Objective
Clinical management of preeclampsia has remained unchanged for almost 5 decades. We now understand that maternal endothelial dysfunction likely arises because of placenta-derived vasoactive factors. Activin A is one such antiangiogenic factor that is released by the placenta and that is elevated in maternal serum in women with preeclampsia. Whether activin has a role in the pathogenesis of preeclampsia is not known.
Study Design
To assess the effects of activin on endothelial cell function, we cultured human umbilical vein endothelial cells in the presence of activin or serum from normal pregnant women or pregnant women with preeclampsia, with or without follistatin, a functional activin antagonist or apocynin, a NADPH oxidase (Nox2) inhibitor. We also administered activin to pregnant C57Bl6 mice, with or without apocynin, and studied maternal and fetal outcomes. Last, we assessed endothelial cell Nox2 and nitric oxide synthase expression in normal pregnant women and pregnant women with preeclampsia.
Results
Activin and preeclamptic serum induced endothelial cell oxidative stress by Nox2 up-regulation and endothelial cell dysfunction, which are effects that are mitigated by either follistatin or apocynin. The administration of activin to pregnant mice induced endothelial oxidative stress, hypertension, proteinuria, fetal growth restriction, and preterm littering. Apocynin prevented all of these effects. Compared with normal pregnant women, women with preeclampsia had increased endothelial Nox2 expression.
Conclusion
An activin-Nox2 pathway is a likely link between an injured placenta, endothelial dysfunction, and preeclampsia. This offers opportunities that are not novel therapeutic approaches to preeclampsia.
Preeclampsia remains a leading cause of maternal and perinatal death and morbidity worldwide. The recognition that much of the maternal syndrome is due to systemic maternal endothelial dysfunction and that this dysfunction likely arises, at least in part, from the excessive placental release of antiangiogenic factors offers the potential for new directed therapies. Candidate targets include soluble fms-like tyrosine kinase 1 (sFlt), a soluble splice variant of the vascular endothelial growth factor receptor, and soluble endoglin (sEng), a circulating form of the transforming growth factor–β coreceptor endoglin. Women with preeclampsia, particularly those with early-onset severe disease, have greatly increased circulating levels of both sFlt and sEng, which, when administered to pregnant rats, are able to induce a preeclampsia-like syndrome. The recent demonstration that removal of sFlt-1 by apheresis in women with early-onset preeclampsia stabilized blood pressure and prolonged pregnancy provided the first proof-of-principle evidence that the mitigation of the effects of the antiangiogenic factor sFlt is therapeutically effective.
Another vasoactive factor that is increased in the maternal circulation in women with preeclampsia is activin. Activin is a member of the transforming growth factor–β family. During pregnancy the majority of activin A in the maternal circulation is derived from the placenta. Circulating levels of activin in women with preeclampsia are approximately 10-fold higher than those in women with a healthy pregnancy. Indeed, elevated maternal levels of activin precede the clinical onset of preeclampsia by many months, which suggests that activin may have a causative role in the pathogenesis of the disease. Accordingly, we hypothesized that the increased circulating level of activin that is observed in women with preeclampsia could contribute to systemic maternal endothelial dysfunction.
Methods
Blood and tissue collection
Blood and tissues were collected from women after written, informed consent was obtained and with the approval of the Southern Health Human Research Ethics Committee. Maternal blood was collected from the antecubital vein of women with a singleton pregnancy in absence of labor. Samples were obtained from women with established preeclampsia, as defined by the Society of Obstetric Medicine of Australia and New Zealand, who were receiving labetalol and/or nifedipine and from women with a healthy singleton pregnancy at matched gestations (26-39 weeks). Serum was separated and pooled into 2 gestational age-matched pools: normal pregnancy pool and preeclampsia pool.
Human umbilical vein endothelial cell (HUVEC) isolation, culture, and treatment
HUVECs were isolated from umbilical cords that were obtained from women with a healthy singleton pregnancy and were cultured, as previously described. Cells were stimulated with 50 ng/mL recombinant human activin A (R&D Systems, Minneapolis, MN), 20% (volume/volume) preeclampsia serum, or 20% (volume/volume) normal pregnancy serum in M199 media. After 24 hours of stimulation with activin A or pooled pregnancy serum, either 600 ng/mL follistatin 288 (R&D Systems) or apocynin (1, 10, or 100 μmol/L; Sigma Aldrich, Castle Hill, NSW, Australia) was added.
