Multi-scale imaging and analysis identify pan-embryo cell dynamics of germlayer formation in zebrafish

The coordination of cell movements across spatio-temporal scales ensures precise positioning of organs during vertebrate gastrulation. Mechanisms governing such morphogenetic movements have been studied only within a local region, a single germlayer or in whole embryos without cell identity. Scale-bridging imaging and automated analysis of cell dynamics are needed for a deeper understanding of tissue formation during gastrulation. Here, we report pan-embryo analyses of formation and dynamics of all three germlayers simultaneously within a developing zebrafish embryo. We show that a distinct distribution of cells in each germlayer is established during early gastrulation via cell movement characteristics that are predominantly determined by their position in the embryo. The differences in initial germlayer distributions are subsequently amplified by a global movement, which organizes the organ precursors along the embryonic body axis, giving rise to the blueprint of organ formation. The tools and data are available as a resource for the community.

). Despite an appreciation of their spatio-temporal overlap and remarkable coordination to accomplish gastrulation, each movement and germlayer have so far only been studied separately within a local region or a temporal window of interest [2][3][4][5] . Hence regardless of the advances in in toto imaging of live embryos 6 , we still fail to comprehend how such a multi-scale process operates as a whole 7 .
Several transgenics and mutants 8,9 have been used to study the formation of germlayers on the basis of cell behavior, signaling and physical properties 1,10-12 . There is strong evidence that specific cell properties (differential adhesion and surface tension) 10,11 and behaviors (random vs. directed cell movement, division and intercalation) 4,13-16 drive the formation of individual germlayers. However given the invasive nature of some of the approaches used, visualizing the interplay of all germlayers within an intact developing embryo has remained out of reach. Therefore, it can only be speculated how dynamics within individual germlayers are coordinated across tissues to drive embryo formation. We addressed this by establishing a workflow for non-invasive simultaneous imaging of all three germlayers. Through analyses of embryo-wide gene expression domains and cell movements, we show that a spatially distinct distribution of cells is established within each germlayer during early gastrulation, which is then amplified by a global morphogenetic movement to position organ progenitors along the embryonic body axis.
Early blastoderm cells are equivalent in their genetic composition and expression, so it has been difficult to generate germlayer-specific transgenic labels. To identify all three germlayers, we co-expressed three fluorescent reporters in single zebrafish embryos: Tg(sox17:H2B-tBFP) (endodermal marker), Tg(mezzo:eGFP) 17 (pan-mesendodermal marker) and Tg(h2afva:h2afva-mCherry) 18 (ubiquitous nuclear marker). A custom 4-lens Selective Plane Illumination Microscopy (SPIM) setup 19 was used to perform in toto imaging of the triply labeled embryos with high acquisition speed (~30s per time-point). The amount of data was reduced by acquiring only a 300 µm thick spherical shell around the embryo surface ( fig. S2).
The raw data from all three channels were then computationally separated into three germlayers, depending on the presence of specific reporters-prospective ectoderm (includes the EVL and YSL cells) During early gastrulation (4-7 hpf), a subpopulation of mezzo expressing cells involuted at 5.5 hpf forming a multi-layered mesendoderm. Subsequently, they moved towards the animal pole, sliding along the outer ectodermal cells undergoing epiboly 2 ( Fig. 2A,B). All three germlayers continued their epiboly movement towards the vegetal pole to spread over the yolk (Fig. 2B). Though endoderm cells were specified around 6.5 hpf, they retained their salt and pepper distribution with mesoderm cells as previously reported 21  Several attempts have been made to find an order in this complex process, assuming germlayer specific cell behaviors: the endoderm has been shown to perform random walk after involution followed by directed dorsal-ward movement 4 , the ectoderm is thought to perform cell intercalation and collective cell movement 1,14,15 , whereas the mesoderm has been shown to exhibit a variety of behaviors along the dorso-ventral axis to pattern its motion [12][13][14]16 . To assess if the characteristic of cell movement (random vs. directed) was indeed germlayer specific, we measured how straight cells in each germlayer moved and correlated the straightness indices (SI) of cell trajectories to each of r (radius), (latitude) and (longitude) coordinates (methods). We found a striking pattern: both mesendoderm and p. ectoderm cell tracks closer to the embryo surface (large r) exhibited a higher SI as compared to those at deeper locations (small r), revealing a radial organization (Fig. 2F-H). Likewise, p. ectoderm cell tracks close to the animal pole ( < -/4) showed a lower SI than cells closer to the margin ( /4 < < -/4), whereas no obvious pattern was observed along (Fig. 2H,I; fig. S5), indicating that the ectoderm cell movement is also organized along the animal-vegetal axis of the embryo. This global perspective of single cell trajectories uncovered that irrespective of their germlayer identity, the straightness of cell movement strongly correlates with the cell's position within the embryo (Fig. 2G,I), implying an emergent position dependent, rather than a germlayer specific, pattern in cell movement that drives local and global cell reorganization during early gastrulation.
