BIOLOGY OF REPRODUCTION 65, 1565–1572 (2001)
Transcriptional Regulation of Cyclooxygenase-2 Gene in Ovine Large Luteal Cells1
Yuh-Lin Wu and Milo C. Wiltbank2
Endocrinology-Reproductive Physiology Program and Department of Dairy Science, University of Wisconsin,
Madison, Wisconsin 53706
ABSTRACT
There is positive feedback pathway in the ovine large luteal
cell, such that prostaglandin (PG) F2a stimulation induces intraluteal PGF2a production as the result of induction of one of the
rate-limiting enzymes in PG production, cyclooxygenase-2 (Cox2). The objective of the present study was to evaluate the intracellular effector systems and important DNA transcriptional element(s) involved in regulating the Cox-2 gene in ovine large
luteal cells. In transient transfection assays, Cox-2 promoter was
rapidly induced (4 h) by phorbol didecanoate (a protein kinase
[PK] C activator), ionomycin, and cloprostenol (PGF2a analogue), with a peak induction at 12 h. Cloprostenol-mediated
promoter activation was not blocked by inhibition of various
second messenger systems, including PKA, calcium calmodulin
kinase II, or mitogen-activated protein kinases. However, myristoylated PKC pseudosubstrate peptide inhibited cloprostenol
stimulation of Cox-2 promoter, indicating the critical role of
PKC in this stimulation. The Cox-2 promoter could be reduced
to 282 base pairs (bp) of the 59 flanking sequence with retention
of full inducibility by cloprostenol. Mutation of three critical cisresponsive elements within this 282-bp region (C/EBP, cAMP responsive element [CRE], and E-box) indicated that E-box was
critical in both basal and cloprostenol-induced promoter activity. However, there was also significant but less dramatic inhibition of cloprostenol stimulation by mutation of C/EBP and CRE
in the Cox-2 promoter, and mutation of all three elements eliminated cloprostenol induction of this promoter. Electrophoretic
mobility shift assays of nuclear extracts from large luteal cells
revealed that upstream stimulatory factor (USF)-1 and USF-2
bound to the E-box in Cox-2. Thus, PKC directly regulates transcription of the Cox-2 gene in large luteal cells by acting through
DNA elements close to the putative transcriptional start point,
particularly an E-box region at 250 bp.
corpus luteum, gene regulation, kinases, signal transduction
INTRODUCTION
An essential, rate-limiting step in the biosynthesis of
prostaglandins (PGs) is the conversion of arachidonic acid
to PGs of the G/H series by cyclooxygenase (Cox), also
known as PG G/H synthase. Two different isoforms of Cox
have been identified. The Cox-1 isoform is expressed in
many tissues under basal conditions and generally is not
inducible [1, 2]. In contrast, Cox-2 is usually absent and is
induced in many different cell types after diverse stimuli,
including mitogens, growth factors, cytokines, and tumor
promoters [1, 2]. The key role of Cox-2 in female reproSupported by NIH grant HD-32623.
Correspondence: Milo C. Wiltbank, University of Wisconsin-Madison,
1675 Observatory Drive, Madison, WI 53706. FAX: 608 263 9412;
e-mail: [email protected]
1
2
Received: 30 April 2001.
First decision: 29 May 2001.
Accepted: 28 June 2001.
Q 2001 by the Society for the Study of Reproduction, Inc.
ISSN: 0006-3363. http://www.biolreprod.org
duction is highlighted by the reduction in efficiency of
many reproductive processes in Cox-2 but not Cox-1
knockout mice, including ovulation, fertilization, implantation, and decidualization [3, 4]. There is a dramatic induction of Cox-2 but not Cox-1 expression in follicular
granulosa cells by the LH surge (hCG) or protein kinase
(PK) A activation in rats [5, 6], horses [7], and cows [8].
There is a substantial delay between hCG treatment of granulosa cells and detection of Cox-2 protein. For example, in
cattle, Cox-2 is first detected about 18 h after hCG or about
10 h prior to expected ovulation [7]. Sirois and Dore [7]
suggested that Cox-2 may be a determinant of the timing
of ovulation because of the surprisingly consistent interval
from induction of Cox-2 until ovulation in species that vary
widely in time from the LH surge until ovulation.
Prostaglandins are also involved in initiation of regression of the corpus luteum (CL) in many species because of
secretion of PGF2a from the uterus. Production of PGF2a
by luteal cells has also been documented in cows [9, 10],
pigs [11, 12], rhesus monkeys [13], sheep [14, 15], and
pseudopregnant rats [16]. Cyclooxygenase-2 mRNA was
induced in the CL [17] or in large luteal cells following
treatment with PGF2a [12, 15]. Thus, a small amount of
PGF2a from the uterus or another source could induce intraluteal production of PGF2a in an autoamplification cascade that may be critical for complete luteolysis [15]. Induction of Cox-2 by PGF2a in large luteal cells is very rapid
[12, 15], similar to the immediate early gene designation
of Cox-2 in other cells but different from the delayed expression pattern in granulosa cells. Thus, although large
luteal cells are generally regarded as having differentiated
from granulosa cells, it appears that regulation of Cox-2
changes from a delayed PKA-dependent regulation in granulosa cells to a rapid PKC-dependent regulation in large
luteal cells [18].
The Cox-2 promoter sequence has been analyzed in rats
[19, 20], mice [21], humans [22–24], and cows [25]. A
variety of cis-responsive elements, e.g., NF-kB, interferonresponse elements, C/EBP (NF-interleukin 6), cAMP response element (CRE), and E-box (-CACGTG-), have been
proposed as key regulatory elements that mediate Cox-2
transcription in various cell types. In rat granulosa cells [19,
20], either C/EBP or E-box are essential elements, but Ebox is the critical regulatory element in bovine granulosa
cells [25]. In luteal cells, the critical regulatory elements of
the Cox-2 promoter have not yet been determined.
