Spatial confinement induces oscillatory migration of epidermal keratinocytes and forms a three-compartment epithelial structure with epithelial–mesenchymal transition dynamics

  1. Takuma Nohara
  2. Sota Itamoto
  3. Junichi Kumamoto
  4. Yosuke Mai
  5. Mayuna Shimano
  6. Sora Kato
  7. Hiroyuki Kitahata
  8. Hideki Nakamura
  9. Shota Takashima
  10. Mika Watanabe
  11. Masaharu Nagayama
  12. Tsukasa Oikawa
  13. Hideyuki Ujiie
  14. Ken Natsuga

  • Curated by eLife

    eLife logo

    eLife Assessment

    The authors established a novel three-dimensional culture system for stratified epithelia that allows epithelial cells to undergo epithelial-to-mesenchymal transition (EMT) and subsequent mesenchymal-to-epithelial transition (MET) while migrating through a membrane with 3.0-µm micropores, and, thus, provides a highly valuable tool to study EMT and possibly wound regeneration or metastasis. Furthermore, a set of experiments provides compelling data suggesting that TGF beta signaling and actin polymerization promote movement of epithelial cells into the pores, while Piezo1 and Keratin 6 prevent keratinocyte migration and EMT.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Epithelial cells undergo epithelial–mesenchymal transition (EMT) during migration and regain their epithelial phenotype in the post-migration phase (mesenchymal–epithelial transition [MET]). We established an experimental system that reproduces a three-compartment epithelial structure comprising the original epithelium, its EMT state, and its MET state. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane, migrated through 3.0 µm or larger micropores. The 3.0-µm-pored membrane induced an epithelial structure with three distinct states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain the three-compartment epithelial structure. Live imaging revealed that KCs moved in an oscillatory manner, with actin-rich filopodia-like structures extending into and out of the 3.0 µm micropores, while the cells migrated unidirectionally into larger micropores. Piezo1 and keratin 6 were identified as negative modulators of KC entry into and exit from the 3.0 µm micropores. These results demonstrate that non-cancerous epithelial cells migrate through confined spaces in an oscillatory manner, which might contribute to the formation of a three-compartment epithelial structure that recapitulates key aspects of wound healing.

Article activity feed

  1. Version published to 10.7554/elife.105192.4 on eLife
  2. Version published to 10.7554/elife.105192 on eLife
  3. Version published to 10.7554/elife.105192.3 on eLife
  4. eLife

    eLife Assessment

    The authors established a novel three-dimensional culture system for stratified epithelia that allows epithelial cells to undergo epithelial-to-mesenchymal transition (EMT) and subsequent mesenchymal-to-epithelial transition (MET) while migrating through a membrane with 3.0-µm micropores, and, thus, provides a highly valuable tool to study EMT and possibly wound regeneration or metastasis. Furthermore, a set of experiments provides compelling data suggesting that TGF beta signaling and actin polymerization promote movement of epithelial cells into the pores, while Piezo1 and Keratin 6 prevent keratinocyte migration and EMT.

  5. eLife

    Reviewer #1 (Public review):

    [Editors' note: this version has been assessed by the Reviewing Editor without further input from the original reviewers. The authors have addressed the comments raised in the previous round of review.]

    Summary:

    The study describes the migration of epidermal keratinocytes through porous membranes and observed a unique size selection whereby only on 3-micron membrane are keratinocytes able to migrate and reform an intact epidermis. The authors propose that the model replicates three cell states of intact epidermis, EMT, and MET. They also show that this response depends on the actin cytoskeleton, keratin 6, and Piezo1, and the migration could be stimulated with TGFbeta ligands.

    Strengths:

    Strengths of the study include the establishment of a simple yet robust in vitro model that capture all three cell states, which could be useful for future investigation of wound healing or metastasis across a range of epithelia. There is also good quantitative characterisation of the pore size effects, providing some interesting observations about the physical regulation of keratinocyte migration, and the study provides some initial insights into the regulatory mechanisms. The images and presentation are clear.

  6. eLife

    Reviewer #2 (Public review):

    Summary:

    The revised manuscript by Nohara et al. presents a novel 3D assay that allows for stratification of epithelia, active EMT through small pores, and active MET. They show that 3um pores allow for keratinocytes to sample the pore through filopodia and up regulate EMT genes to transverse the pores to the other side of the membrane where EMT genes are down regulated as the cells re-establish stratified epithelia. The TGFbeta pathway and actin polymerization promotes movement of cells into the pores and Piezo1 and KRT6 actively block this movement. This work provides a novel 3D assay that is likely to become a benchmark to analyze these processes using a more complex system than other current culture-specific EMT and MET assays.

    Strengths:

    The strengths of the manuscript include the foundational analysis of the pathways involved in establishing the 3-compartment epithelium. The authors have incorporated live imaging, drug studies, KO analysis, and RNA-sequencing to show the relevant pathways involved. The authors were responsive to the original round of reviews and present a much-improved study with greater details on molecular markers, quantification, and clarity.

