A Polymorphism of the Human Matrix (cid:1) -Carboxyglutamic Acid Protein Promoter Alters Binding of an Activating Protein-1 Complex and Is Associated with Altered Transcription and Serum Levels*

Matrix (cid:1) -carboxyglutamic acid protein (MGP) is a mineral-binding extracellular matrix protein synthesized by vascular smooth muscle cells (VSMCs) and chondrocytes that is thought to be a key regulator of tissue calcification. In this study, we identified four polymorphisms in the promoter region of the human MGP gene. Transfection studies showed that the G (cid:2) 7A and T (cid:2) 138C polymorphisms have an important impact on in vitro promoter activity when transiently transfected into VSMCs. We found that one of these polymorphisms (T (cid:2) 138C) is significantly correlated with serum MGP levels in human subjects. Promoter deletion analysis showed that this polymorphism lies in a region of the promoter critical for transcription in VSMCs. This region contains a potential activating protein-1 (AP-1) binding element located between (cid:2) 142 and (cid:2) 136. We have demonstrated that the T (cid:2) for the T 138C polymorphism, 3-min denaturation 94 for 60 Plasmids— into pGL2-Basic luciferase MGP promoter constructs specific for the (cid:1) 7G and (cid:1) 138T were generated with the QuickChange site-directed mutagenesis kit (Stratagene). Mutated constructs were sequenced to confirm successful site-directed mutagenesis. Plasmid DNA was prepared using an endo-toxin-free Maxi Prep kit (Qiagen). dry milk and 0.1% Tween 20 for 1 h. The membrane was then incubated with antibodies specific to pc-Jun, JunB, Fra-1, and Fra-2 (1:5000–1:10,000) for 16 h at room temperature. The membrane was then incubated with horseradish per-oxidase-conjugated goat anti-rabbit Ig antibodies (NA934; Amersham Pharmacia Biotech) for 1 h atroom temperature and washed four times in PBS with 0.05% Tween 20 for 15 min each. The membrane was then incubated with ECL chemiluminescent substrate (Amersham Pharmacia Biotech), and an exposure was made on Kodak XMR film. Statistical Analysis— One-way ANOVA and the Kruskal-Wallis test were used to analyze MGP serum levels between the different geno- types. The observed genotype frequencies for the G (cid:1) 7A and T (cid:1) 138C polymorphisms were compared with those expected under conditions of Hardy-Weinberg equilibrium using the (cid:4) 2 test. Student’s unpaired t tests were used to compare relative luciferase activities in the trans- fection experiments.


Matrix ␥-carboxyglutamic acid protein (MGP) is a mineral-binding extracellular matrix protein synthesized by vascular smooth muscle cells (VSMCs) and
chondrocytes that is thought to be a key regulator of tissue calcification. In this study, we identified four polymorphisms in the promoter region of the human MGP gene. Transfection studies showed that the G؊7A and T؊138C polymorphisms have an important impact on in vitro promoter activity when transiently transfected into VSMCs. We found that one of these polymorphisms (T؊138C) is significantly correlated with serum MGP levels in human subjects. Promoter deletion analysis showed that this polymorphism lies in a region of the promoter critical for transcription in VSMCs. This region contains a potential activating protein-1 (AP-1) binding element located between ؊142 and ؊136. We have demonstrated that the T؊138C polymorphism results in altered binding of an AP-1 complex to this region. The ؊138T allelic variant binds AP-1 complexes consisting primarily of c-Jun, JunB and its partners Fra-1 and Fra-2 in rat VSMC. Furthermore, the ؊138T variant form of the promoter was induced following phorbol 12-myristate 13-acetate treatment, while the ؊138C variant was refractive to phorbol 12-myristate 13-acetate treatment, confirming that AP-1 factors preferentially bind to the ؊138T variant. This study therefore suggests that a common polymorphism of the MGP promoter influences binding of the AP-1 complex, which may lead to altered transcription and serum levels. This could have important implications for diseases such as atherosclerosis and aortic valve stenosis, since it strongly suggests a genetic basis for regulation of tissue calcification. Extracellular calcification is a common and clinically significant component of a number of important human diseases including atherosclerosis and aortic valve stenosis (1,2). The concentration of calcium and phosphate ions in mammalian extracellular fluids are sufficiently high to induce precipitation of apatite, yet widespread tissue calcification does not usually occur in health (3). Protection against calcification during health is thought to be due, at least partly, to the complexing of calcium by organic bone-associated proteins such as matrix Gla protein (MGP) 1 (4). MGP is a mineral-binding extracellular matrix protein synthesized by vascular smooth muscle cells (VSMCs) and chondrocytes (4,5). It belongs to a family of proteins that contain glutamyl residues that are post-translationally modified by a vitamin K-dependent ␥-glutamyl carboxylase into ␥-carboxyglutamic acid (Gla) residues (6). These Gla residues promote binding to calcium and phosphate. Mice lacking MGP develop extensive arterial and valvular calcification, strongly suggesting that it acts as an inhibitor of tissue calcification (7). We have previously demonstrated that MGP is associated with areas of calcification in human atherosclerotic arteries perhaps as a homeostatic response to prevent further pathological calcification (8 -10). Furthermore, we have observed a similar association in cultured human VSMCs, which spontaneously form calcified nodules (11). These studies suggest a key regulatory role for MGP in tissue calcification, particularly in vascular tissues.
