BIOLOGY OF REPRODUCTION 69, 1957–1963 (2003)
Published online before print 6 August 2003.
Bacterially-Induced Preterm Labor and Regulation of Prostaglandin-Metabolizing
Enzyme Expression in Mice: The Role of Toll-Like Receptor 41
Hao Wang and Emmet Hirsch2
Department of Obstetrics and Gynecology, Evanston Northwestern Healthcare, Feinberg School of Medicine,
Northwestern University, Evanston, Illinois 60201
ABSTRACT ﬂammatory stimulus to labor after bacterial exposure re-
main incompletely deﬁned.
Toll-like receptor 4 (TLR-4) is a critical mediator of the cel- Prostaglandins stimulate uterine contractions and cervi-
lular response to lipopolysaccharide. Our purpose was to ex- cal ripening during labor. Both prostaglandin E 2 (PGE 2)
amine the role of TLR-4 in parturition and in the regulation of and prostaglandin F2 (PGF2 ) are produced by maternal
expression of prostaglandin synthase (cyclooxygenase [COX]-1
and COX-2) and 15-hydroxyprostaglandin dehydrogenase
and fetal tissues during parturition, and the concentration
(PGDH) following exposure to heat-killed Escherichia coli (HKE)
of both increases in the amniotic ﬂuid during labor .
in pregnant mice. Inbred TLR-4-mutant C3H/HeJ mice and in- Administration of prostaglandin synthase inhibitors sup-
bred normal C3HeB/FeJ mice on Day 14.5 of a 19- to 20-day presses uterine activity and prolongs the length of pregnan-
gestation received intrauterine injection of either HKE or sterile cy . Primary prostaglandins are formed from arachi-
vehicle (PBS). Preterm or term delivery was recorded for these donic acid through activity of the cyclooxygenase (COX)
animals. Tissues (myometrium, decidual caps, placentas, fetal enzyme complex. The COX catalyzes the ﬁrst committed
membranes, and fetuses) were collected after injection of sterile step of prostaglandin synthesis, the initial conversion of
vehicle or 5 109 HKE bacteria (n 5 mice per strain per arachidonic acid to PGH2. Two isoforms of COX have been
treatment per time point). The COX-1, COX-2, and PGDH gene identiﬁed, COX-1 and COX-2, which are also known as
expression was determined by semiquantitative reverse tran- prostaglandin endoperoxide H synthase (PGHS)-1 and
scription-polymerase chain reaction. We found that 5 109 HKE PGHS-2, respectively. The COX-1 is constitutively ex-
induced preterm delivery in 100% of TLR-4-normal mice but in pressed in many tissues with little regulation in synthesis,
0% of TLR-4-mutant mice. The HKE exposure up-regulated ex- whereas COX-2 is inducible in response to a variety of
pression of COX-2, but not of COX-1, in maternal tissues in both growth factors and inﬂammatory stimuli. Mice lacking the
mouse strains. The prostaglandin-catabolizing enzyme PGDH gene for PGF2 receptor  or COX-1  have delayed
was down-regulated in myometrium, fetal membranes, and fe- onset of labor because of the roles of these factors in reg-
tuses in control mice, but no change was observed in TLR-4- ulating luteolysis and myometrial expression of oxytocin
mutant mice after HKE treatment. These results demonstrate that receptors. Recent studies demonstrate a dramatic increase
a functional TLR-4 is essential for HKE-induced preterm labor
and PGDH down-regulation but is not essential for HKE-induced
of COX-1, but not of COX-2, transcripts and activity in the
COX-2 gene up-regulation. The TLR-4 may mediate bacterially uterus  and fetal membranes  in mice during late
induced preterm labor via regulation of prostaglandin degra- gestation. In one report, both COX-1 and COX-2 expres-
dation rather than prostaglandin synthesis. sion within the uterus were signiﬁcantly altered within 2 h
of lipopolysaccharide (LPS) administration, with COX-2
gene regulation, parturition, pregnancy uterus increasing and COX-1 decreasing .
The NAD -dependent 15-hydroxyprostaglandin dehy-
INTRODUCTION drogenase (PGDH) is responsible for the initial inactivation
of prostaglandins, catalyzing the conversion of primary
Preterm birth occurs in 5–10% of all pregnancies and prostaglandins to their biologically inactive 15-keto deriv-
accounts for 70–75% of early neonatal morbidity and mor- atives. Expression and activity of PGDH have been dem-
tality . Approximately 30–40% of preterm births are as- onstrated in fetomaternal tissues of different species. Re-
sociated with an underlying infectious process. Systemic duced PGDH expression and activity in myometrium and
and local intrauterine infections have been implicated in the chorion have been suggested in association with term and
pathogenesis of preterm labor and delivery [2, 3]. Indeed, preterm birth in humans . However, in the mouse, PGDH
either local or systemic exposure to microbial products mRNA increased in placentas and fetal membranes 
leads to preterm birth in several animal models [4–8]. How- and decreased in uterus  during late gestation. These
ever, the signaling pathways leading from the inciting in- studies suggest that COX and PGDH in the fetomaternal
environment may play important roles in the initiation of
Supported by a grant from March of Dimes (6-FY99-908) and NIH labor, but the relative contributions of the maternal and the
(1RO1HD41689). fetal tissues to COX and PGDH activity during infection
Correspondence: Emmet Hirsch, Department of Obstetrics and Gyne- are still incompletely deﬁned.