  7. eLife

    Reviewer #3 (Public review):

    Summary:

    The authors established an experimental system that reproduced three-dimensional triphasic epithelia, i.e., the original epithelium, its EMT, and MET. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane migrated through 3.0-um or larger micropores. The 3.0-um-pored membrane induced an epithelial structure with three states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain this triphasic epithelia. Live imaging revealed that KCs moved in a reciprocating manner, with actin-rich filopodia-like KC structures extending into and out of the 3.0-um micropores, while the cells migrated unidirectionally into larger micropores. KO of Piezo1 and keratin 6 increased KC entry to and exit from the 3.0-um micropores. Their results demonstrate that benign keratinocytes migrate through confined spaces in a reciprocating manner, which might help form triphasic epithelia, recapitulating wound healing processes.

    Strengths:

    Careful observation of the behaviour of keratinocytes on the different sized pores. CrispR-Cas9 gene editing to KO Piezo 1 and keratin 6 isoforms in HaCaT keratinocytes.

  8. eLife

    Author response:

    The following is the authors’ response to the previous reviews.

    Changes made independently of the reviewers' comments

    In addition to the revisions described below, we have made the following three changes, none of which affects the conclusions of the paper.

    (1) Shared control data (Figure 4c, d, l, m). The parental cells and the PIEZO1 KO and pan-KRT6 KO cell lines were assayed in parallel in the same experiments. We have therefore stated in the figure legend that the parental-cell data are shared between panels c, d and panels l, m.

    (2) Number of DEGs (Figure 4h). We have corrected the number of genes upregulated in cells cultured on the 3.0-µm-pored membrane from 69 to 68.

    (3) Yoda1 data (Figure 4e, f). Some values had been incorrectly transferred during data compilation. We have corrected these values and the corresponding graphs. The statistical results are unchanged.

    We have also provided two new files: Source data 1 (numerical data underlying all graphs) and Source data 2 (uncropped western blot images).

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    The study describes the migration of epidermal keratinocytes through porous membranes and observed a unique size selection whereby only on 3-micron membrane are keratinocytes able to migrate and reform an intact epidermis. The authors propose that the model replicates three cell states of intact epidermis, EMT, and MET. They also show that this response depends on the actin cytoskeleton, keratin 6, and Piezo1, and the migration could be stimulated with TGFbeta ligands.

    Strengths:

    Strengths of the study include the establishment of a simple yet robust in vitro model that capture all three cell states, which could be useful for future investigation of wound healing or metastasis across a range of epithelia. There is also good quantitative characterisation of the pore size effects, providing some interesting observations about the physical regulation of keratinocyte migration, and the study provides some initial insights into the regulatory mechanisms. The images and presentation are clear.

    Weaknesses:

    While there are some insights into the underlying mechanisms, there could be more detailed analysis of the key signalling pathways and functional relationships between actin, keratins, Piezo, and EMT genes. Although the extent of new biological insight is somewhat limited, the study still demonstrates the utility of the model, and these points may be important questions to address in future studies.

    We agree with the reviewer that these are important questions. We will address the functional relationships among actin, keratins, PIEZO1, and EMT genes, and the underlying signaling pathways, in future studies.

    Reviewer #2 (Public review):

    Summary:

    The revised manuscript by Nohara et al. presents a novel 3D assay that allows for stratification of epithelia, active EMT through small pores, and active MET. They show that 3um pores allow for keratinocytes to sample the pore through filopodia and up regulate EMT genes to transverse the pores to the other side of the membrane where EMT genes are down regulated as the cells re-establish stratified epithelia. The TGFbeta pathway and actin polymerization promotes movement of cells into the pores and Piezo1 and KRT6 actively block this movement. This work provides a novel 3D assay that is likely to become a benchmark to analyze these processes using a more complex system than other current culture-specific EMT and MET assays.

    Strengths:

    The strengths of the manuscript include the foundational analysis of the pathways involved in establishing the 3-compartment epithelium. The authors have incorporated live imaging, drug studies, KO analysis, and RNA-sequencing to show the relevant pathways involved. The authors were responsive to the original round of reviews and present a much-improved study with greater details on molecular markers, quantification, and clarity.

    We thank the reviewer for the positive evaluation of our revised manuscript.

    Reviewer #3 (Public review):

    Summary:

    The authors established an experimental system that reproduced three-dimensional triphasic epithelia, i.e., the original epithelium, its EMT, and MET. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane migrated through 3.0-um or larger micropores. The 3.0-um-pored membrane induced an epithelial structure with three states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain this triphasic epithelia. Live imaging revealed that KCs moved in a reciprocating manner, with actin-rich filopodia-like KC structures extending into and out of the 3.0-um micropores, while the cells migrated unidirectionally into larger micropores. KO of Piezo1 and keratin 6 increased KC entry to and exit from the 3.0-um micropores. Their results demonstrate that benign keratinocytes migrate through confined spaces in a reciprocating manner, which might help form triphasic epithelia, recapitulating wound healing processes.

