Pparg promotes differentiation and regulates mitochondrial gene expression in bladder epithelial cells

The urothelium is an epithelial barrier lining the bladder that protects against infection, fluid exchange and damage from toxins. The nuclear receptor Pparg promotes urothelial differentiation in vitro, and Pparg mutations are associated with bladder cancer. However, the function of Pparg in the healthy urothelium is unknown. Here we show that Pparg is critical in urothelial cells for mitochondrial biogenesis, cellular differentiation and regulation of inflammation in response to urinary tract infection (UTI). Superficial cells, which are critical for maintaining the urothelial barrier, fail to mature in Pparg mutants and basal cells undergo squamous-like differentiation. Pparg mutants display persistent inflammation after UTI, and Nf-KB, which is transiently activated in response to infection in the wild type urothelium, persists for months. Our observations suggest that in addition to its known roles in adipogegnesis and macrophage differentiation, that Pparg-dependent transcription plays a role in the urothelium controlling mitochondrial function development and regeneration.

maintenance of urothelial differentiation in adult mice, as well as for urothelial development and regeneration. Perhaps most interestingly (and significantly), the work links deficits in Pparg expression and function to deficits in mitochondrial activity. This work is significant and will be of interest to individuals both within the field of urothelial cell biology, as well as those outside the field, including members of the steroid hormone receptor and developmental biology fields.
Minor concerns 1. It is stated that S cells express the highest level of Pparg. Was this quantified? It appears that other cell populations (I cells?) express Pparg. 2. In 3rd paragraph of the results section the authors state "urothelial cells lining the luminal layer were small compared to controls (Figure 1 l,m,n)." Which cell populations specifically are the authors referring to? These images are quite small and difficult to interpret fully. 3. Figure 1R is not discussed explicitly in the text. 4. In the text it would be helpful if the authors explicitly state if the model is an inducible system (i.e., in areas it is not explicitly stated that the Krt5-Cre ERT2 system is tamoxifen inducible). 5. Figure 2B is discussed first in the text relative to a. The figure should be re-arranged to accommodate the flow of the text, which is quite nice. 6. In the introduction, I thought urothelial cells extend to the proximal urethra, and not just the bladder neck. Is this not the case? 7. PPARG is amplified in a subset of bladder cancers as well, and there are recent studies by Bernard-Pierrot's group (PMID 30604486) that might be worth citing. 8. Were markers of apoptosis additionally examined in the Shh-cre experiments? 9. It would be interesting to see a heatmap showing altered transcription factors in Figure 1. 10. In the section beginning "Analysis of ShhCre;Ppargfl/fl mutants in which Pparg is deleted… were rapidly inactivated in mutants (Fig 3a,b)." I think the authors are referring to 3E, F. Please double check. Also at the end of this page, instead of 3I should it be 3L? 11. What is "over-representation analysis"? 12. What is the definition of an "immature S cell"? Not completely differentiated? 13. Was there any sex differences in the phenotypes in any of the experiments? 14. Are there differences in mitochondrial metabolism in subsets of human bladder cancer included in the TCGA or other cohorts?
Major concerns 1. It is suggestive that luminal markers such as Krt20 are detected at reduced levels in the Shh-Cre mice experiments ( Figure 1). However, it is not clear if these findings are reflective of decreased Krt20 expression, or the fact that pparg KO is toxic to urothelial cells, thus resulting in cell death (and inability to detect Krt20). Is Krt20 completely undetectable? Some of this might be addressed by subsequent experiments shown in the current submission, but I am not sure. 2. The orthogonal approaches shown in Figure 2 (mRNA expression and IF) are a clear strength, but this data should be quantified. While trends are important and support the overall conclusions, I am curious to know how many of these differences are statistically significant. Also-can the IF be quantified (counting positive/negative cells)? 3. While it is almost certainly safe to conclude (based on the findings) that Pparg regulates mitochondrial function in a cell autonomous manner, I am not sure that the same can be said about the cell autonomous manner in which pparg regulates cell differentiation or survival. This is true because it is possible that Pparg ablation influences barrier integrity, resulting in urine leaking through the urothelium, and potentially adversely affecting cell differentiation or survival. Therefore this conclusion should be scaled back a bit. 4. Figure 4 (Krt5-CreERT2/Pparg KO). It is difficult to judge the extent of squamous differentiation. Indeed, I would argue that the upregulation of a few markers of squamous differentiation in isolated cell populations is not indicative of over squamous differentiation (although it might be indicative of early stages of this process). Is hyperkeratinization and/or presence of intercellular bridges observed? Reviewer #3: Remarks to the Author: Liu and co-workers used different cell type-specific inducible PPARgamma knockout mouse models to study the role of PPARgamma in urothelium, both under healthy and UTI conditions. Their data suggest that PPARgamma is needed for normal urothelium development and homeostasis, for mitochondrial function, and is a potent regulator of the inflammatory response after UTI.
Major concern: The authors study the effect of PPARgamma deficiency by using the Cre-Lox model. They use 3 different tamoxifen-inducible Cre mouse lines to drive PPARgamma depletion in distinct urothelial sub-populations. In each case, they use the PPARg fl/fl as control mice. The authors need to proof that these mice are suitable control mice for these studies. Numerous studies have shown that Cre-recombinase expression by itself can have effects on cell differentiation and other aspects evaluated in this manuscript. Therefore, the PPARg fl/fl mice might be the wrong control mice. Have the authors, or others, studied the 3 cre-expressing lines and do they have proof that none of the observed phenotypes are (partially) due to Cre-expression? Moreover, while in the KRT5CreERT2 and UPK2CreERT2 mice the Cre recombinase is expressed as a transgene, in the ShhCre mice the Cre gene actually replaces the endogenous Shh gene (knock in). In the latter case, since Shh has been shown to play a critical role in bladder development (see PMID 26757905), the authors need to provide proof that having only one functional Shh allele does not affect the parameters evaluated in this study. In short, the authors either have to show data from the 3 Cre-expressing lines as controls throughout their study, or be able to cite previous studies (from others) that show that the 3 Cre-expressing mouse strains behave identical to wild-type mice in regards to the parameters studied.
Other concerns: 1) The role of PPARgamma in bladder cancer is very controversial, with numerous studies both claiming an oncogenic or anti-cancerous role for PPARgamma. This is not evident from the way the manuscript is currently written. The short paragraphs in the introduction, results, and discussion that touch upon this subject are vague (using terms like "associated with" and "regulate") and cite few recent publications. The authors should develop this further and more clearly (in a neutral manner), and include recent studies (examples PMIDs 30651555, 30912275, 30845932).
2) In the results (page 10), it is stated that "Upk3, which labels I-cells and S-cells, and Krt20 which labels mature S-cells were both down-regulated ( Fig. 3a-i)". However, Fig. 3i shows an increase in S cells. Please explain.
3) PPARgamma can regulate NFkappaB through different mechanisms, including by sequestering the P65 subunit, but also by regulating I-kappa-B-alpha and IKK-alpha/beta levels. It is not clear what is stained as NFkappaB in Fig. 5 and Supplemental Fig. 4 (no description in Methods section). The whole part on NFkappaB should be developed in more detail, including mechanistic details. Fig. 3 is never mentioned in the text.

