Foxm1 is a critical driver of TGF‐β‐induced EndMT in endothelial cells through Smad2/3 and binds to the Snail promoter

Abstract Endothelial‐to‐mesenchymal transition (EndMT) was first reported in heart development. Recent studies have shown that EndMT also occurs in the progression of cardiac fibrosis. Herein, we demonstrated a critical role of the Forkhead Box M1 (Foxm1) transcription factor in transforming growth factor beta (TGF‐β)‐induced EndMT in endothelial cells (ECs) and a possible underlying molecular mechanism. Foxm1 was induced in ECs following TGF‐β stimulation. Using both pharmacological and molecular approaches to inhibit Foxm1 function can attenuate the TGF‐β‐induced EndMT and cell migration. In contrast, lentivirus‐mediated overexpression of Foxm1 allowed EndMT to proceed despite the absence of TGF‐β in ECs. Moreover, we found that the activation of the Smad2/3 signaling pathway and EndMT‐related transcription factors played important roles in the pathogenesis of Foxm1‐mediated EndMT. Further analysis revealed that Foxm1 bound to and increased the promoter activity of the Snail gene encoding a critical transcriptional regulator of EndMT. In conclusion, our results identify FOXM1 as a driver of TGF‐β‐induced EndMT and underscore the therapeutic potential of targeting FOXM1 for cardiac fibrosis.

Forkhead box M1 (Foxm1) is a transcription factor best-recognized as a master regulator of physiological and pathological processes, including cancer (Aytes et al., 2014), diabetes mellitus (Shirakawa et al., 2017), and fibrosis-related disease (Penke et al., 2018). Among its isoforms, Foxm1b (hereafter designated simply as FOXM1) has been studied most extensively and is considered to be able to activate the expression of multiple target genes critical for normal cell proliferation, survival, and self-renewal as well as cancer initiation, progression, and drug resistance (Koo, Muir, & Lam, 2012;Wang et al., 2018). A growing body of recent evidence has emphasized the potential roles of Foxm1 in organ fibrosis, such as pulmonary fibrosis (Balli et al., 2013) and CF (Sato et al., 2017). However, the underlying molecular mechanisms of Foxm1 in cardiac fibrosis are not clearly elucidated, especially the role of Foxm1 in EndMT, which remains unexplored. Here, we investigated the role of Foxm1 in TGF-β-induced EndMT in ECs and clarified a possible molecular mechanism. The data presented in this report provide new insights into the regulation of FOXM1 in EndMT-induced CF and provides important experimental evidence supporting Foxm1 as a potential therapeutic target for cardiac fibrosis.

| Cell culture
Primary human umbilical vein endothelial cells (HUVECs) were obtained from human umbilical cord veins and maintained in a humidified atmosphere at 37°C in 5% CO 2 using the EGM-2 bullet kit (Lonza, Basel, Switzerland). Cells between passages two and six were plated on vitronectin-coated dishes, and used in each experiment.

| Quantitative real-time PCR
Total RNA was extracted from cultured cells using a TRIzol reagent (Invitrogen). Reverse transcription was performed using a first strand complementary DNA (cDNA) reverse transcription kit, and real-time polymerase chain reaction (RT-PCR) was performed using a SYBR Green/ROX qPCR Master Mix kit according to the manufacturer's instructions (Takara, Otsu, Japan). The relative quantification was determined using the ΔΔCt method with GAPDH as a reference gene.
All primer sequences are shown in Supporting Information Table 1.
2.5 | Knocking down of Foxm1 by lentiviral short hairpin RNA (shRNA) The following Foxm1-targeted short hairpin RNAs (shRNAs) were designed and synthesized by Genechem Co. Ltd (Shanghai, China). The recombinant lentivirus of small interfering RNA (siRNA) targeting Foxm1 (sh-Foxm1) and control lentivirus (sh-NC) were commercially prepared. Briefly, a lentivirus transfer vector (GV118) was con-  The highest dilution producing drug selected colonies were used to transduce primary HUVECs in the presence of polybrene (4 μg/ml, Sigma), and 0.5 μg/ml puromycin was introduced 48 hr postinfection.
Cells were seeded in 6-well plates and the next day media was changed with or without doxycycline (Sigma) to induce transgene expression. Media with or without doxycycline was changed every 24 hr. After 48 hr, cells were prepared for total RNA and protein extractions.