    Strengths:

    Careful observation of the behaviour of keratinocytes on the different sized pores. CrispR-Cas9 gene editing to KO Piezo 1 and keratin 6 isoforms in HaCaT keratinocytes.

    Weaknesses:

    The authors have improved the observed weaknesses including better statistical analysis and more detail on replicates. They have also performed staining for laminin beta3 chain.

    We thank the reviewer for these comments.

    Recommendations for the authors:

    Reviewer #1 (Recommendations for the authors):

    The authors have significantly improved the manuscript in the recent revision. One minor point is to please make the RNA-seq analysis, including gene lists available in the supplementary information.

    As suggested, we have added Figure 4—source data 1, which lists the differentially expressed genes (DEGs) identified by bulk RNA-seq of HaCaT keratinocytes cultured on 0.4-µm- or 3.0-µm-pored membranes.

    Reviewer #3 (Recommendations for the authors):

    I am not suggesting any further changes to the manuscript.

    We thank the reviewer for evaluating our revised manuscript.

  9. Mentioned by @rridrobot.bsky.social on Bluesky

    A preprint using: PIEZO1 Antibody - BS (RRID:AB_11003149) from #Novus was published. SciScore made a table with this resource, see “Automated Services” module (download as csv, xml or #jats) #reproducibility #STMpublishing

  10. eLife

    eLife Assessment

    In this manuscript the authors established a novel three-dimensional culture system for stratified epithelia that allows epithelial cells to undergo epithelial-to-mesenchymal transition (EMT) and subsequent mesenchymal-to-epithelial transition (MET) while migrating through a membrane with 3.0-µm micropores, and, thus, provides a highly valuable tool to study EMT and possibly wound regeneration or metastasis. Furthermore, a set of experiments provides compelling data suggesting that TGF beta signaling and actin polymerization promote movement of epithelial cells into the pores, while Piezo1 and Keratin 6 prevent keratinocyte migration and EMT.

  11. eLife

    Reviewer #1 (Public review):

    Summary:

    The study describes the migration of epidermal keratinocytes through porous membranes and observed a unique size selection whereby only on 3-micron membrane are keratinocytes able to migrate and reform an intact epidermis. The authors propose that the model replicates three cell states of intact epidermis, EMT, and MET. They also show that this response depends on the actin cytoskeleton, keratin 6, and Piezo1, and the migration could be stimulated with TGFbeta ligands.

    Strengths:

    Strengths of the study include the establishment of a simple yet robust in vitro model that capture all three cell states, which could be useful for future investigation of wound healing or metastasis across a range of epithelia. There is also good quantitative characterisation of the pore size effects, providing some interesting observations about the physical regulation of keratinocyte migration, and the study provides some initial insights into the regulatory mechanisms. The images and presentation are clear.

    Weaknesses:

    While there are some insights into the underlying mechanisms, there could be more detailed analysis of the key signalling pathways and functional relationships between actin, keratins, Piezo, and EMT genes. Although the extent of new biological insight is somewhat limited, the study still demonstrates the utility of the model, and these points may be important questions to address in future studies.

  12. eLife

    Reviewer #2 (Public review):

    Summary:

    The revised manuscript by Nohara et al. presents a novel 3D assay that allows for stratification of epithelia, active EMT through small pores, and active MET. They show that 3um pores allow for keratinocytes to sample the pore through filopodia and up regulate EMT genes to transverse the pores to the other side of the membrane where EMT genes are down regulated as the cells re-establish stratified epithelia. The TGFbeta pathway and actin polymerization promotes movement of cells into the pores and Piezo1 and KRT6 actively block this movement. This work provides a novel 3D assay that is likely to become a benchmark to analyze these processes using a more complex system than other current culture-specific EMT and MET assays.

    Strengths:

    The strengths of the manuscript include the foundational analysis of the pathways involved in establishing the 3-compartment epithelium. The authors have incorporated live imaging, drug studies, KO analysis, and RNA-sequencing to show the relevant pathways involved. The authors were responsive to the original round of reviews and present a much-improved study with greater details on molecular markers, quantification, and clarity.

  13. eLife

    Reviewer #3 (Public review):

    Summary:

    The authors established an experimental system that reproduced three-dimensional triphasic epithelia, i.e., the original epithelium, its EMT, and MET. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane migrated through 3.0-um or larger micropores. The 3.0-um-pored membrane induced an epithelial structure with three states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain this triphasic epithelia. Live imaging revealed that KCs moved in a reciprocating manner, with actin-rich filopodia-like KC structures extending into and out of the 3.0-um micropores, while the cells migrated unidirectionally into larger micropores. KO of Piezo1 and keratin 6 increased KC entry to and exit from the 3.0-um micropores. Their results demonstrate that benign keratinocytes migrate through confined spaces in a reciprocating manner, which might help form triphasic epithelia, recapitulating wound healing processes.

