Jag1/2 maintain esophageal homeostasis and suppress foregut tumorigenesis by restricting the basal progenitor cell pool

Basal progenitor cells are crucial for maintaining foregut (the esophagus and forestomach) homeostasis. When their function is dysregulated, it can promote inflammation and tumorigenesis. However, the mechanisms underlying these processes remain largely unclear. Here, we employ genetic mouse models to reveal that Jag1/2 regulate esophageal homeostasis and foregut tumorigenesis by modulating the function of basal progenitor cells. Deletion of Jag1/2 in mice disrupts esophageal and forestomach epithelial homeostasis. Mechanistically, Jag1/2 deficiency impairs activation of Notch signaling, leading to reduced squamous epithelial differentiation and expansion of basal progenitor cells. Moreover, Jag1/2 deficiency exacerbates the deoxycholic acid (DCA)-induced squamous epithelial injury and accelerates the initiation of squamous cell carcinoma (SCC) in the forestomach. Importantly, expression levels of JAG1/2 are lower in the early stages of human esophageal squamous cell carcinoma (ESCC) carcinogenesis. Collectively, our study demonstrates that Jag1/2 are important for maintaining esophageal and forestomach homeostasis and the onset of foregut SCC.

This manuscript by Huang et al carefully characterized the morphological and molecular phenotypes in Jag1/2KO mouse esophagus and showed certain relevance to human esophagitis and esophageal squamous cell carcinoma.This study is an extension of their published work on the role of Notch pathway in embryonic esophageal development (Ref#37).It is, in fact, the first one to systematically understand the role of Notch ligands (Jag1/2) in the homeostasis of adult esophageal epithelium and esophageal diseases.This reviewer has several major concerns and minor concerns as follows: Major concerns: 1.The positional relationship between the ligands, receptors, and effectors of the Notch pathway is not clear.For example, immunofluorescent staining of Jag1, Notch1 (maybe NICD1), and Hes1 may address this issue.2. Clinical relevance of the findings needs to be clearly defined.For example, how are Jag1/2 and Dll1/3/4 expressed in human esophageal epithelium?Do they follow similar expression patterns as in the mouse esophageal epithelium?Are there any differences in Jag1/2 expression in normal vs GERD esophageal epithelium?Does Jag1/2 downregulation take place in human esophageal squamous cell carcinoma in comparison with the normal esophageal epithelium?Is Jag1/2 downregulation an early or late event during esophageal carcinogenesis?One critical and important issue is whether the mouse models (Figure 5a/6a) recapitulate human diseases or just a biological phenomenon."Deregulation of homeostasis", "perturbed epithelium", etc are biological terms, not terms in clinical pathology.A clinical pathologist may be consulted regarding these issues related to clinical relevance.3. Lack of mechanistic in-depth is a concern.For example, how does Jag1/2KO impact the PI3K pathway?Why does Jag1/2KO cause body weight loss and death when the mice were given 4NQO (Figure 5b/6b)?Why does Jag1/2KO cause inflammation?If compromised epithelial barrier function and subsequent bacterial infection are important factors, additional experimental data need to be presented, for example, using germ-free mice.4. Dilated intercellular space and leaky epithelium are possible consequences of Jag1/2KO.TEM and TEER analysis are needed to validate these structural and functional phenotypes in the esophageal epithelium.
Minor concerns: 1.Protein names and gene names need to be expressed consistently and follow the requirements of the journal.For example, most cancer journals may require protein names to be capitalized.2. DCA should be deoxycholic acid instead of dichloroacetate.DCA is rarely detected in the upper gastrointestinal tract in humans.Treatment with gastric content (for example, acid, enzymes, conjugated bile acids) is more clinically relevant than DCA treatment.So, in this study, DCA is mainly an experimental tool to mimic chemical insults.3. Figure 1b/1c shows some staining of Jag1 in the superficial layer.Is that true staining or due to marginal effect of staining?Jag1/2 knockout in p63+ cells needs to be validated by immunostaining.4. Figure 1e/6d appears to show increased keratinization in Jag1/2KO esophagus and forestomach in comparison to WT esophagus.Please explain why. 5. 4NQO is known to produce squamous cell carcinoma in the esophagus and tongue as well as the forestomach.Phenotypes of the 4NQO-treated tongue and esophagus need to be included.6. Jag1/2KO exacerbates "chemically induced inflammation" instead of "inflammatory disease".

