Objective
Preeclampsia (PE) is a hypertensive disorder of pregnancy characterized by widespread maternal endothelial dysfunction. Although clinical signs subside following delivery, long-term risks associated with PE include hypertension, stroke, and cardiovascular disease. MicroRNAs (miRNAs) are emerging as critical regulators of biological function, and while alterations to the miRNAome have been described in the context of pregnancy and PE, the postpartum implications of PE on miRNA expression is unknown. The goal of this study was to characterize circulating miRNA profiles at the time of delivery and at 1 year postpartum for women who did and did not develop PE.
Study Design
Using a targeted reverse transcription polymerase chain reaction approach, selected miRNAs putatively involved in the pathophysiology of PE were examined in 17 normotensive control and 13 PE maternal plasma samples at the time of delivery and 1 year postpartum. Ingenuity Pathway Analysis was used to map putative messenger RNA targets of differentially expressed miRNA to global molecular networks based on gene function.
Results
Significant increases ( P < .05) in 7 miRNAs with antiangiogenic, inflammatory, and apoptotic functions (miR-98-5p, miR-222-3p, miR-210-3p, miR-155-5p, miR-296-3p, miR-181a-5p, miR-29b-3p) were evident in maternal plasma at the time of severe PE compared to time-matched controls. Plasma samples from individuals who developed mild PE exhibited no changes compared to control samples for the subset of miRNAs analyzed here. Differential expression of plasma miRNA at the time of delivery for women with PE were largely resolved at 1 year postpartum, and reduced expression of only miR-221-3p ( P < .05) was evident. Network analysis of putative targets of differentially regulated miRNA identified 11 interacting networks with enrichment for proteins involved in cardiovascular disease, organ system development and function, and cell signaling and interaction.
Conclusion
The systemic effect of PE on maternal systems is evident in the circulating miRNAome with substantial alterations in miRNA expression in women who develop severe PE. In addition we provide novel evidence of disruption to miR-221 expression 1 year postpartum following a pregnancy complicated by PE compared to normotensive time-matched controls, which may allude to persistent inflammation in these women after delivery.
Preeclampsia (PE) is a multisystem disorder of pregnancy arising in 3-5% of pregnancies, contributing to significant maternal and neonatal morbidity and mortality worldwide. Insufficient remodeling of maternal spiral arterioles by trophoblast cells in the earliest stages of placental development, together with immune maladaptations, establishes increasingly hypoxic conditions as fetoplacental demands increase with advancing gestation. Hypoxia-induced up-regulation of placental mediators of inflammation and apoptosis coupled with reduction in the bioavailability of angiogenic molecules such as vascular endothelial growth factor (VEGF) and placental growth factor may contribute to the maternal response characteristic of PE. Maternal constitutional factors (eg, obesity, diabetes, microvascular disease) also contribute to the pathogenesis of PE, and the variable implications of the placental and maternal environments upon which the pregnancy progresses appear to contribute to the heterogeneous nature of the disorder. While hypertension and proteinuria resolve following delivery, it is apparent that the implications of PE extend well beyond that of the peripartum period. Indeed, PE is recognized as one of several pregnancy-related complications that are associated with a woman’s increased risk for future hypertension, cardiovascular disease (including but not limited to myocardial infarction and coronary artery disease), and cerebrovascular disease.
MicroRNA (miRNA) are noncoding RNA transcripts, ∼22 nucleotides long, that provide critical posttranscriptional regulation of gene expression in both health and disease through sequence-specific binding to the 3’-untranslated region of target messenger RNA (mRNA) transcripts. Recent research has identified an abundance of miRNA in the healthy term placenta, and alterations of the miRNAome in cases of placental insufficiency highlighting a role for miRNA signaling in the development of PE.
The identification of stable circulating miRNAs existing in plasma and serum has provided promise for minimally invasive biomarkers for disease prediction, diagnosis, and prognosis. However, there is very limited research in the area of maternal circulating miRNA profiles in PE populations, and none that explore the miRNAome beyond the immediate postpartum period.
