SUMMARY
Germ granules, condensates of phase-separated RNA and protein, are essential for germline development, but how these molecules are organized within the granules and whether such an organization is relevant for germ cell fate is unclear. Combining three-dimensional in vivo structural and functional analyses, we study the dynamic spatial organization of molecules within zebrafish germ granules. We find that the vertebrate-specific Dead end protein is essential for positioning nanos3 RNA at the condensates’ periphery, where ribosomes are located. Without Dead end, or when translation is inhibited, nanos3 RNA translocates into granule interiors, far from the ribosomes’ location. These findings reveal the molecular mechanisms controlling the spatial organization of RNA within the phase-separated organelle and the importance of subgranule RNA localization for preserving germ cell totipotency.
INTRODUCTION
Germline development relies on molecules that reside within germ granules, phase-separated, non-membrane-bound condensates made of RNA and protein (RNP)1–6. However, the mechanisms that control the three-dimensional (3D) organization of the molecules within the condensates and the relevance of their organization for the function of the organelle are not fully understood3–5. An attractive model for studying these processes in vertebrates are zebrafish germ granules, given their relatively large size and accessibility for in vivo imaging7. A key component localized to these organelles is Dead end (Dnd1), a conserved RNA-binding protein that is essential for germline development in vertebrates. Loss of the protein results in sterility and the formation of germ cell tumors8–13 and accordingly recent work showed that germ cells with impaired Dnd1 function fail to maintain their totipotency and acquire somatic fates14. Despite Dnd1’s central role in controlling the fate of the vertebrate germline, the precise molecular function of the protein in the context of the phase-separated structure within which it resides is unknown. Here, we reveal a role for Dnd1 in regulating the spatial distribution and function of nanos RNA within germ granules, thereby ensuring proper development of the germline.
RESULTS
Nanos3 preserves germ cell fate and its mRNA is enriched at the periphery of germ granules
A known target of Dnd1 is the mRNA that encodes for the germ granule-localized protein Nanos3, which is essential for germ cell development in different organisms15–19. Our recent finding that germ cells depleted of Dnd1 adopt somatic cell fates14 prompted us to evaluate whether inhibiting Nanos3 could similarly affect the cells. To investigate this, we employed embryos carrying a mutation in the gene encoding for the chemokine Cxcl12a that prevents germ cells from migrating toward the gonad, such that they reside at ectopic positions throughout the embryo (Fig. 1A)20. Interestingly, while non-manipulated ectopic germ cells maintain their shape independent of the tissue in which they reside, germ cells with depleted Nanos3 protein adopt various somatic cell shapes (e.g. that of a muscle cell, (Fig. 1A,B)), consistent with the idea that Dnd1 and Nanos3 function in the same pathway required for maintaining germ cell fate14.
As the Dnd1 protein was previously shown to reside within germ granules21, we aimed to determine if the granules are spatially organized with respect to the distribution of molecules and if so, whether Dnd1 plays a role in controlling it. To this end, we focused on the distribution of nanos3 mRNA with respect to Dnd1, as well as that of Tudor domain containing 7 (Tdrd7) – encoding RNA and the granule-resident protein Vasa7,22,23. Here, we took advantage of the fact that during early embryonic stages (up to 10 hours post fertilization (hpf)), germ granules can be relatively large (Fig. S1A), ranging at sizes between 5 μm3 and 100 μm3, thus allowing for high spatial resolution analyses. We found that the germline-expressed mRNAs that encode Nanos3 and Tdrd7 are both enriched within the condensates (Fig. 1C), occupying largely non-overlapping punctate sub-granule positions (Fig. 1C). In contrast, the germline-expressed protein Vasa is distributed in a non-punctate pattern throughout the granule, allowing us to employ it for determining the volume and shape of the organelle (Fig. 1C, Fig. S1B, Movie S1).
