Skip to main content
NCBI home page
Nature Communications logoLink to Nature Communications
. 2026 Apr 10;17:5022. doi: 10.1038/s41467-026-71616-4

PGC-derived migrasomes couple PGC proliferation with migration

Boqi Liu 1,2,3, Zheng Jiang 1,2,4, Wenhao Song 1,2,3,6, Zhaocheng Zhai 1,2,3,7, Yaqi Li 1,2,4,8, Weiying Zhang 1,2,4, Anming Meng 1,2,4,5,✉, Li Yu 1,2,3,✉
PMCID: PMC13243449  PMID: 41957360

Abstract

The coordination of cell migration and proliferation is essential for embryogenesis and tissue homeostasis. However, the classical gradient signaling model is insufficient to explain how stable mitogenic signaling is maintained within migratory cells. Here, we reveal that primordial germ cells (PGCs) in zebrafish employ migrasomes—vesicular organelles formed during migration—to couple their proliferation with migration, ensuring germline expansion. Migrasomes, generated at retraction fibers via tspan7-dependent biogenesis, deliver the growth factor GDF3 specifically to neighboring PGCs through contact-dependent interactions. GDF3 activates the TGF-β receptor acvr1ba, driving proliferation in a spatiotemporally restricted manner. This homocrine signaling mechanism allows migrating PGCs to autonomously sustain proliferation, circumventing signal dilution in embryonic environments. This work uncovers migrasomes as a bridge linking migration and proliferation, with implications for understanding collective cell behaviors in development and disease.

Subject terms: Germline development, Cell proliferation, Cell migration

This study reveals migrating zebrafish primordial germ cells use migrasomes as mobile carriers to transport growth signals between neighbours, coupling cell migration with proliferation to ensure robust population expansion during their journey.

Introduction

The orchestration of embryonic development relies on intricate signaling networks, a central principle of which is that positional information is conveyed by morphogen gradients. This concept was first proposed by Thomas Hunt Morgan (1905) and later formalized by Lewis Wolpert in his “French Flag” model, which posits that signals are transmitted by diffusion to provide positional information1–3. Subsequent studies have revealed additional modes of intercellular communication during embryogenesis, including direct cell–cell contact, gap junctions, cytonemes, and tunneling nanotubules4–11. However, these models are limited in explaining signal transmission between migratory cells. Unlike static cells, migrating cells lack stable cell contacts and are constantly exposed to a dynamic environment. Therefore, additional conceptual frameworks are necessary to understand how migrating cells coordinate with each other to ensure robust proliferation, differentiation, and pattern formation.

Migrasomes are vesicular organelles formed on retraction fibers behind migrating cells, representing an evolutionarily conserved structure found in species from dictyostelium to zebrafish, chicken, and mouse12–17. Within them, secretory cargo, including cytokines, chemokines, and growth factors, are actively transported into migrasomes by motor proteins, analogous to the synaptic vesicle transport in neurons17–19. By detaching and targeting specific locations, migrasomes act as mobile signaling hubs to release secretory cargo17,18, making them ideally suited for cell communication within dynamic environments. Previously, three distinct migrasome-mediated signaling modes have been described: the “region cue” mode, where migrasomes provide localized positional signals to guide mesendodermal cell movements during zebrafish gastrulation14; the “vanguard” mode, where migrasomes from leading monocytes guide trailing cells during angiogenesis15; and the “drone” mode, which involves targeted cytokine delivery by circulating monocyte migrasomes during inflammatory responses17. While these mechanisms demonstrate the versatility of migrasome-based signaling, they primarily describe communication between different cell types. This leaves a critical question unanswered: do migrasomes transduce signals within a migratory cell cluster to coordinate behavior?

Primordial germ cells (PGCs), the embryonic precursors of gametes, are among the most actively migratory cells during embryogenesis20–23, making them an excellent in vivo model for studying signal transmission within migratory cell populations. PGCs migrate as clusters20. However, unlike classical collective migration, they do not maintain stable neighbor–neighbor contacts but instead interact only transiently24,25. This poses a significant challenge for communication, as direct signaling through cell adhesion is unlikely. Despite this, PGCs must coordinate their behaviors—migration and proliferation—to reach the gonads in sufficient numbers26–30. How PGCs reliably receive mitogenic signals during migration, and how proliferation and migration are functionally coupled, remain poorly understood.

Here, we identify migrasomes as signaling hubs that couple proliferation and movement within PGC clusters, establishing a fourth mode of migrasome-mediated signaling we term the “renewkit” mode. We demonstrate that PGCs employ migrasomes to deliver the growth factor GDF3—a member of the TGF-β superfamily essential for embryonic development31–35—to neighboring PGCs through a contact-dependent mechanism. This migrasome-mediated transfer of GDF3 activates the receptor acvr1ba, specifically promoting PGC proliferation. Our findings uncover an intrinsic signaling mechanism by which migratory cells reliably propagate proliferative signals within their cohort, ensuring coordinated population expansion during migration through complex environments. Thus, migration itself becomes a driver of proliferation via migrasome-based signaling.

Results

Tspan7 regulates migrasome formation in PGCs

To enable in vivo live imaging of PGCs, we established the Tg(kop:eGFP-caax-nos3UTR) transgenic line. The kop promoter drives germline-specific expression, while the nanos3 3 ′UTR (nos3UTR) stabilizes the transgene mRNA selectively within germ cells36,37. Zebrafish PGC migration initiates at the onset of gastrulation (~5–6 hpf) and continues until the cells colonize the gonadal ridge around 24 hpf20. Consistent with this active migratory phase, we observed migrasome-like vesicles generated by PGCs (Fig. 1a). The size was about 1 µm (Fig. 1b). Live imaging revealed that these vesicles, located at the tips of retraction fibers, emerge at 5 hours post-fertilization (hpf) and peak in abundance by 11 hpf, coinciding with active PGC migration (Fig. 1c, d and Supplementary Movie 1). It is well-established that PGCs form dynamic filopodia, predominantly localized at the front of the cell38. In contrast, retraction fibers are exclusively localized at the rear of the cell and are significantly longer, with an average length of 20 µm, compared to the 10 µm length of filopodia (Supplementary Fig. 1a–c). Additionally, retraction fibers are more stable, with an average duration of 650 s, whereas filopodia last only 211 s (Supplementary Fig. 1d). Filopodia also exhibit full-length polymerized actin, while retraction fibers have polymerized actin confined to their base, as previously reported (Supplementary Fig. 1e, f)13,38,39. These differences clearly indicate that the membrane tubules we observed are retraction fibers, and the vesicles are migrasomes. We also observed some vesicles that were not attached to retraction fibers, suggesting they may either be detached migrasomes or derived from apoptotic cells. To distinguish between these possibilities, we overexpressed the well-established migrasome marker tspan4-mcherry in PGCs14,40. We found that these detached vesicles were tspan4-positive, confirming their identity as migrasomes (Supplementary Fig. 1g). To exclude an apoptotic origin, we expressed secreted AnnexinV-mCherry (SecA5-mcherry), as used in a previous study41, and confirmed that these vesicles lacked phosphatidylserine exposure, thereby excluding the possibility of apoptotic bodies (Supplementary Fig. 1h).

Fig. 1. Tspan7 regulates migrasome formation in PGCs.

