Timothy Fulton, Vikas Trivedi, Andrea Attardi, Kerim Anlas, Chaitanya Dingare, Alfonso Martinez Arias, Benjamin Steventon
Results
]. These gradients establish cell fates in space that in turn lead to the population-specific cell behaviors that dictate the complex cell and tissue rearrangement of gastrulation and axial elongation. In zebrafish, opposing Nodal and BMP signaling gradients are thought to be necessary and sufficient for the establishment of the body plan as shown by experiments in which deployment of such gradients in animal caps leads to the formation of a complete anterior-posterior (AP) axis [
]. In addition to controlling cell fate assignments, a recent study has demonstrated that Nodal signaling is a key driver of convergence and extension movements and is sufficient to generate these behaviors when expressed within zebrafish animal caps [
]. Furthermore, BMP levels have been shown to be important for controlling cell movements during both gastrulation [
] and posterior body elongation [
]. These observations raise the possibility that opposing BMP and Nodal signaling gradients are upstream of both morphogenesis and patterning. However, the causal relationships between these processes are difficult to untangle in situations where continuous external signaling sources are present either from the extra-embryonic signals present during early development or from overexpression experiments. To follow how cells can develop and pattern in the absence of external signals, we used primary culture of cells from zebrafish embryos prior to the formation of the zebrafish extra-embryonic yolk syncytial layer (YSL) that releases signals important for mesendodermal induction [
,
,
,
] and regulation of epiboly [
]. By taking cells at the 512-cell stage and earlier, they are separated from any extra-embryonic signaling source prior to the activation of the zygotic genome at the midblastula transition (Figure 1A). These explants rapidly aggregate and round up and can therefore reveal the sufficiency of early embryonic cells to generate germ layer specification in the context of altered geometry and continued signals from the yolk and YSL.

Figure 1Axial Patterning Can Occur in the Absence of Yolk
(A–K) Explants of early embryonic blastomeres taken at the (A) 256-cell stage demonstrate elongation and mesendodermal induction visualized through expression of a (B–F) Tbx16::GFP reporter (n = 4/8; heterozygous in-cross) and (G–K) tbxta mRNA (n = 1 hpc 4/4; 2 hpc 5/6; 3 hpc 5/6; 5 hpc 6/8; 7 hpc 6/6).
(L and M) At the opposite pole to tbxta expression at 7 hpc, (L) bmp4 expression is observed at the opposing end of the aggregate to tbxta expression with (M) sox32 expression found in the center of some aggregates, but those without sox32 expression are able to elongate (n = 3/8; sox32 positive explants/assayed elongating explants).
(N) In the high bmp4 domain at 7 hpc, gata4 expression is observed (n = 8/9), with protocadherin8 (n = 8/9) expression more posteriorly and noto furthest posterior from the bmp4 expression domain (n = 7/9).
(O) Germ cell markers (O and O’) nanos and vasa are also observed coexpressed in cells within the core of the aggregates in the sox32-positive domain (n = 5/5) at 7 hpc.
] revealed mesoderm specification within the elongating end of the aggregate (Figures 1B–1F; Video S1), accompanied by polarized expression of tbxta (Figures 1G–1K). These results showed how symmetry breaking and mesoderm patterning can occur within explants of embryonic cells separated from the yolk.
-
Video S1. Bright-Field Time Lapse of an Elongating Explant Starting at 1 hpc Followed by an Explant Containing a Tbx16::GFP Reporter Starting at 4 hpc, Related to Figure 1
Both imaged on a widefield microscope at 10X, one frame per 10 minutes.
- Turner D.A.
- Girgin M.
- Alonso-Crisostomo L.
- Trivedi V.
- Baillie-Johnson P.
- Glodowski C.R.
- Hayward P.C.
- Collignon J.
- Gustavsen C.
- Serup P.
- et al.
,
], we refer to explants of early embryonic zebrafish cells as “pescoids.”
-
Video S2. Confocal Stacks Demonstrating the Spatial Organization of the Three Germ Layers, tbxta Mesoderm, sox2 Ectoderm, and sox17 Endoderm, within Explants, Related to Figure 1
Surface renderings of tbxta and sox2 demonstrate the spatial organistion of the ectoderm and mesoderm. This is followed by confocal Stacks and surface renderings demonstrating the spatial organization of expression of the endoderm marker sox32. Expression of the germ cell marker nanos, within the sox32 positive domain is also demonstrated. Lastly confocal stacks demonstrating absence of HCR signal, and hence expression of Otx2. Otx2 positive control embryos also displayed.
