Search

Gastrulation is the fundamental process in all multicellular animals through which the basic body plan is first laid down1–4. It is pivotal in generating cellular diversity coordinated with spatial patterning. In humans, gastrulation occurs in the third week after fertilization. Our understanding of this process in humans is relatively limited and based primarily on historical specimens5–8, experimental models9–12 or, more recently, in vitro cultured samples13–16. Here we characterize in a spatially resolved manner the single-cell transcriptional profile of an entire gastrulating human embryo, staged to be between 16 and 19 days after fertilization. We use these data to analyse the cell types present and to make comparisons with other model systems. In addition to pluripotent epiblast, we identified primordial germ cells, red blood cells and various mesodermal and endodermal cell types. This dataset offers a unique glimpse into a central but inaccessible stage of our development. This characterization provides new context for interpreting experiments in other model systems and represents a valuable resource for guiding directed differentiation of human cells in vitro. The single-cell transcriptional profile of a human embryo between 16 and 19 days after fertilization reveals parallels and differences in gastrulation in humans as compared with mouse and non-human primate models.

The body plan of the mammalian embryo is shaped through the process of gastrulation, an early developmental event that transforms an isotropic group of cells into an ensemble of tissues that is ordered with reference to three orthogonal axes1. Although model organisms have provided much insight into this process, we know very little about gastrulation in humans, owing to the difficulty of obtaining embryos at such early stages of development and the ethical and technical restrictions that limit the feasibility of observing gastrulation ex vivo2. Here we show that human embryonic stem cells can be used to generate gastruloids-three-dimensional multicellular aggregates that differentiate to form derivatives of the three germ layers organized spatiotemporally, without additional extra-embryonic tissues. Human gastruloids undergo elongation along an anteroposterior axis, and we use spatial transcriptomics to show that they exhibit patterned gene expression. This includes a signature of somitogenesis that suggests that 72-h human gastruloids show some features of Carnegie-stage-9 embryos3. Our study represents an experimentally tractable model system to reveal and examine human-specific regulatory processes that occur during axial organization in early development.

Gastruloids are three-dimensional aggregates of embryonic stem cells that display key features of mammalian development after implantation, including germ-layer specification and axial organization1-3. To date, the expression pattern of only a small number of genes in gastruloids has been explored with microscopy, and the extent to which genome-wide expression patterns in gastruloids mimic those in embryos is unclear. Here we compare mouse gastruloids with mouse embryos using single-cell RNA sequencing and spatial transcriptomics. We identify various embryonic cell types that were not previously known to be present in gastruloids, and show that key regulators of somitogenesis are expressed similarly between embryos and gastruloids. Using live imaging, we show that the somitogenesis clock is active in gastruloids and has dynamics that resemble those in vivo. Because gastruloids can be grown in large quantities, we performed a small screen that revealed how reduced FGF signalling induces a short-tail phenotype in embryos. Finally, we demonstrate that embedding in Matrigel induces gastruloids to generate somites with the correct rostral-caudal patterning, which appear sequentially in an anterior-to-posterior direction over time. This study thus shows the power of gastruloids as a model system for exploring development and somitogenesis in vitro in a high-throughput manner. This work was supported by an European Research Council Advanced grant (ERC-AdG 742225-IntScOmics), a Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) TOP award (NWO-CW 714.016.001) and the Foundation for Fundamental Research on Matter, financially supported by NWO (FOM-14NOISE01) to S.C.v.d.B., A.A., V.v.B., M.B., J.V. and A.v.O., a Biotechnology and Biological Sciences Research Council (no. BB/P003184/1), Newton Trust (INT16.24b) and Medical Research Council (MR/R017190/1) grant to A.M.A., a Newnham College Cambridge Junior Research Fellowship to N.M. and a studentship from the Engineering and Physical Sciences Research Council to P.B.-J. The Cambridge Stem Cell Institute is supported by core funding from the Wellcome Trust and Medical Research Council; J.N. was funded by the University of Cambridge and K.F.S. by core funding from the Hubrecht Institute. This work is part of the Oncode Institute, which is partly financed by the Dutch Cancer Society. We thank A. Ebbing and M. Betist for the robotized tomo-seq protocol; G. Keller for the Brachyury-GFP cell line; J. Collignon for the Nodal-YFP line; K. Hadjantonakis for the TCF/LEF-mCherry line; S. van den Brink and E. R. Maandag for the E14-IB10 cells; J. Kress and A. Aulehla for the LfngT2AVenus mouse ES cell line; I. Misteli Guerreiro, J. Peterson-Maduro and J. Hoeksma for suggestions for in situ hybridization experiments; W. Thomas, Y. el Azhar, J. Juksar and J. Beumer for reagents and inhibitors; A. de Graaff and A. Stokkermans for help with multiphoton microscopy and analysis of the microscopy data; D. A. Turner for microscopy panels that were used for tomo-seq validation; J. Korving for help with the somite-size measurements in embryos; the Hubrecht FACS facility and R. van der Linden for FACS experiments; Single Cell Discoveries for 10x Genomics scRNA-seq; the Utrecht Sequencing facility for sequencing; and P. Zeller, H. Viñas Gaza, M. Vaninsberghe, V. Bhardwaj and all members of the van Oudenaarden, Sonnen and Martinez Arias laboratories for discussions.

