Aberrant retinoic acid production in the decidua: Implications for pre-eclampsia

Abstract

Fine-tuning of the endometrium during the evanescent ‘window of implantation’ relies upon an array of diverse and redundant signaling molecules, particularly the ovarian steroids E2 and P4, but also growth factors, eicosanoids, and vitamins including the vitamin A compounds (retinoids). Pregnancy complications such as preeclampsia (PE) can result from aberrations in the production or function of these molecules that arise during this critical period of decidual development. Such aberrations may be reflected by incomplete decidualization, reduced spiral artery modification, and/or loss of immune tolerance to the developing fetus. Our understanding of the role of the active retinoid metabolite all-trans retinoic acid (RA) in maintaining immune balance in certain tissues, along with data describing its role in decidualization, present a compelling argument that aberrant RA signaling in the decidua can play a significant role in the etiology of PE. Recent findings that decidualization and expression of the anti-angiogenic gene product, ‘soluble fms-like tyrosine kinase-1’ (sFLT1) are negatively correlated and that sFLT1 expression is directly inhibited by RA, provide additional evidence of the critical role of this retinoid in regulating early vascular development in the decidua. This review provides insight into the production and function of RA in the decidua and how modifications in its metabolism and signaling might lead to certain pregnancy disorders such as PE.

Pre-eclampsia

Pre-eclampsia (PE) is a life threatening complication of gestation, involving 5–7% of all pregnancies and conferring considerable morbidity and mortality to the mother and infant.^1–3 Additionally, 30% of PE pregnancies result in growth-restricted babies, which is another contributor to neonatal morbidity.⁴ The prevalence of PE appears to have increased ~20% over the past decade and is 60% more likely to occur in African–American than white mothers. Treatments center on managing hypertension and the associated sequelae of the condition and in severe cases, delivery of the preterm infant is the only option to restore the mother’s health. Studies have shown that in addition to medical complications experienced during pregnancy, women with PE have a 7–8-fold greater risk of death and an increased risk of life-long cardiovascular diseases compared to counterparts with normal pregnancies.⁵ In addition to the consequences of prematurity, children and adolescent offspring of PE pregnancies also display increased cardiovascular health risks. The etiology of PE still remains elusive; some prominent theories include those implicating hypoxia resulting in poor placentation, an overexuberant maternal inflammatory response to pregnancy, and an anti-angiogenic state at the fetomaternal interface.^6,7 To this end, multiple lines of evidence support the idea that circulating and local anti-angiogenic factors (particularly soluble fms-like tyrosine kinase-1, sFLT1 and soluble endoglin, sENG) are involved in the genesis of the PE syndrome.⁷ The pathology of PE includes shallow penetration of fetal trophoblasts into the maternal endometrium (decidua) and fewer spiral arteries are remodeled to supply blood to the growing placenta. As a result, placentas in PE pregnancies are smaller, with less capacity for nutrient and gaseous exchange. A deeper understanding of the causes and predictors of PE may afford improved preventive strategies and inform future treatments, thereby leading to improved pregnancy and long-term health for women and infants.

Although the placenta’s role in PE has been extensively studied, contributions from the maternal side of the fetomaternal interface (decidua) have been a lesser focus of investigation. Two compelling lines of investigation regarding the maternal contribution to PE involve defects in decidualization and a heightened maternal immune response directed against the fetus, each of which can lead to inadequate development of blood vessels on both the maternal and fetal sides of the placenta. In recent years, we started exploring events at the maternal–fetal interface to understand the mechanisms of placental pathology leading to diseases like PE. Our studies uncovered certain original and unique events that are dys-regulated at the decidual layer in women who develop this disease. These include reduced levels in the decidua of PE women of certain steroid/retinoid compounds, including the active vitamin A metabolite, all-trans retinoic acid (RA).⁸ By utilizing decidual cells obtained from the fetomaternal interface of the placenta, these studies have also revealed various aspects of the maternal decidualization process that is abnormal in PE. We highlight these novel findings in this review.

