Forkhead box C2 Promoter Variant c.-512C>T Is Associated with Increased Susceptibility to Chronic Venous Diseases

Chronic venous disease (CVD) is one of the most prevalent yet underrated disorders worldwide. High heritability estimates of CVD indicate prominent genetic components in its etiology and pathology. Mutations in human forkhead box C2 (FoxC2) gene are strongly associated with valve failure in saphenous and deep veins of lower extremities. We explored the association of genetic variants of FoxC2 as well as FoxC2 mRNA and protein expression levels with CVD of lower limbs. We systematically sequenced the single coding exon, 5′ and 3′ flanking regions of FoxC2 gene in 754 study subjects which includes 382 patients with CVD and 372 healthy subjects. Four novel and three reported polymorphisms were identified in our cohort. Three variants in 5′ flanking region and one in 3′ flanking region of FoxC2 gene were significantly associated with CVD risk. FoxC2 mRNA in vein tissues from 22 patients was 4±1.42 fold increased compared to saphenous veins from 20 normal subjects (p<0.01). FoxC2 protein was also significantly upregulated in varicose veins compared to control samples. The c.-512C>T (rs34221221: C>T) variant which is located in the FoxC2 putative promoter region was further analyzed. Functional analysis of c.-512C>T revealed increased mRNA and protein expression in patients with homozygous TT genotype compared to heterozygous CT and wild CC genotypes. Luciferase assay indicated higher transcriptional activity of mutant compared to wild genotype of this variant. These findings suggested that c.-512C>T variant of FoxC2 was strongly associated with susceptibility to CVD and also that this variant resulted in FoxC2 overexpression. To obtain a mechanistic insight into the role of upregulated FoxC2 in varicosities, we overexpressed FoxC2 in venous endothelial cells and observed elevated expression of arterial markers Dll4 and Hey2 and downregulation of venous marker COUP-TFII. Our study indicates altered FoxC2-Notch signaling in saphenous vein wall remodeling in patients with varicose veins.


Introduction
Chronic venous disease (CVD) of the lower extremities is one of the most prevalent diseases worldwide though the prevalence estimates differ extensively due to the different disease evaluation methods. CVD comprises of visible venous disorders which are not associated with an identifiable mechanism of venous dysfunction [1]. They are manifested by a variety of signs ranging from telangiectasis and varicose veins to venous ulceration. CVD is generally termed as varicose veins, that being the most common form of clinical manifestation. The great saphenous vein and its tributaries are the anatomic sites frequently affected in CVD of lower limbs [2]. Structural failures of vein such as valve weakness or vein wall dilatation may result in venous retrograde flow in limb leading to distal high venous pressure causing CVD. The primary events resulting in valvular incompetence and primary vein wall changes are not yet elucidated.
Several risk factors contribute to the progression of CVD [3,4,5]. The major risk factors reported are age, sex, pregnancy, family history and life style factors such as occupations which demand prolonged-standing. Evaluations of family history of CVD revealed a high and consistent heritability estimate in this disease [6,7]. Reports suggest that a risk of developing CVD for children with unaffected parents was only 20%. The risk with one affected parent is 25-62% and with both parents suffering with CVD the risk is 90% [6]. These data suggest the presence of genetic components in developing CVD, yet the precise genetic nature and genes involved in the pathogenesis of CVD is not known.
A twin cohort study indicated a link between varicose veins to microsatellite marker D16S520 on chromosome 16 [8]. This chromosomal region contains several genes coding for forkhead box family of proteins such as FoxC2 and FoxF1. FoxC2 gene (FKHL14, MFH1) is located 80 kb distant from this marker. It was also reported that homozygous null mice of FoxC2 (foxc2 2/2 ) had abnormal lymphatic vascular patterning and malfunctioned blood vessels [9]. Even if it is well proved that FoxC2 is a transcription factor involved in cardiovascular development signaling [10] and lymphangiogenesis [11], its exact mode of action in vascular development is yet to be elucidated.
FoxC2 gene variants are strongly associated with lymphedema distichiasis (LD) syndrome where majority of patients develop varicose veins [12]. FoxC2 gene is also implicated in the pathogenesis of saphenous vein and deep vein reflux [13]. Yet there have been no further studies on FoxC2 genetic variants in patients with varicose veins. We investigated the role of FoxC2 genetic variants in the development of CVD of lower limbs in a case-control study. We quantified mRNA and protein expression level of FoxC2 gene in saphenous vein from patients with varicose veins and healthy subjects. FoxC2 expression was highly upregulated in varicose vein tissues compared to normal control veins. Our results demonstrate significant correlation between c.-512C.T, a promoter variant of FoxC2 and the expression levels of

Ethics statement
The study was approved by the human ethics committees of Rajiv Gandhi Centre for Biotechnology, Thiruvananthapuram and all collaborating hospitals. Samples were collected from patients and healthy subjects after obtaining informed written consent.

