Molecular evidence for early deuterostome origins of ovarian cell types and neuroendocrine control of reproduction Authors: Periklis Paganos, Beverly Naigles, Carsten Wolff, Danila Voronov, S Zachary Swartz Published: 2026 ABSTRACT AbstractOogenesis requires complex interactions between germline and somatic cell types in the ovary. How did these cell types and their signaling interactions evolve? Here we use the sea star Patiria miniata as a non-chordate deuterostome representative to define the ovarian cell type toolkit in echinoderms. Sea stars continuously produce millions of new oocytes throughout their lifespan, making them a practical system to understand the mechanisms that drive oogenesis from a biomedical and evolutionary perspective. We performed scRNA-seq combined with high-resolution 3D imaging to reveal the ovarian cell types and their spatial organization. Our data support the presence of putative oogonial stem cells, granulosa-like, and theca-like cells, which display similarities with their mammalian counterparts. We also describe possible signaling interactions between somatic and germ line cells. Lastly, our data may suggest an intrinsic ovarian neuroendocrine system which could potentially regulate oogenesis.Subject terms: Oogenesis, Evolutionary developmental biology, Non-model organisms Introduction Oogenesis is essential for the propagation of most animal species. It takes place within the ovary, where somatic and germline cells interact to form a fundamental reproductive unit: the follicle1. Somatic follicle cells, also called granulosa in mammals, provide signals and metabolic support to regulate growth and meiotic arrest in developing oocytes. Mammalian theca cells in the outer layers of follicles have a key role in steroid hormone production, nutrient delivery to the follicle through the vasculature, and regulating granulosa cell function1. Oogenesis and ovulation are regulated by neuroendocrine signals from the nervous system in both invertebrates and vertebrates. For instance, in the mollusk Lymnaea stagnalis, cells in the cerebral ganglia release an ovulation hormone that triggers egg laying2. In Drosophila melanogaster, juvenile hormone (JH) signaling promotes ovulation3, while the neurohormone methylfarnesoate suppresses sexual maturation in annelids4. In vertebrates, the hypothalamic-pituitary-gonadal (HPG) axis conveys hormonal cues from the central nervous system to the pituitary gland and subsequently to the gonads5. HPG signaling molecules include gonadotropin-releasing hormone (gnrh) and oxytocin/vasopressin (oxy/vp) that are secreted by the hypothalamus and received by gonadotropic cells within the pituitary gland. This induces the secretion of the gonadotropin hormones follicle stimulating hormone (FSH) and luteinizing hormone (LH), that in turn regulate gametogenesis within the gonad through activation of the steroidogenic pathway. In mammals, the secretion of upstream hypothalamic neuropeptides is regulated by several neuromodulators, including serotonin (5HT), dopamine (DA), gamma-aminobutyric acid (GABA), prolactin-releasing peptide (PrRP), and kisspeptin (KPP)6–10. Many germ line processes are broadly conserved, but different organisms have diverse reproductive potentials due to the presence or absence of specific cell types or states across development (Fig. 1a). For instance, the mammalian oocyte reserve is established during embryogenesis, during which time germ cells either enter meiosis and generate oocytes that remain arrested until puberty, or undergo apoptosis11. This limited number of oocytes show signs of developmental decline with age in both mice and humans11. Other animals, such as the fruit fly instead produce oocytes throughout their lifespans with germline stem cells that persist through adulthood12. However, the mechanisms controlling oogenesis and the establishment of reproductive potential in other metazoans remain less understood. Numerous evo-devo studies have identified conserved genes expressed in germline cells13, but much less is known about the evolutionary history of the somatic cell types within ovaries. Molecular characterization of the P. miniata ovary. a Phylogenetic tree showing the evolutionary relationships of P. miniata with other bilaterian animals for which ovarian cell type repertoires have been defined using single-cell transcriptomics. b P. miniata ovarian explant. Scale bar is 1 mm. c Phalloidin and acetylated tubulin staining to label cell boundaries and cilia. Scale bar is 20 μm. d IHC for α-tubulin labeling the ovarian nerve plexus. Scale bar is 40 μm. e Phalloidin staining showing ovarian muscle fibers. Scale bar is 40 μm. f Acetylated tubulin labeling cilia on the follicle cells. Dotted circle indicates oocytes, while germ cell nests are outlined with dotted lines. Scale bar is 20 μm. g Phalloidin staining labeling the cell boundaries within the germ cell nests. Germ cell nests are outlined with dotted lines. Scale bar is 20 μm. h Pan-expansion microscopy showing total protein in the nest. Scale bar is 20 μm. i Ultrastructural expansion microscopy with staining for actin using the HAK-actin probe (black) and DNA (blue) in the nest. Scale bar is 20 μm. j HCR for nanos1 (magenta) and vtg2 (yellow) labeling the entire germ lineage and follicle cells, respectively. Germ cell nests are outlined with dotted lines. Scale bar is 20 μm. Nuclei were labelled with DAPI (cyan). k, l 3D rendering of segmented germ cell nests and oocytes. Scale bar is 100 μm. Ovary seen from the lateral (k) and top (l) views. m Schematic summary of the germ cell nest, young oocyte, and mature oocyte diameters (mean ± SD) in the P. miniata ovary. fc follicle cells, l lobe, o oocytes. Echinoderms are non-chordate deuterostomes, which, together with the hemichordates14, occupy a phylogenetic position informative for both vertebrates and other invertebrates, as well as for the evolution of reproductive cell types (Fig. 1a). Echinoderms, including sea stars, continuously produce millions of oocytes, and some species have documented lifespans that can exceed a century15. However, while echinoderms have been extensively used to study fertilization, gene regulation, and embryogenesis due to the abundance of eggs they provide, we still lack a mechanistic understanding of how they achieve this remarkable reproductive output16. In this work, we asked how the reproductive potential of the sea star Patiria miniata is controlled, and whether sea star ovarian cell types are conserved with distantly related animals. Using high-resolution microscopy and single-cell RNA sequencing (scRNA-seq), we defined the cellular organization of the sea star ovary, determined its cell types, and explored their evolutionary relationship with ovarian cell types of other animals. We found the ovary to consist of epithelial cells, immune cells, neurons, oocytes, three diversified groups of follicle cells, and putative oogonial stem cells or proliferating oocyte progenitors organized in germ cell nests. Our cross-species comparison revealed potentially conserved ovarian cell types, including a granulosa-like population. Lastly, we report conserved gene expression patterns of signaling pathways involved in oogenesis, as well as intrinsic ovarian neurons with a hypothalamic-like and gonadotropic-like molecular signature. Results The structural organization of the Patiria miniata adult ovary We first sought to define the structure of the Patiria miniata ovary, which is multi-lobed and translucent (Fig. 1b). We stained for F-actin as a cell boundary marker and acetylated tubulin to visualize cilia (Fig. 1c). Each lobe is a hollow tube that consists of a thin, tightly packed epithelial cell layer with long motile cilia (Supplementary Movie 1). We identified cells organized in plexus-like structures resembling a nerve net underneath the epithelial layer that are positive for both α-tubulin and synaptotagmin 1, a pan-neuronal marker in echinoderms (Figs. 1d and S1a, e–h). Ovarian neurons appear to be present in the related sea urchin Strongylocentrotus purpuratus (Fig. S1i, j) and the sea cucumber Holothuria tubulosa (Fig. S1k, l). We also found extensive musculature below the nerve net, along with axonal projections, which may regulate their contraction during spawning17 (Figs. 1e and S1b, e, f, i, j). Below the muscle layer and within the lumen, we detected oocytes at different stages of maturation enwrapped by follicle cells, which are filamentous and ciliated (Figs. 1f and S1c, d). We next examined the stages of oocyte development. Early small oogonia are organized in closely associated groups with a diameter range of 8–23 μm and a mean ± standard deviation (SD) of 14.2 ± 2.8 μm (Figs. 1g and S2a, b). These groups were found in all ovaries tested, although their number varied in different individuals. The chromatin morphology of these cells was distinct, with condensed DNA concentrated towards the nuclear periphery (Fig. 1g, i). This may be consistent with the leptotene to zygotene stage of meiosis, in which chromosomes adopt a bouquet configuration18. These cells were compact with intense actin staining at the cell boundaries, and surrounded by small somatic nuclei that do not intercalate within the cluster (Fig. 1g). This organization resembles germ cell nests or cysts that are found in other species, including fruit flies and mammalian fetal ovaries19. Similar structures were observed by electron microscopy in the ovaries of the sea star Asterias rubens20. Co-staining for F-actin and membrane lectins revealed circular cortical structures between distinct oogonia within the germ cell groups, which are reminiscent of the intercellular bridges that connect germ cells within cysts in mammals and protostomes (Fig. S1m–o). To investigate if oogonia in these nests are connected, we used pan-expansion microscopy to visualize total protein, allowing us to trace membranes (Figs. 1h and S1p). We did not see membrane continuity between oogonia using this approach. We also used ultrastructural expansion microscopy (U-ExM) along with the HAK-actin actin probe, but did not observe the putative bridges we had seen with phalloidin staining (Fig. 1i)21. Therefore, while we cannot rule out that oogonia may be connected in cysts, we will refer to the groups as “nests” from here on. To test the germline nature of the nests, we performed HCR for the conserved germ cell factor nanos1. We found that it is specifically expressed in oocytes of all sizes as well as in the nests (Fig. 1j). While sea stars conserve three nanos genes homologous to those in mammals and other organisms, their orthology and correspondence to their human counterparts is unclear22. We also performed HCR for the yolk protein vtg2, which is expressed