Patterned invagination prevents mechanical instability during gastrulation

Abstract

Mechanical forces are crucial for driving and shaping tissue morphogenesis during embryonic development^1-3. However, their relevance for the evolution of development remains poorly understood⁴. Here we show that an evolutionary novelty of fly embryos-the patterned embryonic invagination known as the cephalic furrow^5-7-has a mechanical role during Drosophila gastrulation. By integrating in vivo experiments and in silico simulations, we demonstrate that the head-trunk boundary of the embryo is under increased compressive stress due to the concurrent formation of mitotic domains and germ band extension and that the cephalic furrow counteracts these stresses, preventing mechanical instabilities during gastrulation. Then, by comparing the genetic patterning of species with and without the cephalic furrow, we find evidence that changes in the expression of the transcription factor buttonhead are associated with the evolution of the cephalic furrow. These results suggest that the cephalic furrow may have evolved through the genetic stabilization of morphogenesis in response to the mechanical challenges of dipteran gastrulation. Together, our findings uncover empirical evidence for how mechanical forces can influence the evolution of morphogenetic innovations in early development.

Fig. 1. Formation of ectopic folds in cephalic furrow mutants.

a, Timing of developmental events in wild type and prd, btd and eve mutants. b, Lateral view of btd and eve mutants during gastrulation. Controls are heterozygote siblings. The percentages indicate the extent of germ band extension. Arrows indicate tissue folds; asterisks indicate mitotic domains. Scale bars, 50 µm. c, Profile view of the head–trunk epithelium around stage 8. Scale bars, 20 µm. d, Timing of wild-type cephalic furrow formation compared with btd (P = 0.283), eve (P < 0.001) and prd heterozygotes, and to ectopic folds in btd (P = 0.001), eve (P = 0.001) and prd (P = 0.003) homozygotes. e, Percentage of germ band extension at cephalic furrow formation in wild type compared with ectopic folding in btd, eve and prd homozygotes (P < 0.001). f, Position variability of cephalic furrow formation and ectopic folding in btd heterozygotes and homozygotes, respectively. Scale bars, 50 µm. g, Folded area (yellow outline) of the cephalic furrow (CF) and ectopic folds (EFs) in btd mutants. The numbers show average area and standard deviation. Cartographic projections of lateral views. Scale bars, ≈50 µm. h, Total folded area of wild-type cephalic furrow compared with btd (P = 0.133) eve (P < 0.001) and prd (P < 0.001) heterozygotes, and to ectopic folds in btd, eve and prd homozygotes (P < 0.001 for each). The membrane marker in panels b,c,f,g is Gap43–mCherry. Embryonic staging is based on standard developmental tables^,. Source data

Fig. 2. Role of mitotic domains and germ band in ectopic folding.

a, Folding positions (black lines) relative to mitotic domains (MD) in wild-type, btd and eve embryos. Scale bars, 50 µm. b, Apical cell area during ectopic folding. The white outlines highlight a subset of non-dividing (orange) and dividing (blue) cells. Scale bars, ≈20 µm. c, Mitotic expansion preceding ectopic folding (arrow) in the btd mutant. Scale bar, 20 µm. d, Time of cephalic furrow (CF) and ectopic fold (EF) formation relative to mitotic expansion in wild-type (P < 0.001), heterozygote (P = 0.002) and homozygote (P = 0.002) embryos. e, Lateral and profile views of btd–stg double mutants. Scale bars, 50 µm. f, Tissue compression around the head–trunk (HT) and trunk–germ (TG) regions (white outlines) in the btd mutant. Scale bars, 50 µm. GB, germ band. g, Strain rate at the head–trunk region in btd heterozygotes (n = 3) and homozygotes (n = 3; combined isotropic and anisotropic). Filled triangles denote the formation of CF, MD and EFs, and empty triangles indicate strain rate peaks. A–D refer to frames from Supplementary Fig. 2b. MD_i, metaphase; MD_ii, telophase. h, Dynamics after trunk–germ laser cuts in wild-type embryos. The tracks (rainbow) show the distance between cell vertices (solid white line) in control (n = 3) and ablated (n = 3) embryos. The dashed white line indicates the cut location. Scale bars, 20 µm. The line and shaded area represent mean and 95% confidence interval, respectively (g,h). i, Germ band cauterization (orange circle) in the eve mutant under light-sheet microscopy. Scale bar, 50 µm. j, Compressed non-dividing cells between mitotic domains (from panel i). Scale bar, 50 µm. k, Germ band cauterizations in eve mutants with traces of epithelial deformations over time. Scale bars, 20 µm. l, Tortuosity of epithelial traces in non-cauterized (n = 3) and cauterized (n = 4) eve mutants. m, Summary of the live experiments in cephalic furrow mutants. Source data