Despite the importance of MGP as an inhibitor of calcification and its potential role in diseases such as atherosclerosis, little is known about its transcriptional regulation. We hypothesized that polymorphisms may be present in the promoter region of MGP that could result in interindividual variation in transcription and tissue expression. In this study, we identify a common promoter polymorphism of the MGP gene that leads to altered binding of an activating protein-1 (AP-1) complex and is associated with variations in in vitro expression and serum levels. Furthermore, we show that c-Jun, JunB, Fra-1, and Fra-2 are the dominant components of this AP-1 complex in vascular smooth muscle cells.

EXPERIMENTAL PROCEDURES
Subjects-DNA for the initial polymorphism identification was extracted from 40 subjects. Samples for the MGP assay and subsequent genotyping were obtained from 156 healthy subjects (55-65 years old) in Maastricht as described previously (12). * This work was supported by the British Heart Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  Single-stranded Conformational Polymorphism (SSCP) Analysis-15 overlapping PCR primer pairs were used to amplify 3.3-kilobase pair DNA upstream of the MGP gene in 40 unrelated individuals. 5 l of PCR product was added to 5 l of loading dye (98% formamide, 0.025% bromphenol blue, 0.25% xylene cyanol, 10 mM EDTA), denatured by heating to 95°C for 8 min, and rapidly cooled on ice to preserve single-stranded structure. The denatured samples were loaded onto 6 -8% acrylamide gels and electrophoresed at 5-watt constant power overnight. The gel reactions were carried out either at 4°C or 20°C. Fragments were visualized by silver staining. Whenever a persisting deviant pattern was observed, the samples were sequenced in order to identify the sequence polymorphism causing the shift.
Genotyping-Genotyping for the GϪ7A polymorphism was performed by using a mismatch PCR fragment amplified with the forward primer (5Ј-CTAGTTCAGTGCCAACCCTTCCCCACC-3Ј) and the reverse primer (5Ј-TAGCAGCAGTAGGGAGAGAGGCTCCCA-3Ј), followed by digestion with the restriction enzyme NcoI. The TϪ138C polymorphism was genotyped using a mismatch PCR fragment amplified with the forward primer (5Ј-AAGCATACGATGGCCAAAACTTCT-GCA-3Ј) and the reverse primer (5Ј-GAACTAGCATTGGAACTTTTC-CCAACC-3Ј), followed by digestion with the restriction enzyme BsrSI. The PCRs were performed in a total volume of 25 l of a buffer solution containing the following: 10 mM Tris⅐HCl, pH 8.3, 50 mM KCl, 1.0 mM MgCl 2 , 0.25 mM dNTP, 10 units of Taq DNA polymerase, and 0.25 M forward and reverse primers. For the GϪ7A polymorphism, the reaction was run with a 3-min denaturation at 94°C followed by 30 cycles of 94°C for 30 s, 64°C for 60 s, and 72°C for 60 s. For the TϪ138C polymorphism, the reaction was run with a 3-min denaturation at 94°C followed by 30 cycles of 94°C for 30 s, 57°C for 60 s, and 72°C for 60 s.
Reporter Plasmids-MGP reporter plasmids were a kind gift from Dr. R. Schule (University of Freiburg, Germany). They consisted of progressive MGP promoter deletions (Ϫ3570, Ϫ530, Ϫ270, and Ϫ102) inserted into the pGL2-Basic luciferase reporter plasmid (Promega). MGP promoter constructs specific for the Ϫ7G and Ϫ138T mutations were generated with the QuickChange site-directed mutagenesis kit (Stratagene). Mutated constructs were sequenced to confirm successful site-directed mutagenesis. Plasmid DNA was prepared using an endotoxin-free Maxi Prep kit (Qiagen).