Minor concerns:
The manuscript has been written in a very sloppy manner and includes numerous errors of which a non-exhaustive list follows below. Especially the Figure legends contained many errors and the Methods section is far from complete. Even though these are minor concerns, the authors could have made a bigger effort to submit a version that doesn't come across as an early draft.
1) Introduction (page 5): "Pparg expression is down-regulated in the basal subtype of BC, which has squamous features". It is not clear what BC refers to; bladder cancer or basal cells.

Rebuttal
Thank you, reviewers, for your careful reading and comments. Below, we address each point raised by each reviewer: Reviewer 1: 1. The authors specifically conclude in their discussion that loss of PPARG function alone is not sufficient to promote urothelial/squamous cell carcinoma. But would you expect to see actual carcinoma tumors form only 5-7 days after PPARG knockout? Have the authors followed these PPARG knockout mice out to long-term time points and seen no further phenotypic or gene expression changes? If so, this data should be provided in the supplementary information and a comment should be added somewhere in the manuscript to reference this.

Response:
We have included data showing urothelial differentiation in ShhCre;Pparg fl/fl mutants at 2 weeks, 5 months and 1 year after Tamoxifen induction (now included Supplementary Figure 5). We observe abnormal differentiation in the basal compartment, including an expanded K14-Basal population, however we have not observed signs of invasion or tumors in these animals.
2) Similarly, in the UPEC infection tissue injury work, it is stated that the RNA-Seq analyses were performed 24 hrs, 72 hrs, and 4 weeks post-infection. These seem like completely reasonable time points, but a justification should be provided for how they were chosen.

