Nuclear Magnetic Resonance-Assisted Metabolic Analysis of Plasma for Mild Gestational Diabetes Mellitus Patients

Abstract

Aims:

This study aimed at detecting the difference of metabolic profile in mild gestational diabetes mellitus (GDM) patients, and at identifying the correlation between the disturbed metabolites and insulin resistance pathway induced by the inflammation factors secreted by the placenta and adipocytes.

Materials and Methods:

Blood samples were collected from 36 pregnant women diagnosed with mild GDM and 36 normal glucose-tolerant (NGT) pregnant women. Metabolic profiles were acquired by using proton nuclear magnetic resonance (¹H-NMR) spectroscopy, and the data were analyzed by multivariate statistical data analysis. The messenger RNA (mRNA) expression of CX3CL1, CD14, and CD86 in placenta and omental adipose tissues was detected by quantitative real-time polymerase chain reaction.

Results:

(1) Weight and body mass index in GDM group were remarkably higher at both pre-pregnancy and delivery than those in the NGT group. The levels of fasting blood glucose, insulin, and homeostasis model assessment-2 for insulin resistance in GDM group were also significantly higher than those in the NGT group. (2) The expression of mRNA of CD86 was significantly higher in placenta of GDM patients. However, the gene expression of CX3CL1 and CD86 in omental adipose tissues of GDM was markedly lower as compared with the control group. (3) The GDM group exhibited severely perturbed glucose metabolism, fatty acid metabolism, amino acid metabolism, and activated inflammatory response.

Conclusions:

The ¹H-NMR-based metabonomics approach together with biochemical assay and the mRNA expression analysis provided metabolic information about mild GDM, which indicated disturbed glucose metabolism, fatty acid metabolism, amino acid metabolism, and activated inflammatory response in mild GDM patients. This integrated study provides a new idea and methods to understand underlying mechanisms of mild GDM in clinics.

Introduction

Gestational diabetes mellitus (GDM) is identified as glucose intolerance for a woman who develops high blood glucose levels during pregnancy. It affects ∼17% of all pregnancies worldwide.^1
–3 Most of GDM cases (70%–85%) could be effectively amended by diet control and physical exercise, which were classified as mild GDM.⁴ GDM increases risk factors of poor prognosis in both mother and fetus, such as preeclampsia, macrosomia, operative deliveries, polyhydramnios, shoulder dystocia, birth injuries, and neonatal hypoglycemia.^5
–7

As a metabolic disorder, GDM is of some characteristics similar to obesity and type 2 diabetes.^8,9 It occurs when β cells fail to provide sufficient insulin demands or insulin resistance (IR).¹⁰ The pathogenesis of IR in GDM has not been fully understood. It probably involves the activation of inflammatory response within metabolic and endocrine tissues.^11
–13 During inflammation, macrophage and monocyte infiltrate in the placenta and omental adipose tissue,^14
–16 resulting in increased expression of monocyte and macrophage markers, such as CD86, CD14, and CX3CL1.^16,17 Cytokine secretion in placental and omental adipose tissue might induce inflammation and trigger IR.^16,18 In addition, pregnancy complications still often occurred among part of gestational diabetes patients even if the blood glucose level was satisfactorily controlled.¹⁹ Therefore, it is critical to investigate the relationship between the inflammatory status of placenta and omental adipose tissue and IR in mild GDM.

Metabolomics has been an important tool for diabetes research.^19
–21 Unlike genomics, transcriptomics, and proteomics, metabolomics probes low-molecular-weight metabolites in tissues or biofluids,^22
–24 and it can determine the changes of metabolites in response to pathological events.²⁵ Nuclear magnetic resonance (NMR)-assisted metabolomics, as a non-destructive and high-throughput “omics” tool, makes it highly applicable for disease diagnosis, progression, and prognosis.^25
–27

The metabolome of blood sample can well represent the whole status of the body. In this study, plasma samples from 36 mild GDM patients and 36 normal glucose-tolerant (NGT) pregnancies were collected. ¹H NMR-assisted metabolic analysis combined with biochemical analysis was applied to determine the metabolites in plasma of the mild GDM group and the control group. CD86, CD14, and CX3CL1 messenger RNA (mRNA) expression in placental and omental adipose tissue was used to uncover the inflammatory status of both groups. The aim of this study was to investigate metabolic changes in mild GDM patients and to provide a better understanding of the initial stage of GDM.

Materials and Methods

Study population and experimental design

All subjects were recruited from the department of obstetrics, the first affiliated hospital of Nanjing Medical University, China. GDM was diagnosed according to the International Association of Diabetes and Pregnancy Study Groups (IADPSG).²⁸ Women who had multiple pregnancies and any adverse medical conditions (i.e., including asthma, pre-gestational diabetes, heart disease, and pre-eclampsia) were excluded. GDM patients who were admitted on insulin or glucose controlled unsatisfactorily were also excluded. Finally, 72 pregnancy women with singletons who selected cesarean section for delivery between January and July 2012 were recruited, including two groups of mild GDM (n = 36) and NGT (n = 36). Women in the NGT group had a negative screening result used as a control. Indications of cesarean section included abnormal fetal position, repeat caesarean section, or cephalopelvic disproportion.

This trial was registered with the Ethics committee of the first affiliated hospital of Nanjing Medical University (No. 2013SR061). Written informed consent to participants in the study was obtained from each individual before recruitment.

Demographic data and sample collection

All pregnant women were fasted overnight, and cubital venous blood samples were collected into ethylenediaminetetraacetic acid-prepared tubes before cesarean section. Then, the plasma was obtained by centrifugation (12,000 rpm, 10 min, 4°C) and stored at −80°C before analysis.

