1. Research www. AJOG.org
BASIC SCIENCE: OBSTETRICS
Estrogen receptor-beta mediates cyclooxygenase-2
expression and vascular prostanoid levels
in human placental villous endothelial cells
Emily J. Su, MD, MS; Zhi-Hong Lin, PhD; Rana Zeine, MD, PhD;
Ping Yin, PhD; Scott Reierstad, BS; Joy E. Innes, BS; Serdar E. Bulun, MD
OBJECTIVE: Regulation of fetoplacental blood flow is likely mediated (P .05) in the presence and absence of estradiol. ESR2 knock-down
by factors such as prostanoids. Estrogen and its receptors affect pro- also led to diminished prostacyclin and thromboxane concentrations in
stanoid biosynthesis. Previously, we demonstrated that villous endo- the absence of estradiol (P .005).
thelial cells express estrogen receptor-beta (ESR2), and we sought to
determine its role in the mediation of fetoplacental vascular function. CONCLUSION: ESR2 mediates COX-2 expression levels and both
prostacyclin and thromboxane concentrations in the basal state, which
STUDY DESIGN: Villous endothelial cells from uncomplicated preg-
suggests the possibility of ligand-independent regulation of COX-2 ac-
nancies were isolated, cultured, and treated with estrogen. RNA inter-
tivity and prostaglandin H2 substrate availability. Further investigation
ference, real-time polymerase chain reaction, Western blotting, and en-
regarding ESR2 regulation of prostanoid biosynthesis and its effects on
zyme immunoassays were performed.
the fetoplacental vasculature is warranted.
RESULTS: Cyclooxygenase-2 (COX-2) expression levels were not al-
tered consistently by estrogen. RNA interference of ESR2 led to a con- Key words: cyclooxygenase-2, estrogen receptor-beta, fetoplacental,
comitant decrease in COX-2 messenger RNA (P .0001) and protein villous endothelial cell
Cite this article as: Su EJ, Lin Z-H, Zeine R, et al. Estrogen receptor-beta mediates cyclooxygenase-2 expression and vascular prostanoid levels in human
placental villous endothelial cells. Am J Obstet Gyncol 2009;200:427.e1-427.e8.
A dequate fetoplacental blood flow is a
critical element in the achievement of
successful pregnancy outcome; this circu- topathologic standpoint, placentas that are
ponent in appropriate uteroplacental
function.
The fetoplacental circulation is com-
lation remains incompletely understood. affected by fetal growth restriction with ab- prised of umbilical and fetal vessels as
Doppler studies have established that, in normal umbilical artery Doppler indices well as villous stem vessels and their
fetal growth restriction because of utero- often demonstrate fetal stem vessel vaso- downstream branches. Unlike other vas-
placental insufficiency, aberrant umbilical constriction and luminal obliteration.4,5 cular systems within the human body,
arterial and venous blood flow may exist, Together, these clinical and pathologic the placental portion of this circulation
which leads to potential risks of fetal hy- findings suggest that regulation of fetopla- lacks innervation, which suggests that
poxemia and acidemia.1-3 From a his- cental vascular tone is an important com- autonomic regulation of its tone does
not exist.6 Rather, fetoplacental vasomo-
From the Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology tor tone appears to be controlled primar-
(Dr Su); the Division of Reproductive Biology Research, Department of Obstetrics and ily by humoral factors.7
Gynecology (Drs Lin, Yin, and Bulun, Mr Reierstad, and Ms Innes), and the Department of Prostanoids are 1 group of major reg-
Pathology (Dr Zeine), Feinberg School of Medicine, Northwestern University, Chicago, IL. ulators of the vasculature. They are de-
This research was presented at the 29th Annual Meeting of the Society for Maternal–Fetal rived from metabolism of arachidonic
Medicine, San Diego, CA, Jan. 26-31, 2009.
acid by the endothelium and vascular
Received Nov. 22, 2008; revised Jan. 7, 2009; accepted Jan. 21, 2009.
smooth muscle cells and include the
Reprints: Serdar E. Bulun, 303 E. Superior St., Lurie Building, 4-250, Chicago, IL 60611.
prostaglandins (PGD2, PGE2, PGF2 ),
s-bulun@northwestern.edu.
