Untargeted 1H NMR-Based Metabolomics Analysis of Urine and

High legume intake has been shown to have beneficial effects on the health of humans. ... (9) Discovering and validating specific biomarkers related t...
1 downloads 0 Views 1MB Size
Article Cite This: J. Agric. Food Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/JAFC

Untargeted 1H NMR-Based Metabolomics Analysis of Urine and Serum Profiles after Consumption of Lentils, Chickpeas, and Beans: An Extended Meal Study To Discover Dietary Biomarkers of Pulses Francisco Madrid-Gambin,†,∥ Carl Brunius,§ Mar Garcia-Aloy,†,∥ Sheila Estruel-Amades,† Rikard Landberg,‡,§ and Cristina Andres-Lacueva*,†,∥

Downloaded via UNIV OF SUNDERLAND on July 8, 2018 at 17:37:45 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.



Biomarkers & Nutrimetabolomics Laboratory, Department of Nutrition, Food Sciences and Gastronomy, XaRTA, INSA, Faculty of Pharmacy and Food Sciences, University of Barcelona, Barcelona 08028, Spain ∥ CIBER de Fragilidad y Envejecimiento Saludable (CIBERFES), Instituto de Salud Carlos III, Barcelona, Spain ‡ Department of Molecular Sciences, Swedish University of Agricultural Sciences, Uppsala 750 07, Sweden § Department of Biology and Biological Engineering, Food and Nutrition Science, Chalmers University of Technology, Gothenburg SE-412 96, Sweden S Supporting Information *

ABSTRACT: High legume intake has been shown to have beneficial effects on the health of humans. The use of nutritional biomarkers, as a complement to self-reported questionnaires, could assist in evaluating dietary intake and downstream effects on human health. The aim of this study was to investigate potential biomarkers of the consumption of pulses (i.e., white beans, chickpeas, and lentils) by using untargeted NMR-based metabolomics. Meals rich in pulses were consumed by a total of 11 participants in a randomized crossover study and multilevel partial least-squares regression was employed for paired comparisons. Metabolomics analysis indicated that trigonelline, 3-methylhistidine, dimethylglycine, trimethylamine, and lysine were potential, though not highly specific, biomarkers of pulse intake. Furthermore, monitoring of these metabolites for a period of 48 h after intake revealed a range of different excretion patterns among pulses. Following the consumption of pulses, a metabolomic profiling revealed that the concentration ratios of trigonelline, choline, lysine, and histidine were similar to those found in urine. In conclusion, this study identified potential urinary biomarkers of exposure to dietary pulses and provided valuable information about the time-response effect of these putative biomarkers. KEYWORDS: biomarkers, dietary pulses, legumes, metabolomics, NMR



multivariate food exposure biomarkers in humans.11 Additionally, by reflecting changes in the metabolome induced by diet, new biomarkers of the effect of consuming pulses may reveal potential mechanisms in diet-related physiology.12 One strategy that has proved useful in discovering new dietary biomarkers is the use of untargeted metabolomics in singlemeal studies with a time series of samples taken following intake.13 This technique also enables the study of perturbations in endogenous metabolites due to acute pulse intake, which may have impacts on health. Thus, it has also been suggested that metabolomic approaches be employed in evaluating the relationship between health and nutrition.14 In a previous work on serum biomarkers of consumption following a dietary intervention with dry beans, pipecolic acid and S-methyl cysteine were proposed as biomarker candidates.15 Nevertheless, potential differences between pulses were not investigated in that study as the intervention only involved dry beans. Furthermore, while plasma/serum better reflects modulations in endogenous metabolism, dietary

INTRODUCTION The term “dietary pulses” refers to the dried seeds of legumes including white beans, chickpeas, and lentils.1 A high intake of such pulses generally results in a lower body weight,2 along with lower LDL cholesterol levels3 and an enhancement in glycemic control4 in meta-analyses of controlled feeding studies. Consequently, the role of pulses in promoting health is gaining increasing recognition. In fact, regular consumption of pulses is now included in dietary guidelines across the globe, including the Mediterranean Diet5 and the Dietary Guidelines for Americans,6 not least because pulses have a lower environmental impact than other protein sources.7 However, using dietary assessment tools based on self-reporting to monitor specific food intakes or compliance with dietary components is fundamentally inaccurate.8 Therefore, there is a growing need to discover reliable biomarkers that correctly reflect food intake to both enable accurate risk assessment and to uncover mechanisms of observed effects.9 Discovering and validating specific biomarkers related to pulses could enhance the evaluation of pulse intake in relation to epidemiological studies. By identifying a large number of metabolites using highthroughput techniques,10 metabolomics may prove to be a powerful tool in identifying either single biomarkers or © XXXX American Chemical Society