Reviewer #2 (Remarks to the Author):
In this research, Huang et al. report that Jag1/2 expression in basal cells regulates esophageal homeostasis and suppresses tumorigenesis of foregut.P63Cre mouse was used as a genetic tool to delineate the role of Jag1/2 specifically in basal cells.Loss of Jag1/2 in p63-expressing cells resulted in increased proliferation of basal progenitor cells, defective differentiation of epithelial cells and loss of cell polarity of basal progenitor cells.In a mouse model of DCA-induced gastroesophageal inflammation, P63CreJag1/2f/f mice developed exacerbated disease accompanied with increased myeloid cell infiltration.In 4-NQO-induced SCC model, P63CreJag1/2f/f mice manifested worsened pathology, more weight loss and rate of morality.Finally, reduced Jag2 expression significantly correlated with poor survival of ESCC patients.These findings are of potential interest.However, key weakness in this research was reflected in the following aspects.First, it is not appropriate to use WT mouse as a control for P63Cre-ERT2Jag1/2f/f mouse.According the reference described (PMID: 25210499), Cre-ERT2 was knocked into exon 4 of p63, disrupting one allele of p63.Therefore, this mouse functions as a p63 heterozygous mouse.Although it was not described clearly anywhere in the manuscript, I believe that the authors used P63Cre-ERT2/+Jag1/2f/f for their experiments (described as P63Cre-ERT2Jag1/2f/f).Please describe the genotype clearly in the manuscript and use P63Cre-ERT2/+Jag1/2+/+ mouse as a control for their experiments.This point is critical also because it has been shown previously that the basal cells are heterogeneous containing "p63+/K14-/K5-(3.2%), p63-/K14-/K5+ (5.6%), p63+/K14-/K5+ (4.0%), p63-/K14+/K5-(1.6%),p63+/K14+/K5-(9.6%),p63-/K14+/K5+ (14.6%) and p63+/K14+/K5+ (61.4%)" for the prostate basal cells (PMID: 25210499).This could affect the conclusions drawn from the observation in Fig. 3, as Krt5 could be differentially expressed by p63+ and p63-basal cells.Second, it is not known whether loss of Jag1 and Jag2 alone in p63+ cells is sufficient to cause abnormality of basal cells.Third, mechanism of how Jag1/2 affect the proliferation/differentiation of basal cells is not clearly addressed.As Jag1/2 are ligands of Notch, they are supposed to function through acting on Notch-expressing cells, further leading to changed expression of downstream target genes (such as Hes1, Integrins and ZO-1 etc).Specific points: 1.In this research, the authors deleted Jag1 and Jag2 simultaneously in p63+ cells (including the progenies of p63+ cells).Was Jag1 and Jag2 functionally redundant in regulating basal cell homeostasis?How was the phenotype of p63Cre crossed to Jag1f/f or Jag2f/f mouse (loss of Jag1 or Jag2 alone in p63+ cells)?2. Jag1 and Jag2 are ligands of Notch.Limited evidence supported that Notch signaling were severely impaired upon deletion of Jag1/2, except for reduced Hes1 expression.Jag1/2 were supposed to act on Notch-expressing cells.Upon deletion of Jag1/2, it remains unclear what cells are direct targets of Jag1/2.Deletion of ligands may affect all Notchexpressing cells in the microenvironment, including stroma cells, immune cells etc.In terms of findings of this research, it was unclear if Integrins, E-cadherin, ZO-1 were direct targets of Jag1/2-Notch/RBP-J signaling pathway.Does Jag1/2 expression in basal cells mainly affect Notch signaling within the basal cells themselves?In vitro experiments will be helpful to address this point.