Endothelial integrity
Integrity of an endothelial monolayer was determined with the use of a horseradish peroxidase–based permeability assay, as previously described. Briefly, HUVECs were grown to confluence on gelatinized 8.0 μmol/L pore size polycarbonate transwell inserts (BD Australia, North Ryde, NSW, Australia). Cells were treated with activin A or pooled pregnancy serum for 24 hours before exposure to inhibitors for a further 24 hours. Culture media were then removed from both upper and lower chambers, and 400 μL of 0.5 μg/mL horseradish peroxidase (Sigma-Aldrich) in fresh media was added to the upper chamber. Fresh media (800 μL) was added to the lower chamber and incubated for 60 minutes at 37°C. Media (30 μL) was removed and added to 860 μL reaction buffer (50 mmol/L NaH 2 PO 4 , 5 mmol/L guaiacol; Sigma-Aldrich). H 2 O 2 (100 μL) was added and incubated at room temperature for 30 minutes before absorbance was read at 470 nmol/L.
Transendothelial resistance was determined with the Millicell electrical resistance system (Millipore, Billerica, MA) as previously described. Symmetrically apposing electrodes were placed in the upper and lower chambers that allowed a uniform current to flow across the transwell insert. The electrodes were immersed so that the shorter electrode was in the upper chamber and the longer electrode was in the lower chamber. The resistance was recorded when the meter indicated a stable resistance.
Quantitative real-time polymerase chain reaction on HUVECs
RNA was isolated from the cells with Trizol reagent (Life Technologies, Mulgrave, Victoria, Australia); 1 μg DNase-treated RNA was used for the synthesis of complementary DNA (cDNA) with the Thermoscript real-time polymerase chain reaction (PCR) system (Life Technologies). Messenger RNA (mRNA) expression was determined with quantitative PCR performed on Rotorgene (Qiagen Pty Ltd, Chadstone, Victoria, Australia) with prevalidated primers for NADPH oxidase (Nox2), endothelin-1, and GAPDH (SABiosciences, Qiagen Pty Ltd) at the following cycling conditions: 95°C for 10 minutes then for 40 cycles of 95°C for 20 seconds, 60°C for 30 seconds, and 72°C for 30 seconds. Data were compiled using 3 cell isolates and 4 technical replicates.
Reactive oxygen species measurement
HUVEC were allowed to grow to confluence in 96-well view plates (PerkinElmer, Glen Waverley, Victoria, Australia). Cells were stimulated for 12 hours with either activin A or 20% pooled pregnancy serum with or without inhibitors. The plate was then dark-adapted for 20 minutes in a Chameleon chemiluminescence plate reader (Hidex, Turku, Finland) before the addition of the 1 μmol/L reactive oxygen species (ROS)-detecting chemiluminescent agent, L-012 (Wako Chemicals, Osaka, Japan). Chemiluminescence was measured every 2 minutes for 40 minutes. Each cell preparation was measured in 6 replicate wells. To confirm these findings, ROS were also detected with 10 μmol/L carboxy-H 2 -DCFDA (Life Technologies); fluorescence was measured in a Cellomics ArrayScan VTI HCS plate reader (Thermo Fisher Scientific, Scoresby, Victoria, Australia).
Measurement of zona occludens and Nox2
HUVEC extracts were resolved by sodium dodecylsulfate-polyacrylamide gel electrophoresis, and Western blots were probed overnight at 4°C with primary antibodies anti-Nox2 (1:1000, SAB4200118, Sigma Aldrich), anti-zona occludens (ZO-1) (1:100, 40-2200; Life Technologies) or anti-β actin (1:5000, sc-4778; Santa Cruz Biotechnology Inc, Santa Cruz, CA) and appropriate secondary antibodies. Signals were detected by chemiluminescence (Clarity Western ECL, Biorad, Hercules, CA) and quantified with the ChemiDoc XRS+ (Biorad). Protein levels for Nox-2 and ZO-1 are represented relative to β actin expression.
Measurement of activin, follistatin, and 8-isoprostane
Total activin A in maternal serum and conditioned media was measured with a 2-site enzyme-linked immunosorbent assay (Diagnostic Systems Laboratories, Houston, TX) with minor modifications as previously described. Maternal sera were diluted 1:10; assay diluent and culture supernatant were assayed neat. Follistatin was measured as previously described : 8-isoprostane was measured with a commercial assay (Cayman Chemical, Ann Arbor, MI) according to manufacturer’s instructions.
Small interfering RNA transfection
HUVECs grown to 70-80% confluence were reverse-transfected with either single-stranded small interfering RNA targeted towards Nox2, a scrambled sequence (negative control), or GAPDH (positive control), according to the manufacturer’s instructions (Ambion; Life Technologies). Studies were performed within 4 days of transfection.