Retinoic Acid Production in Human Decidua

It is known that estradiol (E2) and progesterone (P4) are the key steroid hormones involved in the processes of decidualization and regulation of maternal immune tolerance to the developing fetus; E2 and P4 alone can transform the fibroblastic uterine stroma to establish a vascularized, nutritive and immune-privileged secretory platform that permits nidation of the allogeneic blastocyst.⁹ However, it is now understood that the myriad regulatory cellular actions of E2 and P4 on the endometrium are mediated in part by elaboration of local growth factors,¹⁰ eicosanoids,¹¹ vitamins and retinoids¹² that are synthesized or critically metabolized in situ. In this regard, our data and those of others have shown that RA plays a fundamental role regulating optimal decidualization required for effective trophoblast implantation.^13–15 RA is regulated by the enzymatic activity and expression of retinol dehydrogenase (RDH) and retinal dehydrogenase (RALDH) enzymes.¹⁶ As demonstrated in a variety of tissues, RA concentrations can be regulated in vivo through the activity of RALDH enzymes that catalyze the oxidation of retinal to RA^17–20 (Fig. 1). RALDH expression is limited to select cell types and induced by various stimuli, including RA itself.^17,18,21 A primary source of RALDH-expressing cells in the small intestinal mucosa are dendritic cells (DC) that reside in the Peyer’s patches, lamina propria and mesenteric lymph nodes.^20–22 RA produced from these gut-associated DC is required for intestinal leukocyte tropism, differentiation to IgA secreting cells, and determination of Treg induction. The latter serves to inhibit proinflammatory Th17 differentiation.^19,22–25 Gut-associated DC from patients with inflammatory bowel diseases (e.g. Crohn’s disease, ulcerative colitis) express lower RALDH and have reduced RA production and action.²⁶ Based on our understanding of the role of RA in gastrointestinal and other tissues, we have posited that the hyper-inflammatory state occurring in the decidua of PE subjects reflects suppressed RA levels. This hypothesis is consistent with the finding that the concentration and function of Tregs are reduced in PE.^27,28 Figure 2a shows that RA in decidual tissue is reduced by almost 30% in pregnancies affected by PE. In addition to DCs, other cell types produce RA, including epithelial cells, stromal cells as well as monocytes, macrophages and T-cells. To determine the major cellular sources of RA in term decidua, we utilized an Aldefluor assay coupled with flow cytometry.²⁰ As shown in cytograms of Figure 2b, approximately 24% of macrophages/monocytes (CD14+ cells) showed positive RALDH activity compared with less than 4% of stromal cells (CD10+) or T lymphocytes (CD3+). In fact, the subpopulation with the highest RALDH activity shares CD14 and CD163 typical of anti-inflammatory macrophages (so-called ‘M2’ macrophages) and thus appear to be the major source of RA production in the decidua (Fig. 2c).³³ Interestingly, the frequency of decidual M2 cells is reduced in PE compared with normal pregnancies.^34,35 Whether the paucity of decidual M2 cells and reduced RA production in PE are responsible for impaired maternal immune tolerance in PE has yet to be determined.

Figure 1.

Schematic of retinoid metabolism in uterine M2 macrophages. After catabolism from retinol, retinoic acid (RA) can be (i) transported to the nucleus of the RA-producing cell where it binds to nuclear receptors and initiates gene transcription; (ii) transported to neighboring cells to initiate RA-mediated signaling and/or (iii) degraded. Where retinal dehydrogenase (RALDH) expression is restricted to limited cell types, it can confer cells the ability to synthesize RA.

Figure 2.

