A ratchet-like apical constriction drives cell ingression during the mouse gastrulation EMT

  1. Alexandre Francou
  2. Kathryn V Anderson
  3. Anna-Katerina Hadjantonakis

  • Curated by eLife

    eLife logo

    eLife assessment

    This study employs live imaging to investigate the movement of mesodermal cells in early mouse embryos. By examining the dynamics of cell behavior in normal and mutant embryos, the authors propose that apical constriction of cells results from pulsed contraction guided by crumbs2 signals. The paper presents beautiful images and adds to the molecular understanding of cell migration during early development.

Reviewed by

Read the full article See related articles

Listed in

Log in to save this article

Abstract

Epithelial-to-mesenchymal transition (EMT) is a fundamental process whereby epithelial cells acquire mesenchymal phenotypes and the ability to migrate. EMT is the hallmark of gastrulation, an evolutionarily conserved developmental process. In mammals, epiblast cells ingress at the primitive streak to form mesoderm. Cells ingress and exit the epiblast epithelial layer and the associated EMT is dynamically regulated and involves a stereotypical sequence of cell behaviors. 3D time-lapse imaging of gastrulating mouse embryos combined with cell and tissue scale data analyses revealed the asynchronous ingression of epiblast cells at the primitive streak. Ingressing cells constrict their apical surfaces in a pulsed ratchet-like fashion through asynchronous shrinkage of apical junctions. A quantitative analysis of the distribution of apical proteins revealed the anisotropic and reciprocal enrichment of members of the actomyosin network and Crumbs2 complexes, potential regulators of asynchronous shrinkage of cell junctions. Loss of function analyses demonstrated a requirement for Crumbs2 in myosin II localization and activity at apical junctions, and as a candidate regulator of actomyosin anisotropy.

Article activity feed

  1. Version published to 10.7554/elife.84019 on eLife
  2. eLife

    Author Response

    Reviewer #1 (Public Review):

    Francou et al. examine the dynamics of cell ingression at the primitive streak during mouse gastrulation and correlate this with the localization of elements of the apical Crumbs complex and the actomyosin cytoskeleton. Using time-lapse live imaging, they show that cells at the primitive streak ingress in a stochastic manner, by constricting their apical surface through a ratcheting shrinkage of individual junctions. Meticulous evaluation of immunofluorescent staining for many elements of the actomyosin contractile process as well as junctional and apical domain elements reveals anisotropic localization of Crumbs2, ZO1, and ppMLC. In addition, the localization of two groups of proteins showed a close correlation - actomyosin regulators and apical and junctional components - but there was a lack of correlation of localization of these two groups of proteins to each other. The localization of actomyosin and its activity, was altered and more homogeneous in Crumbs2-/- embryos, and there was a significant decrease in aPKC and Rock1. The authors conclude from these observations that Crumbs2 regulates anisotropic actomyosin contractility to promote apical constriction and cell ingression.

    The strengths of this manuscript are the very detailed observations on the process of apical constriction and the meticulous evaluation of the localization of the many proteins likely to be involved in the process. While many of the general observations are not new, Francou et al. provide a much richer understanding of this process, as well as a paradigm with which to evaluate the effects of mutations on the gastrulation process. The figures are beautiful, clear, and informative, and support the conclusions made by the authors. The data provide a very compelling picture of both the dynamics of cell behavior and the anisotropies in protein localization associated with it.

    However, much of the Crumbs2 mutant phenotype is not sufficiently explained by the authors' data or conclusions. First, the loss of Crumbs2 does not prevent ingression, as there are mesoderm cells evident between the epiblast and endoderm (Ramkumar et al., 2016, Xiao et al., 2011). There are certainly fewer, and the biggest effect appears to be during the elongation of the axis from E7.75 onward and not during the earlier migratory period (E6.5-E7.75) according to data from both previously published work (Xiao et al., 2011; Ramkumar et al., 2015, 2016) and the data presented here.

