Hypervariable intronic region in NCX1 is enriched in short insertion-deletion polymorphisms and showed association with cardiovascular traits

We screened NCX1 (also known as SLC8A1) locus for the presence of Conserved Non-coding Regions (CNRs) using the web-based VISTA software [15] with the proposed default parameters (cutoff criteria: 100 bp sliding window; sequence identity 70%; comparison with rat and mouse). The analyzed NCX1 locus (in total 420,181 bp) spanned from 10 kb downstream to 10 kb upstream of the gene [2p22.1; coordinates 40 241 046-40 661 226 according to NCBI Build 35, hg17; GenBank:6546]. All VISTA regions that had any overlap with annotated genes track at UCSC Genome Browser [16] were excluded as potential coding regions. For polymorphism discovery we selected 29 non-coding regions based on the following criteria: the (i) length 50-300 bp; (ii) location >200 bp from the nearest exons, and (iii) sequence identity >70% between human and both rodents (See additional file 1).

The selected NCX1 conserved non-coding segments were targeted to polymorphism screening by Denaturing High-Performance Liquid Chromatography method (DHPLC; Wave Technologies Inc. USA). During the design and experimental screening process of DHPLC products the recommendations of the manufacturer were followed. Details of DHPLC assay and running conditions with appropriate PCR primers and fragment characteristics are given in additional file 2. Among the total 29 CNR-s selected, 16 intronic regions entered the DHPLC screening (See additional file 1) phase. 13 regions were excluded before the laboratory experiments due to failure in DHPLC primer design, inappropriate length of the PCR and CNR fragment (too long >700 bp or short <50 bp), or more than two different Tm melting points for the region of interest. The average length of the screened CNR segments was 163 bp (SD: 64 bp, range: 70-287 bp) and PCR fragments was 334 bp (SD: 87 bp, range 170-489 bp). Polymorphism screening was performed with 15 different DNA pools, each consisting of DNAs of three patients with cardiovascular disease originating from two Eastern European sample sets (n = 22 from HYPEST and n = 24 from CADCZ study; detailed description is given below). Individual DNAs in the pools exhibiting evidence for the presence of a polymorphism were sequenced at least twice on both forward and reverse orientations.

Genotyping of the identified 14 bp indel (rs11274804, NCBI dbSNP database) in NCX1 intron 2 was performed by standard Denaturing Gradient Gel Electrophoresis (DGGE) (Ingeny, Goes, Netherlands). The manufacturer's recommendations were followed in the design of the DGGE assay and in choosing the conditions for the experimental setup. Detailed information of the assay is given in additional file 2. To initially validate the reliability of the DGGE assay, all DNA samples with alternative genotypes previously detected by DHPLC were re-genotyped at the DGGE platform. To further assure DGGE gel typing system's quality in each assay, double positive (product containing 14 bp indel) and negative controls were used. All ambiguous genotypes in DGGE analysis were re-genotyped twice and/or sequenced on both DNA strands by an ABI 377 Prism automated DNA sequencer using ReproGel 377 gels (Amersham Biosciences Inc., USA). The sequences of all novel variants were verified by resequencing twice on both forward and reverse orientation.