Binding of PGF2a to the PGF2a receptor activates a cascade of intracellular effector systems [26, 27], including
PKC, free intracellular calcium, and mitogen-activated protein kinase (MAPK). Pharmacologic activation of PKC increases Cox-2 mRNA, but whether this pathway or other
pathways directly alter luteal Cox-2 gene transcription has
not been addressed. In the present study, we used the 59
upstream region of the ovine Cox-2 gene combined with a
luciferase expression plasmid to investigate the intracellular
effector systems and critical DNA regulatory elements in-
1565
1566
WU AND WILTBANK
synchronized by insertion of a controlled internal drug device (CIDR) for
5 days. Superovulation was stimulated with i.m. injection of 1500 IU eCG
2 days before removal of the CIDR. Prostaglandin F2a (Lutalyse, Kalamazoo, MI) injection was given i.m. when the CIDR was removed. Nine
to 11 days later, the CL were removed by midventral laparotomy, decapsulated, sliced, and dissociated into cell suspensions by incubating at 398C
in a shaking water bath in the presence of collagenase (92.5 mg/ml) and
DNAase (50 mg/ml). Large and small luteal cells were separated by centrifugal elutriation as previously described [15]. The purity and concentration of cells were determined by hemacytometer. Cells with diameters
larger than 22 mm were designated large luteal cells. The purity of large
luteal cells in this project averaged 86.9% 6 4.3% (range, 79%–91%).
Large luteal cells were plated down overnight (50 000/well) in 24-well
tissue culture plates in M199 medium with 0.1% BSA (fraction V powder),
1 mg/ml insulin, and 1% FBS.
Isolation and Mutation of the 59 Flanking Sequence
of the Ovine Cox-2 Gene
FIG. 1. Schematic display of the 59 flanking sequence for the ovine Cox2 gene showing key cis-elements and sites chosen for site-directed mutagenesis. A) Potential cis-elements within the ovine Cox-2 promoter. The
CRE and E-box have an overlapping sequence of two nucleotides. The
random mutation sites are located upstream of C/EBP (random mutant 1)
or between C/EBP and CRE/E-box (random mutant 2). B) The primer sequences (sense strand) used for site-directed mutagenesis of C/EBP, CRE,
and E-box elements. The wild-type sequences for each of the elements
are underlined; lowercase letters indicate the mutated nucleotides.
volved in Cox-2 gene transcription in primary cultures of
ovine large luteal cells.
MATERIALS AND METHODS
Chemicals and Reagents
Fetal bovine serum (FBS) was obtained from Gibco BRL (Rockville,
MD). The luciferase assay kit, pGL2B plasmid, pGEM-T easy vector, T4
polynucleotide kinase, and restriction enzymes were purchased from Promega (Madison, WI). The MicroSpin G-50 column and Poly(dI-dC) oligonucleotides were purchased from Amersham Pharmacia (Piscataway,
NJ). SuperFect transfection reagent and endotoxin-free plasmid purification columns were from Qiagen (Valencia, CA). Intravaginal progesterone
implants (EAZI-BREED CIDR for sheep and goats) were from InterAg
(Hamilton, New Zealand). Pfu DNA polymerase was from Stratagene (La
Jolla, CA). ABI Prism BigDye terminator cycle sequencing kits were from
Applied Biosystems (Foster City, CA). Polyclonal antibodies against upstream stimulatory factor (USF)-1 and USF-2, donkey anti-rabbit IgGhorseradish peroxidase (HRP), and Jurkat whole cell lysates were from
Santa Cruz Biotechnology (Santa Cruz, CA). Cloprostenol was purchased
from Cayman Chemical Company (Ann Arbor, MI). The 18 different kinase inhibitors (KN-93, K-252a, U0126, Lavendustin C, PD 98059, SB
202190, KT5720, H-89, Staurosporine, H-8, 4-cyano-3-methylisoquinoline, myristoylated PKA pseudosubstrate peptide, calphostin C, chelerythrine chloride, Go 6976, RO-32-0432, bisindolylmaleimide I, myristoylated
PKC pseudosubstrate peptide) and the 3-isobutyl-1-methyl-xanthine were
all purchased from Calbiochem (San Diego, CA). Western blot transfer
membrane immobilon-P was from Millipore (Bedford, MA). Western blot
chemiluminescence reagent was purchased from NEN Life Science Products (Boston, MA). Unless otherwise specified, other chemicals and reagents used in these studies were purchased from Sigma (St. Louis, MO).
Animals and Cell Preparation
Western range or polypay ewes from Equity Livestock (Baraboo, WI)
were used in this study. To obtain large luteal cells, mature ewes were
To obtain the 59 flanking sequence of the ovine Cox-2 gene, primers
were designed that corresponded to 21553 base pairs (bp) and 161 bp of
the bovine Cox-2 genomic DNA sequence [25] (59-TGACTAGAGGAGAAAGGCTTCC-39, 59-GAGGAGGGCGGTGCGGAGTT-39). Ovine genomic DNA (100 ng) was amplified by polymerase chain reaction (PCR)
using these primers with Taq DNA polymerase. PCR was done by preheating to 958C for 1 min followed by 35 cycles of 1 min at 958C, 1 min
at 578C, and 1.5 min at 728C, followed by 10 min at 728C. The amplified
product was separated on a 0.7% agarose gel and visualized after staining
with ethidium bromide. The PCR product was treated twice with phenol/
chloroform, precipitated with 100% ethanol, and resuspended in 13 TrisEDTA buffer. The pure PCR product was ligated into a pGEM-T easy vector
through a TA cloning site. The ligation mixture was used to transform competent Escherichia coli (DH5a). Plasmid DNA, isolated from a positive colony, was subjected to DNA sequencing. Plasmid DNA (500 ng) was amplified with BSF primer (59-TTTTCCCAGTCACGACGTTG-39) or BSR
primer (59-TGTGGAATTGTGAGCGGATAA-39) using an ABI Prism
BigDye terminator cycle sequencing kit. The product was purified through
a MicroSpin G-50 column, lyophilized, and sequenced by the Biotech Center at the University of Wisconsin-Madison. The sequence was compared
to the bovine sequence from GenBank (AF031699). Based on the comparison, a second pair of sheep-specific primers was designed (59CTGGTACCTTTGTCTCACGTAGAATCAC-39; 59-AACTCGAGACAACTGGAAGCTAACGGAG-39) with two restriction enzyme sites, KpnI
(59-GGTACC-39) and XhoI (59-CTCGAG-39). These primers were used to
amplify the ovine sequence from sheep genomic DNA by PCR using Pfu
DNA polymerase. The product was digested with restriction enzymes KpnI
and XhoI and subcloned into a luciferase reporter vector pGL2B through
the KpnI and XhoI sites. The entire insert within the pGL2B vector was
subjected to DNA sequencing as described earlier, and the sequence was
submitted to GenBank (AF369795). Deletion mutants of the ovine Cox-2
promoter sequence were generated with different combinations of restriction enzyme digestion. Combinations of KpnI with one of the following
enzymes, PstI, SnaBI, SacI, or EagI, were used to delete a DNA region
between the restriction sites for each pair of enzymes. The digested plasmid DNA was eluted from the agarose gel to remove the deleted fragment.