in sea star follicle cells23, and we detected it in follicle cells surrounding the oocytes. We found that the vtg2+ follicle cells enwrap the nests as a whole, while none of the cells within the nests are individually enwrapped (Fig. 1j). Co-staining for nanos1 and vtg2 enabled us to determine the smallest oocyte diameter necessary to form a primary follicle. We found that the young oocytes adjacent to the nests range from 16 to 42 μm in diameter (Fig. S2c, d), with a mean diameter of 29.0 ± 7.1. This population represents the smallest group of young oocytes, constituting primary follicles that are positive for nanos1. We next asked how oogenesis is spatially organized within the ovarian lobe. Is there a defined stem cell niche, or is oogenesis uniformly distributed? To test this, we used large volume light sheet microscopy (Figs. 1k, l and S2g). Segmentation analysis revealed that oogenesis is spatially polarized from outside to inside. Nests and smaller oocytes were localized entirely to the periphery of the lobe interior, but were broadly distributed. Fully grown oocytes, instead, were centrally located within the lumen (Figs. 1k–m and S2g). Binned segmented cell volumes and manually measured cell sizes (Fig. 1m) showed that the diameter of entire germ cell nests ranges from 27 to 67 μm (40 ± 8.3 μm, mean ± SD), that young oocytes located proximal to the ovarian surface range from 21 to 126 μm in diameter (57.2 ± 21.6 μm, mean ± SD), and that larger oocytes arrested in prophase I of meiosis and containing an intact germinal vesicle (pre–germinal vesicle breakdown) in the center of the lumen range from 138 to 187 μm in diameter (170.3 ± 11.8 μm, mean ± SD) (Fig. S2e, f and Supplementary Data 1). Altogether, these results define the structure of the sea star ovary and the organization of oogenesis (Figs. 1m and S2a–f). The P. miniata ovary consists of nine molecularly distinct cell type groups Having defined the structure of the ovary, we next sought to determine the molecular identities of its cell types with scRNA-seq (Figs. 2a and S3a). We used ovaries from three independent females (Fig. S3b, c and Supplementary Data 1). One ovary originated from an individual that spawned two months prior to ovary isolation (female 1), one was highly enriched in germ cell nests and early oocytes (female 2), and one consisted mostly of fully grown oocytes (female 3) (Fig. S3b, c). Computational analysis and integration of the sequenced libraries yielded an atlas of 35,877 cells distributed in twenty-seven clusters or cell states (Figs. 2a and S3a). Cell type/state diversity of the P. miniata adult ovary. a Integrated UMAP of the three different scRNA-seq libraries. Color code is indicative of different cell type groups. b Dotplot showing the average expression of known echinoderm cell type markers as well as genes labeling ovarian and associated cell types in other organisms. Color code is the same as in (a). Justification for this choice of gene markers can be found in Supplementary Data 3. Asterisks are used to indicate genes whose predicted expression pattern has been confirmed with HCR or IHC. The expression patterns can be found within this figure, Figs. 1, S1, and S3. c, d HCRs for the follicle cells markers tgfb2 (c) and foxl2 (d). e HCR for the follicle cell markers foxl2 and vtg2. Arrowheads indicate foxl2 and vtg2 double positive cells. f FISH using an antisense probe for the transcription factor figla labelling the germ lineage. Germ cell nests in (f) are outlined with dotted lines. g, h HCRs for the germ line markers piwil1 and ddx4. Germ cell nests in (g and h) are outlined with dotted lines. i Schematic representation of the ovary depicting the cell type groups revealed by our single cell RNA sequencing atlas. Color code is the same as in (a). Scale bars are 20 μm. fc follicle cells, o oocytes. Source data are provided as a Source Data file. We next inferred the cell type identities of the twenty-seven clusters using the average expression of differentially expressed marker genes (Supplementary Data 2), and genes known to label ovarian cell types in diverse organisms (Fig. 2b). Details of the gene marker selections and their corresponding references are provided in Supplementary Data 3. We curated 9 cell type groups expressing genes corresponding to epithelial and ciliated cells (btub2/3, cfap77, dhc2, col1a, vwf), mammalian granulosa cells (tgfb2, foxl2, inhbb, cyp26, foxo, gata4), echinoderm follicle cells (vtg2, vtg5), mammalian theca cells (mmp14, angpt1, star, cyp17a1, vegfa, fgf18), muscles (mhc, mlck, tnn1), immune cells (lyz, mpeg1, bpi, cd151), neurons (syt1, smp, pplnp, oxy/vp, hdc) and germ cells/oocytes (nanos1, nanos3, boll, piwil1, msl3, sycp1, figla, ddx4, sox2, pum2). Ultrastructural data from the ovaries of another sea star species supported the presence of immune cells, including coelomocytes and pigment cells, associated with the epithelium20. The identification of two clusters expressing immune pathway genes further supports this conclusion. We found that most cell states were equally distributed amongst the three ovary samples (Fig. S3d), with the exception of clusters 5, 9, 10, 11, 12, and cluster 4, identified as follicle cells and epithelium, respectively, which predominantly originated from the previously spawned ovaries of female 1. Germ cell nest clusters 13 and 14 were also enriched in sample 2, which was used for the 3D volume analysis (Fig. 1k, l) and found to contain numerous germ cell nests. Therefore, our atlas reflects the experimentally validated reproductive states of the ovaries used. With this confidence in our dataset, we next sought to spatially map their location within intact ovaries using HCR and FISH (Figs. 2c–h and S3e–i). We first examined genes expressed in follicle cells. Using HCR, we detected the signaling molecule tgfb2 in a subset of follicle cells (Fig. 2c), similar to what has been reported in mammals24. We also detected foxl2, a known mammalian granulosa cell transcription factor, in follicle cells (Fig. 2d). Moreover, HCR for foxl2 and vtg2 (a previously validated follicle cell marker) revealed partially overlapping expression within follicle cells, as well as distinct subpopulations that were exclusively foxl2-positive or vtg2-positive (Fig. 2e). Therefore, the clusters that contain foxl2 (clusters 5–8, 23–24), tgfb2 (clusters 5, 7–8, 2–25) and vtg2 (clusters 5–8, 9–12, 23) transcripts were confirmed to be follicle cell groups corroborating the scRNA-seq predictions. We next turned to the putative germ line clusters. The transcription factor figla, which functions in mammalian oogenesis, was detected exclusively in the germ cell lineage, including nests and oocytes (Fig. 2f). Similar results were obtained for the germline factors piwil1 and nanos3 (Figs. 2g and S3e–e’). Surprisingly, we did not detect the conserved germ line RNA helicase ddx4/vasa in the germ cell nests, though it was detected in all other oocyte stages (Fig. 2h). This result suggests that ddx4/vasa transcription is activated only once the primary follicle is formed. These expression patterns suggest two cell groups: the nests (clusters 13 and 14) and early oocytes (clusters 15, 16, 17, 18, and 19). To further validate our clustering analysis, we examined syt1 expression, which was unexpectedly predicted to be present not only in the nervous system but also in germ cell nests and oocytes. HCR confirmed syt1 in the nerve net as well as in the germ cell nests and oocytes (Fig. S3f–g’), while a positive control in P. miniata larvae and post-metamorphic juveniles verified the probe’s specificity for the nervous system (Fig. S3h, i). Taken together, our spatial and transcriptomic analyses enabled us to identify and map nine cell type groups distributed in twenty-seven clusters (Fig. 2i). Shared cell type toolkits between P. miniata and other animals How similar are sea star ovarian cell types to those of distantly related animals? Our analysis identified some similarities: for instance, sea star follicle cells (follicle I and follicle III groups) express genes also found in mammalian granulosa (foxl2, inhbb, tgfb2, gata4) and theca (tgfb2, mmp14, angpt1r, star, cyp17a1, vegfa, fgf18) cells. To more broadly test this, we used SAMap25 to compare our atlas with published ovarian datasets from adult fruit fly, sea urchin, zebrafish, mouse, and fetal human (Figs. 3 and S4). As expected, sea star cluster 22 (muscle) aligned with ovarian muscle clusters in all other species. We also identified similarities between the three sea star follicle cell groups and their counterparts in fruit fly and sea urchin (Figs. 3a and S4a). Surprisingly, sea star follicle cell I clusters also aligned with follicle cells in zebrafish, and granulosa cells in both mouse and human, although in the latter with lower alignment scores (Fig. 3b, c and S4b). In addition, the sea star oocyte cluster 19 aligned with germ and oogonial cell clusters in fruit fly, sea urchin, zebrafish and human (Fig. 3a–c and S4a). Zebrafish and mouse theca cells aligned with sea star epithelial clusters, while zebrafish vasculature aligned to one sea star follicle cluster (Figs. 3b and S4b). Similarities became scarcer when we compared sea star with the adult mouse ovary, which does not contain oogonial stem cells26 (Fig. S4b). However, when we compared to the human embryonic ovary, in which early oogenesis may still be occurring, we found that one human germ cell cluster aligned with the sea star oocytes cluster 17 (Fig. 3c). Cross-species comparison of ovarian single cell atlases with SAMap. Sankey plot depicting the alignment scores between the adult sea star ovarian cell clusters with the adult sea urchin ovary92 (a), adult zebrafish ovary91 (b), and fetal human ovarian clusters89 (c). d Summary of the cell types shared between sea star and the taxa used for comparison. Alignment score is defined as the average number of mutual nearest cross-species neighbors of each cell relative to the maximum possible number of neighbors. The width of the alluvial link corresponds to the alignment score between the compared clusters. The greater the width the higher the alignment score. The alignment scores can be found in supplementary material (Supplementary Data 5–7). Source data are provided as a Source Data file. According to our cross-species comparisons, sea star follicle cell II and III clusters exhibit puzzling and heterogeneous alignments across species. Follicle cell II clusters align with follicle and oocyte populations in sea urchin and with oocyte populations in zebrafish; with granulosa populations in humans; with follicle-associated and hemocyte populations in the fruit fly; and with stromal, epithelial, and granulosa populations in mice (Figs. 