Clinical and Population Studies

Identification of Genetic Variants Linking Protein C and Lipoprotein MetabolismHighlights

The ARIC Study (Atherosclerosis Risk in Communities)

James S. Pankow, Weihong Tang, Nathan Pankratz, Weihua Guan, Lu-Chen Weng, Mary Cushman, Eric Boerwinkle, Aaron R. Folsom
https://doi.org/10.1161/ATVBAHA.116.308109
Arteriosclerosis, Thrombosis, and Vascular Biology. 2017;37:589-597
Originally published January 12, 2017

Jump to

Abstract

Objective—Previous studies have identified common genetic variants in 4 chromosomal regions that together account for 14% to 15% of the variance in circulating levels of protein C. To further characterize the genetic architecture of protein C, we obtained denser coverage at some loci, extended investigation of protein C to low-frequency and rare variants, and searched for new associations in genes known to influence protein C.

Approach and Results—Genetic associations with protein C antigen level were evaluated in ≤10 778 European and 3190 black participants aged 45 to 64 years. Analyses included >26 million autosomal variants available after imputation to the 1000 Genomes reference panel along with additional low-frequency and rare variants directly genotyped using the Illumina ITMAT-Broad-CARe chip and Illumina HumanExome BeadChip. Genome-wide significant associations (P<5×10−8) were found for common variants in the GCKR, PROC, BAZ1B, and PROCR-EDEM2 regions in whites and PROC and PROCR-EDEM2 regions in blacks, confirming earlier findings. In a novel finding, the low-density lipoprotein cholesterol–lowering allele of rs12740374, located in the CELSR2–PSRC1–SORT1 region, was associated with lower protein C level in both whites and blacks, reaching genome-wide significance in a meta-analysis combining results from both groups (P=1.4×10−9). To further investigate a possible link between lipid metabolism and protein C level, we conducted Mendelian randomization analyses using 185 lipid-related genetic variants as instrumental variables. The results indicated that triglycerides, and possibly low-density lipoprotein cholesterol, influence protein C levels.

Conclusions—Discovery of variants influencing circulating protein C levels in the CELSR2–PSRC1–SORT1 region may indicate a novel genetic link between lipoprotein metabolism and hemostasis.

Introduction

Protein C is a vitamin K–dependent glycoprotein produced by the liver that circulates in plasma as a zymogen. Once activated by the thrombin–thrombomodulin complex, protein C is a multifunctional molecule with anticoagulant, anti-inflammatory, antiapoptotic, and cytoprotective properties. Activated protein C inactivates factor Va and factor VIIIa, thereby reducing the generation of thrombin. Individuals with protein C deficiency, characterized by protein C antigen or activity levels in the lowest 1% to 2% of the distribution, have been found to have 3- to 7-fold higher rates of venous thromboembolism (VTE) compared with individuals in the normal range of protein C,1,2 and 2% to 3% of VTE cases in the population may be attributable to protein C deficiency.2 Low protein C levels are also associated with higher risk of ischemic stroke3 and presence of cerebral infarctions detected by magnetic resonance imaging.4

Twin and family studies have provided heritability estimates of 40% to 60% for blood levels of protein C,5 and autosomal dominant mutations in the protein C structural gene (PROC) are a known cause of protein C deficiency in some families,6 suggesting important genetic contributions to this trait. Recent genome-wide association analyses in the ARIC study (Atherosclerosis Risk in Communities) identified 504 single-nucleotide polymorphisms (SNPs) associated with protein C levels in whites7 and 79 SNPs associated with protein C levels in blacks.8 In whites, associations with common variants (minor allele frequency >0.05) were found in 4 genomic regions located near PROCR (protein C receptor), PROC, GCKR, and BAZ1B, with a possible second independent association found at EDEM2, located close to PROCR. In blacks, genome-wide significant associations were found at PROCR and PROC but not at GCKR or BAZ1B.8 Genome-wide association studies (GWAS) in French5 and Spanish9 populations have also identified significant associations in the PROCR region. Taken together, these previously discovered variants account for 14% to 15% of the variability in protein C antigen level.7 Of the 8 susceptibility genes found in the largest meta-analysis of VTE GWAS conducted to date, 2 loci (F5 and PROCR) were also significantly associated with protein C level,10 suggesting that protein C is a useful intermediate phenotype to identify genetic determinants of VTE.

Previous GWAS of protein C level used commercial genotyping arrays designed to capture common genetic variation, but targeted sequencing of PROC and other genes involved in coagulation has revealed a spectrum of variation from rare to common.11 To obtain denser coverage at some loci, extend investigation of protein C to low-frequency and rare variants, and search for second independent associations in genes known to influence protein C, we tested >26 million variants available after imputation to the 1000 Genomes Project (1000G) reference panel along with additional variants directly genotyped using the Illumina ITMAT-Broad-CARe (IBC) array12 and Illumina Human Exome BeadChip.13

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.