cology, Evanston Northwestern Healthcare, 2650 Ridge Avenue, Evans-
ton, IL 60201. FAX: 847 733 5083; e-mail: email@example.com
We have reported a model of infection-induced preterm
birth in mice following intrauterine inoculation with live or
heat-killed Escherichia coli (HKE). Intrauterine inoculation
Received: 28 May 2003.
First decision: 14 June 2003.
of pregnant CD-1 mice with HKE on Day 14.5 of a 19- to
Accepted: 6 August 2003. 20-day gestation leads to dose-dependent preterm delivery.
2003 by the Society for the Study of Reproduction, Inc. This process mimics human infection-associated preterm
ISSN: 0006-3363. http://www.biolreprod.org labor in many important ways, such as the expression of
1958 WANG AND HIRSCH
proinﬂammatory cytokines [17–19]. We have demonstrated or HKE suspended in PBS were injected into the midsection of the right
signiﬁcant increases in interleukin (IL)-1, IL-6, and tumor uterine horn at a site between two adjacent fetuses. The abdominal incision
was then closed in two layers using interrupted 5-0 coated vicryl sutures
necrosis factor (TNF) within uteri and fetal membranes fol- through the peritoneum and staples at the skin.
lowing HKE treatment in this mouse model , as has To establish the dose-response relationship between bacterial exposure
been demonstrated in humans. However, several groups and preterm delivery, a group of 40 pregnant C3HeB/FeJ (TLR-4-normal)
have shown that blockade of IL-1 and/or TNF by using IL- mice and 20 pregnant C3H/HeJ (TLR-4-mutant) mice were monitored af-
1-receptor antagonist, soluble TNF-receptor Fc fusion pro- ter inoculation of variable amounts of HKE. These animals underwent
tein, or IL-1 and IL-1-receptor knockout mice does not pre- twice-daily observations in which health status was recorded. Mice that
delivered prematurely (deﬁned as the ﬁnding of at least one pup in the
vent preterm delivery in mice after administration of bac- cage or the lower vagina within 48 h of surgery) or delivered at term
teria or LPS [20–22]. Thus, it remains uncertain whether underwent autopsy when fetuses were found in the cage. Some mice not
inﬂammatory cytokines are critical mediators of the signals delivered prematurely underwent autopsy 72 h after surgery to determine
by which bacterial exposure causes labor. fetal status.
A gram-negative bacterial cell wall component, LPS can For measurement of COX and PGDH transcripts within fetomaternal
induce expression of inﬂammatory cytokines and prosta- tissues, a second group of pregnant C3HeB/FeJ and C3H/HeJ mice were
inoculated with either pyrogen-free PBS or HKE sufﬁcient to cause pre-
glandins and leads to preterm birth in pregnant mice [6, 8] term delivery in 100% of C3HeB/FeJ mice (5 109 organisms). Animals
and rats  following intraperitoneal injection or intra- were killed with carbon dioxide gas, and tissues were collected at 0, 1, 2,
uterine infusion. The current consensus is that LPS signal- and 4 h after surgery for C3HeB/FeJ mice or at 4 h after surgery for C3H/
ing occurs through a heterotrimeric receptor complex, of HeJ mice (5 animals per group per time point). Harvest times early during
which the toll-like receptor 4 (TLR-4) protein is a critical the course of infection were selected, because previous data showed that
component. A point mutation in the intracytoplasmic region levels of cytokines, enzymes, and transcription factors change in myo-
metrium in CD-1 mice within 4 h of inoculation with high-dose HKE
of TLR-4 is responsible for the 20- to 40-fold LPS hypo- . The abdomen was opened, and the injected uterine horn was cut
responsiveness of C3H/HeJ mice . The defective Lps away from its mesometrium and then incised longitudinally along the an-
allele affects the functions of several cell types, including timesenteric border. The gestational sacs and placentas were shelled out,
the macrophage, which on activation normally secretes a and the right uterine horns were then washed in ice-cold PBS. The decid-
large array of proinﬂammatory factors and is critical for the ual cap at each implantation site was removed by sharp dissection, leaving
innate immune response to bacterial pathogens. behind the myometrium. The individual sacs surrounding each fetus were
cut away, and fetus, fetal membranes, and placentas were washed in cold
In the present study, we used inbred pregnant C3HeB/ PBS. These specimens were then minced and immediately frozen in liquid
FeJ mice (TLR-4-normal) and C3H/HeJ (TLR-4-mutant) nitrogen. Decidual caps, fetal membranes, placentas, and fetuses were
mice to test the hypothesis that E. coli induces preterm pooled by tissue for each pregnancy.