Fig. 1

Open in a new tab

a Confocal images of PGCs in Tg(kop:eGFP-caax-nos3UTR) zebrafish showing extracellular vesicle production at indicated time points. Scale bars, 5 μm (main) and 2 μm (insets). b Quantification of vesicle diameters. n = 75 vesicles. c Time-lapse imaging of migrasome formation during PGC migration. Yellow arrow indicates migration direction; white arrow indicates migrasome. Scale bars, 5 μm (main) and 2 μm (insets). d The number of migrasomes per PGC was quantified during PGC migration. n = 30, 5 hpf; n = 44, 7 hpf; n = 64, 9 hpf; n = 41, 11 hpf; n = 53, 13 hpf; n = 37, 15 hpf; n = 57, 17 hpf; n = 49, 19 hpf; n = 50, 21 hpf; n = 35, 23 hpf; n = 40, 25 hpf. e Heat map of top 10 tspan gene expression in PGCs by RNA-seq. n = 3 replicates. FPKM, fragments per kilobase million. f Representative images of migrasomes in WT and mutant embryos at 12 hpf. Scale bar, 10 μm. g Quantification of the number of migrasomes per PGC for the experiments in (f). n = 47 WT; n = 33 MZcd9b; n = 30 MZtspan14; n = 35 MZcd63; n = 49 MZtspan7. h Specific rescue of tspan7 in PGCs by injecting tspan7-nos3UTR mRNA at the one-cell stage. Scale bar, 10 μm. i Quantification of the number of migrasomes per PGC for the experiments in (h). n = 48 WT; n = 44 MZtspan7; n = 39 Rescue (PGC). Data in (b),(d),(g),(i) represent the mean ± s.e.m. from three independent experiments. g, i P values were calculated using a two-tailed, unpaired t-test. NS not significant.

Building on previous studies implicating tetraspanins in migrasome biogenesis40, we performed RNA profiling on sorted PGCs from Tg(kop:eGFP-caax-nos3UTR) embryos, and identified cd9b, tspan14, cd63, and tspan7 as the four most highly expressed tetraspanins in PGCs (Fig. 1e). Then, we generated maternal-zygotic (MZ) mutants for these genes and found that only MZtspan7 mutants exhibited a significant reduction in migrasome numbers (Fig. 1f, g and Supplementary Fig. 2a–d). Using the nos3UTR method, which directs specific expression to PGCs (Supplementary Fig. 2e), we performed rescue experiments by injecting tspan7-nos3UTR at the one-cell stage. This restored migrasome numbers in MZtspan7 mutants, indicating that tspan7 is essential for migrasome formation in PGCs (Fig. 1h, i). Moreover, tspan7 was indeed expressed in PGC (Supplementary Fig. 2f). And protein located in retraction fiber and migrasomes (Supplementary Fig. 2g). Together, these data suggest that tspan7 regulates migrasome formation in PGCs.

PGC-derived migrasomes regulate PGC proliferation

We next investigated the physiological role of PGC-derived migrasomes. Our previous study reported that approximately 20% of MZtspan7 mutants display ectopically localized PGCs14, suggesting potential defects in migration. We confirmed this mislocalization using the Tg(kop:eGFP-caax-nos3UTR) line (Supplementary Fig. 3a). However, detailed analysis revealed no significant differences in PGC migration speed or distance (Supplementary Fig. 3b, c). This suggests that the ectopic phenotype likely reflects impaired guidance rather than reduced migratory ability. Since migratory behavior appeared largely intact, we next examined additional phenotypes and found that the number of PGCs, visualized by the Tg(kop:eGFP-caax-nos3UTR) line, was markedly reduced in MZtspan7 embryos (Fig. 2a, b). At 6 hpf, the number of PGCs in MZtspan7 mutants was not significantly different from WT embryos. However, after the onset of PGC migration, the PGC count began to increase in WT embryos, while it remained virtually unchanged in MZtspan7 mutants. The reduced number of PGCs in MZtspan7 was rescued by injecting tspan7-nos3UTR (Fig. 2c). This difference was not due to cell death, as active Caspase-3 labeling revealed that the vast majority of PGCs in both WT and MZtspan7 mutants remained viable (Supplementary Fig. 3d, e). Closer examination indicated that PGCs in the mutant embryos failed to divide, as confirmed by 5-ethynyl-2′-deoxyuridine (EdU) labeling and PGC live-imaging, which showed a marked reduction in PGC proliferation in MZtspan7 mutants (Fig. 2d–g and Supplementary Movie 2). And the injection of tspan7-nos3UTR recovered the proliferation of PGCs (Fig. 2f, g). These data suggest that tspan7 is essential for PGC proliferation.

Fig. 2. PGC-derived migrasomes regulate PGC proliferation.

Fig. 2

Open in a new tab

a Representative live images of PGCs (green) in WT and MZtspan7 embryos from 6 to 48 hpf. Nuclei are labeled with nls-mCherry (magenta). Scale bar, 100 μm. b Quantification of PGC numbers per embryo for experiments in (a). 6 hpf, n = 37 Control, n = 41 MZtspan7; 13 hpf, n = 57 Control, n = 63 MZtspan7; 24 hpf, n = 58 Control, n = 29 MZtspan7; 48 hpf, n = 45 Control, n = 39 MZtspan7. c PGC numbers at 48 hpf in WT, MZtspan7, and PGC-specific rescue embryos. n = 86 WT; n = 78 MZtspan7; n = 51 Rescue (PGC). d EdU labeling (magenta) of proliferating PGCs (green) from 7 to 10 hpf. Scale bar, 25 μm. e Quantification of EdU-positive PGCs for experiments in (d). n = 39 WT; n = 34 MZtspan7. f Live imaging of PGC divisions from 7 to 10 hpf in WT, MZtspan7, and rescued embryos. Different colors label distinct PGC clones. Scale bar, 20 μm. g Quantification of PGC division percentage for experiments in (f). n = 47 WT; n = 41 MZtspan7; n = 37 Rescue (PGC). h Schematic of PGC transplantation. Donor PGCs (magenta) from WT or MZtspan7 embryos were transplanted into MZtspan7 hosts (green PGCs) at 6 hpf. i Representative images of migrasome formation at 12 hpf after transplantation. Green, host PGCs; magenta, donor PGCs. Scale bar, 10 μm. j Quantification of migrasomes per PGC for experiments in (i). n = 32 WT; n = 22 MZt7; n = 13 WT-MZt7; n = 7 MZt7-MZt7. k Representative images of PGCs at 48 hpf after transplantation. Green indicates host PGCs. Magenta indicates donor PGCs. Scale bar, 100 μm. l Quantification of host PGC numbers for experiments in (k). n = 38 WT; n = 59 MZt7; n = 11 WT-MZt7; n = 11 MZt7-MZt7. Data in (b), (c), (e), (g), (j), and (l) represent the mean ± s.e.m. from three independent experiments. P values were calculated using two-tailed unpaired t-tests.

To determine whether tspan7 regulates PGC proliferation in a migrasome-dependent manner, we performed a transplantation assay. We transplanted 1–3 PGCs from either WT or MZtspan7 (MZt7) mutants into MZtspan7 mutant embryos (Fig. 2h). Upon transplantation, donor WT PGCs actively migrated alongside host PGCs (Supplementary Fig. 3f). These cells maintained their ability to generate migrasomes and even produced more migrasomes compared to WT embryos (Fig. 2i, j and Supplementary Fig. 3g, h). Crucially, the number of mutant PGCs was recovered only when WT PGCs were transplanted, not when MZtspan7 mutant PGCs were transplanted, as the latter failed to generate migrasomes (Fig. 2k, l). Although the MZt7 to MZt7 group shows a modest increase in mean PGC number, this difference is not statistically significant and likely reflects variability inherent to transplantation assays. These results suggest that tspan7 regulates PGC proliferation via migrasomes.

To further investigate the source of the proliferative signal, specifically whether it originates from neighboring PGCs or the surrounding somatic environment, we transplanted MZt7 PGCs into WT host embryos. Following transplantation, we observed a proximity-dependent rescue effect. MZt7 PGCs located at a distance from host PGCs largely retained their proliferation defects, whereas mutant PGCs positioned in close proximity to WT PGCs showed a marked increase in proliferative capacity (Supplementary Fig. 3i, j). These findings demonstrate that tspan7-deficient PGCs receive proliferative signals from neighboring PGCs rather than from surrounding somatic cells.

PGC-derived migrasomes regulate PGC proliferation via GDF3

The proliferation of PGCs during embryonic development is regulated by both cell-autonomous mechanisms and extrinsic signaling cues28,42–44. To identify key factors driving PGC proliferation, we analyzed PGC RNA profiling data to identify highly expressed growth factors (Fig. 3a). We then generated MZ mutants for top 4 growth factors and found that in MZgdf3 mutants, the number of PGCs was significantly reduced (Fig. 3b and Supplementary Fig. 4a–d). Meanwhile, the migration ability and migrasome number did not change in MZgdf3 embryos (Supplementary Fig. 5a–d). Gdf3 is a member of the TGF‑β superfamily that plays a pivotal role in cellular growth, differentiation, and embryonic development32,45–48. Its function in PGC biology has not been previously reported. Although MZgdf3 mutants are embryonically lethal, they can survive up to 24 hpf, providing a time window to study PGCs and PGC-derived migrasomes.