Transient transfection assays were carried out using the Superfect kit (Qiagen). VSMCs plated at 60% confluence on 60-mm plates were exposed for 2 h to 4 g of MGP promoter-luciferase and 0.1 g of Bos-␤-galactosidase constructs (14) in the presence of 20 l of Superfect solution and 1.2 ml of medium 199 (Sigma). Transfected cells were grown for 48 h after transfection before being harvested for analysis of luciferase activity. Briefly, the cells were washed three times with cold phosphate-buffered saline solution and harvested in 1 ml of cell lysis buffer (Roche Molecular Biochemicals) followed by centrifugation to remove cellular debris. 100 l of cell lysate was added to 468 l of luciferase substrate buffer and luminescence measured in a luminometer (LB953; AutoLumat). The ␤-galactosidase was assayed using a standard enzyme-linked immunosorbent assay in microtiter plates (Roche Molecular Biochemicals). Luciferase expression was normalized against ␤-galactosidase activity to account for variation in transfection efficiency. All experiments were conducted in sextuplicate in four independent transfection experiments.
When investigating the effects of PMA, transfected cells were grown for 24 h after transfection as described above, before treating with either 100 nM PMA or 0.1% ethanol vehicle for 2 h. The cells were then washed and grown for a further 24 h before being harvested for analysis of luciferase activity.
Serum MGP Measurements-MGP levels were measured in human serum samples using an enzyme-linked immunosorbent assay as described previously (12).
Electrophoretic Mobility Shift Assay-The sense sequences for the oligonucleotides used were as follows: 138T, 5Ј-TGGAAGGAATGACT-GTTTGGGAAAAGT-3Ј; 138C, 5Ј-TGGAAGGAATGACCGTTTGG-GAAAAGT-3Ј; consensus TRE site incorporated into the MGP oligonucleotide, 5Ј-TGGAAGGAATGACTCATTGGGAAAAGT-3Ј; and mutated consensus TRE oligonucleotide, 5Ј-TGGAAGGAATGACCCATTGG-GAAAAGT-3Ј. The mutated consensus AP-1 oligonucleotide has a substitution of a T for a C in the position corresponding to Ϫ138 in the MGP promoter. The probes were made by end-labeling single-stranded oligonucleotides with [␥-32 P]dATP and T4 polynucleotide kinase. The com-plementary strands were annealed, and the double-stranded oligonucleotides were subsequently purified on Stratagene push columns (Stratagene). Binding reactions were carried out in 10 mM Tris⅐HCl, pH 7.5, 0.5 mmol/liter EDTA, 0.5 mM dithiothreitol, 10% glycerol, 50 mM NaCl, 2 mM MgCl 2 , and 0.05% Nonidet P-40. Labeled DNA probe (100,000 cpm) was added to each reaction mixture containing 1 g of double-stranded poly(dI-dC) and 6 g of protein from crude nuclear extract. Crude nuclear extract was prepared from rat VSMC by the method of Dignam (15). AP-1 control nuclear extracts from 3T3 and 3T6 cells were provided in the AP-1 family nushift kit (Geneka). Reaction mixtures were incubated for 30 min at room temperature. For the competition experiments, the same conditions were used except that the specific cold competitor oligonucleotides were added to the reaction mixture 10 min before the addition of the labeled probe. The sample was loaded on a 4% polyacrylamide (acrylamide/bisacrylamide, 30:1) gel and run at room temperature.
For antibody supershift assays, 4 g of nuclear extract were incubated with 2-4 g of antibody obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) in the binding buffer described above. Alternatively, 4 -6 l of antibody provided in the AP-1 family Nushift kit (Geneka) was incubated in the binding buffer provided by Geneka. Labeled probe was then added for a further 20 min, and the sample run on a precooled gel. pc-Jun (sc-822X) is a monoclonal antibody directed against phosphorylated p39 c-Jun. Pan-cJun/AP-1 (sc-44X) is a polyclonal antibody directed against the highly conserved DNA binding domain of c-Jun and recognizes all known members of the Jun protein family. c-Jun (sc-45X) is a polyclonal antibody recognizing c-Jun only. The Fos gene family antibodies used were a pan-Fos monoclonal antibody (sc-447X), recognizing all members of the Fos protein family and specific polyclonal antibodies recognizing Fra-1 (sc-605X) and Fra-2 (sc-604X). Other antibodies used recognized pan-CREM-1 (sc-440X), ATF2 (sc-187X), pan-C/EBP (sc-746X), and c-Myb (sc-516X). Polyclonal antibodies in the AP-1 family nushift kit (Geneka) were reactive against c-Jun, JunB, JunD, c-Fos, and FosB.
