Interplay between diet and gut microbiome, and circulating concentrations of trimethylamine N-oxide: findings from a longitudinal cohort of US men

Abstract

Objectives: Gut-produced trimethylamine N-oxide (TMAO) is postulated as a possible link between red meat intake and poor cardiometabolic health. We investigated whether gut microbiome could modify associations of dietary precursors with TMAO concentrations and cardiometabolic risk markers among free-living individuals.

Design: We collected up to two pairs of faecal samples (n=925) and two blood samples (n=473), 6 months apart, from 307 healthy men in the Men's Lifestyle Validation Study. Diet was assessed repeatedly using food-frequency questionnaires and diet records. We profiled faecal metagenome and metatranscriptome using shotgun sequencing and identified microbial taxonomic and functional features.

Results: TMAO concentrations were associated with the overall microbial compositions (permutational analysis of variance (PERMANOVA) test p=0.001). Multivariable taxa-wide association analysis identified 10 bacterial species whose abundance was significantly associated with plasma TMAO concentrations (false discovery rate <0.05). Higher habitual intake of red meat and choline was significantly associated with higher TMAO concentrations among participants who were microbial TMAO-producers (p<0.05), as characterised based on four abundant TMAO-predicting species, but not among other participants (for red meat, P-interaction=0.003; for choline, P-interaction=0.03). Among abundant TMAO-predicting species, Alistipes shahii significantly strengthened the positive association between red meat intake and HbA1c levels (P-interaction=0.01). Secondary analyses revealed that some functional features, including choline trimethylamine-lyase activating enzymes, were associated with TMAO concentrations.

Conclusion: We identified microbial taxa that were associated with TMAO concentrations and modified the associations of red meat intake with TMAO concentrations and cardiometabolic risk markers. Our data underscore the interplay between diet and gut microbiome in producing potentially bioactive metabolites that may modulate cardiometabolic health.

Keywords: cardiovascular disease; diet; epidemiology; intestinal microbiology; nutrition.

Figure 1.. Study design and basic characteristics of the study participants.

A. The overall study design, sample collections, and laboratory assays in the Men’s Lifestyle Validation Study. B. Basic characteristics of the study participants, by quartiles of plasma TMAO concentrations assessed in blood samples collected 6 months apart. In the combined dataset of all repeated collections, distribution of each characteristics was compared across TMAO quartiles (P was showed in the column^#) and between the two collections (* indicates the characteristics with a P<0.05) using a univariate generalized linear mixed-effects regression (for continuous variables) or a Chi-square test (for binary variables). Abbreviation: FFQ, food frequency questionnaire; 7DDR, 7-day diet records; TMAO, trimethylamine N-oxide; TC, total cholesterol; TG, triglycerides; HDL-C, high-density lipoprotein-cholesterol; hs-CRP, high-sensitivity C-reactive protein; HbA1c, hemoglobin A1c; MET, metabolic equivalent task; BMI, body mass index; AHEI, alternative healthy eating index; and Q1–Q4; 1^st – 4^th quartiles.

Figure 2.. Red meat and choline intake, microbial communities, and plasma TMAO levels.

A presents association of intakes of red meat and choline, and secondarily, other TMAO precursors, with circulating TMAO concentrations. Generalized linear mixed-effects regressions were adjusted for BMI, smoking, physical activity, calorie intake, cumulatively averaged alternate healthy eating index and/or fish intake (for associations between dietary variables other than fish and TMAO). *, P<0.05 and **, FDR<0.05. In B and C, principal coordinates analysis based on Bray-Curtis dissimilarity were performed using taxonomic data at the species level; the first 2 principal coordinates (PCo) were plotted by plasma TMAO levels and Firmicutes to Bacteroidetes ratio. A PERMANOVA test was used to evaluate the association between the overall microbial compositions and TMAO concentrations. D presents associations between intakes of red meat and choline with TMAO levels, stratified by PCo 1 or 2. In B-D, multivariable models were further adjusted for probiotic use, antibiotic use, and the Bristol stool chart.

Figure 3.. Taxonomic features associated with plasma TMAO concentrations.

A presents the phylogenetic tree for taxonomic features qualified for analysis, highlighting features significantly associated with plasma TMAO concentrations (solid stars with letter labels, FDR<0.05; solid circles, P<0.05). The outside rings denote associations of each species with TMAO levels and intakes of red meat and choline (red, positive associations; blue, inverse associations; color depth, statistical significance), and the abundance of each species are portrayed with grey bars. B compares the selected species for their associations with plasma TMAO concentrations, red meat and choline intake, and other potential TMAO precursors. Species associated with TMAO, or choline or red meat intake at P<0.01 were presented. Colors of cells indicate association coefficients, and asterisks denote association significance (** FDR<0.05, * P<0.05). Generalized linear mixed-effects regressions implemented in MaAsLin2 were adjusted for repeated measurements (participants ID as random intercept), age, BMI, smoking, physical activity, calorie intake, cumulatively averaged alternate healthy eating index, probiotic use, antibiotic use, colonoscopy or acid use in the past 2 months, Bristol stool chart, and/or fish intake (for analyses of TMAO).

Figure 4.. Associations between intakes of red meat or choline with plasma TMAO concentrations, among participants who were categorized as gut microbial TMAO- producers and others.

The microbial TMAO-producer phenotype was categorized based on 4 TMAO-predicting species, including E. hallii (presence), E. biforme (absence), R. hominis (presence), and A. shahii (presence). The y-axis and bars (and lines) indicate the SD-differences (and 95% confidence intervals) in TMAO concentrations that are associated with 1-serving increase in foods intake or 1-SD increase in choline intake or in biomarkers. Generalized linear mixed-effects models were adjusted for age, BMI, smoking, physical activity, calorie intake, cumulatively averaged alternate healthy eating index, probiotic use, antibiotic use, Bristol stool chart, and/or fish intake (for dietary variables other than fish).

Figure 5.. Red meat intake, TMAO levels, and cardiometabolic risk markers.

A presents the associations between plasma TMAO levels and cardiometabolic risk markers. Generalized linear mixed-effect regressions were adjusted for age, BMI, smoking, physical activity, calorie intake, cumulatively averaged alternate healthy eating index. B presented associations of red meat intake with levels of HDL-C and HbA1c, stratified by the microbial TMAO-producer phenotype and A. shahii. Multivariable models were further adjusted for fish intake, refined carbohydrate intake, fiber intake, probiotic use, antibiotic use, and the Bristol stool chart. TC, total cholesterol; HDLC, high-density lipoprotein cholesterol; TG, triglycerides; TC/HDLC, ratio of total to high-density lipoprotein cholesterol; CRP, C-reactive protein; HbA1c, hemoglobin A1c.