Research ArticleDevelopment

Mechanical Signals Trigger Myosin II Redistribution and Mesoderm Invagination in Drosophila Embryos

See allHide authors and affiliations

Sci. Signal.  14 Apr 2009:
Vol. 2, Issue 66, pp. ra16
DOI: 10.1126/scisignal.2000098


During Drosophila gastrulation, two waves of constriction occur in the apical ventral cells, leading to mesoderm invagination. The first constriction wave is a stochastic process mediated by the constriction of 40% of randomly positioned mesodermal cells and is controlled by the transcription factor Snail. The second constriction wave immediately follows and involves the other 60% of the mesodermal cells. The second wave is controlled by the transcription factor Twist and requires the secreted protein Fog. Complete mesoderm invagination requires redistribution of the motor protein Myosin II to the apical side of the constricting cells. We show that apical redistribution of Myosin II and mesoderm invagination, both of which are impaired in snail homozygous mutants that are defective in both constriction waves, are rescued by local mechanical deformation of the mesoderm with a micromanipulated needle. Mechanical deformation appears to promote Fog-dependent signaling by inhibiting Fog endocytosis. We propose that the mechanical tissue deformation that occurs during the Snail-dependent stochastic phase is necessary for the Fog-dependent signaling that mediates the second collective constriction wave.


Embryos develop through the biochemical processes that control the patterning and geometrical morphogenesis of multicellular tissues (14). Animal morphogenesis initiates at gastrulation, a process of genetically controlled active shape changes of the tissue that lead to the formation of the ectoderm, endoderm, and mesoderm layers. In Drosophila embryos, gastrulation begins with mesoderm invagination, which is followed by germ-band extension, and both of these processes are morphogenetic movements that are regulated by polarized distribution of Myosin II (MyoII, encoded by the zipper gene) in the submembrane cortex of the motor cells generating the movements (57). The mechanical strains associated with morphogenesis also serve as signals that feedback into biochemical processes that regulate patterning and into the gene expression profiles that regulate development (811). Here, we report that the mechanical events that occur at the earliest stages of gastrulation also provide a signal that feeds into the posttranslational events actively controlling morphogenesis. The mechanical cues trigger the concentration of MyoII at the apical side of the cells, which constricts the cells to produce mesoderm invagination.

The genetic network that controls the Drosophila embryo mesoderm invagination is activated by the maternally supplied transcription factor Dorsal, which translocates to the nucleus to initiate the invagination process. Dorsal stimulates the expression of the genes Twist (twi) and Snail (sna), which encode transcription factors, in the mesodermal cells (2). Both factors are involved in the genetic control of the cell shape changes necessary for the two constriction waves that lead to mesoderm invagination (1215). Flies with mutations in twi are defective in the second constriction wave owing to the deficiency of a secreted signal protein Fog, which is necessary for stable redistribution of the molecular motor MyoII to the apical side of the cells that triggers the coordinated collective constriction (5, 1517). Flies with mutations in sna, which still express fog (18), are defective in both the first stochastic waves and in the second collective wave. Thus, Snail and Fog are together necessary for the second collective constriction wave (19). Here, we report the existence of a mechanical cue that controls Fog signaling, MyoII redistribution, and constriction of the cells along the apical surface of the embryo, which provides a mechanistic scenario for the interplay between Snail and Fog in controlling mesoderm invagination. The Snail-dependent stochastic phase produces a mechanical signal that feeds into the Twist-dependent collective constriction phase.


Mechanical rescue of mesoderm invagination and apical redistribution of MyoII in sna embryos

To test the hypothesis that a mechanical cue triggers MyoII redistribution, the sna homozygous mutant embryos, which lack the mechanical strain associated with the lack of the 4-min stochastic constriction phase (13), were subjected to a local 3- to 7-μm deformation with a micromanipulated needle. This 4-min indentation was applied to the mesoderm precisely 2 to 3 min after the end of ventral cellularization, the stage at which the first constriction wave would occur in wild-type embryos (Fig. 1A and fig. S1). Local mechanical indentation rescued a complete mesodermal invagination in 67% of the embryos (Figs. 1, B and C, and 2). No mesoderm invagination was observed in nonindented sna homozygous embryos, and ventral flattening was only observed in 25% of cases (Fig. 1C and fig. S2). The invagination initiated 2 to 3 min after indentation and propagated all along the mesoderm.

