Suppression of Extravillous Trophoblast Vascular Endothelial Growth Factor Expression and Uterine Spiral Artery Invasion by Estrogen during Early Baboon Pregnancy

Abstract

We have shown that advancing the increase in maternal serum estrogen levels from the second to the first third of baboon pregnancy suppressed extravillous cytotrophoblast (EVT) spiral artery invasion. Because vascular endothelial growth factor (VEGF) promotes EVT invasion, the present study determined whether EVT VEGF expression is altered by prematurely elevating estrogen in early pregnancy. Placental basal plate was obtained on d 60 of gestation (term is 184 d) from baboons treated daily on d 25–59 with estradiol (0.35 mg/d sc), which increased maternal peripheral serum estradiol levels 3-fold above normal. Overall percentage of uterine arteries (25 to more than 100 μm in diameter) invaded by EVT assessed by image analysis in untreated baboons (29.11 ± 5.78%) was decreased 4.5-fold (P < 0.001) by prematurely elevating estrogen (6.55 ± 1.83%). VEGF mRNA levels in EVT isolated by laser capture microdissection from the anchoring villi of untreated baboons (6.77 ± 2.20) were decreased approximately 5-fold (P < 0.05, ANOVA) by estradiol (1.37 ± 0.29). Uterine vein serum levels of the truncated soluble fms-like receptor, which controls VEGF bioavailability, in untreated baboons (403 ± 37 pg/ml) were increased 3-fold (P < 0.01) by estrogen treatment (1127 ± 197 pg/ml). Thus, placental EVT expression of VEGF mRNA was decreased and serum soluble truncated fms-like receptor levels increased in baboons in which EVT invasion of the uterine spiral arteries was suppressed by advancing the rise in estrogen from the second to the first third of pregnancy. We suggest that VEGF mediates the decline in EVT vessel invasion induced by estrogen in early primate pregnancy.

DURING EARLY HUMAN and nonhuman primate pregnancy, progenitor placental cytotrophoblasts either remain in the fetal compartment and differentiate into the syncytiotrophoblast or aggregate into cell columns of the anchoring villi and differentiate into extravillous cytotrophoblasts (EVT), which infiltrate and remodel the walls of the uterine spiral arterioles and arteries at the placental-decidual junction. Uterine spiral artery remodeling is a critically important event in which EVT differentiate into invasive vascular-like endothelial cells, migrate from anchoring villi to and invade arteries located in the decidua basalis, and replace mural vascular smooth muscle cells and elastic fibers within the spiral arteries (1,2,3,4,5). Consequently, the uterine arteries are transformed from high-resistance low-capacity to low-resistance high-capacity vessels (6), presumably to promote blood flow to support placental and fetal growth and development. Despite the fundamental importance of vessel remodeling to successful pregnancy, relatively little is known about the regulation of this process (reviewed in Ref. 5). Moreover, because most of the research on this aspect of placental biology has been conducted in vitro, translation in vivo to the human or nonhuman primate is unproved. Because EVT migration and remodeling of the uterine arteries does not extend upstream beyond the inner layer of the myometrium, does not involve all of the uterine spiral arteries, and occurs primarily in the first trimester, vessel remodeling appears to be regulated by a balance between stimulatory and inhibitory factors, the levels of which change with advancing gestation. Using the baboon as a nonhuman primate model, we have recently shown that simply advancing the physiological surge in estrogens that occurs with normal gestation from the second to the first trimester markedly suppressed EVT uterine spiral artery invasion (7). The mechanisms and factors that modulate the estrogen-induced regulation of uterine artery invasion, however, have not been elucidated.

Vascular endothelial growth factor (VEGF)-A is expressed by EVT in the anchoring villi and cytotrophoblastic shell and binds to the fms-like (FLT1) and kinase-insert domain-containing tyrosine kinase receptors, which are expressed on placental vascular endothelial and extravillous and villous trophoblast cells (8,9,10,11). VEGF can be sequestered and thus inactivated by a soluble truncated FLT1 receptor (sFLT1), which lacks the transmembrane/cytoplasmic signal transduction domain and is also expressed by placental villous and EVT (12). Importantly, migration and invasion of human EVT as assessed in vitro were suppressed when VEGF receptor binding and signal transduction were blocked by receptor fusion proteins (13). In preeclampsia, which is associated with rudimentary vascular invasion (14), there is EVT down-regulation of VEGF and placental up-regulation of sFLT1 (13,15). These and other in vitro studies have led to the concept that VEGF plays a pivotal role in promoting EVT migration and spiral artery invasion (13,16,17).