prostacyclin (PGI2), and thromboxane
Support for this research was provided by the AAOGF/SMFM Scholarship Award 2007-10 and
Northwestern Memorial Foundation Private Donor Grant 2007-08. (TXA2; Figure 1).8 These substances are
0002-9378/free • © 2009 Mosby, Inc. All rights reserved. • doi: 10.1016/j.ajog.2009.01.025 important in the local regulation of vas-
cular tone in both normal and abnormal
physiologic states. Their mechanism of
For Editors’ Commentary, see Table of Contents action occurs in an autocrine or para-
crine fashion by binding to specific re-
APRIL 2009 American Journal of Obstetrics & Gynecology 427.e1

2. Research Basic Science: Obstetrics www.AJOG.org
respect to vascular physiologic develop-
FIGURE 1
ment, they are known to regulate the ex-
Immunohistochemical distribution of ESR2
pression of multiple vasodilator and va-
soconstrictor proteins.19 ESR2 has been
shown to be critical in maintenance of
normal vascular physiologic develop-
ment, where ESR2 knock-out mice dem-
onstrate significant systolic and diastolic
hypertension.20 Likewise, within endo-
thelial cell cultures from other organs,
both estradiol and an ESR2-specific
agonist have been found to upregu-
late cyclooxygenase-2 (COX-2) inde-
pendently, which leads to increased sub-
strate (ie, PGH2) for the biosynthesis of
the various downstream prostanoids.21-23
We previously demonstrated a
method of isolating endothelial cells
from the fetoplacental vessels; of the 2
estrogen receptors, these cells express
only ESR2.22 The role of estrogen and
ESR2 in balancing vascular prostanoid
biosynthesis remains unclear, both
within the fetoplacental compartment
and within others. Our objective was to
A, Villous endothelial cells demonstrate positivity for CD31, an endothelial cell–specific antigen. B, determine the role of ESR2 in the medi-
These same villous endothelial cells exhibit endogenous expression of ESR2 within the nuclei ation of fetoplacental vascular function,
(magnified inset shows nuclear staining). C, In contrast, there is no evidence of ESR1 expression and we hypothesized that ESR2 upregu-
within term placental sections. D, Breast cancer tissue was used as a positive control for this ESR1 lates COX-2. This, in turn, may lead to
antibody. Of note, there was also positive staining of ESR2 within the nuclei of villous syncytiotro- an alteration in the proper PGI2:TXA2
phoblast. ratio in an in vivo setting, which poten-
Su. ESR2 and villous endothelial cells. Am J Obstet Gynecol 2009. tially leads to a vicious cycle of vessel dys-
function and injury.
ceptors in target cells.8 Although PGE2 tional age–matched normal neonates.15
and PGF2 can contribute to vascular Furthermore, placental TXA2 produc- M ATERIALS AND M ETHODS
phenotype, PGI2 and TXA2 appear to tion is increased in chronic placental in- Cellular isolation and culture
play more vital roles within vascular sufficiency states.16 A balance between Human placental villous endothelial cell
physiology.9 total vascular PGI2 and TXA2 may very isolation was performed, as previously
Although platelet production of TXA2 well regulate the fetoplacental vasculature. described, after approval by the institu-
is well established, recent data have dem- Within the human vasculature, estro- tional review board at Northwestern
onstrated endothelial cell– derived syn- gen plays a vital role in blood vessel University and patient consent.22,24,25
thesis of this mediator.10-13 With the homeostasis. During pregnancy, the Cells were isolated from placentas from
ability of the vasculature itself to synthe- human placenta produces dramatic uncomplicated pregnancies immedi-
size both vasoconstricting and vasodilat- amounts of the various estrogens (ie, es- ately after delivery. None of the subjects
ing mediators, an imbalance between tradiol, estrone, estriol). Estrogens affect were exposed to aspirin or other nonste-
total TXA2 and PGI2 may lead to inap- numerous basic cellular functions that roidal medications throughout preg-
propriate vasoconstriction and endothe- include gene expression, cellular prolif- nancy. Immunofluorescence confirmed
lial cell and platelet activation. For in- eration, and cellular differentiation. purity of the cells (data not shown);
stance, PGI2 limits the vasoconstrictive Their function occurs by activation of 1 based on previous data, primary cells
response to TXA2 within the cardiovas- or both of 2 estrogen receptors (estrogen were used only through the fifth passage
cular system.14 From an obstetric per- receptor-alpha [ESR1] and estrogen re- to avoid changes in phenotype.25
spective, neonatal PGI2 production is ceptor-beta [ESR2]).17,18 These estrogen Cells were cultured and treated with
lower in those neonates with intrauter- receptors are expressed in a wide variety the use of phenol red-free media that was
ine growth restriction because of chronic of tissue that includes endothelial cells supplemented with 5% fetal bovine se-
placental insufficiency than in gesta- and vascular smooth muscle cells. With rum, bovine brain extract with heparin,
427.e2 American Journal of Obstetrics & Gynecology APRIL 2009

3. www.AJOG.org Basic Science: Obstetrics Research
epidermal growth factor, hydrocorti- burg, MD). Specific oligodeoxynucleotide performed after completion of all auto-
sone, and gentamicin/amphotericin B primers for COX-2 were synthesized based radiographic studies with ImageJ soft-
(Lonza, Walkersville, MD). Cells were on its published cDNA sequence (F: ware (National Institutes of Health, Be-
starved in serum-free medium and 5=-GAATCATTTGAAGAACTTACAG- thesda, MD).