The body plan of the mammalian embryo is shaped through the process of gastrulation, an early developmental event that transforms an isotropic group of cells into an ensemble of tissues that is ordered with reference to three orthogonal axes

In vertebrate development, the body plan is determined by primordial morphogen gradients that suffuse the embryo. Retinoic acid (RA) is an important morphogen involved in patterning the anterior-posterior axis of structures, including the hindbrain and paraxial mesoderm. RA diffuses over long distances, and its activity is spatially restricted by synthesizing and degrading enzymes. However, gradients of endogenous morphogens in live embryos have not been directly observed; indeed, their existence, distribution and requirement for correct patterning remain controversial. Here we report a family of genetically encoded indicators for RA that we have termed GEPRAs (genetically encoded probes for RA). Using the principle of fluorescence resonance energy transfer we engineered the ligand-binding domains of RA receptors to incorporate cyan-emitting and yellow-emitting fluorescent proteins as fluorescence resonance energy transfer donor and acceptor, respectively, for the reliable detection of ambient free RA. We created three GEPRAs with different affinities for RA, enabling the quantitative measurement of physiological RA concentrations. Live imaging of zebrafish embryos at the gastrula and somitogenesis stages revealed a linear concentration gradient of endogenous RA in a two-tailed source-sink arrangement across the embryo. Modelling of the observed linear RA gradient suggests that the rate of RA diffusion exceeds the spatiotemporal dynamics of embryogenesis, resulting in stability to perturbation. Furthermore, we used GEPRAs in combination with genetic and pharmacological perturbations to resolve competing hypotheses on the structure of the RA gradient during hindbrain formation and somitogenesis. Live imaging of endogenous concentration gradients across embryonic development will allow the precise assignment of molecular mechanisms to developmental dynamics and will accelerate the application of approaches based on morphogen gradients to tissue engineering and regenerative medicine.

Summary In mammals, specification of the three major germ layers occurs during gastrulation, when cells ingressing through the primitive streak differentiate into the precursor cells of major organ systems. However, the molecular mechanisms underlying this process remain unclear, as numbers of gastrulating cells are very limited. In the E6.5 mouse embryo, cells located at the junction between the extra-embryonic region and the epiblast on the posterior side of the embryo undergo an epithelial-to-mesenchymal transition (EMT) and ingress through the primitive streak (PS). Subsequently, cells migrate, either surrounding the prospective ectoderm contributing to the embryo proper, or into the extra-embryonic region to form the yolk sac (YS), umbilical cord and placenta. Fate mapping has shown that mature tissues such as blood and heart originate from specific regions of the pre-gastrula epiblast1 but the plasticity of cells within the embryo and the function of key cell type-specific transcription factors remain unclear. Here we analyse 1,205 cells from the epiblast and nascent Flk1+ mesoderm of gastrulating mouse embryos using single cell RNA-sequencing, representing the first transcriptome-wide in vivo view of early mesoderm formation during mammalian gastrulation. Additionally, using knock-out mice, we study the function of Tal1, a key hematopoietic transcription factor (TF), and demonstrate, contrary to previous studies performed using retrospective assays2,3, that Tal1 knock out does not immediately bias precursor cells towards a cardiac fate.

During tissue morphogenesis, simple epithelial sheets undergo folding to form complex structures. The prevailing model underlying epithelial folding involves cell shape changes driven by myosin-dependent apical constriction. Here we describe an alternative mechanism that requires differential positioning of adherens junctions controlled by modulation of epithelial apical-basal polarity. Using live embryo imaging, we show that before the initiation of dorsal transverse folds during Drosophila gastrulation, adherens junctions shift basally in the initiating cells, but maintain their original subapical positioning in the neighbouring cells. Junctional positioning in the dorsal epithelium depends on the polarity proteins Bazooka and Par-1. In particular, the basal shift that occurs in the initiating cells is associated with a progressive decrease in Par-1 levels. We show that uniform reduction of the activity of Bazooka or Par-1 results in uniform apical or lateral positioning of junctions and in each case dorsal fold initiation is abolished. In addition, an increase in the Bazooka/Par-1 ratio causes formation of ectopic dorsal folds. The basal shift of junctions not only alters the apical shape of the initiating cells, but also forces the lateral membrane of the adjacent cells to bend towards the initiating cells, thereby facilitating tissue deformation. Our data thus establish a direct link between modification of epithelial polarity and initiation of epithelial folding.

The neural crest is a stem population critical for development of the vertebrate craniofacial skeleton and peripheral ganglia. Neural crest cells originate along the border between the neural plate and epidermis, migrate extensively and generate numerous derivatives, including neurons and glia of the peripheral nervous system, melanocytes, bone and cartilage of the head skeleton. Impaired neural crest development is associated with human defects, including cleft palate. Classically, the neural crest has been thought to form by interactions at the border between neural and non-neural ectoderm or mesoderm, and defined factors such as bone morphogenetic proteins (BMPs) and Wnt proteins have been postulated as neural crest-inducers. Although competence to induce crest cells declines after stage 10 (ref. 14), little is known about when neural crest induction begins in vivo. Here we report that neural crest induction is underway during gastrulation and well before proper neural plate appearance. We show that a restricted region of chick epiblast (stage 3-4) is specified to generate neural crest cells when explanted under non-inducing conditions. This region expresses the transcription factor Pax7 by stage 4 + and later contributes to neural folds and migrating neural crest. In chicken embryos, Pax7 is required for neural crest formation in vivo, because blocking its translation inhibits expression of the neural crest markers Slug, Sox9, Sox10 and HNK-1. Our results indicate that neural crest specification initiates earlier than previously assumed, independently of mesodermal and neural tissues, and that Pax7 has a crucial function during neural crest development.