3. Fig6: It is unclear if the weight loss/lethality of KO mice were due to squamous cell carcinoma.As p63Cre also delete basal cells in the lung, bladder, etc, Jag1/2 KO mice may also have abnormalities in other organs, which is important to look at.4. Figure 4b-f: The authors showed a very small scope of immunofluorescence picture demonstrating the changed expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ, which was hard to reflect the real change of whole esophageal tissue.Please use a smaller magnification to depict a wider scope of the whole tissue.Also, please perform statistical analysis on the expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ.It will also be helpful to analyze the expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ using western blot with the protein lysates of esophageal epithelia, which more accurately represents the change of these molecules in the whole tissue.
Other points: 1.It was not clearly described where the Jag1/Jag2 mice were from and how they were the generated.Line 357. 2. RNA-seq data should be deposited in a public database.3. Y axis label is missing for Figure . 4a.

Reviewer #3 (Remarks to the Author):
In the manuscript by Zhang and colleagues state-of-the-art genetic mouse models were used to define the role of Jag1/2 in esophageal homeostasis and tumor initiation.The manuscript is a strong continuation around the tumor suppressive function of Notch signaling in esophageal cancer.Loss of function of Jag1 and Jag2 describes the importance of this signaling axis for homeostatic conditions and the generation of self-renewing basal progenitors, drives inflammation and tumor initiation of the squamous epithelium of the forestomach.The experiments are sound and the manuscript is clearly written and presented.The major concern is the lack of mechanistic understanding of the observed phenotypes.Therefore, the current version of the manuscript is to descriptive for publication in Nature Communications.
Major comments: Lacking mechanism As mentioned, there are a number of interesting observations assembled, however, which mechanistic determinations are crucial for the observed stem cell dynamics, immune cell infiltration and tumor initiation is not clear.To gain causative understanding, at least one of these aspects should be dissected in much more mechanistic detail.

Human relevance
The presented analysis of human data in Fig. 8 is relatively sparse and should be solidified, also along the lines of the mechanism's aspects.In addition, staining of JAG1 and JAG2 on human ESCC would be highly informative to understand spatial heterogeneity and potential effects of lateral Notch inhibition.
This manuscript by Huang et al carefully characterized the morphological and molecular phenotypes in Jag1/2KO mouse esophagus and showed certain relevance to human esophagitis and esophageal squamous cell carcinoma.This study is an extension of their published work on the role of Notch pathway in embryonic esophageal development (Ref#37).It is, in fact, the first one to systematically understand the role of Notch ligands (Jag1/2) in the homeostasis of adult esophageal epithelium and esophageal diseases.This reviewer has several major concerns and minor concerns as follows: We appreciate the reviewer's constructive comments and, more importantly, the insightful suggestions.We have conducted the recommended experiments and incorporated the valuable suggestions, which have significantly helped strengthen the work, especially in the aspects of mechanistic studies and clinical relevance.
Major concerns: 1.The positional relationship between the ligands, receptors, and effectors of the Notch pathway is not clear.For example, immunofluorescent staining of Jag1, Notch1 (maybe NICD1), and Hes1 may address this issue.
We thank the reviewer's insightful suggestions and have accordingly conducted the recommended immunostaining.The immunostaining shows that Jag1/2 are mainly expressed in the basal cells of the esophageal epithelium (Fig. 1b, c).Notch1 localized to the membrane of basal cells, but mainly in the nucleus of suprabasal cells (Fig. 5d).Western blot analysis revealed a decrease in the protein levels of NICD1 and immunostaining showed a reduction in the number of nuclear Notch1 + suprabasal cells upon Jag1/2 deletion (Fig. 5b, d).Based on these analyses, we hypothesize that Jag1/2 on basal cells activate Notch signaling in suprabasal cells to promote their differentiation, while a deletion of Jag1/2 impairs Notch signaling, resulting in a reduction of epithelial differentiation.We also attempted immunostaining of Hes1 and NICD1 on tissues, but unfortunately, it was unsuccessful.
