Fibulin-4 deficiency increases TGF-β signalling in aortic smooth muscle cells due to elevated TGF-β2 levels

Fibulins are extracellular matrix proteins associated with elastic fibres. Homozygous Fibulin-4 mutations lead to life-threatening abnormalities such as aortic aneurysms. Aortic aneurysms in Fibulin-4 mutant mice were associated with upregulation of TGF-β signalling. How Fibulin-4 deficiency leads to deregulation of the TGF-β pathway is largely unknown. Isolated aortic smooth muscle cells (SMCs) from Fibulin-4 deficient mice showed reduced growth, which could be reversed by treatment with TGF-β neutralizing antibodies. In Fibulin-4 deficient SMCs increased TGF-β signalling was detected using a transcriptional reporter assay and by increased SMAD2 phosphorylation. Next, we investigated if the increased activity was due to increased levels of the three TGF-β isoforms. These data revealed slightly increased TGF-β1 and markedly increased TGF-β2 levels. Significantly increased TGF-β2 levels were also detectable in plasma from homozygous Fibulin-4R/R mice, not in wild type mice. TGF-β2 levels were reduced after losartan treatment, an angiotensin-II type-1 receptor blocker, known to prevent aortic aneurysm formation. In conclusion, we have shown increased TGF-β signalling in isolated SMCs from Fibulin-4 deficient mouse aortas, not only caused by increased levels of TGF-β1, but especially TGF-β2. These data provide new insights in the molecular interaction between Fibulin-4 and TGF-β pathway regulation in the pathogenesis of aortic aneurysms.

A crucial role for the TGF-β pathway in syndromes associated with TAA became evident from both studies in patients and in mouse models [10][11][12][13][14] . Although TAAs are usually associated with increased TGF-β signalling, this association has also been observed with loss of function mutations in TGF-β and TGF-β receptors 15 . The identification of these mutations has led to new insights in the pathogenesis of aneurysm formation, but the molecular mechanism remains to be elucidated.
Mutations in genes of the TGF-β pathway and the ECM lead to phenotypic and functional SMC loss: Tgfbr2 mutations in Loeys-Dietz syndrome lead to decreased expression of SMC contractile proteins 5 . Furthermore, SMCs from mice with Marfan syndrome, another syndromic form of TAAs caused by mutations in the ECM glycoprotein Fibrillin-1, display an altered expression profile with morphological changes, but retain expression of vascular SMC markers 4 . In addition, increased TGF-β signalling inhibits proliferation of SMCs 16 .
Upregulated TGF-β signalling has been observed in another heritable form of TAA caused by a deficiency in the extracellular matrix protein Fibulin-4 13,[17][18][19][20] . Fibulin-4 regulates proper elastogenesis by tethering lysyl oxidase to tropoelastin to facilitate crosslinking 21,22 . In Fibulin-4 deficient patients and mice elevated TGF-β signalling has been shown 12,13,20 . However, the exact mechanism by which Fibulin-4 deficiency leads to increased TGF-β signalling remains to be determined. To further investigate this we isolated SMCs from the aortic arch of hypomorphic Fibulin-4 (Fibulin-4 R/R ) mice, displaying a 4-fold reduction of Fibulin-4 expression. This leads to congenital vascular abnormalities in these mice, including TAAs and vascular tortuosity 12 . Heterozygous Fibulin-4 +/R mice, which have a 2-fold reduced Fibulin-4 expression, show minor irregularities and ECM changes in the aortic wall. Our data reveal that TGF-β signalling is enhanced in isolated SMCs derived from the aortas of Fibulin-4 deficient mice. We observed a decreased proliferation rate in Fibulin-4 R/R SMCs, which could be reverted by addition of TGF-β neutralizing antibodies. We found that this increased TGF-β signal transduction activity is not only associated with increased levels of TGF-β 1, but especially with enhanced TGF-β 2 levels. Increased levels of TGF-β 2 could also be detected in blood and aortic tissue lysates of the Fibulin-4 R/R mice. Treatment of Fibulin-4 R/R mice with losartan, an angiotensin II type-1 receptor blocker, reduced the increased TGF-β 2 levels in blood plasma. This study shows that increased TGF-β signalling in SMCs of Fibulin-4 deficient mice leads to decreased proliferation of SMCs and could be caused by increased bioavailability of TGF-β 1 and especially TGF-β 2.