Fig. 3. Model and simulations of folding mechanics.

a, Region of interest of the model. b, States of the different components based on particles connected by springs. c, Energy equation describing the stretching and bending components and the dimensionless bending rigidity. d, Energy and folding dynamics in simulations. The black line and shaded area represent the mean and standard deviation across simulations (t) (n = 20). The dashed blue line indicates the peak of bending energy, and the dashed pink line denotes the last iteration. Energy values are normalized by the initial total energy. e, Parameter sweep for cephalic furrow mutants without mitotic domains. The heatmap shows the average number of folds for different bending rigidities ($K_b^{*}$) and percentages of germ band extension (g). Outlined in white are conditions without folding (i) and with most folding events (ii). Representative simulations are rendered below. f, Parameter sweep for cephalic furrow mutants with mitotic domains. Outlined in white is a condition with folding events without germ band extension (iii). g, Simulations testing the effect of the cephalic furrow on ectopic folding in three conditions: only mitotic domains (top row), mitotic domains and cephalic furrow (middle row), and delayed mitotic domains and cephalic furrow (bottom row). The added delay mimics the relative timing in wild type. $t_{\mathit{MD}}=1$ corresponds to 10⁵ computational timesteps. The cephalic furrow is ${\kappa }_o^{\mathit{CF}}=2.0$. The black line and shaded area represent the mean and standard deviation across simulations (n = 20). Red bars indicate the position of mitotic domains. The green bar indicates the position of the cephalic furrow. EL, egg length. h, Simulations testing the effect of cephalic furrow position on ectopic folding. Representative samples with the cephalic furrow more anterior (top row), central (middle row) and posterior (bottom row) along the anteroposterior axis. The black line and shaded area represent the mean and standard deviation across simulations (n = 20). Red bars represent the position of mitotic domains. The green bar indicates the position of the cephalic furrow.

Fig. 4. Genetic patterning of the head–trunk boundary in Drosophila, Ceratitis, Anopheles and Clogmia.

a, Expression of btd, eve and slp1 before gastrulation in Drosophila. Early slp1 and eve transcripts demarcate the head–trunk boundary and resolve to sharp stripes with btd transcripts at the interface. The numbers 1 and 2 in orange indicate eve stripe 1 and 2, respectively. Scale bars, 50 µm. ht, btd head–trunk domain; ic, eve stripe 1 with initiator cells; sh, slp head domain. b, Expression patterns at the onset of gastrulation in Drosophila (lateral view). slp1 stripes demarcate the outer edges of the cephalic furrow (dashed lines). Scale bar, 50 µm. ac, btd acron domain. c, Expression patterns at the onset of gastrulation in Drosophila (profile view). eve-expressing initiator cells also express btd and are abutted by slp1 stripes. prd is offset from slp1 by the one-cell row. The dashed lines demarcate the outer edges of the cephalic furrow. Scale bar, 20 µm. d, Expression patterns of the invaginated cephalic furrow in Drosophila. Scale bar, 20 µm. e, Schematic of the combinatorial expression at the head–trunk boundary of Drosophila. f, Expression of btd, eve and slp during nuclear cycles 13 (left) and 14 (right) in Ceratitis, Anopheles and Clogmia embryos. Scale bars, 100 µm. g, btd–eve overlap at the head–trunk boundary of different dipterans; it is present in species with a cephalic furrow (Drosophila and Ceratitis) and absent in species without (Anopheles and Clogmia). Scale bars, 50 µm. fg, btd foregut domain. h, Expression patterns at the onset of gastrulation in Ceratitis, Anopheles and Clogmia (profile view). Scale bars, 20 µm.