Two Eastern-European sample collection, HYPEST and CADCZ were used to conduct association analysis of rs11274804 with cardiovascular traits and serum biomarkers (Table 1). The HYPEST study has been approved by the Ethics Committee on Human Research of University of Tartu (no. 122/13, 22.12.2003; 137/20, 25.04.2005). CADCZ study was approved by the Ethics Committee of Charles University--1st Faculty of Medicine (December 1996) and is published elsewhere [17]. The studies were carried out in compliance with the Helsinki Declaration and all the participants have given their written informed consent. These sample collections have been recruited to target the genetic-epidemiological component of cardiovascular disease in Estonian and Czech populations, respectively. HYPEST subjects were recruited across Estonia during 2004-2007 (1823 individuals, age range 18-85 years) with the aim to evaluate risk factors for essential hypertension and related cardiovascular disease. Details of the recruitment are given in additional file 2. CADCZ subjects (n = 893; n = 296 coronary artery disease patients, n = 597 controls) were recruited by the Cardiology Department of the 2^nd Clinic of Internal Medicine, Faculty Hospital Královské Vinohrady in Prague Czech Republic and Czech heath clinics in years 1998-2000 [17]. In order to exclude obesity and age-related risks, the current study included individuals with BMI <35 kg/m² and age <65 years. From HYPEST individuals 470 hypertensive patients and 652 normotensive controls, and from CADCZ samples 257 CAD patients and 413 controls, were analyzed. The control group for both studies consisted of matched healthy individuals with no personal history of CAD, essential hypertension, MI, peripheral arterial disease, or stroke. As no population differentiation was detected among HYPEST and CADCZ study subjects previously [18], the controls of the two studies were pooled in order to address the association of rs11274804 with cardiovascular traits in general Eastern-European population.

For all subjects in the HYPEST and CADCZ studies resting blood pressure (BP) and heart rate were measured by trained clinicians during recruitment. In both studies BP measurements per subject were obtained after a rest in a sitting position using a standard mercury column sphygmomanometer and size-adjusted cuffs. All HYPEST individuals possessed a documented history of multiple SBP and DBP readings (on average 4.31 readings per individual, range 2-29) during mean 3.17 years (range 1-17 years). To compensate for the variability in heaviness of data per study subjects, we used for the analysis the median across the longitudinal BP readings as well as the median of the subject's age during the readings. Definition of essential hypertension among HYPEST subjects is given in additional file 2. For CADCZ subjects three blood pressure measurements were documented and the median value was recorded.

Coronary artery disease (CAD) in CADCZ study was diagnosed according to WHO criteria, and one or more large stenosis of a major coronary vessel was confirmed by coronarography in all patients details of which have been published elsewhere [17]. Carotid wall intima media thickness (IMT) and the presence of carotid plaque, recorded in the CADCZ subjects were determined by ultrasonography using linear exploring coil 7,5 MHz on the distant interior wall about 1-2 cm distally from the bifurcation. The measurement was performed on the right and the left carotid 5-10 times on each side. Diagnosis of metabolic syndrome was defined based on the criteria appointed by International Diabetes Federation [19].

Altered serum lipid profile is considered as a cardiovascular risk factor - a condition that is associated with an increased risk of developing CVD affecting the heart and blood vasculature. In the current study, lipid measurements (total-cholesterol, HDL-cholestrerol, LDL-cholesterol and triglycerides, TG) were determined from fasting venous blood samples in the HYPEST and CADCZ subjects. For HYPEST total-cholesterol, HDL-cholesterol, LDL-cholesterol, and triglycerides in the serum were measured by standardized assays (Cobas Integra 800^® analytical platform, Roche Diagnostics, Inc., USA) at the United Laboratories, Tartu University Hospital [20] or at the Diagnostics Division Laboratory, the North Estonia Medical Centre [21]. For CADCZ, serum lipids were measured by standard techniques in Institute of Clinical Chemistry of Vinohrady Faculty Hospital.

For all identified polymorphisms, the deviation from Hardy-Weinberg equilibrium and differences in allele frequencies between populations were tested using an exact test implemented in Genepop web Version 3.4 [22]. Four rare population-specific polymorphisms showed differences in allele frequencies between the studied individuals from HYPEST (n = 22) and CADCZ (n = 24) (Fisher's exact test, p < 0.05; data not shown). There was no significant difference between the HYPEST (n = 1122) and CADCZ (n = 670) study groups (Fisher's exact test, p > 0.05; data not shown) for the distribution of the 14 bp indel polymorphism.