    • The reviewer makes a good point regarding the defects observed in Crumbs2 mutant embryos. It is true that in this mutant, a first wave of gastrulation EMT, taking place around E6.5, does not appear to be affected. We interpret this to mean that the gastrulation EMT is a sequential process under differential regulation, and that Crumbs2 is not required for the first wave of cells ingression through the primitive streak, at the onset of gastrulation. Consequently, a small number of early mesodermal cells are produced in Crumbs2 mutants. However, within 24hours of the onset of gastrulation, corresponding to around E7.75, ingression defects are evident in Crumbs2 mutant embryos.

    • For simplicity, these distinct sequential phases of gastrulation regulation, initially independent of Crumbs2, but subsequently dependent, were not initially discussed in our manuscript. We have now elaborated these details in the revised manuscript.

    Nor does the loss of Crumbs2 prevent apical constriction. Ramkumar et al. in their 2016 paper show by live imaging that the major effect of the Crumbs2 mutation is to prevent the cells from detaching from the epithelium, but that the apical domain does undergo constriction, leading to many elongated flask-shaped cells still attached at the apical end. These observations do not fit well with the model proposed by the authors of Crumbs2 regulating anisotropic actomyosin contractility to promote apical constriction and suggest a more complicated story.

    • We thank the reviewer for bringing this up, as it is an important point that we now discuss in greater detail and clarify in the revised manuscript.

    • Importantly, we do not believe our data are in disagreement with the previous study of Ramkumar et al. The precise details of the defect observed in Crumbs2 mutants are still not totally clear. However, we would like to point out that in Ramkumar et al., the timelapse imaging data did not depict cells constricting their surfaces, but rather these data revealed that cells having small apical surfaces failed to detach and delaminate out of the epiblast layer. Thus, this previous study focused on the subsequent step in the process of ingression (delamination), to that being addressed in the present work.

    • Furthermore, epiblast cells outside the domain occupied by the primitive streak, and even some cells positioned on the lateral sides of the embryo, were reported by Ramkumar and colleagues to exhibit abnormally small apical surfaces in Crumbs2 mutants. These cells, at a distance from the primitive streak, will not normally constrict their apical surfaces, since they are not going to undergo the gastrulation EMT, a behavior restricted to the region of the primitive streak. Thus, these previous data do not directly address nor demonstrate that epiblast cells in Crumbs2 mutants undergo apical constriction.

    • Moreover, in Crumbs2 mutants a large number of cells were reported to fail to ingress at the primitive streak, and consequently they were seen to accumulate within the epiblast epithelial layer. Indeed, we believe that the small apical surfaces first reported in Crumbs2 mutants by Ramkumar and colleagues, most likely result from the crowding/jamming of cells within the epiblast layer, and that this causes changes in the shape and volume of cells due to them being spatially constrained. Thus, increased crowding of epithelial cells within a spatially constrained tissue, likely drives a reduction in apical surface area and extensive apico-basal elongation, as observed in Crumbs2 mutants.

    However, the complications of the Crumbs2 mutant do not detract from the value of the basic observations presented in this manuscript, which are solid and well-documented, and will be a valuable resource for the field.

    Reviewer #2 (Public Review):

    In their manuscript, Francou and colleagues study the delamination of epiblast cells into the mesodermal layers using live imaging of mouse embryos cultured ex vivo. By segmenting the apical area of delaminating cells, they quantify extensively the dynamic behavior of delaminating cells. Using immunostaining and crumbs2 mutants, they propose that apical constriction of cells results from pulsed contractions, which could be guided by crumbs2 signals.

    The manuscript is interesting and provides extremely valuable data for our understanding of mouse gastrulation. Occasionally, the manuscript can be a bit confusing and contains a few inaccuracies.

    However, the main issues I have are with some of the interpretations from the authors, which may be incorrect due to limited time resolution (with a 5 min time resolution that was used, it might be difficult to distinguish pulses from measurement noise) and the analysis of immunostaining data, which would require more rigorous quantification.