Subjects and Specimen Collection
382 patients with CVD and 372 control subjects were recruited for this study. Whole blood samples (5 ml) were collected from 360 patients (age -18 to 60 years) with CVD from St.Thomas Hospital, Kerala, India. Diagnosis of CVD was based on physical examination and Doppler ultrasound test. CVD resulting from obstructions such as neoplasm were excluded from the study. Differential diagnosis was performed by an experienced vascular surgeon and presence of distichiasis was ruled out by an ophthalmologist. Patients with type 2 diabetes mellitus were also excluded since genetic variants of FoxC2 have been reported to result in susceptibility to diabetes mellitus. Blood samples were collected from age and gender matched 352 healthy controls with no known family history for CVD.
For tissue level expression analysis, varicose vein tissue samples were collected from 22 patients admitted for treatment of CVD by operative treatments at Kempegowda Institute of Medical Sciences, Bangalore, India. Saphenous control vein samples from 20 patients who underwent coronary artery bypass graft surgery at Sri Jayadeva Institute for Cardiovascular Sciences & Research, Bangalore, India were also collected for the study. Whole blood samples were also collected from these 22 patients and 20 controls for sequencing assays. Relevant data regarding the clinical characteristics of patients were collected from medical records of the hospitals participating in the study.

Data analysis
Demographic data of all study participants and information regarding symptoms such as pain, itching and throbbing sensation in legs and clinical signs such as hemorrhage, lower limb oedema, hyperpigmentation, thrombophlebitis, cellulitis and ulceration were collected for each patient from medical records. Family history, occupational and lifestyle data were collected to examine their influence in aggravating disease manifestation. Disease phenotypes were categorized according to CEAP (clinical severity, etiology, anatomy and pathophysiology) classification system [14]. Varicose veins without odema or pigmentation were classified under C2. Only 2.9% of all our patients were in CEAP Class 3 in which varicose vein with oedema alone are found. The patients in this study were mostly from CEAP Class 4, 5 and 6 who presented various clinical signs such as pigmentation, ulceration along with oedema due to CVD.  Genomic DNA and mRNA extraction Genomic DNA from whole blood samples was extracted using QIAamp DNA blood mini kit (Qiagen, USA) according to the manufacturer's instructions. Genomic DNA and mRNA from vein tissues were extracted by All Prep DNA/RNA/Protein mini kit (Qiagen, USA). Quantification and purity of DNA and mRNA was measured by nanodrop-1000 spectrophotometer (Thermo-Scientific, USA) at 260 nm. RNA was further treated with DNase1 (Ambion, USA) for removing any DNA contamination.

PCR DNA sequencing
A touch-down PCR was performed to amplify the single coding exon, 3 kb of 59 flanking and 200 bp of 39flanking region which includes the 59 and 39 untranslated regions (UTR) of FoxC2 gene from DNA of patients with CVD and healthy subjects. Nine primer pairs (Table S1) to amplify overlapping regions of FoxC2 gene and flanking regions were designed using Primer Premier 5 software (PREMIER Biosoft International, USA). PCR conditions were as follows: Initial denaturation for 5 min at 96uC, 20 cycles of denaturing at 96uC for 30 sec, annealing at 70uC for 40 sec with a touchdown of 0.5uC per cycle and extension at 72uC for 1.5 min. This was followed by 20 cycles at same conditions except that annealing was at 60uC for 40 sec. PCR products were purified using gel band elution kit (GE healthcare, USA). DNA sequencing was carried out on an ABI 3100 DNA analyzer with Bigdye terminator chemistry (Applied Biosystems, USA).

Gene expression analysis of FoxC2 by qRT-PCR
Total RNA (500 ng) from each tissue sample was subjected to reverse transcription with oligodT, dNTPs, and M-MLV reverse transcriptase (Promega, USA). Primers for FoxC2 and GAPDH genes were designed for real time PCR analysis (Table S2). Quantitative RT-PCR was carried out as reported earlier [15]. The temperature conditions were as follows: 48uC, 30 min; 95uC, 10 min; followed by 40 cycles of 95uC,15 s; and 60uC, 1 min and analyzed using ABI Prism 7900HT sequence detection system (Applied Biosystems, CA). Values were normalized with GAPDH mRNA levels. A single peak was observed in the dissociation curve for both genes confirming the specificity of PCR products. Real time mRNA fold change was calculated by the formula, 2 2DDCt [2 (Ct value of target gene-Ct value of control gene) ].

