Cell contacts and pericellular matrix in the Xenopus gastrula chordamesoderm

  1. Olivia Luu
  2. Debanjan Barua
  3. Rudolf Winklbauer

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Convergent extension of the chordamesoderm is the best-examined gastrulation movement in Xenopus. Here we study general features of cell-cell contacts in this tissue by combining depletion of adhesion factors C-cadherin, Syndecan-4, fibronectin, and hyaluronic acid, the analysis of respective contact width spectra and contact angles, and La 3+ staining of the pericellular matrix. We provide evidence that like in other gastrula tissues, cell-cell adhesion in the chordamesoderm is largely mediated by different types of pericellular matrix. Specific glycocalyx structures previously identified in Xenopus gastrula tissues are absent in chordamesoderm but other contact types like 10-20 nm wide La 3+ stained structures are present instead. Knockdown of any of the adhesion factors reduces the abundance of cell contacts but not the average relative adhesiveness of the remaining ones: a decrease of adhesiveness at low contact widths is compensated by an increase of contact widths and an increase of adhesiveness proportional to width. From the adhesiveness-width relationship, we derive a model of chordamesoderm cell adhesion that involves the interdigitation of distinct pericellular matrix units. Quantitative description of pericellular matrix deployment suggests that reduced contact abundance upon adhesion factor depletion is due to some contact types becoming non-adhesive and others being lost.

Article activity feed

  1. 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

    We thank the reviewers for their constructive criticism that helped us to improve the paper. We modified Fig.6I and Fig.7, replaced Fig.8, and added supplementary Figs. 3-5 and supplementary Tables S1-2. The manuscript was extensively re-written. A new paragraph was added in the Discussion section where relative adhesiveness was related to absolute adhesion strength and the cadherin knockdown result to earlier findings.

    Reviewer #1 (Evidence, reproducibility and clarity):

    Summary: This work examines the relationship between cell-cell contacts and pericellular matrix in Xenopus chordamesoderm, which is a tissue actively involved in convergent extension during gastrulation. By lanthanum staining of pericellular materials, the authors found that different types of pericellular matrix are present in cell-cell contacts in the chordamesoderm, which may mediate cell-cell adhesion. Knockdown of C-cadherin, Syndecan-4, fibronectin, and hyaluronic acid leads to the reduced abundance of cell contacts and cell packing density, but this does not seem to affect convergent extension. Based on these observations, the authors propose a model in which cell-cell contacts involve the interdigitation of distinct pericellular matrix units.
    Major points:

    1. Knockdown of adhesion molecules separates cells and leads to wide contacts with large interstitial spaces. Data in figure 1 show loosely packed morphant chordamesoderm cells. Intuitively, these should reduce cell-cell adhesion. However, a main conclusion from this manuscript is that reduced abundance of narrower contacts does not decrease adhesiveness. Although depletion of adhesion molecules modifies but not abolishes a contact, non-attached free surfaces increase significantly in morphant cells. It is therefore not easy to understand that how reduced cell contacts have no effect on cell adhesion.

    We added a section to the Discussion to address this issue (p.11ff). We show in the Results section (modified Fig.7) that relative adhesiveness is indeed significantly reduced in the morphants (Syn-4 always being the exception) when compared in the contact width range of normal chordamesoderm. However, contact width is strongly increased in the morphants, and adhesiveness increases linearly with width. We argue that these effects compensate for the initial lowering of adhesiveness. In other words, adhesive contacts become shorter (more gap surface) but wider (see Fig.6I), and become the more adhesive the wider they become. As in the original version of this paper, we then propose a model that explains the empirically observed increase of adhesiveness with width. How the abundance of cell-cell contact is reduced is less clear yet. Pericellular matrix deployment and structure is strongly affected by adhesion factor knockdown, and contact types are altered. Some contact types seem to widen but remain adhesive, others become non-adhesive, and still others may disappear without being replaced (see last paragraph of Discussion). To add detail to these notions and clarify this important issue to satisfaction will require future research.

    Importantly, the adhesiveness was not experimentally tested.

    Due to external circumstances, we were unable to perform additional experiments. However, we used our previously published quantitative data on adhesion in gastrula tissues including the chordamesoderm to interpret our present results for normal and C-cad-depleted chordamesoderm, and to relate relative adhesiveness to absolute adhesion strength, in a new section of the Discussion (p.11ff).

    1. It is surprising that reduced cell contacts, at least narrower cell contacts, do not affect convergent extension. Does this mean that active cell behavior changes in the chordamesoderm, which are required for convergent extension, are independent of cell contact types?

    We actually claimed that all treatments inhibited convergent extension, except for Syn-4 (Barua et al. 2021, and this manuscript, p.3, Fig.1B,C). Syn-4 knockdown had a dramatic effect on cell contacts, cell density and cell shape but none on convergent extension, at least up to the middle gastrula stage. This is surprising and does not fit easily to current views of cell intercalation during convergent extension, but analysing the underlying cell behaviors is beyond the scope of this article.

