Urinary metabolomics reveals novel interactions between metal exposure and amino acid metabolic stress during pregnancy

Pregnant women are a unique group undergoing profound structural modifications in uterus, breast, adipose tissue, etc.

Abstract

Pregnant women are a unique group undergoing profound structural modifications in uterus, breast, adipose tissue and extracellular fluids. Amino acid metabolic stress is a unique physical process that occurs during pregnancy. Metals constitute a fundamental part of the maternal body and have a universal effect on amino acid metabolism. However, the exact interaction between metals and amino acid metabolism during pregnancy is unknown. The aim of the present study was to determine the correlations of metals with amino acid metabolic intermediates in the urine of 232 healthy pregnant women in their first, second and third trimesters during normal pregnancy. Sixteen metals in the urine of 232 healthy pregnant women in their first, second and third trimesters were quantified using inductively coupled plasma mass spectrometry (ICP-MS). An ultra-performance liquid chromatography quadrupole time-of-flight mass spectrometer (UPLC-QTOFMS)-based metabolomics approach was conducted to detect intermediate products involved in amino acid metabolism during the entire pregnancy period. A panel regression model was established to investigate the relationship between urine metals and amino acid metabolism. Seven metals—cadmium, cobalt, copper, cesium, manganese, thallium and vanadium—showed significant association with amino acid metabolic intermediates, including 2-oxoarginine, 3-indoleacetonitrile, indole, indole-5,6-quinone, N2-succinyl-l-glutamic acid 5-semialdehyde, N-methyltryptamine and N-succinyl-l,l-2,6-diaminopimelate, in the healthy pregnant women. These findings indicated that exposure to cadmium, cobalt, copper, cesium, manganese, thallium and vanadium significantly affected the metabolic status of tryptophan, arginine, proline, tyrosine and lysine metabolism in the maternal body during normal pregnancy.

1. Introduction

Humans have a long history of using metals. However, over the past few decades, excessive amounts of metals have been discharged into the environment due to human activity and have caused a series of environmental problems in both developed and developing countries. The general population has a high risk of exposure to different levels of metals through food, water, and ambient air.1 Certain metals, such as cobalt, vanadium, manganese and copper, are essential to the human body and perform important biological functions, whereas other metals such as cadmium, thallium and lead are nonessential and possess complex toxic properties at different concentrations.2 Although many studies have been conducted to elucidate the physiological or toxic mechanisms of single metals in the human body, studies on the biochemical mechanisms of interactions between metal exposure and the human body when essential and nonessential metals are simultaneously present in the exposure profile are very limited.

Pregnant women are a unique group of people undergoing profound structural modifications in the uterus, breast, adipose tissue and extracellular fluids. During pregnancy, mothers face physiological challenges that require complex adaptation coordinated by both placentally and non-placentally derived hormones to prepare for metabolic stress presented by fetal development and to ensure accurate and adequate nutrient shunting from themselves to the fetus.1,3,4,26 Metals can pass through the placental fetal-maternal barrier and have lasting effects on fetal growth.5 Metals such as iron, calcium, magnesium and zinc are important nutrients for growth and development of the fetus, and their deficiencies can cause serious problems. Other metals such as lead, cadmium, and thallium can cause birth defects, most commonly intrauterine growth restriction and low birth weight,6–12 indicating that metals may interfere with protein metabolism in the maternal and fetal bodies.

Amino acids are the basic materials for protein synthesis. Amino acid metabolism plays an extremely important role in maternal health and fetal growth during pregnancy. Both enhanced maternal and fetal utilization of amino acids to form maternal and fetal tissues, respectively, strongly contributes to amino acid metabolic stress during pregnancy. In the maternal body, increased volume of maternal tissue contributes to approximately 60% of the increase in body weight that occurs during pregnancy.3 In the fetal body, amino acid-induced protein synthesis through the placenta is associated with fetal birth weight, which significantly affects postnatal lifelong metabolic health.13 Emerging evidence suggests that metals such as cadmium, mercury, arsenic, lead, manganese and zinc act as endocrine-disrupting chemicals that can mimic, block, or otherwise alter the activity of hormones, thus disrupting normal growth and development of fetuses.14,15 In addition, metal ions have the potential to damage amino acid metabolic balance by disturbing the activity of metabolic enzymes because one-third of enzymes in the body require the participation of metal ions.16 Our previous study confirmed that alterations in the metabolism of tryptophan, histidine, glutathione, cysteine and methionine occurred during normal pregnancy.17

