(Estarreja) Cohort - ACS Publications - American Chemical Society

Subscriber access provided by UNIV OF NEW ENGLAND ARMIDALE

Article

Assessing exposome effects on pregnancy through urine metabolomics of a Portuguese (Estarreja) cohort Ana M Gil, Daniela Duarte, Joana Pinto, and António S. Barros J. Proteome Res., Just Accepted Manuscript • DOI: 10.1021/acs.jproteome.7b00878 • Publication Date (Web): 09 Feb 2018 Downloaded from http://pubs.acs.org on February 14, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Journal of Proteome Research is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Journal of Proteome Research

Assessing exposome effects on pregnancy through urine metabolomics of a Portuguese (Estarreja) cohort

Ana M. Gil1,*, Daniela Duarte1, Joana Pinto1,2, António S. Barros1,3

1

CICECO- Aveiro Institute of Materials (CICECO/UA), Department of Chemistry,

University of Aveiro, 3810-193 Aveiro, Portugal 2

UCIBIO@REQUIMTE/Laboratório de Toxicologia, Departamento de Ciências Biológicas,

Faculdade de Farmácia, Universidade do Porto, 4050-313 Porto, Portugal 3

Department of Cardiothoracic Surgery and Physiology, Faculty of Medicine, Porto 4200-

319, Portugal

* author to whom correspondence should be addressed. Email: [email protected], Telephone: +351927992240

Keywords: environment, pollutants, exposome, in utero environment, pregnancy, metabolomics, nuclear magnetic resonance (NMR) spectroscopy, maternal urine

ACS Paragon Plus Environment

1

Journal of Proteome Research 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 37

Abstract This nuclear magnetic resonance metabolomics study compared the influence of two different Central Portugal exposomes, one of which comprising an important source of pollutants (the Estarreja Chemical Complex, ECC), on the urinary metabolic trajectory of a cohort of healthy pregnant women (total n=107). An exposome-independent description of pregnancy metabolism was found to comprise a set of 18 metabolites reflecting expected changes in branched-chain-amino acids catabolism, hormone and lipid metabolisms. In addition, a set of small changes in some metabolites were suggested to be exposome-dependent and characteristic of pregnant subjects from the Estarreja region. These results suggested that the Estarreja exposome may impact to a very low extent on pregnancy metabolism, inducing slight changes in amino acid metabolism (alanine, glycine and 3-hydroxyisobutyrate, possibly involved in valine metabolism), tricarboxylic acid (TCA) cycle (cis-aconitate), diet and/or gut microflora (furoylglycine), as well as allantoin, 2-hydroxyisobutyrate and an unassigned resonance at δ 8.45. Furthermore, the urine of Estarreja subjects was found to generally contain higher levels of 4-hydroxyphenylacetate and lower levels of citrate. However, out of the above metabolites, only glycine and citrate seemed to correlate with the proximity to the ECC, with slightly relative higher levels of these compounds found for subjects living closer to the ECC. This suggested possible small effects of local pollutants on energy metabolism, with the remaining exposome-dependent metabolite changes most probably originating from other aspects of the local exposome such as diet and lifestyle. In spite of the limitation of this study regarding the unavailability of objective environmental parameters for the period under study, our results confirm the usefulness of metabolomics of human urine to gauge exposome effects on human health and, particularly, during pregnancy.


2

Page 3 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Introduction The use of metabolomic strategies to assess human susceptibility to health complications as a result of exposure to environmental hazards has gained increasing interest in the last decade, aided by a steady improvement of both analytical tools, based either on Nuclear Magnetic Resonance (NMR) spectroscopy or mass spectrometry (MS), and chemometric tools for data handling, processing and interpretation. In this context, the concepts of environmental or xenobiotic metabolomics as a means of assessing the effects of particular chemicals (pollutants, drugs, cosmetics, diet components) on human health1,2 have supported a large number of studies focusing on metals2-12, natural and synthetic toxins and drugs13–21, smoke22– 24

, pesticides25–28, particulate matter including nanoparticles29–32 and radiation33,34. Such

factors have been tested on a variety of biological systems, spanning from human cell lines13,15,16,20,29,33, to a range of in vivo model systems6-11,14,19-21,24,30,32,35 and human individuals4,5,12,18,22,23,25-28,35. Human studies have gradually unveiled the challenges related to the understanding of the interplay of the complex network of environmental stimuli with inter-subject variability due to distinct lifestyles. The set of stimuli which impact on a certain population has become known as the exposome and this is composed of simultaneously acting nutritional, pharmaceutical and environmental factors36–39. Untargeted metabolomics is a powerful strategy to probe for complex exposure effects on human populations and provide further understanding of the environmental causes underlying diseases. The study of exposome effects is of upmost relevance during the human gestation period, in an attempt to investigate possible relationships between exposure characteristics and potential metabolic disturbances or prenatal disorders. A study investigated the metabolic profile of the urine of pregnant women (11 gestational weeks) of a French cohort (total of n=83) as a function of surface land used for cereal culture (n=20-40 per subgroup) and, hence, influence of multiple pesticides25. The results suggested that complex pesticide mixtures may induce modified


