Placental metabolic profiling in gestational diabetes mellitus: An important role of fatty acids

Abstract Aim Gestational diabetes mellitus (GDM) is the most common metabolic disorder during pregnancy. Accumulating studies have reported metabolites that are significantly associated with the development of GDM. However, studies on the metabolism of placenta, the most important organ of maternal‐fetal energy and material transport, are extremely rare. This study aimed to identify and discuss the relationship between differentially expressed metabolites (DEM) and clinical parameters of the mothers and newborns. Methods In this study, metabolites from 63 placenta tissues (32 GDM and 31 normal controls) were assayed by ultra‐performance liquid chromatography‐high resolution mass spectrometry (UPLC‐HRMS). Results A total of 1297 annotated metabolites were detected, of which 87 significantly different in GDM placenta. Lipids and lipid‐like molecules accounted for 62.1% of DEM as they were significantly enriched via the “biosynthesis of unsaturated fatty acids” and “fatty acid biosynthesis” pathways. Linoleic acid and α‐linolenic acid appeared to be good biomarkers for the prediction and diagnosis of GDM. In addition, the level of PC(14:0/18:0) was negatively correlated with neonatal weight. 14 metabolites significantly different in male and female offspring, with the most increase in female newborns. Conclusion Even if maternal blood glucose level is well controlled, there are still metabolic abnormalities in GDM. Lipids and lipid‐like molecules were the main differential metabolites, especially unsaturated fatty acids.


| INTRODUC TI ON
Pregnancy is accompanied by several metabolic changes which become more intense when women experience complications.
Gestational diabetes mellitus (GDM) is the most common metabolic disorder during pregnancy. Although the molecular mechanism of GDM is still unclear, it is well known that the disease is caused by the complex interaction of genetic and environmental factors, accompanied by a series of metabolic changes. 1,2 Recent advances in metabolomics can provide the most integrated profile of biological status and contribute to better understanding of the etiology and pathogenesis of diseases. GDM has already become one of the subjects of focus in metabonomics with a growing number of studies revealing a range of metabolites, especially including amino acids, organic acids, lipids, and fatty acids, that are significantly associated with the development of GDM. 3,4 The following types of metabonomics studies on GDM are typical: (1) the most common are investigations for predictive biomarkers which can be used for diagnosis and early risk stratification.
Most of these involve studies on blood or urine, as biological specimens. Although many metabolites have been reported, the results are neither consistent nor have they been proven through large sample prospective studies. (2) Exploration of the unique metabolic transition from pregnancy to postpartum. The metabonomic characteristics of the transition from GDM to type 2 diabetes mellitus area recent focus. Allalou 5 identified 21 metabolites that significantly differed when GDM progressed to type 2 diabetes mellitus. The discriminative power was 83.0%, which was far superior to measuring fasting plasma glucose levels alone. The lipid species CE 20:4, PE(P-36:2), and PS 38:4 were also reported as significant risk factors for the progression from GDM to type 2 diabetes. 6 (3) Evaluation of the effect of GDM treatment. Pinto et al., 7 reported that the treatment duration of GDM was related to the urine metabolic profile, when they compared the effects of the excreted metabolome of pregnant GDM women after diet and insulin treatments. (4) Metabolic pathogenesis of GDM. Some dysregulated metabolic pathways were proven to be associated with GDM, such as the metabolism of lipids, amino acids, carbohydrates, and purines. Collectively, these metabolomic-based studies provide new insights into the pathogenesis and potential targets for the treatment and prevention of GDM, mainly through the discovery of useful biomarkers with high sensitivity and specificity.
However, while being the most important organ of maternalfetal energy and material transport, studies of metabolism within the placenta are extremely rare. Some reports have suggested that the placenta has unique metabonomic characteristics in pregnancy complications, including preeclampsia, 8,9 maternal obesity, 10 spontaneous preterm birth, 11 and fetal growth restriction. 12  This study aimed to obtain the comprehensive metabolite profile of GDM placenta through non-targeted metabonomics, plus elucidate their specific metabolic pathways through bioinformatics analysis. The relationship between differentially expressed metabolites (DEM) and clinical data were analyzed comprehensively. We hope these results will contribute to reveal a new molecular mechanism of GDM.

| Ethics approval and consent to participate
The study design and protocol were reviewed and approved by the ethics committee of Changzhou Maternal and Child Health Care Hospital affiliated with Nanjing Medical University. Consent has been obtained from each patient or subject, after full explanation of the purpose and nature of all procedures used.

| Clinical subjects
A total of 63 pregnant women took part in the study, including 32 with GDM and 31 with normal pregnancies. All of the GDM women were diagnosed according to "Guideline No. 393-Diabetes in Pregnancy 14 " The normal pregnant women had neither pregnancy complications nor other basic diseases. Their demographic and clinical assessments are presented in Table 1

| Sample collection
Small pieces of placenta tissue (about 2-5 cm; 1.5 cm 3 ) were harvested by the umbilical cord insertion, immediately (within 5 min) after birth. A total of 20 mg of placental tissue was homogenized, added to tryptophan-d5 and palmitic acid-[13C] 12 as internal standard (1:1, v/v).