FoxC2 protein expression analysis by western blot
Frozen vein tissues were homogenized and incubated in ice-cold RIPA buffer with protease inhibitor cocktail for 90 minutes followed by centrifugation at 15,000 g for 20 min at 4uC to collect   Immunostaining of FoxC2 antigen in tissue specimens 5 mm paraffin embedded tissue sections were de-paraffinized in xylene and endogenous peroxidase activity was quenched with 3% H 2 O 2 in methanol by incubating for 30 minutes. Sections were rehydrated through graded alcohols and antigen retrieval was performed by incubating in 10 mM sodium citrate at 90uC for 15 mins. Sections were washed with TBST and then blocked with 3% BSA for 20 mins. Slides were incubated with anti-FoxC2 antibody (Abcam cat # ab65141) diluted with TBS in 1:100 ratio. Slides were washed thrice for 5 minutes in TBST and incubated for 1 hour with Horse raddish peroxidase conjugated anti rabbit antibody (Abcam cat # ab97051) diluted with TBS in 1:200 ratio. After washing, slides were incubated with 3,39-diaminobenzidine tetrahydrochloride (DAB) (Sigma) and immediately washed under tap water after the color development and were counter stained with haematoxylin. Slides were DPX mounted and observed under light microscope (Nikon, Japan).

Cloning and reporter gene assay
To elucidate the potential effect of c.-512C.T variant of FoxC2 gene in the promoter activity of gene, a luciferase reporter gene assay was performed. A 664 bp DNA sequence from the 2614 bp to +50 bp from the transcription start site of FoxC2 gene encompassing the putative promoter area was amplified from DNA with CC and TT genotype. Primers including KpnI and HindIII restriction sites (Table S2) were used for the amplification. Amplicons were cloned into the KpnI and HindIII sites of the pGL4-basic vector (Promega, USA) and the constructs were verified by DNA sequencing.
EA.hy926 [16] cells were used for expressing the cloned constructs. The cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) containing 10% FBS in a humidified CO 2 chamber at 37uC. 0.8 mg of constructs such as pGL4-Basic, pGL4-FoxC2 promoter (CC genotype) and mutant pGL4-FoxC2 promot- FoxC2 in Chronic Venous Disease PLOS ONE | www.plosone.org er (TT genotype) were transfected into cultured cell lines using Lipofectamine (Invitrogen, USA). Renilla luciferase construct (phRL-TK) was used as a control for transfection efficiency. After 48 h, each group of cells was lysed and luciferase assay was carried out using Dual Luciferase assay kit (Promega, USA) according to manufacturer's instructions and readings were recorded. Technical replicates were performed in triplicate and biological experiments were performed twice.
FoxC2 construct and transfection of EA.hy926 cells FoxC2 -pCAGIG construct was made by inserting FoxC2 coding sequence into EcoRI and XhoI restriction sites of pCAGIG mammalian expression vector (Addgene plasmid 11159) [17]. EA.hy926 cells were plated into 6-well plates and the cells were allowed to adhere for 24 hours. Transfection of FoxC2 -pCAGIG and control empty vector was performed using lipofectamine-2000 (Invitrogen, CA) according to the manufacturer's recommendation. The concentrations of constructs used were 1 mg per well. After 6 hours of transfection, 20% FBS supplemented DMEM medium was added. The assays were carried out 8 days posttransfection. Transfection was performed in triplicates and repeated twice. GFP positive cells were sorted on FACS Aria (Becton Dickinson) and used for further analysis.

Total RNA isolation and qRT PCR analysis
Total RNA from untransfected and transfected EA.hy926 cells was isolated by Allprep RNA/protein kit (Qiagen, USA). After the reverse transcription reaction as described earlier, cDNA was used for quantitative real time PCR for FoxC2, GAPDH, Hey2, Dll4, COUP TFII and Ephrin B4 gene expression. Primers sequences are given in table S2. The conditions for amplifying FoxC2 and GAPDH are as described earlier. For assessing Hey2, Dll4, COUP TFII and Ephrin B4 gene expression, the reactions were performed in triplicate in 96-well plates at 48uC, 30 min; 95uC, 10 min; followed by 40 cycles of 95uC, 15 s; and 61uC, 1 min. The realtime PCR products were re-confirmed by electrophoresis on 2% agarose gels. The amount of the target relative to GAPDH mRNA was expressed as 2 2(DCt) .

