Skip to main content
NCBI home page
The American Journal of Clinical Nutrition logoLink to The American Journal of Clinical Nutrition
. 2017 Aug 30;106(4):1131–1141. doi: 10.3945/ajcn.116.150136

Effects of dietary sodium on metabolites: the Dietary Approaches to Stop Hypertension (DASH)–Sodium Feeding Study

Andriy Derkach 1, Joshua Sampson 1, Justin Joseph 1,2, Mary C Playdon 1, Rachael Z Stolzenberg-Solomon 1,✉
PMCID: PMC5611778  PMID: 28855223

Abstract

Background: High sodium intake is known to increase blood pressure and is difficult to measure in epidemiologic studies.

Objective: We examined the effect of sodium intake on metabolites within the DASH (Dietary Approaches to Stop Hypertension Trial)–Sodium Trial to further our understanding of the biological effects of sodium intake beyond blood pressure.

Design: The DASH-Sodium Trial randomly assigned individuals to either the DASH diet (low in fat and high in protein, low-fat dairy, and fruits and vegetables) or a control diet for 12 wk. Participants within each diet arm received, in random order, diets containing high (150 nmol or 3450 mg), medium (100 nmol or 2300 mg), and low (50 nmol or 1150 mg) amounts of sodium for 30 d (crossover design). Fasting blood samples were collected at the end of each sodium intervention. We measured 531 identified plasma metabolites in 73 participants at the end of their high- and low-sodium interventions and in 46 participants at the end of their high- and medium-sodium interventions (N = 119). We used linear mixed-effects regression to model the relation between each log-transformed metabolite and sodium intake. We also combined the resulting P values with Fisher’s method to estimate the association between sodium intake and 38 metabolic pathways or groups.

Results: Six pathways were associated with sodium intake at a Bonferroni-corrected threshold of 0.0013 (e.g., fatty acid, food component or plant, benzoate, γ-glutamyl amino acid, methionine, and tryptophan). Although 82 metabolites were associated with sodium intake at a false discovery rate ≤0.10, only 4-ethylphenylsufate, a xenobiotic related to benzoate metabolism, was significant at a Bonferroni-corrected threshold (P < 10−5). Adjustment for coinciding change in blood pressure did not substantively alter the association for the top-ranked metabolites.

Conclusion: Sodium intake is associated with changes in circulating metabolites, including gut microbial, tryptophan, plant component, and γ-glutamyl amino acid–related metabolites. This trial was registered at clinicaltrials.gov as NCT00000608.

Keywords: metabolomics, epidemiology, sodium intake, trial, high and low sodium intake, feeding trial, African American

INTRODUCTION

Most adults in the United States, including those with hypertension, consume sodium in excess of the recommended 2300 mg/d, with the mean daily sodium intake being ∼3500 mg (1). High sodium intake is known to be associated with high blood pressure. In 2013, the Institute of Medicine concluded that there was evidence to support an association between sodium intake and an increased risk of cardiovascular disease, stroke, and all-cause mortality in accordance with the known effects of sodium on blood pressure (2). However, associations between sodium intake and other health outcomes have been difficult to identify (2). To our knowledge, sodium intake is poorly measured in epidemiologic studies and this measurement error leads to inaccurate risk estimates (2). In addition, the specific biological effects of high sodium intake, aside from increased blood pressure, are not well understood (2).

We studied the effect of sodium intake on plasma metabolites among participants in the DASH (Dietary Approaches to Stop Hypertension Trial)–Sodium Trial (3), a landmark clinical feeding study. The DASH-Sodium Trial was designed to evaluate the effects of the DASH diet (high in fruits, vegetables, and low-fat dairy products) and 3 levels of sodium intake on blood pressure (3). As part of the study, each subject received, in random order (i.e., a crossover design), diets containing low (1150 mg/d), medium (2300 mg/d), and high (3450 mg/d) sodium (3). The trial demonstrated that lowering sodium intake significantly reduced blood pressure, independent of the dietary patterns (3). Our current study targets a different question and evaluates the effect of sodium intake on metabolomic profiles. This study offers an important step toward determining whether there are associations between sodium intake and health outcomes. First, strongly associated metabolites, if discovered, may eventually be used as surrogate measures of sodium consumption in large epidemiologic studies. Second, these associations might provide further insight into potential biological mechanisms linking dietary sodium intake and disease. We hypothesized that unique metabolic profiles were associated with sodium intake.

METHODS

Study population

The DASH-Sodium Trial (clinicaltrials.gov NCT00000608) was a multicenter, randomized feeding trial. The trial compared the effects of 2 dietary patterns and 3 sodium intake interventions on blood pressure. Details of the methods for the DASH-Sodium Trial were published previously (3–5). Briefly, the original study (conducted between 1997 and 1999) included 412 participants ≥22 y of age who had systolic and diastolic blood pressures between 120 and 159 and 80 and 95 mm Hg, respectively (3). Exclusion criteria included renal insufficiency, diabetes requiring insulin, poorly controlled hyperlipidemia, special dietary requirements, alcohol consumption of >14 drinks/wk, and use of antihypertension drugs or other medicines that would affect blood pressure. Four clinical centers and a coordinating center participated in the trial. All participants signed an informed consent form and the study was approved by the human subject committees of each center (3).

DASH-Sodium Trial participants were randomly assigned to consume either a control diet (i.e., similar to the typical diet in the United States) or the DASH diet (high in fruits, vegetables, and low-fat dairy products) for 12 wk (3, 4). Within their designated dietary pattern, each participant received, in random order, low-, medium-, and high-sodium versions of their diet for 30 d (crossover design, Supplemental Figure 1) (3, 4). The sodium levels designated as low, medium, and high were 50 nmol/d (1150 mg), 100 nmol/d (2300 mg), and 150 nmol/d (3450 mg), respectively. The highest amount of sodium was comparable to typical US intake, the intermediate amount corresponded to the upper range of public health recommendations, and the low amount was lower than current recommendations. Participants’ energy intake was adjusted to ensure that their weight remained constant during the study (3, 4). Per the study protocol, trained staff measured participants’ blood pressure using a random-zero sphygmomanometer while participants were seated (3). Fasting EDTA plasma samples were collected from participants at the end of each sodium intervention and were stored at −80°C at the National Heart, Lung, and Blood Institute repository.

Our metabolomic study includes a subset of 119 participants from the DASH trial. For 73 participants, we measured the metabolomic profiles in plasma collected after the high- and low-sodium interventions. These 73 participants included all individuals who had a low-sodium plasma sample that had not undergone a freeze-thaw cycle, a requirement for all samples used in our study. For the remaining 46 participants, we measured the profiles of plasma collected after the high- and medium-sodium interventions. Approximately half of participants were in each dietary pattern intervention arm (DASH diet, n = 60; typical American control diet, n = 59; Supplemental Figure 2).

Laboratory analysis

All blood samples were thawed, aliquoted, and processed in a controlled and consistent manner. Samples from the same participant were analyzed consecutively within batches, and we included 24 blinded replicate samples for quality control.

The samples were sent to Metabolon Inc. on dry ice. Plasma samples were assayed with untargeted ultra-HPLC coupled to tandem mass spectrometry (MS) and gas chromatography–MS (6). The liquid chromatography–MS portion of the platform was based on a Waters ACQUITY ultra-performance liquid chromatographer and a Thermo Scientific Q-Exactive high-resolution accurate mass spectrometer interfaced with a heated electrospray ionization source and an Orbitrap mass analyzer operated at 35,000 mass resolution. The gas chromatography–MS portion was analyzed on a Thermo Finnigan Trace DSQ fast-scanning single-quadrupole mass spectrometer using electron impact ionization and operated at unit mass resolving power (6). Peaks were identified via linkage to Metabolon’s known chemical reference library. Metabolon grouped the metabolites into 9 chemical classes (amino acids, carbohydrates, cofactors and vitamins, energy metabolites, lipids, nucleotide metabolites, peptides, and xenobiotics) and metabolic subpathways based on Kyoto Encyclopedia of Genes and Genomes classifications (7). Unknown chemical identities were tagged beginning with “X” followed by numbers.