,
,
] and might be important for setting up an initial symmetry-breaking event that leads to the emergence of the opposing signaling gradients described above. To test whether polarized tbxta expression is robust to the disruption of this early positional information, we dissociated explanted cells and reaggregated them to determine whether polarization is observed upon removal of any pre-existing asymmetry in the reaggregates (Figures 2A and 2B ). In many cases, reaggregation was not complete, leading to the formation of smaller pescoids. However, in 7/10 of these reduced-sized pescoids, a polarized tip of tbxta expression was still observed (Figures 2C and 2C’). When most cells were reaggregated, a clear elongated morphology was observed together with tbxta polarization (Figures 2D and 2D’) as in non-dissociated pescoids (Figure 1L).

Figure 2Lineage and Spatial Pre-patterns Are Lost due to Extensive Cell Divisions and Cell Mixing
(A–D’) Disassociation (A) and reaggregation (B) of explanted cells results in tbxta expression (n = 8/8; expressed/total imaged; C and C’) and, infrequently, elongation of the aggregate (n = 2 observed; D and D’). Cells in full embryonic explants undergo rapid cell divisions as seen in images acquired on SPIM.
(E–F’) Number of cells in pescoids (E) counted based on image segmentation (black curve), as seen in images acquired on SPIM (F and F’). Dashed curves are estimates of the number of cells, starting from the number of spots segmented at t = 0, if all cells divided synchronously every 20 min (blue line) or only a random sub-population (<50%) of cells divided every 20 min (orange line).
(F and F’) Cells in pescoid divide randomly with no preference for direction of division, leading to mixing of cells.
(G–I’) Injecting embryos with fluorescent high-molecular-weight dextran at the 64-cell stage, (H–I) labeling marginal cells prior to making pescoids demonstrates that these labeled cells spread across the entire pescoid within (H’–I’) 5 h of explanting. These cells also show a high level of intermixing of labeled and unlabeled cells. n = 10, all demonstrating cell mixing. (G–I) Images shown as maximum projections. (I’) Shown as central 2-μm slice.
(J–L’) This is also shown in pescoids injected at the one-cell stage with Kikume mRNA and then a small population of cells labeled by photo conversion at 1 hpc. Explants were reimaged at 3 hpc to assay for label mixing (n = 6 replicates, all demonstrating cell mixing). (K–K’) Images are shown as maximum projections at 1 and 3 hpc. (L–L’) spot detection of labelled and unlabelled cells in 3D rendering of (K–K’) demonstating cell mixing at 1 and 3 hpc.
(N–Q’) HCR staining of animal cap explants and pescoids at 5 hpc reveals a similar expression of eomesodermin (pescoids 4/4; animal caps 6/6), mxtx2 (pescoids 3/5; animal caps 4/5), tbxta (pescoids 5/6; animal caps 3/4), and goosecoid (pescoids 3/6; animal caps 2/4). In (N)–(Q), n = gene expression observed/total imaged.
(R) This finding is further supported by (R and R’) clear expression of a Tbx16::GFP reporter in both animal caps (4/6; expression/imaged from heterozygous in-cross) and full pescoid explant at 7 hpc (2/4; expression/imaged from heterozygous in cross).
-
Video S3. Ubiquitous H2B-GFP-Labeled Explant Imaged on a SPIM with a Time Resolution of 2.5 min, Related to Figure 2
] (Video S4). Furthermore, blocking Nodal receptor activity between 1 and 3 hpc with SB505124 inhibits pescoid elongation (Figures 3G and 3I), compared to controls (Figures 3F and 3I). The requirement for Nodal activity is increased at early stages as later treatments between 5 and 7 hpc have a lesser effect on elongation (Figure 3H). This importance of Nodal signaling in driving explant elongation is in line with recent work showing that Nodal signaling is sufficient to drive the elongation of animal caps [
] and an inheritance of Nodal activity from the germ ring in blastoderm explants similar to those presented here [
].

Figure 3Nodal Signaling Is Upstream of PCP-Driven Convergence and Extension, which Drives Elongation
(A–C) The first signaling event that polarizes to a single point within the pescoid is that of Nodal signaling, demonstrated through (A–C) polarization of phospho-Smad 2/3 activity. This is shown in composite color images and as pSmad2/3 signal inverted images (A’–C’; 2 hpc n = 4/8; 3 hpc n = 5/8; 5 hpc n = 4/6; total with polarized signal/total imaged).
