Introduction
Preeclampsia (PE), a pregnancy-specific disorder, is one of the most common obstetric disorders [1]. The global morbidity of PE is approximately 3-8%, and it is the leading cause of pregnancy-related maternal and neonatal morbidity and mortality worldwide [2]. PE is typically characterized by new-onset hypertension (≥140/90 mmHg) after 20 weeks of gestation with proteinuria (≥0.3 g protein in 24 h). PE not only affects multiple organs (such as the lungs, liver, kidneys, and heart) and the neurological system of pregnant women, but also causes fetal intrauterine growth restriction, premature birth, death, and other effects [3]. Women with PE are at a greater risk of developing cardiovascular complications later in life. Although the clinical symptoms of patients with PE present after 20 weeks of gestation, the molecular events leading to its onset have been suggested to occur early in pregnancy. As many studies on the mechanism of PE are focused on polymorphisms or chronic hepatitis B infection, RNAs (circRNAs) are rapidly gaining prominence recently [4-7].
It has been shown that circular RNAs (circRNAs) are differentially expressed in the third trimester placenta of patients with PE compared with normal pregnant women and may be involved in PE [8]. CircRNAs are a newly described type of endogenous non-coding RNA that are gaining increasing attention in the field of regulatory RNA species. Compared with traditional linear RNA species (containing 5′ and 3′ ends), circRNA molecules have a closed loop structure, which is not easily degraded by exonuclease RNase R; hence, they are more stable than linear RNA [9-11]. It has been shown that circRNAs can function as microRNA (miRNA) sponges and also regulate the alternative splicing and expression of their parental genes [12-15]. For example, the circRNA ciRS-7/CDR1as (circRNA sponge for miR-7 or CDR1antisense) can reduce the expression of miR-7 target genes if CDR1as is silenced or miR-671 is overexpressed [12, 16]. Furthermore, Cocquerelle et al. demonstrated that circRNAs can have certain tissue and disease specificity [17]. Despite this increase in interest, the mechanism by which circRNAs underlie the pathogenesis of PE, more specifically, prior to the onset of symptoms, has been reported rarely. Moreover, a reliable biomarker for the early diagnosis of PE remains elusive.
Pregnancy-associated plasma protein A (PAPP-A), a macromolecular glycoprotein, is mainly synthesized by placental trophoblast cells and is then secreted into the blood. It is generally used as an index of the first-trimester screening test for Down's syndrome in pregnant women [18]. The amount of PAPP-A can reflect the level of hypoxia and ischemia of the placenta. PAPP-A is closely related to pregnancy because it plays an important role in trophoblast invasion by modulating the activity of insulin-like growth factor (IGF) through the cleavage of binding proteins [19-20]. Previous studies have shown that PAPP-A alters placental trophoblast differentiation and invasion by mediating IGF in the first trimester of pregnancy, which is associated with PE, miscarriage, preterm delivery, fetal growth restriction, and fetal death [21-24].
In this study, we assessed circRNA expression profiles in blood cells of patients with PE before the 20th week of gestation and evaluated the efficacy of circRNAs and circRNA-mediated PAPP-A for diagnosing PE before the onset of clinical symptoms, gaining a new insight into the role of circRNAs in the pathogenesis of PE.
Materials and Methods
Subjects
All of the women who underwent a prenatal examination and subsequently delivered at Guangzhou Women and Children’s Medical Center (GWCMC), China, between April 2012 and July 2015, were enlisted in the study. The ethics committee of GWCMC approved all aspects of this study. Written informed consent was obtained from all subjects.
PE is characterized by new-onset hypertension and proteinuria after the 20th week of gestation and is resolved by 6 weeks postpartum [25-26]. Blood pressure was documented in all subjects on 2 separate occasions that were at least 6 h apart. In the control group, the participants exhibited normal blood pressure without an excess of proteinuria, pregnancy complications, or other fetal malformations. In this study, we excluded women who had multiple pregnancies, gynecological disease, gestational diabetes mellitus, pre-gestational type 1 and type 2 diabetes, chronic hypertension, and cardiovascular, liver, or kidney diseases.