3 and S4). In contrast, follicle cell III clusters align with follicle and epithelial populations, as well as an uncharacterized population (cluster 12) and oocytes in sea urchin; with immune populations in zebrafish; with granulosa and immune populations in humans; with follicle, follicle-related, and hemocyte populations in the fruit fly; and with stromal and immune populations in mice (Figs. 3 and S4). Taken together, these alignment patterns preclude a confident definition of their identities. A summary of the main findings can be found in Fig. 3d. In support of the specificity of these similarities, comparison to the sea urchin larva, which serves as a null comparison, yielded no strong hits apart from muscle (Fig. S4c). Overall, our analysis not only reveals a high degree of transcriptional similarity among ovarian cell types across evolutionarily distant echinoderms, yielding the highest number of shared ovarian clusters among all cross-species analyses, but also identifies conserved ovarian cell type programs across phylogenetically distant organisms. These findings suggest that the origins of several of these cell types, including follicle cells, may predate the divergence of chordates and Ambulacraria, which includes echinoderms and hemichordates. Alternatively, these cell state similarities could arise through convergent evolutionary processes and broader species sampling is therefore required to resolve these possibilities. Indications of a mitotic oogonial stem cell population We next asked what features of the sea star ovary, such as adult oogonial stem cells (OSCs), support its extreme reproductive capability. To identify potential OSC populations, we scored for the average expression of cell cycle-associated and meiotic genes (Figs. 4a and S5a). We found that the germ cell nest clusters 13 and 14 as well as the oocyte clusters 15–19, express genes associated with G1/S phase (including cyclinD2, cyclinE, cdt1, and pcna), whereas oocytes15–19 express genes operating during G2/mitosis (G2/M) phase (including cyclinA, cyclinB, and bub3) (Figs. 4a and S5a). Genes specific to meiosis, such as spo11, mlh1, hormad, sycp1, sycp2, and sycp3, were expressed in the germ cell nest clusters 13–14 and oocyte clusters 15–19 (Fig. 4a and S5a). HCR for one of those genes, mlh1, confirmed the predicted expression in both the germ cell nests and the oocytes (Fig. S5b). To directly test cell proliferation, we cultured ovarian lobes in vitro and performed EdU incorporation assays (Fig. S5c–e). We detected DNA replication in both somatic and germ cell populations. This included the epithelium, follicle cells, and within germ cell nests (Fig. S5c–e). To further define the germ line population, we paired EdU incorporation with HCR for piwil1 and ddx4/vasa (Fig. 4b–e). The EdU signal was exclusively detected in piwil1 positive germ cell nest nuclei, but not in the ddx4 positive growing oocytes. Altogether, these data identified replicating cells within the sea star ovary (Fig. 4f) and support our hypothesis of a self-renewing oogonial stem cell pool in adult sea star ovaries. However, future work will be required to fully elucidate the seasonality, regulation, and lineage of the germ cells in the ovary. Identification of mitotically active germ line cells. a Dotplot showing the average score of genes involved in G1/S, G2/M (mitosis) phases of the cell cycle and meiosis across the P. miniata ovarian cell type groups. b–e HCR for the germline markers piwil1 and ddx4 paired with the EdU proliferation assay. Scale bars are 20 μm. Nuclei were labelled with DAPI (cyan). Dotted lines outline the germ cell nest in (b–d). f Summary of the ovarian cell types undergoing mitosis as revealed by the EdU nuclear incorporation. o oocytes. Source data are provided as a Source Data file. We next asked how gene expression changes as oocytes develop. We used Monocle3 to identify a developmental trajectory linking the 7 germ line cell clusters13–18, from OSC to oocyte. Cluster 13 was chosen as the root because its molecular signature and cellular morphology was most similar to uncommitted germ cells, and because it did not express ddx4, as experimentally validated by HCR (Fig. 5a–e). We found a substantial number of genes that are upregulated in germ cell nests and downregulated in oocytes, and fewer that are more highly expressed in oocytes (3378 genes upregulated in nests, 849 upregulated in small oocytes; Fig. S6a and Supplementary Data 4). In support of this, ddx4 was among the 849 upregulated genes in the oocytes, while it was found significantly downregulated in the germ cell nests (Fig. S6a–b’ and Supplementary Data 4). Gene ontology (GO) enrichment analysis shows that while the top terms are similar between germ cell nests and oocytes, most of the nest genes correspond to transcription and translation factors, while many of the oocyte genes are related to splicing, protein synthesis, and transport (Figs. S6b, b’ and S7). Among the genes also upregulated in the germ cell nests are the transcriptional and epigenetic regulators atrx1 and cxxc1 (Supplementary Data 4). Both genes are important for mammalian sexual differentiation and oogenesis, respectively27,28. Plotting their average expression on the trajectory as well as HCR for both genes confirmed their enriched co-expression in the germ cell nests (Figs. 5f–k and S6c–c”’). Similarly, we performed HCR for cyclin A (cycA), whose protein product is synthesized in late meiosis I29 and, based on our analysis, is downregulated in germ cell nests and upregulated in oocytes (Supplementary Data 4). Consistent with this, our results confirmed the restriction of cyclin A to committed oocytes, with no detectable expression in nests (Fig. 5l–n). Taken together, our data highlight distinct transcriptional differences occurring during the transition from oogonia to early oocytes (Fig. 5o). Reconstruction of the early oogenesis developmental trajectory. a, b HCR for the germline markers piwil1 and ddx4 highlights the absence of ddx4 in the germ cell nests. Dotted lines outline the germ cell nest. Nuclei were labelled with DAPI (cyan). Scale bars are 20 μm. c, d Feature plot showing the expression levels of piwil1 and ddx4 along the germline clusters of the P. miniata single cell atlas. e UMAP depicting the developmental trajectory of the germline clusters reconstructed with Monocle3. The germ cell cluster 13 has been used as a root. f–i HCR for the transcriptional regulators atrx1 and cxxc1 paired with piwil1 confirms their expression in germ cell nests. Scale bars are 5 μm. j, k Feature plot showing the distribution of atrx1 and cxxc1 transcripts along the trajectory confirms their differential expression in clusters corresponding to germ cell nests. Nuclei were labelled with DAPI (cyan). l, m HCR for the germline marker piwil1 and cycA confirms the absence of cycA in the germ cell nests. Dotted lines outline the germ cell nest. Nuclei were labelled with DAPI (cyan). Scale bars are 20 μm. n Feature plot showing the expression levels of cycA along the germline clusters of the P. miniata single cell atlas. o Cartoon highlighting the difference in gene expression of the oogonia in the germ cell nests versus the oocyte stages. o oocytes. Source data are provided as a Source Data file. Signaling pathways across ovarian cell types What are the major signaling pathways that could control oogenesis? We first mined our atlas for gene groups corresponding to the Wnt, Tgfβ, Delta/Notch, Hh, and retinoic acid (RA) signaling pathways, which are important in oocyte development in other organisms30–33 and in oocyte polarity in P. miniata34 (Figs. 6a and S8a). We found that epithelium, oocytes, and immune cells most strongly express Wnt ligands, while follicle I, germ cell nest, oocytes, neurons, and muscles express the corresponding frizzled receptors, dvl3, gsk3, apc, tcf/Lef, and β-catenin (Figs. 6a and S8a). Similarly, Tgfβ ligands are expressed by the epithelium, follicle I, follicle III, and muscle clusters, while their receptors and signaling cascade factors are expressed in germ cell nest, oocytes, neurons, muscles, follicle III, and immune cells (Figs. 6a and S8a). Delta transcripts are expressed in epithelial, follicle II, follicle III, and oocytes cell groups, and the notch receptor and gamma secretase subunits are expressed by follicle I, oocytes, and neurons (Figs. 6a and S8a). Hedgehog (hh) is exclusively expressed by neurons, and the signal may be received by follicle I and germ cell nest type groups that express patched and smoothened receptors (Figs. 6a and S8a). Lastly, retinoic acid synthesis machinery, including sdr7, aldh1, aldh3, and aldh8, is expressed in the follicle II, germ cell nest, and oocyte cell groups, and the receptors rar, rxr, and ppar are expressed by the follicle I, III, germ cell nest, neural, and muscle groups (Figs. 6a and S8a). Degradation of RA is performed by cells expressing cyp26, which includes the follicle I group, while its transport is mediated by cells that express fabpl3, which includes the follicle I and follicle II groups (Figs. 6a and S8a). Our results indicate a complex signaling system within the ovary by which diverse signals are both produced and received by the follicle cells, the cells of the germ lineage, and the ovarian neurons (Fig. 6b). Putative signaling pathways involved in ovarian cell type crosstalk. a Dotplot showing the average score of gene groups corresponding to ligands and signaling cascade genes across the P. miniata ovarian cell type groups. b Schematic representation of the predicted paracrine crosstalk between follicle cells, neurons, and germ cell nests. c HCR for tgfb2 labeling follicle cells. Scale bar is 40 μm. d–f HCRs for tgfbr3, alk2, and smad5 show localization within the germ cell nests and oocytes, spatially confirming the scRNA-seq predictions. Scale bar is 20 μm. Dotted lines outline the germ cell nests. Nuclei were labelled with DAPI (cyan). fc follicle cells, o oocytes. g Box and scatter plot showing the number of TZPs per square micron of follicle imaged. n = 8 follicles per condition from one representative cultured ovary explant; p-value from a two-sided t-test = 0.0236. Center line indicates the median, box limits indicate the upper and lower quartiles, and each dot represents one follicle. In control follicles, the minimum, maximum, median, lower quartile, and upper quartile values were 0.00035, 0.0144, 0.0083, 0.0053, and 0.0091, respectively. In LDN-193189-treated follicles, the minimum, maximum, median, lower quartile, and upper quartile values were 0.0075, 0.0231, 0.0128, 0.0104, and 0.0139, respectively. h Number of follicles with organized or disorganized TZPs, scored blind from a representative culture experiment. i, j