A Dystroglycan–Laminin–Integrin axis controls cell basal geometry remodeling in the developing Drosophila retina

Article activity feed

1. 

   Review Commons

   Apr 4, 2023

   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

   Manuscript number: RC-2022-01771

   Corresponding author(s): Franck Pichaud and Rhian Walther

   1. General Statements [optional]

   We are grateful for the reviewers’ comments and suggestions. Both reviewers agree that our work addresses a poorly understood questions in biology and medicine, and that it will be of interest to the community of cell and developmental biologists.

   We note that most of the comments/suggestions, especially from Rev#2, are concerned with the text. These include suggested references to be added, a need to expand on the Method description and suggested points of discussion. We have addressed all these issues in the revised manuscript.

   Our …

   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

   Manuscript number: RC-2022-01771

   Corresponding author(s): Franck Pichaud and Rhian Walther

   1. General Statements [optional]

   We are grateful for the reviewers’ comments and suggestions. Both reviewers agree that our work addresses a poorly understood questions in biology and medicine, and that it will be of interest to the community of cell and developmental biologists.

   We note that most of the comments/suggestions, especially from Rev#2, are concerned with the text. These include suggested references to be added, a need to expand on the Method description and suggested points of discussion. We have addressed all these issues in the revised manuscript.

   Our work aims to understand which pathways control the basal geometry of epithelial cells, and how cells coordinate remodeling of their basal geometry to organize a tissue in 3D, from apical (top) to basal (bottom). This is a relatively understudied area, especially when compared to the breadth of work related to the pathways that control the apical geometry of epithelial cells.

   The apical geometry of an epithelial cell is a direct function of the number of adherens junctions the cell shares with their neighbors. Suppression or extension of adherens junctions underpins apical geometry remodeling. Basally, this same cell will be attached to the basement membrane though integrin receptors. We use the fly retina, where cells adopt stereotyped basal geometry, to investigate whether and how integrin adhesion might induce cell basal geometry remodeling in morphogenesis.

   The novel finding we report that a temporal sequence of event seems to underpin cell basal geometry remodeling in the retina, whereby i) laminin accumulates at specific location within the basement membrane, which is ii) accompanied by a concomitant accumulation of Dystroglycan (DG), and subsequently iii) integrin receptors are recruited to these sites of high Laminin-DG. This, along with our genetic experiments, suggests that a Laminin-DG-Integrin axis controls the basal geometry of retinal cells. In this axis, we envisage patterning of the basement membrane through Laminin-DG directs integrin recruitment, which in turn induces cell basal geometry remodeling. To our knowledge, this pathway in epithelial morphogenesis, spanning from ECM regulation to integrin polarization, has not been reported before. As the function of these components in basal adhesion is conserved across phyla, we anticipate our findings will be broadly relevant for our understanding of epithelial morphogenesis.

   2. Description of the planned revisions

   The main suggestion, common to both our reviewers, is that we should provide further re-assurance that the RNAi strains we use to target basement membrane components and the DG and integrin pathways are specific, and that these strains do not come with off-target effects.

   We will follow this recommendation by i) including referencing when a line that we have used has been validated elsewhere, ii) by using at least two independent RNAi strains to target a gene of interest, iii) by making use of the deGrad-FP system (Caussinus et al., 2013) to target proteins instead of genes, iv) by making use of available mutant strains. This is all relatively straightforward, and I will detail the proposed experiments as part of the following point-by-point rebuttal and revision plan.

   REVIEWER #1

   Commenting on the need to provide further controls related to some of our RNAi experiments

   1)* All the genetics experiments are based on RNAi induced knock-down approach. Although such an approach is easy to justify for genes associated with lethality when mutated, it becomes less relevant for non-lethal ones as Dystroglycan complex components (Dg, Dys, Sgc) for which null and viable mutants are published and available. The phenotype of such mutants should be provided. *

   AND

   *There is no data explaining how these RNAi lines were validated. The fact that it gives the phenotype expected by the authors is obviously not sufficient. This point is essential to exclude off-target effects and to be able to compare the different genotypes (see #2). For instance, the strong effect of sarcoglycan could be questioned. Is it really specific? If yes, is the difference with other Dystroglycan complex members only due to RNAi efficiency or does it have a specific function? *

   AND

   Line 255, "These perturbations led to a failure of bPS/Mys to accumulate at the grommet". Dg mutants are viable (PMID: 18093579); do they show consistent phenotypes?

   __RE: __Our main methodology has been to use available RNAi strains to perturb composition of the basement membrane and to inhibit the expression of components of the DG and Integrin pathways. As pointed out by the reviewer, this approach allows us to assess the function of genes that might be embryonic lethal and allows us to specifically target the basal geometry remodeling step without perturbing earlier steps of retinal morphogenesis. This is important for the basement membrane and integrins, which are required although retinal tissue development. See for example: (Fernandes et al., 2014, Thuveson, 2019 #3787).

   We are aware that mutant alleles are available for dg, dys and sgc allow for recovering adult homozygous (or trans-heterozygous) animals. However, based on our previous experience using mutants for which only very few flies make it to adulthood, we feel it is best not to examine those animals. Compensatory pathways might be at play that could mask a phenotype (Please see our recent work on the viable roughest null allele in cell intercalation (Blackie et al., 2021).

   Therefore, we propose to induce mutant clones for dg, dys and sgc using the Flp/FRT system, using the strongest alleles that are available to us. Of note, in our experience stable proteins might not show a phenotype in small clones, but will develop a phenotype in larger ones, as the protein becomes further diluted upon multiple rounds of cell division. Bearing this in mind, we will generate animals where the whole retina is mutant for these genes. This will be done using the GMR-hid system (Stowers and Schwarz, 1999).

   Specifically, we will target Dg, Dys and Sgc using:

   Dystroglycan:

   • The dg nonsense mutations, leading to expression of truncated proteins: DgO86 (stop codon at the R87 residue) and *dgO43 *(stop codon at the W462 residue) (Christoforou et al., 2008). While previous studies have suggested that these alleles are homozygous viable (Christoforou et al., 2008; Zhan et al., 2010), we have obtained this strain from the Bloomington Stock Centre, and note that no homozygous flies make it to adult. In preliminary work, we also note that clones mutant for the *dgO86 *allele generated with the flp-FRT system are very small, comprised of only one or two cells. This suggests that DG is required for cell proliferation or viability. These dg alleles are available on the G13 FRT which is not compatible with any FRT system designed to eliminate the wild type cells. To use the GMR-hid system, we will have to first recombine these dg alleles onto the appropriate FRT chromosome. Dystrophin:

   • The dys3397 allele, which is semi-lethal P-element insertion in the dys Very few adult flies homozygous for this allele flies are recovered (Christoforou et al., 2008). We will have to recombine this allele onto an FRT chromosome to generate whole mutant retinas.

   • The deficiency Df(3R)Exel6184, which removes the dys coding frame (Christoforou et al., 2008).

