Cell shape changes indicate a role for extrinsic tensile forces in Drosophila germ-band extension

Abstract

Drosophila germ-band extension (GBE) is an example of the convergence and extension movements that elongate and narrow embryonic tissues. To understand the collective cell behaviours underlying tissue morphogenesis, we have continuously quantified cell intercalation and cell shape change during GBE. We show that the fast, early phase of GBE depends on cell shape change in addition to cell intercalation. In antero-posterior patterning mutants such as Kruppel, defective polarised cell intercalation is compensated by an increase in antero-posterior cell elongation, such that the initial rate of extension remains the same. Spatio-temporal patterns of cell behaviours indicate than an antero-posterior tensile force deforms the germ-band, causing the cells to passively change shape. The rate of antero-posterior cell elongation is reduced in twist mutant embryos which lack mesoderm. We propose that cell shape change contributing to germ-band extension is a passive response to mechanical forces caused by the invaginating mesoderm.

Convergence and extension movements remodel tissues during the morphogenesis of many embryos and organs, including vertebrate axis elongation^1–4. During Drosophila gastrulation, the embryo trunk (the germ-band) elongates in the antero-posterior (AP) axis and narrows in the dorso-ventral (DV) axis⁵. Polarised cell intercalation, which requires antero-posterior (AP) patterning, contributes to germ-band extension (GBE)^6–9. However, the relative importance of cell intercalation versus other possible cell behaviours¹⁰ has not been elucidated. Here we quantify the contribution of cell intercalation and cell shape change to tissue deformation in wild-type and mutant embryos.

We recorded the movement of up to 700 cells at 30-second intervals during GBE, imaging the ventral side of live Drosophila embryos (Fig. 1A). Cell outlines were labelled with DE-Cadherin-GFP¹¹ and cell movements tracked and quantified using a theory and algorithms presented elsewhere¹² (Fig. 1A’-A”). To quantify local tissue deformation, domains defined by a central cell surrounded by a corona of neighbouring cells were followed over 2 minute windows¹². We quantified how fast each of these domains changed dimensions along the AP and the DV axes of the embryo (‘total’ strain rates). Next, we quantified the average rates of shape change for cells belonging to each domain (‘cell shape’ strain rates). Finally, the strain rates attributable to cell intercalation were derived for each domain by subtracting the ‘cell shape’ strain rates from the ‘total’ strain rates (based on the relationship total strain rate = cell shape strain rate + cell intercalation strain rate defined in ¹²). A strength of this method is that it provides a continuous measure of cell intercalation which encompasses every type of cell intercalation described so far in the germ-band ^6,7. This measure also includes subtler cell intercalation movements that do not necessarily lead to cell neighbour exchange, but contribute to tissue deformation nonetheless (Supp. Movie 1).

Figure 1. Relative contribution of cell shape change and cell intercalation to germ-band extension in wild-type and Krüppel- embryos.

A) Lateral view of a Drosophila embryo: the extending germ-band is shaded (anterior to the left, dorsal uppermost). The red box indicates the stack of optical sections taken by confocal imaging, which corresponds to the ventral field of view shown in the adjacent panel. The left-hand side of the field of view is positioned using the cephalic furrow as a landmark. A’) Corresponding movie frame showing a projection of confocal sections through the cell apices labelled with DEcadGFP. The midline bisecting the embryo is the closing furrow through which the mesoderm has invaginated. A”) Tracked movie frame, showing the cell lineages (tracks), cell centroids (end points of tracks) and cell outlines recorded by the tracking software. B) Summary of deformation (strain) rates in the AP axis for 5 wild-type embryos, showing total, cell intercalation and cell shape strain rates (see key). A strain rate is the ratio of the change in length to the original length, divided by the time interval, with units of proportion per minute. The lines show the mean strain rate for all 5 embryos, while the ribbon width represents the average standard error within a data set. The timing of developmental landmarks is shown for the 5 embryos recorded. C) Cumulative representation of the same data, accumulating from t=0. D) Tissue deformation in the AP axis for 5 homozygous Krüppel- embryos. Dotted lines show the wild-type data for comparison. The shaded sections of the ribbons indicate when a given cell behaviour in the mutant is significantly different from the equivalent at that time in wild-type (p<0.05). E) Cumulative representation of the same data: for comparison, wild-type curves are shown as a lighter coloured line, and wild-type developmental landmarks as lighter coloured boxes on the same graph. F) Relative instantaneous contribution of cell shape changes and cell intercalation to extension in wild-type and Kruppel- mutant embryos, for the first 25 minutes of germ-band extension, which corresponds to the period when the total extension rate is the same in both genotypes.