In the screening phase, 478 pregnant women were recruited, in which 5 developed PE in the later phase of gestation, who satisfied all the inclusion and exclusion criteria for the PE group, and the remaining participants were excluded. In the validation phase, 1889 pregnant women were recruited, and 30 of them who developed PE satisfied all the inclusion and exclusion criteria. The remaining subjects were excluded. In both sets, the pregnant women who developed PE were matched at a 1: 1 ratio for age, gestational week, and sample storage date with controls who had pregnancy without complications. The details of the study design are given in Fig. 1.
Specimen collection and preservation
Maternal whole blood samples were obtained from a prospective cohort of unselected women before the 20th week of gestation between April 2012 and July 2015 at GWCMC. Fresh peripheral venous blood samples (3 mL) were collected and then centrifuged (3000 rpm for 10 min at room temperature). Subsequently, the plasma and blood cells were separated immediately and stored at -80°C until use.
Human circRNA microarray analysis
The 5396 labeled circRNAs were hybridized on an Arraystar Human circRNA Array (catalog # AS-CR-001, 8 × 15 K; Arraystar, Rockville, MD, USA). Microarray procedures and data analyses were performed at Kangchen Corporation (Shanghai, China). Briefly, the experiment consisted of the following 5 steps. (1) Total RNA extraction: total RNA from peripheral blood mononuclear cells was extracted using the TRIzol LS Reagent (Invitrogen, Karlsruhe, Germany), following the manufacturer’s instructions; (2) total RNA quality: total RNA quality of the samples was measured using NanoDrop ND-1000 RNA quantity and quality assessment, and RNA integrity was assessed through standard denaturing gel electrophoresis; (3) RNA labeling: labeling was conducted using an Arraystar Super RNA Labeling Kit according to the manufacturer’s instructions, and the total RNA of each sample was amplified using a random primer and reverse transcribed to fluorescence-labeled cRNA; (4) chip hybridization: fluorescence-labeled circRNAs were measured using an Arraystar Human circRNA Array through hybridization and subsequent incubation in an Agilent hybridization oven at 65°C for 17 h; and (5) chip scanning: after washing the chips, they were scanned using an Axon GenePix 4000B Scanner and imported into GenePix Pro 6.0 software (Axon, Foster City, CA, USA).
The CircRNA chip data acquisition and analysis process was as follows. First, the original data extraction was conducted by adding the chip scanning image to GenePix Pro 6 software and reading the original data. Next, circRNA expression profiling was performed using the R software package for data normalization and subsequent data processing. Third, the differential expression of circRNAs between both groups was evaluated by assessing the change in P-value and fold change (FC).
Quantitative reverse transcription PCR
Quantitative reverse transcription PCR (qRT-PCR) was performed with a TIANScript RT Kit (#KR104; Tiangen, Beijing, China) and Talent SYBR Green qPCR Premix (#FP209; Tiangen, Beijing, China) following the manufacturer’s instructions. Divergent primers, instead of the more commonly used convergent primers, were designed. On the basis of the largest FC (> 40) and the P-value (< 0.001), we selected 9 circRNAs as validation genes. The primers for the validation of circRNAs and β-actin were synthesized by the Longsee Biomedical Company (Guangzhou, China). The sequences of β-actin and paired primers are shown in Table 1.
The reaction conditions for the 2 circRNAs were as follows. For hsa_circ_0001855 (FC = 45.77): initial denaturation at 95°C for 3 min; followed by 39 cycles at 95°C for 5 s, pre-selected annealing temperature for 30 s, and 72°C for 20 s; and final extension at 72°C for 10 min. For hsa_circ_0004904 (FC = 47.33): initial denaturation at 95°C for 3 min; followed by 39 cycles at 95°C for 15 s, pre-selected annealing temperature for 30 s, and 72°C for 25 s; and final extension at 72°C for 10 min. The data were analyzed by the ΔΔCT method. The experiments were performed independently three times.
Enzyme-linked immunosorbent assay
PAPP-A reportedly plays an important role in trophoblast differentiation and invasion during early pregnancy. Moreover, bioinformatics analysis predicted that circRNA_0001855 and circRNA_0004904, as well as their miRNA sponges, including miR-29a-5p, miR-765, miR-134-3p, miR-623, miR-138-5p, miR-30c-1-3p, and miR-30c-2-3p, directly target PAPP-A. Hence, its expression in plasma was assessed using enzyme-linked immunosorbent assay (ELISA) (catalog no. CS-human PAPP-A; Chemical Biology Test Systems, Shanghai, China) according to the manufacturer’s instructions. The lowest detectable PAPP-A concentration was 6.25 pg/mL.