   • We will also use dysE17, because it has been used before (Catalani et al., 2021; Cerqueira Campos et al., 2020; Mirouse et al., 2009). This lesion is a Q2807 Stop codon in the C-terminal region common to all 6 dys The Df(3R)Exel6184 and dysE17 alleles have been recombined onto FRT82B, which will allow us to make use of the GMR-hid system to generate whole mutant retinas. Sarcoglycan:

   • Sgc (three subunits in Drosophila) using the deletion allele *dscg169 (Allikian et al., 2007). *We will have to recombine this mutation onto an FRT chromosome to generate whole mutant retinas. In addition, we will reproduce our RNAi phenotypes using additional available RNAi lines from stock centers and from previous studies, targeting different regions of dg, dys and scg. For dys we will use a validated RNAi line. For dg we will use a second RNAi line previously used in (Cerqueira Campos et al., 2020; Villedieu et al., 2023) For dys, we will use a second line previously used in (Cerqueira Campos et al., 2020). For Sarcoglycans, we will complement our work targeting scgd by also targeting scga.

   Moreover, since a functional endogenously GFP-tagged Dg strain is now available (Villedieu et al., 2023) along with the Dys::GFP strain we have already used, we will target these proteins using the DeGrad-FP system (Caussinus et al., 2013). The main advantage with this system is that, as with RNAi, we can target a specific time window without affecting earlier steps in retinal morphogenesis. In addition, these experiments will address the possibility that DG and Dys might be stable in cells – inhibiting genes expression in flp-FRT induced clones does not always correlate with inhibiting protein function. We think that the well-established deGrad-GFP will be useful here to address the reviewer’s comment.

   We trust these complementary approaches will more than address the reviewers’ comment by further ascertaining that the RNAi phenotypes we report here for Laminin, and the DG and integrin pathway, are specific.

   Please note that we show in Fig.3 that the basal geometry phenotype we report for the talin RNAi, using an RNAi line reported in several previous studies (Lemke et al., 2019; Perkins et al., 2010; Xie and Auld, 2011; Xie et al., 2014), is comparable the phenotype we observed using the Flp-FRT system to induce *mys1 *mutant clones. So, we are confident this RNAi line is specific of talin. Nevertheless, we will also show results using second RNAi line targeting *talin. *

   *- Authors claimed that laminin RNAi (or MMPs overexpression) affects cell geometry but why it is not analyzed by PCA? It is not consistent with the other figures. *

   __RE: __To address this comment, we will provide the PCA analysis for the Laminin and MMP phenotypes.

   __REVIEWER #2 __

   • Line 208, "we found that LanB2 RNAi leads to defects in bPS/Mys Integrin localization". Here, because the authors use only single RNAi, there remains the possibility that the observed phenotype was caused by an off-target effect. The authors should exclude this possibility by using another RNAi or mutants. In case of LanB2, however, showing that one RNAi against LanB1 shows the same phenotype would be enough, because LanB1 is another single subunit of fly Laminin __RE: __We have now included loss-of-function mutant clones for LanB1, using the LanB1KG003456 allele, showing defects in integrin localization resembling the LanB2 RNAi (*please refer to section 3: revision already done, Section). *We trust that this is good validation of the LanB2 RNAi strain. These new results have been added to Figure 6 (6E-6F).

   RE: “*This is the same for all the RNAi experiments”. *Please refer to our response to Reviewer 1, above.

   2) *As the authors write "Laminin-rich domains", I suppose that they assume that LanA/B1 accumulates in a restricted region of the BM. However, it has been reported that the majority of Laminin in the fly embryo is soluble and floating in the haemolymph (fly's 'blood' or body fluid) (PMID: 29129537). Therefore, the LanA/B1 observed in the figures might be just floating in the intercellular space and doing nothing on the BM. The authors should exclude this possibility to support their idea that Laminin localised in a specific region of the BM recruits Integrin. For example, does secreted GFP (PMID: 12062063) not behave in the same way as LanA/B1? Can the authors show that the LanA/B1 is indeed incorporated in the BM by FRAP or any methods? *

   RE: While formally possible, our data suggest that it is unlikely that “LanA/B1 is just floating in the intercellular space and doing nothing on the BM”. For instance, our results show that the DG pathway component Scgd is required for accumulation of LanA::GFP (Fig.7E-F). The most likely explanation for this requirement is DG binding to Laminin fibers.

   Nevertheless, we will follow up on the reviewer’s comment and perform FRAP on LanA::GFP, as this is relatively straightforward. We will also try the GFP secretion experiment using the suggested GFPsecr transgene generated by the Vincent lab in 2000.

   3) Line 240. "RNAi against dSarcoglycan led to a decrease in LanA::GFP expression at the presumptive grommet at 20h APF (Figure 7F)". As to this result, the authors seem to interpret that Laminin is not recruited to the "specific BM domain" in grommet in the absence of Dg signalling. However, other possibilities exist, e.g., that the global expression level of Laminin was reduced, or that the intercellular space into which soluble Laminin (see the issue 4 above) flows was narrowed down. The authors should show the data that exclude (or at least reduce) these possibilities.

   __RE: __Addressing Rev2 point (1) will rule out that Laminin is in soluble form. To address the comment that the global expression level of Laminin might be decreased, we will quantify the amount of LanA::GFP that is not at the grommet and compare wild type animals with the scgd ones.

   3. Description of the revisions that have already been incorporated in the transferred manuscript

   __REVIEWER #1 __

   • Line 208, "we found that LanB2 RNAi leads to defects in bPS/Mys Integrin localization". Here, because the authors use only single RNAi, there remains the possibility that the observed phenotype was caused by an off-target effect. The authors should exclude this possibility by using another RNAi or mutants. In case of LanB2, however, showing that one RNAi against LanB1 shows the same phenotype would be enough, because LanB1 is another single subunit of fly Laminin __RE: __We have included new results – LanB2 loss of function – showing the role of Laminin in being required for Integrin localization in the secondary and tertiary pigment cells (revised Figure 6 – panels E-F)

   Line 237: For this, we used both RNAi against LanB2 and a loss-of-function allele of LanB1. Consistent with our model, we found that in both cases bPS/Mys Integrin localization was affected. bPS/Mys failed to accumulate at the grommet, and instead was distributed at the basal plasma membrane into punctate domains (Figure 6A-F). In addition, these perturbation experiments affected cell basal geometry remodeling (Figure 6A, 6C, 6E).

   2)* Methods section describing genetic conditions is really sketchy. The genotype corresponding to each figure is not provided and I guess that GMR-Gal4 has been used in all experiments using the Gal4 system but it is never clearly stated. *

   __RE: __We have revisited the Methods section and Figure Legends to ensure all appropriate information is readily accessible to the reader. The reviewer is correct that the retinal GMR-Gal4 driver was used to express the RNAi used in this study.

   3)* PCA analysis.**

   • In the WT situation it would be really informative to know which variable(s) is/are really discriminant between the two cell populations and then maybe to focus a bit more on these parameters. For instance, a PCA correlation circle plotting both cells and variables would be very helpful.*

   __RE: __We have followed the reviewer’s advice and amended the Methods section accordingly. We now provide the PCA correlation circle plotting both cells and variables in Suppl. Fig. 3, for talin RNAi and MysDN, and Suppl. Fig. 10 for DG and Scgd RNAi

   *Methods: *

   Line 522 : Principle component analysis

   Principal component analysis (PCA) was carried out using the Scikit-learn library in Python. The Standard scaler package was used to standardize the data across all metrics before calculating the principal components. The PCA package was then used to perform the PCA. Metrics included in the PCA were as follows: extent, major axis length, minor axis length, eccentricity, roundness, circularity, area, cell shape index, perimeter.