GBE proceeds rapidly for about 30 minutes (the fast phase), then continues at a slower rate for another 90 minutes (the slow phase)^8,13. GBE has been shown to require cell intercalation, but it is not clear why the extension rate is biphasic^6–10. To understand this, we analysed 5 movies of wild-type embryos during the first 50 minutes of GBE. We confirm that cell intercalation contributes to extension throughout the period analysed (Fig. 1B,C). However, we also find that cell shape change contributes to the first 30 minutes of GBE, with the peak contribution of cell shape change, at around 10 minutes, coinciding with the peak rate of tissue extension (Fig. 1B,C). Over these first 30 minutes, cell shape change contributes about one third of total tissue deformation, identifying cell shape as a novel and significant contributor to GBE (Fig. 1C). We conclude that this additional contribution of cell shape changes explains the higher rate of extension that distinguishes the fast phase from the slow phase.

Mutants in which AP patterning is defective, such as those for the gap gene Kruppel, do not extend their germ-bands fully^6–9. However, in all AP patterning mutants examined, a normal tissue extension occurs initially, despite polarised cell intercalation being defective⁸. To understand this paradox we analysed 5 movies of Kruppel- embryos and found that in the first 25 minutes of GBE, although the rate of cell intercalation is reduced, the rate of cell shape change is increased (Fig. 1D). This increased rate of cell shape change compensates fully for the loss of cell intercalation, resulting in an initial rate of tissue extension indistinguishable from wild-type (Fig. 1E), with cell shape change now accounting for most of tissue extension (Fig. 1F). We find a similar compensation when analysing two other AP patterning mutants: a mutant in the pair-rule gene even-skipped (eve), and a double mutant in the gap genes knirps and hunchback (kni hb) (Supp. Fig. 3). As these analyses were carried out for the whole field of view, we confirmed them by examining the ectodermal cell population alone (Fig. 2 A). In the mutant ectoderms, cell intercalation rates are either reduced (eve and kr mutants) or abolished (kni hb), whereas cell shape change is systematically increased compared to wild-type (Fig. 2B-D). Therefore, increased cell shape change explains the maintenance of an initial rapid rate of tissue extension in AP mutants with defective cell intercalation⁸.

Figure 2. Spatio-temporal analysis of cell behaviours in the ectoderm of wild-type and antero-posterior patterning mutant embryos.

(A) The field of view that we analyse contains a mixture of ectodermal cells (laterally) and mesectodermal cells (at the ventral midline). The ectodermal cells (shaded areas) were selected for further analysis. (B-D) Comparison of the strain rates for total tissue, cell intercalation and cell shape, for wild-type and three antero-posterior patterning mutants, even-skipped (eve), Krüppel (Kr) and knirps hunchback (kni hb). Lines show the mean strain rate for the ectodermal cells for 5 embryos (wild-type, Kr, and eve) or a single embryo in the case of kni hb. (E-E”) Strain rates for cell intercalation contributing to tissue extension in wild-type, eve- and Kr- embryos (average for 5 embryos of each genotype), represented as a function of the cell’s position along the AP axis. Strain rates are colour coded according to the scale shown (note that any values below 0 will be coded blue, while any above 0.03 will be coded red). Cell position is given in µm from the anterior cephalic furrow, which separates head from trunk tissue (see Fig. 1A). The tissue is moving with respect to the cephalic furrow as it extends in the AP axis and the overlaid black lines follow this translation of the tissue over time. The data is shown here for the ectodermal cell population only. (F-F”) Similar analysis for cell shape strain rates.

Our results indicate that an unidentified mechanism independent of AP patterning or polarised cell intercalation causes cell shape change in the germ-band. Cell shape change could be cell-autonomously controlled by a genetic program acting independently of AP patterning. However, this would not explain why cell shape change increases when polarised cell intercalation decreases in AP-patterning mutants. A more plausible scenario is that both in wild-type and AP mutants, an external force is acting on the germ-band cells, causing them to change shape passively in response to tension¹⁴. We hypothesise that autonomous polarised cell intercalation relaxes the stress imposed on the germ-band tissue by this external force. In AP-patterning mutant embryos, this dissipation is reduced because cell intercalation is defective, and cell shape change increases.