    Strengths:

    Careful observation of the behaviour of keratinocytes on the different sized pores. CrispR-Cas9 gene editing to KO Piezo 1 and keratin 6 isoforms in HaCaT keratinocytes.

    Weaknesses:

    The authors have improved the observed weaknesses including better statistical analysis and more detail on replicates. They have also performed staining for laminin beta3 chain.

  14. eLife

    Author response:

    The following is the authors’ response to the original reviews.

    Changes made independently of the reviewers' comments

    In addition to the revisions described below, we have made the following three changes during the revision process. None of them affects the overall claims of the paper.

    (1) Removal of the blebbistatin data. The blebbistatin experiments (Figure 3f and h in the original manuscript) did not reproduce in repeated trials. We have therefore removed these data, together with the related descriptions in the Results, Discussion, and Materials and Methods.

    (2) Removal of the pan-KRT6 KO-2 data. The original manuscript presented two pan-KRT6 KO cell lines. We subsequently found that an insufficient number of biological replicates had been obtained for KO-2, although its results were consistent in direction with those of KO-1. We have removed the KO-2 data and renamed pan-KRT6 KO-1 as pan-KRT6 KO.

    (3) Revised statistics. As described in our response to Reviewer #1's comment (2), we re-analyzed the quantitative data using a generalized linear model that accounts for experimental batch. With this analysis, the effect of ruthenium red on cell exit (Figure 4b) and the effect of one TGF-β receptor inhibitor on cell exit (Figure 3e) no longer reached statistical significance. The corresponding descriptions in the Results have been amended.

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    The study describes the migration of epidermal keratinocytes through porous membranes and observes a unique size selection whereby only on 3-micron membrane are keratinocytes able to migrate and reform an intact epidermis. The authors propose that the model replicates three cell states of the intact epidermis, EMT, and MET. They also show that this response depends on the actin cytoskeleton and Piezo1, and the migration could be stimulated with TGF-beta ligands.

    Strengths:

    Strengths of the study include the establishment of a simple yet robust in vitro model that captures all three cell states, which could be useful for future investigation of wound healing or metastasis. There is also good characterisation of the pore size effects, providing some interesting observations about the physical regulation of keratinocyte migration. The images and presentation are clear.

    Weaknesses:

    (1) Some of the terminology would benefit from better definition or refinement. Triphasic suggests different physical behaviours (e.g. liquid-liquid phase separation) rather than cellular properties. Perhaps it would be better to refer to these as cell states or to describe the model more specifically as an invasion or EMT model. Likewise, the term 'reciprocating' implies two-way communication, but it is used to describe two-way migration or oscillating migration. Here, perhaps oscillatory would be clearer.

    We appreciate the reviewer’s comment. According to the reviewer’s suggestion, we have revised the title, abstract, and main text to avoid the terms “reciprocating” and “triphasic” and have instead adopted “oscillatory” and “three-compartment,” respectively.

    (2) The quantification and statistical analysis of key results could be improved. Notably, quantification of immunostaining in Figures 1 and 2 would strengthen core findings, and greater detail is needed on the sample sizes and number of experiments used for statistical analysis. These details are missing or only appear to N=1 in some places.

    We have added the quantification and statistical analysis to the revised manuscript. We realized that some of the data in the original manuscript combined technical and biological replicates. Therefore, we have revised the statistical analysis and presented the sample size in each experiment in the revised manuscript.

    (3) There is an attempt to analyse the underlying molecular mechanisms, but these studies lack depth and detail. For example, it is not clear how actin, keratins, and piezo1 communicate to regulate cell migration. Are they acting directly on EMT genes such as SNAI1 or through changes in cell mechanics and cell-cell adhesions? Likewise, is TGF-beta signalling active in the system (e.g. nuclear pSMAD during cell migration)? As a result, the new biological insight is somewhat limited and confirms much of what is known about these pathways in keratinocyte migration.

    We have elaborated these matters in the Discussion section of the revised manuscript. Unfortunately, we were not able to specifically determine if nuclear pSMAD is present in the cells within micropores, due to technical issues described below. However, as TGF-β inhibition did not affect the entry into and the exit from the micropores (Figure 3b, e), TGF-β signaling is unlikely to be essential for this process.

    Reviewer #2 (Public review):

    Summary:

    The manuscript by Nohara et al. presents a novel 3D assay that allows for stratification of epithelia, active EMT through small pores, and active MET. They show that 3 µm pores allow for keratinocytes to sample the pore through filopodia and upregulate EMT genes to transverse the pores to the other side of the membrane where EMT genes are downregulated as the cells re-establish stratified epithelia. The TGFbeta pathway and actin polymerization promote the movement of cells into the pores and Piezo1 and KRT6 actively block this movement. This work provides a novel 3D assay that is likely to become a benchmark to analyze these processes using a more complex system than other current culture-specific EMT and MET assays.