Given this fact, we further hypothesized that metals in the maternal body may interact with certain amino acid metabolic pathways during normal pregnancy. In the present study, we evaluated the correlation of urine metals, namely, aluminum, vanadium, manganese, iron, cobalt, copper, zinc, arsenic, selenium, rubidium, strontium, cadmium, cesium, barium, thallium and lead, with amino acid metabolic intermediates using a dynamic panel regression model. An integrated approach of ultra-performance liquid chromatography quadrupole time-of-flight mass spectrometer (UPLC-QTOFMS)-based metabolomics and inductively coupled plasma mass spectrometry (ICP-MS)-based metallomics was conducted to obtain urine amino acid metabolic intermediates and metal profiles in a longitudinal cohort of 232 healthy pregnant women during the course of pregnancy. To our knowledge, this study is the first report of a dynamic interaction between metal profile and amino acid metabolic signature in maternal urine for normal pregnancies.

2. Materials and methods

2.1. Clinical samples

The participants in this study were selected from healthy pregnant women enrolled from the Maternal and Child Health Hospital of Wuhan City, China during November 2013 and July 2014. Non-diabetic subjects who were more than 18 years old with a singleton, intrauterine pregnancy were eligible for inclusion in the present cohort study. All participants were inhabitants of Wuhan City and had no evident occupational exposure to metals. A total of 232 subjects were eventually included in the present study. All participants provided written informed consent and completed an individual questionnaire at the time of urine sample collection for this study. The 232 pregnant women were examined throughout the entire pregnancy and clinical information was obtained from face-to-face interviews and obstetrical and neonatal medical records. Detailed information about demographic characteristics of these subjects is shown in Table 1. The urine samples were collected during the first, second and third trimesters and frozen at –80 °C until analysis. A minimized freeze–thaw cycle was used to reduce the introduced interference as much as possible. The research protocol was approved by the Ethics Committees of the Tongji Medical College, Huazhong University of Science and Technology as well as the study hospital. All experiments were performed in accordance with principles expressed in the Declaration of Helsinki or other relevant guidelines and regulations.

Table 1. Demographic characteristics of pregnant women in this study ^a .

	Characteristics
Number of individuals	232
Maternal age, years	28.17 ± 3.24
Body mass index, kg m–2	20.76 ± 2.62
Gravidity	1.79 ± 1.24
Parity	1.13 ± 0.34
Birth weight of fetuses (g)	3325.95 ± 377.25
Gestational weeks
First trimester	12.78 ± 1.18
Second trimester	23.21 ± 3.00
Third trimester	34.45 ± 4.67

^aValues are presented as mean ± standard deviation.

2.2. Chemicals and instruments

Inductively coupled plasma mass spectrometry (ICP-MS, Agilent 7700, Agilent Technologies, USA), water purifier (Millipore Simplicity, Germany), l-tryptophan d₅ (Cambridge Isotope Laboratories, UK), formic acid (CNW Technologies GmbH, Germany), methanol (Sinopharm, China), acetonitrile (Fisher Scientific, USA), HNO₃ (FisherOptoma, USA), 10 μg mL^–1 multi-metals standard stock solution (Agilent Technologies, USA), urine standard sample (SRM2670a, National Institute of Standards and Technology, USA), Acquity TM ultra performance liquid chromatography system (Waters, USA), Synapt Mass Spectrometry (Waters, USA), Masslynx 4.1 (Waters, USA) and Acquity UPLC HSS T3 Column (2.1 × 100 mm, 1.8 μm, Waters, USA) were used for this study.

2.3. Metals and metalloids measurement18

Six hundred and ninety-six aliquots of 0.5 mL urine samples from all the subjects were nitrified with 2 mL of 1.2% HNO₃ overnight. The resulting samples were digested by ultrasonication at 40 °C for 1 hour and then analyzed by ICP-MS. The operation conditions of ICP-MS were as follows: RF power, 1550 w; plasma gas flow, 15.00 L min^–1; auxiliary gas flow, 0.8 L min^–1; carrier gas flow, 0.25 L min^–1; resolution (peak high 10%), 0.65–0.80 amu, injector flow rate 0.4 mL min^–1; unimodal residence time, 0.1 s.