Quantitative PCR for Nox2 and endothelial nitric oxide synthase on human and mouse blood vessels
Small resistance vessels were isolated from the abdominal fat of women who underwent a cesarean delivery. We obtained samples from 6 women with preeclampsia (mean ± SEM: gestational age, 32 ± 0.6 weeks; body mass index, 35 ± 2.6 kg/m 2 ) and 10 women with a healthy singleton pregnancy (gestational age: 38.0 ± 0.5 weeks; body mass index, 30.4 ± 2.7). Biopsy specimens of abdominal fat were obtained from the site of surgical incision and placed in RNAlater (Ambion; Life Technologies). Blood vessels were then blunt dissected under a dissecting microscope before RNA isolation with the RNAeasy micro kit (Qiagen Pty Ltd). CDNA conversion was performed with the Superscript VILO cDNA synthesis kit (Invitrogen, Life Technologies). Quantitative PCR was performed on Rotorgene (Qiagen Pty Ltd) with the primers detailed in Table 1 . For mRNA expression analysis of mouse blood vessels, the thoracic aorta was isolated under a dissecting microscope at the time of cull. RNA isolation, cDNA synthesis, and quantitative PCR were performed as described earlier with primers as detailed in Table 1 .
| Gene | Primer |
|---|---|
| Human Nox2 forward | 5′-TGG CAC CCT TTT ACA CTG-3′ |
| Human Nox2 reverse | 5′-CCA CTA ACA TCA CCA CCT CA-3′ |
| Human eNOS forward | 5′-GTG ATG GCG AAG CGA GTG AA-3′ |
| Human eNOS reverse | 5′-CCG AGC CCG AAC ACA CAG AA-3′ |
| Mouse eNOS forward | 5′-CAA CGC TAC CAC GAG GAC A-3′ |
| Mouse eNOS reverse | 5′-CTC CTG CAA AGA AAA GCT CTG G-3′ |
| 18s forward | 5′- GTC TGT GAT GCC CTT AGA TGT C-3′ |
| 18s reverse | 5′- AAG CTT ATG ACC CGC ACT TAC-3′ |
Mouse model of preeclampsia
All animal studies were approved by Monash Medical Centre animal ethics committee and conducted in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes (2006). We administered 50 μg recombinant human activin A (0.36 mg/kg/d; R&D Systems, Minneapolis, MN) into 8- to 10-week-old C57Bl6 female mice by means of a subcutaneous miniosmotic pump (AlZET 2001; ALZET, Cupertino, CA) implanted at day 10 of gestation. Control animals received an equal volume of saline solution through an osmotic pump. We confirmed levels of circulating activin A using a commercial enzyme-linked immunosorbent assay kit (R&D Systems). Levels of albumin and creatinine in 24-hour urine sample and sera levels of aspartate transaminase and lactate dehydrogenase were analyzed at the Monash Medical Centre Department of Clinical Biochemistry. Maternal blood pressures were determined by noninvasive tail cuff (MC-4000; Hatteras Instruments, Cary, NC). Animal numbers in each group were as follows: saline solution (n = 6), activin (n = 5), and activin + apocynin (n = 6).
Microvascular reactivity experiments
Relaxation response to acetylcholine was determined in carotid arteries of pregnant mice at embryonic day 16 (E16). Rings of artery (2 mm) were mounted on stainless steel wires between an isometric force transducer and a micrometer in a Mulvany-Halpern myograph filled with carbogen-bubbled Krebs-bicarbonate solution. After 30 minutes equilibration at 37°C, the maximum contraction of the arteries in response to 1 μmol/L U-46619 was determined. Arteries were then stretched to approximately 30% maximal contraction with U46619, then 100 μmol/L N -nitro- l -arginine methyl ester (L-NAME) was added and incubated for a further 30 minutes at 37°C before a final response was recorded.
Immunohistochemistry
Maternal aortas from the mouse studies were embedded in optimal cutting temperature compound, and cryosections were cut at 10-μmol/L thickness. Dihydroethidium staining was performed on freshly cut cryosections, where a prewarmed solution of 10 μmol/L dihydroethidium in Hanks-buffered saline solution was applied directly to the tissue section and incubated at 37°C before imaging. Cryosections were fixed in ice-cold 100% ethanol and air-dried before being immunostained with rabbit polyclonal anti-gp91phox antibody (1:200 ab80508; Abcam, Cambridge, UK) and goat anti-rabbit secondary antibody conjugated to Alexa Fluor 568 (1:200 A-11036; Invitrogen, Life Technologies). Sections were coverslipped in fluorescent mounting media (Dako Australia Pty Ltd, Campbellfield, Victoria, Australia). Paraffin-embedded, formalin-fixed brains were cut at 10-μmol/L thickness then stained with anti-albumin rabbit polyclonal antibody (1:1000, AIA6140; Accurate Chemical & Scientific, Westbury, NY) overnight at 4°C before being probed with a biotinylated anti-rabbit antibody. Staining was visualized with a 2-step LSAB+ kit (Dako Australia Pty Ltd) and liquid 3,3′-diaminobenzidine (Dako Australia Pty Ltd).
