Foxm1 Transcription Factor is Required for Lung Fibrosis and Epithelial to Mesenchymal Transition.

Transcription

1 Manuscript EMBO Foxm1 Transcription Factor is Required for Lung Fibrosis and Epithelial to Mesenchymal Transition. David Balli, Vladimir Ustiyan, Yufang Zhang, I-Ching Wang, Alex J. Masino, Jeffrey A. Whitsett, Vladimir V. Kalinichenko and Tanya V. Kalin Corresponding author: Tanya Kalin, Cincinnati Children's Hospital Medical Center Review timeline: Submission date: 16 July 2012 Editorial Decision: 24 August 2012 Revision received: 07 November 2012 Editorial Decision: 23 November 2012 Accepted: 27 November 2012 Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) 1st Editorial Decision 24 August 2012 Thank you for submitting your research paper on the role of Foxm1 in fibrosis/emt for consideration to The EMBO Journal editorial office. I did in the meantime receive comments from two scientists that reveal a potential interest in your results, but at the same time outline necessary further experimentation before we might be able to offer publication. Major concerns arise from the proposed function of FoxM1 as driver of an EMT in-vivo predominantly in epithelial cells. While ref#1 finds these results still mostly correlative, ref#2 remains unconvinced that the current experimental data support such a conclusion. Recognizing the potential, respective clinical relevance of your results that aslso appear to address current controversies in the field, we would like to offer you the possibility to respond to these critiques by further experimental support that would have to go beyond the enclosed comments. Please do note that The EMBO Journal considers only one round of major revisions with the ultimate decision solely depending on the quality and strength of the final dataset that will have to be reassessed by the original referees. Please do not hesitate to get in touch in case of further questions or with an outline of timeline and feasibility of planned experimental extensions (preferably via ). I am very much looking forward to a revised study and remain with best regards. EMBO 1

2 Yours sincerely, Editor The EMBO Journal REFEREE REPORTS: Referee #1: The authors in this manuscript describe a previously unrecognized function for the transcription factor FOXM1 in promotion of fibrogenesis and implicate activate of EMT as a mechanism. Prior data had implicated FOXM1 in EMT in a cell line but no mechanism for this had been defined and no in vivo data have been reported. The experiments are generally well done and the data both remarkably clear and compelling that EMT is promoted in vitro and in vivo by FOXM1 and that FOXM1 is required for radiation-induced fibrosis. Moreover the authors now define an additional mechanism for activation of the EMT program by showing direct binding of FOXM1 to the Snail reporter. It is a bit less clear that the induction of EMT markers in vivo is mainly in epithelial cells, and in type II cells in particular, but clearly mesenchymal expansion is due to epithelial FOXM1. While these data provide strong supportive evidence for the causal role of activation of a FOXM1- induced EMT program in fibrosis the EMT data are ultimately correlative. But overall these are innovative and important in vivo findings for the field of fibrosis. The implication of FOXM1 signaling in human pulmonary fibrosis is also new. Hence I think this work is very suitable for EMBO. Minor Criticisms Figs 7,8. The relationship between induction of Snail promoter activity in Fig 8 and TGFb1 regulated Snail protein in Fig 7 seems confusing. Clearly expression of FOXM1 in A549 cells is not sufficient to induce Snail protein whereas in the different cell line used for reporter activity there is clear induction of Snail by simple expression of FOXM1, implying the two systems are different. Is induction of Snail promoter activity regulated by FOXM1 regulated by TGFb1? If not, it would seem more logical to measure reporter activity in A549 cells, analogous to Smad reporter measurements, and assess the influence of TGFb1 on FOXM1-induced Snail promoter activity as well as protein. Referee #2: This manuscript describes the function of Foxm1 in a lung radiation injury model. Foxm1 expression is induced in both lung mesenchymal and epithelial cells after ionizing radiation. The authors show that expression of a constitutively active form of Foxm1 in lung epithelium worsens the response of the lung to ionizing radiation where as loss of Foxm1 in lung epithelium helps to protect the epithelium from damage caused by radiation. The authors implicate EMT as a specific process that leads to increased fibrosis after radiation injury. Foxm1 can regulate several EMT related genes including Snail and Zeb. The data up to this point are convincing and well presented. However, the authors try to show using lineage tracing that lung epithelial cells expressing a constitutively active form of Foxm1 generate vimentin positive fibroblasts. These data consist of pictures of single cells that may express both Sp-C and vimentin but are not convincing (there is a graph indicating that more cells are double positive). Even if this does occur the finding of such few cells is unlikely to explain the dramatic increase in fibrosis. Thus, while it appears clear that Foxm1 regulates this injury repair process it is not clear that the mechanism is through EMT. Specific points 1) Foxm1 is known to regulate cell proliferation. How much of the phenotype observed in this model is due to alterations in proliferation? Could increases/decreases in proliferation account for EMBO 2