The digested plasmids were precipitated with ethanol and ligated back to
form intact plasmids. The ligated plasmids contained different sizes of the
59 flanking sequence with deletions from the 59 end: 1046 bp, 770 bp, 461
bp, and 282 bp. Specific mutations were introduced into the C/EBP, CRE,
and E-box regions of the 282-bp sequence using the Quick-change kit
(Stratagene) (Fig. 1). Two random mutations were also introduced at 2192
and 290 bp. Plasmids containing Cox-2 upstream sequences (mutant or
different lengths of native) were amplified in bacteria and column purified
for transfection.
Experiment I
To examine the intracellular effector systems leading to activation of
Cox-2 transcription, large luteal cells were transfected with the Cox-2
promoter construct according to previously described methods used to optimize transfection efficiency in bovine granulosa cells [28]. A luciferasereporter plasmid (1 mg) containing the full-length Cox-2 59 flanking sequence was mixed with transfection reagent SuperFect (2 ml) to form a
DNA-SuperFect complex. Before transfection, large luteal cells that were
attached to 24-well plates were washed three times with M199 1 1% BSA.
The DNA-SuperFect mixture was added directly to cells. After 2 h of
incubation at 398C, the cells were washed three times with M199. Cells
COX-2 TRANSCRIPTION IN LARGE LUTEAL CELLS
were then incubated with 8-bromo-cAMP (1 mM; a PKA activator), phorbol didecanoate (PDD, 10 nM; a PKC activator), ionomycin (1 mM; increases free intracellular Ca21), or cloprostenol (100 nM; a PGF2a analogue). The dose of 8-bromo-cAMP was previously found to stimulate
Cox-2 transcription in bovine granulosa cells (data not shown). The doses
of PDD and cloprostenol were based on results of previous studies on
induction of Cox-2 in ovine large luteal cells [15, 18]. To measure luciferase activity in the cell lysates, medium was removed and cells were
washed once with 13 PBS and lysed with luciferase assay lysis reagent.
Cell lysates were centrifuged at 15 000 rpm for 5 sec, and the luciferase
activity in the supernatant was measured using a commercial luciferase
assay kit (Promega). Transfection efficiency was determined by direct
measurement of the number of transfected plasmids per cell using a PCR
method previously described [28]. Luciferase activity of samples was normalized by transfection efficiencies and expressed as n-fold increases over
promoterless (pGL2B) control.
Experiment II
To determine the intracellular pathways involved in PGF2a stimulation
of Cox-2 transcription, the expression plasmid containing the full-length
Cox-2 promoter construct was analyzed in the presence of various PK
inhibitors. After transfection, inhibitors were added to large luteal cells,
and 1 h later some cells were treated with cloprostenol (100 nM) for
another 12 h. All inhibitors except 3-isobutyl-1-methyl-xanthine (IBMX)
were evaluated in preliminary dose-response experiments to determine
doses that had no detectable cellular toxicity. All doses that were utilized
were greater than reported inhibitory constants (Ki) and most were fivefold
or greater than the reported Ki for the inhibitors. The inhibitors that were
evaluated and their respective final concentrations in the cell cultures were
calcium calmodulin kinase II inhibitors KN-93 (2.5 mM), K-252a (0.5
mM), and Lavendustin C (2.5 mM); MAPK inhibitors PD 98059 (10 mM),
SB 202190 (5 mM), and U0126 (1 mM); PKA inhibitors KT5720 (250
nM), H-89 (0.5 mM), H-8 (2.5 mM), 4-cyano-3-methylisoquinoline (50
nM), and myristoylated PKA pseudosubstrate peptide (25 mM); and PKC
inhibitors Saturosporine (100 nM), Calphostin C (0.5 mM), chelerythrine
chloride (1 mM), Go 6976 (1 mM), RO-32-0432 (0.5 mM), bisindolylmaleimide I (0.5 mM), and myristoylated PKC pseudosubstrate peptide
(25 mM). An inhibitor of cAMP phosphodiesterase, IBMX (1 mM), was
also used as a control. A dose-response evaluation for PGF2a was also
performed in the presence or absence of myristoylated PKA or PKC pseudosubstrate peptide (25 mM). Some of the large luteal cells transfected
with the full-length Cox-2 promoter plasmid were treated for 1 h with
inhibitor and then challenged with different concentrations of cloprostenol
(1, 3, 10, 30, or 100 nM) for another 12 h.
Experiment IIIa
To localize essential regions of the 59 flanking sequence mediating
Cox-2 transcription in large luteal cells, progressive deletion mutants of
the Cox-2 promoter construct were transfected into large luteal cells. The
cells were then treated with 8-bromo-cAMP (1 mM), PDD (10 nM), or
cloprostenol (100 nM) for 12 h. To further identify critical elements within
the Cox-2 promoter region, expression plasmids were prepared with mutations in the C/EBP, CRE, E-box, and two random sites (Fig. 1) and were
transfected into large luteal cells. Following transfection, the cells were
treated with PDD (10 nM) for 12 h.
