This study was permitted by the Human Research Ethics Committee of Nanjing Medical University as well as the Nanjing Maternal and Child Health Hospital and the Obstetrics and Gynecology Hospital Affiliated to Nanjing Medical University.

Samples of amniotic fluid were obtained from women with normal and PE pregnancies at 34–38 weeks of gestation at Nanjing Maternal and Child Health Hospital, Obstetrics and Gynecology Hospital Affiliated to Nanjing Medical University, China, during January 14 and March 27. The pregnant participants were fully informed regarding the aims and scope of the study and freely signed informed consent forms. Preeclampsia patients were diagnosed as first appearance of systolic blood pressure ≥140 mmHg and (or) diastolic blood pressure ≥90 mmHg, companied with urine protein quantification ≥300 mg per 24 h after 20 weeks of gestation. Samples in our study were severe cases that blood pressure more than 160/110 mmHg, proteinuria ≥5,000 mg/24 h, headache or visual impairment, abnormal liver function with or without right upper quadrant pain, and platelet less than 100×10⁹/l. Women suffered from chronic hypertension, acute or chronic heart disease, nephropathy, diabetes mellitus or other maternal autoimmune diseases before pregnant were not included in our research. All amniotic fluids were collected during cesarean sections and placed on ice while en route to the laboratory. The samples of amniotic fluid were centrifuged at 2,000 g for 20 min at 4°C to remove cell debris. The supernatant was then mixed with protease inhibitors (Complete mini EDTA-free; Roche, Mannheim, Germany) and stored at −80°C until used.

The amniotic fluid samples were centrifuged (12,000 g, 4°C, 20 min), and the supernatant was transferred into a new centrifuge tube and mixed with acetonitrile (20% v/v). The mixture was then vortexed and incubated at room temperature for approximately 25 min (19). Molecular weight cut-off (MWCO) filters (10,000 Da; Millipore, Billerica, MA, USA) were washed with H₂O (0.5 ml) before they were used, and the processed samples were centrifuged through the filters according to the manufacturer's recommendations. The concentrations of the peptides in each of the samples were determined using the bicinchonic acid method (BCA) (Pierce Biotechnology, Inc., Rockford, IL, USA). The filtrates were then concentrated and lyophilized.

LC-MS/MS is commonly used as a method for identifying peptides or low molecular weight proteins according to their detection sensitivity and throughput. We labeled the peptides derived from each sample with isotopomeric dimethyl labels, which was based on the inter-reaction of primary amines with deuterated and ¹³C-labeled formaldehyde to generate a Schiff base rapidly reduced by the addition of cyanoborohydride (20,21). The labeled samples were simultaneously detected using LC-MS/MS based on their abundance, which was reflected by mass differences in the dimethyl labels.

The lyophilized samples were dissolved in 0.1% formic acid, filtered through a 0.45-µm membrane and then set aside. We chose an LC Packings C18 trap column (Acclaim PepMap100, 75 µm × 20 mm) to perform reverse-phase chromatography and separated the peptides using an Ultimate 3000 nano-LC system (Dionex) with a linear gradient in liquid phase containing (A) 0.1% formic acid dissolved in water and (B) 0.1% formic acid dissolved in acetonitrile (22,23). After the samples were transferred into the MALDI TOF/TOF (Ultraflextreme; Bruker Daltonics, Bremen, Germany) instrument under the positive ion mode, the mass spectral analysis strategy went as follows: full scan analysis was operated over the m/z range 600–5,000 at 2 spectra/s, with the capillary voltage and cone voltage were maintained at 3.9 kV and 40 V.

For MS analysis, each position (10 random positions on each sample) was irradiated with 200 laser pulses, and 2000 single-shot spectra were collected. Mass Lynx™ software (v.4.1, Waters Corp., Milford, MA, USA) was used to analyze the protein databases using the MS/MS spectral data for the protein ID. The human taxonomy was searched for protein IDs in the NCBI and Matrix Science databases (http://www.matrixscience.com/, ©2016). The MS/MS data were searched using the Mascot database (http://www.matrixscience.com, ©2016).

Each peptide isoelectric point (pI) was calculated using an online pI/Mw tool (http://web.expasy.org/compute_pi/, ©2017). A GO analysis was then performed to investigate whether the identified peptide precursors favored any compartments or protein classes. The peptides were classified according to both their cellular components and their molecular functions using annotations obtained from the UniProt Database (http://www.uniprot.org/, ©2002–2017, last modified January 10, 2017).