Germ cell fate in mice is induced in pluripotent epiblast cells in response to signals from extraembryonic tissues. The specification of approximately 40 founder primordial germ cells and their segregation from somatic neighbours are important events in early development. We have proposed that a critical event during this specification includes repression of a somatic programme that is adopted by neighbouring cells. Here we show that Blimp1 (also known as Prdm1), a known transcriptional repressor, has a critical role in the foundation of the mouse germ cell lineage, as its disruption causes a block early in the process of primordial germ cell formation. Blimp1-deficient mutant embryos form a tight cluster of about 20 primordial germ cell-like cells, which fail to show the characteristic migration, proliferation and consistent repression of homeobox genes that normally accompany specification of primordial germ cells. Furthermore, our genetic lineage-tracing experiments indicate that the Blimp1-positive cells originating from the proximal posterior epiblast cells are indeed the lineage-restricted primordial germ cell precursors.

The human oncogene beta-catenin is a bifunctional protein with critical roles in both cell adhesion and transcriptional regulation in the Wnt pathway. Wnt/beta-catenin signalling has been implicated in developmental processes as diverse as elaboration of embryonic polarity, formation of germ layers, neural patterning, spindle orientation and gap junction communication, but the ancestral function of beta-catenin remains unclear. In many animal embryos, activation of beta-catenin signalling occurs in blastomeres that mark the site of gastrulation and endomesoderm formation, raising the possibility that asymmetric activation of beta-catenin signalling specified embryonic polarity and segregated germ layers in the common ancestor of bilaterally symmetrical animals. To test whether nuclear translocation of beta-catenin is involved in axial identity and/or germ layer formation in 'pre-bilaterians', we examined the in vivo distribution, stability and function of beta-catenin protein in embryos of the sea anemone Nematostella vectensis (Cnidaria, Anthozoa). Here we show that N. vectensis beta-catenin is differentially stabilized along the oral-aboral axis, translocated into nuclei in cells at the site of gastrulation and used to specify entoderm, indicating an evolutionarily ancient role for this protein in early pattern formation.

Now that the mouse and human genome sequences are complete, biologists need systematic approaches to determine the function of each gene1,2. A powerful way to discover gene function is to determine the consequence of mutations in living organisms. Large-scale production of mouse mutations with the point mutagen N-ethyl-N-nitrosourea (ENU) is a key strategy for analysing the human genome because mouse mutants will reveal functions unique to mammals, and many may model human diseases3. To examine genes conserved between human and mouse, we performed a recessive ENU mutagenesis screen that uses a balancer chromosome, inversion chromosome 11 (refs 4, 5). Initially identified in the fruitfly, balancer chromosomes are valuable genetic tools that allow the easy isolation of mutations on selected chromosomes6. Here we show the isolation of 230 new recessive mouse mutations, 88 of which are on chromosome 11. This genetic strategy efficiently generates and maps mutations on a single chromosome, even as mutations throughout the genome are discovered. The mutations reveal new defects in haematopoiesis, craniofacial and cardiovascular development, and fertility.

Twisted gastrulation (TSG) is involved in specifying the dorsal-most cell fate in Drosophila embryos, but its mechanism of action is poorly understood. TSG has been proposed to modify the action of Short gastrulation (SOG), thereby increasing signalling by the bone morphogenetic protein (BMP) Decapentaplegic. SOG, an inhibitor of BMP signalling, is in turn inactivated by the protease Tolloid. Here we identify Tsg gene products from human, mouse, Xenopus, zebrafish and chick. Expression patterns in mouse and Xenopus embryos are consistent with in vivo interactions between Tsg, BMPs and the vertebrate SOG orthologue, chordin. We show that Tsg binds both the vertebrate Decapentaplegic orthologue BMP4 and chordin, and that these interactions have multiple effects. Tsg increases chordin's binding of BMP4, potentiates chordin's ability to induce secondary axes in Xenopus embryos, and enhances chordin cleavage by vertebrate tolloid-related proteases at a site poorly used in Tsg's absence; also, the presence of Tsg enhances the secondary axis-inducing activity of two products of chordin cleavage. We conclude that Tsg acts as a cofactor in chordin's antagonism of BMP signalling.

During neural induction, the 'organizer' of the vertebrate embryo instructs neighbouring ectodermal cells to become nervous system rather than epidermis. This process is generally thought to occur around the mid-gastrula stage of embryogenesis. Here we report the isolation of ERNI, an early response gene to signals from the organizer (Hensen's node). Using ERNI as a marker, we present evidence that neural induction begins before gastrulation--much earlier in development than previously thought. We show that the organizer and some of its precursor cells produce a fibroblast growth factor signal, which can initiate, and is required for, neural induction.

Vertebrate gastrulation involves the specification and coordinated movement of large populations of cells that give rise to the ectodermal, mesodermal and endodermal germ layers. Although many of the genes involved in the specification of cell identity during this process have been identified, little is known of the genes that coordinate cell movement. Here we show that the zebrafish silberblick (slb) locus1 encodes Wnt11 and that Slb/Wnt11 activity is required for cells to undergo correct convergent extension movements during gastrulation. In the absence of Slb/Wnt11 function, abnormal extension of axial tissue results in cyclopia and other midline defects in the head2. The requirement for Slb/Wnt11 is cell non-autonomous, and our results indicate that the correct extension of axial tissue is at least partly dependent on medio-lateral cell intercalation in paraxial tissue. We also show that the slb phenotype is rescued by a truncated form of Dishevelled that does not signal through the canonical Wnt pathway3, suggesting that, as in flies4, Wnt signalling might mediate morphogenetic events through a divergent signal transduction cascade. Our results provide genetic and experimental evidence that Wnt activity in lateral tissues has a crucial role in driving the convergent extension movements underlying vertebrate gastrulation.