labor via TLR-4 and to characterize the role of TLR-4 in
the fetal and maternal expression of prostaglandin meta- RNA extraction
bolic enzymes during the early phase of bacterially induced
preterm labor. Total RNA was extracted from tissue specimens after homogenization
in TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufac-
turer’s instructions. The quantity and quality of the RNA was veriﬁed by
MATERIALS AND METHODS spectrophotometry and formaldehyde gel electrophoresis, respectively.
Reverse Transcription-Polymerase Chain Reaction
All animals were treated in accordance with the Guide for Care and
Use of Laboratory Animals and with the approval of the Animal Care and Five to eight micrograms of total RNA were used as a template for
Use Committee of Evanston Northwestern Healthcare and Northwestern cDNA synthesis. The cDNA was prepared using random primers and the
University. Inbred C3HeB/FeJ mice and C3H/HeJ mice (Jackson Labo- Moloney murine leukemia virus reverse transcriptase system (Invitrogen).
ratory, Bar Harbor, ME) were housed at an ambient temperature of 72 F Polymerase chain reaction (PCR) primers were designed and synthesized
and a 12L:12D photoperiod. Animals had free access to food and water. on the basis of reported mouse cDNA sequences for COX-1, COX-2,
Females (age, 8–14 wk) were mated with fertile males. Mating was veri- PGDH, and -actin. Sequences of the primers and amplicon lengths were
ﬁed by the presence of a vaginal plug. On Day 14.5 after plugging ( 75% as follows: for COX-1, 5 -gcatgtggctgtggatgtca-3 (forward) and 5 -ggt
of the typical 19- to 20-day gestation), surgery was performed (see below). cttggtgttgaggcaga-3 (reverse), with an amplicon of 388 base pairs (bp)
corresponding to nucleotides 1395–1782 in the mouse COX-1 cDNA
Preparation of HKE (GenBank accession no. BC005573.1); for COX-2, 5 -acactctatcactggca
ccc-3 (forward) and 5 -gaagggacaccccttcacat-3 (reverse), with an ampli-
Escherichia coli bacteria (American Type Culture Collection no. con of 585 bp corresponding to nucleotides 1229–1813 in the mouse
12014) were grown to log phase at 37 C in Luria-Bertani broth (Invitro- COX-2 cDNA (GenBank accession no. NM011198.1); for PGDH, 5 -
gen, Carlsbad, CA) and concentrated by centrifugation. They were then atttcggaagattggatattttggtc-3 (forward) and 5 -ttcaatgagatctattaatccattgg-3
washed three times with phosphate-buffered saline (PBS) and suspended (reverse), with an amplicon of 461 bp corresponding to nucleotides 269–
in PBS. Serial dilutions of the E. coli suspension were plated in triplicate 729 in the mouse PGDH cDNA (GenBank accession no. NM008278.1);
to determine the concentration of bacteria by overnight culture. Immedi- and for -actin, 5 -attgtgatggactccggtgacgg-3 (forward) and 5 -atcttgatctt-
ately after plating these dilutions, the E. coli within the PBS suspension catggtgctagg-3 (reverse), with an amplicon of 536 bp corresponding to
were killed by boiling in water for 5 min, and the suspension was then nucleotides 373–908 in the mouse -actin cDNA (GenBank accession no.
frozen at 20 C. Bacterial killing was veriﬁed by lack of growth overnight M12481.1). Each PCR reaction was performed in a 30- l mixture con-
in broth and solid media. After the concentration of the bacterial suspen- taining 1.0 l of cDNA, 10 pmol of each primer, 0.25 mM dNTP, and
sion was determined, the HKE stock was thawed and diluted to a concen- 1.25 U of Taq DNA polymerase (Roche Applied Science, Indianapolis,
tration of 1 1010 organisms/ml. This latter suspension was vortexed, IN). Cycling conditions were as follows: denaturation for 40 sec at 94 C,
aliquoted, and frozen at 80 C. Before each experiment, one of the frozen annealing for 30 sec at 62 C, and extension for 60 sec at 72 C. After
killed bacterial aliquots was thawed, vortexed, and diluted as necessary to sequencing of COX-1, COX-2, PGDH, and -actin PCR products to con-
the desired concentration. ﬁrm the gene speciﬁcity of these fragments, the optimal number of am-
pliﬁcation cycles was determined in pilot reactions and were as follows:
Inoculation procedure and specimens 22 cycles for -actin, 28 cycles for COX-1, 30 cycles for COX-2, and 38
cycles for PGDH.