The significance of the associations between the NCX1 14 bp indel and cardiovascular traits was tested, and odds ratios/effect sizes and confidence intervals were obtained using linear (quantitative traits) and logistic (case-control analysis) regression analyses implemented in the PLINK software [23]. In all study stages the association analyses were performed under additive and dominant genetic models. Additive genetic models assume a trend per copy of the minor allele to contribute to the trait or disease susceptibility on genotype categories, whereas dominant genetic models assume that heterozygotes have the same increased risk as minor homozygous genotypes. Recessive genetic model was not applied, as it requires a large sample size to reach a reasonable statistical power. Meta-analysis was performed using inverse variance method with fixed effect model of both sample sets. Association tests were performed with age, sex and BMI as co-variates. P-values <0.05 were considered statistically significant. Two additional corrections were used in the quantitative parameter analysis: in the serum lipid biomarker analysis, a correction according to Jun Wu was implemented to all of the individuals obtaining lipid-lowering medications [24], and in the blood pressure (SBP and DBP) association test, a correction described by Martin Tobin was used for all subjects receiving antihypertensive treatment [25]. For the multiple comparisons (11 traits tested) a simple Bonferroni correction was used (p < 0.05/11 = 0.0045).

Multiple sequence alignment of the human, common chimpanzee and rhesus macaque DNA sequences of the orthologous regions adjacent to 14 bp indel within NCX1 intron 2 was performed with web-based analysis tool ClustalW2 [26]. LD structure (based on the HapMap variation data) for the analyzed NCX1 genomic region was performed with the Haploview package [27] (See in additional file 3).

We analyzed human cardiovascular candidate gene, NCX1 (Na(+)-Ca(2+) exchanger, 2p22.1), for the conserved non-coding regions using VISTA genome browser [15] with default parameters. In total, we identified 365 conserved non-coding regions between human and mouse and rat (May 2004, NCBI Build 33). These conserved non-coding regions (CNR) covered 15.1% of the analyzed genomic region (gene ± 10 kb). Based on the length (50-300 bp), location (>200 bp from the nearest exon) and sequence conservation (>70% between human and both rodents), 29 regions were selected for further polymorphism screening among 46 East-European cardiovascular disease patients (HYPEST n = 22; CADCZ n = 24). Due to the limitations of the technology and/or complex structure of the DNA sequence, 13 of the selected NCX1 regions did not qualify for the DHPLC screening. The remaining 16 regions were subjected to polymorphism discovery among cardiovascular phenotype patients. Detailed descriptions of the selected regions as well as inclusion/exclusion criteria for the analysis are given in additional file 1. In total, ten genetic variants (including three novel variants) were identified within the 16 analyzed regions (Table 2). Most of the variations were located within the second intron of the gene (first intron relative to ATG initiation codon), including six SNPs, and one 14 bp indel (rs11274804, NCBI dbSNP database). Two SNPs were detected in the NCX1 intron 10 and one SNP in intron 4. Among the screened HYPEST samples (n = 22) three common (minor allele frequency, MAF>10%) and four rare (MAF<10%) variants were identified. Five polymorphisms detected in the studied CADCZ patients (n = 24) were uncommon (MAF<10%) and four were common (MAF>10%). Four variants were specific to one of the studied sample sets (either HYPEST or CADCZ).

Nine of the identified variants in screened NCX1 non-coding regions were either singletons or targeted already by commercial platforms and thus included in large number of studies (Table 2). The current study focused on the 14 bp indel (rs11274804), which was represented with substantial frequency (6.82% in HYPEST; 14.58% in CADCZ) in both study populations. In addition, the location of rs11274804 indel variant in the second intron of NCX1 (the first intron in the coding region) raised the hypothesis about its potential effect on the gene transcription as several gene expression regulatory elements have been mapped within the first introns [28‐30]. Thus, this polymorphism as well as its flanking region was selected for further characterization.