treated with vehicle (ethyl alcohol GAG – 3=; R: 5=-GAGGCTTTTCTACCA-
1:1000; Sigma-Aldrich, St. Louis, MO), GAAGG – 3=). Primers against ESR2, aro- RNA interference
estradiol (10–11 to 10– 6 mol/L; Sigma-Al- matase, and the constitutively expressed RNA oligonucleotides that were directed
drich), the ESR2-specific agonist diaryl- 36B4 were also used as described in previ- against ESR2 and a mismatch negative
propionitrile (10–11to 10– 6 mol/L; Tocris ous reports.26-28 Primer specificity was control small interfering RNA (siRNA)
Bioscience, Ellisville, MO), or lipopoly- confirmed by single peaks demonstrated were purchased from Invitrogen. Pla-
saccharide (from Escherichia coli 026:B6 by dissociation curves after amplification cental endothelial cells were cultured in
100 ng/mL; Sigma-Aldrich). of cDNA and a lack of amplification of media as previously described, but
All experiments were performed on at genomic DNA. lacked gentamicin/amphotericin B. Cells
least 3 representative subject samples, Real-time quantitative polymerase were plated at a density of 4.0 106 cells
with each repeated in triplicate, with the chain reaction (PCR) was used to deter- per 10-cm dish 1 day before transfection
use of cells between the first and fifth pas- mine the relative amounts of each tran- to achieve approximately 30-50% con-
sage. The results of all the experiments script with the use of the DNA-binding fluence at the time of transfection. On
were pooled; Western blots demonstrate dye SYBR green (Applied Biosystems, the day of transfection, the RNAiMAX
representative images from 1 selected Foster City, CA) and the ABI Prism lipofectamine-based reagent (Invitro-
subject. 7900HT Detection System (Applied Bio- gen) was combined in conjunction with
systems). Cycling conditions started at 200 nmol/L siRNA duplexes that were
Primary antibodies 50°C for 2 minutes followed by 95°C for diluted in Opti-Mem I (Invitrogen) and
The following antibodies were used 10 minutes, then 40 cycles of 95°C for 15 applied to the cells. Six hours after the
for immunohistochemistry: monoclonal seconds and 60°C for 1 minute. The cycle start of transfection, complete growth
antibodies against ESR1 (Dako, Carpin- threshold (Ct) was placed at a set level at medium without antibiotics was added,
teria, CA) and endothelial cell-specific which the exponential increase in PCR and cells were allowed to recover and
antigen CD31 (Dako); and polyclonal amplification was approximately parallel proliferate. Of note, additional controls
antibody against ESR2 (BioGenex, San between all samples. Relative fold- were performed simultaneously on cells
Ramon, CA). For immunoblotting, the change was calculated by a comparison from 3 of the subjects. These controls in-
following antibodies were used: mono- of Ct values between the target gene and cluded cells that were exposed only to
clonal antibodies against beta-actin 36B4 as the reference guide. The 2– Ct Opti-Mem I media and cells undergoing
(Sigma-Aldrich), ESR2 (Millipore, Bil- method was used to analyze these rela- mock transfection with exposure to
lerica, MA), and COX-2 (Cell Signaling tive changes in gene expression.29 RNAiMAX lipid reagent in the absence
Technology, Danvers, MA). of RNAi oligos. Cells were starved over-
Immunohistochemistry Protein isolation and immunoblotting night, followed by treatment with vehicle
Placentas from uncomplicated pregnan- Placental endothelial cells were lysed or estradiol (10–7 mol/L) for 24 hours.
cies were obtained after delivery and with the use of Mammalian Protein Ex- This was timed to allow for RNA and
fixed, paraffin-embedded, and processed traction Reagent (M-PER; Pierce, Rock- protein isolation at 48 and 72 hours, re-
by the Pathology Core Facility at North- ford, IL). Protein concentrations were spectively, from the start of transfection.
western University. Antigen retrieval determined by colorimetric BCA Protein
was performed with citrate buffer, and Assay (Pierce); equal concentrations Enzyme immunoassay
the primary antibodies that were de- were loaded in each well. Samples were Villous endothelial cells were treated
scribed earlier were used. Immunoreac- subjected to polyacrylamide gel electro- with vehicle or estradiol or prepared for
tivity was determined with the use of phoresis (Bio-Rac, Hercules, CA) and RNA interference studies as described
horseradish peroxidase-conjugated sec- transferred onto nitrocellulose mem- earlier. Cells were serum- and supple-
ondary antibody, followed by addition of branes (Invitrogen). Membranes were ment-starved in basal media overnight at
diaminobenzidine substrate. probed with antibodies as described 36 hours from the start of transfection.
earlier. Antirabbit and antimouse im- The medium was replaced the next
RNA isolation and real-time munoglobulin G that was conjugated to morning. Twenty-four hours later, cell
polymerase chain reaction horseradish peroxidase (Cell Signaling culture supernatant was collected, cen-
Total RNA from primary endothelial cell Technology) were used as secondary an- trifuged to remove any cellular contam-
cultures was extracted with Tri-Reagent tibodies. Immunoreactive bands were inants, and transferred to a fresh tube.
(Sigma-Aldrich). One microgram of RNA visualized with an enhanced chemilumi- Enzyme immunoassays were performed
was reverse transcribed with the Q-script nescence detection system (GE Health- for thromboxane B2 (TXB2; main me-
Flex complementary DNA (cDNA) Syn- care, Piscataway, NJ). Quantification of tabolite of TXA2) and 6-keto-prosta-
thesis Kit (Quanta Biosciences, Gaithers- chemiluminescence signal intensity was glandin F1 alpha (6-keto-PGF1 ; main
APRIL 2009 American Journal of Obstetrics & Gynecology 427.e3

4. Research Basic Science: Obstetrics www.AJOG.org
FIGURE 2
Cyclooxygenase-2 response to estradiol treatment
A, Real-time PCR demonstrates minimal induction of COX-2 mRNA levels; the 10–7 mol/L dose yields a 1.05-fold change. aP .03. B, Estradiol time-course
demonstrates statistically significant induction of COX-2 mRNA at 1, 4, and 8 hours. bP .05. With the exception of 1.62-fold induction at 1 hour (P .05),
COX-2 mRNA induction varies between 0.98-fold and 1.06-fold change at all other time points. C, COX-2 protein expression is inducible with lipopolysaccharide
(LPS) treatment, but not with estradiol or the ESR2-agonist diarylpropionitrile (DPN). D, Graphic depiction of COX-2 autoradiographic intensities with the use
of ImageJ software confirms that there is significant induction of COX-2 protein with lipopolysaccharide. cP .0005.
