Figures
Abstract
In our attempt to comprehensively understand the nature of association of variants at 11q23.3 apolipoprotein gene cluster region, we genotyped a prioritized set of 96 informative SNPs using Fluidigm customized SNP genotyping platform in a sample of 508 coronary artery disease (CAD) cases and 516 controls. We found 12 SNPs as significantly associated with CAD at P <0.05, albeit only four (rs2849165, rs17440396, rs6589566 and rs633389) of these remained significant after Benjamin Hochberg correction. Of the four, while rs6589566 confers risk to CAD, the other three SNPs reduce risk for the disease. Interaction of variants that belong to regulatory genes BUD13 and ZPR1 with APOA5-APOA4 intergenic variants is also observed to significantly increase the risk towards CAD. Further, ROC analysis of the risk scores of the 12 significant SNPs suggests that our study has substantial power to confer these genetic variants as predictors of risk for CAD, as illustrated by AUC (0.763; 95% CI: 0.729–0.798, p = <0.0001). On the other hand, the protective SNPs of CAD are associated with elevated Low Density Lipoprotein Cholesterol and Total Cholesterol levels, hence with dyslipidemia, in our sample of controls, which may suggest distinct effects of the variants at 11q23.3 chromosomal region towards CAD and dyslipidemia. It may be necessary to replicate these findings in the independent and ethnically heterogeneous Indian samples in order to establish this as an Indian pattern. However, only functional analysis of the significant variants identified in our study can provide more precise understanding of the mechanisms involved in the contrasting nature of their effects in manifesting dyslipidemia and CAD.
Citation: Pranav Chand R, Kumar AS, Anuj K, Vishnupriya S, Mohan Reddy B (2016) Distinct Patterns of Association of Variants at 11q23.3 Chromosomal Region with Coronary Artery Disease and Dyslipidemia in the Population of Andhra Pradesh, India. PLoS ONE 11(6): e0153720. https://doi.org/10.1371/journal.pone.0153720
Editor: Robert M. Lafrenie, Sudbury Regional Hospital, CANADA
Received: January 11, 2016; Accepted: April 1, 2016; Published: June 3, 2016
Copyright: © 2016 Pranav Chand et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files. For individual data from the CAD case control study, the authors may be contacted at bmrisi@gmail.com.
Funding: This work was supported by in-house funding from the Indian Statistical Institute. No specific grant number, etc., is applicable.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Coronary artery disease (CAD) is most predominant and ranks as number one in causing deaths due to cardiovascular diseases (CVDs) in India. Dyslipidemia, diabetes, hypertension, smoking and obesity or overweight were identified as traditional risk factors of CAD across the ethnic groups [1]. The disease occurs due to the process of atherosclerosis, a progressive damage in the blood vessels supplying blood to the heart muscles. The primary event of atherosclerosis is the endothelial injury or dysfunction, which is triggered by abnormal lipoprotein metabolism with subsequent dyslipidemia [2]. It is apparent from the candidate gene association studies that relatively greater number of lipoprotein metabolism related genes is observed to be more consistently associated with CAD as compared to the candidate genes related to other metabolisms [3]. Among the genes that regulate this metabolism, apolipoprotein genes that code for cofactors of several enzymes of cholesterol transport system are the key regulators. These genes are primarily found as clusters on chromosome 11 and 19. The candidate gene [4–7] and GWAS studies [8,9] revealed 11q23.3 Apolipoprotein gene cluster region, in particular, to be more often associated with lipid traits. Exclusive attempts were also made to comprehensively understand the role of genetic variants in this lipid influencing region among the Caucasians [10] and north Indian Punjabis [8] and a couple of polymorphisms were found to be associated with high density lipoprotein cholesterol (HDL-C) and plasma triglyceride (TG) concentrations.
Indians are known to have a unique pattern of dyslipidemia, usually characterized by low levels of low density lipoprotein cholesterol (LDL-C), elevated triglycerides and low HDL with predominantly atherogenic and small-dense LDLs [11, 12]. This characteristic feature is referred to as ‘atherogenic dyslipidemia’ [13, 14]. Although a few of the conventional polymorphisms within the APOAI-CIII-AIV-AV gene cluster region were found to be associated with lipid traits [15, 16], a large number of SNPs in this region remained unexplored for their association with CAD as well as with the lipid traits among Indians. On the other hand, with its reported incidence of 67.6% among the CAD cases, dyslipidemia appears to be the primary cause of CAD in southern Indians [17]. Given the characteristic dyslipidemic feature of Indian populations in general and particularly the South Indians, it is imperative to explore the possible association of polymorphisms at 11q23.3 apolipoprotein gene cluster region with CAD among them. Spanning ~200KB, this chromosomal region contains three regulatory protein coding genes—BUD13, ZPR1 and SIK3—and four apolipoprotein coding genes—APOA1, APOC3, APOA4 and APOA5. In the present study, we made an attempt to explore the association pattern of 96 prioritized SNPs of the 11q23.3 region with CAD in the population of Hyderabad, India.
Materials and Methods
Ethics statement
The study protocol was approved by the Indian Statistical Institute Review Committee for Protection of Research Risks to Humans. Written informed consent for all the participants is obtained as per the guidelines.
