Fibroblast Growth Factor 2 Augments Transforming Growth Factor Beta 1 in Inducing Epithelial-Mesenchymal Transition in Human Lung Epithelial Cells

Background: Epithelial-mesenchymal transition (EMT) is a critical event in wound healing and tissue repair following injury. Transforming growth factor beta-1 (TGFβ1) plays an important role in inducing EMT in lung epithelial cells in vitro and in vivo. As fibroblast growth factor-2 (FGF2) reverses TGFβ1-induced collagen I (COL1A1) and α-smooth muscle actin (Actin alpha 2; ACTA2) expression in primary mouse and human lung fibroblasts, we set out this study to determine the effect of FGF2 on TGFβ1-induced EMT in human lung epithelial cells. Methods: BEAS-2B and A549 cells were treated with recombinant FGF2 (2 nM) with or without TGFβ1 (2 ng/ml) for up to 4 days. The phenotypic alterations associated with EMT were assessed by quantitative real-time PCR and E-cadherin protein expression levels was assayed by western blot and immunofluorescence staining. Cell migration was confirmed using wound-healing assay. Results: TGFβ1 treatment led to significantly reduced expression of E-cadherin (CDH1) and markedly induced expression of mesenchymal proteins such as N-cadherin (CDH2), tenascin C (TNC), fibronectin (FN), ACTA2 and COL1A1. TGFβ1 also induced a morphological change and a significant increase in cell migration. FGF2 did not significantly alter EMT gene expression markers on its own, however enhanced TGFβ1-induced suppression of CDH1 and upregulation of ACTA2, but did not alter TNC, FN and CDH2 gene expression levels induced by TGFβ1. FGF2 maintained TGFβ1-induced morphologic changes as well as increased the migration of TGFβ1-treated cells. Furthermore, FGF2 treatment significantly inhibited TGFβ1-induced COL1A1 expression in both BEAS-2B and A549 cells. FGFR-specific tyrosine kinase inhibitor PD173074 blocked the synergism between these two growth factors. Conclusions: This study suggests a synergistic effect between TGFβ1 and FGF2 in inducing EMT, which may play an important role in wound healing and tissue repair after injury. Our findings provide insight into the effects of FGF2 following lung injury and in pulmonary fibrosis. suggest that generating a well-established EMT phenotype

4 cells are transformed into invasive metastatic mesenchymal cells that underlie cancer progression [8,9]. The mechanism underlying EMT is not well understood in the context of pulmonary fibrosis, and the role of EMT in the pathogenesis of different respiratory diseases such as asthma, COPD and pulmonary fibrosis is under debate [10,11].
It has been reported that FGF2 reduces E-cadherin in human ovarian cancer cells [40], and induces the expression of mesenchymal markers (VIM, FSP1, α-SMA and SNAI1) in corneal endothelial cells [41] and proximal tubular epithelial cells [39,42]. A number of studies have shown the synergistic effect of combined treatment of TGFβ1 and FGF2 in inducing EMT in mouse normal mammary epithelial (NMuMG) cells [43], rat Hertwig's epithelial root sheath (HERS) cells [44], mouse lung epithelial type II cell line MLE-12 [45], human nonsmall cell lung cancer cell lines NCI-H1975 and NCI-H165 [46], and human lung adenocarcinoma cell lines PC-9, HCC-827 and A549 [47,48]. These studies only used the combination of TGFβ1 and FGF2 as profibrotic cytokines to induce type III EMT which is a key component of carcinogenesis. Therefore, inhibition of EMT induction may modify tumor progression and responsiveness to chemotherapy and/or immunotherapy for cancer. 5 No studies have shown the synergistic effect of FGF2 and TGFβ1 in inducing type II EMT in lung epithelial cells that is associated with wound healing and tissue regeneration after injury.
We have previously shown that FGF2 is crucial for epithelial repair and recovery after bleomycin-induced lung injury in mice [49]. We have also found that FGF2 overexpression is protective against bleomycin-induced lung injury in vivo and inhibits TGFβ1-induced collagen I and α-SMA expression in primary mouse and human lung fibroblasts in vitro, suggesting that FGF2 is antifibrotic and protective against lung injury [50]. In this study we investigated the effect of FGF2 on TGFβ1-induced gene expression in both bronchial and alveolar lung epithelial cells in vitro. We hypothesized that FGF2 would induce EMT and may play an important role in wound healing and repair of lung epithelial cells after injury. In vitro, we found that FGF2 did not inhibit the majority of TGFβ1-induced expression of EMT markers in human lung epithelial cells. This study provides insight into the potential use of FGF2 following lung injury and in pulmonary fibrosis.