Estrogen increases VEGF expression by glandular epithelial and stromal cells of the uterus in the baboon (18,19), sheep (20), and rat (21) and placental villous cytotrophoblasts in the baboon (22,23) and decreases VEGF expression by breast cells (24,25). We propose, therefore, that VEGF mediates the estrogen-induced decline in EVT vessel invasion shown in our recent study (7). To begin to assess this possibility, VEGF and sFLT1 mRNA levels in EVT isolated from the placental basal plate and sFLT1 protein levels in uterine vein serum were compared with the level of uterine spiral artery invasion in baboons treated with estradiol early in pregnancy to prematurely elevate estrogen.

Materials and Methods

Animals

Female baboons (Papio anubis), originally obtained from the Southwest Foundation for Biomedical Research (San Antonio, TX) and weighing 13–15 kg, were used in this study. Animals were housed individually in large primate cages and received standard monkey chow (Harlan Primate Diet, Madison, WI) and fresh fruit twice daily, multiple vitamins daily, and water ad libitum. Females were paired with male baboons for 5 d at the time of ovulation, as estimated by menstrual cycle history and perineal turgescence, and d 1 of pregnancy was designated as the day preceding deturgescence. Baboons were cared for and used strictly in accordance with U.S. Department of Agriculture regulations and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (National Academy Press, 1996). The experimental protocol was approved by the Institutional Animal Care and Use Committee of the University of Maryland School of Medicine.

A 2- to 4-ml blood sample was obtained from a maternal peripheral saphenous vein every 1–2 d during the study period from baboons after ketamine HCl sedation (10 mg/kg body weight, im). Placentas were aseptically removed by cesarean section on d 60 of gestation (length of gestation is 184 d) from isoflurane-anesthetized baboons untreated (n = 9) or treated daily on d 25–59 with estradiol benzoate (0.35 mg/d sc injection in 1.0 ml sesame oil, n = 10). At the time of cesarean section, blood samples (1 ml) were obtained from the left and right uterine veins to determine estrogen and sFLT1 levels.

Serum estradiol levels were quantified by RIA using an automated chemiluminescent immunoassay system (Immulite; Diagnostic Products Corp., Los Angeles, CA), as described previously (26). Serum total sFLT1 levels were measured by ELISA (R&D Systems, Inc., Minneapolis, MN).

EVT spiral artery invasion

At least eight randomly selected sections (each 5 mm³) of placental basal plate, comprised of fetal (trophoblast) and maternal uterine decidua basalis tissue that adheres to the placenta at delivery were collected and each section sliced in half. EVT vessel invasion was quantified in each tissue section half in the decidua basalis only because trophoblast invasion of uterine spiral arteries does not extend beyond the decidua in the baboon (27). Tissue was fixed in 10% formalin, embedded in paraffin, sectioned at 4 μm, and processed for hematoxylin/eosin histology and immunocytochemistry. Light microscopy (Nikon Eclipse E 1000 M, Tokyo, Japan) and image analysis (IP Lab, version 3.63; Scanalytics, Inc., Fairfax, VA) were used to assess EVT invasion in all of the arterioles/arteries categorized into diameters of less than 25, 26–50, 51–100, and greater than 100 μm as detailed previously (7). Each observed artery/arteriole was considered a separate vessel, although any single vessel may have appeared more than once as it traversed a given tissue section. The number of vessels exhibiting invasion in cytokeratin immunostained sections was quantified by identifying spiral arterioles/arteries in which the vessel wall was extensively occupied by EVT.

Cell isolation by laser capture microdissection (LCM)

The remaining half of each placental basal plate section was frozen in optimum cutting temperature cryomolds, sectioned (8 μm) via a cryostat at −20 C (Lieca Corp., Deerfield, IL), and mounted onto glass slides (Superfrost Plus; Fisher Scientific, Suwanee, CA) at room temperature. Sections were fixed in 70% ethanol for 30 sec, stained in hematoxylin for 10 sec, dehydrated in 100% ethanol, incubated 5 min with 2× xylene, air dried, and stored in a desiccator containing silica gel before laser capture. An Arcturus PixCell II LCM system equipped with an Olympus microscope (Arcturus Engineering, Inc., Mountain View, CA) set at 71 mW, 40 msec capture duration, and laser spot size of 7.5 or 15.0 μm (depending on tissue area) was used to capture cells from the anchoring villi and cytotrophoblastic shell of the placental basal plate, cytotrophoblast, and syncytiotrophoblast collectively from the outer surface of the floating villi and cells from the decidua basalis excluding observable blood vessels. Cells were collected from 10–12 randomly selected areas of each placental sample standardized with respect to location in the central region of the cell columns of the anchoring villi and cytotrophoblastic shell. Cells were isolated from both the anchoring villi and cytotrophoblastic shell to include proximal and distal areas along the EVT migration pathway. The cell isolates pooled from the eight placental sections from each baboon were extracted via Nonidet P-40 guanidine isothiocyanate silica gel spin column centrifugation (Rneasy; QIAGEN, Valencia, CA) and RNA lysates stored overnight at −80 C.

