Single nucleotide polymorphisms in CETP, SLC46A1, SLC19A1, CD36, BCMO1, APOA5, and ABCA1 are significant predictors of plasma HDL in healthy adults
© Clifford et al.; licensee BioMed Central Ltd. 2013
Received: 13 February 2013
Accepted: 2 May 2013
Published: 8 May 2013
In a marker-trait association study we estimated the statistical significance of 65 single nucleotide polymorphisms (SNP) in 23 candidate genes on HDL levels of two independent Caucasian populations. Each population consisted of men and women and their HDL levels were adjusted for gender and body weight. We used a linear regression model. Selected genes corresponded to folate metabolism, vitamins B-12, A, and E, and cholesterol pathways or lipid metabolism.
Extracted DNA from both the Sacramento and Beltsville populations was analyzed using an allele discrimination assay with a MALDI-TOF mass spectrometry platform. The adjusted phenotype, y, was HDL levels adjusted for gender and body weight only statistical analyses were performed using the genotype association and regression modules from the SNP Variation Suite v7.
Statistically significant SNP (where P values were adjusted for false discovery rate) included: CETP (rs7499892 and rs5882); SLC46A1 (rs37514694; rs739439); SLC19A1 (rs3788199); CD36 (rs3211956); BCMO1 (rs6564851), APOA5 (rs662799), and ABCA1 (rs4149267). Many prior association trends of the SNP with HDL were replicated in our cross-validation study. Significantly, the association of SNP in folate transporters (SLC46A1 rs37514694 and rs739439; SLC19A1 rs3788199) with HDL was identified in our study.
Given recent literature on the role of niacin in the biogenesis of HDL, focus on status and metabolism of B-vitamins and metabolites of eccentric cleavage of β-carotene with lipid metabolism is exciting for future study.
KeywordsSingle nucleotide polymorphism HDL Folate transporter Cholesterol
Understanding the effects of genetic, environmental, and especially of lipid levels on health status, is of wide and significant interest . The relationships between persistent environmental pollutants and micronutrient levels are not well understood; though such speculation includes the role of peroxisome proliferator-activated receptor (PPAR), transcription factors related to lipid homeostasis, or changes in DNA methylation patterns . In our recent work , we investigated 65 single nucleotide polymorphisms (SNP) in 23 candidate genes involved in folate metabolism (8 genes), vitamins B-12, A, and E metabolism (5 genes), and cholesterol pathways or lipid metabolism (10 genes) in a homocysteine/red blood cell folate marker trait association study. Interestingly, a few SNP associated with diabetes mellitus (DM), cardiovascular disease (CVD) and maintenance of the cholesterol pathway or lipid metabolism were identified: serine palmitoyltransferase (SPTLC1 rs11790991), cholesteryl ester transfer protein (CETP rs5882) and scavenger receptor class B type 1 (SCARB1 rs838892). SNP associated with transfer of antioxidant vitamins, including rs2118981 in the cellular retinol binding protein II gene (CRBP2), which is important for vitamin A and retinoid transfer, and rs838892 SCARB1 (for tocopherols and tocotrienols), were also statistically significant predictors in the final model. Significantly, SNP in betaine-homocysteine methyltransferase (BHMT rs3733890) and methylene tetrahydrofolate reductase (MTHFR rs1801131), both of which are involved in one-carbon metabolism, were also included in the final model that was previously described .
Given the establishment of these associations in our previous work , further investigation of the relation of folate, homocysteine (Hcy), and changes in plasma lipid profiles was of significant research interest. High levels of Hcy and changes in plasma lipids are independent risk factors for development of CVD, and there may be a connection between Hcy metabolism and lipid metabolism . Low levels of dietary folate, an important cofactor in the metabolism of Hcy, led to increased levels of serum and liver cholesterol in wild-type mice . As well, cholesterol metabolism may contribute to beneficial effects of dietary folate supplementation . Possible connections, not yet fully supported by experimental data, between folate levels and cholesterol biosynthesis include endoplasmic reticulum stress, which activates cholesterol biosynthesis genes, or through metabolic processes involving choline. Choline provides methyl groups for Hcy metabolism and is a precursor for choline phospholipids, which are required for lipoprotein secretion . Additionally, serum lipid levels are risk factors for a number of adverse health outcomes, including coronary heart disease (CHD), atherosclerosis, type 2 DM, stroke, and metabolic syndrome . Lipid levels of interest include triglycerides, low-density lipoprotein (LDL), and high-density lipoprotein (HDL), especially as recent work has indicated that HDL may not always be protective against atherosclerosis [5, 6]. Particularly, the functionality and quality of HDL  in response to vascular inflammation and oxidative stress  is an interesting area of further study. HDL is required in the reverse cholesterol transport (RCT) mechanism where excess cholesterol is transferred from peripheral cells to the liver for intestinal excretion . Additionally, HDL has anti-inflammatory and antioxidant effects and improves endothelial function .
In this study we selected 65 SNP in 23 candidate genes to perform a marker trait association study with plasma HDL adjusted for gender and body weight in Caucasian male and female study participants from two independent populations (Sacramento, California, and Beltsville, Maryland). Selected SNP corresponded to 8 genes associated with folate metabolism, 5 genes associated with vitamin B-12, vitamin A and vitamin E metabolism, and 10 genes associated with cholesterol pathways or lipid metabolism. The overall objective of this study was to determine if any of these SNP in the 23 candidate genes had a statistically significant association with HDL.
Significant SNP predictors
Statistically significant SNP predictors in the Sacramento and Beltsville populations were similar between both groups though there were some differences in the directionality and magnitude of the association between SNP predictor and HDL. SNP of genes that were statistically significant included: 1) cholesteryl ester transfer protein (CETP; rs7499892 and rs5882); 2) proton-coupled folate transporter (SLC46A1; rs35714695 and rs739439); 3) reduced folate carrier (SLC19A1; rs3788199); 4) thrombospondin receptor (CD36; rs3211956); 5) beta-carotene monooxygenase 1 (BCMO1; rs6564851); 6) apolipoproteinA-V (APOA5; rs662799); and 7) ATP-binding cassette transporter member 1 (ABCA1; rs4149267).