Fig. 5. Interplay between genetics and mechanics during cephalic furrow evolution.

a, Cephalic furrow traits mapped onto a simplified dipteran phylogeny (based on ref. ). Germ band extension and mitotic domains are ancestral, suggesting that compressive stresses at the head–trunk boundary were present since the dawn of Diptera. The cephalic furrow is present in the common ancestor of Megaselia and Drosophila (cyclorrhaphan flies) and correlates with the presence of a btd–eve overlap at the head–trunk boundary. Out-of-plane divisions are present at the head–trunk boundary of the non-cyclorrhaphan flies Clogmia and Chironomus. Data sources are annotated with geometrical symbols: this study (black circle), Dey, Kaul, Kale et al. (black triangle), and Eritano et al. and Vincent et al. (black square). b, Evolutionary scenario for the origin of morphogenetic innovations in Diptera. In the ancestral state (short germ), there was no mechanical instability at the head–trunk boundary during gastrulation (0). The appearance and concurrent formation of mitotic domains and germ band extension increased the compressive stresses and ectopic folding around the head–trunk boundary (1). As epithelial instability could be detrimental to developmental robustness and individual fitness, morphogenetic processes mitigating these effects may have been favoured by natural selection. In this light, the out-of-plane divisions (2a) and the cephalic furrow (2b) might have evolved in response to mechanical instability, representing independent solutions to a common challenge. For the cephalic furrow, this might have occurred through the genetic stabilization of ectopic folds into a patterned embryonic invagination.

Extended Data Fig. 1. Disruption of initiator cell behavior in cephalic furrow mutants.

a, Profile view of the head–trunk boundary epithelium in wildtype embryos and prd, btd, and eve mutants. Samples synchronized by the end of cellularization (0.0 min). Arrow indicates the first infolding of the tissue. Initiator cells (IC) in wildtype embryos are tightly connected to adjacent cells, which become arched in the early invagination of the cephalic furrow (CF). This arrangement is perturbed in mutants. In prd, adjacent cells do not arch over the initiator row (6.1 min), and the invagination is delayed. In btd, there is no initiator shortening, only a partial apical constriction bulging the epithelium (6.0 min). In eve, there is no shortening or apical constriction. Both btd and eve form ectopic folds (EF) about ten minutes after the end of cellularization. Scale bars = 20 µm. b, Surface view of the head–trunk boundary epithelium on 2D cartographic projections showing abnormal apical constriction in prd and btd mutants and absence of this behavior in eve mutants. Scale bars ≈ 20 µm.

Extended Data Fig. 2. Differences between cephalic furrow (CF) and ectopic folds (EF).