Staining for actin (phalloidin) and DNA (DRAQ5) showing TZP organization in representative follicles from control vs 4-day LDN-193189 treated ovarian explants. Scale bar is 20 μm. Arrowheads point to TZPs. Source data are provided as a Source Data file. We next focused on Tgfβ signaling, which has been implicated in primordial germ cell development and follicular growth in diverse animals30,32 (Figs. 6c–f and S8b–d). Of the genes tested, tgfb2 ligand was uniquely expressed in follicle cells enwrapping the oogonia and mature oocytes (Fig. 6c). We found that the type I Activin TGFβ receptor alk2, which binds BMP ligands and signals through the intracellular messengers smad1/5/8; the smad5 ortholog itself; and the tgfbr3 ortholog, a likely TGFβ co-receptor whose vertebrate counterpart lacks intrinsic activity; are all expressed in both germ cell nests and developing oocytes, validating the scRNA-seq predictions. (Fig. 6d–f). In addition, our atlas showed that several Bmp ligands are expressed within the germ cell lineage, suggesting that Tgfβ signaling may also be active between growing oocytes, or acting upon follicle cells, a dual role also reported in mammals30,32 (Figs. 6a and S8a). HCR for bmp1 and bmp4 confirmed these predicted expression patterns (Fig. S8b–d). To test the role for Tgfβ in folliculogenesis, we used the specific Tgfβ inhibitor LDN-193189 that selectively inhibits BMP signaling by blocking SMAD1/5/8 activation through targeting type I BMP receptors ALK1, ALK2, ALK3, and ALK635. We first confirmed that LDN-193189 resulted in a strong reduction in nuclear pSMAD1/5 and dorsal-ventral patterning defects in larvae, as predicted by prior work36, validating its use as a Tgfβ receptor inhibitor in Patiria miniata (Fig. S8e–k). Treating ovaries with this inhibitor resulted in follicles with more transzonal projections (TZPs), which had a disorganized appearance (Fig. 6g–j and S8l, m). In summary, our results support the deep evolutionary conservation and functional role of Tgfβ signaling pathway components within the ovarian follicle. Evidence for a novel ovarian neuronal cell type Vertebrate oogenesis is regulated by endocrine signaling between the hypothalamus, pituitary gland, and ovary. Therefore, we asked how the ovarian nervous system might serve as an intrinsic paracrine regulator of oogenesis. We found that most neuropeptide synthesis genes are specifically expressed by neuronal cluster 20, including tryptophan hydroxylase (tph), tyrosine hydroxylase (th), choline O-acetyltransferase (chat), histidine decarboxylase (hdc), and glutamate decarboxylase (gad), indicating serotonergic, dopaminergic, cholinergic, histaminergic, and GABAergic neurons, respectively (Fig. S9a). Moreover, we found that sixteen different neuropeptides are expressed in the same cell cluster (Fig. S9a). We also found that the relaxin-like gonad-stimulating peptide, known to trigger sea star gamete maturation37, is not present in the neuronal cluster, but is predominantly expressed in clusters corresponding to epithelium, oocytes, and muscles (Fig. S9b). Moreover, its receptor, rgppR, is expressed by distinct epithelial, neuronal, and follicle III clusters (Fig. S9b). The expression of rgpp and rgppR in epithelial and follicle cells, respectively, is in line with previous observations in a related sea star species38. Immunofluorescence for TH and serotonin (5HT) confirmed the presence of extensive dopaminergic and serotonergic neuronal subepidermal networks, respectively (Fig. S9d, e). To our surprise, we also detected serotonin within the germ cell nests (Fig. 7a, b) as well as genes encoding neuromodulators and neuropeptides, including tph, chat, prolactin-releasing peptide (prrp), luqin (lqp), and uncharacterized neuropeptide precursor (np25) (Fig. S9a). The widespread presence of neuromodulators within the germ cell nests prompted us to further assess the link between ovarian neurons and germ cell nests (Fig. 7c). Importantly, most of the ligand/receptor pairs used for this analysis have been previously identified and some even de-orphanized in sea stars and sea urchins39–44. We found that neuropeptides involved in the mammalian HPG axis, including gonadotropin-releasing hormone (gnrh), kisspeptin (kpp), oxy/vp45 and the echinoderm muscle relaxant neuropeptide L-Salmf41 were differentially expressed in the neuron cell type group and in the germ cell nests and/or oocytes (Fig. 7c). Their receptors were predominantly expressed in the germ cell nest clusters, suggesting crosstalk between neurons and germ cell nests (Fig. 7c). Similarly, the receptors for serotonin, dopamine and GABA were found expressed within the germ cell nest clusters (Fig. S9c). On the other hand, we found the ngffyamide (ngffap) neuropeptide to be expressed primarily in the germ cell nest and oocytes cell type groups and not in the neurons, while the latter contained transcripts for its receptor (Fig. 7c). Moreover, we also found neuropeptides including prrp and lqp as well as their receptors expressed by both germ cell nests and neurons (Fig. 7c). Neuropeptides and their receptors are also expressed in additional domains, including follicle cells, muscles, epithelium, and immune tissues. The insulin-like peptide 1 (ilp1) is found in follicle cells and muscles; lqn and gnrhR in muscles; prpp1 in the epithelium, follicle cells II and muscles; F-salmf and L-salmf in the epithelium; prrpR and lqR2 in immune cells; kissR1 and kissR8 in the epithelium; kissR3 in the epithelium, muscles and in follicle III; oxy/vpR in the epithelium and follicle II; salmfR1 and salmfR2 broadly in the epithelium with salmfR1 also expressed in follicle I, muscles and immune cells while samlfR2 in follicle II; and insR in the epithelium, follicle I, and immune cells. These data suggest a pleiotropic role of neuropeptidergic signalling in ovarian physiology (Fig. 7c). Neuropeptidergic profile of the P. miniata intrinsic ovarian nervous system. a, b Immunofluorescence of 5HT and Myosin II. Dotted lines outline the germ cell nests. Scale bars 20 μm. c Dotplot showing the average expression of the reconstructed and/or deorphanized pairs of neuropeptides/receptors. d, d’ HCR for oxy/vp and nanos1 shows the relatively short distance between the oxy/vp positive neurons and the germ cell nests. Scale bars in panels are 20 μm. Dotted lines outline the germ cell nests. e Orthogonal view of the oxy/vp and nanos1 HCR shows the close proximity of the oxy/vp neuron in respect to the nanos1 positive germ cell nests. Dotted lines outline the germ cell nests. Scale bar is 10 μm. f, f’ HCR for gnrh in subepithelial ovarian neurons. Scale bars in panels are 20 μm. Dotted lines outline the germ cell nests. g Orthogonal view of a gnrh HCR, highlights the spatial organization of the gnrh positive neurons in close proximity to the germ cell nests. Scale bar is 10 μm. Dotted lines outline the germ cell nests. h, h’ HCR for gnrhr confirms the single cell predictions of expression in the germ lineage. Scale bars are 20 μm. Dotted lines outline the germ cell nests. i Sankey plot depicting the alignment scores between adult sea star ovarian cell atlas with the adult mouse pituitary gland one. The alignment scores can be found in supplementary material (Supplementary Data 11). j Summary of the components participating in the predicted crosstalk between the intrinsic ovarian neurons and the germ cell nests. Nuclei are labelled with DAPI (cyan). fc follicle cells, o oocytes. Asterisks are used to indicate genes whose predicted expression pattern has been confirmed with HCR. The expression patterns can be found within the same figure and Fig. S9. Source data are provided as a Source Data file. To test these predictions, we examined the expression of oxy/vp as well as the lowly expressed gnrh and gnrhr. Using HCR, we found that oxy/vp are expressed exclusively in rare subepidermal neurons (1–2 per ovarian lobe, Figs. 7d, d’ and S9f, f’). These oxy/vp positive neurons were in close proximity to the germ cell nests, suggesting potential crosstalk (Fig. 7e). Similarly, we detected gnrh+ cells in proximity to germ cell nests, while gnrhr was expressed both in the nests and oocytes (Fig. 7f–h’). Previous studies in the adult sea star A. rubens reported that gnrh is expressed predominantly within the sea star’s nervous system (oral nerve ring and radial nerve cords). Consistent with this prior work, and supporting the specificity of our HCR probes, gnrh and gnrhr were in complementary regions of the oral nerve ring and the radial nerve cord (Fig. S9g, g’). Taken together, our analysis highlights that many known hypothalamic neuromodulators involved in the vertebrate HPG axis, such as 5HT, dopamine, GABA, GNRH, KPP, and OXY/VP45, as well as the pituitary gland secreted PrRP46, are all present in the sea star ovarian neurons. We further tested the similarity between sea star ovarian cell types with the mammalian pituitary gland using SAMap (Fig. 7i). We found that sea star ovarian neurons exclusively align with pituitary gland gonadotropes, which upon stimulation by hypothalamic gnrh produce the gonadotropins that control sex hormone production, oogenesis, and ovulation in mammals47. Our comparison showed alignments between pituitary gland thyrotropes and the sea star follicle cells (follicle I) while the germ cell nests aligned with the pituitary gland lactotropes (Fig. 7i). Altogether, our results reveal a diverse and putatively multifunctional intrinsic ovarian nervous system that may communicate with the germ cell nests (Fig. 7j). Discussion Sexual reproduction depends on the interactions between multiple cell types within the ovary. Ovarian structure varies across animals, reflecting differences in reproductive strategies and adaptations to different environments. For instance, many marine invertebrates follow seasonal patterns and release large quantities of gametes into the water column, whereas in mammals, seasonal mating patterns result in the fertilization of only a limited number of oocytes within the oviduct. Despite these differences, in most animals studied, the ovary serves two critical roles as both an endocrine organ that produces steroid hormones and as the niche for oocyte development. Here, we took advantage of the easily accessible, optically transparent, and highly fecund sea star ovary, representing an extreme case of reproductive output, to explore the identity and the evolutionary origins of the ovarian cell types in a non-chordate deuterostome. Sea star follicle cells are a heterotypic group Follicle cells that enwrap oocytes serve important roles in oogenesis across animals48. In insects, different types of follicle cells produce yolk components or contribute to the formation of egg protective