The two SNP included in this study for CETP (rs7499892 and rs5882) with MAF of ~0.16 and MAF of ~0.04 for the rare allele, respectively, were statistically significantly negatively associated with HDL-C in both the Sacramento and Beltsville populations (Tables 1 and 2) as indicated by the ASE (slope of the regression). It was determined that the rs3788199 SNP in SLC19A1 was positively correlated with HDL levels (ASE = 2.828 for the Sacramento population, Table 1; ASE = 1.796 for the Beltsville population, Table 2). The SNP rs35714695 and rs739439 of SLC46A1 were both negatively associated with HDL levels in the Sacramento population though the Beltsville population had a positive association with HDL levels, which indicates that the allele effects are opposite. The directionality of this association difference is one area of future study. In the present study, a very large positive association between HDL levels and the presence of rs3211956 of CD36 was also identified for both study populations (Table 1 and Table 2). In the Sacramento population, the ASE was 4.372, with similar positive association (but not as large) found in the Beltsville population (ASE = 1.774). The results of the present study support a past finding: in both populations (Sacramento and Beltsville), a positive association, as indicated by the positive ASE values (Table 1, ASE = 0.586 and Table 2, ASE = 1.015), was established for the rs6564851 SNP of BCMO1 and plasma HDL levels. For APOA5, the rs662799 SNP was statistically significantly negatively associated with HDL. The ASE was −4.523 for the Sacramento population (Table 1) and −0.471 for the Beltsville population (Table 2), indicating that there was a negative association of the presence of this SNP with measured HDL concentrations. Finally, the SNP rs4149267 of ABCA1 was associated with HDL-C in both Caucasian populations with similar ASE of −3.236 in the Sacramento population (Table 1) and −2.070 in the Beltsville population (Table 2).
CETP encodes cholesteryl ester transfer protein, which exchanges the triglycerides from VLDL and LDL particles for cholesterol esters from HDL. CETP also selectively enhances liver HDL cholesterol ester uptake . Inhibiting CETP with various pharmacologic agents (torcetrapib, dalcetrapib, anacetrapib, evacetrapib) has been an attractive means to minimize risk for adverse cardiovascular events because of a potential rise in HDL-C and fall in LDL-C [8, 10]. Understanding the molecular basis for cholesteryl ester transport by CETP is of research interest to support the development of other CETP inhibitors . In a previous genome wide association study (GWAS), the SNP rs708272 (or Taq IB SNP) of CETP had the strongest association with HDL-C of all 948 SNPs genotyped in 122 genes though this SNP was not included in our study. As well, this SNP had a strong association with apolipoproteinA-1 (apoA1) . In a population of Chinese Bama Zhuang individuals, the rs708272 SNP was associated with higher HDL-C levels, but the outcome of this SNP on longevity was less clear . A recent meta-analysis of 46 lipid GWAS, six HDL-C loci were identified with at least a second independent association with HDL-C including LPL, ABCA1, APOA1/A4/A5/C3, ZNF664, LIPC, and CETP. Both ABCA1 and CETP were validated in our study. Tietjen and colleagues showed that rare coding and splicing mutations on CETP were enriched in persons with hyperalphalipoproteinemia and segregated with elevated HDL-C in families .
In our prior study, rs5882 of CETP was statistically significantly associated with Hcy normalized by red blood cell folate concentrations . Two of the four SNP included in this study for CETP (rs7499892 and rs5882) were statistically significantly negatively associated with HDL-C in both the Sacramento and Beltsville populations (Tables 1 and 2). The rs5882 SNP has been associated with lower CETP serum concentrations and activity, increased HDL cholesterol levels, and increased lipoprotein sizes, all factors which have been associated with a lower CVD risk . In a recent study in Tunisian population, there was no statistically significant association of the rs5882 SNP with lipoprotein metabolism or atherogenicity .
SLC46A1 and SLC19A1
The proton-coupled folate transporter (PCFT, gene symbol SLC46A1) mediates intestinal folate absorption and folate transport across the choroid plexus. PCFT has an optimal pH transport of 5.0 – 5.5 (jejunum and duodenum), but the role of this transporter in other tissues at normal physiological pH is less clear . Homozygous mutations in SLC46A1 have been associated with a rare disorder, hereditary folate malabsorption . Solute carrier family 19 (folate transporter) member 1, also known as the reduced folate carrier (RFC; gene symbol SLC19A1), is involved in the regulation of intracellular concentrations of folate. Higher serum folate concentrations have been associated with lower levels of LDL-C, lower LDL-C/HDL-C ratio, and higher HDL-C. These associations were independent of gender or age, though influences of medications, diseases, physical exercise, diet, or BMI were not accounted for in that study. Interestingly, vitamin B12 was not associated with the lipoprotein profile in that reported study . Cholesterol may be important for facilitating the import of folate by clustering membrane-bound folate receptors in the cell membrane . Use of clustering membrane-bound folate receptors was favoured when folate status was low . Folate status is inversely associated with obesity , likely due to increased activity of COMT (catechol O- methyltransferase), which uses folate for methyl transfer for metabolism of catechol estrogen produced by adipose tissue. Both obesity  and low folate status  have been associated with reduced HDL cholesterol levels. However, in a recent study examining folate placental transport in obese women, it was determined that while protein expression of folate receptor-α and RFC were altered (PCFT was unchanged), the activities of the transporters was unaltered in obesity and fetal folate serum concentration were not adversely affected . In the present study, it was determined that the rs3788199 SNP in SLC19A1 was positively correlated with HDL levels (ASE = 2.828 for the Sacramento population, Table 1; ASE = 1.796 for the Beltsville population, Table 2), which compares well to the earlier retrospective database study previously described . Surprisingly, it was also determined that the SNP rs35714695 and rs739439 of SLC46A1 were both negatively associated with HDL levels in the Sacramento population whereas the Beltsville population had a positive association with HDL levels, which indicates that the allele effects are opposite. To our knowledge, there are no studies providing information on association studies of SNP in SLC46A1 with HDL cholesterol levels. Given the differences in association between the two populations in this study, investigation of SNP in SLC46A1 with HDL is of further research interest.
A 2010 study examined seven SNP variants in genes involved in Hcy metabolism on the interaction with plasma lipid profile. In that study, SLC19A1 (SNP rs1051266) was not found to have a statistically significant association with blood lipid profiles analysed (fasting lipoprotein, total cholesterol, LDL, HDL, and triglyceride levels) . Additional findings from that study implicated SNP in the genes for transcobalamin II (rs1801198) and MTHFR (rs1801133) were associated with blood lipid profiles. The G allele of rs1801198 was correlated with higher levels of LDL in plasma, lower HDL, higher triglyceride levels, and higher total cholesterol levels . However, in our study, neither rs1801198 (transcobalamin II) nor rs1801133 (MTHFR) were statistically significantly associated with levels of HDL in plasma.