In total, 1027 plasma metabolites were detected in the samples; 613 were chemically identified and 414 were unnamed. We excluded metabolites for which ≥50% of participants had metabolite values below the limit of detection. We focused on the 531 identified metabolites that exceeded the limit of detection in ≥50% of participants. However, in supplementary analyses, we report the statistically significant results for the unidentified metabolites (n = 340) that were present above this limit of detection. Metabolite peak intensity was normalized according to run day by dividing each metabolite observation by the median for that metabolite on that run day. Multiple approaches can validly be used to account for biomarkers below the limit of detection without compromising the associations observed (1–3). We chose to assign the minimum observed value. Metabolites below detection were assigned the minimum observed value for that metabolite. Metabolite measurements were highly reproducible, with a median (IQR) intraclass correlation coefficient of 0.84 (0.62–0.91).

Statistical analysis

We calculated means ± SDs and proportions for the characteristics of the study population shown in Table 1.

TABLE 1.

Baseline characteristics of DASH-Sodium participants (N = 119)1

Sodium intervention
Characteristic All subjects(N = 119) High and medium (n = 46) High and low (n = 73)
Age, y
 18–30 1 (0.8) 0 1 (1.4)
 31–55 76 (63.8) 28 (60.9) 48 (65.8)
 56–65 32 (27.0) 14 (30.4) 18 (24.7)
 >65 10 (8.4) 4 (8.7) 6 (8.2)
Women 65 (54.6) 25 (54.4) 40 (54.8)
Race/ethnicity
 Black or African American 59 (49.6) 4 (8.7) 55 (75.3)
 Non-Hispanic white or other 60 (50.4) 42 (91.3) 18 (24.7)
BMI, kg/m2 29.3 ± 4.1 28.7 ± 3.6 29.8 ± 4.4
Waist circumference, cm 100.8 ± 12.6 97.5 ± 11.1 102.9 ± 13.1
Blood pressure,2 mm Hg
 Systolic 136.0 ± 9.1 136.9 ± 9.9 135.4 ± 8.7
 Diastolic 85.5 ± 4.0 85.7 ± 4.7 85.4 ± 3.6
Hypertension3 47 (39.5) 23 (50.0) 24 (32.9)
Education
 High school graduate or less 19 (16.0) 6 (13.0) 13 (17.8)
 Some college 48 (40.3) 14 (30.4) 34 (46.6)
 College degree 23 (19.3) 8 (17.4) 15 (20.6)
 Postgraduate work/degree 29 (24.4) 18 (39.1) 11 (15.1)
Household income, $/y
 <30,000 34 (28.6) 17 (37.0) 17 (23.3)
 30,000–60,000 46 (38.7) 15 (32.6) 31 (42.5)
 >60,000 35 (29.4) 12 (26.1) 23 (31.5)
 Missing 4 (3.3) 2 (4.3) 2 (2.7)
Smoker
 Never 67 (56.3) 28 (60.9) 39 (53.4)
 Former 41 (34.5) 15 (32.6) 26 (35.6)
 Current 11 (9.2) 3 (6.5) 8 (11.0)
Open in a new tab
1

Values are means ± SDs or n (%). DASH, Dietary Approaches to Stop Hypertension Trial.

2

Blood pressure was the average of 3 screening measurements and 2 measurements during the run-in period.

3

Hypertension was defined as an average systolic blood pressure of 140–159 mm Hg or an average diastolic blood pressure of 90–95 mm Hg during the 3 screening visits.

In our evaluation of the sodium intervention, we first considered 2 groups of participants separately: those with plasma collected after the high- and low-sodium interventions and those with plasma collected after the high- and medium-sodium interventions. We evaluated the effect of sodium intervention with mixed-effect models. We designated Yji as the level of metabolite j in sample i, Xi as a binary indicator equaling 1 if sample i was collected after the lower-sodium diet (medium or low) or 0 if the sample was collected after the high-sodium diet, and Ci = [Ci1…Ci5]T as a vector of covariates that included categorical age (≤55 or >55 y), sex, race (African American, non-Hispanic white, or other), dietary pattern (DASH or control), and sodium intervention order (1, 2, or 3). For each metabolite, we then fit a separate model as follows using mixed-effects linear regression with a subject-specific random intercept:

graphic file with name ajcn150136equ1.jpg

We reported an estimate of the relative change (Rj = 10βj1) in metabolite level when switching from a high- to lower-sodium intervention. For example, an Rj of 1.2 suggests that switching from the high- to lower-sodium intervention increased the metabolite level by a factor of 1.2 or 20%. For metabolites detected in <50% of participants, we modeled the presence or absence of the metabolite with mixed-effects logistic regression.

We then considered all subjects together and assumed that log-metabolite levels changed linearly with sodium intake. We used a similar model to the one above but designated Xi = 0, 1/2, or 1, according to whether sample i was collected after the high-, medium-, or low-sodium interventions and we included an additional covariate for group (i.e., the covariate equaled 1 if the participant’s samples were collected after the high- and low-sodium interventions or 0 if the samples were collected after the high- and medium-sodium interventions). We estimated the relative change (Rj = 10βj1) in metabolite for a 100-nmol/d decrease in sodium (i.e., the difference between the high- and low-sodium interventions). We denote the P and Q values for Inline graphic from this model with Plin and Qlin, respectively, where the subscript “lin” indicates that the sodium level was assumed to have a linear effect on log-metabolite levels. The Q value effectively estimates the proportion of associations with P ≤ Plin that are likely to be false positives (8, 9). We tested for a nonlinear effect of sodium intake with a likelihood ratio test comparing 2 models, and we denote the resulting P and Q values from this likelihood ratio test with Pnl and Qnl, respectively. Here, Qnl is based only on metabolites with Qlin ≤ 0.1. The larger model allows each sodium diet to have its own effect by including sodium as a categorical variable (i.e., the mean log-metabolite for the high-, medium-, and low-sodium diets are βj0, βj0 + γ1, and βj0 + γ2, respectively). The smaller, nested model forces sodium amount to have a linear effect by including sodium as a continuous variable with Inline graphic (i.e., the mean log-metabolite for the high-, medium-, and low-sodium diets are βj0, βj0 + 0.5 γ2, and βj0 + γ2).

We next determined the association between metabolic pathways and sodium intake. The metabolites were divided into 38 defined pathways (Supplemental Table 1). For each pathway, we combined the P values of the included metabolites by Fisher’s method [Fisher’s statistic = Inline graphic]. We obtained the distribution of Fisher’s statistic under the null hypothesis of no association by permutation. We created 105 permuted datasets. For each permuted dataset, we randomly assigned the high- and lower-sodium designations to the 2 measures for every participant. We then calculated Fisher’s statistic for each permuted dataset. Finally, we reported the statistical significance of the pathway association by a pathway P value (Ppath), defined as the proportion of these 105 Fisher’s statistics that are below the value observed for the actual data.

As a second multimetabolite test, we identified the first 10 principal components (PCs) of the metabolomic profile using PC analysis and then tested whether each PC was associated with sodium intervention at the Bonferroni-adjusted significance of 0.005 (0.05/10) with the same linear mixed-effects models used for individual metabolites.

We also performed secondary or sensitivity analyses to test whether covariates (e.g., race, sex, dietary pattern, and age) modified the effect of sodium intake and whether change in blood pressure mediated the effect. Building on the model that assumed a linear trend (i.e., Inline graphic), we considered the following:

graphic file with name ajcn150136equ2.jpg

We then tested whether each covariate modified the effect of intervention (i.e.,Inline graphic). We reported the relative change in effect size (Inline graphic). We also fit the following model:

graphic file with name ajcn150136equ3.jpg

where the covariates Si and Di are the measured systolic and diastolic blood pressure at the end of each sodium intervention for the corresponding individual when the ith sample was collected, respectively. We also reported estimates of the direct effect (Inline graphic) for a 100-nmol/d decrease in sodium intake after adjusting for the mediating effects on blood pressure.

The Bonferroni significance method for adjusting P values when multiple hypotheses are being tested is considered very conservative and may lead to a high number of false negatives; therefore, we used a false discovery rate of 0.10 for statistical significance (Q value). The Bonferroni-adjusted α considering all of the comparisons including the stratified analyses was 3.14 × 10−5 (0.05/1593) for individual metabolites and 0.00044 (0.05/114) for the pathways. However, we noted that the Bonferroni-adjusted α for the combined analyses was 9.4 × 10−5 (0.05/531) for individual metabolites and 0.0013 (0.05/38) for metabolic pathways because this was the focus of our results and discussion.