a, Position variability in eve heterozygotes (n = 6) and homozygotes (n = 5). Scale bars = 50 µm. b, Folded area (yellow outline) in eve heterozygotes (n = 7) and homozygotes (n = 4). Numbers indicate average and standard deviation. Cartographic projections of lateral views. Scale bars ≈ 50 µm. c, Folding dynamics in eve mutant. d, Folding dynamics in btd mutant. e, Folding angle and tortuosity in btd mutants (n = 6). f, Maximum fold depth in btd heterozygotes (n = 32) and homozygotes (n = 7) and eve heterozygotes (n = 20) and homozygotes (n = 4). Ectopic folds are shallower than the cephalic furrow (p < 0.001). Data points correspond to individual folds. g, Area of ectopic folds in wildtype (n = 16), heterozygotes (n = 9), and homozygotes (n = 14) (pooled btd, eve, and prd). Ectopic folds in wildtype are smaller than in mutants (p < 0.001). h, Multiple ectopic folds (arrows) near dividing cells (asterisks) in btd mutant. Scale bar = 20 µm. i, Proportion of fold types in wildtype (n = 36), btd heterozygotes (n = 33) and homozygotes (n = 13), eve heterozygotes (n = 26) and homozygotes (n = 10), prd heterozygotes (n = 26) and homozygotes (n = 14), stg heterozygotes (n = 33) and homozygotes (n = 13). j, Proportion of ectopic folding positions (anterior, middle, posterior) in wildtype (n = 28), btd heterozygotes (n = 6) and homozygotes (n = 12), eve heterozygotes (n = 7) and homozygotes (n = 10), prd heterozygotes (n = 7) and homozygotes (n = 10), and stg heterozygotes (n = 12) and homozygotes (n = 3). k, Lateral view of wildtype and prd embryos. Scale bar = 50 µm. l, Folded area (yellow outline) in wildtype (n = 16) and prd (n = 14) embryos. Numbers indicate average and standard deviation. Cartographic projections of lateral views. Scale bars ≈ 50 µm. m, Ectopic folding in wildtype embryo from l. Scale bar = 10 µm. n, Folded area by fold type in wildtype (n = 16), btd heterozygotes (n = 6) and homozygotes (n = 5), eve heterozygotes (n = 7) and homozygotes (n = 4), prd heterozygotes (n = 9) and homozygotes (n = 5). Source data

Extended Data Fig. 3. Additional in vivo experiments in cephalic furrow mutants.

a, Formation of the cephalic furrow (CF) and mitotic domains (MD) in wildtype embryos. Arrows indicate tissue folds. Asterisks indicate mitotic domains. Scale bars = 50 µm. b, Formation of the cephalic furrow in stg mutants. Scale bars = 50 µm. c, Formation of mitotic domains and ectopic folds (EFs) in btd mutants. Scale bars = 50 µm. d, Absence of ectopic folds in btd–stg double mutants. Scale bar = 50 µm. e, Germ band cauterization (orange) in wildtype embryo under multiview lightsheet microscopy. We quantified and corroborated the phenotype under confocal microscopy (see f). Scale bar = 50 µm. f, Cephalic furrow in non-cauterized and cauterized wildtype embryos. Scale bar = 20 µm. g, Germ band cauterizations in btd mutants under confocal microscopy showing the traces of epithelial deformations over time. Scale bars = 20 µm. h, Tortuosity of epithelial traces in btd non-cauterized (n = 2) and cauterized (n = 3). Plots show mean predicted values from regression with a 95% confidence interval shaded band. Source data

Extended Data Fig. 4. Model properties and parameter sweeps.

a, Reference embryonic proportions in wildtype and cephalic furrow mutants used for the model. Sizes and positions of embryonic traits are relative to embryo length. b, Example simulation ($K_b^{*}=7\times {10}^{-5}$ and $g=0.3$) showing the tissue shape at $t=19$ (blue) and $t=1000$ (pink). These timepoints are marked as dashed lines in the descriptive plots below. $t=1$ corresponds to ${10}^5$ computational steps. The X axis is in $\mathit{lo}{g}_{10}$ scale to improve the visualization. c–e, Parameter sweeps without mitotic domains for the number of folds by bending rigidity using $g=0.3$ (c), the number of folds by germ band extension ($g$) using $K_b^{*}=1.0\times {10}^{-4}$ (d), and the timing of folding by germ band extension using $K_b^{*}=1.0\times {10}^{-4}$ (e). The plot shows the mean value and standard deviation across simulations (n = 20). f–h, Parameter sweeps with mitotic domains for the same conditions as above. i, Parameter sweeps for cephalic furrow formation for different values of ${\kappa }_o^{\mathit{CF}}$ (colors) and germ band extension. j, Fine-grained parameter sweep of ectopic folding at different ${\kappa }_o^{\mathit{CF}}$ values with the simultaneous formation of the cephalic furrow and mitotic domains ($t_{\mathit{MD}}=0$). k, Fine-grained parameter sweep of ectopic folding at different ${\kappa }_o^{\mathit{CF}}$ values with a relative delay between cephalic furrow and mitotic domain formation ($t_{\mathit{MD}}=5$). Values of ${\kappa }_o^{\mathit{CF}}$ are shown in units of $1/L$. $t_{\mathit{MD}}=1$ corresponds to 10⁵ computational timesteps. l, Representative simulations from Fig. 3g at 0 and 20% of germ band extension.