Fig. 3. Gdf3 in PGC migrasomes regulates the proliferation of PGCs.

Fig. 3

Open in a new tab

a Heat map of top 10 growth factor genes expressed in PGCs at 12 hpf. FPKM, fragments per kilobase million. b Quantification of PGC numbers at 14 hpf in indicated mutants. n = 35 WT; n = 58 MZgdf3; n = 64 MZgmfb; n = 72 MZmanf; n = 67 MZmdkb. c EdU labeling (magenta) of proliferating PGCs (green) in WT and MZgdf3 embryos from 7–10 hpf. Scale bar, 10 μm. d Quantification of EdU-positive PGCs for experiments in (c). n = 22 WT; n = 20 MZgdf3. e PGC numbers at 12 hpf in WT, MZgdf3, and rescue (gdf3-nos3UTR) embryos. n = 55 WT; n = 55 MZgdf3; n = 50 Rescue (PGC). f Schematic of PGC-specific gdf3 knockdown strategy using CRISPR/Cas9. g Imaging PGCs in 48 hpf embryos. tdTomato signal marks embryos with PGC-specific gdf3 knockdown. Scale bar, 200 μm. h Quantification of PGC numbers for experiments in (g). n = 51 Control; n = 56 U6:gdf3-gRNA. i Schematic of PGC transplantation. GDF3 is indicated by yellow dots. j Representative images of PGCs at 48 hpf after transplantation. Green, host PGCs; magenta, donor PGCs. Scale bar, 100 μm. k Quantification of host PGC numbers for experiments in (j). n = 47 WT; n = 39 MZt7; n = 18 WT-MZt7; n = 11 MZgdf3-MZt7. l Double fluorescence in situ hybridization showing gdf3 (magenta) expression in PGCs (Vasa-positive) at 9.5 hpf. Nuclei were stained with DAPI. Scale bar, 10 μm. m Immunostaining at 9.5 hpf shows GDF3 enrichment in PGC migrasomes (arrow). Scale bar, 10 μm. n qRT-PCR analysis of gdf3 receptor expression in sorted PGCs at 10 hpf. Data are mean ± s.e.m. from three technical replicates. o Representative images of PGCs at 13 hpf after injection of acvr1baDN-nos3UTR or acvr1ba(K234R)-nos3UTR mRNA. Scale bar, 100 μm. p Quantification of PGC numbers for experiments in (o). n = 55 WT; n = 53 acvr1baDN-nos3UTR; n = 62 acvr1ba(K234R)-nos3UTR. q Representative images of PGCs at 13 hpf after injection of acvr1baCA-nos3UTR mRNA into MZtspan7 embryos. Scale bar, 100 μm. r Quantification of PGC numbers for experiments in (q). n = 34 MZtspan7; n = 42 acvr1baCA-nos3UTR. Data in (b), (d), (e), (h), (k), (p), and (r) represent the mean ± s.e.m. from three independent experiments. P values were calculated using two-tailed unpaired t-tests.

In MZgdf3 mutants, PGC proliferation was significantly reduced, as confirmed by EdU labeling (Fig. 3c, d). Expressing gdf3-nos3UTR in these mutants rescued PGC number, further validating this observation (Fig. 3e). To strengthen these findings, we specifically knocked down gdf3 in PGCs (Fig. 3f and Supplementary Fig. 5e, f). In this system, while gdf3 gRNAs are ubiquitously expressed, cas9 expression is driven by the PGC-specific ziwi promoter, ensuring tissue-specific mutagenesis. As expected, this significantly reduced their number, suggesting that gdf3 expressed in PGCs is crucial for their proliferation (Fig. 3g, h). Migrasome formation was not impacted in embryos with specific knockdown of gdf3 (Supplementary Fig. 5g, h).

To determine whether gdf3 in migrasomes is the key factor driving PGC proliferation, we transplanted WT or MZgdf3 mutant PGCs into MZtspan7 mutants, which have impaired migrasome biogenesis (Fig. 3i). We found that both WT and MZgdf3 mutant PGCs were capable of generating migrasomes when transplanted into MZtspan7 mutant embryos (Supplementary Fig. 5i, j). However, only WT PGCs, not gdf3 mutant PGCs, rescued PGC number in MZtspan7 mutants (Fig. 3j, k). Together, these data suggest that PGC-derived migrasomes maintain PGC proliferation via gdf3.

It has been reported that during early embryonic development, gdf3 is widely expressed throughout the embryo33,49. The fact that PGC-derived migrasomes maintain PGC proliferation via gdf3 prompted us to examine gdf3 expression during PGC migration. If gdf3 were expressed in surrounding cells, it would be difficult to understand why PGC-derived migrasomes are required to maintain PGC proliferation in a GDF3-dependent manner, as surrounding cells could provide GDF3 to PGCs. Using fluorescence in situ hybridization (FISH), we observed that gdf3 transcripts are ubiquitously distributed at early developmental stages but become progressively restricted to PGCs during gastrulation (Supplementary Fig. 6a). At 9.5 hpf, gdf3 expression was predominantly detected in PGCs, whereas the surrounding somatic tissues exhibited minimal signal (Fig. 3l). To quantify this differential expression, we performed RT–qPCR analysis and found that gdf3 expression levels were significantly higher in PGCs than in non-PGC cells during gastrulation, coinciding with the period of PGC proliferation (Supplementary Fig. 6b–e). These results indicate that PGCs represent a major source of gdf3 expression at this stage of development. To further examine the localization of GDF3 protein, we generated a custom antibody and verified its specificity (Supplementary Fig. 6f, g). Immunostaining revealed that GDF3 protein is highly restricted to the PGC cluster (Supplementary Fig. 6h). Within individual cells, the protein predominantly localized to migrasomes rather than the cell body, similar to the enrichment pattern previously reported for other secretory proteins (Fig. 3m)14.

To further confirm the role of gdf3 in PGC proliferation, we targeted gdf3 receptors50–52. We first assessed the expression levels of known gdf3 receptors and found that acvr1ba had the highest expression in PGCs (Fig. 3n). To block gdf3 signaling, we injected dominant-negative acvr1baDN-nos3UTR and kinase-dead acvr1ba(K234R)-nos3UTR53 at the one-cell stage, resulting in a reduction in the number of PGCs (Fig. 3o, p and Supplementary Fig. 6i). Conversely, expression of a constitutively active acvr1ba (acvr1baCA)54, which bypasses ligand-dependent activation, restored PGC numbers in MZtspan7 mutants (Fig. 3q, r). In addition, multiple independent loss-of-function approaches targeting acvr1ba—including F0 CRISPR/Cas9-mediated knockdown55,56, morpholino knockdown, and pharmacological inhibition—consistently led to reduced PGC numbers (Supplementary Fig. 6j–p), further supporting a requirement for acvr1ba signaling in PGC proliferation. Together, these data suggest that gdf3, delivered by PGC-derived migrasomes, regulates PGC proliferation through acvr1ba-mediated signaling.

In vitro reconstitution of migrasome-dependent PGC proliferation

To directly test the role of PGC-derived migrasomes in promoting PGC proliferation, we sought to use PGC-derived migrasomes to stimulate PGC proliferation in vitro. This experiment was initially hindered by the limited number of PGCs and migrasomes in each embryo. To overcome this challenge, we employed a recently developed method for generating induced primordial germ cells (iPGCs), in which somatic cells are reprogrammed to adopt the characteristics of primordial germ cells using specific genetic factors or signaling pathways57. By introducing 13 germplasm factors (GMs) at the one-cell stage, we successfully reprogrammed the majority of cells into iPGCs, which were capable of generating migrasomes (Fig. 4a–c). After purification by differential centrifugation, migrasomes derived from iPGCs were confirmed using scanning electron microscopy (SEM) (Fig. 4d). This approach provided sufficient material to isolate a significant amount of iPGC-derived migrasomes.