Results
Characterization of SMCs derived from Fibulin-4 deficient aortas. To examine TGF-β signalling in Fibulin-4 deficient SMCs, we isolated SMCs from the aortic arches of Fibulin-4 +/+ , Fibulin-4 +/R and Fibulin-4 R/R mice. To confirm that the cells we isolated were SMCs, the cells were analysed for the presence of SMC markers, including α -smooth muscle actin (α -SMA), smooth muscle specific protein-22 (SM22), smooth muscle myosin heavy chain II (MHC II) and fibroblast specific protein 1 (FSP1), which stains SMCs with a rhomboid phenotype 23,24 . Human umbilical vein endothelial cells (HUVECs) were taken along as positive control for CD31 staining and were negative for all other markers, while mouse embryonic fibroblasts (MEFs) were positive controls for FSP1, and SMA and SM22 staining 25 . Isolated SMCs showed positive staining for α -SMA, SM22, MHC II, FSP1 and were negative for CD31 (Fig. 1a) confirming the SMC phenotype. QPCR expression analysis also showed no detectable CD31 and von Willebrand Factor (an additional endothelial marker) mRNA expression (data not shown). α -SMA was highly expressed and seemed somewhat increased in Fibulin-4 +/R and Fibulin-4 R/R SMCs (Fig. 1b). Next, the levels of Fibulin-4 were analysed by QPCR. These data revealed that expression levels of Fibulin-4 mRNA in Fibulin-4 +/R and Fibulin-4 R/R SMCs were downregulated (Fig. 1c). These data show that we isolated a population of SMCs with a gradual reduced Fibulin-4 expression level, which we used for further cell biological and molecular analyses.
Losartan treatment rescues lethality and lowers plasma TGF-β2 levels in Fibulin-4 R/R mice. Next, adult mice were treated with Losartan, an angiotensin-II type-I receptor blocker, which prevents aortic root enlargement and reduces circulating TGF-β 1 in a Marfan mouse model 30 . Compared to the increased secretion of TGF-β 2 in placebo treated Fibulin-4 R/R mice we observed, TGF-β 2 levels were not detectable in the 10 losartan treated Fibulin-4 R/R mice. Consistent with previous studies 31 , Losartan treatment of wild type, Fibulin-4 +/R and Fibulin-4 R/R mice showed improved survival rates of Losartan treated Fibulin-4 R/R mice until at least the age of 160 days compared to placebo treated Fibulin-4 R/R mice, which maximally survive until the age of 100 days (Fig. 5c). All losartan and placebo treated wild type and Fibulin-4 +/R mice survived at least until the duration of the experiment (data not shown). Despite improved survival, 160 days old Losartan treated Fibulin-4 R/R mice developed significantly enlarged aortic diameters compared to losartan treated wild type mice (Fig. 5d), and a thickened and degenerated aortic wall architecture as evidenced by fragmentation of its elastin layers (Fig. 5e). Previously, we showed a reduced SMA staining in the aortic wall of 100 days old Fibulin-4 deficient mice, indicative for SMC loss 31 . This SMC loss is not ameliorated by losartan treatment of Fibulin-4 R/R mice (Fig. 5e). Both placebo and Losartan treated Fibulin-4 +/R mice showed an increase in aortic wall thickness and minor elastin breaks compared to wild type mice, which was also previously observed in non-treated Fibulin-4 +/R mice 12 . These results show that lethality and increased plasma TGF-β 2 levels in Fibulin-4 R/R mice can be reduced by losartan treatment, showing a causal relation between increased TGF-β signalling and lethality in aneurysmal Fibulin-4 mice.

Discussion
In this study we show that TGF-β signalling is gradually enhanced in Fibulin-4 deficient SMCs in a Fibulin-4 dose-dependent manner and influences proliferation of these cells. The increased TGF-β signalling is consistent with increased TGF-β 1 levels, and especially with increased TGF-β 2 levels, detected in plasma from Fibulin-4 deficient mice.