Fig. 1

Mechanical rescue of mesoderm invagination in sna mutant embryos. (A) Indentation of homozygous sna embryos. The top shows the arrangement of the single-cell layer with the nuclei of the cells at the apical side of the embryo. The black circles represent cell nuclei. (B) Mesodermal furrow invagination at stage 7 (black arrow), reflecting mesoderm invagination, occurs in response to mechanical indentation of 67% of sna embryos (n = 15). Embryos are observed in phase contrast. Anterior invagination delay and lack of mesoderm sagittal compression at stage 6 are phenotypes used to select sna homozygotes (see Experimental Procedures section) (fig. S1). (C) Penetrance of the rescue of mesodermal invagination (denoted I) and MyoII redistribution (denoted M) by mechanical disruption. Ventral flattening was the only evidence of partial mesoderm invagination noted in 25% of cases for the sna/sna embryos in the absence of mechanical deformation and these embryos only exhibited slight redistribution of MyoII (n = 8). Number of cases tested in other configurations are sna/sna indented 2 to 3 min after cellularization (n = 15), sna/sna indented 10 min after cellularization (n = 13), sna/sna global deformation at stage 6 (n = 14), sna/sna touched embryos (n = 3), sna/sna heat-shocked embryos (n = 10), twisna/twisna indented (n = 8), twisna/twisna controls (n = 5), twisna/twi indented (n = 3), twisna/twi controls (n = 5), twisna/twi;PE-Fog controls (n = 16), twisna/twi;PE-Fog indented (n = 12), snashi/snashi 20°C (n = 4), snashi/snashi 30°C (n = 7).

Fig. 2

Mechanical rescue of apical redistribution of MyoII in sna mutant embryos. Apical redistribution of MyoII and mesoderm invagination is evident at stage 6 in wild-type (WT) embryos. These phenomena are not observed in sna/sna embryos. Local ventral indentation fully restores apical redistribution of MyoII and mesoderm invagination in sna/sna embryos. When embryos were indented 2 to 3 min after cellularization (n = 15), 67% show this full invagination rescue phenotype, with 26% of the embryos showing a partial or small invagination response and 7% not responding, and with 93% showing apical redistribution of MyoII (red arrows). When embryos were indented 10 min after cellularization (n = 13), 38% show this full invagination rescue phenotype, with a partial or small active invagination in 62% of the embryos, and with 100% showing apical redistribution of MyoII (red arrows). Global deformation by lateral uniaxial compression restores MyoII apical redistribution and unclosed invagination (red arrows) (n = 14; 64% show this phenotype). Anterior invagination delay (yellow arrows) and lack of mesoderm sagittal compression at stage 6 are phenotypes used to select sna homozygotes (see Experimental Procedures section) (fig. S1). To capture mesoderm cells, embryos were imaged at the most ventral cells exhibiting invagination or at most ventral cells in embryos that failed to invaginate. The fluctuations in the cytoplasmic apical intensity of MyoII on the dorsal side of the embryo, which varies as a function of small variations in the orientation of the embryo, are due to variations in the depth of the focal plane imaged to find the apical border of the invaginating or noninvaginating most ventral cells. The white stars denote the position of the local indent. Positions 1 and 2 show the propagation of MyoII redistribution and mesoderm invagination all along the mesoderm from the point of indent.