    Strengths:

    The strengths of the manuscript include the foundational analysis of the pathways involved in establishing the tri-phasic epithelium. The authors have incorporated live imaging, drug studies, KO analysis, and RNA sequencing to show the relevant pathways involved.

    Weaknesses:

    While the authors provide strong evidence that the tri-phasic epithelium represents the EMT process, the MET process is largely relegated to the absence of EMT genes. It would be interesting to know how the stratified MET epithelia submerged in the media is similar or different from the stratified epithelia at the air-liquid interface.

    We appreciate the reviewer's comment and agree that the comparison between the MET-derived epithelium beneath the membrane and the original epithelium at the air–liquid interface is an intriguing question. Our current data provide partial insight: the epithelium beneath the membrane re-expressed E-cad (Figure 1d) and showed laminin β3 deposition comparable to that of the epithelium above the membrane (Figure 1—figure supplement 1e), indicating that it re-establishes at least some features of an epithelial tissue. However, it also retained elevated Vim and N-cad relative to the epithelium on the 0.4-µm-pored membrane (Figure 1—figure supplement 1b, c), suggesting that the MET is not complete. A systematic comparison, including barrier function and terminal differentiation markers, would require a dedicated study, which we plan to undertake.

    Reviewer #3 (Public review):

    Summary:

    The authors established an experimental system that reproduced three-dimensional triphasic epithelia, i.e., the original epithelium, its EMT, and MET. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane migrated through 3.0-um or larger micropores. The 3.0-um-pored membrane induced an epithelial structure with three states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain this triphasic epithelium. Live imaging revealed that KCs moved in a reciprocating manner, with actin-rich filopodia-like KC structures extending into and out of the 3.0-um micropores, while the cells migrated unidirectionally into larger micropores. KO of Piezo1 and keratin 6 increased KC entry to and exit from the 3.0-um micropores. Their results demonstrate that benign keratinocytes migrate through confined spaces in a reciprocating manner, which might help form triphasic epithelia, recapitulating wound healing processes.

    Strengths:

    Careful observation of the behaviour of keratinocytes on the different-sized pores. CrispR-Cas9 gene editing to KO Piezo 1 and keratin 6 isoforms in HaCaT keratinocytes.

    Weaknesses:

    There is no analysis of the matrix produced by the keratinocytes on the different pore sizes as this may influence migration.

    According to the reviewer’s suggestion, we have included images of laminin β3 labeling in 3D epidermis formed on 0.4-µm- or 3.0-µm-pored membranes in the revised manuscript (Figure 1—figure supplement 1e). Laminin β3, a subunit of laminin 332, was detected in the epithelium formed on the 0.4-µm-pored membrane, as well as in all three compartments of the tissue formed on the 3.0-µm-pored membrane. We did not observe an apparent difference in laminin β3 deposition between the two pore sizes, although we note that this assessment is qualitative and that a quantitative comparison of matrix production would require a separate study.

    HaCaT cells are quite different from normal keratinocytes in terms of migration. Pilcher et al. PMID: 9182674

    We have included this limitation in the Discussion section of the revised manuscript and cited the suggested publication (Pilcher et al. 1997).

    Recommendations for the authors:

    Reviewer #1 (Recommendations for the authors):

    (1) Some of the terminology needs better definition or refinement. Triphasic suggests different physical behaviours (e.g. liquid-liquid phase separation) rather than cellular properties. Perhaps it would be better to refer to these as cell states or to describe the model more specifically as an invasion or EMT model. Likewise, the term 'reciprocating' implies two-way communication, but it is used to describe two-way migration or oscillating migration. Here, perhaps oscillatory would be clearer.

    According to the reviewer’s suggestion, we have revised the title, abstract, and main text to avoid the terms “reciprocating” and “triphasic” and have instead adopted “oscillatory” and “three-compartment,” respectively.

    (2) The qualitative images in Figure 1 need better quantitative support through image analysis of the immunostaining to confirm changes in marker expression within the different regions of the epidermis. As this is a core finding of the study, it requires additional quantification beyond the qPCR results, which also do not provide spatial information.

    In accordance with the reviewer's suggestion, we have added quantitative image analysis of E-cad, N-cad, Vim, and Slug in the epithelium on the 0.4-µm-pored membrane and in the compartments above and below the 3.0-µm-pored membrane (Figure 1—figure supplement 1a–d). Vim was significantly more abundant in both compartments of the 3.0-µm-pored condition than in the 0.4-µm-pored condition. N-cad showed the same trend, reaching statistical significance for the comparison between the 0.4-µm-pored membrane and the compartment below the 3.0-µm-pored membrane, and was also significantly more abundant below than above the 3.0-µm-pored membrane. Slug showed a similar trend that did not reach statistical significance, and E-cad did not differ significantly among the three regions. Unlike the qPCR results, these data provide spatial information on marker expression.

