Metabonomic Profiling of Human Placentas Reveals Different

Dec 24, 2013 - WHO Collaborating Center for Reproductive Health and Population Science, ... Health Science Center, Peking University, 38 Xueyuan Road,...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/jpr

Metabonomic Profiling of Human Placentas Reveals Different Metabolic Patterns among Subtypes of Neural Tube Defects Yi Chi,† Lijun Pei,‡ Gong Chen,§,‡ Xinming Song,§,‡ Aihua Zhao,∥ Tianlu Chen,∥ Mingming Su,⊥ Yinan Zhang,∥ Jianmeng Liu,# Aiguo Ren,# Xiaoying Zheng,*,§,‡ Guoxiang Xie,*,∥,⊥ and Wei Jia∥,⊥ †

School of Pharmacy, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai 200240, China Institute of Population Research, Peking University, 5 Yiheyuan Road, Beijing 100871, China § WHO Collaborating Center for Reproductive Health and Population Science, 5 Yiheyuan Road, Beijing 100871, China ∥ Center for Translational Medicine, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, 600 Yishan Road, Shanghai 200233, China ⊥ University of Hawaii Cancer Center, 701 Ilalo Street, Honolulu, Hawaii 96813, United States # Health Science Center, Peking University, 38 Xueyuan Road, Beijing 100871, China ‡

S Supporting Information *

ABSTRACT: Neural tube defects (NTDs) are one of the most common types of birth defects with a complex etiology. We have previously profiled serum metabolites of pregnant women in Lvliang prefecture, Shanxi Province of China, which revealed distinct metabolic changes in pregnant women with NTDs outcome. Here we present a metabonomics study of human placentas of 144 pregnant women with normal pregnancy outcome and 115 pregnant women affected with NTDs recruited from four rural counties (Pingding, Xiyang, Taigu, and Zezhou) of Shanxi Province, the area with the highest prevalence worldwide. A panel of 19 metabolites related to one-carbon metabolism was also quantitatively determined. We observed obvious differences in global metabolic profiles and one-carbon metabolism among three subtypes of NTDs, anencephaly (Ane), spina bifida (SB), and Ane complicated with SB (Ane & SB) via mass-spectrometry-based metabonomics approach. Disturbed carbohydrate, amino acid, lipid, and nucleic acid metabolism were identified. Placental transport of amino acids might be depressed in Ane and Ane & SB group. Deficiency of choline contributes to Ane and Ane & SB pathogenesis via different metabolic pathways. The formation of NTDs seemed to be weakly related to folates. The metabonomic analysis reveals that the physiological and biochemical processes of the three subtypes of NTDs might be different and the subtype condition should be considered for the future investigation of NTDs. KEYWORDS: neural tube defects, metabonomics, placenta, spina bifida, anencephaly, metabolic profiling, one-carbon metabolism, folate, mass spectrometry



INTRODUCTION Neural tube defects (NTDs), arising from the failure of neural tube closure during early pregnancy, are the second most common human birth defects.1 NTDs affect about 1 infant per 1000 births worldwide each year.2 China is among the countries with high incidence of NTDs, approximately 1.3‰.3 Specially, Lvliang prefecture of Shanxi Province in China has been reported to be the most prevalent area of NTDs worldwide with an overall rate of 19.9‰.4 Our previous report revealed that the incidence of NTDs in four rural counties (Pingding, Xiyang, Taigu, and Zezhou) of Shanxi Province in China is also extremely high, with an overall rate of 13.87‰.5 Among all types of NTDs, spina bifida (SB) and anencephaly (Ane) are the two most common NTDs clinically.6 Ane is lethal and most fetuses with Ane would die in utero or shortly after birth. More than 60% of the live-born fetuses with Ane could only survive for several minutes to 1 day, and ∼90% of the live-born fetuses would die within 5 days.7 © 2013 American Chemical Society