Results

Available sample sizes with phenotypic and genotypic data were 10 778 whites and 3190 blacks for the exome array, 9733 whites and 2803 blacks for the IBC array, and 8911 whites and 2706 blacks for the GWAS array. The overlap of participants across the 3 genotyping platforms is summarized in Figure I in the online-only Data Supplement. Characteristics of participants available for exome array, the largest sample size of the 3 platforms, are shown in Table 1. The average age at the time of blood collection was 54 years, and mean protein C levels were similar in whites and blacks.

View this table:
Table 1.

Baseline Characteristics of Participants Included in Exome Array Analyses

Table 2 summarizes top SNP associations with protein C level identified by IBC, exome array, and 1000G GWAS analyses. In whites, all 4 of the most strongly associated regions (2p23/GCKR, 2q13-q14/PROC, 7p13/BAZ1B, and 20q11/PROCR) were the same as those previously identified by previous GWAS analyses in ARIC using the Affymetrix 6.0 array with imputation to a HapMap reference panel.7 In the 2p23 region, the top SNP identified with the IBC array (P=7.8×10−21), exome array (P=7.7×10−24), and 1000G GWAS (P=1.2×10−16) was a nonsynonymous variant (rs1260326; P446L) in GCKR. No other SNPs in this region were genome-wide significant after conditioning on rs1260326. In the 2q13-q14 region, the top SNP identified with the exome array (P=1.4×10−50) was an intronic variant (rs1158867) in PROC. In analysis of the IBC array and GWAS data, another SNP (rs1799810) in tight linkage disequilibrium with rs1158867 (r2=1) had the most extreme P value for association with protein C level. After conditioning on rs1158867 or rs1799810, no other SNPs in this region exceeded the genome-wide significance threshold. In the 7p13 region, the top SNP on the exome array is located in an intron of BAZ1B (rs1178979; P=2.3×10−8); the top SNP in the 1000G GWAS was an intergenic variant near FZD9 (rs42238; P=4.8×10−9). These variants are in moderate-to-high linkage disequilibrium (r2=0.7), and the conditional analyses in the 1000G GWAS data set indicate that they represent the same signal. This region was not captured on the IBC array. No other SNPs in this region reached the genome-wide significance threshold after conditioning on rs1178979 or rs42238. In the 20q11 region, the top SNP identified with both the exome array (P=3.5×10−287) and IBC array (P=1.9×10−249) was a nonsynonymous variant (rs867186) in PROCR; a noncoding SNP (rs11907011) downstream of PROCR had the most extreme P value in the 1000G GWAS (P=1.4×10−237) and was in tight linkage disequilibrium with rs867186 (r2=1). After conditioning on rs867186, additional SNPs in the region were genome-wide significant. These included an intronic variant in EDEM2 on the exome array (rs6120849; conditional P=7.3×10−21), an intergenic variant between EDEM2 and PROCR on the IBC array (rs6060278; conditional P=1.7×10−19), and an intronic variant in EDEM2 in the 1000G GWAS (rs6120848; conditional P=5.3×10−17; Table 3). Pairwise linkage disequilibrium estimates are high (r2>0.9) between these 3 SNPs identified in conditional analysis, suggesting that they were capturing the same second independent association. After adjusting for these additional SNPs, no other SNPs in the 20q11 region met the genome-wide significance threshold.

View this table:
Table 2.

Top Genome-Wide Associations for Protein C Level, by Race/Ethnicity and Genotyping Platform

View this table:
Table 3.

Evidence for a Second Independent Association in the PROCR-EDEM2 Region Identified by Conditional Analyses in Whites

In blacks, SNPs in 3 regions reached the significance threshold (Table 2). Two of these regions (2q13-q14/PROC and 20q11/PROCR) were the same as those identified by the previous GWAS of protein C with HapMap imputation in the ARIC study.8 In the 2q13-q14 region, the top SNP on the IBC array was in the 3′UTR (untranslated region) of MAP3K2 (rs423353; P=7.2×10−11), whereas a low-frequency variant in CYP27C1 (rs140608390) had the most significant association in the 1000G GWAS (P=6.7×10−24). In the 20q11 region, the top SNP (rs867186) in the IBC array, exome array, and 1000G GWAS was the same as the top SNP in whites. In blacks, no other genome-wide significant SNPs were found in the 2q13-q14 or 20q11 regions after conditioning on the top SNPs in each region. In the 1000G GWAS, a potentially novel association was found with rs148639156 (P=4.8×10−9), a low-frequency variant (minor allele frequency=0.02) near SAYSD1 (SAYSVFN motif domain containing 1). This SNP was not available in whites, and so results could not be corroborated.

