Activation of the Receptor for Advanced Glycation End Products System in Women with Severe Preeclampsia

sRAGE is a physiological constituent of maternal serum, amniotic fluid, and fetal blood which undergoes significant modulation in severe preeclampsia.

Abstract

Context:

Activation of the receptor for advanced glycation end products (RAGE) mediates cellular injury. Soluble forms of RAGE [soluble RAGE (sRAGE), endogenous secretory (esRAGE)] bind RAGE ligands, thereby preventing downstream signaling and damage.

Objectives:

The objective of the study was to characterize the changes in maternal serum, amniotic fluid, and cord blood soluble receptor for advanced glycation end products (sRAGE) during physiological gestation and to provide insight into mechanisms responsible for RAGE activation in preeclampsia.

Design and Settings:

This was a cross-sectional study at a tertiary university hospital.

Patients:

We studied 135 women in the following groups: nonpregnant controls (n = 16), healthy pregnant controls (n = 68), pregnant women with chronic hypertension (n = 13), or pregnant women with severe preeclampsia (sPE; n = 38).

Interventions and Main Outcome Measures:

sRAGE and esRAGE levels were evaluated in vivo by ELISA in maternal serum, amniotic fluid, and cord blood and in vitro after stimulation of the amniochorion and placental explants with lipopolysaccharide or xanthine/xanthine oxidase. Placenta and amniochorion were immunostained for RAGE. Real-time quantitative PCR measured RAGE mRNA.

Results:

Pregnant women had significantly decreased serum sRAGE compared with nonpregnant subjects (P < 0.001). sPE women had higher serum and amniotic fluid sRAGE and esRAGE relative to those expected for gestational age (P < 0.001). Cord blood sRAGE remained unaffected by sPE. RAGE immunoreactivity and mRNA expression appeared elevated in the amniochorion of sPE women. Xanthine/xanthine oxidase (but not lipopolysaccharide) significantly up-regulated the release of sRAGE (P < 0.001) in the amniochorion explant system.

Conclusions:

Fetal membranes are a rich source of sRAGE. Elevated maternal serum and amniotic fluid sRAGE and esRAGE, paralleled by increased RAGE expression in the amniochorion, suggest activation of this system in sPE.

Preeclampsia is an idiopathic multisystem disorder that affects 6–8% of all pregnancies (1). It is a major cause of maternal and fetal morbidity and mortality and is classically characterized by hypertension and proteinuria manifesting after 20 wk gestational age (2, 3). Several pathophysiological processes have been linked to preeclampsia, including endothelial dysfunction, angiogenic and inflammatory factor imbalance, oxidative stress, protein misfolding, and recently the activation of the receptor for advanced glycation end products (RAGE) system (4–9).

RAGE is a multiligand cell surface receptor present on numerous cells such as macrophages, endothelial cells, neurons, and smooth muscle cells as well as amnion and choriodecidual tissues (10–12). RAGE was initially identified by its ability to engage advanced glycation end products (AGEs), which are generated when reducing sugars react nonenzymatically with proteins or lipids (11, 13). It has become increasingly clear that a wide range of exogenous and endogenous ligands bind to RAGE and that this system plays key pathogenic roles in chronic inflammatory conditions including diabetes, rheumatoid arthritis, arteriosclerosis, and Alzheimer's disease (14–16). Successful binding of a ligand to the extracellular domain of RAGE activates cell signaling pathways such as nuclear factor-κβ and MAPKs. This leads to the generation of proinflammatory cytokines, prostaglandins, matrix metalloproteases, and free radicals (FRs)/reactive oxygen species (ROS) (17–19). However, studies conducted in vitro and in vivo provide evidence that RAGE signaling can be antagonized by soluble RAGE (sRAGE), which is a molecularly pleiotropic endogenous RAGE antagonist (10). AGE receptor (AGER), the gene encoding RAGE, is highly polymorphic (20). Phenotypes of different AGER allele carriers are associated with reduced levels of sRAGE (20). sRAGE forms are generated by either alternative splicing of RAGE mRNA [i.e. endogenous secretory RAGE (esRAGE)] or cleavage of the extracellular domain of RAGE by the sheddase a disintegrin and metalloproteinase domain-10 (ADAM10) (21, 22).