Fig. 4. In vitro reconstitution of migrasome-dependent PGC proliferation.

Fig. 4

Open in a new tab

a Schematic of iPGC migrasome (iPGC_mig) purification. 13GMs were injected at 1-cell stage to induce iPGCs; iPGC_mig was purified at 7 hpf by differential centrifugation. b Representative images showing iPGC induction in 13GM-injected embryos at 7 hpf. Scale bar, 100 μm. c Mosaic labeling shows iPGC migrasome formation at 7 hpf. Right, magnified view. Scale bars, 10 μm (left), 2.5 μm (right). d SEM image of purified iPGC_mig. Scale bar, 300 nm. e Schematic of in vitro PGC proliferation assay. PGCs sorted from 8.5 hpf embryos were incubated with iPGC_mig purified from WT, MZgdf3, or gdf3-overexpressing iPGCs. f Western blot confirms GDF3 protein in iPGC and iPGC_mig. g Quantification of PGC division percentage during 2 h incubation with indicated iPGC_mig or PBS. Data are mean ± s.e.m. from three independent experiments. P values were calculated using two-tailed unpaired t-tests. h Schematic of RNA-seq sample preparation. PGCs sorted at 8.5 hpf were incubated with or without iPGC_mig for 1.5 h, then collected for RNA-seq. i, j Heat map (i) and volcano plot (j) shown the differentially expressed genes (DEGs) between PGC and PGC (mig). Data from three replications. k KEGG pathway enrichment analysis of DEGs. Enrichment analysis was performed using the hypergeometric test (one-sided). P values adjusted for multiple comparisons using the Benjamini-Hochberg (FDR) method. Data in (b), (c), (d), and (f) are representative of three independent experiments with similar results.

This in vitro system not only allowed us to directly test whether migrasomes promote PGC proliferation but also enabled us to carry out in-depth mechanistic analyses. We first generated iPGCs in WT, MZgdf3 mutant, and gdf3-overexpressing embryos and incubated these iPGC-derived migrasomes with PGCs (Fig. 4e). Western blot analysis confirmed the enrichment of GDF3 in migrasomes compared to iPGC cell bodies, exosomes, and supernatant (Fig. 4f and Supplementary Fig. 7a). Importantly, incubation of these iPGC-derived migrasomes with PGCs promoted proliferation in a gdf3-dose-dependent manner (Fig. 4e–g). There was very little proliferation in PGCs incubated with migrasomes isolated from MZgdf3 mutants. In contrast, we observed significant proliferation in PGCs incubated with WT and gdf3-overexpressing migrasomes, with the gdf3-overexpressing group exhibiting the highest proliferation rate.

To examine the transcriptional response initiated by migrasomes, we incubated isolated PGCs with iPGC-derived migrasomes for 1.5 h, followed by RNA profiling (Fig. 4h). We observed significant changes in the mRNA profile of migrasome-incubated PGCs (Fig. 4i, j). Notably, KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis of differentially expressed genes (DEGs) revealed that the most enriched biological process was the cell cycle, which correlates with the role of PGC-derived migrasomes in regulating proliferation (Fig. 4k). Together, these data further support the role of migrasome-delivered gdf3 in PGC proliferation.

Contact-dependent, specific delivery of GDF3 to PGCs by iPGC-derived migrasomes

Transplantation experiments revealed that PGC-derived migrasomes frequently attached to PGCs (Fig. 5a and Supplementary Movie 3), suggesting that migrasomes may deliver GDF3 to PGCs through direct contact. We first investigated whether PGC-derived migrasomes specifically adhere to PGCs rather than PGC surrounding cells (PSCs) (Fig. 5b). Indeed, migrasomes derived from iPGCs preferentially bind to PGCs rather than to PSCs (Fig. 5c, d). This finding suggests that the delivery of GDF3 to PGCs is specific.

Fig. 5. Contact-dependent, specific delivery of GDF3 to PGCs by iPGC-derived migrasomes.

Fig. 5

Open in a new tab

a Donor PGCs (magenta) transplanted into host embryos (green PGCs). Arrows: attached migrasomes; arrowheads: unattached migrasomes. Scale bar, 10 μm. b Schematic of in vitro attachment assay. PGCs or PSCs were incubated with iPGC_mig for 1 h, washed, and imaged. c Representative images of iPGC_mig (magenta) attached to PGCs or PSCs. Dashed lines indicate cell bodies. Scale bar, 5 μm. d Quantification of attached iPGC_mig number per cell. n = 56 PGC; n = 30 PSC. e Diagram illustrating the system to monitor GDF3-3xFlag signal transfer from migrasome to PGCs. f Live imaging showing Flag signal accumulation on PGCs over 1 h 20 min. Scale bar, 5 μm. g Flag signals on PGC or PSC surfaces after 1.5-h incubation. Scale bar, 5 μm. h Quantification of Flag intensity on cell surface. n = 91 PGC; n = 56 PSC. i Schematic of agarose barrier assay to physically block contact between PGCs and iPGC_mig. j Imaging of PGCs and iPGC_mig under control or agarose conditions. Dashed lines indicate agarose boundary. Scale bar, 5 μm. k Flag signals on PGCs after 1.5 h incubation under control or agarose conditions. Scale bar, 5 μm. l Quantification of Flag intensity on PGC surface. n = 46 Control; n = 40 Agarose. m Model for PGC-derived migrasomes in regulating PGC proliferation. Images in (a), (f), and (j) are representative of three independent experiments. Data in (d), (h), and (l) represent the mean ± s.e.m. from three independent experiments. P values were calculated using two-tailed unpaired t-tests.

To visualize the release of GDF3 from migrasomes and its subsequent binding to PGCs, we generated iPGCs expressing gdf3-3×flag, isolated migrasomes containing GDF3-3×Flag, and incubated them with PGCs in vitro. We added a fluorescently labeled anti-Flag antibody, which can only detect GDF3-3×Flag after its release from membrane-bound migrasomes (Fig. 5e). After 1 h and 20 min of incubation, Flag signals were observed on the surface of PGCs, indicating that GDF3 was released from migrasomes and bound to receptors on PGCs (Fig. 5f and Supplementary Movie 4). In contrast, no Flag signal was detected on PGCs incubated with iPGC-derived migrasomes lacking GDF3-3×Flag, confirming the specificity of the signal (Supplementary Fig. 7b). Strikingly, when PSCs were incubated with these GDF3-3×Flag migrasomes, no Flag signal was observed, further supporting the idea that migrasomes specifically deliver GDF3 to PGCs but not to other cells (Fig. 5g, h).

To test whether GDF3 delivery by migrasomes is contact-dependent, we embedded migrasomes in soft agarose and added PGCs on top of the migrasome-containing agarose (Fig. 5i). In this setup, where migrasomes were physically separated from PGCs, no Flag signal was detected, indicating that direct contact is required for GDF3 delivery (Fig. 5j–l). Transwell experiments showed the same result (Supplementary Fig. 7c–e). Together, these experiments suggest that migrasomes specifically deliver GDF3 to PGCs in a contact-dependent manner, and this interaction is exclusive to PGCs.

Discussion

While signaling gradients are fundamental to embryonic development, the classical model faces limitations in dynamic contexts. Migrasomes, novel organelles released by migrating cells, offer a solution by acting as mobile signaling hubs. Beyond our previously identified “region cue,” “vanguard,” and “drone” modes14,15,17, we now describe a fourth migrasome-mediated signaling mode: the “renewkit” mode, where migrasomes coordinate proliferation and migration within PGC clusters. Specifically, we demonstrate that migrasomes generated by migrating PGCs can directly deliver the signaling molecule GDF3 to neighboring cells within the cell population. Upon delivery, GDF3 activates acvr1ba-mediated signaling in recipient cells, driving their proliferation (Fig. 5m). These findings position migrasomes as precise vehicles for autocrine or homocrine signaling during migration, ensuring that proliferative cues remain confined to the PGC cluster.