layers49. In mammals and fish, granulosa cells are involved in the production of estrogen, nutrient transfer, and regulation of oocyte maturation30,50. Theca cells populate the outer layers of the follicle and produce steroid hormones that are converted to estrogen by granulosa cells. Similarly, in sea stars, follicle cells produce maturation-inducing molecules, such as 1-methyladenine (1-MeAde), upon stimulation by the relaxin-like gonad-stimulating peptide (rgpp) from the nervous system37. Subsequently, 1-MeAde triggers the resumption of meiosis in arrested oocytes51. However, the molecular signature of follicle cells has not been well defined. Here, we provide evidence for three distinct sea star follicle cell groups. One group (follicle I) is transcriptionally similar to follicle cells in zebrafish and to granulosa cells in mice and humans, while expressing genes characteristic of granulosa cells. The second group (follicle II) has a poorly defined identity, computationally aligning with a range of follicle cell types as well as non-follicle types across species, with no clear equivalent outside sea stars. Based on the high expression of vtg genes, we hypothesize that it reflects cell populations dedicated to vitellogenesis. Lastly, the third group (follicle III) also shows ambiguous cross-species similarity, yet expresses genes typically associated with mammalian theca cells; therefore, we speculate that it reflects a possibly diverged sea star theca-like cell type. An open question from our work is whether a pool of pre-follicle cells exists that corresponds to the pregranulosa in the mammalian ovary. Future investigation with potential marker genes and lineage analysis will help resolve this question. Surprisingly, our results also suggest possible evolutionary origins for granulosa and possibly theca cells that predate the divergence of Ambulacraria and chordates. However, future ultrastructural and molecular data from hemichordates will be needed to more fully test this possibility. In support of this hypothesis, our analysis found that several genes associated with mammalian granulosa cells are also expressed by a subset of sea star follicle cells. This includes the transcription factors foxl2 and gata4, and the protein hormone inhibin beta (inhbb)30,52. The forkhead transcription factor foxl2 is a crucial member of the gene regulatory network controlling granulosa cell proliferation and differentiation in mammals30. A similar expression pattern of foxl2 was also reported in planarian ovaries, where it was found to promote oocyte differentiation, suggesting a conserved role of foxl2 in both vertebrate and invertebrate follicular development53. Our cross-species analysis revealed that cluster 6 (follicle I) consistently aligned with granulosa/follicle cell clusters in all of the species tested. Further, mammalian granulosa cells secrete several Tgfβ family ligands that influence oocyte development, and we find that such ligands are expressed by sea star follicle cells54. Lastly, our scRNA-seq data suggest that sea star granulosa-like cells produce and respond to RA signaling, which prior work has found to regulate granulosa cell proliferation and steroidogenesis in mammals55. A previously known function of follicle cells is to receive the rgpp signal from the radial nerve, which activates 1-MeAde production37. Notably, only cluster 25, belonging to the follicle III cell type group (theca-like cells), expressed the Relaxin-like Gonad-Stimulating Peptide receptor (rgppR) (Fig. S9b). This strongly suggests that this population of theca-like cells is a likely candidate for the follicle cell type that produces 1-methyladenine and highlights the subfunctionalization of diverse follicle cell types. We also find that the follicle III cell group expresses orthologs of key mammalian steroidogenesis genes, including star and cyp17a1. However, cyp11-like56, another key steroidogenesis enzyme57, is instead differentially expressed in the epithelium (Supplementary Data 2). Previous studies in sea stars, sea urchins, and sea cucumbers have found strong evidence for the conversion of cholesterol into sex steroids in these animals58–60. Moreover, estradiol has been detected by mass spectrometry in both sea urchins and sea cucumbers60,61. In the sea cucumber H. scabra, progesterone and testosterone, along with key steroidogenesis genes including star and cyp17, have been identified60. Estradiol and progesterone influence vtg levels in sea cucumber ovary cultures, supporting a role in reproduction60. Similarly, in the sea star Sclerasterias mollis, repeated injections of estradiol-17β or estrone during early gametogenesis increased ovarian estrone and progesterone levels, enlarged oocytes, and elevated ovarian protein content62. Comparable results were also observed in the sea star Patiria pectinifera63. Future work will be required to further understand what steroids are synthesized in the ovary, and which cell types contribute. Molecular clues of putative oogonial stem cells A conserved feature of oogenesis is the formation of germ cell nests. Such clusters of oogonia, which in some instances are cytoplasmically connected, have been found in the ovaries of insects, annelids, flatworms, amphibians, reptiles, birds, and mammals19,64–66. Here, we identify germ cell nests in the sea star and define their subcellular morphology using expansion microscopy. However, fully resolving whether they are connected by cytoplasmic bridges, and can thus be called cysts (as in mammals), will require further study. Using 3D segmentation, we also determined the spatial organization of oogenesis, in which germ cell nests are localized to the periphery of the ovarian lobe along with young oocytes organized in primary follicles. Nest-expressed genes include the conserved germline markers piwil1, nanos1, nanos3, boll, msl3, meiotic recombination components sycp1 and mlh1, and transcriptional regulators including figla, atrx1, and cxxc127,28,67–69. We also detect cell proliferation within the nests, supporting them as the likely source of new oocytes70. The presence of mitotically active germ line cells in an adult female is indicative of persistence or reactivation of pre-meiotic germ cells and thus may imply activity of oogonial stem cells or a proliferative pool of committed oocyte progenitors. The conserved germline RNA helicase ddx4/vasa is inactive in the nests, but is expressed in early follicles, similar to what was shown in marsupial fetal germ cells71. This repression may contribute to maintaining an undifferentiated state, as seen in fruit flies, where ddx4 knockout led to arrested oocyte development72. Additionally, we speculate that a potential self-renewal-meiotic entry decision could be regulated by RA, Wnt, Tgfβ, and Hh pathways, whose respective receptors and intracellular components were found expressed in the germ cell nests, but testing this possibility will require future work. These results support an ancient role for these pathways in oogenesis, as demonstrated in other species, including mice and amphibians31,33,73,74. A possible intrinsic ovarian neuroendocrine system How might oogenesis and ovulation be regulated in animals that lack organs such as the brain and the pituitary gland? Previous studies have shown that reproductive neuroendocrine control in invertebrates is strikingly diverse. In ascidians, Cionin (CCK/gastrin homolog) and vasopressin (CiVP) were found to induce ovulation75,76. In mollusks, contraction of the oviduct is regulated by GnRH-like peptides77. In scallops, a GnRH-like peptide stimulated spermatogonial cell division in cultured scallop testis78. In annelids, Corazonin/GnRH-like peptides were found to regulate sexual maturation and growth79, while in crustaceans, a neurohormone in the hyperglycemic hormone-type II (CHH-type II) family was found to have an inhibitory role in vitellogenesis80. Foundational work in sea stars found that signals from radial nerves induce spawning, indicating that oogenesis is under neural control even in invertebrates81. Our findings suggest additional regulatory layers in sea stars. We found that the epithelium, germ cell nests, and oocytes produce neuropeptides such as trh, which is produced by the radial nerve cords and regulates oogenesis in echinoderms82. We also found numerous HPG-associated signaling factors, including 5HT, dopamine, GABA, KPP, GNRH, PrRP, and OXY/VP, to be expressed in the intrinsic ovarian nerve plexus and their respective receptors in the germ cell nests. This molecular signature suggests a dual identity, similar to both chordate hypothalamic neurons and pituitary gonadotrope endocrine cells. We therefore suggest the presence of a hypothalamic-like and gonadotropic-like intrinsic ovarian neuroendocrine signaling system in this non-chordate deuterostome. Signals from this system may be received by germline cells, representing a mechanistic novelty in echinoderms that is distinct from the vertebrate reproductive axis. However, core components of the vertebrate HPG axis, other than FSH and LH, which are absent in invertebrates83, may have been retained in echinoderms, but operate locally within the gonad. In such a scenario, ovarian neurons could be functionally analogous to distinct elements of the mammalian HPG axis. While in mammals the HPG axis flows from the central nervous system, we propose that in sea stars, in addition to regulation by the radial nerves, local neuropeptidergic signaling from ovarian neurons may act directly on germ cells and developing oocytes. In this case, the absence of hypothalamus and pituitary gland could be functionally compensated, at least in part, by ovarian neurons. However, further studies are necessary to explore the function of the sea star ovarian neurons and the directionality of the signaling across cell types. In summary, we speculate that during animal evolution, the regulation of oogenesis may have initially relied on intrinsic neuroendocrine systems within the gonad itself. As organisms became more complex, this regulatory role may have been gradually co-opted by specialized structures, including centralized nervous systems and endocrine glands. Methods Animal husbandry, ovary isolation, in vitro culture Adult Patiria miniata were obtained from South Coast Bio (San Diego, CA, USA) and Monterey Abalone Company (Monterey, CA, USA). No ethical approval was required for this study because sea stars, as marine invertebrates, are not subject to IACUC regulation. The animals were maintained in temperature controlled (15 °C) circulating seawater aquaria at the Marine Resources Center (MBL, MA, USA). Ovarian fragments were surgically isolated through the oral side of the sea star’s arm using a Stoelting™ Sterile