CD36 (thrombospondin receptor), or glycoprotein IIIb/platelet glycoprotein IV is a mediator of platelet adhesion to collagen. The investigation of SNP of CD36 on HDL is not well established, likely because participants in many studies deal with chronic diseases, including DM, CHD, or metabolic syndrome or risk factors for chronic diseases [23–26]. Therefore, CD36 warrants further investigation, with careful statistical control of potentially confounding variables, including environmental factors, dietary factors, and other genotypes. In our study, we enrolled healthy participants as indicated by low use of statins (less than ~7% for overall study population) and healthy BMI (mean ± standard deviation for Sacramento was 24.8 ± 4.2 and 26.6 ± 4.6 for Beltsville). We found that there was a very large positive association between HDL levels and the presence of SNP rs3211956 (Tables 1 and 2). In the Sacramento population the ASE was 4.72, with similar positive association (but not as large) found in the Beltsville population (ASE=1.774).
Beta-carotene monooxygenase 1 (BCMO1) catalyzes the first step in the central cleavage and conversion of dietary provitamin carotenoids to vitamin A (retinal) in the small intestine [27, 28]. Vitamin A is necessary for immune response, vision, embryonic development, cell differentiation, and membrane and skin protection . The statistically significant SNP identified in the present study, rs6564851, is 7.7 kb 5’ upstream from the BCMO1 gene. This particular SNP has been associated with a 48% reduced catalytic activity of converting β-carotene into vitamin A in female participants in a recent study . Other SNP in BCMO1 have been associated with plasma levels of various carotenoids, including β-carotene, lutein, α-carotene, zeaxanthin, and lycopene ; and the G allele of the rs6564851 may explain some of the variance in plasma levels of these provitamin carotenoids. The rs6564851 SNP may be particularly important for individuals at risk for vitamin A deficiency  owing to reduced catalytic activity of BCMO1 . Recently, the rs6564851 SNP had the strongest association (when compared to other SNP: rs11645428, rs6420424, and rs8044334) with fasting β-carotene concentrations in plasma . Higher levels of carotenoids (isomers of β-carotene, lycopene, and β-cryptoxanthin) have been associated with higher levels of HDL and LDL in a recent study involving NHANES participants . In a recent review, the important physiological effects of eccentric cleavage products of beta-carotene were discussed. Considering the effects of BCMO1 SNP as well, there could be some very different physiological effects from beta-carotene consumption owing to genetic influences, oxidative stress, and presence of various beta-carotene metabolites . Interesting recent work has focused on the retinoid receptor antagonist activity of products resulting from β-carotene eccentric cleavage, the β-apocarotenoids [30, 31]. The biological activities and effects of these β-carotene metabolites may be important to consider in evaluating oxidative stress and adverse health outcomes of CVD or cancer . It has been established that there is no association between the BCMO1 SNP rs6564851 and risk of developing type 2 DM .
ApolipoproteinA-V is a protein component of HDL. In this study, the rs662799 SNP of APOA5 was identified as being a significant predictor. The ASE was −4.523 for the Sacramento population and −0.471 for the Beltsville population, indicating that there was a negative association of the presence of this SNP with measured HDL concentrations. In a recent study , the rs662799 SNP was the only SNP (of thirteen evaluated in the APOA1/C3/A4/A5 gene cluster) to be associated with three lipid traits: triglycerides, HDL-C, and LDL-C levels. In this past study, the MAF was statistically significantly associated with familial combined hyperlipidaemia, though the functional effect of this rs662799 SNP may not be well understood . In another recent study, rs662799 was statistically significantly associated with plasma triglycerides in both women and men of the study population and statistically significantly associated with total cholesterol and LDL-C levels in men only. However, the authors concluded that haplotypes for five SNP in the apolipoprotein A1/C3/A5 cluster could explain more serum lipid variation than any one SNP alone, especially for HDL-C . The presence of the rs662799 SNP was statistically significantly associated with lower levels of total cholesterol, triglycerides, and LDL-C in a group of Hei Yi Zhuang Chinese (though this same trend was not observed in Han Chinese), indicating that there may be other gene-gene or gene-environment interactions .
ABCA1 (ATP-binding cassette transporter member 1) plays an important role in cellular cholesterol and phospholipid homeostasis in multiple cell types [36, 37] and is involved in RCT . ABCA1 mediated efflux of cholesterol and phospholipids leads to the formation of nascent HDL via apoA1 [39, 40]; and mutations that disrupt normal ABCA1 function result in little or no circulating HDL . ABC transporter G1 (ABCG1) promotes cholesterol efflux from macrophages to HDL to form mature HDL particles , and hence works in a sequential manner with ABCA1 . All trans-retinoic acid has been shown to increase apoA1/HDL-mediated cholesterol efflux from macrophages by increasing ABCA1 and ABCG1 by regulating promoter activity via liver X receptor-responsive element mechanism . Wiersma and colleagues  also showed that ABCG1 knock-out mice exhibit decreased HDL-C when consuming a high fat diet. In this study, they also demonstrated that ABCG1 mediated cholesterol efflux to HDL . Functional mutations in ABCA1 cause Tangier disease, which is characterized by very low levels of plasma HDL apoA1 [44, 45]. In a recent study investigating exome sequencing, functional rare variants in ABCA1 and LPL (gene for lipoprotein lipase) were identified and explained a major portion of the HDL-C variance in the population enrolled in the study .
Previous studies have found associations between certain SNP in ABCA1 and HDL concentrations [46–48]. Recent GWAS and meta-analysis studies showed that SNP in ABCA1 were significantly associated with HDL-C [14, 49]. In our study, the SNP rs4149267 of ABCA1 was associated with HDL-C in both Caucasian populations with similar ASE of −3.236 in the Sacramento population and −2.070 in the Beltsville population.
It would be significant to understand the effects of apolipoprotein E (apoE), which plays an important role in lipoprotein metabolism and atherosclerosis. ApoE has been shown to promote selective uptake of HDL-C owing to increased ABCA1-mediated cholesterol efflux to plasma . This present work identifies significant SNP in ABCA1 and hence this relation with ABCA1 SNP and apoE facilitated HDL-C transport is of future research interest.