| Chromatin immunoprecipitation assays
After HUVECs had been treated with TGF-β1 (10 ng/ml) or vehicle for 48 hr, protein-DNA complexes were cross-linked using 1% formaldehyde (Thermo Scientific) for 10 min at room temperature, and the reaction was quenched with 125 mM glycine. Cells were washed with PBS and lysed by incubation in immunoprecipitation (IP)-buffer. Chromatin was sheared by sonication (Bioruptor, Diagenode) to obtain an average size of 500-1,000 bps. Protein-DNA complexes were immunoprecipitated overnight using antibodies selective for rabbit polyclonal Foxm1 (CST, #20459), rabbit IgG (Santa Cruz, sc-2072) served as negative controls. The immune complexes were adsorbed with protein A agarose beads (Invitrogen), Immunoprecipitates were washed, eluted, and crosslinks were reversed overnight. The next day, samples were clarified by phenol: chloroform:isoamyl alcohol extraction. IP and non-IP DNA (input) were analyzed by real-time PCR. As a positive control for IP-analysis, primers directed against rat GAPDH promoter were used. PCR are using the following primers: 5′-TCTTACCCCGGGCCTTTCCCCTC G-3′ and 5′-CCGCTCGAGTGGCCAGAGCGACCTAG-3′. Enrichment of specific promoter regions after IP was calculated as fold induction over IgG.

| Migration assay
Cell migration was assessed using transwell chambers (8 μm pore size; Corning, Corning, New York). The samples containing 1 x 10 5 cells were resuspended in EGM-2 serum-free medium and loaded into the upper chamber. The chambers were incubated for 24 hr with complete culture medium added in the lower chamber. Nonmobile cells were removed, and the chambers were stained with crystal violet. Five randomly selected fields were counted under an inverted light microscope.

| Statistical analysis
Data were presented as means ± SEM. All experiments were repeated at least three times. Statistical analysis was performed with SPSS software (v19.0, Chicago, Illinois). Comparisons among groups were made using one-way analysis of variance or the Student's t test. The p values < 0.05 were considered significant.

| TGF-β induces EndMT in cultured HUVECs
We first evaluated the effect of TGF-β1 on EndMT in cultured primary HUVECs. Exposure of HUVECs to TGF-β1 for 48 hr caused an obvious alteration in cellular morphology from a polygonal, cobblestone-like shape to a more spindle-like, fibroblast shape this end, we treated the HUVECs with Sio A and found that Sio A alone had no effect on EndMT. However, the Sio A was sufficient to suppress TGF-β1-induced EndMT. As shown in Figures 2c and 3c, the expression of fibroblast markers vimentin, α-SMA, and FSP1 was significantly reduced compared with levels in the TGF-β1 group.
Meanwhile, inhibition of Foxm1 caused a significant increase in EC markers, VE-cadherin, and CD31 in cells treated with TGF-β1.

| Knockdown of Foxm1 with shRNA inhibits TGF-β1-induced EndMT and HUVEC cell migration
To further investigate the effect of Foxm1 on TGF-β1-induced EndMT, shRNA-mediated transfection was used for 24 hr to downregulate Foxm1 followed by treatment with TGF-β1 for 48 hr. We transfected HUVECs with three Foxm1-specific shRNAs and confirmed gene knockdown efficiency by the qRT-PCR 48 hr posttransfection. We found that sh-Foxm1(3) has the most significant knockdown efficiency, which reaches 70% at the mRNA level (Figure 4a). This knockdown efficiency has also been verified at the protein level ( Figure 4b).
Therefore, the following experiments were performed using sh-Foxm1 (3). Compared with sh-NC, Foxm1 shRNA significantly attenuated the TGF-β1-induced EndMT in HUVECs. As shown in Figure 4c