   The cell types (secondary and tertiary pigment cells) were assigned by following the cells in 3D to the apical surface where the cell types could be identified. Cells that could not be clearly assigned as either secondary or tertiary pigment cells were excluded from the PCA.

   Extent is the area of an object divided by the area or the smallest rectangle (bounding box) that can fit around the object.

   Major axis length is the longest line that can be drawn through an object.

   Minor axis length is the line that can be drawn through an object which is perpendicular to the major axis.

   __Eccentricity __is the ratio of the length of the short (minor) axis to the length of the long

   (major) axis.

   Roundness is a comparison of an object to the best fit circle of an object. The closer the object is to a perfect circle, the more round it will be.

   Circularity is a measure of the smoothness of an object.

   Cell shape index is a dimensionless parameter to describe cell shape. When cells have smaller contacts with their neighbours the cell shape index is small.

   Correlation circle plots were generated using the mlxtend plotting package in python using the plot PCA correlation graph function.

   • Please also see the graphs we now provide in Suppl. Fig.4*. *

   We are also commenting on these results.

   *Line 174: *To understand which parameters explained most of the variance in the PCA analysis we generated correlation circle plots (Supplementary Figure 4). For wildtype cells, perimeter and circularity contribute most to the variance between secondary and tertiary pigment cells along the PC1 axis. Eccentricity and minor axis length contribute most to variance along the PC2 axis (Supplementary Figure 4A). For *talin *RNAi and MysDN cells, the correlation circle plots are remarkably similar (Supplementary Figure 4B-C), indicating that these genetic perturbations have similar effects on cell basal geometry. To confirm this result, we performed PCA comparing secondary and tertiary pigment cells for these two genotypes. In both genotypes, cells fail to form discrete clusters (Supplementary Figure 4D-E). For the secondary pigment cells, expressing *talin *RNAi or MysDN leads to an increase in cell roundness. For the tertiary pigment cell, these genotypes lead to an increase in circularity (Supplementary Figure 4D-E). Examining the original segmentation data confirmed that, relative to wildtype cells, either genetic perturbation has a similar effect on key cell shape parameters (Supplementary Figure 4F-G).

   *- In loss of function conditions, when the tissue is strongly affected, how do the authors recognize the two cell populations if PCA cannot? *

   __RE: __In these genotypes, each cell type is identified based on their apical position and geometry. When a cell cannot be identified it is not included in the analysis. This allowed us to track the cells from apical to basal. We now make this clear in the Methods section.

   Line 529: The cell types (secondary and tertiary pigment cells) were assigned by following the cells in 3D to the apical surface where the cell types could be identified. Cells that could not be clearly assigned as either secondary or tertiary pigment cells were excluded from the PCA.

   - On the opposite, based on the provided image, Dys RNAi seems to have a mild effect and it seems that my eyes can easily recognize those two cell populations based on their shape. So why PCA cannot?

   __RE: __We respectfully disagree with this comment. In the Dys RNAi, one cannot tell which is a secondary and which is a tertiary by visual inspection of the basal surface only. This is consistent with the PCA analysis, now described more thoroughly in Supplemental Figure 4. The Dys RNAi cells tend to remain elongated and they do not round up as much as the Scgd RNAi cells, which gives the false impression that the phenotype is closer to that of the wild type.

   - Based on the proposed images, some phenotypes look clearly different depending on the genotype, e.g. Talin and Mys (figure 3) or Dys and Sgc (Figure 8). In other words, the fact that PCA cannot separate the cell pollutions in these different genotypes does not necessarily mean that their effect is identical. Could authors perform PCA analysis between mutants? If they are different, again it might be very interesting to identify the discriminating parameters.

   RE: We did not claim the defect was identical__. __

   The basal geometries look somewhat different depending on the genotype, and we envisage this is due to differences in RNAi strength and perhaps differences in protein stability. This is the case for Dys and Scgd, as outlined in the preceding point. With respect to talin and mys, none of the authors can distinguish by eye the talin RNAi from mys1 phenotypes. We have informally asked our institutional colleagues, and they were also unable to distinguish these genotypes.

   Nevertheless, we have expanded our PCA analysis between phenotypes, considering one cell type at a time. This analysis shows that these phenotypes show partial overlap, outside of the wildtype range. While there are similarities, it does not reveal, however, any specific relationship between genes of interest (see previous).

   Line 178: For *talin *RNAi and MysDN cells, the correlation circle plots are remarkably similar (Supplementary Figure 4B-C), indicating that these genetic perturbations have similar effects on cell basal geometry. To confirm this result, we performed PCA comparing secondary and tertiary pigment cells for these two genotypes. In both genotypes, cells fail to form discrete clusters (Supplementary Figure 4D-E). For the secondary pigment cells, expressing *talin *RNAi or MysDN leads to an increase in cell roundness. For the tertiary pigment cell, these genotypes lead to an increase in circularity (Supplementary Figure 4D-E). Examining the original segmentation data confirmed that, relative to wildtype cells, either genetic perturbation has a similar effect on key cell shape parameters (Supplementary Figure 4F-G).

   *- From what I can understand, each PCA analysis has been done on a single retina. If true, more replicates should be included. If not true, the number of independent retinas should be mentioned. *

   __RE: __All PCA analyses have been done using multiple retinas from different animals. We have clarified this in the figure legends.

   4) Minor comments:

   • Globally, the article suffers from a lack of details, especially in the methods section and/or in figure legends.

   RE: please see what we have done to address this comment, in section (2) above.

   *- Also, several points could be advantageously discussed. For instance, why MMPs have different effects according to their specificity? Also, what could be the meaning of the nice differential pattern between integrin alpha subunits? *

   __RE: __We were concerned this would be seen as too speculative by our reviewers. Following the reviewer’s advice, we are happy to share our current working model and speculations on this.

   Results:

   Line 242: Moreover, and consistent with basement membrane regulation being important for cell basal geometry remodeling, we found that degrading the basement membrane by expressing Matrix Metalloproteases MMP1 or MMP2 in retinal cells leads to a failure in bPS/Mys localization at the grommet and prevented cell basal geometry remodeling (Figure 6G-J). While recombinant Drosophila MMP1 and 2 can degrade Col-IV, only MMP2 can degrade Laminin (Wen et al., 2020). The MMP2 phenotype we observed in basal surface organization is stronger than that of the MMP1 overexpression. Our results, therefore, suggest that both Col-IV and Laminin play a role in controlling the basal geometry of retinal cells. This suggestion is consistent with our finding that both these basement membrane proteins are enriched at the grommet once cells have acquired their basal geometry.