The Standard Reference Material Human Urine (SRM2670a Toxic Elements in Urine, National Institute of Standards and Technology, USA) was applied as an external quality control in each batch to evaluate instrument performance. The control samples were analyzed for elements after calibration and after every 20th sample, and the concentrations measured were within the certified range recommended by the manufacturer (5%). If concentrations were significantly different from the certified value of SRM2670a, the instrument was recalibrated and the previous batch of samples was reanalyzed. A 1.2% HNO₃ blank was processed in each batch of samples to control possible contamination.

Sixteen metals, namely, aluminum (²⁷Al, Al), vanadium (⁵¹V, V), manganese (⁵⁵Mn, Mn), iron (⁵⁶Fe, Fe), cobalt (⁵⁹Co, Co), copper (⁶³Cu, Cu), zinc (⁶⁶Zn, Zn), arsenic (⁷⁵As, As), selenium (⁷⁸Se, Se), rubidium (⁸⁵Rb, Rb), strontium (⁸⁸Sr, Sr), cadmium (¹¹¹Cd, Cd), cesium (¹³³Cs, Cs), barium (¹³⁷Ba, Ba), thallium (²⁰⁵Tl, Tl) and lead (²⁰⁸Pb, Pb), were rapidly determined and precisely quantified. Values below the LOQ (limit of quantity) were replaced with the LOQ divided by the square root. All measurements were repeated three times and average values were used for statistical analyses.

2.4. Metabolic profiling spectral acquisition17

Chromatographic analysis was performed in an Acquity™ ultra performance liquid chromatography system (Waters, USA) with an ACQUITY UPLC HSS T3 column (2.1 × 100 mm, 1.8 μm, Waters, USA). Mobile phase A was 0.1% formic acid in water (v/v) and mobile phase B was 0.1% formic acid in methanol (v/v); the flow rate was 0.5 mL min^–1. Gradient conditions of the mobile phase in positive and negative modes were as follows: 0–1 min: 1% B; 1–3 min: 1–15% B; 3–6 min: 15–50% B; 6–9 min: 50–95% B; 9–10 min: 95% B; 10–10.1 min: 95–1% B; 10.1–12 min: 1% B. Temperatures of the column and autosampler were maintained at 40 °C and 4 °C, respectively.

A Waters Synapt™ High-Definition Time-of-Flight Mass Spectrometry system (Waters, USA) equipped with an electrospray ionization (ESI) source operating in positive and negative modes connected with UPLC was applied in this study. Capillary voltage was 3.2 kV and 2.4 kV in positive and negative ionization mode, respectively. The desolvation temperature was 350 °C, the sampling cone voltage was 40 V, the extraction cone voltage was 4.0 V, the source temperature was 120 °C, the cone gas flow was 25 L h^–1, and the desolvation gas flow was 900 L h^–1. The mass was corrected during acquisition with leucine-enkephalin to generate reference ion at m/z 556.2771 Da ([M + H]⁺) in positive ion mode before the instrument was used to ensure accurate mass measurement. The metabolites were further identified by comparison of their structural information with those obtained in the HMDB, METLIN or MassBank databases.

2.5. Data model establishment and statistical analysis

Data on urinary metals and amino acid intermediates obtained from 232 healthy pregnant women in three trimesters fit in the panel data model.19 The panel model generally has three basic types: independently pooled panels, random effects models and fixed effects models.19 To select the panel data model, we used the F test to decide whether to use a mixed model or fixed effects model. Then, we used the Hausman test to determine whether we should establish a random effects model or a fixed effects model. In this study, we chose the second type because there was a difference in the levels of metal exposure among different individuals. The regression model was established in EViews 8.0 software. The basic form of the regression equation is as follows:Y_it = α_i + X_it + μ_it

In this equation, Y refers to the dependent variable (urine amino acid metabolic intermediates generated from metabolomics study) and X refers to independent variables (urine metals generated from metallomics study). Subscript i refers to different subjects (i = 1, …, 232) and subscript t refers to different trimesters (t = 1, 2, 3). α refers to the intercept and μ to the residual. Due to the evident covariance relationship among different metals, a method of cross-section weights (estimated generalized least squares, EGLS) was used in the panel model. F tests and t tests were used to test the regression equation and each regression coefficient, respectively. The criterion for determining the statistical significance was p < 0.05. A random forest model was conducted in R (R Core Team (2015). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL ; https://www.R-project.org/). The package version of the random forest model was 4.6–7. The function of Importance and MDSplot() was used to visualize the random forest model.

