Polymorphisms of the Human Matrix Gla Protein (MGP) Gene, Vascular Calcification, and Myocardial Infarction
Abstract—The matrix Gla protein (MGP) is an important inhibitor of vessel and cartilage calcification that is strongly expressed in human calcified, atherosclerotic plaques and could modulate plaque calcification and coronary heart disease risk. Using a genetic approach, we explored this possibility by identifying polymorphisms of the MGP gene and testing their possible association with myocardial infarction (MI) and plaque calcification. Eight polymorphisms were identified in the coding and 5′-flanking sequences of the MGP gene. All polymorphisms were investigated in 607 patients with MI and 667 control subjects recruited into the ECTIM Study (Etude Cas-Témoins de l’Infarctus du Myocarde) and in 717 healthy individuals with echographically assessed arterial calcification and atherosclerosis who were participating in the AXA Study. In the ECTIM Study, alleles and genotypes were distributed similarly in patients and controls in the whole study group; in only 1 subgroup of subjects defined as being at low risk for MI were the concordant A−7 and Ala 83 alleles more frequent in patients with MI than in controls (P<0.003). In the AXA Study among subjects with femoral atherosclerosis, the same alleles were more common in the presence than the absence of plaque calcification (P<0.025). The other MGP polymorphisms were not associated with any investigated clinical phenotype. Transient transfection experiments with allelic promoter-reporter gene constructs and DNA-protein interaction assays were carried out to assess possible in vitro functionality of the promoter variants detected at positions −814, −138, and −7 relative to the start of transcription. When compared with the −138 T allele, the minor −138 C allele consistently conferred a reduced promoter activity of −20% (P<0.0001) in rat vascular smooth muscle cells and of −50% (P<0.004) in a human fibroblast cell line, whereas the other polymorphisms, including −7, displayed no evidence of in vitro functionality. We conclude that the A−7 or Ala 83 alleles of the MGP gene may confer an increased risk of plaque calcification and MI; however, the observed relationships are weak or limited to subgroups of patients and therefore need confirmation.
- Received September 24, 1999.
- Accepted February 23, 2000.
Myocardial infarction (MI) is, in most instances, the consequence of a thrombus forming on a ruptured atherosclerotic plaque.1 Plaque rupture occurs mainly in the shoulder region of the lesion2 and is frequently associated with calcification. Coronary calcification in asymptomatic patients is known to increase the risk of coronary heart disease (CHD),3 and mineral deposits are documented in >90% of patients with CHD.4 A number of studies have demonstrated the presence of bone-associated proteins, such as osteopontin, osteocalcin, bone morphogenic protein 2a, osteonectin, and matrix Gla protein (MGP), in human calcified, atherosclerotic plaques.5 6 7 8 9 Recent evidence suggests that calcification in the arteries may be the consequence of a functional deficit leading to an impaired inhibition of calcification rather than the consequence of an active calcification process.10 11 Under normal circumstances, the arterial wall would be naturally protected from calcium deposition by the presence of 1 or several molecules. In atherosclerotic arteries, Gla-containing proteins may play an important role in clearing calcium phosphate (hydroxyapatite) as a consequence of the strong affinity of Gla residues for this compound. Gla is formed posttranslationally from glutamic acid through γ-carboxylation by the vitamin K–dependent γ-glutamate carboxylase. MGP is a 10-kDa circulating protein that contains 5γ-carboxyglutamic acid residues. The importance of MGP to prevent calcification in soft tissues in vivo is well illustrated in the mgp knockout mouse model, which exhibits intense arterial calcification leading to vessel wall rupture and premature death,10 and in the Keutel syndrome, a rare human recessive disorder characterized by diffuse cartilage calcifications as a consequence of nonsense mutations of the MGP gene.12 It could be hypothesized that during the development of atherosclerotic plaques, MGP is trapped and inactivated or its expression or function is affected by other atheroma-dependent mechanisms; this would account for the tight correlation existing between plaque evolution and the degree of calcification. However, a less-passive role for MGP could also be envisioned as resulting from the modified expression or function dependent on sequence changes in the MGP gene. According to this scenario, production of either less MGP or of a deficient form of MGP could lead to an increased susceptibility to form calcium deposits in the presence of atherosclerosis. An upregulated expression of MGP in the calcified, atherosclerotic intima5 or in vascular smooth muscle cells13 would not contradict this possibility, since it could be the consequence of a feedback mechanism. Our hypothesis in this study was, therefore, that a genetic variability of MGP expression or sequence in humans could play a role in the variable predisposition to plaque calcification and its complications. To test this possibility, we scanned the coding and 5′-flanking regions of the MGP gene for polymorphisms and investigated the detected variants in relation to cardiovascular end points; in addition, we investigated the in vitro functionality of 3 polymorphisms located in the promoter region of the gene.