   Discussion:

   Line 386: Integrins can bind to Col-IV and to Laminin (Hynes, 2002). Our experiments show that MMP2 overexpression leads to a stronger phenotype than MMP1. In addition to catalyzing Collagen-IV proteolysis, MMP2 can degrade Laminin, which is something MMP1 does not seem to be able to do (Wen et al., 2020). Therefore, our results suggest that both Col-IV and Laminin are required for cell basal geometry remodeling.

   Line 408*: *

   The cone cells express two Integrin receptors, ____a____PS1/Mew-____b____PS/Mys and ____a____PS2/if-____b____PS/Mys

   We found that while the interommatidial cells express aPS1/Mew-bPS/Mys, the cone cells express both aPS1/Mew-bPS/Mys and aPS2/if-bPS/Mys. Thus, different cell types express different aPS subunits. It is not clear why the cone cells express two a-subunits. In the developing follicular epithelium of the fly oocyte, cells switch from expressing aPS1/Mew-bPS/Mys, to expressing aPS2/if-bPS/Mys (Delon and Brown, 2009). In this tissue, the developmental switch between aPS1 and aPS2 expression was shown to correlate with a change in stress fiber orientation. In addition, aPS1-bPS/Mys was also shown to be required to control F-actin levels basally. aPS1 mutant cells presented elevated levels of F-actin, a phenotype not seen in aPS2 mutant cells. Remarkably, in this tissue, aPS2-bPS/Mys, but not aPS1/Mew-bPS/Mys was able to recruit the integrin adapter Tensin. The authors envisaged that the aPS2 Tensin interaction might confer robustness in basal surface remodeling. With analogy to the follicular epithelium, we speculate that in the cone cells, aPS1-bPS/Mys and aPS2/Mew-bPS/Mys synergize in mediating robust attachment to the basement membrane, to ensure these cells do not detach as the retina lengthens along the apical-basal axis (Longley and Ready, 1995). We also note that in retinal development, the cone cells form new adherens and septate junctions at their basal feet (Banerjee et al., 2008). These cells, therefore, present two sets of adherens and Septate junctions. It is also possible that the atypical situation seen with the cone cells expressing two a subunits, is linked to the formation of these new junctions at the basal pole of these cells. It will be interesting to examine these possibilities, and to establish the role these two a-subunits play in cone cell morphogenesis. Further, the presence of two distinct integrin subunits within the cone cells may have implications when considering Integrin signaling during cone cell morphogenesis.

   *- In Methods, a list of metrics is given for the PCA analysis but some look very similar and it would be helpful to define them briefly. *

   RE: Please refer to what we have done to address this comment in section (2) above.

   *- Figures are not always color-blind adjusted (e.g. dots on PCA graphs). *

   __RE: __We have rectified this oversight.

   __REVIEWER #2 __

   1)* Line 169, "From these experiments, we conclude that Integrin adhesion is required for cell basal geometry remodeling during retinal morphogenesis". It has been long known that integrin is necessary for the gross morphogenesis of the eye (e.g., Zusman et al. 1993, PMID: 8076515). The authors need to cite these preceding researches and should clarify what new findings this new work adds to the previous knowledge. *

   __RE: __Following the reviewer’s suggestion, we have added this reference which precedes (Longley and Ready, 1995)mentioned in the paper. Both references show that integrins are required for eye integrity and attribute this function to the contraction phase of retinal development. Notably, contraction occurs after cells have remodelled their basal geometry, which we have focused on in this study.

   Line 128: The Integrin* b*PS subunit (Myspheroid, Mys) is required to maintain surface integrity late in retinal development, as the tissue surface undergoes basal contraction (Longley and Ready, 1995; Zusman et al., 1993).

   4) Line 180, "Using available functional GFP protein traps [49, 50]", the authors investigate the behaviour of Laminin subunits LanA and LanB1. First, ref [50] is not relevant here and should be removed. Moreover, the Laminin-GFPs the authors used are not protein traps, but transgenic strains harbouring genes and most of their regulatory information, with the ORFs tagged with GFP [49]. Furthermore, while the ref [49] reported the functionality of LanB1-GFP, this reference did not fully address the functionality of LanA-GFP. The authors need another reference on it (PMID: 29129537), which demonstrated that LanA-GFP rescues LanA mutants.

   5)* Related to the issue above, in addition to LanA and LanB1, the authors examine the localisation of the following BM proteins using GFP-fusion: Perlecan/Trol, Collagen IV/Viking, Nidogen, and SPARC. The authors do not explicitly describe the nature of these GFP fusions, but I am afraid that the authors think all of them are "functional protein traps". However, in fact while Perlecan and Collagen IV are protein traps, Nidogen and SPARC are transgenics including regulatory sequences made in the ref [49]. This must be clarified. Moreover, to rely on the data obtained using these GFP* fusions, their functionality must be confirmed by appropriate references or/and the authors' own data. For information, ref [62] showed the functionality of Perlecan-GFP and Collagen IV-GFP protein traps (they are both homozygous viable), and the Nidogen-GFP transgene rescues the BM deficiency of Ndg mutants (PMID: 30260959). These reports must be explained in the text, and I would like the authors collect and show more information.

   __RE: __We have deleted ref 50. We thank the reviewer for flagging the issue with our referencing. We have now amended this section.

   Line 204: To this end, we examined the localization and requirement of the Laminin A and B1 subunits (Laminina, LanA and Lamininb, LanB1), Perlecan/Trol, Collagen-IV/Viking (Col-IV), the glycoprotein Nidogen (Entactin/Ndg), and the secreted glycoprotein protein-acidic-cysteine-rich (Sparc), which are all components of the basement membrane (Walma and Yamada, 2020). For Laminin, Ndg and SPARC*, *we used strains generated from a fosmid library, and expressing a functional GFP-tagged transgene under the control of their own respective promoter (Dai et al., 2018; Matsubayashi et al., 2017; Sarov et al., 2016). For Col-IV and Perlecan, we used functional GFP exon-trap strains (Morin et al., 2001).

   6) Line 200, "These specific patterns of expression for LamininA/B1, Collagen IV, Perlecan, Nidogen and Sparc". I have several comments here: - 5A. These patterns are discussed only using single optical sections. To highlight the difference in their localisation patterns more objectively, multiple sections and/or 3D images should be shown.

   RE: (a) These are all projections of 3 to 5 confocal sections, and we have amended the manuscript to make this point clearer. (b) Following the reviewer’s advice, we now provide sagittal sections so the reader can better appreciate what is detected above and below the grommet. Please see new Fig. 5.

   *5B. Can the authors discuss, hypothesise, or speculate the biological meaning of the difference? * AND

   *5C. It has been reported that in the mammalian skin BM, different components show distinct localisation patterns (PMID: 33972551). It would be interesting to cite this paper and discuss the generality of the non-uniform distribution of BM components. *

   __RE: __The revised manuscript offers a short discussion in this topic.

   Line: 367 The idea that different cell types in a tissue can express different ECM components, and thus induce localized specialization of a basement membrane is well-supported by recent work in the mouse hair follicle. In this sensory organ, the architecture and composition of the basement membrane is highly specialized depending on the cell-cell and cell-tissue interface considered (Cheng et al., 2018; Fujiwara et al., 2011; Joost et al., 2016). Moreover, different cell populations – epithelial stem cells and fibroblasts, express different ECM components in the hair follicle (Tsutsui et al., 2021), supporting the notion that specific basement membrane organization contributes to cell-cell communication and overall 3D tissue architecture.