The study design and population
A total of 1024 individuals including 508 CAD cases and 516 controls, broadly representing the population of Andhra Pradesh, participated in our case-control study. Patients with characteristic symptoms of stable/unstable angina pectoris along with variable degrees (generally >40%) of stenosis in at least one of the major coronary arteries as determined through angiogram were included in the study. Cases with monogenic diseases, valvular heart disease, cardiomyopathy, renal disease, acute and chronic viral or bacterial infections, asthama, tumours or connective tissue diseases and other vascular diseases were excluded from the study. All the cases were evaluated by interventional cardiologists at the CARE Hospitals, Hyderabad, for the above mentioned criteria. Control samples were recruited by conducting free health camps in and around Hyderabad and broadly representing subjects aged above 45 years and with similar ethnic backgrounds as that of the cases. The individuals with characteristic features of any of the above mentioned disease conditions were not included as part of the controls. However, certain proportions of both the case and control subjects were found with T2DM, dyslipidemia and hypertension.
The population of Hyderabad is a conglomeration of people from different parts of the undivided state of Andhra Pradesh and the mother tongue of most of its population is Telugu, one of the four Dravidian languages. It would be also pertinent to note that despite the subdivision of Telugu population into a number of traditionally endogamous castes and sub castes, Reddy et al. [18] observed genetic differentiation among the populations of Andhra Pradesh to be very low and insignificant; the Markov chain Monte Carlo analysis of population structure, which implements model based clustering method for grouping individuals into populations [19, 20], did not reveal any unique population clusters, suggesting high degree of genetic homogeneity.
Data and sample collection
Data pertaining to present age, sex, age at diagnosis for cases, height, weight, waist circumference, hip circumference and other background information such as history and current status of smoking, alcoholism and food habits were obtained through a detailed questionnaire. The data pertaining to the current status of the subjects on diabetes, dyslipidemia and hypertension were drawn from hospital records for the cases and through personal interviews for controls. About 5-6ml of fasting blood sample was collected peripherally by certified medical lab technicians. Clinical investigations were done for lipid profile and blood sugar for all the samples at Tapadia diagnostic centre, Hyderabad, using AutoAnalyzer.
DNA Isolation, SNP selection and Genotyping
DNAs were isolated from all the blood samples using phenol chloroform method [21] and quantified with the help of Thermo Scientific Varioskan™ Flash Multimode Reader using Quant-iT™ PicoGreen® dsDNA Assay Kit.
Apolipoproteins act as cofactors and inhibitors of several enzymes of lipoprotein metabolism and also regulate the cholesterol transport system by being components of HDL cholesterol, LDL cholesterol and very low density lipoproteins (VLDL). Mutations in the genes coding for these proteins lead to abnormal lipoprotein metabolism resulting in increased plasma lipid levels which inturn cause dyslipidemia and/or CAD. In addition to the SNPs at APOAI-CIII-AIV-AV genes clustered at 11q23.3 chromosomal region, SNPs located at BUD13 and ZPR1 genes were also identified through GWAS [22] as associated with abnormal lipid traits, which have been replicated among Europeans, Chinese and Asian Indians. Except for a couple of conventional polymorphisms related to APOAI-CIII-AIV-AV genes there have been hitherto no genetic studies on the significant GWAS identified SNPs among the south Indian populations. Therefore, to comprehensively genotype the variants at 11q23.3 chromosomal region, we gathered information on SNPs pertaining to this region from earlier candidate gene and sequencing studies and from databases particularly EBI-NHGRI GWAS database, HAPMAP and dbSNP. Given the key role of BUD13 in splicing mechanism and ZPR1 as essential protein for normal cell proliferation and signal transduction, we also included SNPs related to these regulatory protein coding genes. A total of 130 SNPs, mostly studied through candidate gene and GWAS approaches, were subjected to Fluidigm D3 Assay design software [23] and a panel of 96 SNPs with high efficiency for genotyping was chosen (S1 Table). Most of these SNPs were characterized as intronic/intergenic and none were studied for their functional role in the manifestation of dyslipidemia or coronary artery disease, except for rs964184 which is a 500BP downstream variant of ZPR1 gene.
Genotyping was performed using fluidigm nanofluidic SNP genotyping system. Eleven 96.96 IFC chips were utilized for genotyping wherein the selected 96 SNPs were analyzed against 96 samples in each chip. These chips were thermal cycled and the end-point fluorescent values were measured on BiomarkTM system. Final sample wise genotype calls were obtained using Fluidigm SNP Genotyping Analysis software. A subset of 240 samples was genotyped prior to genotyping of the total 1024 samples. The observed call concordance was 100%. Prior to genetic association analysis, data quality control was achieved by limiting the sample wise call rate to ≥90%. This resulted in genotype call rate of 99% for 386 cases and 462 controls, which are considered for further analysis. Further, of the 96 SNPs analyzed, only 75 SNPs were qualified for final analysis, after excluding SNPs that either showed minor allele frequency < 1% or deviated from hardy Weinberg equilibrium (P<0.001). However, we furnished the allele and genotypic data for the entire set of SNPs and for all the cases and controls in S2 Table.
Statistical Methods
The descriptive statistical analysis of the background data on quantitative variables was done using MINITAB (version 17). The significance of the mean difference between two groups was obtained using students T-test. Genotyping quality check and association analysis of alleles were done using PLINK [24]. Genotype-phenotype association analysis assuming different genetic models-dominant, co-dominant, recessive, over dominant and log-additive- and logistic regression analysis with covariates were performed using ‘SNPassoc’ package of R PROGRAM. Analysis of the cumulative effect of risk alleles was done with the help of Microsoft EXCEL. The power of our study in discriminating CAD (Receiver Operating Characteristic curve) based on the cumulative risk scores was estimated using SPSS (Version 21, IBM SPSS). Linkage disequilibrium and haplotype association analyses were done using HAPLOVIEW (Version 4.2). RevMan version 5.3 was used for Meta-analysis.