    1. Although the formation and localization of pericellular materials are differentially affected after knockdown of adhesion molecules, there is no clear evidence showing that different types of pericellular matrix mediate cell-cell adhesion in the chordamesoderm. It is possible that the disrupted distribution of pericellular materials in morphants only represents a secondary consequence of changed cell contacts. This may be supported by the fact that knockdown of adhesion molecules reduces narrow contacts and increases LSM-free gaps.
    2. The relationship between contact width spectra and LSM is also very elusive. Again, changes in contact width or abundance and distribution of LSM may be indirectly caused by loss of adhesion molecules. Therefore, although knockdown of adhesion molecules leads to changes of LSM localization, it cannot be concluded that cell-cell contacts in chordamesoderm are mediated different types of pericellular matrix.

    We find it difficult to interpret for example Fig.5A-F other than assuming an adhesive role for the pericellular matrix, in this case LSM, in normal and morphant tissue. What else would here hold two cells between two gaps together? The contacts are often much too wide for cadherin-cadherin binding. We indeed believe that changes in contact width or abundance are caused by the loss of adhesion molecules, directly or indirectly. Our LSM images show that remarkably, modified contacts (e.g. Fig.3D,F; Fig.5B,C) are still able to keep cells together over some distance, between interstitial gaps, and our quantitative data indicate similarly that e.g. contact widening is consistent with continued adhesion. However, some of the contacts may become non-adhesive, or be lost without being replaced, increasing non-adhesive gap surface. This is discussed now on p.11, middle paragraph.

    1. In contrast to the present observations, works by others using the same morpholinos have shown that Cadherin-dependent cell adhesion, fibronectin-rich extracellular matrix, and Syndecan-4-regulated non-canonical Wnt signaling are required for convergent extension. These discrepancies need to be appropriately addressed.

    As mentioned above, we found that all treatments affected convergent extension, as expected from the work of others and our own, except for Syn-4 depletion. We noticed that in the paper by Munoz et al. on Syn-4 overexpression and knockdown, only late gastrula/early neurula stages were evaluated. Syn-4 knockdown produced moderately strong axis defects, perhaps in part related to impaired neural plate closure. Unfortunately, we did not follow our morphants to these later stages to see whether defects developed then. But our main interest here is cell-cell contacts.

    1. If LSM and LSM-free contacts are similarly adhesive, what will be role of LSM in cell adhesion and how cell adhesion is established in these LSM-free contacts?

    We discuss now more explicitly the notion that gastrula non-epithelial cell adhesion is mediated by a mosaic of pericellular matrix patches of different composition, some containing LSM in different configurations, others not, but each similarly adhesive.

    Minor points:

    1. It may be helpful to clearly define the pericellular matrix in this particular context and its relationship with LSM. It is also necessary to clarify whether the adhesion molecules examined in this work are considered as components of the pericellular matrix.

    We explain the use of these terms at the end of the first paragraph of the Introduction. The most general term is pericellular matrix; part of it is La3+ labeled – LSM; and some of the LSM can be compared to structures which in other systems are termed glycocalyx. We consider the adhesion molecules examined to be part of the pericellular matrix but are aware of other putative functions, like in cell signaling, which may indirectly affect contacts and thus contribute nevertheless to the phenomena studied here.

    1. In figure 1B, it appears that the Cadherin morphant has defects in chordamesoderm elongation and archenteron formation, suggesting impaired convergent extension.

    We find, in agreement with the work of others, that C-cad knockdown impairs convergent extension, and mention this when we describe Fig.1B.

    1. In figure 1C, the Syndecan-4 morphant gastrula clearly shows enhanced anteroposterior elongation of chordamesoderm and archenteron in comparison with the wild-type embryo. This seems to suggest that loss of Syndecan-4 promotes the movements of convergent extension. However, previous studies indicate that both gain and loss of Syndecan-4 impairs convergent extension.

    As mentioned above, late gastrula/early neurula stages were evaluated in the Munoz et al. paper, mid-gastrula stages in our work. One possible explanation would be that mild axis defects develop later, partly in connection with neural tube elongation and closure.

    1. Ideally, in knockdown experiments, control embryos should be injected with corresponding mismatch morpholinos.

    We explain in the Methods section that we only used morpholinos that were extensively characterized in previous publications.

    1. In figure 1E, it is unclear what type of cell contacts the light green arrowheads indicate.

    This is explained now in the figure legend.

    1. Figure 1 legend, "(wt) is from Barua et al. 2021". I am not sure it is appropriate to use previously published data.

    The present data were derived by further evaluations of the same samples and TEM sections as used in Barua et al. 2021. We show the previously published data (acknowledged in the legends) here for easy comparison (instead of citing the previous paper).

    1. There is no light blue arrowhead in figure 2, and in figure 3B and 3I, it seems that the same colored arrows are used to indicate different structures.

