Fibroblast growth factor (FGF) signaling regulates transforming growth factor beta (TGFβ)-dependent smooth muscle cell phenotype modulation

Smooth muscle cells (SMCs) in normal blood vessels exist in a highly differentiate state characterized by expression of SMC-specific contractile proteins (“contractile phenotype”). Following blood vessel injury in vivo or when cultured in vitro in the presence of multiple growth factors, SMC undergo a phenotype switch characterized by the loss of contractile markers and appearance of expression of non-muscle proteins (“proliferative phenotype”). While a number of factors have been reported to modulate this process, its regulation remains uncertain. Here we show that induction of SMC FGF signaling inhibits TGFβ signaling and converts contractile SMCs to the proliferative phenotype. Conversely, inhibition of SMC FGF signaling induces TGFβ signaling converting proliferating SMCs to the contractile phenotype, even in the presence of various growth factors in vitro or vascular injury in vivo. The importance of this signaling cross-talk is supported by in vivo data that show that an SMC deletion of a pan-FGF receptor adaptor Frs2α (fibroblast growth factor receptor substrate 2 alpha) in mice profoundly reduces neointima formation and vascular remodelling following carotid artery ligation. These results demonstrate that FGF-TGFβ signaling antagonism is the primary regulator of the SMC phenotype switch. Manipulation of this cross-talk may be an effective strategy for treatment of SMC-proliferation related diseases.

A number of factors involved in regulation of SMC phenotype plasticity have been identified. Among these, platelet-derived growth factors (PDGFs) and fibroblast growth factors (FGFs) are the most robust in their ability to induce a contractile-to-proliferative phenotype switch, and to control smooth muscle cell growth and proliferation [3][4][5][6] . PDGFs were one of the first growth factors identified on the basis of their ability to stimulate growth and migration of fibroblasts, SMCs, and glial cells 7 . Inhibition of SMC PDGF receptor beta (PDGFRβ) signaling diminishes neointimal hyperplasia in balloon-injured and atherosclerotic arteries, as well as in autologous vein grafts [8][9][10] . The FGFs comprise a 22-member family of proteins that mediate their biological activity by binding to four cell surface receptor tyrosine kinases, designated FGF receptors (FGFR1-FGFR4). This results in assembly of a ternary complex that involves an FGF receptor adaptor protein fibroblast growth factor receptor substrate 2 (FRS2), Grb2 and Sos 11 which leads to activation of the Ras/mitogen-activated protein kinase (MAPK) signaling cascade. Similar to PDGFs, FGFs promote SMCs conversion to the proliferative phenotype and stimulate their growth and migration 12 .
In contrast to FGFs and PDGFs, another growth factor, transforming growth factor beta (TGFβ), mediates a SMC switch from the synthetic to contractile phenotype 13,14 . TGFβ exerts its effects by binding and activating the TGF beta type 2 receptor (TGFβR2). The activated TGFβR2 then associates with and activates TGF beta type 1 receptor (TGFβR1) leading to phosphorylation of downstream targets Smad2 and Smad3, resulting, eventually, in expression of SMC contractile proteins [15][16][17][18] . The TGFβ pathway is critical for vascular development and homeostasis of the adult vasculature. Selective deletions of Tgfbr1 or Tgfbr2 in SMCs cause embryonic lethality due to vascular deformities 19 . Adult mice lacking Tgfbr2 in VSMCs develop vascular defects including aortic wall thickening, dilatation, and dissection 20 .
A number of studies reported the ability of FGFs to antagonize TGFβ-mediated induction of SMC markers expression in pericytes and smooth muscle cells 21,22 . FGFs achieve this effect by repressing expression of TGFβ family 23 . In vitro, inhibition of FGF signaling upregulates TGFβR1 and downstream Smad2/3 activity while in vivo it leads to reduced SMC proliferation 23 . These observations suggest that FGF signaling input may regulate the TGFβ-dependent proliferative to contractile SMC phenotype switch. In this study, we set out to test this hypothesis in vitro and in vivo. We found that deletion of Frs2α in SMC increased SMC TGFβ signaling. This led to a proliferative to contractile phenotype conversion even in the presence of high serum (in vitro) and blood vessel wall injury (in vivo). These results support the notion that FGF-TGFβ signaling antagonism is the key regulator of the SMC phenotype switch; thus, manipulation of this cross-talk may be an effective strategy for treatment of SMC-proliferation related diseases.

