Mechanical competition alters the cellular interpretation of an endogenous genetic program

  1. Sourabh Bhide
  2. Denisa Gombalova
  3. Gregor Mönke
  4. Johannes Stegmaier
  5. Valentyna Zinchenko
  6. Anna Kreshuk
  7. Julio M. Belmonte
  8. Maria Leptin

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

The intrinsic genetic program of a cell is not sufficient to explain all of the cell’s activities. External mechanical stimuli are increasingly recognized as determinants of cell behavior. In the epithelial folding event that constitutes the beginning of gastrulation in Drosophila, the genetic program of the future mesoderm leads to the establishment of a contractile actomyosin network that triggers apical constriction of cells and thereby tissue folding. However, some cells do not constrict but instead stretch, even though they share the same genetic program as their constricting neighbors. We show here that tissue-wide interactions force these cells to expand even when an otherwise sufficient amount of apical, active actomyosin is present. Models based on contractile forces and linear stress–strain responses do not reproduce experimental observations, but simulations in which cells behave as ductile materials with nonlinear mechanical properties do. Our models show that this behavior is a general emergent property of actomyosin networks in a supracellular context, in accordance with our experimental observations of actin reorganization within stretching cells.

Article activity feed

  1. Version published to 10.1083/jcb.202104107
  2. Review Commons

    Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Reviewer #1 (Evidence, reproducibility and clarity):

    Bhide and colleagues present an insightful study of how cellular mechanics influences differential cell behaviour during morphogenesis despite apparent genetic homogeneity of the cellular ensembles. They dissect the extensively studied system of mesoderm invagination in Drosophila, focussing on the differences in cell behaviours between the cells in the middle of the infolding tissue and on the periphery that, as far as we know, share a common gene expression profile. They describe sub-cellular dynamics of major effector of apical constriction morphogenesis, the myosin motor distribution, in the invaginating cells and conclude that differences in myosin levels alone cannot account for the observed differences in cell behaviours. In order to understand the cell behaviour inhomogeneity, they turn to biophysical simulation and in an impressively exhaustive manner substantiate the idea that non-linear effects are required for explaining the phenomenon. This theoretical treatment fits well with the notion that the genetic identity of the cells but rather cell-cell mechanical coupling determine the differences in invaginating cell's behaviours. Additionally, the modelling is consistent with the myosin asymmetry and dynamics in the cells whose behaviours is being contrasted. Complementary, and beautifully executed filament-based modelling of microscopic actomyosin contractility further corroborates this view. Finally, the proposed model of non-linear actomyosin contractility dynamics governing the differential cell behaviour across genetically homogenous cellular field, is challenged by two complementary laser ablation and optogenetic experimental approaches. Overall, the results represent convincing evidence that points the tissue mechanics field of Drosophila mesoderm into an interesting new direction and has general implications for the understanding of the interplay between genetic regulation and emergent behaviours of cells operating in mechanically complex multicellular embryonic context. The study is meticulously executed, highly quantitative and combines effectively experiment and theory. I have only minor comments that concern in particular the presentation of the results.

    The paper is very dense and the text does not complement well the results presented in the main figures. Many panels in the Figures are not referred to explicitly. Figure elements are referenced out of order both within and across Figures. Sometimes, particularly, in the last two Figures (3 and 4) the reader is left alone to figure out what the data show (with the appropriately terse legends and without the clear narrative in the text, it is an uphill battle for non-specialists). Some key results are hidden in the sea of elements within the Figure 2 that contains the most important, relevant and impressive data.

    We have split this figure in two, moved some of the results from Suppl. Fig. 5 into one of its parts and included new calculations and data. We have also extended the description of these results in the main text and in the figure legends.

    As an example, on line the authors point to panel 2F to demonstrate the asymmetry of myosin distribution in some cells. To the best of my understanding, this phenomenon is actually shown in Fig 2E which is curiously not referenced at all.

    We have corrected the references to the panels

    Similarly, Figure 2K and L provide crucial data substantiating much of the conclusions of the paper. It requires a major effort to understand what the graphs mean. The following simulation results are quite impressive and would deserve a separate Figure which could provide more space for explaining what the parameter maps actually show. What is for instance plotted on the Y axis as steepness?

    We have added the following explanation: “The ‘width’ of the profile is the number of cells with maximum value; the ‘steepness’ is the slope between minimal and maximal values (equation 2 in materials and methods).”

    Secondly, I find the overall narrative of the manuscript needing some reorganisation. The main question is set-up extremely well, however in the middle of the manuscript the focus on the connection between cell behaviours and genetic programs is lost. New conclusions on force transmission between cells emerge, however they are not obviously connected with the question posed from the onset and addressed in the discussion section.

    To us, the section on force transmission seemed like an important component of the issue of intrinsic versus extrinsically determined cell behaviours. We had seen that the intrinsic programme of the cells, as reflected in their myosin levels, might not be sufficient to explain the difference between stretching and constricting. If their behaviour is not intrinsically determined, then there must be something acting from the outside, and we are looking here at what that might be, i.e. we need to find out how the potential constriction is influenced. The first model tests under which conditions differential contractility leads to different ‘cell’ behaviours. This in turn leads directly to the question of the forces the cells in the epithelium exert on each other.

    My impression is that the authors are conservative in their reasoning, however it does compromise the overall message of the story that should ideally focus on one subject. I find the combined evidence presented sufficiently supportive of the model that is beautifully and eloquently presented in the concluding sentence of the paper:

    "This mechanism, which we propose corresponds to the non-linear behaviour predicted by the models, would apply both to central and to lateral cells, with a catastrophic 'flip' being stochastic and rare in central cells, but reproducible in lateral cells because of the temporal and spatial gradient in which contractions occur."

    This may not turn out to be the entire story or even entirely correct, but it is certainly and exciting way of thinking about the problem. I wish that the manuscript would stay more on this subject throughout and provide intermediate conclusions supporting this model as the story develops.

    Few more minor comments:

    We have corrected all of the typos, mistakes and omissions and adapted the text, as mentioned below.

    Line 36 - typo > Line 97 - starting bracket missing > Line 126 - data on intensity are presented here. There is also a panel on concentration (Fig 1H). Where is this discussed?

    An explanation (definition) has been added to the main text.

    Line 132 - panel 2G - disruptive out of sequence reference to a future figure > Line 135 - with this regard - please spell out this important conclusion

    We have expanded this part, basically introducing the conclusion more clearly (we hope).

    Line 183 - typo > Line 210 - insects do not have intermediate filaments

    We have added ‘mammalian‘ to the reported experiment in the text, to make it clear that this does not refer to Drosophila cells

    Line 238 - please provide a hint of how such global ablations are performed > We have added this – both explicitly, and the relevant references.