Reverse transcription (RT)-real-time PCR

VEGF primers.

Oligonucleotide primers were designed using LightCycler probe design software (Roche Diagnostics Corp., Penzberg, Germany), based on the VEGF human gene sequence (28), and supplied by Integrated DNA Technologies, Inc (Coralville, IA). The VEGF primers [upstream, 5′-GCATTGGAGCCTTGCCTT-3′ (position 24–41) and downstream, 5′-GCCTTGGTGAGGTTTGAT-3′ (position 342–325)] spanned exons 1, 2, and 3 of the VEGF gene and were upstream of the alternative splice site that generates the different 121, 145, 165, 189, and 206 isoforms of VEGF-A so that a single 323-bp PCR product reflecting all VEGF-A isoforms was formed.

sFLT1 primers.

The sFLT1 primers were: upstream, 5′-CCATCACTAAGGAGCACTCCATCA (positions 2078–2101) and downstream, 5′-AGCCTTTTTGTTGCAGTGCTCACC (positions 2241–2218, accession no. U01134; National Center for Biotechnology Information sequence database, Bethesda, MD).

18S rRNA primers.

The 18S rRNA primers [upstream, 5′-TCAAGAACGAAAGTCGGAGG-3′ (positions 1126–1145) and downstream, 5′-GGACATCTAAGGGCATCACA-3′ (positions 1614–1595)] were based on the human gene sequence (National Center for Biotechnology Information sequence database, accession no. M10098).

RT and real-time PCR.

RT of total RNA from LCM isolates was performed at 42 C for 60 min in a 20-μl reaction volume containing 1 mm each of deoxy (d)-ATP, dCTP, dGTP, and -thymidine triphosphate, 200 U Moloney murine leukemia virus RT, 1× RT buffer, 250 ng random primers (preceding reagents from Invitrogen, Carlsbad, CA), and 40 U RNAguard (Amersham Pharmacia Biotech, Piscataway, NJ). The RT reaction was terminated by heat inactivation of the enzyme at 70 C for 15 min and cooled to 4 C.

VEGF and sFLT1 mRNA levels were quantified by efficiency-corrected calibrator-normalized relative RT-PCR using LightCycler SYBR Green I technology (Roche Diagnostics). An aliquot (1 μl) of the RT reaction mixture was added to 19 μl of LightCycler-FastStart DNA Master SYBR Green I reaction mix containing the gene-specific primers. The reaction profile consisted of denaturation at 95 C for 8 min, 40 cycles of amplification (95 C for 5 sec, 52 C for 5 sec, and 72 C for 16 sec), and product formation measured and displayed in real time. RNA isolated from LCM-captured cells exhibited distinct 28S and 18S rRNA bands, and the input cDNA resulting from RT-PCR was of high quality. However, because the amount of total RNA obtained from LCM samples was too low to measure reliably via spectrophotometry, the levels of 18S rRNA were also simultaneously quantified via real-time PCR and mRNA levels expressed as a ratio of 18S rRNA. Specificity of the products was confirmed by melting curve analysis, agarose gel electrophoresis, and inclusion of negative controls with no template or RT in the reaction. Correction for differences in the efficiencies of target and reference genes and calibrator normalized quantification were automatically calculated by the analysis software (LightCycler version 4) based on the concentrations and efficiencies of standard curves previously made specifically for each product.

Immunocytochemistry

To assess immunocytochemical localization of peptides, placental tissue sections were boiled in 0.01 m Na citrate for antigen retrieval, pretreated with Protease (Biomeda, Foster City, CA) for 5 min at room temperature, incubated in H₂O₂ to inhibit endogenous peroxidase, and blocked with serum-free protein block (Dako Corp., Carpenteria, CA). Tissue sections were then incubated overnight at 4 C with mouse antibodies to Pan-cytokeratin (1:2000 final dilution; Novocastra/Vector, Burlingame, CA), vimentin (1:6000; Dako), and CD68 (1:800; Dako); rabbit antibody to VEGF (1:400; Zymed/Invitrogen, Carlsbad, CA); mouse antibody to estrogen receptor-α (1:80; Novocastra/Vector); and rabbit antibody to estrogen receptor-β (1:200; Abcam, Cambridge, MA).

