Matrix Gla Protein Polymorphisms are Associated with Coronary Artery Calcification in Men

J Nutr Sci Vitaminol, 55, 59 65, 2009 Matrix Gla Protein Polymorphisms are Associated with Coronary Artery Calcification in Men Michael D. CROSIER, Sarah L. BOOTH, *, Inga PETER 2, Bess DAWSON-HUGHES, Paul A. PRICE 3, Christopher J. O DONNELL 4,5, Udo HOFFMANN 5, Matthew K. WILLIAMSON 3 and Jose M. ORDOVAS USDA Human Nutrition Research Center on Aging, Tufts University, 7 Washington St., Boston, MA 02, USA 2 Department of Genetics and Genomic Science, Mount Sinai School of Medicine, 425 Madison Ave., Box 498, New York, NY 0029, USA 3 Division of Biological Sciences, University of California, San Diego, 9500 Gilman Drive, 0368, La Jolla 92093, USA 4 National Heart, Lung and Blood Institute s Framingham Heart Study, 5 Thurber St., Framingham, MA 0702, USA 5 Cardiovascular CT Core Lab, Massachusetts General Hospital, 5 Fruit St., Boston, MA 024, USA (Received July 25, 2008) Summary Matrix Gla protein (MGP) is a key regulator of vascular calcification. Genetic variation at the MGP locus could modulate the development of coronary artery calcification (CAC). Our aim was to examine the cross-sectional association between MGP single nucleotide polymorphisms (SNPs) [rs800802 (T-38C), rs80080 (G-7A), and rs4236 (Ala02Thr)] and CAC. CAC was measured by multidetector computed tomography (MDCT), in older men and women of European descent, (n 386; 60 to 80 y of age). Serum MGP was measured by radioimmunoassay. Linear, Tobit and Ordinal regression analyses all revealed that in men, homozygous carriers of the minor allele of rs800802, rs80080, or rs4236 (minor allele frequency: 2, 38, and 40%, respectively) were associated with a decreased quantity of CAC, relative to major allele carriers. This association was not found in women. Although genetic variation in MGP was associated with serum MGP concentrations, there were no associations between serum MGP and CAC. The results of this study suggest a role for MGP genetic variants in coronary atherosclerosis among men that is not reflected in serum MGP concentrations. Key Words matrix gla protein, coronary artery calcification, polymorphism, vitamin K, atherosclerosis * To whom correspondence should be addressed. E-mail: sarah.booth@tufts.edu Coronary artery calcification (CAC) is representative of the extent of atherosclerotic burden, and is associated with increased risk of cardiovascular morbidity and mortality (). Biochemical and genetic studies have established matrix Gla protein (MGP) as an inhibitor of calcification in cartilage and blood vessels (2). It has been shown that the anticalcifying activity of MGP depends upon the -carboxylation of specific glutamic acid (Glu) residues in MGP. This vitamin K-dependent reaction yields -carboxyglutamic acid (Gla) residues, which are then able to bind calcium (3). In mice, targeted deletion of the MGP gene causes extensive calcification of the elastic lamellae of the abdominal aorta (2). Extensive vascular calcification is also induced when carboxylation of MGP is inhibited using the vitamin K- antagonist, warfarin (4). Recent studies suggest that in addition to modifiable coronary risk factors, such as hypertension, hyperlipidemia, and cigarette smoking (5), there is a strong genetic component to the development of arterial calcification. The heritability of the presence of CAC has been estimated to be up to 50% (6), with similar heritability estimates given for abdominal aortic calcification (7). Genetic variability in MGP is of particular interest because of its role as an inhibitor of calcification. In vitro studies suggest that single nucleotide polymorphisms (SNPs) in MGP are associated with altered promoter activity (8 0). In addition, there is some evidence that MGP SNPs are associated with arterial calcification (0, ), although these results are not consistent (8, 2). In this study we assessed the cross-sectional associations between genetic variations in MGP and measures of serum MGP and CAC, as measured by multidetector computed tomography (MDCT), in older men and women. We put forth two main hypotheses: First, that genetic variation in MGP will influence circulating concentrations of MGP; and second, that variation in MGP will be associated with CAC. 59