While the association between PE and future development of cardiovascular disease is well established, the reasons underlying this relationship remain largely unclear. In consideration of the diverse role of miRNA in a range of biological processes, the expression levels of miRNA regulating angiogenic, inflammatory and apoptotic pathways in the plasma of women with a history of mild PE, severe PE, and uncomplicated pregnancies may provide insight into physiological processes mediating postpartum cardiovascular susceptibility in this population. This study examined targeted circulating miRNA expression profiles by real-time reverse transcription polymerase chain reaction (qRT-PCR) at the time of delivery and at 1 year postpartum in women who did and did not develop PE.
Materials and Methods
Participants and sample collection
Archival plasma samples from the PE-New Emerging Team (NET) prospective longitudinal cohort were used in this study, selected on the basis of availability, plasma volume, and quantity of data available on each sample. Plasma samples were obtained in the peripartum and at 1 year postpartum from women who did and did not develop PE who had delivered at either the Kingston or Ottawa General Hospitals, Ontario, Canada. The PE-NET study was approved by the Queen’s University Research Ethics Board (OBGY-108-03), and all participants provided written informed consent prior to sample collection. Inclusion criteria of the participants have been previously described. In brief, a diagnosis of PE was made based on PE-NET cohort diagnostic criteria: systolic blood pressure ≥140 mmHg and diastolic blood pressure ≥90 mmHg with proteinuria >0.3 g/24 hours or 1+ on repeat dipstick at the time of presentation to clinic or admission/transfer to a participating research hospital. The PE-NET cohort had collected extensive data on the maternal PE symptoms of study participants, and so for this reason, severity of PE was considered in subanalysis of the data collected. Severe PE diagnosis criteria were met if in addition to the criteria for PE, an individual presented with any 1 of the following: 2 stable measurements of systolic blood pressure ≥160 mmHg or diastolic blood pressure ≥110 mmHg a minimum of 6 hours apart, proteinuria of 5 g/24 hours or 3+ on repeat dipstick, oliguria ≤500 mL/24 hours, cerebral or visual disturbances, epigastric pain, thrombocytopenia <150,000 9 /L, aspartate transaminase >46 U/L, alanine transaminase >40 U/L, serum creatinine >106 μmol/L, pulmonary cyanosis, or intrauterine growth-restricted baby. Diagnosis of early (<34 weeks’ gestation) and late-onset (>34 weeks’ gestation) PE was also considered for subanalysis. Maternal nonfasting blood samples were collected into a heparin-containing BD Vacutainer (no. 367878; BD Biosciences, Franklin Lakes, NJ). Plasma was isolated by centrifugation (1000 g for 15 minutes). Aliquots were stored at –80°C until use.
Total RNA including miRNA isolation and heparinase treatment of the plasma samples
Total RNA including miRNA was extracted from archival plasma samples using RNeasy MinElute spin columns (Qiagen, Mississauga, Ontario, Canada), modifying the manufacturer’s instructions to optimize RNA isolation from heparinized samples. In brief, 100 μL of heparinized plasma was diluted in 100 μL RNase-free water and mixed with 1 mL of QIAzol lysis reagent (Qiagen). Samples were incubated at room temperature for 5 minutes before adding 3.5 μL Caenorhabditis elegans miR-39 miRNA mimic spike-in control (no. 219610; Qiagen) and 200 μL chloroform. Samples were vortexed, incubated at room temperature for 3 minutes, and subsequently centrifuged at 12,000 g at 4°C for 15 minutes. The upper phase was collected and mixed with 1.5 vol/vol 100% ethanol and applied to an RNeasy MinElute spin column. After washings, RNA was eluted from the column using ribonuclease (RNase)-free water, and incubated with 0.5 μL human ribonuclease inhibitor (R2520; Sigma Aldrich, Oakville, Ontario, Canada) and heparinase based on protocols previously described (heparinase I from Flavobacterium heparinum , 1 U/μL [H2519; Sigma Aldrich] in 1x reaction buffer [5 mmol/L Tris (hydroxymethyl)aminomethaneaminomethane pH 7.5, 1 mmol/L calcium chloride]) for 2 hours at 25°C. After heparinase treatment samples were stored at –80°C until complementary DNA (cDNA) synthesis.
