Introduction

Single-nucleotide polymorphisms (SNPs) in the apolipoprotein A5 gene (APOA5), located on chromosome 11q23 near the APOA1–C3–A4 gene cluster, have pleiotropic effects on changes in the levels of triglycerides (TG) and high-density lipoprotein cholesterol (HDL-C) [1, 2]. Recently, a genome-wide association study (GWAS) reported that the minor allele of a SNP, rs964184, has a bivariate effect on TG-elevation and HDL-C-decrease [3]. These pleiotropic bivariate associations of APOA5 SNPs, in turn, affect the risk for metabolic syndrome (MS) [1, 2, 4].

Haplotype analyses of association of APOA5 SNPs with lipid levels and risk for MS have focused primarily on variants located within APOA5 or within the APOA1–C3–A4–A5 gene cluster [5–12]. In particular, the APOA5 haplotypes, APOA5*1, and *2, and *3, comprised of rs662799 (-1131T>C), rs3135506 (56C>G), rs2072560 (IVS3 + 476G>A), and rs2266788 (1259T>C), have been studied for association with lipid levels and MS [5, 11–13]. The APOA5*2 haplotype, determined by the minor alleles of rs662799, rs2072560, and rs2266788, has been associated with elevated TG levels and risk for the development of MS [11–13], while the APOA5*3 haplotype, determined by the minor allele of rs3135506, has been shown to associate with increased levels of TG [4, 5].

However, APOA5 is adjacent to the zinc finger protein 259 gene (ZNF259; approximately 1.3 kb downstream of APOA5), so that separation of the effects of these closely located genes is difficult. In fact, it is possible that crosstalk between APOA5 and ZNF259 contributes to modulation of plasma TG levels. Although the precise mechanisms associated with this have yet to be determined, research has indicated that APOA5 interaction with proteoglycan-bound lipoprotein lipase and with low-density lipoprotein receptor can reduce plasma TG levels by hydrolysis of TG-rich lipoproteins and by endocytosis of remnant lipoproteins, respectively [14]. ZNF259 encodes a zinc finger protein ZPR1 that binds to the cytoplasmic tyrosine kinase domain of epidermal growth factor receptor (EGFR) in quiescent cells [15]. When mitogens such as EGF disrupt ZPR1 binding by tyrosine phosphorylation of EGFR, ZPR1 is translocated to the nucleus. It has been reported that remnant lipoproteins, among TG-rich lipoproteins, induce EGFR phosphorylation in smooth muscle cells, which results in smooth muscle cell proliferation and atherosclerosis risk [16] and that EGF concentrations in plasma and peripheral blood mononuclear cells associate with lipid concentrations, including TG and HDL-C [17]. Furthermore, the variants exerting effects on TG and HDL-C levels, including rs662799, rs3135506, rs651821, rs2072560, rs2266788, rs6589566, and rs964184, cluster together and are in strong linkage disequilibrium (LD) with each other around the two genes [3, 18, 19]. Therefore, construction of haplotypes with variants spanning both genes is necessary to estimate the precise TG-elevating and HDL-C-lowering effects of APOA5 SNPs in a population.

The current study aimed to determine the effects of haplotypes constructed from variants located around APOA5 and ZNF259 on the levels of TG and HDL-C, in terms of a ratio of TG to HDL-C, as well as on the risk for MS development. Due to the stronger effects of TG levels on risk of coronary disease in the context of HDL-C levels [20] and the high correlation between the levels of TG and HDL-C, the TG:HDL-C ratio can be a more efficient predictor of the risk for coronary artery diseases [21, 22]. In addition, sex-dependent association patterns of these haplotypes were evaluated due to the reported gender effect of APOA5 haplotypes on MS risk [11].

Results

Characteristics of recruited subjects and constructed haplotypes

Characteristics of the recruited subjects are presented in Table 1 and Additional file 1. Age and the prevalence of MS between men and women, as well as the gender proportion according to MS status, were not statistically different, unlike other clinical characteristics. As the linkage of rs6589566 (A>G) with rs662799 (-1131T>C; r2 = 0.60) and rs651821 (-3A>G; r2 = 0.56) in the studied population was weaker than that in the Asian HapMap population (both r2 = 0.80), three rather than two APOA5–ZNF259 haplotypes with an allele frequency > 5% were constructed, such that TAA (ht1: 68.9%), CGG (ht2: 20.4%), and CGA (ht3: 8.57%) in the order of rs662799, rs651821, and rs6589566 (Figure 1). The genotypes of the three SNPs did not deviate from Hardy–Weinberg equilibrium (p > 0.05). Therefore, the genetic effects of these three haplotypes on a ratio of log[TG] (log-transformed TG): HDL-C and MS risk were evaluated in Koreans.

Table 1 Characteristics of recruited subjects
Full size table
Figure 1
figure 1

Genetic location, linkage disequilibrium (LD), and haplotypes of single-nucleotide polymorphisms (SNPs). (A) The SNPs rs662799 and rs651821 are respectively located in the promoter and 5′ untranslated regions of APOA5 that has 4 exons, whereas SNP rs6589566 is found in intron 12 of ZNF259 that has 14 exons. (B) LD structure of three SNPs: D′ and r2. (C) Haplotype sequence and frequency: haplotypes with frequencies over 5% were named as ht1, ht2, and ht3.