The ECTIM Study
In the present analysis of MGP polymorphisms in the ECTIM Study (Etude Cas-Témoins de l’Infarctus du Myocarde), 607 male patients with MI and 667 male control subjects representative of geographic areas in Northern Ireland (Belfast) and France (Lille, Strasbourg, and Toulouse)14 15 were included. Participants aged 25 to 64 years were recruited between 1988 and 1991 within regions covered by the WHO MONICA (World Health Organization MONItoring trends and determinants in CArdiovascular disease) registers. Cases were recruited 3 to 9 months after the event and had to satisfy the MONICA criteria for definite acute MI. Controls were randomly recruited from the same geographic areas as the cases. All control subjects with CHD were excluded. Written, informed consent was obtained from all subjects.
The AXA Study
The AXA Study,16 which includes 788 healthy volunteers (326 men and 462 women aged 17 to 65 years) from an insurance company (AXA, Paris, France), was designed to investigate risk factors for early stages of atherosclerosis. Investigations included a multifactorial evaluation of cardiovascular risk factors and ultrasound arterial measurements. The study protocol was approved by an ethics committee, and written consent was obtained from all participants.
Ultrasound Arterial Investigation
All echographic investigations of the carotid and femoral arteries were performed by 1 sonographer physician using a real-time B-mode ultrasound imager as described elsewhere.16 For each explored artery, 3 exclusive categories were defined: 1, normal aspect; 2, increased intima-media thickness (IMT); and 3, presence of an atherosclerotic plaque ≥2 mm. In addition, the presence of calcification was defined by the occurrence of a bright, hyperechogenic area inducing a cone-shape echo shadowing. In most analyses relating echographic parameters to MGP genotypes, the right and left arteries in each territory (carotid and femoral) were grouped together. For example, the “presence of plaque with calcification” means either that both arteries carry plaques and 1 or both are calcified or that only 1 artery carries a plaque with associated calcification.
Identification of Polymorphisms of the MGP Gene and Genotyping
The human MGP gene is situated on chromosome 12p, comprises 4 exons, and encodes an 84-residue mature protein. For polymerase chain reaction (PCR)/single-strand conformation polymorphism analysis17 of the MGP gene, 20 individuals with MI were selected from the ECTIM Study. From the published sequences of the MGP gene,18 20 overlapping fragments <300 bp long were amplified to cover the entire coding sequence and 2996 bp of its upstream region.
Single-strand conformation polymorphism analysis, sequencing, and genotyping with the use of allele-specific oligonucleotides were performed as previously described.19 The PCR primers, conditions used for amplifying the regions encompassing the polymorphic sites, and allele-specific oligonucleotides may be found on the CANVAS Internet site (http://genecanvas.idf.inserm.fr).
The MGP promoter was amplified by PCR from genomic DNA of subjects homozygous for the common promoter alleles by using a Tth Advantage polymerase mix (Clontech Laboratories Inc) and the primers MGPfor (5′-GACAGAGCTAATGTACATTGCAA-AGCAC-3′), located between −1946 and −1919 relative to the transcription start point, and MGPrev (5′-GAGTCGACCAG-GGTCTTGTGTAGCAGCAGTAG-3′), located between +10 and +33 and incorporating an SalI (HincII) site at its 5′ end (underlined). The product was blunt-ended and subcloned into an EcoRV-cut pBluescript II KS (Stratagene Inc). For the minor A−7 and C−138 alleles, shorter promoter fragments extending upstream of the StyI site located 593 bp upstream of the transcription start site were prepared by using the primer MGPfor2 (5′-ACTGGATG-GGCAAGTTACAACGGTGT-3′), located between −672 and −647 relative to the transcription start site, and MGPrev, with DNAs from subjects homozygous for these alleles, as templates. The products were cut with StyI and SalI and used to replace the corresponding fragment in the full-length promoter. To prepare a promoter containing the G allele at position −814, the full-length promoter was subcloned as an EcoRI-SalI fragment into pALTER-1 (Promega Corp), and site-directed mutagenesis with the use of primer 5′-CTGCCACTCACTAGAAAGTCTATCAAG-3′ (altered base shown underlined) and an Altered Sites II in vitro mutagenesis kit (Promega) was performed. For functional analysis, each promoter was isolated as a 1.42-kb SspI-HincII fragment and subcloned in the correct orientation upstream of the luciferase reporter gene in SmaI-cut pGL3-Basic (Promega).