60 CROSIER MD et al. EXPERIMENTAL Study population. Free-living men and postmenopausal women, 60 80 y of age, were enrolled in a 3-y, double-blind, placebo-controlled trial to assess the impact of vitamin K supplementation on age-related bone loss and vascular calcification. Subjects in this study were primarily of European descent (93% of entire cohort). Exclusion criteria included a usual dietary intake of calcium,500 mg/d; usual vitamin D intake greater than,500 IU; women 5 y postmenopausal; a terminal illness; renal or liver disease; a kidney stone in the past 5 y; treatment with a bisphosphonate, calcitonin, hormone replacement, or fluoride (other than dental rinse), current oral anticoagulant use and a history of heart disease or osteoporosis. Of the 452 subjects enrolled in this trial, 36 subjects were excluded due to inability to obtain DNA or written consent for genotyping. The Tufts-New England Medical Center and Massachusetts General Hospital Institutional Review Boards approved the protocol. At the time of the baseline visit prior to randomization, information regarding current medication use, medical history, physical activity, and smoking status was collected. Height and weight were measured while the subjects stood. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. Smoking status was defined as current smokers or non-smokers (Y/N). Current smokers were defined as subjects who reported smoking cigarettes on a regular basis during the previous year. Habitual intakes of phylloquinone, calcium, and vitamin D were assessed by the Willett semi-quantitative food frequency questionnaire (FFQ) (3). Information about vitamin and mineral supplement use was documented on medical forms. Physical activity was estimated by PASE (physical activity score for the elderly) (4). Coronary artery calcification. Coronary calcification imaging was performed on an 8-slice MDCT scanner (Lightspeed Ultra, General Electric, Milwaukee, WI) with a gantry rotation of 500 ms using ECG triggered sequential data acquisition. Scans were prospectively initiated at 50% of the RR interval and data were acquired within a window of 330 ms (temporal resolution). Forty-eight contiguous 3 mm thick slices covering the entire heart were acquired during a single breath hold. A scoring method analogous to the Agaston score (5) was used to quantify coronary calcium as previously described (TeraRecon Aquarius, San Matteo, Ca) (6). This method has been shown to be a sensitive and accurate predictor of coronary calcium (6). Biochemical measurements. Fasting blood samples ( 0 h) were collected and shipped in a frozen state to the University of California, San Diego where serum MGP was assayed using a radioimmunoassay, as previously described (7). This assay measures total concentrations of MGP in serum, and does not discriminate between the fully -carboxylated form of MGP (active as a calcification inhibitor) (4) and under- -carboxylated forms of the protein. Plasma 25-hydroxyvitamin D was analyzed by commercially-available 25(OH)D assay kits available from DiaSorin (Stillwater, MN). The DiaSorin 25(OH)D assay consists of a two-step procedure: extraction of 25(OH)D from plasma with acetonitrile, followed by an RIA procedure. Two reference controls provided with the kit and one in-house control were run in each assay and compared to acceptable ranges. Triglycerides were measured on a COBAS Mira (Roche Instruments, Belleville, NJ). Plasma lipid profiles include enzymatic measurement of total, LDL, and HDL cholesterol and triglyceride concentrations (8). Genotyping. Four hundred and sixteen subjects provided written, informed consent for genotyping. Genomic DNA was isolated from blood and purified for PCR analysis using the QIAamp DNA Mini Kit (Qiagen Inc., Chatsworth, CA). Genotyping was carried out using the ABI prism 7900 using single nucleotide polymorphism (SNP) discrimination assays designed by Applied Biosystems (ABI) (Applied Biosystems, Foster City, CA). Accuracy of genotype was tested by introducing 5% dummy duplicates and blank controls. The MGP SNPs genotyped for this study include dbsnp rs800802 (T-38C) located in the promoter region of the gene; dbsnp rs80080 (G-7A); located in the 5 UTR and dbsnp rs4236, a non-synonymous polymorphism located in exon 4 and causing an alanine to threonine amino acid change (Fig. ). PCR conditions and primers and probes used for each assay are Fig.. Scheme of the gene encoding matrix Gla protein. Black boxes indicate exons; solid black lines indicate introns; Single nucleotide polymorphisms (SNP) and minor allele frequency are shown. The entire MGP locus is part of a unique DNA block and the three SNPs are part of this bock with highly significant LD values (Lewontin s D 0.98, p 0.00).