Statistical analysis
Hardy-Weinberg equilibrium was tested for a goodness-of-fit using a Chi square test. Chi-square test was used to investigate the possible association between polymorphisms and CVD in casecontrol studies. Student's t test was used to analyze the difference in luciferase, mRNA transcripts and protein expression levels. Information collected from answered questionnaires and medical records were entered into MS Excel and analyzed using SPSS 16 (SPSS Inc., USA). Differences between groups were considered significant for p values less than 0.05.

Results
Patients with CVD (n = 382) and healthy subjects (n = 372) had a mean age of 48.17 (SD = 13.37) and 37.86 (SD = 13.63) respectively (p,0.001). 53.7% of patients were females and a positive association was seen between female gender and CVD (p = 0.035). A detailed analysis of patients and controls with respect to different age groups and gender is given in table 1. To reduce a possible interference of these confounding variables in the current polymorphism analysis, we used adjusted odds ratio (AOR) with 95% confidence intervals (CI) estimated by multiple logistic regression models in each analysis. Clinical features of the patients with CVD are given in table 2.   Allelic frequencies of these four polymorphisms also differed significantly between patients with CVD and controls (Table 5). Only these four polymorphisms were included in further analysis.
To understand the collective effect of these four significant polymorphisms in the disease, we further classified study subjects into two groups. Subjects with none or either one FoxC2 variant were combined in one group. The second group comprised of subjects with two or more polymorphisms in their FoxC2 gene and flanking sequences. Notably, the second group had 7.20 fold (95% CI: 4.99, 10.39) risk for CVD compared to first group (table S3). DNA was isolated from vein specimens and sequenced to check any genotypes discrepancy between whole blood samples and tissues of same patients. The genotype profiles obtained were similar in both the DNA samples from same patients.
Correlation of FoxC2 genotypes with FoxC2 mRNA transcript levels FoxC2 transcript expression was 461.4 folds increased in venous tissues from patients compared to normal subjects (p,0.05) (Figure 1a). Patients with homozygous mutant TT genotype had higher venous expression of FoxC2 mRNA compared to patients carrying heterozygous CT genotype (p = 0.02) and wild CC genotype (p,0.001) (Figure 1b). The upregulation of FoxC2 in tissue specimens (either at mRNA or protein level) was not significantly altered in patients who had all the four polymorphisms compared to four patients who carried TT genotype of c.-512C.T variant alone (p = 0.65) ( Figure S2). Correlation of FoxC2 genotypes with FoxC2 protein expression levels Densitometry analysis of immunoblots indicated a significant upregulation of FoxC2 protein in varicosed tissues compared to control (Figures 2a and 2b). Correlation of densitometry results of FoxC2 protein expression with FoxC2 genotypes revealed significantly higher protein levels in patients carrying TT genotype compared to patients having heterozygous CT or wild CC genotypes (p,0.01) (Figures 2c and 2d).

Immunostaining of FoxC2 antigen in tissue sections
To confirm the overexpression of FoxC2 in varicose veins we performed an immunostaining on tissue sections of varicose veins from 3 patients and control vein from 3 healthy subjects. There was a marked overexpression of FoxC2 protein in the tissue sections of varicose veins compared to control veins ( Figure 3). Expression of Hey2, Dll4, COUP TFII and Ephrin B4 in FoxC2-Ea.hy926 cell constructs Based on the reports of a potential FoxC2-Notch signaling [18], we focused on the expression patterns of Notch ligand Dll4 and arterial marker Hey2 in the FoxC2 construct transfected Ea.hy926 cells. Expression of venous specific markers such as COUP TFII and Ephrin B4 were also assessed. Ea.hy926 cells were transiently transfected with pCAGIG vector containing FoxC2 construct (FoxC2-pCAGIG) or empty pCAGIG vector and the expression of Hey2 and Dlll4 genes were determined at mRNA levels. FoxC2 activation upregulated (p,0.05) the expression of both these arterial fate specific markers in Ea.hy926 when compared to empty vector transfected cells ( Figure 5). FoxC2 overexpression resulted in a significant downregulation (p = 0.003) of COUP TFII, while Ephrin B4 downregulation was not statistically significant.