    Line 240 - walk us through the Figure, it is too complex to figure it out alone > We have added a more extensive explanation both in the text and in the new figure legend.

    Line 245 - why is the clear hypothesis mentioned above (point 2) rephrased? > Line 273 - vague statement

    We have changed the text in response to these useful pointers.

    **Significance:

    The results represent convincing evidence that points the tissue mechanics field of Drosophila mesoderm into an interesting new direction and has general implications for the understanding of the interplay between genetic regulation and emergent behaviours of cells operating in mechanically complex multicellular embryonic context.

    **Reviewer #2 **

    Bhide and colleagues explore the mechanisms of cell expansion in epithelial morphogenesis. During the invagination of the Drosophila mesoderm, cells in the center of the prospective mesoderm constrict under the action of actomyosin pulses, while lateral cells elongate towards the center of the mesodermal placode to accommodate the reduction in apical surface of the central cells. Central and lateral cells display strong similarities in terms of gene expression. How are thus this different behaviors (contraction and expansion) accomplished? The authors found that both central and lateral cells assemble actomyosin networks, although lateral cells do it with a certain delay. Mathematical models of cell constriction across the mesoderm using different strain-stress responses showed that strain-induced cell softening was necessary recapitulate the patterns of constriction and expansion observed in vivo. Furthermore, modelling predicts that cells can stretch until the actin networks yield and break. Laser ablation and optogenetic reduction of contractility in central cells results in a reduction in the apical surface area of lateral cells. An optogenetic increase in contractility in lateral cells caused an increase in apical area in central cells. Together, these data suggest that mechanical cues can override and contribute to sculpting genetically defined morphogenetic domains.

    I propose to address the following points before further considering the manuscript:

    Major

    1. Figure 3: following laser ablation of central cells, lateral cells reduce their apical surface. How do the authors know that this reduction in lateral cell apical surface area is an active process, driven by actomyosin-based contraction, rather than a passive response to the expansion of the wound induced by laser ablation?

    A similar argument could explain the constriction of lateral cells after optogenetic inhibition of actomyosin networks: the central cells relax, expand and compress the lateral cells.

    With regard to the comparison to wounds, it is important to note that the epithelium is not actually wounded by either ablation method. Thus, while the treatments ablate the actyomyosin meshwork, they do not ablate or kill the cells. Perhaps the term is an unfortunate choice, since it is more commonly used in developmental biology for killing cells. However, here the cells remain intact and when the optogenetic or laser treatment is released the cells resume their physiological activities.

    We have added a note in the text and now refer to ‘laser microdissection’, a term of art in the field, for more clarity.

    Regarding the more important question of what is the active process, expansion of the central cells or constriction of the lateral cells, a contribution from expanding central cells is of course in theory not impossible.

    However, for this scenario to work, in the absence of pulling from the lateral cells, there would have to be a force that is generated in the central cells, in this case a pushing force that would expand the cells and act on the lateral cells. We have shown in our previous work that if the actomyosin is dissected in dorsal cells, which are not surrounded by potentially contractile cells, the cells do not expand (Rauzi et al, 2017). This shows that ‘relaxing’ by itself does not have ‘expansion’ as a consequence. One would therefore have to consider how such a pushing force could arise in these cells. We can think of only two possibilities: hydrostatic pressure or an active force from the subcellular molecular machinery.

    Considering hydrostatic pressure, if the apical actomyosin that is ablated was responsible for maintaining such a pressure inside the cell (a reasonable assumption), then releasing the actomyosin would allow the cell volume to push against the neighbouring cell. However, such a recoil would occur on a very short time scale (seconds), whereas we see the contraction of the lateral cells continuing over extended periods (minutes).

    Alternatively, expansive forces could be generated by the cytoskeleton. Cytoskeletal pushing forces can come from microtubules (classical example: mitotic spindle; epithelial morphogenesis: work from T. Harris and B. Baum labs: PMID 18508861 and 20647372), or from continuous creation of new cross-linked or branching actin networks pushing against plasma membranes, as in the leading edge of crawling cells. But the microtubules in the blastoderm cells are not oriented in such a way they could provide a force in the correct dimension in these cells (the majority is oriented along the apical-basal axis). In addition, the connection between MT and the plasma membrane depends on the cortical actin meshwork (involving, for example, the actin-binding proteins P120-Catenin or patronin/Shot; Roeper lab, PMID 24914560, StJohnston Lab, PMID: 27404359) but the connection of actin with the plasma membrane has been severed in the optogenetically manipulated cells.

    By contrast, we show that normal lateral mesodermal cells possess a contractile actin network. So the only sustained force generated in the system at this point is the contractile force in lateral cells (which is normally counteracted by the stronger contractile force from central cells).

    Thus, we conclude that the expansion of central cells is a passive response to a contractile force from lateral cells, not an active process and conversely, the constriction of lateral cells is an active autonomous process.

    To demonstrate active responses of the lateral cells upon laser ablation and optogenetic manipulations of central cells, at the very least the authors should show the distribution of myosin in the lateral cells that constrict and demonstrate the assembly of contractile networks.

    We have now included the requested data for the experiments with laser ablations. Suppl. Fig. 8 and Suppl. video 3 show the myosin that accumulates in lateral cells. It would be nice also to be able to show this for the optogenetic experiments. However, despite trying hard, we have not succeeded in generating healthy embryos that carry the entire set of transgenes that are necessary to carry out the optogenetic experiments and at the same time visualize myosin (see also response to referee 2, point 3).

    1. Modelling suggests that actin networks yield and break in lateral cells. Does this occur in vivo?

    We postulate that the skewed and inhomogeneous distribution of myosin and the large myosin-free areas in stretched cells (lines 170 – 172 in the original text) are indications of a yielding meshwork, or at least of uneven force distribution in the network that leads to ineffective contraction or even release – i.e. functionally correspond to yielding. We have made this more explicit now.

    We have also added an additional panel quantifying more clearly the proportion of low- myosin areas in lateral cells (now Fig. 3H).

    Work from the Lecuit lab has recently shown beautifully that it is the connectivity of the myosin mesh rather than the underlying actin meshwork that affects apical forces in epithelial cells (PMID: 32483386), and our own findings are entirely consistent with that.

    1. Lines 166-175: The authors propose that constriction of a cell affects the localization of myosin in its neighbors. However, this is not directly measured. The authors should quantify the relative myosin offset in the cells around constricting cells, and show that that offset is greater (and oriented towards the constricting cell) than in cells around expanding cells. There should be a correlation between the relative size change of a cell and the myosin offset (not just concentration) in their neighbours. We now provide measurements of the rate of cell area change against the offset of surrounding myosin (the distance of myosin from a cellular border). We see that surrounding myosin is closer to the border of constricting cells and tends to be further away from the borders of expanding cells.

