Journal of Cardiology 64 (2014) 339–346

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector Contents lists available at ScienceDirect

Journal of Cardiology

journal homepage: www.elsevier.com/locate/jjcc

Original article CELSR2–PSRC1–SORT1 expression and association with coronary artery disease and plasma lipid levels in an Asian Indian cohort

Prathima Arvind (MSc) a, Jiny Nair (MSc) a, Srikarthika Jambunathan (PhD) a, Vijay V. Kakkar (DSc) b,c, Jayashree Shanker (PhD) a,∗

a Mary and Garry Weston Functional Genomics Unit, Thrombosis Research Unit, Bangalore 560099, India b Thrombosis Research Unit, Bangalore 560099, India c Thrombosis Research Institute, London, UK

article info abstract

Article history: Background: Genetic regulation of plasma lipids has been shown to influence the risk of coronary artery Received 6 November 2013 disease (CAD). We analyzed the relationship between rs599839 and rs646776 single nucleotide polymor- Received in revised form 22 January 2014 phisms (SNPs) present in the CELSR2–PSRC1–SORT1 gene cluster, candidate , and their Accepted 15 February 2014 association with CAD and circulating lipid levels in a representative cohort of Asian Indians selected from Available online 24 March 2014 the Indian Atherosclerosis Research Study. Methods: SNPs rs599839 and rs646776 were genotyped by Taqman assay in 1034 CAD patients (cases) and Keywords: 1034 age- and gender-matched controls. Expression of CELSR2, PSRC1, and SORT1 was measured in Cholesterol gene cluster Coronary artery disease 100 cases and 100 controls. Plasma levels of total cholesterol (TC), triglycerides, high-density lipoprotein- Plasma lipid levels cholesterol, and low-density lipoprotein-cholesterol (LDL-c) were measured by enzymatic assay. Single nucleotide polymorphism Results: Both rs646776 and rs599839 were in strong linkage disequilibrium (r = 0.98) and showed signif- icant protective association with CAD (OR = 0.315, 95% CI 0.136–0.728, p < 0.007 and OR = 0.422, 95% CI 0.181–0.981, p = 0.045, respectively). Haplotype TA showed 72% frequency and was associated with CAD (OR 0.77, 95% CI 0.67–0.88, p = 0.0002). PSRC1 gene expression was lower in the cases than in the controls (0.75 ± 0.405 versus 1.04 ± 0.622, p = 2.26 × 10−4). The homozygous variant and heterozygous genotypes showed 30% and 15% higher PSRC1 expression, respectively. Correspondingly, the minor alleles were associated with lower plasma TC and LDL-c levels. Conclusion: PSRC1 in the cholesterol gene cluster shows a significant association with CAD by virtue of the two SNPs, rs646776 and rs599839 that also regulate plasma cholesterol levels. © 2014 Japanese College of Cardiology. Published by Elsevier Ltd. All rights reserved.

Introduction traditional risk factors such as age, hypertension, and diabetes, heritable factors such as genes that regulate plasma lipid levels Coronary artery disease (CAD) is reported to be one of the lead- underscore the significant role of lipids in the development of CAD ing causes of death in young Indians in the 30–50 year age group [5,6]. Studies have identified single nucleotide polymorphisms [1]. Identification of predisposing genetic factors that promote the (SNPs) in novel genetic regions that affect plasma lipid levels [7–9] development of premature CAD should eventually help in reducing and some of those SNPs reside within putative candidate genes the overall mortality rates by promoting targeted therapy. In this that are known to regulate the lipid metabolism [10,11]. regard, genome wide association studies (GWAS) in diverse ethnic The gene cluster, located on 1p13.3 region (in populations have been instrumental in the discovery of numerous Homo sapiens) comprises three distinct genes, CELSR2 (cadherin, genetic variants for CAD [2–4] in addition to confirming previously EGF LAG seven-pass G-type receptor 2), PSRC1 (proline/serine rich identified markers like those in the 9p21 locus [3,4]. Apart from coiled coil 1), and SORT1 (sortilin 1). The CELSR2 gene encodes a non-classic type of cadherin involved in contact-mediated cell adhesion and receptor–ligand interactions [12]. The PSRC1 gene ∗ Corresponding author at: Mary and Garry Weston Functional Genomics Unit, product plays a role in microtubule destabilization [13] while the Thrombosis Research Institute, Narayana Hrudayalaya 258/A, Bommasandra Indus- SORT1 gene encodes for the sortilin that plays an impor- trial Area, Anekal Taluk, Bangalore 560099, Karnataka, India. Tel.: +91 80 27835303; tant role in the uptake of lipids [14]. A variant, rs599839 was fax: +91 80 27835302. initially identified through GWAS study in the intergenic region E-mail address: [email protected] (J. Shanker).