A total of 11 genes in whites and 2 genes in blacks reached genome-wide significance (P<2.5×10−6) in at least one of the gene-based tests (SKAT [sequence kernel association test], T1, or T5 burden tests) used to analyze rare and low-frequency nonsynonymous variants on the exome array (Table I in the online-only Data Supplement). The 2 genes (CPNE1 and C20orf152) that were significant in both SKAT and T5 burden tests were <800 kb from PROCR on chromosome 20. We repeated the gene-based tests after conditioning on top SNP associations for protein C (rs1260326 [GCKR], rs1158867 [PROC], rs1178979 [BAZ1B], and rs867186 [PROCR]) in whites. None of the gene-based tests was significant after adjustment for these 4 SNPs; the gene with the most extreme P value in the conditional analysis was PROCR (P=3.9×10−3 for SKAT; P=5.3×10−5 for T5 burden test). We further examined single SNP tests (unconditional analysis) for 5 rare and low-frequency nonsynonymous variants in PROCR. The strongest association was found for rs150846093 (P=7.4×10−4), which results in a substitution of proline for serine.

Several common variants located at CELSR2–PSRC1–SORT1 in the 1q13 region showed consistent associations with protein C level in both whites and blacks, although they did not reach the threshold for genome-wide significance in either group. Regional association plots are provided in Figure 1 (whites) and Figure 2 (blacks). The variant demonstrating the strongest association in this region was rs12740374, located in the 3′UTR of CELSR2 (Table 2). Using data from the exome array, the minor allele (T) of rs12740374 was associated with a 0.05 μg/mL lower protein C level in whites (P=2.2×10−7) and 0.06 μg/mL lower level in blacks (P=4.5×10−4), and this variant reached genome-wide significance in a meta-analysis combining results from both groups (P=1.4×10−9), with no significant heterogeneity by race/ethnicity (P=0.72). Interestingly, rs12740374 was one of the variants most strongly associated with plasma low-density lipoprotein cholesterol (LDL-C) levels in the GWAS meta-analysis conducted by the Global Lipids Consortium.14 In the ARIC study, the T allele of rs12740374 was associated with an average reduction of LDL-C of 0.18 mmol/L (P=7.2×10−32) in whites and 0.19 mmol/L (P=1.0×10−8) in blacks. The new locus explains an additional 0.3% of the variance of protein C in both whites and blacks. Taken together, rs12740374 (CELSR2–PSRC1–SORT1), rs1260326 (GCKR), rs1158867 (PROC), rs1178979 (BAZ1B), and rs867186 (PROCR) explain 14.1% of the variance in whites and 9.2% in blacks.

Figure 1.
Figure 1.

Regional association plot of observed P values in –log10 scale for association with protein C levels in whites, using single-nucleotide polymorphisms (SNPs) from the IBC (ITMAT-Broad-CARe) array on chromosome 1 (NCBI build 36). Each blue triangle represents the top associated SNP in the corresponding region. Color (red, orange, and yellow) indicates linkage disequilibrium (r2, based on HapMap CEU sample) with the top SNP: red= r2≥0.8, orange= r2≥0.5 but <0.8, yellow=r2≥0.2 but <0.5).

Figure 2.
Figure 2.

Regional association plot of observed P values in –log10 scale for association with protein C levels in blacks, using single-nucleotide polymorphisms (SNPs) from the IBC (ITMAT-Broad-CARe) array on chromosome 1 (NCBI build 36). Each blue triangle represents the top associated SNP in the corresponding region. Color (red, orange, and yellow) indicates linkage disequilibrium (r2, based on HapMap YRI sample) with the top SNP: red= r2≥0.8, orange= r2≥0.5 but <0.8, yellow= r2≥0.2 but <0.5).

Protein C and LDL-C levels were modestly correlated in ARIC (r2=0.23). To further explore possible inter-relationships between protein C and LDL-C levels, we fit a series of linear regression models, first with protein C as the dependent variable and rs12740374 and LDL-C as predictors and then with LDL-C as the dependent variable and rs12740374 and protein C as predictors. When protein C was modeled as the dependent variable (Table 4), the regression coefficient per copy of the T allele at rs12740374 was attenuated by 46% (from −0.0507 to −0.0272 μg/mL) when LDL-C was added as a covariate to age- and sex-adjusted models in whites. By contrast, when LDL-C was modeled as the dependent variable (Table 5), the regression coefficient for rs12740374 was attenuated by only 10% (from −0.188 to −0.169 mmol/L) when protein C was added to the model. Nearly identical patterns of attenuation were observed in blacks (Tables 4 and 5).

View this table:
Table 4.

Linear Regression of Protein C Level on rs12740374 and LDL-C Level, Stratified by Race/Ethnicity

View this table:
Table 5.

Linear Regression of LDL-C Level on rs12740374 and Protein C Level, Stratified by Race/Ethnicity

