Identification of differentially expressed non-peptide molecules with metabolomics approach in pregnancy-induced hypertension

Introduction

Pregnancy-induced hypertension (PIH) is a form of high blood pressure in pregnancy. PIH is also called toxemia or preeclampsia. It is the most common medical condition in pregnancy and is reported to affect up to 9.4% ~10.4% of all gestations. PIH occurs most often in young women with a first pregnancy. Expectant mothers with hypertension are predisposed toward the potential harmful complications, mainly abruptio placentae, disseminated intravascular coagulation, cerebral hemorrhage, hepatic failure, and acute renal failure. These potential adverse effects endanger maternal and fetal health [1]. PIH is a pregnancy syndrome characterized by hypertension and proteinuria after 20 weeks of gestation, which is defined as a systolic pressure 140 mmHg or greater, or a diastolic pressure 90 mmHg or greater. The basic pathology change of PIH is systemic small artery spasm and the clinical manifestations are hypertension, edema, and proteinuria. The National High Blood Pressure Education Program of the National Heart, Lung, and Blood Institute has classified hypertensive disorders of pregnancy into the following categories: chronic hypertension, gestational hypertension, preeclampsia-eclampsia, and preeclampsia superimposed on chronic hypertension [2]. Although physicians for millennia have recognized PIH, comparatively little is known about its pathogenesis and prevention [3]. A unified view accepted by most of physicians is that PIH is caused jointly by multiple factors. In recent years, with the development of molecular biology in obstetrics, great progress has been made in research of etiology of PIH.

Earlier studies showed that several risk factors have been described as predisposing to PIH, which include age of pregnant women, psychological problems during pregnancy, family history of hypertension, income level, weight, and number of pregnancies [4-6]. The knowledge of the most important risk factors could be useful for identifying the patients who have higher chances to develop the hypertensive disorders. Then, adequate prenatal care could contribute to decrease this mortality ratio. The pathophysiology of PIH is not clear yet. Many experts consider that the placenta plays a central role in pathogenesis of PIH, especially preeclampsia, because delivery of the placenta is the only definitive cure for this disease [2, 7]. Thus, many researches have focused on the changes in the maternal blood vessels that supply blood to the placenta [2]. The placenta undergoes dramatic vascularization to enable circulation between fetus and matrix during normal pregnancy. Placental vascularization includes vasculogenesis, angiogenesis, and maternal spiral artery remodeling [8]. These processes need a delicate balance of pro-angiogenic and anti-angiogenic factors. The imbalance of pro-angiogenic and antiangiogenic factors in PIH, especially preeclampsia, is thought to trigger abnormal placental vascularization and disease onset. In general, the angiogenic process is initiated by growth factors such as vascular endothelial growth factor (VEGF), basic fibroblast growth factor (bFGF), and placental growth factor (PlGF). PlGF is produced mainly by the placenta and has potent pro-angiogenic effects. In normal uncomplicated pregnancies, PlGF levels rise until the 32nd week, approximately, and then fall until delivery. In pregnancies complicated by preeclampsia before the 37th week with or without intrauterine growth restriction, PlGF levels are significantly lower [9, 10].

The imbalance of pro- and anti-angiogenic factors is a primary focus of researchers examining the pathogenesis of PIH. In preeclampsia patients, levels of the soluble VEGF receptor-1 (sFlt-1), which binds circulating VEGF and PlGF, are elevated; this in turn leads to a decrease in the concentrations of the bioactive forms of these growth factors. This imbalance is thought to cause the placental and systemic endothelial dysfunction that induces the clinical symptoms of the disease. The role of determining sFlt-1, PlGF and other angiogenic factor levels in maternal peripheral blood in predicting and diagnosing preeclampsia has been extensively studied in recent years [11-13]. Unfortunately, there is still little information available about the levels of these factors in other forms of hypertension in pregnancy or their ability to predict outcomes [14-17]. In addition, vascular endothelium damage factors, such as: interleukin-8, 10 (IL-8, 10) [18, 19], tumor necrosis factor (TNF) [20, 21], lipid peroxidation (LPO) [22], soluble Fms-like tyrosine kinase 1 (sFlt1) [23] and fibronectin degradation product (FnDP), are associated with PIH [24].