Cell Culture and Transient Transfection
A10 and MRC5 cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal calf serum (Life Technologies), 1 mmol/L l-glutamine, and penicillin/streptomycin (100 μg/mL). Lipofectin reagent (Life Technologies) was used to introduce plasmids into cells essentially as previously described,20 except that 500 μL of OPTI-MEM I containing 7.5 μg of promoter construct and 2.5 μg of pRL-TK (Promega) was gently mixed with 500 μL of OPTI-MEM I containing 20 μg of lipofectin for each transfection. Luciferase levels were measured by using a Turner Designs model TD-20/20 luminometer and a dual luciferase assay (Promega).
Electrophoretic Mobility Shift Assay
Nuclear extracts were prepared according to the protocol by Alksnis et al.21 Twenty-five picomoles of annealed oligonucleotides (5′-CTGGAAGGAATGACT/CGTTTGGGAAAAGT-3′ and 5′-ACTTTTCCCAAACA/GGTCATTCCTTCCAG-3′; underlined bases correspond to the different −138 alleles) were 5′-end–labeled with T4 polynucleotide kinase and [γ-32P]ATP. Electrophoretic mobility shift assays (EMSAs) were performed with a binding buffer containing 20 mmol/L HEPES, pH 7.9, 100 mmol/L KCl, 6.25% glycerol, 0.5 mmol/L EDTA, 1 mg of bovine serum albumin, 100 μg of poly(dI-dC), and a 50- to 100-fold molar excess of competitor oligonucleotides, where indicated. Nuclear extract (1.5 μg) was added, followed by 1.25 pmol of [γ-32P] 5′-end–labeled, double-stranded oligonucleotide. Reaction mixtures were incubated at room temperature for 20 minutes, followed by electrophoresis on a 7.0% acrylamide (acrylamide/bisacrylamide 49:1, vol/vol), 0.25× Tris-borate-EDTA gel at 200 V (≈25 mA) for 2.5 hours at 4°C.
DNase I Footprinting
Probes were prepared by PCR and the primers 5′-ACTGCCCACTCAGAGTAGAT-3′ and 5′-GAGAAGGTGGGGA-AGGGTTG-3′, located between −177 and −158 and between −86 and −105, respectively. To label only 1 strand, PCRs were set up with either the forward or the reverse primer 5′-end–labeled and the promoter containing the T −138 allele as a template. Approximately 30 000 counts per minute of each probe was added to 21 μL of binding reactions containing 12.5 μL of 2× binding buffer (40 mmol/L HEPES, pH 7.9, 12.5% glycerol, 100 mmol/L KCl, 1 mmol/L CaCl2, and 200 μg of poly[dI-dC]) and 4.5 μL of nuclear extract. Reactions were incubated at 30°C for 20 minutes before 0.625 U of prewarmed DNase I (Amersham Pharmacia Biotech Inc) was added. After 2 minutes, 3 μL of 250 mmol/L EDTA was added and the reactions were transferred to ice. After electrophoresis on a 5% acrylamide (acrylamide/bisacrylamide 49:1, vol/vol), 0.25× Tris-borate-EDTA gel at 200V for 2 hours, the positions of bound and unbound probe were identified by autoradiography. Gel slices containing bound and unbound probe were excised from the gel, and the probe was electroeluted on to a DEAE NA-45 membrane (Schleicher and Schuell). Membranes were rinsed in low-salt buffer (0.15 mol/L NaCl, 0.1 mmol/L EDTA, and 20 mmol/L Tris HCl, pH 8.0), and the probe was eluted into 200 μL of high-salt buffer (1 mol/L NaCl, 0.1 mmol/L EDTA, and 20 mmol/L Tris HCl, pH 8.0) at 65°C for 45 minutes. The eluted probe was precipitated with ethanol, washed once with 70% ethanol, and resuspended in 10 μL of formamide loading buffer (95% formamide, 20 mmol/L EDTA, 0.05% bromophenol blue, and 0.05% xylene cyanol). Denatured samples were electrophoresed on a 10% acrylamide (acrylamide/bisacrylamide 19:1, vol/vol), 7 mol/L urea gel in a 1× Tris-borate-EDTA buffer alongside a dideoxy sequencing ladder generated by using end-labeled forward and reverse primers.