    We have added these data to the new Fig. 3I.

    In addition, does optogenetic activation of constriction in lateral cells affect the offset of myosin networks in central cells?

    This is technically challenging. For such an experiment we would need an embryo to express membrane and myosin markers in addition to the two optogenetic constructs and the GAL4 driver. We tried multiple times to generate such a cross, but obtained either no embryos or, at best, deformed embryos. We also tried to use the MCP-MS2 system in parallel to CRY2-RhoGEF2 but the crosses had the same problem. This sensitivity to additional genetic load was also observed in the DeRenzis lab, who generated these strains and tested and used them extensively.

    1. Fig. 2E-F: the authors argue that the mean myosin concentration in lateral cells at certain times is equivalent to that of central cells earlier in the invagination process. However, the fraction of apical surface area covered by myosin network is consistently lower for lateral cells (and also for central cells that remain unconstricted!). Have the authors considered this fact, and if not, why? Wouldn't this explain, at least in part, why some cells constrict and others do not, if medial myosin networks drive the disassembly of the apical surface?

    We believe in fact that this is precisely part of the picture and it was what we had meant to propose, but the text was perhaps indeed just to condensed. Thus, we had stated in line of the original document:

    “While the asymmetry is visible in all cell rows, there are larger areas without myosin and the distance of displacement is greater in lateral cells (Fig. 2G-J)”,

    and in the discussion (line 277 – 285):

    “Despite the homogeneous actin meshwork in stretching cells, the areas that are free of active myosin occupy a large proportion of the apical surface – similar to ectodermal or amnioserosa cells in which the connection of pulsatile foci to the underlying actin meshwork is lost. ... Dilution of cortical myosin may compromise a cell’s ability to make sufficient physical connections, in particular along the dorso-ventral axis, so that even if sufficient force is generated, it cannot shorten the cell in the long dimension. In other words, even though the cells have enough myosin to create force, the system is not properly engaged and its force is not transmitted to the cell boundary.”

    However, we didn’t state this with sufficient clarity in the results section and have added an extra sentence to this effect.

    If myosin activity were increased in laterals cells once central cells begin constricting, would that lead to an increased fraction of lateral cell surfaces covered by actomyosin networks and to reduced lateral cell elongation?

    This is a really nice experiment, and we have indeed tried to induce activation at later time points, but unfortunately this did not yield unambiguous results. If we did the manipulation after the central cells had clearly constricted, then activating lateral cells did not lead to their contraction. However, since this is a negative result and we have no independent criterion for knowing how 'strong' the induced contraction was (as explained above, we are unfortunately not able to visualize the myosin in these experiments), and why it might not have been sufficient to overcome the pull from central cells.

    In this context it is worth remembering that in mutants in which myosin is overactivated as a result of defective upstream signalling, lateral cells stretch less or not at all. See PMID: 24026125 for gprk2 mutants and our own results for active Rho1:

    {{images cannot be displayed}}

    Figure: Confocal Z-section of embryos expressing sqh::GFP (myosin; green) and GAP43::mCherry (membrane; magenta) imaged ventrally. A constitutively active form of Rho1 is ectopically expressed using a maternal Gal4 driver, inducing activation of myosin in more lateral cells. White dots mark the mesectoderm determined by backtracing after ventral furrow invagination. Yellow arrows in B are constricted cells in row 7/8.

    Minor

    1. Image panels are missing scale bars in many figures. > 2. Fig. 1C'-D': The authors should include a color bar to provide some indication of the scale of the apical areas measured. Same comment for other figures in which apical area is color-coded.

    We have added the missing elements

    1. Supp. Fig. 2E-F, G-H and Supp. Fig. 6: what is the difference between myosin intensity and myosin concentration? Junctional vs medial localization? Or summed vs mean pixel value? Please be specific, the difference between intensity and concentration is not clear.

    In the cases where we talk about myosin ‘amount’ we have now exchanged the term ‘intensity’, i.e the physical term for the amount of light, for ‘amount’ (i.e. that for which we use the light intensity as a proxy) and have explained in the main text how we define total apical myosin amount and apical myosin concentration (amount over area). However, in the cases where we are describing the actual image analysis, as in Suppl. Fig. 3, we use ‘intensity’ as the term of art that is used for the methods employed here. Similarly, the terms ‘sum intensity’ and ‘mean intensity’ are terms used for image in analysis in Fiji.

    The definitions of “junctional” and “medial” actin were introduced by the Lecuit lab (PMID: 21068726), and we have included the appropriate reference.

    1. Line 118: Supp. Fig. 2 does not have panels I and K. > 5. Line 223: the authors reference data at sec, but Supp. Fig. 6 does not show any images at that time point. They should be added or a different time point indicated.

    These errors have been corrected.

    Typos

    1. Abstract: "[in a supracellular context" should be "in a supracellular context". > 2. Line 145: should this be a reference to Supp. Fig. 5 instead of Supp. Fig. 4? > 3. Line 166: I am not sure how Supp. Fig. 5 supports this statement. Is this the right figure reference? Should it be Supp. Fig. 4 instead? > 4. Line 881: "representing on line" should be "representing one line".

    These errors have been corrected.

    Optional

    Tony Harris' lab showed that the Arf-GEF Steppke antagonizes myosin and facilitates cell deformation at the leading edge of the embryonic epidermis during Drosophila dorsal closure (West et al., Curr Biol, 2017). Does Steppke localize to junctions in lateral but not central mesoderm cells? Does the pattern of Steppke localization in the mesoderm change with manipulations to the contractility of central cells?

    This is certainly interesting, and we have ordered the protein trap, UAS constructs and RNAi lines. However, these will be long-term and time-consuming experiments.

    Significance:

    This is an interesting study, and one that makes uses of beautiful tools, including quantitative microscopy and image analysis, mathematical modeling and optogenetic manipulations. The prediction that embryonic cells display non-linear stress-strain responses is exciting, as linearity has been the predominant assumption so far. However, I find that model predictions are not well supported by the data, and that alternative interpretations of some results are possible. Additionally, the paper lacks insight into the molecular mechanisms that facilitate stretching (although that could be the subject of a follow-up study).

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    Summary:

    In this study, the authors explore potential mechanisms for why some cell constrict while other cells expand, despite similar intrinsic genetic programs, during Drosophila ventral furrow formation at the onset of gastrulation. The authors combine quantitative analyses of cell shapes and myosin levels from multiphoton confocal and Multi-View SPIM imaging, optogenetic and laser perturbation experiments, and mechanical models to argue that nonlinear mechanical interactions between cells are required to explain the cell behaviors. Based on microscopic models of the actomyosin cytoskeleton in the tissue the authors argue that the required nonlinear mechanical behavior is consistent with actomyosin network reorganization.