The region flanking the 14 bp indel (348 bp) in NCX1 intron 2 was targeted for the larger-scale genotyping using Denaturing Gradient Gel Electrophoresis (DGGE). Our study samples, HYPEST (n = 1122) and CADCZ (n = 670), displayed nine different genotype variants of this intronic segment (representing seven novel alternative alleles), detected by DGGE and confirmed by sequencing (Figure 1). The analyzed region appeared to be highly polymorphic in both sample collections. In addition to 14 bp indel, a SNP (C/G), a duplication of 40 bp segment, and four alternative deletions (10 bp, 43 bp and 5 bp) were localized within the 348 bp region (Figure 1, Table 3). The breakpoints of several detected variants co-localized with the 14 bp indel (Figure 2b). The allele frequencies of the detected variants in the full genotyped sample (n = 1792) varied from singletons (10 bp deletion; 40 bp duplication) to common polymorphisms with allele frequencies up to 8.51% (14 bp indel). The 43 bp deletion was enriched in HYPEST samples (20 subjects in HYPEST versus 1 in CADCZ). One HYPEST subject appeared to be a compound heterozygote for 14 bp indel/43 bp deletion. In total, 18.21% of the genotyped HYPEST subjects and 20.49% of CADCZ subjects were carrying alternative variants of the studied NCX1 intron 2 segment (Table 3). Active genome dynamics of the analyzed region is supported by its location between two LD-blocks within the NCX1 gene (See in additional file 3).

In order to determine the ancestral primate variant of the studied 348 bp region, the consensus sequences of human, common chimpanzee and rhesus macaque were aligned. In contrast to the expectations, multiple sequence alignment (ClustalW2) revealed that the ancestral primate variant is actually the minor human allele carrying the 14 bp sequence motif. The major human variant (among East-European subjects) has evolved through a 14 bp human-specific deletion when compared to sister-species chimpanzee and rhesus macaque (Figure 2b).

Other identified short insertion-deletion variants within the studied region have occurred in human lineage on the chromosomal variant carrying the human-specific 14 bp deletion (Figure 2b).

Association of NCX1 intronic 14 bp indel (rs11274804) with cardiovascular disease was studied in two Eastern European sample sets: essential hypertension (HYPEST, Estonia; n = 1122; cases n = 470/controls = 652) and coronary artery disease, CAD (CADCZ, Czech; n = 670; cases n = 257/controls = 413). Associations were assessed using logistic regression under additive and dominant effect models (age, sex, and BMI as covariates; Table 4). NCX1 intronic 14 bp indel region revealed strong association with the diagnosis of CAD (P = 0.0016, OR = 2.02; P = 0.0018, OR = 2.07; additive and dominant models, respectively). As there are various clinical conditions that comprise CAD, additional case-control analysis was performed using patients diagnosed with CAD as well as metabolic syndrome (n = 88). Despite a three fold reduction in sample size compared to the full CADCZ patient group, the analysis of CAD patients with metabolic syndrome revealed highly significant association with increased effect size (P = 0.0014, OR = 2.34; P = 0.0016, OR = 2.41; additive and dominant models, respectively). These associations remained significant after correction for multiple testing (Bonferroni significance level α = 0.05/11 = 0.0045).

NCX1 intronic 14 bp indel was not associated with the diagnosis of essential hypertension in HYPEST sample collection (P > 0.1).

Associations of the 14 bp indel with quantitative cardiovascular parameters [systolic (SBP) and diastolic (DBP) blood pressure, heart rate, Intima-Media Thickness (IMT)], and serum lipid biomarkers [total cholesterol, HDL, LDL, triglycerides (TG)] were evaluated by linear regression under additive and dominant models (Table 5, Table 6). A marginal negative correlation was detected with heart rate (P = 0.04, beta = -1.6; additive model) and LDL (P = 0.04, beta = -0.26; dominant model) among healthy Eastern-European subjects (Table 5, Table 6). Association of the 14 bp indel with serum triglyceride levels reached marginal significance in the CADCZ sample (P = 0.04, beta = 0.25; both models) and a non-significant trend for association in the joint meta-analysis with HYPEST data (P = 0.08, beta = 0.18; P = 0.07, beta = 0.19; additive and dominant models, respectively). No significant association was detected with other studied cardiovascular and serum lipid parameters in separate sample sets and in meta-analysis.