While the movement characteristics of cells appeared to be independent of germlayer identity, we found a unique distribution of cells in each germlayer at the end of early gastrulation, depicted by cell density rendering showing medio-posteriorly located mesendoderm and anteriorly located p. ectoderm cells (Fig.   3A). This suggested that the pattern of initial fate specification is a crucial event in establishing the distribution of cells within germlayers. However, it still remained to be understood if germlayer dynamics are regulated individually or a in a global fashion. To investigate this, we decomposed trajectories 22 of every cell into an epiboly component (movement from animal to vegetal pole) and a convergence component (movement from ventral to dorsal side of the embryo) throughout development. In a spherical coordinate system of a suitably oriented embryo, convergence corresponds to motion along the parallels, whereas epiboly and extension correspond to directed motion along the meridians before and after completion of epiboly, respectively (Fig. 3E',E"; fig. S6). The radial movement of cells, which does not contribute directly to either of the movement components, was ignored in this analysis. Our results show that while the p. ectoderm undergoes a peak of epiboly followed by a peak of convergence in its movement, the mesendoderm has constant epiboly and convergence components throughout the process (Fig. 3B,C; fig. S7). Therefore we hypothesized that the movements of ectoderm and mesendoderm cells are decoupled during late gastrulation as formerly postulated 1,8,23 and that each germlayer might possess a characteristic motion pattern as demonstrated for the endoderm in our previous study 19 .
In order to identify these germlayer specific motion patterns, we merged cell-tracking data from multiple embryos and obtained visualizations of cellular flows 24 for each germlayer individually (methods). To our surprise, they appeared strikingly similar, with a high correlation between the local flow directions of the mesendoderm and p. ectoderm throughout the second phase of gastrulation (Fig. 3D). Hence, in contrast to our hypothesis, we show that a single unifying motion drives the dynamics of all three germlayers ( Fig.   3D; fig. S8). Distinct patterns emerged when the streamlines were scaled to reflect local cell densities of respective germlayers, indicating that the resultant organization of cells in each germlayer is dependent on the cell density distribution (Fig. 3E). Further, we found a separation of flows along the left-right axis between the p. ectoderm and mesendoderm domains, which was strongest around 9 hpf (Fig. 3E"). This boundary explains the prominent convergence of the p. ectoderm towards the anterior and that of mesendoderm towards the medio-posterior region of the embryonic axis, as already apparent in Fig. 1   (Fig. 3E). Taken 19 Further, our study illustrates the power of combining genetic tools, high-speed light sheet microscopy and rigorous data analysis to understand how a process like gastrulation is orchestrated across spatiotemporal scales. Being able to visualize and track cell cohorts from their inception to incorporation into organs will benefit developmental and disease-oriented studies 25 . Conventionally, such questions have been addressed through fate mapping experiments using sparse labeling that report spatial position of cells 26 . Here, we demonstrate that our data contain not only the spatial positions but also the temporal information about large-scale cell movements that assemble specific organ precursors to form organs.
We believe that such a singular dataset offers a crucial resource for the community to reconstruct the emergence and interplay of specific tissues and organs during early zebrafish development.