To investigate this further, we looked at cell behaviours variations along the AP axis in the ectoderm of wild type and AP-patterning mutant embryos (Fig. 2 E-F”). Cell intercalation rates are mostly uniform along this axis in wild-type (Fig. 2E). In kni hb embryos which lack AP patterning in the trunk, cell intercalation is abolished throughout the field of view as expected (Supp. Fig. 3G). In Kruppel- embryos, cell intercalation is abolished in a central region, and reduced elsewhere (Fig. 2E’). This pattern is consistent with a loss of cell intercalation where Kruppel is normally expressed, and a weaker trunk-wide effect perhaps due to mis-expression of pair-rule genes in the Kruppel- mutant. Planar polarisation of Myosin II is associated with polarised cell intercalation in the germ-band^6,7,9. Consistent with this, we found that Myosin II concentration at the apical cell cortex is reduced in the ventral ectoderm of Kruppel- mutant embryos (data not shown). Note that we did not detect a difference in Myosin II localisation between the central and posterior domain to explain the observed difference in cell intercalation rates (Fig. 2E’). In even-skipped- embryos, cell intercalation is reduced uniformly along the length of the trunk (Fig. 2E”), consistent with eve expression in seven closely spaced stripes.

The pattern of cell shape change in wild-type embryos is graded along the AP axis, increasing from the centre towards the posterior end of the embryo between 5 and 20 minutes of GBE (Fig. 2F). There is also a shorter pulse of cell elongation at the anterior for the first 15 minutes of GBE which is also present in Kruppel embryos (Fig. 2F’) but absent in even-skipped- embryos (Fig. 2F”). This suggests that anterior cell elongation is a consequence of cephalic furrow formation, which is defective in eve mutants¹⁵ (Supp. Fig. 3A). In both Kruppel- and even-skipped mutants (Also kni hb, Supp. Fig. 3H), the central to posterior gradient of cell elongation is maintained but the rate of cell shape change is increased (Fig. 2 F’,F”). The presence of a gradient of AP cell elongation in wild-type and all AP mutants examined supports the idea that cell shape change is a passive response to the same extrinsic force.

This extrinsic force could be an AP pulling force or a DV pushing force. To investigate this, we examined the posterior domain of the germ band in Kruppel- mutants, a region with strong deformation (Fig. 3A-C). If cells deform passively in response to mechanical forces, they should behave qualitatively like a passive cellular material such as a foam, in which shape reflects stress¹⁶. Accordingly, different signatures are expected for pulls or pushes: if the tissue experiences a pull in one direction, it will tend to contract in the other two axes, leading to an increase in the cell projected area. Conversely if the tissue is compressed (for instance in the DV axis), cell area is expected to decrease. We found an increase in apical cell area as cells elongate in the posterior tissue of Kruppel- embryos (Fig. 3D,D’; see also E), suggesting that AP cell elongation in this domain results from an AP pull rather than a DV compression.

Figure 3. Spatio-temporal analysis of cell shape and cell area change in the ectoderm of wild-type and Kruppel- embryos.

A) The central and posterior ectodermal cell populations (shaded areas) were analysed separately (see also Supp. Fig. 3I,J). (B-C) Strain rates in the AP axis for total (black), cell intercalation (blue) and cell shape (red), for the central and posterior populations of ectodermal cells in wild-type and Kruppel- embryos. Wild-type is shown as dotted lines and Kruppel as a ribbon, with ribbon shading indicating a significant difference between the two genotypes (p<0.05). (D, D’) Representation of the average cell shape, cell area (quantified by colour coded scale) and cell orientation, according to the cell’s position along the AP axis (cell positions are given in µm from cephalic furrow) and time. Note that ectodermal cells only are analysed in (D) wild-type and (D’) Kruppel- embryos. After 15 minutes of GBE, the wild-type germ-band cells acquire an isotropic shape and maintain it, whereas the cells elongate in the AP orientation in Kruppel- mutants, especially at the posterior, where cell area is also greatest at this time. E) Close-up of movie frame of wild-type and Kruppel- embryos at mid-extension in the posterior domain, showing that whereas wild-type cells retain an isotropic shape, mutant cells are visibly elongated in the AP axis.