To further characterize the distribution of specific mRNAs and proteins within the phase-separated condensates in vivo, we developed a 3D signal-visualization approach that allowed us to determine the spatial organization of molecules within granules in the 5-100 mm3 range (Fig. S1B). Intriguingly, upon evaluating the distribution of nanos3 and tdrd7 along the center-to-periphery axis of such germ granule condensates, we found that both mRNAs are more abundant in the peripheral layers, extending beyond the Vasa protein signal (Fig. 1D upper panels and Fig. 1E left graph, Movie S1). The bias in RNA distribution along this axis is observed in granules of different sizes, as evaluated by the ratio between the signal intensity at the periphery and core regions (Fig. 1F). The largely non-overlapping distribution pattern of the two RNA molecules (Fig. S1A, Movie S1) is similar to that observed in Drosophila, where distinct homotypic RNA clusters were found within germ granules24–27. Together, as exemplified by the analysis of nanos3 and tdrd7 RNAs, these findings demonstrate that zebrafish germ granules are patterned organelles that contain distinct homotypic RNA clusters, organized asymmetrically along concentric 3D layers.
Next, we set out to examine the localization of nanos3 mRNA with respect to the Dnd1 protein, which was shown to interact with this transcript28. Indeed, both molecules exhibit a similar distribution within the granules, with elevated levels observed at the periphery of the condensate (Fig. 1D lower panels, 1e right graph), where clusters of the mRNA and protein are found adjacent to each other, partially overlapping (Fig. S1C, Fig. 2A,B). Interestingly, following the localization of these molecules in live embryos employing the PP7 RNA detection system29,30, we found that Dnd1 protein and nanos3 mRNA clusters are highly dynamic, but maintain their proximity over time (Fig. 2C, Movie S2). These findings highlight the fact that while the distribution of molecules within the granules can differ for different RNAs and proteins, the spatial relationships among them can be maintained. This suggests cross dependency in localization among molecules and functional significance for sub-granule organization.
Dnd1 is required to maintain nanos3 mRNA clusters at the periphery of germ granule condensates
We have previously shown that nanos3 mRNA is a target of Dnd128, that knocking down Dnd1 results in loss of germ cell pluripotency14 and show here that knocking down Nanos3 results in a similar phenotype, where germ cells differentiate into somatic cells (Fig. 1A). These findings, together with the spatial relationships identified here between Dnd1 protein and nanos3 RNA, prompted us to ask whether Dnd1 controls nanos3 distribution within the granule. To investigate this, we examined the organization of germ granules in live embryos following inhibition of dnd1 RNA translation (Fig. 2D,E). Although Dnd1 loss eventually leads to a decrease in germ granule size and number at later stages of development14, we found that, at least until 10 hpf, the volume of large germ granules is not altered (Fig. S2A). Remarkably however, our 3D analysis of the distribution of molecules within the granules revealed a strong reduction in nanos3 mRNA levels at the periphery of the condensate, concomitant with an increase in the mRNA level in the core (Fig. 2D,E left panels and graph). Consistent with findings showing that Dnd1 interacts with the 3’UTR of nanos3 mRNA (nos 3’UTR)28, we found that this region of the transcript is sufficient for Dnd1 to control localization of the RNA within the condensate (Fig. 2F,G). Critically, these results do not represent a global effect on RNA distribution, as the localization of tdrd7 mRNA was not altered under these conditions (Fig. 2D,E right panels and graph). Thus, the distribution of different RNA molecules within germ granules can be regulated by distinct molecular mechanisms. Interestingly however, despite the dramatic effect of Dnd1 knockdown on the positioning of nanos3 mRNA clusters within the granules, the discrete homotypic clustering of tdrd7 and nanos3 molecules was preserved (Fig. S2B). These findings show that the basis for segregating RNA species from one another differs from the mechanisms that control the spatial distribution of the clusters within the granule.
Zebrafish germ granule condensates are enriched with ribosomes at their periphery
The altered distribution of nanos3 mRNA within the germ granule observed upon Dnd1 knockdown led us to examine whether the change in the spatial organization of the RNA could have a functional significance. As a first step in this analysis, we sought to determine the localization of factors that can regulate mRNA function. In line with recent findings in invertebrates suggesting that translationally silent mRNA molecules are stored within phases-separated granules31,32, we found that the RNA-induced silencing complex (RISC) components mir-430 and Argonaute protein 2 (Ago2) are enriched within zebrafish germ granules (Fig. 3A,B). Intriguingly, while the RISC-complex components are found throughout the germ granule, we find that the ribosomes are almost exclusively localized to the edge of the condensate (Fig. 3C, 60S ribosomal protein L10a staining). Ribosomes have also been observed at the periphery of Drosophila polar granules shortly after fertilization, but in this case they disappear from these sites prior to the formation of germ cell precursors33. The distinct L10a localization pattern observed in 10 hpf zebrafish embryos is also found at an earlier stage of germ cell development, shortly after specification of the germline and the initiation of maternal RNA translation (Fig. S3A)34. Importantly, the peripheral ribosome-rich regions partially overlap with the peripheral domains in which nanos3 mRNA clusters reside (Fig. 4B,C). Interestingly, in addition to the peripheral enrichment of ribosomes, we detected intra-granule domains that lack Vasa signal, and instead ribosomes and overlapping nanos3 mRNA are found (Fig. S3B). These observations suggest that the granule-resident Nanos3, a protein essential for germ cell development17 (Fig. 1A), is translated primarily at the borders of the phase-separated material.