Statistical analysis
The data were analyzed by SPSS version 17.0 for Windows (SPSS Inc., Chicago, IL, USA) and GraphPad Prism (version 6.0) software. Continuous variables with a normal distribution are presented as mean (standard deviation), and nonnormal variables are reported as median (inter quartile range). Means of two continuous normally distributed variables were compared by paired samples using Student’s t-test. The MannWhitney U test and KruskalWallis test were, respectively, used to compare the means of two variables not normally distributed. The frequencies of categorical variables were compared using Pearson’s χ2 or Fisher’s exact test, whenever appropriate. A value of P < 0.05 was considered statistically significant.
The miRNA sponges of the circRNAs were predicted in the assay by software from the Arraystar Company (based on miRanda & TargetScan).
The discriminatory performance of the circRNAs associated with PE was assessed using receiver operating characteristic analysis.
Results
Characteristics of the study participants
The demographic characteristics of the study participants are presented in Table 2. No differences in height, primigravida, education > 12 years, and history of spontaneous abortion (≥1) were noted between both groups. However, weight and body mass index before pregnancy, systolic/diastolic blood pressure before the 20th week of gestation, anemia in pregnancy, gestational age at delivery, birth weight of infants, and infant delivery at < 37 weeks were significantly different between both groups (P < 0.05).
Table 2.
Demographic characteristics of the study participants with PE and the controls with normal pregnancy, and characteristics of their infants. Note: *Women who developed PE and women who had a normal pregnancy were matched by age and gestational week. *P < 0.05 vs. Control
CircRNA microarray screening
To assess the differential expression levels of circRNAs in peripheral blood cells from pregnant women with PE before the 20th week of gestation, a human circRNA microarray was utilized. Volcano plots (Fig. 2A) and heat maps (Fig. 2B) at different P-values and FC revealed that 2178 circRNAs were significantly differentially expressed (FC > 2.0; P < 0.05), including 1294 upregulated circRNAs and 884 downregulated circRNAs, when compared between both groups. Only the circRNAs with FC > 40 and P < 0.001 are shown in Table 3. In the hierarchical clustering analysis shown in Fig. 2B, red or green represented upregulated or downregulated circRNAs, respectively.
Table 3.
Circular RNAs that were significantly differentially expressed in maternal blood cells (FC > 40, P < 0.001)
Fig. 2.
Assessment of circRNA expression profiles in the screening phase. (A) Volcano plots (FC > 2, P< 0.01). The red blocks indicate differentially expressed circRNAs; gray blocks indicate circRNAs with no difference in their expression. (B) Hierarchical cluster analysis: circRNA expression profiles in PE of the 10 samples in red (high relative expression) or green (low relative expression) scale.
Validation of circRNAs by qRTPCR
In the validation phase, we selected pregnant women from another cohort and confirmed the expression levels of 18 circRNAs (Fig. 3) in peripheral blood cells using qRT-PCR, based on FC > 40 and P < 0.001. As shown in Fig. 4, for the 18 circRNAs, the expression levels of hsa_ circ_0001855 and hsa_circ_0004904 were markedly increased in women with PE compared with matched healthy women in the validation set (P < 0.05).
Fig. 3.
Histogram for 18 circRNAs in blood cells in the validation phase. The levels of hsa_circ_0001855 and hsa_circ_0004904 were significantly higher in the PE group than in the control group (**P< 0.05). Lower ΔCT values signify higher levels of expression.
Fig. 4.
MRE prediction of circRNAs indicates that circRNA_0001855 and circRNA_0004904 regulate the RNA transcripts of PAPP-A by competing for shared microRNAs.
miRNA recognition element prediction for circRNA_0001855 and circRNA_0004904
On the basis on the interaction between cir cRNA_0001855, circRNA_0004904, and their miRNA sponges (miR-29a-5p, miR-765, miR-134-3p, miR-181d-3p, miR-623, miR-138-5p, miR-30c-1-3p, miR-656-5p, and miR-30c-2-3p), the probable binding sites between the sponge miRNAs and their target genes were predicted using bioinformatics tools (TargetScan and miRanda). The results showed that PAPP-A correlated with most of the miRNA sponges among the lists of both prediction programs (Fig. 4). PAPP-A was thus accepted as a pivotal element in the early differentiation of placental trophoblast cells for elucidating the pathogenesis of PE.