Our group was the first to show that the levels of sRAGE in human amniotic fluid are gestational age regulated and that RAGE is expressed in both amniochorion and choriodecidua (12). Through proteomic mapping of amniotic fluid, we established that S100A12 (a putative RAGE ligand) was up-regulated by inflammation (6, 23). Recently we provided evidence that fetuses born in the setting of severe intraamniotic inflammation/infection have lower sRAGE cord blood levels than those unaffected by inflammation (24). We further demonstrated that RAGE activation is yet another mechanism through which inflammation induces cellular damage to vital fetal organs (24).

Research on the role of RAGE signaling in preeclampsia has been sparse (8, 9). Previous reports suggest that preeclampsia is characterized by increased AGEs in maternal serum and up-regulation of RAGE in the myometrium, omental vessels, and placenta (8, 9). Several studies sought to determine whether maternal serum concentrations of sRAGE are differentially regulated in preeclampsia (25, 26). However, these preliminary studies included few preeclamptic women. The current study tested systematically the hypothesis that preeclampsia is characterized by a heightened state of RAGE system activation (27, 28). We also provided novel insight into the mechanisms responsible for the release of sRAGE in preeclampsia.

Participants and Methods

Participants and study design

We conducted a case-control study using serum samples collected from 135 women enrolled at Yale-New Haven Hospital from May 2004 to January 2008. A flow chart of women and biological samples used in this study is provided in the Supplemental Data, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org. Patients were stratified into the following groups: 1) healthy nonpregnant women of reproductive age (NP-CRL; n = 16); 2) healthy women pregnant with singletons [P-CRL; gestational age median (interquartile range): 30 (24–38) wk, n = 68]; 3) severe preeclampsia [sPE; gestational age: 30 (27–32) wk, n = 38]; and 4) chronic hypertension [crHTN; gestational age: 32 (29–36) wk, n = 13].

After enrollment, P-CRL women remained normotensive and nonproteinuric throughout pregnancy and gave birth at term (≥37 wk) to neonates with normal birth weights. Severe preeclampsia, crHTN, and proteinuria were defined based on American College of Obstetricians and Gynecologists criteria (Supplemental Data) (3). Information regarding the characteristics of crHTN women, our clinical care protocols, and exclusion criteria for the study population is provided in the Supplemental Data.

Biological samples

Maternal serum

Serum samples from sPE and crHTN women were collected at the time of inpatient admission or evaluation and before steroid treatment or initialization of magnesium sulfate seizure prophylaxis, as appropriate.

Amniotic fluid

Amniotic fluid was available from 63% (24 of 38) of women with sPE. Amniotic fluid was collected at the time of cesarean delivery before rupture of the membranes and in the absence of blood contamination. As control, we used amniotic fluid from our bank of biological samples. Amniotic fluid of sPE women was matched for gestational age at delivery (±1 wk difference) with samples obtained from 24 consecutive subjects enrolled during the same study period [gestational age sPE: 30 (26–32) wk vs. preterm group: 30 (25–32) wk, P = 0.975]. Women who provided the control amniotic fluid had a clinically indicated trans-abdominal amniocentesis to rule out infection, had no infection/inflammation, and delivered a healthy baby at term. The control amniotic fluid samples were retrieved before steroids or magnesium sulfate tocolysis.