Our study establishes an unexpected link between PGC migration and proliferation, revealing that migrasomes act as bridges between these processes. Crucially, our data reveal that this relationship is unidirectional: migration functions upstream to trigger proliferation via migrasome formation. Conversely, proliferation itself is dispensable for motility, as evidenced by the unperturbed migration observed in gdf3 mutants. By facilitating the contact-dependent delivery of GDF3 to neighboring PGCs, migrasomes ensure that migration itself promotes germline expansion. This mechanism resolves the spatial and temporal challenges of coordinating motility and division. Beyond germline biology, this paradigm may extend to other migratory cell types, where balancing motility and self-renewal is critical. By coupling proliferation to migration through migrasome-mediated signaling, PGCs exemplify how migrasomes integrate complex cellular behaviors, ensuring developmental robustness in rapidly changing embryonic environments.

Our study expands the known roles of migrasomes in homocrine signaling. Migrasomes selectively bind migrating PGCs via contact-dependent interactions, likely mediated by adhesion molecules, electrostatic interactions, and germline-specific receptors, ensuring that proliferative cues remain restricted to the PGC cluster. This self-sustaining system resolves the challenge of maintaining proliferation in an environment where the only source of ligand is secreted by the PGCs themselves. PGC-derived migrasomes deliver signals directly to other PGCs, thereby preventing dilution and off-target effects. Consistent with recent reports of migrasome-mediated transport of VEGF during angiogenesis and chemokines during gastrulation14,15, our findings suggest that migrasomes may represent a general strategy for spatially concentrating signals to coordinate collective behaviors in migratory cell populations, including neural crest cells and metastatic cancer cells, warranting further investigation.

Our study provides preliminary evidence supporting a specific, contact-dependent mechanism for the delivery of GDF3 via migrasomes. However, the precise molecular machinery governing the physical interaction between migrasomes and PGCs remains unresolved. Notably, migrasomes are enriched with adhesion-associated molecules58, including integrins, which raises the possibility that specific adhesion molecular may act as “recognition tags.” Alternatively, differential surface properties, such as electrostatic interactions, could also contribute to this selective adherence. Future investigation is required to distinguish between these possibilities—whether driven by specific adhesion or biophysical forces—and to determine how such binding triggers the downstream release of GDF3.

Methods

Animal model and subject details

The wild-type (WT) zebrafish utilized in this study were derived from the Tuebingen (Tu) strain. Adult zebrafish were maintained in a water-circulating system at a constant temperature of 28.5 °C. Subsequent to fertilization, the eggs were incubated at 28.5 °C in Holtfreter’s solution (0.059 M NaCl, 0.00067 M KCl, 0.00076 M CaCl2, and 0.0024 M NaHCO3). All procedures involving adult zebrafish and embryos were performed in accordance with the guidelines established by the Animal Care and Use Committee of Tsinghua University. Sex was not considered in this study. All newly generated zebrafish lines are available from the corresponding author Li Yu upon request.

Generation of transgenic fish

The promoter region of kop and the 3′UTR of nanos3 (nos3UTR) were amplified using primers listed in Supplementary Data 1. Subsequently, the kop promoter fragment, nos3UTR, and egfp-caax/mcherry-caax sequences were inserted into a construct containing the tol2 transposon repeat sequence via homologous recombination (D0204P, Beijing LABLEAD Trading Co., Ltd). For generating transgenic fish, purified plasmid DNA (25 pg) and tol2 transposase mRNA (100 pg) were co-injected into 1-cell-stage zebrafish embryos. Transgenic fish expressing egfp-caax or mcherry-caax were identified by fluorescence detection using a stereomicroscope (Olympus, MVX10) on the progeny of injected females.

Transgenic fish for specifically knocking down gdf3 in PGCs were generated in this study. For male fish, three gdf3 gRNAs were ubiquitously expressed throughout the zebrafish using the U6a, U6b, and U6c promoters59. The promoter regions of gsc and tdtomato were amplified using primers listed in Supplementary Data 1. Subsequently, all fragments were inserted into a single construct. Transgenic fish were identified by tdtomato fluorescence.

For female fish, the Tg(ziwi:Cas9NLS-nos3UTR) line was a gift from A. Meng and was crossed with Tg(kop:eGFP-caax-nos3UTR). Double transgenic fish was identified by eGFP fluorescence and PCR amplification of the cas9 gene.

Generation of mutant lines by CRISPR–Cas9

To produce the F0 generation of mutants, 100 pg of each guide RNA (Supplementary Data 1) was co-injected with 300 pg of cas9 mRNA into one-cell-stage Tg(Kop:eGFP-caax-nos3UTR) embryos. F1 generation individuals were genotyped via PCR and sequencing using the primers detailed in Supplementary Data 1. Mutant lines harboring out-of-frame insertions or deletions were retained.

To generate F0 CRISPR/Cas9-mediated mutagenesis of acvr1ba, embryos at the one-cell stage were injected with a mixture of three acvr1ba-specific gRNAs together with Cas9 protein (NEB, M0646T). Embryos were imaged at 12 hpf. The gRNA sequences were as follows: acvr1ba gRNA1, GCTACAGCAGTTCGTCGAGG; acvr1ba gRNA2, TGGGCTGTCCAGGAGGAACC; and acvr1ba gRNA3, GCCTCCACATCCTACATCAA.

mRNA synthesis and injection

In this study, all plasmids were initially linearized using the appropriate New England Biolabs (NEB) restriction enzymes. Subsequently, capped mRNAs were generated with the T7/Sp6 mMessage mMachine Kit (Invitrogen, AM1344 and AM1340) and purified using VAHTS RNA Clean Beads (Vazyme, N412-01).

For rescue experiments, 100 pg of mRNA was injected into embryos at the one-cell stage. For PGC induction experiments, a mixture of 13GMs (50 pg each mRNA), gdf3-3xflag (50 pg), eGFP-caax-nos3UTR (50 pg), or mcherry-caax-nos3UTR (50 pg) was injected into embryos at the one-cell stage57. Additionally, Acvr1baDN-nos3UTR (100 pg) and Acvr1ba(K234R)-nos3UTR (100 pg) mRNA were injected into embryos at the one-cell stage. Acvr1baCA-nos3UTR (10 pg) mRNA was injected into embryos at the eight cells stage. For mosaic expression of egfp-caax/mcherry-caax, mRNA (75 ng) was injected into embryos at the eight cells stage. Lifeact-mcherry-nos3UTR (75 ng) was injected into embryos at the one-cell stage.

Whole-mount RNA ISH

Probe templates were amplified by PCR from complementary DNA (cDNA) derived from embryos at specific developmental stages. The primer sequences are provided in Supplementary Data 1. Probe synthesis was performed using T7 RNA polymerase (Roche, 10881767001). For whole-mount ISH, embryos were rehydrated, permeabilized with proteinase K, and hybridized with DIG-labeled antisense RNA probes overnight at 65 °C. Following post-hybridization washes in SSC buffers, embryos were incubated with anti-DIG-alkaline phosphatase antibody (Roche, 11093274910, 1:5000). Staining was developed using NBT/BCIP substrate (Roche, 11681451001), and embryos were mounted in 70% glycerol for imaging60.

Whole-mount double fluorescent in situ hybridization (FISH) was performed using the same protocol with the following modifications61: DIG- or fluorescein-labeled antisense RNA probes were used for hybridization, followed by incubation with anti-digoxigenin-POD (Roche, 11633716001, 1:1000) and anti-fluorescein-POD (Roche, 11426346910, 1:500) antibodies. Signals were detected using TSA staining (Akoya Biosciences, NEL753001KT).

