Downregulation of Notch Signaling Pathway in Late Preterm and Term Placentas from Pregnancies Complicated by Preeclampsia

Preeclampsia (PE) is a major cause of maternal mortality and morbidity, affecting 3–5% of all pregnancies. The Notch signaling pathway plays an important role during placental development, activating several target genes. Defects in the Notch pathway have adverse effect on placentation. The aim of this study was to investigate the expression of receptors NOTCH1,-2,-3,-4, ligands DLL1,-3,-4, JAG1,-2 and target genes HEY1,-2 in placental tissue samples from 20 late preterm or term pregnancies complicated by PE versus 20 normal pregnancies. mRNA levels of the studied molecules were measured by quantitative Real-Time PCR (qRT-PCR), while the protein expression of the intracellular domain of NOTCH2 (NICD2) and NOTCH3 (NICD3) was measured by Western Blot (WB). qRT-PCR analysis revealed that NOTCH1, NOTCH4 and DLL1 were not expressed in the placenta. On the contrary, NOTCH2, NOTCH3, DLL3, DLL4, JAG1, JAG2, HEY1 and HEY2 mRNA levels were downregulated in PE samples vs. controls (p<0.01). WB confirmed that NICD2 (p = 0.014) and NICD3 (p<0.001) protein levels were also lower in PE specimens. Statistical analysis revealed several significant associations: of NOTCH3 mRNA expression with smoking during pregnancy (p = 0.029), of NICD3 protein levels (p = 0.028) and DLL3 mRNA levels (p = 0.041) with birth weight centile, and of HEY2 transcript levels with parity (p = 0.034) and mode of delivery (p = 0.028). Our results suggest that Notch pathway downregulation is associated with PE. Further studies are required in order to determine the role of these molecules in PE pathogenesis and to evaluate their potential use for the early detection and treatment of PE.


Introduction
Preeclampsia (PE) affects 3-5% of all pregnancies worldwide and is a major cause of maternal mortality and morbidity [1]. PE is a consequence of diverse pathological processes involving and villous placenta, but not the chorionic plate, in order to avoid sample contamination from maternal tissues. Areas involving gross calcifications or infarcts were avoided. Contamination from fetal membranes was also minimized. Tissue biopsies were snap-frozen in liquid nitrogen and stored at -80°C. Three of the above six biopsy specimens were randomly selected for our measurements, while the other three were sent, as is indicative in PE pregnancies, for routine histological examination, which was consistent with the underlying pathology. The Ethics Committee of the University Hospital of Heraklion approved the study, and written informed consent was obtained from all participants.

Clinical Definitions and Sample Description
Preeclampsia was defined as hypertension in previously normotensive women after 20 weeks of gestation (systolic blood pressure of !140 mmHg or diastolic blood pressure of !90 mmHg) on at least two occasions, associated with proteinuria (!300 mg in a 24-hour urine collection or one dipstick measurement of !2+) [13]. Control group included women with uncomplicated, normotensive singleton pregnancies who delivered healthy, appropriate-for-gestational-age babies. Exclusion criteria were stillbirth, multiple gestations, chorioamnionitis, pre-pregnancy hypertension, renal disease, as well as chromosomal abnormalities and fetal anatomical defects.