Results
The means and SDs of age, age at onset of the disease, body mass index (BMI), waist hip circumference ratio (WHR), systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting blood sugar (FBS) and lipid profile are presented separately for CAD patients and controls in Table 1. While the mean values of BMI, DBP, TC, HDLC, LDLC are significantly lower (P = 0.015, 0.028, 0.0001, 0.0001, 0.0001, respectively), the mean systolic blood pressure (P<0.007) and fasting blood sugar (P<0.0001) are significantly higher among the CAD patients. The elevated triglyceride levels with optimum TC and LDLC of the controls represent characteristic atherogenic dyslipidemic feature of the general south Indian population. However, we found average LDLC and HDLC levels to be lower than the NCEP ATPIII defined normal values for the CAD patients. This trend is similar to that observed in a study of large cohort of US CAD subjects, which analyzed admission levels of LDLC and HDLC while adjusting for confounding factors, particularly the lipid lowering medication history before admission [25]. Unfortunately, we could not obtain data on medication, which would have enabled us to explain our findings more specifically.
The allelic association with CAD
The association analysis of 75 SNPs which are in HWE suggests 12 SNPs as significantly associated with CAD at P <0.05 albeit only four of these remained significant after Benjamin Hochberg correction for multiple correction. The nearby genes to which the associated SNPs belong, allelic frequencies, chi-square values, and odds ratios are presented in Table 2. The odds ratios from the logistic regression analysis suggest that the minor alleles of the following SNPs- rs1263171-A, rs664059-T, rs10488699-A, rs6589566-G and rs2075294-Tare associated with increased risk, whereas rs5072-T, rs2849165-A, rs633389-T, rs1263163-A, rs7396835-T, rs17440396-A and rs2187126-Gare associated with decreased risk for CAD. While five of the above 12 SNPs (rs2849165, rs633389, rs1263163, rs1263171, rs7396835) belong to intergenic region of APOA5-APOA4 genes, one SNP (rs5072) belongs to intronic region of APOA1 gene. Of the remaining, four SNPs (rs17440396, rs664059, rs10488699, rs2187126) belong to intronic regions of BUD13 and two (rs6589566 and rs2075294) belong to ZPR1 regulatory gene. However, out of the most prominent four SNPs (rs2849165, rs17440396, rs6589566 and rs633389) that remained significant even after Benjamin Hochberg correction, only one (rs6589566) appears to confer significant risk to CAD, while the others seem to reduce risk for the disease.
Since the 11q23.3 chromosomal region plays a vital role in regulating lipid traits, we analyzed the association of these SNPs adjusting for covariates age, sex and lipid traits, viz. total cholesterol (TC), LDL cholesterol, HDL cholesterol, triglycerides and VLDLs. The results suggest that only 6 out of the 12 SNPs show significant association after adjusting for covariates, suggesting independent effects; while rs664059, rs6589566 and rs2075294 remained significantly associated with increased risk for CAD, rs2849165, rs7396835 and rs17440396 showed decreased risk for CAD (Table 2). In addition to these an APOA5-APOA4 intergenic SNP rs633867-T, which was not associated with CAD earlier, turned out to be significant with increased risk after adjusting for covariates (OR = 1.96(1.17–3.28); p value = 0.009). We further explored possible association of these 12 CAD related variants with the quantitative lipid traits viz., TC, LDLC, HDLC, TG and VLDL, as the main phenotypes. A significant association of rs17440396, rs2187126, rs633389 and rs1262163 is apparent with TC and LDLC implying their risk for dyslipidemia in contrast to rs6589566 that showed negative regression suggesting decreasing risk for dyslipidemia (S3 Table). This trend is consistent with the genotypic distribution of mean levels of LDLC and TC (S1 Fig). However, assuming that the dyslipidemia is the primary trigger for CAD, these trends are somewhat contradictory to the pattern of association of these SNPs with CAD. This may imply that the variants at 11q23.3 chromosomal region exhibit distinct effects in manifesting CAD and/or dyslipidemia.
Genotypic association with CAD
In order to understand the genotype-phenotype relationship, we performed logistic regression of CAD on the genotypes under different genetic models and the results are provided in Table 3. The best genetic model for the genotypic effect was chosen based on the lowest AIC (akaike information criteria) value. We observed three variants that belong to APOA5-APOA4 intergenic region, rs633389 (OR = 0.49; P value = 1.72x10-6), rs1263163 (OR = 0.48; P value = 6.8x10-8) and rs2849165 (OR = = 0.47; P value = 1.6x10-12) to be significantly associated with CAD under log additive model. While the intronic variant rs17440396 of BUD13 showed highly significant protective effect under dominant model (AG-AA vs GG; OR = 0.08, P value = 1.4x10-37), rs2187126 displayed similar association (AG vs AA-GG; OR = 0.63, P value = 0.007) under over-dominant model. On the contrary, rs6589566 that belongs to ZPR1 intronic region is found to increase risk towards CAD (AG-GG vs AA; OR = 2.38, P value = 6.04x10-10), under dominant model. This analysis suggests possible protective role of BUD13 regulatory gene against CAD in our population while the other regulatory gene, ZPR1, confers susceptibility to CAD. Except for the variant rs2187126, the patterns of association remained similar after adjusting for the covariates (Table 3).