Although SB is not fatal, it can lead to lifelong disabilities, and children with SB need multiple surgeries. It was estimated that children with SB needed an average of five operations annually during the first 5 years after birth, and the average annual cost of treatment during the first two decades after birth was up to $70,000.8 NTDs cause a serious burden to many families and the society. Therefore, the screening of potential risk factors and further investigations in the etiology of NTDs are necessary for NTD’s early diagnosis and nutritional intervention. It is widely accepted that NTDs are closely related to the onecarbon metabolism,9,10 and much research has demonstrated that folate2,11,12 and homocysteine13,14 play important roles in the occurrence of NTDs. Supplementation of folate before pregnancy and during the first trimester was proved to be Received: September 26, 2013 Published: December 24, 2013 934

dx.doi.org/10.1021/pr4009805 | J. Proteome Res. 2014, 13, 934−945

Journal of Proteome Research

Article

Table 1. General Information of Pregnant Women in Four Rural Counties (Pingding, Xiyang, Taigu, and Zezhou) of Shanxi Province control number Ane SB Ane & SB other total gestational age (weeks) mean maternal age (years) mean fetal gender male female unknown previous birth defects history yes no unknown fever or flu during early pregnancy Yes No unknown smoking during early pregnancy yes no drinking during early pregnancy yes no folic acid supplementation yes no unknown

NTDs

Pa

144

28 55 20 12 115

35.45 ± 7.02

34.26 ± 7.22

0.141

28.05 ± 5.54

27.43 ± 5.35

0.334

70 (49) 71 (49) 3 (2)

62 (54) 51 (44) 2 (2)

3 (2) 138 (96) 3 (2)

3 (3) 111 (96) 1 (1)

21 (14) 121 (84) 2 (2)

23 (20) 88 (77) 4 (3)

1 (1) 143 (99)

2 (2) 113 (98)

14 (10) 130 (90)

10 (9) 105 (91)

15 (10) 115 (80) 14 (10)

18 (16) 92 (80) 5 (4)

a P values were calculated using Student’s t test. Brackets inside the digital represent the percentage of the total subjects. Ane: anencephaly, SB: spina bifida, other includes Enc (encephalocele, n = 8), SB & Enc (n = 2), Ane & Enc (n = 1), and SB & Ane & SB (n = 1).

effective in preventing NTDs.9,15 Zhang et al.16 quantified the involved metabolites in the folate- and homocysteine-related one-carbon metabolism and found that concentrations of several intermediate metabolites were significantly changed in samples of women with NTD-affected pregnancies compared with those of normal controls. However, the current intervention strategy (folate supplementation) cannot prevent all NTDs, and some research even reported that folate did not play a protective role.17−19 It had been used for nearly 20 years, while the best result is a reduction of no more than 40−70% of NTD occurrences or recurrences.20 One of our published papers21 reported that there was no significant difference in serum total folate and vitamin B12 between pregnant women affected with NTDs and pregnant women with normal outcomes in Lvliang area of Shanxi Province. The etiology of NTDs is complicated, and it has been generally accepted that NTDs were caused by complex interactions between genetic and environmental factors.22 More than 200 genes have been confirmed to be associated with NTDs in mice.23,24 Mutation of 5,10-methylenetetrahydrofolate reductase (MTHFR) that participates in folate cycle was found to be risk factor of NTDs in some human populations, but studies on other populations showed that mutation of MTHFR was irrelevant or a protective factor of NTDs.25 Recently, mutations in Van g-Like 1

(VANGL1) and VANGL2, which affect the planar cell polarity, have been observed in humans.26,27 However, progress in mechanistic studies underlying the molecular basis of human NTDs is limited.25 There’s no doubt that NTDs are multifactorial diseases that may be beyond methylation or one-carbon unit metabolism. Therefore, it is necessary to investigate any other metabolic changes related to NTDs for better understanding. With the development of the analytical technology and chemometrics, metabonomics has become an alternative and promising approach for understanding and elucidating the etiology and mechanisms of human diseases13,28−30 and has been extensively applied to multiple areas of life science, such as toxicology, nutrition, and functional genomics.31−33 Several studies have shown that the metabolic pattern of pregnant women affected with NTDs was distinct from normal women and several altered pathways including inhibited TCA cycle, decreased branched-chain amino acids, neurotransmitters, sphingosine-1-phosphate, galactosylsphingosine, 3-oxohexadecanoic acid, fructose-6-phosphate, docosahexaenoic acid, dehydroepiandrosterone sulfate, and linoleic acid, and increased lysophosphatidylcholine and leukotrienes related to NTDs were identified.32,34−36 However, there are very few reports that 935

dx.doi.org/10.1021/pr4009805 | J. Proteome Res. 2014, 13, 934−945

Journal of Proteome Research

Article

Figure 1. Bar plots of total folate and VB12 in placentas of different trimesters (total = 0−43 weeks, 2nd trimester = 14−26 weeks, 3rd trimester = 27−43 weeks) and in different subtypes (Ane: n = 28, SB: n = 55, SB & Ane: n = 20) from 2nd trimester samples. *P < 0.05 obtained from Wilcoxon−Mann− Whitney test. The bar means the median concentration of the metabolite for each group or trimester, and the error bar represents a 95% confidence interval.

