeres of characteristic sizes and positions, whose fate are strictly determined50. Micro-injection of a fluorescent lineage-tracing dye in individual blastomeres at the earliest stages of the spiral cleavage process, shows that ectodermal, mesodermal, and endodermal trunk tissues of the 4-day, three segmented larva are produced, respectively, by micromere 2d, micromere 4d, and the macromeres 4A-4D of the early embryo50. In another study, individual cells were tracked via live imaging from early embryogenesis into early larval stages to identify the fates of the mesodermal 4d blastomere52. This work revealed that the mesodermal bands and the primordial germ cells are produced by asymmetric divisions of the 4d lineage. In addition, the final divisions of the lineage during embryogenesis forms a group of undifferentiated cells at the posterior end of the hatched larvae, which will possibly become the mesodermal PSCs in later stages. Due to the transient nature of the signal (mRNA or dye injections), tracking the fate of putative PSCs into later juvenile stages was not feasible. However, molecular profiling18 suggests but does not demonstrate the existence of at least two pools of PSCs with specific signatures, ectodermal and mesodermal, organized as two concentric rings anterior to the pygidium, the terminal piece of the Platynereis trunk (Supplementary Data 7, part A). So far, no transgenic lineage tracing technique has been used to clarify the origin of tissues in the posteriorly growing or regenerating juvenile.
To address this gap, we devised a mosaic transgenesis strategy using previously-established Tol2 transgenesis methods15. We constructed a Tol2 transgenesis construct with a nuclear mCherry and a membrane EGFP52, under the control of the ribosomal protein rps9 promoter for ubiquitous expression15 (Fig. 3a). We injected several batches of zygotes at the one-cell stage with the donor plasmid containing rps9::H2A:mCherry:F2A:mGFP transgene and transposase mRNA. These G0 worms typically show mosaic integration of the transgene. We raised the G0 batches that showed high numbers of surviving juveniles (Supplementary Data 7) (Fig. 3b–j). To screen these individuals for fluorescence patterns and identify which clonal lineages had the transgene integration, we amputated juvenile worms when they reached 6 weeks. These original tails (pygidium + a few growing segments) were imaged via confocal microscopy from both the dorsal and ventral sides. The amputated worms were further raised in individual containers and allowed to regenerate their posterior parts for three weeks. They were then amputated again one segment anterior to the regenerated part to collect the regenerated posterior parts for imaging. The whole cycle was repeated once. For each transgenic individual, we thus collected pictures of the primary clones derived from transgenic blastomeres as a result of normal development, as well as pictures of two reiterative, independent regeneration events from the same primary clones originating from the non-regenerated trunk (Supplementary Data 7, part B).
a Summary sketch of the protocol for creating large embryonic clones with transposase. b–g Examples of simple germ layer– or tissue-specific primary clones observed on posteriorly growing worms; 6 weeks, ventral views. Observed frequencies of integration patterns for ectoderm = 29/61, mesoderm = 6/61, pygidial ectoderm = 53/61, median neural = 13/61, endoderm = 16/61 (Supplementary Data 7). b Ectodermal clone, no pygidial or internal cell labeled. c Mesodermal clone, only the muscles are clearly visible. d Pygidial and median neural lineage clones. Most median neurites are emanating from pygidial sensory neurons. e Endodermal clone. f–i Dorsal views of a primary clone in ectoderm-derived PSCs. f General confocal stack projection, showing the position of the ectoderm-derived PSCs and uniformly labeled nascent segments; the pygidium is labeled with independent clones. Magnified confocal section views of the SAZ region, at 2 µm z-depth (g), 6.5 µm z-depth (h) and and y-z section (i). g–i show the continuity of the clonal expression of the transgene in bottleneck-shaped ectoderm-derived PSCs with large nuclei-nucleoli (yellow arrowheads), transversely elongated columnar progenitor cells and squamous epidermal differentiated cells. For all panels, green labelings are cell membranes, magenta labelings are cell nuclei. White arrows: position of the PSCs. White asterisks: background staining. Scale bars b–f = 100 µm; g–I = 10 µm.