Disposable Scalpel, No.11 and ovaries were collected using forceps in Eppendorf tubes containing filtered sea water (FSW). For the single cell RNA sequencing experiments, the ovaries were used immediately. For the rest of the experiments of the manuscript, ovaries were kept in FSW plus 10 μg/ml trimethoprim (Sigma # S7507) and 50 μg/ml sulfamethoxazole (Sigma # T7883-5G) at 15 °C as previously described84. Ovary dissociation Dissociation of isolated sea star intact ovarian lobes into single cells was performed as previously described with minor modifications85. Briefly, ovarian fragments were transferred with forceps to a petri dish containing Ca2+ Mg2+-free artificial sea water (CaMg-free ASW) and allowed to settle for 3 min (min). Then the ovary was transferred to an Eppendorf tube containing the dissociation buffer (1 M glycine, 0.02 M EDTA, in CaMg-free ASW) and incubated for 10 min. Dissociation was promoted through gentle pipette aspiration every 2 min. Once dissociation was complete, cells were spun down with a swing bucket centrifuge at 500 × g for 10 min and washed twice with CaMg-free ASW. Propidium Iodide (Sigma #25535-16-4) and Fluorescein diacetate (Sigma # 0596-09-08) were used to assess cell viability and only cell suspensions with cell viability ≥ 90% were further processed. Single cells were passed through a 40 μm cell strainer (Pluriselect # NC1789427) to remove aggregates. The number of cells was estimated using a hemocytometer and diluted according to the manufacturer’s protocol (10× Genomics). During the entire process, specimens were kept on ice, except for the centrifugations in which they were kept at 4 °C. The isolated cells were loaded on the 10× Genomics Chromium Controller according to the manufacturer’s instructions and cDNA libraries were generated using the Chromium Single Cell 3′ Reagent Kit (v3.1 Chemistry Dual Index). Specimens were sequenced by the Azenta Life Sciences (Burlington, MA) (https://www.genewiz.com/) sequencing service using the Illumina NextSeq 550 with a maximum read length 2 × 150 bp at a resolution of 350 M PE reads. Single cell RNA sequencing mapping of reads and data analysis Single cells from ovaries originating from three different animals were processed using the 10× Genomics Chromium scRNA-seq capturing system. The libraries generated were biased towards early oogenesis since the 10× Genomics Chromium capturing system allows for the capture of cells up to 40 μm, which means that in our single cell atlas, the maximum oogonium size represented is 40 μm. Fully grown oocytes in Patiria miniata are ~170 μm. Patiria miniata genome version 3.086 was used for single cell RNA sequencing read mapping. The corresponding annotation was modified to contain only protein coding entries using AGAT v0.7.087 and Linux command line tools GNU coreutils v8.26. Cell Ranger Software Suite v7.1.0 was used to create the index from the genome FASTA file and the annotation GTF file as well as for mapping with –no-bam and –nosecondary flags, with all other settings left as default, so the number of cells was estimated automatically by Cell Ranger. Cell Ranger output files were loaded in R v4.4.1 and processed with the Seurat v5.3.0 pipeline88. Seurat objects were created by excluding cells containing genes that are transcribed in less than three cells and cells that have less than a minimum of 400 and a maximum of 5000 transcribed genes. Next, the Seurat objects were normalized, and variable genes were found through the variance stabilizing transfer (VST) method with a maximum of 2000 variable features. The three different objects were integrated through identification of gene anchors (FindIntegrationAnchors). Following integration, scaling, and principal component analysis (PCA) were performed. Jackstraw was used to evaluate PCA significance. Uniform Manifold Approximation and Projection (UMAP) was used to perform clustering dimensionality reduction. The final object consisted of 35,877 cells. The FindAllMarkers command was used to identify differentially expressed marker genes. The average expression of genes of interest were visualized using the DotPlot and DoHeatmap commands included in the Seurat R package. To compare the Patiria miniata ovary integrated single cell atlas with the publicly available human89, mouse90, zebrafish91, sea urchin92, and fruit fly93 ovarian atlases as well as with the mouse pituitary gland atlas94 and the sea urchin larva one85 SAMap v1.02 was used as previously described in ref. 25. For the mouse, zebrafish, fruit fly, and sea urchin (ovary and larva) datasets, the already available rds files were used, while for the human fetal ovary and the mouse pituitary gland datasets, clustering analysis was performed similar to what was previously described89,94. SAMap output mapping tables can be found in supplementary material (Supplementary Data 5–11). The average score of genes of interest was determined using the AddModuleScore function incorporated in the Seurat R package. The average score for each cell was defined as the average expression of the gene set of interest adjusted by subtracting the average expression of randomly selected, matched control gene sets to correct for expression biases. These values were binned based on their averaged expression and stored as a new metadata column in the Seurat object. The oogenesis trajectory was reconstructed using the R package Monocle3 v1.3.795. The package EnhancedVolcano v1.22.0 was used to visualize the differentially expressed genes between the germ cell nests and oocytes along the trajectory, while the R packages topGO v.2.56.0 and clusterProfiler v4.12.6 were used for the GO enrichment analysis. Light sheet microscopy/segmentation 3D analysis Ovarian samples were mounted individually in Fluidflon FEP tubes (Proliquid 1.6 × 2.4 mm) using 1% low gelling agarose (Sigma-Aldrich A0701, type VII-A) with their elongated morphology parallel to the tube. A Zeiss Lightsheet 7 was used to acquire multiple-view data. It was equipped with a 10×/0.5 W Plan-Apochromat detection objective and two 10×/0.2 air illumination objectives producing two light sheets 5.8 µm thick at the waist and 12 µm thick at the edges of a 1100 × 1100 µm field of view. Illumination mode was dual (mean fused) with active pivot scan. Samples were imaged from five views (72 degrees apart) and Zoom factor 0.8. Each z-stack has a pixel frame of 1920 × 1920 and a voxel size of 0.57 × 0.57 × 1.8 µm. Each optical slice was acquired with a combination of 405, 488, and 561 nm laser with following filter settings: 405 nm: EF1—BP 420–470, CAM BS—SBS LP 490; 488 nm: EF1—BP 505–545, CAM BS—SBS LP 560, 561 nm: EF1—BP 575–615, CAM BS—SBS LP 640. For registration of multi-view acquisitions and fusion plus deconvolution of the registered data, the Fiji plugin BigStitcher v2.3.496 was used with standard settings. Image segmentation, 3D-reconstructions and data analysis were performed using the Amira software package (FEI Visualization Sciences Group Thermo Fisher Scientific), version 2024.1. Morphological features such as muscles and nuclei were segmented semi-automatically using various tools in Amira, such as “Thresholding” and “Magic Wand”. Oocytes and germ cell nests were primarily segmented manually. Volumes of oocytes and germ cell nests were measured using the “Material Statistic” tool. Visualization of all segmented reconstructions was achieved using tools like “Volren”, “Surface Rendering”, and “Surface View” within the software package. Ovarian immunohistochemistry (IHC) and fluorescent staining Ovarian fragments were fixed with 4% paraformaldehyde (PFA) in FSW for 20 min at room temperature (RT). Specimens were washed twice with FSW and five times with PBS supplemented with 0.1% Tween (1× PBST). Next, samples were incubated in blocking solution [1 mg/ml Bovine Serum Albumin (BSA) and 4% sheep serum (SS)] in PBST for one hour at RT. Blocking solution was removed and primary antibodies were added in the appropriate dilution in blocking solution and incubated overnight at 4 °C. 1E11 (gift from Dr. Ina Arnone) was used in a 1:20 dilution, DM1α (Santa Cruz Biosciences #sc-32293) 1:100, anti-acetylated tubulin (Sigma # T67930) 1:100, anti-alpha tubulin (Sigma # T5168) 1:100, anti-MHC (custom-made anti-SpMHC, PRIMM, Italy, gift from Dr. Ina Arnone) 1:50, anti-serotonin (Sigma # S5545) 1:500, anti-TH (Sigma #AB152) 1:100, anti-RNA polymerase II (Abcam #ab5408) 1:250, and anti-Myosin II (Developmental Studies Hybridoma bank # CMII 25) (1:50). Samples were washed five times with 1× PBST (15 min each) and incubated for one hour at RT with Alexa Fluor™ secondary antibodies (Goat anti-Mouse Secondary antibody Alexa fluor 488 and Goat anti-rabbit Secondary antibody Alexa Fluor 546; Invitrogen # A-11001 and # A11035 respectively) diluted 1:1000 in PBST. Samples were washed five times with 1× PBST. In cases where phalloidin (Invitrogen # A22283) was used, it was diluted 1:20 in 1× PBST and allowed to incubate overnight at 4 °C. For WGA staining (Invitrogen # W11261), a working solution of 5 μg/mL was prepared, and the specimens were incubated overnight at 4 °C. Ovaries were washed five times with 1× PBST, nuclei were stained with DAPI (1 μg/mL), and specimens were mounted for imaging. During the entire procedure, the interval of each wash was 15 min. Specimens were imaged on a Zeiss Cell Observer Spinning Disk as well as on Zeiss LSM 700 and 780 confocal microscopes. Images were processed in Fiji/ImageJ v2.16.097. Whole mount hybridization chain reaction (HCR) HCR was performed according to previous protocols98 with minor modifications. Probes were designed either by Molecular Instruments or HCR 3.0 Probe Maker v2.099. Further details regarding the probes can be found in Supplementary Data 12. Ovarian fragments were fixed in 4% PFA in Fixative Buffer (0.1 M MOPS, 0.5 M NaCl, 2 mM EGTA, 1 mM MgCl2, and 1× PBS) for 1 h at room temperature (RT). Specimens were washed three times with Fixative Buffer (without PFA) at RT. Samples were washed twice with 100% methanol, transferred into fresh 100% methanol, and placed at −20 °C for long-term storage. Specimens were rehydrated in 75, 50, 25% methanol in RNAase free water on ice (each for 10 min). Next, ovaries were washed four times for 5 min each in RNase free 1× PBST. Ovaries were treated with permeabilization buffer (1.0% SDS, 0.5% Tween, 50 mM Tris-HCl (pH 7.5), 1.0 mM EDTA (pH 8.0), 150 mM NaCl) and incubated for 1 h at RT. Next, specimens were washed four times for 5 min each in 1× PBST. Pre-heated (37 °C) Probe Hybridization Buffer (PHB) was added to the samples and incubated for 3 h at 37 °C. PHB was exchanged with PHB containing in total 0.04 μM of each probe and specimens were left to incubate for one day at 37 °C. Following