http://dx.doi.org/10.1016/j.jjcc.2014.02.012 0914-5087/© 2014 Japanese College of Cardiology. Published by Elsevier Ltd. All rights reserved. 340 P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 between CELSR2 and PSRC1 and was associated with increased quantitated using NanoDrop ND-1000 Spectrophotometer (Nano- risk of CAD [15]. Subsequently, variants in this cholesterol gene Drop, Wilmington, DE, USA). RNA samples were further digested cluster (rs646776 and rs599839) have been associated with low with DNase I followed by first strand synthesis of cDNA using the low-density lipoprotein cholesterol (LDL-c) levels [8,16]. Further, cDNA Archive kit (Applied Biosystems, Foster City, CA, USA). a recent replication study conducted on Asian Indians from the Sikh Diabetes Study (SDS) cohort reported a significant associa- Lipid assays tion between variants in the lipid regulating gene loci and the lipid profiles including high-density lipoprotein cholesterol (HDL-c) and Serum total cholesterol (TC) and TG were estimated by standard triglyceride (TG) levels [17]. These studies highlight the significant enzymatic analysis on the Cobas-Fara II Clinical Chemistry Auto role played by these SNPs and emphasize the need to replicate analyzer (F. Hoffman La Roche Ltd., Basel, Switzerland). HDL-c these findings in other populations with the possibility of trans- concentrations were estimated after precipitating the non-HDL lating these findings into clinically meaningful outcomes. We have fractions with a mixture of 2.4-mmol/l phosphotungstic acid and examined the association of two SNPs, rs599839 and rs646776 with 39 mmol/l magnesium chloride (Bayer Diagnostics, Gujarat, India). CAD, the expression of CELSR2, PSRC1, and SORT1 genes and fasting Plasma LDL-c was calculated using the Friedewald formula [21]. lipid levels in a representative cohort of Asian Indians. The inter-assay co-efficient of variation for the commercial con- trols and normal serum pool ranged from 4.9% to 7.0% for TC, 6.1% to 7.7% for TG, and 7.1% to 12.2% for HDL-c. Materials and methods The atherogenic index of plasma (AIP) was estimated using the formula [AIP = log(TG/HDL-c)] [22]. Subjects were classified into Study population three groups based on their AIP scores as follows: Group 1 (low atherogenic risk): <0.011; Group 2 (intermediate risk): 0.11–0.21; The study subjects (cases and controls) were selected from Group 3 (high risk): >0.21. Further, all the participants in the asso- the ongoing Indian Atherosclerosis Research Study (IARS), an epi- ciation cohort were classified into those with or without metabolic demiological study that was initiated in January 2004 with the syndrome based on standard Adult Treatment Panel (ATP) III guide- objective of understanding the contributions of traditional and lines (National Cholesterol Education Program-ATP-III, 2001) [23] emerging risk factors of premature CAD in Asian Indians residing as well as modified ATPIII criteria [24]. The primary difference in the Indian sub-continent. The design and overview of the IARS between the two methods of classification is that abdominal obe- have been previously published [18]. Briefly, CAD patients were sity defined by waist circumference of WC >102 cm (40 in.) in men enrolled from two major cardiology specialty hospitals, Narayana and >88 cm (35 in.) in women was revised to >90 cm for men and Health, Bangalore and the Asian Heart Institute and Research Cen- >80 cm for women based on publications that have shown that the ter, Mumbai as well as from the local clinics/hospitals in Bangalore revised cut-offs represent abdominal obesity better for the Asian between 2004 and 2011. CAD patients had a strong family his- Indian population due to their unique pattern of fat distribution tory of cardiovascular disease, with males having age at CAD onset [25]. <60 years and females <65 years. Presence of CAD was based on the report of cardiologists’ diagnosis and included any of the fol- Genotyping of CELSR2–PSRC1–SORT1 variants lowing criteria – angiographically confirmed presence of disease, with/without history of myocardial infarction and treated with Two SNPs in the CELSR2–PSRC1–SORT1 gene loci namely standard medications, percutaneous coronary intervention (PCI), or rs646776 and rs599839 were genotyped by TaqMan allelic discrim- coronary angiography bypass graft (CABG). Controls were matched ination assay on a 7900HT Fast Real Time PCR instrument (Applied for gender and age (±3 years) and were enrolled from the same Biosystems). In-house controls with known genotype were used geographical area as that of the cases. They were healthy, clinically as positive control while a no-template reaction was assigned asymptomatic for CAD, showed normal electrocardiography (ECG) as the negative control and run with every batch of Taqman readings and did not have a family history of cardiovascular disease. assay. All the participants were above 18 years, with family living in the Indian sub-continent for at least two generations, did not have any Gene expression assay other major illness such as cancer, liver disease, or primordial heart diseases such as cardiomyopathies or congenital heart disease, and Expression levels of CELSR2, PSRC1, and SORT1 genes were mea- were free of concomitant infections such as cold and cough at the sured using Taqman real-time PCR assay. The FAM and TAMRA time of enrolment. Written informed consent was obtained from dye labeled probe-primer mix (20×) and the Universal Master Mix all the participants. The IARS protocol has been approved by the (2×) were purchased from ABI (Applied Biosystems). ␤-Actin was Thrombosis Research Institute Ethics committee and is based on used as an endogenous control and run along with every sample. the bioethics guidelines of the Indian Council of Medical Research All assays were done in duplicates and according to manufac- [19]. A total of 1034 cases and 1034 age- and gender-matched con- turer’s instructions. Data were processed using SDS v2.3 (sequence trols, were selected for the association study, while an additional detection software) (Applied Biosystems). RQ Manager was used to 100 cases and 100 controls were included for the gene expression calculate the relative changes in mRNA expression levels between study. the cases and the controls. The assay was repeated in duplicates for samples showing skewed expression pattern while the persistent Sample preparation outliers were excluded from further analysis.

Venous blood samples were collected after an overnight fast Network analysis from all the study participants. Aliquots of serum and plasma were stored at −80 ◦C. Genomic DNA was extracted by a modi- We generated a protein interaction network with the fied salting out procedure [20]. Total RNA was extracted from fresh PSRC1 as seed gene using the software Search Tool for the 3 ml ethylenediaminetetraacetic acid whole blood sample using the Retrieval of Interacting Genes/ (STRING) version 9.05 QIAmp RNA blood mini kit (Qiagen, Valencia, CA, USA), following (http://string-db.org) [26,27] in order to identify its interacting manufacturer’s instructions. Both genomic DNA and total RNA were partners. P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 341