3 the majority of the phenotype? 2) Some of the co-localization studies are not clear i.e. Fig. 5C and 9C and E. Better confocal images would be important to include. 3) The graph showing percentages of SMA+/lacZ+ cells in Figure 9 does not appear to correlate with the co-localization of these 2 markers by immunostaining. Might be better to perform these studies using FACS analysis. 4) I'm not sure I see the relevance of the immunology data in Figure 4. Any injury will lead to an immune response and sense the data focus on gain or loss of Foxm1 in the epithelium such a response is clearly secondary to the primary insult. Might be worth moving to supplemental data section. 1st Revision - authors' response 07 November 2012 Referee #1: The authors in this manuscript describe a previously unrecognized function for the transcription factor Foxm1 in promotion of fibrogenesis and implicate activate of EMT as a mechanism. Prior data had implicated Foxm1 in EMT in a cell line but no mechanism for this had been defined and no in vivo data have been reported. The experiments are generally well done and the data both remarkably clear and compelling that EMT is promoted in vitro and in vivo by Foxm1 and that Foxm1 is required for radiationinduced fibrosis. Moreover the authors now define an additional mechanism for activation of the EMT program by showing direct binding of Foxm1 to the Snail reporter. It is a bit less clear that the induction of EMT markers in vivo is mainly in epithelial cells, and in type II cells in particular, but clearly mesenchymal expansion is due to epithelial Foxm1. While these data provide strong supportive evidence for the causal role of activation of a Foxm1-induced EMT program in fibrosis the EMT data are ultimately correlative. But overall these are innovative and important in vivo findings for the field of fibrosis. The implication of Foxm1 signalling in human pulmonary fibrosis is also new. Hence I think this work is very suitable for EMBO. Minor Criticisms Figs 7,8. The relationship between induction of Snail promoter activity in Fig 8 and TGFb1 regulated Snail protein in Fig 7 seems confusing. Clearly expression of Foxm1 in A549 cells is not sufficient to induce Snail protein whereas in the different cell line used for reporter activity there is clear induction of Snail by simple expression of Foxm1, implying the two systems are different. Is induction of Snail promoter activity regulated by Foxm1 regulated by TGFb1? If not, it would seem more logical to measure reporter activity in A549 cells, analogous to Smad reporter measurements, and assess the influence of TGFb1 on Foxm1-induced Snail promoter activity as well as protein. We agree and would like to thank the Reviewer for this suggestion. To determine if TGF-β signalling is important for Foxm1 mediated regulation of the Snail1 promoter, we performed additional experiments in A549 cells using CMV-Foxm1, Snail1-Luc reporter and TGF-β. In the absence of TGF- β, CMV-Foxm1 induced Luc activity driven by the -720bp mouse Snail1 promoter region (new Fig. 7C). Targeting mutagenesis of the Foxm1-binding site significantly reduced transcriptional activity of the Snail1 promoter in A549 cells (Fig. 7C) and U2OS cells (Suppl. Fig. 5). Thus, Foxm1 alone was capable of inducing transcriptional activity of the -720bp Snail1 promoter region in co-transfection experiments. Interestingly, TGF-β did not synergize with Foxm1 to enhance Snail1 promoter activity (Fig. 7C). Published studies have demonstrated that AP-1/4 binding sites (located between -900 and bp upstream of the ATG start site) are required for TGF-beta-mediated induction of Snail1 (Peinado et al, 2003; Medici et al, 2006). Consistent with these studies, the -720 bp Snail1-Luc construct was incapable of responding to TGF-β stimulation (Fig. 7C). While our co-transfection experiments are important (1) to establish that Foxm1 directly stimulates Snail1 promoter and (2) to identify a functional Foxm1 binding site, we believe that the -720bp mouse Snail1-Luc construct is not suitable to examine transcriptional synergy between Foxm1 and TGF-β due to the absence of TGFb-responsive regions. EMBO 3