Response:
We have analyzed urothelial regeneration in ShhCre;Pparg fl/fl mutants and controls at 12h, 24h, 72h, 2 weeks, 4 weeks, 6 weeks and 1 year post-infection using the UTI model. We chose the 24h and 72h time-points for comparison because 24h represents the peak of urothelial proliferation post-infection and 72h is the time-point when newly formed Superficial cells are observed in the luminal layer in controls. Superficial cells when formed do not immediately express Krt20; they undergo maturation, which takes about 2 weeks, at which time Krt20 is detectable throughout the luminal layer (which is why we use the 2-week time point). There is a transient increase in the K14-Basal cell population in both controls and mutants during regeneration. Urothelial differentiation returns to normal in controls by but in mutants, K14basal cell expansion continues, and is accompanied by expression of squamous markers, a condition that persists for a year or more. We chose the 1 month time point versus 2 weeks as a representative stage for comparison to maximize the possible differences in gene expression between controls and mutants. In summary, we have not observed signs of tumor formation in these animals after a year or more, either during homeostasis or after UTI. This data is included in Supplementary Figure 5.
3) In the statistical methods section, the authors state that comparisons were made with n=3 or greater animals per genotype. Three mice seems like a very low number. How many comparisons were made with only three mice per genotype? Was it only an outlier analysis that was not feasible to replicate with additional mice, or were most analyses done with only 3 mice per genotype? If only 3 mice per genotype was the majority of comparisons, then please provide some justification for the small sample size and explanation for how such small sample sizes can provide statistical confidence in the data generated from them.
Response: This is a good point. We have actually analyzed a minimum of 3 samples for each experiment, however we generally use more than 3 animals in a cohort. The numbers of animals analyzed are now included in figure legends for each experiment. In most cases, we determine phenotypes by counting cells expressing a given marker on many sections from mutants and controls which gives the analysis considerable power. With the exception of RNA-seq data, where we analyzed 5 or more samples, we did not observe significant variability in the phenotypes examined in this manuscript.

Reviewer 2
It is stated that S cells express the highest level of Pparg. Was this quantified? It appears that other cell populations (I cells?) express Pparg.
Response: During homeostasis, Pparg expression is detected throughout the urothelium, however expression is consistently much higher in S-cells compared to I-cells and Basal cells (Figure 1, Panels A and C). Consistent with this, Fabp4, a direct target of Pparg is present at high levels in S-cells and is expressed at much lower levels in the I-cell and Basal populations (Panel B).
In 3rd paragraph of the results section the authors state "urothelial cells lining the luminal layer were small compared to controls (Figure 1 l,m,n)." Which cell populations specifically are the authors referring to? These images are quite small and difficult to interpret fully.

Response:
We have increased the resolution of the Figure 1h 3. Figure 1R is not discussed explicitly in the text.

Response:
We have now included 1r in the text describing the data in Figure 1  cells and I-cells (Fig. 4e,h). Immunostaining showed basal cell abnormalities in Krt5CreERT2;Pparg fl/fl mutants similar to those observed in ShhCre;Pparg fl/fl mice, including an expanded K14-basal population and up-regulation of Krt10 ( Fig.  4e-j)." For the UP2CreERT2 line, we have added to the text: "To begin to address this, we used the Tamoxifen inducible Upk2CreERT2 line 34 to selectively delete Pparg in S-cells and I-cells, then we analyzed the effects on urothelial homeostasis." 5. Figure 2B is discussed first in the text relative to a. The figure should be re-arranged to accommodate the flow of the text, which is quite nice.

Response:
We have changed the order of 2a and 2b components in Figure 2 as suggested.
6. In the introduction, I thought urothelial cells extend to the proximal urethra, and not just the bladder neck. Is this not the case?
Response: The proximal bladder neck/trigone is a transition zone between the bladder and urethra. There is some overlap between urothelial and urethral markers at this site, however it is not clear how far the physical features of the urothelial barrier (high-resistance tight junctions and apical plaque) extend into the proximal urethra/prostatic urethra in males.
7. PPARG is amplified in a subset of bladder cancers as well, and there are recent studies by Bernard-Pierrot's group (PMID 30604486) that might be worth citing.