(a) Retinoic acid (RA) levels are reduced in preeclampsia (PE) decidua. (a) RA concentration (pmol/g protein) in whole decidual tissue was assessed by liquid chromatography with tandem mass spectrometry (LC-MS/MS)-based analyses in normal pregnant (NP) patients (n = 9) and PE patients (n = 10). Graph is a modified version of our published work.⁸ LC-MS/MS-based analyses were performed as described previously.^29–32. (b) retinal dehydrogenase (RALDH) activity (Aldefluor fluorescence) in subpopulations of decidual mononuclear cells. Dispersed mononuclear cells isolated from term placenta decidua were incubated for 45 min at 37°C in Aldefluor assay buffer containing activated Aldefluor in the presence (+) or absence (−) of the RALDH inhibitor diethylaminobenzaldehyde (DEAB) (Stemcell Technologies). Subsequently, the cells were stained with specific anti-bodies to determine their lineage (CD expression) and analyzed by multicolor flow cytometry. Following reactions by cellular RALDH, the amount of fluorescent product in the cell is proportional to the RALDH activity. DEAB-treated cells were used to evaluate background fluorescence to set negative control levels. CD3, T-lymphocytes; CD10, stromal cells; CD14, macrophages/monocytes.³³ (c) Correlation between RALDH activity (Aldefluor fluorescence) and RA levels in subpopulations of macrophages. RALDH was assessed in CD14 + CD163neg and CD14 + CD163+ decidual cells by the Aldefluor assay. The ability of the sorted populations to synthesize RA was assessed by using LC-MS/MS-based analysis following addition of 2-μM retinol for 6 h under serum-free conditions (red box).³³ LC-MS/MS-based analyses as described previously.^29–32 (b) and (c) are taken from our published work.³³

How RA Is Involved in the Formation of Healthy Decidua

The concept that defective decidualization could be a precursor of pregnancy complications such as PE has been postulated for several years^36,37 but only recently have data been produced to support the hypothesis.^38–40 We have shown that secretory endometrium in vivo²⁹ and decidualized endometrial stromal cells (ESC) in vitro,³⁰ are associated with increased local RA biosynthesis. Unfortunately, RA signaling is complex and mediated via two competing pathways involving multiple members of the nuclear receptor family of transcription factors: retinoic acid receptors (RAR), retinoic X receptors (RXR) and peroxisome proliferator-activated receptor (PPARβ/δ (Fig. 3a).^41–43 The RA-RAR and RA-PPARβ/δ pathways have opposing effects, alternately promoting cell cycle arrest and apoptosis versus cell survival and differentiation, respectively.^41–43 In vitro studies with ESC have shown that during decidualization, initial dominance of the cellular retinoic acid binding protein 2 (CRABP2)-RAR pathway is shifted to the fatty acid binding protein 5, (FABP5)-PPARβ/δ.⁴⁴ This shift in signaling pathways explains why the effects of RA on decidualization (inhibition or enhancement) may be dependent on the timing of RA production. To this end, some studies indicate that RA-RAR complexes inhibit ESC decidualization,^44,45 whereas others show the converse effect.¹⁵ Utilizing an in vitro culture model of differentiating human endometrial stromal cells (ESC) exposed to E2 (10 nM), progesterone (100 nM) and cAMP (0.5 mM) (cocktail containing estradiol, progesterone and cAMP (EPC)), our initial experiments indicated that exposure to RA throughout the entire culture period suppressed decidualization and prolactin (PRL) secretion, but studies have suggested that delaying the addition of RA to later time points during the decidualization process (when RA signaling shifts to the FABP5-PPARβ/δ pathway⁴⁴) results in an increase in PRL secretion. This interpretation is consistent with our observation that addition of the potent RAR antagonist, BMS493, to block RA signaling through the CRABP2-RAR pathway that is mediated by endogenous RA production significantly increases PRL secretion.⁴⁶ The data shown in the experiment of Fig. 3b support this hypothesis, as the addition of GW0742 (a pure PPARβ/δ agonist)^47,48 to the cell cultures doubled the PRL response over EPC alone. Similar quantitative effects of BMS493 and GW0742 on the secretion of another decidual marker protein, IGFBP1 were also noted (data not shown). Together, these findings are consistent with the hypothesis that RA synthesis during discrete stages of the decidualization process plays a modulating role in the optimal differentiation of ESC to form competent decidua for trophoblast attachment and implantation, and placental vascularization.

Figure 3.