4 In the context of endogenous Snail1 promoter, TGF-β increased Foxm1 binding to the Snail1 promoter DNA as demonstrated by ChIP assay (Fig. 7B), suggesting that there is a crosstalk between TGF- β signalling and Foxm1 in the regulation of endogenous Snail1 promoter. These results are consistent with our western blot data demonstrating that sirna-mediated depletion of Foxm1 from TGF-β-treated A549 cells prevented up regulation of the Snail1 protein (Fig. 6B). Finally, we would like to point out that the Foxm1 protein is present in many epithelial cell lines, including A549 cells (Fig. 6B, (Kalin et al, 2011). Despite high levels of Foxm1 in A549 cells, they do not express Snail1 protein or undergo EMT in the absence of TGF-β (Fig.6B, line 1). These data suggest that the endogenous Snail1 promoter is repressed at basal conditions and the addition of TGF-β alleviates this repression. Consistent with this hypothesis, MTA2/ Mi-2/ NuRD repressor complex was bound to the Snail1 promoter in epithelial cells and the activation of TGF-β pathway eliminated this repression mechanism, stimulating Snail1 transcription (Dhasarathy et al, 2007). Since this repression complex has two binding sites within the first 500 bp of the Snail1 promoter (Fujita et al, 2003), it is possible that TGFmediated removal of this repression mechanism can promote the binding of Foxm1 to the Snail1 promoter DNA as seen by ChIP (Fig. 7B). We have modified the Discussion section to provide this information (page 9). Referee #2: This manuscript describes the function of Foxm1 in a lung radiation injury model. Foxm1 expression is induced in both lung mesenchymal and epithelial cells after ionizing radiation. The authors show that expression of a constitutively active form of Foxm1 in lung epithelium worsens the response of the lung to ionizing radiation where as loss of Foxm1 in lung epithelium helps to protect the epithelium from damage caused by radiation. The authors implicate EMT as a specific process that leads to increased fibrosis after radiation injury. Foxm1 can regulate several EMT related genes including Snail and Zeb. The data up to this point are convincing and well presented. However, the authors try to show using lineage tracing that lung epithelial cells expressing a constitutively active form of Foxm1 generate vimentin positive fibroblasts. These data consist of pictures of single cells that may express both Sp-C and vimentin but are not convincing (there is a graph indicating that more cells are double positive). Even if this does occur the finding of such few cells is unlikely to explain the dramatic increase in fibrosis. Thus, while it appears clear that Foxm1 regulates this injury repair process it is not clear that the mechanism is through EMT. Based on the Reviewer s critique, we provided additional data to demonstrate that Foxm1 induces EMT in vivo (see specific points below). In addition, we also found that the number of proliferating fibroblasts was increased in Foxm1 over-expressing mice. Therefore, we believe that both fibroblast proliferation and EMT contribute to aberrant fibrosis in Foxm1-overexpressing transgenic mice. We modified our manuscript to clarify this point (page 6, Fig. 3D, Supplemental Fig. 4A-B). Specific points 1) Foxm1 is known to regulate cell proliferation. How much of the phenotype observed in this model is due to alterations in proliferation? Could increases/decreases in proliferation account for the majority of the phenotype? We agree. To address the question of proliferation in the Foxm1 over-expressing mice, we performed additional immunostaining using proliferation-specific marker Ki-67. Six months after irradiation, the number of Ki-67-positive cells was increased in the lungs of epifoxm1- N mice (new Fig. 3D). In nonirradiated lungs, proliferation was unaltered. Since Foxm1 was over-expressed in type II lung epithelial cell (SPC-positive cells) and is known to regulate proliferation, we performed additional co-localization studies to identify the number of type II cells undergoing proliferation. Six months after irradiation, the number of proliferating type II cells (double positive for SPC and Ki-67) were similar in epifoxm1- N and control mice (new Supplemental Fig. 4A-B). In addition, we isolated type II epithelial cells from bleomycin-treated control lungs and Foxm1-DN transgene(gfp+)-positive type II cells from epifoxm1- N lungs. We used mrna from these cells for qrt-pcr analysis. No changes in cyclin B1 and cyclin D1 mrnas were observed (new Supplemental Figure 4C). Thus, Foxm1 over-expression in type II cells does not influence their proliferation in fibrosis model. EMBO 4

