Post-translational modifications regulate matrix Gla protein function: importance for inhibition of vascular smooth muscle cell calcification


L.J. Schurgers, VitaK, University Maastricht, Universiteitssingel 50, 6200MD Maastricht, the Netherlands.
Tel.: +31 433881680; fax: + 31 433884160; e-mail:


Summary. Background: Matrix Gla protein (MGP) is a small vitamin K-dependent protein containing five γ-carboxyglutamic acid (Gla) residues that are believed to be important in binding Ca2+, calcium crystals and bone morphogenetic protein. In addition, MGP contains phosphorylated serine residues that may further regulate its activity. In vivo, MGP has been shown to be a potent inhibitor of vascular calcification; however, the precise molecular mechanism underlying the function of MGP is not yet fully understood. Methods and results: We investigated the effects of MGP in human vascular smooth muscle cell (VSMC) monolayers that undergo calcification after exposure to an increase in Ca2+ concentration. Increased calcium salt deposition was found in cells treated with the vitamin K antagonist warfarin as compared to controls, whereas cells treated with vitamin K1 showed decreased calcification as compared to controls. With conformation-specific antibodies, it was confirmed that warfarin treatment of VSMCs resulted in uncarboxylated (Gla-deficient) MGP. To specifically test the effects of MGP on VSMC calcification, we used full-length synthetic MGP and MGP-derived peptides representing various domains in MGP. Full length MGP, the γ-carboxylated motif (Gla) (amino acids 35–54) and the phosphorylated serine motif (amino acids 3–15) inhibited calcification. Furthermore, we showed that the peptides were not taken up by VSMCs but bound to the cell surface and to vesicle-like structures. Conclusions: These data demonstrate that both γ-glutamyl carboxylation and serine phosphorylation of MGP contribute to its function as a calcification inhibitor and that MGP may inhibit calcification via binding to VSMC-derived vesicles.


Vascular calcification is a common feature of atherosclerosis, aging, uremia and diabetes, and it is clearly associated with an increased risk of morbidity and mortality [1–3]. In the aged human population, arterial calcification is almost ubiquitous, both in the media and in the intima, where it occurs in association with atherosclerotic lesions [4]. For decades it was thought that ectopic calcification was a passive end-process, having no clinical relevance. However, it is now clear that the development of calcium deposits in the vasculature is an active and highly regulated process with similarities to bone formation.

Matrix Gla protein (MGP) is a vitamin K-dependent protein expressed by vascular smooth muscle cells (VSMCs) [5,6]. MGP is thought to be an inhibitor of vascular calcification, and evidence for this stems from its Ca2+ -binding γ-carboxyglutamic acid (Gla) motif [7,8], inhibition of cartilage calcification [9], binding and antagonizing BMP-2 [10,11], and work with MGP-knockout mice [12]. These mice developed severe vascular calcification that resulted from replacement of the medial VSMCs by chondrocyte-like cells, accompanied by deposition of a calcified matrix. Recently, the same group showed that restoration of MGP expression in VSMCs rescued the calcification phenotype [13].

The function of MGP depends on a post-translational modification by which specific glutamate residues are converted into Gla residues by a vitamin K-dependent carboxylase (Table 1). Carboxylation of Gla proteins is essential for their function [14]. Vitamin K hydroquinone (KH2) is used as the active cofactor in the carboxylation reaction. Upon oxidation of KH2, the released energy is used to introduce an extra carboxyl group at the γ-position of a glutamate residue. Oxidized vitamin K can be reused after reduction by vitamin K epoxide reductase (VKOR). The synthesis of Gla proteins can be blocked by 4-hydroxycoumarin derivatives such as warfarin, which inhibit the VKOR enzyme. Price et al. [15,16] showed that warfarin induced calcification of the elastic lamellae in arteries and heart valves of rats within 3–5 weeks through inhibition of the γ-glutamyl carboxylation of MGP.

Table 1.   Sequences and molecular masses of used peptides Thumbnail image of

The function of the serine residues at positions 3, 6 and 9 (Table 1), representing the sites for potential phosphorylation, is still unclear. The recognition motif found for phosphorylation at these residues in MGP, Ser-X-Glu/Ser(P), has been seen in a number of regulatory peptides [17], and the formation of phosphoserine (pSer) residues was proposed as a mechanism for regulation of activity of proteins secreted in the extracellular matrix. Another pSer-containing protein is osteopontin, a non-collagenous glycosylated protein associated with biomineralization in osseous tissues as well as ectopic calcification. The calcification inhibitory function of osteopontin depends on the phosphorylation of a number of well-defined serine residues [18,19].