Whole-mount immunofluorescence

To stain endogenous GDF3, Tg(Kop:eGFP-caax-nos3UTR) embryos at 9.5 hpf were initially fixed in 4% paraformaldehyde for 1 day at 4 °C. Subsequently, the fixed embryos were dehydrated using a graded methanol series and stored in −20 °C methanol overnight. The dehydrated embryos were then rehydrated through a graded 0.1% PBST (Triton X-100; Solarbio, T8200-500ml). Following rehydration, embryos were incubated in blocking solution (1% BSA and 10% goat serum in PBST) for 1 h at room temperature on a slow shaker, followed by overnight incubation with primary antibodies diluted 1:250 (Rabbit anti-GDF3, CUSABIO company) and 1:500 (Chicken anti-GFP, Abcam, ab13970) in blocking solution at 4 °C on a slow shaker. After removing the primary antibody solution, embryos were washed in 0.1% PBST for at least 3 × 1 h at room temperature on a shaker. Secondary antibodies, Alexa Fluor 488-conjugated Anti-Chicken IgY H&L (1:400; ab150173, Abcam) and Alexa Fluor 546-conjugated anti-Rabbit IgG (1:250; A11010, Invitrogen), were diluted in blocking solution and incubated for 1 day at 4 °C. The embryos were then washed in 0.1% PBST as before and finally mounted in 1% low-melting-point agarose for imaging.

For caspase3 staining, rabbit anti-caspase3 (BD biosciences, 559565) was diluted 1:250 in blocking solution.

The anti-GDF3 (zebrafish) antibody was generated by CUSABIO company in this study.

Transplantation assay

For PGCs transplantation experiments, PGCs were aspirated in situ from 6 hpf WT, MZtspan7, and MZgdf3 embryos using a micropipette under a fluorescence stereomicroscope (Olympus, MVX10). The isolated PGCs were then transplanted into the PGC region of MZtspan7 embryos at the same developmental stage. Recipient embryos were subsequently cultured to 12 hpf for the quantification of migrasome numbers and to 48 hpf for the quantification of PGC numbers.

EdU staining

The EdU Cell Proliferation Image Kit (Abbkine, KTA2031) was used to label the DNA of proliferating cells with AbFluor 545 azide. The protocol was modified for zebrafish as follows:

Tg(Kop:eGFP-caax-nos3UTR) embryos were dechorionated using Pronase (Protease from Streptomyces griseus, Sigma, cat. no. P5147) and incubated in EdU solution (1 µM) at 28 °C for 3 h. The embryos were then fixed in 4% PFA overnight, dehydrated using a graded methanol series, and stored in −20 °C methanol overnight. The dehydrated embryos were rehydrated through a graded methanol series (75%, 50%, and 25% methanol in PBST containing 0.1% Triton X-100), followed by washes in PBST. The samples were then stained with anti-GFP antibodies following the immunofluorescence protocol and labeled with AbFluor 545 azide according to the manufacturer’s instructions.

Prepare cDNA and qPCR analysis

The Single Cell Full Length mRNA Amplification Kit (Vazyme, N712) was used to prepare cDNA from PGCs. The protocol was shown as follows:

Approximately 500 PGCs were sorted from Tg(kop:eGFP-caax-nos3UTR) embryos using a flow cytometer. These cells were lysed in 5 µL of sample buffer (supplied in the kit), and cDNA was prepared according to the manufacturer’s protocol. The expression levels were quantified by qPCR using the primers listed in Supplementary Data 1.

For qPCR in Supplementary Fig. 5f and Supplementary Fig. 6c, about 500 PGCs or 2500 non-PGC cells were used to prepare qPCR template by using the Single Cell Sequence Specific Amplification Kit (Vazyme, P621).

iPGC induction and iPGC_migrasome purification

The RNA of 13GMs (germplasm factors) was synthesized using the mMESSAGE mMACHINE T7 Transcription Kit (Thermo Fisher Scientific, AM1344) and injected into zebrafish embryos at the 1-cell stage to induce PGCs.

Approximately 100 iPGC-induced embryos were collected at 7 hpf and dechorionated using pronase. The embryos were washed in Holtfreter’s solution and homogenized by pipetting. The samples were then centrifuged at 500 × g for 10 min at 4 °C to remove cells, followed by 2000 × g for 20 min at 4 °C to remove cell debris, and finally at 20,000 × g for 1 h at 4 °C. The pellet was resuspended in culture medium for in vitro incubation or in 2.5% Gluta (LEAGENE, DF0151) for Scanning Electron Microscopy.

In vitro culture of PGC/PSC and iPGC_migrasomes

Approximately 300 PGCs were sorted from Tg(Kop:eGFP-caax-nos3UTR) embryos by using a flow cytometer. Mcherry-caax mRNA was injected into Tg(Kop:eGFP-caax-nos3UTR) embryos at one-cell stage. PSCs were aspirated in situ from these mRNA injected embryos using a micropipette under a stereomicroscope. iPGC_migrasomes were purified according to above description. PGCs/PSCs were incubated with iPGC_migrasomes in 20 µL of culture medium. The medium was covered with mineral oil (Nanjing Aibei Biotechnology Co. Ltd, M2460) to inhibit evaporation. Cells were imaged using confocal microscopy.

For flag staining, 13GMs, gdf3-3xflag, and mcherry-caax/egfp-caax mRNA were co-injected in wild-type embryos at one-cell stage. Then, iPGC_migrasomes were purified and co-cultured with PGCs/PSCs. Anti-Flag-647 (Invitrogen, MA1-142-A647, 1:50 dilution) was added to the culture medium. Cells were imaged using confocal microscopy.

For agarose experiments, iPGC_migrasomes were mixed with low-melting-point agarose (VWR Life Science, 9012-36-6). The mixture was immediately dropped onto the bottom of a confocal dish. After 3 min of solidification, add PGCs, anti-Flag-647, and culture medium on top.

For transwell experiments, PGCs and iPGC_migrasomes were separated by a 0.4-µm filter membrane (Corning, 3413). The system was cultured at 30 °C for 1.5 h. After incubation, PGCs were transferred to a confocal dish and imaged using confocal microscopy.

RNA sequencing

The Single Cell Full Length mRNA Amplification Kit (Vazyme, N712) and TruePrep RNA Library Prep Kit for Illumina (Vazyme, TR503) were used to prepare RNA library. The protocol was shown as follows:

Approximately 200 Tg(kop:eGFP-caax-nos3UTR) embryos were collected at 12 hpf. The embryos were dechorionated using pronase and homogenized by pipetting. The dispersed cells were centrifuged at 500 × g to remove the supernatant, and the pellet was resuspended in PBS. Sorting was performed based on eGFP fluorescence intensity to isolate the eGFP-positive PGC population from eGFP-negative somatic cells using a flow cytometer. About 500 PGCs were collected. These cells were lysed in 5 µL of sample buffer (supplied in the kit) and subjected to mRNA amplification and reverse transcription using the first kit (Vazyme, N712). The amplified cDNA was then used to prepare the RNA library according to the manufacturer’s protocol for the second kit (Vazyme, TR503). RNA sequencing was performed by Azenta Life Sciences company.

About 500 primordial germ cells (PGCs) were co-incubated with iPGC migrasomes in vitro for 1.5 h. Subsequently, the cells were resuspended in 500 µL of culture medium, centrifuged at 500 × g to remove the supernatant, and the pelleted cells were lysed with sample buffer for RNA library construction and sequencing.

Imaging

To acquire z-stack images of living embryos for the statistical analysis of PGC number or migrasomes, embryos were embedded in 1% low-melting-point agarose, and z-stack imaging was performed using spinning disk microscopy.

To acquire images of FISH, IF, or EdU-stained embryos, embryos were embedded in 1% low-melting-point agarose and imaged using a Nikon AX microscope.

To acquire images of in vitro cultured PGCs or PSCs, the sorted cells and purified migrasomes were resuspended in 20 µL of culture medium. The medium was covered with mineral oil to inhibit evaporation. Cells were imaged using spinning disk microscopy.

Image processing

Time-lapse multiple-view z-stack embryo images were processed using Imaris software 8.1.4 (Bitplane AG).

Images of FISH, IF, and EdU staining were processed using NIS-Elements 5.4.

Fluorescence intensity was quantified using ImageJ, and statistical analyses were conducted using GraphPad Prism 8.

To quantify the length of filopodia, retraction fibers and lifact, as well as the duration of filopodia and retraction fibers, images were processed using Imaris software 8.1.4.