Linkage Disequilibrium and Haplotype analysis
Pair wise LD analysis of all the SNPs with their D’ values (Fig 1) revealed 55 SNP pairs with r2> 0.8, 63 pairs with r2 between 0.5 and 0.8 and another 68 SNP pairs with r2 < 0.5. Overall, a disrupted LD pattern is seen in this population with 46 tag SNPs identified at r2 ≥ 0.8. Using Gabriel et al [26] haplotype block definition criteria, 11 haplotype blocks were identified in this chromosomal region. Table 4 provides the information on SNPs and genomic region covered in each of the haplotype blocks. Table 5 provides frequencies of the associated haplotypes with the respective blocks and constituent SNPs along with the results of logistic regression analysis.
Pair wise LD among 63 SNPs is indicated by diamonds shaded in white–pink–red, the strength of LD reflected in the intensity of colour, from D’ = 0 in white to D’ = 1 in red. The 11 blocks represent the presence of strong LD.
We found haplotypes AT (OR = 1.28; p value = 0.01) and CGTTAT (OR = 1.57; P value = 0.03) carrying risk alleles of the SNPs, rs664059 and rs2075294,from the H3 and H5 haplotype blocks, respectively, as associated with increased risk for CAD. The TC haplotype with SNPs rs633389 and rs633867 that belong to H6 block is found to show decreased risk for CAD (OR = 0.04; P value = 6.97x10-10). These patterns of haplotypic association are similar after adjusting for age and sex as covariates. However, after adjusting for age and sex we observed TT haplotype of H6 block to be associated with increased risk for CAD (OR = 1.49; P value = 0.047). This is similar to the allelic association finding from logistic regression analysis, where SNP rs633867 exhibited significantly increased risk for CAD after adjusting for covariates. On the other hand, GTCT haplotype of block H4, which was significantly associated (OR = 1.36; P value 0.049) with increased risk for the disease turns out to be non-significant (P value = 0.1) after adjusting for the covariates. However, no SNP that belongs to this block is found associated at allelic or genotypic level with the disease.
SNP-SNP interactions
Given that the SNPs under study are within 200kb of 11q23.3 chromosomal region, we attempted to understand the interaction effects among these SNPs on the CAD by standard parametric approach implemented through pair wise logistic regression analysis of the associated 12 SNPs using PLINK. From the pairs of SNPs that showed significant interactions along with the associated P values and odds ratios (Table 6), the following three categories of interactions can be inferred:
- Interactions between intronic SNPs of BUD13 regulatory gene i.e., rs17440396, rs10488699, rs2187126 with intergenic SNPs rs633389, rs1263163, rs1263171 of APOA5-APOA4 genes.
- Intronic SNP, rs6589566 of ZPR1 gene and APOA5-APOA4 Intergenic SNP rs1263163.
- Interactions within APOA5-APOA4 intergenic region involving SNP pairs rs1263163-rs123171 and 1263163-rs2849165
Among the identified SNP-SNP interactions, rs2187126-rs1263171 (OR = 2.681, P Value = 1.35 x10-05) and rs6589566-rs1263163 (OR = 3.548, P value = 2.91 x10-07) that belong to categories one and two above show increased risk for CAD. Both the regulatory genes BUD13 and ZPR1 are involved in disease manifestation through their significant interactions with APOA4-APOA5 intergenic region.
Generalized Multifactor Dimensionality Reduction (GMDR) approach for Interaction analysis:
GMDR is an agnostic and nonparametric approach to determine the gene-gene interactions. It is widely used because of its power to detect interactions in small sample sizes while permitting adjustment for covariates. Here we carried out an exhaustive search over all possible combinations of interactions to an order of six attributes with 10 fold cross validation. However, significant interactions were observed only up to five loci combination, with interaction models above the threshold testing balance accuracy of 0.55 with significant test p-values and high cross validation consistency are selected as the best models. With their consistent occurrence in all interaction models, rs17440396 and rs2849165 emerge as the most prominent variants in the interaction analysis. This implies the presence of significant interaction between the intronic region of BUD13 gene and intergenic region of APOA4-APOA5 genes, which persists even after adjusting for age and sex (Table 7). Additionally, interaction of rs6589566 of ZPR1 regulatory gene appeared in 5 loci model. The results of parametric and non-parametric interaction analysis are concurrent in suggesting that the interaction between intergenic variants of APOA5-APOA4 genes and intronic variants of BUD13, ZPR1 regulatory genes plays a major role in the manifestation of CAD in this population.
Results of cumulative risk score analysis for CAD associated variants
In order to determine the combined risk effect of the 12 significant SNPs, we computed the weighted mean proportion of the risk alleles at these 12 SNPs by taking 2 for two risk alleles, 1 for one risk allele and 0 for no risk alleles with weights as relative log odds ratios of different SNPs. The cumulative risk allele score for each individual is obtained by multiplying with 12 (no of SNPs in analysis). The individuals with the risk scores ranging from 2.7 to 20.9 were grouped into 12 risk categories as shown in Table 8. Given the very low frequency of Individuals within the risk score range of 2–9 and 19–21, these were merged into the categories 1 and 12, respectively. Fig 2 shows the distribution of CAD cases and controls according to the risk category. A clear trend of increased percentage of cases with increasing risk score is observed. With reference to risk category 1, we computed odds ratios for each of the remaining 11 risk categories. An increasing trend of OR values with increasing number of risk alleles is apparent from the plot (Fig 3). However, the odds ratios suggest significant association of CAD only with risk categories ranging from 7–12. Further, to understand the discriminative power of the risk scores, we constructed the ROC (receiver operating curve) plot (Fig 4) for the risk scores and CAD status, which yielded area under curve (AUC) as 0.763 (95% CI: 0.729–0.798, p = <0.0001). Given these highly significant results, the observed AUC probably indicates that this study has substantial power to confer these genetic variants as predictors of risk for CAD.