Dnd1 activity is essential for nanos3 mRNA translation at the periphery of germ granule condensates
To gain further insight into the mechanisms by which Dnd1 controls the function of nanos3 mRNA within germ granules, we next examined the effect of Dnd1 depletion on the translation of the RNA. Indeed, Dnd1-depleted germ cells show a strong reduction in Nanos3 protein levels, as evaluated by the translation of injected nos 3’UTR-containing RNA encoding for a GFP-Nanos3 fusion protein (Fig. 4A)14. Importantly, the strong reduction in Nanos3 protein synthesis was associated with a loss of the overlap between nanos3 mRNA and the ribosomes at the germ granule periphery (Fig. 4B,C). This observation is consistent with the idea that Dnd1 aids in the association of the nanos3 mRNA with ribosomes and, consequently, to mRNA translation, thereby facilitating Nanos3 protein production. To determine whether the translational status of the RNAper se could be instructive regarding the localization of nanos3 mRNA, we inhibited translation initiation using a derivative of the natural product pateamine A, des-methyl, des-amino pateamine A (DMDAPatA)35–37 (Fig. S4A). Focusing on germ granules, we observed a strong depletion of nanos3 mRNA from the periphery of the granule following DMDAPatA treatment, with a concomitant increase in the level of the RNA in the core of the condensate (Fig. 4D,E). In addition to the effect of global inhibition of translation, specific inhibition of nanos3 mRNA translation by mutating the first Kozak sequence-containing ATG led to a similar effect. Specifically, egfp-nos 3’UTR RNA, in which the translation start codon was mutated, accumulated in the center of the granule and was largely absent from the periphery (Fig. 4F,G). Importantly however, mutating all ATGs within the open reading frame eliminated the accumulation of the RNA within germ granules (Fig. S4B-D). This finding suggests that the stability and localization of nanos3 mRNA molecules to germ granules is controlled by interactions with translation initiation factors. Indeed, we found that translation initiation factors eIF4G, eIF4E and PolyA-binding protein (PABP) are enriched within the condensates, where they largely overlap with nanos3 mRNA clusters (Fig. S4B,E).
Together, these findings are consistent with the idea that germ granules function as storage sites for mRNAs that are already associated with translation initiation complex proteins38, whereby translation of these RNA molecules occurs specifically at the periphery of the condensate, where the ribosomes and translation initiation factors are detected.
Dnd depletion causes a translocation of nanos3 mRNA to the condensate core rather than enhanced degradation at the periphery
Considering the effect of the translation status on nanos3 mRNA localization as described above, we proceeded to investigate the mechanisms by which Dnd1 regulates the localization of nanos3 mRNA to the granule periphery. Dnd1 was previously shown to interfere with the function of the micro-RNA (miRNA) mir-430, thereby increasing mRNA stability and translation28. Indeed, measuring the overall intensity levels of nanos3 RNA, we found that the total amount of the mRNA is reduced within Dnd1-depleted germ cells at 10 hpf (Fig. 5A). Interestingly however, in contrast to the decrease in the overall mRNA levels within the cells, we observed a specific and strong elevation of the amount of nanos3 mRNA within germ granules in the absence of Dnd1 (Fig. 5B). This increase in mRNA abundance suggests that a portion of the accumulated transcripts in the granule core is derived from the cytoplasm. More importantly, our findings indicate that the loss of transcripts from the periphery of the condensate (Fig. 2D) is a consequence of translocation to the core domain rather than merely representing a decay of mRNA at the granule periphery.