The vertebrate body plan is established during gastrulation, when cells move inwards to form the mesodermal and endodermal germ layers. Signals from a region of dorsal mesoderm, which is termed the organizer, pattern the body axis by specifying the fates of neighbouring cells. The organizer is itself induced by earlier signals. Although members of the transforming growth factor-beta (TGF-beta) and Wnt families have been implicated in the formation of the organizer, no endogenous signalling molecule is known to be required for this process. Here we report that the zebrafish squint (sqt) and cyclops (cyc) genes have essential, although partly redundant, functions in organizer development and also in the formation of mesoderm and endoderm. We show that the sqt gene encodes a member of the TGF-beta superfamily that is related to mouse nodal. cyc encodes another nodal-related proteins, which is consistent with our genetic evidence that sqt and cyc have overlapping functions. The sqt gene is expressed in a dorsal region of the blastula that includes the extraembryonic yolk syncytial layer (YSL). The YSL has been implicated as a source of signals that induce organizer development and mesendoderm formation. Misexpression of sqt RNA within the embryo or specifically in the YSL induces expanded or ectopic dorsal mesoderm. These results establish an essential role for nodal-related signals in organizer development and mesendoderm formation.

Zebrafish cyclops (cyc) mutations cause deficiencies in the dorsal mesendoderm1,2 and ventral neural tube3,4, leading to neural defects and cyclopia5,6. Here we report that cyc encodes a transforming growth factor-β (TGF-β)-related intercellular signalling molecule that is similar to mouse nodal7. cyc is expressed in dorsal mesendoderm at gastrulation and in the prechordal plate until early somitogenesis. Expression reappears transiently in the left lateral-plate mesoderm, and in an unprecedented asymmetric pattern in the left forebrain. Injection of cyc RNA non-autonomously restores sonic hedgehog -expressing cells of the ventral brain and floorplate that are absent in cyc mutants, whereas inducing activities are abolished by cycm294, a mutation of a conserved cysteine in the mature ligand. Our results indicate that cyc provides an essential non-cell-autonomous signal at gastrulation, leading to induction of the floorplate and ventral brain.

During gastrulation in vertebrates, dorsal ectoderm is induced to form neural tissue that later gives rise to the brain and spinal cord. This induction depends on signals arising from a group of cells on the dorsal side of the gastrula. This group of cells constitutes the organizer. It is thought that the organizer initially induces neural tissue with anterior, or forebrain, character, and that other signals subsequently posteriorize neural tissue in the trunk. Here we show that development of the anterior central nervous system of the zebrafish embryo also depends on a small group of ectodermal cells located in the prospective head region. Removal of these ectodermal cells during gastrulation perturbs subsequent neural patterning and results in widespread cell death. Transplantation of these cells shows that they can induce forebrain-specific gene expression in more posterior regions of the neural plate. Our results indicate that an early step in neural patterning is the establishment of a small population of signalling cells within the most anterior region of the embryo. These cells are required for patterning and survival of the anterior brain.

The haemangioblast is a biopotential precursor of both blood and endothelial cells that has been hypothesized to exist for almost 100 years, but for which formal proof has been lacking. Vogeli et al. now provide in vivo evidence for its existence. High-resolution, single cell lineage studies of zebrafish embryos show that the haemangioblast is present in the early fish embryo, and also reveal new information about its structure and function. In vivo evidence details the existence of the haemangioblast: a bipotential precursor of both blood and endothelial cells that has been hypothesized to exist for almost 100 years, but for which formal proof has so far been lacking. It has been proposed that haematopoietic and endothelial cells share a common progenitor, termed the haemangioblast. This idea was initially conceived as a result of the observation that these two cell types develop in close proximity to each other within the embryo1,2. Support for this hypothesis was provided by studies on single-cell-derived colonies that can produce both haematopoietic and endothelial cells in vitro3,4,5,6,7. Although these data point towards the existence of a common progenitor for these two lineages, the presence of a bipotential progenitor cell has yet to be demonstrated in vivo. Through the construction of single-cell-resolution fate maps of the zebrafish late blastula and gastrula, we demonstrate that individual cells can give rise to both haematopoietic and endothelial cells. These bipotential progenitors arise along the entire extent of the ventral mesoderm and contribute solely to haematopoietic and endothelial cells. We also find that only a subset of haematopoietic and endothelial cells arise from haemangioblasts. The endothelial descendants of the haemangioblasts all clustered in a specific region of the axial vessels regardless of the location of their progenitors. Our results provide in vivo evidence supporting the existence of the haemangioblast and reveal distinct features of this cell population.

An abundant cDNA enriched in Spemann's organizer, cerberus, was isolated by differential screening. It encodes a secreted protein that is expressed in the anterior endomesoderm. Microinjection of cerberus mRNA into Xenopus embryos induces ectopic heads, and duplicated hearts and livers. The results suggest a role for a molecule expressed in the anterior endoderm in the induction of head structures in the vertebrate embryo.

The notochord is a midline mesodermal structure with an essential patterning function in all vertebrate embryos. Zebrafish floating head (flh) mutants lack a notochord, but develop with prechordal plate and other mesodermal derivatives, indicating that flh functions specifically in notochord development. We show that floating head is the zebrafish homologue of Xnot, a homeobox gene expressed in the amphibian organizer and notochord. We propose that flh regulates notochord precursor cell fate.

Specification of dorsal-ventral cell fate during Drosophila embryogenesis is mediated by a signal transduction pathway. Asymmetry of cell fates arises through the spatially restricted production of a ligand in an extracellular compartment called the perivitelline space. The snake and easter genes are required for the production of the ligand and they encode the proenzyme form of secreted extracellular serine proteases. We have examined the effect of producing a preactivated form of the snake protease on the generation of dorsal-ventral polarity. SP6 RNA microinjection experiments reveal that different cell fates acquired at cellular blastoderm can be specified by the amount and spatial distribution of activated snake protein. Our results support a protease cascade model in which localized activation of uniformly distributed protease proenzymes leads to the spatially restricted production of ligand in the perivitelline space on the ventral side of the embryo.