Discussion
The apo genes clustered at 11q23.3 chromosomal region codes for plasma apolipoproteins APOA1, APOC3, APOA4 and APOA5. APOA1, as a major component of HDL and cofactor for lecthin cholesterol acyl transferase (LCAT) enzyme, is an established biochemical marker of CAD. APOC3 is a component of VLDL and inhibits lipase enzyme. APOA4 was found to activate LCAT enzyme. APOA5, being a component of VLDL, HDL and chylomicrons, was observed to influence plasma triglyceride levels. With these functional implications in lipoprotein metabolism, several polymorphisms at 11q23.3 were associated with elevated plasma lipoprotein and cholesterol levels and in turn also with CAD. So far, Indian studies pertaining to the above genes could validate only the conventional polymorphisms of APOA5-1131T>C (rs662799), −3A>G (rs651821), S19W (rs3135506), G185C (rs2075291) and APOC3-SacI SNP (rs5128) [6, 15, 16, 27]. On the other hand, a GWAS conducted on a sample of glycemic and non glycemic subjects, as part of Sikh Diabetes Study (SDS),Sanghera et al [8] observed an intronic variant of ZPR1 regulatory gene rs12286037 to be associated with triglycerides. This study also validated rs964184, the most consistently replicated ZPR1-BUD13 intergenic SNP across the ethnic groups. Further, exploring the association pattern of 45 SNPs located between the above two SNPs, they found five SNPs- rs7350481, rs180326, rs618923, rs10047459 and rs533556- to be associated with plasma triglycerides [8]. Similarly, Gagandeep et al. [28] observed rs964184 and rs662799 to be associated with triglycerides among the Indian subjects screened as part of the Indian migration study (IMS). However, despite their characteristic atherogenic dyslipidemic feature, the south Indian populations were not hitherto explored for these variants. Our attempt to comprehensively explore the pattern of association of variants at 11q23.3 chromosomal region with CAD, however, yielded a different set of SNPs in the population of Hyderabad. Of the genetic variants that belong to APOA5-APOA4 intergenic region, rs2849165-A, rs1263163-A and rs633389-T show a profound risk reducing effect on CAD in our population, which is perhaps a pioneering observation. That the association of rs633867-Twas significant only when adjusted for lipid traits may imply its independent risk conferring nature to CAD, which was probably masked by the confounding effects of lipid traits. This variant along with another tag SNP rs633389-T forms a significant risk conferring haplotype as well. There is also an interaction among the variants of APOA5-A4 intergenic region on the one hand and these variants with the variants of BUD13 and ZPR1 regulatory genes on the other. Of the four BUD13 SNPs, we observed rs17440396-A to be significant even after adjusting for lipid traits and regulates several associated SNPs through its interaction. Further, a bivariate genome wide approach among Europeans with metabolic syndrome has identified three SNPs (rs11825181, rs11820589, and rs10790162) of this gene to be associated with trait combinations TG-BP, TG-GLUC, and TG-HDLC, MetS and WC-TG, respectively [29]. Subsequently, one of these variants (rs10790162) was also found associated with TC among the Chinese [30]. However, none of these were found to be associated with CAD in our population. With its evolutionarily conserved function of forming pre m-RNA Retention and Splicing (RES) complex [31], BUD13 appears to be one of the key regulating genes of 11q23.3 chromosomal region. Our results of both GMDR and SNP-SNP pair wise interaction analysis are congruent with this in showing significant interaction of rs17440396 BUD13 intronic variant with that of intergenic APOA5-APOA4 and intronic variantsofZPR1 genes. Conversely, an intronic variant rs6589566-G of ZPR1gene is observed to exhibit high risk for CAD in the population of AP. This variant is found to interact independently with APOA5-APOA4 intergenic region as well as in combination with variants of BUD13 gene. It is found to be associated with triglycerides and LDL cholesterol among Europeans [32, 33] and with triglycerides, in conjunction with a downstream 500bp variant rs964184 (D’ = 1, r2 = 1), among the Han Chinese. A Meta analysis of 14 GWAS studies on Europeans identified SNP rs964184 of ZPR1 gene to be significantly associated with HDL cholesterol, coronary artery disease [22] and other lipid traits among Asian Indians and Chinese [8,30]. However, we did not find any association of this SNP with either CAD or any of the lipid traits in the present study. Another variant of this gene, rs2075294-T, is shown to be susceptible to CAD in our population at allelic level albeit not at the genotype level. It was observed that the deficiency of ZPR1 protein causes defects in transcription and cell cycle progression [34] due to improper localization of survival motor neuron (SMN1) protein, gems and cajal bodies. Therefore, it is imperative to understand the precise role of the variants at ZPR1 gene in regulating the apolipoprotein levels. In particular, the risk conferring nature of rs6589566 of this gene towards CAD both at the allelic and genotypic levels as well as by its interaction with rs1263163 of APOA5-APOA4 intergenic region in our population probably warrants functional validation. Overall, we found that the variants associated with CAD in the present study are unique. By far the most significant result could be the high discriminative power of the risk scores as illustrated by AUC (0.763 (95% CI: 0.729–0.798, p = <0.0001), suggesting that our study has substantial power to confer these genetic variants as predictors of risk for CAD.