2. Clinical relevance of the findings needs to be clearly defined.For example, how are Jag1/2 and Dll1/3/4 expressed in human esophageal epithelium?Do they follow similar expression patterns as in the mouse esophageal epithelium?Are there any differences in Jag1/2 expression in normal vs GERD esophageal epithelium?Does Jag1/2 downregulation take place in human esophageal squamous cell carcinoma in comparison with the normal esophageal epithelium?Is Jag1/2 downregulation an early or late event during esophageal carcinogenesis?
We thank the reviewer's insightful questions.To answer these questions, we have analyzed single-cell RNA sequencing data of human esophageal epithelial cells of normal, inflammation, neoplasia, and ESCC (Fig. 9a, b).First, we agreed with the reviewer's insight below that the DCA-treatment mouse model serves as an experimental tool to mimic chemical insults, which is characterized by inflammation.
As shown in Fig. 9a, b, both JAG1 and JAG2 exhibited lower expression in human esophageal epithelial cells under inflammatory conditions compared to the normal state.Moreover, there is also a downregulation of JAG1/2 in esophageal epithelial cells with neoplasia (Fig. 9a, b).However, JAG1/2 are unregulated in the ESCC compared to the normal esophageal epithelium, as evidenced by both single-cell RNA sequencing and immunostaining of human ESCC samples (Fig. 9a-d).These findings demonstrated that JAG1/2 downregulation is an early event in ESCC carcinogenesis.Our findings corroborate previous studies that have established Notch signaling as a tumor suppressor of squamous cell carcinoma (SCC) initiation [1][2][3][4] .Nevertheless, contrasting evidence suggests a pro-tumorigenic role of Notch signaling during SCC progression 5,6 .In line with this, we found that the expression levels of JAG1/2 were elevated in the human ESCC samples compared to normal samples.It is possible that Jag1/2-mediated Notch signaling exerts distinct functions at different stages of ESCC carcinogenesis, which is suppressing the onset of ESCC while promoting its progression.However, validating these hypotheses necessitates further investigation through comprehensive mouse genetic studies by modulating Jag1/2-Notch signaling at various stages of carcinogenesis.We have included these findings and details in the results (lines 297-308) and discussion sections (lines 388-399).Again, we appreciate the reviewer's insightful suggestions, which have significantly improved the clinical relevance of our studies.
One critical and important issue is whether the mouse models (Figure 5a/6a) recapitulate human diseases or just a biological phenomenon."Deregulation of homeostasis", "perturbed epithelium", etc are biological terms, not terms in clinical pathology.A clinical pathologist may be consulted regarding these issues related to clinical.
The 4-NQO mouse model has been widely utilized in the field to investigate the mechanisms underlying ESCC carcinogenesis.As the reviewer suggested, we have consulted a pathologist and confirmed that the histological features observed in this model resemble those of human ESCC.Regarding the DCA-treatment mouse model, we agreed with the reviewer's assessment that it serves as an experimental model to mimic chemical insults.We have integrated this insight into the revised manuscript.
Overall, these mouse models bear significant clinical relevance in addition to their biological significance.
3. Lack of mechanistic in-depth is a concern.For example, how does Jag1/2KO impact the PI3K pathway?
We thank the reviewer's insightful questions.To elucidate the molecular mechanisms, we have conducted the following experiments.Given that Jag1/2 are important ligands for activating Notch signaling, which has been shown essential for esophageal epithelial cell differentiation 7,8 , we set out to determine whether the ablation of Jag1/2 impairs the Notch signaling pathway to reduce cell differentiation.We first conducted gene set enrichment analysis (GSEA) on the RNA-sequencing data by comparing the esophageal epithelium of the Jag1/2 cKO mutants to the control.The analysis revealed a downregulation of Notch signaling in the Jag1/2 cKO mutants (Fig. 5a).Subsequent western blot analysis demonstrated a reduction in the protein levels of Notch1 intracellular domain (NICD1), the activated form of Notch1, upon Jag1/2 deletion (Fig. 5b).The expression of the differentiation marker Krt4 was decreased (Fig. 5c), in line with the immunostaining analysis (Fig. 3e, f).Furthermore, immunostaining showed that the nuclear Notch1 + cells are reduced in the esophageal epithelium (Fig. 5d, e).These results demonstrated that Jag1/2 deletion impaired the activation of Notch signaling.