The distribution of MyoII in wild-type, sna/sna, and indented sna/sna embryos was investigated by immunolabeling (Fig. 2) and is summarized in Fig. 1C. Embryos were considered positive for apical redistribution of MyoII if the protein concentrated at the apical membrane cortex of the mesodermal cells, leading to a thin concentration of signal at the membrane (5). Similarly to the mesodermal cells of the wild type (Fig. 2), MyoII became concentrated in the apical portion of the cells in 93% of indented sna/sna embryos (Fig. 2). No apical redistribution of MyoII was observed in 75% of the unperturbed sna embryos (Fig. 2) (with the remaining 25% exhibiting only a slight redistribution of MyoII and ventral flattening; fig. S2).

When the mesoderm was locally indented 10 min after the end of ventral cellularization (equivalent to the mid-mesoderm invagination stage for a wild type), the indented embryo showed a partial or small active invagination in 62% of the embryos and a complete invagination in 38% of cases, with apical redistribution of MyoII occurring only in the invagination domains in all cases (Fig. 2). When the mesoderm was touched by the needle without deformation, neither mesoderm invagination nor apical redistribution of MyoII was observed, showing that mechanical deformation of mesoderm tissues is required to rescue the sna phenotype (fig. S3). Thus, a mechanical cue rescues apical MyoII redistribution and mesoderm invagination in sna mutant embryos, with the most efficient rescue occurring when the mechanical indentation is applied to the mesodermal cells at stages equivalent to that of the Snail-dependent endogenous deformation in the wild type.

As MyoII apical relocation was observed in the mesoderm only, we investigated whether this feature was specific to exogenous indentation of the mesoderm or if global deformation of the tissue could trigger apical redistribution of MyoII outside of the mesoderm in sna mutants. Applying a 4-min-long global uniaxial deformation to the sna mutant embryo (10) at the beginning of stage 6, which increases the embryo’s dorsoventral size 10 to 20%, led to MyoII apical redistribution in mesodermal cells only and to the rescue of a thin and unclosed invagination in 64% of the embryos (Figs. 1C and 2).

Fog in the mechanical rescue of apical redistribution of MyoII in sna embryos

Because the mechanical activation of MyoII redistribution that drives the invagination of sna mutants is restricted to the mesoderm cells, we investigated if the presence of Twist was necessary for the mechanical rescue of sna mutant invagination. After indentation of sna twi double mutants, neither apical MyoII redistribution (Fig. 3A) nor mesoderm invagination was rescued (Fig. 1C), showing the necessity of the Twist pathway for MyoII redistribution.

Fig. 3

The Fog signaling pathway is necessary for mechanical induction of apical redistribution of MyoII. (A) Apical redistribution of MyoII is not rescued in twisna homozygous embryos indented ventrally, with only MyoII remaining junctional and cytoplasmic, which is characteristic of stage 7 nonmesodermal cells (17). On the left side are control embryos that were not indented (n = 5, all embryos). The right side shows embryos indented 2 to 3 min after cellularization. Indented embryos failed to invaginate (n = 8, all embryos). (B) Left: twi sna/twi;PE-Fog control embryos that were noninvaginating (n = 16, all embryos). Right: Local ventral indentation rescues mesoderm invagination and MyoII apical redistribution in twi sna/twi;PE-Fog embryos all along the mesoderm (n = 12, all embryos) (red arrows). Anterior invagination delay (yellow arrows) and lack of mesoderm sagittal compression at stage 6 (not shown) are phenotypes used to select sna twi homozygous and sna twi/twi PE-Fog noninvaginating phenotypes (see Experimental Procedures section) (fig. S1). The white star denotes the position of the indent.