Pregnant mice were anesthetized with 16–18 g/g body weight of a After ampliﬁcation, the PCR products were resolved by 1.2% agarose
mixture of 2.5% (w/v) tribromoethyl alcohol and 2.5% (v/v) tert-amyl gel electrophoresis and visualized by staining with SYBR green-1 (Mo-
alcohol (Aldrich Chemical, Milwaukee, WI) in PBS. A 1.5-cm midline lecular Probes, Eugene, OR). Cycle numbers were determined empirically
incision was made in the lower abdomen, and 100 L of pyrogen-free PBS to yield amplicon bands of moderate intensity that represented a linear
TLR-4, COX, AND PGDH IN BACTERIALLY-INDUCED LABOR 1959
FIG. 1. Expression of COX-1 and COX-2
transcripts in maternal tissues of C3HeB/
FeJ (TLR-4-normal) mice receiving PBS
(Control) or HKE. The COX-1, COX-2, and
-actin transcripts were detected by RT-
PCR. Typical gels are shown (A). Ratios of
the relative signal intensities of COX-1/ -
actin (B) and COX-2/ -actin (C) in myo-
metrium and decidual caps (n 5 in each
group) are depicted. *P 0.05 compared
with the value of vehicle-treated controls.
relationship between the number of cycles and the logarithm of the number RESULTS
of target molecules. The density of each DNA band was evaluated with a
STORM-860 PhosphorImager and analyzed using the ImageQuantTM soft- Dose-Response Relationship between HKE and Incidence
ware package (both from Molecular Dynamics, Sunnyvale, CA) as re- of Preterm Delivery in TLR-4-Normal (C3HeB/FeJ) Mice
ported previously . The ratios of the signals for COX-1, COX-2, and
PGDH to that of -actin were used as to determine the relative level of Six of six LPS-sensitive C3HeB/FeJ mice treated with
transcription expression. We have shown that -actin levels remain rela- vehicle delivered normal litters at term. Intrauterine injec-
tively stable within gestational tissues following HKE exposure in our tion of HKE demonstrated a dose-response relationship be-
mouse model over at least 4 h (unpublished data).
tween inoculum size and incidence of preterm delivery (Ta-
ble 1). Administration of the higher inocula (5 109 or 1
10 10 bacteria) resulted in 100% preterm delivery within
an average of 17.5 h after surgery. All offspring delivered
All values in the ﬁgures and text are expressed as the mean SEM. preterm were dead. Among animals treated with the lower
Data sets were examined by one-way analysis of variance, and individual
group means were compared with the Student unpaired t-test. For preterm
HKE inocula, intrauterine demise was observed in 20% of
delivery or fetal death, chi-square analyses was used with the Fisher exact fetuses exposed to 1 107 bacteria (48 live fetuses and 12
correction when necessary. Differences between groups were considered dead fetuses in a total of six mothers killed on Day 3 after
to be signiﬁcant when P 0.05. surgery) and in 55% exposed to 1 108 bacteria (17 live
TABLE 1. Relationship between inoculum size of HKE and incidence of preterm delivery in C3HeB/FeJ (normal) and C3H/HeJ (TLR-4-mutant) mice
treated on Day 14.5 of 20-day gestation.a
HKE in 100 l of PBS 0 1 107 1 108 1 109 5 109 1 1010
C3HeB/FeJ mice (TLR-4 normal) 6 6 8 7 7 6
Preterm delivery 0% 0% 25% 71% 100% 100%
C3H/HeJ mice (TLR-4 mutant) 5 8 7
Preterm delivery 0% 0% 28%
a Preterm delivery was deﬁned as expulsion of one or more pups within 48 h of surgery.
1960 WANG AND HIRSCH
FIG. 2. Expression of COX-1 and COX-2
transcripts in fetal tissues of C3HeB/FeJ
(TLR-4-normal) mice receiving PBS (Con-
trol) or HKE. The COX-1, COX-2, and -
actin transcripts were detected by reverse
transcription-PCR. Typical gels are shown
(A). Ratios of the relative signal intensities
of COX-1/ -actin (B) and COX-2/ -actin
(C) in placentas, fetal membranes, and fe-
tuses (n 5 in each group) are depicted.
*P 0.05 compared with the value of ve-
fetuses and 21 dead fetuses in a total of four mothers killed was observed in 7% of fetuses exposed to 5 109 bacteria
on Day 3 after surgery). In general, animals that were ad- (26 live fetuses and two dead fetuses in a total of ﬁve moth-
ministered a delivery inoculum at or below 5 109 organ- ers killed on Day 3 after surgery) and in 34% exposed to
isms appeared to be healthy or mildly ill (exhibiting mild 1 1010 bacteria (23 live fetuses and 12 dead fetuses in a
piloerection, decreased mobility, and anorexia). Recovery total of ﬁve mothers killed on Day 3 after surgery). Ma-
after delivery appeared to be complete in most cases. The ternal illness was not seen with 5 109 bacteria and was
HKE inoculation at a dose of 1 1010 organisms caused only mild with 1 1010 bacteria.
