Single-cell transcriptomics identifies Keap1-Nrf2 regulated collective invasion in a Drosophila tumor model
Curation statements for this article:-
Curated by eLife
Evaluation Summary:
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and epithelial tumorigenesis. The analysis of single-cell transcriptomes of follicle cells now shows that transcriptionally distinct cell populations emerge shortly after induction of loss of polarity. Strengths of the work include the use of advanced single cell omics and imaging analyses to identify cell types and factors playing a role the disruption of polarity and the implications of this work for epithelial cancers. The authors' claims are generally well supported by the data and analyses. Weaknesses include the lack of high magnification images and need to clarify motivation for the study and highlight the biology rather than technical advances in the results section. Overall, this work is viewed as an important contribution to cell biologists who work on the epithelial morphogenesis or tumorigenesis.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)
This article has been Reviewed by the following groups
Listed in
- Evaluated articles (eLife)
Abstract
Apicobasal cell polarity loss is a founding event in epithelial–mesenchymal transition and epithelial tumorigenesis, yet how pathological polarity loss links to plasticity remains largely unknown. To understand the mechanisms and mediators regulating plasticity upon polarity loss, we performed single-cell RNA sequencing of Drosophila ovaries, where inducing polarity-gene l(2)gl -knockdown (Lgl-KD) causes invasive multilayering of the follicular epithelia. Analyzing the integrated Lgl-KD and wildtype transcriptomes, we discovered the cells specific to the various discernible phenotypes and characterized the underlying gene expression. A genetic requirement of Keap1-Nrf2 signaling in promoting multilayer formation of Lgl-KD cells was further identified. Ectopic expression of Keap1 increased the volume of delaminated follicle cells that showed enhanced invasive behavior with significant changes to the cytoskeleton. Overall, our findings describe the comprehensive transcriptome of cells within the follicle cell tumor model at the single-cell resolution and identify a previously unappreciated link between Keap1-Nrf2 signaling and cell plasticity at early tumorigenesis.
Article activity feed
-
-
Author Response
Reviewer #1 (Public Review):
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and tumorigenesis of epithelia. Studies have found that lack of proper cell polarity causes invasive delamination of cells and formation of multilayered epithelia, reminiscent of Epithelial-Mesenchymal Transition (EMT). Using this system, the authors analyzed the single-cell transcriptome of follicle cells and show that distinct cell populations emerge shortly after induction of polarity loss. Authors identified dynamic activation of Keap1-Nrf2 pathway Finally, subpopulation classification and analysis of regulon activity identified that Keap1-Nrf2 pathway is responsible for epithelial multilayering caused by polarity loss.
Strengths:
The authors characterized the single-cell transcriptome of …
Author Response
Reviewer #1 (Public Review):
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and tumorigenesis of epithelia. Studies have found that lack of proper cell polarity causes invasive delamination of cells and formation of multilayered epithelia, reminiscent of Epithelial-Mesenchymal Transition (EMT). Using this system, the authors analyzed the single-cell transcriptome of follicle cells and show that distinct cell populations emerge shortly after induction of polarity loss. Authors identified dynamic activation of Keap1-Nrf2 pathway Finally, subpopulation classification and analysis of regulon activity identified that Keap1-Nrf2 pathway is responsible for epithelial multilayering caused by polarity loss.
Strengths:
The authors characterized the single-cell transcriptome of follicle cell subpopulations after induction of polarity loss. Using temperature-inducible driver, they can induce the polarity loss in a short period of time, which enables detection of epithelial populations in various transition stages. Detected cell-heterogeneity could be caused intrinsically or by environmental cues within in vivo tissue. Therefore, it is likely well recapitulating tumorigenesis in vivo.
Weaknesses:
- Authors should show cells corresponding to identified key cell clusters within the tissue by immunostaining, GFP-trap, or RNA FISH.
We thank the reviewer for their comment. However, for this particular case, we would like to underscore the observation that the clusters derived from our integrated analysis do not exhibit mutually exclusive gene expression. This is unlike other studies where different clusters exhibit unique markers. The different clusters in this study represent distinguishable cell states and not distinct cell types. Even though the Lgl-KD follicle cells transcriptomically deviate from their corresponding cells of origin to form their own clusters, they continue to express several markers that show gene-expression overlap with normal follicle cells. This overlap exacerbates the problem of identifying distinct cells using differentially-enriched markers.