    This has been corrected.

    1. Triple-layered contacts are not clearly defined.

    We define this term now repeatedly, as consisting of two LSM layers enclosing a non-labeled layer between them.

    1. Page 2, "based on driven by" should be either "based on" or "driven by".

    Has been corrected.

    1. Page 8, "selectin" should be "selecting".

    Has been corrected.

    Reviewer #1 (Significance):

    Strengths:
    Demonstrated the effects of several adhesion molecules on the formation of cell contacts and pericellular matrix in Xenopus chordamesoderm.
    Limitations:
    The significance of chordamesoderm cell contact changes in convergent extension or gastrulation is not clear;

    Effects on gastrulation of PCM or membrane adhesion molecule depletion have very often been described as mediated by effects on cell signaling. Without excluding such possibilities, we liked to redirect attention here to other putative mechanisms by describing basic effects of treatments on cell-cell contacts including PCM deployment and structure. Future work must relate the specific, often dramatic, contact changes upon depletion of a specific factor to cell behavior during convergent extension and other tissue movements.

    there is no direct evidence showing the functional link between pericellular matrix, cell contacts and cell adhesion;

    Please see our response to main points 3 and 4 above.

    the absence of effects on convergent extension after depletion of several adhesion molecules is not fully consistent with previous reports.

    Please see our response to main points 2 and 5 and minor point 3 above.

    Advance: This work likely provides some fundamental and methodological advances for studying cell-cell adhesion. It shows promise for elucidating mechanisms underlying the regulation of cell contact changes in tissues involved in morphogenetic movements.
    Audience:
    This work likely interests readership studying embryonic cell adhesion in the field of developmental biology and cell biology. It may be also potentially interesting for people working on glycocalyx pericellular matrix in adult tissues.

    Reviewer #2 (Evidence, reproducibility and clarity):

    Summary: During gastrulation, cells within vertebrate embryos require the ability to both adhere to one another and rearrange with their neighbors to shape the emerging body plan. These authors posit that such flexible adhesive contacts are mediated in part by the pericellular matrix (PCM), including multiple types of glycocalyces containing molecules such as fibronectin, hyaluronic acid, and syndecans, which they previously characterized in multiple embryonic tissues (Barua et al, PNAS, 2021). Here, in a follow-up to their 2021 study, the authors use electron microscopy to characterize the pericellular matrix within the chordamesoderm of Xenopus gastrulae. They identify several types of adhesive contacts within the chordamesoderm and assess how they are altered in the absence of key PCM molecules via morpholino knock-down. They conclude that syndecan-4 and hyaluronic acid comprise and promote assembly of PCM plaques whereas fibronectin and C-cadherin anchor them to cell surfaces. Cell packing density is decreased upon loss of all 4 of these molecules, which the authors attribute to a decrease in the number of cell contacts without affecting the strength of the remaining contacts. They further conclude that adhesiveness increases linearly with contact width, and that this relationship is unaffected by loss of any aforementioned adhesive/ PCM molecules.

    Major comments:
    Many conclusions in this manuscript are based on measurements of cell contact angles, which indicate the reduction of tension at cell contacts vs. free cell surfaces and thus relative adhesive strength. While this lab previously applied the same approach to live tissues (David et al, 2014), it is not clear to what extent such measurements accurately reflect adhesive strength in fixed tissues and/or electron micrographs. Especially given the issue of random sectioning planes, which cause distortion of contact angles. Although a correction was applied, the authors note this is not theoretically derived because the heterogeneity of gap sizes made such calculations too difficult. Indeed, it appears that the large gaps between cells within morphant embryos affect contact angle measurements, but if this is corrected for in any way, it is not mentioned.

    Geometrically determined contact angle distortion should affect angle or relative adhesiveness distributions in all conditions or treatments similarly and thus should not or only little affect comparisons of distribution peaks, averages, etc. Beyond this effect of random sectioning planes, we don’t see how large contact width should by itself affect measurements of angles.

    Because this is the sole measure of cell adhesion provided in the study, this reviewer is not convinced of the conclusion that loss of PCM components does not affect adhesive strength.

    In response to this criticism, we re-evaluated our adhesiveness-width data (Fig.7A-E). We noticed that there is indeed a reduction of relative adhesiveness when morphants are compared to normal chordamesoderm within the width range of the latter. But the addition of increased widths in the morphants and the linear increase of adhesiveness with width compensated or overcompensated the initial reduction of adhesiveness.

    Could such measurements not be made from live cells/tissues after manipulating PCM components, as the lab has done previously? Because the lab already has the necessary reagents and expertise for such experiments, the time and resources needed for such measurements shouldn't be prohibitive.

    Due to circumstances, we were unable to perform additional experiments. However, we used our previously published quantitative data on adhesion in gastrula tissues including the chordamesoderm to analyze our present results for normal and C-cad-depleted chordamesoderm, and to relate relative adhesiveness to absolute adhesion strength, in a section added to the Discussion (p.11ff).