maternal death or severe illness requiring ﬁve of six mice
to be killed. Effects of HKE on COX-1, COX-2, and PGDH Transcripts
in Maternal and Fetal Tissues
Effect of TLR-4 on Bacterially Induced Preterm Delivery
To determine whether transcription of the rate-limiting
To test the hypothesis that HKE induces preterm deliv- enzymes of prostaglandin synthesis (COX-1, COX-2, and
ery via TLR-4, C3H/HeJ (TLR-4-mutant) mice were in- PGDH) is regulated via TLR-4 during the early phase of
oculated with quantities of HKE sufﬁcient to cause preterm bacterially induced preterm labor, semiquantitative reverse
delivery in 100% of control C3HeB/FeJ mice. As shown transcription-PCR was performed in pregnancy tissues ob-
in Table 1, preterm delivery occurred in none of eight TLR- tained within 4 h after administration of either PBS or 5
4-mutant mice after treatment with 5 109 HKE and in 109 HKE. In both mouse strains, bacterial exposure caused
two of seven after treatment with 1 1010 HKE. Among signiﬁcant increases in COX-2 mRNA in myometrium and
TLR-4-mutant mice treated with HKE, intrauterine demise
FIG. 3. Comparison of COX-2 transcripts in fetomaternal tissues of nor- FIG. 4. Comparison of COX-1 transcripts in fetomaternal tissues of nor-
mal (C3HeB/FeJ) and TLR-4-mutant (C3H/HeJ) mice 4 h after HKE injec- mal (C3HeB/FeJ) and TLR-4-mutant (C3H/HeJ) mice 4 h after HKE injec-
tion. For purposes of standardization, all ratios are expressed as fold- tion. For purposes of standardization, all ratios are expressed as fold-
change from the baseline measurement in control fetuses. *P 0.05 com- change from the baseline measurement in control fetuses. No signiﬁcant
pared with the value of vehicle-treated controls for the same tissue type. differences are present between the two treatment groups.
TLR-4, COX, AND PGDH IN BACTERIALLY-INDUCED LABOR 1961
FIG. 5. Comparison of PGDH transcripts
in fetomaternal tissues of normal (C3HeB/
FeJ) and TLR-4-mutant (C3H/HeJ) mice 4 h
after HKE injection. The PGDH and -ac-
tin transcripts were detected by reverse
transcription-PCR. Typical gels are shown
(upper panels) as well as quantitative anal-
ysis of relative signal intensities of PGDH/
-actin ((lower panels); n 5 in each
group). *P 0.05 compared with the val-
ue of vehicle-treated controls.
decidual caps but not in placentas and fetuses (Figs. 1–3). sheep, increased COX-2 expression in endometrium, pla-
After HKE treatment, COX-2 mRNA becomes higher in centa, and myometrium was tightly associated with the on-
maternal tissues (myometrium and decidual caps) than in set of betamethasone-induced premature labor as well as
fetal tissues in C3HeB/FeJ (normal) mice but not in C3H/ spontaneous term labor . In the baboon, COX-2 ex-
HeJ (LPS-resistant) mice (Fig. 3). Basal levels of COX-1 pression was increased in the lower uterine segment, cervix,
mRNA in maternal tissues are higher than in fetal tissues and decidua but not in the uterine fundus, chorion, and
in C3H/HeJ mice but not in C3HeB/FeJ mice. Levels of placenta during late pregnancy and labor . In humans,
COX-1 remained unchanged in all tested maternal and fetal a large increase in COX-2 mRNA was found throughout
tissues in both strains (Figs. 1, 2, and 4). late gestation in fetal tissues . Both COX-1 and COX-
In normal (C3HeB/FeJ) mice, basal levels of PGDH 2 have been reported to decrease or remain unchanged in
were detectable but were lower in maternal tissues (myo- the myometrium at the onset of labor  and to increase
metrium and decidual caps) than in fetal tissues (Fig. 5). in the amnion during term labor .
Levels of PGDH decreased signiﬁcantly after HKE treat- The present results show that the levels of COX-2 tran-
ment in fetuses, fetal membranes, and myometrium, re- scripts in myometrium and decidua increase sharply in re-
spectively, compared to those after control injections. In sponse to HKE treatment in both C3HeB/FeJ and C3H/HeJ
decidual caps, however, PGDH mRNA increased over its mice, whereas COX-1 mRNA levels remain unchanged in
low basal level. In TLR-4-mutant (C3H/HeJ) mice, basal the tissues tested for both mouse strains. These data suggest
levels of PGDH were detectable in fetal tissues but not in that TLR-4 signaling may not be essential for bacterially
maternal tissues (Fig. 5). Levels of PGDH remained un- induced COX-2 gene expression, and they support the no-
changed in fetal tissues after HKE treatment. tion that COX-2, not COX-1, is the enzyme primarily re-
sponsible for increased prostaglandin biosynthesis during
DISCUSSION bacterially induced preterm labor. Several studies have
The present study demonstrates that TLR-4 is a critical shown that COX-2 antagonism may inhibit preterm labor
mediator of labor signals in the murine bacterial infection in different species, such as human , sheep , rat
model. Although HKE induces premature delivery in nor- , and mice [15, 38]. This conclusion is also supported
mal mice in a dose-dependent manner, this phenomenon by other data, such as a report that pretreatment of pregnant
does not occur, or is greatly obtunded, in TLR-4-mutant mice with COX-2 inhibitor, but not COX-1 inhibitor, pre-
mice. The small effect observed with very high numbers vented LPS-induced preterm labor .