Cord blood

Cord blood was available for analysis in 71% (27 of 38) of the sPE women. As controls, we used cord blood from our bank of biological samples. Cord blood of fetuses delivered by sPE women was matched for gestational age at delivery (±1 wk difference) with samples obtained from 27 consecutive fetuses delivered by women who had an idiopathic preterm birth defined as absence of intraamniotic infection, histological chorioamnionitis or funisitis, abruption, and/or polyhydramnios [gestational age sPE: 30 (27–32) wk vs. idiopathic preterm birth: 31 (30–32) wk, P = 0.197]. Details of the assessment of the fetal acid-base status at birth, preparation and storage of the biological samples are presented in the Supplemental Data.

sRAGE, esRAGE, IL-8, and total protein assays

Total sRAGE levels were measured in maternal blood, amniotic fluid, cord blood, and explant medium using an immunoassay that measures all sRAGE forms (R&D Systems, Minneapolis, MN). esRAGE was measured in maternal blood and amniotic fluid by a specific immunoassay (B-Bridge International, Cupertino, CA). Additionally, explant media were immunoassayed for the proinflammatory chemokine IL-8 (R&D Systems). Each sample was tested in duplicate according to the manufacturer's protocol. The intra- and interassay coefficients of variation values were less than 10% for all analytes. The minimal detectable sRAGE, esRAGE, and IL-8 reported by the manufacturer were 4.1, 10.0, and 1.5–7.5 pg/ml, respectively. Total protein levels in the explant media were measured using a bicinchoninic acid/cupric sulfate reagent (BCA kit; Pierce, Rockford, IL).

RAGE immunohistochemistry

Sections of amniochorion and placental villous tissue from sPE women (n = 10) were stained for RAGE using a goat antihuman polyclonal antibody (R&D Systems). Amniochorion and placental tissues from a subgroup of idiopathic preterm deliveries matched for gestational age at birth with sPE cases were used as controls (n = 10). Membranes exposed in vitro to oxidative stress mimics (see below) were also stained for RAGE. Staining intensity was evaluated using a histological scoring algorithm (12). Additional details are provided in the Supplemental Data.

Quantitative real-time RT-PCR

Immediately after delivery, the tissues (placenta and amniochorion) of sPE (n = 10) and idiopathic preterm birth cases (n = 10) were frozen in liquid nitrogen and kept at −80 C for mRNA studies. Two probes (Applied Biosystems, Carlsbad, CA) were used for RAGE analysis: transmembranar full-length RAGE receptor (RAGE1, Hs00153957_m1) and RAGE2 (esRAGE, Hs00542590_m1). Technical details are provided in the Supplemental Data.

Placental explant experiments

Placentas and amniochorion were obtained from healthy women (n = 4) undergoing elective cesarean delivery in the absence of labor (38–40 wk gestational age). Tissues were incubated in the presence of 1 μg/ml lipopolysaccharide (LPS) to mimic inflammation. Oxidative stress was generated by the reaction of 2 mmol/liter xanthine (X) with 20 mU/ml xanthine-oxidase (XO), as previously described (29). Explant medium was collected at 1, 3, 6, 18, and 24 h of incubation for measurements of sRAGE and IL-8 by ELISA. Additional details are provided in the Supplemental Data.

Statistical analysis

The data were tested for normality using the Kolmogorov-Smirnov test and reported as either mean ± sem or as median and interquartile range, as appropriate. A P < 0.05 value was considered statistically significant. Other statistical methods are provided in the Supplemental Data.

Results

Clinical characteristics of the women

The demographic and outcome characteristics of the cases enrolled in this study are provided in Table 1. Women with crHTN were significantly older and of higher gravidity and parity compared with all other groups. Among the pregnant groups, there were no significant differences in gestational age at the time of serum collection. As expected, women with sPE had higher blood pressure values and were significantly proteinuric both by urine dipstick and 24-h urine protein excretion. sPE women also had a higher frequency of neurological manifestations, HELLP (hemolysis, elevated liver enzymes, and low platelet count) syndrome and intrauterine growth restriction. All sPE women had an indicated premature delivery, which in 92% of cases occurred at less than 34 wk, whereas the controls delivered at a median gestational age of 39 (39–40) wk.

Table 1.