Veh, vehicle.
Su. ESR2 and villous endothelial cells. Am J Obstet Gynecol 2009.
metabolite of PGI2) with a commer- R ESULTS cular endothelium of certain organs.23,30
cial kit that used competitive substrate- In vivo distribution of ESR2 Within our model of fetoplacental endo-
acetylcholinesterase assays (Cayman in term placentas thelial cells, there was no consistent
Chemical, Ann Arbor, MI). Concentra- Cellular distribution of ESR1 and ESR2 COX-2 induction with estradiol treat-
tions were normalized to total protein was evaluated by immunohistochemistry ment in a dose- or time-dependent fash-
concentrations. in uncomplicated, term placentas. Immu- ion (Figure 2, A and B). This was repro-
noreactive ESR2 was detectable readily in duced with another less potent ESR2
Statistical analysis the nuclei of the villous endothelial cells ligand, estrone, and diarylpropionitrile
The data from all experiments were and syncytiotrophoblast (Figure 1). Sur- treatment (data not shown). We did note
pooled; numeric data are reported as prisingly, ESR1 was not detected. Of note, a 1.62-fold increase in COX-2 messenger
means of the 3 replicates performed human breast cancer tissue was used as a RNA (mRNA) after 1 hour of estradiol
within 1 subject, with error bars that rep- positive control for ESR1. treatment (P .05). Similarly, statistical
resent SEM. Statistical analysis for com- COX-2 is not induced by estradiol analysis also demonstrated significant
parison of treatment groups was per- or diarylpropionitrile in villous changes with 10–7 mol/L treatment at 4
formed with the Student t test or analysis endothelial cells and 8 hours, although the maximal fold-
of variance followed by the Scheffe mul- Previous studies have demonstrated es- change that was seen was only 1.06. Al-
tiple comparison test, when appropriate. tradiol induction of COX-2 within vas- though statistically significant, this de-
427.e4 American Journal of Obstetrics & Gynecology APRIL 2009

5. www.AJOG.org Basic Science: Obstetrics Research
gree of induction was minimal and of
FIGURE 3
uncertain relevance from a mechanistic
ESR2 knock-down effects on cyclooxygenase-2 expression
standpoint. From a protein standpoint,
there did not appear to be any COX-2
induction with estradiol or diarylpropio-
nitrile, whereas COX-2 protein levels
were inducible when lipopolysaccharide
was used as a positive control (Figure 2,
C and D).
Human placental villous endothelial
cells do not express endogenous
aromatase
Endothelial cell aromatase expression
has been described previously and ap-
pears organ dependent.31 With real-time
PCR, aromatase cycle threshold expres-
sion was undetectable in our villous en-
dothelial cell model in 5 subjects. Of
note, cDNA from MCF-7 breast cancer
cells that were used as a positive control
demonstrated the presence of low, but
detectable, expression (average Ct value,
34.98).
COX-2 expression is decreased
in the setting of ESR2 knock-down
The lack of COX-2 induction with estra-
diol treatment or in the setting of an
ESR2-specific agonist was surprising. To
confirm that estrogen and ESR2 did not
mediate COX-2 induction within villous
endothelial cells, we used RNA interfer-
ence to knock down ESR2 within our
cultured villous endothelial cells. Trans-
fection of ESR2 siRNA led to a consistent
knock-down of ESR2 mRNA by no A, Real-time PCR demonstrates that ESR2 mRNA expression is decreased by no less than 84% after
84%, and these results were not altered ESR2 knock-down in comparison with transfection with nontarget, mismatch siRNA. aP .0001.
by estradiol treatment (P .0001; Figure This effect occurs regardless of vehicle or estradiol (E2) treatment. B, COX-2 mRNA levels are also
3, A). Similarly, COX-2 mRNA levels ablated significantly in the setting of ESR2 knock-down; again, this appears independent of estradiol
also were ablated significantly in the set- treatment. bP .001. C and D, Representative Western blots with graphic depiction of pooled data
ting of ESR2 knock-down; again, this ap- show that there are no significant differences in COX-2 or ESR2 protein levels in the absence of
peared independent of estradiol treat- transfection reagent with exposure to lipid transfection reagent alone or transfection with nontarget
ment (P .001; Figure 3, B). Although siRNA (lanes 1-3). In contrast, there is a trend toward ESR2 protein expression decrease with ESR2
ESR2 protein knock-down was less dra- knock-down with a concomitant significant decrease in COX-2 protein (lane 4). As shown in lanes
matic, pooled image analyses of all West- 5-8, COX-2 protein levels are decreased significantly in the setting of ESR2 knock-down, regardless
ern blots demonstrated an approximate of vehicle treatment (V) or estradiol (E2) treatment. cP .06; dP .05.