However, we have shown the antibody staining against Drpr to identify cluster 8 follicle cells that associate with Dcp1+ dying germline cells. We have used GstD-lacZ reporter (cluster 7 marker, specifically cluster 7_3) to show pathway activity within the multilayer. Besides GstD-lacZ, we also show F-Actin enrichment in cluster 7 (specifically 7_3) cells, that is significantly enriched in invasive cells. Additionally, we now have added images depicting the cell/stage specific expression pattern of JNK pathway components pJNK and puc, as well as that of Thor (4E-BP) which is expressed at high levels in cluster 8 and medium levels in cluster 7, and Xbp1-GFP (UPR stress sensor) that marks late stages of Lgl-KD cells.
- Images are low magnification and difficult to see individual cells.
We have replaced several such images in the revised manuscript. Specifically, the revised manuscript has entirely new (or improved versions of) image panels in figure 5. In figure 1A, the focus is the entire ovariole and therefore, we have only highlighted the enrichment of Hnt and pH3 antibody staining separately for a subsetted region of interest (ROI). The ROI panels are included within the larger image itself. For figure 6, we have converted the LUTs of panels showing distinct channels for RFP and Shg/Arm antibody stainings to grayscale.
- Manuscript is written weighted toward the technical aspect and more biology behind this study has to be discussed.
We have added new paragraphs to discuss the evidence supporting the loss of polarity, specifically that of Lgl, in human cancers. Additionally, we have also discussed how our results regarding Keap1 relates to what is already known about it and the implications of our results in context to cancer progression and metastasis.
Reviewer #2 (Public Review):
Chatterjee et al. perform extensive image and single-cell RNA sequencing (scRNA-seq) analysis of Drosophila ovaries with and without knockdown of a gene, Lethal giant larvae (Lgl), which is known to establish apical-basal polarity as well as controlling proliferation of epithelial tissues. The goal of the study is to characterize the effect of apicobasal-polarity loss in epithelial cells via Lgl knockdown on Drosophila ovaries at the phenotypic, cellular, single-cell gene expression and regulatory level. By focusing on single-cell gene expression clusters that are unique to Lgl-KD compared to those from flies without the knockdown, they were able to identify a highly transient cluster (cluster 7) which consists of tumorigenic cells. Differential markers within a sub-cluster (cluster 7_3) of this cluster followed by validation using a GstD-lac-Z enhancer-trap reporter assay lead to their conclusion that cluster 7 represents the cells of multilayering phenotype (i.e., the major Lgl-KD phenotype observed from image analysis) where activation of Keap1-Nrf2 signaling was observed. The KEAP1-NRF2 pathway is associated with protecting cells from oxidative stress. KEAP1 forms part of an E3 ubiquitin ligase, which controls NRF2, a transcription factor, by targeting it for ubiquitin-mediated proteasomal degradation. Surprisingly, inducing loss of function of both Keap1 and separately NRF2 (cnc in Drosophila) in Lgl-KD cells resulted in the same phenotype/rescue, loss of the multilayering phenotype. Over expression of Keap1 in Lgl-KD induced increased multilayer volume compared to Lgl-KD alone further supporting the role of Keap1 in cellular invasion and possibly early stages of tumorigenesis when epithelial cells start losing their polarity.
The strengths of this paper are:
The mutually reinforcing advanced imaging, scRNA-seq and genetic manipulation (knockdown and over expression) experiments/analyses that largely support the major conclusions of the manuscript which are summarized above as well as more minor observations that the authors make.
The systems biology flow of the study from broad to a specific gene/pathway implicated in the phenotype. The authors start with a clear phenotypic characterization of Lgl-KD and genome-wide scRNA-seq analysis. This leads to regulatory factor enrichment and further identification of a cluster (cluster 7) and then to a sub-cluster (cluster 7_3). This is followed by the identification of the KEAP1-NRF2 pathway and demonstration that KEAP1 knockdown and overexpression in Lgl-KD rescues and aggravates the cell multilayering phenotype, respectively.
The multilayering phenotype, genes and regulatory factors associated with loss of polarity are known to play an important role in the epithelial to mesenchymal transition (EMT). For example, this includes the enrichment of AP-1 family members, which are known to regulate EMT, in the regulon analyses as well as identification of KEAP1-NRF2.