VSMCs in culture constitutively express MGP, and MGP mRNA expression is upregulated in calcifying VSMC nodular cultures [20]. MGP mRNA expression was also upregulated in response to an increase in Ca2+ concentrations, a mechanism thought to promote inhibition of calcification [21]. However, MGP may not always be able to inhibit calcification, as was shown by immunohistochemical studies where MGP protein was abundant at sites of calcification in human arteries [22,23]. More recently, it was demonstrated that MGP in the γ-carboxylated form was present in normal arteries, whereas uncarboxylated MGP (ucMGP) was only detected in atherosclerotic plaques in areas of calcification [24]. This raised the possibility that the local vitamin K stores are too low to ensure full carboxylation of upregulated MGP. In addition, in vitro experiments have suggested that MGP may be able to stimulate calcification: blocking the function of MGP using a polyclonal antibody against the C-terminus of MGP reduced pericyte calcification [25]; and addition of exogenous MGP to calcifying VSMCs in high concentrations led to an increase in calcification, possibly via regulation of BMP-2 function [26].

To address the question of whether endogenous MGP carboxylation or exogenous MGP species may affect human VSMC calcification, and to obtain insights into the importance of the various post-translational modifications for MGP activity, we have synthesized peptides similar to MGP domains containing or not containing the various unusual amino acid residues Gla or pSer. We have tested these peptides for their effects on VSMC calcification. In addition, the contribution of endogenously expressed MGP was tested in the presence of vitamin K or its antagonist, warfarin.

Materials and methods


All reagents were of analytical grade or better and were obtained from commercial suppliers. Phylloquinone (vitamin K1) and warfarin were obtained from Sigma (St Louis, MO, USA). Full-length carboxylated (five Gla residues) MGP (cMGP) and ucMGP (no Gla residues) were made using solid-phase peptide chemistry and native chemical ligation [8] of cMGP residues 1–53 (cMGP1–53) and the C-terminal fragment MGP54–84. All other peptides were obtained from Pepscan (Lelystad, The Netherlands), and included: cMGP35–54, ucMGP35–54, the phosphorylated (three pSer residues) pMGP3–15, and the non-phosphorylated (no pSer residues) peptide npMGP3–15. All synthetic peptides that did not contain the full-length N-terminal tyrosine or C-terminal lysine residues were either N-acetylated or C-amidated or both (see Table 1 for peptide details). Peptides were biotinylated using an EZ-Link NHS-Biotin kit according to the manufacturer’s instructions (Pierce, Etten-Leur, The Netherlands). Synthetic full-length osteocalcin in its carboxylated (three Gla residues) form was used as a negative control and was also purchased from Pepscan. The amino acid sequences of synthetic peptides and proteins were similar to the sequences found in the corresponding native human proteins. The preparation of conformation-specific antibodies and the specificity against MGP has been described elsewhere [24,27].

Cell culture

The culture medium used was M199 (Gibco, Breda, The Netherlands) buffered with 3.7 g L–1 NaHCO3 and supplemented with 105 IU L–1 penicillin (Sigma), 100 g L–1 streptomycin, 250 mg L–1 amphotericin, and 4 mm l-glutamine (Sigma), and incubated at 37 °C in humidified air with 5% CO2. Heat-inactivated fetal bovine serum (FBS) was purchased from Gibco and contained approximately 30 pm vitamin K1. Human VSMCs were obtained from non-atherosclerotic areas of aortas from organ donors of various ages (both males and females; age range 15–65 years). The cells were prepared from explants of aortic tissue, and were identified as smooth muscle cells by positive staining with monoclonal antibodies (mAb) against α-SM actin (A2547, Sigma). Cells were maintained in M199 medium containing 20% (v/v) FBS and were used between passages 3 and 15. At least three different isolates from individuals were used in each experiment.