of HKE may be caused either by residual signaling via the To our knowledge, this is the ﬁrst study to report that
mutant TLR-4 or other receptors or by bacterial factors oth- the basal levels of PGDH mRNA in mid- to late-pregnancy
er than LPS. Parallel observations were also made in rela- mice are higher in fetal tissues (fetus, fetal membranes, and
tion to maternal illness and fetal demise resulting from bac- placenta) than in maternal uterine tissues in both C3HeB/
terial exposure. We have previously demonstrated that fetal FeJ and C3H/HeJ mice. This tissue distribution of PGDH
death per se in the absence of a bacterial stimulus does not may allow for ﬁnely controlled regulation of prostaglandin
result in labor in the mouse . activity in individual tissues during pregnancy. After HKE
What is the mechanism by which bacteria cause labor? treatment, PGDH mRNA decreases signiﬁcantly in the fe-
Previous studies have suggested a central role for prosta- tus, fetal membranes, and myometrium in TLR-4-normal
glandins and their synthetic and catabolic enzymes in this mice but not in TLR-4 mutant mice. These data suggest
process. Variable reports have appeared concerning changes that TLR-4 signaling may be involved in HKE-induced
in COX expression at parturition in different species and in PGDH down-regulation. The role of prostaglandins gener-
different circumstances. In the mouse, increased COX-1, ated in the fetus itself during gestation is still not clear.
but not COX-2, mRNA was reported during late gestation When the fetus is infected, fetal cortisol, cytokine, and
in the uterus  and fetal membranes , with increased prostaglandin production are increased [39, 40]. The cur-
COX-2 expression during term labor as well as ethanol- rently accepted hypothesis is that PGDH represents a met-
and LPS-induced preterm labor [26, 27]. In the rat, one abolic barrier in fetal tissues, either to prevent the passage
study found that both COX-1 and COX-2 increased in myo- of prostaglandins generated in these tissues to the uterus or
metrium with the onset of labor , but a different study to prevent prostaglandins generated in other tissues from
reported increased expression of COX-2 only . In damaging the fetus. During HKE-induced preterm labor,
1962 WANG AND HIRSCH
this functional barrier may break down, perhaps accounting 18. Muhle RA, Pavlidis P, Grundy WN, Hirsch E. A high-throughput
for the fetal inﬂammatory response that has been observed study of gene expression in preterm labor with a subtractive microar-
ray approach. Am J Obstet Gynecol 2001; 185:716–724.
during infection [40, 41]. 19. Mussalli GM, Blanchard R, Brunnert SR, Hirsch E. Inﬂammatory cy-
In summary, the present study suggests that TLR-4 sig- tokines in a murine model of infection-induced preterm labor: cause
naling is a critical factor in bacterially induced preterm la- or effect? J Soc Gynecol Investig 1999; 6:188–195.
bor. During bacterially induced preterm labor, TLR-4 sig- 20. Reznikov LL, Fantuzzi G, Selzman CH, Shames BD, Barton HA, Bell
naling mediates PGDH gene down-regulation but is not es- H, McGregor JA, Dinarello CA. Utilization of endoscopic inoculation
sential for COX-2 gene up-regulation. Bacterially induced in a mouse model of intrauterine infection-induced preterm birth: role
of interleukin 1beta. Biol Reprod 1999; 60:1231–1238.
prostaglandin activity may be mediated primarily by in- 21. Fidel PL Jr, Romero R, Cutright J, Wolf N, Gomez R, Araneda H,
creased synthesis in maternal tissues and decreased degra- Ramirez M, Yoon BH. Treatment with the interleukin-I receptor an-
dation in fetal tissues. tagonist and soluble tumor necrosis factor receptor Fc fusion protein
does not prevent endotoxin-induced preterm parturition in mice. J Soc
ACKNOWLEDGMENT Gynecol Investig 1997; 4:22–26.
22. Hirsch E, Muhle RA, Mussalli GM, Blanchard R. Bacterially induced
The authors thank Yana Filipovich for technical assistance. preterm labor in the mouse does not require maternal interleukin-1
signaling. Am J Obstet Gynecol 2002; 186:523–530.