    • As mentioned above, these authors previously measured adhesive strength in live Xenopus cells and tissues (David et al, 2014). In that study, they found that C-cadherin MO reduced relative adhesiveness whereas the current study found that relative adhesiveness actually increases in this condition. What explains this discrepancy?

    We explain now in the new Discussion section (p.11ff) and with the help of supplementary Figure S5 how adhesion strength and relative adhesiveness are related overall (tissue surface vs. cell contacts) and at gaps within a tissue (gap free cell surface vs. cell contacts). In the previous study (David et al, 2014), we discussed relative adhesiveness in relation to overall adhesion strength, and both are decreased upon C-cad knockdown. Here we examined these parameters at interstitial gaps, where we find a small increase of relative adhesiveness, due to overcompensation caused by a strong increase of adhesiveness with contact width. Using our David et al, 2014 data we quantitated the effects. We previously found a similar increase of relative adhesiveness at gaps in C-cad morphant ectoderm (Barua et al. 2017) which we could not explain at the time, but explain now by analogy to our chordamesoderm results.

    • No control morpholinos are used, and for the morpholinos that are used, the doses are very large. An equally high dose of control MO should be used to ensure that all observed phenotypes are specific.

    We detail in the Methods section that we used here and in previous publications only previously characterized morpholinos.

    • It appears that all the images analyzed were collected in the sagittal plane, and the analyses don't seem to consider the intrinsic polarity of the chordamesoderm. For example: cells in different positions within the tissue (basal vs. apical), or that WT chordamesoderm cells are mediolaterally polarized and actively intercalating whereas disruption of PCM components like fibronectin disrupts cell intercalation and randomizes cell polarity. It is possible that 1) cell-matrix (in basal cells) and 2) cell-cell (during intercalation) interactions may affect the measurements made in this study. In other words, that cell contacts could differ by position within the embryo and intercalation/polarity status... have such effects been accounted for in the current analysis?

    Here we only analyzed cell contacts deep in the chordamesoderm. Basal contacts were examined to some extent in Barua and Winklbauer, 2022, apical contacts not yet. Our present analysis is based on sagittal sections. The cells in the chordamesoderm are elongated and aligned mediolaterally but not in register, i.e. they are randomly wedged between each other. Thus, all mediolateral positions in cells should be present in our samples. Nevertheless, trends in the occurrence of contacts related to medial-to-lateral positions on cells (e.g. recognizable in spindle-shaped cells as wide vs narrow cell cross-sections) may have escaped our attention, and in particular, the protrusion-bearing medial and lateral ends of cells may develop special contacts. However, our goal in this study was to analyse basic properties of cell-cell contacts in this tissue, as a foundation for further detailed studies.

    • In this study, the authors state that chordamesoderm movements are preserved in syndecan-4 morphants, and in their 2021 article (Barua et al) they state that convergent extension movements are accelerated. But another study describing this MO found that it causes severe convergent extension defects (Munoz et al, NCB, 2006). What explains this discrepancy?

    In their knockdown experiments, Munoz et al. find relatively mild axis defects in late gastrula/early neurula stage embryos while we studied the mid-gastrula. Perhaps defects develop during later stages in Syn-4 morphant embryos.

    Also, the syn-4 morphant showed in Fig. 1 appears more developmentally advanced than the other embryo... if the embryos are not stage matched it could affect the measurements and conclusions drawn from them.

    Stage matching was not possible since C-cad and FN morphants did not involute or engage in convergent extension (i.e. were arrested at the initial gastrula stage), Syn-4 morphants appeared to gastrulate faster than normally. Therefore, embryos were strictly time matched. A limitation remains, that the time course of cell contact development over gastrulation was considered low priority in this initial study and was thus not determined.

    • In figure 7, the authors plot relative adhesion (measured from contact angles) vs. contact width, then fit regression lines to the lower boundaries of these scatter plots. It is not clear why this analysis is focused only on the lower boundaries rather than considering the full spread of the data. Particularly for syn-4 morphants, whose values do not appear to be concentrated along the lower boundary. This analysis is further confused by the introduction of alpha*, which represents relative adhesiveness relative to the regression.

    The lower boundary line is most convenient to extract (Fig.7A’-E’). But we agree that the “interior” of the scatter plot distribution should also be analyzed. Using average adhesiveness gives rise to artifacts since the density of data points decreases strongly with contact width but also with distance from the lower boundary, leading to the preferential disappearance of large adhesiveness values for higher widths. Instead, we constructed a line tracing the highest density in the scatter plot near the lower boundary (Fig.7B’’-E’’), by determining the positions of adhesiveness distribution peaks in consecutive width brackets (new Fig.8, Fig.S3). We abstained from introducing alpha*.

    • Based on these regression lines alone, the authors conclude that all 4 conditions are similar enough to pool the data for further analysis. If these contacts have different properties, which the data in Figures 1-6 suggest they do, it seems inappropriate to pool them together.