Western Blotting of VSMC Extract-Whole cell lysate was prepared from rat VSMCs that had been cultured for 24 h in medium containing 100 nM PMA or the same amount of ethanol vehicle. 30 g of whole cell lysate was denatured in 2.5% SDS, 30 mM dithiothreitol, and 62.5 mM Tris⅐HCl (pH 6.8) at 65°C for 15 min, fractionated in 10% SDS-polyacrylamide gel electrophoresis, and then electrotransferred onto a polyvinylidene difluoride membrane. The molecular mass measurements were based on Kaleidoscope prestained standards (Bio-Rad). The membrane was incubated in PBS with 5% nonfat dry milk and 0.1% Tween 20 for 1 h. The membrane was then incubated with antibodies specific to pc-Jun, JunB, Fra-1, and Fra-2 (1:5000 -1:10,000) for 16 h at room temperature. The membrane was then incubated with horseradish peroxidase-conjugated goat anti-rabbit Ig antibodies (NA934; Amersham Pharmacia Biotech) for 1 h at room temperature and washed four times in PBS with 0.05% Tween 20 for 15 min each. The membrane was then incubated with ECL chemiluminescent substrate (Amersham Pharmacia Biotech), and an exposure was made on Kodak XMR film.
Statistical Analysis-One-way ANOVA and the Kruskal-Wallis test were used to analyze MGP serum levels between the different genotypes. The observed genotype frequencies for the GϪ7A and TϪ138C polymorphisms were compared with those expected under conditions of Hardy-Weinberg equilibrium using the 2 test. Student's unpaired t tests were used to compare relative luciferase activities in the transfection experiments.

Identification of Four Common Polymorphisms in the Promoter
Region of MGP-SSCP analysis of 15 overlapping PCR fragments upstream of the MGP gene in a cohort of 40 individuals revealed the presence of four deviant bands (Fig. 1a). Subsequent sequencing of these bands confirmed four polymorphic sites at nucleotide positions Ϫ7 (G or A), Ϫ138 (T or C), Ϫ514 (C or T), and Ϫ2447 (G or A). Fig. 2 shows the approximate location of these polymorphisms on a schematic diagram of the MGP gene.
The Region of the Promoter in the Vicinity of the TϪ138C Polymorphism Is Essential for Transcription in VSMCs-Progressive MGP promoter deletion constructs transiently transfected into rat VSMCs revealed a major loss of transcription following deletion of the Ϫ270 to Ϫ102 region (Fig. 3). This suggests that the sequence between Ϫ270 and Ϫ102 is critical for MGP transcription in VSMCs. Interestingly, the TϪ138C polymorphism lies in this region, suggesting that it may have significant effects on MGP expression.
Promoter Polymorphisms Are Associated with Modulated Gene Expression in Vitro-The influence of the two most common polymorphisms on gene expression was examined by using reporter gene constructs transiently transfected into VSMCs. These demonstrated independent impact of both common polymorphisms on transcriptional activity of the MGP gene (Fig. 4). The Ϫ7A variant had ϳ1.5-fold higher activity than the Ϫ7G variant (p Ͻ 0.001), whereas the Ϫ138C variant had ϳ4-fold higher activity than the Ϫ138T variant (p Ͻ 0.001).
The TϪ138C Polymorphism Is Associated with Variations in Serum MGP Levels-To test whether the MGP promoter polymorphisms had any effects on MGP levels in vivo, restriction fragment length polymorphism analysis was performed for the TϪ138C and GϪ7A polymorphisms on a sample of 156 healthy subjects in whom serum MGP had been assayed (Fig. 1b). The TϪ514C and AϪ2447G polymorphisms were not investigated in this respect, since they were much less common (incidence of the rarer allele was Ͻ5%). The population frequencies of the GϪ7A and TϪ138C polymorphisms were not significantly different from the distribution expected from Hardy-Weinberg equilibrium (p Ͼ 0.5 and p Ͼ 0.1, respectively). There were highly significant variations in serum levels of MGP as a function of the TϪ138C polymorphism (ANOVA, p Ͻ 0.0001; Kruskal-Wallis test, p Ͻ 0.0001) but not the GϪ7A polymorphism (ANOVA, p ϭ 0.67; Kruskal-Wallis test, p ϭ 0.759) ( Table I and Fig. 5). Thus, the CC variant at Ϫ138 was associated with higher mean serum levels of MGP (124.6 units/ml) than subjects with the TT variant (96.4 units/ml). A gene dose effect is also evident, with the CT heterozygotes having intermediate values (101.9 units/ml).
The TϪ138C Polymorphism Leads to Altered Binding Affinity for VSMC Nuclear Proteins-To investigate the possibility that the TϪ138C polymorphism alters binding of nuclear pro- teins, gel mobility shift assays were performed using oligonucleotides corresponding to the Ϫ138T and Ϫ138C alleles. Incubation of these oligonucleotides with nuclear extract from rat VSMCs produced a different pattern of protein-DNA complex formation for each variant of the TϪ138C polymorphism (Fig.  6). The Ϫ138C variant preferentially binds to complex B, whereas the Ϫ138T variant preferentially binds to complex A. Furthermore, complex A is preferentially eliminated by competition with cold Ϫ138T compared with competition with cold Ϫ138C. Similarly, complex B is preferentially reduced by competition with cold Ϫ138C compared with competition with cold Ϫ138T. These competition experiments demonstrate that the TϪ138C polymorphism alters the binding affinity of nuclear proteins for that region of the promoter.