cysteine, cystathionine, histidine, serine, choline, betaine, N,Ndimethylglycine, acetylcholine, ethanolamine, 5′-cytidine diphosphocholine (CDP-choline), citric acid (CA), ascorbic acid (AA), dithiothreitol (DTT), methoxylamine, and N,O-bistrimethylsilyl-trifluoroacetamide (BSTFA) (containing 1% trimethylchlorosilane (TMCS)) were purchased from Sigma− Aldrich (St. Louis, MO). Formic acid, acetonitrile, methanol, pyridine, and chloroform for HPLC grade were purchased from CNW Technologies (German); ultrapure water (18.2 MΩ) was prepared with a Milli-Q water purification system (Millipore, Billerica, MA).

systemically delineated the placental metabolic profiles in NTDs, especially among different subtypes. Here we conducted a metabonomics study of 144 pregnant women with normal outcome and 115 pregnant women affected with NTDs with different subtypes in four rural counties (Pingding, Xiyang, Taigu, and Zezhou) of Shanxi Province using gas chromatography−time-of-flight mass spectrometry (GCTOFMS) and ultra-performance liquid chromatography− quadrupole time-of-flight mass spectrometry (UPLCQTOFMS) to profile the metabolic composition of placental samples. A panel of 19 metabolites involved in one-carbon metabolism was quantified using UPLC−triple-quadrupole mass spectrometry (UPLC-TQMS) due to the vital role of one-carbon metabolism in NTDs formation.37 Total folate and vitamin B12 (VB12) were also quantified via chemiluminescence immunoassay (CLIA).



Human Subjects

The study protocol was reviewed and approved by the Institutional Review Board of Peking University, and written informed consent was obtained from all subjects. The diagnosis and classification of NTDs were performed through a population-based birth defects surveillance program by trained clinicians. Information on the mother’s socio-demographic characteristics, lifestyle, reproductive history, periconceptional use of folate supplements, smoking, and so on was obtained with a questionnaire (Supplementary Appendix 1, Supporting Information). In this study, placental samples of pregnant

MATERIALS AND METHODS

Chemicals and Reagents

Folate, tetrahydrofolate (THF), 5-methyltetrahydrofolate (5MTHF), vitamin B6 (VB6), S-adenosylhomocysteine (SAH), Sadenosylmethionine (SAM), homocysteine, methionine, cystine, 936

dx.doi.org/10.1021/pr4009805 | J. Proteome Res. 2014, 13, 934−945

Journal of Proteome Research

Article

Figure 2. PLS-DA scores plots of (A) global placental metabolic profiles and (B) 19 quantified metabolites of three subtypes of NTDs. The scores plots clearly showed the differences among the three subtypes.

differential metabolites between each subtype of NTDs and controls. Default seven-fold cross-validation procedure was automatically carried out to avoid model overfitting and two fundamental parameters R2Y and Q2Y were attained accordingly in SIMCA-P. The value of Q2Y ≥ 0.4 is generally considered to be a reliable model.40 Receiver operating characteristic (ROC) curve analysis was performed to assess the reliability of the OPLS-DA model.41 Metabolites with variable importance in the projection (VIP) > 1 and P < 0.05 from nonparametric Wilcoxon−Mann−Whitney test were considered significant. Quantified results via UPLC-TQMS analysis were processed using Wilcoxon−Mann−Whitney test with the critical P value set to 0.05. Heatmap was created to visualize variations of quantified metabolites in each subtype of NTDs. Heatmap and ROC curve analysis were carried out using R (www.r-project.org). Pearson correlation analysis was performed with IBM SPSS Statistics 21.0.

women with normal outcomes (n = 144) and NTDs (n = 115) were collected from four rural counties (Pingding, Xiyang, Taigu, and Zezhou) of Shanxi province, China (Table 1). Placenta was collected at delivery or termination of NTD-affected pregnancies and kept at −80 °C until analysis. Immunoassay