    We no longer pooled the data, except in supplementary Fig.S4 where we consider angle distortion. Instead, we show in Fig.8 relative-adhesiveness frequency distributions for different treatments and width brackets. This emphasizes differences between the different adhesion factor depletions and shows that adhesiveness is not simply normal or log-normal distributed, in agreement with different contact types contributing differently though similarly to overall adhesion. It also allows to follow main peaks as they shift position with width, roughly in proportion to the lower surface boundary.

    Based on this pooling, the authors then conclude that relative adhesiveness increases linearly with contact width over the entire width range, regardless of adhesion factor depletion. This again assumes that all contacts (morphant and WT) are functionally equivalent, and that what is observed in morphant embryos in very wide contacts would also hold true in WT contacts. But because WT contacts occupy only a small portion of the width range, we cannot know how they would behave if scaled to be wider, and I am not convinced that very wide morphant contacts are representative of or functionally equivalent to WT. In other words, we cannot know that contact width is the only factor increasing their relative adhesion, given the experimental manipulations that structurally alter these contacts.

    Although differences between contact types are apparent, we think that the contacts function very similarly. We still hold that relative adhesiveness increases with contact width, as seen in each of the separate plots for wt and adhesion factor depletions. But re-evaluating the alpha-width scatter plots now we show that in the narrow width range of normal chordamesoderm, C-cad, FN and Has depletions show similar, significantly decreased relative adhesiveness (Fig.7A-E). With alpha proportional to width, and width strongly increased in morphants, this initial decrease is compensated in total adhesiveness averages. The relative independence of adhesiveness from contact type could hint at non-specific PCM-PCM adhesion (Winklbauer, 2019). We think that although adhesion factor depletion leads to the loss of some contact types or renders others non-adhesive (thus lowering contact abundances), it modifies some contact types (e.g. by widening them) while only moderately lowering their adhesiveness per unit interaction surface.

    Minor comments

    • In their descriptions of PCM in different experimental conditions, the authors overstate some conclusions drawn from EM data. For example, that type I glycocalyces are absent in chordamesoderm (although this signal is only reduced),

    We qualified the statement.

    or that because the Has2 morphant phenotype is intermediate between C-cad and fibronectin morphants this indicates an adhesive role for hyaluronic acid.

    Overall, Has2MO increases the abundance of gaps, i.e. HA normally reduces gaps between cells, strongly suggesting an adhesive role of HA. HA is also required for the formation of 10-20 nm gaps, again proposing a direct or at least indirect adhesion-promoting role.

    • The authors state of the data in figure 1 that "All treatments significantly increase the size of non-adhesive gaps", but they don't show a quantification of the gaps size (they show the abundance).

    Has been corrected.

    • The authors state that LSM contacts exist as 10-20 and 20-50 nm subtypes. It is not clear what about the data suggest this division.

    In the LSM width difference spectra, CadMO and SynMO both increase the abundances of ≤ 20 nm contacts and decrease those of 20-50 nm contacts (Fig.4). The different response suggests at least two differently reacting subtypes.

    • In the same paragraph, the authors state that "C-cad and Syn-4... favor LSM width between 20-50 nm." What is meant by "favor"? Given that the number of 20 nm contacts is increased and 50 nm contacts is decreased in both conditions, this statement is unclear.

    The whole paragraph has been reworded.

    • On page 7, the authors say that the size of LSM structures is "consistent with larger plaques being assembled from small units", but if that were the case, wouldn't the plaque sizes be multiples of the size of a single unit? I.e. 100, 200, and 300 nm peaks? Because this is not the case, the data seem more consistent with a continuous range of LSM plaque sizes than with discrete units.

    The size of the units has a peak at 100 nm but a long tail (Fig.6F-H). Moreover, we discuss lateral compression (piling up of PCM material) or active stretching of plaques (to separate units for interdigitation), all factors that would blur plaque length patterns, i.e. we did not expect plaque sizes to be multiples of 100 nm.

    • On page 8, the authors refer repeatedly to LSM volume. Given that these measurements are made from TEM sections, how is volume being measured?

    This is explained now (p.7).

    • The authors present a model in which PCM interdigitates within cell contacts, but this is based on measurements from static tissues alone. Could the measurements of contact width instead be explained by compression of the PCM or some other mechanism? The data as presented don't rule out such possibilities.

    The model is in agreement with the linear increase of relative adhesiveness with contact width, with LSM height at gap surfaces not adding up to adjacent contact width, with visible interdigitation of glycocalyx units (“bushes”) described previously for prechordal mesoderm (Barua et al. 2021), and with the good agreement of calculated unit size with the size of measured LSM units. In addition, it agrees with literature data on endothelial glycocalyx plaques being composed of 100 nm units and of complete interpenetration of glycocalyces during blood cell adhesion.

    Some terms used are not clear, for example: "partial LSM", "triple layer contact", "random removal [of LSM plaques]".