   7) Line 215, "However, inhibiting the expression of Collagen IV, Ndg, Perlecan and Sparc individually, by expressing RNAi against these genes in all retinal cells, did not lead to defects in bPS/Mys localization". To conclude so, the authors must demonstrate that the used RNAis efficiently removed its target proteins.

   __RE: __We have removed this section referring to Collagen IV, Ndg, Perlecan and Sparc.

   Instead, we now focus solely on Laminin. Because Laminin accumulation at the presumptive grommet precedes that of the other ECM factors examined in our study, we favor a model in which Laminin plays a key role in promoting integrin localization.

   8)* Line 222, "DG is required to organize the ECM in several experimental settings [42, 43, 45, 51]". Here, the authors must mention to a preceding paper that reported the eye deficiency of Dg mutant flies (PMID: 20463973), and discuss what new findings authors can add to the previous report. *

   __RE: __We have followed this recommendation.

   Line 441: We also note that a previous study showed that early in retinal development, DG localizes at the apical membrane of the photoreceptors. This study proposed that DG promotes elongation of these sensory neurons, independently to any potential role this surface receptor might play in basement membrane organization (Zhan et al., 2010). This conclusion was based on Df(2R)Dg248 mutant clones and trans-heterozygous retinas, where DG function was impaired not only in photoreceptors, but in all interommatidial cell types. Moreover, the basement membrane was not examined in this study. Our work, and the fact the bulk of retinal cell elongation occurs late in retinal development(Longley and Ready, 1995), is consistent with DG playing a role in retinal cell elongation and overall tissue thickening.

   Under “Advance”:

   *The 3D imaging of ommatidia development is beautiful and of good descriptive value. ** However, as mentioned in the major comments 1, 2, 3, and 8 above, I am afraid that the search of preceding literature seems insufficient, and it is often unclear what this manuscript add to existing knowledge. *

   __RE: __The logic of how the reviewer links points 2, and 3 they raise as part of their review, to their assessment of how our work advances the field, is unclear to me. Their Points 2 and 3 have to do with making sure we better explain how the functional ECM transgenes were generated and by whom. The importance the reviewer places on points 2, 3 when considering the Advance our work provides to the field does not appear justified to me.

   Point 1 refers to a previous study by Zusman et al., published in 1993. Using partial loss of function alleles and heat-shock inducible rescue constructs they show that bPS/Mys plays a role in eye development. They note that in adult eyes, retinal cells are not attached to their basement membrane. They show this is accompanied by a failure for the retina to elongate along the apical-basal axis. These phenotypes are consistent with a role for integrins in mediating attachment of epithelial cells to the basement membrane, and we are now referring to this work in the revised manuscript. A much more relevant reference to our work however, is (Longley and Ready, 1995), which we have used repeatedly in our manuscript to stress what was novel about our work.

   Point 8 refers to a previous report implicating DG in photoreceptor elongation, which is a developmental phase that mostly occurs after the process we are studying here (please see Fig.3 of (Longley and Ready, 1995) for quantification using sections). The photoreceptors do no contribute basal profiles at the basal surface of the retina. The DysGFP signal we detect at this tissue surface, in the presumptive and established grommet, is clearly coming from the pigment cells, not from the photoreceptor axons which are found at this basal location. We now discuss this previous report, to make what is clearer what is novel about our own work.

   .

   Minor comments: - Line 85, "This is the case in the follicular epithelium for example". Here, the text would be more reader-friendly if the authors could clarify this is the follicular epithelium of the fly ovary.

   __RE: __We have modified the text to address this comment.

   - Line 203-, regarding all the experiments involving the Gal4-UAS system. Not all the readers are familiar with the system. A brief explanation on it should be added in the main text. Moreover, in the Results section, not in the Methods, the authors should show what Gal4 they used, and where is the Gal4 expressed.

   __RE: __We have amended the manuscript accordingly.

   *- Line 239, "We found that inhibiting the expression of the DG cofactor, dSarcoglycan [53] was most effective in inhibiting this pathway in retinal cells". Here, the authors should show the data. *

   __RE: __This statement is based on the results shown in Fig.8 and Suppl. Fig.9, which make use of a PCA representation to quantify the Dg, Dys and dScg RNAi phenotypes in cell basal geometry. We have re-phrased this statement to make it clear that we are referring to the RNAi-based perturbation of these genes’ expression.

   4. Description of analyses that authors prefer not to carry out

   We will address all the reviewer comments as they will consolidate our findings.

   Our further validation of the few RNAi lines used in our study that have not been used before in publications will also be valuable to the community.

   References:

   Allikian, M.J., Bhabha, G., Dospoy, P., Heydemann, A., Ryder, P., Earley, J.U., Wolf, M.J., Rockman, H.A., and McNally, E.M. (2007). Reduced life span with heart and muscle dysfunction in Drosophila sarcoglycan mutants. Hum Mol Genet* 16*, 2933-2943.

   Banerjee, S., Bainton, R.J., Mayer, N., Beckstead, R., and Bhat, M.A. (2008). Septate junctions are required for ommatidial integrity and blood-eye barrier function in Drosophila. Dev Biol* 317*, 585-599.

   Blackie, L., Tozluoglu, M., Trylinski, M., Walther, R.F., Schweisguth, F., Mao, Y., and Pichaud, F. (2021). A combination of Notch signaling, preferential adhesion and endocytosis induces a slow mode of cell intercalation in the Drosophila retina. Development* 148*.

   Catalani, E., Bongiorni, S., Taddei, A.R., Mezzetti, M., Silvestri, F., Coazzoli, M., Zecchini, S., Giovarelli, M., Perrotta, C., De Palma, C., et al. (2021). Defects of full-length dystrophin trigger retinal neuron damage and synapse alterations by disrupting functional autophagy. Cell Mol Life Sci* 78*, 1615-1636.

   Caussinus, E., Kanca, O., and Affolter, M. (2013). Protein knockouts in living eukaryotes using deGradFP and green fluorescent protein fusion targets. Current protocols in protein science / editorial board, John E Coligan [et al]* 73*, Unit 30 32.

   Cerqueira Campos, F., Dennis, C., Alegot, H., Fritsch, C., Isabella, A., Pouchin, P., Bardot, O., Horne-Badovinac, S., and Mirouse, V. (2020). Oriented basement membrane fibrils provide a memory for F-actin planar polarization via the Dystrophin-Dystroglycan complex during tissue elongation. Development* 147*.

   Cheng, C.C., Tsutsui, K., Taguchi, T., Sanzen, N., Nakagawa, A., Kakiguchi, K., Yonemura, S., Tanegashima, C., Keeley, S.D., Kiyonari, H., et al. (2018). Hair follicle epidermal stem cells define a niche for tactile sensation. Elife* 7*.