MGP Polymorphisms 6 Table. Study participant characteristics. Men (n 60) Women (n 226) p Age, y 68.9 5.6 67.8 5.3 0.038 Serum matrix Gla protein, ng/ml 200 43.8 20 5 0.862 Agaston score 34.5 623.9 25 286 0.00 BMI, kg/m 2 28. 4.3 28.0 5.4 0.845 Total cholesterol, mg/dl 88.6 33.8 24.9 36.6 0.00 HDL cholesterol, mg/dl 50.3 2.8 6. 5.3 0.00 LDL cholesterol, mg/dl 4.5 29.2 3.5 29.5 0.00 Triglycerides, mg/dl 20.5 76.3 3.6 66. 0.309 Serum 25(OH)D, ng/ml 22.6 8.7 24. 8.3 0.084 Systolic blood pressure, mmhg 35 6 30 6 0.004 Current smoking, % 4.9 7.0 0.595 Diabetic, % 5.0 7.9 0.257 Statin use, % 24 26 0.648 Values are presented as mean SD unless otherwise noted. Only participants of Northern European descent were included in this study. Continuous traits were compared using the t-test; smoking status, diabetes status and statin use were compared using the 2 test. listed in Supplemental Table. Statistical analysis. The genotype frequencies for all markers were evaluated for Hardy-Weinberg Equilibrium (HWE) using the 2 tests. Linkage disequilibrium (LD) between each pair of SNPs within MGP was evaluated using Lewontin s D (9). Haplotypes were inferred by the EM-algorithm using the Helixtree software package (Golden Helix, Inc., Boseman, MT). The prevalence of coronary calcification has been shown to differ among ethnic groups by others, with men and women of European descent having the highest prevalence (20). Therefore these analyses were limited to only participants of European descent (n 386). Due to the skewed distribution of CAC, Tobit regression of the logged Agaston score [ln(cac )], proven to accurately identify associations of CAC with risk factors (2, 22), was our primary analytical approach. Linear regression of the logged Agaston score [ln(cac )] and ordinal logistic regression with the CAC outcome expressed in ordinal categories have also been shown to accurately identify associations of CAC with risk factors (2, 22). For consistency of interpretation and comparison of results across studies, we also used these additional analytical approaches in our analyses. All analyses were performed for men and women separately because previous reports indicated sex-dependent associations between risk factors and CAC (2, 23). All models are adjusted for age, BMI, LDL cholesterol, HDL cholesterol, smoking status, systolic blood pressure, diabetes (Y/N), and use of statins. The results of Tobit regression are presented as the ratio of CAC score for presence or absence of a particular genotype (e.g. ratio of CAC score in carriers of genotype A versus carriers of genotype B). For ordinal regression, CAC was divided into 5 categories, using published criteria to approximate the presence of no, minimal, mild, moderate, and severe coronary atherosclerosis (CAC scores of 0, 0, 00, 0 400, and 400) (24). SNP-gender interactions were explored in fully-adjusted multivariate Table 2. MGP SNP MGP allele frequency. Allele analysis that included both men and women. To evaluate the relationships between serum levels of MGP and polymorphisms in the MGP gene, we used analysis of covariance (ANCOVA) while adjusting for age, gender, serum triglycerides, BMI, smoking status, and serum 25(OH)D. Serum MGP was the outcome variable of interest and each MGP marker was used as a primary independent variable. The SNPs associated with serum MGP (p 0.0) in the fully adjusted models were further analyzed assuming dominant, recessive, or additive modes of inheritance. Statistical analysis of Tobit regression was performed using SAS (v. 9, Cary, NC). Linear regression of ln(cac ) and ordinal regression was performed in SPSS for Windows (v. 4, Chicago, IL). Tests with p- value 0.05 were considered statistically significant. RESULTS Men n (%) Women n (%) rs800802 (T-38C) AA 03 (64) 35 (60) AG 47 (29) 82 (36) GG 0 (6) 0 (4) rs80080 (G-7A) GG 57 (36) 94 (4) AG 77 (48) 00 (44) AA 26 (6) 33 (5) rs4236 (Ala02Thr) TT 54 (34) 87 (38) TC 78 (49) 02 (45) CC 28 (7) 38 (7) Reverse DNA orientation used for genotyping. Characteristics of the 226 and 60 women and men, respectively, who were exclusively of Northern European descent and were selected for the present study are presented in Table. The mean age of the subjects was 68 5.5 y. Although subjects were excluded from this