Response:
We have replaced the text with a new paragraph describing the potential role for Pparg in bladder cancer subtypes, citing the work from Bernard-Pierrot's group: "Mapping of the mutational landscape of muscle-invasive bladder cancers (MIBC) together with unsupervised clustering analysis of the whole genome expression data revealed that MIBC can be sub-categorized into luminal and basal subtypes. These subtypes are histologically distinct and display discrete sets of mutations and gene expression signatures [14][15][16][17][18][19] . These analyses reveal alterations in PPARG expression and signaling, suggesting that PPARG-dependent transcriptional regulation may be important in the etiology of urothelial carcinoma. Supporting this, PPARG copy number expansion and increased expression of FABP4, a direct PPARG transcriptional target, were observed in luminal-tumors 20-22 . Activating mutations in PPARG and RXRA, a PPARG binding partner, were also observed in luminal MIBC 23,24 . In addition, upregulation of these gene sets important for lipid metabolism and adipogenesis in patients that harbor PPARG gain-of function mutations suggest that PPARG may be an important regulator of lipid metabolism in the luminal subtype of MIBC.
The exact contribution of PPARG to the etiology of the basal subtype of urothelial carcinoma is less clear. PPARG expression is low in basal subtype tumors compared to healthy urothelium, and PPARG is down-regulated in Claudin-low tumors, which have basal-like features. Interestingly, genes encoding cytokines and chemokines are upregulated in Claudin-low basal-like tumors, which may reflect unregulated NF-B signaling due to low levels of PPARG 25 . Expression of PPARG and its binding partner RXRA, are reduced in the SCC-like (SCCL) subtype of MIBC, which shares many features with the basal subtype, including gene expression signatures and common mutations. Transcriptional analysis of those basal or basal-like subtype of tumors revealed a large cluster of genes important for lipid metabolism that were down-regulated compared to the luminal subtype of tumors. In silico Chip-Seq analysis revealed that many of these down-regulated genes contained PPARG binding sites in their regulatory regions, suggesting that these are PPARG-transcriptional targets 26 ." 8. Were markers of apoptosis additionally examined in the Shh-cre experiments?
Response: We performed TUNEL staining and immunostaining for activated caspase 3 at several stages after Tamoxifen induction but we did not detect evidence for apoptosis. 9. It would be interesting to see a heat map showing altered transcription factors in Figure 1.

Response: The heat map showing changes in expression of Superficial cell markers and transcription factors in
ShhCre;Pparg fl/fl mutants is included in Figure 2, which also includes basal cell markers.
10. In the section beginning "Analysis of ShhCre;Pparg fl/fl mutants in which Pparg is deleted… were rapidly inactivated in mutants (Fig 3a,b)." I think the authors are referring to 3E, F. Please double check. Also, at the end of this page, instead of 3I should it be 3L?

Response:
Here is the corrected paragraph: "indicating that both Pparg expression and signaling were decreased in mutants compared to controls (Fig. 3a,b,e,f). This analysis also revealed that Upk3a, which is highly enriched in S-cells, and Krt20, which labels mature S-cells, were both down-regulated (Fig. 3a-h; Supplementary Fig. 3a-f), suggesting that Pparg-dependent transcription is important for proper maintenance of S-cells." In addition, we replaced incorrect labeling of 3i with 3l.
Response: Over-representation analysis (OAR) was performed using gene set analysis with the ConsensusPathDB-mouse database (http://cpdb.molgen.mpg.de/MCPDB). We uploaded lists of genes from RNA-seq analysis of ShhCre;Pparg fl/fl mutant urothelium with p-values of 0.05 or higher, which were 1.5-fold up or down-regulated. For OAR, we used a pvalue of 0.01 for analysis. According to the website: "The gene identifiers are mapped to physical entities in ConsensusPathDB. Over-represented sets are searched among currently three categories of predefined gene sets: network neighborhood-based sets, pathway-based sets and Gene Ontology-based sets." 12. What is the definition of an "immature S cell"? Not completely differentiated?
Response: Despite the lack of expression of Krt20, a marker of mature S-cells, and low levels of Upk family members, which are highly expressed in wild type S-cells, we use the term "Immature S-cells" since: (i) they express ZO1, which is specifically expressed in tight junctions of S-cells, not in I-cells; and because S-cells in mutants lack detectable expression of p63, which marks I-cells.
13. Was there any sex differences in the phenotypes in any of the experiments?