    Major comments:

    • Although the area of investigation is exciting and the results are interesting, unfortunately the quality of the results and comparison between experiment and modeling in the current version of the manuscript are not convincing. Although it is not clearly explained in the manuscript, the experimental results on cell shapes, myosin intensity, laser manipulation, optogenetic perturbations appear to be from a single embryo or small number of embryos for each experiment (Figures 1, 3, 4).

    We had analysed a much larger number of embryos, but only included those for presentation that provided the most extensive data. It is extremely difficult to obtain absolutely ‘perfect’ embryos at high resolution for full quantification over long periods. ‘Perfect’ means that the embryos are mounted in such a way that they are imaged from an angle of 45 degrees off the dorso-ventral axis, so that initially mesodermal rows 3 to 7 are seen, and then, as furrow formation progresses, the more lateral rows move through the field of vision. It is difficult to mount in this perfect manner for two reasons: the shape of the embryo means that the embryo does not ‘like’ to be balanced in this position, but instead prefers to fall back on its side. Secondly, the embryo has to be mounted at a time point before visible differentiation along the D-V axis, so no visual cues exist to get the positioning right. This means that many of our recordings lack either the more ventral or the lateral cell rows. While the findings for these more restricted observations are fully consistent with our reports, they cannot be quantified with a full comparison across all cell rows over the entire imaging period. Nevertheless, we have processed and analysed further examples which we have now included in Suppl. Fig. 2 and Suppl. Fig. 8.

    The authors state that the cell stretching pattern "was best recapitulated by a superelastic response", but did not provide direct quantitative comparisons of the different mechanical models to the experimental data to clearly demonstrate this.

    Data that illustrate this were shown in Suppl. Fig 5 – but, admittedly, were not well explained, or rather, not at all. We have now added better explanations, expanded the figure, included new analyses, and now present some of these data in the new Fig. 2. Briefly, the figure shows that superelastic and elastoplastic responses are the only curves that successfully reproduce the pattern of stretching lateral cells (last 3 cells stretching with the inner cell stretching most and the last cell stretching least) while at the same time matching the ratio between the cell sizes of the most stretching cells to the least stretching cell.

    The top row of the parameter scans in Suppl. Fig. 5 (now Fig. 2) shows how many cells stretch for each combination of myosin curve steepness (y-axis) and width (x-axis) with shades of blue indicating the number of cells, and the red outline in the field where 3 cells stretch outlining those conditions where the inner cell stretches most. The bottom row shows the resulting size ratios of largest to smallest cell. High ratios in the region outlined in red in the top row are only reached for the superelastic and elastoplastic responses, with the elastomeric tending in the right direction.

    We have now also quantified a goodness-of-fit (root mean squared error, RMSE) measurement between our experimental data and the simulated data of all our models. This is shown now in the new Fig. 2.[1]

    We also note that only the parameter maps of the superelastic and elastoplastic models (Fig. 2J,K) resemble the equivalent parameter maps of the microscopic model (Fig. 3Q).

    Moreover, the local optogenetic myosin recruitment experiments in Figure 4 do not provide sufficient information on optogenetic tool recruitment,

    We have included images that illustrate the optogenetic construct in the illuminated cells, but not in the central cells in Suppl. Fig. 8. It is impossible to show the construct in the ‘dark’ cells, because illuminating them would activate the construct.

    myosin localization,

    As explained above, this is unfortunately technically not feasible. The best we can do is refer to the description of the construct by Izquierdo et al. (PMID: 29915285), which shows the accuracy of the tool and the highly specific membrane recruitment of myosin.

    or cell behaviors

    We have added quantitative comparisons between the experimental and control areas. to justify the claim that the central cells are not activated by the optogenetic perturbation and are only responding to the forces from neighboring cells.

    • The authors should provide direct quantitative comparisons of the models and experiments to clearly demonstrate their claims that the superelastic model is better than the linear model or other nonlinear models.

    See response above.

    • The authors should do additional experiments and/or provide more details for the existing experiments (to include several embryos per condition) on myosin quantification, photo-manipulation, and optogenetics experiments.

    We have provided data for more embryos for all cases.

    Additional controls would like be necessary for claims resulting from the optogenetics experiments in Figure 4.

    This has been addressed above – we have provided additional data and controls.

    • The additional time and resources required to address these concerns would depend on the experimental details, N values, and statistics in the current studies, which unfortunately were not described in the current manuscript.

    We have been able to add substantial additional data and have added the requested numbers. For many of the experiments each recording can be very time consuming and for the reasons explained in this response, it is not always easy to obtain precisely the desired recording from the desired imaging angle with the manipulations having been done precisely in the desired position. The numbers of embryos are therefore not high, but multiple shorter recordings provide a body of results that support the findings, but are not easily comparable statistically.

    • Methods descriptions for reproducibility are generally adequate, with the exception of N values and statistics

    See above.

    • Are the experiments adequately replicated and statistical analysis adequate?

    No, see above.

    Minor comments:

    1. Scale bars for images are missing throughout.

    We have added these

    1. Number of embryos and cells analyzed missing throughout text and figure legends. We have added additional embryos for all conditions and have included the numbers of cells analysed for all quantifications (except in cases where each data point represents a cell).
    1. Units are missing for many quantities in figures and tables throughout.

    We have added these

    1. Many figure references in the main text are incorrect, pointing either to the wrong figure or wrong figure panel.

    These have been corrected

    1. Line 728. What time point was used for myosin concentrations used in the model?

    We have added this information to the figure legend.

    How might myosin dynamics influence these findings?

    As regards the subcellular dynamics of myosin, these are included in the microscopic model (see ref Belmonte et al.;PMID: 28954810). Preliminary results showed that small changes in myosin stall force and unloaded myosin speed have little effect in our general results. This is now shown in a new supplemental figure (Suppl. Fig. 6). However, if the referee is referring to the dynamics of myosin accumulation over time, this is an interesting question.

    We had begun to explore this topic, but then realized for the linear stress-strain model that it is in fact expected that myosin accumulation would ultimately not affect the outcome. This is because in a linear model the final state of the system is determined by the final shape of the governing myosin profile regardless of the time evolution of the profile, and our simulations confirm this. A systematic analysis for all other stress- strain curves with temporal changes in myosin profiles (where a dependency on the profile temporal evolution is expected) is very time-consuming and will be interesting to pursue in future.

    The main conclusion here that linear models do not recapitulate the observed data as well as the non-linear ones stands regardless of how the temporal dynamics of myosin accumulation may affect the non-linear systems.