3. Results

3.1. Amino acid metabolic intermediates generated from metabolomics study

We reported the dynamic profiles of urine metabolites involved in amino acid metabolism in different trimesters during normal pregnancy using a UPLC-QTOFMS analysis protocol in a previous study.17 In the present study, we selected metabolites involved in amino acid metabolism that were relatively higher in abundance and mutually exist in all three trimesters to establish the panel regression model. The baseline information of these metabolites is shown in Table 2. Intermediates involved in tryptophan metabolism, tyrosine metabolism, lysine biosynthesis, and arginine and proline metabolism were investigated. Eight metabolites—3-indoleacetonitrile, indole, indole-5,6-quinone, 2-oxoarginine, N2-succinyl-l-glutamic acid 5-semialdehyde, N-methyltryptamine, N-succinyl-l,l-2,6-diaminopimelate and creatinine—were selected to detect the interaction with the metals.

Table 2. Baseline data of urine metabolites of healthy pregnant women involved in panel regression analysis.

HMDB ID	Metabolite name	Primary ID in T1 a	Primary ID in T2 a	Primary ID in T3 a	Adduct ion	Metabolic pathway
HMDB04225	2-Oxoarginine	4.09_347.168	4.09_347.169	4.09_347.169	[2M + H]+	Arginine and proline metabolism
HMDB06524	3-Indoleacetonitrile	2.09_157.077	2.11_157.078	2.13_157.078	[M + H]+	Tryptophan metabolism
HMDB00738	Indole	0.51_156.020	0.51_156.021	0.51_156.020	[M + K]+	Tryptophan metabolism
HMDB06779	Indole-5,6-quinone	0.88_148.040	0.88_148.041	0.89_148.040	[M + H]+	Tyrosine metabolism
HMDB01180	N2-Succinyl-l-glutamic acid 5-semialdehyde	5.02_463.157	5.02_463.157	5.03_463.156	[2M + H]+	Arginine and proline metabolism
HMDB00562	Creatinine	0.54_227.126	0.54_227.126	0.54_227.126	[2M + H]+	Arginine and proline metabolism
HMDB04370	N-Methyltryptamine	3.03_175.123	3.04_175.124	3.05_175.124	[M + H]+	Tryptophan metabolism
HMDB12267	N-Succinyl-l,l-2,6-diaminopimelate	6.16_291.121	6.17_291.122	6.18_291.122	[M + H]+	Lysine metabolism

^aValues are presented as retention time (minute)_mass (Da); T1, T2 and T3 denote the first trimester, the second trimester and the third trimester, respectively.

3.2. Urine metal profile during normal pregnancy

By using an ICP-MS approach, sixteen metals in maternal urine in different trimesters were quantified. Table 3 shows the detailed information on urine metal concentrations within each trimester and comparisons of median values between trimesters. As shown in Table 3, the levels of five metals—vanadium, manganese, copper, barium, and lead—increased significantly in the second trimester and this high level was maintained in the third trimester. The levels of six metals—aluminum, iron, zinc, arsenic, strontium, and thallium—did not change significantly in the second trimester but increased significantly in the third trimester. Maternal urine metal profile was further ranked by the random forest model in R software. As shown in Fig. 1A, cobalt was assigned the highest importance score. As the most important metal according to the random forest model, the level of cobalt was found to increase time-dependently in maternal urine during the course of normal pregnancy (Fig. 1B).

Table 3. Urine metals concentrations (μg g^–1 creatinine) within each trimester and comparison of median values between trimesters.