The weaknesses of the paper are:
The framing/motivation of the study could be improved especially for those who study EMT/metastasis in humans. Given that loss of polarity is one of many events associated with tumorigenesis and metastatic progression, the claims made that studying Lgl-KD in Drosophila ovaries directly leads to insights into tumor cell invasiveness, early stages of tumorigenesis and EMT may leave some readers doubtful if they are not familiar with Lgl. Reviewing major findings that show that Lgl is a tumor suppressor as is its human homologue Hugl-1 as well as making a stronger case that studying Lgl-KD in Drosophila is relevant for tumorigenesis and EMT would be helpful.
We thank the reviewer for these suggestions. Accordingly, we have added new paragraphs to the Discussion section, where how the Lgl-KD mediated polarity loss links to mammalian tumorigenesis, as well as the implications of our results, have been discussed.
Given that Keap1 antagonizes NRF2, the apparent contradictory result that inducing loss of function of both Keap1 and separately NRF2 (cnc in Drosophila) in Lgl-KD cells resulted in the same phenotype/rescue (loss of the multilayering phenotype) is not fully addressed. Keap1 over expression revealed it aggravates multilayering. NRF2 over expression experiments were not performed. In addition, it was shown that over expression and knockdown of Keap1 did not affect NRF2 gene expression (Figure 5C); however, Keap1 regulates Nrf2 at the protein level directly via ubiquitin-mediated proteasomal degradation. Nrf2 protein levels in flies with and without Lgl-KD with various manipulations of Keap1 including control, KD and OE were not measured.
As the Keap1-Nrf2 pathway is widely studied in context of oxidative-stress response signaling, Keap1 is widely accepted as a negative regulator of Nrf2-driven transcription. However, Nrf2 has been found to positively drive the expression of Keap1 (Sykiotis and Bohmann, 2008), and that manipulating Keap1 did not change Nrf2 expression (Fig.5C). In response to this comment however, we performed additional experiments driving the ectopic expression of Nrf2 (CncC-OE) in Lgl-KD cells, which increased the invasiveness of Lgl-KD cells, similar to that by Keap1-OE. Since the UAS-CncC line has been shown to upregulate Keap1 expression (Sykiotis and Bohmann, 2008), we concluded that this increase in invasiveness is indirectly due to the increase in Keap1 expression itself.
Given that the antagonizing relationship of Keap1 and Nrf2 is only relevant to oxidative-stress response pathway, the genetic epistasis experiments in this study render that relationship irrelevant in context to the observed phenotype, as KD or OE of both components result in comparable phenotypes. Previous studies showing that Keap1 plays a role in cytoskeletal regulation (which is in agreement with our observation) also add weight to the argument that the observed phenotype is likely an indirect consequence of Keap1-Nrf2 signaling activation.
Many of the conclusions in early Results paragraphs are purely technical and not biological. For example, "These observations highlight the limitations of marker validation to identify specific cells of the differential Lgl-KD phenotype" and "SCENIC was able to detect the common as well as distinct transcriptomic states of the cells in unique Lgl-KD clusters, while also highlighting the heterogeneity among them". Some of these technical conclusions could be part of brief discussions in the Methods section.
For those not familiar with various detailed scRNA-seq analysis approaches (e.g., RNA velocity analysis), a brief description of how they should be interpreted biologically in Methods would be helpful. This might help resolve what appear to be contradictory/confusing results. First, the upper branch of cluster 7 (which is a focus of the study) shown in Fig. 3B is in a "late" stage based on Velocity Pseudotime analysis (left panel) and a "root" or an early stage based on Terminal end-points of differential analysis (right panel). The bottom branch of cluster 7 is "late"/"stable end point" based on these two analyses which is now consistent. Second, given these differences between the upper and lower branch of cluster 7, how is cluster 7 biologically the same cluster? Third, the bottom branch of cluster 7 bleeds into cluster 8 and while Ets21C is uniquely expressed in the bottom branch of 7, important markers of the study including Jra, kay (AP-1 family members), grnd, cnc (NRF2), Keap1, and the genes shown in Fig. 6F are all robustly expressed in clusters 7 (bottom branch) and 8. The biologically relevant distinction between the bottom branch of cluster 7 and 8 is not clear. Is cluster 8 important/relevant to the phenotypes observed as well?