Cell culture
Human virus-transformed bronchial epithelial cell line (BEAS-2B) and Human alveolar type II epithelial carcinoma cell line (A549) were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA; #CRL9609, #CCL185, respectively). BEAS-2B cells were plated on pre-coated plates with a mixture of 0.01 mg/ml fibronectin, 0.03 mg/ml bovine collagen type I and 0.01 mg/ml bovine serum albumin (BSA) dissolved in BEBM medium and were grown in bronchial epithelial growth medium bullet kit (BEGM; Lonza, Walkersville, MD, USA). A549 cells were plated in Ham's F-12 medium (Life Technologies, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum and 1% penicillinstreptomycin (10,000 U/ml). Cells were incubated in a humidified incubator at 37°C with 6 5% CO 2 .

Epithelial-mesenchymal transition induction
To induce epithelial-mesenchymal transition, cells were plated at ~30-40% confluence in 6-well plates. After overnight culture, cells were treated with 2 ng/ml of TGFβ1 (Fisher Scientific, Fair Lawn, NJ, USA, #PHG9204), 2 nM of recombinant human low-molecularweight FGF2 (PeproTech, Rocky Hill, NJ, USA, #100-18B) + 1 nM heparin sulfate (Fisher Scientific, #BP2524), or TGFβ1 + FGF2 + heparin for 4 days in complete medium. Medium with or without treatments was changed after 48 hours. The experiments were designed so that the cells reached confluence one day prior to harvesting and were conducted independently 3-6 times each in duplicate.

RNA isolation and quantitative real-time PCR
Cells were lysed in RLT buffer and total RNA was prepared from the cells using the RNeasy plus mini kit (Qiagen, Valencia, CA, USA, #74136) according to the manufacturer's instructions. RNA concentration was determined utilizing a Nanodrop spectrophotometer.
cDNA was made using the BioRad iScript Reverse Transcription Supermix for RT-qPCR kit (BioRad, Hercules, CA, USA, #170-8841). Quantitative RT-PCR was performed on an Applied Biosystems StepOne thermocycler using ABI Taqman® Fast Advanced Master Mix (Applied Biosystems, Foster City, CA, USA, #4444557) and Taqman® gene expression assays. All samples were normalized to GAPDH and then scaled relative to controls using the standard delta-Ct (∆Ct) method. Data are reported as fold change over control.
Membranes were blocked for one hour at room temperature with gentle shaking in Tris- Immunoblotting for β-tubulin (Abcam, Cambridge, MA, USA, #ab6046) was used as loading control. After three rinses in TBST, membranes were incubated for one hour at room temperature in horseradish peroxidase-linked secondary antibodies in TBST with 5% nonfat milk, rinsed again in TBST, and developed using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Scientific, #34096) or Pico (Thermo Scientific, #34580).

Migration assay
BEAS-2B cells were cultured in 6-well plates and grown in complete media. Cells werẽ 90% confluent at the time of scratch wounding. Five 1 mm diameter circular wounds were created using a custom-made rubber tool [51]. The non-adherent cells washed off with medium and fresh medium with the same treatments as described previously was added to the wells. The wound closure was measured immediately after scratch wounding

Immunofluorescence staining for E-cadherin
BEAS-2B cells were grown on coated glass coverslips and stimulated with the same treatments for 4 days as described above. Cells were fixed with 4 % paraformaldehyde for 10 minutes at room temperature. After washing with ice-cold PBS, cells were blocked in 0.1% BSA and 5% serum prepared in PBS + 0.2% Tween-20 for 1h at room temperature.