Procedure of tissue sample pretreatment for immunoassay is provided in the Supplementary Methods in the Supporting Information. The concentrations of total folate and VB12 in placentas were determined by an immunoassay analyzer ARCHITECT i200 (Abbott Laboratories, Abbott Park, IL) following manufacturer’s protocols. Metabolic Profiling

Placentas were extracted using our published two-step extraction method38 with minor modifications. GC-TOFMS and UPLCQTOFMS were used for detection of small-molecule metabolites. The acquired MS data files were extracted and pretreated following standard operation procedure of our laboratory. Details are provided in the Supporting Information.



RESULTS

Total Folate and Vitamin B12

Quantitative Analysis of Metabolites involved in One-Carbon Metabolism by UPLC-TQMS

Concentrations of total folate and VB12 in all 259 human placentas were quantified using chemical luminescence immune test. Neither total folate nor VB12 shows significant difference between NTDs group and controls. We then compared these two parameters at different pregnancy trimesters, and no significant differences were observed between the two groups (Figure 1A,B). However, as illustrated in Figure 1C,D, total folate and VB12 were significantly increased in placentas of all pregnant women affected with SB & Ane as compared with controls (P = 0.031 for total folate, P = 0.028 for VB12) without significant difference between SB (or Ane) and normal controls.

Sample preparation for UPLC-TQMS analysis is provided in the Supporting Information. A Waters Acquity UPLC system (Waters, Milford, MA) coupled to an Applied Biosystems SelexION 5500 triple-quadrupole mass spectrometry (AB Sciex, Dublin, CA) was used for the quantification of selected metabolites related to one-carbon metabolism. Multiple reaction monitoring (MRM) mode was applied. MRM transitions and MS parameters were optimized for each analyte individually (Supplementary Table S1 in the Supporting Information.). The acquired MS data were analyzed using Analyst 1.5.2 (AB Sciex).

Placenta Metabonomic Profiling of Human NTDs

Statistical Analysis

As shown in Supplementary Figure S1A in the Supporting Information, no clear separation was observed between NTDs and controls from the OPLS-DA model (R2X = 0.184, R2Y = 0.276, Q2Y = 0.115 with one predictive component and one orthogonal component). However, obvious differences in metabolic profiles between each subtype of NTDs (Ane, SB, Ane & SB) and controls (Supplementary Figure S1B−D in the Supporting Information) were observed, particularly between Ane & SB group and normal controls. The scores plot of PLS-DA model (Figure 2A) clearly showed the separations among the three different subtypes (R2X = 0.376, R2Y = 0.86, Q2Y = 0.51).

The raw MS data were pretreated following our published protocols28,39 (see Supplementary Methods in the Supporting Information). The combined data set resulting from GCTOFMS and UPLC-QTOFMS was normalized, mean-centered, and UV-scaled (unit variance scaling) before multivariate statistical analysis (SIMCA-P 12.0.1, Umetrics, Umeå, Sweden). Partial least-squares−discriminant analysis (PLS-DA) was used to visualize the different metabolic profiles among all groups including subtypes of NTDs. Orthogonal partial least-squares− discriminant analysis (OPLS-DA) was used to identify the 937

dx.doi.org/10.1021/pr4009805 | J. Proteome Res. 2014, 13, 934−945

Journal of Proteome Research

Article

Figure 3. OPLS−DA scores plots and ROC curves of training and test set samples. (A,D) Ane versus controls. (B,E) SB versus controls. (C,F) Ane & SB versus controls. The area under the curves was larger than 0.8, indicating reliable OPLS-DA models.