All statistical analyses were performed using R programming language (R Foundation for Statistical Computing) (10).

RESULTS

Table 1 shows the characteristics of the study participants. Overall, the majority of participants were 31–55 y of age, were women, were not hypertensive, were never smokers, attained some college education, and had an income of $30,000–60,000/y. The mean BMI (in kg/m2) was 29.3 (overweight) and the mean systolic and diastolic blood pressure was 136.0 and 85.5 mm Hg, respectively. Compared with participants in the high- to medium-sodium group, participants in the high- to low-sodium groups were more often African American and had less education but were similar with respect to other characteristics. Compared with the high- to low-sodium group, participants in the high- to medium-sodium group had a larger proportion of clinically defined high blood pressure; however, both groups had similar mean systolic and diastolic blood pressure and the difference could be attributable to chance because of the small number of participants in the medium- to high-sodium group (n = 43).

Six metabolic pathways were associated with sodium intake at a Bonferroni-corrected threshold of 0.0013 (i.e., fatty acid, food component or plant, benzoate, γ-glutamyl amino acid, methionine, and tryptophan; Table 2). Although 82 metabolites were associated with sodium intake at a false discovery rate of 0.1 (Qlin ≤ 0.1), only 4-ethylphenylsufate, a xenobiotic related to benzoate metabolism, was significant at a Bonferroni-corrected threshold (Plin < 10−5) (Table 3). The relative change in metabolite levels was larger in participants switching from the high- to low-sodium interventions than the high- to medium-sodium interventions. For many metabolites (including our most strongly associated metabolite, 4-ethylphenysulfate), there was little change when participants switched from the high- to medium-sodium interventions. However, in general, there was not enough evidence to confidently reject the hypothesis that sodium intake was linearly related to log-metabolite levels (Q > 0.10; Supplemental Table 2). In secondary analyses, the effect of reducing sodium intake was not significantly modified by dietary pattern, sex, race, or age (Q > 0.90; Supplemental Table 3). The effect was not significantly altered after adjusting for blood pressure (Supplemental Tables 4 and 5), with 68 of the initial 82 metabolites remaining statistically significant (Qlin ≤ 0.10) after adjustment. Despite these strong and robust associations, the top 10 PCs were not significantly associated with sodium intake. Additional analyses on unknown metabolites detected 2 statistically significant metabolites (Plin < 10−5) that were highly correlated to γ-glutamyliosoleucine and γ-glutamylleucine (Supplemental Table 6); no metabolite detected in <50% of participants was significant at Q < 0.1.

TABLE 2.

Metabolic pathways associated with sodium intervention1

Sodium intervention, P value2
Pathway group Metabolites, n High and medium High and low All subjects
Fatty acid 15 0.10 0.00041 0.00007
Food component or plant group 27 0.57 0.00005 0.00010
Benzoate metabolism 15 0.01 0.00061 0.00013
γ-Glutamyl amino acid 11 0.55 0.00106 0.00017
Methionine metabolism 14 0.11 0.00051 0.00020
Tryptophan metabolism 18 0.23 0.00010 0.00020
Phenylalanine and tyrosine metabolism 28 0.24 0.017 0.002
Others3 53 0.17 0.022 0.006
Purine metabolism 17 0.16 0.026 0.007
Valine, leucine, and isoleucine metabolism 30 0.54 0.029 0.009
Sterol or steroid 36 0.75 0.008 0.010
Pentose metabolism 9 0.67 0.023 0.010
Alanine and aspartate metabolism 8 0.58 0.020 0.013
Glutamate metabolism 6 0.25 0.052 0.021
Glycolysis, gluconeogenesis, and pyruvate metabolism 7 0.48 0.13 0.04
Chemical group 15 0.97 0.08 0.04
Histidine metabolism 7 0.76 0.11 0.05
Fatty acid, dicarboxylate 13 0.10 0.21 0.06
Sphingolipid metabolism 10 0.90 0.07 0.06
Fatty acid, monohydroxy 13 0.37 0.25 0.07
Urea cycle 13 0.60 0.13 0.10
Long-chain fatty acid; PUFA [n–3 (ω-3) and n–6 (ω-6)] 14 0.04 0.27 0.10
Fructose, mannose, galactose, starch, and sucrose metabolism 7 0.22 0.17 0.10
Tocopherol metabolism 6 0.36 0.02 0.12
Lysolipid 31 0.92 0.23 0.12
Dipeptide group 9 0.26 0.40 0.12
Pyrimidine metabolism 12 0.77 0.12 0.12
Carnitine metabolism 12 0.56 0.40 0.18
Drug 5 0.03 0.50 0.20
Krebs or tricarboxylic acid cycle 6 0.48 0.22 0.23
Medium-chain fatty acid group 9 0.31 0.40 0.24
Long-chain fatty acid 15 0.14 0.81 0.36
Lysine metabolism 10 0.21 0.52 0.39
Monoacylglycerol 10 0.92 0.53 0.41
Glycine, serine, and threonine metabolism 10 0.65 0.73 0.62
Secondary bile acid metabolism 12 0.32 0.90 0.62
Primary bile acid metabolism 6 0.26 0.60 0.77
Xanthine metabolism 12 0.91 0.88 0.95
Open in a new tab
1

Pathways are based on the Kyoto Encyclopedia of Genes and Genomes and are described in Supplemental Table 1.

2

P values describe the significance of the association between sodium intake and metabolic pathway among individuals receiving the high- and medium-sodium interventions, high- and low-sodium interventions, and all subjects. The reported pathway-level P values combine metabolite-level P values obtained from mixed models adjusted for age, sex, race (African American, or non-Hispanic white or other), dietary pattern (Dietary Approaches to Stop Hypertension Trial or control), sodium intervention order (1, 2, or 3), and group (when using all subjects; e.g., high and low or high and medium) covariate. P values are not adjusted for multiple comparisons. Bonferroni-corrected significance for the 38 pathways was 0.05/114 = 0.00044 for all of the comparisons considering the stratified analyses and 0.05/38 = 0.0013 for the combined analyses alone.

3

“Others” includes categories with <5 metabolites that could not otherwise be categorized.

TABLE 3.