RNA extraction and cDNA preparation
Tissues (~100mg) were homogenized with a Mortar and Pestle with the help of liquid nitrogen. RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA), followed by chloroform addition and centrifugation. Total RNA was precipitated from the supernatant with isopropanol, washed with 75% ethanol and resuspended in 30μl of diethylpyrocarbonate (DEPC)-treated water. RNA concentration and purity were calculated after measuring on a Nanodrop 1000 spectrophotometer (NanoDrop Products, Wilmington, DE) its 260 nm absorbance and 260/ 280 nm absorbance ratio, respectively. cDNA was synthesized using the PrimerScript RT reagent kit (Takara Bio, Otsu, Japan). In detail, 1μg of total RNA, 2μl of random hexamers and 1μl dNTPs were heated at 65°C for 5min, in order to remove RNA secondary structures, and placed on ice until the addition of cDNA synthesis mix, which contained 5× PrimerScript Buffer, 20 units RNase Inhibitor and 200 units PrimerScript RTase. The final mix (volume 20μl) was incubated for 10min at 30°C for primer extension, and cDNA synthesis was conducted at 50°C for 50min. The reaction was terminated by heating at 95°C for 5min. cDNA was stored at -20°C until use.
Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) assay NOTCH1, NOTCH2, NOTCH3, NOTCH4, DLL1, DLL3, DLL4, JAG1, JAG2, HEY1 and HEY2 mRNA expression was measured using a qRT-PCR assay with the SYBR Green I dye. The housekeeping gene, beta-Actin (ACTB), was used as an internal control, in order to normalize the studied genes' expression levels. The mRNA-specific primers, which were designed with the Lasergene software (DNASTAR, Madison, WI) and span at least one intron with an average length >800 bp, are listed in Table 1. Their specificity was verified with the BLAST program (http://www.ncbi.nlm.nih.gov/BLAST). cDNA (1μl) from PE or control samples was amplified in a PCR reaction containing 2× Maxima SYBR Green qPCR Master Mix (Fermentas, Vilnius, Lithuania) and 300nM of each primer in a final volume of 20μl. To ensure the accuracy of the quantification measurements, a representative pool of all the samples was diluted in a series of six 2× dilutions and was run on the same plate, in order to construct a standard curve for the quantification process. After initial denaturation at 95°C for 10min, samples were subjected to 40 cycles of amplification, comprised of denaturation at 95°C for 20s, annealing at 60°C for 30s and elongation at 72°C for 30s, followed by a melt curve analysis in which the temperature was increased from 60°C to 95°C at a linear rate of 0.2°C/sec. Data collection was performed during both annealing and extension, with two measurements at each step, and at all times during the melt curve analysis. PCR experiments were conducted on an Mx3000P real-time PCR thermal cycler using software version 4.01 (Stratagene, La Jolla, CA). After amplification, standard curves were constructed from the samples used in the series of consecutive dilutions. Subsequently, using these standard curves and the Ct value of the samples, we calculated the mRNA expression of the genes studied. Samples with no amplification plots or with dissociation curves that exhibited signs of primer-dimer formation or by-products were excluded. To normalize the mRNA expression of each gene, its value was divided by the β-Actin mRNA value. The normalized values of PE samples were divided by the average normalized values of normal samples. The result of this division provided the relative expression of PE specimen in relation to the control group. This mathematical process is summarized in the following formula: A 2-fold increased (a value !2) or decreased (a value 0.5) expression was considered biologically significant (overexpression or downregulation respectively). In each PCR reaction two negative controls were included, one with no cDNA template and one with no reverse transcription treatment. All qRT-PCR measurements were conducted in triplicates. PCR products were electrophorized on 2% (w/v) agarose gels and stained with ethidium bromide, in order to verify that the corresponding product band (as visualized with the help of a UV transilluminator) had the correct size. Representative bands of all PCR products were extracted from the agarose gels, purified and sequenced, as a final confirmation step that the appropriate gene was being amplified at each PCR reaction.

Statistical analysis
NOTCH1, NOTCH2, NOTCH3, NOTCH4, DLL1, DLL3, DLL4, JAG1, JAG2, HEY1 and HEY2 mRNA levels, as well as NICD2 and NICD3 protein levels, were first evaluated by the one-sample Kolmogorov-Smirnov goodness-of-fit test, in order to determine whether they followed a normal distribution pattern. Depending on the results, Pearson's or the non-parametric Spearman's rank test was used in order to examine their relationship pair-wise and their association with continuous variables (maternal age, BMI, weight gain, gestational age at delivery, birth weight). Moreover, their association with categorical data (smoking habit, mode of delivery, child gender, parity) was examined using Student's T test (after an assessment of the equality of variances using Levene's test), or its non-parametric equivalents Mann-Whitney U and Kruskal-Wallis H tests. Additionally, the Chi-square (χ 2 ) test, replaced by Fisher's exact test when indicated by the analysis, was used to examine the studied genes' expression status with the various clinicopathological parameters after stratification. Finally, univariate analysis, with gestation age, mode of delivery and smoking during pregnancy as co-factors, was used in order to correct the results for these differences among the study groups. Statistical analyses were 2-sided and were performed with the SPSS 11.5 software (SPSS, Chicago, IL). Statistical significance was set at the 95% level (p-value<0.05).

Clinical Data
Baseline demographic characteristics and medical history information (maternal weight, height, age, parity, smoking, mode of delivery, fetal gender, birth weight and its calculated centiles) were recorded for all study subjects (Table 2). Women with PE gained less weight during their pregnancies than women with normal pregnancies (p = 0.015). Gestation age was also shorter by two weeks (p = 0.012) and babies born from women with PE weighted 800g less that than those born from women with normal pregnancies (p<0.001). Additionally, due to higher pregnancy risks, women with PE were more often subjected to caesarean section than women with uncomplicated pregnancies (p = 0.025). Finally, women that developed PE during pregnancy were more often smokers prior to conceiving than women that had normal pregnancies (p = 0.001).

mRNA expression profiling
Our results revealed that NOTCH1, NOTCH4, and DLL1 were not expressed in the late preterm or term placentas of either women with normal pregnancies or women with pregnancies    Fig 1). However, the studied genes were not down-regulated in all PE samples (Fig 2). Compared to controls, the mRNA expression levels of NOTCH2 and NOTCH3 were downregulated in 100% (20/20)