The analyses of cell shape and cell area also show that 10 minutes after GBE initiation, cells in wild-type embryos acquire an isometric apical shape which they maintain for the remainder of GBE (Fig. 3D, E). In Kruppel- mutants, however, cell shapes are briefly isometric at around 10 minutes then become elongated in the AP axis (Fig. 3D’, E). We hypothesise that in wild-type embryos, autonomous polarised cell intercalation relaxes the stress imposed externally on the germ-band, allowing stretched cells to return to isometric shapes. In Kruppel- embryos, the reduction in polarised intercalation caused by the defect in AP patterning upsets this balance, and cells stretch passively in the AP axis in response to tension. When this tension ceases, at about 25 minutes into GBE, the cells contract back for another 25 minutes, suggesting that the tissue relaxes as in an elastic response (Fig. 3C; see also Fig. 1D).

Other morphogenetic movements happening concurrently to GBE are good candidates to cause an antero-posterior tensile force. The sharp increase in the rate of cell shape change at the beginning of GBE overlaps with mesoderm invagination (Fig. 1B,D). Since twist is required for mesoderm specification and invagination^5,17–19, we analysed 5 movies of twist- mutant embryos (Fig. 4). Mesoderm invagination causes the cells on both sides of the furrow to become elongated towards the midline^5,11,20,21 (See also Fig. 3D,D’). The peak of cell elongation in the DV axis, which is lost as expected in twist- mutants (Fig. 4A’; see also 4F), is tightly temporally correlated with the onset of GBE (Fig. 4A and Supp. Fig. 1). Moreover, in twist- embryos where mesoderm invagination failed, the rate of both cell shape and cell intercalation strains in the AP axis are decreased in the first 20 minutes of GBE (Fig. 4B,C), even though twist- mutants ultimately extend their germ-bands as far as wild-type⁸ (Suppl. Fig. 2). This indicates that twist is required for the fast rate of extension at the beginning of GBE. In contrast to wild-type, small, non-invaginated ‘mesodermal’ cells are present in the field of view in twist- mutants. In order to compare the same population of cells in twist- and wild-type, we analysed the ectodermal cell population alone and confirmed these results (Fig. 2A and Supp. Fig. 4A,B). Given that ectodermal cells do not express twist, this demonstrates an indirect (non cell-autonomous) requirement for twist in the fast phase of GBE.

Figure 4. Relationship between mesoderm invagination and germ-band extension.

A) Tissue deformation in the DV axis (convergence, in grey) versus AP axis (extension, in black). The tissue deformation peak in the DV axis at the onset of germ-band extension corresponds to the ventro-lateral tissue stretching in the DV axis in response to mesoderm invagination. A’) As expected, this peak of DV stretching is absent in twist- embryos. B) Tissue deformation in the AP axis for 5 homozygous twist- embryos in which there is no mesodermal invagination (total strain rate in black, cell shape in red, cell intercalation in blue). Dotted lines show the wild-type data for comparison. Ribbon shading indicates a significant difference between mutant and wild-type (p<0.05). C) Cumulative representation of the same data, with lighter colours for corresponding wild-type data. D) Strain rates for cell intercalation contributing to tissue extension in twist- embryos are represented as a function of the cell’s position along the AP axis. E) Similar analysis for cell shape strain rates. F) Representation of averages for cell shape, area and orientation, along the AP axis and throughout germ-band extension for twist- embryos.

Cell intercalation in twist- mutants is decreased almost uniformly along the AP axis (Fig. 4D), but cell shape change is not (Fig. 4E). The anterior pulse of cell elongation is maintained, as expected since twist mutants have normal cephalic furrows, but the steep gradient of cell shape change towards the posterior is substantially reduced. Thus twist is required non cell-autonomously for cell shape change that is patterned in the AP axis. Since twist is not known to regulate AP patterning ^5,22,23, this suggests that the cell shape change gradient and the AP tensile force depends upon the invaginating mesoderm itself. Moreover, ectodermal cells in twist- embryos maintain an isometric shape throughout extension despite reduced cell intercalation (Fig 4F), consistent with the hypothesis that the tensile force is attenuated in this mutant.