Statistical Analysis
The data showed the mean ± standard deviation and the significant differences in mean values were determined using one-way ANOVA. A p -value of less than 0.05 was considered to be significant. Statistical analysis was performed using GraphPad Prism software.

Recombinant FGF2 does not alter morphological changes in bronchial epithelial cells induced by TGFβ1.
Addition of FGF2 to BEAS-2B cells did not alter their cobblestone-like morphology (Figure 1A, B). Upon treatment with TGFβ1, cells developed a fibroblast-like shape ( Figure 1C). Compared to cells treated with TGFβ1 alone, the addition of FGF2 maintained the fibroblast-like shape of BEAS-2B cells (Figure 1 D).

Recombinant FGF2 enhances TGFβ1 induced EMT-related gene expression.
To determine the minimal concentration of TGFβ1 and the time point required to induce EMT in BEAS-2B cells, BEAS-2B cells were treated with 2 ng/ml or 5 ng/ml of TGFβ1 for 3, 4 and 5 days and total RNA was collected. Expression of CDH1, ACTA2, and COL1A1 mRNA were then assessed by qRT-PCR. We found that 2 ng/ml was sufficient to repress CDH1 and induce ACTA2 and COL1A1, but ACTA2 started to be only detectable after 4 days of treatment (data not shown). We then determined whether the addition of FGF2 to BEAS-2B cells leads to altered gene expression. BEAS-2B cells were treated for 4 days with FGF2 (2 nM), TGFβ1 (2 ng/ml), or TGFβ1 + FGF2. FGF2 treatment alone led to a non-significant decrease in CDH1 (P value = 0.2225) and a non-significant increase in both ACTA2 ( P value = 0.2239) and CDH2 (P value = 0.7515) when compared to control. We observed a significant decrease in CDH1 (P value = <0.0001) (Figure 2A)  To confirm these mRNA changes, we assessed the effects of TGFβ1 and FGF2 on Ecadherin protein levels in BEAS-2B cells. Immunoblotting of total cell lysates obtained after 4 days of incubation with TGFβ1, FGF2, or TGFβ1 + FGF2 demonstrated that Ecadherin protein levels were not significantly altered by FGF2 treatment alone and were significantly decreased (P value = 0.0012) in response to TGFβ1 (Figure 3A, B). Addition of FGF2 to TGFβ1 led to further suppression of E-cadherin (Figure 3A, B).
Immunofluorescence for E-cadherin revealed a loss of grid-like localization of E-cadherin at the cell-cell contact surface following FGF2 treatment and further loss of cell-cell contact induced by TGFβ1 that was not altered by the addition of FGF2 ( Figure 3C).

FGF2 inhibits TGFβ1-induced collagen, but not fibronectin or tenascin C.
We then tested whether FGF2 has an effect on expression of extracellular matrix (ECM) proteins such as fibronectin (FN), tenascin C (TNC) and collagen I (COL1A1). BEAS-2B cells were treated with FGF2, TGFβ1, or TGFβ1 + FGF2 for 4 days prior to total mRNA collection.
Treatment with FGF2 alone did not significantly alter expression of FN, TNC, or COL1A1, but TGFβ1 treatment led to a highly significant induction of each of these genes (P values = <0.0001) (Figure 4A-C). FGF2 treatment had no effect on TGFβ1-induced expression of FN ( Figure 4A) and TNC ( Figure 4B), but interestingly there was a significant decrease in the expression of COL1A1 (P value = 0.0008) ( Figure 4C) compared to TGFβ1 treatment alone unlike other EMT genes studied in this report.