Results
FGF signaling is downregulated and TGFβ signaling activity is upregulated during smooth muscle cell differentiation. Primary human aortic smooth muscle cells (HASMCs) rapidly lose expression of contractile SMC proteins when cultured in the high serum ("growth") medium, but retain their expression in the low-serum ("differentiation") medium 24 . As expected, switching cultured HASMC from the growth to differentiation medium resulted in a gradual increase in expression of smooth muscle contractile markers (SM α-actin, SM22α, SM-calponin, SM-MHC) and cell cycle inhibitors (p21 and p27), while the expression of the proliferative marker Cyclin D1 gradually declined (Fig. 1A). Western blot analysis demonstrated increased expression levels of SMC contractile markers (SM α-actin, SM22α, SM-calponin, caldesmon, SM-MHC) and cell cycle inhibitor proteins p21 and p27 (Fig. 1B,C), which was consistent with the qRT-PCR data.
Activation of TGFβ signaling has been linked with the induction of SMC differentiation and cell cycle exit 13,25 . Therefore, we examined this signaling cascade in HASMC during their switch from growth to differentiation. There was a gradual increase in TGFβR1 protein expression and activation of TGFβ signaling (increased p-Smad2 and p-MLC) (Fig. 1D), beginning on day 2 after the medium switch. There was a simultaneous decrease in FGFR1 activity, as measured by FGFR1 tyrosine kinase phosphorylation, while its expression remained unchanged (Fig. 1D). The downregulation of FGF signaling and upregulation of TGFβ activity during smooth muscle cell differentiation process was further confirmed in mouse vascular smooth muscle cells (VSMCs) isolated from the mouse aorta ( Supplementary Fig. 1). We used two different experimental approaches to test if TGFβ activity is required for SMC differentiation. First, treatment of HASMCs with the TGFβR1 kinase inhibitor SB431542 during differentiation induction, significantly decreased SM α-actin, SM22α, SM-calponin, and SM-MHC expression compared to DMSO-treated controls (Fig. 1E). Second, transduction of HASMCs with a dominant negative TGFβR1 construct (TGFβR1 K230R) led to an almost a complete block of differentiation (Fig. 1F).

FGF signaling inhibits TGFβ activity and promotes smooth muscle cell proliferative phenotype induction.
Recent studies demonstrated that FGFs regulate both endothelial cell and SMC TGFβ signaling 23,26 .
FGF signaling input was inhibited by a short hairpin RNA (shRNA) to reduce expression of FRS2α, the key adaptor molecule involved in FGF receptor signal transduction 27 . Knockdown of FRS2α, in HASMCs cultured under growth conditions, led to increased expression of TGFβ-related genes (Table 1) and increased expression of TGFβ2 protein in the extracellular medium ( Fig. 2A). Western blotting experiments confirmed the increased TGFβR1 expression and activation of TGFβ signaling (p-Smad2) (Fig. 2B). Confocal microscopy imaging showed a strong nuclear staining for p-Smad2/3 after FRS2α knockdown compared to control cells following TGFβ1 stimulation ( Supplementary Fig. 2). FGFR1 is the most abundant FGF receptor in smooth muscle cells 23 . Knockdown of FGFR1 was associated with an increase in TGFβ1-induced Smad2 phosphorylation and TGFβR1 expression (Fig. 2C), similar to FRS2α knockdown (Fig. 2B).
