Genetics meets metabolomics: a genome-wide association study of metabolite profiles in human serum

Abstract

The rapidly evolving field of metabolomics aims at a comprehensive measurement of ideally all endogenous metabolites in a cell or body fluid. It thereby provides a functional readout of the physiological state of the human body. Genetic variants that associate with changes in the homeostasis of key lipids, carbohydrates, or amino acids are not only expected to display much larger effect sizes due to their direct involvement in metabolite conversion modification, but should also provide access to the biochemical context of such variations, in particular when enzyme coding genes are concerned. To test this hypothesis, we conducted what is, to the best of our knowledge, the first GWA study with metabolomics based on the quantitative measurement of 363 metabolites in serum of 284 male participants of the KORA study. We found associations of frequent single nucleotide polymorphisms (SNPs) with considerable differences in the metabolic homeostasis of the human body, explaining up to 12% of the observed variance. Using ratios of certain metabolite concentrations as a proxy for enzymatic activity, up to 28% of the variance can be explained (p-values 10(-16) to 10(-21)). We identified four genetic variants in genes coding for enzymes (FADS1, LIPC, SCAD, MCAD) where the corresponding metabolic phenotype (metabotype) clearly matches the biochemical pathways in which these enzymes are active. Our results suggest that common genetic polymorphisms induce major differentiations in the metabolic make-up of the human population. This may lead to a novel approach to personalized health care based on a combination of genotyping and metabolic characterization. These genetically determined metabotypes may subscribe the risk for a certain medical phenotype, the response to a given drug treatment, or the reaction to a nutritional intervention or environmental challenge.

Figure 1. Schematic illustration of the role of intermediate phenotypes (IPs), such as metabolic traits, demonstrated at the examples of two genes that code for major enzymes of the long-chain fatty acid metabolism (FADS1 and LIPC).

We show that new information on the functional basis of the observed associations can be inferred from the biochemical properties of the affected metabolites. Moreover, both genes were previously reported to be associated with common clinical phenotypes, FADS1 in an extent which would not attract immediate attention for follow-up in a genome-wide context. Since several genes and pathways are involved in the development of a clinical endpoint, the IP focuses on one pathway (e.g., cholesterol or a given metabotype) which is already known to be involved in the clinical endpoint (e.g. coronary artery disease (CAD)). It is much easier to identify the genes which are associated with the IP since the associations of genetic variation with the IP is much stronger than with the clinical endpoint. Environmental factors interact at different levels with the IPs and thereby add to the variability in the system. The closer the IP is related to the genetic polymorphism, the stronger the association is expected to be. In our case the association reflects enzymatic activity of FADS1 and LIPC which results in very strong effect sizes of the genetically determined metabotype.

Figure 2. P-values of association assuming an additive genetic model, superposing the results obtained from all genome-wide tested metabolic traits.

Chromosomal location is indicated by different colors on the x-axis, negative logarithmic p-values are reported on the y-axis. The top ranking SNPs together with the closest gene and the most significant associating metabolite(s) are indicated. A complete list of all associations with p<10⁻⁶ is provided in Table S1, together with significant associations from previous GWA studies with medical phenotypes. Metabolite abbreviations are explained in the material and methods section and a full list of all measured metabolites is provided as supplementary data.

Figure 3. Boxplots of the metabolite concentrations of five top ranking associations as a function of genotype.

They show the differentiation of the population that is induced by these genetically determined metabotypes (0 = major allele homozygote, 1 = heterozygote, 2 = minor allele homozygote). Boxes extend from 1^st quartile (Q₁) to 3^rd quartile (Q₃); median is indicated as a horizontal line; whiskers are drawn to the observation that is closest to, but not more than, a distance of 1.5(Q₃-Q₁) from the end of the box. Observations that are more distant than this are shown individually on the plot. The number of individuals in each group is given below the boxes. P-values for these associations are given in Table 1.

Figure 4. Boxplots of the strongest associations of metabolite concentration ratios with polymorphisms in the FADS1 (A; p = 2.4×10⁻²²), SCAD (B; p = 9.3×10⁻¹⁷), and MCAD (C; p = 7.6×10⁻¹⁷) genes (see legend to Figure 3 for details).

The metabolic efficiencies of the reactions that are catalyzed by these three enzymes differ considerably between individuals of different genotype.