Statistical methods and the age at onset of CAD were around 50 years. Classical risk factors such as diabetes and hypertension were more prevalent in Statistical analysis was done using SPSS v17.0 software (SPSS the cases than in the controls. Measurement of waist–hip ratio and Inc., Chicago, IL, USA). Continuous variables were expressed as measurement of waist circumference were comparable between mean ± standard error of the mean (SEM). Chi-square test was the two groups whereas the mean BMI (kg/m2) was significantly used for testing the association of CELSR2–PSRC1–SORT1 vari- higher in cases as compared to the controls in the association study ants with CAD and for estimating the odds ratios (ORs) and cohort (25.97 ± 0.13 kg/m2 versus 25.47 ± 0.13 kg/m2, p = 0.008). 95% confidence intervals (CIs), while Student’s t-test and uni- TC and LDL-c levels were found to be lower in the cases than in the variate analysis were used to test for the mean differences controls, which might be due to the usage of statins in the former in the expression levels and other quantitative traits between group. the cases and the controls. Age, gender, diabetes, hyperten- In the association study cohort, there were 464 cases of chronic sion, plasma lipids, body mass index (BMI) and smoking were stable angina (CSA), 61 with unstable angina (UA), and 482 cases considered as covariates and appropriately adjusted for during with myocardial infarction (MI). Information was not available multivariate analysis to test for the association of the two SNPs for 27 cases. Severity of CAD based on the number of diseased with plasma lipid levels. SNPstats online software was also used vessels showed the following distribution: 1-vessel disease: 186; to calculate allele and genotype frequencies, Hardy–Weinberg 2-vessel disease: 278; 3- or more vessel disease: 501; data miss- equilibrium and genotype/haplotype associations with CAD ing: 69. With regard to medical management of CAD, 169 patients (http://bioinfo.iconcologia.net/SNPstats) [28]. Estimation of link- underwent PCI, 692 patients underwent CABG, while 173 were age disequilibrium (LD) was performed using the Haploview v3.32 only on standard drug therapy. Medication details are provided in software (http://www.broad.mit.edu/mpg/haploview) [29]. Table 1 and primarily include statins, beta-blockers, angiotensin- converting enzyme inhibitors, calcium-channel blockers, nitrates, Results antiplatelets, fenofibrates, and hypoglycemic agents. In the gene expression study cohort, there were 51 cases of CSA Clinical characteristics of study participants and 46 cases of MI. Information was missing for 4 cases. Frequency distribution values based on the number of diseased vessels were The clinical profile of subjects enrolled in (a) given as follows: 1-vessel disease: 10; 2-vessel disease: 33; 3- or genotype–phenotype association study and (b) gene expres- more vessel disease: 53; data not available: 5 cases. 88 patients sion study is summarized in Table 1. In the gene expression study underwent CABG; three had PCI while nine patients were treated cohort, after removal of outliers and individuals with missing with standard medication only. As there is a restricted access to information, data on 98 cases and 97 controls were included in the the old patient records at Narayana Health due to the ongoing pro- final analysis. Males showed higher representation (76–92%) than cess of digitization of hospital records, we have been unable to females (24–8%). The average age of the cases and the controls retrieve medical information for all the participants. Further, as

Table 1 Clinical characteristics of study participants included in the association and expression study.

Association study cohort p-Value Gene expression study cohort p-Value

Cases (n = 1034) Control (n = 1034) Cases (n = 98) Control (n = 97)

Clinical factors Age (years)a 50.13 ± 0.261 50.13 ± 0.252 0.711 50.13 ± 0.651 49.88 ± 0.651 0.780 Age at onset (years)a 48.56 ± 0.740 – Males, N (%) 801(77.5%) 785 (75.9%) – 82 (84.5) 89 (91.7) – Females, N (%) 233 (22.5%) 249 (24.1%) – 16 (16.49) 8 (8.2) – BMI (kg/m2)a 25.97 ± 0.13 25.47 ± 0.13 0.008 24.82 ± 0.394 24.43 ± 0.392 0.491 Waist circumference (cm)a 89.90 ± 0.327 90.48 ± 0.362 0.234 89.51 ± 0.899 91.13 ± 0.904 0.205 Waist/hip ratio (cm)a 0.95 ± 0.003 0.95 ± 0.002 0.19 0.99 ± 0.006 0.98 ± 0.006 0.287 Laboratory studies TC (mg/dl)a 152.42 ± 1.33 176.71 ± 1.209 9.74 × 10−4 129 ± 3.89 179 ± 3.87 1.86 × 10−16 TG (mg/dl)a 165.55 ± 2.41 167.04 ± 3.88 0.744 154.75 ± 9.11 194.71 ± 9.06 0.002 HDL-c (mg/dl)a 37.52 ± 0.27 40.30 ± 0.317 4.244 × 10−11 31.02 ± 1.028 37.84 ± 1.023 5.02 × 10−6 LDL-c (mg/dl)a 81.79 ± 1.146 104.62 ± 1.012 1.12 × 10−47 68.30 ± 3.29 103.87 ± 3.35 1.79 × 10−12 Medical history Smoking, N (%) 406 (39.3) 232 (22.5) 8.96 × 10−17 59 (62.1) 25 (26.3) 5.50 × 10−7 Hypertension, N (%) 557 (54) 156 (15.3) 2.98 × 10−14 38 (39.6) 11 (11.5) 5.76 × 10−6 Diabetes mellitus, N (%) 450 (43.6) 164 (16.1) 1.92 × 10−43 49 (51) 17 (17.7) 8.87 × 10−7 Metabolic syndrome MS-ATPIII 205 (19.8) 221 (21.4) 0.415 33 (32.7) 39 (40.6) 0.300 MS-modified ATPIII 556 (53.8) 499 (48.3) 0.014 42 (41.6) 55 (57.3) 0.033 Medicationsb Statin, N (%) 775 (75) 2 (0.5) – 53 (56.7) – – Fenofibrates, N (%) 20 (1.93) – – – – – Beta blockers 660 (63.83) – – 58 (57.43) – – ACE-inhibitors 329 (31.82) – – 30 (30) – – Calcium-channel blockers 285 (27.56) – 8 (8) – – Nitrates 377 (36.46) – – 42 (42) – – Antiplatelets 790 (76.40) 1 (0.09) – 51 (51) – – Hypoglycemic agents 201 (19.44) – – 8 (8) – –