3


Page 4 of 37

urinary metabolic fingerprints, with increased levels of glycine, threonine, lactate and glycerophosphocholine, and decreased citrate levels, possibly reflecting increased oxidative stress and energy metabolism disturbances. Another study of urine profiling of pregnant women in their 1st trimester has recently assessed the effects of low-dose arsenic exposure by MS metabolomics (total n=246, 82 subjects per each of three levels of exposure)13. The authors advanced several possible biomarkers of arsenic exposure: lysophosphocholine (14:0), glutathione, 18-carboxy-dinor-leukotriene E4 (LTE4), 20-COOH-LTE4, cystathionine ketimin, 1-(β-D-ribofuranosyl)-1,4-dihydronicotinamide, thiocysteine, p-cresol glucuronide and vanillactic acid. These compounds were suggested to relate to oxidative stress and liver and kidney disorders in pregnant women. Also recently, arsenic levels measured in the drinking water and urine of pregnant women (n=50, collection at delivery) have been found to be associated with changed levels of several metabolites in neonate cord blood, as assessed by NMR metabolomics40. Such results suggested a slowing down of energy metabolism as a result of arsenic exposure, and deviant behaviors in vitamin and amino acid metabolisms. The above results demonstrate the value of prenatal and neonatal biofluid metabolites as markers of environmental impact on in utero development. Hair metabolome has also been suggested as a valuable source of markers of the contributions of environment to prenatal development and, in particular, fetal growth restriction (FGR)41. This exploratory work suggested a predictive model of FGR based on 5 metabolites present in hair extracts (lactate, levulinate, 2methyloctadecanate, tyrosine and margarate), and the increased levels of exogenous heptadecane found in FGR cases were putatively associated to air pollution and/or food contamination. The above studies relate to a fixed point in pregnancy, whereas a recent report has addressed pregnancy progression by sampling urine at 1st and 3rd trimesters for each subject42. Such work compared two different large Spanish cohorts (n ≥ 400 each), suggesting possible relationships between different lifestyles and environmental conditions (e.g. smoke


4

Page 5 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and chemical exposure, surrounding greenness), and metabolic phenotype and clinical outcomes of pregnancy. Increased branched chain amino acids (BCCA) were suggested as a marker of positive fetal growth in more favorable environments. In the present exploratory work, the changes in urinary metabolic profile across the three trimesters of pregnancy of women (n=52) residing near a chemical industrial site (the Estarreja Chemical Complex (ECC) in central Portugal) were characterized by NMR metabolomics. Located at ca. 3 km north of the urban center of Estarreja (Figure 1), the EEC is one of the largest clusters of chemical industry in Portugal and comprises industrial units manufacturing polyurethanes, thermoplastic materials, chlorine-alkali, aniline and derivatives and industrial gases43. Following a cross-sectional design, urine samples were collected for all pregnancy trimesters, in order to establish an average metabolic trajectory of pregnancy in Estarreja and, thus, allow for the identification of time-dependent effects. The urinary metabolic trajectory of healthy pregnant women in Estarreja was compared with data reported earlier44 for a similar-sized cohort (n=55) of pregnant women residing in the Coimbra region, another central Portugal area, located ca. 80 km south of the ECC (Figure 1). Such comparison enabled the identification of 1) a common urinary metabolic signature globally descriptive of pregnancy and named as exposome-independent signature of pregnancy and, 2) differing metabolites composing an exposome-dependent signature descriptive of pregnancy in Estarreja and subsequently tested for its association to distance to the EEC site.