Experiment IIIb
An electrophoretic mobility shift assay (EMSA) was used to define
potential transcription factors capable of binding to the E-box element in
the Cox-2 promoter region. Nuclear extracts were prepared according to
the method of Schreiber et al. [29]. A double-strand oligonucleotide containing the CRE/E-box (59-CAGTCATGCCGTCACGTGGGCTATTT-39)
was end-labeled with [g-32P]ATP using T4 polynucleotide kinase and purified in a MicroSpin G-50 column. The 32P-labeled probe was mixed with
nuclear extracts (5 mg) in 10 mM Hepes (pH 7.9) containing 50 mM KCl,
2.5 mM MgCl2, 10% glycerol (v/v), 0.1 mg/ml BSA, 1 mM dithiothreitol,
and 0.25 mg/ml poly(dI-dC) and incubated in the presence or absence of
unlabeled competitor oligonucleotide for 20 min at room temperature. The
unlabeled competitor sequences were either wild type (same as probe) or
had an E-box mutation (59-CAGTCATGCCGTCAatgaGGCTATTT-39) or
a CRE mutation (CAGTCATGCaagCACGTGGGCTATTT). For supershift assays, nuclear extracts were incubated with USF-1 or USF-2 antibodies for 10 min at room temperature prior to addition of the labeled
oligonucleotide. The mixtures were analyzed on a 4% acrylamide gel in
1567
22.5 mM Tris-borate containing 0.5 mM EDTA. The gels were dried and
analyzed with the Cyclone Storage Phosphor System (Packard Instrument
Company, Meriden, CT).
To examine USF-1 and USF-2 in nuclear extracts, nuclear proteins
(10 mg) were mixed in 20 ml 23 SDS-loading buffer, and samples were
steamed for 10 min. Samples were then loaded on a 10% polyacrylamide
gel and separated at a constant current of 30 mA for 2 h. Proteins were
transferred from gels to Immobilon-P membranes (Millipore) using the
miniprotean II gel transfer system (Bio-Rad, Hercules, CA). Following
transfer, blots were incubated in blocking buffer (10 mM Tris, pH 8.0,
150 mM NaCl, 0.05% Tween-20, and 5% nonfat dry milk) at 48C overnight. The blots were incubated with anti-USF-1 or anti-USF-2 antibodies (Santa Cruz Biotechnology) at 1:5000 dilution for 2 h at room temperature, followed by three washes (10 mM Tris, pH 8.0, 150 mM NaCl,
0.05% Tween-20). Anti-rabbit HRP (Santa Cruz Biotechnology) was
added at 1:10 000 dilution for 1 h at room temperature, followed by three
washes. The intensity of the bands was increased with enhanced chemiluminescence reagent (NEN Life Science Products). The developed blots
were exposed to x-ray film for 1 min.
Statistical Analyses
Results were analyzed by one-way ANOVA using the general linear
model of the Statistical Analysis System (SAS, Cary, NC) to test for treatment effects. If a significant F-test was detected by ANOVA, a Fisher
least significant difference (LSD) test was used to compare different treatments, using an a-value of 0.05. In experiment II, means were compared
for different inhibitors in the absence (control) or presence of cloprostenol.
In the experiment using different doses of cloprostenol, treatments (no
inhibitor, PKC pseudosubstrate peptide, or PKA pseudosubstrate peptide)
were compared only within each dose of cloprostenol (0, 1, 3, 10, 30, and
100 nM) by ANOVA and LSD. In experiment IIIa, luciferase activity from
different deletion mutants was compared with that of the full-length promoter within untreated (control) and 8-bromo-cAMP, PDD, and cloprostenol treatment groups. The luciferase activity for each group of different
site-directed mutants was compared with that of the full-length promoter
within control or PDD treatment groups.
RESULTS
Experiment I
In large luteal cells, there was a significant induction of
luciferase activity by PDD, ionomycin, and cloprostenol but
not by 8-bromo-cAMP (Fig. 2). The increase in promoter
activity induced by PDD, ionomycin, and cloprostenol was
more dramatic at 12 h (PDD: 303; ionomycin: 153; cloprostenol: 403) than at 4 h (PDD: 93; ionomycin: 73;
cloprostenol: 93).
Experiment II
In large luteal cells transfected with the full-length Cox2 promoter construct, the calcium calmodulin kinase II inhibitors (KN-93, K-252a, and Lavendustin C), MAPK inhibitors (PD 98059, SB 202190, and U0126), PKA inhibitors (KT5720, H-89, H-8, 4-cyano-3methylisoquinoline,
and PKA pseudosubstrate peptide), and cAMP phosphodiesterase inhibitor did not have any effect on either untreated
or cloprostenol-treated cells (Fig. 3). The PKC inhibitors
chelerythrine chloride, RO-32-0432, and PKC pseudosubstrate peptide significantly decreased the cloprostenol-mediated promoter activation but not the basal promoter activity (Fig. 4). However, the PKC inhibitors Staurosporine,
bisindolylmaleimide I, Calphostin C, and Go 6976 did not
alter either basal or cloprostenol-stimulated luciferase activity (Fig. 4). In the presence of different concentrations
of cloprostenol, there was a dose-dependent induction of
Cox-2 promoter activity by cloprostenol from 1 nM to 100
nM at both 4 h (Fig. 5A) and 12 h (Fig. 5B). The PKC
pseudosubstrate peptide, but not the PKA pseudosubstrate
peptide, dramatically reduced cloprostenol-mediated lucif-
1568
WU AND WILTBANK
FIG. 2. Functional analysis of ovine Cox-2 promoter in ovine large luteal
cells. Cells were transfected with the luciferase expression vector driven
by the Cox-2 promoter sequence (1500 bp) and treated with 8-bromo
cAMP (1 mM), PDD (10 nM), ionomycin (1 mM), or cloprostenol (100
nM) for 4 or 12 h. Data are normalized for transfection efficiency and
represent the mean 6 SEM for three different experiments, with each
treatment evaluated in triplicate wells in each experiment. Differences (P
, 0.05) between treatments and control within a time point are indicated
by an asterisk.
erase activity by 64%–68% at 4 h (Fig. 5A) and by 74%–
84% at 12 h (Fig. 5B).