Placenta (maternal side) and omental adipose tissues were collected from each pregnant woman during operation, ∼1 cm³ in volume. Tissues were quickly frozen in liquid nitrogen and stored at −80°C until use for quantitative real-time polymerase chain reaction (qRT-PCR).

Blood biochemical analysis

Fasting blood glucose (FBG) concentrations were measured by the Olympus AU2700 fully automatic biochemical analyzer. Fasting insulin (FIN) and C-peptide were assayed by the electrochemiluminescence technique (Roche, Germany). IR was calculated by using the homeostasis model assessment for insulin resistance (HOMA-IR) method, where HOMA-IR = FBG (mmol/L) × FIN (μU/mL)/22.5.

RNA extraction and qRT-PCR

Total RNA of placenta and omental adipose tissues were isolated by using Trizol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. The concentrations and purities of RNA were assessed by NanoDrop^® ND-1000 spectrophotometry (Thermo Scientific, Rockford, IL). According to the manufacturer's instructions, 1 μg of total RNA was converted to complementary DNA by using the Primescript™ RT Master Mix (Takara Bio, Inc., Otsu, Shiga, Japan). qRT-PCR was performed with LightCycler 480 (Mannheim, Germany) by using the SYBR green qRT-PCR mix (Takara Bio, Inc.) according to the manufacturer's protocol. The sequences of primers used for qRT-PCR were designed and synthesized by Invitrogen as shown in Table 1.

Table 1.

Real-Time Polymerase Chain Reaction Primer Sequences

Genes	Forward primer (5′–3′)	Reverse primer (5′–3′)
CX3CL1	ACCACGGTGTGACGAAATG	TGTTGATAGTGGATGAGCAAAGC
CD14	CAACCTAGAGCCGTTTCTAAAGC	GCGCCTACCAGTAGCTGAG
CD86	CTGCTCATCTATACACGGTTACC	GGAAACGTCGTACAGTTCTGTG
GAPDH	GCACCGTCA AGGCTGAGA AC	TGGTGAAGACGCCAGTGG A

¹H-NMR spectroscopy

Before the NMR study, 300 μL of each plasma sample was added to 300 μL D₂O phosphate buffer [0.2 mol L⁻¹ Na₂HPO₄ and 0.2 mol L⁻¹ NaH₂PO₄, pH 7.4, including 0.05% 3-trimethylsilylpropionic acid (TSP)]. The samples were mixed and centrifuged at 12,000 g for 10 min at 4°C, and then, the supernatants were placed in a 5 mm wide NMR tube for NMR analysis.

¹H-NMR spectra were recorded on a 500 MHz flow-injection NMR spectrometer (Bruker AV500 spectrometer; Bruker, Karlsruhe, Germany) at 300K. The Carr-Purcell-Meiboom-Gill spin-echo pulse sequence (RD-90°-(τ-180°-τ)_n-ACQ) with a total spin-echo delay (2nτ) of 40 ms was used to suppress broad signals from macromolecules (i.e., proteins or lipoproteins).¹⁶ A total of 128 scans were collected in 32K data points over a spectral width of 10,000 Hz for each ¹H-NMR spectrum. Before Fourier transformation, a line broadening of 0.5 Hz was applied to the free induction decays. All plasma ¹H-NMR spectra were referenced to TSP (CH₃, δ 0.0), followed by manual phase and baseline correction by using Topspin 3.0 software (Bruker GmbH).

Data processing and analysis

All ¹H-NMR spectra were exported to ASCII files by using MestReNova (Version 8.0.1; Mestrelab Research S.L.), which were then imported into “R” (http://cran.r-project.org) for multivariate data analysis with an in-house developed R-script. The NMR spectra were binned into an average of 0.015 ppm integrated spectral bucket by using an adaptive algorithm between 0.70 and 8.00 ppm, and the signals of residual water (4.20–5.25 ppm) were excluded. To compensate different sample dilutions, the integral values of the residual bins were probability quotient normalized. Before multivariate data analysis, the values of all binned spectra were mean-centered and pareto-scaled.²⁹

Spectral data were first analyzed by an unsupervised principal component analysis (PCA) to give an overview of the metabolic patterns and trends, and to detect any outliers among samples. However, no obvious separation was observed in the PCA score (data not shown). A supervised orthogonal partial least-squares discriminant analysis (OPLS-DA) was used to maximize covariance between the measured data and the response variable.³⁰ The OPLS-DA method was built by using an orthogonal filter to remove systematic variations unrelated to interested status.

The scores plots displayed the patterns of samples, and the corresponding loading plots were used to reveal variables contributing to grouping. The loadings plot were color coded according to the absolute value of the coefficient values: The warm-colored (e.g., red) metabolites were more significant than the cold-colored (e.g., blue) ones. A repeated two-fold cross validation and 2000 times permutation test were used to further validate the OPLS-DA modes. The validity of the models avoiding over-fitting was assessed by the parameter R², and the predictive ability was described by Q².

Statistical analysis

The differences between GDM and NGT groups were evaluated by a parametric Student's t test or nonparametric Mann–Whitney test (according to the conformity of the data to normal distribution). A P value of ≤0.05 was considered statistically significant. The fold change (FC) values of metabolites and their associated P values corrected by Benjamini–Hochberg-adjusted method³¹ were calculated and visualized in Table 3.