The strength of our study is the cross-validation between two independent populations (one on the West coast, from Sacramento, CA, and one from the East coast, from the Washington, D.C. area) with similar SNP associations identified in both populations. It would be of significant research interest to focus on the relation of B vitamins on HDL status. In this work, we have identified SNP in two folate transporters (SLC46A1 rs35714695 and rs739439; SLC19A1 rs3788199) having statistically significant ASE in relation to HDL status in both study populations. Cholesterol may be important for facilitating the import of folate across the cell membrane and higher serum folate concentrations have been associated with lower levels of LDL-C and higher levels of HDL-C . Past work by Kitami et al. focused on the importance of the homeostatic role of cholesterol metabolism on folate retention in mouse strains, so there has been an established relationship between cholesterol and folate in the mouse . Recent work by Zhang et al. identified the role of niacin (also known as vitamin B3) on early hepatic HDL formation through transcription of ABCA1. In that study, apoA1 lipidation and formation of nascent HDL was mediated and stabilized by niacin, which may prevent premature HDL catabolism .
Finally, the identification of the positive association of the BCMO1 SNP rs6564851 with HDL levels was of significance. This SNP has a high MAF in the two independent study populations of this work (Tables 1 and 2). Additionally, the presence of this SNP has been associated with a 48% reduction in activity of converting β-carotene into vitamin A through central cleavage, resulting in higher circulating levels of plasma carotenoids . These higher levels of carotenoids may be associated with higher levels of HDL and LDL . The biological effects of the eccentric cleavage products of β-carotene (the apo-β-carotenoids), especially on lipid metabolism and oxidative stress, are an exciting area of future study.
The Institutional Review Board of the University of California, Davis, approved the study, which was conducted according to Good Clinical Practice guidelines and the Declaration of Helsinki, version 1989. Written informed consent was obtained from each participant before enrollment in the study. Women and men ranging in age from 18 to 67 years were recruited by posted, published, and mailed advertisements in the California counties of Sacramento, San Joaquin, Solano, and Yolo from May 2004 through August 2005. Persons were excluded for any history of a serious medical condition, for using medications that could interfere with folate metabolism (including phenytoin, primidone, fosphenytoin, carbamazepine, phenobarbital, ranitidine, trimethoprim, metformin, sulfasalazine, triamterene or methotrexate, beta blockers, and angiotensin inhibitors), for using tobacco, or for heavy consumption of alcohol (≥ 3 drinks per day). Responders to advertisements enrolled in a single clinic visit at the Ragle Human Nutrition Research Center at the University of California, Davis. Before the clinic visit, participants in the study received via US mail a packet containing information about the study, consent forms, instructions to fast for 8 – 10 h prior to the clinic visit, and two dietary intake instruments to assess folate intake from the diet and supplements. Folate intake assessed by the Block Dietary Folate Equivalents (DFE) Screener is available through Supporting Information (Additional file 1: Figure S1). At the time of the scheduled visit, participants were interviewed about general medical, personal, and family histories. Gender, age, and BMI (calculated from height and weight measurements) were also recorded. Fasting whole blood samples were then drawn as described below. A $15 gift certificate to a local supermarket or department store was given to each person at the end of the clinic visit. Two hundred forty-nine individuals were enrolled in the Sacramento population.
The Johns Hopkins University Bloomberg School of Public Health Committee on Human Research approved the study procedures and protocol, which was in accord with Good Clinical Practice guidelines and the Declaration of Helsinki, version 1989. The population consisted of 510 adults (257 females, 253 males) aged 30 – 69 years residing in the greater Washington, D.C. area. Participants were recruited to participate in a study of dietary recall between July 2002 and June 2004  and were included in the present study as an adjunct to the previous study. Potential participants were recruited through email, advertisements in local newspapers, and announcements on USDA ARS websites. Participants attended an informational meeting concerning study procedures before completing a health history questionnaire. The questionnaire covered general medical, personal, and family histories. Folate intake assessed by the Block Dietary Folate Equivalents (DFE) Screener is available through Supporting Information (Additional file 1: Figure S1). A medical screening evaluation included measurement of height, weight, blood pressure, and laboratory analysis of fasting blood and urine. Age, BMI (calculated from height and weight measurements) and gender were also recorded. The medical history and laboratory results were reviewed by study investigators and a cooperating physician to confirm good health of the participants and that there was no evidence of underlying disease, untreated thyroid disorders, gastrointestinal disease, malabsorption syndromes, history of eating disorders, cancer, or DM. All participants were weight stable, were not actively pursuing a weight-loss regimen, were not taking medications known to affect food intake or appetite, and were not taking diuretics or other medications that may affect water balance. Pregnant and lactating women were excluded from the study. Participants were compensated according to the requirements of the main study .
Blood analysis and genotyping
Fasting whole blood was collected in triplicate. It was collected 1) into spray-dried K2EDTA tubes for assessment of complete blood count, red blood cell (RBC) folate, plasma vitamin B6, and plasma Hcy; 2) into serum separator tubes for the assessment of serum vitamin B12, serum folate, and lipid panel; and 3) into 8.5-mL whole blood DNA tubes for the assessment of genomic DNA. The preparation and analysis of the collected blood samples has been previously described . The results of RBC folate concentrations and plasma vitamin B12 levels are included in the online Supporting Information (Additional file 1: Figures S2 and S3). Briefly and specifically to this study, the lipid panel was completed using the Beckman LXI and LX20 Pro (Beckman Instruments, Brea, CA) at the Department of Pathology laboratory of the University of California, Davis. Results of serum triglycerides, total cholesterol, HDL-C, and LDL-C are provided in the online Supporting Information (Additional file 1: Figures S4 S5, S6, S7). Additionally, genomic DNA was extracted from whole blood with the use of a PAXgene Blood DNA kit (PreAnalytix; Qiagen, Inc., Valencia, CA).