Distinct patterns of association of SNPs at 11q23.3 region with CAD and Dyslipidemia
Except for the association of APOA1 gene polymorphism rs5072 (756 C>T) at the allelic level, the earlier reported conventional polymorphisms of this chromosomal region are not associated in the present study. Although this is consistent with the findings of a study from Western India [16], the urban study of Chennai on the contrary suggests association of two conventional polymorphisms with CAD [35]. However, the meta analysis on the 3 SNPs (rs5128, rs662799 and rs651821) for which comparative data were available from the above two populations [16, 35] (S4 Table) suggests the overall effect (under the random effect model) to be non-significant for any of the three SNP which implies that these conventional polymorphisms may not have a direct risk conferring effect on CAD in the studied populations. This may suggest that the atherogenic dyslipidemia, with putative role in manifesting CAD in this population, has its specific genetic etiological background. In this context, it may be plausible to surmise if the regulatory genes of 11q23.3 chromosomal region that are found associated with CAD contain a pool of both upregulating and downregulating genetic variants for the lipid traits. Our quantitative trait analysis is concurrent with this conjecture, albeit the risk reducing variants of CAD are risk conferring towards dyslipidemia by their effect in elevating LDLC and TC. However, this might be a result of confounding effects of lower average levels of these lipid traits observed in the case cohort. In order to determine true association of the above variants with lipid traits we tested for their association with dyslipidemia in the control cohort and found that the results are congruent with the susceptible nature of the CAD protective variants to dyslipidemia (S5 Table). Further, the test of genetic association with CAD cases and nondyslipidemic controls yielded identical results compared to what was obtained with pooled sample of controls that included significant proportion of dyslipidemic subjects. The distinct effects of these variants observed in manifesting CAD and dyslipidemia suggest that there is a direct role of the APOA4-APOA5 intergenic variants of this region in regulating lipid traits. Given the pleiotropic nature of BUD13 and ZPR1 regulatory genes, an indirect risk reducing effect of these genes towards CAD could possibly be a result of their action mediated through other metabolic pathways such as inflammation, oxidative stress, thrombosis, blood glucose homeostasis etc. The interactions between the genes might also cause these varying effects of SNPs. For example, rs2187126, which is a protective variant of BUD13 regulatory gene show profound risk towards CAD through its interaction with APOA5-APOA4 intergenic risk variant rs1263171 (OR = 2.68; P value = 1.35 x10-05). Also interaction between rs6589566 and rs1263163 is observed to confer increased risk towards CAD (OR = 3.54; P value = 2.91 x10-07). Functional analysis of these variants would probably provide better insights into the precise roles of these genes in the 11q23.3 chromosomal region.
Conclusion
We found unique variants of 11q23.3 apo gene cluster region as associated with coronary artery disease in this population, which could be because of the differential genetic predisposition of Indians to complex genetic disorders, as compared to other ethnic groups. This was amply demonstrated in a few earlier studies with reference to cardiomyopathies [36], recurrent miscarriages [37], PCOS [38] and T2DM [39, 40, 41]. Most of the variants we observed to be associated with CAD in the present study belong to APOA5-APOA4 intergenic region and intronic regions of BUD13 and ZPR1 genes. Although these genes were reported earlier to have remarkable associations with lipid traits, we validated their significant effect in manifesting CAD. The most significant finding of our study is the high discriminative power of the risk scores based on the 12 significant SNPs, as illustrated by AUC (0.763 (95% CI: 0.729–0.798, p = <0.0001). This may suggest that these genetic variants possess substantial power to be the predictors of risk for CAD.
This study also demonstrated significant interaction between regulatory genes of 11q23.3 region and intergenic APOA5-APOA4 protein coding genes to show distinct effects in manifesting CAD and dyslipidemia. Future studies on genetic association of CAD should include SNPs related to other metabolisms along with the above regulatory genes to get better insights into the mechanisms leading to the distinct effects of the genes observed in the present study. We are in the process of generating high throughput genotype data representing important pathways responsible for atherosclerosis as well as the other genes of lipoprotein metabolism. It may be pertinent to extrapolate this approach of focusing on disease specific domains of genome to other complex diseases which may help identifying important population specific and disease specific genetic markers. More aptly, a GWAS covering the broad ethnic elements in the Indian population with sizeable sample would provide genetic susceptibility profile for CAD at large.
Supporting Information
S1 Fig. Genotype wise mean levels of quantitative lipid traits for significant SNPs of 11q23.3 chromosomal region associated with them.
https://doi.org/10.1371/journal.pone.0153720.s001
(TIF)
S1 Table. Localized gene, nucleotide position and biological relevance of SNPs at 11q23.3 chromosomal region.
https://doi.org/10.1371/journal.pone.0153720.s002
(DOCX)
S2 Table. Allelic and genotype data for CAD cases and controls for all the SNPs under study.
https://doi.org/10.1371/journal.pone.0153720.s003
(XLSX)
S3 Table. Results of linear regression of CAD related variants at 11q23.3 with quantitative lipid traits in the pooled sample of cases and controls.
Footnote: β (Beta) indicates the linear regression coefficient. SE indicates standard.
https://doi.org/10.1371/journal.pone.0153720.s004
(DOCX)
S4 Table. Meta-analysis of SNPs at 11q23.3 chromosomal region analyzed under random effect model.
Footnote: *Conventional ID of the SNP, MAF Minor Allele Frequency.
https://doi.org/10.1371/journal.pone.0153720.s005
(DOCX)
S5 Table. Comparative association analysis of CAD cases and controls in three sets of subjects.