Metabolites associated with sodium intervention1

Sodium intervention
High to medium
 High to low
 All subjects w, 100-nm/d decrease
Pathway or metabolite Class R (95% CI) P value R (95% CI) P value R (95% CI) P value Q value Rank2
Fatty acid
 Isovalerate3 Lipid or amino acid 1.04 (0.98, 1.12) 0.20 0.55 (0.39, 0.78) 1.49 × 10−3 0.61 (0.47, 0.79) 2.82 × 10−4 0.01 7
 Butyrylcarnitine Lipid 1.07 (0.97, 1.18) 0.19 1.26 (1.02, 1.55) 7.08 × 10−3 1.27 (1.11, 1.46) 7.45 × 10−4 0.01 20
 Valerylcarnitine Lipid 1.40 (0.81, 2.42) 0.23 1.66 (1.13, 2.45) 1.28 × 10−2 1.69 (1.16, 2.47) 7.16 × 10−3 0.06 49
Food component or plant group
 4-Allylphenol sulfate Xenobiotic 1.02 (0.88, 1.18) 0.78 1.30 (1.12, 1.51) 7.35 × 10−4 1.26 (1.11, 1.44) 5.05 × 10−4 0.01 10
 Methyl glucopyranoside (α plus β) Xenobiotic 1.10 (0.93, 1.29) 0.27 1.29 (1.09, 1.53) 5.37 × 10−3 1.28 (1.11, 1.48) 9.02 × 10−4 0.01 24
 N-Acetylalliin Xenobiotic 0.84 (0.49, 1.44) 0.54 2.18 (1.48, 3.20) 9.33 × 10−5 1.91 (1.31, 2.78) 1.06 × 10−3 0.02 26
 Methyl indole-3-acetate4 Xenobiotic 1.13 (0.98, 1.3) 0.10 1.25 (1.07, 1.47) 6.40 × 10−3 1.25 (1.09, 1.43) 1.77 × 10−3 0.02 33
 Gluconate Xenobiotic 1.04 (0.96, 1.13) 0.32 1.14 (1.03, 1.25) 1.24 × 10−2 1.13 (1.05, 1.22) 2.55 × 10−3 0.03 34
 Homostachydrine Xenobiotic 1.08 (0.75, 1.56) 0.70 1.50 (1.15, 1.95) 4.60 × 10−3 1.47 (1.13, 1.91) 4.95 × 10−3 0.04 44
 Ergothioneine Xenobiotic 1.01 (0.81, 1.27) 0.92 1.45 (1.06, 1.98) 1.74 × 10−2 1.39 (1.08, 1.77) 1.02 × 10−2 0.07 55
Benzoate metabolism
 4-Ethylphenysulfate Xenobiotic 0.96 (0.82, 1.13) 0.61 2.03 (1.67, 2.47) 2.03 × 10−9 1.78 (1.51, 2.11) 5.85 × 10−10 2.24 × 10−7 1
 4-Methycatechol sulfate Xenobiotic 1.09 (0.90, 1.32) 0.36 1.28 (1.06, 1.54) 1.20 × 10−2 1.23 (1.06, 1.43) 7.87 × 10−3 0.06 51
 O-Methylcatechol sulfate Xenobiotic 1.26 (1.00, 1.58) 0.06 1.26 (0.99, 1.60) 5.39 × 10−2 1.29 (1.05, 1.58) 1.45 × 10−2 0.08 66
γ-Glutamyl amino acid
 γ-Glutamylvaline Peptide 0.99 (0.94, 1.03) 0.55 0.59 (0.43, 0.80) 1.45 × 10−3 0.63 (0.51, 0.79) 1.08 × 10−4 0.01 2
 γ-Glutamylisoleucine Peptide 0.98 (0.93, 1.03) 0.35 0.59 (0.43, 0.80) 1.57 × 10−3 0.63 (0.51, 0.79) 1.09 × 10−4 0.01 3
 γ-Glutamylleucine Peptide 0.98 (0.93, 1.03) 0.48 0.65 (0.51, 0.84) 1.95 × 10−3 0.69 (0.58, 0.83) 1.46 × 10−4 0.01 5
 γ-Glutamylmethionine Peptide 0.94 (0.85, 1.04) 0.22 0.78 (0.66, 0.91) 3.22 × 10−3 0.79 (0.70, 0.90) 4.40 × 10−4 0.01 9
 γ-Glutamylglutamate Peptide 0.98 (0.86, 1.13) 0.81 0.40 (0.22, 0.73) 4.10 × 10−3 0.45 (0.29, 0.7) 5.43 × 10−4 0.01 12
 γ-Glutamylphenylalanine Peptide 0.97 (0.94, 1.01) 0.17 0.82 (0.71, 0.95) 9.07 × 10−3 0.84 (0.76, 0.94) 1.75 × 10−3 0.02 32
 γ-Glutamyltyrosine Peptide 0.97 (0.89, 1.05) 0.46 0.84 (0.74, 0.95) 7.43 × 10−3 0.86 (0.78, 0.95) 3.19 × 10−3 0.03 37
Methionine metabolism
 Methionine sulfone Amino acid 1.09 (0.98, 1.21) 0.10 1.13 (1.05, 1.22) 1.63 × 10−3 1.14 (1.06, 1.22) 5.20 × 10−4 0.01 11
 α-Ketobutyrate Amino acid 1.08 (0.87, 1.34) 0.48 1.86 (1.18, 2.93) 1.17 × 10−2 1.75 (1.25, 2.46) 1.63 × 10−3 0.02 31
 S-Adenosylhomocysteine Amino acid 0.91 (0.78, 1.07) 0.28 1.35 (1.11, 1.63) 3.91 × 10−3 1.26 (1.08, 1.47) 4.48 × 10−3 0.04 40
 Methionine sulfoxide Amino acid 1.02 (0.95, 1.09) 0.63 0.67 (0.48, 0.92) 2.03 × 10−2 0.71 (0.56, 0.90) 6.19 × 10−3 0.05 48
 N-Formylmethionine Amino acid 0.96 (0.90, 1.03) 0.25 1.15 (1.04, 1.28) 1.79 × 10−2 1.12 (1.03, 1.22) 8.06 × 10−3 0.06 52
 N-Acetylmethionine Amino acid 0.97 (0.90, 1.05) 0.49 1.13 (1.03, 1.24) 2.09 × 10−2 1.10 (1.02, 1.19) 1.37 × 10−2 0.08 63
 Methionine Amino acid 0.94 (0.87, 1.02) 0.13 1.18 (1.04, 1.34) 2.64 × 10−2 1.13 (1.03, 1.25) 1.58 × 10−2 0.09 67
Tryptophan metabolism
 Indoleacetate Amino acid 1.07 (0.97, 1.18) 0.21 1.15 (1.05, 1.26) 2.93 × 10−3 1.15 (1.06, 1.24) 6.49 × 10−4 0.01 17
 Indolebutyrate Amino acid 1.05 (0.85, 1.30) 0.65 1.27 (1.13, 1.44) 1.95 × 10−4 1.25 (1.10, 1.43) 8.96 × 10−4 0.01 23
 Methyl indole-3-acetate4 Amino acid 1.13 (0.98, 1.30) 0.10 1.25 (1.07, 1.47) 6.40 × 10−3 1.25 (1.09, 1.43) 1.77 × 10−3 0.02 33
 Tryptophan betaine Amino acid 1.11 (0.98, 1.25) 0.09 1.15 (1.04, 1.28) 9.85 × 10−3 1.15 (1.05, 1.25) 3.05 × 10−3 0.03 36
 Indoleacetylglutamine Amino acid 0.96 (0.80, 1.15) 0.64 1.27 (1.08, 1.49) 4.82 × 10−3 1.22 (1.06, 1.41) 7.30 × 10−3 0.06 50
 C-Glycosyl tryptophan Amino acid 1.01 (0.93, 1.08) 0.88 1.08 (1.02, 1.14) 1.05 × 10−2 1.07 (1.02, 1.13) 9.59 × 10−3 0.07 54
Phenylalanine and tyrosine metabolism
 4-Hydroxyphenylpyruvate Amino acid 1.34 (1.02, 1.76) 0.04 1.72 (1.22, 2.44) 2.10 × 10−3 1.76 (1.33, 2.34) 1.39 × 10−4 0.01 4
 N-Formylphenylalanine Amino acid 0.92 (0.77, 1.11) 0.39 0.76 (0.59, 0.97) 4.32 × 10−2 0.76 (0.63, 0.92) 5.58 × 10−3 0.05 47
 Phenyllactate Amino acid 1.01 (0.97, 1.06) 0.65 1.06 (1.01, 1.12) 2.74 × 10−2 1.06 (1.01, 1.11) 1.08 × 10−2 0.07 57
 Phenylpyruvate Amino acid 0.93 (0.58, 1.48) 0.75 1.55 (1.16, 2.07) 4.46 × 10−3 1.44 (1.06, 1.96) 1.97 × 10−2 0.098 77
Others (pathway)
 Acisoga (polyamine metabolism) Amino acid 1.04 (0.97, 1.11) 0.25 1.10 (1.04, 1.17) 2.15 × 10−3 1.10 (1.04, 1.16) 6.10 × 10−4 0.01 16
 HWESASXX (polypeptide) Peptide 1.11 (0.97, 1.26) 0.13 0.58 (0.42, 0.8) 1.65 × 10−3 0.64 (0.50, 0.82) 6.64 × 10−4 0.01 18
 Prostaglandin E2 (eicosanoid) Lipid 1.00 (0.91, 1.09) 0.92 0.39 (0.20, 0.77) 7.65 × 10−3 0.43 (0.27, 0.7) 7.96 × 10−4 0.01 22
 Leukotriene B4 Lipid 1.06 (0.97, 1.14) 0.19 0.26 (0.10, 0.68) 8.00 × 10−3 0.30 (0.15, 0.6) 9.55 × 10−4 0.01 25
 4-Guanidinobutanoate (guanidino and acetamido metabolism) Amino acid 0.82 (0.58, 1.18) 0.29 0.51 (0.32, 0.83) 7.61 × 10−3 0.55 (0.38, 0.81) 2.86 × 10−3 0.03 35
 Erythrulose (advanced glycation end product) Carbohydrate 1.28 (1.07, 1.55) 0.01 1.19 (0.99, 1.43) 6.14 × 10−2 1.27 (1.08, 1.49) 3.60 × 10−3 0.04 39