MGP Promoter Sequence
In the course of this work, DNA sequence analysis of the MGP promoter revealed a discrepancy with the published sequence (EMBL/GenBank accession No. M55270). A 60-bp sequence located at approximately −1.2 kb (from bp 2221 to 2280 in M55270) was found to be completely different from that published. The altered sequence was identified in PCR subclones of the promoter and when sequencing was performed directly on several different genomic DNAs. The reason for the discrepancy is unclear. However, basic local alignment search tool analysis of the EMBL and GenBank databases indicated that the 60-bp sequence present in the original database entry can also be found at a similar location in the database entry for another gene, human histone H2A.Z (EMBL/GenBank accession No. L10137). A revised sequence for the MGP promoter has been submitted to GenBank (EMBL/GenBank accession No. AF067176).
Data were analyzed with sas statistical software (SAS Institute Inc). Pairwise linkage disequilibrium coefficients were estimated in the control samples; coefficients are reported as the ratio of the unstandardized coefficients to their minimal/maximum value (|D′|).22 The frequency of the most frequent haplotypes was estimated by the maximum-likelihood method by using the myriad program.23 Hardy-Weinberg equilibrium was tested with a χ2 test with 1 df. Genotype and allele frequencies were compared between cases and control subjects in the ECTIM Study and between groups defined in the AXA Study with a χ2 test. Odds ratios (ORs) and their 95% confidence intervals are provided. Homogeneity of the relationship between MGP polymorphisms and MI according to low-risk status was evaluated with the sas/catmod procedure by testing the interaction term crossing genotypes (3 categories) and low-risk status (2 categories).
Polymorphisms of the MGP Gene: Frequency, Pairwise Linkage Disequilibrium, and Haplotypes
Eight polymorphisms were identified by scanning 40 alleles: 2 in exons, Lys34Glu and Thr83Ala; and 6 in the upstream region of the gene: G−7A, T−138C, C−514T, A−814G, G−2447A, and C−2682T (Figure 1⇓). All 8 polymorphisms were investigated in the ECTIM and AXA studies. C−2682T and Lys34Glu were relatively rare, their minor allele frequencies being 0.1% and 0.7%, respectively, in the control group of the ECTIM Study. Because these polymorphisms were unrelated to any relevant phenotype, they were not studied further. Table⇓ I (available online at http://atvb.ahajournals.org) reports the pairwise linkage disequilibrium coefficients between the 6 most common polymorphisms; because the values of the coefficients did not significantly differ between the groups from Belfast and France, pooled values are shown. Table⇓ II (available online at http://atvb.ahajournals.org) reports the most common haplotypes combining the 6 MGP polymorphisms and their estimated frequencies in control subjects from the Belfast and French populations.
Distribution of MGP Genotypes in Patients With MI and Control Subjects: the ECTIM Study
The mean age of cases and control subjects in the ECTIM Study was 54.0 (SD 8.2) and 53.0 (SD 8.5) years, respectively. The distribution of MGP genotypes and alleles in cases and control subjects is shown in Table 1⇑. There was no significant deviation of genotype frequencies from those expected under Hardy-Weinberg equilibrium. The Ala83, A−7, and G−2447 alleles were slightly more frequent in France than in Belfast (P<0.05 for cases and controls combined). Genotype and allele frequencies did not differ between case and control subjects in France or Northern Ireland in the entire ECTIM population (Table 1⇑) as well as between subjects younger or older than 55 years, the median age in the study (not shown). To assess whether background risk could have affected the results, a group of subjects at low risk of CHD was defined as previously in the ECTIM Study15 by the absence of treatment for hyperlipidemia, a plasma apoB level <1.25 g/L (the median of apoB in control subjects), and a body mass index <26 kg/m2 (the median body mass index in control subjects). The heterogeneity of the ORs for MI between the low- and high-risk groups was significant for G−7A (P<0.008) and Thr83Ala (P<0.002). In the low-risk subgroup, the Ala83 and A−7 alleles were more frequent in patients than in control subjects (Table 2⇓). The ORs for MI of heterozygotes and homozygotes for A −7 relative to GG −7 homozygotes were 1.34 and 3.82, respectively (P for trend <0.002); similar results were observed for Thr83Ala.