62 CROSIER MD et al. study for major health problems such as liver, renal, or heart disease, a mean BMI of 28 kg/m 2 suggests that a large number were overweight or obese. Gender-specific differences were detected in total, HDL- and LDL-cholesterol and systolic blood pressure. Serum MGP concentrations were not significantly different between men and women. However, as expected, men had a significantly higher mean Agaston score than women (Table ). Genotype frequencies are presented in Table 2. Genotype frequency distributions were in accordance with HWE expectations (p 0.05). Our analyses revealed a highly significant LD (p 0.00) among the three SNPs examined with D values higher than 0.98. These three SNPs are part of the same DNA block encompassing the entire MGP locus. Determinants of serum MGP and CAC including age, serum cholesterol, triglyceride, BMI, 25(OH)D, systolic blood pressure and smoking status were not significantly different in participants of European descent genotyped for this study as compared to those for whom there were no genotype data available (data not shown). Table 3. Tobit analysis of the association between MGP SNPs and CAC.,2 MGP rs Alleles Ratio ln(cac ) (CI) Men rs800802 3 AA/AG vs. GG 0. (0.03 0.49) 0.004 rs80080 GG/GA vs. AA 0.28 (0. 0.69) 0.005 rs4236 TT/TC vs. CC 0.32 (0.3 0.77) 0.0 Women rs800802 3 AA/AG vs. GG 5.94 (0.63 55.7) 0.2 rs80080 GG/GA vs. AA 0.65 (0.6 2.74) 0.56 rs4236 TT/TC vs. CC 0.88 (0.23 3.38) 0.85 Results of Tobit regression are presented as the ratio of CAC score for presence or absence of a particular genotype (e.g. ratio of CAC score in carriers of genotype A versus carriers of genotype B). 2 Results are relative to carriers of the major allele for each SNP (rs800802: A; rs80080: G; rs4236: T). All analyses are adjusted for age, diabetes status, smoking status, BMI, HDL cholesterol, LDL cholesterol, systolic blood pressure, and use of statins. 3 Significant SNP * gender interaction (p 0.000). p All three MGP SNPs analyzed in this study were significant predictors of CAC in men, either alone or after adjusting for other CHD risk factors (Table 3). These SNPs were not significantly associated with CAC in women. Significant associations remained in men after adjusting for serum MGP in addition to other CHD risk factors. The strongest associations were found with the recessive mode of inheritance (recessive for the minor allele). In men, minor allele homozygotes of rs800802, rs80080, or rs4236 were associated with a decreased quantity of CAC, relative to carriers of the major alleles. A significant gender*snp interaction was identified for SNP rs800802. Minor allele homozygotes were associated with lower CAC in men and higher (although not significant) CAC in women. Additional analyses using linear regression of ln(cac ) and ordinal regression supported the results of Tobit analysis (data not shown). General linear models showed significant associations of MGP rs80080 and rs4236 with serum concentrations of MGP in men and women combined (overall p 0.002 and p 0.028, respectively). Genderspecific analyses showed similar trends, and there was not a significant gender*snp interaction with any SNP. Therefore, to increase power, analyses were conducted in men and women combined. The minor allele homozygotes (AA for rs80080 and CC for rs4236) had lower serum concentrations of MGP compared to major allele homozygotes, with heterozygotes having intermediate values [(adjusted mean MGP