Complex A Contains the AP-1 Transcription Factor Complex-A data base search (TRANSFAC) (16) with the sequence of the MGP promoter in the vicinity of the TϪ138C polymorphism identified a potential AP-1 binding site between Ϫ136 and Ϫ142. Competition experiments demonstrated that increasing concentrations of cold consensus TRE oligonucleotides were able to compete effectively with the Ϫ138T radiolabeled oligonucleotide for binding complex A (Fig. 7). In fact, cold consensus TRE oligonucleotide was a better competitor for the radiolabeled Ϫ138T oligonucleotide than either cold Ϫ138T or cold Ϫ138C (Fig. 7). Furthermore, when nuclear extract from 3T3 and 3T6 cells (which are known to contain AP-1 complexes) was used, the labeled consensus TRE oligonucleotide formed an intense complex in the same region as complex A (observed with labeled Ϫ138T oligonucleotide) (Fig. 8). In contrast, labeled Ϫ138C oligonucleotide was a weak binder. Binding of the intense complex A is significantly reduced by mutation of T in the TRE consensus oligonucleotide (TGACTCA) to C (TGAC-CCA), corresponding to the TϪ138C polymorphism (Fig. 9). These experiments strongly suggest that AP-1 complex proteins are components of complex A in VSMCs and that the T in position Ϫ138 is critical for binding of the factors within complex A.
The TϪ138C Polymorphism Leads to Altered Responses of the MGP Promoter to PMA-Since AP-1 binding sites mediate the effects of phorbol esters, we evaluated the effects of the TϪ138C polymorphism on PMA stimulation of the Ϫ270 MGP promoter construct (Fig. 10). 2 h of treatment with 100 nM PMA resulted in a statistically significant 1.6-fold increase in transcription of the Ϫ138T variant of the promoter (p ϭ 0.008) transiently transfected into rat VSMCs. However, treatment with 100 nM PMA had no significant effect on transcription of the Ϫ138C variant of the promoter (p ϭ 0.9).
The AP-1 Complex Binding in the Region of the TϪ138C Polymorphism Contains pc-Jun, JunB, Fra-1, Fra-2, and FosB-The components of the AP-1 complex binding to the Ϫ138T oligonucleotide were initially investigated by performing supershifts using antibodies reactive to c-Myb, pc-Jun, pan-Fos, pan-Jun, ATF-2, pan-CREM, and pan-C/EBP (Fig.  11). The pan-Jun antibody blocked the appearance of complex A, and anti-pc-Jun caused a strong clear supershift. The pan-Fos antibody also reduced the intensity of the shift, while those antibodies recognizing other factors had no effect (Fig. 11).
Antibodies recognizing other components of AP-1 were used to further probe the identity of the constituent proteins in the rat VSMC complex A (Fig. 12). A blockshift was observed with the c-Jun polyclonal antibody, and the presence of c-Jun was confirmed by a strong supershift seen with the anti pc-Jun monoclonal antibody. Antibodies to JunB caused a supershift indicating its presence within the complex. Of the Fos family antibodies, FosB caused a weak supershift, whereas Fra-1 and Fra-2 blocked the lower and upper components of complex A, respectively (Fig. 12). The addition of both Fra-1 and Fra-2 antibodies resulted in an additive effect with a very extensive block shift (Fig. 12). A similar pattern of antibody reactivity and AP-1 binding was observed when labeled consensus TRE was used with rat nuclear extract (data not shown).
pc-Jun, JunB, Fra-1, and Fra-2 Are Detected by Western Blotting in Rat VSMC, and the Levels of pc-Jun Increase following PMA Treatment-Western blotting of rat VSMC with antibodies recognizing pc-Jun, JunB, Fra-1, and Fra-2 confirmed the presence of these proteins (Fig. 13). Furthermore, the level of pc- Jun   FIG. 4. a, the GϪ7A polymorphism had statistically significant effects on in vitro transcription when inserted into the MGP Ϫ270 promoter luciferase construct (p Ͻ 0.001). b, the TϪ138C polymorphism had statistically significant effects on in vitro transcription when inserted into the MGP Ϫ270 promoter luciferase construct (p Ͻ 0.001). VSMCs were exposed for 2 h to 4 g of MGP promoter-luciferase and 0.1 g of Bos-␤-galactosidase constructs in the presence of 20 l of Superfect solution and 1.2 ml of medium 199. Transfected cells were grown for 48 h after transfection before being harvested for analysis of luciferase activity. Luciferase expression was normalized against ␤-galactosidase activity to account for variation in transfection efficiency. The activity of the more active polymorphism is expressed relative to that of the less active variant (mean Ϯ S.D.), which was attributed a value of 100. Basic represents VSMCs transfected with the luciferase reporter plasmid (pGL2) lacking any MGP promoter inserts. All constructs were tested in sextuplicate in four independent transfection experiments. protein increased following treatment with 100 nM PMA, when compared with ethanol vehicle alone (Fig. 13).