    1. The authors show a few examples of myosin pulsing in lateral cells and then conclude that myosin pulsing is not qualitatively different from central cells (lines 135- 136). The author should quantify the number of pulsing lateral cells as well as period and amplitude of pulsing, or discuss relevant results from prior studies in more detail to justify this conclusion.

    By ‘not qualitatively different’ we had meant only ‘in the sense that they are capable of generating contractile forces’, and we have made that more explicit in the text now. The quantitative differences have already been analysed and reported by the Martin lab (https://doi.org/10.1101/2020.04.15.043893; the pulses are slower and less persistent), and our point was that in spite of these known differences, the pulses are able to mediate constriction.

    1. Lines 145-150. The authors very briefly describe the results of the linear-stress strain response and conclude this did not yield outputs corresponding to in vivo data and leave this largely to the supplementary figures. This is a key point in the paper and deserves much more discussion and space in the main text.

    We have included a more extensive description and interpretation of the results in the main text, as detailed in several responses above

    As mentioned in main comments above, a quantitative comparison of the different mechanical models to show that the superelastic model better describes the observations should be included (potentially as an inset to Fig 2D showing a quantitative measure of the quality of model fit to the data).

    These comparisons have now been expanded and explained more extensively and moved to the main Figures.

    1. Lines 162-163. Provide more rationale for why strain-softening would most likely manifest as permanent or reversible cytoskeletal reorganization.

    The only component of the cell that can likely mediate this physical property and also respond at the observed time scales is the cytoskeleton. In these cells it is the main mechanical determinant. Other components that could in principle contribute to the nonlinearity of stress-strain response might be the viscosity of the cytosol, or the plasma membrane. However, stress responses of fluids to shear are usually in the direction of increasing stiffness, and rarely, if ever, with shear thinning. The same is mostly true for colloidal solutions. Therefore it is more likely that the stress-strain relationships at the apical surface of the cells are dominated by the dynamics of the actin cytoskeleton given that even the shape of the plasma membrane is in general determined by the cytoskeleton. We have added a note to this effect in the text.

    1. Lines 187-188. "This shows that forces acting on each cell from its neighbors have an important role in determining the cell's behavior." This seems somewhat obvious; perhaps a bit more explanation would help the reader to understand the importance of these results.

    We have expanded the explanations of these findings and added a sentence to relate them to the main model of the paper

    1. Lines 196-198. How were the concentrations and lengths of F-actin chosen? How were the concentration and properties of linkers chosen?

    The parameters were chosen on the basis of our earlier studies on simulated contractile meshworks and the theory underlying their behaviour. We had reported the conditions under which such meshes are able to contract, and also shown that the underlying theory correctly predicts behaviour of experimental meshworks (for those few conditions for which they have been reported).

    Unfortunately, there are practically no measurements for the length of F-actin filaments in vivo and estimates vary widely. Reliable data on the density of the cortical network are equally sparse.

    Based on our own previous work we chose concentrations of cross-linkers, myosin motors and transmembrane connectors that are able to ensure optimal contraction and force. Our in vivo measurements reported here show that the amounts of F-actin do not vary significantly across the mesoderm, so we used the same concentration of actin, crosslinkers and membrane connectors in all cells of the model, varying only myosin concentration. Taking into account the cell diameter of the mesodermal cells (~7um) and to ensure that the meshwork is sufficiently cross-connected (dense) to generate contraction and transmit forces between cells we used a model where each cell contains F-actin filaments of 1.5 um.

    We have expanded our supplemental material to make these points clearer.

    How sensitive are the results to these details of the cytoskeletal composition?

    We varied both the amounts of cytoskeletal components and the parameters controlling their dynamics (such as myosin stall force and viscosity) and found little impact on model predictions. These data are now presented in Suppl Fig. 6.

    1. Lines 238-244. It would be helpful to include some additional quantification that clearly shows the reader the differences in cell behaviors in control and perturbed tissue.

    We have added quantitative comparisons of the cells in the perturbed region with cells in an equivalent control region, together with evaluations of two additional embryos.

    For the optogenetics experiment, it would be important to show quantification that the lateral cells are not being directly perturbed during photoactivation of neighboring cells (e.g. due to light leakage).

    We have included this information, as described above.

    In both perturbations, it would be helpful to quantify how many cells in rows 7 and 8 constricted and by how much did they constrict? How reproducible were these effects?

    The perturbation experiments were those where it was most difficult to obtain a large number of identical-looking embryos that would allow broad statistics to be applied. For this to work, we would have to have embryos that were identically mounted and illuminated in the identical area of precisely rows 1 to 6 on each side of the midline – at a resolution of one cell row of 6.2 um width. And all this blind, because at the start of the manipulation there are no visual cues for orientation. Morphology gives no cues at this stage. The MS2-MCP-GFP works for laser ablations, but cannot be used for the optogenetics, because the embryo must not be exposed to blue light. This means we cannot predetermine precisely which rows we target.

    We have however added data and quantifications for the control and two further laser- manipulated embryos, which are now shown in suppl. Fig. 8. It is evident from both that our perturbations were slightly asymmetric and included the outer rows on only one side and on that side several cells that would normally have stretched are now strongly constricted. While by no means true for all lateral cells, this is a case of one black swan disproving the hypothesis that all swans are white: any constricting cell within two cell diameters of the mesectoderm, i.e. ones that would normally stretch proves that lateral cells do have the capacity to constrict.

    1. Lines 245-252. A key assumption in interpreting this experiment seems to be that the central cells are not directly perturbed by the optogenetic activation. Additional quantifications of RhoGEF2-CRY2 and/or myosin should be shown to support this.

    We have included an image of the optogenetically activated construct in this experiment in Fig. 5, but we cannot show its behaviour in the non-activated part because if we illuminated it, it would be activated. We were unable to create the embryos necessary to document the behaviour of myosin.

    It would be helpful to include some additional quantification that clearly shows the reader the differences in cell behaviors in control and experimental regions. How reproducible were these effects?

    We now provide the results from two additional embryos in Suppl. Fig. 8, and include quantitative comparisons between the control and experimental regions for these and for the embryos that are currently shown in Fig. 5 E.

    1. A section on statistics is missing from the methods section.

    We have added descriptions of the quantifications and statistics.

    1. Line 615. Ensure that Eq. 1 is dimensionally consistent; crucially, what units are used for 'M'? If the model is non-dimensionalized, provide the reference scales.

    Apart from the initial distance between membrane positions (set to 6.2 um) all other units in our visco-elastic model are arbitrary. In order to make this clearer, instead of using the term “viscosity” in equation 1, we now call it a “damping constant”.

    1. Line 675: The investigated stress-strain relationships are presented in Table S1. What are the definitions of xpl and xsh?