Blood was collected into serum separator tubes after a 12 h fast. Blood samples were allowed to sit for 30 min before centrifugation at 2000 × g for 10 min at 4°C then aliquotted and stored at −80°C until analysis. Plasma-deplete samples were analysed for folate concentration as described previously . The results of RBC folate concentrations and plasma vitamin B12 levels are included in the online Supporting Information (Additional file 1: Figures S2 and S3). Thawed samples were analyzed in duplicate for HDL cholesterol on a Dade Behring Dimension xPand clinical chemistry analyzer (Siemens, Erlangen, Germany). Results of serum triglycerides, total cholesterol, HDL-C, and LDL-C are provided in the online Supporting Information (Additional file 1: Figures S4, S5, S6, S7). High quality DNA was extracted from white blood cells using the Gentra PureGene Blood DNA Purification Kit (Qiagen, Inc.) as was previously described  prior to SNP analysis.
Genotyping of SNP
SNP included in analyses
APOA1, protein component of HDL, promotes cholesterol efflux from tissues to liver
APOA5, protein component of HDL, regulates plasma triglycerides
rs3135506, rs662799, rs12272004
ATP-binding cassette, subfamily A, member 1
ABCA1, transport AAs, sugars, vitamins, cholesterol, pt-choline to lipid acceptor apoA1
rs2230806, rs2230808, rs4149267, rs4149327
β-Carotene 15,15’-monooxygenase 1
BCMO1, cleave β-carotene
rs12934922, rs6564851, rs7501331
β-Carotene oxygenase 2
BCMO2, cleave β-carotene
rs11214139, rs2250417, rs35361223
Betaine-homocysteine S-methyl transferase
BHMT, methylate Hcy
Cholesteryl ester transfer protein, plasma
CETP, transfer cholesterol
rs12708980, rs5880, rs5882, rs7205804, rs7499892
CBS, synthesize cystathionine
Cytochrome P450, family 4, subfamily F, polypeptide 2
CYP4F2, involved in cholesterol synthesis
rs3093156, rs2108622, rs3093168, rs3093194
Folate hydrolate (prostate specific membrane antigen) 1
FOLH1 (GCPII), hydrolyze folate
Methionine synthase reductase
MTRR, remethylate Hcy
MTHFR, distribute one carbon units
MTR, remethylate Hcy
Microsomal triglyceride transfer protein
MTTP, lipoprotein assembly
rs10516445, rs1057613, rs3828542, rs881980
Niemann-Pick disease, type C1, gene-like 1
NPC1L1, absorption of intestinal cholesterol and α-tocopherol transport
rs11763759, rs217420, rs217430, rs217432
Retinol binding protein 1, cellular
rs2118981, rs35674260, rs3772875
Scavenger receptor class B, member 1
SCARB1, intestinal absorption of carotenoids
rs10773105, rs12582221, rs7306660, rs7967521, rs838892
Serine palmitoyltransferase, long chain base subunit 1
SPTLC1, sphingolipid biosynthesis
rs11790991, rs2297568, rs7858659
Solute carrier family 19 (folate transporter), member 1
SLC19A1, reduced folate carrier
rs13050920, rs3788205, rs3788199
Solute carrier family 46 (folate transporter), member 1
SLC46A1, proton coupled folate transporter (PCFT)
rs11080058, rs1128162, rs17719944, rs35714695, rs739439
CD36 Thrombospondin receptor
CD36, platelet surface glycoprotein, binds to oxidized LDL
rs1358337, rs3211834, rs3211931, rs3211956
α-Tocopherol transfer protein
α-TTP, transport α-tocopherol
rs4501570, rs4587328, rs4606052
TCII, transport vitamin B12
Total SNP included in analysis
Data processing and statistical analysis
Where y was the adjusted phenotype (HDL adjusted for gender and weight), b 1 x + b 0 represented the model, and the error term, ϵ, expressed the random residual effect.
Statistical significance of fixed effects
Participant data (gender, age, height, weight, and calculated BMI) were tested to adjust phenotypes for systematic effects using a full (including covariates) versus reduced model regression equation. The regression sums of squares were calculated both for a reduced and for the full model. An F test was then performed to find the significance of the full versus the reduced model. A P-value threshold of 0.01 was used to establish significant associations. Gender and body weight effects were statistically significant; therefore, adjusted phenotypes were obtained for all samples.
The linear regression was also performed including SNP interactions using the SVS version 7 regression module from Golden Helix. FDR was controlled according to a previous method  and a cutoff for a significant association value was set at FDR q value < 0.01.
Single nucleotide polymorphism(s)
- HDL and HDL-C:
High-density lipoprotein high-density lipoprotein cholesterol
Gene symbol for cholesteryl ester transfer protein
Gene symbol for solute carrier family 46 (folate transporter), member 1 or proton-coupled folate transporter
Gene symbol for solute carrier family 19 (folate transporter), member 1, or reduced folate carrier
Gene symbol for thrombosponding receptor or glycoprotein IIIb/platelet glycoprotein IV
Gene symbol for beta-carotene monooxygenase 1
Gene symbol for apolipoproteinA-V
Gene symbol for ATP-binding cassette transporter member 1
Peroxisome proliferator-activated receptor
Gene symbol for serine palmitoyltransferase
Gene symbol for scavenger receptor class B type 1
Gene symbol for cellular retinol binding protein
Gene symbol for betaine-homocysteine methyltransferase
Gene symbol for methylene tetrahydrofolate reductase
Coronary heart disease
- LDL and LDL-C:
Low-density lipoprotein and low-density lipoprotein cholesterol
Reverse cholesterol transport
Allele substitution effect
False discovery rate
Minor allele frequency
Very low-density lipoproteins
Cholesteryl ester transfer protein
Genome-wide association study
Proton-coupled folate transporter
Reduced folate carrier
Thrombosponding receptor or glycoprotein IIIb/platelet glycoprotein IV
Body mass index
Beta-carotene monooxygenase 1
National Health and Nutrition Examination Survey
ATP-binding cassette transporter member 1
ATP-binding cassette transporter G1
Gene symbol for lipoprotein lipase
Potassium ethylenediaminetetraacetic acid
Red blood cell
Matrix-assisted laser desorption ionization time-of-flight
Dietary Folate Equivalents.
We are grateful to Alma Islas-Trejo for her work on the genotyping of the BHMT polymorphism. We thank the reviewers for their careful consideration of the manuscript and their helpful and perceptive comments.
This work was supported by grant no. DK 45939 from the National Institutes of Health, Grant/Cooperative Agreement no. RO1 8928 from the Centers for Disease Control and Prevention, and Regional Research Grant no. W2002, Project #CA-D*-NTR-7778-H from the US Department of Agriculture. The contents are the sole responsibility of the authors and do not necessarily represent the official views of the NIH, CDC, or USDA.