BMI, body mass index; FBS, fasting blood sugar; HDL-c, high-density lipoprotein cholesterol; LDL-c, low-density lipoprotein cholesterol; MS, metabolic syndrome; ATP, Adult Treatment Panel; TC, total cholesterol; TG, triglycerides. a Continuous variables are expressed as mean ± standard error. b Information on medication was missing for 1 case in the association study cohort and 1 case in the gene expression study cohort. 342 P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 some of the patients underwent angiogram procedure or percu- taneous interventional surgery elsewhere and came to Narayana Health for a general consultation, we were unable to procure rele- vant test records from such patients. The AIP was estimated for all study participants. In the associa- tion study cohort, the mean AIP was significantly higher in the cases (0.614) than in the controls (0.557) (p = 1.26 × 10−6). Mean AIP was higher in the males (n = 1572) (0.608) than in the females (n = 478) (0.608 versus 0.515, p = 4.18 × 10−11). There was a significant asso- ciation between three AIP groups and CAD (p = 1.29 × 10−12) and the percent frequencies in the cases and controls were as follows: low risk: 1.3% and 4.9%; intermediate risk: 1.6% and 6.5%; high risk: 97.0% and 88.6%. Logistic regression showed an OR of 3.988 (95% CI 2.011–7.909) for CAD when the 1st and 3rd groups were com- pared after adjusting for age, gender, diabetes, and hypertension. Significant higher mean AIP values were noted in the cases in spite of 73.4% of the cases being on statin medication while only 0.5% of the controls were on any lipid-lowering drug. The overall trend was similar in the gene expression cohort; however the differences or association was not statistically signif- Fig. 1. Linkage disequilibrium plot for rs646776 and rs599839 SNPs in the 1p13.3  icant possibly due to the small sample size. gene loci. Haploview linkage disequilibrium plot with D value of 98 between the two SNPs In the association study cohort of 2068 subjects, there was no significant difference in the proportion of subjects with metabolic syndrome (MS) in the cases (n = 205, 19.8%) and the controls (n = 221, 21.4%) (p = 0.415) when standard ATPIII guidelines were heterozygote genotypes showed 15% higher expression (Fig. 2). applied. However, the pattern changed with the application of the There was however no significant change in the expression levels revised WC cut-offs where the cases had higher proportion of MS of CELSR2 and SORT 1 genes across all the three genotypes. (n = 556, 53.8%) than among the controls (n = 499, 48.3%) (p = 0.014). In the gene expression study cohort, MS information was avail- able for 100 cases and 96 controls. While the standard ATPIII Association of CELSR2–PSRC1–SORT1 gene expression with CAD classification did not show any significant difference in the preva- and lipids lence of MS in the cases (n = 33, 32.7%) and the controls (n = 39, 40.6%) (p = 0.300), with the revised ATPIII, MS prevalence was inter- Next, we looked at the association of CELSR2–PSRC1–SORT1 estingly higher in the controls (n = 55, 57.3%) than in the cases (42, gene expression in CAD. Expression levels of PSRC1 gene tran- 41.6%) (p = 0.033). script were significantly lower in the cases than in the controls (0.75 ± 0.405 versus 1.04 ± 0.622, p = 2.255 × 10−4). On the other Association of SNPs with CAD hand, the expression of CELSR2 and SORT1 genes did not show any significant difference between the cases and the controls. Further, rs646776 and rs599839 SNPs in the CELSR2–PSRC1–SORT1 gene the PSRC1 gene showed significant protective association with CAD loci were in Hardy–Weinberg equilibrium in the control group (OR = 0.216, 95% CI 0.087–0.535, p = 0.001). We observed strong (p > 0.05). rs599839 (OR = 0.315, 95% CI 0.136–0.728, p < 0.007) and correlation of PSRC1 mRNA expression with SORT1 (r = 0.69) and rs646776 (OR = 0.422, 95% CI 0.181–0.981, p = 0.045) showed signif- CELSR2 (r = 0.46). PSRC1 expression also showed modest corre- icant protective association with CAD and were in strong LD with lation with TC (r = 0.22, p = 0.003), LDL-c (r = 0.17, p = 0.024), and each other (r = 0.98) (Fig. 1). The common ‘TA’ haplotype for the two HDL-c (r = 0.16, p = 0.032) levels while CELSR2 showed correlation variants, rs646776 and rs599839 showed a frequency of 0.73. The with TC (r = 0.22, p = 0.003), LDL-c (r = 0.18, p = 0.011), and TG levels alternate ‘CG’ haplotype (0.27) showed protection against CAD (OR (r = 0.18, p = 0.013). 0.77, 95% CI 0.67–0.88, p = 0.0002).

Association of SNPs with plasma lipid levels Table 2 Mean plasma lipid levels for rs646776 and rs599839 SNPs in the association study. We examined the effect of rs646776 and rs599839 on plasma lipid levels. Both variants showed a significant association with TC rs646776 (p = 0.007) and LDL-c (p < 0.01), with the minor alleles being associ- CC TC TT p-Value ated with lower lipid levels. This association remained significant TC (mg/dl) 155.68 163.35 166.60 0.007 even after adjusting for age, gender, diabetes, hypertension, lipids, TG 150.63 169.03 166.50 0.131 BMI, and smoking (p < 0.002). Homozygotes for the minor allele HDL-c 38.52 39.41 38.6 0.172 (CC or GG) showed an overall reduction of approximately 11 mg/dl LDL-c 87.38 90.99 95.21 0.006 (0.29 mmol/l) in TC levels and around 8 mg/dl (0.21 mmol/l) in rs599839 LDL-c levels (Table 2). AA AG GG p-Value

Association between genotype and CELSR2–PSRC1–SORT1 gene TC (mg/dl) 166.75 163.15 156.13 0.007 expression TG 166.91 168.74 149.94 0.106 HDL-c 38.67 39.27 38.74 0.387 As compared to the common TT genotype in rs646776 or AA LDL-c 95.09 91.15 87.74 0.01 genotype in rs599839, the homozygous variant genotype (CC or TC, total cholesterol; TG, triglycerides; HDL-c, high-density lipoprotein cholesterol; GG) showed 30% increase in PSRC1 mRNA expression while the LDL-c, low-density lipoprotein cholesterol. P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 343

Fig. 2. Mean mRNA expression levels of CELSR2, PSRC1 and SORT1 genes across rs646776 and rs599839 genotypes. mRNA expression level of CELSR2, PSRC1 and SORT1 in the presence of homozygote (major and minor) and heterozygote alleles for rs646776 and rs599839 SNPs. Note: for rs646776, TT: major allele, CC: minor allele; for rs599839, AA: major allele, GG: minor allele.