To further investigate whether Jag1/2 directly modulate epithelial cell differentiation, we generated 3D organoids from mouse esophageal epithelial cells (Fig. 5f).Deletion of Jag1/2 in the organoids led to the upregulation of the basal cell marker p63 but a decrease in the expression of the suprabasal cell marker Krt4 upon Jag1/2 deletion (Fig. 5g).Western blot analysis consistently showed a downregulation of Krt4 protein levels (Fig. 5h).Moreover, immunostaining on organoids revealed an increase in p63 + basal cells while showing a reduction in squamous differentiation with Jag1/2 deletion (Fig. 5i-k).Moreover, the protein levels of NICD1 were downregulated (Fig. 5h), and nuclear Notch1 + cells were reduced in Jag1/2 KO organoids (Fig. 5l).Together, these results suggest that Jag1/2 within esophageal epithelial cells promote squamous differentiation through the Notch signaling pathway.
In terms of PI3KCG, PI3KCG has been identified as a direct target of Notch/Rbpjk signaling (PMID: 25808869) 9 .While the study revealed that Notch signaling enhanced PI3KCG expression in breast cancer, our findings showed reduced PI3KCG expression in Jag1/2 cKO mutants.This discrepancy likely arises because Notch signaling functions differently in a context-dependent manner.Additionally, we have also attempted to conduct Rbpjk and NICD1 ChIP-seq studies.Regrettably, the experiments did not yield conclusive results.It will be of significance to investigate whether this gene and other differentially expressed genes (DEGs) are direct targets of the Jag1/2mediated Notch signaling pathway in the esophageal epithelium in future studies.
Overall, our mechanistic studies support that Jag1/2 KO reduces squamous epithelial differentiation, leading to basal cell hyperplasia, by impairing Notch signaling activation.
Why does Jag1/2KO cause body weight loss and death when the mice were given 4NQO (Figure 5b/6b)?Why does Jag1/2KO cause inflammation?If compromised epithelial barrier function and subsequent bacterial infection are important factors, additional experimental data need to be presented, for example, using germ-free mice.
We thank the reviewer's questions.Our initial studies showed that Jag1/2 KO leads to increased intercellular space in the foregut squamous epithelium, resulting in enhanced epithelial permeability upon insults (Supplementary Fig. S3).Both DCA and 4-NQO are potent squamous epithelial damaging agents, and the increased permeabilization of these agents in Jag1/2 KO mutants exacerbated injury, leading to increased inflammation (Fig. 6d, h, i).Additionally, the expression levels of pro-inflammatory cytokines were increased in the epithelial cells of Jag1/2 mutants (Supplementary Fig. 7).Therefore, the enhanced inflammation likely resulted from a combined effect involving both chemical insult and the inflammatory cytokines.Moreover, both the control and Jag1/2 KO mutants were maintained under specific-pathogen-free conditions, and increased inflammation in Jag1/2 KO mutants was only observed following treatment with DCA or 4-NQO.Hence, subsequent infections are not assumed to be the primary factors for the in these models.That being said, it will be of significance to determine whether Jag1/2 KO mice are more susceptible to the pathogenic infection in future studies.This insight was incorporated into the discussion section (lines 375-378).
4. Dilated intercellular space and leaky epithelium are possible consequences of Jag1/2KO.TEM and TEER analysis are needed to validate these structural and functional phenotypes in the esophageal epithelium.
We thank the reviewer's insightful suggestion.We conducted the analyses as suggested.Transmission electron microscope (TEM) analysis showed a consistent dilated intercellular space between basal cells in Jag1/2 cKO mutants (Fig. 4f).Regarding the functional analysis, performing TEER measurements on mouse foregut mucosa is highly challenging.Thus, we turned to Evans blue dye staining as an alternative method 10,11 .The staining revealed increased permeability in the Jag1/2 cKO forestomach mucosa treated with DCA (Supplementary Fig. 3).
Minor concerns: 1.Protein names and gene names need to be expressed consistently and follow the requirements of the journal.For example, most cancer journals may require protein names to be capitalized.