The non-invaginated ‘mesodermal’ cells in twist mutants could act as a brake and decrease an AP tensile force generated by morphogenetic movements elsewhere in the embryo. Alternatively, the invaginating mesoderm could directly cause the AP tensile force deforming the germ-band in the fast phase. Note that AP cell elongation in wild-type and AP-patterning mutants embryos might also result from a relaxation of the ectoderm stretched along DV during mesoderm invagination. However, tissue relaxation is unlikely to explain the observed increase in cell area and cell elongation in the posterior of the germ-band, suggesting that here an AP tensile force predominates.

Aside mesoderm invagination, three other morphogenetic movements could conceivably produce the stress deforming the germ-band: cephalic furrow formation, posterior midgut invagination and amnioserosa cell elongation. While cephalic furrow formation does correlate with the pulse of cell elongation in the anterior domain (Fig. 2F-F”), the central to posterior gradient of cell elongation is intact in eve mutants, which do not have cephalic furrows. Posterior midgut invagination (PMI), which initiates on the dorsal posterior side, could pull the ectodermal tissue around the posterior of the embryo⁵. However, analysis of extension rates in a folded gastrulation (fog) mutant which blocks PMI formation¹⁹, shows that tissue deformation, cell intercalation and cell shape change are indistinguishable from wild-type during the first 20 minutes of extension (Supp. Fig. 5). Finally, DV elongation of the amnioserosa cells on the dorsal side²⁴ could push the ectodermal tissue towards the midline. However, cell area measurements suggest that AP tension rather than DV compression explains cell deformation in the posterior domain (See Fig. 3D’, E). Based on the above, we propose that mesoderm invagination is the main source of AP tensile force.

In support of an instructive role for the invaginating mesoderm, we found that the germ-band extends fastest close to the midline during the fast phase, suggesting an axial pull (Supp. Fig. 4 C, D). A possible mechanism is that once invaginated, the mesoderm undergoes convergence and extension (through an AP-patterning independent mechanism), elongating the tube in the anterior-posterior axis. This convergence and extension of the mesoderm (which is consistent with earlier observations^5,19), could drag the midline cells and adjacent ectodermal tissue, as reported for amphibian gastrulation^14,25.

In summary, we have found in contrast to what was previously thought^6,8,13, that cell shape change contributes significantly to the fast phase of germ-band extension. While cell intercalation during GBE requires AP patterning^6–9, cell shape change is under DV patterning control since it is reduced in twist- mutants. Our analysis of cell shape change and cell intercalation shows that at the beginning of the fast phase, the decrease of cell intercalation in AP mutants is compensated by an increase in AP cell elongation such that the overall rate of tissue extension can remain initially the same. These findings parallel our earlier observations on notochord extension in zebrafish¹, another tissue previously thought to extend as a sole consequence of active convergence. In both studies, quantitative analyses reveal redundancy in the forces extending the body axis, highlighting the robustness of convergence and extension movements.

We propose that cell shape change are a passive response to an AP tensile force resulting from convergence and extension of the internalising mesoderm. Intriguingly, in addition to a decrease in the rate of cell shape change, the rate of intercalation is reduced by about 30% in the first 25 minutes of GBE in twist- mutants (Fig. 4B,C and Supp. Fig. 4A), even though the AP patterning system upon which the polarised exchange of cell neighbours depends is not affected in twist- mutants^5,8. One possible explanation is that the AP tensile force that normally stretches the ectodermal cells also increases the rate of polarised cell intercalation. Stretching cells in the AP axis could make it mechanically favourable for cell junctions to shorten in the orthogonal DV axis, as suggested by^26,27. DV junction shortening is the first step leading to cell neighbour exchange in the germ-band^6,7. Thus our analysis provides an opportunity to explore how the autonomous cell behaviours described previously^6–9 interact with mechanical forces generated by morphogenetic movements¹⁴.

Materials and Methods

Fly strains

Apical cell membranes were labelled with ubi-DE-CadherinGFP (abbreviated as DEcadGFP), a fusion between DE-Cadherin and GFP¹¹. We used the null mutant alleles eve[3], Kr[1], kni[10]hb[4], twist[1] and fog[S4]. To label cells in these mutant backgrounds, we made the recombinant chromosomes eve[3]DEcadGFP, Kr[1]DEcadGFP and twist[1]DEcadGFP and the balanced stocks DEcadGFP; kni[10]hb[4] and fog[S4]; DEcadGFP.