 l-Urobilin (hemoglobin and porphyrin metabolism) Cofactors and vitamins 1.09 (0.72, 1.64) 0.69 0.61 (0.45, 0.83) 2.24 × 10−3 0.66 (0.49, 0.87) 4.74 × 10−3 0.04 41
 Bradykinin, hydroxy-pro(3) (polypeptide) Peptide 0.85 (0.61, 1.18) 0.34 0.54 (0.30, 0.95) 3.62 × 10−2 0.55 (0.35, 0.85) 8.13 × 10−3 0.06 53
 Bilirubin (Z,Z) (hemoglobin and porphyrin metabolism) Cofactors and vitamins 1.09 (0.67, 1.77) 0.74 1.84 (1.08, 3.12) 2.74 × 10−2 1.77 (1.14, 2.75) 1.21 × 10−2 0.08 59
 Creatinine (creatine metabolism) Amino acid 0.97 (0.86, 1.10) 0.68 1.15 (1.04, 1.28) 9.24 × 10−3 0.46 (0.24, 0.88) 1.40 × 10−2 0.08 64
 Myo-inositol (inositol metabolism) Lipid 1.03 (0.98, 1.09) 0.24 1.08 (0.99, 1.17) 8.90 × 10−2 1.08 (1.02, 1.15) 1.43 × 10−2 0.08 65
 4-Acetamidobutanoate (polyamine metabolism) Amino acid 1.03 (0.92, 1.16) 0.62 1.15 (1.01, 1.31) 3.29 × 10−2 1.12 (1.02, 1.23) 1.92 × 10−2 0.10 75
 HWESASLLR (polypeptide) Peptide 0.96 (0.88, 1.04) 0.33 0.46 (0.20, 1.10) 8.42 × 10−2 0.46 (0.24, 0.88) 1.95 × 10−2 0.10 76
 Erythronate (aminosugar metabolism) Carbohydrate 1.01 (0.97, 1.04) 0.75 0.93 (0.87, 1.00) 4.22 × 10−2 0.94 (0.89, 0.99) 2.14 × 10−2 0.10 80
Purine metabolism
 Urate Nucleotide 1.03 (0.99, 1.07) 0.17 1.07 (1.03, 1.11) 1.62 × 10−3 1.06 (1.03, 1.1) 2.35 × 10−4 0.01 6
 N6-Carbamoylthreonyladenosine Nucleotide 1.05 (0.94, 1.16) 0.39 1.17 (1.07, 1.28) 9.08 × 10−4 1.15 (1.07, 1.24) 4.36 × 10−4 0.01 8
Valine, leucine, and isoleucine metabolism
 Isovalerate3 Amino acid 1.04 (0.98, 1.12) 0.20 0.55 (0.39, 0.78) 1.49 × 10−3 0.61 (0.47, 0.79) 2.82 × 10−4 0.01 7
 2-Methylbutyrylcarnitine Amino acid 0.97 (0.89, 1.06) 0.48 1.18 (1.05, 1.34) 7.71 × 10−3 1.11 (1.04, 1.19) 3.39 × 10−3 0.03 38
 β-hydroxyisovaleroylcarnitine5 Amino acid 1.00 (0.95, 1.06) 0.93 1.07 (1.01, 1.14) 2.08 × 10−2 1.07 (1.01, 1.12) 1.36 × 10−2 0.08 61
 Ethylmalonate Amino acid 1.04 (0.99, 1.09) 0.15 1.06 (0.98, 1.15) 1.60 × 10−1 1.07 (1.01, 1.13) 2.25 × 10−2 0.10 82
 Isobutyrylcarnitine Amino acid 1.08 (0.98, 1.18) 0.13 1.11 (0.99, 1.24) 8.64 × 10−2 1.11 (1.02, 1.22) 2.13 × 10−2 0.10 79
Sterol or steroid
 Etiocholanolone glucuronide Lipid 1.05 (0.95, 1.16) 0.34 1.19 (1.07, 1.32) 2.15 × 10−3 1.17 (1.07, 1.28) 6.03 × 10−4 0.01 14
 4-Androsten-3β,17β-diol monosulfate(1) Lipid 1.00 (0.96, 1.04) 0.91 1.10 (1.04, 1.17) 1.53 × 10−3 1.08 (1.04, 1.14) 7.96 × 10−4 0.01 21
 5α-Androstane-3β, 17β-diol monosulfate Lipid 1.00 (0.93, 1.08) 0.95 1.11 (1.05, 1.18) 5.38 × 10−4 1.1 (1.04, 1.16) 1.06 × 10−3 0.02 27
 Andro steroid monosulfate(1) Lipid 0.93 (0.78, 1.10) 0.39 1.34 (1.16, 1.55) 1.51 × 10−4 1.26 (1.10, 1.44) 1.26 × 10−3 0.02 28
 Epiandrosterone sulfate Lipid 1.01 (0.95, 1.06) 0.81 1.09 (1.02, 1.17) 9.60 × 10−3 1.08 (1.02, 1.14) 5.51 × 10−3 0.05 46
 Androsterone sulfate Lipid 1.02 (0.98, 1.07) 0.32 1.07 (1.01, 1.15) 3.79 × 10−2 1.07 (1.02, 1.13) 1.16 × 10−2 0.08 58
 5α-Androstane-3-α,17β-diol disulfate Lipid 1.07 (0.88, 1.29) 0.51 1.21 (1.03, 1.41) 2.10 × 10−2 1.19 (1.03, 1.37) 1.92 × 10−2 0.10 74
Pentose metabolism
 Arabitol Carbohydrate 0.97 (0.90, 1.04) 0.39 1.16 (1.07, 1.25) 3.97 × 10−4 1.11 (1.04, 1.18) 1.32 × 10−3 0.02 29
Alanine and aspartate metabolism
 Asparagine Amino acid 1.03 (0.96, 1.11) 0.47 1.19 (1.06, 1.34) 4.33 × 10−3 1.17 (1.07, 1.28) 6.07 × 10−4 0.01 15
Glutamate metabolism
 N-Acetylglutamate Amino acid 1.34 (1.03, 1.74) 0.03 1.25 (0.99, 1.57) 6.25 × 10−2 1.30 (1.06, 1.60) 1.34 × 10−2 0.08 60
 Glutamine Amino acid 1.03 (0.97, 1.10) 0.34 1.12 (0.96, 1.30) 1.52 × 10−1 1.13 (1.02, 1.25) 1.86 × 10−2 0.10 72
Glycolysis, gluconeogenesis, and pyruvate metabolism
 Glucose Carbohydrate 1.01 (0.99, 1.04) 0.24 1.03 (1.00, 1.06) 2.35 × 10−2 1.03 (1.01, 1.06) 4.91 × 10−3 0.04 43
Chemical group
 O-Sulfo-l-tyrosine Xenobiotics 1.00 (0.95, 1.05) 1.00 1.09 (1.03, 1.15) 2.04 × 10−3 1.08 (1.03, 1.13) 6.76 × 10−4 0.01 19
 Phenylcarnitine Xenobiotics 1.13 (0.97, 1.31) 0.12 1.14 (0.99, 1.30) 6.55 × 10−2 1.15 (1.03, 1.3) 1.90 × 10−2 0.10 73
Histidine metabolism
 trans-Urocanate Amino acid 1.04 (0.93, 1.16) 0.48 1.14 (1.02, 1.28) 2.50 × 10−2 1.13 (1.03, 1.23) 1.37 × 10−2 0.08 62
 Imidazole lactate Amino acid 1.04 (0.92, 1.18) 0.53 1.14 (1.02, 1.28) 2.48 × 10−2 1.13 (1.02, 1.26) 1.78 × 10−2 0.10 70
Fatty acid, dicarboxylate
 3-Carboxy-4-methyl-5-propyl-2-furanpropanoate Lipid 1.05 (0.96, 1.14) 0.29 1.19 (1.06, 1.32) 2.86 × 10−3 1.17 (1.07, 1.27) 5.46 × 10−4 0.01 13
Sphingolipid metabolism
 Myristoyl sphingomyelin Lipid 1.02 (0.92, 1.14) 0.68 1.13 (1.03, 1.24) 8.79 × 10−3 1.13 (1.04, 1.22) 4.88 × 10−3 0.04 42
Fatty acid, monohydroxy
 13-hode, 9-hode Lipid 1.01 (0.93, 1.10) 0.78 0.56 (0.36, 0.85) 9.07 × 10−3 0.60 (0.44, 0.82) 1.53 × 10−3 0.02 30
 5-Hydroxyhexanoate Lipid 1.16 (0.98, 1.38) 0.09 1.15 (0.97, 1.36) 1.05 × 10−1 1.17 (1.03, 1.33) 1.83 × 10−2 0.10 71
Urea cycle
 Citrulline Amino acid 1.00 (0.92, 1.1) 0.95 1.08 (1.02, 1.15) 1.57 × 10−2 1.08 (1.01, 1.14) 1.77 × 10−2 0.10 69
Fructose, mannose, galactose, starch, and sucrose metabolism
 Maltotriose Carbohydrate 1.24 (0.79, 1.94) 0.36 1.87 (1.17, 3.00) 1.14 × 10−2 1.78 (1.20, 2.64) 5.16 × 10−3 0.04 45
Lysolipid
 1-Linoleoylglycerophosphoinositol Lipid 1.06 (0.97, 1.15) 0.20 1.12 (1.00, 1.26) 4.77 × 10−2 1.13 (1.03, 1.23) 1.05 × 10−2 0.07 56
Dipeptide group
 cys-Gly, oxidized (glutathione metabolism) Amino acid 1.10 (0.82, 1.46) 0.52 1.59 (1.00, 2.52) 5.43 × 10−2 1.54 (1.07, 2.21) 2.03 × 10−2 0.10 78
Lysine metabolism
 Glutarylcarnitine (C5) Amino acid 0.96 (0.93, 1.00) 0.05 1.06 (1.02, 1.10) 4.47 × 10−3 1.04 (1.01, 1.08) 2.18 × 10−2 0.10 81
Secondary bile acid metabolism
 Ursodeoxycholate Lipid 0.66 (0.37, 1.19) 0.17 0.52 (0.27, 0.99) 5.19 × 10−2 0.51 (0.30, 0.88) 1.62 × 10−2 0.09 68
Open in a new tab
1

