Local cell interactions and self-amplifying individual cell ingression drive amniote gastrulation

Abstract

Gastrulation generates three layers of cells (ectoderm, mesoderm, endoderm) from a single sheet, while large scale cell movements occur across the entire embryo. In amniote (reptiles, birds, mammals) embryos, the deep layers arise by epithelial-to-mesenchymal transition (EMT) at a morphologically stable midline structure, the primitive streak (PS). We know very little about how these events are controlled or how the PS is maintained despite its continuously changing cellular composition. Using the chick, we show that isolated EMT events and ingression of individual cells start well before gastrulation. A Nodal-dependent 'community effect' then concentrates and amplifies EMT by positive feedback to form the PS as a zone of massive cell ingression. Computer simulations show that a combination of local cell interactions (EMT and cell intercalation) is sufficient to explain PS formation and the associated complex movements globally across a large epithelial sheet, without the need to invoke long-range signalling.DOI: http://dx.doi.org/10.7554/eLife.01817.001.

Keywords: EMT; cell movements; computer simulation; epithelial–mesenchymal interactions; modelling; primitive streak.

Figure 1.. Diagrams depicting the early stages of chick development.

The upper row of diagrams shows embryos at stages XI-XIV (pre-primitive streak), 2 (early streak), 3 (mid-streak) and 3⁺ (mid- to late streak), viewed from the dorsal (epiblast) side. The arrows denote the main morphogenetic movements (‘Polonaise’) occurring within the plane of the epiblast. After stage 4 (end of gastrulation), convergence of cells towards and ingression through the anterior part of the streak slows down or ceases (although these movements continue through the middle and posterior parts of the streak), while the epiblast anterior to the streak (prospective neural plate) elongates (Sheng et al., 2003); later, the streak starts to regress, further lengthening the neural plate posteriorly (Spratt, 1947). The lower row of diagrams shows an exploded view of the embryos at each of the above stages, with the top row of diagrams representing the upper layer (epiblast, shades of yellow), the bottom row showing the lower layer (shades of blue/green: hypoblast in dark green, endoblast in light green, definitive or gut endoderm in blue) and the centre row showing the middle (mesodermal) layer (primitive streak, in red). Within the epiblast, the central (yellow) region is the area pellucida and the outer (mustard) region the extraembryonic, area opaca. DOI: http://dx.doi.org/10.7554/eLife.01817.003

Figure 2.. EMT in the formation of the primitive streak (PS).

(A–E) Images from a time-lapse sequence of entire embryos (Video 1), showing the uniform epiblast 6 hr (A, stage EG&K XII) and just before primitive streak formation (B, stage EG&K XIV), the first appearance of the primitive streak (C, stage HH2), accumulation of mesoderm beneath the flat streak (D, stage HH3), appearance of a groove in the PS and emigration of mesoderm (E, stage HH3+). (F–K) SEM of fractured embryos before (F–H) and after (I–K) streak formation. White arrows indicate possible EMT before PS formation. (L–P) SEM of fractured PS, showing EMT cells with various degrees of apical constriction and basolateral expansion (classified as ‘ingression stages 1–5’). (Q) This embryo was cultured for 1 hr after electroporation of a control, fluorescent morpholino into the entire epiblast at stage XI, then sectioned sagitally and viewed under fluorescence. Labelled cells in the epiblast show similar morphologies to those in SEMs (panels L–P, ‘ingression stages 1–5’). (R) This embryo was cultured for 4 hr after electroporation of a control, fluorescent morpholino into the entire epiblast at stage XI, then fixed (at stage XII), sectioned sagitally and stained with anti-fluorescein antibody (brown). The section shows several cells that have left the epiblast and are now in the underlying space throughout the anterior-posterior extent of the embryo (arrows). DOI: http://dx.doi.org/10.7554/eLife.01817.005

Figure 3.. Clustering of seemingly stochastic EMT underpins the formation of PS.