Experiment IIIa
FIG. 3. Luciferase activity in lysates of ovine large luteal cells transfected
with luciferase expression vector driven by the ovine Cox-2 promoter
(1500 bp) and treated with or without cloprostenol (100 nM) for 12 h in
the presence or absence of inhibitors of calcium calmodulin kinase II,
MAPKs, and PKA. Data are normalized for transfection efficiency and
represent the mean 6 SEM for four experiments, with each treatment
evaluated in triplicate wells in each experiment. There were no differences (P , 0.05) between cells treated with inhibitors and untreated cells in
the control or cloprostenol-treated groups.
50-fold excess of unlabeled oligonucleotide that contained
an E-box mutation eliminated all bands except the uppermost band (Fig. 7, lane 6). Addition of a 50-fold excess of
unlabeled oligonucleotide containing a CRE mutation but
not an E-box mutation eliminated all bands (data not
Progressive deletion of the ovine Cox-2 promoter region
revealed that deletion from full-length 1500 bp to 282 bp
did not result in any change in basal or PDD- and cloprostenol-induced luciferase activity (data not shown). Cyclic
AMP did not alter luciferase activity from any deletion mutants (data not shown). In untreated large luteal cells, single
or combined mutations of C/EBP and CRE did not cause
any change in basal luciferase activity (Fig. 6). Mutation
of the E-box element reduced the basal luciferase expression (P , 0.05), and this reduction was similar whether it
was caused by the E-box mutation alone or by the E-box
mutation combined with mutations of the C/EBP and/or
CRE. In the presence of the PKC activator PDD, the scenario was more complicated. The PDD-stimulated luciferase activity was reduced (P , 0.05) by single mutations in
C/EBP (55%), CRE (32%), or E-box (68%). Combined mutations produced more dramatic inhibition of PDD-stimulated luciferase activity: C/EBP plus CRE (60%), C/EBP
plus E-box (78%), CRE plus E-box (83%), and C/EBP plus
CRE plus E-box (95%). The two random mutants did not
alter basal or PDD-stimulated luciferase activity (Fig. 6).
Experiment IIIb
There were four distinct bands that could be identified
after incubation of nuclear extracts from untreated or
PGF2a-treated large luteal cells with the 32P-labeled oligonucleotide containing the CRE and E-box element (Fig. 7,
lane 2). Addition of a 50-fold excess of unlabeled wildtype oligonucleotide to the incubation mixture completely
eliminated all of the bands (Fig. 7, lane 5). Addition of a
FIG. 4. Luciferase activity in lysates of ovine large luteal cells transfected
with luciferase expression vector driven by the ovine Cox-2 promoter
(1500 bp) and treated with or without cloprostenol (100 nM) in the presence of PKC inhibitors for 12 h. Data are normalized for transfection
efficiency and represent the mean 6 SEM for four experiments, with each
treatment evaluated in triplicate wells in each experiment. Differences (P
, 0.05) due to inhibitors in cloprostenol-treated cells are indicated by an
asterisk.
COX-2 TRANSCRIPTION IN LARGE LUTEAL CELLS
1569
FIG. 6. Identification of critical element(s) in ovine Cox-2 promoter mediating PDD-regulated luciferase activity in ovine large luteal cells. Large
luteal cells were transfected with the full-length construct (Cox-2/1500
bp), the 282-bp wild-type construct (Cox-2/282 bp), and various mutations of the 282-bp construct, including mutations of C/EBP, CRE, and Ebox and combinations of these mutations and two random mutations (see
Fig. 1). Following transfection, cells were treated with or without PDD
(10 nM) for 12 h. Data are normalized for transfection efficiency and
represent the mean 6 SEM for four experiments, with each treatment
evaluated in triplicate wells in each experiment. Differences (P , 0.05)
between mutants and full-length promoter construct in the presence or
absence of PDD are indicated by an asterisk.
only the uppermost band (Fig. 7, lanes 3 and 4). The banding pattern was similar whether nuclear proteins were from
untreated or PGF2a-treated large luteal cells (Fig. 7, A vs.
B). Western blotting analysis revealed no detectable differences in amounts of USF-1 or USF-2 proteins in nuclear
extracts from control or PGF2a-treated large luteal cells
(Fig. 8, A and B).
DISCUSSION
FIG. 5. The PKC pseudosubstrate peptide blocks PGF2a-mediated Cox-2
promoter activity in ovine large luteal cells at 4 (A) or 12 (B) h after
cloprostenol treatment. Ovine large luteal cells were transfected with the
luciferase expression vector driven by the ovine Cox-2 promoter (1500
bp). These cells were treated with or without the myristolated PKC pseudosubstrate peptide (25 mM) or the myristolated PKA pseudosubstrate
peptide (25 mM); 1 h later increasing doses of cloprostenol were added
for an additional 4 or 12 h. Data are normalized for transfection efficiency
and represent the mean 6 SEM for three experiments, with each treatment
evaluated in triplicate wells in each experiment. Differences (P , 0.05)
between inhibitory peptide treatments within each dose of cloprostenol
are indicated by an asterisk at 4 or 12 h.
shown). Use of a 50-fold excess of unrelated oligonucleotide as a competitor eliminated all bands except the uppermost band (data not shown). Incubation of the nuclear extract with antibodies against USF-1 or USF-2 eliminated
Prostaglandin F2a is the initiator of luteolysis in sheep.
An autoamplification pathway has been reported in which
PGF2a treatment will induce Cox-2 and PGF2a production
within large luteal cells, although no previous study has
defined the potential transcriptional mechanisms involved
in this pathway. In the present study, we established that
induction of Cox-2 mRNA by PGF2a in large luteal cells
is mediated by direct control of gene transcription induced
by a short (less than 300 bp) region of DNA immediately
59 to the Cox-2 gene. The action of PGF2a on Cox-2 transcription was mediated by three critical elements within this
Cox-2 promoter, with regulation through the PKC intracellular effector system.