VSMC calcification model

VSMCs were seeded at a density of 10 000 cells per well in 24-well plates (Falcon, Breda, the Netherlands) and incubated in M199 medium containing 20% (v/v) FBS in the presence of vitamin K1 or warfarin (all 10 μm). After 72 h, VSMCs were incubated with 1.8 mm CaCl2 (control) or extra CaCl2 (3.6 mm) in the presence of vitamin K1 or warfarin for 48 h. Synthetic MGP peptides were added to monolayer cultures 1 day after seeding cells, together with 3.6 mm CaCl2 and 20% FBS/M199, and incubated for 48 h. As a cell-free negative control, collagen-coated wells (50 mg L–1, Sigma) were used.

Detection of calcification

Von Kossa and Alizarin Red staining were performed as described previously [20]. Calcification of VSMC monolayers was measured by the cresolphthalein method or using atomic absorption spectrometry (Department of Clinical Chemistry, University Hospital Maastricht, The Netherlands). Protein concentrations were determined using a BCA kit from BioRad (Veenendaal, The Netherlands).

Scanning electron microscopy (SEM) and energy dispersive X-ray (EDX) analysis

VSMCs were washed in ice-cold de-ionized water, frozen in melting propane, and lyophilized. SEM and EDX were performed as described previously [20].

Western blotting

Samples were applied to a 14% polyacrylamide gel in 8 m urea. Gels were transferred to Immobilon-P (Millipore, Bedford, MA, USA) by using an electroblotting system (Bio-Rad, Veenendaal, the Netherlands). After blotting, membranes were incubated with mAb against either npMGP3–15 (α-npMGP3–15) or ucMGP35–53 (α-ucMGP35–53). Antibodies bound to the membrane were detected with a second antibody [rabbit anti-(mouse IgG)] conjugated with horseradish peroxidase, and the immune complexes were visualized by enhanced chemiluminescence (Amersham Pharmacia, Uppsala, Sweden).

Cellular localization of added biotinylated peptides

Cells were grown on glass coverslips in the presence of 20% FBS/M199 medium. At confluence, the CaCl2 concentration in the medium was increased to 3.6 mm, and biotinylated MGP peptides were added, after which the cells were grown for another 48 h. After being washed twice, the cells were fixed by incubation in 4% (v/v) formaldehyde in phosphate-buffered saline (PBS; 10 mm sodium phosphate, 0.14 mm NaCl, pH 7.4) for 45 min at 4 °C and treated either with or without permeabilization [0.5% (v/v) Nonidet P40 and 0.1% (v/v) Triton X-100 in PBS]. Glass coverslips were incubated with blocking buffer (1% bovine serum albumin, 0.1% Nonidet P40 in PBS) for 1 h at room temperature. After washing twice with PBS, streptavidin–fluorescein isothiocyanate (FITC) (Dako, Heverlee, The Netherlands) was added in blocking buffer for 1 h at 37 °C. After stringent washing with PBS, the coverslips were mounted in Mowiol (Calbiochem, Darmstadt, Germany) containing 1 mg L–1 of 4′,6-deamidino-2-phenylindole (DAPL). Samples were viewed using a Zeiss axiovert 10 microscope with a plan-APOCHROMATE 63 × /1.40-oil immersion lens and equipped with a Digital Pixel Instruments 12-bit CCD camera. Images were captured using IP Lab Scientific Imaging Software (Scanalytics, Fairfax, VA, USA). Additionally, a Zeiss LSM 410 confocal laser scanning microscope was used (63 × /1.40-oil immersion lens) for imaging MGP within the cells.

Statistical analysis

The Mann–Whitney test was used to test for differences between groups. Results are expressed as mean ± SD.


Effect of γ-glutamyl carboxylation of endogenous MGP on human VSMC calcification

To examine the role of γ-glutamyl carboxylation in human VSMC calcification, either vitamin K1 or warfarin was added to VSMC cultures (Fig. 1A). Extracellular Ca2+ levels were raised to 3.6 mm, resulting in heavy deposits of diffuse calcification within 48 h, as determined by Alizarin Red and Von Kossa staining (Fig. 1B,C). The composition of the calcium crystals was similar to calcium apatite, and calcification occurred as cell- and matrix-associated vesicle-like structures (Fig. 1D,E). This model of calcification is similar to that described previously in human VSMCs [28,29] and osteoblasts [30]. In addition, VSMC calcification induced by elevated Ca2+ concentrations was increased by 10 μm warfarin, whereas 10 μm vitamin K1 inhibited calcification as compared with controls (containing 30 pm vitamin K1; see Fig. 1G). This suggested that in both models of calcification, fully carboxylated vitamin K-dependent proteins, such as MGP, inhibit VSMC calcification.