| Ultra-performance liquid chromatography separation
Ultra-performance liquid chromatography-high resolution mass spectrometry (UPLC-HRMS) was used in this study. The samples were analyzed with the combined system of Ultimate TM 3000 ultra-performance liquid chromatography and Q Exactive quadrupole-Orbitrap (Thermo Scientific,).
The positive ion detection mode metabolites were separated on an Acquity TM HSS C18 column (Waters Co., 1.7 μm, 2.1 × 100 mm) with 0.1% formic acid aqueous solution and acetonitrile as eluents in linear gradient elution mode. The mobile phases in positive ion detection mode were water, acetonitrile, and methanol, and 5 mM ammonium bicarbonate buffer was added to all solvents, and the metabolites were separated on Acquity TM BEH C18 column (WatersCo., 1.7 μm, 2.1 × 100 mm) with the following elution gradient: The organic phase increased from 2% to 100% in 10 min, and the extra 5 min was used for flushing and balancing the column. The flow rate, sample volume, and column temperature of the two methods were set as 0.4 ml/min, 5 μl, and 50℃, respectively.

| Detection parameters
Heating electrospray ionization was used in the two detection modes, and the other mass spectrum parameters were the same except for the ionization voltage, which was 4KV in the positive ion detection mode and 3.5KV in the negative ion detection mode.
Other mass spectrum parameters were as follows: The sheath gas flow rate is 45 arb, the auxiliary gas flow rate is 10 arb, and the heater temperature is 355℃, the capillary temperature is 320℃, and the ion transmission lens RF level is 55%. Metabolites were

| Statistical analysis
The group differences in maternal age, BMI, gestational age of delivery, newborns' birth weight, and placental metabolomes were analyzed with Student's t test to compare normally distributed data and with Mann-Whitney test to compare non-normal distributions. Chi-square test was used to compare the distribution of newborns' gender between the groups. Since it is not known if metabolites can actually predict GDM, we plotted a receiver operating characteristic curve (ROC), and the overall diagnostic accuracy of the test was summarized by area under the ROC curve (AUC).

The strength of the linear relationship between metabolites and
birth weight of neonates was tested by Spearman rank-based correlation coefficient. All analyses were performed by IBM SPSS Statistics 26.

| Characteristic of GDM placental metabolome
Placental metabolites were determined by UPLC-HRMS, and data were analyzed by PCA and (O)PLS-DA ( Figure 1A). OPLS-DA score plots showed that two groups (GDM vs control) could be clearly separated ( Figure 1B). In total, 1297 annotated metabolites were were thus considered as DEM(VIP>1.0 and p < 0.05) (Table S1). Their volcano plot is presented in Figure 2A, wherein 72 increased and 15 decreased. Figure 2B showed the classification of these DEM. The predominant DEMs were lipids and lipid-like molecules, accounting for 54/87 types (62.1%). Of which, fatty acyls, glycerophospholipids, and prenol lipids were the major three subtypes, comprising 50.0%, 24.1%, and 14.8%, respectively. After hierarchical clustering, a heatmap ( Figure 2C) showed relative abundance of the top 50 DEM in each sample, which clearly displayed the significant separation between GDM cases and control.
Based on the DEM, pathway enrichment was performed by KEGG pathway maps. The top 5 statistically significant pathways of the DEM are presented in Figure 3A. Notably, the DEM in GDM placenta was significantly enriched to "biosynthesis of unsaturated fatty acids" and "fatty acid biosynthesis" pathways, involving 6 lipid molecules (arachidonic acid, palmitic acid, oleic acid, stearic acid, linoleic acid, α-linolenic acid). Compared to that of normal pregnant women, these six types of fatty acids significantly increased in the placenta of GDM women (p < 0.05). The relationship between metabolites and pathways was then analyzed with R pack (3.6.2) ( Figure 3B), and it clearly demonstrated how the fatty acids were involved in the metabolic pathways.

| Clinical evaluation and analysis
We compared the abundance of the six fatty acids in placental tissues. The levels of arachidonic acid, palmitic acid, oleic acid, stearic acid, linoleic acid, and α-linolenic acid significantly increased in GDM placenta compared with those of normal pregnant women (p < 0.05) ( Figure 4A). We then evaluated whether these six fatty acids could also predict the occurrence of GDM. ROC curve analysis confirmed the association. The areas under the curve, 95% confidence interval, and p value are listed in Figure 4B. AUC of linoleic and α-linolenic acid were higher than 0.75 ( Figure 4B).
Secondly, we analyzed the correlation between the different metabolites and some clinical parameters of mothers and newborns. Table 2 showed the statistically significant results. There were many correlations between the DEM and maternal age, gestational age of delivery, newborn birth weight. Due to timely prenatal intervention of GDM women, the birth weight of newborns increased slightly, although it was not statistical significant.
The incidence of macrosomia also did not significantly increase.
However, it is still worth noting that there were four metabolites The levels of α-linolenic acid were also slightly higher in females.
Only four metabolites increased in male newborns, including MG In conclusion, we described the placental metabolic profiling of GDM confirmed that lipids and lipid-like molecules were the main differential metabolites, especially unsaturated fatty acids.
Furthermore, even if maternal blood glucose level is well controlled, TA B L E 2 Results of correlation analysis between metabolites and clinical parameters (p < 0.05) there are still metabolic abnormalities in GDM and sex-specific alterations. Placental metabolites such as PC(14:0/18:0) correlate with the birth weight of newborns. These findings will contribute to reveal new molecular mechanisms of GDM.

ACK N OWLED G M ENTS
We thank all the project participants for their contributions.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

E TH I C A L A PPROVA L
The study design and protocol were reviewed and approved by the ethics committee of Changzhou Maternal and Child Health Care Hospital affiliated with Nanjing Medical University (2021140).
Consent has been obtained from each patient or subject after full explanation of the purpose and nature of all procedures used.

DATA AVA I L A B I L I T Y S TAT E M E N T
The questionnaire and datasets used are available from the corresponding author on request. TA B L E 3 Sex-specific differences in placental metabolomes (10E8)