Statistical analysis

Statistical analyses were conducted using an unpaired two-tailed t-test in GraphPad Prism 8 software. Data are presented as the mean ± SEM. A description of each statistical test, including n and P values, is provided in the figure legends and figures. The number of experimental repeats is specified in the figure legends. Technical replicates are performed for the qPCR experiments.

Reporting summary

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

Supplementary information

Supplementary Information (1.7MB, pdf)
41467_2026_71616_MOESM2_ESM.pdf (35.4KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (11.8KB, xlsx)
Supplementary Movie 1 (4.6MB, mp4)
Supplementary Movie 2 (19.7MB, mp4)
Supplementary Movie 3 (4.2MB, mp4)
Supplementary Movie 4 (17.9MB, mp4)
Reporting Summary (1.9MB, pdf)
Transparent Peer Review file (1.9MB, pdf)

Source data

Source data (1.1MB, xlsx)

Acknowledgements

We are grateful to the members of the Meng and Yu groups for their helpful discussions. This research was supported by the New Cornerstone Science Foundation (NCI202529, L.Y.), the National Natural Science Foundation of China (grant no. 32330025, L.Y.), Tsinghua University Dushi Program (grant no. 20251080019, L.Y.), Scientific and Technological Innovation Project of China Academy of Chinese Medical Sciences (grant no. CI2023C024YL, L.Y.), and the Ministry of Science and Technology of the People’s Republic of China (grant nos. 2024YFF1502900, 2024YFA1307301, L.Y.). We would like to acknowledge the State Key Laboratory of Biomembrane and Membrane Biotechnology for their support with confocal microscopy imaging and flow cytometry analysis. We thank Hui Zhang, Fan Lei, and Hongshuang Li for their technical support and help with data analysis. We thank Ying Li and Xiaomin Li (Cryo-EM Facility of China National Center for Protein Sciences) for their technical assistance with SEM and for their help with the Helios G3 UC electron microscope (Thermo Fisher Scientific). B.L. was funded by the Tsinghua University-Peking University Joint Center for Life Sciences. B.L. was supported by the Advanced Innovation Fellow Program of the Beijing Frontier Research Center for Biological Structure.

Author contributions

L.Y. and A.M. conceived the experiments, wrote the paper and supervised the project. B.L. carried out the experiments. Z.J. helped with the experiment of transplantation. W.S. and Z.Z. contributed to the mutant-zebrafish verification. Y.L. and W.Z. generated the Tg(ziwi:Cas9NLS-nos3UTR) line.

Peer review

Peer review information

Nature Communications thanks Carl-Philipp Heisenberg and the other anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The raw sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under accession code PRJNA1405346. All data needed to evaluate the conclusions are present in the article, the supplementary information, and source data file provided with this paper. Source data are provided with this paper.

Competing interests

Li Yu is the scientific founder of Migrasome Therapeutics Ltd. All other authors declare no conflicts of interest.

Footnotes

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

Contributor Information

Anming Meng, Email: mengam@mail.tsinghua.edu.cn.

Li Yu, Email: liyulab@mail.tsinghua.edu.cn.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-71616-4.