The relative change (R) in metabolite level is reported when switching from the high- to medium-sodium intervention and the high- to low-sodium intervention and when decreasing sodium intake by 100 nm/d (e.g., assuming a linear relation and combining data from all participants). R and its associated 95% CI, P value, and Q value were calculated using mixed-effect models adjusted for age, sex, race (African American, or non-Hispanic white or other), dietary pattern (Dietary Approaches to Stop Hypertension Trial or control), sodium intervention order (1, 2, or 3) and group (when using all subjects; e.g., high-low or high-medium) covariate. Metabolites with Qlin < 0.1 are included. The P values were not adjusted for multiple comparisons. The Bonferroni-adjusted α level considering all the comparisons including the stratified analyses was 3.14 × 10–5 (0.05/1593) and is 9.4 × 10–5 (0.05/531) for the combined analyses alone.

2

Relative rank of metabolites based on Plin.

3

Isovalerate is in the fatty acid and the valine, leucine, and isoleucine metabolism group pathway.

4

Methyl indole-3-acetate is in the tryptophan metabolism and the food component pathways.

5

β-Hydroxyisovaleroylcarnitine is also in carnitine metabolism group.

The most strongly associated pathway was the fatty acid group, in which the short-chain fatty acid isovalerate decreased and the lipids butyrylcarnitine and valerylcarnitine increased with sodium reduction. Seven xenobiotic metabolites in the food component or plant group increased with sodium reduction, including 4-allylphenol sulfate, methyl glucopyranoside (α plus β), N-acetylalliin, methyl indole-3-acetate, gluconate, homostachydrine, and erothioneine (Table 3). Three xenobiotic metabolites increased with low sodium intake within the benzoate metabolism pathway, including 4-ethylphenysulfate, which reached Bonferroni significance (Plin < 5.85 × 10−10); other significant metabolites in this pathway included 4-methylcatechol sulfate and O-methylcatechol sulfate. In the γ-glutamyl amino acid group, 7 peptide metabolites (γ-glutamylvaline, γ-glutamylisoleucine, γ-glutamylleucine, γ-glutamylmethionine, γ-glutamylglutamate, γ-glutamylphenylalanine, and γ-glutamyltyrosine) decreased with lower sodium intake. In the methionine metabolism pathway, 6 amino acid metabolites (methionine sulfone, α-ketobutyrate, S-adenosylhomocysteine, N-formylmethionine, N-acetylmethionine, and methionine) increased, whereas amino acid methionine sulfoxide decreased with the low-sodium intervention. Finally, in the tryptophan group, 6 amino acid metabolites (namely, indoleacetate, indolebutyrate, methyl indole-3-acetate, tryptophan betaine, indoleacetylglutamine, and C-glycosyl tryptophan) increased with lower sodium intake. The results for all 531 known metabolites contributing to each pathway are shown in Supplemental Table 7.

A few additional metabolites were among the most strongly associated metabolites (P < 0.0005). 4-Hydroxyphenylpuruvate, an amino acid related to phenlyanalanine and tyrosine metabolism, increased with the low-sodium intervention, as did 2 nucleotides in the purine metabolism pathway (urate and N6-carbamoylthreonyladenosine).

DISCUSSION

In this controlled feeding study, we observed that changes in sodium intake resulted in numerous changes in plasma metabolite concentrations. The strongest associations were observed in the fatty acid, food plant component, benzoate, methionine, and tryptophan pathways with metabolites that mostly increased with the low-sodium intervention and metabolites in the γ-glutamyl amino acid pathway that decreased with sodium restriction. The significant relative changes in metabolites were primarily driven by the high- to low-sodium interventions. 4-Ethylphenysulfate was associated with changes in sodium intake after correction for multiple comparisons. The most significant metabolite changes were not mediated by coinciding changes in blood pressure and adjustment for blood pressure did not substantially change their associations.

Our study provides evidence that suggests that sodium intake may affect the gut microbiome. 4-Ethylphenylsulfate increased with sodium restriction and was the most strongly associated with change in sodium intake. This metabolite is produced by the gut microflora (11), and higher serum concentrations are associated with various health outcomes such as periodontal disease and lean body mass in animals or humans (12–16). 4-Ethylphenylsulfate is correlated with the consumption of soy products (17, 18) and is elevated in the plasma of vegans compared with omnivores; a vegan diet increases concentrations of several products produced by the gut microbiota (19). Salt has known antibacterial properties (20) and has been used for medicinal purposes for centuries (21, 22). Others have suggested that both diet and the microbiome mediate associations between metabolites and blood pressure (23). Although speculative, the effect of sodium restriction on the gut microbiome may contribute to the increases in 4-ethylphenylsulfate and possibly to changes in other metabolites (e.g., 4-hydroxyphenylpyruvate, indole-related metabolites, and isovalerate) produced by gut bacterial metabolism (11).

Metabolites in the γ-glutamyl amino acid metabolite group decreased with lower sodium intake. These metabolites are formed using the enzyme γ-glutamyl transferase, which transfers a γ-glutamyl moiety from glutathione to amino acids and peptides (24). γ-Glutamyltransferase is a cell-surface protein that is important for maintaining glutathione homeostasis and its activity increases in response to oxidative stress (25). In an in vitro experiment, the sodium chloride concentration was directly proportional to γ-glutamyl transferase activity (26). To our knowledge, most epidemiologic studies have shown blood concentrations of γ-glutamyltransferase to be positively associated with prehypertension (27), hypertension (28), and cardiovascular disease (29, 30). There is evidence that excess salt intake induces oxidative stress in salt-sensitive (SS) hypertension (31). Although speculative, it is possible that compared with high sodium intake, sodium restriction decreases γ-glutamyl transferase activity via reduction of oxidative stress and accounts for the significant decreases in γ-glutamyl amino acid metabolites we observed.

Metabolites within the tryptophan metabolism pathway, particularly indole-related metabolites, significantly increased with sodium restriction. Although some endogenous production occurs in mammals, commensal enteric bacteria catabolize tryptophan to indole compounds and derivatives, which are absorbed passively through the colonic epithelia into the portal circulation (32). Indole and its metabolites are metabolized through several different pathways that have distinct biological activity, including maintaining mucosal reactivity and homeostasis, expression of pro- and anti-inflammatory genes in intestinal epithelial cells, and modulating the release of intestinal glucagon-like peptide-1 that acts to signal satiety, reducing food intake and obesity (32). 4-Hydroxyphenylpyruvate was also significantly increased with sodium restriction in our study. 4-Hydroxyphenylpyruvate is a keto acid that is an intermediate in the metabolism of phenylalanine and synthesized from tyrosine. Studies of patients with disorders of tyrosine metabolism who are treated with the drug nitisinone suggest that 4-hydrophenylpyruvate can stimulate the production of indole metabolites from tryptophan via tryptophan aminotransferase (33). More research is needed to understand how sodium intake impacts tryptophan-related metabolites, 4-hydroxyophenylpyruvate, and the biochemical relation between these metabolites.