Response:
We did not observe sex differences in our experiments.
14. Are there differences in mitochondrial metabolism in subsets of human bladder cancer included in the TCGA or other cohorts?
Response: Eriksson et al (Eriksson, P., Aine, M., Veerla, S., Liedberg, F., Sjodahl, G., and Hoglund, M. (2015). Molecular subtypes of urothelial carcinoma are defined by specific gene regulatory systems. BMC Med Genomics 8, 25) identified a cluster of genes down-regulated in the LUND cohort, that regulate lipid metabolism in the Basal and SCCL subtypes of urothelial carcinoma. These were not down-regulated in luminal or UroA tumors. This has been included in the introduction: "Expression of PPARG and its binding partner RXRA, are reduced in the SCC-like (SCCL) subtype of MIBC, which shares many features with the basal subtype, including gene expression signatures and common mutations. Transcriptional analysis of those basal or basal-like subtype of tumors revealed a large cluster of genes important for lipid metabolism that were down-regulated compared to the luminal subtype of tumors. In silico Chip-Seq analysis revealed that many of these down-regulated genes contained PPARG binding sites in their regulatory regions, suggesting that these are PPARG-transcriptional targets 26 ." Major concerns 1. It is suggestive that luminal markers such as Krt20 are detected at reduced levels in the Shh-Cre mice experiments ( Figure 1). However, it is not clear if these findings are reflective of decreased Krt20 expression, or the fact that pparg KO is toxic to urothelial cells, thus resulting in cell death (and inability to detect Krt20). Is Krt20 completely undetectable? Some of this might be addressed by subsequent experiments shown in the current submission, but I am not sure.
Response: Krt20 is not detectable in S-cells of ShhCre;Pparg fl/fl mice, we hypothesize because they fail to mature. Supporting this, we do observe expression of other markers including ZO1, which is specifically expressed in S-cells, as well as Gata3 and Foxa1 (please see below). Figure 2 (mRNA expression and IF) are a clear strength, but this data should be quantified. While trends are important and support the overall conclusions, I am curious to know how many of these differences are statistically significant. Also-can the IF be quantified (counting positive/negative cells)?

The orthogonal approaches shown in
Response: p-values for each of the genes shown as up or down-regulated in Figure 2 are now included in Table 1. All have a p-value  0.001, with the exception of Atp5s, Mcee, Ndufa12, Cpt1a and Gata3 which have a p-value  0.05. Pvalues for changes in expression of Krt5, Krt6b, Foxa1 and Ucp, p-values were not statistically significant (0.08, 0.09, 0.18 and 0.4, respectively, however down-regulation of Krt5 and Foxa1, and up-regulation of Krt6b were validated with immunostaining. 3. While it is almost certainly safe to conclude (based on the findings) that Pparg regulates mitochondrial function in a cell autonomous manner, I am not sure that the same can be said about the cell autonomous manner in which pparg regulates cell differentiation or survival. This is true because it is possible that Pparg ablation influences barrier integrity, resulting in urine leaking through the urothelium, and potentially adversely affecting cell differentiation or survival. Therefore, this conclusion should be scaled back a bit. Figure 3 shows expression Gata3, Foxa1 and ZO1 in S-cells from mutants versus controls. The robust expression of Gata3 and Foxa1, and specific expression of ZO1 in S-cells is evident in both mutants and controls, suggesting that S-cells are still viable in the mutants, despite the loss of Krt20 and other urothelial markers. We have not observed differences in permeability between mutants and controls using methylene blue staining, but we have not performed more exhaustive assays of permeability; hence we will tone down the text as follows: The title of the section "Pparg regulates differentiation, mitochondrial functions and survival of S-cells in a cell autonomous manner" is now: "Pparg regulates mitochondrial functions in S-cells."