To further corroborate our hypothesis, we analyzed the effect of Dnd depletion in maternal zygotic (MZ) dicer mutants, which lack mature miRNAs39,40 (Fig. 5C,D). Indeed, we found that nanos3 mRNA localization to the condensate periphery was not restored in these mutants, indicating that the loss of nanos3 mRNA from these sites is independent of miRNA activity. Rather, our findings suggest that the positioning of nanos3 mRNA within the condensates is primarily controlled by its interaction with the active translation machinery. Together, we conclude that inhibition of Dnd1 renders nanos3 mRNA inaccessible to the translation machinery, leading to its accumulation in the granule core, and consequently to a drop of Nanos3 protein levels.
DISCUSSION
Our findings reveal how phase-separated vertebrate germ granules are organized concerning the spatial distribution of RNAs, proteins and the translation machinery. Importantly, we found that following the formation and segregation of germ plasm into the future germ cells6,41, patterning of the condensates is controlled by the vertebrate-specific RNA-binding protein Dead end, and we present the relevance of this arrangement for cell fate and germline development (Fig. 6).
For many organisms, post-transcriptional regulation and properties of phase separated condensates are key for specifying and preserving the fate of germ cells42,43. Indeed, recent studies have shown that translationally repressed mRNAs can accumulate in germ granules, where they are stored and protected from degradation31,32. At the same time, translation of such RNA molecules requires interactions between the transcripts and the translation machinery. In the case of early germ cell development, RNA translation must occur rapidly to ensure the maintenance of germline fate and avoid trans-fating into cells of other lineages, yet sufficient RNA silencing must be sustained for translation at later stages. Generating and maintaining such sub-granule organization of molecules is especially challenging in vertebrate germ cells, where the volume of the condensates is exceptionally large (e.g., up to 6 μm in diameter in zebrafish vs. up to 500 nm in Drosophila in early germline development44).
Since we found that the ribosomes are located at the periphery of the phase-separated germ granules, the translation of RNAs, such as those encoding for Nanos, could be controlled by the degree of localization of the RNAs to the border of the granules. Our study shows that the vertebrate-specific protein Dnd1 plays a key role in controlling the accessibility of specific granule-localized RNA to the translation machinery, thereby ensuring the maintenance of the germline (Fig. 6). Similar to the effect of Dnd1 knockdown, inhibiting the translation caused nanos RNA to accumulate at internal locations of the granules, suggesting that the capability of the RNA to be translated is important for its localization. The mechanism we describe allows for the production of proteins required for germ cell specification and fate maintenance, which has to occur rapidly at this stage to avoid the trans-fating of germline cells.
While the precise mechanisms by which Dnd1 controls nanos3 localization are unknown, our findings suggest that the formation of an active translation complex is required for maintaining mRNA clusters at the condensate periphery. Recent findings in C. elegans link P granule localization of nanos mRNA to translation repression31,32. Accordingly, association of repressors of nanos3 translation in Dnd1-depleted germ cells could maintain the transcript at the core of the granule. Consistently, in Xenopus, Dnd1 was show to prevent association of translational repressor eIF3f with nanos mRNA45. It would be interesting to determine whether a similar mechanism functions in zebrafish as well.
Our findings thus suggest that while the function of Nanos is conserved across the animal kingdom, rapid localization and maintenance of the RNA to the periphery is more challenging in larger vertebrate germ granules. We provide evidence that Dnd1 is a key protein that ensures the proper localization and function of fate determinants such as nanos RNA. Consistently, we found that knocking down Nanos results in trans-fating of germ cells into somatic cells, a phenotype originally described for Dnd114.
AUTHOR CONTRIBUTIONS
E.R. supervised the project. K.J.-W. and E.R. designed the work and wrote the manuscript. K.J.-W. performed all of the experiments except for the following: T.G.-T. quantified the cell shape of Nanos-depleted germ cells and A.G. acquired the respective microscopy images; K.T. generated MZdicer mutants; M.Z. synthesized DMDAPatA, K.H. and D.R. supervised the synthesis of the translation inhibitor, DMDAPatA. All authors read and approved the final manuscript with minor revisions.
DECLARATION OF INTERESTS
The authors declare that they have no competing interests.
METHODS
Resource availability
All source data are provided with this paper. Any information or requests for resources and reagents should be directed to the Lead Contact, Erez Raz (erez.raz{at}uni-muenster.de).