Demographic, clinical, laboratory, and outcome characteristics of the patients

Variable	NP-CRL (n = 16)	P-CRL (n = 68)	sPE (n = 38)	crHTN (n = 13)	P value
Maternal and laboratory characteristics
Age (yr)a	32 (30–37)	29 (22–31)	26 (24–35)	36 (34–38)	<0.001
Graviditya	1 (0–3)	2 (1–3)	1 (1–2)	4 (4–6)	<0.001
Paritya	0 (0–1)	1 (0–1)	0 (0–1)	2 (1–2)	0.009
Nulliparityb	9 (56)	32 (47)	23 (60)	2 (11)	0.038
Gestational age (wk)a		30 (22–39)	30 (27–32)	32 (29–36)	0.386
Systolic blood pressure (mm Hg)a		115 (100–125)	169 (160–182)	152 (143–161)	<0.001
Diastolic blood pressure (mm Hg)a		61 (60–72)	102 (92–110)	91 (90–100)	<0.001
Urinary dipsticka		0 (0–0)	3 (2–4)	0 (0–0)	<0.001
24-h protein excretion (g)c			2.5 (1.2–4.2)	0.2 (0.1–0.3)	<0.001
Neurological manifestationsb		0 (0)	19 (50)	3 (23)	<0.001
IUGRb		0 (0)	10 (26)	0 (0)	<0.001
HELLPb		0 (0)	10 (26)	0 (0)	<0.001
Pregnancy outcome characteristics
Gestational age at delivery (wk)a		39 (39–40)	30 (27–32)	37 (35–38)	<0.001
Delivery < 34 wk)b		0 (0)	35 (92)	2 (15)	<0.001
Indicated deliveryb		0 (0)	38 (100)	3 (23)	<0.001

^aData presented as median (interquartile range) and analyzed by Kruskal-Wallis one-way ANOVA.

^bData presented as n (percent) and analyzed by χ² tests.

^cData presented as median (interquartile range) and analyzed by Mann-Whitney rank sum test.

Pregnancy and gestational age regulation of maternal serum sRAGE levels

This analysis was restricted to samples retrieved from NP-CRL and P-CRLs. Overall, P-CRL women had lower sRAGE serum levels compared with NP-CRLs (Fig. 1A, one-way ANOVA, P < 0.001). When pregnancy data were analyzed based on gestational age intervals, we found that the maternal serum sRAGE levels were not decreased early in pregnancy (NP-CRL vs. P-CRL < 20^0/7 wk, P = 0.116) but became significantly lower at midgestation (P-CRL < 20^0/7 vs. P-CRL 20^0/7 to 33^6/7 wk, P = 0.005). sRAGE remained low up to term (P-CRL 20^0/7 to 33^6/7 vs. P-CRL ≥ 37^0/7 wk, P = 0.125).

Fig. 1.

Serum total sRAGE levels in nonpregnant, healthy pregnant, and preeclamptic women across gestation. Compared with NP-CRL (n = 16), the levels of sRAGE were significantly decreased in healthy P-CRL women (n = 68) (A). Compared with NP-CRL, maternal serum sRAGE levels were significantly lower at midgestation (NP-CRL: 710 ± 57 vs. P-CRL 20^0/7 to 33^6/7 wk: 414 ± 29 pg/ml, n = 33, P < 0.001) but not early gestation (NP-CRL vs. P-CRL < 20^0/7 wk: 603 ± 57 pg/ml, n = 16). sRAGE levels remained low thereafter (NP-CRL vs. P-CRL ≥ 37^0/7 wk, 499 ± 36 pg/ml, n = 19, P < 0.001; P-CRL 20^0/7-33^6/7 vs. P-CRL ≥ 37^0/7 wk, P = 0.125]. Maternal serum sRAGE levels among P-CRL (n = 33), sPE (n = 38), and crHTN (n = 13) groups at midgestation are shown (B). sPE women had significantly elevated sRAGE serum concentrations compared with P-CRL (P-CRL > 20^0/7 to 33^6/7 wk vs. sPE: 1111 ± 159 pg/ml, P < 0.001) and crHTN groups (sPE vs. crHTN: 505 ± 524 pg/ml, P = 0.009). sRAGE levels were consistently higher in sPE cases compared with controls (CRLs) for each gestational age interval (two way ANOVA, P < 0.001) (C). Data are presented as mean ± se.