Response:
4. Figure 4 (Krt5-CreERT2/Pparg KO). It is difficult to judge the extent of squamous differentiation. Indeed, I would argue that the upregulation of a few markers of squamous differentiation in isolated cell populations is not indicative of over squamous differentiation (although it might be indicative of early stages of this process). Is hyperkeratinization and/or presence of intercellular bridges observed?
Response: Point well taken. We have changed the text from : "Analysis of adult Krt5CreERT2;Pparg fl/fl mutants 14 days after Tamoxifen induction revealed that Pparg was selectively down-regulated in Basal cells, and mutants displayed squamous differentiation in a similar pattern as observed in ShhCre;Pparg fl/fl mice including an expanded K14-basal population and up-regulation of squamous markers such as Krt10, that are not present in the healthy urothelium" to "Analysis of adult Krt5CreERT2;Pparg fl/fl mutants 14 days after Tamoxifen induction revealed down-regulation of Pparg in basal cells, while the expression level remained the same in S-cells and I-cells (Fig. 4e,h). Immunostaining showed basal cell abnormalities in Krt5CreERT2;Pparg fl/fl mutants similar to those observed in ShhCre;Pparg fl/fl mice, including an expanded K14-basal population and up-regulation of Krt10 (Fig. 4e-j).

Reviewer #3
Liu and co-workers used different cell type-specific inducible PPARgamma knockout mouse models to study the role of PPAR gamma in urothelium, both under healthy and UTI conditions. Their data suggest that PPARgamma is needed for normal urothelium development and homeostasis, for mitochondrial function, and is a potent regulator of the inflammatory response after UTI.
Major concern: The authors study the effect of PPARgamma deficiency by using the Cre-Lox model. They use 3 different tamoxifeninducible Cre mouse lines to drive PPARgamma depletion in distinct urothelial sub-populations. In each case, they use the PPARg fl/fl as control mice. The authors need to proof that these mice are suitable control mice for these studies. Numerous studies have shown that Cre-recombinase expression by itself can have effects on cell differentiation and other aspects evaluated in this manuscript. Therefore, the PPARg fl/fl mice might be the wrong control mice. Have the authors, or others, studied the 3 cre-expressing lines and do they have proof that none of the observed phenotypes are (partially) due to Cre-expression? transgene, in the ShhCre mice the Cre gene actually replaces the endogenous Shh gene (knock in). In the latter case, since Shh has been shown to play a critical role in bladder development (see PMID 26757905), the authors need to provide proof that having only one functional Shh allele does not affect the parameters evaluated in this study. In short, the authors either have to show data from the 3 Cre-expressing lines as controls throughout their study, or be able to cite previous studies (from others) that show that the 3 Cre-expressing mouse strains behave identical to wildtype mice in regards to the parameters studied. "In an effort to fate map cells that have expressed Shh in the mouse limb, we used gene targeting to insert a gene that encodes a gfpcre fusion protein into the Shh locus (see Experimental Procedures). The gfpcre cassette contained a nuclear localization signal and was inserted at the ATG of Shh. In addition, during the construction of the Shhgfpcre allele, the first 12 amino acids of Shh were removed to create a Shh null allele. ES cells in which the gfpcre cassette was correctly targeted were used to make mice, and these animals were then analyzed. Klf5 flox/flox Shh GfpCre-, Klf5 flox/wt Shh GfpCre-, and Klf5 flox/wt Shh GfpCre+ littermates were used as the controls. Urothelial differentiation in mutants and controls was examined, by immunostaining, RTPCR and microarray analysis. There were no noted differences between control groups. R. Balsara, Warren G. Hill, Xue Li. (2017). Development 144, p400-

408.
Eed fl/+; ShhGC/ + or Ezh2 fl/+; ShhGC/ + mice were used as controls in this paper. The authors analyzed the developing and regenerating urothelium in Eed and Eed/Ezh2 mutants. In situ hybridization revealed robust expression of Shh, Ptched1 and Ptched2 in controls. In a series of experiments, were no evident defects in urothelial development or regeneration in control mice.   Figure 1 shows Characterization comparing wild type urothelium with urothelium from Krt5CreERT2;mTmG and Up2CreERT2;mTmG lines (the mTmG is a Rosa26 reporter) that were used in lineage analysis studies of carcinogen induced bladder cancer. There were no differences observed between controls.

In vivo replacement of damaged bladder urothelium by
Other concerns: 1) The role of PPARgamma in bladder cancer is very controversial, with numerous studies both claiming an oncogenic or anti-cancerous role for PPARgamma. This is not evident from the way the manuscript is currently written. The short paragraphs in the introduction, results, and discussion that touch upon this subject are vague (using terms like "associated with" and "regulate") and cite few recent publications. The authors should develop this further and more clearly (in a neutral manner), and include recent studies (examples PMIDs 30651555, 30912275.