    • We acknowledge the reviewer’s comments and agree that a shorter time resolution would be ideal to facilitate the detection of constriction pulses of apical surfaces. However, we need to consider that imaging the apical surface of cells within the epiblast layer, which constitutes the most internal surface inside the embryo, is technically challenging in a gastrulating mouse embryo.

    • As suggested by the reviewer, we attempted to image with a shorter time interval than 5min on several different microscope systems and modalities available at our institution (including two different laser point scanning confocals, a spinning disc system, as well as light-sheet microscopes with both upright and inverted configurations) and were not successful in acquiring usable images (having a shorted time-resolution) with the ZO1GFP knock-in reporter. We also need to consider that single-copy GFP knock-in reporters are often dim, thereby exacerbating the issue. In our hands, a high-speed resonant scanning confocal (Nikon A1RHD25) was the system that gave us the best signal-to-noise ratio, spatial resolution and temporal resolution, and was the set-up we used for our most recent live imaging experiments. Using this system, we were able to acquire a limited number of time-lapses with a time resolution of 2min, but none with a shorter time interval, and from our analyses, we determined that movies with a 2min time interval did not yield increased detail over movies with 5min time intervals to warrant a detailed reanalysis. We have provided additional detail relating to these technical issues within the revised manuscript and edited some of the conclusions.

    • We acknowledge that immunostaining is not the most quantitative method, but we were unable to come up with alternative methods that can be used with our samples. We believe the junctional reduction of Myosin, aPKC and Rock1 is generally due to a nonrecruitment or activation of these proteins at junctions, and do not reflect their reduced expression at the gene or protein level. We do not believe that methods such as RTqPCR or Western blotting would be informative in the context in which we are looking, especially since they do not yield spatial resolution. Furthermore, we would need to isolate primitive streak cells to consider applying these methods, and we do not believe they would provide a sufficient improvement over immunostaining.

    • By contrast to the live imaging, which was performed by placing the objective at the posterior side of the embryo in closest proximity to the outer visceral endoderm layer, for fixed tissue imaging, embryos were microdissected to recover the posterior side containing the primitive streak. Microdissected posterior regions were imaged on the side of the cavity by placing the objective in closest proximity to the inner epiblast layer, which permitted direct access to the apical surface of epiblast cells at the primitive streak. In this fixed tissue imaging configuration, the apical surfaces of cells in WT and Crumbs2 mutants were in closest proximity to the imaging objective and thus directly accessible. Thus, any difference in tissue thickness on the other side of the epithelium did not interfere with light penetration. We have edited the figures and include schematics to clarify how the objective positions are flipped with respect to the primitive streak regions at the embryo’s posterior for live vs. fixed tissue imaging.

    • We have now measured the signal intensity in the cytoplasmic region of WT and Crumbs2 mutant embryos, and junctional intensity measurements have been normalized to cytoplasmic intensities.

    Reviewer #3 (Public Review):

    The manuscript by Francou et al investigated cellular mechanisms of epiblast ingression during mouse gastrulation. The authors wanted to know whether/how epiblast cell-cell junctional dynamics correlate with apical constriction and subsequent ingression. Because mouse gastrula adopts an inverted-cup morphology (as a result of differential invasive behavior of polar and mural trophoblast cells), epiblast cells are located in the innermost position and are difficult to image. This is more so when one wants to perform live imaging of epiblast cells' apical surface. The authors tackled such problems/limitations by using a combination of ZO-1 GFP line, confocal time-lapse microscopy, fixed embryo immunostaining, and Crumbs2 mutant embryos. The authors observed that apical constriction was associated with cell ingression, that this constriction occurred in a pulsed fashion (i.e., 2-4 cycles with phases of contraction and expansion, eventually leading to reduction of apical surface and ingression), that this constriction took place asynchronously (i.e., neighboring epiblast cells did not exhibit coordinated behavior) and that junctional shrinkage during apical constriction also occurred in a pulsed and asynchronous manner. The authors also investigated localization/co-localization of several apical proteins (Crumbs2, Myosin2B, pMLC, ppMLC, Rock1, F-actin, PatJ, and aPKC) in fixed samples, uncovering somewhat reciprocal distribution of two groups of proteins (represented by Myosin2B in one group, and Crumbs2 in the other). Finally, the authors showed that Crumbs2 -/- embryos had disturbed actomyosin distribution/levels without affecting junctional integrity (partially explaining the ingression defect reported in Crumbs2 -/- mutant embryos). Overall, this manuscript offers high-quality live imaging data on the dynamic remodeling of epiblast apical junctions during mouse gastrulation.

