Retinoic Acid Is a Negative Regulator of sFLT1 Expression in Decidual Stromal Cells, and Its Levels Are Reduced in Preeclamptic Decidua

Abstract

sFLT1 (soluble VEGF [vascular endothelial growth factor] receptor-1) levels are increased in preeclampsia—a pathological condition of pregnancy. The mechanism of sFLT1 overexpression by gestational tissues, particularly the decidua, remains unknown. Mass spectrometry measurement of the active retinoid metabolite, all-trans retinoic acid (RA), showed significantly lower levels of RA in preeclamptic versus normotensive decidua. In this study, we investigated the involvement of RA in regulating decidual sFLT1 expression. When decidual stromal cells (DSCs) isolated from the decidua basalis of normotensive and preeclampsia placentas were treated with BMS493—a pan-RAR (RA nuclear receptor) antagonist—upregulation of sFLT1 expression was observed. Conversely, treatment with RA resulted in downregulation of sFLT1 in normotensive DSCs and preeclampsia DSCs. Unlike treatment with cAMP, which induces decidualization while downregulating sFLT1, RA treatment did not alter DSC expression of prolactin—a marker of decidualization—or FOXO1 (forkhead box protein 01)—a transcription factor required for prolactin upregulation. TFAP2A (transcription factor AP-2-alpha [activating enhancer-binding protein 2 alpha]), a different transcription factor was upregulated in normotensive DSCs but not in preeclampsia DSCs after RA treatment. Collectively, our data show that RA suppresses sFLT1 expression in DSCs independently of cellular decidualization. These findings suggest that reduced decidual RA levels may contribute to preeclampsia pathogenesis by allowing sFLT1 accumulation at the maternal-fetal interface.

Preeclampsia—a hypertensive disorder of pregnancy that is estimated to affect 2% to 8% of pregnant women worldwide—is generally diagnosed after the 20th week of pregnancy with the onset of hypertension leading to systemic inflammation and kidney, liver, and neurological dysfunction.¹ Recent studies suggest that a maternal antiangiogenic state, which is a result of overexpression of sFLT1 (soluble VEGF [vascular endothelial growth factor] receptor-1), plays a major role in preeclampsia pathology.^2–4 In addition, oxidative stress and failed immune adaptation have all been suggested to be responsible for pathological characteristics of preeclampsia, such as poor trophoblast invasion, placental hypoxia, and malformation of spiral arteries.^5–8

In preparation for pregnancy, the human endometrium that will later become the maternal-fetal interface (MFI) first undergoes a dramatic cellular transformation. Uterine stromal cells enter a state of mesenchymal-to-epithelial transition called decidualization that is essential for embryo implantation and subsequent fetal development. It is well known that progesterone is a primary mediator of this cellular process, and in a recent publication, we showed that, in addition to upregulation of prolactin (PRL), IGFBP1 (insulin-like growth factor-binding protein), and Cx43 (connexin 43), decidualization of endometrial stromal cells is also associated with downregulation of sFLT1.⁹ sFLT1 is an antiangiogenic protein that binds VEGF and PlGF (placental growth factor). It is produced and secreted in a controlled manner during healthy pregnancies by various cell types, including decidual, trophoblastic, endothelial, and immune cells. Its implication in the pathogenesis of preeclampsia reflects the increased circulating levels of sFLT1 in plasma of affected mothers. Recently, Garrido-Gomez et al^l0 showed that decidual stromal cells (DSCs) isolated from the decidua basalis and parietalis of placentas from patients with a history of preeclampsia produce lower levels of PRL and IGFBP1 than DSCs isolated from healthy patient placentas. Further, our group has found that DSCs isolated from the decidua of term preeclampsia placenta fail to fully decidualize and exhibit reduced sFLT1 downregulation when stimulated by hormones in vitro.¹¹ While the presence of FLT (vascular endothelial growth factor receptor-1)/sFLT1 in decidual tissue has been established by immunohistochemistry,^12–14 gene array analysis,¹⁵ and in stromal cells by polymerase chain reaction (PCR),¹⁶ ours is the first study to show the effect of decidualization on sFLT1 expression.⁹ Thus, although the mechanisms driving overexpression of sFLT1 during preeclampsia are not known, these observations suggest that defective decidualization at the MFI predisposes preeclampsia patients to impaired decidual differentiation.