Effect of FGF2 on TGFβ1-treated cells is inhibited by PD173074 in both types of epithelial cells.
To confirm that the effect of FGF2 is mediated by FGF receptor (FGFR) signaling, BEAS-2B and A549 cells were treated with the FGFR-specific tyrosine kinase inhibitor PD173074 (0.1 µM) in combination with FGF2, TGFβ1, or TGFβ1 + FGF2. We found that PD173074 significantly blocked the effect of FGF2 on TGFβ1-induced repression of CDH1 (P value =0.0002) ( Figure 6A) and induction of ACTA2 (Figure 6B) , and reversed FGF2-mediated inhibition of TGFβ1 induction of COL1A1 expression non-significantly (P value = 0.8644) in BEAS-2B cells (Figure 6C). Similarly, PD173074 inhibited FGF2 effect on TGFβ1-induced EMT in A549 cells significantly for CDH1 (P value = <0.0001) ( Figure   6D) and non-significantly for ACTA2 ( P value = 0.5619) ( Figure 6E) and COL1A1 (P value = 0.0641) (Figure 6F). PD173074 alone did not alter the expression of the above genes in both types of epithelial cells.

Discussion
This study has demonstrated that TGFβ1 induces EMT in human bronchial epithelial cells 12 (BEAS-2B) and alveolar type II epithelial cells (A549), as shown by morphological and EMTrelated gene expression. We found that FGF2 enhances TGFβ1 induced EMT-related gene expression except for collagen I, which is inhibited by FGF2. Additionally, epithelial migration was enhanced by pre-treatment with FGF2 and/or TGFβ1.
Although the synergistic effect between TGFβ1 and FGF2 in inducing EMT [43,44,47] and proliferation [54][55][56] has been described in other cell types, to our knowledge, the synergistic effect between FGF2 and TGFβ1 in inducing type II EMT in lung epithelial cells in vitro had not been described previously.
In this study, we stimulated both BEAS-2B and A549 cells with TGFβ1, FGF2 or both for up to 4 days. In response to TGFβ1, BEAS-2B cells lost their cobblestone morphology and adopted an elongated spindle-like shape, and this shape was unaltered by the addition of FGF2. TGFβ1 treatment also led to downregulation of the epithelial cell-specific adherence junction protein CDH1 and upregulation of the mesenchymal markers ACTA2 and CDH2. In accordance with the present results, Shirakihara et al. [43] showed that there was a 13 synergistic effect between FGF2 and TGFβ1 on the induction of EMT in the mouse normal mammary epithelial (NMuMG) cells, without evidence of induction of EMT by FGF2 alone.
They found that the morphology of NMuMG cells clearly changed from a cobblestone-like shape to a fibroblastic spindle shape with TGFβ1 treatment. Although FGF2 alone did not alter this shape, the addition of FGF2 to TGFβ1-treated cells maintained the spindledshape morphology. Li et al. [46] and Kurimoto et al. [48] indicated that treatment with FGF2 + TGFβ1 downregulated E-cadherin, upregulated vimentin and N-cadherin and increased migration ability in A549 cells. These findings suggest that the combination of FGF2 and TGFβ1 treatment and not FGF2 alone is an effective way of promoting the induction of an EMT phenotype.
The present study showed that FGF2 alone or in combination with with TGFβ1, increased the migratory capacity of BEAS-2B cells after 24 h, 40 h and 48 h that were pre-treated for 3 days prior to wounding, however, the cell motility of immediately treated cells after wounding was not accelerated significantly when measured after 24 h, 48 h and 72 h.
These results show that treatment with TGFβ1, FGF2, or both requires at least 3 days to promote increased motility and almost complete wound closure after injury. These findings are consistent with a previous study which showed that TGFβ1 treatment alone increased cell motility, and the addition of FGF2 to TGFβ1-treated cells treated for 4 days strongly enhanced the motility of NMuMG cells [43]. These results match those observed in Chen et al. [44] study who also reported that treatment with TGFβ1, FGF2, or both generate EMT phenotype in rat Hertwig's epithelial root sheath (HERS) cells. They found that there was no significant difference in the EMT markers mRNA expression levels after 3 days and highly appeared after 7 days of induction. Also, they found the migratory capacity highly increased after 48h and 72 h of the pretreated cells with TGFβ1, FGF2, or both for 3 days.
These findings suggest that generating a well-established EMT phenotype in epithelial 14 cells using the combined treatments of TGFβ1 and FGF2 require prolonged induction.
Several studies have shown increased COL1A1 expression in BEAS-2B [28,29,57] as well as in A549 [26,32,35] epithelial cells in response to TGFβ1. This also accords with our observations, which showed that TGFβ1 treatment significantly increases COL1A1 expression in both BEAS-2B and A549 cells. However, interestingly, COL1A1 expression was dramatically suppressed with the addition of FGF2 to TGFβ1-treated BEAS-2B and A549 cells. These findings mirror those of our previous study demonstrating inhibition of TGFβ1-induced collagen expression by FGF2 in primary mouse and human lung fibroblasts in vitro [50]. These results may provide an important insight into the antifibrotic effect of FGF2 through suppression of TGFβ1-induced collagen expression in both lung fibroblast and epithelial cells.
The FGFR-specific tyrosine kinase inhibitor PD173074 was used to block the inductive effect of FGF2 on TGFβ1-treated cells. PD173074 inhibitor was reported to show both high affinity and selectivity for the FGF receptor (FGFR) family [58]. We found that the addition of PD173074 attenuates the inductive effect of FGF2 as indicated by reversing reduction of CDH1 expression and COL1A1, and the induction of ACTA2. This finding suggests that the effect of FGF2 is dependent upon canonical signaling through FGFRs.
Numerous studies have demonstrated that EMT is implicated in pulmonary fibrosis and cancer metastasis in mouse models [59][60][61] and in humans [62,63]. The importance of EMT in pulmonary fibrosis has been challenged by other studies in animal models [64][65][66] and humans [67]. Whilst this study did not investigate the cellular mechanism underlying the effect of FGF2 in lung epithelial cells, it does demonstrate that FGF2 has an antifibrotic effect in part through decreasing collagen expression in epithelial cells and promotion of TGFβ1-induced gene expression required for migration of epithelial cells in wound repair after injury. 15