Metals	LOQ (μg l–1)	Trimester	Total population (N = 232)		Between-trimester comparison‡
			Median	IQR	Pair	p value
Al	5.50 × 10–01	T1	37.96	91.52	T2–T1	1.22 × 10–04
		T2	64.63	462.72	T3–T2	2.66 × 10–12
		T3	829.25	2683.39	T3–T1	2.20 × 10–16
V	7.67 × 10–03	T1	0.37	0.23	T2–T1	2.20 × 10–16
		T2	0.68	0.57	T3–T2	8.22 × 10–01
		T3	0.70	0.57	T3–T1	2.20 × 10–16
Mn	3.71 × 10–02	T1	0.89	1.42	T2–T1	3.31 × 10–07
		T2	1.70	2.6	T3–T2	5.32 × 10–02
		T3	2.12	3.03	T3–T1	8.06 × 10–10
Fe	1.94 × 10–01	T1	22.21	27.31	T2–T1	7.91 × 10–04
		T2	33.10	44.27	T3–T2	3.10 × 10–05
		T3	49.23	76.7	T3–T1	6.42 × 10–11
Co	4.48 × 10–04	T1	0.18	0.17	T2–T1	2.20 × 10–16
		T2	0.67	0.69	T3–T2	1.35 × 10–09
		T3	1.15	1.24	T3–T1	2.20 × 10–16
Cu	8.63 × 10–02	T1	9.37	4.5	T2–T1	2.20 × 10–16
		T2	15.82	7.91	T3–T2	3.06 × 10–02
		T3	17.20	12.5	T3–T1	2.20 × 10–16
Zn	1.53 × 10–02	T1	235.00	163.4	T2–T1	1.64 × 10–03
		T2	285.15	230.62	T3–T2	6.91 × 10–10
		T3	400.41	335.71	T3–T1	2.20 × 10–16
As	2.00 × 10–03	T1	23.65	14.29	T2–T1	1.17 × 10–04
		T2	30.42	16.94	T3–T2	6.14 × 10–02
		T3	31.55	20.92	T3–T1	9.88 × 10–07
Se	2.87 × 10–02	T1	17.15	5.34	T2–T1	6.69 × 10–02
		T2	18.55	6.63	T3–T2	7.63 × 10–03
		T3	20.00	9.17	T3–T1	8.07 × 10–05
Rb	2.66 × 10–03	T1	1874.50	885.1	T2–T1	8.94 × 10–04
		T2	2002.90	873.8	T3–T2	1.45 × 10–01
		T3	1930.40	906.4	T3–T1	2.19 × 10–01
Sr	1.28 × 10–02	T1	170.27	148.43	T2–T1	1.50 × 10–04
		T2	228.00	164.49	T3–T2	2.90 × 10–01
		T3	230.97	215.49	T3–T1	1.52 × 10–06
Cd	1.00 × 10–03	T1	0.47	0.41	T2–T1	1.50 × 10–02
		T2	0.57	0.37	T3–T2	1.65 × 10–01
		T3	0.58	0.56	T3–T1	1.37 × 10–03
Cs	4.60 × 10–04	T1	9.77	3.87	T2–T1	1.53 × 10–03
		T2	10.64	4.15	T3–T2	9.57 × 10–01
		T3	10.62	4.61	T3–T1	1.93 × 10–02
Ba	3.77 × 10–01	T1	2.92	3.47	T2–T1	2.20 × 10–16
		T2	9.81	11.99	T3–T2	9.27 × 10–01
		T3	9.67	10.66	T3–T1	2.20 × 10–16
Tl	6.67 × 10–04	T1	0.37	0.25	T2–T1	1.32 × 10–04
		T2	0.44	0.21	T3–T2	2.54 × 10–02
		T3	0.47	0.31	T3–T1	1.98 × 10–07
Pb	2.30 × 10–02	T1	2.60	1.87	T2–T1	1.38 × 10–07
		T2	3.47	3.14	T3–T2	2.80 × 10–01
		T3	3.10	3.44	T3–T1	9.61 × 10–05

Fig. 1. Total importance of maternal urine metals generated by random forest model. A: Random forest scores of maternal urine metals profile. B: Box plot of the summary of the urine concentration of Co, the most important metal according to the random forest model. T1, T2 and T3 denote the first trimester, the second trimester and the third trimester, respectively. **p < 0.05, vs. T1; ##p < 0.05, vs. T2.