Importantly, the effect of FGF signaling inhibition on TGFβ activity was specific. Knockdowns of epidermal growth factor receptor (EGFR), insulin-like growth factor-1 receptor (IGF1R), and platelet-derived growth factor receptor beta (PDGFRβ) had little effect on TGFβR1 protein expression or Smad2 phosphorylation in response to TGFβ1 treatment ( Fig. 2D-F). Since TGFβ has been reported to directly regulate expression of vascular SMC differentiation markers 28-31 , we next examined if FGF/TGFβ cross-talk affects this process. To this end, we used Scientific RepoRts | 6:33407 | DOI: 10.1038/srep33407 xFGFR1 K562E and FRS2α 8V mutant constructs which have been shown to constitutively activate FGF signaling in the absence of ligand stimulation 5,32 . Overexpression of either xFGFR1 K562E or FRS2α 8V completely suppressed growth arrest-induced SMC differentiation. This resulted in a decrease in TGFβR1 levels, p-Smad2 activity, and smooth muscle marker gene expression (SM α-actin, SM-calponin, SM-MHC, Notch3) (Fig. 3A,B). Addition of FGF2 to HASMC, cultured under growth arrest conditions, also blocked their differentiation (Fig. 3C). To further assess the role of FGF signaling in the cell cycle regulation, HASMCs were transduced with control, xFGFR1 K562E, or FRS2α 8V lentiviruses under growth conditions and then switching to the differentiation medium for 1 to 8 days. Increased FGF signaling activity upregulated Cyclin D1 while downregulating p21 and p27 expression (Fig. 3D,E). Conversely, inhibition of FGF signaling by a knockdown of the adaptor protein FRS2α, further increased p21 and p27 protein expression compared to controls (Fig. 3F). Thus, inhibition of FGF signaling overrides the proliferative phenotype-inducing effect of the growth medium while FGF stimulation overrides the contractile phenotype-inducing effects of the differentiation medium.
Blockade of SMC FGF signaling in mice decreases neointima formation in the carotid artery ligation model. To investigate whether FGF signaling affects SMC differentiation and phenotype switch in vivo, we used a recently reported mouse line with the SMC deletion of Frs2α 23 . The carotid ligation model was used to assess the effect of this deletion in the Frs2α SMCKO and control mice on smooth muscle proliferation and vascular remodeling. In this blood-flow cessation model, after an early phase of inflammatory cell recruitment, medial SMCs rapidly proliferate and migrate toward the lumen. This results in extensive neointima formation four weeks later ( Fig. 4A) 33 . Morphometric analysis of carotid arteries 4 weeks after the ligation demonstrated a marked decrease in the amount of neointima (62% reduction) and an increase in lumen area (51%) and lumen diameter (28%) in the left (ligated) carotid arteries of Frs2α SMCKO compared to control mice ( Fig. 4B-D). In addition, there was a significant decrease in vessel remodeling of ligated arteries in Frs2α SMCKO mice. This resulted in smaller media and total vessel areas in the ligated arteries of Frs2α SMCKO mice (Fig. 4E). There were no significant differences between non-ligated (right) carotids in the two groups.
SMC proliferation is a major contributor to neointima formation in this model 33 . We used 5-Bromo-2′ -deoxyuridine (BrdU) to assess its extent over the 2 week time course. To this end, BrdU was injected subcutaneously every other day for 2 weeks starting 2 weeks after the ligation. Quantification of the BrdU + signal revealed that in control mice, 21.5% of cells in the neointima and 3.4% of cells in the media of control mice stained positive for BrdU 28 days post-ligation, while 6.4% of the neointimal SMCs and 0.5% of medial SMCs were BrdU positive in Frs2α SMCKO mice (Fig. 5A,B). Similarly, quantification of the α-SMA + BrdU + signal revealed that 16.7% of cells in the neointima and 1.8% of cells in the media of control mice stained positive for α-SMA + BrdU + 28 days post-ligation, while 4.7% of the neointimal SMCs and 0.3% of medial SMCs were BrdU positive in Frs2α SMCKO mice ( Fig. 5A,C). BrdU + labeling was not detected in contralateral non-ligated arteries and in uninjured arteries in control and Frs2α SMCKO mice (Fig. 5A).