We thank the reviewer's comments.The spelling of the protein and gene names has been corrected.To clarify, the first letter was capitalized for mouse protein and gene names, while all letters were capitalized for human protein and gene names.
2. DCA should be deoxycholic acid instead of dichloroacetate.DCA is rarely detected in the upper gastrointestinal tract in humans.Treatment with gastric content (for example, acid, enzymes, conjugated bile acids) is more clinically relevant than DCA treatment.So, in this study, DCA is mainly an experimental tool to mimic chemical insults.
We appreciate the reviewer's corrections and insightful suggestions.We have corrected the spelling to "deoxycholic acid".We have incorporated the reviewer's insight that DCA treatment is mainly utilized as an experimental tool to mimic chemical insults.1b/1c shows some staining of Jag1 in the superficial layer.Is that true staining or due to marginal effect of staining?Jag1/2 knockout in p63+ cells needs to be validated by immunostaining.

Figure
We thank the reviewer's questions.The staining on the superficial layers is marginal staining, and we have added annotations to clarify in the figure legends (Fig. 1b, c and Supplementary Fig. 1c, d).The immunostaining confirming the Jag1/2 knockout in the esophageal epithelium in Jag1/2 cKO mutants was included (Supplementary Fig. 1e, f).1e/6d appears to show increased keratinization in Jag1/2KO esophagus and forestomach in comparison to WT esophagus.Please explain why.

Figure
We thank the reviewer's questions.Original Figure 1e: WT and Jag1/2KO esophagus displayed similar keratinization.We have re-cropped the representative enlarged image for WT for clarification.Original Figure 6d: large tumors developed in the Jag1/2 KO mutants, with tumor cells expanding towards the lumen and undergoing apoptosis and eventually keratinization.Therefore, the increased keratinization is a result of the expansion of tumor cells in the mutants.5. 4NQO is known to produce squamous cell carcinoma in the esophagus and tongue as well as the forestomach.Phenotypes of the 4NQO-treated tongue and esophagus need to be included.analysis (GSEA) on the RNA-sequencing data by comparing the esophageal epithelium of the Jag1/2 cKO mutants to the control.The analysis revealed a downregulation of Notch signaling in the Jag1/2 cKO mutants (Fig. 5a).Subsequent western blot analysis demonstrated a reduction in the protein levels of Notch1 intracellular domain (NICD1), the activated form of Notch1, upon Jag1/2 deletion (Fig. 5b).The expression of differentiation marker Krt4 was decreased (Fig. 5c), in line with the immunostaining analysis (Fig. 3e, f).Furthermore, immunostaining showed that the nuclear Notch1 + cells are reduced in the esophageal epithelium (Fig. 5d, e).These results demonstrated that the Jag1/2 deletion impaired the activation of Notch signaling.Previous studies conducted by our group and others have shown that Notch signaling is required for esophageal epithelial cell differentiation 7,8 .Therefore, these findings demonstrated that the Jag1/2 ablation impaired Notch signaling leading to reduced cell differentiation.
To further investigate whether Jag1/2 directly modulate epithelial cell differentiation, we generated 3D organoids from mouse esophageal epithelial cells (Fig. 5f).Deletion of Jag1/2 in the organoids led to the upregulation of the basal cell marker p63 but a decrease in the expression of the suprabasal cell marker Krt4 upon Jag1/2 deletion (Fig. 5g).Western blot analysis consistently showed a downregulation of Krt4 protein levels (Fig. 5h).Moreover, immunostaining on organoids revealed an increase in p63 + basal cells while showing a reduction in squamous differentiation with Jag1/2 deletion (Fig. 5i-k).Moreover, the protein levels of NICD1 were downregulated (Fig. 5h), and nuclear Notch1 + cells were reduced in Jag1/2 KO organoids (Fig. 5l).These results together suggest that Jag1/2 within esophageal epithelial cells promote squamous differentiation through the Notch signaling pathway.