    It would be interesting to see whether phenomena reported in this manuscript can be extended to the entire primitive streak (or are they specific only to a subset of mesoderm precursors) and to the entire period of mesendoderm formation. More importantly, it would be interesting to see whether the ingression behavior seen here is representative of all eutherian mammals regardless of their gastrular topography.

    • The reviewer raises a very interesting and important point. We focused our data analysis on a middle region in the proximo-distal axis of the embryo, because this is the most optically accessible and the flattest region of the posterior of the embryo to analyze. We also focused on the E7.5 stage of development when the primitive streak is fully elongated, so as to capture as many ingression events within a single time-lapse experiment as possible. Due to the difficulties associated with live imaging the apical epiblast layer of embryos at these stages, we chose to focus our analysis on a defined region of the embryo and a defined period of time. We acknowledge that it will be important to analyze different regions of the primitive streak and at different stages of gastrulation to glean any general versus more distinct modes of epiblast cell ingression, but given the technical difficulties discussed we believe that any extended analysis is beyond the scope of the current study.

    • We also agree that it would be interesting to know if the ingression behavior we observe in the mouse embryo is representative of all mammals, and even more generally of amniotes, but this is beyond the scope of our study.

  3. eLife

    eLife assessment

    This study employs live imaging to investigate the movement of mesodermal cells in early mouse embryos. By examining the dynamics of cell behavior in normal and mutant embryos, the authors propose that apical constriction of cells results from pulsed contraction guided by crumbs2 signals. The paper presents beautiful images and adds to the molecular understanding of cell migration during early development.

  4. eLife

    Reviewer #1 (Public Review):

    Francou et al. examine the dynamics of cell ingression at the primitive streak during mouse gastrulation and correlate this with the localization of elements of the apical Crumbs complex and the actomyosin cytoskeleton. Using time-lapse live imaging, they show that cells at the primitive streak ingress in a stochastic manner, by constricting their apical surface through a ratcheting shrinkage of individual junctions. Meticulous evaluation of immunofluorescent staining for many elements of the actomyosin contractile process as well as junctional and apical domain elements reveals anisotropic localization of Crumbs2, ZO1, and ppMLC. In addition, the localization of two groups of proteins showed a close correlation - actomyosin regulators and apical and junctional components - but there was a lack of correlation of localization of these two groups of proteins to each other. The localization of actomyosin and its activity, was altered and more homogeneous in Crumbs2-/- embryos, and there was a significant decrease in aPKC and Rock1. The authors conclude from these observations that Crumbs2 regulates anisotropic actomyosin contractility to promote apical constriction and cell ingression.

    The strengths of this manuscript are the very detailed observations on the process of apical constriction and the meticulous evaluation of the localization of the many proteins likely to be involved in the process. While many of the general observations are not new, Francou et al. provide a much richer understanding of this process, as well as a paradigm with which to evaluate the effects of mutations on the gastrulation process. The figures are beautiful, clear, and informative, and support the conclusions made by the authors. The data provide a very compelling picture of both the dynamics of cell behavior and the anisotropies in protein localization associated with it.