Full size image

Associations of APOA5-ZNF259 haplotypes with TG:HDL-C ratio and MS risk

Carriage of haplotypes ht2 or ht3, versus the ht1 haplotype, was significantly associated with increased levels of the log[TG]: HDL-C ratio (for ht2 vs ht1, effect = 0.00222, 95% CI = 0.00130 - 0.00315; for ht3 vs ht1, effect = 0.00460, 95% CI = 0.00313 - 0.00607), whereas carriage of the ht3 haplotype versus the ht2 haplotype showed only a tendency toward elevating the log[TG]:HDL-C ratio (Table 2). When the population was stratified by gender, the genetic effects of the two haplotypes, ht2 and ht3, versus ht1 haplotype on log[TG]:HDL-C ratio remained significant, in both genders (for ht2 vs ht1: effect = 0.00328, 95% CI = 0.00154 - 0.00502, in men and effect = 0.00157, 95% CI = 0.00507 - 0.00264 in women; for ht3 vs ht1: effect = 0.00606, 95% CI = 0.00324 - 0.00888 in men and effect = 0.00391, 95% CI = 0.00223 - 0.00558 in women). Interestingly, haplotype ht3, with reference to haplotype ht2, was significantly associated with elevation of the log[TG]:HDL-C ratio only in women (effect = 0.00473, 95% CI = 0.00130 - 0.00816). Therefore, the ht3 haplotype showed a sex-dependent association with the log[TG]:HDL-C ratio.

Table 2 Association of haplotypes in APOA5 and ZNF259 gene regions with log[TG]:HDL-C ratio
Full size table

The sex-dependent association of the haplotype ht3 was more marked in analyses of the risk of MS (Table 3). The genetic effect of the ht2 haplotype versus the ht1 haplotype on an increased risk for MS remained significant in men (odds ratio (OR) = 1.50, 95% CI = 1.16 – 1.94), while the effect of ht3 haplotype versus ht1 haplotype remained significant in women (OR = 1.41, 95% CI = 1.00 – 1.97). Moreover, association of the ht3 haplotype versus the ht2 haplotype was significant only in women (OR = 2.67, 95% CI = 1.25 – 6.07). Taken together, unlike men, women harboring the ht3 haplotype were the most susceptible to MS risk and elevation of the log[TG]:HDL-C.

Table 3 Association of haplotypes in APOA5 and ZNF259 gene regions with metabolic syndrome (MS) risk
Full size table

Discussion

Recent GWAS findings indicated that genes with variants influencing MS per se included the APOA5, ZNF259, and BUD13 homolog); these genes are known to play an important role in lipid metabolism [2]. We also recently reported that the APOA5 - 1131T>C polymorphism (rs662799) is associated with the risk of MS because of its marked effect on serum TG and HDL-C levels in Korean subjects [1]. The current study has afresh identified that haplotype constructed by SNPs in the closely positioned APOA5–ZNF259 gene region was sex-dependently associated with the ratio of TG to HDL-C and the risk of MS development in Koreans. Specifically, haplotype ht3, viz., C–G–A at rs662799, rs651821, and rs6589566, respectively, was most significantly associated with elevation of the log[TG]:HDL-C ratio, as well as MS susceptibility, in women. We suggest that a causal variant in LD with this APOA5–ZNF259 haplotype may give rise to TG and HDL-C variations and thus contribute to the risk of MS and coronary artery disease.

In terms of sex-specific influences of the APOA5–ZNF259 haplotype on the TG:HDL-C ratio and MS risk, rs6589566, which is located in the ZNF256 intron, will have an important role only in women, such that the major allele of the ZNF256 SNP in combination with the minor alleles of the two APOA5 SNPs can result in greater susceptibility to elevation of the TG:HDL-C ratio and MS risk (Tables 2 and 3). In contrast, in men, the minor alleles of the two SNPs located in the APOA5 locus (rs662799 and rs651821), rather than the rs6589566 allele, will produce the dominant effects. In other words, sex-dependent association of the APOA5 SNPs with the TG:HDL-C ratio and MS risk can be determined by using haplotype analyses of the SNPs in the APOA5–ZNF259 locus. These gender effects are different from those previously reported [11], although we also found greater susceptibility to MS with the presence of the ht2 haplotype in men than in women (Table 3), which is similar to the previously reported gender effect of the APOA5*2 haplotype on MS risk [11].

In a study involving haplotype analysis of a 150-kb region spanning the apolipoprotein gene cluster (APOA1–C3–A4–A5), the SNPs in ZNF259 were separated from those in the APOA5 haplotype block [6]. However, in this haplotype analysis, rs1942478 and rs603446 in the ZNF259 intron, were weakly correlated with rs6589566, which is in strong LD with APOA5 SNPs: r2 = 0.125 for rs1942478 and r2 = 0.164 for rs603446 in the Chinese HapMap population [6]. Therefore, the APOA5 haplotype block can be extended to the ZNF259 gene region using rs6589566 and rs964184 located in the ZNF259 intron.