50% knock-down in ESR2 protein (Fig- Su. ESR2 and villous endothelial cells. Am J Obstet Gynecol 2009.
ure 3, C and D). This led to concomitant
decreases in COX-2 protein levels that
occurred in the presence and absence of TXB2 and 6-keto-prostaglandin endothelial cells and platelets locally and
treatment (P .05; Figure 3, C and D). F1 levels are decreased exerts its effect in an autocrine or para-
Of note, there was no effect noted on after ESR2 knock-down crine manner.8,32 It then is converted
COX-2 or ESR2 expression in the absence TXA2, which is one of the main prosta- rapidly (nonenzymatically) to its metab-
of transfection conditions or in the pres- noids responsible for vasoconstriction olite TXB2.33 Similarly, PGI2, which is
ence of lipid transfection reagent alone. and platelet activation, is formed by both synthesized primarily by endothelial
APRIL 2009 American Journal of Obstetrics & Gynecology 427.e5

6. Research Basic Science: Obstetrics www.AJOG.org
within the 12 subjects without significant
FIGURE 4
COX-2 induction, lipopolysaccharide
6-keto-PGF1 and TXB2-concentrations after ESR2 interference was able to stimulate a response, thereby
demonstrating the potential to capture
COX-2 induction within our model.35
Another possible explanation was that
endogenous estrogen exists within the
culture system. Certain endothelial cells
are capable of synthesizing estrogen
through aromatase; in our system, this
may have negated any further exogenous
estradiol that was added to the culture
medium. To test this hypothesis, we per-
formed real-time PCR using validated
In the absence of estradiol, ESR2 knock-down leads to significant decreases in endothelial cell
aromatase primer-probes on endothelial
contributions of both A, 6-keto-PGF1 (aP .0001) and B, TXB2 (bP .005). The percent change
cell cDNA from 5 separate subjects. Aro-
in both is similar (67% vs 64%).
Su. ESR2 and villous endothelial cells. Am J Obstet Gynecol 2009.
matase Ct values were undetectable in all,
whereas cDNA from MCF-7 breast can-
cer cells that were used as a positive con-
cells and leads to vascular smooth mus- sults in this study confirm endogenous trol demonstrated the presence of low,
cle relaxation, is converted rapidly to its expression of ESR2 within villous endo- but detectable, expression. Without the
metabolite 6-keto-PGF1 .32 Enzyme thelial cells and negate the possibility of presence of endogenous estrogen, an-
immunoassays of cells that were treated induction from the isolation process it- other possibility was that our in vitro
with a dose-dependent course of estra- self. Thus, with the presence of ESR2, it culture system was inadvertently and in-
diol did not demonstrate any significant was anticipated that estrogenic com- consistently affecting coactivators or
changes in TXB2 or 6-keto-PGF1 levels pounds would induce COX-2 within vil- corepressors. This could explain the fact
(data not shown). To elucidate down- lous endothelial cells. that 5-fold induction was seen in 2 sub-
stream effects of ESR2 and COX-2 abla- It should be noted that there did ap- jects. The final possibility was that ESR2
tion, we performed enzyme immunoas- pear to be a significant induction of itself was not an actual mediator of
says for TXB2 and 6-keto-PGF1 in the COX-2 at 1 hour. However, this value COX-2 and prostanoid biosynthesis. To
setting of nontarget siRNA transfection was driven primarily by results from 2 address this, we performed RNA inter-
and ESR2 siRNA transfection in the ab- subjects that demonstrated approxi- ference studies knocking down ESR2;
sence of treatment. These results demon- mately 5-fold induction of COX-2 at 1 surprisingly, we found striking decreases
strated corresponding decreases in both hour. Within the remaining 12 subjects, in COX-2 mRNA and protein levels with
TXB2 and 6-keto-PGF1 concentra- fold-induction varied from 0.92-1.31 at ESR2 knock-down. This occurred both
tions in cell culture supernatant (P 1 hour, which was unlikely to be biolog- in the basal untreated state and under es-
.0001 and P .005, respectively; Figure 4). ically relevant. Similarly, there was statis- tradiol-treatment conditions.
tical significance noted at the 10–7 mol/L The implications of ESR2 mediation
C OMMENT dose point, the 4-hour time point, and of COX-2 within fetoplacental endothe-
In this study, we found that, unlike other the 8-hour time point. Of these 3 specific lial cells, even in the untreated basal
endothelial cells such as those within points, maximal fold-induction was state, are important for a number of rea-
uterine vasculature, COX-2 was not in- 1.06, which again, although statistically sons. It is possible that ESR2 is a major
duced reliably by direct estrogenic stim- significant, likely did not carry biologic regulator of COX-2, where a 50% decre-
uli within villous endothelial cells. Yet, 1 significance. ment in expression of a major transcrip-
of the main receptors for the various es- The discrepancy between subjects at 1 tion factor may be adequate to induce
trogens, ESR2, appears to be an impor- hour was surprising, especially because significant changes to downstream
tant mediator in the prostanoid biosyn- ESR2 was expressed endogenously, and genes.36 Furthermore, to our knowledge,
thesis pathway. various possible explanations were con- ligand-independent regulation of vascu-
We previously demonstrated ESR2 ex- sidered. First, nongenomic induction of lar function and vasomotor tone has not
pression as the sole estrogen receptor COX-2 was considered, and treatment at been described previously, and this pos-
within total cellular protein of cultured 5, 15, and 30 minutes was performed, sibility warrants further investigation.