Received: Revised: Accepted: Published: A

January 3, 2018 June 18, 2018 June 19, 2018 June 19, 2018 DOI: 10.1021/acs.jafc.8b00047 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry exposure is often better reflected in the urinary metabolome.16 Our group previously explored urinary exposure markers of pulses in free-living subjects,17 but to the best of our knowledge the effects of exposure on both urinary and serum metabolome after a controlled dietary intervention with specific pulses have not been evaluated in any study. We applied an untargeted 1H NMR metabolomics approach in the present study with a view to discovering dietary and endogenous biomarkers related to a single-meal intake of white beans, chickpeas, and lentils and to investigate the timedependent profiles of these potential biomarker candidates in urine and serum. This original work was previously presented in modified form in a doctoral thesis published open access as regulated by the University of Barcelona, Spain.



buffer (37.5 mM sodium phosphate buffer, 0.02% NaN3, 0.25 mM DSS-d6, and 1 mM imidazole in deuterium oxide (D 99%)) before being shaken for 30 min. The samples (180 μL) were then transferred into 3 mm SampleJet NMR tubes in readiness for analysis. Serum (150 μL) was transferred into 96-deepwell plates, with each well containing 150 μL of ultrapure water. Then 600 μL of pure cold methanol (−20 °C) was added to each well prior to incubation in a thermomixer at 12 °C and 800 rpm for 10 min. The samples were kept at −20 °C for 30 min before being centrifuged at 2250g and 4 °C for 60 min. Supernatants (600 μL) were lyophilized, washed with 50 μL of MeOD, and then relyophilized to remove solvent remains. Dried samples were reconstituted in 200 μL of buffer (37.5 mM sodium phosphate, 0.02% NaN3, 0.25 mM DSS-d6, and 1 mM imidazole in deuterium oxide (D 99%)). Prior to analysis, the samples (180 μL) were then transferred into 3 mm SampleJet NMR tubes. After the urine samples had been thawed at 4 °C and spun down, 180 μL was added into a 96-deepwell plate containing 20 μL of internal standard solution (1.5 M potassium phosphate, 0.5% NaN3, and 0.5 mM DSS-d6 in deuterium oxide (D 99%)). The mixture was shaken in an Eppendorf thermomixer (22 °C, 800 rpm for 30 s) before being transferred into 3 mm SampleJet NMR tubes with the aid of a SamplePro L liquid-handling robot (Bruker BioSpin, Rheinstetten, Germany). NMR Spectroscopy Analysis. An Oxford 800 MHz magnet equipped with a Bruker Avance III HD console and a 3 mm TCI cryoprobe was used to record spectra, with the acquisition temperature being set at 298 K. A water suppression pulse program was utilized to acquire 1H spectra, with 128 scans collected into 65 K data points in a spectral width of 20 ppm, an acquisition time of 4 s, and a relaxation delay of 1 s. A line broadening of 0.3 Hz was applied to the FIDs before undergoing Fourier transformation. For identification purposes, 1H−13C HSQC spectra were obtained, recorded into 16 scans, with spectral widths of 20 (1H) and 100 (13C) ppm and a relaxation delay of 3 s. TopSpin 3.5pI6 (Bruker GmbH, Rheinstetten, Germany) software was used for processing spectra. Processing of NMR Data. After being processed, the spectra were aligned using the “speaq” R-package version 1.2.1.19 The features were acquired using continuous wavelet transformation peak picking and extracting intensities of peaks at corresponding ppms (chemical shifts). Data sets were normalized by “probabilistic quotient normalization”20 for the adjustment of dilution factors. The area under the curve (AUC) of each variable was calculated between t0h and t6h. Identification. Chenomx NMR Suite 8.2 profiler (Chenomx Inc., Edmonton, Canada) was used for metabolite identification and deconvolution, and a comparison of the chemical shifts with those available in the Human Metabolome Database (http://www.hmdb. ca) provided further contributions to the proton peak assignment. Finally, tentative metabolite signals were assigned to specific metabolites through the use of an in-house R script for statistical total correlation spectroscopy (STOCSY)25 and a heteronuclear 1 H−13C HSQC experiment. Chenomx Profiler was also used for identifying and quantifying compounds detected in the foods employed in the study, and the identified food compounds were quantified through the use of a known reference signal (TSP) to quantify. Statistical Data Analysis. To assess differences between the profiles of pulses, one-way ANOVA followed by the BenjaminiHochberg procedure to control the false discovery rate (FDR) was carried out on metabolic profiles. A FDR-corrected p-value of