cDNA preparation and RT-PCR assays for miRNA analysis
Total RNA including miRNA were reverse transcribed using the miScript reverse transcription kit (no. 218161; Qiagen, Mississauga, Ontario, Canada) according to the manufacturer’s instructions. In brief, 9 μL of RNA plasma preparation was combined with 4 μL 5x miScript HiSpec buffer, 10x miScript nucleic mix, 2 μL miScript reverse transcriptase mix, and 3 μL RNase-free water for a total volume of 20 μL per reverse transcription reaction. Each reaction was incubated at 37°C for 60 minutes followed by 95°C for 5 minutes to inactivate the reverse transcription reaction. cDNA was then diluted in 200 μL RNase-free water and frozen at –20°C until PCR amplification and quantification of gene products.
Custom miScript miRNA PCR arrays (no. CMIHS02264; Qiagen) were used to quantify miRNA gene expression using the manufacturer’s protocol (Qiagen). In brief, a reaction mix containing 344 μL of 2x QuantiTect SYBR green PCR master mix, 69 μL of 10x miScript universal primer, 25 μL of diluted template cDNA, and 250 μL of RNase-free water was prepared. Twenty-five μL of the reaction mix was applied to each well of the custom array containing specific miRNA primer assays.
PCR was carried out on a LightCycler 480 real-time PCR system (LC-480; Roche Applied Science, Laval, Quebec, Canada) using the cycling conditions set by the manufacturer (heat activation 95°C, 15 minutes followed by 45 cycles of denaturation: 94°C, 15 seconds; annealing: 55°C, 30 seconds; extension: 70°C, 30 seconds. Ramp rate was set at 1°C/s). SYBR green fluorescence data were collected during each extension cycle, and a dissociation curve analysis was run at the end of the cycling program for each array (70-95°C, ramp rate 0.1°C/s) to confirm the formation of specific amplified products.
Selection of miRNAs
miRNA for analysis were carefully selected following extensive review of the literature based on relevance to vascular biology and cardiovascular disease; placenta and pregnancy-specific markers were excluded from analysis to justify the reexamination of these miRNA at 1 year postpartum. Those included in the final analysis were those regulating angiogenic, inflammatory, and apoptotic pathways: hsa-let-7f-5p , hsa-miR-98-5p , hsa-miR-221-3p , hsa-miR-222-3p , hsa-miR-126-3p , hsa-miR-130a-3p , hsa-miR-210-3p , hsa-miR-155-5p , hsa-miR-17-5p , hsa-miR-18a-5p , hsa-miR-19a-3p , hsa-miR-29a-3p , hsa-miR-92a-3p , hsa-miR-20a-5p , hsa-miR-20b-5p , hsa-miR-15b-5p , hsa-miR-16-5p , hsa-miR-296-3p , hsa-miR-181a-5p , hsa-miR-195-5p , and hsa-miR-29b-3p . Primer assays for specific target miRNAs were designed using sequences obtained from the miRBase sequence database ( http://www.mirbase.org ) version 21, and sent to Qiagen for synthesis of a customized miScript PCR array. The miRBase accession numbers for the target miRNA used in this study are listed in Table 1 .