    (3) More detail is needed regarding the number of independent experiments and cells analysed. This appears to be missing from both the methods and figure captions.

    We apologize for the lack of detail. In the Materials and Methods, we have added a full description of the experimental design (measurement points per biological replicate, composition of each experimental batch) and of the statistical analysis, including the gamma generalized linear model with experimental batch as a categorical fixed effect, and the software used. In the figure legends, we have added the number of biological replicates, the statistical test, and the meaning of each graph element for every quantitative panel, as well as the number of replicates underlying each representative image. For Figure 2g and 2h, we have stated the number of independent experiments. We note that our quantification is based on the fraction of the imaging field occupied by GFP fluorescence rather than on individual cell counts; we therefore report the number of biological replicates and measurement points rather than the number of cells analyzed.

    (4) Figure 2i,j needs quantification to support the conclusions. Were the studies in 2g,h replicated in independent experiments? Likewise, could KCs in the micropores simply appear more positive for Vimentin because they are smaller? It would help to show quantification of a marker that does not change or goes down in the channel.

    In response to the reviewer's suggestion, we quantified the Vim/E-cad fluorescence intensity ratio, using E-cad as the internal reference marker (Figure 2—figure supplement 1d). The ratio was significantly lower within the micropores than above the membrane, indicating that Vim signals were not enriched within the micropores relative to E-cad. We therefore agree with the reviewer that the apparent increase in Vim positivity within the micropores in the original images was, at least in part, attributable to the smaller apparent size of the cells in this region. The quantification instead indicates that the partial EMT phenotype was already present in the cells above the membrane rather than being acquired upon entry into the micropores, which is consistent with the elevated Vim intensity both above and below the 3.0-µm-pored membrane in primary KCs (Figure 1—figure supplement 1c). We have revised the description of Figures 2i and 2j accordingly.

    The experiments shown in Figures 2g and 2h were performed three times independently, and both panels summarize the pooled data from all three experiments. This is now stated in the Materials and Methods and in the figure legends.

    (5) How does the oscillatory migration response depend on the length of the channel? If the channels were longer, would they not be able to exit and reform the epithelium? This would require a different experimental system but could be important for understanding the fundamental principles of the observed migratory responses.

    We agree with the reviewer that the length of the channel may influence cellular behavior. In our study, we used 3.0-µm-pored membranes from Millicell (Merck) and Transwell (Corning), both of which have a membrane thickness of approximately 10 µm. Greiner also offers 3.0-µm-pored membranes with longer channels; however, the pore orientation in these products is not uniform, making them unsuitable for our experimental setup. As these are the only commercially available options accessible to us, we are currently unable to perform experiments addressing the effect of channel length. We agree that this is an important question and warrants investigation in future studies using appropriately engineered systems.

    (6) The underlying mechanisms are not described in much detail and there is no linkage between the pathways and molecules investigated. How exactly do actin, keratins, and piezo1 communicate to regulate cell migration? Are they acting directly on EMT genes such as SNAI1 or through changes in cell mechanics and cell-cell adhesions? Likewise, is TGF-beta signalling active in the system (e.g. nuclear pSMAD during cell migration)?

    Our data support a model in which mechanosensing and cytoskeletal remodeling, rather than transcriptional EMT programs alone, primarily regulate keratinocyte migration through confined micropores. Notably, inhibition of TGF-β signaling, a major driver of canonical EMT programs, did not impair keratinocyte entry into or exit from the micropores (Figure 3). These findings suggest that while EMT-associated pathways can modulate the process, as evidenced by the enhanced migration observed upon addition of TGF-β ligand, keratinocyte migration through micropores does not depend on classical TGF-β–driven EMT signaling. We could not directly assess nuclear pSMAD localization within the micropores because autofluorescence of micropore walls hindered the detection of nuclear signals in whole-mount imaging. Moreover, sagittal sectioning of frozen or FFPE-reconstructed 3D epidermis did not yield a sufficient number of nuclei in the micropores for assessing pSMAD positivity. Therefore, we cannot exclude context-dependent activation of TGF-β signaling. However, the lack of functional inhibition upon TGF-β receptor blockade indicates that canonical TGF-β signaling is not essential for this process. Our experiments further indicate that actin polymerization is essential for both entry into and exit from 3.0-µm micropores, suggesting that dynamic protrusive activity is a prerequisite for confined migration. Piezo1 and keratin 6 appear to function as negative modulators of this process. Piezo1, a mechanosensitive ion channel, likely influences migration by regulating cytoskeletal tension and cell shape, consistent with previous reports (Holt et al., 2021). Keratin 6, a stress-induced intermediate filament, may modulate actin organization and the mechanical properties of keratinocytes, as suggested by increased stress fiber formation in Krt6a/Krt6b-null cells (Wong & Coulombe, 2003).