For years, the promotion of endometrial decidualization by estrogen and progesterone has been recognized, but recently, mediation by local cofactors, including retinoic acid (RA), has emerged.¹⁷ This observation is supported by the presence of RARs (RA nuclear receptors) and CRABPs (cellular RA-binding proteins) in DSCs, as well as by the highly orchestrated changes that these proteins undergo during the decidualization process.^17,18 There is also a clear precedent for the relationship between deficient RA levels and proinflammatory conditions in the gastrointestinal tract.^19,20 It is known that inflammatory bowel disorders are associated with reduced suppressor T regulatory cells,²¹ which is notably also a characteristic feature of preeclampsia.^22,23 This context suggests that RA may play a regulatory role in decidualization and the proinflammatory preeclampsia state.^24–26 Given this background, we seek to characterize the effect of RA on stromal cell decidualization and sFLT1 production.

Materials and Methods

The data that support the findings of this study are available from the corresponding author on reasonable request.

Patient Samples

Decidual samples were obtained from term placentas of preeclamptic (n=13) and normotensive (n=11) pregnant women at Emory University Hospital Midtown (Atlanta, GA) with written informed consent. The study protocol was approved by the Emory Institutional Review Board (IRB00078902). The 2017 ACOG Hypertension Guidelines were used to define preeclampsia.²⁷ Our inclusion criteria include nonsmoking, English-speaking women >18 years. Patients were excluded if they had preexisting chronic conditions (such as heart disease, diabetes mellitus, pregestational hypertension, and autoimmune disorders), pregnancy complications besides preeclampsia, or any known fetal anomalies. Normotensive pregnancies served as controls.

Quantification of Retinoids in Decidua

Retinoid concentrations were measured as described previously^28–30 using placental decidua that were collected within 30 minutes of delivery and flash-frozen in liquid nitrogen. Detailed methodology is provided in the online-only Data Supplement.

Isolation of DSCs

Immediately after delivery, a portion of the dissected decidua basalis was transferred to our laboratory within 30 minutes in PBS. DSCs were isolated by enzymatic digestion of the decidua using 0.1% collagenase.⁹ Growth medium (DMEM/Ham F-12 supplemented with L-glutamine containing 10% fetal bovine serum, 1.0 nmol/L sodium pyruvate, 1% nonessential amino acid, 1% penicillin-streptomycin, and 1% amphotericin B) was added to the digest, and cells were seeded in a T75 flask. After a 24-hour attachment period, supernatant containing unbound cells and tissue was replaced by fresh medium. The cells were subcultured twice for stromal cell enrichment and then trypsinized and stored in liquid nitrogen (10% fetal bovine serum and 90% dimethyl sulfoxide) as aliquots. Stromal cell homogeneity was confirmed with vimentin staining.³¹

Treatment With BMS, RA, and cAMP

For experimental setup, cells were expanded in T75 flasks and seeded onto 6-well plates and grown to 90% confluence. All experiments were conducted in cells before the fifth passage. Cells were treated with 0.1 μM of RA, 1 μM BMS493 (a pan-RAR inhibitor³²), or 0.5 mmol/L cAMP (N6,2′-O-Dibutyryladenosine 3′,5′-cyclic monophosphate sodium salt; Sigma Chemicals, St. Louis, MO) for 12 days. Media was changed every fourth day. On day 12, cell supernatant was collected for ELISA and Western blot analysis, while cells were washed with 1× PBS, collected, and stored as snap-frozen pellets for RNA studies.