Tissues were incubated 1 h (room temperature) with either biotinylated antimouse or antirabbit immunoglobulin (Novocastra/Vector) and 1 h with an avidin-biotin-peroxidase complex (ABC Elite; Novocastra/Vector). Tissue sections were developed using diaminobenzidine (Sigma, St. Louis, MO) and lightly counterstained with hematoxylin. Negative controls for immunocytochemistry included omission of the primary antibody, preabsorption of the primary antibody with an excess of the respective peptide, and/or substitution of goat immunoglobulin (Dako) for primary antibody.

Statistical analyses

Data were expressed as the means ± se. Serum estradiol, placental VEGF mRNA, and sFLT1 levels were analyzed by ANOVA with post hoc comparison of the means by Newman-Keuls multiple comparison test or by Student’s unpaired t test. Data of spiral artery invasion were analyzed using the generalized estimating equation to control for within animal correlation and Genmod procedure within SAS (Cary, NC). Spiral artery invasion proportions were analyzed using generalized linear models with a log link and binomial error distribution. Results are presented as relative risks of arterial invasion and compared data from estradiol treated and untreated animals within arterial diameter groupings.

Results

Serum estradiol and weights

Maternal peripheral serum estradiol concentrations in untreated baboons exhibited a very gradual increase from approximately 0.10 ng/ml on d 25 to more than 0.40 ng/ml on d 60 of gestation (Fig. 1). To place the levels of estrogen at this time in early pregnancy into perspective, maternal serum estradiol levels rapidly increase to more than 2.5 ng/ml on d 65–85 (i.e. during second third of gestation) and more than 4.0 ng/ml on d 184, i.e. near term (26). The administration of estradiol on d 25–59 increased maternal peripheral vein serum estradiol on each of the days between d 27 and 60 to levels that were greater (P < 0.01) than in the untreated animals (Fig. 1). Estrogen treatment also elevated serum estradiol levels on d 60 in the maternal saphenous (1.07 ± 0.12 ng/ml) and uterine (1.27 ± 0.23 ng/ml) veins to values that were approximately 3-fold (P < 0.001) and 2-fold (P < 0.05) greater, respectively, than in untreated animals (0.30 ± 0.06 and 0.62 ± 0.12 ng/ml, respectively, Fig. 1 and Table 1).

Figure 1.

Maternal peripheral serum estradiol levels (means ±se) in baboons untreated (n = 9) or treated daily on d 25–59 of gestation (term is 184 d) with estradiol benzoate (0.35 mg/d sc, n = 10). Serum estradiol levels on each of the days between d 27 and 60 are greater (P < 0.01) in estradiol-treated than untreated animals.

Table 1.

Maternal serum estradiol levels and placental and fetal body weights in baboons

		Estradiol (ng/ml)
Treatment	n	Saphenous vein	Uterine vein	Placental weight (g)	Fetal body weight (g)	Maternal body weight (kg)
Untreated	9	0.30 ± 0.06	0.62 ± 0.12	32.2 ± 2.4	11.8 ± 0.8	15.3 ± 0.9
Estradiol	10	1.07 ± 0.12a	1.27 ± 0.23b	29.0 ± 1.3	11.5 ± 0.8	15.1 ± 0.6

Mean ± se maternal saphenous and uterine vein serum estradiol levels and placental and fetal body weights on d 60 of gestation in baboons untreated or treated daily on d 25–59 of gestation (term = 184 d) with estradiol benzoate (0.35 mg/d, sc). 

^aP < 0.001, values different from untreated baboons (ANOVA and Newman-Keuls multiple comparison test). 

^bP < 0.05, values different from untreated baboons (ANOVA and Newman-Keuls multiple comparison test). 

Placental and fetal body weights on d 60 were not significantly different in untreated vs. estrogen-treated baboons (Table 1). Moreover, maternal body weights and indices of maternal cardiovascular function were similar on d 60 in untreated and estradiol-treated baboons, including arterial systolic (69 ± 2 and 69 ± 3 mm Hg) and diastolic (26 ± 1 and 30 ± 3 mm Hg) blood pressure and heart rate (112 ± 2 and 98 ± 6 beats/min), respectively.

Spiral artery invasion

Figure 2, A and B, shows representative histology of the floating chorionic villi, which contain the villous cytotrophoblasts and the placental basal plate comprised of anchoring villi, which are the principal source of EVT, cytotrophoblastic shell, and decidua basalis. Illustrated are spiral arteries invaded by EVT in untreated baboons (Fig. 2, A and B) and arteries not invaded in estradiol-treated baboons (Fig. 2, C and D). The striking architectural modification of the arterial vessel wall resulting from cytotrophoblast invasion in the untreated baboons contrasts with the tightly organized structure of noninvaded arteries in estradiol-treated animals.