Although the underlying cause of PIH is still unknown, the most widely accepted hypothesis is that placental ischemia/hypoxia results from inadequate uteroplacental vascular remodeling, which leads to a decrease in placental blood flow. The ischemic placenta releases factors such as the soluble VEGF receptor-1 (sFlt-1), the angiotensin II type-1 receptor autoantibody (AT1-AA), and cytokines such as TNF-α and Interleukin 6 which cause maternal endothelial dysfunction characterized by elevated circulating endothelin (ET), reactive oxygen species (ROS), and enhanced vascular sensitivity to angiotensin II. The family of endothelins comprise three isoforms (ET-1, ET-2, and ET-3) and each of them has 21 amino acids. ET-1 is the most abundant member of this family [24]. ET-1 has been suggested to contribute to hypertension in preeclampsia. Recently, ET-1 has also been implicated in the induction of both oxidative stress and endoplasmic reticulum stress in preeclampsia; each of them has been proposed to contribute to many of the clinical manifestations of this disorder. In plasma from healthy pregnant women, the concentration of ET-1 ranges from 5 to 10 pg/ml, whereas the concentration of ET-1 is from 20 to 50 pg/ml in the presence of preeclampsia [25]. The potent vasodilator NO is produced in the presence of L-arginine and oxygen by one of three NO synthase (NOS) isoforms encoded by at least three different genes. A major contributor to vascular hemodynamic regulation is the endothelial NOS (eNOS) isoform [26]. Multiple posttranslational modifications have been reported to determine NO production. In preeclampsia, systemic vascular resistance strikingly increases resulting in compromised uteroplacental perfusion and high blood pressure. The expression of eNOS in the uteroplacental compartment is controversially reported. While some authors identify a decreased expression [27-29], others report on a normal [30] or even enhanced level of the enzyme [31]. An indirect evidence for a major role of NO is the beneficial effect of L-arginine supplementation to preeclamptic women with an improved neonatal outcome in the treatment group, which might be due to an enhanced NO production in response to an enhanced substrate supply [32]. These studies suggested that detection of metabolites could present a valuable strategy to investigate the metabolic risk factors associated with HIP. However, most of the metabolites reported were related to larger molecule metabolism, especially to peptide(s) and/or protein(s) but a few have been connected with non-peptide small molecules that have been differentially metabolized in the placentas of women with PIH. The aim of our study was to investigate the differential metabolomic non-peptide molecules presented in PIH applying a metabolomic approach.

Materials and Methods

1) Specimens: A total number of 29 placentas were taken including 14 from women with PIH and 15 from women with normal pregnancies. According to national ethic regulation, informed consent was obtained from the pregnant women. The research project was approved by the Hospital Ethic Review Committee. Detailed information about the placentas obtained from the patients affected by PIH and women from the control group with normal blood pressure is provided in Table 1.

2) Procedures for collection of the placentas: Placentas were taken immediately after delivery, flushed with cold distilled water to clean the blood from the surface of the placenta, dissected into a cube of 1x1 cm2 (2 cm from the edge of the placenta) that was immersed in the placental membrane, and frozen immediately in the liquid nitrogen for at least 30 minutes (the entire process should be done within 30 minutes before the placenta is dropped into the liquid nitrogen), and then kept at -80oC (without frozen- thaw) until use.

3) Extraction of non-peptide small molecules: 150 mg placenta tissue was ground in liquid nitrogen to a powder, soaked in 0.6 ml of methanol extraction solvent at 4oC, homogenized in an ice-cold mortar and pestle, and transferred to an Eppendorf (EP) tube. The EP was centrifuged at 4oC (15 min, 3,000g). The top layer was transferred to fresh EP tubes for lyophilization.