   Christoforou, C.P., Greer, C.E., Challoner, B.R., Charizanos, D., and Ray, R.P. (2008). The detached locus encodes Drosophila Dystrophin, which acts with other components of the Dystrophin Associated Protein Complex to influence intercellular signalling in developing wing veins. Dev Biol* 313*, 519-532.

   Dai, J., Estrada, B., Jacobs, S., Sanchez-Sanchez, B.J., Tang, J., Ma, M., Magadan-Corpas, P., Pastor-Pareja, J.C., and Martin-Bermudo, M.D. (2018). Dissection of Nidogen function in Drosophila reveals tissue-specific mechanisms of basement membrane assembly. PLoS Genet* 14*, e1007483.

   Delon, I., and Brown, N.H. (2009). The integrin adhesion complex changes its composition and function during morphogenesis of an epithelium. J Cell Sci* 122*, 4363-4374.

   Fernandes, V.M., McCormack, K., Lewellyn, L., and Verheyen, E.M. (2014). Integrins regulate apical constriction via microtubule stabilization in the Drosophila eye disc epithelium. Cell reports* 9*, 2043-2055.

   Fujiwara, H., Ferreira, M., Donati, G., Marciano, D.K., Linton, J.M., Sato, Y., Hartner, A., Sekiguchi, K., Reichardt, L.F., and Watt, F.M. (2011). The basement membrane of hair follicle stem cells is a muscle cell niche. Cell* 144*, 577-589.

   Hynes, R.O. (2002). Integrins: bidirectional, allosteric signaling machines. Cell* 110*, 673-687.

   Joost, S., Zeisel, A., Jacob, T., Sun, X., La Manno, G., Lonnerberg, P., Linnarsson, S., and Kasper, M. (2016). Single-Cell Transcriptomics Reveals that Differentiation and Spatial Signatures Shape Epidermal and Hair Follicle Heterogeneity. Cell Syst* 3*, 221-237 e229.

   Lemke, S.B., Weidemann, T., Cost, A.L., Grashoff, C., and Schnorrer, F. (2019). A small proportion of Talin molecules transmit forces at developing muscle attachments in vivo. PLoS Biol* 17*, e3000057.

   Longley, R.L., Jr., and Ready, D.F. (1995). Integrins and the development of three-dimensional structure in the Drosophila compound eye. Dev Biol* 171*, 415-433.

   Matsubayashi, Y., Louani, A., Dragu, A., Sanchez-Sanchez, B.J., Serna-Morales, E., Yolland, L., Gyoergy, A., Vizcay, G., Fleck, R.A., Heddleston, J.M., et al. (2017). A Moving Source of Matrix Components Is Essential for De Novo Basement Membrane Formation. Curr Biol* 27*, 3526-3534 e3524.

   Mirouse, V., Christoforou, C.P., Fritsch, C., St Johnston, D., and Ray, R.P. (2009). Dystroglycan and perlecan provide a basal cue required for epithelial polarity during energetic stress. Dev Cell* 16*, 83-92.

   Morin, X., Daneman, R., Zavortink, M., and Chia, W. (2001). A protein trap strategy to detect GFP-tagged proteins expressed from their endogenous loci in Drosophila. Proc Natl Acad Sci U S A* 98*, 15050-15055.

   Perkins, A.D., Ellis, S.J., Asghari, P., Shamsian, A., Moore, E.D., and Tanentzapf, G. (2010). Integrin-mediated adhesion maintains sarcomeric integrity. Dev Biol* 338*, 15-27.

   Sarov, M., Barz, C., Jambor, H., Hein, M.Y., Schmied, C., Suchold, D., Stender, B., Janosch, S., K, J.V., Krishnan, R.T., et al. (2016). A genome-wide resource for the analysis of protein localisation in Drosophila. Elife* 5*, e12068.

   Stowers, R.S., and Schwarz, T.L. (1999). A genetic method for generating Drosophila eyes composed exclusively of mitotic clones of a single genotype. Genetics* 152*, 1631-1639.

   Tsutsui, K., Machida, H., Nakagawa, A., Ahn, K., Morita, R., Sekiguchi, K., Miner, J.H., and Fujiwara, H. (2021). Mapping the molecular and structural specialization of the skin basement membrane for inter-tissue interactions. Nat Commun* 12*, 2577.

   Villedieu, A., Alpar, L., Gaugue, I., Joudat, A., Graner, F., Bosveld, F., and Bellaiche, Y. (2023). Homeotic compartment curvature and tension control spatiotemporal folding dynamics. Nat Commun* 14*, 594.

   Walma, D.A.C., and Yamada, K.M. (2020). The extracellular matrix in development. Development* 147*.

   Wen, D., Chen, Z., Zhang, Z., and Jia, Q. (2020). The expression, purification, and substrate analysis of matrix metalloproteinases in Drosophila melanogaster. Protein Expr Purif* 171*, 105629.

   Xie, X., and Auld, V.J. (2011). Integrins are necessary for the development and maintenance of the glial layers in the Drosophila peripheral nerve. Development* 138*, 3813-3822.

   Xie, X., Gilbert, M., Petley-Ragan, L., and Auld, V.J. (2014). Loss of focal adhesions in glia disrupts both glial and photoreceptor axon migration in the Drosophila visual system. Development* 141*, 3072-3083.

   Zhan, Y., Melian, N.Y., Pantoja, M., Haines, N., Ruohola-Baker, H., Bourque, C.W., Rao, Y., and Carbonetto, S. (2010). Dystroglycan and mitochondrial ribosomal protein L34 regulate differentiation in the Drosophila eye. PLoS One* 5*, e10488.

   Zusman, S., Grinblat, Y., Yee, G., Kafatos, F.C., and Hynes, R.O. (1993). Analyses of PS integrin functions during Drosophila development. Development* 118*, 737-750.

   Read the original source
2. 

   Review Commons

   Apr 4, 2023

   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:

   Cell shape remodelling is essential for tissue morphogenesis. To model this event, the fruit fly Drosophila melanogaster has been widely used. In the pupal retina, ommatidial cells change their structure to form the photo-sensing machinery in the compound eye. Previous studies investigating this event mainly focused on the cell shape change at the apical plane. However, the cell shape at the basal side and the three-dimensional (3D) structure of the cells have been little studied.

   In this manuscript, the authors address this issue by combining state-of-art 3D imaging and fly genetics. They report that at the initial stage …

   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:

   Cell shape remodelling is essential for tissue morphogenesis. To model this event, the fruit fly Drosophila melanogaster has been widely used. In the pupal retina, ommatidial cells change their structure to form the photo-sensing machinery in the compound eye. Previous studies investigating this event mainly focused on the cell shape change at the apical plane. However, the cell shape at the basal side and the three-dimensional (3D) structure of the cells have been little studied.

   In this manuscript, the authors address this issue by combining state-of-art 3D imaging and fly genetics. They report that at the initial stage of eye development, a basement membrane (BM) component Laminin accumulates at the basal side of the ommatidial cells in a manner dependent on the BM-receptor molecule dystroglycan (Dg). The authors propose that this Dg-dependent Laminin accumulation induces the polarisation of integrin at the basal surface, which is essential for proper ommatidia morphogenesis.