    We have incorporated these points into the revised Discussion accordingly.

    (7) While the culture model is interesting and a simple yet useful tool to study EMT/MET, the new insight into these processes is somewhat limited and confirms much of what we already know about Piezo1, TGF-beta, and the cytoskeleton in KC migration. A clearer demonstration of new insight would help to strengthen the utility of the model.

    We agree that Piezo1, TGF-β signaling, and cytoskeletal dynamics have previously been implicated in keratinocyte migration. Our intention was not to claim the identification of entirely new molecular regulators of migration. Rather, the conceptual advance of our study lies in demonstrating how confined spatial architecture alone can generate a spatially organized epithelium comprising spatially graded EMT and MET states within a single continuous tissue.

    Specifically, our model reveals that keratinocytes undergoing confined migration exhibit oscillatory protrusive behavior that preserves the integrity of the overlying epithelium while enabling the formation of a new stratified epithelium beneath the membrane. To our knowledge, this dynamic integration of confined migration, partial EMT, and subsequent MET within a stratified epithelial context has not been previously demonstrated. The oscillatory behavior itself represents a distinct migratory mode compared to the predominantly unidirectional migration described in two-dimensional or Transwell assays.

    In summary, the novelty of this study lies in revealing how physical confinement can organize epithelial states and migratory dynamics in a self-organizing manner. We have included these conceptual advances in the revised manuscript.

    Reviewer #3 (Recommendations for the authors):

    In the introduction the authors state that 3D epidermis is not useful for wound healing. There are lots of papers using 3D epidermis for wound healing/re-epithelialization experiments.

    We agree with the reviewer’s point. Our original intention was to emphasize the utility of our new 3D model; however, we have moderated the tone in the Introduction to avoid underestimating the applicability of existing 3D epidermis models for wound healing and re-epithelialization studies and cited the relevant papers in the revised manuscript.

    As the cultures on the different membrane pore sizes are in culture for 14 days, a substantial matrix is produced. I think it would be useful to show staining for laminin 332 at the minimum as this may regulate migration.

    Production may also be influenced by the pore size/substrate.

    According to the reviewer’s suggestion, we have included images of laminin β3 labeling in 3D epidermis formed on 0.4-µm- or 3.0-µm-pored membranes in the revised manuscript (Figure 1—figure supplement 1e). Laminin β3, a subunit of laminin 332, was detected in the epithelium formed on the 0.4-µm-pored membrane, as well as in all three compartments of the tissue formed on the 3.0-µm-pored membrane. We did not observe an apparent difference in laminin β3 deposition between the two pore sizes, although we note that this assessment is qualitative and that a quantitative comparison of matrix production would require a separate study.

    HaCat cells are a slightly flawed keratinocyte model but I can see why you used them for gene manipulation, actin expression. Their flaws should be acknowledged in the Discussion.

    We have included this limitation in the Discussion section of the revised manuscript and cited the suggested publication (Pilcher et al. 1997).

    The RNA-seq dataset should be publicly available.

    We have made this RNA-seq dataset available publicly (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE282553) and have presented this information in the Data availability section.

    In Figure 2, i and j in the legend the authors refer to 3 and 8-micron pores, but these are not labelled in the Figure.

    This was our error in mentioning 8.0-µm micropores in the Figure 2 legend. We intended to show only the 3D structure formed by the 3.0-µm-pored membrane. We have removed the reference to “8.0-µm micropores” in the revised manuscript.

  15. Version published to 10.7554/elife.105192.2 on eLife
  16. Version published to 10.1101/2024.11.12.623158v3 on bioRxiv
  17. Version published to 10.1101/2024.11.12.623158v2 on bioRxiv
  18. eLife

    eLife Assessment

    In this manuscript the authors established a novel three-dimensional culture system for stratified epithelia that allows epithelial cells to undergo epithelial-to-mesenchymal transition (EMT) and subsequent mesenchymal-to-epithelial transition (MET) while migrating through a membrane with 3.0-µm micropores, and, thus, provides a valuable tool to study EMT and possibly wound regeneration or metastasis. Furthermore, a set of experiments provides solid data suggesting that TGF beta signaling and actin polymerization promote movement of epithelial cells into the pores, while Piezo1 and Keratin 6 prevent keratinocyte migration and EMT.

  19. eLife

    Reviewer #1 (Public review):

    Summary:

    The study describes the migration of epidermal keratinocytes through porous membranes and observes a unique size selection whereby only on 3-micron membrane are keratinocytes able to migrate and reform an intact epidermis. The authors propose that the model replicates three cell states of the intact epidermis, EMT, and MET. They also show that this response depends on the actin cytoskeleton and Piezo1, and the migration could be stimulated with TGFbeta ligands.