hybridization, samples were washed 4 times with pre-heated Probe Wash Buffer (PWB) for 5 min each, followed by two washes with PWB for 30 min each at 37 °C. Samples were then washed two times with 5× SSCT (5× Saline Sodium Citrate and 0.1% Tween) and two times with 1× SSCT (1× Saline Sodium Citrate and 0.1% Tween) each for 5 min. Next, specimens were incubated with Amplification Buffer for 30 min at RT, hairpins were added to the specimens according to manufacturer’s guidelines, and incubated overnight at RT in the dark. Lastly, samples were washed twice with 1× SSCT for 5 min, two more times with 1× SSCT for 30 min, and four times with 1× PBST. Nuclei were stained with DAPI (1 μg/mL), and specimens were mounted for imaging. Specimens were imaged on a Zeiss Cell Observer Spinning Disk and Zeiss LSM 780 confocal microscope. Images were processed in Fiji/ImageJ v2.16.097. Fluorescent in situ hybridization (FISH) FISH and antisense RNA probe design was performed as previously described100,101. In brief, ovaries were fixed in 4% PFA in MOPS Buffer for 1 h at RT. Samples were washed five times with MOPS buffer, gradually dehydrated (20%, 50% ethanol), and stored in 70% ethanol and kept at −20 °C for long term storage. Specimens were gradually rehydrated (50%, 20% ethanol) and washed five times with MOPS Buffer. At this step ovaries were dissected to smaller pieces using micro scissors. Specimens were then pre-hybridized for 3 h at 65 °C, followed by a 48 h period of probe hybridization. Next, specimens were washed five times with MOPS Buffer and incubated in blocking solution (Akoya Biosciences) for 30 min at RT. Next, specimens were transferred to a solution containing the anti-Digoxigenin (Sigma # 11207733910) antibody diluted 1:1000 in blocking solution. The antibody solution was removed; samples were washed with MOPS Buffer five times and the signal was developed using fluorophore-conjugated tyramide technology (Tocris # 64581). DNA was stained through IHC for anti-RNA polymerase II. Immunohistochemistry was performed after the completion of the FISH procedure as described in ref. 101. Specimens were imaged on a Zeiss 710 confocal microscope. Images were processed in Fiji/ImageJ v2.16.097. EdU labeling Click-iT EdU Cell Proliferation Kit for Imaging Alexa Fluor 488 (Thermo Fisher Scientific # C10337) was used and was combined with either HCR or IHC as previously described in refs. 85,102. Isolated intact ovarian fragments were treated with EdU at a final concentration of 20 μM in FSW with 10 μg/ml trimethoprim and 50 μg/ml sulfamethoxazole and allowed to incubate for 2 days at 15 °C. Depending on whether the EdU cell proliferation was combined with HCR or IHC specimens were fixed accordingly as described in their respective sections above. To develop the EdU signal, the Click-iT reaction mix was prepared according to the manufacturer’s guidelines. In cases where the EdU assay was combined with HCR, after the washes with PWB, specimens were washed twice with 3% BSA in PBST for 15 min at RT. Next, the Click-iT reaction mix was added and incubated for 30 min in the dark at RT. After this step, the HCR procedure was carried out as described above. For combination of the EdU Cell Proliferation assay with immunostaining, the ovaries were fixed in 4% PFA in FSW for 15 min at RT and washed five times with PBST. PBST was removed and the reaction mix was added to the samples for 30 min (RT). Specimens were washed five times with PBST, mounted for imaging. Specimens were imaged on a Zeiss Cell Observer Spinning Disk and Zeiss LSM 780 confocal microscope. Images were processed in Fiji/ImageJ v2.16.097. Pan-expansion microscopy We followed a modified version of the protein-retention expansion microscopy protocol used in ref. 103. Ovary fragments were fixed in 4% PFA in FSW for 1 h at RT, then washed 3 times for 20 min each in 1× PBST (0.1% Triton-X-100). They were then washed 3 times for 5 min each in acryloyl-X (AcX) solution (0.1 mg/mL AcX in PBS, Invitrogen # A20770) and incubated in AcX solution overnight at RT in a microcentrifuge tube with rocking. The next day, fragments were washed 2 times for 15 min each with PBS. A gelation chamber was prepared by supergluing an 18 mm coverslip to a glass side, and supergluing two glass slides on top of each other on either side of the coverslip, resulting in a chamber 2 glass slides minus one coverslip thickness thick. The gelation solution consisted of 8.625% sodium acrylate (Sigma # 408220), 20% acrylamide (Sigma # A4058), 0.075% N,N’-methylenebisacrylamide (Sigma # M1533), 2 M NaCl, 1 × PBS, 0.2% TEMED, and 0.2% ammonium persulfate. The ovary fragment was placed on the coverslip in the gelation chamber, and 200 μL of the gelation solution was added and mixed with a pipette tip. A 22 mm square coverslip was placed on top of the chamber as a lid. The gelation chamber was placed in a humidified box at 37 °C for 1 h. The gel with the ovary fragment was then removed from the chamber, excess gel was cut from the sides, and the gel containing the fragment was transferred to a glass bottle containing 5 mL alkaline denaturation buffer (100 mM Tris base, 5% w/v Triton-X-100, 1% SDS in ddH2O). The gels were incubated in this bottle for 15 min at RT, and then the bottle was autoclaved at 121 °C for 1 h. The gel was then washed 5 times for 10 min each in PBS and left at 4 °C overnight. The next day, NHS ester dye ATTO594 (Sigma # 08741) was diluted to a final concentration of 20 μg/mL in 100 mM sodium bicarbonate, and the gel was incubated in this solution for 1.5 h in the dark at RT on a rocker. Gels were then washed 3 times for 20 min each with PBST. Next, gels were stained with SYTOX green (Thermo Fisher Scientific # S7020) by diluting SYTOX green 1:3000 in HBSS buffer without calcium or magnesium and incubating the gels in this solution for 30 min at RT in the dark on a rocker. Gels were then washed 3 times for 20 min each with PBST. Gels were placed in a 50 mL conical tube wrapped in aluminum foil and filled with ddH2O. Water was exchanged for 2 × 30 min washes, and then exchanged again and gels were left overnight at RT for the final expansion. Gels were imaged on a Leica Stellaris confocal microscope. Expansion factor was calculated by comparing the size of expanded follicle cell nuclei to the size of non-expanded follicle cell nuclei, and here was calculated to be 3.4. Scale bar on images is corrected for the expansion factor. HAK-actin expansion microscopy Ovary fragments were fixed in 4% PFA in FSW for 1 h at RT, then washed 3 times for 20 min each in 1× PBST (0.1% Triton). Following the protocol in ref. 21, we incubated the fragments in PBST with 200 nM of HAK-actin probe (gift from Luc Reymond and Spirochrome) for 1 h, followed by a 15 min post-fix incubation in 4% paraformaldehyde/0.0125% glutaraldehyde in PBS, and a quick PBST wash. We then followed the U-ExM protocol from ref. 104. Ovary fragments were incubated in 0.7% formaldehyde/1% acrylamide overnight with rocking. The gelation chamber was made as described above. The gelation solution (23% sodium acrylate (Sigma # 408220), 10% acrylamide (Sigma # A4058), 0.1% N,N’-methylenebisacrylamide (Sigma # M1533), 0.5% TEMED, 0.5% APS in PBS) was made, and 200 μL were added to an ovary fragment on the gelation chamber with a coverslip lid placed on top as described above. The gelation chamber was placed in a humidified box at 37 °C for 1 h. The gel with the ovary fragment was then removed from the chamber, excess gel was cut from the sides, and the gel containing the fragment was transferred to a glass bottle containing 5 mL U-ExM alkaline denaturation buffer (50 mM Tris base, 200 mM SDS, 200 mM NaCl in ddH2O). The gel was incubated at room temp for 15 min, changed to fresh alkaline denaturation buffer, and the bottle was partially submerged to above the liquid line in a boiling water bath for 90 min. The gel was then expanded in a 50 mL conical tube with ddH2O at room temp for 2 × 30 min, followed by 2 h, changing the water between each incubation. The gel was then washed 2 × 15 min with PBS. Rabbit monoclonal anti-HA antibody (Cell Signaling Technology # C29F4) was added at a 1:300 dilution in 2% BSA in PBS and gels were incubated at 37 °C overnight with shaking. The next day, the gel was washed 3 × 10 min in PBST (with 0.1% Tween) at RT with shaking. Secondary goat anti-rabbit Alexa Fluor 546 antibody (Invitrogen # A11035) was added at a 1:300 dilution in 2% BSA in PBST and the gel was incubated at 37 °C overnight in the dark with shaking. The next day, the gel was washed 3 × 10 min with PBST, followed by a 30 min incubation at RT in SYTOX Green diluted 1:1500 in HBSS buffer without calcium or magnesium. The gel was then washed 3 × 20 min in PBST and expanded in ddH2O for 2 × 30 min followed by overnight incubation at RT for the final expansion. The next day, gels were imaged on a Leica Stellaris confocal microscope. Expansion factor was calculated as discussed above. LDN-193189 treatment and validation Ovary explants were cultured for four days in either FSW + sulfamethoxazole/trimethoprim (S/T, as described above) or FSW + S/T + 1 μM LDN-193189 (Selleckchem # S2618), being transferred to fresh FSW + S/T and LDN-193189 each day. For counting transzonal projections (TZPs) per follicle, follicles were isolated from ovaries and cultured for four hours in the dark at 15 °C in either FSW + S/T or FSW + S/T + LDN-193189 with SPY555-FastAct (Cytoskeleton # CY-SC205) at a 1:1000 dilution. SPY555-FastAct labels follicle cell actin but cannot penetrate the oocyte. Follicles were then imaged on a Zeiss Cell Observer Spinning Disk confocal microscope. TZPs were counted manually from the z-slice containing the most TZPs for each follicle using Fiji/ImageJ v2.16.097. The area of the follicle in this z-slice was measured, and the number of TZPs was divided by this area to normalize. For counting follicles with organized vs disorganized TZPs, ovary explants were fixed and stained as described above for DM1alpha, phalloidin, and DRAQ5 (at 1:1000) in place of DAPI. Explants were imaged on a Zeiss Cell Observer Spinning Disk. Images were blinded and each follicle with visible TZPs was manually scored by eye for organized vs disorganized TZPs. For validation experiments, embryos were cultured in either FSW or FSW + 1 μM LDN-193189 from 24 h post-fertilization (post-hatching, at the very start of gastrulation) until 3 days post-fertilization. For brightfield images, larvae were immobilized using 2× seawater and imaged on a Zeiss Axiovert. This was repeated 3 times, and in each experiment larvae were scored for whether they showed the radialized phenotype. For pSMAD1/5 staining, larvae were fixed with 4% PFA in FSW for 20 min at RT, washed 3 times