ELISA

Supernatant collected from control and treated cells was subjected to Flt1/Vascular endothelial growth factor receptor-1 ELISA (R&D Systems, Minneapolis, MN) according to manufacturer’s instructions.³³ The assay was performed in duplicate and averaged for analysis.

Western Blot Analysis

For Western blot analysis of sFLT1 in the medium, 50 μL of Heparin-Agarose (Sigma) was added to the cell supernatant (5 mL), and bound sFLT1 was eluted in 25 μL of SDS sample buffer. After electrophoresis and transfer of protein to nitrocellulose membranes, sFLT1 was detected using a vascular endothelial growth factor receptor-1 antibody from Sigma Chemical Company (catalog No. V4262) and a fluorescent tagged secondary antibody using Odessey CLx (LiCor, Lincoln, NE).⁹

Quantitative PCR

RNA was isolated from cells using Trizol (Invitrogen, Carlsbad, CA). Next, cDNA was prepared and subjected to SYBR Green quantitative PCR as reported previously.⁹ hRPL17 (housekeeping gene human ribosomal protein L17) was used as an internal control. Primer sequences and annealing temperatures are shown in Table S1 in the online-only Data Supplement.

Statistical Analysis

Given the relatively small number of observations, and the difficulty to prove the data had a Gaussian distribution, we chose to use conservative, nonparametric measures to compare our results. Clinical parameters and RA estimation (Figure 1) were analyzed with Mann-Whitney U test and are presented as mean±SD. When >2 treatment groups were compared, Kruskal-Wallis H tests are presented as median±25th to 75th percentile (interquartile range) or as percentage of control. GraphPad PRISM software, version 7.0 (GraphPad Software, Inc, San Diego, CA), was used for analysis, and a P of ≤0.05 was considered statistically significant. Treated samples were standardized to their own controls.

Figure 1.

Retinoic acid (RA) levels are reduced in preeclamptic (PE) decidua. A, RA concentration (pmol/g protein) in whole decidual tissue was assessed by liquid chromatography with multiple reaction monitoring (3) in normotensive (NT) patients (n=9) and PE patients (n=10). B, Retinol (ROL) and retinyl esters (RE) were quantified via high performance liquid chromatography with ultra-violet detectors (nmol/g protein) in NT and PE patients. C, RA concentrations in PE, shown as >34 wk gestation (n=5) and <34 wk gestation (n=5). *P<0.05, **P≤0.005.

Results

Clinical Characteristics of Patients

Decidua were collected from a total of 24 women (11 normotensive and 13 preeclampsia). Clinical characteristics are shown in Table S2. Individual decidual RA levels were determined by LC-MS/MS analysis of the decidua of 19 patients (normotensive [n=9] and preeclampsia [n=10]). For experimental in vitro studies, stromal cells were isolated from normotensive (n=6) and preeclamptic (n=6) decidua. Maternal age and maternal body mass index were comparable between normotensive and preeclampsia patients. Measures of blood pressures, gestational age (GA), neonatal weight, and placental weight differed significantly between groups.

Endogenous RA in Normal and Preeclamptic Decidua

We determined the levels of RA in decidua of normotensive and preeclampsia patients using liquid chromatography with multiple reaction monitoring (3) and high performance liquid chromatography with ultra-violet detectors. RA levels in preeclamptic decidua were 30% lower than those detected in normotensive decidua (P=0.002; Figure 1A). RA isomers without known biological activity—13-cis and 9,13-di-cis RA—were also detected, but they did not differ based on source (not shown). No other RA isomers, including 9-cis RA, were detected above the liquid chromatography with multiple reaction monitoring (3) assay limit of detection in biologic matrices (≈12 pmol/g protein). Retinol and retinyl ester levels were comparable in preeclamptic and normotensive decidua (Figure 1B), suggesting that RA biosynthesis is not rate limited by substrate availability.