(D and E) We also observe polarized (D) ndr1 and (E) ndr2 expression in the elongation at 5hpc (ndr1 n = 4/6; ndr2 n = 4/6; expression observed/total imaged; total with polarized signal/total imaged).
(F–I) Treatment with the Nodal inhibitor SB-505124 between (G) 1 and 3 hpc inhibits elongation of the explants and (H) to a lesser extent when applied between 5 and 7 hpc when compared to controls at 7 hpc (F and I).
(J) The PCP components wnt11f2 and fzd2 are observed expressed in the elongating end of the pescoid with the spatial organization of these domains reflected in (J’) surface renderings of the HCR signal.
(K and K’) Inhibition of Nodal signaling between 1 and 3 hpc results in loss of wnt11f2 and frz2 expression.
(L–O) Inhibition of convergence and extension movements using (M) blebbistatin or (N) injection at the one-cell stage of a dominant-negative dishevelled construct blocks elongation when compared to controls (L and O), further supporting that elongation is caused by convergence and extension movements.
-
Video S4. Bright-Field and Fluorescence Time Lapse of an Explant Containing an Activin Response Element Fusion to GFP Followed by an Explant Containing a 7xTCF-Xla.Sia:GFP Reporter, Followed by a BMP Response Element Fusion to RFP, Imaged at 10×, One Frame per 10 min, Related to Figures 3 and 4
], we next sought to determine whether this is acting downstream of Nodal signaling in driving pescoid elongation. In the control situation, both the PCP ligand, Wnt11f2, and its receptor, Frizzled2, are expressed in the elongating tip (Figures 3J and 3J’). This expression is reduced upon SB-505124 treatment (Figures 3K and 3K’). Pescoid elongation is not associated with polarized proliferation, suggesting that polarized growth is not a major driver of elongation in pescoids (Figure S2C). Inhibition of the non-muscle myosin, Myosin II, with the inhibitor blebbistatin is sufficient to block elongation, suggesting that dynamic actomyosin contractions are important for this process. To assess whether this is controlled downstream of the PCP pathway, we specifically inhibited the PCP pathway using a dominant negative version of dishevelled (Dsh-DEP+) that in Xenopus and zebrafish inhibits axial elongation without perturbing Wnt/β-catenin activity [
]. This reduced pescoid elongation (Figures 3N and 3O), in a similar manner to that of both Nodal inhibition and blebbistatin treatments (Figure 3M). Taken together, this suggests to us that Nodal is an upstream mediator of pescoid elongation through the control of Wnt11f2 and Frizzled expression.
,
,
,
], yet how this temporal modulation of these signaling pathways is achieved is unknown. One possibility could be that gastrulation movements are themselves important for regulating signal exposure, by spatially separating cells expressing signals and their secreted inhibitors. Our ability to observe the progressive morphogenesis of pescoids along one primary axis of elongation offers a unique opportunity to investigate this hypothesis.
] for both bmp7 expression and GFP mRNA at successive stages post-culture (Figures 4A–4F). Both bmp7 and tcf.:gfp are uniformly expressed at low levels prior to explant elongation (Figures 4A, 4B, and 4D) but then generate opposing levels of activity at 7 hpc (Figures 4C and 4F). This correlation between elongation and the signal activity polarization can also be observed in time-lapse movies of both the TCF:GFP (Video S4) and of a BMP-responsive element driving RFP expression [
] (Video S4). This progressive polarization of Wnt and BMP activity poles also occurs with the progressive addition of bands of krox20 expression, a marker of rhombomeres 3 and 5 [
] (Figures 4G–4I). Based on these results, we propose that pescoid elongation is important for the temporal control of BMP and Wnt exposure and for controlling the onset of krox20 expression. To test this, we blocked elongation using either blebbistatin treatment or the injection of DEP+ and assayed the effect on bmp7, chordin, tcf.:gfp, and krox20 expression (Figures 4J–4X). Neither treatment effects the level of expression of either TCF activity (Figures 4O, 4T, and 4U) or the expression of BMP pathway components (Figures 4P, 4Q, 4U, and 4V). Instead, the tcf.:gfp and bmp7 expression domains are not as spatially separated as they are in the control situation (Figures 4J and 4K), meaning that a region of low tcf.:gfp and low bmp7 expression is no longer observed (Figures 4S and 4X). This results in a reduction of krox20 expression (Figures 4R, 4W, and 4X) compared with controls (Figure 4M), with treated pescoids showing only one stripe of expression or less (Figures 4S’ and 4X’). These results demonstrate that elongation is an important regulator of pattern formation through the spatial and temporal regulation of BMP and Wnt signal activity.