23. Bennett WA, Terrone DA, Rinehart BK, Kassab S, Martin JN Jr,
REFERENCES Granger JP. Intrauterine endotoxin infusion in rat pregnancy induces
1. Challis JR, Smith SK. Fetal endocrine signals and preterm labor. Biol preterm delivery and increases placental prostaglandin F2 metabolite
Neonate 2001; 79:163–167. levels. Am J Obstet Gynecol 2000; 182:1496–501.
2. Romero R, Mazor M, Wu YK, Sirtori M, Oyarzun E, Mitchell MD, 24. Beutler B, Poltorak A. Positional cloning of LPS, and the general role
Hobbins JC. Infection in the pathogenesis of preterm labor. Semin of toll-like receptors in the innate immune response. Eur Cytokine
Perinatol 1988; 12:262–279. Netw 2000; 11(2):143–152.
3. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection and 25. Wang H, Tan X, Chang H, Gonzalez-Crussi F, Remick DG, Hsueh W.
preterm delivery. N Engl J Med 2000; 342:1500–1507. Regulation of platelet-activating factor receptor gene expression in
4. McDufﬁe RS Jr, Sherman MP, Gibbs RS. Amniotic ﬂuid tumor ne- vivo by endotoxin, platelet-activating factor and endogenous tumour
crosis factor- and interleukin-1 in a rabbit model of bacterially in- necrosis factor. Biochem J 1997; 322:603–608.
duced preterm pregnancy loss. Am J Obstet Gynecol 1992; 167:1583– 26. Cook JL, Zaragoza DB, Sung DH, Olson DM. Expression of myo-
1588. metrial activation and stimulation genes in a mouse model of preterm
5. Gravett MG, Witkin SS, Haluska GJ, Edwards JL, Cook MJ, Novy labor: myometrial activation, stimulation, and preterm labor. Endocri-
MJ. An experimental model for intraamniotic infection and preterm nology 2000; 141:1718–1728.
labor in rhesus monkeys. Am J Obstet Gynecol 1994; 171:1660–1667. 27. Swaisgood CM, Zu HX, Perkins DJ, Wu S, Garver CL, Zimmerman
6. Fidel PL Jr, Romero R, Wolf N, Cutright J, Ramirez M, Araneda H, PD, Iams JD, Kniss DA. Coordinate expression of inducible nitric
Cotton DB. Systemic and local cytokine proﬁles in endotoxin-induced oxide synthase and cyclooxygenase-2 genes in uterine tissues of en-
preterm parturition in mice. Am J Obstet Gynecol 1994; 170:1467– dotoxin-treated pregnant mice. Am J Obstet Gynecol 1997; 177:1253–
7. Hirsch E, Saotome I, Hirsh D. A model of intrauterine infection and 28. Dong YL, Gangula PR, Fang L, Yallampalli C. Differential expression
preterm delivery in mice. Am J Obstet Gynecol 1995; 172:1598–1603. of cyclooxygenase-1 and -2 proteins in rat uterus and cervix during
8. Kaga N, Katsuki Y, Obata M, Shibutani Y. Repeated administration the estrous cycle, pregnancy, labor and in myometrial cells. Prosta-
of low-dose lipopolysaccharide induces preterm delivery in mice: a glandins 1996; 52:13–34.
model for human preterm parturition and for assessment of the ther- 29. Arslan A, Zingg HH. Regulation of COX-2 gene expression in rat
apeutic ability of drugs against preterm delivery. Am J Obstet Gynecol uterus in vivo and in vitro. Prostaglandins 1996; 52:463–481.
1996; 174:754–759. 30. Wu WX, Ma XH, Yoshizato T, Shinozuka N, Nathanielsz PW. In-
9. Patel FA, Challis JR. Prostaglandins and uterine activity. Front Horm crease in prostaglandin H synthase 2, but not prostaglandin F2 syn-
Res 2001; 27:31–56. thase, mRNA in intrauterine tissues during betamethasone-induced
10. Jeyabalan A, Caritis SN. Pharmacologic inhibition of preterm labor. premature labor and spontaneous term labor in sheep. J Soc Gynecol
Clin Obstet Gynecol 2002; 45:99–113. Investig 2001; 8:69–76.
11. Sugimoto Y, Yamasaki A, Segi E, Tsuboi K, Aze Y, Nishimura T, 31. Wu WX, Ma XH, Smith GC, Koenen SV, Nathanielsz PW. A new
Oida H, Yoshida N, Tanaka T, Katsuyama M, Hasumoto K, Murata concept of the signiﬁcance of regional distribution of prostaglandin H
T, Hirata M, Ushikubi F, Negishi M, Ichikawa A, Narumiya S. Failure synthase 2 throughout the uterus during late pregnancy: investigations
of parturition in mice lacking the prostaglandin F receptor. Science in a baboon model. Am J Obstet Gynecol 2000; 183:1287–1295.