   Major comments:

   The beautiful images presented here provide interesting descriptions of the events occurring during eye development. Also, the authors propose an attractive and simple hypothesis that the Dg-dependent recruitment of Laminin leads to integrin polarisation and tissue morphogenesis. However, I'm afraid that this hypothesis is not supported enough by the presented data. In addition, the novelty of some conclusions and the reliability of a number of reagents used are unclear. Specific concerns are described below:

   1. Line 169, "From these experiments, we conclude that Integrin adhesion is required for cell basal geometry remodeling during retinal morphogenesis". It has been long known that integrin is necessary for the gross morphogenesis of the eye (e.g., Zusman et al. 1993, PMID: 8076515). The authors need to cite these preceding researches and should clarify what new findings this new work adds to the previous knowledge.
   2. Line 180, "Using available functional GFP protein traps [49, 50]", the authors investigate the behaviour of Laminin subunits LanA and LanB1. First, ref [50] is not relevant here and should be removed. Moreover, the Laminin-GFPs the authors used are not protein traps, but transgenic strains harbouring genes and most of their regulatory information, with the ORFs tagged with GFP [49]. Furthermore, while the ref [49] reported the functionality of LanB1-GFP, this reference did not fully address the functionality of LanA-GFP. The authors need another reference on it (PMID: 29129537), which demonstrated that LanA-GFP rescues LanA mutants.
   3. Related to the issue above, in addition to LanA and LanB1, the authors examine the localisation of the following BM proteins using GFP-fusion: Perlecan/Trol, Collagen IV/Viking, Nidogen, and SPARC. The authors do not explicitly describe the nature of these GFP fusions, but I am afraid that the authors think all of them are "functional protein traps". However, in fact while Perlecan and Collagen IV are protein traps, Nidogen and SPARC are transgenics including regulatory sequences made in the ref [49]. This must be clarified. Moreover, to rely on the data obtained using these GFP fusions, their functionality must be confirmed by appropriate references or/and the authors' own data. For information, ref [62] showed the functionality of Perlecan-GFP and Collagen IV-GFP protein traps (they are both homozygous viable), and the Nidogen-GFP transgene rescues the BM deficiency of Ndg mutants (PMID: 30260959). These reports must be explained in the text, and I would like the authors collect and show more information.
   4. Line 182-, LanA and LanB1 "accumulate at the center of the ommatidium, in a pattern resembling the grommet structure (Figure 4A and Supplementary Figure 4)"... "LamininA/B1 accumulation at the presumptive grommet precedes Integrin accumulation at this location. It suggests that localized Laminin might control Integrin localization in the interommatidial cells". Based on these results, the authors discuss that "generating specific polygonal geometries at the basal surface of cells starts with organizing the ECM to establish a pattern of Laminin-rich domains, distributed across the tissue basal surface" (Line 267).

   As the authors write "Laminin-rich domains", I suppose that they assume that LanA/B1 accumulates in a restricted region of the BM. However, it has been reported that the majority of Laminin in the fly embryo is soluble and floating in the haemolymph (fly's 'blood' or body fluid) (PMID: 29129537). Therefore, the LanA/B1 observed in the figures might be just floating in the intercellular space and doing nothing on the BM. The authors should exclude this possibility to support their idea that Laminin localised in a specific region of the BM recruits Integrin. For example, does secreted GFP (PMID: 12062063) not behave in the same way as LanA/B1? Can the authors show that the LanA/B1 is indeed incorporated in the BM by FRAP or any methods?

   1. Line 200, "These specific patterns of expression for LamininA/B1, Collagen IV, Perlecan, Nidogen and Sparc". I have several comments here: 5A. These patterns are discussed only using single optical sections. To highlight the difference in their localisation patterns more objectively, multiple sections and/or 3D images should be shown. 5B. Can the authors discuss, hypothesise, or speculate the biological meaning of the difference? 5C. It has been reported that in the mammalian skin BM, different components show distinct localisation patterns (PMID: 33972551). It would be interesting to cite this paper and discuss the generality of the non-uniform distribution of BM components.
   2. Line 208, "we found that LanB2 RNAi leads to defects in bPS/Mys Integrin localization". Here, because the authors use only single RNAi, there remains the possibility that the observed phenotype was caused by an off-target effect. The authors should exclude this possibility by using another RNAi or mutants. This is the same for all the RNAi experiments. In case of LanB2, however, showing that one RNAi against LanB1 shows the same phenotype would be enough, because LanB1 is another single subunit of fly Laminin.
   3. Line 215, "However, inhibiting the expression of Collagen IV, Ndg, Perlecan and Sparc individually, by expressing RNAi against these genes in all retinal cells, did not lead to defects in bPS/Mys localization". To conclude so, the authors must demonstrate that the used RNAis efficiently removed its target proteins.
   4. Line 222, "DG is required to organize the ECM in several experimental settings [42, 43, 45, 51]". Here, the authors must mention to a preceding paper that reported the eye deficiency of Dg mutant flies (PMID: 20463973), and discuss what new findings authors can add to the previous report.
   5. Line 240. "RNAi against dSarcoglycan led to a decrease in LanA::GFP expression at the presumptive grommet at 20h APF (Figure 7F)". As to this result, the authors seem to interpret that Laminin is not recruited to the "specific BM domain" in grommet in the absence of Dg signalling. However, other possibilities exist, e.g., that the global expression level of Laminin was reduced, or that the intercellular space into which soluble Laminin (see the issue 4 above) flows was narrowed down. The authors should show the data that exclude (or at least reduce) these possibilities.
   6. Line 255, "These perturbations led to a failure of bPS/Mys to accumulate at the grommet". Dg mutants are viable (PMID: 18093579); do they show consistent phenotypes?

   Minor comments:

   1. Line 85, "This is the case in the follicular epithelium for example". Here, the text would be more reader-friendly if the authors could clarify this is the follicular epithelium of the fly ovary.
   2. Line 203-, regarding all the experiments involving the Gal4-UAS system. Not all the readers are familiar with the system. A brief explanation on it should be added in the main text. Moreover, in the Results section, not in the Methods, the authors should show what Gal4 they used, and where is the Gal4 expressed.
   3. Line 239, "We found that inhibiting the expression of the DG cofactor, dSarcoglycan [53] was most effective in inhibiting this pathway in retinal cells". Here, the authors should show the data.

   Referee cross-commenting

   This session includes comments from both reviewers

   Reviewer 2: I almost totally agree with Reviewer 1, who is also mainly concerned about the functional analyses part of the paper while being impressed by the authors' beautiful imaging. One issue that Reviewer 1 and I apparently disagree with is the Estimated time to Complete Revisions: while they say 1-3 months, I say 3-6. However, actually I don't think this is a serious discrepancy. Thinking of the time to obtain flies and carry out their crosses necessary for the requested experiments, I'm afraid that the revision cannot be done in 1 month. However, if the authors are fortunate, they may finish the revision in 2-3 months. As I still think that the authors may struggle, I would say the time 2-6 months. I'd be glad if the comments of Reviewer 1 and me could complement with each other to help the revision of the manuscript.