Western Blot measurements
Subsequently, since NOTCH1 and NOTCH4 were not expressed, the protein expression of the intracellular domains of only NOTCH2 (NICD2) and NOTCH3 (NICD3) receptors were measured by WB. As expected, we detected a decrease in the expression of NICD2 and NICD3 in PE samples compared with controls. The average expression of NICD2 and NICD3 in PE specimens was at 24% and 15% of the same protein's expression in normal samples (p = 0.014 and p<0.001, respectively) ( Table 3 & Fig 1). In accordance to the mRNA data, NICD2 was downregulated in 70% (14/20) of PE samples, while NICD3 was universally downregulated in 100% (20/20) of PE specimens.

Statistical analysis
Univariate analysis. Univariate analysis of the data, using gestation age, mode of delivery and smoking during pregnancy as co-founding factors, revealed that the observed statistical differences in NOTCH2, NOTCH3, DLL3, DLL4, JAG1, JAG2, HEY1, HEY2, NICD2 and NICD3 levels between normal and PE-complicated placentas remained, even after correcting the data for possible discrepancies caused by the differences in the gestation period, delivery method and smoking status during pregnancy between the two groups (Table 3).
Associations with clinicopathological parameters. Statistical analysis revealed significant associations between the placental mRNA and protein levels of the studied molecules and  certain pathological parameters in cases with PE (Fig 3). NOTCH3 receptor was not expressed in women with PE that smoked during gestation compared with women with PE who did not smoke (0.00±0.00 versus 0.29±0.09, p = 0.029). Additionally, in babies born from women with PE with birth weight centile (BWC) <5 NICD3 protein levels and DLL3 mRNA levels were higher compared with babies born from women with PE with BWC>5 (NICD3: 1. Co-expression analysis. Using Spearman's rank test, we tested the mRNA co-expression pattern of the eight expressed genes in a pair-wise manner in both PE and control samples (after normalization). This test examines whether two molecules are upregulated or downregulated together (positive association), or whether when one is overexpressed and the other has a reduced expression (negative association).
In term placentas from women with normal pregnancies we observed 12 co-expression pairs. NOTCH3 was positively associated with NOTCH2. DLL4 was positively associated with receptors NOTCH2, NOTCH3 and ligands JAG1, JAG2 and DLL3. JAG1 was positively associated with NOTCH3. JAG2 was positively associated with NOTCH3 and JAG1. HEY1 target gene was positively associated with ligands JAG1, JAG2 and DLL4. HEY2 was not co-expressed with any of the other molecules (Table 4).
In late preterm and term placentas from women with PE we observed more (19 instead of 12) and different co-expressions between the studied genes than in normal samples. Specifically, the positive associations of NOTCH3 with NOTCH2 and DLL4 with JAG1 were lost in PE samples. Additionally, both JAG1 and JAG2 were not co-expressed with NOTCH3, but with  Notch Pathway Downregulation in Preeclampsia NOTCH2 instead. HEY1 was additionally associated with NOTCH2 and DLL3, and HEY2 was co-expressed with the other 7 molecules (NOTCH2,-3, JAG1,-2, DLL3,-4 and HEY1) ( Table 5).