EXPERIMENTAL MODEL AND SUBJECT DETAILS
Zebrafish strains and handling
The following lines were used to obtain zebrafish (Danio rerio) embryos: wild type of the AB or AB/TL genetic background, cxcr4bt26035 46, medusaNYO45 47 and transgenic kop-egfp-f-nos-3’UTR43. Maternal zygotic (MZ) dicer embryos (dicerhu715) were generated by conducting a germline replacement as described previously39. Embryos were raised in 0.3× Danieau’s solution (17.4 mM NaCl, 0.21 mM KCl, 0.12 mM MgSO4·7H2O, 0.18 mM Ca(NO3)2) at 25°C, 28°C or 31°C. The fish were handled according to the regulations of the state of North Rhine-Westphalia, supervised by the veterinarian office of the city of Muenster.
METHOD DETAILS
Cloning of DNA constructs
To follow mRNA using the PP7 detection system in live embryos, the sequence of 24 PP7 stem loops49 was cloned into a Nanos3-nos 3’UTR plasmid17 downstream of the open reading frame (ORF) sequence. The Dnd-mScarlet-nos 3’UTR plasmid was generated by replacing the GFP sequence of a Dnd-GFP-nos 3’UTR plasmid50 with that of mScarlet. The nos 3’UTR-containing mRNAs used in Fig. 2F-G, Fig. 4F-G and Fig. S4B-C all carry an EGFP sequence in the ORF to allow for detection by RNAscope probes. For the generation of a non-translatable nos 3’UTR-containing mRNA used in Fig. 4 F-G, a 1-base pair deletion was introduced into the KOZAK-flanked ATG start codon of an EGFP coding sequence located upstream of the nos 3’UTR (EGFP-TG-nos 3’UTR). For this purpose, site-specific mutagenesis was employed51. For the generation of a translatable nos 3’UTR-containing mRNA used in Fig. 2F-G and Fig. 4F-G, the sequence encoding for amino acids 2-15 was deleted from a full-length EGFP sequence-containing plasmid employing a Restriction free cloning approach52 to prevent fluorescence emission of the EGFP protein (EGFPΔ2-15-nos 3’UTR). This step was required to simultaneously visualize multiple RNA transcripts without emission of the EGFP protein. To generate a nos 3’UTR-containing mRNA lacking all start codons in the ORF used in Fig. S4B-C, all ATGs within the EGFPΔ2-15-nos 3’UTR sequence were mutated to GTGs. Guanine was chosen to replace Adenine based on the structural similarity of the two bases. The respective ORF sequence carrying the ATG > GTG conversions was synthesized by Eurofins Genomics and cloned into a nos 3’UTR-containing plasmid. To generate the GFP-globin 3’UTR plasmid for global GFP expression in zebrafish, the GFP sequence was cloned upstream of a Xenopus globin 3’UTR.
Microinjection into zebrafish embryos
Capped sense mRNAs were synthesized using the mMessage mMachine kit (Ambion) according to the protocol of the manufacturer. 1 nL of mRNA and / or Morpholino-containing solution were injected into the yolk of 1-cell stage embryos. To mark the membrane of somatic cells, 45 ng/μl of mcherry-f’-glob 3’UTR (A709) were injected53. For visualization of protein within germ granules, vasa-gfp-vasa 3’UTR mRNA (291)54 was injected at a concentration of 100 ng/μl, dnd-gfp-nos 3’UTR (516) mRNA50 at 180 ng/μl and mgfp-nanos-nos 3’UTR (356) mRNA at 90 ng/μl 17. For the evaluation of Dnd protein and nanos3 mRNA movement, 140 ng of nanos-nos 24xPP7 3’UTR (D016) and 8 ng of nls-ha-tdpcp-yfp-glob 3’UTR (C987)55 were co-injected with 100 ng of dnd-mscarlet-nos 3’UTR (E513). To visualize nos 3’UTR-containing mRNA localization within germ granules, egfp-TG-nos 3’UTR, egfpΔ215-nos 3’UTR and egfpΔ2-15_mut-ATG-nos 3’UTR (D366, E801, E845) transcripts were injected at an equimolar concentration of 100 ng. For the evaluation of DMDAPatA drug efficiency, gfp-globin 3’UTR mRNA (B289) was injected at a concentration of 50 ng/μl. Nanos3 Morpholino (TGAATTGGAGAAGAGAAAAAGCCAT) was injected at a concentration of 60 μM17, Dnd Morpholino (GCTGGGCATCCATGTCTCCGACCAT) at 20 μM or 30 μM (Fig. 4A)50 and Cxcl12a MO (TTGAGATCCATGTTTGCAGTGTGAA) at 200 μM20.