Statistical analysis
Data were analyzed with GraphPad Prism 5 software for Mac OS X (GraphPad Software, La Jolla, CA). Data are expressed as means ± SEM. Pair-wise comparisons were performed with a 1-tailed t test. One-way analysis of variance was used for multiple group comparisons. Two-way analysis of variance and post-hoc Bonferroni’s test were used for multiple comparisons. In vitro data were collected from 3 separate experiments in which there were 3 samples, unless otherwise stated.
Results
Effects of activin on HUVECs in vitro
Activin A (50 ng/mL) increased the production of ROS ( Figure 1 , A) and the lipid peroxidation product 8-isoprostane ( Figure 1 , B) from HUVECs, increased transendothelial permeability, and reduced transendothelial resistance in a HUVEC monolayer ( Figure 1 , C and D). All of these effects were prevented by the addition of follistatin (FS288), which is an activin-binding protein that blocks activin activity ( Figure 1 ). Exploring the mechanism of impaired endothelial integrity after activin exposure, we assessed the expression of the cell junctional protein zona occludens 1 (ZO-1). Compared with controls, activin reduced ZO-1 expression, which is an effect that is prevented by FS288 ( Figure 1 , E and F).
To determine the source of the ROS from HUVECs, we measured Nox2 mRNA expression that showed that activin A significantly increased Nox2 expression, which is an effect that is abolished by either FS288, or apocynin Nox inhibitor ( Figure 2 , A). Apocynin also prevented the production of ROS, lipid peroxidation, and impaired endothelial functions that were induced by activin ( Figure 1 ). To confirm that the increased endothelial ROS production and impaired endothelial integrity after exposure to activin were due to increased Nox2 activity, we showed that increased Nox2 mRNA expression, increased ROS production, and decreased transendothelial resistance that is induced by activin were all mitigated by Nox2 gene silencing ( Figure 2 , B-D). To explore the upstream signal transduction mechanisms by which activin A induced Nox2 activity, we assessed Smad2/3 phosphorylation and nuclear localization. Smad2/3 is the postreceptor signaling pathway for activin. Compared with controls, treatment with activin A induced Smad2/3 phosphorylation and nuclear localization in HUVECs, which are effects blocked by FS288 but not apocynin ( Figure 3 , A and B).
Effects of normal and preeclamptic pregnancy serum on HUVECs in vitro
To explore whether activin may be a component of preeclamptic serum that induces endothelial dysfunction, we repeated the in vitro experiments by culturing HUVECs in either pooled serum that was collected from women with a healthy singleton pregnancy or from women with established preeclampsia. The characteristics of the women from whom serum was collected are outlined in Table 2 . The level of maternal serum activin in the women with preeclampsia was approximately 4-fold higher than that in women with a healthy pregnancy. Compared with being cultured in media with 20% (volume/volume) normal pregnancy sera, HUVECs that were cultured in media with 20% (volume/volume) preeclamptic serum had significantly increased Nox2 mRNA expression ( Figure 4 , A) and ROS production ( Figure 4 , C), decreased endothelial resistance ( Figure 4 , D), and reduced ZO-1 expression ( Figure 4 , E) and protein ( Figure 4 , F), which were effects that were prevented by the addition of either FS288 or apocynin ( Figure 4 ). Follistatin, but not apocynin, decreased Nox2 protein levels in HUVECs ( Figure 4 , B). Consistent with these effects being mediated by activin through Nox2, culturing HUVECs in preeclamptic serum increased nuclear localization of phosphorylated Smad2/3, which is an effect that is prevented by FS288 but not apocynin ( Figure 3 , C and D); all effects of preeclamptic serum were mitigated by Nox2 gene silencing ( Figure 2 ).
| Variable | Normal | Preeclamptic | P value |
|---|---|---|---|
| Maternal age, y | 32.5 ± 1.1 | 30.2 ± 1.1 | .1 |
| Gestational age, wk | 31.5 ± 0.8 | 33.5 ± 0.9 | .1 |
| Body mass index, kg/m 2 | 25.2 ± 1.7 | 26.4 ± 2.2 | 1.0 |
| Parity, n | 0.6 ± 0.2 | 0.6 ± 0.2 | .8 |
| Systolic blood pressure, mm Hg | 107.3 ± 2.2 | 157.6 ± 3.5 | < .0001 |
| Diastolic blood pressure, mm Hg | 62.4 ± 2.5 | 102.4 ± 1.7 | < .0001 |
| Proteinuria, g/mL/24hr | 0 ± 0 | 1.0 ± 0.3 | < .0001 |
| Activin A, ng/mL | 12.0 ± 1.6 | 49.5 ± 7.0 | < .0001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