villous endothelial cells.22 Other investi- with no induction noted of COX-2 (data To accomplish this, it will be important
gators, however, have described induc- not shown). To confirm that COX-2 was to rule out less traditional ligands such as
ible ESR2 expression under certain con- indeed inducible in villous endothelial estriol.37 It will also be essential to eluci-
ditions that include shear laminar cells of all subjects, lipopolysaccharide date the regions of the COX-2 promoter
stress.34 Our immunohistochemical re- was used as a positive control. Even that are stimulated by ESR2. Several po-
427.e6 American Journal of Obstetrics & Gynecology APRIL 2009

7. www.AJOG.org Basic Science: Obstetrics Research
tential candidate DNA-binding sites ex- noid biosynthesis. In the future, it will be standing of the pathophysiologic mech-
ist and include multiple estrogen re- important to investigate the direct effects anisms behind chronic placental insuffi-
sponse element (ERE) half-sites (eg, of ESR2 on TXA2 synthase and PGI2 ciency states. f
– 832/– 827, –1493/–1398, – 4671/– 4666) synthase gene expression. Furthermore,
based on results from the “Transcription even if ESR2 does not mediate these REFERENCES
Element Search System,” a computer- genes directly and affects TXA2 and 1. Baschat AA. Pathophysiology of fetal growth
assisted homology search.38 Although PGI2 production solely through control restriction: implications for diagnosis and sur-
veillance. Obstet Gynecol Surv 2004;59:
there do not appear to be any classic ERE of PGH2 substrate, investigation sur-
617-27.
consensus sequences (aGTTCAnnnT- rounding the effects of altered PGH2 2. Bilardo CM, Nicolaides KH, Campbell S.
GACCt), there is increasing evidence concentration within the vasculature is Doppler measurements of fetal and uteropla-
that nonclassic binding of transcription necessary and will help delineate the cental circulations: relationship with umbilical
factors to DNA occurs. For instance, 1 paracrine interactions between endothe- venous blood gases measured at cordocente-
sis. Am J Obstet Gynecol 1990;162:115-20.
group of investigators has shown that lial cells and its adjacent neighbors,
3. Weiner CP. The relationship between the
many natural EREs deviate substantially which include platelets and vascular umbilical artery systolic/diastolic ratio and um-
from the classic consensus sequence, smooth muscle cells. bilical blood gas measurements in specimens
which suggests that ESR binding to half- In summary, ESR2 is expressed endo- obtained by cordocentesis. Am J Obstet Gy-
sites may occur.39 Furthermore, the pos- genously within villous placental endo- necol 1990;162:1198-202.
4. Mitra SC, Venkataseshan VS, von Hagen S,
sibility of transcription factor cross-talk thelial cells and appears to mediate
Barton PT, Delshad G, Gil J. Morphometric
also exists, and multiple authors have COX-2 expression and TXA2 and PGI2 study of the placental vessels and its correlation
found circumstances in which estrogen production. These are novel findings for with umbilical artery Doppler flow. Obstet Gy-
receptors interact with other transcrip- 2 reasons. First, ESR2 appears to mediate necol 1997;89:238-41.
tion factors, which include activating COX-2 expression even in the absence of 5. Raio L, Ghezzi F, Di Naro E, Duwe DG, Cromi
A, Schneider H. Umbilical cord morphologic
protein-1 and stimulating protein 1.40 treatment, which suggests that a ligand-
characteristics and umbilical artery Doppler pa-
Other candidate binding sites have been independent mechanism may exist to rameters in intrauterine growth-restricted fe-
described in nonvascular tissue. For ex- regulate COX-2. Second, ESR2 regula- tuses. J Ultrasound Med 2003;22:1341-7.
ample, a variant nuclear factor- B site tion of TXA2 production has not been 6. Carter AM, Myatt L. Control of placental
within the proximal COX-2 promoter described previously. Although its medi- blood flow: workshop report. Reprod Fertil Dev
1995;7:1401-6.
has been described to be critical for ation of TXA2 may occur solely through
7. Dong YL, Vegiraju S, Chauhan M, et al. In-
COX-2 induction by malignant endo- its effect on COX-2 and substrate avail- volvement of calcitonin gene-related peptide in
metrial epithelial cells within endome- ability, it will be critical to further eluci- control of human fetoplacental vascular tone.
trial stromal cells.41 Furthermore, ESRs date potential paracrine interactions. Am J Physiol Heart Circ Physiol 2004;
have been shown to prevent nuclear fac- For instance, the rate of platelet TXA2 286:H230-9.