THE vertebrate body is organized along three geometric axes: anterior–posterior, dorsal–ventral and left–right. Left–right axis formation, displayed in heart and gut development, is the least understood, even though it has been studied for many years1–4. In Xenopus laevis gastrulae, a fibronectin-rich extracellular matrix is deposited on the basal surface of ectoderm cells5,6 over which cardiac and visceral primordia move during development. Here I report experiments in which localized perturbation of a small patch of extracellular matrix by microsurgery was correlated with localized randomization of left–right asymmetries. Global perturbation of the extracellular matrix by microinjection of Arg-Gly-Asp peptides or heparinase into the blastocoel resulted in global randomization of left-right asymmetries. From these observations, I suggest that left–right axial information is contained in the extracellular matrix early in development and is independently transmitted to cardiac and visceral primordia.

The vertebral column exhibits segmentation and regionalization along the antero-posterior axis. During embryogenesis, the rhythmic production of the precursors of the vertebrae, the somites, imposes a segmented aspect to the spine, whereas the spine's regional differentiation is controlled by Hox genes. Here we show that in the paraxial mesoderm, Hoxb genes are first activated in a temporal collinear fashion in precursors located in the epiblast lateral to the primitive streak. Our data suggest that collinear activation of Hoxb genes regulates the flux of cells from the epiblast to the streak and thus directly controls the establishment of the genes' characteristic nested expression domains in the somites. This suggests that establishment of the spatial co-linearity in the embryo is directly controlled by the Hox genes themselves.

Haematopoietic and vascular cells are thought to arise from a common progenitor called the haemangioblast. Support for this concept has been provided by embryonic stem (ES) cell differentiation studies that identified the blast colony-forming cell (BL-CFC), a progenitor with both haematopoietic and vascular potential1,2. Using conditions that support the growth of BL-CFCs, we identify comparable progenitors that can form blast cell colonies (displaying haematopoietic and vascular potential) in gastrulating mouse embryos. Cell mixing and limiting dilution analyses provide evidence that these colonies are clonal, indicating that they develop from a progenitor with haemangioblast potential. Embryo-derived haemangioblasts are first detected at the mid-streak stage of gastrulation and peak in number during the neural plate stage. Analysis of embryos carrying complementary DNA of the green fluorescent protein targeted to the brachyury locus demonstrates that the haemangioblast is a subpopulation of mesoderm that co-expresses brachyury (also known as T) and Flk-1 (also known as Kdr). Detailed mapping studies reveal that haemangioblasts are found at highest frequency in the posterior region of the primitive streak, indicating that initial stages of haematopoietic and vascular commitment occur before blood island development in the yolk sac.

Oriented cell division is an integral part of pattern development in processes ranging from asymmetric segregation of cell-fate determinants to the shaping of tissues. Despite proposals that it has an important function in tissue elongation, the mechanisms regulating division orientation have been little studied outside of the invertebrates Caenorhabditis elegans and Drosophila melanogaster1. Here, we have analysed mitotic divisions during zebrafish gastrulation using in vivo confocal imaging and found that cells in dorsal tissues preferentially divide along the animal–vegetal axis of the embryo. Establishment of this animal–vegetal polarity requires the Wnt pathway components Silberblick/Wnt11, Dishevelled and Strabismus. Our findings demonstrate an important role for non-canonical Wnt signalling in oriented cell division during zebrafish gastrulation, and indicate that oriented cell division is a driving force for axis elongation. Furthermore, we propose that non-canonical Wnt signalling has a conserved role in vertebrate axis elongation, orienting both cell intercalation and mitotic division.

Vertebrate gastrulation is a critical step in the establishment of body plan. During gastrulation, epithelial-mesenchymal transition (EMT) occurs. EMT is one of the central events of embryonic development, organ and tissue regeneration, and cancer metastasis. Signal transducers and activators of transcription (STATs) mediate biological actions such as cell proliferation, differentiation and survival in response to cytokines and growth factors, in a variety of biological processes. STATs are also important in EMT during gastrulation, organogenesis, wound healing and cancer progression. We previously showed that STAT3 is activated in the organizer during zebrafish gastrulation and its activity is essential for gastrulation movements. The requirement for STAT3 is cell-autonomous for the anterior migration of gastrula organizer cells, and non-cell-autonomous for the convergence of neighbouring cells. The molecular mechanisms of STAT's action in EMT, however, are unknown. Here we identify LIV1, a breast-cancer-associated zinc transporter protein, as a downstream target of STAT3 that is essential and sufficient for STAT3's cell-autonomous role in the EMT of zebrafish gastrula organizer cells. Furthermore, we demonstrate that LIV1 is essential for the nuclear localization of zinc-finger protein Snail, a master regulator of EMT. These results establish a molecular link between STAT3, LIV1 and Snail in EMT.

During vertebrate embryo development, the breaking of the initial bilateral symmetry is translated into asymmetric gene expression around the node and/or in the lateral plate mesoderm. The earliest conserved feature of this asymmetric gene expression cascade is the left-sided expression of Nodal, which depends on the activity of the Notch signalling pathway. Here we present a mathematical model describing the dynamics of the Notch signalling pathway during chick embryo gastrulation, which reveals a complex and highly robust genetic network that locally activates Notch on the left side of Hensen's node. We identify the source of the asymmetric activation of Notch as a transient accumulation of extracellular calcium, which in turn depends on left-right differences in H+/K+-ATPase activity. Our results uncover a mechanism by which the Notch signalling pathway translates asymmetry in epigenetic factors into asymmetric gene expression around the node.