Figure 1.

 Effect of warfarin and vitamin K on vascular smooth muscle cell (VSMC) monolayer calcification, quantitated by the cresolphthalein method. Human VSMCs in typical ‘hill and valley’ morphology but not yet in nodules (referred to here as monolayers) were incubated in either control medium containing 1.8 mm Ca2+ (A) or medium containing 3.6 mm Ca2+ (B, C) in 20% fetal bovine serum for 48 h. The cultures treated with 3.6 mm Ca2+ were stained for calcification using the Alizarin red (B) or von Kossa technique (C). The scanning electron micrograph image (D) shows that vesicular structures are produced by VSMCs exposed to 3.6 mm Ca2+. Arrows indicate vesicular structures on or within the VSMCs (filled arrows) and extracellular matrix (open arrow), and the backscattered image (E) suggests that these vesicles contain crystalline material (scale bar represents 20 μm). The vesicle indicated by an arrow in (E) was analyzed for elemental content, and the graph in (F) shows that this area contained Ca, P and O, probably representing apatite mineral. In this model, warfarin (10 μm) increased calcification (< 0.01) and vitamin K1 level, and inhibited calcification (< 0.01), in comparison with VSMCs treated with 3.6 mm Ca2+ alone (control) (G), as measured using the cresolphthalein method. Each measurement represents three isolates, measured in triplicate (= 9; mean ± SD).

To determine whether endogenous MGP carboxylation was affected by these treatments, cell extracts were collected and analyzed for ucMGP and npMGP (the latter does not discriminate between cMGP and ucMGP). Cells treated with warfarin showed increased ucMGP content, whereas cells treated with vitamin K1 had lower levels of ucMGP (Fig. 2).

Figure 2.

 Effect of warfarin and vitamin K on matrix Gla protein (MGP) carboxylation. Vascular smooth muscle cells treated with warfarin showed upregulation of uncarboxylated MGP (ucMGP) at 14 kDa (most likely mature MGP) and 11 kDa (probably a truncated form of MGP), whereas the cells treated with vitamin K1 had low levels of ucMGP. When α-npMGP3–15 was used, no change was observed in warfarin or vitamin K-treated cells, indicating that overall MGP production was not affected. Equal amounts of protein were loaded onto the gel (10 μg), and measured using a Bio-Rad protein assay kit; Ponseau S staining was used to check blots for equal loading.

Effect of MGP-derived peptides on calcification of human VSMCs

To determine whether MGP affects cell calcification, synthetic MGP variants were added to human VSMCs cultured in 3.6 mm Ca2+. Full-length cMGP inhibited calcification in a dose-dependent manner, whereas ucMGP had no significant effect (Fig. 3A). As MGP is poorly soluble under physiologic conditions, we confirmed that MGP is still soluble in culture medium up to 200 nm (data not shown). The N-terminal peptide cMGP1–53 inhibited calcification, but a higher dose was needed than that of cMGP to obtain a significant effect, whereas the C-terminal peptide MGP54–84 had no effect on calcification (Fig. 3B). The cMGP35–54 and pMGP3–15 peptides inhibited calcification, whereas the corresponding peptides ucMGP35–54 and npMGP3–15 had no significant effect (Fig. 3C,D). In all calcification studies, cell-free negative controls (collagen; 50 mg L–1) were incubated in parallel, but in none of these cases did we find calcium salt precipitation (data not shown). Although it is difficult to quantitate the effects of the various peptides, it is obvious that the inhibitory effects of cMGP1–53 and cMGP35–54 are similar, but substantially lower, than that of full-length cMGP1–84, suggesting a contribution to MGP activity of the C-terminal domain. Moreover, the difference between the calcification inhibitory activity of pMGP3–15 and npMGP3–15 suggests that phosphorylation also contributes to the activity of mature MGP.

Figure 3.