Figure 2.

Photomicrographs of histology of the placental basal plate and chorionic villi on d 60 of gestation in baboons untreated (A and B) and treated daily on d 25–59 with estradiol (C and D). Magnifications, ×200 (A and C) and ×400 (B and D). FV, Floating villi; AV, anchoring villi; V, vessel wall with trophoblast cells and fibrinoid; CS, cytotrophoblastic shell; DB, decidua basalis; SA, spiral artery.

Approximately 50% of the uterine spiral arteries on d 60 of gestation were 26–50 μm in diameter, approximately 20% were less than 25 or 51–100 μm in diameter, whereas 9% were greater than 100 μm in diameter (Table 2). This arterial size distribution was similar in untreated and estrogen-treated baboons.

Table 2.

Percent distribution of uterine spiral arteries in the placental basal plate of baboons

		Arterial diameter (μm)
	n	<25	26–50	51–100	>100
Untreated	9	22 ± 3.3	49 ± 3.0	20 ± 2.4	9 ± 1.0
Estradiol	10	23 ± 2.2	51 ± 3.3	17 ± 3.1	9 ± 2.0

Mean ± se distribution of uterine spiral arteries in the placental basal plate catagorized by diameter on d 60 in baboons untreated or treated daily on d 25–59 with estradiol as detailed in the legend of Table 1. 

The number of spiral arterioles/arteries invaded by EVT, expressed as the mean (±se) percentage of the total number of vessels counted, is shown in Fig. 3A. Arterial invasion in untreated baboons for vessels of less than 25 μm (5.30 ± 2.78%), 26–50 μm (5.16 ± 1.74%), 51–100 μm (20.44 ± 8.30%), and greater than 100 μm (67.72 ± 6.30%) in diameter was decreased (P < 0.01) to 0.20 ± 0.20, 0.66 ± 0.23, 3.18 ± 1.14, and 15.80 ± 4.07%, respectively, by the administration of estradiol. Collectively, the level of spiral artery invasion for all vessels analyzed (i.e. including the smaller resistance and larger conduit arteries) was 29.11 ± 5.78% in untreated baboons and 6.55 ± 1.83% or 5-fold lower (P < 0.001) in estradiol-treated animals. Moreover, the decline in cytotrophoblast invasion of spiral arteries of each vessel size class in estrogen-treated (P < 0.01), compared with untreated baboons, was confirmed when vessel invasion was calculated on the basis of decidua basalis area, i.e. number of arteries invaded per 10⁶ μm² decidua (Fig. 3B). The relative risks and 95% confidence intervals for percent invasion of spiral arteries as determined by the generalized estimating equation and Genmod procedure are shown in Table 3.

Figure 3.

Percent invasion (A) and number invaded per square micrometer decidua (B) of uterine spiral arterioles/arteries grouped by vessel diameter on d 60 in baboons untreated (n = 9) or treated daily on d 25–59 with estradiol benzoate (n = 10). *, Values different (P < 0.01) from untreated animals for each arterial diameter as analyzed by the generalized estimating equation and Genmod procedure (Table 3).

Table 3.

Relative risks and 95% confidence intervals for invasion of spiral arteries on d 60 of baboon pregnancy^a

	Arterial diameter (μm)
Treatment	<25	26–50	51–100	>100
Untreated	1.0	1.0	1.0	1.0
Estradiol, percent invasion	0.037 (0.004, 0.306)b	0.114 (0.053, 0.247)b	0.171 (0.039, 0.738)b	0.188 (0.110, 0.319)b
Estradiol, number invaded per square micrometer decidua	0.031 (0.004, 0.270)b	0.100 (0.051, 0.195)b	0.141 (0.038, 0.527)b	0.188 (0.101, 0.351)b

^aData analyzed using the generalized estimating equation and Genmod procedure with SAS statistical software. Relative risks were determined for percent invasion of spiral arteries categorized by diameter and number of arteries invaded per square micrometer decidua. Treatment groups were adjusted for arterioles/arteries analyzed per tissue section and within-animal correlation. 

^bValues with confidence intervals (in parentheses) of less than 1.00 are different (P < 0.01) from those in untreated controls. 