8. Smith WL. Prostaglandin biosynthesis and its
tor- B– dependent gene activation by release is highly modifiable and appears
compartmentation in vascular smooth muscle
preventing DNA-binding directly, which to be driven largely by enhanced sub- and endothelial cells. Annu Rev Physiol
may explain the decrease in COX-2 in- strate availability.45 Furthermore, within 1986;48:251-62.
duction with ESR2 interference.42 Other a population of subjects with unstable 9. Smith W. Cellular and subcellular compart-
response elements, which include a cy- angina whose platelet COX-1 has been mentation of prostaglandin and thromboxane
synthesis. In: Lands W, Biochemistry of arachi-
clic adenosine monophosphate response inhibited irreversibly by daily aspirin
donic acid metabolism. vol 1. Boston: Kluwer-
element and a CCAAT/enhancer-bind- therapy, 1 group of investigators has Nijhoff; 1985:246-7.
ing protein regulatory element have also demonstrated the ability of platelets to 10. Flavahan NA. Balancing prostanoid activity
been deemed important in COX-2 trans- transcellularly convert endothelial cell- in the human vascular system. Trends Pharma-
activation.43,44 With regard to the effects derived PGH2 to TXA2.46 Thus, it is pos- col Sci 2007;28:106-10.
11. Fulton CT, Stallone JN. Sexual dimorphism
of ESR2 on prostanoid biosynthesis it- sible that ESR2 mediation of COX-2 af-
in prostanoid-potentiated vascular contraction:
self, TXB2 levels were decreased in the fects PGH2 substrate availability; in an in roles of endothelium and ovarian steroids. Am J
setting of ESR2 knock-down, which sug- vivo setting in which multiple cellular in- Physiol Heart Circ Physiol 2002;283:H2062-73.
gests that ESR2 mediates endothelial cell teractions occur (including those be- 12. Ramadan FM, Upchurch GR Jr, Keagy BA,
TXA2 production. This regulation may tween endothelial cells and platelets), Johnson G Jr. Endothelial cell thromboxane
production and its inhibition by a calcium-chan-
occur through a direct effect on TXA2 substrate availability ultimately may
nel blocker. Ann Thorac Surg 1990;49:916-9.
synthase gene expression. However, be- have a more significant effect on total 13. Sellers MM, Stallone JN. Sympathy for the
cause 6-keto-PGF1 levels were compa- TXA2 values in comparison with total devil: the role of thromboxane in the regulation
rably decreased with ESR2 knock-down, PGI2 levels. In total, further investiga- of vascular tone and blood pressure. Am J
a more likely explanation of ESR2 medi- tion of the mechanisms that surround Physiol Heart Circ Physiol 2008;294:H1978-86.
14. Cheng Y, Austin SC, Rocca B, et al. Role of
ation on prostanoid biosynthesis is its ef- prostanoid biosynthesis and other hu-
prostacyclin in the cardiovascular response to
fect on COX-2 and the concomitant de- moral, vasomotor mediators within the thromboxane A2. Science 2002;296:539-41.
crease in PGH2 formation, which is the fetoplacental vasculature is warranted 15. Stuart MJ, Clark DA, Sunderji SG, et al. De-
main substrate for downstream prosta- and hopefully will help enhance under- crease prostacyclin production: a characteristic
APRIL 2009 American Journal of Obstetrics & Gynecology 427.e7

8. Research Basic Science: Obstetrics www.AJOG.org
of chronic placental insufficiency syndromes. agonistic versus antagonistic activities by gene 38. Schug J, Overton GC. TESS: transcriptional
Lancet 1981;1:1126-8. expression profiling in breast cancer cells. Can- element search software on the www: technical
16. Poston L. The control of blood flow to the cer Res 2004;64:1522-33. report. vol 2008. Philadelphia, PA. University of
placenta. Exp Physiol 1997;82:377-87. 28. Fujimoto J, Nakagawa Y, Toyoki H, Sak- Pennsylvania; 1997.
17. Kuiper GG, Enmark E, Pelto-Huikko M, Nil- aguchi H, Sato E, Tamaya T. Estrogen-related 39. O’Lone R, Knorr K, Jaffe IZ, et al. Estrogen
sson S, Gustafsson JA. Cloning of a novel re- receptor expression in placenta throughout receptors alpha and beta mediate distinct path-
ceptor expressed in rat prostate and ovary. gestation. J Steroid Biochem Mol Biol ways of vascular gene expression, including
Proc Natl Acad Sci U S A 1996;93:5925-30. 2005;94:67-9. genes involved in mitochondrial electron trans-
18. Walter P, Green S, Greene G, et al. Cloning 29. Livak KJ, Schmittgen TD. Analysis of rela- port and generation of reactive oxygen species.
of the human estrogen receptor cDNA. Proc tive gene expression data using real-time quan- Mol Endocrinol 2007;21:1281-96.
Natl Acad Sci U S A 1985;82:7889-93. titative PCR and the 2(-delta delta C[T]) method. 40. Bjornstrom L, Sjoberg M. Mechanisms of
19. Mendelsohn ME, Karas RH. The protective Methods 2001;25:402-8. estrogen receptor signaling: convergence of
effects of estrogen on the cardiovascular sys- 30. Akarasereenont P, Techatraisak K, Tha- genomic and nongenomic actions on target
tem. N Engl J Med 1999;340:1801-11. worn A, Chotewuttakorn S. The induction of cy- genes. Mol Endocrinol 2005;19:833-42.
