Abstract 002: DNA Sequence Variation and the Human Metabolome among African Americans in the Atherosclerosis Risk in Communities (ARIC) Study
Background: Chronic cardiovascular diseases result from the interaction of genes and the environment, and the human metabolome is an ultimate measured outcome of gene-environment interaction.
Methods: We designed a large-scale genomic study of the human serum metabolome in 1260 deeply phenotyped African-Americans (AAs) in the Atherosclerosis Risk in Communities (ARIC) Study. This analysis presents the results from 308 known serum metabolites, measured by GC-MS and LC-MS. To detect common and rare variations across the genome influencing the human metabolome we carried out both array-based GWAS analyses and whole exome sequencing analyses.
Results: We identified 19 distinct loci (p < 1.610-10) affecting the human metabolome. Depending on the particular metabolite, these loci were associated with 7-50% of the difference in metabolite levels per allele copy, with an average effect of 25%. To further promote novel discovery, we investigated unknown metabolite structures and disease etiology. For example, a missense mutation in N-acetyltransferase 8 (NAT8), rs13538 (MAF = 0.48, G → A) was significantly associated with N-acetylornithine (p = 4.010-66). We then examined this locus with 201 additional unnamed metabolites, and discovered it contributed to the unknown compounds X-11333 and X-11787 (p = 1.010-61 and p = 2.510-25, respectively). By a mass spec-based structural approach, X-11333 was determined to be N-acetyl-1-methylhistidine. Given that NAT8 is a known GWAS locus for chronic kidney disease, we assessed the association between these two N-acetyl amino acids and kidney function. A pronounced association with estimated glomerular filtration rate (eGFR) was found for both N-acetylornithine and N-acetyl-1-methylhistidine, whereby higher levels of N-acetyl molecules were associated with lower eGFR (p = 2.710-14 and p = 9.510-23, respectively). Moreover, the association persisted for incident chronic kidney disease (CKD), whereby higher levels of N-acetylornithine and N-acetyl-1-methylhistidine were associated with the onset of CKD (p = 0.004 and p = 0.02, respectively).
Conclusion: These studies demonstrate the utility of -omic studies in deeply phenotyped cohorts. We identified 19 loci that affected human metabolite levels among AAs. Our results provided new insights into gene function annotation and disease processes. Further use of -omic technologies will facilitate replication of our findings.
- © 2013 by American Heart Association, Inc.