DA model using one predictive component and one orthogonal components (Figure 3C, R2X = 0.255, R2Y = 0.904, Q2Y = 0.645) derived from Ane & SB group and control group also indicated significant metabolic relationships between the two groups. To evaluate the robustness and predictive ability of the models, we performed ROC curve analysis using cross-validated predicted Y of the training sets and predicted Y of the test sets (Figure 3D− F). Area under ROC curves (AUC) was 0.930, 0.876, and 0.961 in the training sets and 0.887, 0.811, and 0.877 in the test sets for Ane, SB and Ane & SB group, respectively, demonstrating the good performance of the OPLS-DA models.42 On the basis of VIP > 1 of the OPLS-DA models and P < 0.05 of nonparametric Wilcoxon−Mann−Whitney test, a total of 29, 16, and 56 differentially expressed metabolites in each subtype of

To test the performance of the models derived from each subtype of NTDs and controls, ∼80% of the subjects of each subtype were randomly selected, and an equal number of controls with matched gestational weeks and mother ages was picked out as the training set. Residual subjects of each subtype and controls were considered as the test set (Supplementary Table S2 in the Supporting Information). Distinct metabolic profiles between Ane group and controls could be observed from OPLS-DA scores plot (Figure 3A, R2X = 0.303, R2Y = 0.905, Q2Y = 0.489 with one predictive component and two orthogonal components) and metabolic profiles between SB group and controls as well (Figure 3B, R2X = 0.227, R2Y = 0.83, Q2Y = 0.409 with one predictive component and two orthogonal components). The cross-validated OPLS938

dx.doi.org/10.1021/pr4009805 | J. Proteome Res. 2014, 13, 934−945

Journal of Proteome Research

Article

Table 2. Differential Metabolites Derived from OPLS-DA Model with Wilcoxon−Mann−Whitney Test in Each Subtype of NTDsa Ane versus control metabolites amino acid metabolism asparagineLC prolineGC ornithineGC quinaldateLC tryptophanGC N-acetyl-5-hydroxytryptamineLC dopamineLC dopaLC 4-hydroxy-3-methoxymandelateLC N-acetyl-tyrosineLC normetanephrineLC tyrosineGC 2-hydroxyisocaprateNIST isoleucineGC leucineGC valineGC phenylalanineGC hydrocinnamateGC alanineGC cadaverineGC glutamineGC methionineLC 2-aminobutyrateGC glutamateGC homocysteineGC N-acetyl-asparatelc O-phospho-serineLC pyroglutamateGC asparateGC lysineLC norleucineGC histidineGC N-acetyl-lysineLC threonineLC glycineGC serineGC carbohydrate metabolism pyruvateGC glycerol phosphateGC ribulose-5-phosphateLC sorbitolGC ribitolGC xylitolGC glucoseLC gluconateGC myo-inositolGC malateGC cis-aconitateLC succinateGC lipid metabolism 1-monopalmitinGC 1-stearoyl-glycerolGC arachidonic acidGC caproic acidGC cholesterolGC docosahexaenoicacidGC linoleic acidGC

SB versus control

Ane & SB versus control

VIPb

Pc

FCc

VIPb

Pc

FCc

VIPb

Pc

FCc

0.92 0.35 1.22 1.58 1.50 0.04 1.43 0.68 0.60 0.13 0.83 1.07 1.33 1.93 1.90 1.80 1.65 0.76 1.46 1.11 1.85 1.67 0.88 0.20 0.63 1.53 0.13 0.01 1.00 0.60 0.32 1.35 0.59 1.29 0.70 0.60

0.302 0.227 0.542 0.049 0.034 0.532 0.007 0.814 0.433 0.009 0.346 0.105 0.017 0.003 0.002 0.009 0.008 0.004 0.028 0.002 0.004 0.018 0.213 0.981 0.581 0.063 0.805 0.787 0.185 0.925 0.341 0.080 0.051 0.082 0.336 0.398

1.20 1.23 1.11 1.40 1.45 1.11 1.60 1.04 0.88 0.64 1.14 1.32 1.52 1.69 1.72 1.58 1.59 1.65 1.47 1.72 1.66 1.51 1.24 1.00 0.91 1.33 0.96 1.05 1.26 1.02 1.18 1.35 1.40 1.35 1.18 1.16

0.71 0.03 0.33 0.42 0.60 1.71 0.69 0.45 0.84 0.51 0.29 0.95 0.58 0.22 0.27 0.01 0.40 0.30 0.27 0.82 1.85 1.24 2.19 1.91 1.76 1.20 2.67 2.32 0.81 0.72 0.71 0.53 0.89 1.05 0.67 0.33

0.169 0.720 0.884 0.172 0.200 0.038 0.355 0.559 0.463 0.738 0.408 0.249 0.287 0.867 0.499 0.881 0.418 0.450 0.780 0.223 0.003 0.223 0.008 0.011 0.043 0.010 0.001