RNA detection and immunostaining of whole-mount embryos
For all treatments, embryos were fixed in 4% PFA overnight at 4°C (6 hpf and 10 hpf stage) or for 2 h at room temperature (RT, 10 hpf stage) under slow agitation. Embryos were subsequently rinsed in 1x PBT, dechorionated and transferred to and stored in 100% methanol. Detection of mir-430 RNA in whole-mount embryos was performed using an in-situ hybridization procedure optimized for miRNA detection as previously described56,57. Dual labeled miRCURY LNA miRNA detection probes of dre-mir-430a-3p and scramble-miR (Qiagen, catalogue no. YD00613610-BCO and YD00699004-BCO) were hybridized at a concentration of 250 nmol overnight at 45°C. To co-visualize proteins with mRNAs within germ cells, an RNAscope in situ hybridization procedure was conducted using the RNAscope Multiplex Fluorescent Reagent Kit (ACD Bio) as previously described58. RNAscope probes Probe Diluent (ACD Bio 300041), Dr-nanos-C2 (ACD Bio 404521-C2), Dr-tdrd7-C2 (ACD Bio 536551-C2), EGFP-C1 (ACD Bio 400281-C1) and Dr-nanos3-CDS-C3 (ACD Bio 431191-C3) were hybridized overnight at 40°C. The embryos were then subjected to an immunostaining procedure as previously described, with a few modifications58. Briefly, embryos were permeabilized in 1x PBS containing 0.1% Tween-20 and 0.3% Triton X 100 (Carl Roth) for 1 h at RT and subsequently incubated in blocking solution (0.3% Triton X 100 and 4% BSA in 1x PBS) for 6 h at RT. Incubation with primary antibodies was then performed for 24 h at 4°C and incubation with secondary antibodies was done overnight at 4°C. The treatment with a GFP-targeting antibody was performed to enhance the fluorescent signal of expressed GFP-tagged Dnd protein. For the detection of Ago2 protein in germ cells, embryos were incubated in Dents solution (20% DMSO in methanol) for two days at 4°C, rehydrated in increasing amounts of 1x PBT in methanol (25%, 50%, 75% and 100%), re-fixed in 4% PFA for 20 minutes at RT and rinsed in 1x PBT. Embryos were then incubated in 1x PBT + 0.3% Triton X 100 for 1 h at RT and subsequently incubated for 5 h in blocking solution (1 x PBT + 20% goat serum + 5% DMSO) at RT. Subsequent treatment with primary and secondary antibodies was conducted as above.
Primary antibodies used: anti-Vasa (rabbit polyclonal, 1:400 dilution, GeneTex catalogue no. GTX128306), anti-GFP-1020 (polyclonal chicken, 1:600 dilution, Aves labs: SKU GFP-1020), anti-eIF4G (monoclonal mouse, 1:250 dilution, Sigma catalogue no. SAB1403762-100U), anti-PABP (polyclonal rabbit, 1:250 dilution, ThermoFisher catalogue no. PA5-17599), anti-eIF4E (polyclonal rabbit, 1:250 dilution, ThermoFisher catalogue no. PA5-86047), anti-L10a (monoclonal mouse, 1:100 dilution, Sigma catalogue no. WH0004736M1-100UG), anti-Ago2 (monoclonal mouse, 1:500 dilution, Abcam catalogue no. ab57113)
Secondary antibodies used: to visualize GFP, a 1:600 dilution of a 488 nm anti-chicken secondary antibody was employed (Jackson Immuno-Research); in all other cases, secondary antibodies from Thermo Fisher Scientific, catalogue no. A-11031, A-11034, A-11036 were employed using a 1:500 dilution for visualizing Vasa and a 1:1000 dilution for visualizing the remaining proteins.