Distribution of MGP Genotypes in Subjects With Echographically Assessed Atherosclerosis and Calcification in the Carotid and Femoral Arteries: the AXA Study
In Table 3⇓ are shown the prevalences of increased IMT and atherosclerotic plaque(s) in the femoral arteries and of associated calcification (see Methods for the definition of categories). Among 722 subjects with appropriate echographic information available, 22.2% had atherosclerotic plaque(s) ≥2 mm and 18.3% had an increased IMT without identifiable plaque. In these 2 categories, echographically assessed calcification was present in 57.5% and 13.6% of the subjects, respectively, whereas in those without plaque and with a normal IMT, only 5.1% had calcifications. In the presence of femoral atherosclerotic plaque, calcifications were more prevalent in carriers of the A−7 allele (64.1%) than in GG−7 homozygotes (45.6%, P<0.025), with an OR for femoral atherosclerosis–associated calcification of 2.1 (95% confidence interval, 1.1 to 4.3). The other polymorphisms were unrelated to femoral artery calcification, and none of the polymorphisms was related to carotid artery calcification.
Effect of MGP Promoter Polymorphisms on MGP Promoter Activity
Reporter-gene constructs were prepared to determine whether the polymorphisms at positions −7, −138, and −814 affected MGP promoter activity. The resulting constructs pLuc-MGPfreq (containing the common alleles at each position), pLuc-MGP7A, pLuc-MGP138C, and pLuc-MGP814G (containing the minor alleles at −7, −138, and −814, respectively) were used to transiently transfect the rat vascular smooth muscle cell line A10 together with the control Renilla luciferase plasmid pRL-TK (Promega). Forty-eight hours later, luciferase levels were measured and standardized to the level of Renilla luciferase. Luciferase levels in cells transfected with pLuc-MGPfreq (275.3, n=12), pLuc-MGP7A (275.8, n=10), and pLuc-MGP814G (268.5, n=2) were not significantly different, indicating that the minor alleles at −7 and −814 did not affect MGP promoter activity under these experimental conditions. However, in cells transfected with pLuc-MGP138C, luciferase levels were consistently 20% lower than in cells transfected with pLuc-MGPfreq (219.1, n=12 versus 275.3, n=12; Kruskal-Wallis rank-sum test, P<0.0001). Luciferase levels for pLuc-MGPfreq, pLuc-MGP7A, and pLuc-MGP138C were also measured in the human fibroblast cell line MRC5. As in A10 cells, luciferase levels in cells transfected with pLuc-MGPfreq and pLuc-MGP7A were not significantly different (253.8, n=6 and 296.5, n=6, respectively). However, in cells transfected with pLuc-MGP138C, luciferase levels were approximately 50% of those in cells transfected with pLuc-MGPfreq (124.2, n=6 versus 253.8, n=6; Kruskal-Wallis rank-sum test, P<0.004).
The T−138C Polymorphism Is Located Within the Binding Site for a Nuclear Protein
To determine whether the T−138C polymorphic site affects the binding of a nuclear protein to the MGP promoter, EMSAs were performed with nuclear extracts from A10 cells (Figure 2⇓). The data indicate that a nuclear protein binds in the region encompassing the T−138C polymorphic site and that the extent of this binding is reduced when the minor C allele is present (see Figure 2⇓ for detailed discussion). Similar results were obtained with nuclear extracts from MRC5 cells and from the mouse macrophage cell line MALU (data not shown).
To identify the sequence bound by the nuclear protein(s), DNase I footprinting analysis was performed. An extensive footprint was obtained for both strands (Figure 3⇓), and the similarity between the footprints for the 2 complexes suggests that the same protein(s) binds in both cases. The clearer footprint for CII may indicate that proteins are more tightly associated in this complex or perhaps that they exist in a different form. The footprints obtained actually extended beyond the limits of the probes used in EMSAs. On this basis, the binding site of the protein affected by the T−138C polymorphism may not be located entirely within the EMSA probe, or it may be masked by the binding of additional proteins to the footprint probe. Further experiments are required to identify the binding site and nature of the protein interaction affected by the T−138C polymorphism.