3.3. The interaction between urine metals and amino acid metabolic intermediates during normal pregnancy

We conducted a regression fitting using the data of sixteen metals in the urine of pregnant women as independent variables and intermediates of amino acid metabolism (2-oxoarginine, 3-indoleacetonitrile, creatinine, indole, indole-5,6-quinone, N2-succinyl-l-glutamic acid 5-semialdehyde, N-methyltryptamine and N-succinyl-l,l-2,6-diaminopimelate) as the dependent variables using the EGLS panel model. Among the sixteen metals, seven metals were found to be significantly related to metabolites of certain amino acid metabolism pathways (Table 4). Previously, we demonstrated that 3-indoleacetonitrile increased in the second trimester and that indole and indole-5,6-quinone increased in the third trimester according to partial least-squares-discriminant analysis.17 In the present study, 3-indoleacetonitrile was found to be significantly correlated with cobalt, indole was found to be significantly correlated with cadmium, cobalt and thallium, and indole-5,6-quinone was found to be significantly correlated with cobalt and vanadium. Cobalt, which was assigned the highest importance score in the random forest model, had significant correlations with all the selected metabolites (2-oxoarginine, 3-indoleacetonitrile, creatinine, indole, indole-5,6-quinone, N2-succinyl-l-glutamic acid 5-semialdehyde, N-methyltryptamine and N-succinyl-l,l-2,6-diaminopimelate) in the present study. Cadmium significantly affected six metabolites: 2-oxoarginine, creatinine, indole, N2-succinyl-l-glutamic acid 5-semialdehyde, N-methyltryptamine and N-succinyl-l,l-2,6-diaminopimelate. Cesium significantly affected urine levels of N2-succinyl-l-glutamic acid 5-semialdehyde. Copper significantly affected urine levels of creatinine, N2-succinyl-l-glutamic acid 5-semialdehyde and N-succinyl-l,l-2,6-diaminopimelate. Manganese significantly affected the urine levels of creatinine and N2-succinyl-l-glutamic acid 5-semialdehyde. Thallium significantly affected urine levels of 2-oxoarginine, indole, N2-succinyl-l-glutamic acid 5-semialdehyde and N-methyltryptamine. Vanadium significantly affected the urine levels of 2-oxoarginine, indole-5,6-quinone, N2-succinyl-l-glutamic acid 5-semialdehyde and N-succinyl-l,l-2,6-diaminopimelate. Fig. 2 shows the interaction network of urine metals and the intermediates involved in amino acid metabolism.

Table 4. Panel regression analysis of the relationship between specific metabolites and urine metals of the normal pregnant cohort generated from the panel model.

	2-Oxoarginine	3-Indoleacetonitrile	Creatinine	Indole	Indole-5,6-quinone	N2-Succinyl-l-glutamic acid 5-semialdehyde	N-Methyltryptamine	N-Succinyl-l,l-2,6-diaminopimelate
R 2	0.877	0.936	0.836	0.881	0.808	0.851	0.818	0.808
F	12.908	26.518	9.250	13.411	7.616	10.396	8.127	7.620
p a	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Al	0.000	0.000	0.002	0.000	0.000	–0.001	–0.000	–0.000
As	0.002	0.002	–0.015	–0.000	0.000	0.012	–0.002	0.015
Ba	–0.001	0.000	–0.003	0.001	0.000	–0.021	0.000	–0.002
Cd	0.249**	0.042	–3.953**	–0.166**	–0.021	8.724**	–0.235**	2.292**
Co	–0.162**	0.521**	8.864**	0.152**	0.201**	–8.361**	0.811**	–1.850**
Cs	0.012	–0.028	0.163	0.017	–0.004	0.525**	0.025	0.020
Cu	0.029	0.016	–0.134**	–0.016	0.008	–0.441**	0.007	–0.156**
Fe	0.000	–0.000	0.016	0.002	0.000	–0.008	0.000	–0.001
Mn	0.006	–0.015	–0.331**	–0.005	–0.007	0.305**	–0.009	0.058
Pb	–0.002	–0.001	–0.045	–0.004	–0.003	0.021	0.004	0.002
Rb	0.000	0.000	–0.007	0.000	–0.000	0.008	–0.001	0.001
Se	–0.025	0.002	0.014	–0.032	–0.012	–0.071	0.036	0.083
Sr	0.000	0.001	–0.021	0.001	–0.000	–0.001	0.003	0.000
Tl	0.352**	0.145	4.522	–0.301**	0.170	10.247**	2.056**	0.844
V	–0.110**	–0.078	0.042	0.014	–0.122**	–1.611*	–0.277	–0.737**
Zn	0.000	0.001	0.024	–0.000	0.000	–0.004	–0.000	0.002

^a p value for F test; *p < 0.05 for t test; **p < 0.01 for t test.