    Strengths:

    Strengths of the study include the establishment of a simple yet robust in vitro model that captures all three cell states, which could be useful for future investigation of wound healing or metastasis. There is also good characterisation of the pore size effects, providing some interesting observations about the physical regulation of keratinocyte migration. The images and presentation are clear.

    Weaknesses:

    (1) Some of the terminology would benefit from better definition or refinement. Triphasic suggests different physical behaviours (e.g. liquid-liquid phase separation) rather than cellular properties. Perhaps it would be better to refer to these as cell states or to describe the model more specifically as an invasion or EMT model. Likewise, the term 'reciprocating' implies two-way communication, but it is used to describe two-way migration or oscillating migration. Here, perhaps oscillatory would be clearer.

    (2) The quantification and statistical analysis of key results could be improved. Notably, quantification of immunostaining in Figures 1 and 2 would strengthen core findings, and greater detail is needed on the sample sizes and number of experiments used for statistical analysis. These details are missing or only appear to N=1 in some places.

    (3) There is an attempt to analyse the underlying molecular mechanisms, but these studies lack depth and detail. For example, it is not clear how actin, keratins, and piezo1 communicate to regulate cell migration. Are they acting directly on EMT genes such as SNAI1 or through changes in cell mechanics and cell-cell adhesions? Likewise, is TGF-beta signalling active in the system (e.g. nuclear pSMAD during cell migration)? As a result, the new biological insight is somewhat limited and confirms much of what is known about these pathways in keratinocyte migration.

  20. eLife

    Reviewer #2 (Public review):

    Summary:

    The manuscript by Nohara et al. presents a novel 3D assay that allows for stratification of epithelia, active EMT through small pores, and active MET. They show that 3um pores allow for keratinocytes to sample the pore through filopodia and up-regulate EMT genes to transverse the pores to the other side of the membrane where EMT genes are downregulated as the cells re-establish stratified epithelia. The TGFbeta pathway and actin polymerization promote the movement of cells into the pores and Piezo1 and KRT6 actively block this movement. This work provides a novel 3D assay that is likely to become a benchmark to analyze these processes using a more complex system than other current culture-specific EMT and MET assays.

    Strengths:

    The strengths of the manuscript include the foundational analysis of the pathways involved in establishing the tri-phasic epithelium. The authors have incorporated live imaging, drug studies, KO analysis, and RNA sequencing to show the relevant pathways involved.

    Weaknesses:

    While the authors provide strong evidence that the tri-phasic epithelium represents the EMT process, the MET process is largely relegated to the absence of EMT genes. It would be interesting to know how the stratified MET epithelia submerged in the media is similar or different from the stratified epithelia at the air-liquid interface.

  21. eLife

    Reviewer #3 (Public review):

    Summary:

    The authors established an experimental system that reproduced three-dimensional triphasic epithelia, i.e., the original epithelium, its EMT, and MET. Keratinocytes (KCs), skin epithelial cells, placed on a microporous membrane migrated through 3.0-um or larger micropores. The 3.0-um-pored membrane induced an epithelial structure with three states: stratified KCs above the membrane, KCs showing EMT within the micropores, and a new stratified epithelium under the membrane. The membrane with larger micropores failed to maintain this triphasic epithelium. Live imaging revealed that KCs moved in a reciprocating manner, with actin-rich filopodia-like KC structures extending into and out of the 3.0-um micropores, while the cells migrated unidirectionally into larger micropores. KO of Piezo1 and keratin 6 increased KC entry to and exit from the 3.0-um micropores. Their results demonstrate that benign keratinocytes migrate through confined spaces in a reciprocating manner, which might help form triphasic epithelia, recapitulating wound healing processes.

    Strengths:

    Careful observation of the behaviour of keratinocytes on the different-sized pores. CrispR-Cas9 gene editing to KO Piezo 1 and keratin 6 isoforms in HaCaT keratinocytes.

    Weaknesses:

    There is no analysis of the matrix produced by the keratinocytes on the different pore sizes as this may influence migration.

    HaCaT cells are quite different from normal keratinocytes in terms of migration. Pilcher et al. PMID: 9182674

  22. Version published to 10.7554/elife.105192.1 on eLife
  23. Mentioned by @biorxiv-cellbio.bsky.social on Bluesky

    Spatial confinement induces reciprocating migration of epidermal keratinocytes and forms triphasic epithelia https://www.biorxiv.org/content/10.1101/2024.11.12.623158v1

  24. Mentioned by @biorxivpreprint.bsky.social on Bluesky

    Spatial confinement induces reciprocating migration of epidermal keratinocytes and forms triphasic epithelia https://www.biorxiv.org/content/10.1101/2024.11.12.623158v1

  25. Version published to 10.1101/2024.11.12.623158v1 on bioRxiv