(a) Schematic showing the dual transcriptional pathways of retinoic acid (RA) activity. CRABP2 and FABP5 target RA to retinoic acid receptors (RAR) and peroxisome proliferator-activated receptor (PPARβ/δ, respectively. In endometrial stromal cells (ESC) that express a high CRABP2/FABP5 ratio, RA is channeled to RAR, resulting in growth arrest and apoptosis. Conversely, in the presence of a low CRABP2/FABP5 expression ratio, RA is targeted to PPARβ/δ which promotes cellular differentiation. Adapted from the published work of Pavone et al.¹³ (b) Modulation of ESC decidualization by targeting the RAR and PPARβ/δ pathways. ESC were treated with the EPC decidualization cocktail in absence or present of RA (0.1 μM), BMS493 (BMS, 1 μM) to block RAR, or GW0742 (GW) to activate PPARβ/δ for 8 days.

In previous studies, we determined that gap junction intercellular communications (GJIC) are critically involved in ESC decidualization and that RA plays a key role in this process by regulating Cx43 dephosphorylation, resulting in increased GJIC function.^46,49–51 Although the role of Cx43 in pregnancy has been most extensively investigated in the setting of uterine contractility, our group established that ESC expression of Cx43 is critical for decidualization, and placental angiogenesis.^46,49,51,52 In the mouse, where global Cx43 knock-out results in embryonic cardiac death, the Cre-Lox approach was used to selectively delete Cx43 in PR-expressing uterine cells (Cx43^d/d mice).⁴⁹ The findings substantiated an obligatory role for Cx43 and functional GJIC in normal implantation. Aberrant Cx43 expression has been documented in a variety of reproductive disorders associated with defective decidualization including endometriosis, recurrent miscarriage, PCOS^51,53–56 and provide a rationale for Cx43 dysregulation in the context of PE. RA regulation of Cx43 in endometrial cells was first addressed by Tanmahasamut and Sidell,⁵⁷ who showed an RA dose-responsive increase in Cx43 mRNA and protein. Subsequent studies demonstrated that RA reduced the (inactive) phosphorylated species (P1 + P2) of Cx43 and increased the (active) dephosphorylated (P0) species (Fig. 4a). Increased GJIC activity by RA was proven with dye diffusion experiments.⁴⁶ That this effect of RA on Cx43 is mediated through the FABP5-PPARβ/δ pathway was suggested by the finding that blockade of the CRABP2-RAR pathway with the RAR antagonist BMS493 resulted in more dephosphorylated Cx43, and stimulated GJIC and biomarkers of decidualization (PRL and VEGF secretion).⁴⁶ We now have evidence that the PPARβ/δ agonist GW0742 also potently upregulates nonphosphorylated (active) Cx43 compared to control, an effect that is indistinguishable from RA (Fig. 4b). These data suggest that local RA biosynthesis and action through the FABP5-PPARβ/δ pathway increases dephosphorylated Cx43 to activate gap junction function in ESC and thereby promote decidualization at the fetomaternal interface. By contrast, in settings where local RA biosynthesis is impaired, as in cases of PE (Fig. 2a), it would be predicted that GJIC are impeded and decidual differentiation is retarded, resulting in reduced trophoblast invasion and poor decidual vascular remodeling. In addition to this consequence of reduced RA biosynthesis on competent decidua formation, faulty RA generation and signaling may contribute to alterations of immune pathways through its direct regulation of Tregs,⁵⁸ theoretically explaining several of the immunopathological features characteristic of the PE syndrome. The translational significance of this knowledge depends upon an understanding of how PPARβ/δ impacts placental development. PPARs are expressed in human trophoblasts and ESC⁵⁹ and have been implicated in placental vascular development relevant to PE.⁶⁰ Whereas embryonic PPARβ/δ is vital for placentation, maternal PPARβ/δ is critical for implantation.⁶¹ Circulating fatty acid activators of PPARs are reduced in PE pregnancy⁶² and some of these (e.g., linoleic acid) are abnormally sequestered onto albumin in preeclamptic plasma.⁶³ Consistent with this hypothesis is the observation that among women undergoing IVF, those with high levels of linoleic acid had significantly improved implantation and pregnancy outcomes.⁶⁴ These studies support the concept that PPARs maintain normal decidual function and might avert PE. Synthetic ligands of PPARβ/δ are currently under pharmaceutical development. Among their salutary activities in preclinical models are increased fatty acid oxidation and reduction in triglycerides; both of these metabolic alterations are known to be associated with PE.⁶⁵ Obviously, the safety profiles of such new therapeutics will need to be assiduously vetted before clinical trials in human pregnancy could be considered.