Correlation network

Correlation network of the GDM group was built by using the R package igraph. In the network, metabolites with correlation coefficients above a threshold (0.65) are joined by solid lines that are color coded according to the correlation coefficients (a cool color for negative and a warm color for positive correlations). The nodes (framed in circle) represent metabolites or other parameters, and the dotted lines represent a similar structure of connected metabolites. The names of metabolites were colored in red or blue to denote the statistically significant increase and decrease of metabolites as compared with the NGT group, respectively.

This improved correlation network presents both the Pearson's correlation coefficients among the levels of metabolites and their structural similarity. Pearson's correlation networks generally show only the links between metabolites with diverse status. But, such correlations do not show causal connections. The addition of structural similarity information improves the networks since the substrates and products of nearly all of the represented biochemical reactions should be analogous. In this context, a high correlation between metabolites with a high structural similarity might demonstrate a theoretically conceivable biochemical reaction between them and, consequently, a causative effect.

Correlation network between metabolites and other parameters

To integrate metabolites and other parameters including IR, all data were subjected to canonical analysis with the metabolites as X variables and other parameters as Y variables. The nodes in the network denote metabolites and the related parameters, and the colored lines between the nodes represent a correlation between them (red and blue means positive and negative correlation, respectively).

Shared and unique structure plot

In the shared and unique structure (SUS) plots, the metabolites scattered across the red dotted line represent those that were positively correlated (++/−−) and those scattered across the green dotted line represent those that were negatively correlated (±/±). Those that are found in the red boxes across the plot axes are either positive or negative for that particular model and represent the “unique structures.” It was also important to note that metabolites that were on the extreme ends of the axes contribute more significantly than those close to the center.

Results

Demographic and biochemical analysis

Demographic data and biochemical analysis of the two groups are summarized in Fig. 1. Compared with the NGT group, the levels of weight and body mass index (BMI) in the GDM group were remarkably high, at both pre-pregnancy and delivery. The levels of FBG, 1 hr OGTT (oral glucose tolerance test), 2 hr OGTT, FIN (fasting insulin), and HOMA2-IR in the GDM group were significantly high as compared with the NGT group. No differences were detected in maternal age, height, BMI gain, systolic and diastolic blood pressure, blood glucose, fasting C-peptide, gestational age, and fetal weight between GDM cases and controls.

FIG. 1.

Boxplots for age, height, pre-pregnancy weight, pre-pregnancy BMI, pre-delivery weight, pre-delivery BMI, BMI gain, systolic blood pressure and diastolic blood pressure, FBG, 1 hr OGTT, 2 hr OGTT, blood glucose, FIN, HOMA2-IR, C-peptide, gestational age, and fetal weight. The bottom of each box, the line drawn in the box, and the top of the box represent the first, second, and third quartiles, respectively. The whiskers extend to 1.5 times the interquartile range (from the first to third quartile). Data in top represent the P values. BMI, body mass index; BMI gain, BMI gain during pregnancy; FBG, fasting blood glucose; FIN, fasting insulin; HOMA2-IR, homeostasis model assessment-2 of insulin resistance; 1 hr OGTT, 1-hour oral glucose tolerance test; 2 hr OGTT, 2-hour oral glucose tolerance test. Color images available online at www.liebertpub.com/met

Expression of CX3CL1, CD14, and CD86 mRNA in placenta and omental adipose tissue

The mRNA expression of CD86 (an indicator of macrophage and monocyte infiltration) was significantly higher in placenta of the GDM group than that in the NGT group. The mRNA levels of CX3CL1 and CD14 in placenta of the GDM group were slightly higher. However, the mRNA expression of CX3CL1 and CD86 in omental adipose tissues of the GDM group was lower than that in the control group (Table 2).

Table 2.

Comparison of CX3CL1, CD14, and CD86 Messenger RNA Expression in Human Placenta and Omental Adipose Tissues for Gestational Diabetes Mellitus Patients and Normal Glucose-Tolerant Controls

	Placenta			Adipose tissues
Genes	NGT (n = 20)	GDM (n = 20)	P	NGT (n = 20)	GDM (n = 20)	P
CX3CL1	0.99 ± 0.13	1.58 ± 1.33	0.06	2.61 ± 1.79	1.57 ± 0.75	0.02
CD14	1.08 ± 0.89	1.98 ± 2.45	0.31	1.93 ± 1.31	2.07 ± 1.22	0.73
CD86	2.65 ± 1.78	12.39 ± 9.89	<0.01	2.78 ± 1.27	1.65 ± 1.07	<0.01

GDM, gestational diabetes mellitus; NGT, normal glucose tolerant.

¹H-NMR spectra analysis

Representative ¹H-NMR spectra of plasma samples from control and GDM groups with major metabolites labeled are shown in Fig. 2. Resonances were assigned according to database querying (Madison: http://mmcd.nmrfam.wisc.edu, KEGG: www.genome.jp/kegg, HMDB: www.hmdb.ca, etc.), aided by STOCSY and Chenomx NMR suit (Version 7.5; Chenomx, Inc.).^28,32

FIG. 2.