Footnote: *Set 1 indicates CAD cases vs controls, Set 2 indicates CAD cases vs non dyslipidemic controls, Set 3 indicates dyslipidemic controls vs non dyslipidemic controls. # indicates p value not significant.
https://doi.org/10.1371/journal.pone.0153720.s006
(DOCX)
Acknowledgments
We are thankful to the Director, Indian Statistical Institute for logistics support and the participants in the study for their cooperation.
Author Contributions
Conceived and designed the experiments: BMR RPC. Performed the experiments: BMR RPC ASK KA. Analyzed the data: RPC. Contributed reagents/materials/analysis tools: BMR. Wrote the paper: BMR RPC SV. Reading and approving the manuscript, evaluation of patients: ASK KA VS.
References
- 1. Wong ND. (2014) Epidemiological studies of CHD and the evolution of preventive cardiology. Nat Rev Cardiol. 11: 276–289. pmid:24663092
- 2. Ross R, Glomset J, Harker L. (1977) Response to injury and atherogenesis. Am J Pathol. 86: 675–684. pmid:842616
- 3. Pranavchand R, Reddy B M (2013) Current status of understanding of the genetic etiology of coronary heart disease. J Postgrad Med 59:30–41 pmid:23525056
- 4. Yamada Y, Matsuo H, Warita S, Watanabe S, Kato K, Oguri M, et al. (2007) Prediction of genetic risk for dyslipidemia. Genomics. 90: 551–558. pmid:17919884
- 5. Yin RX, Li YY, Lai CQ. (2011) Apolipoprotein A1/C3/A5 haplotypes and serum lipid levels. Lipids Health Dis. BioMed Central Ltd 10: 140.
- 6. Henkhaus RS, Dodani S, Manzardo AM, Butler MG. (2011) APOA1 gene polymorphisms in the South Asian immigrant population in the United States. Indian J Human Genet. 17(3):194–200
- 7. Moreno-Luna R, Perez-Jimenez F, Marin C, Perez-Martinez P, Gomez P, Jimenez-Gomez Y, et al. (2007) Two Independent Apolipoprotein A5 Haplotypes Modulate Postprandial Lipoprotein Metabolism in a Healthy Caucasian Population. J Clin Endocrinol Metab. 92: 2280–2285. pmid:17356049
- 8. Braun TR, Been LF, Singhal A, Worsham J, Ralhan S, Wander GS, et al. (2012) A replication study of GWAS-derived lipid genes in Asian Indians: the chromosomal region 11q23.3 harbors loci contributing to triglycerides. PLoS One. 7(5):e37056. Epub 2012 May 18. pmid:22623978
- 9. Willer CJ, Schmidt EM, Sengupta S, Peloso GM, Gustafsson S, Kanoni S, et al. (2013) Discovery and Refinement of Loci Associated with Lipid Levels. Nat Genet 45: 1–24.
- 10. Olivier M, Wang X, Cole R, Gau B, Kim J, Rubin EM, et al. (2004) Haplotype analysis of the apolipoprotein gene cluster on human chromosome 11. Genomics. 83(5):912–23. pmid:15081120
- 11. Sekhri T, Kanwar RS, Wilfred R, Chugh P, Chhillar M, Aggarwal R, et al. (2014) Prevalence of risk factors for coronary artery disease in an urban Indian population. BMJ Open. 4: e005346–e005346. pmid:25488095
- 12. Joshi SR, Anjana RM, Deepa M, Pradeepa R, Bhansali A, et al. (2014) Prevalence of Dyslipidemia in Urban and Rural India: The ICMR–INDIAB Study. PLoS ONE 9(5): e96808. pmid:24817067
- 13. Misra A, Luthra K, Vikram NK. (2004) Dyslipidemia in Asian Indians: Determinants and significance. J Assoc Physicians India. 52: 137–142. pmid:15656049
- 14. Chandra KS, Bansal M, Nair T, Iyengar SS, Gupta R, Manchanda SC, et al. (2014) Consensus statement on management of dyslipidemia in Indian subjects. Indian Heart J. 66: S1–S51.
- 15. Shanker J, Perumal G, Rao VS, Khadrinarasimhiah NB, John S, Hebbagodi S, et al. (2008) Genetic studies on the APOA1-C3-A5 gene cluster in Asian Indians with premature coronary artery disease. Lipids Health Dis. 7: 33. pmid:18801202
- 16. Bhanushali AA, Das BR. (2010) Influence of genetic variants in the apolipoprotein A5 and C3 gene on lipids, lipoproteins, and its association with coronary artery disease in Indians. J Community Genet. 1: 139–148. pmid:22460246
- 17. Kumar BAV, Kaujalgi RI, Sharma AK, Basavaraj Y, Deepa VA, Kumar SV. (2014) Significance of Dyslipidemia in First Incidence Coronary Syndrome : A Retrospective Cohort Study. Int J Sci Stud. 2: 73–77.
- 18. Reddy BM, Naidu VM, Madhavi VK, Thangaraj LK, Kumar V, Langstieh BT, et al. (2005) Microsatellite diversity in Andhra Pradesh, India: genetic stratification versus social stratification. Hum. Biol. 77, 803–823. pmid:16715839
- 19. Falush D., Stephens M., Pritchard J.K., 2003. Inferences of population structure using multilocus genotype data: linked loci and correlated allele frequencies. Genetics. 164: 1567–1587. pmid:12930761
- 20. Pritchard J.K., Stephens M., Donnelly P. (2000). Inference of population structure using multilocus genotype data. Genetics. 155: 945–959. pmid:10835412
- 21.