Figure 4.

(a) Effect of retinoic acid (RA) on Cx43 phosphorylation. Total cellular protein was isolated from endometrial stromal cells (ESC) and assayed for Cx43 expression by Western blotting. Quantitation was obtained by densitometry. Columns represent the mean (±SEM) of at least three determinations for each treatment condition. ESC treated with increasing concentrations of RA for 24 h showing progressive changes in the band intensity of the nonphosphorylated (P0) and phosphorylated (P1 and P2) species of Cx43 (inset). Columns in the lower panel quantify changes in expression of P0, P1 and P2 in ESC treated with 10 μM RA relative to untreated control cells (dashes). All Cx43 phospho-species treated with RA were significantly different from untreated controls (P < 0.05). From our published work and with permission from Journal of Cellular Physiology (2013).⁴⁶ (b) Increased dephosphorylation of Cx43 by RA and GW0742 (GW). ESC were treated with the PPARβ/δ agonist GW0742 (10 μM) or RA (10 μM) for the times indicated or solvent control (C) for 24 h. Western blot analysis was performed with an antibody specific for nonphosphorylated forms of Cx43 at Ser368.

How Defective Decidual Development Inhibits Blood Vessel Formation

Numerous studies have shown that an excess of anti-angiogenic sFLT1 leads to PE pathology. A seminal finding by Maynard et al., in 2003 showed that experimental overexpression of sFLT1 in pregnant rats produced the classic phenotype of PE, namely increased blood pressure, proteinuria and renal dysfunction.⁶⁶ Following this report, sFLT1 levels have been analyzed in multiple groups of pregnant women across the globe and its association with the disease has been validated and accepted. The hypothesis is that excess circulating sFLT1 contributes to maternal endothelial dysfunction by binding VEGF and PLGF and preventing their functions. Recent studies have targeted sFLT1 as a treatment for PE, using plasmapheresis to reduce their circulating levels.⁶⁷ Currently, there is very limited information regarding the regulation and expression of sFLT1 within the placenta or endometrium.

Our recent publication showed that in vitro decidualization of ESC is associated with a dramatic (>90%) inhibition of sFLT1 protein production and secretion.⁶⁸ Similar effects were observed in cultured decidual stromal cells (DSC) from term placenta that are induced to re-decidualize in vitro. Since decidualization is a prerequisite for successful implantation, this observation suggests that repression of maternal sFLT1 at the maternal-fetal interface may be necessary for optimal implantation. This thesis is intuitive given the critical need for brisk angiogenesis to establish maternal blood flow to the developing placenta by the 10–12th week of gestation. In addition to the effects of low RA levels on sFLT1 production via effects on decidualization, we have determined that RA also has a direct inhibitory effect on sFLT1 transcription and secretion from ESC and DSC⁸ (Fig. 5). Thus, reduced RA levels by these multiple mechanisms can promote overexpression of sFLT1, with its anti-angiogenic mechanisms that further encourage shallow implantation and reduced spiral arteriole modification, leading to serious gestational consequences.

Figure 5.

Retinoic acid (RA) and cAMP-mediated decidualization of decidual stromal cell (DSC) reduce soluble fms-like tyrosine kinase-1 (sFLT1) protein and mRNA levels. (a) DSC show reduced levels of sFLT1 protein as quantified by enzyme-linked immunosorbent assay in culture medium on day 8 of either RA or cAMP treatment. (b) Relative changes in sFLT1 mRNA levels in cells treated for 8 days with RA or cAMP, n = 5, *P < 0.05 by anova. Similar effects of RA and cAMP on sFLT1 levels were obtained with endometrial stromal cells.⁸