Network analysis alleles of both rs599839 and rs646776 with low plasma LDL-c lev- els [10,39,40]. In our preliminary analysis of the protein interaction A bionetwork was generated using STRING to understand the network generated using STRING v9.05, with PSRC1 as the seed relationship of the PSRC1 gene with other interactors where PSRC1 gene, we observed that the PSRC1 appeared to be a key protein was used as the seed gene. This gene showed interactions with linking CELSR2 and SORT1 on one side and factors affecting mitosis various partners including genes involved in lipoprotein uptake and cell division on the other side (Fig. 3). Further, the link of PSRC1 and transport (SORT1 and SORL1), mitosis (CDCA3 and CDCA8), and Golgi transport (KIF20A) while PSRC1 itself is responsible for microtubule destabilization and spindle assembly (Fig. 3).

Discussion

Based on the literature search from GWAS, we identified two SNPs, rs599839 and rs646776 present in the CELSR2–PSRC1–SORT1 cholesterol gene cluster on chromosome 1p13.3 and examined their effect on gene expression, association with CAD and cir- culating lipid levels in a representative cohort of Asian Indians. We observed a strong independent association of both the vari- ants with the disease, lower expression of PSRC1 gene transcript in the cases and the minor alleles of both SNPs being associ- ated with lower level of circulating lipids, particularly TC and LDL-c. These two variants that were initially identified through GWAS in the Caucasian population showed an association with CAD and the lipid phenotype [15,30–32] and have been subse- quently validated in other populations [33] including the Sikh Diabetes Study cohort [34]. Thus, the most plausible conclu- sion derived from these reported studies including the present study is that the CELSR2–PSRC1–SORT1 locus exerts its effect on CAD through a well-established risk factor, namely plasma LDL-c. It is interesting to note that of the three genes in the above- mentioned cluster, SORT1 has been reported to be directly involved in lipid regulatory metabolism [14,35,36] while the primary role of CELSR2 and PSRC1 is cell adhesion and cytoskeleton stabiliza- tion, respectively [13,37]. To the contrary, we observed a significant association of PSRC1 mRNA expression with CAD as well as the association of the two SNPs with TC and LDL-c levels in the present study. This was comparable to a report by Wild et al. [38] who showed a significant association of rs599839 SNP with expression Fig. 3. Network analysis with PSRC1 gene as the seed gene. Network constructed of PSRC1 transcripts in the monocytes and plasma LDL-c levels. A using STRING, a web-based tool with PSRC1 as the protein of interest shows inter- number of studies have also shown an association of the variant actions (edges) of PSRC1 with other proteins. 344 P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 with SORL1 is interesting in that the latter encodes for a sortilin- their protective association with CAD and circulating cholesterol, related receptor protein, which is located on chromosome 11q23, it is a matter of time before rs599839 and rs646776 can be consid- yet another important locus regulating plasma lipids [34]. From this ered as novel variants with potential clinical utility. However, much it appears that the relationship among the candidate genes in the work needs to be done to understand the underlying mechanism by 1p13.3 locus is complex and remains to be fully elucidated. which these SNPs actually contribute to the clinical manifestation The fact that lipid is an important factor for the development of of the disease. atherosclerosis is understood from the fact that there was a strong We acknowledge some of the limitations of this study. To begin association of atherogenic index and CAD. Further, MS is considered with, the controls were selected from among individuals who to be a precursor for CAD. Asian Indians have a greater predispo- were clinically asymptomatic for the disease and showed normal sition to MS and the need for ethnicity-specific cut-offs has been ECG readings. This does not exclude the presence of an under- extensively discussed [25,41,42]. This was evident from the fact lying silent disease. Nonetheless, as the only fool-proof method that there was high MS prevalence both among the cases and con- to rule out the presence/absence of disease is through a coro- trols and the frequency differences in the cases and controls were nary angiogram, which is not ethically feasible in an experimental statistically significant when the revised cut-off was used for waist setting, this could to some extent have an impact on the signif- circumference. Thus both lipids and visceral adiposity appear to icance of our study findings. Importantly, our initial findings on be significant factors contributing to the development of metabolic the association of the two SNPs with CAD have not been vali- diseases and CAD in this population. dated in an independent cohort, considering that the contribution Both rs646776 and rs599839 are located in an intergenic region of genetic factors to CAD risk is rather small and is primarily between CELSR2 and PSRC1 genes. SNP rs599839 was identified based on differences in the risk allele frequency in cases versus first to be located in the 3-untranslated region (UTR) of PSRC1, controls. In this context it is worthwhile to recall that this locus and showed association with CAD in the British and German cohort retained genome-wide significance (>10−8) even after stringent study [43]. Subsequently, a meta-analysis carried out on the British Bonferroni correction in the Caucasian [47] and other Asian [48] population demonstrated that the rs646776 and rs599839 SNPs populations demonstrating the robustness of their association were significantly associated with lower levels of circulating LDL-c with CAD, primarily mediated through the lipids [8]. Further, only [16]. In the present study, which includes a representative cohort one locus, namely CELSR2–PSRC1–SORT1, was considered for this of Asian Indians living in the Indian sub-continent, we observed a study while a recent study has identified around 157 loci associ- similar pattern, wherein the minor alleles of both the SNPs were ated with the lipid trait [49]. The predominant usage of statins, associated with reduced LDL-c levels. This demonstrates a protec- a popular lipid-lowering drug, might to some extent have influ- tive role of both the variant alleles. We also noted higher expression enced the outcome of the findings. However, as the genotype–lipid of PSRC1 mRNA in the carriers of the minor alleles, C and G. Con- association pattern did not change even after the data were strat- trary to this, in an earlier study, higher mRNA expression levels of ified according to statin usage, there is no apparent confounding SORT1 were shown to induce a significant cellular uptake of LDL-c effect of statins on the outcome. Statins are known to influence [10] and the homozygous minor allele of rs646776 SNP was asso- gene expression, particularly those associated with inflamma- ciated with 12-fold higher SORT1 and PSRC1 expression in human tion [50,51]. Again the gene expression results were comparable liver samples [44]. In fact, this was the first study on a series of when the data were stratified based on statins thus alleviating our human cohorts and human-derived hepatocytes to provide func- concerns. tional evidence on the association of CELSR2–PSRC1–SORT1 locus In conclusion, this pilot study reiterates the involvement of the with CAD. CELSR2–PSRC1–SORT1 locus in modulating the risk of CAD through A non-coding SNP located in the 3-UTR of angiotensin II type I hitherto unknown mechanisms that regulate plasma lipid levels. receptor gene has been shown to affect the miRNA-mediated reg- While the association of the two SNPs, rs646776 and rs599839 with ulation of the mRNA [45], while in the case of human LDL receptor CAD and lipids corroborate with other published studies, PSRC1 and gene, a SNP present in the 3-UTR was associated with plasma lipid SORT1 appear to be potential gene candidates in the lipid pathway. levels in Caucasian males [46]. This provides an insight into the However, until the functional relationship among the members of post-transcriptional regulation brought about by these non-coding this cholesterol locus and other reported lipid genes is established, SNPs. rs599839 is also a non-coding SNP located in the 3-UTR of the the true clinical utility of the CELSR2–PSRC1–SORT1 locus will not PSRC1 gene which could possibly play a regulatory role at the mRNA be fully realized. level. Given that the increased expression level of PSRC1 mRNA transcripts and the reduced levels of LDL-c and TC were detected in the presence of minor alleles of both rs599839 and rs646776 Conflict of interest SNPs, it might indicate a possible role of the PSRC1 gene in the direct or indirect regulation of plasma lipid levels. The PSRC1 and The authors have no conflicts of interest to disclose. SORT1 genes lay in close proximity in the 1p13.3 genomic locus, which was reflected by the presence of strong correlation in the gene expression between the microtubule destabilizing PSRC1 and Acknowledgments plasma lipid regulating SORT1 genes in our study. Furthermore, in the light of rs646776 and rs599839 SNPs being present as a tight We gratefully acknowledge the support extended by the intergenic cluster, it would be interesting to understand the under- trustees of the Thrombosis Research Institutes in London and lying mechanisms, the functional relationship, and the manner in Bangalore, the W. Garfield Weston Foundation, UK, Founda- which the PSRC1 and SORT1 genes might complement each other in tion Bay, Switzerland, the Tata Social Welfare Trust, India modulating CAD risk through their combined effect on cholesterol (TSWT/IG/SNB/JP/Sdm) and the Department of Biotechnology, metabolism. Ministry of Science and Technology, Government of India A number of GWAS studies have confirmed the consistent yet (BT/01/CDE/08/07). We thank the study participants and the doc- modest association of the 9p21.3 common variants with incident tors; without whom this work would not have been possible. The CAD. There is however an urgent need to identify additional such sponsors did not participate in the design, conduct, sample collec- variants that can improve risk prediction. Given the published evi- tion analysis, and interpretation of the data or in the preparation, dence on the role of these SNPs in modulating PSRC1 expression, review, or approval of the manuscript. P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346 345

References Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treat- ment of High Blood Cholesterol in Adults (Adult Treatment Panel III). J Am Med [1] Panwar RB, Gupta R, Gupta BK, Raja S, Vaishnav J, Khatri M, Agrawal A. Assoc 2001;285:2486–97. Atherothrombotic risk factors & premature coronary heart disease in India: [24] Kanjilal S, Shanker J, Rao VS, Khadrinarasimhaih NB, Mukherjee M, Iyengar a case–control study. Indian J Med Res 2011;134:26–32. SS, Kakkar VV. Prevalence and component analysis of metabolic syndrome: [2] Helgadottir A, Thorleifsson G, Manolescu A, Gretarsdottir S, Blondal T, an Indian atherosclerosis research study perspective. Vasc Health Risk Manag Jonasdottir A, Sigurdsson A, Baker A, Palsson A, Masson G, Gudbjartsson DF, 2008;4:189–97. Magnusson KP, Andersen K, Levey AI, Backman VM, et al. A common variant on [25] Misra A, Khurana L. The metabolic syndrome in South Asians: epidemi- chromosome 9p21 affects the risk of myocardial infarction. Science 2007;316: ology, determinants, and prevention. Metab Syndr Relat Disord 2009;7: 1491–3. 497–514. [3] Lu X, Wang L, Chen S, He L, Yang X, Shi Y, Cheng J, Zhang L, Gu CC, Huang J, [26] Szklarczyk D, Franceschini A, Kuhn M, Simonovic M, Roth A, Minguez P, Doerks Wu T, Ma Y, Li J, Cao J, Chen J, et al. Genome-wide association study in Han T, Stark M, Muller J, Bork P, Jensen LJ, von Mering C. The STRING database Chinese identifies four new susceptibility loci for coronary artery disease. Nat in 2011: functional interaction networks of proteins, globally integrated and Genet 2012;44:890–4. scored. Nucleic Acids Res 2011;39:D561–8. [4] CADCDG Consortium. A genome-wide association study in Europeans and [27] von Mering C, Jensen LJ, Snel B, Hooper SD, Krupp M, Foglierini M, South Asians identifies five new loci for coronary artery disease. Nat Genet Jouffre N, Huynen MA, Bork P. STRING: known and predicted protein–protein 2011;43:339–44. associations, integrated and transferred across organisms. Nucleic Acids Res [5] Namboodiri KK, Kaplan EB, Heuch I, Elston RC, Green PP, Rao DC, Laskarzewski 2005;33:D433–7. P, Glueck CJ, Rifkind BM. The Collaborative Lipid Research Clinics Family Study: [28] Sole X, Guino E, Valls J, Iniesta R, Moreno V. SNPStats: a web tool for the analysis biological and cultural determinants of familial resemblance for plasma lipids of association studies. Bioinformatics 2006;22:1928–9. and lipoproteins. Genet Epidemiol 1985;2:227–54. [29] Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD [6] Weiss LA, Pan L, Abney M, Ober C. The sex-specific genetic architecture of and haplotype maps. Bioinformatics 2005;21:263–5. quantitative traits in humans. Nat Genet 2006;38:218–22. [30] Kathiresan S, Voight BF, Purcell S, Musunuru K, Ardissino D, Mannucci PM, [7] Asselbergs FW, Guo Y, van Iperen EP, Sivapalaratnam S, Tragante V, Anand S, Engert JC, Samani NJ, Schunkert H, Erdmann J, Reilly MP, Rader DJ, Lanktree MB, Lange LA, Almoguera B, Appelman YE, Barnard J, Baumert J, Morgan T, Spertus JA, et al. Genome-wide association of early-onset myocardial Beitelshees AL, Bhangale TR, Chen YD, Gaunt TR, et al. Large-scale gene-centric infarction with single nucleotide polymorphisms and copy number variants. meta-analysis across 32 studies identifies multiple lipid loci. Am J Hum Genet Nat Genet 2009;41:334–41. 2012;91:823–38. [31] Kleber ME, Renner W, Grammer TB, Linsel-Nitschke P, Boehm BO, [8] Kathiresan S, Melander O, Guiducci C, Surti A, Burtt NP, Rieder MJ, Cooper GM, Winkelmann BR, Bugert P, Hoffmann MM, Marz W. Association of the single Roos C, Voight BF, Havulinna AS, Wahlstrand B, Hedner T, Corella D, Tai ES, nucleotide polymorphism rs599839 in the vicinity of the sortilin 1 gene with Ordovas JM, et al. Six new loci associated with blood low-density lipoprotein LDL and triglyceride metabolism, coronary heart disease and myocardial infarc- cholesterol, high-density lipoprotein cholesterol or triglycerides in humans. tion. The Ludwigshafen Risk and Cardiovascular Health Study. Atherosclerosis Nat Genet 2008;40:189–97. 2010;209:492–7. [9] Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, [32] Saade S, Cazier JB, Ghassibe-Sabbagh M, Youhanna S, Badro DA, Kamatani Y, Pirruccello JP, Ripatti S, Chasman DI, Willer CJ, Johansen CT, Fouchier SW, Isaacs Hager J, Yeretzian JS, El-Khazen G, Haber M, Salloum AK, Douaihy B, Othman A, Peloso GM, Barbalic M, et al. Biological, clinical and population relevance of R, Shasha N, Kabbani S, et al. Large scale association analysis identifies three 95 loci for blood lipids. Nature 2010;466:707–13. susceptibility loci for coronary artery disease. PLoS ONE 2011;6:e29427. [10] Linsel-Nitschke P, Heeren J, Aherrahrou Z, Bruse P, Gieger C, Illig T, Prokisch [33] Zhou L, Ding H, Zhang X, He M, Huang S, Xu Y, Shi Y, Cui G, Cheng L, Wang H, Heim K, Doering A, Peters A, Meitinger T, Wichmann HE, Hinney A, Reinehr QK, Hu FB, Wang D, Wu T. Genetic variants at newly identified lipid loci are T, Roth C, et al. Genetic variation at chromosome 1p13.3 affects sortilin mRNA associated with coronary heart disease in a Chinese Han population. PLoS ONE expression, cellular LDL-uptake and serum LDL levels which translates to the 2011;6:e27481. risk of coronary artery disease. Atherosclerosis 2010;208:183–9. [34] Braun TR, Been LF, Singhal A, Worsham J, Ralhan S, Wander GS, Chambers [11] Waterworth DM, Ricketts SL, Song K, Chen L, Zhao JH, Ripatti S, Aulchenko YS, JC, Kooner JS, Aston CE, Sanghera DK. A replication study of GWAS-derived Zhang W, Yuan X, Lim N, Luan J, Ashford S, Wheeler E, Young EH, Hadley D, lipid genes in Asian Indians: the chromosomal region 11q23.3 harbors loci et al. Genetic variants influencing circulating lipid levels and risk of coronary contributing to triglycerides. PLoS ONE 2012;7:e37056. artery disease. Arterioscler Thromb Vasc Biol 2010;30:2264–76. [35] Dube JB, Johansen CT, Hegele RA. Sortilin: an unusual suspect in cholesterol [12] Vincent JB, Skaug J, Scherer SW. The human homologue of flamingo, EGFL2, metabolism: from GWAS identification to in vivo biochemical analyses, sor- encodes a brain-expressed large cadherin-like protein with epidermal growth tilin has been identified as a novel mediator of human lipoprotein metabolism. factor-like domains, and maps to chromosome 1p13.3-p21.1. DNA Res Bioessays 2011;33:430–7. 