Characteristics of MGP Gene Polymorphisms
Eight new polymorphisms were detected, 6 in the 5′-flanking region of the gene and 2 in its coding sequence. As has been reported for other genes,24 most polymorphisms were in tight linkage disequilibrium (online Table⇑ I). Eight haplotypes accounted for most of the variability of the MGP gene, and only 3 of them, haplotypes 1, 2, and 3, accounted for 88% of this variability (online Table⇑ II). Haplotype 1 (35% of all haplotypes) is a combination of the most frequent alleles of all polymorphisms. Haplotype 2 (34% of all haplotypes) differs from haplotype 1 at positions −7 and 83, these 2 polymorphisms being in almost complete association (online Table⇑ I). Haplotypes 3, 5, and 7 differ from haplotype 1 at a single site, and the information conveyed by these haplotypes is very similar to that provided by the C−138, A−2447, and G−814 alleles, respectively. Haplotype 4 differs from haplotype 2 at the −514 site. Finally, haplotype 6 differs from haplotype 1 at the 83 polymorphic site, but the corresponding Ala83 allele is also present on haplotypes 2 and 4. From this pattern of association among polymorphisms, we concluded that haplotypes 1 and 2 are ancestral, whereas haplotypes 3, 5, and 7 descended from haplotype 1 and haplotype 4 descended from haplotype 2, as the result of a single mutation. If this scheme is correct, haplotypes 6 and 8 would most likely have resulted from a recombination between haplotypes 1 and 2, although other genealogies could be proposed if it is assumed that the 2 most frequent haplotypes are not the most ancient. The overall heterozygosity, inferred from the haplotype frequencies provided by the 6 most common polymorphisms, was 0.72.
MGP Gene Polymorphisms and Cardiovascular End Points
In the ECTIM Study, the genotype distributions did not differ between patients with MI and control subjects. Only in a group of low-risk subjects were the Ala83 and A−7 alleles more frequent in cases than in control subjects (P<0.002 and P<0.004, respectively); low-risk status was defined as in a previous report of the ECTIM Study.15 In the AXA Study, none of the MGP polymorphisms was related to calcification or atherosclerosis of the carotid artery. On the other hand, the Ala83 and A−7 alleles were associated with femoral calcification in the presence of atherosclerotic plaques (P<0.025). The difference in findings between the carotid and femoral arteries might be explained by the significantly lower frequency of atherosclerotic/calcified plaques in the carotid than in the femoral arteries or by a possible heterogeneity of effect according to anatomic site.25 According to these observations, the G−7A or Thr83Ala polymorphism could influence the calcification process affecting atherosclerotic plaques, and it might contribute to the risk of MI in low-risk individuals. However, it will be necessary to verify these results in other studies before any definitive conclusion can be drawn, especially since the associations were observed in subgroups of patients and not in the whole population.
The T−138C but Not the G−7A Polymorphism Is Functional In Vitro
The results of our experiments in rat vascular smooth muscle cells and the human fibroblast cell line MRC do not indicate that the G−7A polymorphism may be functional in vitro. However, because this polymorphism is located at a possible transcription start site,18 it might be functional in other cell types than those studied here. The Thr83Ala polymorphism represents a change from a polar to a nonpolar amino acid, which might give rise to an altered form of the mature MGP. This alteration, which affects the Gla-binding domain 1 amino acid away from the carboxyl end of the MGP, could diminish its capacity to bind Ca2+, leading to enhanced deposition of free Ca2+ in the arterial wall. It is also conceivable that the G−7A and Thr83Ala polymorphisms are only markers in linkage disequilibrium with another, as-yet-unidentified functional variant.