| Overexpression of Foxm1 promotes EndMT in ECs
To determine whether FOXM1 is sufficient to drive EndMT in | 9059 implicated in EndMT (Sabbineni et al., 2018). Next, we wanted to examine whether Foxm1 regulated the EndMT process through these three transcription factors. We found that silencing Foxm1 can reverse increases in Snail, Slug, and Twist mRNA expression induced by TGF-β1 stimulation (Figure 7a). This effect was also verified by western blot analysis (Figure 7b). Canonical Smad signaling appears to be involved in TGF-β1-induced EndMT. We also found that TGF-β1 factor. Our results confirmed that TGF-β-exposed ECs could undergo EndMT and that Foxm1 was significantly upregulated in multiple ECs after the TGF-β treatment. Using both molecular and pharmacological approaches to interrogate the functional roles of Foxm1, we found that Foxm1 inhibition prevented TGF-β-induced EndMT as well as cell migration. In contrast, lentivirus-mediated overexpression of Foxm1 allowed EndMT to proceed despite the absence of TGF-β in ECs. We have suggested that Foxm1 is indeed involved in the EndMT process by using both knockdown and overexpression approaches.
We also found that Foxm1 regulates the primary mechanism of EndMT via the Smad2/3 pathway induced by TGF-β and direct binding to the promoter region of the Snail gene. These results suggest that Foxm1 may be an important mediator of EndMTassociated CF.
EndMT plays an important role in the pathogenesis of CF by associating with the emergence of fibroblasts of an EC origin (Zeisberg et al., 2007). During EndMT, several molecular and structural rearrangements take place leading to the cellular changes necessary to switch to a mesenchymal phenotype. EndMT results in ECs without cell-cell adhesion, with high migratory potential and with expression of specific mesenchymal cell markers, such as vimentin, α-SMA, and FSP1 (Cooley et al., 2014). Concurrently, ECs undergoing EndMT lose the expression of characteristic surface EC markers, such as VE-cadherin and CD31 (Figure 1). Zeisberg et al. demonstrated that TGFβ was implicated in a signaling pathway that stimulated EndMT in cardiac injury. TGF-β induced ECs to undergo EndMT, which provided evidence of the EndMT role in aortic bandedinduced cardiac fibrosis (Zeisberg et al., 2007). In addition, Sabbineni et al. demonstrated that TGF-β2 was the most potent inducer of EndMT (Sabbineni et al., 2018). Our results also demonstrate that in the human adult heart and animal disease models (Xu et al., 2015a). Development of technological innovations to better visualize the EndMT is a key for therapeutic targeting of EndMT in human CF.
EndMT is related to the more widely known mechanism of EMT.
There is strong evidence that Foxm1 is a key regulator of EMT, and it is well recognized as a driver of transcriptional activation of EMTregulator expression as well as the expression of typical mesenchymal markers (Huang et al., 2012). To our knowledge, our data are the first to establish the importance of Foxm1 in EndMT-mediated fibrosis. The antifibrotic function of Foxm1 has been reported in cardiac hypertrophy and fibrosis, diabetes, and pulmonary fibrosis (Penke et al., 2018;Yang et al., 2016). This study showed that Foxm1 expression was upregulated in TGF-β1-induced EndMT and that Foxm1 knockdown could inhibit TGF-β1-induced EndMT and cell migration. Balli et al. reported that one of the Foxm1 roles during lung fibrosis was to induce EMT through direct transcriptional activation of Snail and promote pulmonary inflammation through increased expression of inflammatory mediators (Balli et al., 2013). In addition, a previous study showed that Foxm1 overexpression significantly polymerizes actin assembly and impairs E-cadherin expression, resulting in EMT, and metastasis in a xenograft mouse model, whereas Foxm1 knockdown has the opposite effect (Zhang et al., 2017). The above results demonstrated that Foxm1 is indeed involved in the EMT process. However, the potential mechanism of TGF-β1 regulation of the EndMT through Foxm1 expression is still not understood.