4) High pressure liquid chromatography-mass spectrometry (HPLC–MS) analysis: Lyophilized small molecule extracts derived from the nonpolar chloroform extract were reconstituted in 100ml of 95:5 water: acetonitrile, vortex, and centrifuged (15min, 13,000g). The supernatant was transferred to a 2 ml chromatography vial, sealed with a septum containing screw cap and stored in an autosampler at 4oC. 15ml samples were analyzed on an HPLC system (UltiMate3000, DIONEX) coupled to a micrOTOF-Q II mass spectrometer (https://www.bruker.com, Bruker, Germany). Chromatographic separations were performed with a HPLC 3.5um C18 column (Agilent) at a flow rate of 0.25 ml/min in positive ion modes. A 15ml sample was introduced into the column. Centroid and accurate mass spectra were acquired in the 50–1000 m/z range and a scan time of 0.2 sec.

5) Compound formula assignment

The molecular formula assignment made for the selected small molecule biomarkers was conducted through a combination of liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS) fragmentation using a quadrupole-TOF MS (micrOTOF-Q II, Bruker, USA). For the QTOF analysis chromatographic conditions were identical to those reported for the profiling experiment. The average m/z and retention time of each of the biomarkers obtained through XCMS analysis (an innovative platform with an intuitive graphical interface which allows users to easily upload and process LC/MS data for untargeted metabolomics profiling, https://xcmsonline.scripps.edu) were used for targeted MS/MS analysis with a starting collision-induced dissociation energy of 10eV; 20eV; 40eV. Fragmentation patterns were analyzed with the Bruker Compass DataAnalysis 4.0 using the targeted MS/MS and formula generation algorithms and compared with the MS/MS fragment data in the METLIN (METabolite LINk) database (http://metlin.scripps.edu) [33].

5) Data processing and data analysis: Raw data files (.raw format) acquired from the HPLC–MS platform were converted to the Network Common Data Form (NetCDF) format using the File Converter program in the Data analysis software (Bruker). The data were processed with XCMS software running on R version 2.6.0, an open-source deconvolution program available for LC–MS data. Linear mixed models were applied to the twin-pair data using the R statistical language v 2.60, implemented in R package NLME. Intra-pair correlations were treated as random effects in the model. The overall group difference was tested using the F-statistic. The adjusted P-value for each comparison was computed using multiple testing procedures under default combinations (R package multicomp).

Results

Five hundred and thirty-seven molecule fractions, determined by isoelectric point (PI) and molecular weight (MW), have been detected and computerized at the p < 0.05 value (Figure 1). Among these fractions, 84 were identified to be shared by all PIH1, PIH2, and PIH3 groups; 10 fractions were common in PIH1 and PIH2 groups; 123 fractions were common in PIH2 and PIH3 groups, and 6 fractions were in PIH1 and PIH3 groups. There were 150 fractions only in PIH1 group, 50 fractions in PIH2 group, and 114 fractions only in PIH3. The level of 113 peaks decreased and 37 peaks increased when the control group was compared with that of PIH1. The level of 259 peaks was lower and that of 8 peaks was higher in PIH2 group, and the level of 311 peaks decreased and 16 peaks increased in PIH3 group.

To search for the differentially metabolized molecules in the PIH, we analyzed the two groups of PIH vs. control group. As Figure 2 shows, 10 fractions, with a statistical significance (p < 0.05), were selected automatically with preset parameters. However, value of the small molecules detected by HPLC/MS is not significantly different in PIH and control groups (Figure 2a), which did not meet our criteria for a purpose of identifying a significant bio-signature for the PIH. Similar results were noticed when patients who had proteinuria were compared to controls (Figure 2b).