To determine whether other lipid-related genes might demonstrate similar associations with protein C levels, we extracted genotype data for 185 SNPs associated with LDL-C, high-density lipoprotein cholesterol (HDL-C), or triglycerides in a large GWAS meta-analysis conducted by the Global Lipids Genetics Consortium.14 As shown in Figure II in the online-only Data Supplement, 37 genetic variants were associated with 2 lipid fractions in the Global Lipids Genetics Consortium, and 8 variants were associated with all 3 lipid fractions. In ARIC study, 12 of 82 LDL-related SNPs, 11 of 96 HDL-related SNPs, and 10 of 60 triglyceride-related SNPs demonstrated at least nominal significance (P<0.05) in association with protein C level in whites. Among LDL-related SNPs, the largest effect sizes were observed for rs1270374 along with HFE (rs1800562; −0.056 μg/mL per A allele; P=0.003) and NPC1L1 (rs2072183; +0.033 μg/mL per G allele; P=0.005). Because individuals homozygous for the minor allele of HFE rs1800562 (C282Y homozygotes) account for the majority of cases of hereditary hemochromatosis in populations of European ancestry, we further characterized associations by genotype: mean protein C was 2.98, 3.13, and 3.17 μg/mL and mean LDL-C was 3.18, 3.52, and 3.55 mmol/L, respectively, among individuals with AA, GA, and GG genotypes for rs1800562, indicating that both protein C and LDL-C levels were lower in C282Y homozygotes compared with the other 2 genotypes.

We summarize Mendelian randomization (MR) analyses to estimate possible effects of lipid fractions on protein C levels in Table 6. Weighted genetic risk scores explained 8% of the variance in LDL-C (n=82 SNPs), 5% of the variance in HDL-C (n=96 SNPs), and 5% of the variance in triglycerides (n=60 SNPs). Weighted genetic risk scores for LDL-C, HDL-C, and triglycerides were each positively and statistically significantly associated with protein C level. The magnitude of association was largest for triglycerides, with an estimated increase in protein C of 0.15 μg/mL (95% confidence interval, 0.10–0.21) for every 1-SD increment in triglyceride level (0.73 mmol/L). Inverse-variance weighted meta-analyses of the genetic instrumental variables indicated associations between LDL-C and protein C (P<0.001) and between triglycerides and protein C (P<0.001) but not between HDL-C and protein C (P=0.12). Sensitivity analyses were conducted using MR-Egger and MR-weighted median approaches. For LDL-C, the intercept term from MR-Egger regression was statistically significant (P=0.03), suggesting that some LDL-related genetic instruments affect protein C level through biological pathways not involving LDL-C (ie, horizontal pleiotropy). The regression coefficient for LDL-C obtained from Egger regression was attenuated compared with the IVW (inverse variance weighted) estimate and not statistically significant (β=0.07; 95% confidence interval, −0.02 to 0.16; P=0.15). For triglycerides, the intercept term estimated by MR-Egger regression was not statistically significant (P=0.29), suggesting little evidence of directional pleiotropy. Results of Egger and weighted median approaches generally corroborated the results of IVW analyses and suggest that higher triglyceride levels increase circulating protein C.

View this table:
Table 6.

Associations of Lipid Fractions with Protein C Levels Estimated From MR in Whites

We summarize MR analyses to estimate possible effects of protein C on lipid fractions in Table 7. Taken together, the 4 SNPs selected as genetic instruments explained 14% of the variance in protein C level. None of the MR results was statistically significant (P>0.05), and there was little support for the hypothesis that circulating protein C influences LDL-C, HDL-C, or triglyceride levels.

View this table:
Table 7.

Associations of Protein C Levels With Lipid Fractions Estimated From MR in Whites

To test whether the variant rs12740374 at the CELSR2–PSRC1–SORT1 locus is also associated with risk of VTE, we conducted an analysis of 9185 whites in ARIC study, among whom there were 248 events of VTE identified over an average follow-up of 16 years.15 The T allele of rs12740374, which was associated with lower mean protein C level, was not significantly associated with incident VTE (hazard ratio: 1.19; 95% confidence interval: 0.97–1.46).

Discussion

To obtain denser coverage at some loci, extend investigation to low-frequency and rare variants, and search for second independent associations in genes previously reported to influence protein C, we evaluated associations with SNPs genotyped with 3 different candidate gene or genome-wide arrays. Associations with common variants were confirmed in or near GCKR, PROC, BAZ1B, and PROCR–EDEM2 in whites and PROC and PROCR–EDEM2 in blacks, but no new independent associations with common, low-frequency, or rare variants were detected in these regions. The primary novel finding was the discovery of an association of protein C with common variants in the CELSR2–PSRC1–SORT1 region on chromosome 1, with the strongest signal at rs12740374, an important functional variant known to influence LDL-C. The magnitude of association with protein C was consistent in whites and blacks and reached genome-wide significance in a meta-analysis combining both groups. The result suggests that variability at this locus influences both LDL-C and protein C levels; furthermore, these apparent effects on protein C may extend to other loci involved in lipid metabolism. MR analyses indicated that triglycerides, and possibly LDL-C, influence protein C levels.

Protein C is a multifunctional molecule with anticoagulant, anti-inflammatory, antiapoptotic, and cytoprotective properties. Protein C deficiency is an established predictor of VTE.1,2 At least 2 susceptibility genes for VTE (F5 and PROCR) are also significantly associated with protein C level,10 highlighting the importance of protein C as an intermediate phenotype that can provide insights into the genetic architecture of VTE.