Previous analyses on aortas from Fibulin-4 deficient mice showed increased TGF-β signalling associated with aneurysm formation by gene expression analysis and increased nuclear pSMAD2 staining in the SMCs of these aortas 12 . Isolation of aortic SMCs from these mice provided the opportunity to assess TGF-β signalling in vitro. Fibulin-4 R/R SMCs have a reduced proliferation rate compared to Fibulin-4 +/R and wild type SMCs, which is reversed by TGF-β inhibition. Reduced proliferation only takes place after a prolonged incubation time, which is most probably caused by the requirement of certain levels of TGF-β before it affects the proliferation rates of the SMCs. However, in the aortic wall local active TGF-β concentrations can be much higher, due to local activation of the ECM bound TGF-β . In our previous studies we found a hyperproliferation of SMCs as well as a decreased SMC content in the aortic wall of Fibulin-4 deficient mice 12,31 . The hyperproliferation of SMCs was specifically found in the adventitial layers of the aortic wall of newborns. Tsai et al. showed that TGF-β can transform from an inhibitor to a stimulant of SMC proliferation in the context of elevated SMAD3 32 . We observed a gradual increase in TGF-β signalling in Fibulin-4 deficient SMCs, which could be reverted by inhibition of TGF-β . These data indicate that the proliferation of Fibulin-4 deficient SMCs is reduced due to increased TGF-β signalling, thereby potentially contributing to aortic aneurysm formation.
ELISA analyses point to increased TGF-β 1 levels in Fibulin-4 deficient SMCs and strongly increased TGF-β 2 levels, also detected in plasma of Fibulin-4 R/R mice. The three TGF-β isoforms are involved in both overlapping and divergent roles. While Tgf-β1 null mice develop an autoimmune-like inflammatory disease 33 and Tgf-β3 knockout mice show abnormal lung development and cleft palate 34 , Tgf-β2 knockout mice have multiple developmental defects, including cardiovascular, pulmonary, skeletal, ocular, inner ear and urogenital manifestations 35 . Tgf-β2 heterozygous mutations in patients result in a different phenotype compared to Tgf-β2 knock-out mice 15 . TGF-β 2 haplo-insufficiency predisposes for adult-onset vascular disease, including aortic tortuosity and dilation, cerebrovascular disease and mitral valve disease, which overlaps with the phenotype of Fibulin-4 deficient patients. The phenotype of the TGF-β 2 deficient patients also shows overlap with other TGF-β signalopathies including Marfan  syndrome, Loeys-Dietz syndrome, the aneurysm-osteoarthritis syndrome and similarly present with a paradoxical, probably compensatory, local increase in TGF-β 1 and TGF-β 2. Furthermore, increased Tgf-β2 expression has been detected in patients with the Loeys-Dietz syndrome 36 . The fact that TGF-β 2 haplo-insufficiency results in a cardiovascular phenotype and local increased TGF-β 2, stresses the potential importance of TGF-β 2 in the vasculopathy. For various TGF-β superfamily members it is known that their effects are very concentration dependent 37,38 . Very high or very low levels of these cytokines can have similar or opposite effects on cells. In addition, crucial in the regulation of TGF-β activity is its activation from the latent ECM-bound complexes. This might explain the, at first sight, contradictory findings. This also suggests that increased TGF-β 2 expression is part of a common pathophysiologic process involved in aortic aneurysm formation in these syndromes. Whether it is a direct or indirect consequence of Fibulin-4 deficiency has to be determined in further studies.
We observed that specifically the TGF-β 2 isoform is elevated and detected at higher levels in the conditioned medium from these cells and in vivo in aortic lysates and blood. TGF-β 2 differs in its receptor binding properties from TGF-β 1 and TGF-β 3. While TGF-β 1 and TGF-β 3 have a high affinity for binding to Tβ RII, TGF-β 2 primarily binds to the transforming growth factor type-III receptor (Tβ RIII), also called betaglycan, after which it presents the ligand to the Tβ RI-Tβ RII signalling complex 39 . Bee et al. showed a specific regulatory role for the TGF-β receptor-IIb (Tβ RIIb), an alternatively spliced variant of Tβ RII, in TGF-β 2 signal transduction. Tβ RIIb mutations result in TGF-β 2 dependent increased SMAD2 phosphorylation, which is involved in aortic aneurysm progression 40 . Human SMCs express Tβ RI, Tβ RII and Tβ RIII, while in SMCs derived from atherosclerotic lesions Tβ RII expression is decreased 41 . This indicates that alterations in TGF-β receptor expression probably contribute to the regulation of the TGF-β pathway. As our data point to markedly increased TGF-β 2 levels in Fibulin-4 deficient SMCs, analyses on TGF-β receptors on these SMCs might further clarify the process of TGF-β regulation and determine its role in the pathogenesis of Fibulin-4 associated aortic aneurysms.