Validation of PAPP-A levels by ELISA
The results implied that pregnant women who later developed PE had significantly increased PAPP-A levels before the 20th week of gestation compared with the corresponding healthy pregnant women (65.78 versus 41.38 pg/mL, respectively, P < 0.05, Fig. 5C). The above results suggest that circRNA_0001855 and circRNA_0004904 may negatively regulate miRNA sponges and upregulate PAPP-A expression. PAPP-A, the target gene of the cir-cRNA_0001855 and circRNA_0004904 miRNA sponges, promotes trophoblast differentiation toward the invasive pathway in pregnancy.
Fig. 5.
Expression levels of hsa_ circ_0001855 and hsa_circ_0004904 in blood cells and those of PAPP-A in serum in the validation phase. (A) Expression of hsa_circ_0001855 (FC = 5.6) was significantly higher in patients with PE than in the controls (*P< 0.01) as determined by qRT-PCR. (B) Expression of hsa_circ_0004904 (FC = 7.7) was significantly higher in patients with PE than in controls (*P< 0.01) as determined by qRT-PCR. (C) Women with PE showed significantly higher serum PAPP-A levels than controls (P = 0.001).
Predictive performance of the circRNAs and PAPP-A
The area under the curve (AUC) of hsa_circ_0001855 was 0.621 (95% confidence interval [CI] 0.478–0.764); its sensitivity and specificity were 53.33 and 70.00%, respectively; and the cut-off value was 7.13 (cycle threshold [CT] value). The predictive performance of hsa_circ_0004904 was similar to that of hsa_circ_0001855; however, using plasma protein PAPP-A as a single predictor yielded 76.67% sensitivity, 60.00% specificity, 39.67 pg/mL cut-off value, and 0.728 AUC (95% CI 0.601–0.855) in PE. When PAPP-A was utilized in combination with hsa_ circ_0001855 and hsa_circ_0004904, the AUC was markedly improved to 0.940 (95% CI 0.869–1.000), with 86.67% sensitivity and 96.67% specificity (Fig. 6).
Fig. 6.
Receiver operating characteristic curves showed that serum protein PAPP-A levels combined with circRNAs improved the possibility of early PE detection in pregnant women. CircRNA_0001855: AUC 0.621 (95% CI 0.478–0.764); circRNA_0004904: AUC 0.611 (95% CI 0.466–0.756); PAPP-A alone: AUC 0.728 (95% CI 0.601–0.855); and combination model (PAPP-A + circRNAs): AUC 0.940 (95% CI 0.869–1.000).
Discussion
The present study documented the circRNA expression profiles of blood cells before the 20th week of gestation in women with PE for the first time. Although circRNAs have frequently been reported to be formed due to incorrect splicing during transcription, substantial evidence has suggested that they adsorb miRNAs during transcription, blocking their inhibitory effect on target mRNAs, thereby regulating mRNA expression. Moreover, several studies have indicated that circRNAs can act as molecular sponges that bind miRNAs, which have been implicated in various biological functions.
It was not until 1993 that scientists discovered numerous circRNAs [17] in human cells, but nowadays, the widespread and substantial presence of circRNAs has been demonstrated in eukaryotic organisms [27]. In recent years, with the rapid development and wide application of RNA sequencing [9], researchers have successfully detected more than 25000 types of circRNA in human fibroblast cells and proposed that circRNAs from exons and introns could be highly stable, and formed as conserved products after RNA splicing.
CircRNAs are gaining increasing attention in the field of RNA due to the following reasons: (1) most circRNAs are the products of exons, and some are directly formed by introns; (2) circRNAs are extensively found in mammalian cells, and sometimes, there are over 10 times more circRNAs than linear RNAs [28]; (3) circRNAs produced by a special variable shear have a certain organization, sequence, and disease specificity, and they are found in the cytoplasm of eukaryotic cells, but the circRNAs contained in introns are evident in nucleic acids [29]; (4) most circRNAs are conserved [9]; (5) unlike traditional linear molecules that include a 3′ head and 5′ tail, circRNAs have a closed ring structure and cannot be degraded easily by exonuclease RNase R, and hence they are more stable than linear RNA species; and (6) circRNAs have a high amount of MREs and can act as endogenous competitive RNAs and serve as natural microRNA “sponges,” and thus they can induce the expression of miRNA target genes [12, 16].