2000;7:233–5. [36] Strong A, Rader DJ. Sortilin as a regulator of lipoprotein metabolism. Curr [13] Hsieh PC, Chang JC, Sun WT, Hsieh SC, Wang MC, Wang FF. down- Atheroscler Rep 2012;14:211–8. stream target DDA3 is a novel microtubule-associated protein that interacts [37] Takeichi M, Nakagawa S, Aono S, Usui T, Uemura T. Patterning of cell assemblies with end-binding protein EB3 and activates beta-catenin pathway. Oncogene regulated by adhesion receptors of the cadherin superfamily. Philos Trans R Soc 2007;26:4928–40. Lond B Biol Sci 2000;355:885–90. [14] Kjolby M, Andersen OM, Breiderhoff T, Fjorback AW, Pedersen KM, Madsen P, [38] Wild PS, Zeller T, Schillert A, Szymczak S, Sinning CR, Deiseroth A, Schn- Jansen P, Heeren J, Willnow TE, Nykjaer A. Sort1, encoded by the cardiovascu- abel RB, Lubos E, Keller T, Eleftheriadis MS, Bickel C, Rupprecht HJ, Wilde S, lar risk locus 1p13.3, is a regulator of hepatic lipoprotein export. Cell Metab Rossmann H, Diemert P, et al. A genome-wide association study identifies 2010;12:213–23. LIPA as a susceptibility gene for coronary artery disease. Circ Cardiovasc Genet [15] Samani NJ, Braund PS, Erdmann J, Gotz A, Tomaszewski M, Linsel-Nitschke 2011;4:403–12. P, Hajat C, Mangino M, Hengstenberg C, Stark K, Ziegler A, Caulfield M, [39] Schunkert H, Konig IR, Kathiresan S, Reilly MP, Assimes TL, Holm H, Preuss M, Burton PR, Schunkert H, Tobin MD. The novel genetic variant predisposing Stewart AF, Barbalic M, Gieger C, Absher D, Aherrahrou Z, Allayee H, Altshuler D, to coronary artery disease in the region of the PSRC1 and CELSR2 genes on Anand SS, et al. Large-scale association analysis identifies 13 new susceptibility associates with serum cholesterol. J Mol Med (Berl) 2008;86: loci for coronary artery disease. Nat Genet 2011;43:333–8. 1233–41. [40] Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, [16] Sandhu MS, Waterworth DM, Debenham SL, Wheeler E, Papadakis K, Zhao JH, Pirruccello JP, Ripatti S, Chasman DI, Willer CJ, Johansen CT, Fouchier SW, Isaacs Song K, Yuan X, Johnson T, Ashford S, Inouye M, Luben R, Sims M, Hadley D, A, Peloso GM, Barbalic M, et al. Biological, clinical and population relevance of McArdle W, et al. LDL-cholesterol concentrations: a genome-wide association 95 loci for blood lipids. Nature 2011;466:707–13. study. Lancet 2008;371:483–91. [41] Misra A. Revisions of cutoffs of body mass index to define overweight and [17] Berry JD, Dyer A, Cai X, Garside DB, Ning H, Thomas A, Greenland P, Van Horn obesity are needed for the Asian-ethnic groups. Int J Obes Relat Metab Disord L, Tracy RP, Lloyd-Jones DM. Lifetime risks of cardiovascular disease. N Engl J 2003;27:1294–6. Med 2012;366:321–9. [42] Misra A, Shrivastava U. Obesity and dyslipidemia in South Asians. Nutrients [18] Shanker J, Maitra A, Rao VS, Mundkur L, Dhanalakshmi B, Hebbagodi S, Kakkar 2013;5:2708–33. VV. Rationale, design & preliminary findings of the Indian Atherosclerosis [43] Samani NJ, Erdmann J, Hall AS, Hengstenberg C, Mangino M, Mayer B, Dixon Research Study. Indian Heart J 2010;62:286–95. RJ, Meitinger T, Braund P, Wichmann HE, Barrett JH, Konig IR, Stevens SE, [19] Kumar NK. Bioethics activities in India. East Mediterr Health J 2006;12(Suppl. Szymczak S, Tregouet DA, et al. Genomewide association analysis of coronary 1):S56–65. artery disease. N Engl J Med 2007;357:443–53. [20] Miller SA, Dykes DD, Polesky HF. A simple salting out procedure for extracting [44] Musunuru K, Strong A, Frank-Kamenetsky M, Lee NE, Ahfeldt T, Sachs KV, Li X, DNA from human nucleated cells. Nucleic Acids Res 1988;16:1215. Li H, Kuperwasser N, Ruda VM, Pirruccello JP, Muchmore B, Prokunina-Olsson [21] Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of L, Hall JL, Schadt EE, et al. From noncoding variant to phenotype via SORT1 at low-density lipoprotein cholesterol in plasma, without use of the preparative the 1p13 cholesterol locus. Nature 2010;466:714–9. [45] Haas U, Sczakiel G, Laufer SD. MicroRNA-mediated regulation of gene expres- ultracentrifuge. Clin Chem 1972;18:499–502.  [22] Frohlich J, Dobiasova M. Fractional esterification rate of cholesterol and ratio sion is affected by disease-associated SNPs within the 3 -UTR via altered RNA of triglycerides to HDL-cholesterol are powerful predictors of positive findings structure. RNA Biol 2012;9:924–37. on coronary angiography. Clin Chem 2003;49:1873–80. [46] Muallem H, North KE, Kakoki M, Wojczynski MK, Li X, Grove M, Boerwinkle E, Wilhelmsen KC, Heiss G, Maeda N. Quantitative effects of common genetic [23] Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol  in Adults. Executive Summary of the Third Report of the National Cholesterol variations in the 3 UTR of the human LDL-receptor gene and their associations 346 P. Arvind et al. / Journal of Cardiology 64 (2014) 339–346