    We point out the meaning of the terms now more clearly. That “partial LSM” is identical with “triple layer contact” (but shorter, for use in figure) is explained in the legend to fig.6.

    • In figure 5, the graphs depict negative "abundance". Recommend "difference in abundance" instead.

    Done. For shortness, Δ Abundance.

    • Statistics: In figure 1I, it is not clear what the asterisk in this graph means or if statistical differences between these groups was determined. And in figure 6, some groups are marked as n.s., but P values for groups that are statistically different are not presented.

    The asterisk in fig.1I was meant to indicate that this column is from Debanjan et al. 2021, but this is indicated by different shading and mentioned in the legend. The non-used n.s. marks were removed.

    Reviewer #2 (Significance):

    This detailed electron microscopy study advances our understanding of pericellular matrix within vertebrate embryos and how loss of its constituent molecules affects cell interactions. It further addresses the relationship between structurally distinct pericellular matrices and their adhesive properties, although this analysis is less convincing. This study adds to a body of literature in which cell-cell and cell-matrix adhesion are known to regulate morphogenetic cell movements, but how such contacts are remodeled as cells rearrange is poorly understood. Previous work has also used measurements from live cells, embryos, and tissues to infer physical forces within embryos such as adhesive strength, cortical tension, and viscosity. This work follows up directly on a previous study from this group that characterized glycocalyces within various tissues within Xenopus gastrulae by electron microscopy. The hypothesis that pericellular matrix enables flexible/fluid adhesion within highly dynamic embryonic tissues is exciting, and is likely to be of interest to developmental biologists - particularly those who apply mechanical concepts to embryos. However, additional evidence, preferably from live tissues and embryos, is needed to support this hypothesis. This assessment is based on over 15 years' experience studying gastrulation morphogenesis in multiple vertebrate species.

  2. 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

    Summary:

    During gastrulation, cells within vertebrate embryos require the ability to both adhere to one another and rearrange with their neighbors to shape the emerging body plan. These authors posit that such flexible adhesive contacts are mediated in part by the pericellular matrix (PCM), including multiple types of glycocalyces containing molecules such as fibronectin, hyaluronic acid, and syndecans, which they previously characterized in multiple embryonic tissues (Barua et al, PNAS, 2021). Here, in a follow-up to their 2021 study, the authors use electron microscopy to characterize the pericellular matrix within the chordamesoderm of Xenopus gastrulae. They identify several types of adhesive contacts within the chordamesoderm and assess how they are altered in the absence of key PCM molecules via morpholino knock-down. They conclude that syndecan-4 and hyaluronic acid comprise and promote assembly of PCM plaques whereas fibronectin and C-cadherin anchor them to cell surfaces. Cell packing density is decreased upon loss of all 4 of these molecules, which the authors attribute to a decrease in the number of cell contacts without affecting the strength of the remaining contacts. They further conclude that adhesiveness increases linearly with contact width, and that this relationship is unaffected by loss of any aforementioned adhesive/ PCM molecules.

    Major comments:

    • Many conclusions in this manuscript are based on measurements of cell contact angles, which indicate the reduction of tension at cell contacts vs. free cell surfaces and thus relative adhesive strength. While this lab previously applied the same approach to live tissues (David et al, 2014), it is not clear to what extent such measurements accurately reflect adhesive strength in fixed tissues and/or electron micrographs. Especially given the issue of random sectioning planes, which cause distortion of contact angles. Although a correction was applied, the authors note this is not theoretically derived because the heterogeneity of gap sizes made such calculations too difficult. Indeed, it appears that the large gaps between cells within morphant embryos affect contact angle measurements, but if this is corrected for in any way, it is not mentioned. Because this is the sole measure of cell adhesion provided in the study, this reviewer is not convinced of the conclusion that loss of PCM components does not affect adhesive strength. Could such measurements not be made from live cells/tissues after manipulating PCM components, as the lab has done previously? Because the lab already has the necessary reagents and expertise for such experiments, the time and resources needed for such measurements shouldn't be prohibitive.
    • As mentioned above, these authors previously measured adhesive strength in live Xenopus cells and tissues (David et al, 2014). In that study, they found that C-cadherin MO reduced relative adhesiveness whereas the current study found that relative adhesiveness actually increases in this condition. What explains this discrepancy?
    • No control morpholinos are used, and for the morpholinos that are used, the doses are very large. An equally high dose of control MO should be used to ensure that all observed phenotypes are specific.
    • It appears that all the images analyzed were collected in the sagittal plane, and the analyses don't seem to consider the intrinsic polarity of the chordamesoderm. For example: cells in different positions within the tissue (basal vs. apical), or that WT chordamesoderm cells are mediolaterally polarized and actively intercalating whereas disruption of PCM components like fibronectin disrupts cell intercalation and randomizes cell polarity. It is possible that 1) cell-matrix (in basal cells) and 2) cell-cell (during intercalation) interactions may affect the measurements made in this study. In other words, that cell contacts could differ by position within the embryo and intercalation/polarity status... have such effects been accounted for in the current analysis?
    • In this study, the authors state that chordamesoderm movements are preserved in syndecan-4 morphants, and in their 2021 article (Barua et al) they state that convergent extension movements are accelerated. But another study describing this MO found that it causes severe convergent extension defects (Munoz et al, NCB, 2006). What explains this discrepancy? Also, the syn-4 morphant showed in Fig. 1 appears more developmentally advanced than the other embryo... if the embryos are not stage matched it could affect the measurements and conclusions drawn from them.
    • In figure 7, the authors plot relative adhesion (measured from contact angles) vs. contact width, then fit regression lines to the lower boundaries of these scatter plots. It is not clear why this analysis is focused only on the lower boundaries rather than considering the full spread of the data. Particularly for syn-4 morphants, whose values do not appear to be concentrated along the lower boundary. This analysis is further confused by the introduction of alpha*, which represents relative adhesiveness relative to the regression.
    • Based on these regression lines alone, the authors conclude that all 4 conditions are similar enough to pool the data for further analysis. If these contacts have different properties, which the data in Figures 1-6 suggest they do, it seems inappropriate to pool them together. Based on this pooling, the authors then conclude that relative adhesiveness increases linearly with contact width over the entire width range, regardless of adhesion factor depletion. This again assumes that all contacts (morphant and WT) are functionally equivalent, and that what is observed in morphant embryos in very wide contacts would also hold true in WT contacts. But because WT contacts occupy only a small portion of the width range, we cannot know how they would behave if scaled to be wider, and I am not convinced that very wide morphant contacts are representative of or functionally equivalent to WT. In other words, we cannot know that contact width is the only factor increasing their relative adhesion, given the experimental manipulations that structurally alter these contacts.