A number of intracellular effectors are activated by
PGF2a in large luteal cells, including free intracellular calcium, PKC, and MAPKs [26, 27]. Treatment of ovine large
luteal cells with a PKC activator increased steady state concentrations of Cox-2 mRNA [15]. However, no previous
investigators have utilized PKC inhibitors to evaluate
1570
WU AND WILTBANK
FIG. 8. Western Blotting of USF-1 and USF-2 proteins in nuclear extracts
from ovine large luteal cells untreated (control) or treated with PGF2a (100
nM) for 1 h. The blot was analyzed with anti-USF-1 antibody (A) or antiUSF-2 antibody (B). Whole cell lysates from Jurkat cells are used as a
positive control for USF-2. Molecular masses of USF-1 (43 kDa) and USF2 (44 kDa) are indicated with an arrow.
FIG. 7. EMSAs using a 32P-labeled CRE/E-box oligonucleotide probe incubated with nuclear extracts from ovine large luteal cells that were untreated (A) or treated for 1 h with 100 nM cloprostenol (B). Lane 1 is
probe only without nuclear extract, and lanes 2–6 contain probe incubated with nuclear extracts. Lanes 3 and 4 had nuclear extract preincubated with antibodies against USF 1 (lane 3) or USF-2 (lane 4). Lane 5
contains labeled probe and nuclear extract incubated with a 50-fold excess of wild-type unlabeled probe, and lane 6 contains a 50-fold excess
of unlabeled probe with mutation in the E-box element. The uppermost
band that appears to be specific for E-box and USF proteins is indicated
by an arrow.
whether the PGF2a-induced increase in Cox-2 mRNA is
mediated through this pathway. We utilized a number of
different inhibitors in an attempt to dissect the pathways
involved in PGF2a induction of Cox-2 gene transcription.
Three different calcium calmodulin kinase inhibitors (KN-
93, K252a, and Lavendustin C) did not alter induction of
Cox-2 by PGF2a. Likewise, inhibition of MAPK kinase
(MEK inhibited by PD 98059), P38 MAPK (by SB
202190), or P42/44 MAPK (by U0126) did not alter PGF2a
induction of Cox-2 gene transcription in spite of the elegant
studies that clearly show that PGF2a activates the Raf/MEK
and P42/44 MAPK pathways [27]. In contrast, PKC inhibitors inhibit Cox-2 induction by PGF2a, with the most dramatic inhibition by the highly specific myristolated PKC
pseudosubstrate peptide. This pseudosubstrate peptide was
designed based on the autoinhibitory domains present in
the PKCa and PKCb isoforms [30]. This autoinhibitory
domain has a peptide sequence that resembles a PKC phosphorylation site (a set of basic amino acids), except an alanine is substituted for the serine or threonine residue that
would normally be the phosphate acceptor. Thus, inhibition
of PGF2a action by this highly specific PKC inhibitor is
taken as strong evidence that PKC is involved in Cox-2
activation in large luteal cells. We did not clearly delineate
which of the 12 identified PKC isoforms [31] are involved
in this action of PGF2a. However, this pseudosubstrate
should inhibit PKCa and PKCb, because the sequence exactly matches their pseudosubstrate sequences, and should
inhibit the PKCg isoform, which contains a pseudosubstrate
with only a single amino acid substitution [30]. Other subtypes of PKC show considerable variability in their pseudosubstrate sequences and may not be inhibited by this peptide [30].
Expression of the Cox-2 gene is regulated in a myriad
of different cell types by numerous regulators during many
physiologic and pathophysiologic processes. Many different DNA response elements (see Fig. 1) are utilized in a
remarkably cell-specific and activator-specific manner. For
example, interleukin-1b and dexamethasone acted through
the NF-kB pathway in human amnion-derived cells [32],
whereas the actions of platelet-derived growth factors and
dexamethasone did not utilize the NF-kB site in renal mesangial cells [33]. Neither basal nor PGF2a-induced expression of Cox-2 utilized the NF-kB site in ovine large luteal
cells because this site was eliminated from the shorter promoter sequence (282 bp), yet the promoter remained fully
functional. Other more distal response elements such as the
COX-2 TRANSCRIPTION IN LARGE LUTEAL CELLS
interferon response element [34] were critical for Cox-2
regulation in other cells but were not essential for basal or
PGF2a-induced Cox-2 expression in large luteal cells. Similar to the effect in large luteal cells, a short region of 59
DNA was sufficient for complete Cox-2 promoter activity
in granulosa cells [19, 20, 25], hepatocytes [35], microvascular endothelial cells [36], and osteoblastic cells [37].
There are a number of potential DNA regulatory elements
in the 282-bp promoter sequence. We focused on the Ebox, an adjacent CRE, and a C/EBP element because previous research indicated the regulatory importance of these
elements in other cell types.
The E-box is critical for basal and forskolin-induced expression of Cox-2 in bovine granulosa cells [25] and for
basal and LH- or GnRH-stimulated expression of Cox-2 in
rat granulosa cells [20]. Our results showed that E-box is
critical for basal and PGF2a-stimulated expression of the
Cox-2 gene in ovine large luteal cells. In the EMSA, nuclear proteins from large luteal cells specifically bound to
an oligonucleotide that spanned the E-box sequence. This
binding was eliminated by incubation with an unlabeled Ebox oligonucleotide but not with a similar oligonucleotide
with a mutated E-box. Disruption of the E-box band by
antibodies to either USF-1 or USF-2 indicates that both
USF-1 and USF-2 are involved in the complex formation
between E-box and large luteal cell nuclear extracts. The
USF proteins are basic-helix-loop-helix-leucine zipper transcription factors that have polypeptide structures similar to
those of Myc oncoproteins, and both Myc and USF bind
to E-box. Overexpression of c-Myc often contributes to rapid proliferation of tumor cells, whereas overexpression of
USF is known to inhibit cellular proliferation in a number
of cancer cell lines [38, 39]. There appears to be ubiquitous
expression of USF-1 and USF-2 in most tissues, with perhaps some differences among tissues in the ratio of USF-1
to USF-2 [40]. There was no detectable difference in USF
proteins between untreated and PGF2a-treated large luteal
cells using Western blotting techniques. In addition, there
were no differences in banding patterns between untreated
and PGF2a-treated large luteal cells in EMSA using labeled
oligonucleotides spanning the E-box region. Although similar binding of USF to E-box in transcriptionally active and
inactive cells seems paradoxical, it is consistent with the
results of the elegant studies of Qyang et al. [41] on USF/
E-box transcriptional activation. They showed that USF
proteins (USF-1 or USF-2) activated transcription in HeLa
cells but not in Saos-2 cells. The inactivity of USF proteins
in Saos-2 cells was not due to a lack of proper localization
of USF to the nucleus or lack of USF binding to the DNA
promoter. Their results are most consistent with mediation
of the transcriptional activity of USF by interaction with a
not yet identified coactivator that was inactivated or lacking
in Saos-2 cells [41]. Although the E-box region appears to
be crucial for either basal or PGF2a-induced Cox-2 gene
transcription in large luteal cells, factors other than binding
of USF proteins to E-box may be crucial for explaining Ebox-mediated transcriptional activation.