Maternal circulating levels of total sRAGE in sPE and crHTN

We compared the maternal serum sRAGE levels among midgestation P-CRLs (n = 33), sPE (n = 38), and women with crHTN (n = 13). We found that sPE women had significantly elevated sRAGE serum concentrations compared with P-CRL (P < 0.001) and crHTN groups (P = 0.009) (Fig. 1B), independent of gestational age at evaluation. When the midpregnancy interval was broken down into additional gestational age intervals (Fig. 1C), we found that sRAGE levels were consistently higher in sPE cases compared with CRLs (two way ANOVA, P < 0.001 for sPE, P = 0.317 for pregnancy intervals with no interaction between the two variables, P = 0.198).

Maternal circulating levels of esRAGE in sPE and crHTN

Women with preterm sPE had significantly higher maternal serum esRAGE compared with both the P-CRL and crHTN groups (P-CRL: 96 ± 7 vs. crHTN: 108 ± 33 vs. sPE: 337 ± 57 pg/ml, P < 0.001), even after correction for gestational age. In P-CRLs, the contribution of esRAGE to total sRAGE was 24 ± 1%, whereas in sPE this increased significantly to 29 ± 2% (P = 0.012).

Amniotic fluid and cord blood total sRAGE and esRAGE concentrations in sPE and controls

Next, we evaluated whether preeclampsia is associated with changes in amniotic fluid concentration of sRAGE and esRAGE. Compared with their respective controls, we found that sPE women had significantly elevated amniotic fluid total sRAGE (P = 0.011) (Fig. 2A). The amniotic fluid esRAGE was also significantly increased in sPE patients (control: 1248 ± 149 vs. sPE: 1959 ± 238 pg/ml, P = 0.015). In controls, the contribution of esRAGE to the total amniotic fluid sRAGE was 17 ± 1%. This proportion remained unaffected by sPE (16 ± 1%, P = 0.375). In cord blood, sPE per se did not induce notable differences in total sRAGE (P = 0.704) (Fig. 2B).

Fig. 2.

Amniotic fluid and cord blood total sRAGE concentrations in controls (CRLs) and sPE women at midgestation (20^0/7 to 33^6/7 wk). ELISA for total sRAGE showed that compared with CRL cases, sPE women had significantly elevated amniotic fluid (CRL: 7,762 ± 899, n = 24 vs. sPE: 12,224 ± 141 pg/ml, n = 24, t test, P = 0.011) (A) but not cord blood (CRL: 2,575 ± 238, n = 27 vs. sPE: 2,449 ± 229 pg/ml, n = 27, P = 0.704) sRAGE concentrations (B). The highest concentration of sRAGE was measured in amniotic fluid followed by cord blood and maternal serum compartments in both CRL (C) and sPE (D) cases. Data are presented as mean ± se.

Because the activity and response of the RAGE system may vary based on the compartmentalization process characteristic of human pregnancy, we compared the sRAGE concentrations among maternal, amniotic fluid, and cord blood compartments. Among controls, the amniotic fluid sRAGE concentration was about 3-fold higher than in cord blood, which in turn was about 6-fold higher than the maternal blood level (Fig. 2C). Similar relative differences were identified in sPE (Fig. 2D).

Relationships between maternal serum, amniotic fluid, and cord blood sRAGE levels and clinical and biochemical indicators of preeclampsia severity

This analysis was limited to women with hypertensive disorders (sPE and crHTN) who had a clinical and laboratory work-up for preeclampsia at midgestation. Using multivariate analysis, we determined that maternal circulating sRAGE levels could be modeled by a combination of gestational age (R = −0.216, P = 0.024), maternal hematocrit (R = −0.199, P = 0.011), and uric acid concentration (R = 0.375, P = 0.001), which emerged as the strongest direct correlate (multivariate F-ratio: 6.456, P = 0.001). Variables excluded from the model were maternal age, parity, race, blood pressure, proteinuria, intrauterine growth restriction, or HELLP syndrome. When the analysis was confined only to the sPE group, the relationship between sRAGE and uric acid remained significant (R = 0.337, P = 0.038), independent of gestational age.