Discussion
This report provides several lines of evidence to support the hypothesis that FGF-TGFβ signaling cross-talk is the key regulator of vascular SMC phenotype modulation. Suppression of FGF signaling in primary SMC in vitro activated TGFβ signaling and induced SMC differentiation, even in the presence of a high concentration of serum. At the same time, activation of FGF signaling inhibited TGFβ activity and induced conversion to the proliferative phenotype, even when cells were cultured under the low serum (differentiation medium) conditions. This was further verified in vivo where SMC deletion of Frs2α induced TGFβ signaling and prevented normal contractile media SMC from acquiring a proliferative phenotype, even in the presence of severe vascular injury (carotid artery ligation). The end result was a marked reduction in medial SMC proliferation and a decrease in the neointima formation. Interestingly, Frs2α SMCKO mice were viable and born at the expected Mendelian frequency. This suggests that the absence of FGF signaling in smooth muscle cells can be compensated by other signaling pathways during development, maturation, and adult homeostasis 23 . Many growth factors are capable of inducing SMC contractile to synthetic phenotype switch, which stimulates SMC growth and migration. However, FGF regulation of TGFβ signaling in this context is unique. That is, activation of TGFβ activity, due to reduced FGF signaling input, could not be overridden by high serum concentration in vitro or by vascular injury in vivo; thus, SMC remained in the contractile state. Similarly, FGF stimulation of growth arrested SMCs induced a proliferative phenotype by suppressing TGFβ signaling in the absence of any other growth factor.
The biological significance of this cross-talk is demonstrated by the in vivo results. Smooth muscle cells in the arterial media, in the absence of disease or injury, are found in the G0/G1 phase of the cell cycle. They exhibit high expression levels of contractile smooth muscle marker genes. SMC attain a synthetic state in response to vascular injury, characterized by a high rate of cell proliferation, migration, and production of extracellular components. This leads to formation of neointima, a feature of a number of vasculo-proliferative conditions ranging from atherosclerosis to restenosis 1,2,33 . This switch to the synthetic (proliferative) phenotype is thought to be triggered by changes in local environmental cues, including an increase in the local concentration of mitogens such as EGF, IGF, PDGF, and FGFs. Previous studies showed that inhibition of FGF signaling by an anti-FGF2 antibody or an FGFR tyrosine kinase inhibitor SU5402, resulted in decreased SMC proliferation and attenuation of neointimal thickening 4,34,35 . However, given the broad nature the anti-FGF2 antibody or FGFR inhibitor activity, the mechanism of these effects remained elusive. The results of our study indicate that it is the SMC FGF-TGFβ cross-talk that is central to these effects. The nature of FGF-dependent regulation of TGFβ signaling is related to the previously demonstrated FGF dependence of let-7 miRNA expression 23 . The let-7 miRNA binds to TGFβR1 3′ UTR and inhibits TGFβR1 expression at mRNA levels 26 . Thus, changes in let-7 levels have a profound effect on TGFβR1 expression. The decline in let-7 levels in the absence of the FGF signaling input activates TGFβ signaling while the opposite is true during FGF stimulation.
It is important to note that SM22α-Cre activity is not limited to SMC and it is also expressed in cardiomyocytes and in myeloid cells 36,37 . Bone marrow-derived cells contribute to neointima formation [38][39][40] . Therefore, we cannot definitively rule out the possibility that the decreased neointima formation in Frs2α SMCKO mice was due, in part, to the effect of FGF-TGFβ cross-talk in myeloid cells. This issue can be resolved using a more SMC specific Cre such as Myh11-CreER T2 20,41 , and will be the subject of future studies. This study also provides an interesting insight into the origin of neointimal SMCs. Several sources for these cells have been proposed, including blood and bone marrow derived precursor cells [38][39][40] , dedifferentiated medial VSMCs, resident progenitor cells, adventitial fibroblasts 42-44 and the endothelium due to endothelial-to-mesenchymal transition (EndMT) 26,45,46 . Importantly, there is little quantitative information regarding contribution of each of these potential sources.