We have also endeavored to conduct Rbpjk and NICD1 ChIP-seq studies to determine whether Integrins, E-cadherin, ZO-1 are direct targets of the Jag1/2-Notch/RBP-J signaling pathway.Regrettably, the experiments did not yield conclusive results.Notably, previous studies have demonstrated that the Notch signaling pathway negatively regulates the expression of the basal cell master transcription factor p63, leading to cell differentiation in skin keratinocytes 12 .Additionally, p63 has been shown to directly transactivate the expression of Itga6, Itgb1, and Itgb4 13 .Our findings revealed an increase in the expression levels of p63, and expansion of p63 + basal cells with Jag1/2 deletion (Figs.3b-d, 5g, i, j).It is possible that the reduced epithelial differentiation and increased expression of integrins in the Jag1/2 KO mutants are attributed to the upregulation of p63.We have included this potential explanation in the discussion section (lines 360-366).Overall, our studies elucidated a mechanism that Jag1/2 deficiency diminishes esophageal epithelial differentiation through the Notch signaling pathway.
3. Fig6: It is unclear if the weight loss/lethality of KO mice were due to squamous cell carcinoma.As p63Cre also delete basal cells in the lung, bladder, etc, Jag1/2 KO mice may also have abnormalities in other organs, which is important to look at.
We thank the reviewer's insightful suggestions.We have conducted the suggested studies.Jag1/2 KO mice developed severe SCC in the forestomach at around 4-5 weeks (Fig. 6b-c).Once tumors developed, Jag1/2 KO mice exhibited illness, ceased food intake, and eventually succumbed (Supplementary Fig. 4a).Histological analyses showed that there were no apparent phenotypic changes in the lung and bladder of both the Jag1/2 cKO mutants and the control group (Supplementary Fig. 6c, d).Based on these analyses, we reasoned that mouse mortality is majorly attributed to the tumors developing in the forestomach.4. Figure 4b-f: The authors showed a very small scope of immunofluorescence picture demonstrating the changed expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ, which was hard to reflect the real change of whole esophageal tissue.Please use a smaller magnification to depict a wider scope of the whole tissue.Also, please perform statistical analysis on the expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ.It will also be helpful to analyze the expression of Itga6, Itga1, Itgb4 and E-cadherin and PKCζ using western blot with the protein lysates of esophageal epithelia, which more accurately represents the change of these molecules in the whole tissue.
We appreciate the reviewer's suggestions.As suggested, we included large images for all staining of Itga6, Itgb1, Itgb4 and E-cadherin and PKCζ (Fig. 4b-e, g).Quantification of fluorescence intensity, showing the upregulation of the protein Itga6, Itgb1, and Itgb4 proteins, has also been included (Fig. 4b-d).Additionally, we attempted western blot analysis using antibodies for Itga6, Itgb1, and Itgb4 from at least two sources, but regrettably, none yielded successful results.Furthermore, the immunostaining of E-cadherin and PKCζ was conducted to illustrate a dilated intercellular space and disrupted basal cell orientation, respectively.Their expression levels were not changed, and therefore quantification of their fluorescence intensity was not included.Overall, we are grateful for the reviewer's suggestions, which helped enhance the clarity of the data presentation.
Other points: 1.It was not clearly described where the Jag1/Jag2 mice were from and how they were the generated.Line 357.The references have been added accordingly.
2. RNA-seq data should be deposited in a public database.The RNA-seq data have been deposited in the Genome Sequence Archive (GSA) database under accession number CRA011089 and can be accessed for review through the following link: https://ngdc.cncb.ac.cn/gsa/s/UDe7kx7h.
3. Y axis label is missing for Figure . 4a.We have added the Y axis label accordingly.

Reviewer #3 (Remarks to the Author):
In the manuscript by Zhang and colleagues state-of-the-art genetic mouse models were used to define the role of Jag1/2 in esophageal homeostasis and tumor initiation.The manuscript is a strong continuation around the tumor suppressive function of Notch signaling in esophageal cancer.Loss of function of Jag1 and Jag2 describes the importance of this signaling axis for homeostatic conditions and the generation of selfrenewing basal progenitors, drives inflammation and tumor initiation of the squamous epithelium of the forestomach.The experiments are sound and the manuscript is clearly written and presented.The major concern is the lack of mechanistic understanding of the observed phenotypes.Therefore, the current version of the manuscript is too descriptive for publication in Nature Communications.