DISCUSSION
The present study has identified four novel polymorphisms in the promoter region of the human MGP gene. Transfection studies showed that the two most common of these have an important impact on in vitro promoter activity in rat VSMCs. Using MGP promoter deletion constructs, a region of the promoter critical for transcription was identified as being between Ϫ270 and Ϫ102. Interestingly, this key region contained the TϪ138C polymorphism. The importance of this region is consistent with the finding that the TϪ138C polymorphism has the greatest effect on in vitro transcription in VSMCs. Furthermore, the functional relevance of the TϪ138C polymorphism was demonstrated by the presence of significant differences in serum concentrations of MGP between the different allelic variants. This study therefore strongly suggests a genetic basis for variations in MGP transcription and serum levels.
The differing constitutive expression of the Ϫ138T and Ϫ138C luciferase constructs as well as the different serum levels associated with the TϪ138C polymorphic variants appears to be paralleled by altered ability to bind nuclear factors (Fig. 6). Nuclear complex A binds preferentially to the Ϫ138T variant, whereas nuclear complex B binds preferentially to the Ϫ138C variant.  1 versus lane 8). Both oligonucleotides form two main gel retardation complexes, A and B. However, the Ϫ138C variant preferentially binds to complex B, whereas the Ϫ138T variant preferentially binds to complex A. Furthermore, complex A formation is preferentially reduced by competition with excess cold Ϫ138T (lanes 9 -11), compared with competition with excess cold Ϫ138C (lanes [12][13][14]. Similarly, complex B is preferentially reduced by competition with cold Ϫ138C (lanes 5-7) compared with competition with cold Ϫ138T (lanes 2-4).

FIG. 7. Gel mobility shift assay using ␥-32 P-labeled doublestranded oligonucleotides corresponding to the ؊138T allele.
Complex A formation is preferentially reduced by competition with excess cold Ϫ138T (lanes 2-4), compared with competition with excess cold Ϫ138C (lanes 5-7). However, excess cold consensus TRE in the context of the MGP-binding oligonucleotide (MGP-TRE) competes much more effectively for complex A than cold Ϫ138T, suggesting that complex A contains AP-1 (lanes 8 -10).
The TϪ138C polymorphism occurs in a region (Ϫ142 to Ϫ136) with partial homology to a TRE. The consensus TRE sequence TGA(C/G)TCA is known to bind the AP-1 transcription factor complex in many genes and is defined by its ability to mediate phorbol ester-dependent induction of transcription (17). AP-1 is a collective term for a range of nuclear factors that bind the TRE as dimers. Members of the Jun gene family (c-Jun, JunB, and JunD) bind as homo-or heterodimers, while members from the Fos gene family (c-Fos, FosB, Fra-1, and Fra-2) form heterodimers with Jun (18). In addition, certain members of the ATF, CREB/CREM, Maf, Nrl, and JBP1/2 transcription factors can form leucine zipper dimers with Jun and/or Fos and bind a more diverse range of cis elements (18,19).
The MGP promoter sequence in the region of the Ϫ138T polymorphic variant (TGACTGT) has a 5-nucleotide identity FIG. 8. Gel mobility shift assay using nuclear extract from 3T3, 3T6, and VSMC cells. The extracts were incubated with ␥-32 P-labeled double-stranded oligonucleotides corresponding to the Ϫ138T allele, Ϫ138C allele, and consensus TRE binding site (MGP-TRE). Incubation of the Ϫ138T and consensus TRE oligonucleotides with nuclear extract from the different cell types produced protein-DNA complex with the same mobility shift as the Ϫ138T oligonucleotide. In contrast, the Ϫ138C labeled oligonucleotide does not bind, or only weakly binds, the nuclear proteins comprising complex A.