In the current study we observed a dramatic reduction in medial SMC proliferation after carotid ligation in Frs2α SMCKO mice. Indeed, there was essentially no BrdU-positive cells in the media of these animals. This translated in to a 62% reduction in the neointima size. This implies that ~40% of the neointima in this model forms in a non-SMC, non-myeloid cell-dependent manner. Endothelial-to-mesenchymal transition likely accounts for the lion share of these cells 45,47 . In summary, this study provides in vitro and in vivo evidence that FGF signaling is the primary controller of SMC phenotype modulation and that this control is achieved via direct regulation of TGFβ activity.   were maintained in Dulbecco's modified Eagle's medium (Gibco 11965-092) with 10% fetal bovine serum (Life Technologies 16000-044) and penicillin-streptomycin (15140-122, Gibco), and were grown at 37 °C, 5% CO2. Human aortic SMCs (#C-007-5C), media (#M231-500), and supplements (SMGS: S-007-25; SMDS: S-008-5) were purchased from Life Technologies. The cells were grown at 37 °C, 5% CO2 in Medium 231 supplemented with smooth muscle growth supplement (4.9% FBS, 2 ng/ml FGF2, 0.5 ng/ml EGF, 5 ng/ml heparin, 2 μg/ml IGF-1, and 0.2 μg/ml BSA). For SMC differentiation, HASMC were incubated with Medium 231 containing smooth muscle differentiation supplement (1% FBS and 30 μg/ml heparin) for different lengths of time. Primary human aortic SMCs between passages 6 and 10 were used in all experiments. Primary mouse SMCs were isolated from mouse aorta as previously described 20 . In brief, aortas were predigested for 5 minutes at 37 °C in HBSS (Gibco 14170-112) solution containing 1 mg/ml collagenase type A (Sigma 10103578001) to facilitate removal of the adventitia under a dissecting microscope. The denuded vessels were transferred into 3 ml HBSS solution containing 2 mg/ml collagenase type A and 0.5 mg/ml elastase (Worthington LS006365) and incubated at 37 °C for 30 minutes. Digestion was stopped with growth medium, the mixture was centrifuged, and the cells were resuspended in Claycomb medium (Sigma 51800C) supplemented with 10% fetal bovine serum (Life Technologies 16000-044), 10 units/ml Penicillin/10 μg/ml Strep (Gibco 15140-122), 2 mM L-glutamine (Gibco 25030-081), and cultured in 35 mm dishes. For mouse SMC differentiation, the cells were maintained in Dulbecco's modified Eagle's medium (Gibco 11965-092) with 0.5% fetal bovine serum (Life Technologies 16000-044). Primary mouse SMCs between passages 2 and 3 were used in all experiments.

Generation of lentiviruses.
Human EGFR, FGFR1, IGF1R, and PDGFRβ shRNA lentiviral constructs were purchased from Sigma while human FRS2α shRNA lentiviral construct was purchased from Open Biosystems. For the production of shRNA lentivirus, 3.7 μg of Δ 8.2, 0.2 μg of VSVG, and 2.1 μg of pLKO.1 carrying the control, EGFR, FGFR1, FRS2α, IGF1R, or PDGFRβ shRNAs were co-transfected into 293T cells using X-tremeGENE 9 DNA Transfection Reagent (Sigma 6365787001). Forty-eight hr later the medium was harvested, cleared by 0.45 μm filter (PALL Life Sciences 4184), mixed with polybrene (5 μg/ml) (Sigma H9268), and applied to cells. After 6 hr of incubation, the virus-containing medium was replaced by the fresh medium. For the production of xFGFR1 K562E 5 , FRS2α 8V 32 , or TGFβR1 K230R 26 , 10 μg of pLVX-IRES carrying the xFGFR1 K562E, FRS2α 8V, or TGFβR1 K230R, 5 μg of pMDLg/PRRE, 2.5 μg of RSV-REV, and 3 μg of pMD.2G were co-transfected into 293T cells using X-tremeGENE 9 DNA Transfection Reagent (Sigma 6365787001). Forty-eight hr later, the medium was harvested, cleared by 0.45 μm filter (PALL Life Sciences PN4184), mixed with 5 μg/ml polybrene (Sigma H9268), and applied to the cells. After 6 hr incubation, the virus-containing medium was replaced by fresh medium. TGFβ2 content. Cultured HASMCs from control and FRS2α knockdown were seeded at a confluent cell density (150000 cells in 6 cm culture plate) and then maintained in culture for three days without changes of the Scientific RepoRts | 6:33407 | DOI: 10.1038/srep33407 culture medium. Thereafter, the respective extracellular medium was collected and a commercially available sandwich ELISA kit (R&D Systems #DB250) was used to evaluate the secretion of TGFβ2 in the extracellular medium. The absorbance (450 nm) for each sample was analyzed by a microplate reader (BioTek Synergy2 multi-Mode) and was interpolated with a standard curve.