Discussion
In the present study, we found that receptors NOTCH2 and NOTCH3 and their ligands DLL3, DLL4, JAG1 and JAG2 had reduced mRNA expression in PE placentas compared to controls. Deficiencies in NOTCH1, NOTCH2, JAG1 and DLL4 result in the failed incorporation of the placental arterial vasculature into the maternal circulation [14]. Additionally, JAG1, along with VEGF, are significantly decreased in PE placentas, a finding believed to play a role in the onset of this pregnancy complication [15]. Other studies, however, have failed to detect any differences in NOTCH2 and DLL4 expression between PE and normal placentas [8], have observed an increase in JAG1 expression through pregnancy [16], or while agreeing with our observations regarding NOTCH2 downregulation, they found NOTCH3 to be overexpressed [17], albeit in early-onset severe preeclampsia (gestation period 24-33 wks). Receptors NOTCH1 and NOTCH4 and the ligand DLL1 were not expressed in the placenta of both normal and PE-complicated pregnancies. This is in accordance with a previous study in which NOTCH1 was not expressed in CTBs and NOTCH4 levels were noticeably reduced in extravillous trophoblast cells [8]. However, other reports suggest that both NOTCH1 and NOTCH4 are expressed in the placenta [6,15,16]. Possible explanations regarding expression variations between our study and previous ones are the different experimental procedures used, since we measured expression with qRT-PCR while those studies used either Western Blot or immunohistochemistry, the different ethnic background of the study populations, and the temporal and spatial heterogeneity of the placental structures, since it is believed that CTBs alter the expression of Notch receptors and ligands as they differentiate.
NICD2 and NICD3 protein levels were also lower in PE placentas versus normal ones, a finding in accordance with the reduced mRNA levels of NOTCH2 and NOTCH3 in our PE cohort. A significant decrease in the immunoreactivity of NOTCH2 and JAG2 proteins has been observed before in human term placentas complicated by fetal growth restriction (FGR) or hypertension, compared to normal placentas [18]. Furthermore, our findings revealed that the Notch target genes HEY1 and HEY2 had a decreased mRNA expression in PE placentas compared with controls. Although no previous data on HEY genes expression in PE exist, their reduced mRNA levels are probably a direct result of the universal downregulation of the Notch signal transduction mechanism. This can be verified by our co-expression analysis, especially for HEY2, which was co-downregulated with NOTCH2, NOTCH3, JAG1, JAG2, DLL3, DLL4 and HEY1 in PE placentas, while it was not coexpressed with any of the aforementioned genes in normal placentas. However, the different interaction of HEY2 with the other Notch pathway molecules in PE-complicated placentas versus controls could also indicate that it plays a significant role in the development and progression of PE, a finding which warrants further investigation.
According to our results, some of the studied genes in the PE and control samples were positively associated with each other. However, certain co-expressions were lost, while others were gained in placentas from women with PE versus control placentas. Additionally, in PE cases a greater number of co-expressions (19 instead of 12) were observed compared to those in normal samples. The diverse interplay between these molecules in PE-affected placentas compared to normal ones could be a key event in PE pathogenesis.
Statistical analysis revealed that PE-affected pregnancies with BWC<5 had increased DLL3 mRNA and NICD3 protein levels compared with PE-affected pregnancies with BWC>5. This observation is similar to one in our previous study on pregnancies complicated by FGR, in which PHD3, a member of the HIF hypoxia pathway, was higher in babies born from FGR pregnancies with BWC<0.5 compared to babies born from FGR pregnancies with BWC>0.5 [19]. Additionally, in women with PE that were on their first parity, HEY2 transcript levels were increased compared with PE women that gave birth to their 2 nd or 3 rd child. Several studies suggest that gene expression alters between the 1 st , 2 nd and 3 rd pregnancy, as in the case of sFLT1, whose levels are higher in the 1 st pregnancy compared with its levels in subsequent ones [20]. Moreover, Hey-2 mRNA expression was higher in women with PE that gave birth with a cesarean section compared with women with PE that gave birth naturally, a finding also observed for PHD3 in our FGR study [19], and can probably be attributed to less stress during delivery and the fact that women that gave birth with a cesarean section had more severe preeclampsia than women who gave birth naturally. Finally, in women with PE that smoked during their gestation, receptor NOTCH3 was not expressed, while in women with PE that did not smoke NOTCH3 was expressed. Tobacco smoking downregulates the expression of Notch pathway members in the lung epithelium, both in smokers and chronic obstructive pulmonary disease (COPD) patients [21], and causes maternal endothelial dysfunction and abnormal placentation [22], a finding consistent with our results.
The main strength of our research is that it is the first study to concurrently measure the mRNA levels of all key Notch signaling pathway molecules, and especially target genes HEY1 and HEY2, whose expression in human placenta have never been studied before, and to determine the associations between them. Study limitations include the fact that our research was conducted only on late preterm and term placentas, and therefore we were not able to measure the expression of the Notch pathway molecules during the earlier stages of pregnancy, and that our samples were whole tissue homogenates and not micro-dissected, which means that the measured expression of the Notch signaling pathway genes was their average expression among the various placenta cell types and not their expression in each cell type individually. Evidently, these limitations could also explain the differences between our results and those of some previously conducted studies. Another potential limitation would be the differences in gestation period (shorter in PE cases), mode of delivery (since elective caesarean sections were more common in PE samples than in controls) and smoking status (since more women with PE smoked) between the two groups, which could introduce a possible bias in the results. However, since univariate analysis, with the aforementioned parameters as co-factors, found that the differences in the expression of all studied genes between PE cases and controls retained their statistical significance, our findings cannot be attributed to sampling bias, but are linked to PE pathophysiology.

Conclusions
Our findings, that receptors NOTCH2,-3, ligands DLL3,-4 and JAG1,-2 and target genes HEY1,-2 are downregulated in PE-complicated late preterm and term placentas, provide evidence that the Notch signaling pathway is associated with this pregnancy complication (visual depiction of the pathway in Fig 4). Further research regarding the function of Notch receptors, ligands and target genes in the placenta is required in order to elucidate their role in PE pathogenesis and to determine whether one or more of these genes could be used as biomarkers for the early detection of PE, or for new preventive and therapeutic strategies.