A few additional metabolites that changed with sodium restriction deserve mention. The increase in food component or plant group metabolites could be related to the low-sodium diet food preparation. For example, 4-allylphenol sulfate (chavicol) and homostachydrine are found in herbs and are flavorings that may have been added to enhance the palatability of the low-sodium diet. It is uncertain how sodium restriction contributes to the observed reduction in the short-chain fatty acid isovalerate. Isovalerate is a degradation product of the branched-chain amino acid valine and is modulated by both mammalian and bacteria metabolism (11). Alterations in the gut microbiota owing to changes in salt intake might alter carbohydrate fermentation and contribute to increased production of short-chain fatty acids such as isovalerate (34). Urate (a purine derivative) increased with sodium restriction, as observed previously (35). Humans produce only small quantities of uric acid, with excess accumulation leading to gout (36). Uric acid is also a natural antioxidant produced endogenously that may protect against oxygen radicals (37). HWESASXX (a peptide linked to inflammation and positively associated with blood pressure) (38), and 2 eicosanoids (prostaglandin E2 and leukotriene B4) decreased with sodium restriction. Eicosanoids are known to play a role in the regulation of blood pressure (39).

Three previous studies (40–42), 2 in humans and 1 in mice, examined the association between sodium intake or salt sensitivity and metabolites. First, in a 10-wk double-blind randomized crossover study of 17 middle-aged adults with elevated blood pressure, participants were instructed to consume a low-sodium diet and were given both a placebo and a slow-release NaCl (150 nmol/d) tablet. Metabolites were measured in 24-h urine samples collected at the end of each intervention. Sodium restriction was associated with increases in succinate, methionine sulfoxide, S-adenosylhomocysteine, d-gluconate, and asparagine (40). Our study of plasma metabolites replicated the increases in succinate, gluconate, and asparagine, but we did not observe the same direction of associations for succinate or methionine sulfoxide. Another study of 13 treated patients with hypertensive heart failure compared 152 metabolites measured in 24-h urine samples at baseline and after participants consumed a DASH sodium-restricted diet (50 nmol/d) for 21 d (41). Similar to our study, this study showed decreases in isovalerate with the sodium-restricted diet (41). Finally, an experimental study that explored differences in metabolites owing to genetic variation in Dahl SS rats compared with normotensive consomic SS.13 rats showed that 23 metabolites differed (P < 0.05) between the 2 animal groups (42). Metabolites that were intermediates or linked (glycolysis) to the tricarboxylic acid cycle (cis-aconitate, isocitrate, citrate, succinate, fumarate, and pyruvate) were upregulated in SS rats (42). Valine, isoleucine, methionine, and urea were increased in SS rats, whereas glutamate, asparagine, glycine, and tyrosine were decreased (42). We observed significant associations for some of these metabolites or molecules in related pathways (e.g., succinate, methionine, glutamate, urea, asparagine, branched-chain amino acids, glucose, and tyrosine) (42).

This study has several strengths. First, this was a carefully conducted feeding study. Dietary intake was controlled, the participants maintained a constant weight, and nonadherence was minimized to facilitate measuring only the true biological effects of sodium intake (3). The sodium intervention was a crossover design: each participant served as his or her own control, diminishing the effects of between-person genetic variation and confounding by race, sex, age, and other exposures, thereby increasing study power. Blood pressure was measured in a standard manner, allowing us to attempt to untangle the effect of sodium intake from that of changes in blood pressure. Given the original design of the DASH-Sodium Trial, our results are likely generalizable to many people in the United States (3). The absence of the PC analysis not being associated with sodium intake suggests that sodium is not the strongest factor driving the variability of the metabolites. Correctly identifying metabolites is important for biochemical interpretation. Although they were not subsequently validated with pure compounds, the methods used to identify metabolites in our study are based on a multiple orthogonal criteria linkage to a unit mass spectral library built from authentic standards, which has been shown to accurately identify metabolites (6, 43). Limitations of our study include its relatively modest sample size, the lack of an independent replication sample, and the inability to directly examine metabolites for medium to low sodium intake in the same individuals.

In conclusion, metabolites and associated metabolic pathways differed by sodium intake. Some of our observed associations may be related to salt’s effect on the gut microbiome; however, more research is needed to understand the biochemical and molecular basis of our observations.

Acknowledgments

The authors’ responsibilities were as follows—RZS-S: conceived the study and acquired biomarker data for the research; RZS-S and JS: designed the study; AD: analyzed the data and performed the statistical analysis; JS: supervised the statistical analysis; AD, JS, and RZS-S: wrote the paper; JS, JJ, and MCP: provided intellectual content to revise the manuscript; JJ: conducted literature reviews for understanding metabolites; and all authors: read and approved the final manuscript and assumed full responsibility for the analyses, interpretation of these data, and final content of the manuscript. None of the authors reported a conflict of interest related to the study.

Footnotes

Abbreviations used: DASH, Dietary Approaches to Stop Hypertension Trial; MS, mass spectrometry; PC, principal component; SS, salt sensitive.