Overall, we found that most individuals showed complex patterns of fluorescent primary clones. Although we cannot exclude that some of the patterns observed may be due to enhancer trapping, we see no indication that this phenomenon occurs significantly in our complete set of 62 transgenic individuals, presumably due to the relative strength of the ubiquitous promoter we have used (rps9). All six individual primary clonal patterns we deduce from observations are obtained multiple times (from 4 to 53 times, in 62 individuals, Supplementary Data 7, part C), practically excluding that they may be due to neighboring endogenous enhancers. The complexity of patterns likely results from a combination of reasons: Firstly, multiple blastomeres were transformed (Supplementary Data 7, part B, N09 and N35 for examples) resulting in combinations of tissues labeled. Secondly, only a part of a germ layer-derived tissue may be labeled. This is most evident in cases where only a bilateral half of the tissues is fluorescent because transgenesis happened in only one of the bilateral descendants of the germ layer founding blastomere (e.g. 4d divides bilaterally to give the precursors of the right and left mesoderm, Supplementary Data 7 part B, M24 and N23). Thirdly, some tissues were labeled in a stochastic, salt-and-pepper manner. This phenomenon is known as variegation55 and presumably happens when a transgene is inserted near or within a heterochromatic region that imposes unstable transcriptional repression on it. This was particularly recurrent at the level of ectodermal tissues (Supplementary Data 7, part B, N25 and N33 for examples).
Despite this complexity, simpler patterns were also recovered in several individuals corresponding to the labeling of the whole trunk ectodermal tissues (Fig. 3b), the whole trunk mesodermal tissues (Fig. 3c) and the entire gut endoderm (Fig. 3e). The clonal nature of the ectodermal patterns is indicated by the continuity of expression of the transgene in PSCs, segmental precursors and differentiated segmental cells (Fig. 3f–i). Ectodermal PSCs, corresponding in location and cytological characteristics to the ring of cells identified by molecular signature before (Gazave et al.18), are easily identifiable (Fig. 3h). Potential mesodermal PSCs are also tentatively imaged in locations already identified molecularly (Supplementary data 7, part B, M03). These primary clones support the aforementioned concept that separate pools of precursor cells generate these sets of tissues during the life-long process of posterior addition of segments. As for the endoderm, so far, no endodermal PSCs have been identified by molecular signature, and it is possible that endodermal precursors or stem cells are spread in a diffuse way along the length of the trunk56.
In addition to the trunk germ layer-derived tissues, several primary clonal patterns were obtained repeatedly either alone or in combination with others (Supplementary Data 7). The pygidial ectoderm was often labeled independently of the trunk ectoderm (Fig. 3d). This demonstrates that the pygidial ectoderm is derived from blastomeres different from the trunk ectoderm and that the anterior border of the pygidial ectoderm with the trunk ectoderm is a compartment border with no contribution of the pygidial cells to the growth of the trunk ectoderm (Fig. 3b, d). The one exception to the pygidium/trunk compartmentalization is the presence of a median neural lineage (Fig. 3d), composed of two pairs of cells per new segment, that is also identified alone, sometimes unilaterally (Supplementary Data 7, part B, M09, M10, N04, N11, N14, N26, N40 and O15). These cells are probably produced by independent median specialized posterior stem cells that segregate from the 2d lineage in the early embryo. Lastly, a lineage of amoeboid, presumably phagocytic cells, possibly derived from anterior embryonic mesoderm, was observed several times (Supplementary Data 7, part B, M09, N03, N05, N22).
Most importantly, germ-layer compartmentalization is fully conserved during regenerative events (Fig. 4), with each germ layer of the regenerate originating exclusively from cells of its kind in the neighboring non-regenerated trunk. The clonal nature of the transgene expression is again illustrated by the continuous transgene expression in the differentiated epidermal cells, blastemal cells, and the regenerated PSCs (Fig. 4a–f). Ectodermal regenerated PSCs are clearly identifiable as soon as 4 days post amputation (Fig. 4c, d). Lineage restrictions in the regeneration blastema (Fig. 4g–l) are in agreement with the distinct source populations of stem cells suggested by our scRNAseq analyses. A diagram of the whole set of primary clones obtained fully supports this interpretation of compartmentalization (Supplementary Data 7, part C).
a Time lapse ventral confocal stack projections of the regenerating tail tip of a worm displaying clonal transgene expression in the ectodermal lineage. b–d magnified confocal sections of the same individual. e, f interpretative schemes of c and d. The time-lapsed views illustrate the continuity of clonal expression of the transgene in epidermal cells (a, stage 1), undifferentiated blastema cells (b), regenerated PSCs (c, d) and progenitor cells (c, d). g–l Regeneration experiments on animals bearing simple clones. Dorsal views of confocal stack projections, with pre-amputation views on top and the matching full regenerates (3 weeks post amputation) on the bottom. This series illustrates the strict compartment restriction in the regeneration of ectoderm-derived and mesoderm-derived PSCs, as well as gut endodermal lineage. Pygidial ectoderm, entirely removed upon amputation, is regenerated exclusively from trunk ectoderm precursors (g, j). For all panels, green labelings are cell membranes, magenta labelings are cell nuclei. White arrows: position of rings of PSCs in the respective focal plane. Scale bars a = 100 µm; b, c = 10 µm; d = 20 µm; g–l = 100 µm;.