Typical 500 MHz ¹H-NMR spectra of plasma samples obtained from the NGT group (black line), GDM (red line) pregnant women. Metabolites: 1. LDL/VLDL, 2. Leucine/isoleucine, 3. Valine, 4. 3-Hydroxybutyrate, 5. Lactate, 6. Alanine, 7. Arginine, 8. Pyruvatoxime, 9. Acetoacetate, 10. Pyruvate, 11. Creatinine, 12. Trimethylamine N-oxide, 13. GPC, 14. Triethanolamine, 15. Glycerol-3-phosphate. GDM, gestational diabetes mellitus; GPC, glycerophosphorylcholine; ¹H-NMR, proton nuclear magnetic resonance; LDL/VLDL, low-density lipoprotein/very low-density lipoprotein; NGT, normal glucose tolerant. Color images available online at www.liebertpub.com/met

Metabolic changes in GDM group

In the scores plot (Fig. 3A), the two groups were well separated from each other. The corresponding loadings plot and S-plot (Fig. 3B, C) revealed increased levels of low-density lipoprotein/very low-density lipoprotein (LDL/VLDL), pyruvatoxime, arginine, acetoacetate, and triethanolamine, and decreased levels of leucine/isoleucine, valine, 3-hydroxybutyrate, lactate, alanine, pyruvate, creatinine, trimethylamine N-oxide, glycerophosphorylcholine (GPC), and glycerol-3-phosphate in the GDM group. The OPLS-DA models revealed significant variations among groups (Fig. 3D) and were validated by a repeated twofold cross-validation and 2000 times permutation test (Fig. 3E). The corresponding FC plots and table are shown in Fig. 3F and Table 3, respectively.

FIG. 3.

OPLS-DA scores plot (A), color-coded coefficient loading plots (B), S-plot (C) of metabolomic profiles in plasma. OPLS-DA scatter plots from plasma (D) of statistical validation obtained by 2000 times permutation test, with R² and Q² values in the vertical axis, the correlation coefficient (between the permuted and true class) in the horizontal axis, and the OLS line for the regression of R² and Q² on the correlation coefficients. Histograms for permutation test scores of OPLSDA model of NGT and GDM group on the basis of 2000 permutations: the red arrows reveal the performance based on the original labels, significant for a P-value less than 0.05 (E). Fold change plots color coded according to the corresponding P values adjusted by the BH method, indicating significance of altered metabolites in plasma (F). The blue dotted lines and red dashed lines represent an increase or decrease of 20% and 100%, respectively. BH, Benjamini–Hochberg; OLS, ordinary least squares; OPLS-DA, orthogonal partial least-squares discriminant analysis; TMAO, trimethylamine N-oxide. Color images available online at www.liebertpub.com/met.

Table 3.

Identified Metabolites with the Fold Change Between Different Groups and P Value

			GDM vs. NGT
Metabolite	Assignments	Chemical shift (ppm)	Fold change	P
LDL/VLDL	γCH₃, βCH	0.89 (m), 1.20–1.30 (m)	1.21	0.285
Leucine/isoleucine	δCH₃, γCH₃, αCH	0.94 (t), 0.96 (t)	0.75	0.004
Valine	γCH₃, γCH₃	0.99 (d), 1.04 (d)	0.48	0.004
Lactate	βCH₃, αCH	1.32 (d), 4.1 (q)	0.86	0.043
Alanine	CH₃	1.47 (d), 3.77 (q)	0.91	0.159
Arginine	γ-CH₂, β-CH₂	1.70–1.75 (m), 1.91–1.95 (m)	1.13	0.960
Pyruvatoxime	CH₃	2.02 (m)	1.10	0.320
Acetoacetate	CH₂	2.35 (s)	1.28	0.189
Pyruvate	CH₃	2.36 (s)	0.97	0.960
Creatinine	CH₃	2.72–2.74 (s)	0.90	0.331
TMAO	NH₂-CH₂, SO₃-CH₂	3.27 (s)	0.91	0.320
GPC	CH₃	3.21 (s)	0.72	0.004
Triethanolamine	CH₃	3.39–3.41 (s)	8.25	<0.001
Glycerol-3-phosphate	CH	3.25 (t), 3.41 (t)	0.72	0.006

Color coded according to log₂ (fold change) using color bar . A p-value of p ≤ 0.05 was considered statistically significant. Multiplicity: s, singlet; d, doublet; t, triplet; q, quartet; m, multiplet.

GPC, glycerophosphorylcholine; LDL/VLDL, low-density lipoprotein/very low-density lipoprotein; TMAO, trimethylamine N-oxide.

Correlation network of differential metabolites in GDM group

To evaluate the relations of metabolites, a correlation network analysis of the GDM group (Fig. 4) was performed by using the R package igraph. Located in the center of the network, acetoacetate shows its significance in GDM patients, whose glucose utilization is greatly hampered. Acetoacetate is one of the ketone bodies that could be used as an important energy material to supplement insufficient energy supply in the GDM group. LDL/VLDL, another important energy material, can be oxidized to afford energy and rescue energy crisis, so LDL/VLDL is positively correlated to acetoacetate.

FIG. 4.

Correlation network analysis of the GDM group, performed by using the igraph R package. The nodes (framed in circle) represent metabolites, and correlations above a threshold (0.65) are shown to be connected by solid colored lines (color coded according to the correlation coefficients, a cool color for negative and a warm color for positive correlations). The dotted lines represent the compounds of a similar structure. Color images available online at www.liebertpub.com/met

Branched-chain amino acids (BCAAs) (valine, leucine, and isoleucine) act as the essential amino acids that play an important role in body protein synthesis, which showed strong mutual positive correlations in the correlation network. Arginine is strongly and positively correlated with pyruvate, and they are known to have a close relationship in a biochemical reaction. Arginine is a glycogenic amino acid and can be used to synthesize glucose that is further decomposed into pyruvate. Our results demonstrated that the energy and amino acid metabolism might be perturbed in GDM pregnant women.

Correlation network between metabolites and other parameters

Several important parameters were filtered out, including IR-related parameters 1 and 2 hr OGTT; and risk factors such as BMI and weight (Fig. 5).

FIG. 5.