 Effect of synthetic matrix Gla protein (MGP) peptides on vascular smooth muscle cell (VSMC) calcification as determined with atomic absorption spectrometry. VSMCs were incubated in 3.6 mm Ca2+ with or without MGP peptides, and calcification was measured. Synthetic full-length cMGP (○) inhibited calcification in a dose-dependent manner from 10 to 200 nm, as compared with controls (no MGP added) (200 nm, < 0.01; 50 nm, < 0.02; 10 nm, < 0.05), whereas uncarboxylated MGP (ucMGP) (□) had no significant effect on calcification (A). Carboxylated MGP (cMGP1–53) (○) inhibited calcification at 200 nm (< 0.02), whereas the C-terminal peptide MGP54–84 (□) had no effect on calcification (B). cMGP35–54 (○) inhibited calcification at 200 nm (< 0.02), whereas ucMGP35–54 (□) had no effect (C). Phosphorylated MGP3–15 (○) inhibited calcification (200 nm, < 0.01; 50 nm, < 0.05), whereas non-phosphorylated MGP3–15 (□) had no measurable effect (D). This indicates that the Gla and the phosphoserine domains are the active regions in inhibiting VSMC calcification. Each measurement represents four isolates, measured in triplicate (= 12; mean ± SEM).

Site of action of added synthetic peptides

The way in which MGP or MGP-derived peptides may modulate cellular activity may be based on either adhesion to the outer cellular/vesicular membrane or internalization by the cells or vesicles. In a first attempt to find their site of action, biotinylated MGP peptides were added to confluent VSMC monolayers in culture medium containing 1.8 or 3.6 mm Ca2+. Using FITC-labeled streptavidin with immunofluorescence and confocal microscopy, the FITC label was not detected within cells under normal or high-Ca2+ conditions, suggesting that the peptides had not been taken up by VSMCs (data not shown). However, at the extracellular surface, clear staining was observed during the high-Ca2+ treatment only, with most prominent staining of biotinylated cMGP1–84, cMGP1–53, cMGP35–54 and pMGP3–15 being seen (Fig. 4A–D), whereas ucMGP35–54 (Fig. 4E) and npMPGP3–15 (data not shown) did not bind. Gla-containing peptides were found to be present both at the cell surface and in association with vesicular structures (Fig. 4A–C, see arrows), whereas pMGP3–15 was found to be exclusively associated with vesicular structures (Fig. 4D). Biotinylated osteocalcin, a Gla protein specifically synthesized by osteoblasts in bone, did not bind to cells or vesicles under any condition (Fig. 4F).

Figure 4.

 Vascular smooth muscle cells (VSMCs) were cultured in the presence of 20% fetal bovine serum and 3.6 mm CaCl2 for 48 h. To each well, biotinylated matrix Gla protein (MGP)-derived peptides were added at a concentration of 500 μg L–1. (A) Full-length carboxylated MGP (cMGP1–84), confocal section within VSMCs. (B) cMGP1–53. (C) cMGP35–54. (D) Phosphorylated MGP3–15. (E) Uncarboxylated MGP35–54. (F) Synthetic full-length carboxylated osteocalcin. The peptides were detected using streptavidin–fluorescein isothiocyanate. Counterstaining was performed with DAPI to localize nuclei of cells.


Although in vivo studies clearly demonstrate that MGP is a potent inhibitor of calcification, its mechanism of action has not been fully elucidated [12,13]. MGP is thought to bind Ca2+ or calcium crystals and to inhibit calcium crystal growth and/or BMP signaling [8,11,24]. Conflicting results were obtained with in vitro studies. Two reports suggest that, under certain conditions, MGP may stimulate calcification [25,26], whereas a recent report demonstrated that MGP concentrations up to 10 μm completely inhibit calcification [31].

We and others have previously shown that inhibiting extrahepatic γ-glutamyl carboxylation by warfarin induces vascular media calcification in rats [15,32]. This is consistent with the finding that in humans high vitamin K intake is associated with low aorta calcification [33] and has beneficial effects on the elastic properties of the vessel wall [34]. Recently, we showed that high vitamin K intake could even reverse preformed arterial calcification in rats [35].

In the present study, we found that vitamin K (both K1 and K2) and its antagonist warfarin oppositely affect calcification in VSMC cultures exposed to high extracellular Ca2+ concentrations. These effects are probably due to the extent of MGP carboxylation, as we demonstrated the presence of ucMGP in VSMCs treated with warfarin. Vitamin K significantly inhibited calcification when compared to cells in standard medium (vitamin K concentration approximately 30 pm), demonstrating suboptimal vitamin K concentrations in the control medium. The low vitamin K content of regular cell culture medium is consistent with the findings of other studies [36], and we suggest that vitamin K should be routinely added as an essential vitamin to cell culture media. VSMCs grown in the presence of warfarin under high-Ca2+ conditions had significantly higher calcium crystal accumulation in the extracellular matrix as compared to control VSMCs.