Microscopy and image processing in zebrafish embryos
Images of germ granules in germ cells were acquired using a confocal microscope (LSM 710, Zeiss) equipped with 405 nm, 488 nm, 561 nm and 633 nm lasers and a 40x W N-Achroplan and a 63x W Plan-Apochromat objective (Zeiss). The ZEN software (Zeiss, version 2010B SP1, 6.0) was used to control the microscopy setup. Acquisition of different samples was conducted using the same pinhole size, optical slice thickness, gain and offset parameters. For images showing protein and RNA localization within germ granules, images were acquired with varying laser intensities (except for data presented in Fig. 4A and Fig. 5A-B) due to variability in germ cell depth within the organism and in germ granule size. This facilitated an optimal resolution without over exposure, which is required for accurately resolving mRNA structures. In this case, the mRNA signal intensities across the germ granule were expressed in relative amounts to compare the mRNA distribution among samples. For the 3D granule layer analysis in Fig. 1C-E and Fig. 2D-E, images of granules were acquired with a z-distance of 300 nm. To measure signal intensities of nanos3 mRNA within germ cells (Fig. 5A), images were acquired with a z-distance of 1 μm. For the time lapse data shown in Fig. 2C and Movie S1, stacks with a z-distance of 500 nm were acquired with a Yokogawa CSU-X1 Spinning Disk microscope equipped with a Piezo-controlled 63x W Plan-Apochromat objective (Zeiss) and connected to a Hamamatsu Orca Flash 4.0 V3 camera. The VisiView software (Visitron, version 4.0.0.14) was used to control the setup of the microscope. For the verification of Pateamine A drug efficiency, data was acquired with a 5x EC Plan-Neofluar objective (Zeiss) on the Spinning Disk. Images were deconvolved using the Huygens Professional version 19.04 software (Scientific Volume Imaging, The Netherlands, http://svi.nl), employing the CMLE algorithm, with a maximum of 40 iterations.
DMDAPatA treatment of zebrafish embryos
To block translation of mRNAs, 8 hpf zebrafish embryos were incubated in Danieau’s medium containing 10 μM DMDAPatA, a derivative of Pateamine A (synthesized by the lab of Daniel Romo, Baylor University) or DMSO for 2 h at 28°C. Embryos were then fixed in 4% PFA overnight at 4°C under slow agitation. Embryos were subsequently rinsed in 1x PBT, dechorionated and transferred to and stored in 100% methanol until subjected to RNAscope and immunostaining procedures. For the evaluation of drug efficiency, embryos were injected with 50 ng/μl of gfp-globin 3’UTR mRNA at the 1-cell stage and incubated with DMDAPatA or DMSO from the 16-cell stage until 5 hpf. Global expression of GFP was then compared between drug- and DMSO-treated embryos. Since DMDAPatA treatment stops embryonic development, the incubation period of these embryos was timed by monitoring the development of DMSO-treated control embryos, as described previously59.
QUANTIFICATION AND STATISTICAL ANALYSIS
Evaluation of somatic cell shape acquisition by germ cells
To analyze the effect of Nanos3 depletion on the transfating potential of germ cells, medusaNYO45 embryos that lack expression of the guidance cue Cxcl12a were employed. In such embryos germ cells are localized at ectopic locations. Control or Nos Morpholino were injected at the 1-cell stage into these embryos, and the shape of germ cells was assessed at 24 hpf based on EGFP-F’ expression. To have a clear definition of a somatic cell shape, germ cells were compared to three common somatic cell types (muscle, notochord or neuronal cells) and counted as acquiring a somatic cell shape only if their outline matched with either of these cell types. Subsequently, the ratio between the number of germ cells with somatic cell shape and that with germ cell shape was calculated. For the images shown in Fig. 1A, Cxcl12a depletion was done by Morpholino injection.