Although this is a relatively small study, given the significant difference in GA observed in our normotensive and preeclampsia cohorts, we delineated the potential effect of GA on RA levels in preeclampsia samples. Of the 10 patients in the preeclampsia group, 5 were >34 weeks GA (37.5±2.2) and the other 5 were <34 weeks GA (31.22±3.0). Each group’s RA level was significantly different from the normotensive group: at >34 weeks, preeclampsia was 163.45±15.8 (P=0.002) and at <34 weeks, it was 197.0±36.0 (P=0.04), each compared with 247.4±33.1 pmol/g protein (Figure 1C). Despite marginal differences in RA levels between the preeclampsia groups, we did not observe a significant effect of GA on RA levels (P=0.22).

Flt1 Promoter and Identification of a Putative RAR/RXR-Binding Site

Bioinformatic analysis of the human FLT promoter using MotifMap^34,35 revealed a RXR/RAR-binding element at −522 bp from the transcriptional start site and an activator protein 2 site (TFAP2A [transcription factor AP-2-alpha (activating enhancer-binding protein 2 alpha)]) at −611.

BMS493—a Pan-RAR Inhibitor—Upregulated sFLT1 Expression in DSCs

We first sought to determine whether the RA pathway has any biological role in sFLT1 regulation. Exposure to BMS493—a pan-RAR inhibitor that blocks the action of endogenously produced RA—resulted in upregulation of sFLT1 expression in both normotensive DSCs (NT-DSCs) and preeclampsia DSCs (PE-DSCs). Using ELISA analysis of sFLT1 levels in supernatant, we found that the upregulation was 3.67-fold (2.46–6.14; P=0.002) in NT-DSCs and 2.28-fold (1.69–3.87; P=0.02) in PE-DSCs (Figure 2A). Western blot analysis of the supernatant also showed that treatment with BMS493 significantly upregulated sFLT1 expression in both NT-DSCs (3.50-fold [1.69–5.48]; P=0.005) and PE-DSCs (3.29-fold [2.08–5.13]; P=0.007; Figure 2B and 2C) compared with untreated controls. This increase in sFLT1 production with addition of BMS493 suggests that the RA pathway is involved in the regulation of sFLT1 production by DSCs.

Figure 2.

BMS493 treatment of decidual stromal cells (DSCs) causes upregulation of sFLT1 (soluble vascular endothelial growth factor receptor-1). DSCs isolated from normotensive (NT) and preeclamptic (PE) decidua treated with BMS493–a pan-RAR (retinoic acid nuclear receptor) antagonist. A, Soluble Flt1 expression in the supernatant quantified by ELISA in both NT- and PE-DSCs. B, Supernatants of treated DSCs were concentrated by Heparin-Agarose and subjected to Western blot analysis. C, Quantified intensity of protein bands. *P<0.05, **P≤0.005. BMS indicates BMS493; and Ctrl, control.

RA Downregulated sFLT1 Production in DSCs

NT-DSCs and PE-DSCs treated with RA significantly downregulated the expression of sFLT1 in both cells (Figure 3; P<0.05). RA treatment reduced sFLT1 mRNA expression to 0.55-fold (0.37–0.81; *P=0.048) in NT-DSCs and 0.34-fold (0.28–0.75; **P=0.026) in PE-DSCs (Figure 3A) relative to respective controls (1.0). cAMP treatment also downregulated sFLT1 mRNA to 0.28-fold (0.13–0.40; P=0.0004) in NT-DSCs and 0.23-fold (0.20–0.67; P=0.001) in PE-DSCs relative to untreated controls (1.0; Figure 3A). However, after cAMP treatment over 12 days, NT-DSCs showed a 10% higher downregulation of sFLT1 than the PE-DSCs (as we have previously reported in a 8-day treatment protocol¹¹).

Figure 3.

