Loss of Cardioprotective Effects at the ADAMTS7 Locus as a Result of Gene-Smoking Interactions

Abstract

Background: Common diseases such as coronary heart disease (CHD) are complex in etiology. The interaction of genetic susceptibility with lifestyle factors may play a prominent role. However, gene-lifestyle interactions for CHD have been difficult to identify. Here, we investigate interaction of smoking behavior, a potent lifestyle factor, with genotypes that have been shown to associate with CHD risk.

Methods: We analyzed data on 60 919 CHD cases and 80 243 controls from 29 studies for gene-smoking interactions for genetic variants at 45 loci previously reported to be associated with CHD risk. We also studied 5 loci associated with smoking behavior. Study-specific gene-smoking interaction effects were calculated and pooled using fixed-effects meta-analyses. Interaction analyses were declared to be significant at a P value of <1.0×10^-3 (Bonferroni correction for 50 tests).

Results: We identified novel gene-smoking interaction for a variant upstream of the ADAMTS7 gene. Every T allele of rs7178051 was associated with lower CHD risk by 12% in never-smokers (P=1.3×10^-16) in comparison with 5% in ever-smokers (P=2.5×10^-4), translating to a 60% loss of CHD protection conferred by this allelic variation in people who smoked tobacco (interaction P value=8.7×10^-5). The protective T allele at rs7178051 was also associated with reduced ADAMTS7 expression in human aortic endothelial cells and lymphoblastoid cell lines. Exposure of human coronary artery smooth muscle cells to cigarette smoke extract led to induction of ADAMTS7. CONCLUSIONS: Allelic variation at rs7178051 that associates with reduced ADAMTS7 expression confers stronger CHD protection in never-smokers than in ever-smokers. Increased vascular ADAMTS7 expression may contribute to the loss of CHD protection in smokers.

Keywords: ADAMTS7 protein; coronary artery disease; gene-environment interaction; genome-wide association study; smoking.

Figure 1

(a) Regional association analyses at the chromosome 15q25.1 locus in association with CHD risk stratified by smoking status. Association P-values for genetic variants with CHD risk in “never-smokers” (green squares) and “ever-smokers” (red triangles). (b) Longitudinal bars represent gene-smoking CHD interaction P-values at the chromosome 15q25.1 locus; bars in blue are P-values for variants listed in Table-1 and each variant has been assigned a unique identification number based on its physical location; (c) LD-blocks at the 15q25.1 locus visualized through HAPLOVIEW using LD estimates in the HapMAP-2 CEU reference population.

Figure 2

Several variants at chromosome 15q21.1 have stronger effects on CHD risk in “never-smokers” compared to “ever-smokers”. Variants with the strongest interaction P-value are displayed.

Figure 3

Step-wise conditional analysis of genetic variation at the chromosome 15q21.1 locus with CHD (red triangles) and smoking behavior (cigarettes per day, CPD; grey circles). At the chromosome 15q21.1 locus, analyses adjusted for rs7178051 and rs11072794 completely attenuated the gene-CHD associations whereas gene-smoking remained unchanged. Analyses adjusted for rs1051730 and rs684513 completely attenuated the gene-smoking associations whereas gene-CHD effect remained unchanged.

Figure 4

Analyses mutually adjusted for rs7178051, rs11072794, rs1051730 and rs684513 at 15q21.1 on CHD and smoking behavior; gene-CHD interaction analyses were only found significant for rs7178051. Analyses on the left panel show associations of rs7178051, rs11072794, rs1051730 and rs684513 with CHD risk mutually adjusted for each other. Analyses on the right panel show associations of rs7178051, rs11072794, rs1051730 and rs684513 with smoking behavior mutually adjusted for each other.

Figure 5

(a) ADAMTS7 and CHRNB4-A3-A5 mRNA levels were measured in HCASMC. Cells were cultured to confluence, total RNA was extracted and cDNA generated. q-PCR was performed for ACTB, GAPDH, TBP, ADAMTS7, CHRNB4, CHRNA3, CHRNA5 (95°C 15s, 60°C 1min). Delta Cts were calculated as follows: (Ct_ACTB + Ct_GAPDH + Ct_TBP)/3 – Ct_{TARGET GENE}). Fold changes are derived from delta delta Cts based on formula FC = 2^−dCt. (b) Confluent HCASMC were exposed to cigarette smoke extract. Serum starved (x24 hrs.) confluent HCASMC were treated with 0.5% or 1.0% cigarette smoke extract (v/v) for 4, 12, and 24 hrs. in serum reduced conditions (0.5% FBS in DMEM). Total RNA was extracted, cDNA generated preparation and q-PCR performed for ADAMTS7 by Taqman and normalized to GAPDH. The Average Ct for ADAMTS7 at baseline was 28.25. Results were presented as means ± SEM, and data were analyzed using Student’s t-Test. (c) expression and eQTL Data from the GTEx consortium, the HapMap consortium (restricted to European populations), the Multiple Tissue Human Expression Resource (MuTHER) and in 147 donor HAoEC lines. Association of the independent lead variants identified in our conditional analyses with expression of ADAMTS7 and genes in the CHRNB4-A3-A5 cluster. A P-value threshold of 0.002 was set to account for multiple testing involved in the eQTL analyses.

Figure 6

Genome browser view of regulatory features at rs7178051 on Chr15q21.1. ChIP-seq experiments were performed on confluent HCASMC for TCF21, Jun, JunD, CEBP and H3K4me1, H3K27me3, H3K27ac. DNAaseI hypersensitivity data for human AoSMC were acquired from the ENCODE project. Human aortic tissue H3K4me1, H3K9me3, H3K27me3, and H3K36me3 ChIP-seq data were acquired from the NIH Roadmap Epigenomics Project. HCASMC = human coronary artery smooth muscle cells; AoSMC = human aortic smooth muscle cells.