5 In contrast, the number of cells double positive for Ki-67 and mesenchymal marker asma was increased in epifoxm1- N mice (new Supplemental Figure 4A-B). These data suggest that over-expression of Foxm1 in type II cells indirectly influenced proliferation of fibroblasts, possibly contributing to the fibrotic phenotype in epifoxm1- N mice. Our studies demonstrated the increased levels of IL-1b, Ccl2 and Cxcl5 in irradiated epifoxm1- N lungs. These proinflammatory mediators induced proliferation of fibroblasts in vitro and in vivo (Ekert et al, 2011; Kawamura et al, 2012; Moore et al, 2001; Quan et al, 2006), suggesting that they can contribute to increased fibrogenesis in epifoxm1- N mice. We modified our manuscript to incorporate these data (page 7). 2) Some of the co-localization studies are not clear i.e. Fig. 5C and 9C and E. Better confocal images would be important to include. We agree. We performed additional co-localization experiments to address this comment. Images in Fig. 4C and 8A and F have been replaced with better quality confocal images. We have also added additional higher magnification images demonstrating that Foxm1- N transgene co-localized with a-smooth muscle actin (asma) in lungs of irradiated epifoxm1- N mice (Figure 4C, upper panels). 3) The graph showing percentages of SMA+/lacZ+ cells in Figure 9 does not appear to correlate with the co-localization of these 2 markers by immunostaining. Might be better to perform these studies using FACS analysis. We would like to thank the reviewer for this suggestion. To address the comment, we performed a flow cytometry analysis to count the number of epithelial cell that co-expressed mesenchymal marker asma and epithelial marker prospc in fibrotic lungs. To induce lung fibrosis we utilized a well-established bleomycin-induced model of pulmonary fibrosis. We already found that Foxm1 promotes lung fibrosis in both irradiation model and bleomycin models (Supplemental Fig. 7). Single cell suspensions were permeabilized and used for intracellular staining with prospc and αsma antibodies. After bleomycine treatment, 5.46 ±0.71% of cells in control mice expressed both markers, suggesting that a population of type II epithelial cells were undergoing EMT (Figure 8C-D). The number of prospc+/αsma+ cells in Foxm1-over-expressing mice was increased to 8.31 ±1.03% (P < 0.05) (Figure 8C-D). Thus, overexpression of Foxm1 increased the number of type II cells undergoing EMT, findings consistent with our co-localization data in radiation-induced (Fig.8B and Supplemental Fig. 6E) and bleomycin-induced (Supplemental Fig. 7D) models of pulmonary fibrosis. Moreover, we performed additional experiments to clearly demonstrate that Foxm1 induces expression of mesenchymal genes in purified population of type II cells. We have isolated a pure population of type II cells from bleomycin-treated control and epifoxm1- N lungs (see modified Methods section, page 22, and (Rice, 2002)). Since Foxm1-DN transgene has a GFP tag, we used cell sorting to isolate Foxm1DNexpressing type II cells [epifoxm1-dn (GFP+) cells]. EMT-associated gene expression was compared in GFP-positive epifoxm1-dn cells and GFP-negative control type II cells. Over-expression of Foxm1 in type II cells dramatically increased mrna levels of Vimentin and Snail1, and decreased E-cadherin mrna (Figure 9E). These data further support our conclusion that over-expression of Foxm1 promotes EMT in a subset of type II cells during progression of pulmonary fibrosis in vivo. We incorporated these data into revised manuscript. 4) I'm not sure I see the relevance of the immunology data in Figure 4. Any injury will lead to an immune response and sense the data focus on gain or loss of Foxm1 in the epithelium such a response is clearly secondary to the primary insult. Might be worth moving to supplemental data section. We agree with the reviewer. As the Reviewer suggested, we placed these data into Supplemental section (Supplemental Figure 3). 2nd Editorial Decision 23 November 2012 I just received final remarks from one of the original referees that is fully satisfied with the revisions provided and thus supports publication in The EMBO Journal. Before formal acceptance, I would be grateful for the provision of original source data, particularly electrophoretic blots in figure 6 and 7. This is an accord with our policy to make uncropped/unprocessed original results accessible for the community and thus increase reliability of EMBO 5

6 published results. We would welcome one PDF-file per figure that combines this information. These will be linked online as supplementary "Source Data" files. Please allow me to congratulate you to this study at this point. The editorial office will soon be in touch with an official acceptance letter and necessary post-acceptance paperwork. Yours sincerely, Editor The EMBO Journal EMBO 6