- Patel CJ, Cullen MR, Ioannidis JP, Butte AJ: Systematic evaluation of environmental factors: persistent pollutants and nutrients correlated with serum lipid levels. Int J Epidemiol. 2012, 41: 828-843. 10.1093/ije/dys003PubMed CentralView ArticlePubMed
- Clifford AJ, Chen K, McWade L, Rincon G, Kim S-H, Holstege DM, Owens JE, Liu B, Muller H-G, Medrano JF: Gender and single nucleotide polymorphisms in MTHFR, BHMT, SPTLC1, CRBP2, CETP, and SCARB1 are significant predictors of plasma homocysteine normalized by RBC folate in healthy adults. J Nutr. 2012, 142: 1764-1771. 10.3945/jn.112.160333PubMed CentralView ArticlePubMed
- Semmler A, Farmand S, Moskau S, Stoffel-Wagner B, Linnebank M: The G allele of transcobalamin 2 c.776C→G is associated with an unfavorable lipoprotein profile. Ann Nutr Metab. 2010, 57: 112-115. 10.1159/000320418View ArticlePubMed
- Kitami T, Rubio R, O'Brien W, Quackenbush J, Nadeau JH: Gene-environmental interactions reveal a homeostatic role for cholesterol metabolism during dietary folate perturbation in mice. Physiol Genomics. 2008, 35: 182-190. 10.1152/physiolgenomics.00294.2007PubMed CentralView ArticlePubMed
- Corsetti JP, Ryan D, Moss AJ, McCarthy J, Goldenberg I, Zareba W, Sparks CE: CD3, CD36 Thrombospondin-4 polymorphism (A387P) predicts cardiovascular risk in postinfarction patients with high HDL cholesterol and C-reactive protein levels. Thromb Haemost. 2011, 106: 993-1234. 10.1160/TH11-09-0677View Article
- Kim DS, Burt AA, Ranchalis JE, Richter RJ, Marshall JK, Nakayama KS, Jarvik ER, Eintracht JF, Rosenthal EA, Furlong CE, Jarvik GP:Dietary cholesterol increases paraoxonase 1 enzyme activity. J Lipid Res. 2012, 53: 2450-2458. 10.1194/jlr.P030601PubMed CentralView ArticlePubMed
- Zelijkovic A, Vekic J, Spasojevic-Kalimanovska V, Jelic-Ivanovic Z, Peco-Antic A, Kostic M, Vasic D, Spasic S: Characteristics of LDL- & HDL-subclasses in pediatric renal transplant patients. Transpl Int. 2011, 24: 1094-1102. 10.1111/j.1432-2277.2011.01313.xView Article
- Hewing B, Fisher EA: Rationale for cholesteryl ester transfer protein inhibition. Curr Opin Lipidol. 2012, 23: 372-376. 10.1097/MOL.0b013e328353ef1dPubMed CentralView ArticlePubMed
- Gauthier A, Lau P, Zha X, Milne R, McPherson R: Cholesteryl ester transfer protein directly mediates selective uptake of high density lipoprotein cholesteryl esters by the liver. Arterioscler Thromb Vasc Biol. 2005, 25: 2177-2184. 10.1161/01.ATV.0000183613.13929.13View ArticlePubMed
- Davidson MH: HDL and CETP Inhibition: Will this DEFINE the Future?. Curr Treatm Options Cardiovasc Med. 2012, 14: 384-390. 10.1007/s11936-012-0191-8.View Article
- Zhang L, Yan F, Zhang S, Lei D, Charles MA, Cavigiolio G, Oda M, Krauss RM, Weisgraber KH, Rye KA: Structural basis of transfer between lipoproteins by cholesteryl ester transfer protein. Nat Chem Biol. 2012, 8: 342-349. 10.1038/nchembio.796PubMed CentralView ArticlePubMed
- Drenos TF, Talmud PJ, Casas JP, Smeeth L, Palmen J, Humphries SE, Hingorani AD: Integrated associations of genotypes with multiple blood biomarkers linked to coronary heart disease risk. Hum Mol Genet. 2009, 18: 2305-2316. 10.1093/hmg/ddp159PubMed CentralView ArticlePubMed
- Pan SL, Wang F, Lu ZP, Liu CW, Hu CY, Luo H, Peng JH, Luo XQ, Pang GF, Lu SH: Cholesteryl ester transfer protein Taq IB polymorphism and its association with serum lipid levels and longevity in Chinese Bama Zhuang population. Lipids Health Dis. 2012, 11: 10.1186/1476-511X-11-26.
- Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, Pirruccello JP, Ripatti S, Chasman DI, Willer CJ: Biological, clinical, and population relevance of 95 loci for blood lipids. Nature. 2010, 466: 707-713. 10.1038/nature09270PubMed CentralView ArticlePubMed
- Tietjen I, Hovingh GK, Singaraja RR, Radomski C, Barhdadi A, McEwen J, Chan E, Mattice M, Legendre A, Franchini PL: Segregation of LIPG, CETP, and GALNT2 mutations in Caucausian familesi with extremely high HDL cholesterol. PLoS One. 2012, 7: e37437- 10.1371/journal.pone.0037437PubMed CentralView ArticlePubMed
- Thompson A, Angelantonio ED, Sarwar N, Erquo S, Saleheen D, Dullart RPF, Keavney B, Ye Z, Danesh J: Association of cholesteryl ester transfer protein genotypes with CETP mass and activity, lipid levels, and coronary risk. J Amer Med Assoc. 2008, 299: 2777-2788. 10.1001/jama.299.23.2777.View Article
- Rejeb J, Omezzine A, Boumaiza I, Rebhi L, Rejeb NB, Nabli N, Abdelaziz AB, Boughzala E, Bouslama A: Four polymorphisms of cholesteryl ester transfer protein gene and coronary stenosis in a Tunisian population. J Cardiovasc Med (Hagerstown). 2012, 13: 546-553.