The mechanisms involved in the regulation of sFLT1 have not been addressed in placenta, especially during decidualization. Earlier studies established cAMP response elements (CRE) and ETS1 sites in the FLT gene promoter that are essential for active FLT gene transcription in endothelial cells.^69–72 Studies also suggest the presence of a negative regulatory element in the first exon of the FLT gene.⁷² FLT transcriptional regulation during pregnancy and decidualization is not known. Our comprehensive analysis of the FLT gene promoter sequence using MotifMap^73,74 showed an RXR/RAR binding element at −522 and an activator protein 2 (TFAP2A) at −603. Both of these motifs are regulated by RA (Fig. 6). It is presently unknown whether inhibitory action by RA vs. cAMP-induced decidualization is governed by the same or different regulatory elements in the FLT promoter. By cDNA array analysis, we have identified a number of transcription factors and accessory proteins expressed in ESC (both decidualized and nondecidualized) known to be involved in RA- and cAMP-mediated gene regulation and, correspondingly, have cognate response elements located in the immediate 5’-promoter of the FLT gene (Fig. 6). These factors include: CRABP1 and 2; RARα, β and γ; FABP5, PPARβ/δ, RXRα, β and γ; HIF1A; Ets1 and 2; CREB1; CREBBP; NCoR; TFAP2A and SMRT and are currently being characterized.

Figure 6.

Fms-like tyrosine kinase (FLT) gene promoter. Using deletional constructs of the human FLT1 gene promoter, a region of about 1.0 Kb has been found to contain maximum transcription activity.⁷⁵ Sequence analysis show putative transcription factor binding sites for CREB/ATF, ETS, EGR and ERG. A hypoxia response element (HRE) has also been identified proximal to the core promoter. The FLT1 gene promoter has an authentic TATA sequence and CREB/ATF and the adjacent ETS binding region control the basal promoter activity. Our search using Motifmap (Region marked in Red), shows the presence of putative TFAP2A and RAR/RXR binding sites. (Source: http://motifmap.ics.uci.edu/).

Summary of our Model for RA Action in the Decidua

Our studies have determined the importance of RA in promoting healthy decidua through suppression of sFLT1 and regulation of decidualization via Cx43. We propose the following steps by which RA is involved in early placentation and fetal development and how abnormal RA signaling may promote the development of PE (Fig. 7): The cells that are primarily responsible for the production of RA are M2 macrophages – step (1) in schematic. These cells secrete RA (2) into the surrounding uterine tissue that includes ESC (3), the precursors of decidual cells, as well as a variety of immune cells, including T-lymphocytes (4). By stimulating Treg differentiation (5) and activating the PPARβ/δ molecular pathway (6), RA may control immune tolerance (7) and augment the decidualization response of ESC in combination with ovarian hormones (8). RA production and signaling appear to be defective in PE, as indicated by reduced RA levels in decidua dissected from PE placentas compared with normal pregnancy tissues. This defect may be at least partly responsible for maternal hyper-inflammatory responses and suboptimal decidualization of ESC, the latter resulting in the continued production and secretion of the anti-angiogenic factor sFLT1 (9). High concentrations of sFLT1 in the vicinity of decidualized stromal cells (DSC) during embryo attachment block the action of VEGF and PLGF (10), preventing new blood vessel formation and modification, resulting in an inadequate vascular network necessary for healthy placental development and growth. This scenario gives rise to smaller placentas, with less capacity for nutrient and gaseous exchange necessary to support a normal pregnancy. Hypoxic conditions in the placenta result from these processes, which adversely affect the development of spiral arteries that further enhance the severity of the PE condition. Maternal compensation gives rise to the symptoms of PE (high blood pressure, protein in the urine, liver and kidney problems). This scenario suggests the possibility that PE might be reversed or prevented by treatment with potentially ‘safe’ PPARβ/δ activators to counteract inadequate RA production and/or activity.

Figure 7.

Schematic of our proposed steps by which retinoic acid (RA) is involved in early placentation and fetal development and how abnormal RA signaling may promote the development of preeclampsia. See ‘Summary’ text for a detailed description of the steps.