ADORSALIZING signal acts during gastrulation to change the specification of lateral mesodermal tissues from ventral (blood, mesenchyme) to more dorsal fates (muscle, heart, pronephros)1–3. This signal, from Spemann's organizer, cannot be mimicked by the mesoderm inducers activin and fibroblast growth factor2. The gene noggin is expressed in the organizer4, and could be the dorsalizing signal. Here we show that soluble noggin protein added to ventral marginal zones during gastrulation induces muscle, but that activin does not. Dorsal pattern can be partially rescued in ventralized embryos by injection of a plasmid that expresses noggin during gastrulation. The results suggest that the noggin product may be the dorsalizing signal from the organizer.

Cell signalling through Frizzled receptors has evolved to considerable complexity within the metazoans. The Frizzled-dependent signalling cascade comprises several branches, whose differential activation depends on specific Wnt ligands, Frizzled receptor isoforms and the cellular context. In Xenopus laevis embryos, the canonical beta-catenin pathway contributes to the establishment of the dorsal-ventral axis. A different branch, referred to as the planar cell polarity pathway, is essential for cell polarization during elongation of the axial mesoderm by convergent extension. Here we demonstrate that a third branch of the cascade is independent of Dishevelled function and involves signalling through trimeric G proteins and protein kinase C (PKC). During gastrulation, Frizzled-7 (Fz7)-dependent PKC signalling controls cell-sorting behaviour in the mesoderm. Loss of zygotic Fz7 function results in the inability of involuted anterior mesoderm to separate from the ectoderm, which leads to severe gastrulation defects. This result provides a developmentally relevant in vivo function for the Fz/PKC pathway in vertebrates.

Although cell movements are vital for establishing the normal architecture of embryos, it is unclear how these movements are regulated during development in vertebrates. Inhibition of Xenopus Dishevelled (Xdsh) function disrupts convergent extension movements of cells during gastrulation, but the mechanism of this effect is unclear, as cell fates are not affected. In Drosophila, Dishevelled controls both cell fate and cell polarity, but whether Dishevelled is involved in controlling cell polarity in vertebrate embryos has not been investigated. Here we show, using time-lapse confocal microscopy, that the failure of cells lacking Xdsh function to undergo convergent extension results from defects in cell polarity. Furthermore, Xdsh mutations that inhibit convergent extension correspond to mutations in Drosophila Dishevelled that selectively perturb planar cell polarity. Finally, the localization of Xdsh at the membrane of normal dorsal mesodermal cells is consistent with Xdsh controlling cell polarity. Our results show that polarized cell behaviour is essential for convergent extension and is controlled by vertebrate Dishevelled. Thus, a vertebrate equivalent of the Drosophila planar cell polarity signalling cascade may be required for normal gastrulation.

The earliest cell fate decision in the mammalian embryo separates the extra-embryonic trophoblast lineage, which forms the fetal portion of the placenta, from the embryonic cell lineages. The body plan of the embryo proper is established only later at gastrulation, when the pluripotent epiblast gives rise to the germ layers ectoderm, mesoderm and endoderm. Here we show that the T-box gene Eomesodermin performs essential functions in both trophoblast development and gastrulation. Mouse embryos lacking Eomesodermin arrest at the blastocyst stage. Mutant trophoectoderm does not differentiate into trophoblast, indicating that Eomesodermin may be required for the development of trophoblast stem cells. In the embryo proper, Eomesodermin is essential for mesoderm formation. Although the specification of the anterior-posterior axis and the initial response to mesoderm-inducing signals is intact in mutant epiblasts, the prospective mesodermal cells are not recruited into the primitive streak. Our results indicate that Eomesodermin defines a conserved molecular pathway controlling the morphogenetic movements of germ layer formation and has acquired a new function in mammals in the differentiation of trophoblast.

In zebrafish, as in Xenopus, the well-orchestrated cell movements of gastrulation can be dissected into several components, including epiboly, involution, convergence and extension. Embryos homozygous for the recessive lethal mutation spt-1(b104) or 'spadetail' have a complex set of defects in the trunk of the embryo that may arise secondarily after loss of one of these movements, convergence, from those precursors that would normally have given rise to trunk somitic mesoderm. We have now tested this hypothesis by transplanting cells between wild-type and mutant embryos, to identify the cells that spt-1 affects directly. Our results show that the mutation autonomously affects only those mesodermal precursors located along the lateral margin of the early gastrula blastoderm. Other mesodermal cells and all ectodermal precursors seem not to require function of the wild-type gene. Our findings reveal an unexpectedly delicate genetic control of vertebrate gastrulation.

The anterior–posterior axis of the mouse embryo is established by two distinct organizing centres in the anterior visceral endoderm and the distal primitive streak1,2,3,4,5,6,7. These organizers induce and pattern the head and trunk respectively, and have been proposed to be localized through coordinate cell movements that rotate a pre-existing proximal–distal axis6,8. Here we show that correct localization of both head- and trunk-organizing centres requires Cripto9,10, a putative signalling molecule that is a member of the EGF-CFC gene family11,12. Before gastrulation, Cripto is asymmetrically expressed in a proximal–distal gradient in the epiblast, and subsequently is expressed in the primitive streak and newly formed embryonic mesoderm. A Cripto null mutation generated by targeted gene disruption results in homozygous Cripto−/− embryos that mostly consist of anterior neuroectoderm and lack posterior structures, thus resembling a head without a trunk. Notably, markers of the head organizer are located at the distal end of the embryo, whereas markers of the primitive streak are absent or localized to the proximal side. Our results indicate that Cripto signalling is essential for the conversion of a proximal–distal asymmetry into an orthogonal anterior–posterior axis.