Sambrook J, Fritschi EF, Maniatis T.(1989). Molecular Cloning: A Laboratory Manual. Newyork: Cold Spring Harbor Laboratory Press.
- 22. Schunkert H, Konig IR, Kathiresan S, Reilly MP, Assimes TL, Holm H, et al. (2011) Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet. 43(4):333–338. pmid:21378990
- 23. Wang J, Lin M, Crenshaw A, Hutchinson A, Hicks B, Yeager M, et al. (2009) High-throughput single nucleotide polymorphism genotyping using nanofluidic Dynamic Arrays. BMC Genomics. 10: 561. pmid:19943955
- 24. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MAR, Bender D, et al. (2007) PLINK: a toolset for whole-genome association and population-based linkage analysis. American Journal of Human Genetics, 81.
- 25. Sachdeva A, Cannon CP, Deedwania PC, LaBresh KA, Smith SC, Dai D, et al. (2009) Lipid levels in patients hospitalized with coronary artery disease: An analysis of 136,905 hospitalizations in Get With The Guidelines. Am Heart J.157: 111–117. pmid:19081406
- 26. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, et al. (2002) The structure of haplotype blocks in the human genome. Science. 296(5576): 2225–2229. pmid:12029063
- 27. Chhabra S, Narang R, Lakshmy R, Das N. (2005). APOA1-75 G to A substitution associated with severe forms of CAD, lower levels of HDL and apoA-I among northern Indians. Dis Markers. 21: 169–74. pmid:16403951
- 28. Walia GK, Gupta V, Aggarwal A, Asghar M, Dudbridge F, Timpson N, et al. (2014) Association of Common Genetic Variants with Lipid Traits in the Indian Population. PLoS ONE 9(7): e101688. pmid:24991929
- 29. Kraja AT, Vaidya D, Pankow JS, Goodarzi MO, Assimes TL, Kullo IJ, et al. (2011) A Bivariate Genome-Wide Approach to Metabolic Syndrome: STAMPEED Consortium. Diabetes. 60: 1329–1339. pmid:21386085
- 30. Aung LH, Yin RX, Wu DF, Wang W, Liu CW, Pan SL. (2014) Association of the variants in the BUD13-ZNF259 genes and the risk of hyperlipidaemia. J Cell Mol Med. 18(7):1417–28. Epub 2014 Apr 30. pmid:24780069
- 31. Dziembowski A, Ventura AP, Rutz B, Caspary F, Faux C, Halgand F, et al. (2004) Proteomic analysis identifies a new complex required for nuclear pre-mRNA retention and splicing. EMBO J.23: 4847–56. Available: http://doi:10.1038/sj.emboj.7600482 pmid:15565172
- 32. Drenos F, Talmud PJ, Casas JP, Smeeth L, Palmen J, Humphries SE, et al. (2009). Integrted associations of genotypes with multiple blood biomarkers linked to coronary heart disease risk. Hum Mol Genet. 18(12):2305–16. Epub 2009 Mar 31. pmid:19336475
- 33. Wallace C, Newhouse SJ, Braund P, Zhang F, Tobin M, Falchi M, et al. (2008) Genome-wide Association Study Identifies Genes for Biomarkers of Cardiovascular Disease: Serum Urate and Dyslipidemia. Am J Hum Genet. 82: 139–149. pmid:18179892
- 34. Gangwani L (2006). Deficiency of the zinc finger protein ZPR1 causes defects in transcription and cell cycle progression. J Biol Chem. 281: 40330–40. pmid:17068332
- 35. AshokKumar M, Subhashini NG, SaiBabu R, Ramesh A, Cherian KM, Emmanuel C. (2010) Genetic variants on apolipoprotein gene cluster influence triglycerides with a risk of coronary artery disease among Indians. Mol Biol Rep. 37(1):521–527. pmid:19701693
- 36. Dhandapany PS, Sadayappan S, Xue Y, Powell GT, Rani DS, Nallari P, et al. (2009) A common MYBPC3 (cardiac myosin binding protein C) variant associated with cardiomyopathies in South Asia. Nat Genet. 41:187–91 pmid:19151713
- 37. Aruna M, Nagaraja T, Andal S, Tarakeswari S, Reddy AG, Thangaraj K, et al. (2011) Novel Alleles of HLA-DQ and DR Loci Show Association with Recurrent Miscarriages among South Indian Women. Human Reproduction 26: 765–774. pmid:21325036
- 38. Dasgupta S, Sirisha P, Neelaveni K, Anuradha K, Sudhakar G, Reddy BM. (2012) Polymorphisms in the IRS-1 and PPAR-γ genes and their association with polycystic ovary syndrome among South Indian women. Gene. 503(1):140–6. Epub 2012 May 2. pmid:22575725
- 39. Uma Jyothi K, Reddy BM. (2015) Gene–gene and gene–environment interactions in the etiology of type 2 diabetes mellitus in the population of Hyderabad, India. Meta Gene. 5: 9–20. pmid:26042206
- 40. Saxena R, Saleheen D, Been LF, Garavito ML, Braun T, Bjonnes A, et al. (2013) Genome-wide association study identifies a novel locus contributing to type 2 diabetes susceptibility in Sikhs of Punjabi origin from India. Diabetes 62:1746–55. pmid:23300278
- 41. Tabassum R, Chauhan G, Dwivedi OP, Mahajan A, Jaiswal A, Kaur I, et al.(2013) Genome-wide association study for type 2 diabetes in Indians identifies a new susceptibility locus at 2q21. Diabetes. 62:977–86. pmid:23209189