Conclusions
In conclusion, the data presented in this report suggest that FGF2 is synergistic with TGFβ1 to drive type II EMT associated with wound healing and tissue repair after injury.
The mechanism driving this synergy needs more investigation to identify possible therapeutic targets and the future uses of FGF2 following lung injury and in pulmonary fibrosis.

Availability of data and materials
The datasets used and analyzed during the current study available from the corresponding author on reasonable request.

Competing interests
No conflicts of interest, financial or otherwise, are declared by the authors.

Funding
This work was funded by NIH grant K08HL125910, and AHA grants 14FTF19840029. LMFB was supported by The Culture Affairs and Mission Sector, Ministry of Higher Education and Scientific Research, Egypt.

Author Contributions
LMFEB performed all experiments, analyzed data, and wrote the manuscript. NMS, MLS, and HSH contributed to manuscript revision. RDG designed, supervised all experiments, and contributed to manuscript revision.  Statistical significance was determined by one-way ANOVA followed by Tukey's multiple comparisons test; ns = not significant, * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001, and **** indicates p < 0.0001.   FGF2, but not TGFβ1, increases BEAS-2B migration of epithelial cells when added immediately after wounding. BEAS-2B were grown in complete media to confluence, and 1mm diameter circular wounds were generated and the cells were treated immediately with TGFβ1 (2 ng/ml), FGF2 (2 nM), or TGFβ1 + FGF2 (n =3). (A) Wound area was imaged at 0 h, 24 h, 48 h and 72 h then the % of wound closure was measured using ImageJ software. Statistical significance was determined by one-way ANOVA followed by Tukey's multiple comparisons test; Red asterisks mark significant differences for FGF2 vs. TGFβ1, blue asterisks mark significant differences for TGFβ1 + FGF2 vs. FGF2, and green asterisks mark significant differences for TGFβ1 vs. control. * indicates p < 0.05, *** indicates p  Statistical significance was determined by one-way ANOVA followed by Tukey's multiple comparisons test; Red asterisks mark significant differences for FGF2 vs.