    We have included these definitions in materials and methods:

    All stress-strain curves are linear for extensive strains (∆𝑥) lower than the proportionality limit (𝑥!"), with some curves (elastoplastic and superelastic) undergoing a strain-softening to strain-hardening change after a given strain-hardening limit (𝑥#$).

    1. Line 678: Parameter values for the stress-strain relationships are given in Table S2. Can you provide more information on how these values were selected and their units? How sensitive are the results to changes in these values? Provide references when possible.

    The values for xpl and xsh were chosen to be within the range of the observed lengths of stretching cells, with xpl < xsh. Changing the values of each parameter listed in Table S2 does change the results quantitatively, but over the ranges we tested them, never to the point of making the linear or the other non-linear models reproduce the target pattern of stretching.

    We have stated this in the materials and methods section.

    1. Line 697. Please comment on why the embryo appears skewed to the right. Embryos are not always ‘perfect’, unfortunately. In addition, they can get slightly squashed during mounting and imaging. In spite of its imperfection, we showed this particular one, because we had imaging data for a long period without drift or other interference, and with good contrast at great depth.
    1. Line 712. A color-bar corresponding to this color-code is missing in the figure.

    This has been corrected.

    1. Lines 715-717. It seems panels E and E' are swapped in the legend.

    corrected

    1. Line 724 (Fig 2). It is difficult to read anything in panel K inset or Panel L inset.

    We have rearranged this figure and replaced some panels for greater clarity, and to remove redundancy.

    1. Line 728. What does "embryo 1" refer to?

    This was a remainder from an old plan where each embryo was numbered and listed in a table so that it could be cross-referred to. We have now described in the supplementary table the genotypes and imaging technique for each group of embryos. Where we show data or analyses of the same embryo in different figures, we refer directly to the relevant panels. We have made sure the embryos are referred to correctly in the figure legends.

    1. Line 732. A quantitative measure of the quality of the fits of the models to the experimental data should be included.

    We have done this, and the new data are now included in the new Figure 2.

    1. Line 739. What exactly does "Embryo 2" refer to?

    See comment 21

    1. Line 779. Why is a z-plane of 15 microns below surface chosen? > 25) Line 797. Why is a z-plane of 25 microns below the surface chosen?

    The planes were chosen in each case to show the reader in one single plane rows 7 and 8 along with the central cells > 26) Line 900. Panel G in Supp Fig 5 is not described in figure description.

    The panel captions were wrongly numbered. This has now been corrected, and more information on this figure has been included in the text. > - Are prior studies referenced appropriately?

    Yes.

    • Are the text and figures clear and accurate?

    No (see details listed above).

    • It would be very helpful to the reader to show direct quantitative comparison of the different mechanical models with the experimental observations to show how much better the nonlinear model is compared to the linear model.

    We have included this.

    An extended explanation of experiments and experimental results within the main text would improve the manuscript.

    We have expanded our explanations in many places.

    Significance:

    The key advance in this work is in identifying a potential role of nonlinear mechanical properties in contributing to distinct cell behaviors within a tissue during development in vivo. This contributes to a growing body of work highlighting the importance of cell and tissue mechanical properties in regulating cell behaviors during the formation of tissue structure.

    This work adds to a growing body of work connecting actomyosin contractility in cells to tissue-scale behavior during development. This work provides a unique mechanical modeling perspective to the study of apical constriction during Drosophila ventral furrow invagination, highlighting a potential role for superelastic cell mechanical behaviors during morphogenesis in vivo.

    The finding would be of interest to researchers working in the areas of morphogenesis, mechanobiology, the cytoskeleton, and active matter.

    This reviewer's expertise is in experimental studies of the cytoskeleton and cell mechanics during morphogenesis.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary:

    In this study, the authors explore potential mechanisms for why some cell constrict while other cells expand, despite similar intrinsic genetic programs, during Drosophila ventral furrow formation at the onset of gastrulation. The authors combine quantitative analyses of cell shapes and myosin levels from multiphoton confocal and Multi-View SPIM imaging, optogenetic and laser perturbation experiments, and mechanical models to argue that nonlinear mechanical interactions between cells are required to explain the cell behaviors. Based on microscopic models of the actomyosin cytoskeleton in the tissue the authors argue that the required nonlinear mechanical behavior is consistent with actomyosin network reorganization.

    Major comments:

    • Although the area of investigation is exciting and the results are interesting, unfortunately the quality of the results and comparison between experiment and modeling in the current version of the manuscript are not convincing. Although it is not clearly explained in the manuscript, the experimental results on cell shapes, myosin intensity, laser manipulation, optogenetic perturbations appear to be from a single embryo or small number of embryos for each experiment (Figures 1, 3, 4). The authors state that the cell stretching pattern "was best recapitulated by a superelastic response", but did not provide direct quantitative comparisons of the different mechanical models to the experimental data to clearly demonstrate this. Moreover, the local optogenetic myosin recruitment experiments in Figure 4 do not provide sufficient information on optogenetic tool recruitment, myosin localization, or cell behaviors to justify the claim that the central cells are not activated by the optogenetic perturbation and are only responding to the forces from neighboring cells.
    • The authors should provide direct quantitative comparisons of the models and experiments to clearly demonstrate their claims that the superelastic model is better than the linear model or other nonlinear models.
    • The authors should do additional experiments and/or provide more details for the existing experiments (to include several embryos per condition) on myosin quantification, photo-manipulation, and optogenetics experiments. Additional controls would like be necessary for claims resulting from the optogenetics experiments in Figure 4.
    • The additional time and resources required to address these concerns would depend on the experimental details, N values, and statistics in the current studies, which unfortunately were not described in the current manuscript.
    • Methods descriptions for reproducibility are generally adequate, with the exception of N values and statistics - see above.
    • Are the experiments adequately replicated and statistical analysis adequate? No, see above.

    Minor comments:

    1. Scale bars for images are missing throughout.

    2. Number of embryos and cells analyzed missing throughout text and figure legends.

    3. Units are missing for many quantities in figures and tables throughout.

    4. Many figure references in the main text are incorrect, pointing either to the wrong figure or wrong figure panel.

    5. Line 728. What time point was used for myosin concentrations used in the model? How might myosin dynamics influence these findings?

    6. The authors show a few examples of myosin pulsing in lateral cells and then conclude that myosin pulsing is not qualitatively different from central cells (lines 135-136). The author should quantify the number of pulsing lateral cells as well as period and amplitude of pulsing, or discuss relevant results from prior studies in more detail to justify this conclusion.

    7. Lines 145-150. The authors very briefly describe the results of the linear-stress strain response and conclude this did not yield outputs corresponding to in vivo data and leave this largely to the supplementary figures. This is a key point in the paper and deserves much more discussion and space in the main text. As mentioned in main comments above, a quantitative comparison of the different mechanical models to show that the superelastic model better describes the observations should be included (potentially as an inset to Fig 2D showing a quantitative measure of the quality of model fit to the data).