Both in vitro and in vivo studies have shown that TGF-β plays a central role in endocardial EndMT. Sabbineni et al. proved that all three TGF-β isoforms (TGF-β1, TGF-β2, and TGF-β3) induced EndMT in HMECs after 72 hr, which resulted in phosphorylation of Smad2 and Smad3 (Sabbineni et al., 2018). Activated Smad2 and Smad3 form a complex with Smad4. It has also been reported that FOXM1 interacts with Smad3 to sustain activation of the Smad2/Smad3/ Smad4 complex in the nucleus (Xue et al., 2014). The complex translocates to the nucleus and activates ECM associated genes, as well as Snail 1, 2, and Twist genes, which triggers a cascade of signaling pathways that culminate in EndMT (Sabbineni et al., 2018;Yoshimatsu & Watabe, 2011). Cooley et al. (2014) have shown that the TGF-β-Smad2/3-Slug signaling pathway plays a pivotal role in regulating vein graft EndMT, with in vivo reduction of TGF-β signaling decreasing both neointimal formation and the relative contribution of endothelial lineage-derived cells to the neointima.
Interestingly, the two most recent studies indicate the unique role of the Smad1/5 pathway in EndMT (Ramachandran et al., 2018;Sniegon, Priess, Heger, Schulz, & Euler, 2017), in addition to the observed role of the Smad2/3-Snail pathway. All studies listed above suggested that TGF-β signaling plays a critical role in EndMT. Our data showed that TGF-β1 increased phosphorylation of Smad2/3 in HUVECs, whereas inhibition of Foxm1 attenuated this activation.
Loss of VE-cadherin is a hallmark of EndMT. In our study, Foxm1 knockdown with shRNA resulted in the loss of EndMT-related transcription factors after the TGF-β1 treatment. Previous studies demonstrated that the loss of VE-cadherin was mediated by the inhibitor binding to the E-box sequences within the VE-cadherin promoter (Hennig, Lowrick, Birchmeier, & Behrens, 1996). Snail represses VE-cadherin expression through direct binding to these E-boxes (Cano et al., 2000). Balli et al. reported that Foxm1 bound to and increased promoter activity of the Snail gene in alveolar epithelial cells (Balli et al., 2013). Consistent with the important role of Snail in EndMT, the present studies demonstrated that Foxm1 induced Snail mRNA and protein in HUVECs. We investigated whether Snail was a direct transcriptional target of Foxm1 in HUVECs. In the context of endogenous Snail promoter, TGF-β1 increased Foxm1 binding to the Snail promoter DNA as demonstrated by ChIP assay, suggesting that F I G U R E 8 Diagrammatic sketch of the working hypothesis. Foxm1 promotes the TGF-β1-induced EndMT process in ECs through Smad2/3 signaling pathway and direct binding to and inducing the transcriptional activity of the Snail gene: an EndMT-promoting transcription factor. ECs: endothelial cells; EndMT: endothelial-mesenchymal transition; Foxm1: forkhead box M1; Sio A: Siomycin A; TGF-β, transforming growth factor beta; TGF-βR: transforming growth factor beta receptor; TSS: transcriptional start site [Color figure can be viewed at wileyonlinelibrary.com] there was crosstalk between TGF-β1 signaling and Foxm1 in the regulation of the Snail promoter. The finding that Foxm1 directly bound to and increased the activity of the Snail promoter demonstrates that Snail is a direct transcriptional target of Foxm1 providing a mechanism by which Foxm1 induces EndMT and potentially contributes to TGF-β1-induced CF.
In conclusion, Foxm1 promotes TGF-β1-induced EndMT via direct activation of the Smad2/3 signaling pathway and binding to the Snail promoter DNA. Even though TGF-β has been identified as the single most important growth factor that can induce EndMT, there may be other signaling pathways, such as Wnt and MAPK, that can also regulate EndMT (Gonzalez & Medici, 2014). Whether Foxm1 regulates EndMT through other signaling pathways needs further investigation.
Our findings suggest that Foxm1 might be a prospective target for TGF-β1-induced EndMT and a potential target in CF therapy.