VEGF mRNA expression

The majority of cells in the anchoring villi and cytotrophoblastic shell on d 60 of gestation were EVT as evidenced by extensive immunocytochemical expression of epithelial cell-specific cytokeratin (Fig. 4A) and limited expression of fibroblast-specific vimentin (Fig. 4B, labeled decidual basalis) and macrophage marker CD68 (Fig. 4C). Although vimentin-positive mesenchymal core tissue (Fig. 4B) was present within the anchoring villi of the placenta, this was easily visualized and not included in the LCM isolates. Figure 5 illustrates histology before (Fig. 5A) and after (Fig. 5B) LCM isolation of cells specifically from the anchoring villi and cytotrophoblastic shell of the baboon placenta and presence of isolated cells on the LCM cap (Fig. 5C) for RNA analysis.

Figure 4.

Immunocytochemical localization (brown precipitate) of cytokeratin (A), vimentin (B), and CD68 (C) in the placental basal plate on d 60 of baboon pregnancy. D, Vimentin primary antibody replaced with immunoglobulin control. AV, Anchoring villi; CS, cytotrophoblastic shell; DB, decidua basalis: MC, mesenchymal core. Magnification scale bar, 100 μm.

Figure 5.

Placental histology on d 60 of baboon pregnancy before (A) and after (B) LCM of cells from anchoring villi (AV) and cytotrophoblastic shell (CS). C, Isolated cells in LCM cap. LCM, area of cell isolation; MC, mesenchymal core; DB, decidua basalis; VS, venous sinusoid.

As shown in Fig. 6, VEGF mRNA in untreated baboons was abundantly expressed in EVT isolated from the anchoring villi and cytotrophoblastic shell. Most importantly, VEGF mRNA levels (expressed as a ratio of 18S rRNA) in untreated baboons in cells of the anchoring villi (6.77 ± 2.20, Fig. 6) and trophoblastic shell (5.28 ± 2.34) were decreased approximately 5-fold (P < 0.05) by estradiol administration (1.37 ± 0.29 and 1.04 ± 0.49, respectively). Moreover, VEGF mRNA levels in cells isolated from the decidua were also substantially lower (P < 0.01) in estrogen-treated (0.07 ± 0.01) than untreated (0.35 ± 0.06) baboons. However, the decline in EVT VEGF expression in the latter tissues appears to be cell specific because VEGF mRNA levels in syncytiotrophoblast and cytotrophoblast cells isolated from the floating villi were similar in the same untreated (3.20 ± 1.35) and estradiol-treated (2.75 ± 0.52) baboons (Fig. 6).

Figure 6.

VEGF mRNA levels (means ± se) in cells isolated by LCM from the anchoring villi, cytotrophoblastic shell, and floating villi of the placenta on d 60 in baboons untreated (n = 5) or treated daily on d 25–59 with estradiol (n = 5). *, Different (P < 0.05) from values in untreated animals (ANOVA and Newman-Keuls multiple statistic).

sFLT1 serum and mRNA levels

The levels of serum sFLT1 were 3-fold greater (P < 0.01) on d 60 in the uterine veins (mean of left and right veins that drain placenta) of estrogen treated (1127 ± 197 pg/ml) than untreated (403 ± 37 pg/ml) baboons (Fig. 7A). Coinciding with the increase in serum, sFLT1 mRNA levels in cells isolated from the cytotrophoblastic shell were almost 3-fold greater (P < 0.059) in estradiol-treated than untreated baboons (Fig. 7B). However, sFLT1 mRNA expression by cells of the anchoring villi was low and similar in untreated and estradiol-treated animals (Fig. 7C).

Figure 7.

Serum sFLT1 levels (means ±se) in uterine vein (mean of left and right, A) and sFLT1 mRNA levels in cells isolated by LCM from the cytotrophoblastic shell (B) and anchoring villi (C) on d 60 in baboons untreated (n = 4) or treated daily on d 25–59 of gestation with estradiol (n = 5). *, Different (P < 0.01) from value in untreated animals (Student’s t test).