    8. Lines 162-163. Provide more rationale for why strain-softening would most likely manifest as permanent or reversible cytoskeletal reorganization.

    9. Lines 187-188. "This shows that forces acting on each cell from its neighbors have an important role in determining the cell's behavior." This seems somewhat obvious; perhaps a bit more explanation would help the reader to understand the importance of these results.

    10. Lines 196-198. How were the concentrations and lengths of F-actin chosen? How were the concentration and properties of linkers chosen? How sensitive are the results to these details of the cytoskeletal composition?

    11. Lines 238-244. It would be helpful to include some additional quantification that clearly shows the reader the differences in cell behaviors in control and perturbed tissue. For the optogenetics experiment, it would be important to show quantification that the lateral cells are not being directly perturbed during photoactivation of neighboring cells (e.g. due to light leakage). In both perturbations, it would be helpful to quantify how many cells in rows 7 and 8 constricted and by how much did they constrict? How reproducible were these effects?

    12. Lines 245-252. A key assumption in interpreting this experiment seems to be that the central cells are not directly perturbed by the optogenetic activation. Additional quantifications of RhoGEF2-CRY2 and/or myosin should be shown to support this. It would be helpful to include some additional quantification that clearly shows the reader the differences in cell behaviors in control and experimental regions. How reproducible were these effects?

    13. A section on statistics is missing from the methods section.

    14. Line 615. Ensure that Eq. 1 is dimensionally consistent; crucially, what units are used for 'M'? If the model is non-dimensionalized, provide the reference scales.

    15. Line 675: The investigated stress-strain relationships are presented in Table S1. What are the definitions of xpl and xsh?

    16. Line 678: Parameter values for the stress-strain relationships are given in Table S2. Can you provide more information on how these values were selected and their units? How sensitive are the results to changes in these values? Provide references when possible.

    17. Line 697. Please comment on why the embryo appears skewed to the right.

    18. Line 712. A color-bar corresponding to this color-code is missing in the figure.

    19. Lines 715-717. It seems panels E and E' are swapped in the legend.

    20. Line 724 (Fig 2). It is difficult to read anything in panel K inset or Panel L inset.

    21. Line 728. What does "embryo 1" refer to?

    22. Line 732. A quantitative measure of the quality of the fits of the models to the experimental data should be included.

    23. Line 739. What exactly does "Embryo 2" refer to?

    24. Line 779. Why is a z-plane of 15 microns below surface chosen?

    25. Line 797. Why is a z-plane of 25 microns below the surface chosen?

    26. Line 900. Panel G in Supp Fig 5 is not described in figure description.

    • Are prior studies referenced appropriately? Yes.
    • Are the text and figures clear and accurate? No (see details listed above).
    • It would be very helpful to the reader to show direct quantitative comparison of the different mechanical models with the experimental observations to show how much better the nonlinear model is compared to the linear model. An extended explanation of experiments and experimental results within the main text would improve the manuscript.

    Significance

    The key advance in this work is in identifying a potential role of nonlinear mechanical properties in contributing to distinct cell behaviors within a tissue during development in vivo. This contributes to a growing body of work highlighting the importance of cell and tissue mechanical properties in regulating cell behaviors during the formation of tissue structure.

    This work adds to a growing body of work connecting actomyosin contractility in cells to tissue-scale behavior during development. This work provides a unique mechanical modeling perspective to the study of apical constriction during Drosophila ventral furrow invagination, highlighting a potential role for superelastic cell mechanical behaviors during morphogenesis in vivo.

    The finding would be of interest to researchers working in the areas of morphogenesis, mechanobiology, the cytoskeleton, and active matter.

    This reviewer's expertise is in experimental studies of the cytoskeleton and cell mechanics during morphogenesis.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    Bhide and colleagues explore the mechanisms of cell expansion in epithelial morphogenesis. During the invagination of the Drosophila mesoderm, cells in the center of the prospective mesoderm constrict under the action of actomyosin pulses, while lateral cells elongate towards the center of the mesodermal placode to accommodate the reduction in apical surface of the central cells. Central and lateral cells display strong similarities in terms of gene expression. How are thus this different behaviors (contraction and expansion) accomplished? The authors found that both central and lateral cells assemble actomyosin networks, although lateral cells do it with a certain delay. Mathematical models of cell constriction across the mesoderm using different strain-stress responses showed that strain-induced cell softening was necessary recapitulate the patterns of constriction and expansion observed in vivo. Furthermore, modelling predicts that cells can stretch until the actin networks yield and break. Laser ablation and optogenetic reduction of contractility in central cells results in a reduction in the apical surface area of lateral cells. An optogenetic increase in contractility in lateral cells caused an increase in apical area in central cells. Together, these data suggest that mechanical cues can override and contribute to sculpting genetically defined morphogenetic domains.

    I propose to address the following points before further considering the manuscript:

    MAJOR

    1. Figure 3: following laser ablation of central cells, lateral cells reduce their apical surface. How do the authors know that this reduction in lateral cell apical surface area is an active process, driven by actomyosin-based contraction, rather than a passive response to the expansion of the wound induced by laser ablation? A similar argument could explain the constriction of lateral cells after optogenetic inhibition of actomyosin networks: the central cells relax, expand and compress the lateral cells. To demonstrate active responses of the lateral cells upon laser ablation and optogenetic manipulations of central cells, at the very least the authors should show the distribution of myosin in the lateral cells that constrict and demonstrate the assembly of contractile networks.
    2. Modelling suggests that actin networks yield and break in lateral cells. Does this occur in vivo?
    3. Lines 166-175: The authors propose that constriction of a cell affects the localization of myosin in its neighbors. However, this is not directly measured. The authors should quantify the relative myosin offset in the cells around constricting cells, and show that that offset is greater (and oriented towards the constricting cell) than in cells around expanding cells. There should be a correlation between the relative size change of a cell and the myosin offset (not just concentration) in their neighbours. In addition, does optogenetic activation of constriction in lateral cells affect the offset of myosin networks in central cells?
    4. Fig. 2E-F: the authors argue that the mean myosin concentration in lateral cells at certain times is equivalent to that of central cells earlier in the invagination process. However, the fraction of apical surface area covered by myosin network is consistently lower for lateral cells (and also for central cells that remain unconstricted!). Have the authors considered this fact, and if not, why? Wouldn't this explain, at least in part, why some cells constrict and others do not, if medial myosin networks drive the disassembly of the apical surface? If myosin activity were increased in laterals cells once central cells begin constricting, would that lead to an increased fraction of lateral cell surfaces covered by actomyosin networks and to reduced lateral cell elongation?