Data from quantitative experiments were analyzed using unpaired Student's t-tests where appropriate in Statistical Program for Social Sciences (SPSS) v.22 and GraphPad Prism 5.0. The statistical significance of the results was determined using multiple comparisons followed by Student's t-tests. Significance was set at P<0.05. All data are shown as the means ± standard deviations (SDs). The heat map was generated by software HemI Version 1.0.0.

The basic information relating to the subjects involved in our study are summarized in Table I. There was no difference in the ages, number of pregnancies, BMI and gestational week between the groups (P>0.05). Every subject delivered by caesarean section at the appropriate time, and the newborn babies were all healthy. No cases of asphyxia neonatorum were observed. There were differences between the two groups in systolic pressure, diastolic pressure, proteinuria level and neonatal weight (P<0.05).

After LC-MS/MS, a total of 352 peptides were collectively detected in the human amniotic fluid samples (Fig. 1). The peptidome contained peptides spanning a wide range of molecular weight (Mw) and pI values. In addition, the analysis revealed that most of the peptides were between 2800–3200 Da in MW and between 4–6 and 8–10 in pI (Fig. 1A and B). The categories of detected peptides clustered into two general groups according to the range of acidic and basic pH. All of the Mw of the peptides analyzed using LC-MS/MS were under 3500 Da, in agreement with previous reports.

Among the 352 detected peptides, 23 were found to be differentially expressed (6 were downregulated, and 17 were upregulated) (P≤0.05) between PE and normal pregnancies. Table II lists these 23 peptides and their respective precursor proteins. Further analyses revealed that most of the upregulated peptides were distributed between 2900–3100 Da in Mw and that they were predominantly acidic in pI, whereas the corresponding distributions of the downregulated peptides were more average (Fig. 1D and E). A heat map of the six individuals is shown to clarify these differences in expression (Fig. 2).

LC-MS/MS identified 352 peptides, and the bioinformatics analysis showed the specificity of the four cleavage sites (Fig. 1C). Leucine (L) was the most common amino acid at the C-terminal cut site of the preceding peptide, and lysine (K) was most commonly the C-terminal amino acid of the identified peptide. Glutamic acid (E) was most commonly observed at the N-terminal of the preceding peptide, and serine (S) and tryptophan (M) were most commonly observed at the N terminus of the identified peptides.

The distribution of the cut sites in the differentially expressed peptides were also variable in the PE amniotic fluid samples. To further investigate the peptidome in these samples, we next used bioinformatics analyses to determine the amino acids at the cleavage sites of both the N- and C-termini (Fig. 1F). The results of our investigation of these four cut sites generally reflected the findings reported in endogenous proteolytic enzymes in human amniotic fluids. The results revealed that glutamic acid (E), glycine (G) and leucine (L) were the most common amino acids at the C-terminal cut site of the preceding peptide, while phenylalanine (F) and serine (S) were most commonly observed at the C-terminal of the identified peptide. However, alanine (A), cysteine (C) and glutamine (G) dominated the cleavage site at the N terminus of the identified peptide, while arginine (R) was the most common N-terminal pre-cleavage amino acid.

To determine whether these peptide precursors were biased toward specific protein compartments or protein classes, we combined Pathway and GO analyses to investigate the probable roles of the endogenous peptidome and the precursor proteins of peptidome components. The 352 identified peptides were categorized according to their cellular components and their molecular functions using the UniProt Database (http://www.uniprot.org/, ©2002–2017, last modified January 10, 2017). The most likely cellular component and molecular function categories are shown in Fig. 3A and B, respectively. We found that the majority of the proteins identified in the present study were cytoplasmic (17%) or nuclear (20%) proteins and that the most common functions were the regulation of gene expression (20%) and enzymatic functions (13%). Because late pregnancy amniotic fluid consists of the metabolites of a variety of both maternal and neonatal organs, the results of this analysis also demonstrated the condition of the mothers and babies.

We also summarized the cellular components and molecular functions of the peptides found to be differentially expressed, as shown in Table II. These peptides were mainly located in the nucleus and have been shown to play roles in gene expression, consistent with the results of our analysis of the human amniotic fluid peptidome. The different categories of the listed protein precursors demonstrated no existence of the preferential extraction from specific cellular components or with functions and the successful of the previous ultrafiltration.