Figure 4PCP-Dependent Elongation Is Required for Regulating Exposure to BMP and Wnt Activity
(A–F) Expression of bmp7 (A–C) and TCF::GFP (D–F) as a time course at 2 hpc, 5 hpc, and 7 hpc reveals both signaling domains are spread across the explant evenly at early time points and are restricted to either end of the explant by 7 hpc. TCF::GFP assayed by HCR against GFP mRNA for immediate reporter activity readout is shown (2 hpc bmp7 2/3, tcf.:gfp 4/6; 5 hpc bmp7 5/5, tcf.:gfp 10/10; 7 hpc bmp7 4/4, tcf.:gfp 4/5; expression observed/total imaged).
(G–M, O–R, and T–W) The time course reveals the isolation of the BMP and Wnt domains to either end of the explant allows expression of hindbrain marker (G–I) krox20 in a characteristic two-stripe pattern. (G) Initially no expression is observed (n = 0/6; expression/total imaged), followed by (H) expression of a single stripe (n = 7/8; expression/total imaged) at 7 hpc and then (I) two stripes by 10 hpc (n = 5/8 expression/total imaged). Inhibition of convergence and extension using dominant-negative dishevelled injected at the one-cell stage (O–R) and treatment with 2.5 μM blebbistatin (T–W) reveals that the Wnt/TCF and BMP domains do not separate as observed in the (J–M) control.
(N, S, and X) Description of these profile quantitatively through normalization of the long axis of the explant and normalization of signal intensity between 0 and 1 (n = minimum 7 per condition; line represents mean). The lack of low BMP moderate Wnt/TCF domain can be observed in the central region of the explants compared to control. (N) displays a control profile, (S) displays the profile of a DEP+ explant, and (X) displays a profile of an explant treated with Blebbistatin.
]. Whether the absence of forebrain specification in pescoids is due to the lack of additional extra-embryonic signals or due to the fact that additional morphogenetic events are required to separate organizer-derived signals is an open question. The requirement for a precise temporal modulation of BMP and Wnt activity during the specification and patterning of the ectoderm is well known [
,
,
,
,
,
]. Here, we provide evidence for a role of global tissue morphogenesis in regulating the exposure to ligands and secreted inhibitors of these pathways, potentially providing a mechanism by which patterning and morphogenesis is coordinated during gastrulation.
Discussion
,
,
], an early dorsal pole of β-catenin activity [
,
,
,
,
], and the release of Nodal signals from the yolk syncytial layer (YSL) [
,
,
,
]. That explanted cells from teleost embryos can break morphological symmetry and elongate has been known for some time [
,
,
]. These findings have been further confirmed recently by showing that whole-embryo explants in zebrafish also require the polarized inheritance of maternal factors from the dorsal marginal zone [
]. Here, we extend these findings to show that mesoderm specification and explant elongation can occur even when the early positional information established by these early signals is disrupted through forced dissociation and reaggregation or through the cell mixing that is occurring normally in the explants. In our hands, elongation was not observed when the cells were centrifuged following dissociation, as also reported in a similar experiment where dissociated animal cap cells were treated with activin [
] or in whole-embryo zebrafish explants [
]. This suggests that centrifugation might disrupt some component of cytoskeletal structure important for later convergence and extension movements, as suggested previously [
]. We observe a difference in the extent of mixing of labeled cells between 1 and 3 hpc (Figure 2J) compared to a similar labeling experiment of Schauer et al. [
]. However, marginal zone cells are clearly dispersed across the pescoids by 6 hpc (Figure 2G). Therefore, the cells that are local to the early polarized region of Nodal activity observed between 1 and 3 hpc will be distributed across the pescoid by the time elongation and germ layer patterning can be seen at 7 hpc (Figures 2L–2N). Despite these continual cell rearrangements, a polarized Wnt/Nodal activity can be maintained at the elongating pole of the pescoid. How this region of polarized signaling activity is maintained as cells continually move in and out of that domain is unknown. Understanding this process will be of importance as it results in the localized expression of multiple markers of the gastrula organizer, drives embryo elongation, and thereby mediates anterior-posterior patterning of the early nervous system. The phenomena of robust organizer gene expression in the context of cell movement has also been observed in chick embryos during organizer formation [
] and is likely to be a general attribute of the gastrula-stage organizer [
]. A complete understanding of this highly dynamic cellular and molecular process will require models that specifically incorporate morphogenesis in the investigation of axial patterning during early development.