Extended Data Fig. 5. Analyses of slp mutants in Drosophila.

a, Lateral view of slp mutants at the onset of initiator cell behavior. Cephalic furrow formation is delayed and shifted forward. Scale bars = 50 µm. b, Position of the cephalic furrow (CF) and germ band (GB) in slp mutants. CF is displaced anteriorly, and GB is further extended (p < 0.001) between heterozygotes (n = 26) and homozygotes (n = 7). c, Expression of btd and eve in slp mutants before gastrulation. Scale bars = 50 µm. d, Expression of btd and eve in the head–trunk boundary of slp mutants. Distance between eve stripe 1 and 2 (dashed lines) is larger in slp embryos. Asterisk indicates ectopic expression of btd between acron (ac) and head–trunk (ht) domains. Scale bars = 50 µm. e, Dorsal view of expression domains in slp mutants. sh, slp1 head domain. Asterisks indicate ectopic expression of btd. Scale bars = 20 µm. f, Profile view showing the increased number of cell rows between eve stripe 1 and 2 in slp mutants. ic, initiator cells. Scale bars = 20 µm. g, Behavior of initiator (orange) and dividing (blue) cells in slp mutants. Scale bar = 20 µm. h, Profile view showing the asymmetric cephalic furrow in slp mutants. Scale bar = 20 µm. Source data

Extended Data Fig. 6. Genetic patterning of the head–trunk boundary in Drosophila.

a, Expression of slp1, eve, and btd in wildtype from nuclear cycle (nc) 11 to gastrulation. Scale bars = 50 µm. b, Expression of prd, slp1, and eve in wildtype embryos. Scale bars = 50 µm. c, Expression of prd, slp1, and eve in wildtype embryos in a lateral view of the head and a profile view of the head–trunk epithelium. Scale bars = 20 µm.

Extended Data Fig. 7. Disruption of gene expression patterns at the head–trunk boundary of cephalic furrow mutants in Drosophila.

a, Expression of btd, eve, slp1, and prd in btd, eve, and prd mutants. Asterisks indicate altered expression patterns. Stripe numbers are color-coded for each gene. Scale bars = 50 µm. b, Profile views of g showing the altered gene expression patterns (asterisks) of epithelial cells at the head–trunk boundary of cephalic furrow mutants. Scale bars = 20 µm. c, Lateral views of btd, eve, and slp1 expression in btd, eve, and prd mutants after gastrulation. Scale bars = 50 µm. ac: btd acron domain, sh: slp head domain, ht: btd head–trunk domain.

Extended Data Fig. 8. Genetic patterning of the head–trunk boundary in other dipteran species.

a, Expression of slp1, eve, and btd in Ceratitis developmental stages before gastrulation. Scale bars = 100 µm. b, Expression of slp2, eve, and btd in Anopheles developmental stages before gastrulation. Scale bars = 50 µm. c, Expression of slp1, eve, and btd in Clogmia developmental stages before gastrulation. Scale bars = 50 µm. d, Expression of slp1, eve, and prd in Clogmia before gastrulation. Scale bars = 50 µm. e, Expression of slp1, eve, and prd at the head–trunk boundary of Clogmia showing a lateral and profile views of the epithelium. Stripe numbers are color-coded for each gene. Scale bars = 20 µm.

Extended Data Fig. 9. Extended summary figure of cephalic furrow evolution including data from Anopheles and Ceratitis.

Please refer to the legend of Fig. 5 for the full details. a, Cephalic furrow traits mapped onto a simplified dipteran phylogeny. b, Evolutionary scenario for the origin of morphogenetic innovations in Diptera.