Genome-wide conditional association study reveals the influences of lifestyle cofactors on genetic regulation of body surface area in MESA population

Abstract

Body surface area (BSA) is an important trait used for many clinical purposes. People's BSA may vary due to genetic background, race, and different lifestyle factors (such as walking, exercise, reading, smoking, transportation, etc.). GWAS of BSA was conducted on 5,324 subjects of four ethnic populations of European-American, African-American, Hispanic-American, and Chinese-American from the Multi-Ethnic Study of Atherocloris (MESA) data using unconditional and conditional full genetic models. In this study, fifteen SNPs were identified (Experiment-wise PEW < 1×10-5) using unconditional full genetic model, of which thirteen SNPs had individual genetic effects and seven SNPs were involved in four pairs of epistasis interactions. Seven single SNPs and eight pairs of epistasis SNPs were additionally identified using exercise, smoking, and transportation cofactor-conditional models. By comparing association analysis results from unconditional and cofactor conditional models, we observed three different scenarios: (i) genetic effects of several SNPs did not affected by cofactors, e.g., additive effect of gene CREB5 (a≙ -0.013 for T/T and 0.013 for G/G, -Log10 PEW = 8.240) did not change in the cofactor models; (ii) genetic effects of several SNPs affected by cofactors, e.g., the genetic additive effect (a≙ 0.012 for A/A and -0.012 for G/G, -Log10 PEW = 7.185) of SNP of the gene GRIN2A was not significant in transportation cofactor model; and (iii) genetic effects of several SNPs suppressed by cofactors, e.g., additive (a≙ -0.018 for G/G and 0.018 for C/C, -Log10 PEW = 19.737) and dominance (d≙ -0.038 for G/C, -Log10 PEW = 27.734) effects of SNP of gene ERBB4 was identified using only transportation cofactor model. Gene ontology analysis showed that several genes are related to the metabolic pathway of calcium compounds, coronary artery disease, type-2 Diabetes, Alzheimer disease, childhood obesity, sleeping duration, Parkinson disease, and cancer. This study revealed that lifestyle cofactors could contribute, suppress, increase or decrease the genetic effects of BSA associated genes.

Fig 1. Genetic architecture of detected QTSs for Body Surface Area (BSA) in both base model and five cofactor models with experiment-wise significance (−Log₁₀ P_EW > 5).

BSA: base model; BSA|Walk = walk cofactor model; BSA|Exer = exercise cofactor model; BSA|Read = read cofactor model; BSA|Smoke = smoke cofactor model; and BSA|Trans = transportation cofactor model. Note: The left axis is the QTS IDs: Chromosome-SNP-Alleles; Red circle dot: QTS with additive effects; Green circle dot: QTS with ethnic-specific additive effects; Blue circle dot: QTS with both additive and ethnic-specific additive effects; Red square dot: QTS with dominance effects; Green square dot: QTS with ethnic-specific dominance effects; Blue square dot: QTS with both dominance and ethnic-specific dominance effects; Line between two QTSs = epistasis effects; Red color = QTS with general effects for two race groups; Green color = QTS with ethnicity-specific effects; Blue color = QTS with both general and ethnicity-specific effects; Black color = QTS with ethnicity-specific effects but without detected individual effects.

Fig 2. Genetic effects matrix (G + GE) image plot of BSA loci.

(A) Identified loci using the base model; and (B) suppressed loci by cofactors. Vertical axis for the size of genetic effects including four ethnic groups: HA = Hispanic-American, AA = African-American, CA = Chinese-American, EA = European-American; horizontal axis for the individual and epistasis loci; different color present different genetic effects according to a color scale, where gray color = no significant effects.

Fig 3. Genetic architecture of detected quantitative trait signal nucleotide polymorphism (QTSs) for BSA.

(A) QTSs detected in both the BSA symptom count (non-cofactor) model and individual cofactor models with high significance (–Log₁₀P_EW > 5); (B) QTSs detected for BSA symptom count only in a suppressed model with high significance (–Log₁₀P_EW > 5). The size of balls and thickness of lines stand for a number of publications related. Red balls represent seed genes detected; orange balls represent association diseases; Olive balls represent association functions; Brown balls represent association chemicals; Royal blue balls represent association genes. Red-orange lines represent gene-disease association; Paris green lines represent gene ontology; Dark blue lines represent protein-protein interaction; Bronze lines represent pathway interaction; Dark gray lines represent the database.