We have now added the following paragraph elaborating the logical choices made within the analytical pipeline in our Methods section:
In this study, we have highlighted RNA velocity-derived interpretations that strictly agree with the other analytical perspectives pursued in this study. We applied scVelo to obtain information on the underlying lineage for (1) all unique Lgl-KD clusters, and (2) cluster-7 cells. The cells of the unique Lgl-KD clusters represent a mixed population of mitotic, post-mitotic, border-follicle cells and dying germline-cell associating cells that depict inconsistent transcriptional lineages. In this group of cells, the true developmental end-point of the observed Lgl-KD lineage is cluster 8 (germline-cell death occurs at the end of Lgl-KD follicular development), which likely consists of a mixed population of cells from the lateral epithelia as well as the multilayered epithelia, all responding to germline-cell death. Indeed, certain sections of cluster 7 appear more similar to cluster 8 and others seem comparable to that of cluster 13. These observations underscore our conclusions that the unique Lgl-KD clusters exhibit distinguishable gene expression, representing different cell states. For cluster 7, the state of transcriptomic heterogeneity is what defines its unique state of gene expression and we have assessed this heterogeneity by specifically sub-setting those cells.
For a comprehensive interpretation of the results of the RNA-velocity based analysis, more information can be found in the scVelo tutorial (https://scvelo.readthedocs.io/).
-
Evaluation Summary:
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and epithelial tumorigenesis. The analysis of single-cell transcriptomes of follicle cells now shows that transcriptionally distinct cell populations emerge shortly after induction of loss of polarity. Strengths of the work include the use of advanced single cell omics and imaging analyses to identify cell types and factors playing a role the disruption of polarity and the implications of this work for epithelial cancers. The authors' claims are generally well supported by the data and analyses. Weaknesses include the lack of high magnification images and need to clarify motivation for the study and highlight the biology rather than technical advances in the results section. Overall, this work is viewed as an important …
Evaluation Summary:
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and epithelial tumorigenesis. The analysis of single-cell transcriptomes of follicle cells now shows that transcriptionally distinct cell populations emerge shortly after induction of loss of polarity. Strengths of the work include the use of advanced single cell omics and imaging analyses to identify cell types and factors playing a role the disruption of polarity and the implications of this work for epithelial cancers. The authors' claims are generally well supported by the data and analyses. Weaknesses include the lack of high magnification images and need to clarify motivation for the study and highlight the biology rather than technical advances in the results section. Overall, this work is viewed as an important contribution to cell biologists who work on the epithelial morphogenesis or tumorigenesis.
(This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their name with the authors.)
-
Reviewer #1 (Public Review):
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and tumorigenesis of epithelia. Studies have found that lack of proper cell polarity causes invasive delamination of cells and formation of multilayered epithelia, reminiscent of Epithelial-Mesenchymal Transition (EMT). Using this system, the authors analyzed the single-cell transcriptome of follicle cells and show that distinct cell populations emerge shortly after induction of polarity loss. Authors identified dynamic activation of Keap1-Nrf2 pathway Finally, subpopulation classification and analysis of regulon activity identified that Keap1-Nrf2 pathway is responsible for epithelial multilayering caused by polarity loss.
Strengths: The authors characterized the single-cell transcriptome of follicle cell …
Reviewer #1 (Public Review):
Drosophila ovarian follicle cells have been utilized as a model system to study organogenesis and tumorigenesis of epithelia. Studies have found that lack of proper cell polarity causes invasive delamination of cells and formation of multilayered epithelia, reminiscent of Epithelial-Mesenchymal Transition (EMT). Using this system, the authors analyzed the single-cell transcriptome of follicle cells and show that distinct cell populations emerge shortly after induction of polarity loss. Authors identified dynamic activation of Keap1-Nrf2 pathway Finally, subpopulation classification and analysis of regulon activity identified that Keap1-Nrf2 pathway is responsible for epithelial multilayering caused by polarity loss.
Strengths: The authors characterized the single-cell transcriptome of follicle cell subpopulations after induction of polarity loss. Using temperature-inducible driver, they can induce the polarity loss in a short period of time, which enables detection of epithelial populations in various transition stages. Detected cell-heterogeneity could be caused intrinsically or by environmental cues within in vivo tissue. Therefore, it is likely well recapitulating tumorigenesis in vivo.