Because Fog is downstream of Twist and necessary for MyoII apical redistribution (5, 14), we investigated if Fog was required for MyoII apical redistribution in response to mechanical cues. We mechanically manipulated twi sna/twi;PE-Fog embryos, in which Fog expression, altered by twi mutations, is rescued under the control of the proximal element of the twi promoter (PE) in the mesoderm only (19). The expression of snail is strongly reduced in twist mutants at stage 5 (20); thus, we anticipated that the expression of snail in these embryos lacking both copies of twi and one copy of sna would be reduced enough to provide a background phenotype equivalent to a sna twi mutant at stage 6. Indeed, all Twist-labeled embryos deficient in Twist did not invaginate and showed no anterior midgut invagination from stage 6 to stage 8, as the sna twi mutants showed (fig. S4A). The Myo-labeled furrow formation–defective embryos lack any apical MyoII redistribution (Figs. 1C and 3B). Thus, forced expression of Fog in the mesoderm of embryos deficient in Twist and Snail fails to rescue invagination and MyoII apical redistribution in the mesoderm. In contrast, local deformation of these embryos in the mesoderm rescued ventral furrow formation and apical MyoII redistribution in all cases (Figs. 1C and 3B and fig. S4B). No such rescue was observed in the absence of forced expression of Fog in twi sna/twi indented embryos (Fig. 1C and fig. S5). Because twi sna/twi;PE-Fog embryos express Fog but do not express other targets downstream of twi, the Fog signaling pathway is the relevant target involved in the mechanotransduction pathway that stimulates MyoII apical redistribution. Therefore, mechanical strain and Fog are necessary and sufficient for the rescue of apical MyoII redistribution and mesoderm invagination in sna-deficient embryos.

Endocytosis in mesoderm invagination and apical redistribution of MyoII

Because Fog is required for mechanical cues to trigger MyoII apical redistribution, we investigated the underlying mechanism that translates the mechanical strain signal into activation of the Fog signaling pathway. Fog is an apically secreted protein that most likely signals by binding to a receptor (5, 14). That the Fog-receptor complex may be regulated by endocytosis and receptor recycling or ligand and receptor lysosomal degradation is also expected. The initiation of mesoderm invagination correlates with external blebs of apical membranes (16), consistent with an increase of mesoderm cell internal pressure associated with apical flattening and cell shape changes associated with constriction (21). These processes alter membrane tension and increase outward membrane curvature, which could block endocytosis and thereby enhance the Fog signaling pathway by blocking the endocytosis of the putative Fog-receptor complex (22, 23). shibire is a temperature-sensitive conditional mutant of dynamin and shifting flies with this mutation to 28°C rapidly and reversibly blocks endocytosis. Thus, we constructed a shi sna double mutant line to mimic the putative mechanical inhibition of endocytosis in mesodermal cells that would be lacking in the sna embryos because they lack the first constriction wave. After the shi sna embryos were laid at 20°C, they were shifted to 30°C for 5 min and then returned to 20°C for 15 min and then fixed. Double mutant homozygous embryos fixed at late stage 6 and stage 7 exhibited both a rescue of the MyoII apical redistribution and the invagination of the mesoderm, compared to homozygous embryos developing at 20°C in which only small fluctuations in mesoderm curvature initiated (Figs. 1C and 4A). No such rescue was observed in heat-shocked sna homozygous embryos (Fig. 1C and fig. S6A). Thus, blocking endocytosis rescues MyoII apical redistribution and mesoderm invagination in sna-defective mutants, suggesting that this may be a mechanism by which mechanical cues control the second wave of cell constriction that mediates mesoderm invagination.