References

  • 1.Wolpert, L. Positional information and the spatial pattern of cellular differentiation. J. Theor. Biol.25, 1–47 (1969). [DOI] [PubMed] [Google Scholar]
  • 2.Wolpert, L., Tickle, C. & Arias, A. M. Principles of Development (Oxford University Press, 2015).
  • 3.Keros, T. et al. The centenary progress of molecular genetics. A 100th anniversary of T. H. Morgan’s discoveries. Coll. Antropol.34, 1167–1174 (2010). [PubMed] [Google Scholar]
  • 4.Kumar, N. M. & Gilula, N. B. The gap junction communication channel. Cell84, 381–388 (1996). [DOI] [PubMed] [Google Scholar]
  • 5.Singer, S. J. Intercellular communication and cell-cell adhesion. Science255, 1671–1677 (1992). [DOI] [PubMed] [Google Scholar]
  • 6.Kornberg, T. B. & Roy, S. Cytonemes as specialized signaling filopodia. Development141, 729–736 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Yamashita, Y. M., Inaba, M. & Buszczak, M. Specialized intercellular communications via cytonemes and nanotubes. Annu. Rev. Cell Dev. Biol.34, 59–84 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Hall, E. T. et al. Cytoneme signaling provides essential contributions to mammalian tissue patterning. Cell187, 276–293 e223 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Zhang, C., Brunt, L., Ono, Y., Rogers, S. & Scholpp, S. Cytoneme-mediated transport of active Wnt5b-Ror2 complexes in zebrafish. Nature625, 126–133 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Inaba, M., Buszczak, M. & Yamashita, Y. M. Nanotubes mediate niche-stem-cell signalling in the Drosophila testis. Nature523, 329–332 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Miao, L. et al. Tunneling nanotube-like structures regulate distant cellular interactions during heart formation. Science387, eadd3417 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Yu, J. et al. Inchworm migration facilitates amoeboid cell adaptation to high-adhesion environments. Dev. Cell.10.1016/j.devcel.2025.06.035 (2025). [DOI] [PubMed]
  • 13.Ma, L. et al. Discovery of the migrasome, an organelle mediating release of cytoplasmic contents during cell migration. Cell Res. 25, 24–38 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Jiang, D. et al. Migrasomes provide regional cues for organ morphogenesis during zebrafish gastrulation. Nat. Cell Biol.21, 966–977 (2019). [DOI] [PubMed] [Google Scholar]
  • 15.Zhang, C. et al. Monocytes deposit migrasomes to promote embryonic angiogenesis. Nat. Cell Biol.24, 1726–1738 (2022). [DOI] [PubMed] [Google Scholar]
  • 16.Jiang, D. et al. Neutrophil-derived migrasomes are an essential part of the coagulation system. Nat. Cell Biol.26, 1110–1123 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Jiao, H. et al. Packaged release and targeted delivery of cytokines by migrasomes in circulation. Cell Discov.10, 121 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Jiao, H. et al. Localized, highly efficient secretion of signaling proteins by migrasomes. Cell Res.34, 572–585 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Lin, R. C. & Scheller, R. H. Mechanisms of synaptic vesicle exocytosis. Annu Rev. Cell Dev. Biol.16, 19–49 (2000). [DOI] [PubMed] [Google Scholar]
  • 20.Cantu, A. V. & Laird, D. J. A pilgrim’s progress: seeking meaning in primordial germ cell migration. Stem Cell Res.24, 181–187 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Aalto, A., Olguin-Olguin, A. & Raz, E. Zebrafish primordial germ cell migration. Front Cell Dev. Biol.9, 684460 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Doitsidou, M. et al. Guidance of primordial germ cell migration by the chemokine SDF-1. Cell111, 647–659 (2002). [DOI] [PubMed] [Google Scholar]
  • 23.Knaut, H., Werz, C., Geisler, R., Nusslein-Volhard, C. & Tubingen Screen, C. A zebrafish homologue of the chemokine receptor Cxcr4 is a germ-cell guidance receptor. Nature421, 279–282 (2003). [DOI] [PubMed] [Google Scholar]
  • 24.Reichman-Fried, M., Minina, S. & Raz, E. Autonomous modes of behavior in primordial germ cell migration. Dev. Cell6, 589–596 (2004). [DOI] [PubMed] [Google Scholar]
  • 25.Weidinger, G. et al. Regulation of zebrafish primordial germ cell migration by attraction towards an intermediate target. Development129, 25–36 (2002). [DOI] [PubMed] [Google Scholar]
  • 26.Nakamura, Y. et al. Migration and proliferation of primordial germ cells in the early chicken embryo. Poult. Sci.86, 2182–2193 (2007). [DOI] [PubMed] [Google Scholar]
  • 27.Farini, D., Scaldaferri, M. L., Iona, S., La Sala, G. & De Felici, M. Growth factors sustain primordial germ cell survival, proliferation and entering into meiosis in the absence of somatic cells. Dev. Biol.285, 49–56 (2005). [DOI] [PubMed] [Google Scholar]
  • 28.Wang, P. et al. NRF1 promotes primordial germ cell development, proliferation and survival. Cell Prolif.57, e13533 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Runyan, C. et al. Steel factor controls midline cell death of primordial germ cells and is essential for their normal proliferation and migration. Development133, 4861–4869 (2006). [DOI] [PubMed] [Google Scholar]
  • 30.Cantu, A. V., Altshuler-Keylin, S. & Laird, D. J. Discrete somatic niches coordinate proliferation and migration of primordial germ cells via Wnt signaling. J. Cell Biol.214, 215–229 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Bisgrove, B. W., Su, Y. C. & Yost, H. J. Maternal Gdf3 is an obligatory cofactor in Nodal signaling for embryonic axis formation in zebrafish. Elife10.7554/eLife.28534 (2017). [DOI] [PMC free article] [PubMed]
  • 32.Montague, T. G. & Schier, A. F. Vg1-Nodal heterodimers are the endogenous inducers of mesendoderm. Elife10.7554/eLife.28183 (2017). [DOI] [PMC free article] [PubMed]
  • 33.Pelliccia, J. L., Jindal, G. A. & Burdine, R. D. Gdf3 is required for robust Nodal signaling during germ layer formation and left-right patterning. Elife10.7554/eLife.28635 (2017). [DOI] [PMC free article] [PubMed]
  • 34.Peterson, A. G., Wang, X. & Yost, H. J. Dvr1 transfers left-right asymmetric signals from Kupffer’s vesicle to lateral plate mesoderm in zebrafish. Dev. Biol.382, 198–208 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Levine, A. J. & Brivanlou, A. H. GDF3, a BMP inhibitor, regulates cell fate in stem cells and early embryos. Development133, 209–216 (2006). [DOI] [PubMed] [Google Scholar]
  • 36.Blaser, H. et al. Transition from non-motile behaviour to directed migration during early PGC development in zebrafish. J. Cell Sci.118, 4027–4038 (2005). [DOI] [PubMed] [Google Scholar]
  • 37.Koprunner, M., Thisse, C., Thisse, B. & Raz, E. A zebrafish nanos-related gene is essential for the development of primordial germ cells. Genes Dev.15, 2877–2885 (2001). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Meyen, D. et al. Dynamic filopodia are required for chemokine-dependent intracellular polarization during guided cell migration in vivo. Elife10.7554/eLife.05279 (2015). [DOI] [PMC free article] [PubMed]
  • 39.Stanganello, E. et al. Filopodia-based Wnt transport during vertebrate tissue patterning. Nat. Commun.6, 5846 (2015). [DOI] [PubMed] [Google Scholar]
  • 40.Huang, Y. et al. Migrasome formation is mediated by assembly of micron-scale tetraspanin macrodomains. Nat. Cell Biol.21, 991–1002 (2019). [DOI] [PubMed] [Google Scholar]
  • 41.van Ham, T. J., Mapes, J., Kokel, D. & Peterson, R. T. Live imaging of apoptotic cells in zebrafish. FASEB J.24, 4336–4342 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Resnick, J. L., Bixler, L. S., Cheng, L. & Donovan, P. J. Long-term proliferation of mouse primordial germ cells in culture. Nature359, 550–551 (1992). [DOI] [PubMed] [Google Scholar]
  • 43.Whyte, J. et al. FGF, Insulin, and SMAD signaling cooperate for avian primordial germ cell self-renewal. Stem Cell Rep.5, 1171–1182 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Okamura, D. et al. Cell cycle gene-specific control of transcription has a critical role in proliferation of primordial germ cells. Genes Dev.26, 2477–2482 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Masurkar, N. et al. Novel cardiokine GDF3 predicts adverse fibrotic remodeling after myocardial infarction. Circulation147, 498–511 (2023). [DOI] [PubMed] [Google Scholar]
  • 46.Tykwinska, K., Lauster, R., Knaus, P. & Rosowski, M. Growth and differentiation factor 3 induces expression of genes related to differentiation in a model of cancer stem cells and protects them from retinoic acid-induced apoptosis. PLoS ONE8, e70612 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Levine, A. J., Levine, Z. J. & Brivanlou, A. H. GDF3 is a BMP inhibitor that can activate Nodal signaling only at very high doses. Dev. Biol.325, 43–48 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Ehira, N. et al. An embryo-specific expressing TGF-beta family protein, growth-differentiation factor 3 (GDF3), augments progression of B16 melanoma. J. Exp. Clin. Cancer Res.29, 135 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Dohrmann, C. E., Kessler, D. S. & Melton, D. A. Induction of axial mesoderm by zDVR-1, the zebrafish orthologue of Xenopus Vg1. Dev. Biol.175, 108–117 (1996). [DOI] [PubMed] [Google Scholar]
  • 50.Martinez-Hackert, E., Sundan, A. & Holien, T. Receptor binding competition: a paradigm for regulating TGF-beta family action. Cytokine Growth Factor Rev.57, 39–54 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Andersson, O., Korach-Andre, M., Reissmann, E., Ibanez, C. F. & Bertolino, P. Growth/differentiation factor 3 signals through ALK7 and regulates accumulation of adipose tissue and diet-induced obesity. Proc. Natl. Acad. Sci. USA105, 7252–7256 (2008). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Preiss, H. et al. Regulation of Nodal signaling propagation by receptor interactions and positive feedback. Elife10.7554/eLife.66397 (2022). [DOI] [PMC free article] [PubMed]
  • 53.Jazwinska, A., Badakov, R. & Keating, M. T. Activin-betaA signaling is required for zebrafish fin regeneration. Curr. Biol.17, 1390–1395 (2007). [DOI] [PubMed] [Google Scholar]
  • 54.Renucci, A., Lemarchandel, V. & Rosa, F. An activated form of type I serine/threonine kinase receptor TARAM-A reveals a specific signalling pathway involved in fish head organiser formation. Development122, 3735–3743 (1996). [DOI] [PubMed] [Google Scholar]
  • 55.Wu, R. S. et al. A Rapid Method for Directed Gene Knockout for Screening in G0 Zebrafish. Dev. Cell46, 112–125.e114 (2018). [DOI] [PubMed] [Google Scholar]
  • 56.Kroll, F. et al. A simple and effective F0 knockout method for rapid screening of behaviour and other complex phenotypes. Elife10.7554/eLife.59683 (2021). [DOI] [PMC free article] [PubMed]
  • 57.Wang, X. et al. Induced formation of primordial germ cells from zebrafish blastomeres by germplasm factors. Nat. Commun.14, 7918 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Wu, D. et al. Pairing of integrins with ECM proteins determines migrasome formation. Cell Res.27, 1397–1400 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Yin, L. et al. Multiplex conditional mutagenesis using transgenic expression of Cas9 and sgRNAs. Genetics200, 431–441 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Thisse, C. & Thisse, B. High-resolution in situ hybridization to whole-mount zebrafish embryos. Nat. Protoc.3, 59–69 (2008). [DOI] [PubMed] [Google Scholar]
  • 61.Brend, T. & Holley, S. A. Zebrafish whole mount high-resolution double fluorescent in situ hybridization. J. Vis. Exp.10.3791/1229 (2009). [DOI] [PMC free article] [PubMed]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information (1.7MB, pdf)
41467_2026_71616_MOESM2_ESM.pdf (35.4KB, pdf)

Description of Additional Supplementary Files

Supplementary Data 1 (11.8KB, xlsx)
Supplementary Movie 1 (4.6MB, mp4)
Supplementary Movie 2 (19.7MB, mp4)
Supplementary Movie 3 (4.2MB, mp4)
Supplementary Movie 4 (17.9MB, mp4)
Reporting Summary (1.9MB, pdf)
Transparent Peer Review file (1.9MB, pdf)
Source data (1.1MB, xlsx)

Data Availability Statement

The raw sequencing data generated in this study have been deposited in the NCBI Sequence Read Archive (SRA) under accession code PRJNA1405346. All data needed to evaluate the conclusions are present in the article, the supplementary information, and source data file provided with this paper. Source data are provided with this paper.

Articles from Nature Communications are provided here courtesy of Nature Publishing Group

RESOURCES