Retinoic acid (RA) treatment reduces sFLT1 (soluble vascular endothelial growth factor receptor-1) expression in decidual stromal cells (DSCs). A, Quantitative polymerase chain reaction analysis of sFLT1 mRNA. B, ELISA levels of sFLT1 in supernatant. C, Representative Western blot image showing the sFLT1 bands during RA and cAMP treatment. D, Quantification of band intensity in Western blot. Together, these data show that both RA and cAMP treatment significantly downregulates the expression of sFLT1 by both normotensive (NT) and preeclampsia (PE) DSCs. *P=0.05, **P<0.05, ***P≤0.005. Ctrl indicates control.

Secreted sFlt1 protein in supernatant was quantified with ELISA. Here, RA treatment lowered sFLT1 protein levels to 0.35 (0.25–0.72; P=0.026) in NT-DSCs and 0.22 (0.16–0.33; P=0.006) in PE-DSCs in comparison with untreated control (1.0). Similarly, cAMP downregulated sFLT1 protein secretion to 0.20 (0.09–0.37; P=0.001) of untreated control (1.0) in NT-DSCs and 0.24 (0.10–0.47; P=0.006) of untreated control (1.0) in PE-DSCs after 12 days of treatment (Figure 3B)

Western blot quantification of sFLT1 in supernatant shows that sFLT1 levels are reduced in both NT-DSCs (0.48 [0.26–0.66]; P=0.09) and PE-DSCs (0.31 [0.15–0.46]; P=0.022) treated with RA (Figures 3C and 3D). cAMP treatment also resulted in sFLT1 downregulation in NT-DSCs (0.07 [0.06–0.09]; P=0.0001) and PE-DSCs (−0.18 [0.15–0.19]; P=0.0013; Figure 3C and 3D) compared with controls (1.0).

Downregulation of sFLT1 by RA Is Independent of Decidualization

Because decidualization is a known modulator of sFLT1 expression, we sought to determine whether RA-induced sFLT1 downregulation is mediated by decidualization. Although there was no significant change in the expression of PRL in RA-treated NT-DSCs (0.79 [0.52–1.17]; P>0.99; Figure 4A), RA-treated PE-DSCs did show significant downregulation of PRL (0.69 [0.46–0.85]; P=0.047; Figure 4A). In contrast, cAMP supplementation resulted in PRL upregulation in both NT-DSCs (20.14 [17.26–45.65]; P=0.016) and PE-DSCs (7.79 [2.70–24.77]; P=0.041; Figure 4A). After 12 days of cAMP treatment, PRL levels were 3× higher in NT-DSCs than in PE-DSCs (Figure 4A).

Figure 4.

Effect of retinoic acid (RA) on prolactin (PRL), FOXO1 (forkhead box protein 01), and TFAP2A (transcription factor AP-2-alpha [activating enhancer-binding protein 2 alpha]) mRNA expression. A, To determine whether sFtl1 modulation is decidualization dependent, normotensive (NT) decidual stromal cells (DSCs; n=6) and preeclampsia (PE) DSCs (n=6) were treated with RA or cAMP A, PRL, (B) FOXO1, and (C) TFAP2A mRNA expression were then measured by quantitative polymerase chain reaction and presented relative to controls (1.0). Effect of BMS493 on DSCs is shown in Figure S1. hRPL17 indicates housekeeping gene human ribosomal protein L17. *P≤0.05, **P<0.005. BMS indicates BMS493; and Ctrl, control.