Immunocytochemistry

In untreated baboons, VEGF protein (Fig. 8A, identified by brown staining) was localized by immunocytochemistry to cells within the anchoring villi but not the pale-staining mesenchymal core (inset), cytotrophoblastic shell (demarcated by extensive extracellular matrix that lacked staining), and decidua basalis as well as trophoblasts covering the outer surface of the floating villi. Estrogen receptor-β (brown immunostaining, Fig. 8C) was also abundant in nuclei of trophoblast cells covering the outer surface of the floating villi and in the anchoring villi (inset) and present, although lighter, in the cytotrophoblastic shell and the decidua basalis of untreated animals. In contrast, estrogen receptor-α immunocytochemical staining (Fig. 8D) appeared less abundant throughout the placental basal plate, including in cells of the anchoring villi (Fig. 8D, inset). In estradiol-treated baboons, the level of immunostaining for VEGF protein (Fig. 8B) appeared to be reduced throughout the placental basal plate, notably in the blue counterstained cells within the anchoring villi (Fig. 8B, inset) when contrasted with that present in untreated animals (Fig 8A, inset); however, VEGF expression was retained in the trophoblast cells covering the anchoring villi (Fig. 8B, inset). The low level of VEGF immunostaining in cells within the anchoring villi and cytotrophoblastic shell did not seem to reflect a loss of EVT in these regions because the extensive amount of epithelial cell-specific brown cytokeratin immunoreactivity and lack of staining in the light blue-colored decidua was similar in the placental basal plate of untreated (Fig. 8E) and estradiol-treated (Fig. 8F) baboons. When the VEGF primary antibody was preabsorbed with 100-fold excess of human recombinant VEGF (Fig. 8G) or the estrogen receptor-β primary antibody replaced with control immunoglobulin (Fig. 8H), there was no immunostaining, indicating specificity of the antibodies employed.

Figure 8.

Representative photomicrographs of placental immunocytochemical localization (brown precipitate) on d 60 of gestation of VEGF (A), estrogen receptor-β (C), estrogen receptor-α (D), and cytokeratin (E) in untreated baboons and VEGF (B) and cytokeratin (F) in baboons treated with estradiol on d 25–59. Preabsorption of VEGF antibody with recombinant VEGF protein (G) and replacement of primary antibody for estrogen receptor-β with immunoglobulin (H). Insets in A–D represent higher magnification of anchoring villi region illustrating immunocytochemistry for each respective protein. FV, Floating villi; AV, anchoring villi; CS, cytotrophoblastic shell; DB, decidua basalis. Magnification bar (G), 250 μm.

Figure 9A shows epithelial-specific cytokeratin immunostaining, indicative of the presence of EVT, in the wall of a remodeled spiral artery in an untreated baboon and lack of staining in the blue-colored stroma. Although VEGF immunostaining remained extensive in EVT after vessel invasion (Fig. 9B), estrogen receptor-β (Fig. 9C), and estrogen receptor-α (Fig. 9D) protein was absent or very low.

Figure 9.

Immunocytochemical localization of cytokeratin (A), VEGF (B), estrogen receptor-β (C), and estrogen receptor-α (D) within cells of the wall of a remodeled spiral artery in an untreated baboon on d 60 of gestation. Magnification scale bar, 100 μm.

Discussion

The present study shows that placental EVT expression of VEGF mRNA was decreased and placental trophoblastic shell sFLT1 mRNA and uterine vein serum sFLT1 levels increased in baboons in which EVT invasion of the uterine spiral arteries was suppressed by prematurely elevating estrogen levels early in baboon pregnancy. In vitro studies have indicated that VEGF has a pivotal role in promoting EVT migration and invasion (13,16,17). Presumably, the decrease in EVT VEGF expression and increase in levels of the truncated FLT1 receptor that sequesters VEGF would decrease the bioavailability of VEGF required for EVT migration to and invasion of the uterine spiral arteries. Moreover, estrogen regulates VEGF expression in other tissues (18,19,20,21,24,25). Collectively, based on the results of the latter and the present studies, we propose that VEGF mediates the decline in EVT spiral artery invasion and remodeling induced by estrogen in early primate pregnancy.

We believe our study is the first to show that estrogen, a hormone fundamentally important to the physiology of normal primate pregnancy (29,30), has a central role in regulating EVT uterine artery development. We propose that the very low level of endogenous estrogen exhibited in early gestation ensures normal spiral artery invasion by EVT and the increase in estrogen that occurs during advancing pregnancy has a physiologically important role in regulating the extent to which uterine arteries are remodeled by inhibiting/restraining EVT VEGF expression. Therefore, we suggest that the proper level of uterine vascular remodeling during normal pregnancy is temporally coordinated by estrogen.

Insufficient trophoblast invasion and remodeling of the uterine arteries may impair blood flow to the placenta and lead to miscarriage and complications during pregnancy, e.g. preeclampsia (31,32) and fetal growth restriction, which result in neonatal morbidity and mortality. Conversely, excessive EVT migration and replacement of the smooth muscle layer of uterine arteries upstream presumably would disrupt uteroplacental blood flow dynamics and subsequently impair essential vasoregulatory processes, e.g. uterine artery vasoconstriction, which along with myometrial contraction prevents excessive blood loss at delivery and vasomotor function within the remaining endometrial basalis after delivery. Although a definitive cause-and-effect link between spiral artery invasion/remodeling and delivery of blood to and growth of the fetus has not been demonstrated, uteroplacental vessel remodeling as well as villous vasculogenesis early in pregnancy almost certainly impacts pregnancy outcome (33,34) and fetal growth (35,36). Although placental and fetal body weights were not significantly altered through d 60 of gestation in baboons in which vessel invasion was suppressed by estrogen, it is likely that fetal growth and development would be adversely impacted later in pregnancy when growth and physiological demands of the fetus become exponential. The baboon provides an excellent in vivo primate model to investigate the impact of vessel remodeling on placental/fetal blood flow and development.