concentrations (ng/ml): rs80080 GG 20.4, AG 96.0, AA 89.8; rs4236 TT 208.3, TC 97.7, CC 92.3)]. Further analysis of these SNPs showed the best performing model with the additive mode of inheritance (rs80080 p 0.00, rs4236 p 0.009) (Fig. 2). In contrast, MGP rs800802 did not show a significant association with serum MGP concentrations. To further explore the associations among MGP SNPs, CAC and serum MGP, we performed haplotype analyses. In haplotype analysis of the MGP gene using the three SNPs, we identified three common haplotypes (frequency 5%): Hap-rs800802 A-rs80080 A- rs4236 C (frequency 37%); Hap2-rs800802 A- rs80080 G-rs4236 T (frequency 37%), and Hap3- Fig. 2. Histograms of mean SE MGP concentrations (ng/ml) relative to MGP SNPs. General linear models were used to compare serum concentrations of MGP by MGP SNPs in men and women combined. The additive model for rs80080 (p 0.00); rs4236 (p 0.009); and rs800802 (p 0.860) Genotypes with different superscripts are significantly different (p 0.05), with serum MGP concentrations indicated by a higher than those indicated by b.

MGP Polymorphisms 63 Table 4. CAC. Associations between MGP haplotypes and Haplotype Parameter estimate p Men Hap vs. Hap 2 3.4 0.023 Hap vs. Hap 3.35 0.60 Hap 2 vs. Hap 3 2.4 0.39 Women Hap vs. Hap 2.24 0.648 Hap vs. Hap 3.52 0.456 Hap 2 vs. Hap 3 0.82 0.733 Hap-rs800802 A-rs80080 A-rs4236 C; Hap2-rs800802 A-rs80080 G-rs4236 T; Hap3-rs8008024 A-rs80080 A-rs4236 C. Table 5. Associations between MGP haplotypes and serum MGP concentrations. Haplotype Parameter estimate p Hap vs. Hap 2 27.3 0.00 Hap vs. Hap 3 4. 0.096 Hap 2 vs. Hap 3 3.2 0.34 Hap-rs800802 A-rs80080 A-rs4236 C; Hap2-rs800802 A-rs80080 G-rs4236 T; Hap3-rs8008024 A-rs80080 A-rs4236 C. rs8008024 A-rs80080 A-rs4236 C (frequency 22%) (Fig. ). In terms of CAC, Hap2 was significantly higher than Hap in men only (3.4 unit difference per each copy; p 0.023) (Table 4). Haplotype analysis for serum MGP concentrations was conducted in men and women combined for consistency with the analyses of individual SNPs for this outcome. For serum MGP concentrations, Hap2 was significantly higher than Hap (27.3 ng/ml difference per each copy; p 0.00) (Table 5) The direction of haplotype analyses is consistent with the single SNP results; however, haplotype analyses did not increase the variance explained by the models relative to analyses of single SNPs in any of the models. DISCUSSION In the present study, we found gender specific associations between genetic variation in MGP and CAC. In our study, we used robust measurement and statistical analysis of CAC to show associations with genetic variation in MGP in older, healthy men and women of European descent. Our measurement of CAC, multidetector computed tomography, has been shown to be a sensitive and accurate predictor of coronary calcium in older men and women (6). Tobit and linear regression and ordinal logistic regression of the skewed CAC scores have proven to accurately identify associations of CAC with atherosclerotic risk factors (2, 22). We used these methods to show significant associations of MGP SNPs with CAC in men, such that homozygous carriers of the minor alleles of MGP rs800802, rs80080, or rs4236 are associated with decreased quantity of CAC, relative to carriers of the major alleles. Previous studies have been inconsistent in their findings of an association between MGP genetic variation and atherosclerosis or vascular calcification. In one case-control study, the outcome measures of myocardial infarction and ultrasound femoral and carotid artery calcification were assessed for an association with MGP rs4236 and rs80080. These SNPs were weakly associated with femoral calcification in a subset of subjects who had the presence of atherosclerotic plaques (0). In the Coronary Artery Risk Development in Young Adults (CARDIA) study, a population-based