Effect of RA on TFAP2A and FOXO1

To further elucidate the mechanism of RA-mediated sFLT1 transcription changes in DSCs, we analyzed expression of various transcription factors using a quantitative PCR array (PAHS-075Z, Qiagen). Based on results of this quantitative PCR array, we characterized the RA responses of a select number of transcription factors: TFAP2A, FOXO1 (forkhead box protein 01), PPAR (peroxisome proliferator-activated receptors [PPARα, PPARβ/δ, and PPARγ), and CYP26A1 (cytochrome p450 family 26 A1). The expression levels of these transcription factors after RA exposure were then quantified, with expression levels after cAMP treatment serving as positive control. Effect of BMS493 treatment on PRL, FOXO1, and TFAP2A mRNA expression is shown in Figure S1. While the PRL levels showed slight decrease, there was a downregulation of FOXO1 in DSC and selective overexpression of TFAP2A in NT-DSC (P<0.05) under BMS493 treatment. RA treatment did not result in any significant change in FOXO1 expression in either NT-DSCs (1.15 [0.85–1.68]; P=0.99; Figure 4B) or PE-DSCs (0.80 [0.30–1.90]; P>0.99; Figure 4B). However, cAMP treatment significantly upregulated FOXO1 expression in both NT-DSCs (18.10 [7.05–23.08]; P=0.001) and in PE-DSCs (6.49 [4.43–9.50]; P=0.022; Figure 4B). FOXO1 induction in PE-DSCs, although significant, is 2.5-fold less than the induction seen with NT-DSCs.

RA treatment significantly upregulated the expression of TFAP2A in NT-DSCs (3.84 [2.79–4.42]; P=0.0049; Figure 4C) but not in PE-DSCs (0.59 [0.41–1.09]; P=0.67; Figure 4C). Treatment with cAMP also increased expression of TFAP2A in NT-DSCs (2.55 [1.03–4.11]; P=0.11; Figure 4C). No significant changes in TFAP2A levels were observed in cAMP treatment of PE-DSCs (0.50 [0.27–2.88]; P=0.62; Figure 4C). The other genes investigated, specifically PPARα, PPARβ/δ and PPARγ, and CYP26A1, did not show measurable changes with either RA or cAMP (data not shown).

Discussion

Two major new findings are presented in this study. First, RA levels are significantly lower in preeclamptic compared with normotensive decidua. Second, RA-mediated downregulation of sFLT1 in DSCs occurs independently of decidualization and appears to be mediated by a repressive RXR/RAR element located at −522 bp in the FLT gene promoter.

The MFI is composed of an admixture of trophoblasts, endothelial cells, and DSCs, which are all indispensable for embryo implantation and placental development. Defects in this process affect the growth and development of the fetus.^36,37 Experimental evidence has recently begun to implicate defective decidualization in preeclampsia and other pregnancy complications. Garrido-Gomez et al¹⁰ showed that both DSCs obtained from women with current preeclampsia and endometrial stromal cells from women with a remote history of preeclampsia have defects in their ability to decidualize. Our own investigations of DSC response to decidualization stimuli have shown that subtotal decidualization in DSCs from preeclampsia pregnancies is linked to incomplete suppression of sFLT1.¹¹ Preeclampsia pathology has long been characterized by elevated circulating levels of the antiangiogenic protein sFLT1. When characterizing the promoter sequence of sFLT1[34, 35], we identified a putative binding site for RAR/RXR (−522 bp) and TFAP2A (−611 bp) transcription factors involved in the RA pathway, suggesting that other regulatory mechanisms may contribute to controlling sFLT1 production, particularly in DSCs. Earlier reports by Morishita et al³⁸ and Wakiya et al³⁹ have suggested the presence of negative regulatory elements in the Flt promoter. This is in addition to the previously described TATA box, a GC-rich region and putative binding site for CREB (cAMP response element-binding protein)/ATF (activating transcription factor-3), multiple Ets-1 (avian erythroblastosis virus E26 oncogene homolog-1), and Egr-1 (early growth response-1 motifs).^38,40 In addition, a hypoxia-inducible enhancer element (−976 to −937) was also found in the Flt promoter.⁴¹

RA is known to play a crucial role in fetal development of the eyes, heart, and kidneys.⁴² Fetal malformations, including sensorineural hearing and vision loss, renal hypoplasia, and impaired lung maturation, have been associated with inadequate levels of RA during growth and development.^43,44 RA is also known to be involved in important checkpoints of inflammation and tolerance.⁴⁵ This background paired with the chronic inflammation observed during preeclampsia prompted us to investigate a possible association between RA and preeclampsia.⁷ The results presented here demonstrate significantly reduced levels of RA in preeclamptic decidua compared with normotensive decidua. Relative tissue concentrations of retinol and retinyl ester were not different in preeclamptic versus normotensive decidua, indicating that changes in RA were not caused by reduced vitamin A availability but by alterations in retinoid metabolism. And despite the GA difference in preeclampsia samples, comparison of early versus late preeclampsia showed no effect on RA levels.