Mutations of the CRE and C/EBP elements did not reduce basal Cox-2 gene activity but did reduce PKC-mediated transcriptional activation. Mutation of all three critical
elements, E-box, CRE, and C/EBP, eliminated PGF2a induction of the Cox-2 gene. Thus, there was a combinatorial
action of all three cis-acting elements in regulating transcriptional activation of the Cox-2 gene in large luteal cells.
Combinatorial involvement of more than one DNA element
in Cox-2 transcription has been reported in other cell types,
1571
such as AP-2, C/EBP, and CRE in human microvascular
endothelial cells [36], CRE and AP-1 in osteoblastic cells
[37], CRE and C/EBP in macrophage cells [42], and C/
EBP and E-box in skin carcinoma cells [43].
This study provides clear evidence that PGF2a activates
Cox-2 gene transcription through the PKC signaling pathway. An E-box DNA element is critical for both basal and
PGF2a-induced Cox-2 gene transcription. However, CRE
and C/EBP DNA elements are also involved in PGF2a-induced Cox-2 gene transcription, acting in combination with
E-box. Further studies are needed on the proteins involved
in regulating Cox-2 transcription, particularly the proteins
that interact with USF/E-box, to precisely characterize the
transcriptional mechanisms involved in PGF2a induction of
Cox-2 in large luteal cells.
REFERENCES
1. Smith WL, DeWitt DL. Prostaglandin endoperoxide H synthase-1 and
-2. Adv Immunol 1996; 62:167–215.
2. Vane JR, Bakhle YS, Botting RM. Cyclooxygenase 1 and 2. Annu
Rev Pharmacol Toxicol 1998; 38:97–120.
3. Lim H, Paria BC, Das SK, Dinchuk JE, Langenbath R, Trzaskos JM,
Dey SK. Multiple female reproductive failures in cyclooxygenase 2deficient mice. Cell 1997; 91:197–208.
4. Dinchuk JE, Car BD, Focht RJ, Johnston JJ, Jaffee BD, Covington
MB, Contel NR, Eng VM, Collins RJ, Czernlak PM, Gorry SA, Trzaskos JM. Renal abnormalities and an altered inflammatory response in
mice lacking cyclooxygenase II. Nature 1995; 378:406–409.
5. Wong WY, Dewitt DL, Smith WL, Richards JS. Rapid induction of
prostaglandin endoperoxide synthase in rat preovulatory follicles by
luteinizing hormone and cAMP is blocked by inhibitors of transcription and translation. Mol Endocrinol 1989; 3:1714–1723.
6. Wong WY, Richards JS. Evidence for two antigenically distinct molecular weight variants of prostaglandin H synthase in the rat ovary.
Mol Endocrinol 1991; 5:1269–1279.
7. Sirois J, Dore M. The late induction of prostaglandin G/H synthase2 in equine preovulatory follicle supports its role as a determinant of
the ovulatory process. Endocrinology 1997; 38:4427–4434.
8. Sirois J. Induction of prostaglandin endoperoxide synthase-2 by human chorionic gonadotropin in bovine preovulatory follicles in vivo.
Endocrinology 1994; 135:841–848.
9. Milvae RA, Hansel W. Prostacyclin, prostaglandin F2a and progesterone production by bovine luteal cells during the estrous cycle. Biol
Reprod 1983; 29:1063–1068.
10. Pate JL. Regulation of prostaglandin synthesis by progesterone in the
bovine corpus luteum. Prostaglandins 1988; 36:303–315.
11. Guthrie HD, Rexroad CE Jr, Bolt DJ. In vitro synthesis of progesterone and prostaglandin F by luteal tissue and prostaglandin by endometrial tissue from the pig. Prostaglandins 1978; 16:433–440.
12. Diaz FJ, Crenshaw T, Wiltbank MC. Prostaglandin F2a induces distinct
physiological responses in porcine corpora lutea after acquisition of
luteolytic capacity. Biol Reprod 2000; 63:1504–1512.
13. Johnson MS, Ottobre AC, Ottobre JS. Prostaglandin production by
corpora lutea of rhesus monkeys: characterization of incubation conditions and examination of putative regulation. Biol Reprod 1988; 39:
839–846.
14. Rexroad CE Jr, Guthrie HD. Prostaglandin F2a and progesterone release in vitro by ovine luteal tissue during induced luteolysis. Adv
Exp Med Biol 1979; 112:639–644.
15. Tsai SJ, Wiltbank MC. Prostaglandin F2a induces expression of prostaglandin G/H synthase-2 in the ovine corpus luteum: a potential positive feedback loop during luteolysis. Biol Reprod 1997; 57:1016–
1022.
16. Olofsson J, Norjavaara E, Selstam G. Synthesis of prostaglandin F2a,
E 2 and prostacyclin in isolated copora lutea of adult pseudopregnant
rats through the luteal life-span. Prostaglandins Leukotrienes Essent
Fatty Acids 1992; 46:151–161.
17. Tsai SJ, Wiltbank MC. Prostaglandin F2a regulates distinct physiological changes in early and mid-cycle bovine corpora lutea. Biol Reprod
1998; 58:346–352.