Response:
We have included the following paragraph in the introduction: "Mapping of the mutational landscape of muscle-invasive bladder cancers (MIBC) together with unsupervised clustering analysis of the whole genome expression data revealed that MIBC can be sub-categorized into luminal and basal subtypes. These subtypes are histologically distinct and display discrete sets of mutations and gene expression signatures [14][15][16][17][18][19] . These analyses reveal alterations in PPARG expression and signaling, suggesting that PPARG-dependent transcriptional regulation may be important in the etiology of urothelial carcinoma. Supporting this, PPARG copy number expansion and increased expression of FABP4, a direct PPARG transcriptional target, were observed in luminal-tumors 20-22 . Activating mutations in PPARG and RXRA, a PPARG binding partner, were also observed in luminal MIBC 23,24 . In addition, upregulation of these gene sets important for lipid metabolism and adipogenesis in patients that harbor PPARG gain-of function mutations suggest that PPARG may be an important regulator of lipid metabolism in the luminal subtype of MIBC.
The exact contribution of PPARG to the etiology of the basal subtype of urothelial carcinoma is less clear. PPARG expression is low in basal subtype tumors compared to healthy urothelium, and PPARG is down-regulated in Claudin-low tumors, which have basal-like features. Interestingly, genes encoding cytokines and chemokines are upregulated in Claudin-low basal-like tumors, which may reflect unregulated NF-B signaling due to low levels of PPARG 25 . Expression of PPARG and its binding partner RXRA, are reduced in the SCC-like (SCCL) subtype of MIBC, which shares many features with the basal subtype, including gene expression signatures and common mutations. Transcriptional analysis of those basal or basal-like subtype of tumors revealed a large cluster of genes important for lipid metabolism that were down-regulated compared to the luminal subtype of tumors. In silico Chip-Seq analysis revealed that many of these down-regulated genes contained PPARG binding sites in their regulatory regions, suggesting that these are PPARG-transcriptional targets 26 ." We have included the following paragraph in the discussion: "Positive and negative Pparg signaling can have profound effects on bladder cancer cells 56 and on immune functions in MIBC 57 . Pparg expression is down-regulated in basal/squamous subtypes of MIBC compared to the healthy urothelium, and up-regulated in the luminal subtype of MIBC. We observed a number of abnormalities in Pparg mutants that are similar to those observed in bladder cancers with low Pparg expression, in particular the basal/squamous subtype of MIBC and in Claudin-low (basal-like) subtype tumors. Similarities include up-regulated expression of markers of basal differentiation (Krt14, Krt6, Krt5) and up-regulation of NF-B and expression of inflammatory molecules 25,58 . We also observed alterations in the urothelium that are associated with tumor formation, including up-regulation of Snail1, Slug and Vimentin, and a compromised basement membrane (Fig. 5). Despite these similarities, we did not observe invasion or signs of tumor formation in ShhCre;Pparg fl/fl mutants, suggesting that Pparg mutations are unlikely to be primary drivers of tumor formation. We believe future studies should focus on whether Pparg mutations promote tumorigenesis in cooperation with other mutations, or whether Pparg acts down-stream regulating cellular differentiation or immune functions after tumor initiation." 2) In the results (page 10), it is stated that "Upk3, which labels I-cells and S-cells, and Krt20 which labels mature S-cells were both down-regulated (Fig. 3a-i)". However, Fig. 3i shows an increase in S cells. Please explain.

Response:
The problem here is that we have included numbers of immature S-cells in the analysis of S-cell numbers, which is not informative, since the main point is that luminal cells are present in mutants but have down-regulated S-cell markers. We have eliminated the top panel graph in Fig. 3i, which we agree, is confusing, along with the associated text, which has been changed to: "This analysis also revealed that Upk3a, which is highly enriched in S-cells, and Krt20, which labels mature S-cells, were both down-regulated ( Fig. 3a-h; Supplementary Fig. 3a-f), suggesting that Pparg-dependent transcription is important for proper maintenance of S-cells. Unexpectedly, we observed Ki67 expression in 20% of Scells and 10% of I-cells in Upk2CreERT2;Pparg fl/fl mutants during homeostasis (Fig. 3c,g,i)." 3) PPARgamma can regulate NFkappaB through different mechanisms, including by sequestering the P65 subunit, but also by regulating I-kappa-B-alpha and IKK-alpha/beta levels. It is not clear what is stained as NFkappaB in Fig. 5 and Supplemental Fig. 4 (no description in Methods section). The whole part on NFkappaB should be developed in more detail, including mechanistic details.