In vitro analysis of MGP promoter activity revealed that the C−138 allele reduced promoter activity by ≈20% in rat vascular smooth muscle cells and by up to 50% in a human fibroblast cell line. Moreover, with the use of EMSAs, it was possible to demonstrate that a nuclear protein(s) specifically binds in the region covering the T−138C polymorphic site and that binding is enhanced in the presence of the T allele. Thus, the difference in promoter activity may be explained by differential binding of a nuclear protein that is important in MGP transcription. DNase I footprint analysis in fact indicated that the region between −160 and −104 is protected from DNase I digestion by several bound proteins. The −138 polymorphism is located in a retinoic acid (RA) negative-responsive element, which maps between positions −138 and −102 from the transcription start site of the MGP gene and also contains a CCAAT box (−123 to −118). This element is bound by 2 members of the nuclear receptor family, the RA receptor (RAR) and the retinoid X receptor (RXR), both forming the heterodimeric complex RAR/RXR. In the rat kidney cell line NRK52E, Kirfel and colleagues26 have shown that RA, via the RAR/RXR receptor complex, repressed MGP gene expression to ≈30% of that in untreated controls. However, using a 1.42-kb MGP promoter that contained the polymorphic sites at positions −7, −138, and −814, we did not observe a significant effect of RA on MGP gene expression in rat vascular smooth muscle cells and the fibroblast cell line MRC5 (not shown), although this discrepancy might be due to the fact that Kirfel et al used cell lines different from ours in their experiments. The RAR/RXR heterocomplex competes with the CAAT binding protein C/EBPβ (CCAAT/enhancer binding protein), since both factors recognize the same consensus site (CCAAT box) within the negative-response element. This might explain tissue-specific differences in MGP gene expression; ie, overexpression of C/EBPβ might overcome the repressive effect of RA on MGP expression. It is not clear which if any of these proteins binds to the footprint probe. However, preliminary EMSAs have indicated that the protein(s) binding to the probe does not change in the presence of RA (data not shown). Analysis of the region between −160 and −104 for potential cis-regulatory sequences revealed the presence of a sequence, 5′-CTGGAA-3′, between −152 and −147 that resembles the consensus for the interleukin-6–responsive transcription factor, IL-6 REBP.27 A sequence identical to this in the gene for the γ-chain of human fibrinogen has been shown to be a functional element in interleukin-6 response.28 It remains to be established whether the same sequence is functional in the MGP promoter. Despite the identification of a functional effect on MGP promoter activity in vitro, the T−138C polymorphism was not related to calcification, femoral artery atherosclerosis, or MI in our studies. This result may indicate that the reduction in absolute levels of MGP production caused by the C−138 allele may not be sufficient to affect these phenotypes.
In this study, the possible impact of MGP polymorphisms on artery calcification was investigated in the femoral and carotid arteries only. However, it is possible that the MGP variants are associated with calcification in the coronary arteries or with calcification-related diseases other than those studied here.27 29 It is also important to remember that the lethality associated with MI is ≈40% within the first month after the event.30 This is a general problem in case-control studies of MI, which might introduce a selection bias when a risk factor preferentially associated with mortality is under investigation.
In conclusion, we have described a number of polymorphisms of the MGP gene. The observed associations of the G−7A and Thr83Ala polymorphisms with MI in low-risk individuals and with echographically assessed, atherosclerosis-associated calcification suggest possible involvement of these polymorphisms in coronary artery disease calcification; however, these results must be substantiated in other studies focusing on appropriate end points, especially since the associations were weak and limited to subgroups of individuals. Our in vitro studies were unable to demonstrate any functional impact of the G−7A polymorphism, whereas the −138C allele was associated with reduced promoter activity.
Recruitment in the ECTIM Study was supported by grants from the Squibb Laboratory, the British Heart Foundation (program grant PG95008), INSERM, and the Institut Pasteur-Lille. Stefan-Martin Herrmann and Eva Brand are supported by grants from the Deutsche Forschungsgemeinschaft (HE 2852/1-1 and Br 1589/1-1, respectively).
Shanahan CM, Cary NR, Metcalfe JC, Weissberg PL. High expression of genes for calcification-regulating proteins in human atherosclerotic plaques. J Clin Invest.. 1994;93:2393–2402.
Giachelli CM, Bae N, Almeida M, Denhardt DT, Alpers CE, Schwartz SM. Osteopontin is elevated during neointima formation in rat arteries and is a novel component of human atherosclerotic plaques. J Clin Invest.. 1993;92:1686–1696.
Raines EW, Lane TF, Iruela-Arispe ML, Ross R, Sage EH. The extracellular glycoprotein SPARC interacts with platelet-derived growth factor (PDGF)-AB and -BB and inhibits the binding of PDGF to its receptors. Proc Natl Acad Sci U S A.. 1992;89:1281–1285.
