Serum proteins differentially expressed in early- and late-onset preeclampsia assessed using iTRAQ proteomics and bioinformatics analyses

Introduction

Preeclampsia (PE), a hypertensive disorder complicating pregnancy, is the main cause of increased perinatal mortality among mothers and infants worldwide. The rate of PE in the United States is approximately 3.4% (Ghulmiyyah & Sibai, 2012), and it is higher in developing countries (Ananth, Keyes & Wapner, 2013). PE is marked by new-onset hypertension occurring after 20 weeks of gestation, accompanied by either new-onset proteinuria or systemic multiple organ damage (Committee on Practice Bulletins—Obstetrics, 2019). Severe PE may lead to convulsions, coma, cerebral hemorrhage, heart failure, placental abruption, disseminated intravascular coagulation, and even death (Bibbins-Domingo et al., 2017). Fetal delivery is the most effective treatment for PE, with other treatments of relieving symptoms used only as an attempt to gain time for enabling further maturity of the fetus. As a result, PE is the leading cause of premature birth and low birth weight, especially severe PE (Ananth & Vintzileos, 2006).

The exact pathogenesis of PE remains unclear. However, PE is considered a placenta-derived disease because the syndrome resolves once the placenta is removed. There are two subtypes of PE: early- and late-onset preeclampsia. Their pathogeneses are not identical. Early-onset preeclampsia (EOPE) has a higher degree of placental damage, whereas late-onset preeclampsia (LOPE) may focus on the interaction between the normal senescence of the placenta and a maternal genetic susceptibility to cardiovascular and metabolic disease (Burton et al., 2019). However, pathophysiological changes of the placenta further lead to endothelial dysfunction and systemic inflammatory response, which are their common pathogenesis link (Young, Levine & Karumanchi, 2010).

The placenta is in direct contact with the maternal circulation. Therefore, any changes in the protein expression of placental tissue are reflected in maternal serum proteins (Norwitz, 2007). If the types and levels of differentially expressed proteins (DEPs) in the serum of patients with PE can be researched quantitatively and holistically, specific serum biomarkers may be found for predicting PE and for further study of its pathogenesis. Compared with other proteomics methods, isobaric tags for relative and absolute quantitation (iTRAQ) combined with liquid chromatography–tandem mass spectrometry (LC-MS/MS) is considered more effective for searching for serum or plasma biomarkers (Moulder et al., 2018). Therefore, in the present study, we used iTRAQ and LC-MS/MS methods to identify serum proteins differentially expressed between EOPE and LOPE and between women with PE and healthy pregnant women. We then used bioinformatics analyses to determine the key proteins and related signaling pathways associated with the development of EOPE and LOPE.

Materials & Methods

LC-MS/MS analysis

The labeled samples were redissolved in 40 µL of 0.1% formic acid aqueous solution. The peptides were loaded onto a C18-reversed phase column (3 µm C18 resin, 75 µm ×15 cm). The mobile phases consisted of 2% methyl cyanide/0.1% formic acid/98% water and 80% methyl cyanide/0.08% formic acid/20% water. The gradient for the B phase increased linearly at 0–68 min from 7% to 36% and at 68–75 min, from 36% to 100%. Each sample was separated by capillary high-performance liquid chromatography and was analyzed by an Orbitrap Fusion Lumos mass spectrometer (Thermo Science). After the data were collected, they were processed according to the flowchart shown in Fig. 1. Protein identification and quantification were done by using the ProteinPilot Software, version 4.2 (Sciex). Human proteome databases containing UniProt sequences were used to perform peptide identification. Serum proteins with fold changes in their levels compared with control groups of at least 1.2 (in two of two replicates) and that were also statistically significantly different between the groups at P < 0.05 were further analyzed.

Results

SDS-PAGE

The total proteins in the molecular weight range of 10–220 kDa from 16 samples were effectively separated by SDS-PAGE. The proteins were not degraded, and the high-abundance proteins were obtained (Fig. 2).