Fig. 2. Interaction network between maternal urine metals and amino acid metabolic intermediates. The circle nodes denote the corresponding metals and the diamond nodes denote the intermediates involved in amino acid metabolism. Edges among metals are constructed based on Pearson's correlation coefficients (r ≥ 0.6 and p < 0.05). Edges between metals and metabolites are constructed based on the significance (p < 0.05) of the regression coefficients from panel regression analysis.

4. Discussion

The biggest challenge in accurately quantifying the dynamic relationship between metal exposure and health effects is the lack of an effective data model. Panel data refer to the data obtained by repeated measurements of the same cross section in a given period of time.19 The panel data model can effectively overcome the multicollinearity of time series analysis and has robust quality and relatively high efficiency in quantifying the dynamic relationship between different stochastic variables.19 The panel data model has already been applied to the monitoring of environmental pollutants and epidemiological studies in recent years.20,21 To date, large-scale panel data on the interaction between urine metals and amino acid metabolic stress in a cohort of healthy pregnant women are lacking. In this study, a UPLC-QTOFMS-based metabolomic approach and ICP-MS-based metallomic approach were employed to investigate the dynamic interaction between metals and amino acid metabolism in a cohort of 232 healthy pregnant women during the course of pregnancy using a panel model. Among sixteen metals, seven metals—cadmium, cobalt, copper, cesium, manganese, thallium and vanadium—showed significant associations with certain amino acid metabolic pathways. Furthermore, two metabolic pathways, tryptophan metabolism and arginine and proline metabolism, were significantly affected by metal exposure during normal pregnancy.

Many neurotransmitters or neuroactive compounds are synthesized in the tryptophan metabolic pathway.22 Previous studies have confirmed that tryptophan metabolism changed significantly in the maternal body during normal pregnancy.17,22 The significant change in the status of tryptophan metabolism indicates fluctuations of neurotransmitters due to metabolic adaption in the maternal body for growth of the fetus during normal pregnancy.22 Indole, 3-indoleacetonitrile and N-methyltryptamine are intermediates involved in tryptophan metabolism. Cobalt, cadmium and thallium showed significant correlations with 3-indoleacetonitrile, indole and N-methyltryptamine, which indicated that these metals may affect tryptophan metabolism through interactions with key enzymes, such as tryptophanase, methyltransferase, and indoleacetaldoxime dehydratase, in the metabolic pathway of 3-indoleacetonitrile, indole and N-methyltryptamine in the maternal body.

N2-Succinyl-l-glutamic acid 5-semialdehyde, 2-oxoarginine and creatinine are intermediates involved in arginine and proline metabolism. N2-Succinyl-l-glutamic acid 5-semialdehyde is a substrate for succinate semialdehyde dehydrogenase (SSADH). SSADH is a key enzyme in γ-aminobutyric acid (GABA) shunting, regulating the generation of inhibitory neurotransmitter GABA in the brain. Metals including thallium, cadmium, cesium and manganese, which result in distinct characteristics of neurotoxicity, showed a significant positive correlation with N2-succinyl-l-glutamic acid 5-semialdehyde, while the essential trace elements, including cobalt, vanadium and copper, showed a significant negative correlation in this study; the underlying mechanism may be related to the regulation of SSADH activity. Creatinine has been found to be decreased in Parkinson's disease.23 In this study, cadmium, copper and manganese were negatively correlated with creatinine, while cobalt showed an opposite tendency. Additionally, 2-oxoarginine is a neurotoxicity-related marker24 and can cause central nervous system damage.25 In this study, cadmium and thallium were positively correlated with 2-oxoarginine, while cobalt and vanadium showed the opposite tendency. These findings may provide a new clue for studies in neurodegenerative diseases.

In addition, intermediates indole-5,6-quinone and N-succinyl-l,l-2,6-diaminopimelate, which are involved in tyrosine metabolism and lysine metabolism, were affected by metals during pregnancy. Indole-5,6-quinone participates in the synthesis of eumelanin and N-succinyl-l,l-2,6-diaminopimelate participates in the synthesis of lysine. Cobalt had a positive correlation with indole-5,6-quinone and a negative correlation with N-succinyl-l,l-2,6-diaminopimelate, while vanadium had a negative correlation with both indole-5,6-quinone and N-succinyl-l,l-2,6-diaminopimelate, indicating that cobalt and vanadium may play an active role in the synthesis of eumelanin and lysine.