- Desmoulin SK, Hou Z, Gangjee A, Matherly LH: The human proton-coupled folate transporter: Biology and therapeutic applications to cancer. Cancer Biol Ther. 2012, 13: 1-19. 10.4161/cbt.13.1.18438View Article
- Semmler A, Moskau S, Grigull A, Farmand S, Klockgether T, Smulders YM, Blom HJ, Zur B, Stoffel-Wagner B, Linnebank M: Plasma folate levels are associated with the lipoprotein profile: a retrospective database analysis. Nutr J. 2010, 9: 31-34. 10.1186/1475-2891-9-31PubMed CentralView ArticlePubMed
- Mauritz R, Peters GJ, Kathmann I, Teshale H, Noordhuis P, Comijn EM, Pinedo HM, Jansen G: Dynamics of antifolate transport via the reduced folate carrier and the membrane folate receptor in murine leukaemia cells in vitor and in vivo. Cancer Chemother Pharmacol. 2008, 62: 937-948. 10.1007/s00280-008-0683-0View ArticlePubMed
- Mahabir S, Ettinger S, Johnson L, Baer DJ, Clevidence BA, Hartman TJ, Taylor PR: Measures of adiposity and body fat distribution in relation to serum folate levels in postmenopausal women in a feeding study. Eur J Clin Nutr. 2008, 62: 644-650. 10.1038/sj.ejcn.1602771PubMed CentralView ArticlePubMed
- Carter MF, Powell TL, Li C, Myatt L, Dudley D, Nathanielsz P, Jansson T: Fetal serum folate concentrations and placental folate transport in obese women. Am J Obstet Gynecol. 2011, 205: 83.e17-83.e25. 10.1016/j.ajog.2011.02.053.View Article
- Rac ME, Suchy J, Kurzawski G, Kurlapska A, Safranow K, Rac M, Sagasz-Tysiewicz D, Krzystolik A, Poncyljusz W, Jakubowska K: Polymorphism of the CD36 gene and cardiovascular risk factors in patients with coronary artery disease manifested at a young age. Biochem Genet. 2012, 50: 103-111. 10.1007/s10528-011-9475-zPubMed CentralView ArticlePubMed
- Knowles JW, Wang H, Itakura H, Southwick A, Myers RM, Iribarren C, Fortmann SP, Go AS, Quertermous T, Hlatky MA: Association of polymorphisms in platelet and hemostasis system genes with acute myocardial infarction. Am Heart J. 2007, 154: 1052-1058. 10.1016/j.ahj.2007.05.021PubMed CentralView ArticlePubMed
- Heni M, Mussing K, Machicao F, Machann J, Schick F, Claussen CD, Stefan N, Fritsche A, Haring HU, Staiger H: Variants in the CD36 gene locus determine whole-body adiposity, but have no independent effect on insulin sensitivity. Obesity. 2011, 19: 1004-1009. 10.1038/oby.2010.251View ArticlePubMed
- Chien KL, Hsu HC, Liu PH, Lin HJ, Chen MF: Common sequence variants in CD36 gene and the levels of triglyceride and high-density lipoprotein cholesterol among ethnic Chinese in Taiwan. Lipids Health Dis. 2012, 11: 10.1186/1476-511X-11-174.
- Ferrucci L, Perry JR, Matteini A, Perola M, Tanaka T, Silander K, Rice N, Melzer D, Murray A, Cluett C: Common variation in the beta-carotene 15, 15'-monooxygenase 1 gene affects circulating levels of carotenoids: a genome-wide association study. Am J Hum Genet. 2009, 84: 123-133. 10.1016/j.ajhg.2008.12.019PubMed CentralView ArticlePubMed
- Wang CX, Wongsiriroj N, Deckelbaum RJ, Blaner WS, Harrison EH: New findings on apo-carotenoid metabolites of B-carotene. Sight and Life. 2012, 26: 18-27.
- Lietz G, Oxley A, Leung W, Hesketh J: Single nucleotide polymorphisms upstream from the β-carotene 15, 15'-monoxygenase gene influence provitamin A conversion efficiency in female volunteers. J Nutr. 2012, 142: 161S-165S. 10.3945/jn.111.140756View ArticlePubMed
- Eroglu A: Hruszkewycz DP, dela Sena C, Narayanasamy S, Riedl KM, Kopec RE, Schwartz SJ, Curley RWJ, Harrison EH: Naturally occuring eccentric cleavage products of provitamin A B-carotene function as antagonists of retinoic acid receptors. J Biol Chem. 2012, 287: 15886-15895. 10.1074/jbc.M111.325142PubMed CentralView ArticlePubMed
- Harrison EH: dela Sena C, Eroglu A, Fleshman MK: The formation, occurrence, and function of B-apocarotenoids: B-carotene metabolites that may modulate nuclear receptor signaling. Am J Clin Nutr. 2012, 96: 1189S-1192S. 10.3945/ajcn.112.034843PubMed CentralView ArticlePubMed
- Perry JR, Ferrucci L, Bandinelli S, Guralnik J, Semba RD, Rice N, Melzer D, Saxena R, Scott LJ, McCarthy MI: Circulating beta-carotene levels and type 2 diabetes - cause or effect?. Diabetologia. 2009, 52: 2117-2121. 10.1007/s00125-009-1475-8PubMed CentralView ArticlePubMed
- Liu ZK, Hu M, Baum L, Thomas GN, Tomlinson B: Associations of polymorphisms in the apolipoprotein A1/C3/A4/A5 gene cluster with familial combined hyperlipidaemia in Hong Kong Chinese. Atherosclerosis. 2010, 208: 427-432. 10.1016/j.atherosclerosis.2009.08.013View ArticlePubMed
- Yin RX, Li Y-Y, Lai C-Q: Apolipoprotein A1/C3/A5 haplotypes and serum lipid levels. Lipids Health Dis. 2011, 10: 1-16. 10.1186/1476-511X-10-1View Article
- Li Y-Y, Yin RX, Lai C-Q, Li M, Long X-J, Li K-L, Liu W-Y, Zhang L, Wu J-Z: Association of apolipoprotein A5 gene polymorphisms and serum lipid levels. Nutr Metab Cardivasc Dis. 2011, 21: 947-956. 10.1016/j.numecd.2010.04.004.View Article
- Vedhachalam C, Duong PT, Nickel M, Nguyen D, Dhanasekaran P, Saito H, Rothblat GH, Lund-Katz S, Phillips MC: Mechanism of ATP-binding cassette transporter A1-mediated cellular lipid efflux to apolipoprotein A-1 and formation of high density lipoprotein particles. J Biol Chem. 2007, 282: 25123-25130. 10.1074/jbc.M704590200View ArticlePubMed