KEGG signaling pathway analysis

The KOBAS online analysis tool was used to identify the functions associated with the DEPs and the KEGG signaling pathways. Results of the KEGG pathway enrichment analysis showed that complement and coagulation cascades, proteoglycans in cancer, and metabolic pathways were the main signaling pathways associated with the EOPE group, whereas complement and coagulation cascades were the main signaling pathways associated with the LOPE group (Fig. 6). Thus, the results indicated that a functional change in coagulation was the main finding for both EOPE and LOPE.

Table 2:

List of differentially expressed proteins in the late-onset preeclampsia group.

Regulated type	Protein accession	GENE NAME	PROTEIN DESCRIPTION	FC
UP	Q9BXP8	PAPPA2	Pappalysin-2	5.27
UP	A0A0S2Z3F6	CETP	Cholesteryl ester transfer protein plasma isoform 1	3.51
UP	P08709	F7	Coagulation factor VII	2.8
UP	Q6NS95	IGL@	IGL@ protein	2.67
UP	P04275	VWF	von Willebrand factor	2.46
UP	C0JYY2	APOB	Apolipoprotein B	1.93
UP	A0A140VK24	-	Testicular secretory protein Li 24 protein 1	1.74
UP	P05546	SERPIND1	Heparin cofactor 2	1.52
UP	O95445	APOM	Apolipoprotein M	1.34
UP	P80108	GPLD1	Phosphatidylinositol-glycan-specific phospholipase D	1.26
DOWN	P05543	SERPINA7	Thyroxine-binding globulin	0.82
DOWN	Q9NZP8	C1RL	Complement C1r subcomponent-like protein	0.63
DOWN	A8K2T7	-	Receptor protein-tyrosine kinase	0.61
DOWN	Q00887	PSG9	Pregnancy-specific beta-1-glycoprotein 9	0.6
DOWN	P07333	CSF1R	Macrophage colony-stimulating factor 1 receptor	0.56
DOWN	Q2L9S7	AAT	Alpha-1-antitrypsin MBrescia variant receptor 1	0.53
DOWN	A0A2S0BDD1	ATIII-R2	Antithrombin III isoform	0.49
DOWN	Q8WW79	SELL	L-selectin	0.48
DOWN	P40189	IL6ST	Interleukin-6 receptor subunit beta (Fragment)	0.47
DOWN	B2R7Y0	-	cDNA, FLJ93654	0.41
DOWN	Q12860	CNTN1	member 2 (SERPINB2)	0.36
DOWN	Q8WWZ8	OIT3	Oncoprotein-induced transcript 3 protein	0.31
DOWN	A0A140TA29	C4B	Complement C4-B	0.21
DOWN	B2R950	-	cDNA, FLJ94213	0.18

DOI: 10.7717/peerj.9753/table-2

PSG9 protein level in serum

Serum level of PSG9 was measured by ELISA. The data showed a significant difference between LOPE group and control group (4.54 ± 1.22 vs. 6.32 ± 1.73, Fig. 7). The result is consistent with the study of mass spectrometry.

Discussion

Proteomics methods have provided some important information regarding PE. For example, Blumenstein et al. (2009) used a differential in gel electrophoresis–based approach to identify changes in the plasma proteome of pregnant women who subsequently developed PE. They found that those DEPs are mainly involved in lipoprotein metabolism, the blood coagulation system, and the complement system. The results of the present study were consistent with those of Blumenstein et al. (2009), but we also made some new discoveries (discussed below). These results suggest that the pathogenesis of PE is complex and is associated with multiple proteins and signaling pathways.