The amniotic fluid sRAGE level was modeled by the combination of a diagnosis of sPE (R = 0.365, P = 0.002) and gestational age at evaluation (R = 0.621, P < 0.001). However, gestational age was the strongest covariate in multivariate regression (F-ratio: 23.29, R = 0.713, P = 0.001). Within the sPE group, there was no correlation between maternal and cord blood sRAGE levels (R = 0.122, P = 0.305). Cord blood sRAGE did not correlate with gestational age at delivery (R = 0.154, P = 0.267).

Amniochorion and placental expression of RAGE isoforms in sPE

Next, we quantified RAGE1 (full length transmembrane receptor) and RAGE2 (encoding the alternative spliced soluble isoform esRAGE) mRNA in sPE and gestational age matched controls. The real-time RT-PCR experiments demonstrated that: 1) RAGE1 and RAGE2 transcripts were expressed in both amniochorion and placental tissues (Fig. 3A); 2) relative to housekeeping gene transcripts, the expression of RAGE1 and RAGE2 was similar in placenta and amniochorion; 3) compared with RAGE1, the RAGE2 transcript was significantly less abundant in both placenta and fetal membranes; and 5) women with sPE have a significant elevation in RAGE2 expression in amniochorion (P = 0.032) but not placental villous tissue (Fig. 3B, P = 0.975).

Fig. 3.

Expression of RAGE1 and RAGE2 (esRAGE) transcripts in fetal membranes and placental villous tissue in pregnancies complicated by sPE. RAGE1 (full length RAGE receptor) and RAGE2 (splice variant esRAGE) mRNA expression in fetal membranes and placental tissue of controls (CRLs) are shown. Relative quantitation (RQ) change in cycle threshold value (ΔCT) values are reported relative to expression of the housekeeping genes β-2 microglobulin and ribosomal protein L30 for each tissue (A). Estimate of relative RAGE1 and RAGE2 mRNA abundance (ΔΔCT) of women with sPE compared with CRLs. ΔΔCT RQ values were reported relative to a reference RNA pool of the same tissue (B). Data are presented as mean ± se.

Immunostaining of RAGE in human amniochorion and placenta

Overall, RAGE immunostaining was more conspicuous in fetal membranes compared with placental sections (Fig. 4). Within the fetal membranes of control samples, the amnion (Fig. 4A) stained less RAGE positive than the choriodecidua (Fig. 4B). In choriodecidua, the RAGE signal appeared localized both to extravillous cytotrophoblasts and decidual cells. Within the placental villi, both syncytiotrophoblasts and cytotrophoblasts displayed RAGE immunostaining, albeit stronger in the latter (Fig. 4C). In sPE, the amnion stained intensely RAGE positive (Fig. 4D). Choriodecidual (Fig. 4E) and placental tissue (Fig. 4F) RAGE immunostaining appeared to be less impacted by preeclampsia. Our histological scoring analysis (Fig. 4I) indicated that in control samples RAGE immunostaining intensity was strongest in choriodecidua compared with amnion epithelium or placental villous trophoblast (one way ANOVA, P = 0.007). Moreover, sPE was characterized by an up-regulation in RAGE immunostaining in the amnion epithelium (control vs. sPE, P = 0.006) but not choriodecidua (control vs. sPE, P = 0.486) or villous tissue (control vs. sPE, P = 0.051).

Fig. 4.