While these transgenic clones do not demonstrate the embryonic germ layer origins, they show that tissues remain strictly compartmentalized during posterior segment addition, similar to embryonic/larval development. Taken together, all these results are compatible with the presence of the two rings of ectodermal and mesodermal PSCs immediately anterior to the pygidium/trunk border, while the unsegmented endoderm may grow diffusely or through the activity of specific endodermal PSCs yet to be identified. After amputation (which removes all PSCs), ectodermal and mesodermal PSCs, as well as endoderm, are regenerated exclusively from precursors of their kind in the uncut segments, either from dedifferentiating cells, or from unknown resident lineage restricted precursor cells, in complete agreement with the single-cell transcriptomics clustering.
Activation of PSCs and regeneration requires TOR signaling
The protein kinase Target Of Rapamycin (TOR) has been implied in wound response and blastemal signaling in planarians, zebrafish and axolotl57,58,59,60,61, reviewed in ref. 62. Our in silico GO-analysis revealed increased expression of TOR- related transcripts in cells as they acquire PSC identity in response to injury in Platynereis (Supplementary Data 4). These include the Platynereis orthologs of genes encoding TOR (Supplementary Data 6) as well as components of the ragulator complex (lamtor 1, 2, 3, 4 and 5), which is involved in TOR complex activation and localisation, and therefore might influence cell metabolism and proliferation63. Additionally, many biological processes known to be controlled by TOR activity were found enriched in our GO-term analysis of genes associated with high CytoTRACE scores (e.g. translation, rRNA processing, ribosome biogenesis, see Supplementary Data 4).
Moreover, increased TOR signaling has also been observed in hypertranscriptomic cells45, as we observe them in the Platynereis regeneration process, and TOR complex activity has been shown to be a key requirement for maintaining a hypertranscriptomic state in embryonic stem cells64. We therefore investigated whether posterior regeneration in Platynereis dumerilii also required a functional TOR signaling system.
In our single-cell atlas, Platynereis tor was broadly expressed, including in the tentative PSC subpopulations (Fig. 5a–c). To assess whether or not TOR signaling was required for regenerating PSCs after amputation, we treated amputated animals with the ATP competitive TOR inhibitor AZD805565 and compared their regenerative success to DMSO-treated controls (Fig. 5d–l). Already at 24 hpa, so before the formation of a significant blastema or a strong increase in cell proliferation21, in situ HCR revealed that myc expression was strongly reduced and hox3 expression was completely undetectable in treated animals (Fig. 5e, f), whereas control animals successfully established a zone of myc+ and hox3+ cells (Fig. 5i, j). Additionally, treated animals did not reach a stage 3 regenerate at 72 hpa (Fig. 5g, h). DMSO-treated control animals progressed normally and regenerated a blastema and early developing anal cirri within the same timespan (Fig. 5k, l). We therefore conclude that, upon TOR inhibition, Platynereis fails to regenerate PSCs, and subsequently does not develop a blastema or differentiated posterior tissues.
a–c UMAP visualization of tor expression; a comparison between 0 hpa and 12-72 hpa; b, c enlarged views of cluster 0 and cluster 1 in 12–72 hpa samples. d Scheme of posterior amputation, TOR inhibition and posterior regeneration after 24 and 72 h, highlighting region used for assessing stem cell gene expression 24 hpa. e–h Analysis of amputated animals (n = 6) treated with AZD8055 TOR inhibitor at 24hpa and 72hpa. e, f Confocal images of in situ HCR stainings detecting expression of hox3 and myc at 24 hpa; g, h brightfield images at 72 hpa; i–l. Equivalent analyses in DMSO-treated controls. (n = 6) (i, j) Confocal images of in situ HCR stainings for hox3 and myc at 24 hpa; k, l brightfield images of posterior regenerates at 72 hpa. Scale bars = 25 µm (e, f, i, j); = 250 µm (g, h, k, l).
- SEO Powered Content & PR Distribution. Get Amplified Today.
- PlatoData.Network Vertical Generative Ai. Empower Yourself. Access Here.
- PlatoAiStream. Web3 Intelligence. Knowledge Amplified. Access Here.
- PlatoESG. Carbon, CleanTech, Energy, Environment, Solar, Waste Management. Access Here.
- PlatoHealth. Biotech and Clinical Trials Intelligence. Access Here.
- Source: https://www.nature.com/articles/s41467-024-54041-3