Fig. 4

Fog and MyoII show apical accumulation of mesoderm cells in response to mechanical inhibition of endocytosis in early stage 6 embryos. (A) Blocking endocytosis of a temperature-sensitive dynamin mutant (shi) rescues apical redistribution of MyoII and mesoderm invagination of sna shi mutants (n = 7, all embryos) (red arrows). The left image shows a sna shi double mutant in which endocytosis was not blocked (n = 4, all embryos). On the right, endocytosis was blocked by transient heat shock. (B) Wild-type embryos at stage 5 show little apical cytoplasmic accumulation of Fog in mesoderm cells (red arrows). By mid stage 6, Fog appears at the plasma membrane (red arrows) and accumulates on the apical side of the cell. (C) Mid stage 6 sna embryos lack apical plasma membrane accumulation of Fog (red arrows), and Fog fails to redistribute at the apical membrane. Local indentation (red asterisk) at early stage 6 restores Fog accumulation at the plasma membrane of the mesoderm (red arrows) and apical redistribution. Indented sna embryos were imaged from stage 6 to stage 7. (D) Heat-shocked sna shi embryos at stage 6 (right) show plasma membrane accumulation of Fog (red arrows) and accumulation of cytoplasmic Fog at the apical side of the mesoderm, in contrast to stage 6 control embryos (left). Black asterisks denote the location of pole cells initiating dorsal movements at mid stage 6 (at 10 min after initiation of mesoderm invagination into the wild type). Anterior invagination delay and lack of mesoderm sagittal compression at stage 6 are phenotypes used to select sna homozygotes (see Experimental Procedures section) (fig. S1).

The intracellular distribution of Fog at stage 5 and stage 6 is consistent with a block in Fog endocytosis occurring as the mechanical signal is generated in response to the first stochastic cell constriction. Whereas at stage 5, Fog was diffusely localized throughout the cytoplasm of mesoderm cells with little apical concentration in all 12 embryos observed, by mid stage 6, embryos having initiated apical flattening and constriction showed an apical redistribution of cytoplasmic Fog toward the membrane, and Fog was visible at the apical cell surface that is directly adjacent to cytoplasmic concentration just beneath the plasma membrane in all 11 embryos observed (Fig. 4B). In contrast, the Fog distribution of mid stage 6 sna homozygous mutants embryos, which do not flatten and do not constrict, remained diffusely localized throughout the cytoplasm, with poor redistribution toward the apical side of the cell and no Fog at the cell surface in all 10 embryos observed (Fig 4C). Both the retention of secreted signaling proteins at the cell surface and the intracellular apical accumulation of secreted vesicles are consistent with a mechanical strain on the plasma membrane blocking endocytosis (22, 23) and driving exocytosis (24), leading to an increase of the association of Fog with the cell surface, presumably bound to its receptor, in early stage 6 wild-type embryos, but not in early stage 6 sna mutant embryos that are defective in mechanical deformations associated with apical flattening and cell constriction.

To confirm that a block in endocytosis altered Fog distribution in mesoderm cells, we analyzed Fog distribution after heat shock in the sna shi mutants. We found Fog accumulated on the apical side of the cells in all 10 embryos examined (Fig. 4D) and accumulated on the cell surface of apically flattened cells in 70% of the embryos examined after a 5-min incubation at 30°C at mid stage 6 (Fig. 4D). In contrast, sna shi embryos that were not subjected to heat shock failed to exhibit redistribution of Fog in all 7 embryos examined (Fig. 4D). Even though some cytoplasmic apical increase of concentration was observed, possibly due to an increase in endocytosis and exocytosis activity at 30°C, no massive apical attraction of Fog and no plasma membrane accumulation were observed in all of 10 heat-shocked sna homozygous embryos at stage 6 (fig. S6B). The shi mutation blocks endocytosis at 28°C; thus, plasma membrane retention of Fog likely reflects a block in Fog endocytosis in sna shi stage 6 mutants at 28°C. Accumulation of Fog at the apical side of the cells is also observed, which is consistent with an enhancement of exocytosis due to the increase in membrane tension triggered by the apical flattening and contraction of the cells after activation of Fog signaling.

To check if mechanical cues could block off Fog endocytosis, we monitored Fog distribution in stage 6 sna homozygous mutant embryos that were subjected to local indentation. Local indentation triggered the accumulation of Fog on the apical side of the cells in all six embryos observed (Fig. 4C), with Fog accumulating at the cell surface in four of the six embryos (Fig. 4C). This condition shows a mechanical rescue of the inhibition of Fog endocytosis by indentation in mid stage 6 sna mutant embryos that are defective in stochastic constriction and the associated mechanical deformation.