Signalling by the transforming growth factor-β (TGF-β) superfamily of proteins depends on the phosphorylation and activation of SMAD proteins by heteromeric complexes of ligand-specific type I and type II receptors with serine/threonine-kinase activity1. The vertebrate SMAD family includes at least nine members, of which Smad2 has been shown to mediate signalling by activin and TGF-β2,3,4,5. In Xenopus, Smad2 can induce dorsal mesoderm, mimicking Vg-1, activin and nodal2,4. Here we investigate the function of Smad2 in mammalian development by generating two independent Smad2 mutant alleles in mice by gene targeting. We show that homozygous mutant embryos fail to form an organized egg cylinder and lack mesoderm, like mutant mice lacking nodal6,7 or ActRIB, the gene encoding the activin type-I receptor8. About 20 per cent of Smad2 heterozygous embryos have severe gastrulation defects and lack mandibles or eyes, indicating that the gene dosage of Smad2 is critical for signalling. Mice trans-heterozygous for both Smad2 and nodal mutations display a range of phenotypes, including gastrulation defects, complex craniofacial abnormalities such as cyclopia, and defects in left–right patterning, indicating that Smad2 may mediate nodal signalling in these developmental processes. Our results show that Smad2 function is essential for early development and for several patterning processes in mice.

DURING gastrulation in vertebrates, ectodermal cells choose between two fates, neural and epidermal. The nervous system forms in response to signals from the Spemann organizer1,2; ectoderm that does not receive these signals becomes epidermis. Unexpectedly, however, in Xenopus, neural tissue also forms when cell-cell communication within the ectoderm is disrupted by cell dissociation3,4 or by antagonists of the growth factor activin5-7. These observations suggest that epidermal specification depends on local signalling, by activin or a close relative, and that neural tissue forms when this communication is blocked6. Here we report that bone morphogenesis protein 4 (Bmp-4), a relative of activin that is expressed in the embryo at the time of ectodermal fate determination8,9, is a potent epidermal inducer and neural inhibitor, the first reported in any vertebrate. Activin can inhibit neuralization by inducing mesoderm, but does not induce epidermis. Moreover, the dominant-negative activin receptor, which stimulates neuralization when expressed in the embryo5,6, blocks Bmp-4 in our assay. Our findings demonstrate that epidermal fate can be induced, and thus provide further evidence that neural specification is under inhibitory control in vertebrates.

A similar mechanism may underlie the handedness seen in all vertebrate body plans. How left–right handedness originates in the body plan of the developing vertebrate embryo is a subject of considerable debate1,2. In mice, a left–right bias is thought to arise from a directional extracellular flow (nodal flow) that is generated by dynein-dependent rotation of monocilia on the ventral surface of the embryonic node3,4. Here we show that the existence of node monocilia and the expression of a dynein gene that is implicated in ciliary function are conserved across a wide range of vertebrate classes, indicating that a similar ciliary mechanism may underlie the establishment of handedness in all vertebrates.

AT the midblastula transition in the zebrafish, three, and only three, spatially separate mitotic domains arise with distinctive cycle lengths and rhythms. As in Drosophila1 and at about the equivalent stage, the mitotic domains reflect the fate map, but they do so only very crudely: two are extraembryonic and the third forms the entire embryo. The domains appear not to subdivide during gastrulation, when the germ layers form and when cells probably commit to their eventual fates. The domains may signal specification of morphogenesis rather than cell fate, because, shortly after they appear, each assumes a different role during epiboly, the first morphogenetic movement of the embryo. During meroblastic cleavage, and continuing in the early blastula, zebrafish blastomeres divide rapidly and synchronously. At the time of the tenth cleavage, the beginning of the midblastula transition, the cell cycle lengthens, and, as in Xenopus2 and Drosophila3, cycle length comes under nucleocytoplasmic control (D.A.K. and C.B.K., manuscript in preparation). This nucleocytoplasmic control seems to be maintained during cycle lengthening in the next 2 or 3 cycles, comprising a midblastula transition period. We now show that functionally distinct subsets of cells that arise during this period have reproducibly different mitotic cycle lengths.

INTERMEDIATE filaments are ubiquitous in eukaryotic cells, but their functions are poorly understood. The Xenopus oocyte contains both messenger RNA and protein products of cytokeratin and vimentin genes'1–5 in non-overlapping arrays. The cytokeratin filaments contain dimers of the type I (acidic) subunit XLK3a(19), and the type II (basic) subunit XCK1(8), polymerized to form a cortical network. These are homologues of the human simple epithelial keratins 19 and 8, respectively. After the first few cell cycles following fertilization these filaments become restricted to the superficial cells of the blastula. We have depleted the oocyte's store of the type II cytokeratin mRNA by injecting antisense oligodeoxynucleotides (oligos) and studied the effect on embryonic development. As zygotic transcription does not commence until the late blastula stage6, there are at least 9 hours in which to see the effect of loss of function of this mRNA. We report here that the cytokeratin filaments become depleted in the cortical cells of the embryo. As a result, there is a loss of the 'compacted' epithelial surface of the blastula, an inability to close a wounded surface and defective gastrulation.