The most obvious weakness of our study is that it is cross-sectional; thus, it would be necessary to confirm the sex-dependent effects of the APOA5–ZNF259 haplotypes on cardiometabolic risk via longitudinal analyses. Although haplotype ht3 and the TG:HDL-C ratio appear to function in identifying cardiometabolic risk, we cannot conclude that the TG:HDL-C ratio will be as effective in predicting clinical outcomes as a diagnosis of MS. However, despite these weaknesses, the simplicity of the TG:HDL-C ratio approach has obvious advantages: the TG:HDL-C ratio, which relies on two commonly used laboratory measurements, can provide a relatively simple way to identify apparently healthy insulin-resistant persons with increased cardiometabolic risk, since the TG:HDL-C concentration ratio is significantly related to a direct measure of insulin-mediated glucose disposal [23–25].

Conclusions

In conclusion, we found that haplotypes formed from three SNPs located in APOA5 and ZNF259, viz., rs662799, rs651821, and rs6589566, were significantly associated with susceptibility to MS and elevation of the log[TG]:HDL-C ratio, sex-dependently, in Koreans; in particular, the ht3 haplotype was associated with these traits in women. We believe that our findings will help to further develop the clinical utility of the APOA5 haplotypes and the TG:HDL-C ratio as a way to identify cardiometabolic risk.

Methods

Study population

All subjects provided written informed consent to participate in the study, and the study was approved by the Institutional Review Board of the Korea Institute of Oriental Medicine.

We recruited 2,949 subjects from 22 oriental medical clinics who were a part of the Korea Constitution Multicenter Study for 6 years since 2006 [1, 26]. This population contained 690 patients who had hypertension, dyslipidemia, and/or diabetes. None of the subjects had a history of cancer treatment, thyroid dysfunction, or postmenopausal hormonal therapy. The subjects with MS were defined according to the modified guidelines of National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) [27], which stipulated that at least three of the following five criteria had to be met: (1) abdominal obesity with waist circumference ≥ 90 cm for men and ≥ 80 cm for women [28], (2) systolic blood pressure ≥ 130 mmHg, diastolic blood pressure ≥ 85 mmHg, or medication for hypertension, (3) TG ≥ 150 mg/dL, (4) HDL-C < 40 mg/dL for men and < 50 mg/dL for women, and (5) fasting blood glucose ≥ 110 mg/dL or medication for hyperglycemia.

Genotyping of SNPs

Since TG-elevating and HDL-C-lowering SNPs in APOA5 and ZNF259 (rs662799, rs651821, rs2072560, rs2266788, rs6589566, and rs964184) are strongly correlated with each other in Asian HapMap population (r2 > 0.8), we selected three SNPs across the two genes, viz., rs662799 (-1131 of APOA5; one of representative SNPs for APOA5*2 haplotype), rs651821 (-3 of APOA5; a coding SNP in the 5′ untranslated region and one of the representative SNPs for APOA5*2 haplotype), and rs6589566 (a SNP in ZNF259 intron 12). For example, rs6589566 was strongly linked both with rs964184 (D′ = 0.98, r2 = 0.97 in Asian HapMap population) and with rs2266788 (D′ = 1.0, r2 = 0.98 in Asian HapMap population). The rs3135506 was not selected as a representative SNP for APOA5*3 haplotype, as it is monomorphic in Asian populations.

The genotypes of the three SNPs were determined using two genotyping methods, by using the Affymetrix Genome-Wide Human SNP array 5.0 (Affymetrix, Santa Clara, CA) and unlabeled oligonucleotide probes (UOPs) on the given polymorphic nucleotides [29]. The Affymetrix SNP array 5.0 was used for rs6589566 in 882 subjects [30], and the UOP genotyping method was used for the three SNPs (in the rest of the 2,067 subjects for rs6589566). The details of the genotyping method using UOPs (oligonucleotide sequences: Additional file 2) have been described in a previous report [1].

Statistical analysis

The chi-squared test was used to determine whether the SNPs deviated from Hardy–Weinberg equilibrium in studied population. LD (Lewontin’s D′ = D/|D max | and r2) was obtained from Haploview program, version 4.2 (Daly Lab at the Broad Institute, Cambridge, MA) [31]. Construction of haplotypes of APOA5 SNPs was implemented with Phase, version 2.1, which is a Bayesian algorithm-based program [32, 33]. Multiple linear regression analyses were performed for a ratio of log[TG] to HDL-C, after adjusting for age, sex, physical activity (three categories), daily food intake (three categories), clinical history of hypertension, dyslipidemia, and/or diabetes, and medications for the three diseases. Multiple logistic regression analyses with the same adjustments, except medications, used in the multiple linear regressions were used for determining the effects of SNPs on MS risk. The p-values of < 0.05 were considered significant. All statistical analyses were performed using R, version 2.15.2 (http://www.r-project.org/).