The proposal that mechanical inhibition of Fog endocytosis is the primary mechanotransduction mechanism activating signaling downstream of Fog and leading to the apical redistribution of MyoII to in response to Snail-dependent stochastic constrictions is supported by three lines of evidence: (i) A Fog-dependent mechanotransduction pathway regulates MyoII apical redistribution (Fig. 3). (ii) Blocking endocytosis phenocopies the mechanical rescue of MyoII apical redistribution and mesoderm invagination of sna mutants (Fig. 4D). (iii) The apical membrane accumulation of Fog that is observed in stage 6 wild types (Fig. 4A) is mechanically rescued in sna stage 6 mutants by indentation (Fig. 4C).

We thus propose that the first stochastic constriction of 40% of mesodermal cells inhibits Fog endocytosis in the other 60% of cells as a result of the increase in apical membrane tension induced by apical surface stretching or an increase in cell pressure, and the enhanced Fog signaling then triggers the redistribution of MyoII in these cells.

Simulating the interaction between Fog and Snail in MyoII apical redistribution

Ectopic expression of Fog leads to ectopic MyoII apical redistribution (5) and, in domains other than the mesodermal domain defined by the presence of Snail, to apical constriction (18). Thus, in apparent contradiction with the necessity for Snail and Fog to generate apical constriction, Fog without Snail can also trigger apical constriction. The explanation for how both Fog and Snail are necessary for MyoII apical redistribution and collective apical contraction in mesoderm cells, but not in nonmesodermal cells in which Fog alone is sufficient, remained a missing link in understanding the control of the early Drosophila embryo mesoderm invagination. Because mechanical strains propagate rapidly over long distances, Snail and Fog could be necessary for MyoII apical redistribution without being present in the same cells. Indeed, in embryos ectopically expressing Fog outside of the mesoderm, Snail-initiated stochastic deformations of the mesoderm, which normally lead to the wild-type mesoderm collective constriction, could be transmitted mechanically into regions of the embryo where Snail is not present but where Fog is overexpressed (5). We tested the plausibility of this scenario in silico by predicting the phenotype of embryos constitutively overexpressing Fog with a previously developed simulation of the multicellular invaginating embryo (21) to which we incorporated a Snail-dependent stochastic apical constriction process, a mechanosensitive activation of the Fog signaling pathway, and the presence of Fog in all of the modeled cells. The simulation predicted several previously reported phenotypes that have been reported in vivo (5, 18).

In contrast to finite element simulations in sea urchin embryo (25), which introduce a direct mechanical activation of a change in cell shape initiated by the shape change of the central cell of the invaginating domain, we introduced a stochastic concentration parameter “Csna” (to monitor Snail-dependent stochastic constriction of cells), a concentration parameter “Cfog” (to monitor Fog-dependent apical tension from actin-myosin that was above a membrane tension threshold), and an “Endo” parameter (to monitor the value of the membrane tension threshold, which is dependent on the apical surface tension). The values of the three parameters were defined for mesoderm cells only to mimic a transition from stochastic constriction to collective constriction in the wild-type embryo, before introducing the Cfog and Endo parameters into the Fog+ simulation. Mesoderm stochastic apical flattening and constriction was observed at early stage 6 (Fig. 5A), followed by collective apical flattening and constriction that initiated in all mesodermal cells, which was followed by apical flattening in all nonventral cells. The simulation that included Fog in all of the cells, not just the mesoderm, predicted incomplete invagination at stage 6 as a result of Fog-dependent mechanical activation of lateral and dorsal apical constriction tension, whereas the simulation with Fog only present in the mesoderm reproduced the wild-type phenotype (Fig. 5A). The model predicts a mechanical activation of MyoII apical redistribution that initiates in mesodermal cells, then occurs in lateral and dorsal cells all around the embryo (5). The model also predicts apical flattening in sna embryos that are expressing Fog ectopically all around the embryo (18) (fig. S7).