To maintain pregnancy and fetal growth progress within normal reference ranges, healthy pregnant women develop physiological hyperlipidemia. However, they do not experience angiopathy because high-density lipoprotein (a vascular protection factor) and low-density lipoprotein (LDL, an atherogenic factor) also increase to protect the vascular endothelium from injury (Jin et al., 2016; Wang et al., 2017). Compared with healthy pregnant women, pregnant women with PE display significantly increased serum triglycerides and LDLs, which may enhance oxidative stress and ultimately lead to vascular endothelial cell injury (Huda, Sattar & Freeman, 2009; Pohanka, 2013). Our study found that there were six DEPs associated with lipid regulation increased in the PE groups, including ApoB, LRP1, ApoL3, ApoC-I, CETP, and ApoM. ApoB, the main carrier protein of LDL, is an atherogenic risk factor. Previous evidence supports that ApoB is significantly upregulated in the plasma of patients with PE (Lin et al., 2019). CETP plays an important role in high-density lipoprotein metabolism and reverse cholesterol transport and is upregulated in the third trimester of pregnancy. Another study has reported that the TaqIB polymorphism of the CETP gene is significantly correlated with the triglyceride and total cholesterol levels in patients with severe PE (Belo et al., 2004). Here, we reported for the first time, to our knowledge, that LRP1, ApoC1, and ApoL3 are increased in the EOPE group. LRP1 and ApoC1 accelerate the development of atherosclerosis through different pathways (Mueller et al., 2018; Westerterp et al., 2007). Endothelial function decreases during PE, and endothelial dysfunction is a characteristic of atherosclerosis. ApoL3, as a regulator of MAPK and FAK signaling in endothelial cells, has been shown to be involved in angiogenesis in vitro (Khalil et al., 2018). Increased ApoL3 in PE may be a compensatory response to endothelial dysfunction. Therefore, the current evidence suggests that dyslipidemia may be related to the development of PE, but the specific pathogenesis remains to be explored in future studies.

Physiological hypercoagulability in healthy pregnant women can prevent intrapartum and postpartum hemorrhage. However, abnormalities in coagulation, the anti-coagulation system, and the fibrinolytic system lead to pathological hypercoagulability in PE (Dusse et al., 2011). In the LOPE group of the present study, coagulation factors VII, vWF, and SRPIND1 were upregulated group, whereas AAT and ATIII were downregulated compared with healthy controls. In the EOPE group, coagulation factor V, plasminogen, and ADAMTS13 were upregulated. These results are consistent with those of many previous trials (discussed below). The level of plasma FVII in patients with severe PE is significantly higher than that in healthy pregnant women; therefore plasma FVII levels may show high sensitivity and specificity in differentiating between PE and normal pregnancy (Dusse et al., 2016). The coagulation factor vWF is a specific marker that reflects damage of endothelial cells; thus, damaged microvascular endothelial cells in PE promote the expression of vWF. Owing to the activation of the intrinsic and exogenous coagulation pathways by damaged endothelial cells, a large number of coagulation factors are activated. This leads to a massive generation of thrombin, which antagonizes a large amount of anti-thrombin III, resulting in a significant decrease in its level (Demir & Dilek, 2010; Pottecher et al., 2009). In response to abnormalities in the coagulation mechanism of PE, evidence-based medical study has also shown that oral administration of low-dose aspirin during early pregnancy can significantly reduce the incidence rate of PE (Rolnik et al., 2017).

A growing number of studies have shown that abnormal expression with the complement system is associated with PE (Agostinis et al., 2010; Derzsy et al., 2010). Our results support this evidence, showing that expression levels of CFD, CIQC, CIRL, C9, C1S, and C1R were significantly increased, while expression levels of C4B, FCN2, and FCN3 were significantly decreased in the EOPE group compared with their control group. Moreover, the expression levels of complement C1RL and C4B were decreased in the LOPE group compared with their controls.

Downregulation of PSG1 and PSG9 expressed in the PE groups is an interesting finding in our study. PSG is a pregnancy-specific glycoprotein that is synthesized and secreted into the blood by placental syncytiotrophoblast cells. Rong et al. (2017) found that PSG9 significantly promotes the angiogenesis of human umbilical vein endothelial cells. Chang et al. (2016) found a significant increase in the number of cases with deletions in the PSG gene locus among patients with PE. We hypothesize that downregulation of the PSG protein family may be involved in the pathogenesis of PE by affecting the proliferation of endothelial cells although no specific experimental evidence currently supports or disproves this hypothesis.