for 5 min each with FSW, then fixed with ice cold methanol for 10 min. They were then washed 3 times for 5 min each with PBST (0.1% Tween) and then permeabilized for 30 min in PBST (0.1% Triton). Larvae were blocked in SuperBlock (Thermo Fisher # 37580) for 30 min at RT and then incubated overnight at 4 °C with anti-pSMAD1/5 (41D10) (Cell Signaling Technology #9516T) diluted 1:200 in SuperBlock. Larvae were then washed 3 times for 10 min each with PBST (0.1% Triton), incubated for 2 h at RT in the dark with goat anti-rabbit Alexa Fluor 488 secondary antibody (Invitrogen # A-11034) diluted 1:200 and DRAQ5 diluted 1:1000 in PBST (0.1% Triton), washed 3 times for 10 min each with PBST, and imaged using a Zeiss 780 point scanning confocal microscope. To quantify nuclear pSMAD1/5 signal, z-stacks were taken through the dorsal side of the larva and max-projected in Fiji. Nuclei were segmented by taking the DRAQ5 channel, subtracting the background using a rolling-ball radius of 50, applying a Gaussian blur of radius 1, and then using Fiji’s Otsu thresholding. This mask was used to measure total nuclear fluorescence in both the DRAQ5 channel and the pSMAD1/5 channel after max-projecting and subtracting the background using a rolling ball radius of 50 for each channel. Total nuclear pSMAD1/5 signal was then divided by total nuclear DRAQ5 signal for each larva to obtain the normalized nuclear signal value. For phalloidin staining, larvae were fixed for 20 min in 4% PFA in FSW, then washed 3 times for 10 min each with PBST (0.1% Triton), blocked for 30 min in blocking solution [1 mg/ml Bovine Serum Albumin (BSA) and 4% sheep serum (SS)] in PBST, and then incubated overnight at 4 °C with DM1α as described above for ovaries. Larvae were then washed 3 times for 10 min each with PBST, incubated for 1 h with secondary antibody, phalloidin (1:500), and DRAQ5 (1:1000) diluted in PBST, washed 3 times for 10 min each with PBST, and imaged using the Zeiss LSM 780 confocal microscope. Statistics and reproducibility No statistical method was used to predetermine sample size. Sample sizes were based on standard practices in the field and reproducibility across biological replicates. Three independent adult female sea stars were used for single-cell RNA sequencing experiments. Statistical analyses associated with the single-cell datasets, including differential gene expression, trajectory reconstruction, and GO enrichment analyses, were performed using established computational frameworks implemented in Seurat, Monocle3, topGO, and clusterProfiler. HCR, FISH, IHC, and fluorescent staining experiments were independently replicated at least 3 times using independent biological replicates and at least three ovarian lobes per experiment or, in the case of larvae and juveniles, more than 50 larvae and more than 6 juveniles per experiment. In detail, panels c–j in Fig. 1, c–h in Fig. 2, b–e in Fig. 4, a–b, f–i, and l–m in Fig. 5, c–f in Fig. 6, and a–b and d–h′ in Fig. 7 are representative of multiple independent biological replicates (5, 5, 7, 3, 3, and 4 replicates, respectively), each consisting of at least three distinct ovarian lobes. Fig. 1 panel h and Fig. S1 panel p are representative of at least four distinct imaging areas from at least two ovarian lobes collected from three different females. Fig. 1 panel i is representative of six distinct regions across two ovarian lobes. Fig. 6 panel g is quantified from a single representative dataset and was independently performed a total of six times using multiple experimental formats to assess reproducibility. Quantification of TZPs in Fig. 6 panel g included n = 8 follicles per condition, and statistical significance was determined using a two-sided t-test (p = 0.0236). In control follicles, the minimum, maximum, median, lower quartile, and upper quartile values were 0.00035, 0.0144, 0.0083, 0.0053, and 0.0091, respectively. In LDN-193189-treated follicles, the minimum, maximum, median, lower quartile, and upper quartile values were 0.0075, 0.0231, 0.0128, 0.0104, and 0.0139, respectively. Fig. 6 panel h was independently repeated using three different females. In Fig. 6 panel h, blinded scoring of follicles with organized or disorganized TZPs was performed from a representative culture experiment. Fig. 6 panels i–j are representative images from these experiments. Figure S8 panels e–f were independently repeated three times, with quantified sample numbers indicated in the panels. Figure S8 panels g–h were performed once for staining analysis but were consistent with the morphological phenotypes observed across all three independent experiments shown in Fig. S8 panels e–f. Fig. S8 panels i–j were independently repeated twice. For larval LDN-193189 experiments, control and treated larvae were derived from the same biological replicate and processed in parallel under identical conditions. Quantification of dorsal pSMAD1/5 signal in Fig. S8 panel k included n = 6 larvae per condition, and statistical significance was determined using a two-sided t-test (p = 0.000235). No data were excluded from the analyses. The experiments were not randomized because animals were selected from the available pool of healthy wild-caught individuals. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5 Supplementary Data 6 Supplementary Data 7 Supplementary Data 8 Supplementary Data 9 Supplementary Data 10 Supplementary Data 11 Supplementary Data 12 Supplementary Movie 1 Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements We thank Jamie MacKinnon, Akshay Kane, and Karen Velez, current and past members of the Swartz lab, for advice and fruitful discussions throughout the duration of this project. We thank Maria Ina Arnone (Stazione Zoologica Anton Dohrn) for helping by providing the 1e11 (Syt1) and Mhc antibodies, and S. purpuratus ovaries. We also thank Rossella Annunziata (Stazione Zoologica Anton Dohrn) for kindly providing us with H. tubulosa ovaries. We thank Luc Reymond and Spirochrome for the gift of HAK-Actin. We thank Yamini Vadapalli, Pooja Mehta, and Peter Lénárt for sharing their expansion microscopy protocol and the HCR probe for cycA. We thank Kalynn Fette for performing preliminary experiments with LDN-193189. We would also like to thank Louie Kerr, the Central Microscopy Facility at the Marine Biological Laboratory (Woods Hole, MA, USA), and Chris Bjornsson (ZEISS) for imaging assistance with confocal microscopy. We are grateful to Lisa Abbo and the Marine Resources Center staff for providing excellent animal care and husbandry of adult P. miniata used in this work. Author contributions Conceptualization: P.P., S.Z.S.; Methodology: P.P., B.N., S.Z.S.; Formal analysis: P.P., B.N., C.W., D.V.; Investigation: P.P., B.N., S.Z.S.; Writing–original draft: P.P.; Writing–review & editing: P.P., B.N., C.W., D.V., S.Z.S.; Visualization: P.P., C.W.; Supervision: S.Z.S.; Project administration: S.Z.S.; Funding acquisition: S.Z.S. Peer review Peer review information Nature Communications thanks Bruce Draper, Honoo Satake and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Funding The authors are grateful for the support of this work by the National Institutes of Health (R00HD099315 to S.Z.S.) and by a Faculty Recruitment Gift from the Hibbitt Foundation. P.P. was supported by a Postdoctoral fellowship awarded by the Global Consortium for Productive Health (previously known as Global Consortium for Reproductive Longevity and Equality) part of the Buck Center for Healthy Aging in Women. B.N. was supported by NIH F32HD117487. Data availability Source data for all figures and Supplementary Figs. are provided with the paper. The scRNA-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus database under accession number GSE292963. An interactive single cell atlas for the P. miniata integrated dataset generated in this study is available through the Broad Institute’s Single Cell Portal at https://singlecell.broadinstitute.org/single_cell/study/SCP3291. Previously published datasets used in this study include human prenatal ovary data (GSE143380; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE143380), mouse ovary data (GSE232309; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE232309), zebrafish ovary data (GSE191137; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE191137), sea urchin ovary data (GSE246430; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE246430), fruit fly ovary data (GSE136162; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE136162), mouse pituitary gland data (GSE153045; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE153045), and sea urchin larva data (GSE241045; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE241045). Source data are provided with this paper. Code availability Scripts used for data analysis can be found on GitHub (https://github.com/Perpag/Pm-ovary-clustering-analysis) and Zenodo (https://zenodo.org/records/20185904)105. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-74401-5. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Supplementary Data 2 Supplementary Data 3 Supplementary Data 4 Supplementary Data 5 Supplementary Data 6 Supplementary Data 7 Supplementary Data 8 Supplementary Data 9 Supplementary Data 10 Supplementary Data 11 Supplementary Data 12 Supplementary Movie 1 Reporting Summary Transparent Peer Review file Source Data Data Availability Statement Source data for all figures and Supplementary Figs. are provided with the paper. The scRNA-seq data generated in this study have been deposited in the NCBI Gene Expression Omnibus database under accession number GSE292963. An interactive single cell atlas for the P. miniata integrated dataset generated in this study is available through the Broad Institute’s Single Cell Portal at https://singlecell.broadinstitute.org/single_cell/study/SCP3291. Previously published datasets used in this study include human prenatal ovary data (GSE143380; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE143380), mouse ovary data (GSE232309; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE232309), zebrafish ovary data (GSE191137; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE191137), sea urchin ovary data (GSE246430; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE246430), fruit fly ovary data (GSE136162; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE136162), mouse pituitary gland data (GSE153045; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE153045), and sea urchin larva data (GSE241045; https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE241045). Source data are provided with this paper. Scripts used for data analysis can be found on GitHub (https://github.com/Perpag/Pm-ovary-clustering-analysis) and Zenodo (https://zenodo.org/records/20185904)105.