Response:
We have included the following paragraph in the discussion section: Pparg is known to be an important regulator of inflammatory response, in part by regulating transcriptional activity of NF-B, which among other things, controls the innate immune response to UPEC infection. p65/Rela, one of five NF-B family members, is transiently up-regulated in the wild type urothelium in response to UPEC infection, but persists in ShhCre;Pparg fl/fl mutants for months accompanied by edema and leukocyte infiltration (Fig 5), suggesting that mutants fail to resolve the NF-B response. NF-B is a transcription factor that regulates expression of immune genes, and also plays an important role in epithelial barriers such as the skin, gut and esophagus, controlling recognition and response to invading pathogens 47-49 . The NF-B family has 5 members, including p65/Rela, which contains a transactivation domain that can bind to DNA and positively regulate transcription. NF-B signaling is mediated by heterodimers in the urothelium (most likely including p65/Rela) that are sequestered in the cytoplasm in an inactive state in resting cells, where they are bound to the inhibitor Ib. Stimulus induces degradation of Ib and release of NF-B-heterodimers, which move to the nucleus and bind to response elements in target genes, activating their transcription 50-52 . NF-B signaling, as evidenced by p65 expression, is activated rapidly in the wild type urothelium in response to UPEC infection. Recent studies suggest that its initial activation may be triggered by binding of the bacterial adhesin, Fimh, to Upk1a, which is expressed on the surface of S-cells. This interaction triggers TLR4 mediated pattern recognition 53 . Pparg is known to be an important regulator of NF-B-transcriptional activity, acting as a transrepressor. Pparg can bind directly to the NF-B protein, preventing its interaction with promoter regions of target genes. SUMOylation of the Pparg ligand binding domain enables Pparg to bind to regulatory sequences of NF-B target genes, preventing disassociation of co-repressors and resulting in repression of NF-B-dependent transcription 54,55 . While the direct mechanism by which Pparg controls NF-B in the urothelium is unclear, our studies provide strong evidence that Pparg regulates p65/Rela expression and is required in urothelial cells to suppress the innate immune response induced by UPEC infection. Fig. 3 is never mentioned in the text.

4) Supplement
Response: Please see response to point 2, including a reference to Supplementary Figure 3 which has been added to the text.
The manuscript has been written in a very sloppy manner and includes numerous errors of which a non-exhaustive list follows below. Especially the Figure legends contained many errors and the Methods section is far from complete. Even though these are minor concerns, the authors could have made a bigger effort to submit a version that doesn't come across as an early draft. 1) Introduction (page 5): "Pparg expression is down-regulated in the basal subtype of BC, which has squamous features". It is not clear what BC refers to; bladder cancer or basal cells.
Response: This is a typo; basal cells are detectable at E16. Sentence now reads: "Pparg expression is undetectable in the Basal cell compartment until E16 (Fig. 1g, green arrow)" 3)Page 9: "However, the mitochondrial membrane is impermeable to fatty acids and a specialized carnitine carrier system consisting of Cpt1, Slc25a20 and Cpt2 controls fatty acid transport (Spinelli and Haigis, 2018). We observed downregulation of all 3 genes in ShhCre;Ppargfl/fl mutants (Fig. 2b,e). In addition, 15 genes that encode members of complex 1 NADH ubiquinone oxidoreductase were down-regulated, including Nd4, Nd5 and Nd6 that are transcribed from mitochondrial DNA (Fig, 2b; Cpt2 immunostaining is shown in Fig. 2e)" It would make more sense to write: "However, the mitochondrial membrane is impermeable to fatty acids and a specialized carnitine carrier system consisting of Cpt1, Slc25a20 and Cpt2 controls fatty acid transport (Spinelli and Haigis, 2018). We observed down-regulation of all 3 genes in ShhCre;Ppargfl/fl mutants (Fig. 2b, Cpt2 immunostaining is shown in Fig. 2e). In addition, 15 genes that encode members of complex 1 NADH ubiquinone oxidoreductase were down-regulated, including Nd4, Nd5 and Nd6 that are transcribed from mitochondrial DNA (Fig, 2b)"

Response:
The original paragraph has been replace with the suggested paragraph.