Discussion
Human FoxC2 is a forkhead/winged helix transcription factor coding gene located on chromosome 16q24.1 (NCBI Reference Sequence No: NG_012025.1). FoxC2 is implicated in vascular development especially in arterial and lymphatic differentiation [19]. Foxc2 deficiency in mouse results in abnormal lymphatic patterning and failure of lymphatic valve formation [9]. Mutations in the FoxC2 coding sequences were reported in patients with lymphoedema-distichiasis (LD) which is an autosomal dominant form of primary lymphoedema with majority of patients developing varicose veins in lower limbs [12]. Several studies have associated mutations in FoxC2 gene with LD risk and suggested a role of FoxC2 in the pathogenesis of varicose veins [12]. Mellor et al linked FoxC2 mutations to venous valve failure and reflux using conventional colour Doppler duplex ultrasound in patients with lymphodema distichiasis [13]. The role of FoxC2 gene however has not yet been well-defined in patients with varicose veins or CVD. We report for the first time a positive association between genetic variants of FoxC2 and chronic venous diseases and a mechanistic insight on the role of FoxC2 in pathogenesis of CVD.
FoxC2 polymorphisms and abnormal protein expression have been implicated with insulin sensitivity in patients with obesity and diabetes mellitus [20]. We hence excluded patients with diabetes and obesity from this study population to get an unbiased data of FoxC2 polymorphism pattern in patients with CVD alone. Patients with lymphoedema distichiasis were also not included in our subjects.
We initially sequenced the 1.5 kb single coding exon of FoxC2 gene from DNA isolated from whole blood samples of 382 patients with CVD and 372 control subjects. DNA sequencing revealed the presence of only two rare synonymous variants, c.354C.T (novel variation) and c.426G.A (rs41312258) with a frequency of 0.02 of 0.04 respectively. Hence these variations were not followed up for further studies. Variants such as c.-2647A.T and c.-1538A.G were not further experimentally validated as they lacked the putative binding sites for transcription factors. Transcription factor binding affinity was evaluated by TF SEARCH version 1.3 computer program (http://www.cbrc.jp/research/db/TFSEARCH. html) [21]. C.*126G.A variant is located 126 bp downstream to translation termination codon and 35 bp downstream to 39UTR sequence of FoxC2 gene. C.*126G.A was consistently present in either heterozygous GA or wild GG genotypes but never in homozygous mutant AA genotype in our cohort. Prediction of microRNA/target duplexes for C.*126G.A variant was analyzed by miRNA prediction tools (http://bibiserv.techfak.uni-bielefeld. de/rnahybrid) [22] and miRBase database [23]. Even though a putative binding site for Has-mir-4732-5p was obtained at this variant's nucleotide position from miRBase database, further in silico analysis by RNAhybrid tool gave a very weak binding probability. C.-512C.T variant is present in the highly conserved proximal promoter of the FoxC2 gene [20]. This variant can possibly alter transcription factor binding and subsequent gene expression and hence was selected for further tissue centric expression analysis.
FoxC2 mRNA and protein were over expressed in vein tissues of patients with CVD compared to normal saphenous vein specimens. The FoxC2 mRNA transcript and protein upregulation in vein tissues positively correlated with the presence of TT genotype of c.-512C.T polymorphism in all the patients with CVD. Our observations are in concordance with an earlier report that variations outside the forkhead domain of FoxC2 result in a gain of function [24]. A slight increase in gene expression was observed with reporter luciferase assays using mutant construct (p = 0.051) which indicates the contribution of other polymorphisms and factors in this upregulation as well. Since this is an initial study with 754 subjects, further studies in multiple cohorts is essential to verify our conclusion.
FoxC1 and FoxC2 transcription factors promote arterial specification during vascular development by acting upstream of Notch [19]. Arterial specific markers such as Dll4 and Hey2 were found overexpressed and venous marker COUP TFII was found downregulated in vein endothelial cells transfected with FoxC2 overexpressing mammalian construct. Our observations support the earlier reports on Hey2 and Dll4 based inhibition of Coup TFII in vitro [25]. As Hey2 is an important regulator of smooth muscle proliferation [26], we assume an altered FoxC2-Notch signaling in vein wall thickening in varicose veins. While arterial markers, Hey2 and Dll4 expression was upregulated in RNA samples from patients with CVD and controls ( Figure S1), venous markers did not show any differential expression in RNA samples from patients with CVD and controls (data not included).
Taken together, our results suggest c.-512C.T variant can contribute to the upregulation of FoxC2 in vein tissues. This possibly triggers an altered FoxC2-Notch signaling cascade which results in the remodeling of saphenous vein in patients with CVD. Figure S1 Hey2 and Dll4 mRNA expression in vein tissues of patients with CVD and healthy subjects.    Table S3 Patients and controls categorized as first group with neither or any single polymorphism and second group with 2, 3 or more polymorphisms. Percentages were taken from the column totals. Chi-square test for measure of association was used to derive p value. a Odds ratio and 95% confidence intervals of individual polymorphisms. b Adjusted odds ratio and 95% confidence intervals is obtained adjusting for age group and sex in multiple logistic regression model. (DOC)