Several reports have described the expression and modulation of RA pathway proteins, such as PPARs and RXR, in the placenta during healthy and preeclampsia pregnancies.^46,47 In 2017, Ozaki et al¹⁷ published a comprehensive report of altered regulation of RA pathway proteins throughout decidualization in human endometrial stromal cells. Here, we investigated whether sFLT1 overexpression in preeclampsia may be associated with suppressed decidual production of RA—a relationship that had not yet been characterized. Our results clearly establish that both NT-DSCs and PE-DSCs downregulate sFLT1 expression when treated with the RA.

Further, our findings demonstrate that RA acts as a regulator of sFLT1 independent of the well-described cAMP-mediated decidualization pathway. We have recently published data showing that, when treated with cAMP, PE-DSCs do not fully decidualize or downregulate sFLT1 compared with cAMP-treated NT-DSCs. Our results show that cAMP-mediated upregulation of the decidualization marker PRL is roughly 3× greater in NT-DSCs than in PE-DSCs, which corroborates results of our previous studies and those of Garrido-Gomez et al¹⁰ suggesting that PE-DSCs are deficient in their decidualization potential.¹¹ Because these results reflect changes late in pregnancy or after a history of preeclampsia, it is not known whether this is an inherent or an acquired defect.¹⁰

In contrast to the reduced inhibition of sFLT1 observed in cAMP-mediated decidualization, our findings in this study show that RA treatment leads to more robust downregulation of sFl1 in PE-DSCs than in NT-DSCs. PE-DSCs may be more responsive to RA-mediated sFLT1 modulation because, as we have determined, they come from an environment with lower levels of RA in vivo. There appear to be other compensatory mechanisms that suppress sFLT1 in preeclampsia. When studying decidualization response to RA treatment in DSCs, we found slightly lower levels of the decidualization marker PRL in RA-treated DSCs, whereas cAMP treatment upregulated PRL in both NT-DSCs and PE-DSCs. These findings indicate that regulation of sFLT1 by RA in DSCs occurs independently of decidualization and may involve a distinct pathway of action.

In this study, we also investigated potential pathways of RA-mediated sFLT1 modulation by screening for expression changes in transcription factors. One transcription factor, FOXO1, was found to be upregulated by cAMP but not by RA. FOXO1 belongs to the FOX (forkhead box) family of transcription factors and plays an essential role in regulating decidualization of endometrial stromal cells.⁴⁸ Both PRL and IGFBP1, common markers of decidualization, have FOXO1-binding sites in their promoter regions.⁴⁹ Its upregulation with cAMP but not RA treatment reaffirms that fact that RA fails to induce PRL induction and decidualization in DSCs.

Our analysis of the Flt promoter region revealed the presence of a TFAP2A-binding site (−618 to −603) adjacent to the RXR/RAR-binding region (−522). TFAP2A—a bifunctional, RA-inducible transcription factor—is a member of AP-2-alpha family of genes. In the placenta, TFAP2A regulates hormonal secretions involved in trophoblast differentiation by binding to RAR and acting on the RA pathway.⁵⁰ Previous reports also demonstrate TFAP2A overexpression in preeclampsia villous placentas.^51,52 In our study, despite distinct expression patterns of TFAP2A in NT-DSCs and PE-DSCs—specifically, overexpression in normotensive and no change in PE-DSCs when treated with RA—sFLT1 was inhibited in both cell populations, indicating that there are additional mechanisms responsible for sFLT1 downregulation.