Fig. 5

Fog-dependent mechanical induction of collective MyoII apical redistribution and apical constriction by Snail-dependent stochastic constrictions. (A) In silico simulations from (21) implemented with a mix of Snail-dependent stochastic contraction of cells and Fog-dependent apex constriction of mesodermal cells in response to stress, with Fog ectopically expressed homogeneously (Fog+), compared to simulation of the wild type (WT). Lines show hydrodynamics velocity field. Colors code for the direction of the velocities along the lines. The flow is directed along the line from yellow and pale blue to red and dark blue colors. (B) A schematic showing that the second collective constriction wave depends on both input from Twist and from a mechanical signal that arises from the Snail-induced stochastic constriction wave.


Mechanical shape and strains play an important role in the maintenance and the development of organ morphology (26, 27). Embryogenesis is finely controlled by reciprocal mechanical and transcriptional cues that link developmental gene expression and tissue morphogenesis (8, 9, 16). Here, we suggest the existence of a posttranslational interplay of active morphogenetic movements with regulation of signaling and protein localization involved in morphogenesis. The mechanical deformations of a given stage of development activate the molecular processes that generate the active forces that control the morphogenetic movements of the next stage. We show that the apical redistribution of Myosin is mechanosensitive through a process that is dependent on a Fog signaling mechanotransduction pathway. We find that mechanical cues block Fog endocytosis at the apical membranes of mesoderm cells in response to Snail-dependent apical flattening and constriction, either because of direct membrane stretching or because of an increase in cell volume pressure. The block in Fog endocytosis likely activates the Fog signaling pathway leading to MyoII apical redistribution in all mesoderm cells. In this model, forced Fog expression would be countered by endocytosis-mediated degradation and would not be strong enough to activate the Fog signaling pathway, as indicated by the lack of MyoII apical redistribution and mesoderm invagination in embryos defective in both twi and sna but that have Fog present in the mesoderm (Fig. 3B). Thus, the presence of Fog without the mechanical strain that blocks its endocytosis is not sufficient to activate MyoII apical redistribution and mesoderm invagination. Experiments with stronger overexpression of Fog might eventually compensate for degradation and activate the pathway.

In this mechanotransduction process, the sensor would not be a protein that changes conformation in response to the strain [such as a transmembrane pore (28) or a protein linked to the cytoplasmic cytoskeleton (29)], but the sensor would be the membrane itself (22), whose strain state modulates the efficiency of the endocytosis of Fog signaling molecules, thereby enhancing the downstream signaling pathway. The modulation of T48 endocytosis, the other protein that is downstream of Twist and is known to participate in mesoderm invagination in parallel with Fog (20, 30), might additionally enhance MyoII apical redistribution, even though only Fog is required in the mechanotransduction process (Fig. 3). We thus propose a model in which the Fog-dependent mechanical activation of the apical redistribution of MyoII is used to produce a collective apical constriction leading to an efficient invagination in response to the stochastic contraction of Snail-containing cells (Fig. 5B).

Interestingly, response to stress appears to also be involved during plant early morphogenesis, during which microtubules orientate along the mathematically predictable stress field lines of the developing meristem (31), with this orientation following a potentially active (32) or passive (subcellular minimization of elastic energy) process.

Standard views of embryonic development focus on the control of patterning and of morphogenesis by the genome (33), with demonstrations of a reversal interaction from morphogenesis to the expression status of the genome (9, 10). The existence of such a bidirectional interplay between the genome and the macroscopic morphology of the embryo shows that some vital aspects of embryogenesis belong to emerging properties of a reciprocal coupling between the morphology and the genome (8). However, among the schemes of interplays controlling embryonic development including these previously unidentified concepts (gene-to-gene, patterning; gene-to-shape, morphogenesis; and shape-to-gene, mechanical induction of gene expression), evidence for the shape-to-shape interplay (mechanical activation of biochemical pathways triggering active morphogenetic movements of embryogenesis) has been lacking. Here, we describe such a link, through the Fog-dependent mechanical activation of collective and synchronized constriction in response to the Snail-dependent stochastic constriction of mesoderm cell apexes, which allows mesoderm invagination.