The current study further shows that estrogen decreased VEGF mRNA levels in cytotrophoblasts within the extravillous compartment but had no effect on VEGF mRNA levels in cytotrophoblasts and syncytiotrophoblast collectively obtained from the villous compartment. Moreover, VEGF mRNA levels were increased by estrogen in cytotrophoblasts selectively isolated from the placental villi of baboons (22,23) and uterine endometrial cells in nonpregnant baboons (18,19), rats (21), and sheep (20). Thus, estrogen differentially regulates in a cell-specific manner VEGF mRNA expression within different compartments of the placenta and the uterus. These cell-specific divergent effects of estrogen on VEGF expression are also exhibited in other estrogen-responsive systems. For example, estrogen increased VEGF transcription and cell proliferation in estrogen receptor-α- and -β-positive MCF-7 and decreased VEGF expression and cell proliferation in estrogen receptor-α-negative -β-positive MDA-MB breast cancer cells (24,25,37). This differential regulation may reflect cell-specific presence/absence of estrogen receptor subtypes as well as functional enhancers, coactivators, and/or corepressors that modulate estrogen-regulated target gene transcription (37,38,39) and be particularly manifest in placental trophoblasts that undergo marked cellular differentiation as they migrate down either the villous or extravillous pathways.

The molecular mechanisms by which estrogen regulates placental VEGF expression remain to be elucidated. The present study showed that estrogen receptor-β was abundantly expressed in cells of the anchoring villi and cytotrophoblastic shell, areas of the placental basal plate heavily populated by EVT. This would provide a mechanism for estradiol to exert an inhibitory action on expression of VEGF by EVT located in this region of the placenta. However, as shown in this study, once EVTs have invaded the uterine arteries, they no longer expressed estrogen receptor and thus may no longer respond to estrogen. VEGF may regulate expression of key integrins, cell adhesion molecules, matrix metalloproteinases, and other cell surface-extracellular matrix remodeling molecules, which are integral to EVT migration and invasion (13,40,41). Although there are no consensus estrogen response elements in the VEGF promoter, estrogen gene regulation is a multifactorial process controlled via interaction between estrogen receptors and transcription factors (42,43), notably hypoxia-inducible factor-1, which is recruited along with estrogen receptor-α to the VEGF promoter in the rat uterus (44). Hypoxia is a potent stimulus of VEGF expression in many cell systems (10,45,46), and HIF-1 modulates physiological responses to hypoxia (47). Therefore, it is possible that the effects of estrogen on EVT VEGF expression and spiral artery invasion involve indirect hypoxia-modulated as well as direct actions of estrogen. Because VEGF is typically up-regulated by hypoxia, whereas estrogen acutely increases blood flow and thus delivery of oxygen within the uterine artery in sheep (48,49), the suppression of EVT VEGF expression and vessel invasion might be attributed to an increase in uterine blood flow/oxygen levels in estrogen-treated baboons of the present study. However, this seems unlikely because recent in vitro studies show that EVT invasion is actually increased by elevated oxygen levels (50,51,52).

Other growth factors have been shown by in vitro studies to modulate trophoblast invasion, including TGF-β (53), IGF-2 (54), epidermal growth factor (55), placental GH (56), and placenta growth factor (10,57). Whether the latter factors are operable in vivo, are regulated by estrogen, and/or potentially interact with VEGF with respect to EVT vessel invasion during primate pregnancy remain to be elucidated.

In summary, the present study shows that placental EVT VEGF mRNA expression was suppressed, and levels of uterine vein serum sFLT1, which sequesters and thus governs VEGF bioavailability, were increased in baboons in which EVT uterine spiral artery invasion was decreased by advancing the increase in estrogen levels from the second to the first third of gestation. Therefore, we propose that VEGF mediates the decline in EVT spiral artery invasion induced by estrogen, whereby the low level of endogenous estrogen in early gestation ensures normal EVT spiral artery invasion and the increase in estrogen of advancing pregnancy has a physiologically important role in regulating the extent to which uterine arteries are remodeled by inhibiting EVT VEGF expression.