Experimental Section Cohort definition Urine samples were collected under the approval of the Ethical Committee of the Portuguese Administração Regional de Saúde da Região Centro (ref. 55/2014) and the National Commission for Data Protection (ref. 4944/2015). Informed consents were obtained from


5


Page 6 of 37

each woman participating in the study. For each subject, pregnancy was followed up to term and only individuals with normal pregnancy outcomes were considered for the study, since the aim was to characterize healthy pregnancies alone and not study the incidence of derived complications. This was mostly a cross-sectional study (i.e. most women contributed only once to sampling, mostly due to logistic reasons and occasional opting for private care) carried out between February 2015 and May 2015, at routine pregnancy follow-up appointments, at three different health centers in the Estarreja municipality: Avanca (ca. 3.9 km from ECC), Estarreja-town (ca. 2.5 km from ECC) and Salreu (ca. 4.8 km from ECC) (Figure 1). The low population density in out of-town areas was a limiting factor for cohort size. Most samples were collected under fasting conditions, although some (17 in 52) had to be collected non-fasting due to medical appointment logistics. As the effect of non-fasting on maternal urine has been studied before44, the previously noted changes were taken into account at the time of data interpretation. The results obtained here for Estarreja were compared with those obtained previously for a Coimbra cohort, a central Portugal region located ca. 80 km south of EEC44, the study of which involved identical standard operating procedures. The data from the two cohorts were here partially reanalyzed together, for the sake of comparison. This enabled the identification of common and differing metabolite variations, respectively reflecting exposome-independent and exposome-dependent metabolic signatures of pregnancy. Table 1 shows the population demographics data for both cohorts. Sample collection, storage and preparation for analysis Urine samples (ca. 50 mL) were collected in each of the three health centers of Avanca, Estarreja-town and Salreu (Figure 1) and stored at 4ºC up to a maximum of 24 hours. Samples were transported in ice to the University of Aveiro (within a maximum of 3 hours) and stored at -80ºC. Before analysis, samples were thawed and 800 µL were centrifuged (4500 g, 5 min, 25ºC). Then, 60 µL of 1.5 M KH2PO4/D2O buffer pH 7, 0.1%Na+/3-trimethylsilyl-propionate


6

Page 7 of 37 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(TSP) were added to 540 µL of supernatant, followed by pH readjustment to 7.00±0.02 with KOD (4 M) or DCl (4 M). The mixture was centrifuged (4500g, 5 min, 25ºC) and 550 µL were transferred to a 5 mm NMR tube. NMR Spectroscopy NMR spectra were recorded on a Bruker Avance DRX 500 spectrometer at 300 K. Standard proton 1D spectra were acquired, using a noesy 1D pulse sequence with 100 ms mixing time, a 3 µs t1 delay, and water suppression during relaxation delay and mixing time. 128 transients were acquired into 64 k complex data points, with 10080.65 Hz spectral width, 4s relaxation delay and 3.25 s acquisition time. Each free-induction-decay was multiplied by a 0.3 Hz exponential line-broadening function prior to Fourier transformation. Spectra were manually phased and baseline corrected. Chemical shifts were referenced internally to TSP at δ=0.0 ppm. Peak assignments were carried out with basis on literature45, 2D NMR experiments (Total Correlation Spectroscopy-TOCSY, Heteronuclear Single Quantum Coherence- HSQC and J-resolved), spectral databases (Bruker Biorefcode database and the human metabolome database (HMDB)46. Data analysis Multivariate analysis was applied to the full resolution 1H NMR urine spectra, after exclusion of water (4.60-5.05 ppm) and urea (5.50-6.20 ppm) regions. Spectra were aligned using a recursive segment-wise peak alignment47 and normalized to total area, to account for sample concentration differences. Principal component analysis (PCA)48 and partial least squares discriminant analysis (PLS-DA)49 were performed after unit variance (UV) scaling, using SIMCA-P 11.5 (Umetrics, Sweden). The corresponding loading weights were obtained by multiplying each variable by its standard deviation and were colored according to each variable importance to the projection (VIP). PLS-DA model validation was carried out by Monte-Carlo cross-validation (MCCV) (7 blocks) with 500 runs, with recovery of Q2 values


7


Page 8 of 37

and confusion matrices. Classification rates, specificity and sensitivity were computed (in %) and the predictive power of each model was further assessed using a receiver operating characteristic (ROC) map, a function of the true positive rate (TPR or sensitivity) and false positive rate (FPR or 1-specificity). PLS-DA models were considered robust when minimal overlapping of the distribution of true and permuted Q2 was achieved50,51 . To improve PLSDA models, variable selection was applied as previously described52. For valid models, the relevant peaks were integrated (Amix 3.9.14, BrukerBioSpin, Rheinstetten, Germany), normalized to total area, assessed using the non-parametric t-test (Wilcoxon test). Univariate analysis was reapplied to the spectra obtained previously for the Coimbra cohort44, using the same data analysis protocol. Effect sizes were computed for all relevant resonances (p

(Estarreja) Cohort - ACS Publications - American Chemical Society

Recommend Documents