| Accession no. | Mature miRNA |
|---|---|
| MIMAT0000067 | hsa-let-7f-5p |
| MIMAT0000096 | hsa-miR-98-5p |
| MIMAT0000278 | hsa-miR-221-3p |
| MIMAT0000279 | hsa-miR-222-3p |
| MIMAT0000445 | hsa-miR-126-3p |
| MIMAT0000425 | hsa-miR-130a-3p |
| MIMAT0000267 | hsa-miR-210-3p |
| MIMAT0000646 | hsa-miR-155-5p |
| MIMAT0000070 | hsa-miR-17-5p |
| MIMAT0000072 | hsa-miR-18a-5p |
| MIMAT0000073 | hsa-miR-19a-3p |
| MIMAT0000086 | hsa-miR-29a-3p |
| MIMAT0000092 | hsa-miR-92a-3p |
| MIMAT0000075 | hsa-miR-20a-5p |
| MIMAT0001413 | hsa-miR-20b-5p |
| MIMAT0000417 | hsa-miR-15b-5p |
| MIMAT0000069 | hsa-miR-16-5p |
| MIMAT0004679 | hsa-miR-296-3p |
| MIMAT0000256 | hsa-miR-181a-5p |
| MIMAT0000461 | hsa-miR-195-5p |
| MIMAT0000100 | hsa-miR-29b-3p |
Selection of putative mRNA target genes
To determine the biological significance of differentially expressed miRNAs, the miRNA-target prediction tool miRecords ( http://mirecords.biolead.org/ ) was used to generate a list of mRNA targets for those miRNA determined to be differentially expressed between PE and control subjects. miRecords integrates results from 11 independent resources (DIANA-microT, MicroInspector, miRanda, miRDB/miRTarget2, miTarget, NBMirTar, PicTar, PITA, RNA22, RNAhybrid, and TargetScan/TargetScanS), each with their own unique target prediction algorithm. Targets were selected based on the following criteria: validated or predicted with agreement from ≥6/11 target prediction resources used by miRecords. For miR-296, these criteria were widened to targets predicted by ≥4/11 prediction tools. Although we initially attempted qRT-PCR of validated and highly predicted mRNA targets to determine the biological implications of differentially expressed miRNA in maternal plasma, amplification of mRNA targets from RNA samples proved insufficient for analysis. Given the long-term instability of mRNA and the archival nature of the current samples however, it is not surprising that our samples had negligible mRNA signal. Putative targets identified were instead evaluated using Ingenuity Pathway Analysis (IPA) (Qiagen, www.qiagen.com/ingenuity ). Selected genes were filtered through the World Wide Web–based Ingenuity Knowledge Database, a repository of curated biological interactions and functional annotations and subsequently mapped to global molecular networks based on gene function.
Statistical analysis
Continuously distributed demographic variables are presented as mean ± SD. Variables were compared using an unpaired t test or Fisher exact test where appropriate (GraphPad Prism 5 software; La Jolla, CA, www.graphpad.com ). PCR array data analysis was performed using PCR Array Data Analysis software (SABiosciences, Mississauga, Ontario, Canada, http://www.sabiosciences.com/dataanalysis.php ) with data normalized by geometric mean to the Caenorhabditis elegans miR-39 spike-in control and a threshold cycle (C T ) cutoff at 35 cycles. Enrichment of putative mRNA genes within global networks mapped by IPA was evaluated using right-tailed Fisher exact tests, considering the number of focus genes participating in a given biological process and comparing it to the total number of genes identified by the Ingenuity Knowledge Database to be associated with that process.