Previous studies have reported that a common haplotype including rs12740374 in the CELSR2–PSRC1–SORT1 region of chromosome 1p13.3 is associated with multiple cardiovascular phenotypes, including LDL-C,16 coronary heart disease events,17,18 subclinical atherosclerosis,19 and abdominal aortic aneurysm.20 Our results suggest that the phenotypic effects of this locus also extend to protein C levels. In vitro and mouse experiments found that the SNP alters a binding site for C/EBP (CCAAT-enhancer–binding protein) transcription factors and provided considerable evidence that rs12740374 is a causal variant associated with SORT1 expression in hepatocytes, LDL-C levels, and CHD events.21

Our results indicate that the minor allele at rs12740374, which falls in the 3′UTR of CELSR2, is associated with lower levels of both protein C and LDL-C. Cross-sectional epidemiological studies had previously reported modest positive correlations of LDL-C with protein C, as well as with other anticoagulant proteins such as antithrombin, protein S, and tissue factor pathway inhibitor, but the mechanism responsible for these associations is unclear.22,23 Both randomized trials24 and observational studies25 have found that individuals on statin therapy are at lower risk of VTE, but it is unclear whether this apparent benefit of statins can be attributed to their LDL-lowering effects or to their effects on inflammation, coagulation, and fibrinolysis pathways. The association of variants in the CELSR2–PSRC1–SORT1 region with both protein C and LDL-C in our study provides a further evidence of a biological link between these 2 phenotypes, supported by several observations: (1) associations of rs12740374 with protein C and LDL-C were remarkably similar in magnitude between whites and blacks; (2) rs12740374 has already been established as a likely causal variant for LDL-C based on extensive functional experiments, likely mediated by expression of sortilin, the gene product of SORT1 that serves as a multiligand receptor with diverse cellular functions21; (3) even excluding rs12740374, 21% of LDL-related gene variants identified in published GWAS also showed nominal (P<0.05) associations with protein C level; and (4) the direction of the estimated protein C effect was positively correlated with the direction of estimated LDL-C effect across established 82 LDL-related loci. Furthermore, our results are more consistent with LDL-C being an upstream determinant of protein C level than protein C being a determinant of LDL-C. First, adjustment for LDL-C substantially attenuated the association between rs12740374 and protein C, suggesting that LDL-C could be a mediator of the association; by contrast, adjustment for protein C resulted in little attenuation of the association between rs12740374 and LDL-C. Second, although SNPs in several known LDL-influencing genes such as HFE and NPC1L1 were also nominally associated with protein C in our data, SNPs in known protein C–influencing genes (ie, GCKR, PROC, BAZ1B, and PROCR-EDEM2) were not associated with LDL-C.

Motivated by the discovery of variants at the CELSR2–PSRC1–SORT1 locus associated with protein C, we explored other lipid-related genes, including those influencing HDL-C or triglyceride levels. A previous investigation in the ARIC study found LDL-C, HDL-C, and triglycerides were all positively and independently associated with protein C in a multivariate model that adjusted for demographic, lifestyle, and biochemical factors.23 Of the 3 lipid fractions, triglycerides demonstrated the largest effect size for association with protein C. Furthermore, LDL-C and triglycerides, but not HDL-C, were significantly associated with protein C deficiency, defined as ≥2 SDs below the mean level of protein C.23 Other population-based studies have reported similar patterns of association between triglycerides and protein C.26,27 Our MR analyses using lipid-related genetic instruments provided some support for a causal effect of triglycerides and LDL-C on protein C but not of HDL-C on protein C. By contrast, bidirectional MR analyses did not support a causal effect of protein C on any of the lipid fractions. It is possible that triglycerides and LDL-C modulate the synthesis or degradation of protein C.23 Further research is needed to elucidate possible mechanisms linking lipid metabolism and protein C.

Strengths of our study include a large population-based sample, standardized data collection and laboratory measurements at a central laboratory, and extensive genetic data obtained from 3 different genotyping platforms. Limitations include modest reproducibility for the protein C assay in participants measured several times over several weeks, measurements of protein C and LDL-C in samples collected at the same exam, making it difficult to establish temporal relationships between the 2 phenotypes, and lack of information on protein C activity, which may have different determinants than protein C antigen level. Findings were not replicated externally as almost no other population-based studies have measured protein C, yet results could be internally corroborated across the 2 racial/ethnic subgroups represented in the study.

In conclusion, comprehensive genomic analysis of protein C in the ARIC study led to discovery of a novel association with variants in the CELSR2–PSRC1–SORT1 region on chromosome 1, with the strongest signal at rs12740374, an important functional SNP known to influence LDL-C. MR analyses indicated that triglycerides, and possibly LDL-C, influence protein C levels. These findings need to be replicated in other populations. Some clinical trials and observational studies of statins have found a reduction in VTE risk, suggesting that lipid lowering may have effects on coagulation or anticoagulation, including enhanced thrombomodulin expression resulting in increased protein C activation and factor Va inactivation.28 Studies of protein C in individuals with more severe lipid disorders such as familial hypercholesterolemia or among patients initiating lipid-lowering therapies may provide additional insights into possible biological links between protein C and cholesterol metabolism.