Weaknesses:
- Authors should show cells corresponding to identified key cell clusters within the tissue by immunostaining, GFP-trap, or RNA FISH.
- Images are low magnification and difficult to see individual cells.
- Manuscript is written weighted toward the technical aspect and more biology behind this study has to be discussed.
-
Reviewer #2 (Public Review):
Chatterjee et al. perform extensive image and single-cell RNA sequencing (scRNA-seq) analysis of Drosophila ovaries with and without knockdown of a gene, Lethal giant larvae (Lgl), which is known to establish apical-basal polarity as well as controlling proliferation of epithelial tissues. The goal of the study is to characterize the effect of apicobasal-polarity loss in epithelial cells via Lgl knockdown on Drosophila ovaries at the phenotypic, cellular, single-cell gene expression and regulatory level. By focusing on single-cell gene expression clusters that are unique to Lgl-KD compared to those from flies without the knockdown, they were able to identify a highly transient cluster (cluster 7) which consists of tumorigenic cells. Differential markers within a sub-cluster (cluster 7_3) of this cluster …
Reviewer #2 (Public Review):
Chatterjee et al. perform extensive image and single-cell RNA sequencing (scRNA-seq) analysis of Drosophila ovaries with and without knockdown of a gene, Lethal giant larvae (Lgl), which is known to establish apical-basal polarity as well as controlling proliferation of epithelial tissues. The goal of the study is to characterize the effect of apicobasal-polarity loss in epithelial cells via Lgl knockdown on Drosophila ovaries at the phenotypic, cellular, single-cell gene expression and regulatory level. By focusing on single-cell gene expression clusters that are unique to Lgl-KD compared to those from flies without the knockdown, they were able to identify a highly transient cluster (cluster 7) which consists of tumorigenic cells. Differential markers within a sub-cluster (cluster 7_3) of this cluster followed by validation using a GstD-lac-Z enhancer-trap reporter assay lead to their conclusion that cluster 7 represents the cells of multilayering phenotype (i.e., the major Lgl-KD phenotype observed from image analysis) where activation of Keap1-Nrf2 signaling was observed. The KEAP1-NRF2 pathway is associated with protecting cells from oxidative stress. KEAP1 forms part of an E3 ubiquitin ligase, which controls NRF2, a transcription factor, by targeting it for ubiquitin-mediated proteasomal degradation. Surprisingly, inducing loss of function of both Keap1 and separately NRF2 (cnc in Drosophila) in Lgl-KD cells resulted in the same phenotype/rescue, loss of the multilayering phenotype. Over expression of Keap1 in Lgl-KD induced increased multilayer volume compared to Lgl-KD alone further supporting the role of Keap1 in cellular invasion and possibly early stages of tumorigenesis when epithelial cells start loosing their polarity.
The strengths of this paper are:
The mutually reinforcing advanced imaging, scRNA-seq and genetic manipulation (knockdown and over expression) experiments/analyses that largely support the major conclusions of the manuscript which are summarized above as well as more minor observations that the authors make.
The systems biology flow of the study from broad to a specific gene/pathway implicated in the phenotype. The authors start with a clear phenotypic characterization of Lgl-KD and genome-wide scRNA-seq analysis. This leads to regulatory factor enrichment and further identification of a cluster (cluster 7) and then to a sub-cluster (cluster 7_3). This is followed by the identification of the KEAP1-NRF2 pathway and demonstration that KEAP1 knockdown and overexpression in Lgl-KD rescues and aggravates the cell multilayering phenotype, respectively.
The multilayering phenotype, genes and regulatory factors associated with loss of polarity are known to play an important role in the epithelial to mesenchymal transition (EMT). For example, this includes the enrichment of AP-1 family members, which are known to regulate EMT, in the regulon analyses as well as identification of KEAP1-NRF2.
The weaknesses of the paper are:
The framing/motivation of the study could be improved especially for those who study EMT/metastasis in humans. Given that loss of polarity is one of many events associated with tumorigenesis and metastatic progression, the claims made that studying Lgl-KD in Drosophila ovaries directly leads to insights into tumor cell invasiveness, early stages of tumorigenesis and EMT may leave some readers doubtful if they are not familiar with Lgl. Reviewing major findings that show that Lgl is a tumor suppressor as is its human homologue Hugl-1 as well as making a stronger case that studying Lgl-KD in Drosphila is relevant for tumorigenesis and EMT would be helpful.