    Minor comments

    • In their descriptions of PCM in different experimental conditions, the authors overstate some conclusions drawn from EM data. For example, that type I glycocalyces are absent in chordamesoderm (although this signal is only reduced), or that because the Has2 morphant phenotype is intermediate between C-cad and fibronectin morphants this indicates an adhesive role for hyaluronic acid.
    • The authors state of the data in figure 1 that "All treatments significantly increase the size of non-adhesive gaps", but they don't show a quantification of the gaps size (they show the abundance).
    • The authors state that LSM contacts exist as 10-20 and 20-50 nm subtypes. It is not clear what about the data suggest this division.
    • In the same paragraph, the authors state that "C-cad and Syn-4... favor LSM width between 20-50 nm." What is meant by "favor"? Given that the number of 20 nm contacts is increased and 50 nm contacts is decreased in both conditions, this statement is unclear.
    • On page 7, the authors say that the size of LSM structures is "consistent with larger plaques being assembled from small units", but if that were the case, wouldn't the plaque sizes be multiples of the size of a single unit? I.e. 100, 200, and 300 nm peaks? Because this is not the case, the data seem more consistent with a continuous range of LSM plaque sizes than with discrete units.
    • On page 8, the authors refer repeatedly to LSM volume. Given that these measurements are made from TEM sections, how is volume being measured?
    • The authors present a model in which PCM interdigitates within cell contacts, but this is based on measurements from static tissues alone. Could the measurements of contact width instead be explained by compression of the PCM or some other mechanism? The data as presented don't rule out such possibilities.
    • Some terms used are not clear, for example: "partial LSM", "triple layer contact", "random removal [of LSM plaques]".
    • In figure 5, the graphs depict negative "abundance". Recommend "difference in abundance" instead.
    • Statistics: In figure 1I, it is not clear what the asterisk in this graph means or if statistical differences between these groups was determined. And in figure 6, some groups are marked as n.s., but P values for groups that are statistically different are not presented.

    Significance

    This detailed electron microscopy study advances our understanding of pericellular matrix within vertebrate embryos and how loss of its constituent molecules affects cell interactions. It further addresses the relationship between structurally distinct pericellular matrices and their adhesive properties, although this analysis is less convincing. This study adds to a body of literature in which cell-cell and cell-matrix adhesion are known to regulate morphogenetic cell movements, but how such contacts are remodeled as cells rearrange is poorly understood. Previous work has also used measurements from live cells, embryos, and tissues to infer physical forces within embryos such as adhesive strength, cortical tension, and viscosity. This work follows up directly on a previous study from this group that characterized glycocalyces within various tissues within Xenopus gastrulae by electron microscopy. The hypothesis that pericellular matrix enables flexible/fluid adhesion within highly dynamic embryonic tissues is exciting, and is likely to be of interest to developmental biologists - particularly those who apply mechanical concepts to embryos. However, additional evidence, preferably from live tissues and embryos, is needed to support this hypothesis. This assessment is based on over 15 years' experience studying gastrulation morphogenesis in multiple vertebrate species.