Acknowledgments

We thank the staff and participants of the ARIC study for their important contributions. This work was performed in part using computing resources at the University of Minnesota Supercomputing Institute.

Sources of Funding

The ARIC study (Atherosclerosis Risk in Communities) is performed as a collaborative study supported by the National Heart, Lung, and Blood Institute contracts (HHSN268201100005C, HHSN268201100006C, HHSN268201100007C, HHSN268201100008C, HHSN268201100009C, HHSN268201100010C, HHSN268201100011C, and HHSN268201100012C), R01HL087641, R01HL59367, and R01HL086694; National Human Genome Research Institute contract U01HG004402; and National Institutes of Health contract HHSN268200625226C. Infrastructure was partly supported by grant number UL1RR025005, a component of the National Institutes of Health and NIH Roadmap for Medical Research. Funding support for Building on GWAS for NHLBI-diseases: the US CHARGE consortium was provided by the NIH through the American Recovery and Reinvestment Act of 2009 (5RC2HL102419). CARe is supported by National Institutes of Health/National Heart, Lung, and Blood Institute (N01HC65226) through the Broad Institute of Harvard University and the Massachusetts Institute of Technology. A full listing of the grants and contracts that have supported individual cohorts of the CARe is provided at http://public.nhlbi.nih.gov/GeneticsGenomics/home/care.aspx. Part of this study was supported by National Heart, Lung, and Blood Institute grants (R01HL095603 and R01HL59367).

Disclosures

None.

Footnotes

  • The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.116.308109/-/DC1.

  • Nonstandard Abbreviations and Acronyms
    ARIC
    Atherosclerosis Risk in Communities
    GWAS
    genome-wide association studies
    HDL-C
    high-density lipoprotein cholesterol
    IBC
    ITMAT-Broad-CARe
    MR
    Mendelian randomization
    SNP
    single-nucleotide polymorphism
    VTE
    venous thromboembolism
    1000G
    1000 Genomes Project

  • Received July 1, 2016.
  • Accepted December 30, 2016.