Increased TGF-β levels or TGF-β signalling is associated with multiple diseases. Enhanced TGF-β signalling is known to mediate a pathologic increase in ECM secretion and deposition and is causative for fibrosis in multiple disorders throughout the body 42 . Overexpression of TGF-β 2 is likely to induce trabecular meshwork ECM deposition 43 and increased ECM deposition is also observed in aortic aneurysm formation. TGF-β 2 is also frequently overexpressed in malignant cancers, where it induces immunosuppression and stimulates metastasis formation 44 . TGF-β 2 expression can be targeted with antisense oligonucleotides, which are currently under investigation in clinical trials 45 . As inhibition with pan TGF-β neutralizing antibodies is likely to induce side effects, aortic aneurysms associated with increased TGF-β 2 might benefit from a TGF-β 2 specific intervention decreasing systemic side effects by targeting the other isoforms. In Marfan patients, mouse models for Loeys-Dietz syndrome and transverse aortic constriction (TAC), losartan treatment prevents aortic aneurysm formation accompanied by reduced TGF-β 1 levels in patients with Marfan syndrome, and reduced TGF-β 1 and TGF-β 2 levels in Loeys-Dietz syndrome and TAC mice 30,36,46,47 . Our data indicate that losartan could also serve as an important therapeutic agent. The exact mechanism how losartan treatment leads to reduced TGF-β signalling needs to be determined.
Fibulin-4 binds LTBP-1 with high affinity and therefore an important role for Fibulin-4 in the association of LTBP-1 with microfibrils is predicted. The large latent complex (LLC), which is formed by LTBP and LAP-bound TGF-β , is linked to microfibrils through binding of LTBP-1 to Fibrillin-1. Therefore, Fibulin-4 might be additionally involved in sequestering of the LLC through LTBP-1 binding 48 . We speculate that reduced Fibulin-4 levels lead to defective sequestering to the ECM and thereby increased free TGF-β 1 and TGF-β 2. In conclusion, these data show that SMC derived TGF-β 2 is associated with aortic aneurysm formation and levels decrease upon losartan treatment, which improves survival of Fibulin-4 deficient mice. Specific intervention on TGF-β 2 could provide more information on its role in the pathogenesis of aortic aneurysm formation. In vitro analyses on isolated SMCs provide the opportunity to determine the molecular link between Fibulin-4 and TGF-β pathway regulation, and to further unravel its role in aortic aneurysm formation.

Material and Methods
Animals. Mice containing the Fibulin-4 R allele were generated as previously described 12 . All mice used were bred in a C57BI/6J background and were kept in individually ventilated cages to keep animals consistently micro-flora and disease free. To avoid stress-related vascular injury, mice were earmarked and genotyped 4 weeks after birth. Animals were housed at the Animal Resource Centre (Erasmus University Medical Centre), which operates in compliance with the "Animal Welfare Act" of the Dutch government, using the "Guide for the Care and Use of Laboratory Animals" as its standard. As required by Dutch law, formal permission to generate and use genetically modified animals was obtained from the responsible local and national authorities. All animal studies were approved by an independent Animal Ethical Committee (Dutch equivalent of the IACUC).
Proliferation assay. Fibulin-4 +/+ , Fibulin-4 +/R and Fibulin-4 R/R SMCs were seeded in triplicate in 6 cm dishes (5000 cells/well) and allowed to attach overnight. Next, cells were treated with TGF-β neutralizing antibodies (kindly provided by Dr. E. de Heer, Leiden University Medical Centre, Dept. of Pathology 26,27 ) and counted every day using a Burker cell counting chamber. Medium was replaced every other day. The MTS proliferation assay was performed according to the manufacturer's instructions (Promega, Madison, USA). In short, SMCs were seeded in 96-well plates (1500 cells/well) and allowed to attach overnight. At day-1, -2 and -3 medium was changed to 100 μ l complete DMEM + 20 μ l MTS substrate and the metabolic activity of the cells was analysed by absorbance change at 490 nm after 2 hours. TGF-β response assay. TGF-β response in SMCs was determined using (CAGA) 12 − MLP− Luciferase promoter reporter construct 28 . This construct contains 12 palindromic repeats of the SMAD3/4 binding element derived from the PAI-1 promoter and was shown to be highly specific and sensitive to TGF-β . The assay was performed as described previously 29 . In short SMCs were seeded in 1% gelatin coated 24-well plates and allowed to attach