(A and B) Uniform distribution of EMT in the epiblast before PS formation (A) and acceleration of EMT as the PS appears (B). Locations are plotted from 6 hr time-lapse sequences (see Videos 3 and 4, respectively) and the time of ingression is colour-coded (numbers represent minutes). Each field of view is 600 × 600 μm, in the central posterior epiblast (where the primitive streak arises). (C and D) Apical surface of the epiblast seen in SEM at PS formation stages. (E) An individual epiblast cell followed in time-lapse before (see Video 5) undergoing repeated attempts at full EMT. (F) Multi-photon time-lapse sequence of EMT at PS stages. The top left-hand panel shows a diagram of the embryo with the area imaged enclosed in a square. The other panels represent views in the x-z (top right), y-z (bottom-left) and x-y (bottom-right). The positions of selected, colour-coded cells at successive time points (10 min intervals) are connected with lines. DOI: http://dx.doi.org/10.7554/eLife.01817.007

Figure 4.. Quantification of ingression from the epiblast with time.

Cell ingression accelerates as the PS forms and cells approach its midline. (A) the first and last frames of Video 4 (left and right panels, respectively), highlighting the triangles used for measuring. (B) relative change in surface area of each triangle over time (min). (C) relative change in surface area of each triangle as a function of distance to the midline (in μm). (D) correlation coefficient (r²) of the size reduction plotted against initial distance to the midline (μm). DOI: http://dx.doi.org/10.7554/eLife.01817.009

Figure 5.. Cells in EMT trigger a chain reaction of EMT in a Nodal-dependent manner.

(A) EMT cells from the early PS of a quail embryo (left) are grafted under the epiblast of a pre-PS chick embryo (right). (B) Grafted cells (brown stain, thin black arrow) upregulate EMT markers (cSnail2, purple) and trigger EMT from the epiblast above, after 4 hr. (C) Grafted embryo after 15 hr. The grafted quail cells (brown) have migrated away, and the new PS they triggered (‘2^o PS’) is composed of host cells. The PS developing along the original orientation is labelled ‘1^o PS’. In grafts combined with COS cells secreting Cerberus (E) or Cer-S (F), or beads soaked in SB431542 (H) or SB505124 (I), EMT (thickening) from the epiblast and induction of cSnail2 (purple) in the epiblast (red arrows) are abolished. Control COS cells (D) or beads soaked in DMSO (G) do not abolish the induction by the grafted mesoderm (black arrows). Mesoderm from a region lateral to the PS cannot induce EMT or cSnail2 either alone (not shown) or in the presence of GFP-transfected COS cells (J) or beads soaked in solvent alone (K) (red arrows). DOI: http://dx.doi.org/10.7554/eLife.01817.015

Figure 6.. Different views of a simulation of normal development.