The correlation network of differential metabolites and other parameters in plasma. The network is graphically represented with metabolites and other parameters as nodes, and the colored lines indicate correlations (red for positive and blue for negative correlations, color coded according to the correlation coefficients). Only correlations with absolute values of correlation coefficients that exceeded 0.6 were kept. Color images available online at www.liebertpub.com/met

SUS plot analysis

The whole data were subdivided into different subset data, cut in half to afford young and old halves, and low and high halves for age and BMI, respectively. Metabolomics analyses were made on these subset, which were compared with the results obtained on the whole data by SUS analysis.

The SUS plots revealed most metabolites distributed near the red dotted line, indicating that the results of these metabolite changes were mostly independent of age, pre-pregnancy BMI, pre-delivery BMI, and BMI gain during pregnancy (Fig. 6). Only few metabolites, such as GPC and glycerol-3-phosphate in Fig. 6F, were found unique metabolites for pre-delivery BMI. For the unique metabolites, their levels in subgroups were visualized by boxplots (Fig. 6).

FIG. 6.

The SUS plots of plasma show the shared and unique correlation metabolites. (A, B) SUS-plot indicated the comparisons of metabolites between all pregnancies and old pregnancies, all pregnancies and young pregnancies, respectively. (C, D) SUS-plot indicated the comparisons of metabolites between all pregnancies and high pre-pregnancy BMI pregnancies, all pregnancies and low pre-pregnancy pregnancies, respectively. (E, F) SUS-plot indicated the comparisons of metabolites between all pregnancies and high pre-delivery BMI pregnancies, all pregnancies and low pre-delivery BMI pregnancies, respectively. (G, H) SUS-plot indicated the comparisons of metabolites between all pregnancies and high BMI gain pregnancies, all pregnancies and low BMI gain pregnancies, respectively. Boxplots for unique metabolites (GPC and glycerol-3-phosphate) of pre-delivery BMI pregnancies (I, J). The bottom of each box, the line drawn in the box, and the top of the box represent the first, second, and third quartiles, respectively. The dotted line represents average value. The whiskers extend to 1.5 times the interquartile range (from the first to third quartile). Data in top represent the P values. SUS, shared and unique structure. Color images available online at www.liebertpub.com/met

The levels of GPC and glycerol-3-phosphate in the GDM group were significantly lower than the NGT group in the high half of pre-delivery BMI divided subset data, but not significant in the low half, which suggested that the decreases of GPC and glycerol-3-phosphate were strongly associated with high pre-delivery BMI GDM patients. As an important gluconeogenesis intermediate, the pre-delivery BMI-dependent change of GPC might suggest that glucose absorption and utilization capacity in GDM patients with high pre-delivery BMI was weaker than those with low pre-delivery BMI. Glycerol-3-phosphate is an intermediate of glycolysis; its pre-delivery BMI-dependent change suggested that GDM patients with high pre-delivery BMI are prone to undergo insufficient energy supply.

Discussion

In this study, a ¹H-NMR-based metabolomics approach was used to study the plasma metabolic profile change associated with GDM, which revealed obvious pattern differences between GDM and NGT groups. Significant perturbations in energy metabolism, amino acid metabolism (isoleucine, leucine, valine, and arginine), as well as inflammatory response were observed in GDM patients. The disturbed metabolism in GDM pregnant women was also investigated by biochemical analysis and mRNA expression.

IR may play an important role in the process of GDM, whereas the pathogenesis of GDM is still unclear.^33,34 Several biochemical parameters were measured to evaluate the status of insulin, including HOMA2-IR, FIN, and C-peptide. HOMA2-IR, an index of IR, can be used to predict GDM.³⁵ FIN also can be used to evaluate insulin sensitivity.³⁶ The levels of HOMA2-IR and FIN were noticeably elevated in the GDM group, demonstrating that insulin sensitivity was significantly reduced in GDM patients and the bodies could not utilize glucose efficiently.

C-peptide is a peptide that is made when proinsulin is split into insulin and C-peptide. The plasma concentrations of C-peptide may provide insights into β cell function.³⁷ In our study, the level of fasting insulin in mild GDM increase was not accompanied by the level of fasting C-peptide increase, which could be explained by reduced insulin clearance (reflecting hepatic IR) rather than β cell dysfunction.³⁸ In most individuals, IR and obesity or higher BMI coexist. The significantly increased weight and BMI in both pre-pregnancy and pre-delivery were observed in the GDM group, indicating that high weight and BMI were risk factors of GDM. Pregnancy overweight and obesity can cause GDM.^32,39
–41

Metabolic analysis indicated that GDM severely perturbed glucose metabolism and fatty acid metabolism, revealing that energy metabolism was disturbed in GDM patients. Compared with the NGT group, levels of glycerol-3-phosphate, lactate, leucine, isoleucine, and valine were markedly decreased in GDM patients. Glycerol-3-phosphate, a glycolysis intermediate, is synthesized by reducing dihydroxyacetone phosphate with glycerol-3-phosphate dehydrogenase. The decrease of glycerol-3-phosphate in the plasma suggested that the production of glycerol-3-phosphate was inhibited. In addition, a notable decrease of lactate, the anaerobic products also indicated that the glycolysis process was inhibited in GDM patients. The insufficient glucose metabolism in the GDM group may be due to the impaired glucose uptake and utilization caused by IR.