  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 #1

    Evidence, reproducibility and clarity

    Summary:

    This work examines the relationship between cell-cell contacts and pericellular matrix in Xenopus chordamesoderm, which is a tissue actively involved in convergent extension during gastrulation. By lanthanum staining of pericellular materials, the authors found that different types of pericellular matrix are present in cell-cell contacts in the chordamesoderm, which may mediate cell-cell adhesion. Knockdown of C-cadherin, Syndecan-4, fibronectin, and hyaluronic acid leads to the reduced abundance of cell contacts and cell packing density, but this does not seem to affect convergent extension. Based on these observations, the authors propose a model in which cell-cell contacts involve the interdigitation of distinct pericellular matrix units.

    Major points:

    1. Knockdown of adhesion molecules separates cells and leads to wide contacts with large interstitial spaces. Data in figure 1 show loosely packed morphant chordamesoderm cells. Intuitively, these should reduce cell-cell adhesion. However, a main conclusion from this manuscript is that reduced abundance of narrower contacts does not decrease adhesiveness. Although depletion of adhesion molecules modifies but not abolishes a contact, non-attached free surfaces increase significantly in morphant cells. It is therefore not easy to understand that how reduced cell contacts have no effect on cell adhesion. Importantly, the adhesiveness was not experimentally tested.
    2. It is surprising that reduced cell contacts, at least narrower cell contacts, do not affect convergent extension. Does this mean that active cell behavior changes in the chordamesoderm, which are required for convergent extension, are independent of cell contact types?
    3. Although the formation and localization of pericellular materials are differentially affected after knockdown of adhesion molecules, there is no clear evidence showing that different types of pericellular matrix mediate cell-cell adhesion in the chordamesoderm. It is possible that the disrupted distribution of pericellular materials in morphants only represents a secondary consequence of changed cell contacts. This may be supported by the fact that knockdown of adhesion molecules reduces narrow contacts and increases LSM-free gaps.
    4. The relationship between contact width spectra and LSM is also very elusive. Again, changes in contact width or abundance and distribution of LSM may be indirectly caused by loss of adhesion molecules. Therefore, although knockdown of adhesion molecules leads to changes of LSM localization, it cannot be concluded that cell-cell contacts in chordamesoderm are mediated different types of pericellular matrix.
    5. In contrast to the present observations, works by others using the same morpholinos have shown that Cadherin-dependent cell adhesion, fibronectin-rich extracellular matrix, and Syndecan-4-regulated non-canonical Wnt signaling are required for convergent extension. These discrepancies need to be appropriately addressed.
    6. If LSM and LSM-free contacts are similarly adhesive, what will be role of LSM in cell adhesion and how cell adhesion is established in these LSM-free contacts?

    Minor points:

    1. It may be helpful to clearly define the pericellular matrix in this particular context and its relationship with LSM. It is also necessary to clarify whether the adhesion molecules examined in this work are considered as components of the pericellular matrix.
    2. In figure 1B, it appears that the Cadherin morphant has defects in chordamesoderm elongation and archenteron formation, suggesting impaired convergent extension.
    3. In figure 1C, the Syndecan-4 morphant gastrula clearly shows enhanced anteroposterior elongation of chordamesoderm and archenteron in comparison with the wild-type embryo. This seems to suggest that loss of Syndecan-4 promotes the movements of convergent extension. However, previous studies indicate that both gain and loss of Syndecan-4 impairs convergent extension.
    4. Ideally, in knockdown experiments, control embryos should be injected with corresponding mismatch morpholinos.
    5. In figure 1E, it is unclear what type of cell contacts the light green arrowheads indicate.
    6. Figure 1 legend, "(wt) is from Barua et al. 2021". I am not sure it is appropriate to use previously published data.
    7. There is no light blue arrowhead in figure 2, and in figure 3B and 3I, it seems that the same colored arrows are used to indicate different structures.
    8. Triple-layered contacts are not clearly defined.
    9. Page 2, "based on driven by" should be either "based on" or "driven by".
    10. Page 8, "selectin" should be "selecting".

    Significance

    Strengths:

    Demonstrated the effects of several adhesion molecules on the formation of cell contacts and pericellular matrix in Xenopus chordamesoderm.

    Limitations:

    The significance of chordamesoderm cell contact changes in convergent extension or gastrulation is not clear; there is no direct evidence showing the functional link between pericellular matrix, cell contacts and cell adhesion; the absence of effects on convergent extension after depletion of several adhesion molecules is not fully consistent with previous reports.

    Advance:

    This work likely provides some fundamental and methodological advances for studying cell-cell adhesion. It shows promise for elucidating mechanisms underlying the regulation of cell contact changes in tissues involved in morphogenetic movements.

    Audience:

    This work likely interests readership studying embryonic cell adhesion in the field of developmental biology and cell biology. It may be also potentially interesting for people working on glycocalyx pericellular matrix in adult tissues.

  4. Version published to 10.1101/2023.07.05.547782v1 on bioRxiv