Results
Participants
Of 44 control and 52 PE RNA extractions, 38 and 36 samples, respectively, met array quality-control criteria. As a result, paired delivery and postpartum samples from a total of 17 normotensive pregnant controls and 13 women who developed PE were available for data analysis. Clinical characteristics of these patients are listed in Table 2 . PE subjects exhibited elevated prepregnancy body mass indices (BMI) (kg/m 2 ) and delivered early in gestation with significant hypertension, and smaller infants compared to control subjects. At 1 year postpartum BMI and systolic and diastolic blood pressures remained higher in subjects with a history of PE. Nearly 50% of PE subjects delivered with a diagnosis of severe PE. These findings are consistent with reported population characteristics of the larger PE-NET cohort. Due to a loss of samples derived from patients who had developed early-onset PE after array-based quality-control assessment, only 15.4% (n = 2) early-onset PE samples were included in the final analysis, preventing reliable subanalysis of the effect of PE onset on maternal miRNA profiles.
| Characteristic | Control (n = 17) | PE | |||||
|---|---|---|---|---|---|---|---|
| All (n = 13) | P value | Mild (n = 7) | P value | Severe (n = 6) | P value | ||
| Index characteristics | |||||||
| Maternal age, y | 28.2 ± 4.1 | 30.4 ± 7.3 | .0336 a | 32.4 ± 6.9 | .0759 | 28.0 ± 7.6 | .0817 |
| Smoking, n (%) | 1 (5.9) | 2 (15.4) | .4675 | 2 (28.6) | .3000 | 0 (0) | .7391 |
| Nulliparous, n (%) | 11 (64.7) | 8 (61.5) | .8656 | 3 (42.8) | .3711 | 5 (83.3) | .4615 |
| Prepregnancy BMI, kg/m 2 | 25.2 ± 3.5 | 28.9 ± 7.6 | .0049 a | 32.1 ± 8.3 | .0713 | 25.1 ± 5.1 | .9579 |
| GA at delivery, wk | 40.2 ± 1.5 | 37.0 ± 2.7 | .0305 a | 38.0 ± 1.3 | .0026 a | 35.9 ± 3.6 | .0296 a |
| Infant birthweight, g | 3623.2 ± 351.4 | 2965.8 ± 965.8 | .0003 a | 3332.4 ± 808.6 | .3899 | 2538 ± 1022.8 | .0515 |
| Postpartum characteristics | |||||||
| 1-y postpartum BMI, kg/m 2 | 25.9 ± 3.9 | 31.3 ± 8.7 | .0036 a | 34.2 ± 10.1 | .0727 | 27.9 ± 5.1 | .3788 |
| 1-y postpartum SBP, mm Hg | 111.0 ± 7.8 | 114.5 ± 10.9 | .3192 | 117.7 ± 12.6 | .1390 | 110.7 ± 7.8 | .9362 |
| 1-y postpartum DBP, mm Hg | 72.1 ± 7.8 | 77.5 ± 9.5 | .0939 | 81.4 ± 10.5 | .0251 a | 73.0 ± 6.4 | .8027 |
| Early-onset PE <34 wk, n (%) | – | 2 (15.4) | – | 0 (0) | – | 2 (33.3) | – |
| Lifetime risk of CVD | |||||||
| Low risk <39%, n (%) | 16 (94.1) | 8 (61.5) | .0423 a | 4 (57.1) | .0626 | 4 (66.7) | .1666 |
| High risk ≥39%, n (%) | 1 (5.9) | 5 (38.5) | 3 (42.8) | 2 (33.3) | |||
| Positive for MetS, n (%) | 2 (11.8) | 2 (15.4) | 1.000 | 2 (28.6) | .3847 | 0 (0) | .5375 |
miRNA expression in PE compared to control subjects at delivery
All the selected miRNAs were expressed in the plasma samples obtained from both PE and uncomplicated pregnancies at delivery. Of the 21 mature miRNAs examined, we found differences in several miRNA with 2- to 4-fold change compared to controls: miR-222 (2.45), miR-210 (2.216), miR-16 (–2.526), miR-296 (3.058), miR-181a (2.000), miR-29b (2.133). In particular, miR-296 and miR-181a were significantly increased ( P < .05) in PE compared to uncomplicated pregnancies at the time of delivery. Differential expression of an additional 3 miRNA neared significance (miR-222, P = .097; miR-210, P = .055; miR-29b, P = .098) ( Figure 1 , A).