To supplement the insufficient energy source, fat fatty acid β oxidation comes to the rescue in GDM patients. The high levels of FBG, 1 and 2 hr OGTT indicate IR or impaired β cell function, leading to an increased risk of diabetes. In the correlation network, triethanolamine exhibited strong positive correlations with 1 and 2 hr OGTT, demonstrating its important role in mild GDM. Triethanolamine was conspicuously increased in the GDM group as compared with the NGT group. Triethanolamine is a precursor of acetaldehyde, which is one important intermediate in the process of glycolysis. The marked increase of triethanolamine suggested that the bioconversion of triethanolamine to acetaldehyde might be blocked, thus contributing to the aggravated energy crisis of the mild GDM group.

The increased levels of lipids (LDL/VLDL) and acetoacetate showed that fat fatty acid β oxidation was accelerated in the GDM group. The increased LDL/VLDL in the GDM group indicated accelerated fat mobilization to produce fatty acids for β oxidation. Acetoacetate was the metabolite of fatty acid β oxidation and could be transformed to acetyl-CoA for the TCA cycle, which could also come to rescue the insufficient energy supply. Glycerol-3-phosphate could be synthesized by fats with glycerol kinase; the significantly deceased glycerol-3-phosphate also indicates an increase in fatty acid β oxidation to supply energy.

GDM is defined as glucose intolerance with onset or first recognition during pregnancy that can disorder glucose metabolism.^42,43 IR leads to insufficient supply of glucose, and the bodies need to improve endogenous glucose production by gluconeogenesis. Alanine is a major glycogenic amino acid that can be used for gluconeogenesis to produce glucose. Compared with the NGT group, the decrease of alanine in the GDM group demonstrated its great consumption for gluconeogenesis. In Fig. 5, the levels of branched-chain amino acids (isoleucine, leucine, and valine) were negatively correlated with pre-pregnancy weight and pre-pregnancy BMI. As important risk factors for GDM,^44,45 high pre-pregnancy weight and BMI were markedly elevated in the GDM group.

As glycogenic amino acids, BCAAs are important for the maintenance of glucose homeostasis.⁴⁶ BCAAs can decomposed in skeletal muscle and transferred to liver by blood, and then produce alanine via a range of transamination.⁴⁷ Finally, the produced alanine was used to maintain the levels of glucose. The significantly decreased levels of BCAAs might be utilized to replenish the diminished glucose in GDM patients who have higher pre-pregnancy weight and BMI than NGT pregnancies. GPC, a precursor of glycerophosphate, is converted to glycerophosphate under glycerophosphocholine phosphodiesterase.⁴⁸ Further, glycerophosphate plays an important role in gluconeogenesis. The markedly decreased levels of GPC could be explained as a consequence of the enhanced gluconeogenesis.⁴⁹

The pathogenesis of IR among GDM involves the triggering of low-grade inflammation.^50
–52 Omental adipose tissue and placenta are believed to be important contributors to inflammatory transcriptomes.¹⁶ CD86 encodes the corresponding protein expressed in antigen-presenting cells, which provides costimulatory signals that are necessary for T cell activation and survival. The mRNA expression of CD86 was conspicuously decreased in adipose tissue of the GDM group, indicating the disruption of T cell homeostasis. The impaired T cell activation resulted in adipose inflammation enhancement that was also reported in type 2 diabetes.⁵³ The stimuli generated by the low-grade inflammation of obese women will challenge the placenta to develop an anti-inflammatory response.¹⁶

Compared with the NGT group, the expression of mRNA of CD86 was significantly higher in placenta of the GDM group than the NGT group, demonstrating that placental tissue accelerated T cell activation to strengthen the immune response for resisting inflammation.^16,30,54 It is reported that the administration of BCAAs could attenuate inflammation.⁵⁵ Therefore, the decreased BCAAs in the GDM group might be ascribed to their use to fight inflammation.⁵⁶

In this study, the ¹H-NMR-assisted metabolomics approach together with biochemical assay and the mRNA expression analysis was first used to estimate the pathogenesis of GDM. These findings indicated disturbed glucose metabolism, fatty acid metabolism, amino acid metabolism, and activated inflammatory response in mild GDM patients.

Footnotes

Acknowledgments

This study was supported by the Natural Science Foundation of China (grant no. 81270710), Jiangsu provincial Science and Technology Projects (grant no. XK201112). The authors thank everyone who helped with this study.

Compliance with Ethical Standards

This study was conducted with the guidance of the Helsinki Declaration and the International Conference on Harmonization-Good Clinical Practices (ICH-GCP), and it was approved by the Ethics Committee of The First Affiliated Hospital of Nanjing Medical University, China.

Author Disclosure Statement

No conflicting financial interests exist.

References

Guariguata

, Linnenkamp

, Beagley

, et al. Global estimates of the prevalence of hyperglycaemia in pregnancy. Diabetes Res Clin Pract, 2014; 103:176–185.

American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care, 2013; 36(Suppl 1):S67–S74.

Mueller-Wieland

, Petermann

, Nauck

, et al. Definition, classification and diagnosis of diabetes mellitus. Diabetes Metab, 2016; 11:S78–S81.

Pinsker

, Shank

, Dassau

, et al. Comment on American diabetes association: Approaches to Glycemic Treatment. Sec. 7. In Standards of Medical Care in Diabetes-2015. Diabetes Care, 2015; 38(Suppl. 1):S41–S48. Diabetes Care, 2015; 38:E174–E174.

Cosson

, Valensi

, Carbillon

. Screening for dysglycaemia during pregnancy: Proposals conciliating International Association of Diabetes and Pregnancy Study Group (IADPSG) and US National Institutes of Health (NIH) panels. Diabetes Metab, 2015; 41:239–243.

Cosson

, Cussac-Pillegand

, Benbara

, et al. The diagnostic and prognostic performance of a selective screening strategy for gestational diabetes mellitus according to ethnicity in Europe. J Clin Endocrinol Metab, 2014; 99:996–1005.