The observation that warfarin stimulates calcification is consistent with several other studies. Price et al. [15,20] showed that the Gla residues of MGP are important for inhibiting medial calcification in warfarin-treated rats. In addition, warfarin accelerated calcification in human VSMCs in response to calcium and phosphate ion levels in uremic patients [28]. Furthermore, it was recently demonstrated in humans that blocking vitamin K function by coumarins resulted in increased coronary artery and aortic valve calcification [37,38].

To specifically address which MGP domains are important for human VSMC calcification, we used synthetic MGP-derived peptides and tested their effects in a model of calcification in which extracellular Ca2+ concentrations were elevated. The phosphate concentration and the FBS concentration were kept constant. FBS contains calcification inhibitors, such as fetuin. Because only the Ca2+ concentration was raised, MGP peptides were the only variables. In the cell-free control, calcification was absent under high-Ca2+ conditions, which suggests that the trigger for calcification was cell-derived and not due to precipitation of calcium salts in the medium. Using this model, we found that full-length cMGP inhibited calcification. We used concentrations at which cMGP and the various MGP-derived peptides were fully soluble in culture medium. At present, it is technically very difficult to synthesize large peptides (>15 amino acid residues) containing the pSer motif. Hence, the effect of serine phosphorylation could only be monitored by comparing pMGP3–15 and npMGP3–15. From Fig. 3, it can be seen that none of the peptides containing an uncarboxylated Gla domain showed calcification inhibitory activity. Although it is not possible to accurately quantify the activities displayed by the various peptides, it seems that the carboxylated peptides cMGP1–53 and cMGP35–54 had comparable activity, whereas similar calcification inhibition was obtained at 4- to 5-fold lower concentrations of the full-length cMGP1–84 (cf. Fig. 3A). This suggests that the non-phosphorylated pSer domain does not contribute to calcification inhibition, but that the C-terminal half of the molecule is also of major importance for full MGP activity. By comparing phosphorylated and non-phosphorylated MGP3–15, it was found that the pSer residues also actively participate in MGP’s calcification inhibitory activity. Taken together, these data strongly suggest that the pSer domain, the Gla domain and the invariable C-terminus are all important for MGP function. This is the first demonstration that the phosphorylated region of MGP adds to the calcification inhibitory activity of MGP. By analogy, phosphorylation of osteopontin is also required for its calcification inhibitory activity [18,19].

Studies were performed to identify the site at which MGP interacts with VSMC cells. Biotinylated peptides were not taken up by VSMCs but adhered to the cell surface and/or to vesicle-like structures, but only at elevated Ca2+ concentrations. This implies that MGP may inhibit calcification by interacting with the cell surface or cell-derived vesicles, but that such binding is markedly stimulated by conditions mimicking increased risk for tissue calcification. It was recently shown that MGP binds to vitronectin, and it was proposed that this extracellular matrix interaction may alter its function in modulating BMP-2 or transforming growth factor β activity [39]. High levels of extracellular Ca2+ and phosphate were previously shown to induce apoptosis in addition to vesicle release from viable cells [28]. In the current study, 20% serum was present in the cultures, and no apoptosis was observed over the 48 h studied, as determined by time-lapse videomicroscopy (data not shown). The vesicles bound by biotinylated MGP peptides in this study are therefore likely to represent matrix vesicles. As Gla proteins are known to bind to phosphatidylserine (PS), a component of apoptotic bodies and matrix vesicles, MGP could potentially bind vesicles via PS [40]. Although the precise interaction between MGP peptides, the cell surface and cell-derived vesicles requires further investigation, MGP may bind directly to calcium crystals associated with matrix vesicles.

In conclusion, the pSer domain, the Gla domain and the invariable C-terminus of MGP all contribute to the calcification inhibitory activity of MGP, which might be mediated through binding of MGP to cellular structures through its Gla/pSer domains.


The work described in this article was supported by the British Heart Foundation and grant 2001.033 of the Netherlands Heart Foundation.

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.