20. Zhu Y, Bian Z, Lu P, et al. Abnormal vascu- clooxygenase-2 by 17beta-estradiol in endo- 41. Tamura M, Sebastian S, Yang S, et al. Up-
lar function and hypertension in mice deficient in thelial cells is mediated through protein kinase regulation of cyclooxygenase-2 expression and
estrogen receptor beta. Science 2002;295: C. Inflamm Res 2000;49:460-5. prostaglandin synthesis in endometrial stromal
505-8. 31. Murakami H, Harada N, Sasano H. Aro- cells by malignant endometrial epithelial cells. A
21. Gibson LL, Hahner L, Osborne-Lawrence matase in atherosclerotic lesions of human paracrine effect mediated by prostaglandin E2
S, et al. Molecular basis of estrogen-induced aorta. J Steroid Biochem Mol Biol 2001;
and nuclear factor-kappa B. J Biol Chem 2002;
cyclooxygenase type 1 upregulation in endo- 79:67-74.
277:26208-16.
thelial cells. Circ Res 2005;96:518-25. 32. Egan K, FitzGerald GA. Eicosanoids and the
42. Guzeloglu-Kayisli O, Halis G, Taskiran S,
22. Su EJ, Cheng YH, Chatterton RT, et al. vascular endothelium. Handb Exp Pharmacol
Kayisli UA, Arici A. DNA-binding ability of NF-
Regulation of 17-beta hydroxysteroid dehydro- 2006;176(pt 1):189-211.
kappaB is affected differently by ERalpha and
genase type 2 in human placental endothelial 33. Patrono C, Ciabattoni G, Pugliese F,
ERbeta and its activation results in inhibition of
cells. Biol Reprod 2007;77:517-25. Pierucci A, Blair IA, FitzGerald GA. Estimated
estrogen responsiveness. Reprod Sci 2008;15:
23. Tamura M, Deb S, Sebastian S, Okamura rate of thromboxane secretion into the circula-
493-505.
K, Bulun SE. Estrogen up-regulates cyclooxy- tion of normal humans. J Clin Invest 1986;
43. Tamura M, Sebastian S, Yang S, Gurates B,
genase-2 via estrogen receptor in human uter- 77:590-4.
ine microvascular endothelial cells. Fertil Steril 34. Lindner V, Kim SK, Karas RH, Kuiper GG, Fang Z, Bulun SE. Interleukin-1beta elevates
2004;81:1351-6. Gustafsson JA, Mendelsohn ME. Increased ex- cyclooxygenase-2 protein level and enzyme ac-
24. Jinga VV, Gafencu A, Antohe F, et al. Es- pression of estrogen receptor-beta mRNA in tivity via increasing its mRNA stability in human
tablishment of a pure vascular endothelial cell male blood vessels after vascular injury. Circ endometrial stromal cells: an effect mediated by
line from human placenta. Placenta 2000; Res 1998;83:224-9. extracellularly regulated kinases 1 and 2. J Clin
21:325-36. 35. Sunday L, Tran MM, Krause DN, Duckles Endocrinol Metab 2002;87:3263-73.
25. Lang I, Pabst MA, Hiden U, et al. Heteroge- SP. Estrogen and progestagens differentially 44. Zhu Y, Saunders MA, Yeh H, Deng WG, Wu
neity of microvascular endothelial cells isolated modulate vascular proinflammatory factors. KK. Dynamic regulation of cyclooxygenase-2
from human term placenta and macrovascular Am J Physiol Endocrinol Metab 2006; promoter activity by isoforms of CCAAT/en-
umbilical vein endothelial cells. Eur J Cell Biol 291:E261-7. hancer-binding proteins. J Biol Chem 2002;
2003;82:163-73. 36. Trukhacheva E, Lin Z, Reierstad S, Cheng 277:6923-8.
26. Chen D, Reierstad S, Lin Z, et al. Prosta- YH, Milad M, Bulun SE. ER{beta} regulates 45. Patrono C. The PGH-synthase system and
glandin E(2) induces breast cancer related ER{alpha} expression in stromal cells derived isozyme-selective inhibition. J Cardiovasc
aromatase promoters via activation of p38 and from ovarian endometriosis. J Clin Endocrinol Pharmacol 2006;47(suppl):S1-6.
c-Jun NH(2)-terminal kinase in adipose fibro- Metab 2008. E-pub ahead of print. 46. Camacho M, Vila L. Transcellular formation
blasts. Cancer Res 2007;67:8914-22. 37. Mishra RG, Stanczyk FZ, Burry KA, et al. of thromboxane A(2) in mixed incubations of
27. Frasor J, Stossi F, Danes JM, Komm B, Metabolite ligands of estrogen receptor-beta endothelial cells and aspirin-treated platelets
Lyttle CR, Katzenellenbogen BS. Selective es- reduce primate coronary hyperreactivity. Am J strongly depends on the prostaglandin I-syn-
trogen receptor modulators: discrimination of Physiol Heart Circ Physiol 2006;290:H295-303. thase activity. Thromb Res 2000;99:155-64.
427.e8 American Journal of Obstetrics & Gynecology APRIL 2009