with plasma lipid levels in the Atherosclerosis Risk in Communities study. Hum [49] Willer CJ, Schmidt EM, Sengupta S, Peloso GM, Gustafsson S, Kanoni S, Ganna Genet 2007;121:421–31. A, Chen J, Buchkovich ML, Mora S, Beckmann JS, Bragg-Gresham JL, Chang HY, [47] Samani NJ, Deloukas P, Erdmann J, Hengstenberg C, Kuulasmaa K, McGinnis R, Demirkan A, Den Hertog HM, et al. Discovery and refinement of loci associated Schunkert H, Soranzo N, Thompson J, Tiret L, Ziegler A. Large scale association with lipid levels. Nat Genet 2013;45:1274–83. analysis of novel genetic loci for coronary artery disease. Arterioscler Thromb [50] Pereira MM, Sant’ana Santos TP, Aras R, Couto RD, Sousa Atta ML, Atta Vasc Biol 2009;29:774–80. AM. Serum levels of cytokines and chemokines associated with cardiovascu- [48] Lee JY, Lee BS, Shin DJ, Woo Park K, Shin YA, Joong Kim K, Heo L, Young Lee J, lar disease in Brazilian patients treated with statins for dyslipidemia. Int Kyoung Kim Y, Jin Kim Y, Bum Hong C, Lee SH, Yoon D, Jung Ku H, Oh IY, et al. A Immunopharmacol 2014;18:66–70. genome-wide association study of a coronary artery disease risk variant. J Hum [51] Vaughan CJ, Murphy MB, Buckley BM. Statins do more than just lower choles- Genet 2013;58:120–6. terol. Lancet 1996;348:1079–82.