overnight. Subconfluent cells were transfected using Lipofectamin 2000 (Invitrogen, Carlsbad, California, USA) according to the manufacturer's protocol. A β -galactosidase plasmid was co-transfected to correct for transfection efficiency. After 6 hours, medium was changed to DMEM containing 10% FCS and the cells were incubated for 24 hours. Next, cells were serum-starved overnight and stimulated with 5 ng/ml TGF-β 3 (kindly provided by Kenneth K. Iwata, OSI, Inc., New York, USA) in the presence or absence of 10 uM SB431542 (Tocris/R&D systems, Abington, UK) for 6 hours. After stimulation the cells were washed, lysed and luciferase activity was determined according to the manufacturer's protocol (Promega). β -Galactosidase activity in the lysates was determined using β -gal substrate (0.2 M H 2 PO 4 , 2 mM MgCl 2 , 4 mM ortho-nitrophenyl-phosphate, 0.25% β -mercaptoethanol) and measuring absorbance change at 405 nm. The luciferase count was corrected for β -galactosidase activity. The relative increase in luciferase activity was calculated versus controls. All experiments were performed at least three times in triplicate. To determine the transfection efficiency of Fibulin-4 +/+ , Fibulin-4 +/R and Fibulin-4 R/R SMCs they were transfected with a GFP plasmid as described above, trypsinised and fixed with 1% PFA. Subsequently, SMCs were analysed with flow cytometry for the percentage of GFP transfected SMCs compared to the total amount of SMCs.
Western blot analysis. Western blot analysis was performed as described before 50 . In short, equal amounts of protein (DC protein assay, Bio-Rad Laboratories, Hercules, CA, USA) were separated on 10% SDS− polyacrylamide gel electrophoresis under reducing conditions. Proteins were transferred to nitrocellulose membranes (Whattman, Dassel, Germany) and blocked with 5% milk powder in Tris-HCl buffered saline containing 0.05% Tween-20 (Merck, Darmstadt, Germany). After washing, blots were overnight incubated with rabbit anti-pSMAD2 (Cell signaling Technologies, USA) and rabbit anti-pSMAD3 (kindly provided by Dr. E. Leof, Mayo Clinic, Rochester, MN, USA) followed by horseradish peroxidase-conjugated secondary antibodies (all GE Healthcare, Waukesha, WI, USA). Detection was Scientific RepoRts | 5:16872 | DOI: 10.1038/srep16872 performed by chemoluminescence according to the manufacturer's protocol (Pierce, Rockford, IL, USA). Afterwards, blots were stripped and reprobed with mouse anti-β -actin antibodies as a loading control.
RNA isolation and real-time PCR. RNA was isolated using RNeasy Mini Kit according to the manufacturer's instructions (Qiagen, Hilden, Germany). RNA concentration and purity was determined spectrometrically. Complementary DNA synthesis was performed using random primers. cDNA samples were subjected to 40 cycles real-time PCR analysis using maxima SYBR Green qPCR Master Mix 2× (Fermentas, Vilnius, Lithuania) and primers shown in Table 1. Reactions were performed in triplicates for each sample. Product specificity was determined by melting curve analysis and gel electrophoresis. The average Ct values of the triple reactions were calculated for each gene and all values were normalized for cDNA content by Hprt expression. The levels of fold-change for each gene were calculated relative to the gene expression levels in baseline wild type SMCs. RNA isolated from HUVECs and fibroblasts were used as controls for the genes analysed.
TGF-β ELISAs. SMC conditioned medium was prepared by seeding the cells and growing them to subconfluence. Medium was changed to serum-free DMEM, containing antibiotics as described above, and incubated for 4 days. Samples were collected every day and frozen at − 20 until analysis. Lysates were prepared from aortic arches of 14-15 week old Fibulin-4 +/+ , Fibulin-4 +/R and Fibulin-4 R/R mice and protein amounts were determined (Pierce BCA protein assay kit, Thermo Scientific). Total TGF-β 1, TGF-β 2, and TGF-β 3 levels in CM samples, aortic arch lysates and plasma samples were determined by commercially available duo-sets (R&D Systems) using transient acidification as described before 51 . Statistical analysis. Data are presented as mean ± SEM. The non-parametric Mann-Whitney U-test and unpaired student's t-test were performed to analyse the specific sample groups for significant differences. The two-way ANOVA test was used to test significant differences between independent variables. A p-value of < 0.05 was considered to indicate a significant difference between groups. All analyses were performed using IBM SPSS Statistics version 20.0 and 22.0 (SPSS Inc., Chicago, IL, USA).