3D layer analysis and volume measurement of germ granules
The analysis of mRNA and protein distribution within defined regions of the germ granule was conducted by employing a 3D image segmentation and processing workflow using the Software Imaris (Imaris 8.4.2, Oxford Instruments). The following steps were carried out to divide the germ granule volume into 3D layers (see also Fig. S1B): Germ granules imaged in the form of confocal stacks with a z-distance of 300 nm were pre-selected based on a minimum radius length of 750 nm and a clearly visible mRNA accumulation. A 3D segmentation of the granule was conducted based on an automated thresholding of the Vasa protein signal. Subsequently, the segmented granule was expanded or shrunk to several sizes using the ‘Increase Surfaces’ plugin (Imaris XTension, Dominic Filion). For this purpose, the ellipsoid minor axis length value was increased by 25% and then divided by five to obtain the value required for the plugin to generate five surfaces with an equal distance of the borders along the granule axis. Subtracting the masked molecule signal of each of the created surfaces created layers with equal thickness. Importantly, the outermost layer covers the space directly surrounding the granule, thus considering molecules that are associated with the outer granule surface. After evaluating the integrity of the layers, the mean molecule signal intensities were extracted for each layer and the share of signal relative to the total amount in the granule was calculated and presented in percentage. Finally, the germ granule size was determined by extracting the volume computed from the segmented Vasa protein signal. For the comparison of mRNA localization among different germ granule sizes, the ratio of the mean mRNA intensity in the peripheral layer to that in the core was determined. The dataset was divided into three subgroups according to the radius of the germ granules.
3D co-localization analysis among germ granule components
For the co-localization analysis of different germ granule components, the Coloc module of the software Imaris was used. To ensure an unbiased and reliable analysis, the following steps were implemented: a surface was created based on the signal of one of the two granule components (e.g., nanos3 mRNA) and the automatically computed surface threshold value was extracted. The same process was repeated for the second granule component (e.g., tdrd7 mRNA). The two extracted values were then multiplied by a factor of 1.7 (determined in pre-trials) and served as a minimum intensity threshold for co-localization. Values extracted from the co-localization analysis where the Pearson’s Coefficient (PC) (graphs in Fig. S2B) and the amount of material colocalized above threshold (region of overlap in the graph in Fig. S1C).
Measurement of the periphery-core ratio of mRNA within germ granules
To quantify the relative distribution of mRNA in germ granules as shown in Fig. 2G, Fig. 4E,G and Fig. 5D, a ratio of the relative mRNA abundance between the periphery of the granule and the core region was determined. For this purpose, the following analysis was performed using the Fiji software (National Institutes of Health, version 2.0.0-rc-43/1.51a): A binary image of the Vasa granule marker channel was created using an auto intensity threshold, followed by a duplication of the image. One of the duplicated versions was dilated (3 iterations) to upscale the granule area, the other version is eroded (5 iterations) to downscale it. A region of interest (ROI) was defined around the dilated signal and inversed. This inversed ROI was combined with the ROI of the eroded signal. The combined ROI was inversed again, resulting in a layer-like space covering the periphery of the granule. Another ROI was created around the eroded signal to cover the core region. Finally, using the two ROIs, the ratio of the mean mRNA intensity in the periphery to that in the core of the granule was determined.
Evaluation of absolute RNA and protein signal intensities within germ cells
RNA and protein signal intensity measurements shown in Fig. 4A, Fig. 5A,B and Fig. S4D were conducted with the Fiji software. For the measurement of mRNA intensities within germ granules, a cross section of a granule was segmented based on the Vasa protein signal as described above. The segmented granule outline was dilated with 3 iterations to upscale the granule area and thereby incorporate all peripheral mRNA signal. The dilated outline was used as a ROI to measure the mean intensity of the mRNA in this region. To measure mRNA intensities in the whole germ cell volume, cells were segmented across a 3D image stack of 1 μm-distanced slices based on a membrane marker signal that had been pre-processed with a median filter (2 x 10) for noise reduction. The resulting outlines were used as a ROI to extract the mean intensity of mRNA signal in the germ cell. Nanos3 protein levels within germ cells were measured employing the same approach, but focusing on a single cross section of the cell that covered the area of nucleus and germ granules.
Statistical analysis
The Prism software (version 9.4.0, Graphpad) was used to perform statistical analysis. Three biological replicates were conducted for each experiment, and the data pooled into one sample. Significance was evaluated using the Mann-Whitney U test for non-parametric data sets. Embryos and cells were analyzed with non-biased approaches.
ACKNOWLEDGMENTS
We thank Celeste Brennecka for editing and Michal Reichman-Fried for critically revising the manuscript. We thank Ursula Jordan, Esther Messerschmidt and Ines Sandbote for technical assistance. This work was supported by funding from the University of Münster (KJW, KT, ER, AG, TGT), the German Research Foundation grant CRU 326 (P2) RA863/12-2 (ER), Baylor University (KH, DR) and the National Institutes of Health grant R35 GM 134910 (KH, DR).