   Reviewer 1:As Reviewer #2 mentioned, there is a strong convergence of our opinions on this article, which should make the work of the authors easier. In fact, I hesitated between 1-3 or 3-6 months for the estimated revision time.

   Reviewer2: Thank you Reviewer #1 for your response. I guess we (Reviewers #1 and #2) have reached an agreement now, haven't we?

   Significance

   General assessment:

   The beautiful images presented here provide interesting descriptions of the events occurring during eye development. Also, the authors' hypothesis on the Dg-dependent recruitment of Laminin leading to integrin polarisation and tissue morphogenesis is simple and attractive. However, I'm afraid that this hypothesis is not supported enough by the presented data. In addition, the novelty of some conclusions and the reliability of a number of reagents used are unclear. Therefore, I cannot say that the conclusions of this manuscript are solid.

   Advance:

   The 3D imaging of ommatidia development is beautiful and of good descriptive value. However, as mentioned in the major comments 1, 2, 3, and 8 above, I am afraid that the search of preceding literature seems insufficient, and it is often unclear what this manuscript add to existing knowledge.

   Audience:

   If the issues mentioned above have been solved, this manuscript would be of general interest to researchers in various fields in cell and developmental biology. Would not be restricted to those using Drosophila.

   Read the original source
3. 

   Review Commons

   Apr 4, 2023

   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

   Walther and colleagues address the role of the cell/ECM interface for cell shape, focusing on the basal domain of epithelial cells. More specifically, they present a thorough descriptive approach on the intricate morphology of the neuroepithelial cells composing Drosophila ommatidia in the retina and relate it to the organization of the basal lamina and the complexes interacting with it, Dystroglycan and integrins. Based on genetics and quantitative imaging approaches, they propose a linear mechanism where 1) Dystroglycan organizes the basal membrane 2) this organization guides the localization of integrins 3) integrin localization …

   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

   Walther and colleagues address the role of the cell/ECM interface for cell shape, focusing on the basal domain of epithelial cells. More specifically, they present a thorough descriptive approach on the intricate morphology of the neuroepithelial cells composing Drosophila ommatidia in the retina and relate it to the organization of the basal lamina and the complexes interacting with it, Dystroglycan and integrins. Based on genetics and quantitative imaging approaches, they propose a linear mechanism where 1) Dystroglycan organizes the basal membrane 2) this organization guides the localization of integrins 3) integrin localization defines the shape of the basal domain.

   Major comments:

   First, the 3D description of the ommatidia organization is really nice and interesting, refreshing quite old data using better imaging tools. In particular, it illustrates the extreme difference in morphology between the apical pole and the basal pole of the cells composing the ommatidia, making it a paradigm for understanding how the basal shape is defined independently of the apical one. It also provides a nice and detailed spatiotemporal cartography of basement membrane components, integrin and dystroglycan. However, functional data are less convincing, mainly for technical issues, but could be really improved in a reasonable timeline. My criticisms mainly converge on two aspects of the experimental work :

   #1) Genetics :

   • All the genetics experiments are based on RNAi induced knock-down approach. Although such an approach is easy to justify for genes associated with lethality when mutated, it becomes less relevant for non-lethal ones as Dystroglycan complex components (Dg, Dys, Sgc) for which null and viable mutants are published and available. The phenotype of such mutants should be provided.
   • There is no data explaining how these RNAi lines were validated. The fact that it gives the phenotype expected by the authors is obviously not sufficient. This point is essential to exclude off-target effects and to be able to compare the different genotypes (see #2). For instance, the strong effect of sarcoglycan could be questioned. Is it really specific? If yes, is the difference with other Dystroglycan complex members only due to RNAi efficiency or does it have a specific function?
   • Methods section describing genetic conditions is really sketchy. The genotype corresponding to each figure is not provided and I guess that GMR-Gal4 has been used in all experiments using the Gal4 system but it is never clearly stated.

   #2) Image analysis by PCA After segmentation, authors analyzed their images by PCA using various parameters, which allows them to discriminate between two cell populations that correspond to SC and TC. Then, whatever the genotype they studied, PCA failed to separate those two cell populations leading the authors to propose that they all lead to similar morphological defects, arguing for a linear pathway Dystroglycan / BM / integrins. However, this approach raises many questions:

   • In the WT situation it would be really informative to know which variable(s) is/are really discriminant between the two cell populations and then maybe to focus a bit more on these parameters. For instance, a PCA correlation circle plotting both cells and variables would be very helpful.
   • In loss of function conditions, when the tissue is strongly affected, how do the authors recognize the two cell populations if PCA cannot? On the opposite, based on the provided image, Dys RNAi seems to have a mild effect and it seems that my eyes can easily recognize those two cell populations based on their shape. So why PCA cannot?
   • Based on the proposed images, some phenotypes look clearly different depending on the genotype, e.g. Talin and Mys (figure 3) or Dys and Sgc (Figure 8). In other words, the fact that PCA cannot separate the cell pollutions in these different genotypes does not necessarily mean that their effect is identical. Could authors perform PCA analysis between mutants? If they are different, again it might be very interesting to identify the discriminating parameters.
   • Authors claimed that laminin RNAi (or MMPs overexpression) affects cell geometry but why it is not analyzed by PCA? It is not consistent with the other figures.
   • From what I can understand, each PCA analysis has been done on a single retina. If true, more replicates should be included. If not true, the number of independent retinas should be mentioned.

   Minor comments:

   • Globally, the article suffers from a lack of details, especially in the methods section and/or in figure legends.

   Also, several points could be advantageously discussed. For instance, why MMPs have different effects according to their specificity? Also, what could be the meaning of the nice differential pattern between integrin alpha subunits?

   In Methods, a list of metrics is given for the PCA analysis but some look very similar and it would be helpful to define them briefly.

   Figures are not always color-blind adjusted (e.g. dots on PCA graphs).

   Significance

   This article addresses a relevant and still poorly answered question, which is how the shape of epithelial cells is defined on the side of their basal domain. Indeed, the vast majority of studies tackle this issue for the apical domain with the role of adherens junction and their regulators. Instead, here, the authors explore the role of BM/cell interface. They especially propose a specific sequence of events leading in fine to the proper subcellular targeting of integrins. Whereas many studies on other systems have reached similar conclusions on each of the different steps of this sequence, the main interest of the paper is to bring them together, allowing the proposal of a general framework. Of notice, they made it possible by first doing a nice description of their system. However, functional analysis is somehow superficial and does not really provide mechanistic clues for each step (i.e. how Dystroglycan allows BM assembly and/or secretion, how Integrins controls cell shape.... ).

   Nonetheless, such an article might interest anyone working on tissue morphogenesis in vivo or ex vivo and wondering what the role of cell/BM interplay could be in its own system. Moreover, these protein complexes are highly conserved and involved in many diseases in humans. Thus, getting a more global understanding of their relationships is also relevant for readers working on the aetiology and pathophysiology of those diseases.

   I am a developmental biologist interested in morphogenesis.

   Read the original source
4. 

   Version published to 10.1101/2022.11.28.518187v1 on bioRxiv

   Nov 28, 2022