Abstract
Objective: The human metabolic profile is affected by host genetic, microbiome, environmental/lifestyle, and dietary factors which may all influence human raised blood pressure (BP). We aim to delineate metabolite-BP relationships by evaluating the effect of single-nucleotide polymorphisms (SNPs) mapped to enzyme-coding genes involved in metabolic pathways on this association. We specifically focus on human commensal urinary gut microbial co-metabolites.
Design and Method: Two cross-sectional datasets of over 4,000 individuals in total were used in a two-stage design. The first cohort uses two timed 24-hr urine samples for the discovery and replication of urinary metabolites associated with BP. The second cohort uses spot urine samples to validate the associations and subsequently investigates the influence of BP and other metabolic syndrome-related SNPs on these associations. Urine samples from both cohorts were measured with 600 MHz 1H Nuclear Magnetic Resonance spectroscopy. The SNPs were mapped to metabolites using a bioinformatic approach, which also incorporates the microbiome in establishing the metabolic distance between SNPs and metabolites through biochemical reactions and pathways.
Results: 11 urinary metabolites, replicated within the first cohort and validated in the second cohort (adjusted P-values 1.68 × 10–15 to 1.08 × 10–4) were associated with systolic BP (SBP); 9 of these were also associated with diastolic BP (DBP) in both cohorts as well as 2 further metabolites with DBP only (adjusted P-values 9.18 × 10–13 to 1.23 × 10–4). The analysis of the host-genomic (SNPs linked to GLYAT, KYAT3 and SULT13A genes) effect on gut microbial co-metabolites phenylacetylglutamine, hippurate and 4-cresylsulfate and their association with BP revealed a significant contribution of SNPs on the metabolite-BP association (P-values 1.08 × 10–133 to 2.87 × 10–29). Other metabolite-BP associations, such as for amino acids tyrosine, glycine, lysine, and leucine, also demonstrated a significant effect of individual SNPs (P-values 1.54 × 10–303 to 2.04 × 10–8) on the relationship. These associations were validated with null models of random SNPs with the same minor allele frequency also mapping to enzyme-coding genes, as well as a null model demonstrating that these metabolites have more SNPs linked to them than other metabolites with similar network topologies in other metabolic pathways.
Conclusions: We have shown that human commensal urinary gut microbial co-metabolites are reliably associated with BP differences in two cohorts, and we demonstrated a significant host-genetic component that contributes to these associations. These relationships indicate potential complex mechanisms involving host genetics, the gut microbiome, and metabolic changes associated with BP that may lead to new ways to prevent high BP.