References

  1. 1.↵
    1. Koster T,
    2. Rosendaal FR,
    3. Briët E,
    4. van der Meer FJ,
    5. Colly LP,
    6. Trienekens PH,
    7. Poort SR,
    8. Reitsma PH,
    9. Vandenbroucke JP
    . Protein C deficiency in a controlled series of unselected outpatients: an infrequent but clear risk factor for venous thrombosis (Leiden Thrombophilia Study). Blood. 1995;85:2756–2761.
    OpenUrlAbstract/FREE Full Text
  2. 2.↵
    1. Folsom AR,
    2. Aleksic N,
    3. Wang L,
    4. Cushman M,
    5. Wu KK,
    6. White RH
    . Protein C, antithrombin, and venous thromboembolism incidence: a prospective population-based study. Arterioscler Thromb Vasc Biol. 2002;22:1018–1022.
    OpenUrlAbstract/FREE Full Text
  3. 3.↵
    1. Folsom AR,
    2. Ohira T,
    3. Yamagishi K,
    4. Cushman M
    . Low protein C and incidence of ischemic stroke and coronary heart disease: the Atherosclerosis Risk in Communities (ARIC) Study. J Thromb Haemost. 2009;7:1774–1778. doi: 10.1111/j.1538-7836.2009.03577.x.
    OpenUrlCrossRefPubMed
  4. 4.↵
    1. Knuiman MW,
    2. Folsom AR,
    3. Chambless LE,
    4. Liao D,
    5. Wu KK
    . Association of hemostatic variables with MRI-detected cerebral abnormalities: the atherosclerosis risk in communities study. Neuroepidemiology. 2001;20:96–104. doi: 54767.
    OpenUrlCrossRefPubMed
  5. 5.↵
    1. Oudot-Mellakh T,
    2. Cohen W,
    3. Germain M,
    4. Saut N,
    5. Kallel C,
    6. Zelenika D,
    7. Lathrop M,
    8. Trégouët DA,
    9. Morange PE
    . Genome wide association study for plasma levels of natural anticoagulant inhibitors and protein C anticoagulant pathway: the MARTHA project. Br J Haematol. 2012;157:230–239. doi: 10.1111/j.1365-2141.2011.09025.x.
    OpenUrlCrossRefPubMed
  6. 6.↵
    1. Couture P,
    2. Demers C,
    3. Morissette J,
    4. Delage R,
    5. Jomphe M,
    6. Couture L,
    7. Simard J
    . Type I protein C deficiency in French Canadians: evidence of a founder effect and association of specific protein C gene mutations with plasma protein C levels. Thromb Haemost. 1998;80:551–556.
    OpenUrlPubMed
  7. 7.↵
    1. Tang W,
    2. Basu S,
    3. Kong X,
    4. Pankow JS,
    5. Aleksic N,
    6. Tan A,
    7. Cushman M,
    8. Boerwinkle E,
    9. Folsom AR
    . Genome-wide association study identifies novel loci for plasma levels of protein C: the ARIC study. Blood. 2010;116:5032–5036. doi: 10.1182/blood-2010-05-283739.
    OpenUrlAbstract/FREE Full Text
  8. 8.↵
    1. Munir MS,
    2. Weng LC,
    3. Tang W,
    4. Basu S,
    5. Pankow JS,
    6. Matijevic N,
    7. Cushman M,
    8. Boerwinkle E,
    9. Folsom AR
    . Genetic markers associated with plasma protein C level in African Americans: the atherosclerosis risk in communities (ARIC) study. Genet Epidemiol. 2014;38:709–713. doi: 10.1002/gepi.21868.
    OpenUrl
  9. 9.↵
    1. Athanasiadis G,
    2. Buil A,
    3. Souto JC,
    4. Borrell M,
    5. López S,
    6. Martinez-Perez A,
    7. Lathrop M,
    8. Fontcuberta J,
    9. Almasy L,
    10. Soria JM
    . A genome-wide association study of the protein C anticoagulant pathway. PLoS One. 2011;6:e29168. doi: 10.1371/journal.pone.0029168.
    OpenUrlPubMed
  10. 10.↵
    1. Germain M,
    2. Chasman DI,
    3. de Haan H,
    4. et al
    ; Cardiogenics Consortium. Meta-analysis of 65,734 individuals identifies TSPAN15 and SLC44A2 as two susceptibility loci for venous thromboembolism. Am J Hum Genet. 2015;96:532–542. doi: 10.1016/j.ajhg.2015.01.019.
    OpenUrlCrossRefPubMed
  11. 11.↵
    1. Wu C,
    2. Dwivedi DJ,
    3. Pepler L,
    4. Lysov Z,
    5. Waye J,
    6. Julian J,
    7. Desch K,
    8. Ginsburg D,
    9. Weitz JI,
    10. Kearon C,
    11. Liaw PC
    . Targeted gene sequencing identifies variants in the protein C and endothelial protein C receptor genes in patients with unprovoked venous thromboembolism. Arterioscler Thromb Vasc Biol. 2013;33:2674–2681. doi: 10.1161/ATVBAHA.113.302137.
    OpenUrlAbstract/FREE Full Text
  12. 12.↵
    1. Keating BJ,
    2. Tischfield S,
    3. Murray SS,
    4. et al
    . Concept, design and implementation of a cardiovascular gene-centric 50 k SNP array for large-scale genomic association studies. PLoS One. 2008;3:e3583. doi: 10.1371/journal.pone.0003583.
    OpenUrlCrossRefPubMed
  13. 13.↵
    1. Grove ML,
    2. Yu B,
    3. Cochran BJ,
    4. et al
    . Best practices and joint calling of the HumanExome BeadChip: the CHARGE Consortium. PLoS One. 2013;8:e68095. doi: 10.1371/journal.pone.0068095.
    OpenUrlCrossRefPubMed
  14. 14.↵
    1. Willer CJ,
    2. Schmidt EM,
    3. Sengupta S,
    4. et al
    . Discovery and refinement of loci associated with lipid levels. Nat Genet. 2013;45:1274–1283. doi: 10.1038/ng.2797.
    OpenUrlCrossRefPubMed
  15. 15.↵
    1. Tang W,
    2. Teichert M,
    3. Chasman DI,
    4. et al
    . A genome-wide association study for venous thromboembolism: the extended cohorts for heart and aging research in genomic epidemiology (CHARGE) consortium. Genet Epidemiol. 2013;37:512–521. doi: 10.1002/gepi.21731.
    OpenUrlCrossRefPubMed
  16. 16.↵
    1. Teslovich TM,
    2. Musunuru K,
    3. Smith AV,
    4. et al