Formation of mesoderm is a crucial event in vertebrate development, establishing many of the important features of the body. Recent studies have implicated molecules that are similar to growth factors in mesoderm formation in Xenopus, but other gene products involved in this process have yet to be identified. Genetic evidence indicates that in the mouse the T gene (Brachyury) has a role in the formation and organization of mesoderm. Mice homozygous for mutant alleles of the T gene do not generate enough mesoderm, and show severe disruption in morphogenesis of mesoderm-derived structures, in particular the notochord. The cloning of the T gene has now allowed us to examine its expression pattern. We report that T-gene expression occurs in both early stage mesoderm and its epithelial progenitor, and then becomes restricted to the notochord. This expression pattern correlates with the tissues affected in the T-gene mutant, and indicates that the T gene has a direct role in the early events of mesoderm formation and in the morphogenesis of the notochord.

Contact guidance has been implied in various morphogenetic movements including neural crest cell migration, primordial germ cell migration and guidance of axonal growth cone. In urodele gastrulae, we reported the presence of an aligned network of extracellular fibrils on the inside of the ectodermal layer and suggested that it guides the migration of the presumptive mesodermal cells from the blastopore towards the animal pole. We also reported in vitro experiments in which the fibril network of the ectodermal layer was transferred onto the surface of a coverslip. Dissociated mesodermal cells attach to and locomote actively on such conditioned surfaces in an oriented fashion along the blastopore-animal pole axis (bp-ap axis) of the ectodermal layer that conditioned the surface. Recent reports suggest that these fibrils contain fibronectin. We now report that the fibril network on the conditioned surface can be artificially aligned in any orientation by exerting mechanical tension on the ectodermal layer during the conditioning. Such prepared surfaces cause cell movements aligned along the tension axis, even when the tension axis is perpendicular to the natural axis of alignment along the bp-ap axis. These results suggest that the extracellular matrix fibrils aligned by the mechanical stress that arises in embryos during development can orient cell migration by the contact guidance, in a similar manner to that reported in the collagen gel and fibroblasts system.

It has recently been suggested that the Arg-Gly-Asp sequence (RGD) forms part of a widespread cell–extracellular matrix recognition system (see for example ref. 1). Analysis of the cell binding sites of vertebrate fibronectin2,3 and other extracellular proteins that interact with cell surfaces implicate the same amino acid triplet1. Peptides containing this sequence inhibit certain developmental events such as cell–matrix adhesion or cellular migration in vitro2–7 and in vivo8. The RGD-sequence is also part of the cellular recognition site of the aggregation protein discoidin I in Dictyostelium9 suggesting that the RGD-recognition system could be universally used. In Drosophila, despite its advanced genetics, very little is known about the extracellular components that are involved in cell movements and morphogenesis. We report here that peptides containing the RGD-sequence prevent gastrulation of Drosophila embryos. The phenotypic effect is similar to that observed in the dorsal-group mutants: no ventral furrow is formed and the embryos lack dorsal–ventral polarity. The specificity of the inhibiting action suggests that the RGD-sequence may also be used by invertebrates to mediate cell-attachment phenomena.

Cell lineage may contribute to determining the numbers, positions and types of cells formed during embryogenesis. In vitro clonal analyses show that vertebrate cells can autonomously maintain lineage commitments to single fates1–5 and that terminal development may include an invariant sequence of cell divisions6,7. In addition, in vivo studies with Xenopus8 led to the proposal that clonal restrictions to spatial 'compartmental' domains arise during early development, analogous to what is observed in insects9,10. In the zebrafish, individual gastrula cells generate clones of progeny that are confined within single tissues11, but spatial restrictions have not been described. We now have examined the in vivo terminal cell lineages of zebrafish axial muscles. We obtained no evidence either for strict developmental regulation of division pattern or for spatial compartmentation within muscle lineages.

Segmentation in the insect embryo is initiated by maternally provided information, which is stored in the developing oocyte1–6. In Drosophila, the genes necessary for this process have been genetically characterized2–6. The anterior segmented region is organized by the bicoid (bcd) gene product2. The posterior segmented region is organized by several interacting gene products, among them the oskar (osk)3–5 gene product. The first zygotic group of genes, which are thought to respond to the spatial cues provided by the maternal genes, are the gap genes7–9, whose members include hunchback (hb), Kruppel (Kr) and knirps (kni)7. To elucidate the role played by the maternal genes in expression of the gap gene hb, antibodies were raised against a fusion protein and were used for the cytological localization of the hb gene product in wild-type and mutant embryos. The hb protein is pre-dominantly located in the nucleus. Its spatial expression includes the formation of an anterior-posterior gradient during the early cleavage stages and a strong zygotic expression in the anterior half of the embryo. Analysis of embryos mutant for the maternal genes affecting the anterior-posterior segmentation pattern shows that the formation of the early gradient is controlled by the osk group of genes, whereas efficient activation of the zygotic anterior expression domain is dependent on bcd activity.

Gastrulation and formation of the neural plate are major steps in early vertebrate embryogenesis. Although morphogenetic movements leading to the formation of the primary germ layers have been extensively described1–3, the mechanisms governing migration of mesodermal cells and their interactions with ectoderm remain ill-defined. A large body of evidence indicates that fibronectin (FN), a high molecular weight cell-surface-associated glycoprotein, promotes cell adhesion and cell migration throughout embryogenesis4–6. FN has been detected at an early blastula stage in Pleurodeles waltlii7. We now show that FN is a component of a dense fibrillar matrix underlying the blastocoel roof; in contrast, the exterior surface of the embryo is devoid of FN. Microsurgical inversion of part of the blastocoel roof does not prevent mesodermal cell migration except at the site of inversion where no FN matrix is available. Perturbation experiments using antibodies to FN demonstrate that the invagination of presumptive mesodermal cells does not occur when the monovalent antibodies are injected before or at the onset of gastrulation; on the other hand, the formation of a neural plate is not prevented when late gastrula stage embryos are treated with antibodies to FN. We conclude that the presence of FN is required for cell migration during gastrulation.