In this study, elements such as cesium and cadmium did not change significantly from T2 to T3. However, they still significantly affected the metabolism of amino acids in the maternal body. This deviation could be linked to the effect of hormesis, an interesting phenomenon in which sub-inhibitory amounts of toxin could stimulate growth, physiological or metabolic changes, longevity, and reproductive activity in the body.2 Hormesis reflects the plasticity of biological processes and systems to adapt and respond to different types of stressors.27 The mechanism of hormesis may be related to the activation of protective systems, including the DNA repair system and antioxidant system, and is induced by low-dose exposure to toxins.28,29 During normal pregnancy, a significant correlation between amino acid metabolism and a relatively stable exposure level of metals may be the result of the adaptive process of environmental toxicant stimulation.

Dynamic correlations between metals and intermediates in amino acid metabolic pathways provide important baseline data on the mechanism of metal action in the maternal body of healthy pregnant women during the course of normal pregnancy. As potential biomarkers, these intermediates of amino acid metabolic pathways may have great value in evaluating metal exposure in healthy pregnant women. Although the kidneys process amino acids and organic acids uniquely and differently from metals, significant associations between metals and amino acid metabolic intermediates generated from a cohort of healthy pregnant women with a relatively large sample size will undoubtedly provide future direction for studies on the interaction mechanisms between metals and the human body. Further molecular biology studies on the molecular mechanisms of the effect of metals on amino acid metabolism during pregnancy may be beneficial.

5. Concluding remarks

In this study, we employed an integrated approach of UPLC-QTOFMS-based metabolomics and ICP-MS-based metallomics to obtain urine amino acid metabolic intermediates and metal profiles in a longitudinal cohort of 232 healthy pregnant women during the course of pregnancy and then evaluated the dynamic interactions between metals and amino acid metabolic intermediates using a panel model. The results obtained identified eight intermediates involved in amino acid metabolism. In addition, their correlations with different metals throughout pregnancy were revealed, thus displaying a detailed signature of amino acid metabolism related to metal exposure in normal pregnancy. Within the correlations discovered, metabolites in metabolic networks of tryptophan metabolism, arginine and proline metabolism, tyrosine metabolism and lysine biosynthesis were observed using the most comprehensive set of current data on metal exposure during normal pregnancy. Herein, we illustrated the key pathways regulating neurotransmitter generation and neurotoxicity to explore the roles of metals in the mechanisms of pregnancy-induced adaptation of amino acid metabolism. Our findings will help generate novel hypotheses of the mechanisms of metal action in metabolic adaptation of healthy pregnant women.

Funding

This study was supported by the National Natural Science Foundation of China (21437002, 81372959, 81402649), the R&D Special Fund for Public Welfare Industry (Environment) (201309048), the Natural Science Foundation of Hubei Province (2016CFB541), the Applied Basic Research Program of Wuhan Municipal Science and Technology Bureau (2016010101010003), and the independent innovation research fund, Huazhong University of Science and Technology (2017KFYXJJ069).

Author contributions

Mu Wang performed the experiments, analyzed the data and wrote the manuscript. Han Li, Huailong Chang, Wenyu Liu, Xiaojie Sun, Yangqian Jiang, Hongxiu Liu, Chuansha Wu and Xinyun Pan performed the experiments. Wei Xia, Fang Liu, Hongbin Liu, Jie Sun, Yuanyuan Li, Weiqing Rang and Songfeng Lu reviewed and edited the manuscript. Shunqing Xu designed the experiment, reviewed and edited the manuscript, and is the guarantor of this study. All authors reviewed and approved the final manuscript.

Ethical approval

All procedures performed in this study were in accordance with the ethical standards of the Ethics Committees of the Tongji Medical College, Huazhong University of Science and Technology, and the Study Hospital of the Maternal and Child Health Hospital of Wuhan City in China as well as with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

Written informed consent was obtained from all individual participants included in the study.

Conflicts of interest

The authors declare no conflict of interest.