investigation of cardiovascular disease in younger African-American and non-hispanic white participants, MGP rs800802 (T-38C) was analyzed for association with presence or absence of CAC (2). This SNP was also studied in autopsy samples for an association with several measures of atherosclerotic calcification, including chest and abdominal X-rays (8). However, no association with MGP rs800802 and measures of calcification was found in either of these studies. This could be due to population stratification or sensitivity of calcification measurement. Vascular calcification progresses with aging; therefore a young cohort, as used in the CARDIA study, may have little calcification, which would decrease the likelihood of finding an association with MGP genotype. Furthermore, X-rays and other measures of calcification such as the Gore index, as used by Kobayashi et al. (8), may not be sensitive enough for this type of outcome measure. In our study, significant associations were found in men only. Men initiate vascular calcification earlier in life than women, and in our study, men had a higher calcification score than women (mean Agaston score in men: 34; women: 25). Therefore it is plausible that the associations between these MGP SNPs and CAC may only manifest in subjects with high amounts of calcification, which in this study was limited to men. A significant gender*snp interaction was found for MGP rs800802 but not the other SNPs analyzed in this study. Although there is evidence of altered promoter activity for MGP rs80080 and rs800802, and MGP rs4236 causes a non-polar to polar amino acid change (alanine to threonine), it is unknown whether the associations with CAC are due to functional consequences of these polymorphisms or if these polymorphisms are in linkage with other functional loci. In addition to associations with CAC, genetic variation in MGP was associated with serum MGP concentrations. In a previous in vitro study of MGP SNPs, the AA variant of rs80080 (G-7A) was shown to have a.5 higher transcriptional activity compared to the GG variant (9). In addition, the CC variant of rs800802 (T-38C) was shown to have a 4-fold higher transcriptional activity compared to the TT variant (9). The increased expression of the rs800802 CC allele corresponded to a higher mean serum MGP concentration in a gene-dose relationship (9). Although we did not find a significant association between MGP rs800802 and serum MGP, we did find significant associations with rs80080 and rs4236 (A83T). Like the associations

64 CROSIER MD et al. found with CAC, it is not known whether these associations are due to functional consequences of these polymorphisms, or if these polymorphisms are in linkage with other functional loci. The significance of an association of MGP SNPs with serum MGP is not fully understood. Serum MGP has been shown to be elevated in atherosclerotic patients (25), while inversely correlated with the severity of CAC in another study (26). Expression of MGP increases in calcified regions of the vascular wall and MGP has been shown to accumulate at sites of calcification (4). It is thought that this increased accumulation of MGP slows the progression of calcification, and that a fraction of this increased synthesis escapes to serum where it causes an elevation in serum MGP. However, it has been shown that increased serum levels of MGP failed to rescue vascular calcification due to targeted MGP deletion (3). MGP is a vitamin K-dependent protein, and as such, must be carboxylated to be fully functional. The currently available MGP assay measures total serum MGP, and does not discriminate between the under-carboxylated and fully-carboxylated forms of MGP, making the interpretation of these findings more difficult. A modest, but inconsistent association between increasing concentrations of MGP and higher levels of coronary