A caveat of these findings is the general premise that conditions or events that predispose to preeclampsia arise early in pregnancy and precede the clinical manifestations of the disease.⁵³ Although our findings are consistent with the hypothesis that defects in decidual RA biosynthesis may be involved in the development of preeclampsia, our analysis of the term decidua does not allow us to distinguish the cause and effect relationship between the development of preeclampsia and RA production.

In conclusion, these data demonstrate for the first time that RA levels are lower in preeclamptic decidua and that RA downregulates sFLT1 expression in both NT-DSCs and PE-DSCs. Our characterization of PRL and FOXO1 expression shows that mechanisms of RA- and cAMP-mediated sFLT1 regulation are distinct and independent processes. These results suggest that multiple regulatory mechanisms of sFLT1 expression exist in stromal cells to coordinate the potent suppression of the antiangiogenic effect of sFLT1 at the MFI in healthy pregnancies. We speculate that, in the event of any defect in these mechanisms, insufficient downregulation of stromal cell sFLT1 production leads to higher-than-normal sFLT1 levels at the MFI. Aberrantly high levels are likely to lead to further dysfunction in sFLT1 regulation, potentially leading to cross talk with the adjacent placenta to drive the excess plasma sFLT1 seen in preeclampsia.

Perspectives

Maternal risk factors have been recognized for many years, yet only recently has experimental evidence of maternally derived mediators of preeclampsia been reported. We have used tissue samples from late pregnancy to identify fundamental differences among maternally derived tissues and cells of preeclampsia and normotensive pregnancies. Despite the challenges of separating cause and effect in cases of clinically manifested preeclampsia, our data indicate that defective decidualization, in part, explained by reduced RA concentrations in PE-DSC, appears to contribute to excessive sFLT1 at the MFI in these cases.

Novelty and Significance.

What Is New?

• Retinoic acid (RA) levels are lower in decidua from preeclamptic than decidua from normotensive control pregnancies.

• Treatment of decidual stromal cells (DSCs) with RA downregulates sFLT1 (soluble vascular endothelial growth factor receptor-1) expression.

• RA regulation of sFLT1 appears to be mediated through a suppressive RXR/RAR (RA nuclear receptor) element in FLT1 (vascular endothelial growth factor receptor-1) promoter at −522 bp.

• cAMP-mediated decidualization of DSCs also causes a reduction in sFLT1 production, but unlike RA, cAMP is associated with an increase in the production of decidualization markers prolactin and IGFBP1 (insulin-derived growth factor-binding protein 1).

• FOXO1 (forkhead box protein 01)—a key regulator of decidualization markers prolactin and IGFBP1—is not upregulated in DSCs exposed to RA, corroborating that RA-mediated downregulation of sFLT1 acts independently of decidualization.

• TFAP2A (transcription factor AP-2-alpha [activating enhancer-binding protein 2 alpha])—a bifunctional transcription regulator—is upregulated in control DSCs in response to RA and cAMP treatment.

What Is Relevant?

• The etiology of preeclampsia remains unknown. Although overexpression of sFLT1 is manifest in the circulation and associated with preeclampsia pathology, the precise cellular sources and exact causes of excessive sFLT1 are not known at present.

• Our observation that RA levels are reduced in preeclamptic decidua, coupled with the demonstrated ability of RA to downregulate in DSCs, supports the hypothesis that these phenomena are mechanistically linked. When considered in conjunction with evidence from us and others that decidualization in preeclampsia DSCs is defective, the present study supports a uterine origin of factors that predispose to preeclampsia.

Summary

Reduced decidual RA levels are insufficient to suppress sFLT1 expression in DSC, leading to accumulation of sFLT1 levels at the maternal-fetal interface and interference with angiogenesis, as observed in cases of preeclampsia.