Genome-wide association study of plasma amino acids and Mendelian randomization for cardiometabolic traits

Abstract

Plasma amino acids (AAs) have emerged as promising biomarkers for metabolic disorders, yet their causality remains unclear. We aimed to investigate the genetic determinants of AA levels in a cohort of 10,333 individuals and their causal effects on cardiometabolic traits using Mendelian randomization (MR). Plasma levels of 20 AAs were quantified using capillary electrophoresis mass spectrometry. Genome-wide association studies were conducted using BOLT-LMM and heritability estimation via LDSC analysis. Causal effects of AAs on 11 cardiometabolic traits were examined using two-sample MR analyses. We identified 85 locus-metabolite associations across 43 genes for 18 AAs, including 44 novel loci linked to metabolic genes. Heritability for AAs was estimated at 16%. MR analysis demonstrated cystine to positively associate with systolic blood pressure (SBP) (β = 0.056, SE = 0.010), while serine indicated protective effects on SBP (β = - 0.040, SE = 0.011), diastolic BP (β = - 0.044, SE = 0.010), and coronary artery disease (odds ratio 0.888, SE = 0.028). We identified potentially novel genetic loci associated with AA levels and demonstrated robust causal associations between several AAs and cardiometabolic traits. These findings reinforce the importance of AAs as potential biomarkers and therapeutic targets in cardiometabolic health.

Keywords: Amino acids; Cardiometabolic diseases; Genome-wide association study; Mendelian randomization; Metabolomics.

Fig. 1

Chromosomal map of the genomic risk loci associated with the 18 amino acids levels. PhenoGram (http://visualization.ritchielab.org/phenograms/plot) was used to create the map, and the novel genetic loci associated with amino acids reported in this study are depicted in red.

Fig. 2

List of pleiotropic loci associated with multiple amino acids and Spearman’s correlation coefficients between amino acid concentrations. The values in the upper table in the figure are coefficient βs (estimated from the GWAS summary), and those in the lower table are Spearman’s correlation coefficients. The dendrogram is based on the Euclidean distance calculated on the basis of Spearman’s correlation coefficient. Positive associations (β, Spearman’s correlation coefficient > 0) are highlighted in red, and negative associations (β, Spearman’s correlation coefficient < 0) are highlighted in blue. Ala, alanine; Arg, arginine; Asn, asparagine; Asp, aspartic acid; Gln, glutamine; Glu, glutamic acid; Gly, glycine; His, histidine; Ile, isoleucine; Leu, leucine; Lys, lysine; Met, methionine; Phe, phenylalanine; Pro, proline; Ser, serine; Thr, threonine; Trp, tryptophan; Tyr, tyrosine; Val, valine.

Fig. 3

Volcano plots of Mendelian randomizations illustrating causal associations between amino acids and cardiometabolic traits. (A) Cardiometabolic traits as continous variables (BMI, SBP, DBP, TG, LDL-C, HDL-C, and HbA1c) (B) Cardiometabolic traits as categorical variables (Hypertension, Hyperlipidemia, T2DM, and CAD). The x-axis represents either the beta estimate (β) or the odds ratio (OR), and the y-axis represents the negative logarithm of the P-value. Each point represents a single SNP associated with each amino acid indicated by color coding. Associations were estimated using two MR methods: inverse variance weighting (IVW) and weighted median (WM), and points were indicated in squares and triangles, respectively. Solid vertical lines indicate null effect (β = 0 for continuous traits, OR = 1 for categorical traits), and dashed horizontal lines represent the threshold of P-value at 0.05 based on crude P-values. Points shaded in light gray represent associations with P-values > 0.5. Annotations highlight amino acids with strong associations based on crude P-values, effect size magnitude, and consistency between methods. BMI, body mass index; CAD, coronary artery disease; DBP, diastolic blood pressure; HbA1c, hemoglobin A1c; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low density lipoprotein-cholesterol; SBP, systolic blood pressure; T2DM, type 2 diabetes; TG, triglyceride.

Fig. 4

Flowchart for comparing the GWAS summary data of this study with previous studies. eQTL, expression quantitative trait locus; GWAS, genome-wide association study; LD, link disequilibrium; MGWAS, metabolite genome-wide association study; SNP, single nucleotide polymorphism; TMM, Tohoku Medical Megabank.