    MINOR

    1. Image panels are missing scale bars in many figures.
    2. Fig. 1C'-D': The authors should include a color bar to provide some indication of the scale of the apical areas measured. Same comment for other figures in which apical area is color-coded.
    3. Supp. Fig. 2E-F, G-H and Supp. Fig. 6: what is the difference between myosin intensity and myosin concentration? Junctional vs medial localization? Or summed vs mean pixel value? Please be specific, the difference between intensity and concentration is not clear.
    4. Line 118: Supp. Fig. 2 does not have panels I and K.
    5. Line 223: the authors reference data at 175 sec, but Supp. Fig. 6 does not show any images at that time point. They should be added or a different time point indicated.

    TYPOS

    1. Abstract: "[in a supracellular context" should be "in a supracellular context".
    2. Line 145: should this be a reference to Supp. Fig. 5 instead of Supp. Fig. 4?
    3. Line 166: I am not sure how Supp. Fig. 5 supports this statement. Is this the right figure reference? Should it be Supp. Fig. 4 instead?
    4. Line 881: "representing on line" should be "representing one line".

    OPTIONAL

    Tony Harris' lab showed that the Arf-GEF Steppke antagonizes myosin and facilitates cell deformation at the leading edge of the embryonic epidermis during Drosophila dorsal closure (West et al., Curr Biol, 2017). Does Steppke localize to junctions in lateral but not central mesoderm cells? Does the pattern of Steppke localization in the mesoderm change with manipulations to the contractility of central cells?

    Significance

    This is an interesting study, and one that makes uses of beautiful tools, including quantitative microscopy and image analysis, mathematical modeling and optogenetic manipulations. The prediction that embryonic cells display non-linear stress-strain responses is exciting, as linearity has been the predominant assumption so far. However, I find that model predictions are not well supported by the data, and that alternative interpretations of some results are possible. Additionally, the paper lacks insight into the molecular mechanisms that facilitate stretching (although that could be the subject of a follow-up study).

  5. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Bhide and colleagues present an insightful study of how cellular mechanics influences differential cell behaviour during morphogenesis despite apparent genetic homogeneity of the cellular ensembles. They dissect the extensively studied system of mesoderm invagination in Drosophila, focussing on the differences in cell behaviours between the cells in the middle of the infolding tissue and on the periphery that, as far as we know, share a common gene expression profile. They describe sub-cellular dynamics of major effector of apical constriction morphogenesis, the myosin motor distribution, in the invaginating cells and conclude that differences in myosin levels alone cannot account for the observed differences in cell behaviours. In order to understand the cell behaviour inhomogeneity, they turn to biophysical simulation and in an impressively exhaustive manner substantiate the idea that non-linear effects are required for explaining the phenomenon. This theoretical treatment fits well with the notion that the genetic identity of the cells but rather cell-cell mechanical coupling determine the differences in invaginating cell's behaviours. Additionally, the modelling is consistent with the myosin asymmetry and dynamics in the cells whose behaviours is being contrasted. Complementary, and beautifully executed filament-based modelling of microscopic actomyosin contractility further corroborates this view. Finally, the proposed model of non-linear actomyosin contractility dynamics governing the differential cell behaviour across genetically homogenous cellular field, is challenged by two complementary laser ablation and optogenetic experimental approaches. Overall, the results represent convincing evidence that points the tissue mechanics field of Drosophila mesoderm into an interesting new direction and has general implications for the understanding of the interplay between genetic regulation and emergent behaviours of cells operating in mechanically complex multicellular embryonic context.

    The study is meticulously executed, highly quantitative and combines effectively experiment and theory. I have only minor comments that concern in particular the presentation of the results.

    The paper is very dense and the text does not complement well the results presented in the main figures. Many panels in the Figures are not referred to explicitly. Figure elements are referenced out of order both within and across Figures. Sometimes, particularly, in the last two Figures (3 and 4) the reader is left alone to figure out what the data show (with the appropriately terse legends and without the clear narrative in the text, it is an uphill battle for non-specialists). Some key results are hidden in the sea of elements within the Figure 2 that contains the most important, relevant and impressive data. As an example, on line 168 the authors point to panel 2F to demonstrate the asymmetry of myosin distribution in some cells. To the best of my understanding, this phenomenon is actually shown in Fig 2E which is curiously not referenced at all.

    Similarly, Figure 2K and L provide crucial data substantiating much of the conclusions of the paper. It requires a major effort to understand what the graphs mean.

    The following simulation results are quite impressive and would deserve a separate Figure which could provide more space for explaining what the parameter maps actually show. What is for instance plotted on the Y axis as steepness?

    Secondly, I find the overall narrative of the manuscript needing some reorganisation. The main question is set-up extremely well, however in the middle of the manuscript the focus on the connection between cell behaviours and genetic programs is lost. New conclusions on force transmission between cells emerge, however they are not obviously connected with the question posed from the onset and addressed in the discussion section. My impression is that the authors are conservative in their reasoning, however it does compromise the overall message of the story that should ideally focus on one subject. I find the combined evidence presented sufficiently supportive of the model that is beautifully and eloquently presented in the concluding sentence of the paper:

    "This mechanism, which we propose corresponds to the non-linear behaviour predicted by the models, would apply both to central and to lateral cells, with a catastrophic 'flip' being stochastic and rare in central cells, but reproducible in lateral cells because of the temporal and spatial gradient in which contractions occur."

    This may not turn out to be the entire story or even entirely correct, but it is certainly and exciting way of thinking about the problem. I wish that the manuscript would stay more on this subject throughout and provide intermediate conclusions supporting this model as the story develops.

    Few more minor comments:

    Line 36 - typo Line 97 - starting bracket missing Line 126 - data on intensity are presented here. There is also a panel on concentration (Fig 1H). Where is this discussed? Line 132 - panel 2G - disruptive out of sequence reference to a future figure Line 135 - with this regard - please spell out this important conclusion Line 183 - typo Line 210 - insects do not have intermediate filaments Line 238 - please provide a hint of how such global ablations are performed Line 240 - walk us through the Figure, it is too complex to figure it out alone Line 245 - why is the clear hypothesis mentioned above (point 2) rephrased? Line 273 - vague statement

    Significance

    The results represent convincing evidence that points the tissue mechanics field of Drosophila mesoderm into an interesting new direction and has general implications for the understanding of the interplay between genetic regulation and emergent behaviours of cells operating in mechanically complex multicellular embryonic context.

  6. Version published to 10.1101/2020.10.15.333963v3 on bioRxiv
  7. Version published to 10.1101/2020.10.15.333963v2 on bioRxiv
  8. Version published to 10.1101/2020.10.15.333963v1 on bioRxiv