1997; 277:681–683. 32. Olson DM, Mijovic JE, Zaragoza DB, Cook JL. Prostaglandin endo-
12. Gross GA, Imamura T, Luedke C, Vogt SK, Olson LM, Nelson DM, peroxide H synthase type 1 and type 2 messenger ribonucleic acid in
Sadovsky Y, Muglia LJ. Opposing actions of prostaglandins and oxy- human fetal tissues throughout gestation and in the newborn infant.
tocin determine the onset of murine labor. Proc Natl Acad Sci U S A Am J Obstet Gynecol 2001; 184:169–174.
1998; 95:11875–11879. 33. Giannoulias D, Patel FA, Holloway AC, Lye SJ, Tai HH, Challis JR.
13. Winchester SK, Imamura T, Gross GA, Muglia LM, Vogt SK, Wright Differential changes in 15-hydroxyprostaglandin dehydrogenase and
J, Watanabe K, Tai HH, Muglia LJ. Coordinate regulation of prosta- prostaglandin H synthase (types I and II) in human pregnant myo-
glandin metabolism for induction of parturition in mice. Endocrinol- metrium. J Clin Endocrinol Metab 2002; 87:1345–1352.
ogy 2002; 143:2593–2598. 34. Mijovic JE, Zakar T, Angelova J, Olson DM. Prostaglandin endoper-
14. Gupta DK, Sato TA, Keelan JA, Marvin KW, Mitchell MD. Expres- oxide H synthase mRNA expression in the human amnion and decidua
sion of prostaglandin H synthase-1 and -2 in murine intrauterine and during pregnancy and in the amnion at preterm labour. Mol Hum
gestational tissues from mid pregnancy until term. Prostaglandins Oth- Reprod 1999; 5:182–187.
er Lipid Mediat 2001; 66:17–25. 35. Stika CS, Gross GA, Leguizamon G, Gerber S, Levy R, Mathur A,
15. Gross G, Imamura T, Vogt SK, Wozniak DF, Nelson DM, Sadovsky Bernhard LM, Nelson DM, Sadovsky Y. A prospective randomized
Y, Muglia LJ. Inhibition of cyclooxygenase-2 prevents inﬂammation- safety trial of celecoxib for treatment of preterm labor. Am J Obstet
mediated preterm labor in the mouse. Am J Physiol Regul Integr Gynecol 2002; 187:653–660.
Comp Physiol 2000; 278:R1415–R1423. 36. Scott JE, Grigsby PL, Hirst JJ, Jenkin G. Inhibition of prostaglandin
16. Sato TA, Gupta DK, Keelan JA, Marvin KW, Mitchell MD. Cytosolic synthesis and its effect on uterine activity during established prema-
phospholipase A2and 15-hydroxyprostaglandin dehydrogenase mRNA ture labor in sheep. J Soc Gynecol Investig 2001; 8:266–276.
expression in murine uterine and gestational tissues during late preg- 37. Okawa T, Suzuki H, Yaanagida K, Sato A, Vedernikov Y, Saade G,
nancy. Prostaglandins Leukot Essent Fatty Acids 2001; 64:247–251. Garﬁeld R. Effect of lipopolysaccharide on uterine contractions and
17. Hirsch E, Blanchard R, Mehta SP. Differential fetal and maternal con- prostaglandin production in pregnant rats. Am J Obstet Gynecol 2001;
tributions to the cytokine milieu in a murine model of infection-in- 184:84–89.
duced preterm birth. Am J Obstet Gynecol 1999; 180:429–434. 38. Sakai M, Tanebe K, Sasaki Y, Momma K, Yoneda S, Saito S. Eval-
TLR-4, COX, AND PGDH IN BACTERIALLY-INDUCED LABOR 1963
uation of the tocolytic effect of a selective cyclooxygenase-2 inhibitor 40. Romero R, Gomez R, Ghezzi F, Yoon BH, Mazor M, Edwin SS, Berry
in a mouse model of lipopolysaccharide-induced preterm delivery. SM. A fetal systemic inﬂammatory response is followed by the spon-
Mol Hum Reprod 2001; 7:595–602. taneous onset of preterm parturition. Am J Obstet Gynecol 1998; 179:
39. Yoon BH, Romero R, Jun JK, Maymon E, Gomez R, Mazor M, Park 186–193.
JS. An increase in fetal plasma cortisol but not dehydroepiandroster- 41. Chaiworapongsa T, Romero R, Kim JC, Kim YM, Blackwell SC,
one sulfate is followed by the onset of preterm labor in patients with Yoon BH, Gomez R. Evidence for fetal involvement in the pathologic
preterm premature rupture of the membranes. Am J Obstet Gynecol process of clinical chorioamnionitis. Am J Obstet Gynecol 2002; 186:
1998; 179:1107–1114. 1178–1182.