We appreciate the reviewer's constructive comments.To strengthen the mechanistic studies, following the suggestions, we have employed esophageal-derived 3D organoids and biochemical approaches to demonstrate that Jag1/2 directly regulates foregut epithelial differentiation by activating Notch signaling.Furthermore, we also follow the reviewer's suggestion to look into the expression of Jag1/2 in human ESCC using single cell-seq data and immunostaining of human ESCC tissue arrays.Significantly, we have found that Jag1/2 are down-regulated at the early stages of tumor initiation, which are inflammation and neoplasia, while their expression is increased in the tumors.These data suggest a distinct role of Jag1/2 in ESCC tumor initiation and progression.A detailed response to these questions well described as below.
Major comments: Lacking mechanism As mentioned, there are a number of interesting observations assembled, however, which mechanistic determinations are crucial for the observed stem cell dynamics, immune cell infiltration and tumor initiation is not clear.To gain causative understanding, at least one of these aspects should be dissected in much more mechanistic detail.
We thank the reviewer's insightful questions.As suggested, we chose to further investigate the mechanisms by focusing on the stem cell dynamics utilizing 3D organoids and biochemical approaches.
Given that Jag1/2 are important ligands for activating Notch signaling which has been shown essential for esophageal epithelial cell differentiation 7,8 , we set out to determine whether the ablation of Jag1/2 impairs the Notch signaling pathway to reduce cell differentiation.We first conducted gene set enrichment analysis (GSEA) on the RNA-sequencing data comparing the esophageal epithelium of the Jag1/2 cKO mutants to the control.The analysis revealed a downregulation of Notch signaling in the Jag1/2 cKO mutants (Fig. 5a).Subsequent western blot analysis demonstrated a reduction in the protein levels of Notch1 intracellular domain (NICD1), the activated form of Notch1, upon Jag1/2 deletion (Fig. 5b).The expression of the differentiation marker Krt4 was decreased (Fig. 5c), in line with the immunostaining analysis (Fig. 3e, f).Furthermore, immunostaining showed that the nuclear Notch1 + cells reduced in the esophageal epithelium (Fig. 5d, e).These results demonstrated that Jag1/2 deletion impaired the activation of Notch signaling.
To further investigate whether Jag1/2 directly modulate epithelial cell differentiation, we generated 3D organoids from mouse esophageal epithelial cells (Fig. 5f).Deletion of Jag1/2 in the organoids led to the upregulation of the basal cell marker p63 but a decrease in the expression of the suprabasal cell marker Krt4 upon Jag1/2 deletion (Fig. 5g).Western blot analysis consistently showed a downregulation of Krt4 protein levels (Fig. 5h).Moreover, immunostaining on organoids revealed an increase in p63 + basal cells while showing a reduction in squamous differentiation with Jag1/2 deletion (Fig. 5i-k).Moreover, the protein levels of NICD1 were downregulated (Fig. 5h), and nuclear Notch1 + cells were reduced in Jag1/2 KO organoids (Fig. 5l).These results together suggest that Jag1/2 within esophageal epithelial cells promote squamous differentiation through the Notch signaling pathway.
Overall, our mechanistic studies support that Jag1/2 KO reduces squamous epithelial differentiation, leading to basal cell hyperplasia, by decreasing Notch signaling.This in line with the overall conclusion that the disruption of Jag1/2-mediated Notch signaling reduces the differentiation of the basal cells, leading to expansion of basal progenitor cells and facilitating SCC initiation.We thank the reviewer's suggestion which have helped strengthen the mechanisms of the study.

Human relevance
The presented analysis of human data in Fig. 8 is relatively sparse and should be solidified, also along the lines of the mechanism's aspects.In addition, staining of JAG1 and JAG2 on human ESCC would be highly informative to understand spatial heterogeneity and potential effects of lateral Notch inhibition.