calcium deposit were reported in this entire cohort (27). Analysis of the subset of subjects of European descent for whom genotype data were available in our study, revealed no significant associations between serum MGP and CAC in men or women. These inconsistencies in the association between serum MGP and CAC underscore the conclusions of the original study that serum MGP is not a robust biochemical measure for vascular calcification. In addition, the significant associations between MGP SNPs and CAC remained unchanged after adjusting for serum MGP concentrations, in addition to other cardiovascular risk factors, suggesting that the associations between these MGP SNPs and CAC are independent of serum MGP. Previous studies have shown ethnic differences in the prevalence and quantity of calcification (20). We therefore limited our analyses to only subjects of European descent 60 80 y of age. This strengthens our analyses; however, at the same time it limits how generalizable our results are to other ethnicities and age groups. The results of our study lend further support for a mechanistic connection between MGP genetic variants and coronary atherosclerosis that are not reflected in serum MGP concentrations. Further studies are necessary for replication and to fully elucidate the mechanisms behind these associations. These associations also need to be explored among different populations. Acknowledgments Based upon work supported by the U.S. Department of Agriculture, Agricultural Research Service under Cooperative Agreement No. 58-950-7-707, National Institute of Aging (AG4759, HL69272 and NIDDK T32DK62032). Any opinions, findings, conclusions or recommendations expressed in this publication are those of the authors, and do not necessarily reflect the view of the U.S. Department of Agriculture. REFERENCES ) Stanford W. 999. Coronary artery calcification as an indicator of preclinical coronary artery disease. Radiographics 9: 409 49. 2) Luo G, Ducy P, McKee MD, Pinero GJ, Loyer E, Behringer RR, Karsenty G. 997. Spontaneous calcification of arteries and cartilage in mice lacking matrix GLA protein. Nature 386: 78 8. 3) Murshed M, Schinke T, McKee MD, Karsenty G. 2004. Extracellular matrix mineralization is regulated locally: different roles of two gla-containing proteins. J Cell Biol 65: 625 630. 4) Price PA, Faus SA, Williamson MK. 998. Warfarin causes rapid calcification of the elastic lamellae in rat arteries and heart valves. Arterioscler Thromb Vasc Biol 8: 400 407. 5) Wilson PW, D Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. 998. Prediction of coronary heart disease using risk factor categories. Circulation 97: 837 847. 6) Post W, Bielak LF, Ryan KA, Cheng YC, Shen H, Rumberger JA, Sheedy PF 2nd, Shuldiner AR, Peyser PA, Mitchell BD. 2007. Determinants of coronary artery and aortic calcification in the Old Order Amish. Circulation 5: 77 724. 7) O Donnell CJ, Chazaro I, Wilson PW, Fox C, Hannan MT, Kiel DP, Cupples LA. 2002. Evidence for heritability of abdominal aortic calcific deposits in the Framingham Heart Study. Circulation 06: 337 34. 8) Kobayashi N, Kitazawa R, Maeda S, Schurgers L, Kitazawa S. 2004. T-38C polymorphism of matrix gla protein promoter alters its expression but is not directly associated with atherosclerotic vascular calcification. Kobe J Med Sci 50: 69 8. 9) Farzaneh-Far A, Davies JD, Braam LA, Spronk HM, Proudfoot D, Chan SW, O Shaughnessy KM, Weissberg PL, Vermeer C, Shanahan CM. 200. A polymorphism of the human matrix gamma-carboxyglutamic acid protein promoter alters binding of an activating protein- complex and is associated with altered transcription and serum levels. J Biol Chem 276: 32466 32473. 