]. How the diffusion of Ndr1 and Ndr2 and their inhibitor Lefty interact in the context of extensive cell mixing will be essential to obtain a complete picture of how axis specification occurs in both whole embryos and explants. During normal development, it is likely that early cell rearrangements and YSL signal release act as a precise balance to ensure that embryo patterning is both robust to alterations in external environment and the initial conditions of the fertilized egg. In the context of early embryonic explants, it is clear that the size of the explanted tissue is an important factor in allowing the tissue to polarize and elongate. Animal caps refer to explants taken of approximately 50% of the animal-most portion of the early embryo and have previously been shown to be unable to elongate in vitro [
,
]. Here, we observe the same result with either small- or medium-sized animal explants failing to elongate (Figure S3C). In our hands, we do see some differences when animal cap explants are taken and cultured, as the expression of gsc, tbxta, and tbx16 (Figure 2M) has not been observed when analyzed by colorimetric in situ [
]. It is possible that these differences may lie in the media used to culture animal caps, though in both cases, explants were cultured in the presence of serum that may contain additional growth factors (either 3% fetal bovine serum [here] or newborn calf serum were added) [
]. Despite these apparent differences in mesoderm specification, our results show an additional size dependency for explant elongation, as this requires either all embryonic cells to be explanted, for multiple small animal caps to be combined, or for large regions that lack either the vegetal or animal-most pole to be taken. Importantly, this size dependency for elongation can be rescued via the overexpression of Nodal activity, in PCP-dependent manner, demonstrating that these signals act together to drive convergence and extension movements [
].
- Turner D.A.
- Girgin M.
- Alonso-Crisostomo L.
- Trivedi V.
- Baillie-Johnson P.
- Glodowski C.R.
- Hayward P.C.
- Collignon J.
- Gustavsen C.
- Serup P.
- et al.
,
]. Such behaviors resemble those that we have described here. Similar emergence of embryonic pattern has been observed in dissociated and reaggregated cells from other metazoan species, such as hydra [
], Xenopus [
,
], and occurs naturally in Killifish [
]. Investigating how each of these examples differ in the precise mechanisms of axial patterning is likely to reveal further insight into how morphogenesis, morphogens, and gene-regulatory networks interact to generate pattern during complex morphogenesis.
Acknowledgments
We would like to thank Carolina Monck and Rohan Sanghera for helping with some experiments and the quantification of pescoid elongation, respectively. Many thanks to Caroline Hill for sharing the Nodal and BMP reporter lines, Michael Lardelli and Simon Wells for sharing the Tbx16::GFP reporter line, and the Steven Wilson lab for sharing the Tg(7xTCF-Xla.Sia:GFP) reporter zebrafish. We also thank Gopi Shah at the Mesoscopic Imaging Facility at EMBL Barcelona for help with SPIM imaging. V.T. was supported by a Herchel Smith Postdoctoral Fellowship, University of Cambridge ; a John Henry Coates Fellowship, Emmanuel College, Cambridge ; and by the European Molecular Biology Laboratory (EMBL) Barcelona . B.S. and T.F. are supported by a Henry Dale Fellowship jointly funded by the Wellcome Trust and the Royal Society ( 109408/Z/15/Z ) and T.F. by a scholarship from the Cambridge Trust . A.A. was supported by the Erasmus+ Traineeship scheme of the European Commission . K.A. was supported by the European Molecular Biology Laboratory (EMBL) Barcelona . A.M.A. was supported by the Biotechnology and Biological Sciences Research Council (BBSRC ) ( BB/M023370/1 ).
Author Contributions
Conceptualization, V.T. and B.S.; Funding Acquisition, V.T., A.M.A., and B.S.; Investigation, T.F., V.T., A.A., K.A., and C.D.; Methodology, V.T., A.A., B.S., and T.F.; Project Administration, V.T. and B.S.; Resources, B.S.; Supervision, V.T. and B.S.; Validation, T.F. and K.A.; Visualization, V.T. and T.F.; Writing – Original Draft, V.T. and B.S.; Writing – Review and Editing, A.M.A., T.F., V.T., and B.S.
Declaration of Interests
The authors declare no competing interests.