Cosson

, Bihan

, Reach

, et al. Psychosocial deprivation in women with gestational diabetes mellitus is associated with poor fetomaternal prognoses: An observational study. BMJ Open, 2015; 5:e007120.

Russo

, Ross

, Cowart

. Sphingolipids in obesity, type 2 diabetes, and metabolic disease. Handb Exp Pharmacol, 2013; 216:373–401.

Chang

Y-H

, Chang

D-M

, Lin

K-C

, et al. Visfatin in overweight/obesity, type 2 diabetes mellitus, insulin resistance, metabolic syndrome and cardiovascular diseases: A meta-analysis and systemic review. Diabetes Metab Res Rev, 2011; 27:515–527.

10.

Buchanan

, Xiang

. Gestational diabetes mellitus. J Clin Invest, 2005; 115:485–491.

11.

Lekva

, Norwitz

, Aukrust

, et al. Impact of systemic inflammation on the progression of gestational diabetes mellitus. Curr Diab Rep, 2016; 16:26.

12.

Chokwiriyachit

, Dasanayake

, Suwannarong

, et al. Periodontitis and gestational diabetes mellitus in non-smoking females. J Periodontol, 2013; 84:857–862.

13.

Kjos

, Buchanan

, Greenspoon

, et al. Gestational diabetes mellitus: The prevalence of glucose intolerance and diabetes mellitus in the first two months post partum. Am J Obstet Gynecol, 1990; 163(1 Pt 1):93–98.

14.

Chavan

, Ashford

, Wiggins

, et al. The early bioinflammatory milieu in gestational diabetes mellitus (GDM). Am J Obstet Gynecol, 2016; 214:S96–S96.

15.

, Kim

C-S

, Kwon

B-S

, et al. Mesenteric adipose tissue-derived monocyte chemoattractant protein-1 plays a crucial role in adipose tissue macrophage migration and activation in obese mice. Obesity, 2006; 14:1353–1362.

16.

Challier

, Basu

, Bintein

, et al. Obesity in pregnancy stimulates macrophage accumulation and inflammation in the placenta. Placenta, 2008; 29:274–281.

17.

Mrizak

, Grissa

, Henault

, et al. Placental infiltration of inflammatory markers in gestational diabetic women. Gen Physiol Biophys, 2014; 33:169–176.

18.

Lappas

. Activation of inflammasomes in adipose tissue of women with gestational diabetes. Mol Cell Endocrinol, 2014; 382:74–83.

19.

Desoye

, Mouzon

SH-D

. The human placenta in gestational diabetes mellitus. Diabetes Care, 2007; 30:S120–S126.

20.

Ashwal

, Hod

. Gestational diabetes mellitus: Where are we now?. Clin Chim Acta, 2015; 451(Pt A):14–20.

21.

Scifres

, Parks

, Feghali

, et al. Placental maternal vascular malperfusion and adverse pregnancy outcomes in gestational diabetes mellitus. Placenta, 2017; 49:10–15.

22.

Rochfort

. Metabolomics reviewed: A new “Omics” platform technology for systems biology and implications for natural products research. J Nat Prod, 2005; 68:1813–1820.

23.

Coen

, Lenz

, Nicholson

, et al. An integrated metabonomic investigation of acetaminophen toxicity in the mouse using NMR spectroscopy. Chem Res Toxicol, 2003; 16:295–303.

24.

Jordan

, Nordenstam

, Lauwers

, et al. Metabolomic characterization of human rectal adenocarcinoma with intact tissue magnetic resonance spectroscopy. Dis Colon Rectum, 2009; 52:520–525.

25.

Huynh

, Xiong

, Bentley-Lewis

. A systematic review of metabolite profiling in gestational diabetes mellitus. Diabetologia, 2014; 57:2453–2464.

26.

Diaz

, Pinto

, Graca

, et al. Metabolic biomarkers of prenatal disorders: An exploratory NMR metabonomics study of second trimester maternal urine and blood plasma. J Proteome Res, 2011; 10:3732–3742.

27.

Sachse

, Sletner

, Morkrid

, et al. Metabolic changes in urine during and after pregnancy in a large, multiethnic population-based cohort study of gestational diabetes. PLoS One, 2012; 7:e52399.

28.

Pinto

, Almeida

, Martins

, et al. Prediction of gestational diabetes through NMR metabolomics of maternal blood. J Proteome Res, 2015; 14:2696–2706.

29.

Sun

, Li

, Zhou

, et al. Metabolomic analysis reveals metabolic disturbance in the cortex and hippocampus of subchronic MK-801 treated rats. PLoS One, 2013; 8:e60598.

30.

Craig

, Cloareo

, Holmes

, et al. Scaling and normalization effects in NMR spectroscopic metabonomic data sets. Anal Chem, 2006; 78:2262–2267.

31.

Wei

D-D

, Wang

J-S

, Li

M-H

, et al. A pilot study of the onset of hepatic encephalopathy (OHE) in mice induced by thioacetamide and the protective effect of taurine by holistic metabolic characterization. Metabolomics, 2015; 11:559–570.

32.

Catalano

, McIntyre

, Cruickshank

, et al. The hyperglycemia and adverse pregnancy outcome study associations of GDM and obesity with pregnancy outcomes. Diabetes Care, 2012; 35:780–786.

33.

Sivan

, Boden

. Free fatty acids, insulin resistance, and pregnancy. Curr Diabetes Rep, 2003; 3:319–322.

34.