Movie acquisition

Embryos at the end of cellularisation were mounted on O₂-permeable membrane (Sartorius) and covered in Voltalef oil (Attachem). The embryo to image was rolled to have the ventral side in the field of view, and the stage was rotated to capture the equivalent field of view for each embryo, with the cephalic furrow just visible to the left and the ventral furrow bisecting the field of view horizontally (Fig 1A’). The tissue which undergoes polarised intercalation at gastrulation is the ventro-lateral ectoderm, and this tissue is about 40 cell diameters long in the AP axis, and 20 to 23 cell diameters wide in the DV axis at the beginning of GBE¹³. Our field of view captures 12-15 cell diameters on either side of the invaginated mesoderm (midline) at the onset of GBE (Fig 1A’, see also Supp. Movies 2-4). This corresponds to the ventral-most 60% of the ventro-lateral ectoderm on either side of the midline. Homozygous mutant embryos were identified by the absence of GFP expression from the balancer chromosome CyOKrGal4UASGFP²⁸ and by their phenotypes. Note that twist null mutants show variable mesoderm invagination defects which range from very shallow furrows to almost wild-type like deep furrows¹⁸. To ensure a consistent phenotype, we analysed 5 twi mutant embryos exhibiting the most severe of these phenotypes. Movies were recorded at 20.5°C +/-1 °C, measured using a high-resolution thermometer (Checktemp1). Time points were taken every 30 seconds on a MRC1024 Biorad confocal coupled to an upright Nikon Eclipse E800 microscope (objective: 40X oil Plan/Fluor NA=1.3). For each time point, a z-stack of 10 optical slices separated by 1µm were taken starting from the apical surface of the cells (Supp. Movies 2-4). After each movie, embryos were left to develop to check that they survive normally until the end of embryogenesis. In the case of mutant embryos, the cuticle was mounted in Hoyers medium to verify that it showed the expected phenotype.

Movie tracking

The confocal Z stacks were converted into stacks of curved 2D representations, the outermost of which followed the surface of the embryo. The section giving the clearest view of cell apices was selected for tracking. Tracking software identifies cells and links them in an iterative process using a watershedding algorithm¹². For each cell at each timepoint, the program stores the coordinates of the centroid, and those of a polygon describing the cell as well as information about the cell lineage (Supp. Movies 5-7).

Movie analysis

Using the relative movement of cell centroids, the strain rate for a small domain of tissue composed of one corona of cell neighbours is calculated for each cell and for each 2 minute interval (corresponding to 4 movie frames)¹² (Supp. Movies 8-10). A direct measure of cell shape change is calculated by first approximating each cell from this small region to an ellipse, then finding the strain rates that best map a cell’s elliptical shape as it evolves over time (Supp. Movies 11-13). The difference between the total strain rate and cell shape strain rate is attributed to cell intercalation¹² (Supp. Movies 14-16). These strain rates are then projected onto the embryonic axes, AP and DV. Cell area change is calculated as the mean of the AP and DV cell shape strain rates.

Intragenotype movie synchronization

Within the same genotype, movies were synchronized in most cases by making a quantitative comparison, using a given threshold of proportional change per minute for instantaneous proportional extension in the AP axis. We use a synchronization threshold of 0.01 for wild-type and Kruppel- and 0.005 for twist- embryos (which have a slower initial rate of extension). In the case of even-skipped- embryos, we made a qualitative comparison, using the peaks of extension and convergence to synchronize the movies.

Intergenotype movie synchronization

Average strain rate curves weighted by planar cell area were calculated for each genotype. Genotypes were then synchronized using the time zero of average instantaneous extension in the AP axis. Note that contrarily to mutants embryos analysed in this study, most wild-type embryos roll on their sides during filming, around 40 minutes after the onset of germ-band extension (See Fig. 1B). Thus the field of view incorporate progressively more lateral cells from this point on, in wild-type compared to the mutants analysed.

Statistical validation

To test for evidence of differences between data derived from different embryos, we constructed a mixed-effects model²⁹ using the R software³⁰. We estimated the p-value associated with a fixed effect of differences between genotypes, allowing for random effects contributed by differences between embryos within a given genotype. Error ribbons show the typical standard error for data from one embryo, averaged across all embryos of that group³¹. Sections of ribbons are shaded when p < 0.05, signifying evidence for a difference between genotypes.