REFERENCES

  • 1.Jackson SL, King SM, Zhao L, Cogswell ME. Prevalence of excess sodium intake in the United States - NHANES, 2009-2012. MMWR Morb Mortal Wkly Rep 2016;64:1393–7. [DOI] [PubMed] [Google Scholar]
  • 2.Institute of Medicine. Sodium intake in populations: assessment of evidence Strom BL, Yaktine AL, Oria M, editors. Washington (DC): The National Academies Press; 2013. [PubMed] [Google Scholar]
  • 3.Sacks FM, Svetkey LP, Vollmer WM, Appel LJ, Bray GA, Harsha D, Obarzanek E, Conlin PR, Miller ER III, Simons-Morton DG, et al. ; . DASH-Sodium Collaborative Research Group. Effects on blood pressure of reduced dietary sodium and the Dietary Approaches to Stop Hypertension (DASH) diet. N Engl J Med 2001;344:3–10. [DOI] [PubMed] [Google Scholar]
  • 4.Svetkey LP, Sacks FM, Obarzanek E, Vollmer WM, Appel LJ, Lin PH, Karanja NM, Harsha DW, Bray GA, Aickin M, et al. ;. DASH-Sodium Collaborative Research Group. The DASH Diet, Sodium Intake and Blood Pressure Trial (DASH-Sodium): rationale and design. J Am Diet Assoc 1999;99:S96–104. [DOI] [PubMed] [Google Scholar]
  • 5.Appel LJ, Moore TJ, Obarzanek E, Vollmer WM, Svetkey LP, Sacks FM, Bray GA, Vogt TM, Cutler JA, Windhauser MM, et al. ;. DASH Collaborative Research Group. A clinical trial of the effects of dietary patterns on blood pressure. N Engl J Med 1997;336:1117–24. [DOI] [PubMed] [Google Scholar]
  • 6.Evans AM, Bridgewater BB, Liu Q, Mitchell MW, Robinson RJ, Dai H, Stewart SJ, DeHaven CD, Miller LAD. High resolution mass spectrometry improves data quantity and quality as compared to unit mass resolution mass spectrometry in high-throughput profiling metabolomics. Metabolomics 2014;4:132. [Google Scholar]
  • 7.Ogata H, Goto S, Sato K, Fujibuchi W, Bono H, Kanehisa M. KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res 1999;27:29–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Storey JD. A direct approach to false discovery rates. J R Stat Soc Series B 2002;64:479–98. [Google Scholar]
  • 9.Bass AJ, Dabney A, Robinson D. Q-value: Qvalue estimation for false discovery rate control. R package version 260. Vienna (Austria): R Foundation for Statistical Computing; 2015. [Google Scholar]
  • 10.R Development Core Team. R: a language and environment for statistical computing. Vienna (Austria): R Foundation for Statistical Computing; 2016. [Google Scholar]
  • 11.Guo L, Milburn MV, Ryals JA, Lonergan SC, Mitchell MW, Wulff JE, Alexander DC, Evans AM, Bridgewater B, Miller L, et al. Plasma metabolomic profiles enhance precision medicine for volunteers of normal health. Proc Natl Acad Sci USA 2015;112:E4901–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Hsiao EY, McBride SW, Hsien S, Sharon G, Hyde ER, McCue T, Codelli JA, Chow J, Reisman SE, Petrosino JF, et al. Microbiota modulate behavioral and physiological abnormalities associated with neurodevelopmental disorders. Cell 2013;155:1451–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Ilievski V, Kinchen JM, Prabhu R, Rim F, Leoni L, Unterman TG, Watanabe K. Experimental periodontitis results in prediabetes and metabolic alterations in brain, liver and heart: global untargeted metabolomic analyses. J Oral Biol (Northborough) 2016;3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Kikuchi K, Itoh Y, Tateoka R, Ezawa A, Murakami K, Niwa T. Metabolomic search for uremic toxins as indicators of the effect of an oral sorbent AST-120 by liquid chromatography/tandem mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci 2010;878:2997–3002. [DOI] [PubMed] [Google Scholar]
  • 15.Lustgarten MS, Price LL, Phillips EM, Kirn DR, Mills J, Fielding RA. Serum predictors of percent lean mass in young adults. J Strength Cond Res 2016;30:2194–201. [DOI] [PubMed] [Google Scholar]
  • 16.Itoh Y, Ezawa A, Kikuchi K, Tsuruta Y, Niwa T. Protein-bound uremic toxins in hemodialysis patients measured by liquid chromatography/tandem mass spectrometry and their effects on endothelial ROS production. Anal Bioanal Chem 2012;403:1841–50. [DOI] [PubMed] [Google Scholar]
  • 17.Guertin KA, Moore SC, Sampson JN, Huang WY, Xiao Q, Stolzenberg-Solomon RZ, Sinha R, Cross AJ. Metabolomics in nutritional epidemiology: identifying metabolites associated with diet and quantifying their potential to uncover diet-disease relations in populations. Am J Clin Nutr 2014;100:208–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Pallister T, Jennings A, Mohney RP, Yarand D, Mangino M, Cassidy A, MacGregor A, Spector TD, Menni C. Characterizing blood metabolomics profiles associated with self-reported food intakes in female twins. PLoS One 2016;11:e0158568. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Wu GD, Compher C, Chen EZ, Smith SA, Shah RD, Bittinger K, Chehoud C, Albenberg LG, Nessel L, Gilroy E, et al. Comparative metabolomics in vegans and omnivores reveal constraints on diet-dependent gut microbiota metabolite production. Gut 2016;65:63–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Parish M. How do salt and sugar prevent microbial spoilage? Sci Am 2006;294:98. [PubMed] [Google Scholar]
  • 21.Wang G, Nauseef WM. Salt, chloride, bleach, and innate host defense. J Leukoc Biol 2015;98:163–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Cirillo M, Capasso G, Di Leo VA, De Santo NG. A history of salt. Am J Nephrol 1994;14:426–31. [DOI] [PubMed] [Google Scholar]
  • 23.Holmes E, Loo RL, Stamler J, Bictash M, Yap IK, Chan Q, Ebbels T, De IM, Brown IJ, Veselkov KA, et al. Human metabolic phenotype diversity and its association with diet and blood pressure. Nature 2008;453:396–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Orlowski M, Meister A. The gamma-glutamyl cycle: a possible transport system for amino acids. Proc Natl Acad Sci USA 1970;67:1248–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Emdin M, Pompella A, Paolicchi A. Gamma-glutamyltransferase, atherosclerosis, and cardiovascular disease: triggering oxidative stress within the plaque. Circulation 2005;112:2078–80. [DOI] [PubMed] [Google Scholar]
  • 26.Bindal S, Gupta R. Hyper production of gamma-glutamyl transpeptidase from Bacillus licheniformis ER15 in the presence of high salt concentration. Prep Biochem Biotechnol 2017;47:163–72. [DOI] [PubMed] [Google Scholar]
  • 27.Shankar A, Li J. Association between serum gamma-glutamyltransferase level and prehypertension among US adults. Circ J 2007;71:1567–72. [DOI] [PubMed] [Google Scholar]
  • 28.Kunutsor SK, Apekey TA, Cheung BM. Gamma-glutamyltransferase and risk of hypertension: a systematic review and dose-response meta-analysis of prospective evidence. J Hypertens 2015;33:2373–81. [DOI] [PubMed] [Google Scholar]
  • 29.Kunutsor SK, Bakker SJ, Kootstra-Ros JE, Gansevoort RT, Dullaart RP. Circulating gamma glutamyltransferase and prediction of cardiovascular disease. Atherosclerosis 2015;238:356–64. [DOI] [PubMed] [Google Scholar]
  • 30.Kunutsor SK, Apekey TA, Khan H. Liver enzymes and risk of cardiovascular disease in the general population: a meta-analysis of prospective cohort studies. Atherosclerosis 2014;236:7–17. [DOI] [PubMed] [Google Scholar]
  • 31.Majid DS, Prieto MC, Navar LG. Salt-sensitive hypertension: perspectives on intrarenal mechanisms. Curr Hypertens Rev 2015;11:38–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Zhang LS, Davies SS. Microbial metabolism of dietary components to bioactive metabolites: opportunities for new therapeutic interventions. Genome Med 2016;8:46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Gertsman I, Gangoiti JA, Nyhan WL, Barshop BA. Perturbations of tyrosine metabolism promote the indolepyruvate pathway via tryptophan in host and microbiome. Mol Genet Metab 2015;114:431–7. [DOI] [PubMed] [Google Scholar]
  • 34.Upadhyaya B, McCormack L, Fardin-Kia AR, Juenemann R, Nichenametla S, Clapper J, Specker B, Dey M. Impact of dietary resistant starch type 4 on human gut microbiota and immunometabolic functions. Sci Rep 2016;6:28797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Juraschek SP, Gelber AC, Choi HK, Appel LJ, Miller ER III. Effects of the Dietary Approaches to Stop Hypertension (DASH) diet and sodium intake on serum uric acid. Arthritis Rheumatol 2016;68:3002–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Choi HK, Mount DB, Reginato AM; American College of Physicians; American Physiological Society. Pathogenesis of gout. Ann Intern Med 2005;143:499–516. [DOI] [PubMed] [Google Scholar]
  • 37.Ames BN, Cathcart R, Schwiers E, Hochstein P. Uric acid provides an antioxidant defense in humans against oxidant- and radical-caused aging and cancer: a hypothesis. Proc Natl Acad Sci USA 1981;78:6858–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Menni C, Graham D, Kastenmuller G, Alharbi NH, Alsanosi SM, McBride M, Mangino M, Titcombe P, Shin SY, Psatha M, et al. Metabolomic identification of a novel pathway of blood pressure regulation involving hexadecanedioate. Hypertension 2015;66:422–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Swan CE, Breyer RM. Prostaglandin E2 modulation of blood pressure homeostasis: studies in rodent models. Prostaglandins Other Lipid Mediat 2011;96:10–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Jablonski KL, Klawitter J, Chonchol M, Bassett CJ, Racine ML, Seals DR. Effect of dietary sodium restriction on human urinary metabolomic profiles. Clin J Am Soc Nephrol 2015;10:1227–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Mathew AV, Seymour EM, Byun J, Pennathur S, Hummel SL. Altered metabolic profile with sodium-restricted dietary approaches to stop hypertension diet in hypertensive heart failure with preserved ejection fraction. J Card Fail 2015;21:963–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Wang L, Hou E, Wang Z, Sun N, He L, Chen L, Liang M, Tian Z. Analysis of metabolites in plasma reveals distinct metabolic features between Dahl salt-sensitive rats and consomic SS.13(BN) rats. Biochem Biophys Res Commun 2014;450:863–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Evans AM, DeHaven CD, Barrett T, Mitchell M, Milgram E. Integrated, nontargeted ultrahigh performance liquid chromatography/electrospray ionization tandem mass spectrometry platform for the identification and relative quantification of the small-molecule complement of biological systems. Anal Chem 2009;81:6656–67. [DOI] [PubMed] [Google Scholar]

Articles from The American Journal of Clinical Nutrition are provided here courtesy of American Society for Nutrition

RESOURCES