Given that Keap1 antagonizes NRF2, the apparent contradictory result that inducing loss of function of both Keap1 and separately NRF2 (cnc in Drosophila) in Lgl-KD cells resulted in the same phenotype/rescue (loss of the multilayering phenotype) is not fully addressed. Keap1 over expression revealed it aggravates multilayering. NRF2 over expression experiments were not performed. In addition, it was shown that over expression and knockdown of Keap1 did not affect NRF2 gene expression (Figure 5C); however, Keap1 regulates Nrf2 at the protein level directly via ubiquitin-mediated proteasomal degradation. Nrf2 protein levels in flies with and without Lgl-KD with various manipulations of Keap1 including control, KD and OE were not measured.
Many of the conclusions in early Results paragraphs are purely technical and not biological. For example, "These observations highlight the limitations of marker validation to identify specific cells of the differential Lgl-KD phenotype" and "SCENIC was able to detect the common as well as distinct transcriptomic states of the cells in unique Lgl-KD clusters, while also highlighting the heterogeneity among them". Some of these technical conclusions could be part of brief discussions in the Methods section.
For those not familiar with various detailed scRNA-seq analysis approaches (e.g., RNA velocity analysis), a brief description of how they should be interpreted biologically in Methods would be helpful. This might help resolve what appear to be contradictory/confusing results. First, the upper branch of cluster 7 (which is a focus of the study) shown in Fig. 3B is in a "late" stage based on Velocity Pseudotime analysis (left panel) and a "root" or an early stage based on Terminal end-points of differential analysis (right panel). The bottom branch of cluster 7 is "late"/"stable end point" based on these two analyses which is now consistent. Second, given these differences between the upper and lower branch of cluster 7, how is cluster 7 biologically the same cluster?. Third, the bottom branch of cluster 7 bleeds into cluster 8 and while Ets21C is uniquely expressed in the bottom branch of 7, important markers of the study including Jra, kay (AP-1 family members), grnd, cnc (NRF2), Keap1, and the genes shown in Fig. 6F are all robustly expressed in clusters 7 (bottom branch) and 8. The biologically relevant distinction between the bottom branch of cluster 7 and 8 is not clear. Is cluster 8 important/relevant to the phenotypes observed as well?
-
Reviewer #3 (Public Review):
In this manuscript, the authors aim to identify the regulators of epithelial invasiveness upon Lethal giant larvae (Lgl), a basolateral polarity protein, knockdown in the follicular epithelium of the Drosophila ovaries, which can serve as a model system to investigate cellular plasticity when apical-basal polarity is lost. Knockdown (KD) of Lgl causes a multilayered epithelium and through extensive single cell RNA-seq analyses, the authors demonstrate that Lgl-KD triggers the appearance of groups of cells exhibiting tumor-associated molecular signatures and invasive behaviour. Overall, the manuscript is technically sound and the combination of computational and experimental approaches results in a thorough characterisation of the earliest steps of epithelial de-stabilisation upon the loss of apical-basal …
Reviewer #3 (Public Review):
In this manuscript, the authors aim to identify the regulators of epithelial invasiveness upon Lethal giant larvae (Lgl), a basolateral polarity protein, knockdown in the follicular epithelium of the Drosophila ovaries, which can serve as a model system to investigate cellular plasticity when apical-basal polarity is lost. Knockdown (KD) of Lgl causes a multilayered epithelium and through extensive single cell RNA-seq analyses, the authors demonstrate that Lgl-KD triggers the appearance of groups of cells exhibiting tumor-associated molecular signatures and invasive behaviour. Overall, the manuscript is technically sound and the combination of computational and experimental approaches results in a thorough characterisation of the earliest steps of epithelial de-stabilisation upon the loss of apical-basal polarity. In my view, the aims set by the authors are met and the experimental data provided support the claims. Interpretations are balanced and the display items are presented logically and informatively for even non-experts. Together, this work will set the basis for further investigations using apical-basal destabilisation of the follicle epithelium as a model of epithelial tumorigenesis.
-