0) Herrmann SM, Whatling C, Brand E, Nicaud V, Gariepy J, Simon A, Evans A, Ruidavets JB, Arveiler D, Luc G, Tiret L, Henney A, Cambien F. 2000. Polymorphisms of the human matrix gla protein (MGP) gene, vascular calcification, and myocardial infarction. Arterioscler Thromb Vasc Biol 20: 2386 2393. ) Brancaccio D, Biondi ML, Gallieni M, Turri O, Galassi A, Cecchini F, Russo D, Andreucci V, Cozzolino M. 2005. Matrix GLA protein gene polymorphisms: clinical correlates and cardiovascular mortality in chronic kidney disease patients. Am J Nephrol 25: 548 552. 2) Taylor BC, Schreiner PJ, Doherty TM, Fornage M, Carr JJ, Sidney S. 2005. Matrix Gla protein and osteopontin genetic associations with coronary artery calcification and bone density: the CARDIA study. Hum Genet 6: 525 528. 3) Willitt W. 990. Nutrition Epidemiology. Oxford Press, New York. 4) Washburn RA, Ficker JL. 999. Physical activity scale for the elderly (PASE): the relationship with activity measured by a portable accelerometer. J Sports Med Phys

MGP Polymorphisms 65 Fitness 39: 336 340. 5) Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. 990. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 5: 827 832. 6) Cury RC, Ferencik M, Hoffmann U, Ferullo A, Moselewski F, Abbara S, Booth SL, O Donnell CJ, Brady TJ, Achenbach S. 2004. Epidemiology and association of vascular and valvular calcium quantified by multidetector computed tomography in elderly asymptomatic subjects. Am J Cardiol 94: 348 35. 7) Otawara Y, Price PA. 986. Developmental appearance of matrix GLA protein during calcification in the rat. J Biol Chem 26: 0828 0832. 8) McNamara JR, Schaefer EJ. 987. Automated enzymatic standardized lipid analyses for plasma and lipoprotein fractions. Clin Chim Acta 66: 8. 9) Lewontin RC. 964. The interaction of selection and linkage. Optimum models. Genetics 50: 757 782. 20) Budoff MJ, Nasir K, Mao S, Tseng PH, Chau A, Liu ST, Flores F, Blumenthal RS. 2006. Ethnic differences of the presence and severity of coronary atherosclerosis. Atherosclerosis 87: 343 350. 2) Reilly MP, Wolfe ML, Localio AR, Rader DJ. 2004. Coronary artery calcification and cardiovascular risk factors: impact of the analytic approach. Atherosclerosis 73: 69 78. 22) Reilly MP, Wolfe ML, Localio AR, Rader DJ. 2003. C- reactive protein and coronary artery calcification: The Study of Inherited Risk of Coronary Atherosclerosis (SIRCA). Arterioscler Thromb Vasc Biol 23: 85 856. 23) Newman AB, Naydeck BL, Sutton-Tyrrell K, Feldman A, Edmundowicz D, Kuller LH. 200. Coronary artery calcification in older adults to age 99: prevalence and risk factors. Circulation 04: 2679 2684. 24) Rumberger JA, Brundage BH, Rader DJ, Kondos G. 999. Electron beam computed tomographic coronary calcium scanning: a review and guidelines for use in asymptomatic persons. Mayo Clin Proc 74: 243 252. 25) Braam LA, Dissel P, Gijsbers BL, Spronk HM, Hamulyak K, Soute BA, Debie W, Vermeer C. 2000. Assay for human matrix gla protein in serum: potential applications in the cardiovascular field. Arterioscler Thromb Vasc Biol 20: 257 26. 26) Jono S, Ikari Y, Vermeer C, Dissel P, Hasegawa K, Shioi A, Taniwaki H, Kizu A, Nishizawa Y, Saito S. 2004. Matrix Gla protein is associated with coronary artery calcification as assessed by electron-beam computed tomography. Thromb Haemost 9: 790 794. 27) O Donnell CJ, Shea MK, Price PA, Gagnon DR, Wilson PW, Larson MG, Kiel DP, Hoffmann U, Ferencik M, Clouse ME, Williamson MK, Cupples LA, Dawson- Hughes B, Booth SL. 2006. Matrix Gla protein is associated with risk factors for atherosclerosis but not with coronary artery calcification. Arterioscler Thromb Vasc Biol 26: 2769 2774. Supplemental Table. Details of SNP genotyping. SNP Primers Probe MGP rs800802 MGP rs80080 MGP rs4236 F: GCACTGAACTAGCATTGGAACTTT R: GAAAACTGCCCACTCAGAGTAGATA F: TGTAGCAGCAGTAGGGAGAGA R: CCTCTCAACTGCTCTGGTTCTTAT F: GGTGCCAGCCTCCAGAAA R: ATGCTGCCTATAATCGCTACTTCAG TCCCAAAC[A/G]GTCATTC CACTAACATCCC[G/A]TAGGAG TCTCATTTGG[T/C]CCCTCG 5 ng DNA per 5 L reaction. 40 cycles (92 C denature 5 s; 60 C anneal/extend 60 s).