PE is considered a placenta-derived disease. In the classic two-stage model, placental stress leads to dysfunction of maternal peripheral endothelial cells, systemic inflammatory response, and the clinical syndrome of PE (Staff, 2019). This model is also reflected in our mass spectrometry results. In both the EOPE and LOPE groups, abnormalities of coagulation cascade activation, lipid metabolism, and complement activation were found. However, the main stressor associated with EOPE is placental hypoperfusion secondary to impaired spiral artery remodeling; by contrast, in LOPE, the cause is more likely attributable to a mismatch between normal maternal perfusion and the metabolic demands of the placenta and fetus (Burton et al., 2019). Spiral artery remodeling is completed by the placental trophoblast cells continuously invading and destroying the uterine spiral arterial wall and gradually replacing the endothelial cells. Many studies have focused on the control of this invasion (Pollheimer et al., 2018; Wagner, Otomo & Christians, 2011; Wu et al., 2016).

Unexpectedly, we also found 19 DEPs in the EOPE group that were closely related to placental development or invasion of tumor cells. This result was not detected in the LOPE group, which showed an increase only in PAPPA2, which affects the invasion and metastasis of placental trophoblasts and leads to a gradual decline in placental function (Wagner, Otomo & Christians, 2011). LYVE1, CST3, and NCAM1 are also reportedly involved in placental vascular remodeling (Pawlak et al., 2019; Song et al., 2010; Zhang, Xu & Han, 2019). GRN, CD5L, ENO1, CHL1, CRISP3, VASN, and TNIK promote the invasive ability of tumor cells (Aran et al., 2018; Bhandari et al., 2019; Buhusi et al., 2003; Chon et al., 2016; Song et al., 2014; Voshtani et al., 2019; Wang et al., 2019). By contrast, SEMA4B, KRT1, KRT9, FBLN1, and PTGDS are involved in the anti-invasive activity of tumor cells (Blanckaert et al., 2015; Jian et al., 2015; Marano et al., 2018; Zhang et al., 2018). MMP2 plays an important role in trophoblast invasion and is generally thought to be downregulated in PE (Wu et al., 2016). We found the opposite result, that is, the expression of MMP2 was upregulated in EOPE; however, expression of PI16, which inhibits MMP2 activity (Hazell et al., 2016), was also upregulated. PEPD is a hydrolase that affects collagen biosynthesis, cell proliferation, and matrix remodeling. Pehlivan et al. (2017) have shown that PEPD activity in the plasma, umbilical cord, and placental tissue of women with pregnancy-induced hypertension is higher than that of healthy pregnant women. Our results also indicated an upregulation of PEPD in the EOPE group. The dysregulation of the expression of these proteins may be an important cause of placental dysfunction in EOPE.

Furthermore, previous studies have shown that soluble form of placenta-derived endoglin (sENG) was upregulated in the serum of EOPE, which is involved in endothelial dysfunction in coordination with soluble fms-like tyrosine kinase (Venkatesha et al., 2006). Our mass spectrometry results were similar to that previous finding showing that endoglin was upregulated in EOPE but not in LOPE group. This finding may also indicate more severe endothelial dysfunction in EOPE.

Conclusions

In summary, the use of iTRAQ combined with LC-MS/MS was effective for screening serum for DEPs in PE. We used bioinformatics to analyze the DEPs that showed significant changes in their expression levels to provide potential indicators for detecting PE and ideas for the study of pathogenesis in PE. In conclusion, the pathogenesis of EOPE and LOPE appeared to be related to dysfunctions in coagulation cascade activation, lipid metabolism, and complement activation. However, compared with that in LOPE, the pathogenesis of EOPE was associated with greater placental injury. This study provides several new proteins in the serum which may potentially have value in clinical diagnosis of EOPE and LOPE and offers potential mechanisms for the development of EOPE and LOPE, contributing to the future research on prediction, prevention, and pathogenesis of PE.

Supplemental Information