Although there has been minimal direct evidence of a correlation between circRNAs and PE, studies on miRNAs in PE have been widely reported. Their findings suggested that miRNAs may have lost control during the process of developing the placenta, fetus, and environment of spiral arteries in the uterus [30, 31]. Several research groups that analyzed miRNA expression in placental tissue found that the expression of gene transcription groups was significantly different in patients with PE [32, 33]. In addition, with the use of qRT-PCR or high-throughput technology, researchers have found that the expression of miRNAs (miR-29a-5P, miR-103, miR-130b, miR-181a, miR-342-3p, and miR-574-5p) was associated with a high risk of PE, and previous studies have reported that miR-29a-5p is intimately related to PE [33, 34].
On the basis of bioinformatics analysis, the miRNA recognition elements (MREs) of hsa_circ_0001855 and hsa_circ_0004904 were hsa-miR-138-5p, hsa-miR-30c-1-3p, hsa-miR-623, hsa-miR-30c-2-3p, hsa-miR-134-3p, hsa-miR-29a-5p, and hsa-miR-765, which all target PAPP-A. Our results showed that the expression levels of hsa_circ_0001855 and hsa_ circ_0004904 were increased significantly in patients with PE compared with the control group. More importantly, the expression levels of PAPP-A were higher in patients with PE compared to healthy pregnant women. These results implied that hsa_circ_0001855 and hsa_circ_0004904 may act as sponges of their MREs and then block the inhibition of MREs on their target mRNA of PAPP-A, thereby mediating the regulation of plasma PAPP-A protein expression in the pathogenesis of PE. PAPP-A combined with hsa_circ_0001855 and hsa_ circ_0004904 can be used as a potential early biomarker for predicting PE with an AUC of 0.940 (95% CI 0.869–1.000), sensitivity of 86.67%, and specificity of 96.67%.
The biological effects of PAPP-A play an important role in targeting the IGF axis. The IGF axis is composed of insulin-like growth factors, including the IGF binding proteins (IGFBPs) and IGFBP hydrolases [35, 36]. Protein hydrolysis of IGFBP splits and releases IGF. The effects of IGF-I are mediated through the IGF-I receptor by complex interactions with multiple IGFBPs. PAPP-A mediates the increase in bioactive IGF-I. IGF-I likely contributes to the proliferation and migration of endovascular extravillous trophoblast cells [37].
Overall, we screened differentially expressed circRNAs between normal and PE samples in this study and selected 18 circRNAs as representatives for validation. We also validated PAPP-A, a protein that may be closely related to hsa_circ_0001855 and hsa_circ_0004904. PAPP-A combined with hsa_circ_0001855 and hsa_circ_0004904 could be used as a potential early biomarker for predicting PE.
This study has some limitations. First, the sample size was small because our study included pregnant women before the 20th week of gestation and because of the difficulty in obtaining specimens. What’s more? The samples were not matched for BMI because the PE samples before 20 weeks’ gestation were not easy to acquire when already matched with age and gestation. Previous literature showed no correlation between PAPP-A and BMI [38]. When a multiple regression analysis was performed including gender, age, BMI, waist-hip ratio, homoeostasis model assessment of insulin resistance (HOMA-IR), adiponectin and leptin as confounders, PAPP-A was independently correlated with adiponectin and leptin in cardiovascular disease. Moreover, we needed to verify the miRNAs that were related to circRNAs and PAPP-A. This verification would more clearly explain the pathogenesis of PE and would help us find suitable early prediction biomarkers. In the future, we will carry out these experiments in multiple centers with a larger number of subjects.
Acknowledgements
We thank all the pregnant women who participated in our study and the research nurses at Guangzhou Women and Children’s Medical Center.
This study was supported by the Natural Science Fund of Guangdong Province (2014A030313004), the Science and Technology Fund of Guangzhou (201300000135 and 201707010019), and the Science and Technology Fund of Guangdong Province (2016A020218004). The sponsors had no role in the design and conduct of the study, the collection, analysis, and interpretation of the data, in the preparation, review, or approval of the manuscript, or the decision to submit the manuscript for publication.
Disclosure Statement
The authors declare that they have no conflict of interests.


Get Permission