MGP-TRE(T) is identical with
Ϫ138T except for the substitution of the TRE consensus binding sequence (-TGACTCA-) in place of the MGP putative AP-1 binding site (-TGACTGT-). MGP-mutTRE(C) is identical with MGP-TRE(T) except for substitution of a T for a C in a position corresponding to the TϪ138C polymorphic site (-TGACCCA-). Labeled DNA probe was added to each reaction mixture containing nuclear extract from rat VSMC cells. The consensus MGP-TRE(T) oligonucleotide forms a more intense complex in the same region as complex A when compared with MGP-TRE(C) and Ϫ138T labeled oligonucleotides. The binding in this intense complex is significantly reduced by mutation of T in the TRE consensus oligonucleotide (-TGACTCA-) to C (-TGACCCA-). These experiments strongly suggest that AP-1 complex proteins are components of complex A in VSMCs and that the T in position Ϫ138 is critical for binding of the factors within complex A. with the consensus TRE sequence (TGA(C/G)TCA). We performed cold competition experiments demonstrating that increasing concentrations of cold consensus TRE oligonucleotides were able to compete effectively with the Ϫ138T radiolabeled oligonucleotide for the binding complex A in VSMC nuclear extracts (Fig. 7). In fact, cold consensus TRE was a better competitor for the radiolabeled Ϫ138T oligonucleotide than cold Ϫ138T, showing it to be a stronger binding site. In contrast, the Ϫ138C oligonucleotide competed very poorly, indicating weak or no binding to complex A (Fig. 7). When control AP-1 containing nuclear extracts derived from 3T3 and 3T6 cells were analyzed for factor binding to both the consensus TRE site and Ϫ138T labeled oligonucleotides, the protein complexes observed were of identical mobility in the gel (Fig. 8). However, the Ϫ138C variant did not exhibit similar binding intensity patterns, indicating that it bound AP-1 very poorly or not at all (Fig. 8).
Furthermore, we have shown that the T nucleotide at position 5 of the consensus TRE site is necessary for strong binding.
Nuclear factor binding was significantly reduced by mutation of T in (TGACTCA) to C (TGACCCA). This site corresponds to the TϪ138C polymorphism in the MGP promoter (Fig. 9). As mentioned above, the Ϫ138T variant appears to be a stronger binder of the AP-1 complex than the Ϫ138C variant. It therefore seems that the mutation of a T to a C nucleotide in the Ϫ138 position results in significant loss of AP-1 binding affinity. These experiments strongly suggest that AP-1 complex proteins bind the MGP promoter in the vicinity of the TϪ138C polymorphism and that the T in position Ϫ138 is critical for effective binding.
Antibodies to various AP-1 component proteins were used in gel mobility shift assays to determine the composition of the nuclear complexes binding in the region of the Ϫ138T variant of the polymorphism (Figs. 11 and 12). Antibodies against c-Jun, pc-Jun, and JunB blocked or supershifted complex A bound to Ϫ138T, indicating that these proteins were present. The polyclonal antibodies to Fra-1 and Fra-2 blocked the upper and lower regions of complex A, respectively. When both Fra-1 and Fra-2 antibodies were added in combination, this resulted in a considerable reduction of complex A formation (Fig. 12). Furthermore, FosB antibodies caused a weak but detectable supershift. Moreover, Western blot analysis confirmed the presence of Fra-1, Fra-2, pc-Jun, and JunB in rat VSMC (Fig.  13). Therefore, the AP-1 complex binding in the Ϫ138T region of the MGP promoter in rat VSMCs contains pc-Jun, JunB, Fra-1, and Fra-2 proteins. The levels of FosB are probably low but need to be further quantified. Interestingly, although AP-1 complexes are usually involved in trans-activation processes, the presence of JunB, Fra-1, and Fra-2 in AP-1 complexes has been associated with repression as well as activation (20 -24). The ability of AP-1 complexes containing Fra-1 and Fra-2 to act as repressors/weak activators is in part thought to be due to the absence of activation domains in these transcription factors. (25,26). This may well account for the lower transcription levels of the Ϫ138T variant when compared with the Ϫ138C form in normal rat VSMCs.
AP-1 binding sites mediate phorbol ester-dependent induction of transcription in many genes (17). We therefore investigated the functional effects of the TϪ138C polymorphism by examining the effects of PMA on the Ϫ138T and Ϫ138C MGP promoter variants (Fig. 10). Incubation of VSMCs for 2 h with 100 nM PMA resulted in induction of transcription of the Ϫ138T variant of the MGP promoter but not the Ϫ138C variant. This indicates that the TϪ138C polymorphism has significant functional effects on MGP transcription by causing altered responsiveness of the promoter to phorbol ester. The mechanism of this activation may be due to an increase in the levels of phosphorylated c-Jun (pc-Jun), since Western blotting showed that the levels of pc-Jun increases as a result of PMA treatment (Fig. 13). Therefore, the TϪ138C polymorphism appears to alter the ability of the promoter to respond to certain secondary messenger pathways by altering AP-1 binding.