18. Wu YL, Wiltbank MC. Differential regulation of prostaglandin endoperoxidesynthase-2 transcription in ovine granulosa and large luteal
cells. Prostaglandins Other Lipid Mediators 2001; 65:103–116.
19. Sirois J, Richards JS. Transcriptional regulation of the rat prostaglan-
1572
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
WU AND WILTBANK
din endoperoxide synthase 2 gene in granulosa cells. J Biol Chem
1993; 268:21931–21938.
Morris JK, Richards JS. An E-box region within the prostaglandin
endoperoxide synthase-2 (PGS-2) promoter is required for transcription in rat ovarian granulosa cells. J Biol Chem 1996; 271:16633–
16643.
Fletcher BS, Kujubu DA, Perrin DM, Herschman HR. Structure of
the mitogen-inducible TIS 10 gene and demonstration that TIS 10encoded protein is functional prostaglandin G/H synthase. J Biol
Chem 1992; 267:4338–4344.
Tazawara R, Xu XM, Wu KK, Wang LH. Characterization of the
genomic structure, chromosomal location and promoter of human
prostaglandin H synthase-2 gene. Biochem Biophys Res Commun
1994; 203:190–199.
Inoue H, Tanabe T. Transcriptional regulation of human prostaglandinendoperoxide synthase-2 gene in vascular endothelial cells. Adv Exp
Med Biol 1997; 407:139–144.
Inoue H, Tanabe T. Transcriptional role of the nuclear factor kappa B
site in the induction by lipopolysaccharide and supression by dexamethasone of cyclooxygenase-2 in U937 cells. Biochem Biophys Res
Commun 1998; 244:143–148.
Liu J, Antaya M, Boerboom D, Lussier JG, Silversides DW, Sirois J.
The delayed activation of the prostaglandin G/H synthase-2 promoter
in bovine granulosa cells is associated with down-regulation of truncated upstream stimulator factor-2. J Biol Chem 1999; 274:35037–
35045.
Wiltbank MC, Diskin MG, Niswender GD. Differential actions of second messenger system in the corpus luteum. J Reprod Fertil Suppl
1991; 43:65–75.
Chen DB, Westfall SD, Fong HW, Roberson MS, Davis JS. Prostaglandin F2a stimulates the Raf/MEK1/mitogen-activated protein kinase
signaling cascade in bovine luteal cells. Endocrinology 1998; 139:
3876–3885.
Wu YL, Wiltbank MC. Determination of transfection efficiency by
direct polymerase chain reaction. Anal Biochem 1999; 271:108–111.
Schreiber E, Matthias P, Muller MM, Schaffner W. Rapid detection
of octamer binding proteins with mini-extracts, prepared from a small
number of cells. Nucleic Acids Res 1989; 17:6419.
Eichholtz T, de Bont DB, de Widt J, Liskamp RM, Ploegh HL. A
myristoylated pseudosubstrate peptide, a novel protein kinase C inhibitor. J Biol Chem 1993; 268:1982–1986.
Way KJ, Chou E, King GL. Identification of PKC-isoform-specific
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
biological activities using pharmacological approaches. Trends Pharmacol Sci 2000; 21:181–187.
Wang Z, Tai HH. Interleukin-1b and dexamethasone regulate gene
expression of prostaglandin H synthase-2 via the NF-kB pathway in
human amnion derived WISH cells. Prostaglandins Leukotrienes Essent Fatty Acids 1998; 59:63–69.
Goppelt-Struebe M, Rehm M, Schaefers HJ. Induction of cyclooxygenase-2 by platelet-derived growth factor (PDGF) and its inhibition
by dexamethasone are independent of NF-kB/IkB transcription factors. Naunyn-Schmiedebergs Arch Pharmakol 2000; 361:636–645.
Blanco JCG, Contursi C, Salkowski CA, DeWitt DL, Ozato K, Vogel
SN. Interferon regulatory factor (IRF)-1 and IRF-2 regulate interferon
g-dependent cyclooxygenase 2 expression. J Exp Med 2000; 191:
2131–2144.
Callejas NA, Bosca L, Williams CS, Dubois RN, Martin-Sanz P. Regulation of cyclooxygenase 2 expression in hepatocytes by CCAAT/
enhancer-binding proteins. Gastroenterology 2000; 119:493–501.
Kirtikara K, Raghow R, Laulederkind SJF, Goorha S, Kanekura T,
Ballou LR. Transcriptional regulation of cyclooxygenase-2 in the human microvascular endothelial cell lines, HMEC-1: control by the
combinatorial actions of AP2, NF-IL-6 and CRE elements. Mol Cell
Biochem 2000; 203:41–51.
Okada Y, Voznesensky O, Herschman H, Harrison J, Pilbeam C. Identification of multiple cis-elements mediating the induction of prostaglandin G/H synthase-2 by phorbol ester in murine osteoblastic cells.
J Cell Biochem 2000; 78:197–209.
Koskine PJ, Alitalo K. Role of myc amplification and overexpression
in cell growth, differentiaton and death. Semin Cancer Biol 1993; 4:
3–12.
Luo X, Sawadogo M. Antiproliferative properties of the USF family
of transcription factors. Proc Natl Acad Sci U S A 1996; 93:1308–
1313.
Sirito M, Lin Q, Maity T, Sawadogo M. Ubiquitous expression of the
43- and 44-kDa forms of transcription factor USF in mammalian cells.
Nucleic Acids Res 1994; 22:427–433.
Qyang Y, Luo X, Lu T, Ismail PM, Krylov D, Vinson C, Sawadogo
M. Cell-type-dependent activity of the ubiquitous transcription factor
USF in cellular proliferation and transcriptional activation. Mol Cell
Biol 1999; 19:1508–1517.
Wadleigh DJ, Reddy ST, Kopp E, Ghosh S, Herschman HR. Transcriptional activation of the cyclooxygenase-2 gene in endotoxin-treated RAW 264.7 macrophages. J Biol Chem 2000; 275:6259–6266.
Kim Y, Fisher SM. Transcriptional regulation of cyclooxygenase-2 in
mouse skin carcinoma cells. J Biol Chem 1998; 273:27686–27694.