    However, much of the Crumbs2 mutant phenotype is not sufficiently explained by the authors' data or conclusions. First, the loss of Crumbs2 does not prevent ingression, as there are mesoderm cells evident between the epiblast and endoderm (Ramkumar et al., 2016, Xiao et al., 2011). There are certainly fewer, and the biggest effect appears to be during the elongation of the axis from E7.75 onward and not during the earlier migratory period (E6.5-E7.75) according to data from both previously published work (Xiao et al., 2011; Ramkumar et al., 2015, 2016) and the data presented here. Nor does the loss of Crumbs2 prevent apical constriction. Ramkumar et al. in their 2016 paper show by live imaging that the major effect of the Crumbs2 mutation is to prevent the cells from detaching from the epithelium, but that the apical domain does undergo constriction, leading to many elongated flask-shaped cells still attached at the apical end. These observations do not fit well with the model proposed by the authors of Crumbs2 regulating anisotropic actomyosin contractility to promote apical constriction and suggest a more complicated story. However, the complications of the Crumbs2 mutant do not detract from the value of the basic observations presented in this manuscript, which are solid and well-documented, and will be a valuable resource for the field.

  5. eLife

    Reviewer #2 (Public Review):

    In their manuscript, Francou and colleagues study the delamination of epiblast cells into the mesodermal layers using live imaging of mouse embryos cultured ex vivo. By segmenting the apical area of delaminating cells, they quantify extensively the dynamic behavior of delaminating cells. Using immunostaining and crumbs2 mutants, they propose that apical constriction of cells results from pulsed contractions, which could be guided by crumbs2 signals.

    The manuscript is interesting and provides extremely valuable data for our understanding of mouse gastrulation. Occasionally, the manuscript can be a bit confusing and contains a few inaccuracies. However, the main issues I have are with some of the interpretations from the authors, which may be incorrect due to limited time resolution (with a 5 min time resolution that was used, it might be difficult to distinguish pulses from measurement noise) and the analysis of immunostaining data, which would require more rigorous quantification.

  6. eLife

    **Reviewer #3 (Public Review):
    **
    The manuscript by Francou et al investigated cellular mechanisms of epiblast ingression during mouse gastrulation. The authors wanted to know whether/how epiblast cell-cell junctional dynamics correlate with apical constriction and subsequent ingression. Because mouse gastrula adopts an inverted-cup morphology (as a result of differential invasive behavior of polar and mural trophoblast cells), epiblast cells are located in the innermost position and are difficult to image. This is more so when one wants to perform live imaging of epiblast cells' apical surface. The authors tackled such problems/limitations by using a combination of ZO-1 GFP line, confocal time-lapse microscopy, fixed embryo immunostaining, and Crumbs2 mutant embryos. The authors observed that apical constriction was associated with cell ingression, that this constriction occurred in a pulsed fashion (i.e., 2-4 cycles with phases of contraction and expansion, eventually leading to reduction of apical surface and ingression), that this constriction took place asynchronously (i.e., neighboring epiblast cells did not exhibit coordinated behavior) and that junctional shrinkage during apical constriction also occurred in a pulsed and asynchronous manner. The authors also investigated localization/co-localization of several apical proteins (Crumbs2, Myosin2B, pMLC, ppMLC, Rock1, F-actin, PatJ, and aPKC) in fixed samples, uncovering somewhat reciprocal distribution of two groups of proteins (represented by Myosin2B in one group, and Crumbs2 in the other). Finally, the authors showed that Crumbs2 -/- embryos had disturbed actomyosin distribution/levels without affecting junctional integrity (partially explaining the ingression defect reported in Crumbs2 -/- mutant embryos). Overall, this manuscript offers high-quality live imaging data on the dynamic remodeling of epiblast apical junctions during mouse gastrulation. It would be interesting to see whether phenomena reported in this manuscript can be extended to the entire primitive streak (or are they specific only to a subset of mesoderm precursors) and to the entire period of mesendoderm formation. More importantly, it would be interesting to see whether the ingression behavior seen here is representative of all eutherian mammals regardless of their gastrular topography.

  7. Version published to 10.1101/2022.04.30.489707v4 on bioRxiv
  8. Version published to 10.1101/2022.04.30.489707v3 on bioRxiv
  9. Version published to 10.1101/2022.04.30.489707v2 on bioRxiv
  10. Version published to 10.1101/2022.04.30.489707v1 on bioRxiv