Bostrom K, Watson KE, Horn S, Wortham C, Herman IM, Demer LL. Bone morphogenetic protein expression in human atherosclerotic lesions. J Clin Invest.. 1993;91:1800–1809.
Shanahan CM, Cary NR, Osbourn JK, Weissberg PL. Identification of osteoglycin as a component of the vascular matrix: differential expression by vascular smooth muscle cells during neointima formation and in atherosclerotic plaques. Arterioscler Thromb Vasc Biol.. 1997;17:2437–2447.
Proudfoot D, Skepper JN, Shanahan CM, Weissberg PL. Calcification of human vascular cells in vitro is correlated with high levels of matrix Gla protein and low levels of osteopontin expression. Arterioscler Thromb Vasc Biol.. 1998;18:379–388.
Parra HJ, Arveiler D, Evans AE, Cambou JP, Amouyel P, Bingham A, McMaster D, Schaffer P, Douste-Blazy P, Luc G, Richard JL, Ducimetiere P, Fruchart JC, Cambien F. A case-control study of lipoprotein particles in two populations at contrasting risk for coronary heart disease: the ECTIM Study. Arterioscler Thromb.. 1992;12:701–707.
Cambien F, Poirier O, Lecerf L, Evans A, Cambou JP, Arveiler D, Luc G, Bard JM, Bara L, Ricard S, Tiret L, Amouyel P, Alhenc-Gelas F, Subrier F. Deletion polymorphism in the gene for angiotensin-converting enzyme is a potent risk factor for myocardial infarction. Nature.. 1992;359:641–644.
Gariepy J, Salomon J, Denarie N, Laskri F, Megnien JL, Levenson J, Simon A. Sex and topographic differences in associations between large-artery wall thickness and coronary risk profile in a French working cohort: the AXA Study. Arterioscler Thromb Vasc Biol.. 1998;18:584–590.
Cancela L, Hsieh CL, Francke U, Price PA. Molecular structure, chromosome assignment, and promoter organization of the human matrix Gla protein gene. J Biol Chem.. 1990;265:15040–15048.
Herrmann SM, Ricard S, Nicaud V, Mallet C, Evans A, Ruidavets JB, Arveiler D, Luc G, Cambien F. The P-selectin gene is highly polymorphic: reduced frequency of the Pro715 allele carriers in patients with myocardial infarction. Hum Mol Genet.. 1998;7:1277–1284.
Ye S, Eriksson P, Hamsten A, Kurkinen M, Humphries SE, Henney AM. Progression of coronary atherosclerosis is associated with a common genetic variant of the human stromelysin-1 promoter which results in reduced gene expression. J Biol Chem.. 1996;271:13055–13060.
Alksnis M, Barkhem T, Stromstedt PE, Ahola H, Kutoh E, Gustafsson JA, Poellinger L, Nilsson S. High level expression of functional full length and truncated glucocorticoid receptor in Chinese hamster ovary cells: demonstration of ligand-induced down-regulation of expressed receptor mRNA and protein. J Biol Chem.. 1991;266:10078–10085.
Qiao JH, Xie PZ, Fishbein MC, Kreuzer J, Drake TA, Demer LL, Lusis AJ. Pathology of atheromatous lesions in inbred and genetically engineered mice: genetic determination of arterial calcification. Arterioscler Thromb.. 1994;14:1480–1497.
Kirfel J, Kelter M, Cancela LM, Price PA, Shule R. Identification of a novel negative retinoic acid responsive element in the promoter of the human matrix Gla protein gene. Proc Natl Acad Sci U S A.. 1997;94:2227–2232.
Mizuguchi J, Hu CH, Cao Z, Loeb KR, Chung DW, Davie EW. Characterization of the 5[prime]-flanking region of the gene for the γ-chain of human fibrinogen. J Biol Chem.. 1995;270:28350–28356.
Lehto S, Niskanen L, Suhonen M, Ronnemaa T, Lakso M. Medial artery calcification: a neglected harbinger of cardiovascular complications in non-insulin-dependent diabetes mellitus. Arterioscler Thromb Vasc Biol.. 1996;16:978–983.
Tunstall-Pedoe H, Kuulasmaa K, Amouyel P, Arveiler D, Rajakangas AM, Pjak A. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project: registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation.. 1994;90:583–612.