These diagrams provide an explanatory key for the simulation videos and illustrate the principal signals, cell behaviours and the major tissues involved in gastrulation. Three time points are shown: stage XI, stage 2 and stage 3+. The upper 7 rows are dorsal views onto the epiblast; the lower 3 rows are oblique views. Colours are additive when a cell is positive for more than one displayed state (see e.g., the row labelled ‘combined’, which symbolises the sum of all features in the rows above it for the forming primitive streak). Nodal(+) cells are shown in red (top row), Wnt-PCP(+) cells in yellow (second row). Cells positive for both Nodal and Wnt-PCP appear orange (third row). At Stage XI all cells in the future streak-forming region are Nodal and Wnt-PCP positive. Later, most continue to have both activities but some cells are only positive for Nodal (red). Cells undergoing EMT are shown in blue and ‘mesendodermal’ cells in aquamarine (fourth row). For combinations of Nodal, Wnt-PCP, EMT and mesendoderm, note that Nodal(+)-EMT cells appear purple (red + blue); if also Wnt-PCP(+) then approximately violet (red + yellow + blue) (‘combined’). The hypoblast is shown chocolate-coloured and the endoblast greenish-slate (rows 6 and 8). Hypoblast displacement by the endoblast (at stage XIV; between stages XI and 2 in the Figure) disinhibits Nodal in the overlying epiblast (see text). Sequential cell positions are integrated by remembering all previous time points to form ‘trails’, as shown in row 7. For clarity, trails made from 15% of the cells are shown. The last three rows depict the embryo viewed from an oblique angle. In row 8 (‘hypoblast and endoblast’), the position of the lower layer can be seen (also see above, lower layer). Initially this consists only of hypoblast (chocolate). At later stages, endoblast (greenish-slate) partially displaces the hypoblast. The epiblast is also seen from below (‘epiblast ventral view’, row 9), allowing clear visualization of EMT (blue/purple/violet) and emerging and emerging middle layer (aquamarine) cells. The final row, ‘epiblast dorsal view’ (row 10), displays the epiblast from above with a pseudo-surface applied, simulating indentations caused by ingressing cells. These indentations sum as cells approach the posterior midline, generating a midline groove at the PS. The pseudo-surface is created by tessellating points representing the top of each epithelial cell (using the cell body for cells undergoing EMT). DOI: http://dx.doi.org/10.7554/eLife.01817.016

Figure 6—figure Supplement 1.. Modelling: hierarchical time implementation.

Time is represented as ‘ticks’. Each simulation tick executes activities that include a set of actions for the entire embryo (‘organism tick’). The organism tick in turn executes activities including a cell tick for each cell in the organism. Cell ticks calculate and execute activities for each cell. Note that many of these calculations and activities are themselves iterative. DOI: http://dx.doi.org/10.7554/eLife.01817.017

Figure 6—figure Supplement 2.. Modelling: MZ displacement vectors.

For each MZ cell a displacement vector (white arrowhead) is calculated as the vector sum of ‘curvature’ (orange), ‘density’ (green) and ‘area correction’ (blue) vectors. A mark (red dot) identifies the common origin of each. Vectors are shown magnified 50x for illustration. DOI: http://dx.doi.org/10.7554/eLife.01817.018

Figure 6—figure Supplement 3.. Modelling: schematic representation of EMT and Nodal expression.

(A) Shown is a schematic representation of EMT in the model. Right side: Nodal(−) epithelial cells (grey) may convert to emt cells (blue) which at first are tethered to the epithelium (t-emt) but then become untethered (u-emt) as they descend into the middle layer. They complete the transition as mesenchymal (meso) cells (aquamarine). While still tethered and with cell body above the basement membrane (BM), some will revert and rejoin the epithelium (double-headed arrow). Left side: Nodal(−) epithelial cells convert to Nodal(+) (red) in the region of the PS. The rate of EMT increases with increasing Nodal activity from the cell and its neighbours; Nodal-active emt cells (red + blue = purple) lose the ability to rejoin the epithelium (thicker, single-headed arrow). Conversion to Nodal-positivity and the enhanced rate of EMT is inhibited by the hypoblast and disinhibited when the endoblast displaces the hypoblast. (B) Shown are cell interactions leading to Nodal expression. Nodal(−) epithelial cells (grey) are converted to Nodal(+) cells by near-neighbour Nodal(+) epithelial cells (red) and local neighbour Nodal(+) emt cells (purple = blue[emt] + red[Nodal]). For local neighbours the effect falls off with distance but is particularly enhanced for near-neighbour epithelial cells (arrow widths). A similar scheme (not shown) applies to Wnt-PCP conversion and to EMT recruitment. Numbers of cells, distances and proportions not to scale. BM: basement membrane, t-emt: tethered emt cell, u-emt: untethered emt cell, meso: mesenchymal cell. DOI: http://dx.doi.org/10.7554/eLife.01817.019

Figure 6—figure Supplement 4.. Modelling: cell movements within the plane of the epiblast.