We speculate that the use of mechanical activation of MyoII apical redistribution during mesoderm invagination might have evolved from an active invagination feeding reflex of early embryos, putatively dependent on external mechanical strains interrupting the endocytic flow of a secreted protein such as Fog (34, 35). Such a scenario, already suggested at the origin of the mechanical induction of twi during early gastrulation (10), would involve a more rapid response through posttranscriptional rather than transcriptional mechanisms.

Experimental Procedures

Strains and labeling

Wild types Oregon R, SnaIIG/Cyo sna mutants, and Shi1 shi mutant were from Bloomington, and snaIIG twiIIH/SM1 sna twi double mutants and twiEY53;PE-Fog/SM1 were a gift of M. Leptin. The sna*shi cross was realized through classical procedures and took advantage of the “sleeping” phenotype of shi mutants at 28°C to identify the presence of the mutation at any step of the cross. Labeling procedures were classical, with rabbit antibodies against MyoII (36), rabbit antibodies against Twist (37), and rabbit antibodies against Fog (5) provided by R. Karess, S. Roth, and E. Wieschaus, respectively. The Alexa488 secondary antibody against rabbit IgG was from Molecular Probes. Labeling methods with formaldehyde fixation were classical, showing labeling of the complete cell (38), and differ from those in (5) in that a heat-methanol fixation method was used to specifically preserve protein labeling at plasma membrane (39).

Kymographs, determination of phenotypes, and mechanical perturbations

Two complementary methods were used to define the phenotype of the sna homozygous embryos. The first one consists of tracing the kymographs of the mesoderm movements during the first 10 min of stage 6, which shows a contraction wave associated with apical constriction in the wild type or in the heterozygous (fig. S1A), whereas the homozygous that does not constrict does not show any contraction wave (fig. S1A). The second one is the timing of the invagination of the cells in the anterior midgut, which is delayed in snail homozygous compared to the timing in the heterozygous and wild-type embryos (fig. S1B). The same methods were used to define twi sna homozygous phenotypes. The absence of Twist in the indented invaginating embryos was checked by immunolabeling of Twist (fig. S4). It was not possible to similarly check for the absence of Snail in sna embryos owing to the poor quality of the currently available antibody against Drosophila Snail.

Kymographs were captured with Image J, and needle indentation was performed with a Nashigire classical micropipette micromanipulator.


40. This work was funded by the Human Frontier Science Program (RGP0014/2006) and ANR (06PCVI001302) and by a Microsoft PhD fellowship grant to P.A.P. We thank J. Whitehead for her suggestions and reading of the manuscript. P.A.P. realized the simulations, mechanical cuts, and imaging; P.A., the indent experiments and imaging; and A.C.B., the ShiSna experiments. P.A.P. and E.F. designed the experiments with P.A., and E.F. coordinated the work.

Supplementary Materials

Fig. S1. Morphological phenotypes of sna homozygous mutants.

Fig. S2. A minority of sna/sna embryos exhibit ventral flattening and limited MyoII apical concentration.

Fig. S3. sna mutants do not rescue in response to simple contact.

Fig. S4. twi mutants that are defective in sna expression that express Fog in the mesoderm lack Twist.

Fig. S5. Mechanical deformation fails to rescue twi mutants that are also defective in sna expression.

Fig. S6. Fog and MyoII localization is not altered by heat shock in sna mutant embryos.

Fig. S7. sna and sna Fog+ phenotypes predicted by simulations.

References and Notes

  1. P. A. Lawrence, The Making of a Fly. The Genetics of Animal Design (Blackwell Scientific Publications, Oxford, 1992).
View Abstract

Navigate This Article