- Denis M, Landry YD, Zha X: ATP-binding cassette A1-mediated lipidation of apolipoprotein A-1 occurs at the plasma membrane and not in the endocytic compartments. J Biol Chem. 2008, 283: 16178-16186. 10.1074/jbc.M709597200PubMed CentralView ArticlePubMed
- Kolsch H, Lutjohann D, Jessen F, Von Bergmann K, Schmitz S, Urbach H, Maier W, Heun R: Polymorphism in ABCA1 influences CSF 24S-hydroxycholesterol levels but is not a major risk factor of Alzheimer's disease. Int J Mol Med. 2006, 17: 791-794.PubMed
- Sorci-Thomas MG, Owen JS, Fulp B, Bhat S, Zhu X, Parks JS, Shah D, Jerome WG, Gerelus M, Zabalawi M, Thomas MJ: Nascent high density lipoproteins formed by ABCA1 resemble lipid rafts and are structurally organized by three apoA-1 monomers. J Lipid Res. 2012, 53: 1890-1909. 10.1194/jlr.M026674PubMed CentralView ArticlePubMed
- Kennedy MA, Barrera GC, Nakamura K, Baldan A, Tarr P, Fishbein MC, Frank J, Francone OL, Edwards PA: ABCG1 has a critical role in mediating cholesterol efflux to HDL and preventing cellular lipid accumulation. Cell Metab. 2005, 1: 121-131. 10.1016/j.cmet.2005.01.002View ArticlePubMed
- Linga Reddy MV, Iatan I, Weissglas-Volkov D, Nikkola E, Haas BE, Juvonen M, Ruel I, Sinsheimer JS, Genest J, Pajukanta P: Exome sequencing identifies two rare variants for low HDL-C in an extended family. Circ Cardiovasc Genet. 2012, 5: 538-546. 10.1161/CIRCGENETICS.112.963264View Article
- Ayaori M, Yakushjii E, Ogura M, Nakaya K, Hisada T, Uto-Kondo H, Takiguchi S, Terao Y, Sasaki M, Komatsu T: Retinoic acid receptor agonists regulate expression of ATP-binding cassette transporter G1 in macrophages. Biochim Biophys Acta. 1821, 2012: 561-572.
- Wiersma H, Nijstad N, de Boer JF, Out R, Hogewerf W, van Berkel TJ, Kuipers F, Tietge UJ: Lack of Abcg1 results in decreased plasma HDL cholesterol levels and increased biliary cholesterol secretion in mice fed a high cholesterol diet. Atherosclerosis. 2009, 206: 141-147. 10.1016/j.atherosclerosis.2009.02.022View ArticlePubMed
- Bodzioch M, Orso E, Klucken J, Langmann T, Bottcher A, Diederich W, Drobnik W, Barlage S, Buchler C, Porsch-Ozcurumez M: The gene encoding ATP-binding cassette transporter 1 is mutated in Tangier disease. Nat Genet. 1999, 22: 347-351. 10.1038/11914View ArticlePubMed
- Brooks-Wilson A, Marcil M, Clee SM, Zhang L-H, Roomp K, van Dam M, Brewer C, Collins JA, Molhuizen HOF, Loubser O: Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficienc. Nat Genet. 1999, 22: 336-345. 10.1038/11905View ArticlePubMed
- Porchay I, Pean F, Bellili N, Royer B, Cogneau J, Chesnier M-C, Caradec A, Tichet J, Balkau B, Marre M, Fumeron F: ABCA1 single nucleotide polymorphisms on high-density lipoprotein-cholesterol and overweight: the D.E.S.I.R. study. Obesity. 2006, 14: 1874-1879. 10.1038/oby.2006.217View ArticlePubMed
- Soro-Paavonen A, Naukkarinen J, Lee-Rueckert M, Watanabe H, Rantala E, Soderlund S, Hiukka A, Kovanen PT, Jauhianen M, Peltonen L, Taskinen M-R: Common ABCA1 variants, HDL levels, and cellular cholesterol efflux in subjects with familial low HDL. J Lipid Res. 2007, 48: 1409-1416. 10.1194/jlr.P600012-JLR200View ArticlePubMed
- Wang J, Burnett JR, Near S, Young K, Zinman B, Hanley AJG, Connelly PW, Harris SB, Hegele RA: Common and rare ABCA1 variants affecting plasma HDL cholesterol. Arterioscler Thromb Vasc Biol. 2000, 20: 1983-1989. 10.1161/01.ATV.20.8.1983View ArticlePubMed
- Voight BF, Peloso GM, Orho-Melander M, Frikke-Schmidt R, Barbalic M, Jensen MK, Hindy G, Hólm H, Ding EL, Johnson T: Plasma HDL cholesterol and risk of myocardial infarction: a mendelian randomisation study. Lancet. 2012, 380: 572-580. Erratum in: Lancet, 2012 Aug 2011;2380(9841):2564, 10.1016/S0140-6736(12)60312-2PubMed CentralView ArticlePubMed
- Annema W, Dikkers A: Freark de Boer J, Gautier T, Rensen PC, Rader DJ, Tietge UJ: ApoE promotes hepatic selective upatke but not RCT due to increased ABCA1-mediated cholesterol efflux to plasma. J Lipid Res. 2012, 53: 929-940. 10.1194/jlr.M020743PubMed CentralView ArticlePubMed
- Zhang LH, Kamanna VS, Ganji SH, Xiong XM, Kashyap ML: Niacin increases HDL biogenesis by enhancing DR4-dependent transcription of ABCA1 and lipidation of apolipoprotein A-1 in HepG2 cells. J Lipid Res. 2012, 53: 941-950. 10.1194/jlr.M020917PubMed CentralView ArticlePubMed
- Moshfegh AJ, Rhodes DG, Baer DJ, Murayi T, Clemens JC, Rumpler WV, Paul DR, Sebastian RS, Kuczynksi KJ, Ingwersen LA: The US Department of Agricultur Automated Multiple-Pass Method reduces bias in the collection of energy intakes. Am J Clin Nutr. 2008, 88: 324-332.PubMed
- dbSNP Home Page: Short Genetic Variations. http://www.ncbi.nlm.nih.gov/projects/SNP/; Last accessed 06/21/2012
- e!Ensembl. http://uswest.ensembl.org/index.html; Last accessed 06/21/2012,
- Storey J: A direct approach to false discovery rates. J R Statist Soc. 2002, B 64: 479-498.View Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.