(A) Schematic diagram of the equilibration algorithm. Cells (solid circles) are distributed in a hexagonal array. Voronoi regions (VR) are transiently generated around these cells (solid lined hexagons). Cell centroids and VR centroids correspond (dot) and the tissue is at equilibrium. When cell ‘A’ shifts to a new position (dashed circle), new VR's are generated (dotted hexagons), making the centroid of the new VR for neighbouring cell ‘B’ move to a new position (cross). Cell ‘B’ then shifts towards this new position to reestablish equilibrium (arrow). The vector from the original centroid of cell ‘B’ to the centroid of the new VR of cell B is the equilibrium displacement vector (v_equil). (B) Propagation of oriented intercalation orientation vectors. MZ-cells maintain a reference OI-orientation vector state perpendicular to the MZ (determined by the local curvature). Epiblast cells calculate their individual OI-orientation vector states by averaging their current vector with the consensus of their near-neighbours, including MZ-cells (see text). Since the MZ-cell vectors are fixed, epiblast cells abutting the MZ will tend to align their vectors to those of the MZ-cells. Note that although this state is stored as a vector in the model, it has angular but not heads vs tails orientation. (C) Diagram of the oriented intercalation algorithm. A cell and its near-neighbour (NN) both possess OI-orientation information (double-headed arrows). A sequential displacement vector is calculated, oriented from the cell to its target and with a magnitude equal to |sinθ|. This is applied iteratively for all cells. DOI: http://dx.doi.org/10.7554/eLife.01817.020

Figure 7.. A model based on local cell behaviours explains the global movements in the epiblast and experimental conditions.

(A–E) Epithelial intercalation in a posterior domain (orange) and EMT (blue, isolated events, cooperative in the pink domain) are sufficient to explain the formation of the PS. (A–C) sequence in time, vertical view; (D) ventral view of the epiblast; (E) apical view of the epiblast. (F–H) Sequence from a time-lapse experiment, with cells in the intercalation domain electroporated with control morpholino (green) and other locations in the epiblast labelled with DiI (red). (F) initial condition, 6 hr before streak formation; (G) movements prior to streak formation; (H) movements over 6 hr after PS forms. (I and J) Movements observed in the same time-frame as in F–H, when intercalation is blocked by electroporating morpholinos (green) against the Wnt-PCP pathway. (K–O) The computer model correctly simulates the observed movements both in normal embryos (K–M) and in intercalation-compromised condition (N and O). (P–R) Hypoblast rotation at pre-PS stages leads to bending of the PS. (P) Experimental embryo, with the PS marked by Bra expression; the model accounts for this result (red in Q) by the induction of a new intercalation domain (yellow in R) which deforms the original one and the field of cooperative ingression (orange in R). (S and T) EMT cells can trigger a chain reaction of EMT and initiate a new PS in both experimental embryos (S) and in the model (T). DOI: http://dx.doi.org/10.7554/eLife.01817.023

Figure 7—figure Supplement 1.. Effect of key parameters on the behaviour of the computer simulation model.

The figure shows the composite effects of changing the value of m_N and σ_d on PS morphology. DOI: http://dx.doi.org/10.7554/eLife.01817.024

Figure 7—figure Supplement 2.. Model predictions compared with the results of Spratt (1946).

Spratt: global epiblast movements as described by Spratt from carbon-particle marking experiments (Spratt, 1946) (adapted from Spratt 1946). The diagrams combine a movement schematic and a representation of the PS. Model: a series of stages in a normal simulation showing the fates of horizontal bands of marked cells (upper row) and formation of the PS (lower row). The simulated pattern is also consistent with more recent analysis of epiblast cell movements and the Polonaise (Foley et al., 2000; Wei and Mikawa, 2000; Voiculescu et al., 2007). DOI: http://dx.doi.org/10.7554/eLife.01817.025