Representative photomicrographs of RAGE immunostaining in fetal membranes and placental villous tissue of sPE and control (CRL) women. In CRL samples, the amnion epithelium (A; Am) stained less RAGE positive compared with fetal membrane choriodecidua (B) and placental villi (C). In sPE, the amnion epithelium stained intensely RAGE positive (D), whereas the fetal membrane choriodecidua (E) and placental villi (F) did not appear to change in RAGE staining intensity. RAGE detection was with Vector NovaRed and tissues were counterstained with hematoxylin. Negative controls of amnion (G) and placenta (H) had the primary antibody omitted. Bar, 50 μm for all panels. The results of the histological scoring analysis (mean ± se) are shown in I. Statistical analysis: one-way ANOVA followed by Student-Newman-Keuls tests.

Effect of inflammation and oxidative stress on sRAGE and IL-8 immunoreactivity in amniochorion and placental explant systems

We explored the role of inflammation and oxidative stress (X/XO) as potential triggers for sRAGE release by using amniochorion and placental villous explants. In unstimulated culture conditions, over the course of a 24-h period, we observed a significant increase in the sRAGE release in the amniochorion but not placental villous medium in which sRAGE remained undetectable (Fig. 5A). In the medium of amniochorion explants, the maximal sRAGE concentration was measured at 18 h.

Fig. 5.

Effect of LPS and X/XO on sRAGE and IL-8 production in a fetal membrane and placental villous explant system. Time course of sRAGE immunoreactivity in the amniochorion and placental villous explant medium during 24 h of incubation is shown (A). The fetal membrane explant medium concentration of sRAGE peaked at 18 h. sRAGE remained undetectable in the placental villous tissue system. The release of sRAGE was augmented by X/XO 2 mmol/liter (X) per 20 mU/ml (XO) (CRL vs. X/XO, P < 0.001) but not by LPS (1 μg/ml) (CRL vs. LPS, P = 0.716) (B). In contrast, LPS stimulated the release of the chemokine IL-8 (CRL vs. LPS, P < 0.001), whereas X/XO had no effect (CRL vs. X/XO P = 0.948) (C). A marked increase in RAGE immunoreactivity was noted in the amnion epithelium (Am) following exposure of fetal membranes to X/XO (D). Bar, 50 μm for all panels.

In Fig. 5B we show the differential effect of LPS vs. X/XO on sRAGE release by the amniochorion explants. X/XO, but not LPS, significantly stimulated the release of sRAGE in the medium at 18 h (control vs. X/XO, P < 0.001; control vs. LPS, P = 0.716). However, as shown in Fig. 5C, LPS was effective in stimulating the release of IL-8, whereas X/XO had no effect (control vs. X/XO, P = 0.948; control vs. LPS, P < 0.001). A marked increase in RAGE immunostaining was noted in the amnion epithelium after exposure of the fetal membrane explants to X/XO (Fig. 5D).

Discussion

As our understanding continues to evolve, we have gained a greater appreciation for not only the pathological but also the physiological role of the RAGE axis in human gestation. In our study we showed that, beginning with midpregnancy, total maternal sRAGE was significantly decreased compared with NP-CRLs, and its concentration remained low for the remainder of gestation. Thus, our results are slightly different from two prior studies that evaluated changes in the maternal serum sRAGE levels during normal pregnancy (25, 26). Sample size differences and evaluation of populations with diverse polymorphism of the AGER gene may account for this (20).

We posit that the delicate changes in the oxidative processes in healthy human pregnancy may at least partially account for our results. Endogenous production of FR/ROS via diverse enzymatic sources is recognized as crucial to normal development and homeostasis (16). Studies comparing healthy nonpregnant and pregnant women found greater lipid peroxide levels (product of the oxidative process) at the beginning of pregnancy, which may taper off after the second trimester (30). This is consistent with the hypothesis that sRAGE is needed in the first half of gestation to dampen activation of the RAGE signaling pathways secondary to an increased state of oxidative stress. It is also tempting to speculate that the fine regulation of production and clearance of AGEs may play a role, but further studies are needed to provide clarification.