    . Biological, clinical and population relevance of 95 loci for blood lipids. Nature. 2010;466:707–713. doi: 10.1038/nature09270.
    OpenUrlCrossRefPubMed
  17. 17.↵
    1. Samani NJ,
    2. Erdmann J,
    3. Hall AS,
    4. et al
    ; WTCCC and the Cardiogenics Consortium. Genomewide association analysis of coronary artery disease. N Engl J Med. 2007;357:443–453. doi: 10.1056/NEJMoa072366.
    OpenUrlCrossRefPubMed
  18. 18.↵
    1. Wang AZ,
    2. Li L,
    3. Zhang B,
    4. Shen GQ,
    5. Wang QK
    . Association of SNP rs17465637 on chromosome 1q41 and rs599839 on 1p13.3 with myocardial infarction in an American caucasian population. Ann Hum Genet. 2011;75:475–482. doi: 10.1111/j.1469-1809.2011.00646.x.
    OpenUrlCrossRefPubMed
  19. 19.↵
    1. O’Donnell CJ,
    2. Kavousi M,
    3. Smith AV,
    4. et al
    ; CARDIoGRAM Consortium. Genome-wide association study for coronary artery calcification with follow-up in myocardial infarction. Circulation. 2011;124:2855–2864. doi: 10.1161/CIRCULATIONAHA.110.974899.
    OpenUrlAbstract/FREE Full Text
  20. 20.↵
    1. Jones GT,
    2. Bown MJ,
    3. Gretarsdottir S,
    4. et al
    . A sequence variant associated with sortilin-1 (SORT1) on 1p13.3 is independently associated with abdominal aortic aneurysm. Hum Mol Genet. 2013;22:2941–2947. doi: 10.1093/hmg/ddt141.
    OpenUrlAbstract/FREE Full Text
  21. 21.↵
    1. Musunuru K,
    2. Strong A,
    3. Frank-Kamenetsky M,
    4. et al
    . From noncoding variant to phenotype via SORT1 at the 1p13 cholesterol locus. Nature. 2010;466:714–719. doi: 10.1038/nature09266.
    OpenUrlCrossRefPubMed
  22. 22.↵
    1. Sakkinen PA,
    2. Cushman M,
    3. Psaty BM,
    4. Kuller LH,
    5. Bajaj SP,
    6. Sabharwal AK,
    7. Boineau R,
    8. Macy E,
    9. Tracy RP
    . Correlates of antithrombin, protein C, protein S, and TFPI in a healthy elderly cohort. Thromb Haemost. 1998;80:134–139.
    OpenUrlPubMed
  23. 23.↵
    1. Conlan MG,
    2. Folsom AR,
    3. Finch A,
    4. Davis CE,
    5. Sorlie P,
    6. Wu KK
    . Correlation of plasma protein C levels with cardiovascular risk factors in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study. Thromb Haemost. 1993;70:762–767.
    OpenUrlPubMed
  24. 24.↵
    1. Glynn RJ,
    2. Danielson E,
    3. Fonseca FA,
    4. Genest J,
    5. Gotto AM Jr.,
    6. Kastelein JJ,
    7. Koenig W,
    8. Libby P,
    9. Lorenzatti AJ,
    10. MacFadyen JG,
    11. Nordestgaard BG,
    12. Shepherd J,
    13. Willerson JT,
    14. Ridker PM
    . A randomized trial of rosuvastatin in the prevention of venous thromboembolism. N Engl J Med. 2009;360:1851–1861. doi: 10.1056/NEJMoa0900241.
    OpenUrlCrossRefPubMed
  25. 25.↵
    1. Smith NL,
    2. Harrington LB,
    3. Blondon M,
    4. Wiggins KL,
    5. Floyd JS,
    6. Sitlani CM,
    7. McKnight B,
    8. Larson EB,
    9. Rosendaal FR,
    10. Heckbert SR,
    11. Psaty BM
    . The association of statin therapy with the risk of recurrent venous thrombosis. J Thromb Haemost. 2016;14:1384–1392. doi: 10.1111/jth.13334.
    OpenUrl
  26. 26.↵
    1. Woodward M,
    2. Lowe GD,
    3. Rumley A,
    4. Tunstall-Pedoe H,
    5. Philippou H,
    6. Lane DA,
    7. Morrison CE
    . Epidemiology of coagulation factors, inhibitors and activation markers: The Third Glasgow MONICA Survey. II. Relationships to cardiovascular risk factors and prevalent cardiovascular disease. Br J Haematol. 1997;97:785–797.
    OpenUrlCrossRefPubMed
  27. 27.↵
    1. Rodeghiero F,
    2. Tosetto A
    . The VITA Project: population-based distributions of protein C, antithrombin III, heparin-cofactor II and plasminogen–relationship with physiological variables and establishment of reference ranges. Thromb Haemost. 1996;76:226–233.
    OpenUrlPubMed
  28. 28.↵
    1. Undas A,
    2. Brummel-Ziedins KE,
    3. Mann KG
    . Anticoagulant effects of statins and their clinical implications. Thromb Haemost. 2014;111:392–400. doi: 10.1160/TH13-08-0720.
    OpenUrlCrossRefPubMed

Highlights

  • Comprehensive genetic analyses confirmed associations with single-nucleotide polymorphisms in 4 regions and protein C level but did not identify new independent associations with common, low-frequency, or rare variants in these regions.

  • In a novel finding, variants in the CELSR2/PSRC1/SORT1 region were associated with protein C level, with the strongest signal at rs12740374, an important functional single-nucleotide polymorphism known to influence low-density lipoprotein cholesterol levels.

  • These findings may indicate a novel genetic link between lipoprotein metabolism and hemostasis.

View Abstract

This Issue

Jump to

Article Tools

Share this Article

  • Email
  • Identification of Genetic Variants Linking Protein C and Lipoprotein MetabolismHighlights
    James S. Pankow, Weihong Tang, Nathan Pankratz, Weihua Guan, Lu-Chen Weng, Mary Cushman, Eric Boerwinkle and Aaron R. Folsom
    Arteriosclerosis, Thrombosis, and Vascular Biology. 2017;37:589-597, originally published January 12, 2017
    https://doi.org/10.1161/ATVBAHA.116.308109
    Permalink:
    del.icio.us logo Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo

Related Articles

Cited By...

Subjects