Quantification of Western blots. Images of blot signals on HyBlot ES ® Autoradiography Film (DENVILLE E3218) were scanned on a CanoScan LiDE 200 scanner. Images were then viewed in ImageJ software for data analysis. Signal intensities of individual bands were determined using Gel Analysis followed the ImageJ user's guide. Data were exported to Prism 7 to generate the plot. Data are presented as fold change in protein expression for the experimental groups compared to the control group after normalized to loading controls (GAPDH, HSP90, β-tubulin or total phospho-protein).
Generation of mice. All mice were maintained on a C57BL/6 background. Frs2α flox/flox mice were previously described 49 . Frs2α flox/flox mice were crossed with mice expressing Cre recombinase under the control of SM22α promoter 50 . PCR genotyping analysis was done using the following primers: Frs2α flox/flox (5′-GAGTG TGCTGTGATTGGAAGGCAG-3′ and 5′-GGCACGAGTGTCTGCAGACACATG-3′), SM22α-Cre (5′-GCG GTCTGGCAGTAAAAACTATC-3′, 5′-GTGAAACAGCATTGCTGTCAC TT-3′, 5′-CTAGGCCACAGA ATTGAAAGATCT-3′, and 5′-GTAGGTGGAAATTCTAGCATCATCC-3′). All animal procedures were performed under protocols approved by Yale University Institutional Animal Care and Use Committee and carried out in accordance with the approved protocol and the relevant University guidelines regarding the use of anesthesia, post-operative monitoring and performance of surgical procedures. Mouse carotid ligation model. To achieve carotid artery ligation, the left carotid artery of an anesthetized 8-10 week-old mouse was isolated and ligated with a 6-0 silk suture just proximal to the carotid bifurcation through a midline cervical incision (Fig. 4A) 51 . The wound was close with a 5-0 prolene suture. Four weeks after injury, all animals were anesthetized and perfused with PBS, followed by 4% paraformaldehyde for 5 min. Both left and right carotid arteries were excised and embedded in paraffin. Lumen diameter, lumen area, neointima area, media area, and total vessel area were measured using NIH Image J. Animals also received 8 subcutaneous injections of the thymidine analog 5-Bromo-2′ -deoxyuridine (BrdU) (Sigma B9285; 1 mg per injection).
Histology and Immunohistochemical staining. In ligated left common carotid arteries, cross sections from the predefined proximal distances from the ligation site (500 μm) were analyzed. In non-ligated right common carotid arteries or in no injured left and right common carotid arteries, sections from 500 μm distance to the bifurcation of internal and external carotid artery were analyzed. Blocks were sectioned at 5 μm intervals using a Microtome. Slides were dewaxed in xylene, boiled for 20 min in citrate buffer (10 mM, pH 6.0) for antigen retrieval, and rehydrated. After washing three times with phosphate-buffered saline, tissue sections were incubated with primary antibodies diluted in blocking solution (10% BSA and horse serum in PBS) overnight at 4 °C in a humidified chamber. For BrdU staining, slides were denatured with 1.5 M HCl for 20 min prior to antibody labeling. Sections were washed three times with tris-buffered saline, incubated with appropriate Alexa Fluor-conjugated secondary antibody (1:500) (Invitrogen) for 1 hr at room temperature, washed again three times, and mounted on slides with ProLong Gold mounting reagent with DAPI (Life Technologies P36931). All immunofluorescence micrographs were acquired using a Zeiss microscope. Images were captured using Volocity software and quantifications performed using ImageJ software (NIH).

Statistical analysis.
Graphs and statistical analysis were prepared using GraphPad Prism. Data are expressed as mean ± SD. The level of statistical significance was determined by 2-tailed Student's t test as appropriate using the GraphPad Prism software. A P value less than 0.05 was considered significant (*P < 0.05, **P < 0.01, ***P < 0.001). All results were confirmed by at least 3 independent experiments. Error bars represent mean ± SD.