Evidence derived from in vitro and in vivo experimental models have placed RAGE at the center of many conditions that are characterized by heightened oxidative and inflammatory stress, such as preeclampsia (8, 9). Similar investigational data have implicated sRAGE as a decoy protective molecule against cell injury and death (12, 18, 21). Recently Germanová et al. (25) reported elevated maternal serum sRAGE levels in the third trimester of women with preeclampsia and gestational hypertension. Our study expanded on these previous observations by demonstrating increased maternal serum sRAGE and esRAGE levels in women with sPE, but not crHTN, earlier than the third trimester. Interestingly, sRAGE levels of women with sPE were already elevated at 20 wk, which is traditionally recognized as the earliest diagnostic cutoff point for this disease (3). The pathophysiological implication of this observation is that, in preeclampsia, the maternal RAGE axis is functionally active at an early gestational age and sRAGE may exhibit a cell protective antioxidant and antiinflammatory activity, even in the absence of overt clinical symptoms.

Although long regarded as an antioxidant, uric acid can initiate intracellular redox signaling and renders cells vulnerable to oxidative damage (31). Injured cells release damage-associated molecular-pattern proteins (DAMPs) such as high-mobility group box-1 and S100 proteins, which are well-recognized RAGE ligands (32). Uric acid is itself considered to be a DAMP (32). Based on our results, it is tempting to propose that in sPE, the direct correlation between uric acid and sRAGE is part of the molecular mechanisms responsible for the FR/ROS-RAGE interaction. For now, there is no direct evidence that RAGE serves as sensor for uric acid. Still, it is possible that uric acid may play a direct or an indirect role by synergistically reinforcing the activity of other DAMPs, both at the receptor or transcriptional level. For example, up-regulation of sRAGE in women with sPE may be needed to counteract activation of the inflammatory and oxidative RAGE positive feedback loops caused by uric acid.

This study examines in parallel the levels of sRAGE in relevant gestational compartments. An unexpected finding was that in healthy pregnancy the highest levels of sRAGE were present in amniotic fluid and not in maternal blood. Preeclampsia displays an identical pattern, but similar to maternal serum, the total sRAGE amniotic fluid levels were higher compared with those expected for gestational age. The concept of a higher sRAGE clearance in the maternal vs. the amniotic fluid compartment will have to be further investigated. Along this line in concurrent studies (data not shown), we determined that sRAGE is undetectable in the urine of both healthy and sPE women, despite proteinuria.

At the DNA level, AGER consists of 11 introns/exons that can be spliced alternatively into different variants (33). esRAGE is a major splice variant of RAGE (11, 33). In considering mechanisms that could explain the increased levels of sRAGE in the amniotic fluid, we searched for the transcription levels of RAGE1 (encodes the transmembranar full length of the RAGE receptor) and RAGE2 (encodes esRAGE) in both placental villous trophoblast and fetal membranes. We found that RAGE1 and RAGE2 are constitutively expressed in both tissues but at higher levels in the placental trophoblast compared with the amniochorion. Interestingly, a higher level of RAGE2 expression was detected in the amniochorion of women with sPE. Our in vivo and in vitro data point toward fetal membranes as a source of amniotic fluid sRAGE. In this study we assessed the total sRAGE, which represents the sum of esRAGE and that of other RAGE variants including proteolytically derived forms (22). The RAGE2 mRNA results imply that the splicing cellular machinery, which is responsible for the release of the esRAGE, is an important mechanism for the synthesis and release of the RAGE antagonist in both maternal circulation and amniotic fluid of women with sPE.

The biological significance of oxidative stress in triggering RAGE activation is difficult to differentiate from that of inflammation alone, given that FR/ROS are known byproducts of inflammatory processes (34, 35). Yet the results of our explant experiments favor oxidative stress and not inflammation as the primary stimulus for the expression of RAGE in amniochorion and for RAGE activation in sPE. This would be consistent with the presence of oxidative modifications of the amniotic fluid proteins in preeclamptic women (36). Hence, it could be hypothesized that the oxidative stress conditions prevailing in preeclampsia are responsible for the activation of the signaling pathways responsible for the release of sRAGE as a compensatory mechanism against amniochorion tissue damage.