The ability of the MGP promoter to respond to extracellular calcium is clearly important for its function as a regulator of extracellular calcification. We have recently shown that MGP transcription is regulated in VSMCs by a G protein-coupled cation-sensing mechanism that is functionally related to the cell surface calcium-sensing receptor (10). We demonstrated that increases in extracellular calcium concentration in the range between 2.2 and 6.0 mM result in induction of MGP transcription. Interestingly, Ng et al. (27) recently identified an AP-1 binding extracellular calcium-responsive element in the promoter of the keratinocyte gene Involucrin. Using supershift EMSAs, they showed that JunD, Fra-1, and Fra-2 were the major factors that bound this AP-1 element. Furthermore, they showed changes in the levels of these AP-1 components following changes in extracellular calcium concentration. The current study has demonstrated the involvement of c-Jun, JunB, Fra-1, and Fra-2 in VSMC MGP transcriptional regulation. It therefore remains to be seen if the effects of extracellular calcium on MGP transcription are mediated via these transcription factors acting through the AP-1 site in the vicinity of the TϪ138C polymorphism.
Other workers have studied the MGP promoter and characterized response elements close to where the TϪ138C polymorphisms. C/EBP and retinoic acid response elements exist immediately downstream and an ETS site immediately upstream of the TϪ138C polymorphism. We have confirmed that the MGP promoter is repressed by retinoic acid as shown by others (28) and that the polymorphism has no effect on this response (29). The transcription factor C/EBP␤ has previously been shown to be a strong stimulator of MGP transcription acting through the Ϫ138 to Ϫ102 region of the promoter (28). We have shown in this paper that C/EBP does not bind in the region of the polymorphism (Fig. 11). This may be because the labeled oligonucleotide used lacked the CCAAT site immediately downstream (Fig. 2). The presence of these response elements so close to the TϪ138C polymorphism, however, could be of functional significance due to differential protein-protein interactions resulting from the presence or absence of the AP-1 complex and need to be investigated further.
The main sources, pharmacodynamics, and pharmacokinetics of circulating MGP are currently unknown. Data from rats and mice suggest that MGP is mainly produced in vascular smooth muscle cells and developing cartilage (7,30). Therefore, it seems reasonable to assume that VSMCs are a major source for the circulating serum MGP that we measured in our adult population.
Since MGP is an important regulator of tissue and vascular calcification, our study suggests that there may be a genetic basis to tissue and vascular calcification in vivo. It is interesting to hypothesize that these promoter polymorphisms may affect an individual's susceptibility to vascular or valvular calcification. It is possible that the Ϫ138C variant provides protection against tissue calcification in VSMC by resulting in higher levels of MGP transcription. Equally, the responsiveness of the Ϫ138T site to extracellular stimuli mediated via AP-1 may result in altered susceptibility to calcification. We are therefore currently investigating the effects of the TϪ138C polymorphism on in vivo vascular and coronary artery calcification as measured by electron beam CT. Furthermore, since calcification occurs as part of the atherosclerotic process, it will be of interest for future studies to look for any associations of the TϪ138C polymorphism with atherosclerosis.
A recent study by Herrmann et al. (31) analyzed the effects of MGP polymorphisms on vascular calcification and myocardial infarction. They found that the Ϫ7A allele was more frequent in patients with myocardial infarction and among individuals with femoral calcification. In addition, they performed transient transfection experiments with MGP promoter constructs showing that the Ϫ138T variant had a 20% greater activity than the Ϫ138C variant and no effect of the GϪ7A polymorphism. The disparity with our results may be related to their use of rat VSMC lines in contrast to our primary adult rat VSMCs. It is known that the composition of the AP-1 complex varies between cell types. Since the AP-1 binding site in the MGP promoter is important for transcription, the presence of different AP-1 complexes would be expected to result in altered transcription. It would therefore be of interest to determine if the AP-1 composition reported here for primary adult rat VC-MCs is maintained in the cell lines used by Herrmann et al.
The correlation of polymorphisms with clinical phenotype described by Herrmann et al. (31) were weak or limited to subgroups of patients. Moreover, calcification was measured using ultrasound of the carotid and femoral arteries. This is a relatively crude measure of calcification (32) and may explain why the observed correlation with the Ϫ7A allele was only found in femoral calcification and not in carotid calcification. The ideal way to measure calcification in such a study would be by electron beam CT. Not only does this provide an accurate quantitative measure of calcification, but it is also capable of measuring calcification in the coronary arteries directly.
This study has demonstrated that a novel promoter polymorphism of MGP alters binding of an AP-1 transcription factor complex and is associated with altered serum levels in man. We have identified the components of the AP-1 complex in rat vascular smooth muscle cells as consisting primarily of c-Jun, JunB, Fra-1, and Fra-2. This has important implications for understanding the mechanisms underlying conditions that involve vascular calcification, such as atherosclerosis and aortic valve stenosis.