To investigate if the severity of blood pressure (BP) may have a strong association with PIH, the PIH was divided into three subgroups on a basis of severity of BP and the cut-off was raised to p < 0.01. The PIH1 represents a mild group with BP 140-159/95-100 (Figure 4C), PIH2 represents the moderate group with BP 160-179/105-125 (Figure 4B), and PIH3 represents the severe group with a BP of 180/100 (Figure 4A). As shown in Figure 3, once the p value was raised to p < 0.01 and the PIH was divided into three subgroups, non-peptide fractions were able to be separated among PIH groups and control group without common fractions. Twelve fractions were further subjected to the analysis to identify their molecular formula and to characterize the molecular structure. Six small molecular metabolites that had been found as increased in PIH were characterized. These include C-6 NBD Ceramide, N-pentadecanoyl-L-Homoserine lactone, Granaticin and Leukotriene D4 in PIH1 and Bromotetralone in PIH2 (Table 2). Eight metabolites that had been identified as decreased in PIH were characterized, one from PIH2 and three from PIH2.

Discussion

Metabolomics, the scientific study of chemical processes involving metabolites, represents the collection of all metabolites in a biological cell, tissue, organ or organism, which are the end products of cellular processes [34, 35]. Biomarkers are biological characteristics that are objectively measured and evaluated as indicators of normal biological processes, pathological processes or pharmacologic responses to a therapeutic intervention. Biomarkers are widely used in clinical practice for the diagnosis, assessment of severity and response to therapy in a number of clinical diseases. In human studies, metabolomics has been applied to define biomarkers related to prognosis or diagnosis of a disease or drug toxicity/efficacy and to provide greater pathophysiological understanding of disease or therapeutic toxicity/efficacy [36]. For PIH, metabolomic technology is a potentially useful screening tool, especially in diagnosing the preeclampsia. For example, through high-resolution gas chromatography time-of-flight mass spectrometry (GC-tof-MS) on 87 plasma samples from women with preeclampsia and 87 matched controls, Kenny et al. developed a metabolomic pattern that could differ preeclampsia from normal pregnant controls using just 3 of the metabolite peak variables, with a sensitivity of 100% and a specificity of 98% [37]. Then, using an ultra-high performance liquid chromatography - mass spectrometry, they compared 60 plasma samples from women with late pregnancy preeclampsia and 60 matched controls; they found 14 kinds of metabolites of the high occurrence rat and finally confirmed the metabolites related to preeclampsia [38]. Similarly, Luo et al. distinguished 51 women with severe preeclampsia from 45 women with normal pregnancy with the high performance liquid chromatography tandem 3200 Q trap mass spectrometry, and found 16 biomarkers, which could better represent the metabolic characteristic of severe preeclampsia [39]. These findings all justified a prospective assessment of metabolomic technology as a screening tool for preeclampsia, and in future may lead to an improved method for monitoring preeclampsia using several biochemical markers.

Ceramide is an important bioactive small molecule compound. Ceramide can regulate Akt/eNOS signaling pathway [40-41]. In Akt/eNOS signaling pathway, Ceramide regulates the function of endothelial cells. Endothelial cells perform important functions in immunity, their dysfunction caused by abnormal cells may result in the occurrence of preterm labor. In addition, ceramide can regulate lysosome mediated cell death and autophagy [42-43]. Programmed cell death and cell autophagy play an important role in the occurrence of preterm labor. Animal experiments show that ceramide can affect the contractility of pregnant rat uterus [44]. Phosphocholine proved relevant to angiotensin [45]. Angiotensin plays an important role in the occurrence of PIH. Studies have found that phosphocholine transferase, a placental-specific enzyme post-translationally modifying neurokinin B, is essential for the pathogenic role of C-reactive protein in preeclampsia through activation of the neurokinin 3 receptor [46].

In summary, we have applied metabolomics approach to study non-peptide small molecules that could be distinguished in PIH from control groups. This is the first report on the non-peptide molecules that are differentially metabolized in PIH. Further investigation of compounds characteristic of PIH and possibility of using metabolomic profile in PIH screening has proved to be prospective. It may improve the diagnosis of PIH and preeclampsia using non-peptide biochemical markers.