Verma

, Boney

, Tucker

, et al. Insulin resistance syndrome in women with prior history of gestational diabetes mellitus. J Clin Endocrinol Metab, 2002; 87:3227–3235.

35.

Ghasemi

, Tohidi

, Derakhshan

, et al. Cut-off points of homeostasis model assessment of insulin resistance, beta-cell function, and fasting serum insulin to identify future type 2 diabetes: Tehran lipid and glucose study. Acta Diabetol, 2015; 52:905–915.

36.

Druet

, Tubiana-Rufi

, Chevenne

, et al. Characterization of insulin secretion and resistance in type 2 diabetes of adolescents. J Clin Endocrinol Metab, 2006; 91:401–404.

37.

Mykkanen

, Zaccaro

, Hales

, et al. The relation of proinsulin and insulin to insulin sensitivity and acute insulin response in subjects with newly diagnosed Type II diabetes: The Insulin Resistance Atherosclerosis Study. Diabetologia, 1999; 42:1060–1066.

38.

Retnakaran

, Hanley

AJG

, Sermer

, et al. The impact of insulin resistance on proinsulin secretion in pregnancy—Hyperproinsulinemia is not a feature of gestational diabetes. Diabetes Care, 2005; 28:2710–2715.

39.

Kim

, England

, Wilson

, et al. Percentage of gestational diabetes mellitus attributable to overweight and obesity. Am J Public Health, 2010; 100:1047–1052.

40.

Mottola

. The role of exercise in the prevention and treatment of gestational diabetes mellitus. Curr Diabetes Rep, 2008; 8:299–304.

41.

Catalano

, deMouzon

. Maternal obesity and metabolic risk to the offspring: Why lifestyle interventions may have not achieved the desired outcomes. Int J Obes (Lond), 2015; 39:642–649.

42.

Retnakaran

, Qi

, Sermer

, et al. Glucose intolerance in pregnancy and future risk of pre-diabetes or diabetes. Diabetes Care, 2008; 31:2026–2031.

43.

Dempsey

, Butler

, Sorensen

, et al. A case-control study of maternal recreational physical activity and risk of gestational diabetes mellitus. Diabetes Res Clin Pract, 2004; 66:203–215.

44.

Persson

, Pasupathy

, Hanson

, et al. Pre-pregnancy body mass index and the risk of adverse outcome in type 1 diabetic pregnancies: A population-based cohort study. BMJ Open, 2012; 2:e000601.

45.

Ray

, Burrows

, et al. MOS HIP: McMaster outcome study of hypertension in pregnancy. Early Hum Dev, 2001; 64:129–143.

46.

Layman

, Walker

. Potential importance of leucine in treatment of obesity and the metabolic syndrome. J Nutr, 2006; 136:319S–323S.

47.

. Intestinal mucosal amino acid catabolism. J Nutr, 1998; 128:1249–1252.

48.

Daimon

, Soga

, Hozawa

, et al. Serum glycerophosphate levels are increased in Japanese men with type 2 diabetes. Intern Med, 2012; 51:545–551.

49.

Chen

, Boyko

, Bergstrom

, et al. Earlier appearance of impaired insulin-secretion than of visceral adiposity in the pathogenesis of niddm-5-year follow-up of initially nondiabetic Japanese-American men. Diabetes Care, 1995; 18:747–753.

50.

Bastard

, Maachi

, Lagathu

, et al. Recent advances in the relationship between obesity, inflammation, and insulin resistance. Eur Cytokine Netw, 2006; 17:4–12.

51.

, Barnes

, Yang

, et al. Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance. J Clin Invest, 2003; 112:1821–1830.

52.

Dandona

, Aljada

, Bandyopadhyay

. Inflammation: The link between insulin resistance, obesity and diabetes. Trends Immunol, 2004; 25:4–7.

53.

Zhong

, Rao

, Braunstein

, et al. T-cell costimulation protects obesity-induced adipose inflammation and insulin resistance. Diabetes, 2014; 63:1289–1302.

54.

Gillum

, Kotas

, Erion

, et al. SirT1 regulates adipose tissue inflammation. Diabetes, 2011; 60:3235–3245.

55.

Ohno

, Tanaka

, Sugauchi

, et al. Suppressive effect of oral administration of branched-chain amino acid granules on oxidative stress and inflammation in HCV-positive patients with liver cirrhosis. Hepatol Res, 2008; 38:683–688.

56.

Atashak

, Baturak

. The effect of BCAA supplementation on serum C-reactive protein and creatine kinase after acute resistance exercise in soccer players. Ann Biol Res, 2012; 3:1569–1576.

Nuclear Magnetic Resonance-Assisted Metabolic Analysis of Plasma for Mild Gestational Diabetes Mellitus Patients

Abstract

Aims:

Materials and Methods:

Results:

Conclusions:

Introduction

Materials and Methods

Study population and experimental design

Demographic data and sample collection

Blood biochemical analysis

RNA extraction and qRT-PCR

1 H-NMR spectroscopy

Data processing and analysis

Statistical analysis

Correlation network

Correlation network between metabolites and other parameters

Shared and unique structure plot

Results

Demographic and biochemical analysis

Expression of CX3CL1, CD14, and CD86 mRNA in placenta and omental adipose tissue

1 H-NMR spectra analysis

Metabolic changes in GDM group

Correlation network of differential metabolites in GDM group

Correlation network between metabolites and other parameters

SUS plot analysis

Discussion

Footnotes

Acknowledgments

Compliance with Ethical Standards

Author Disclosure Statement

References

¹H-NMR spectroscopy

¹H-NMR spectra analysis