Molecular Human Reproduction Vol.7, No.3 pp. 293–300, 2001

Different regulation of oestrogen receptors α and β in thehuman cervix at term pregnancy

Hong Wang1, Ylva Stjernholm2, Gunvor Ekman2, Håkan Eriksson1 and Lena Sahlin1,3

1Division for Reproductive Endocrinology and 2Division for Obstetrics and Gynecology, Department of Woman and Child Health,Karolinska Institutet, Stockholm, Sweden

3To whom correspondence should be addressed at: Division for Reproductive Endocrinology, Karolinska Hospital, L5:01,S-171 76 Stockholm, Sweden. E-mail: Lena.Sahlin@kbh.ki.se

During pregnancy, a cervical connective tissue remodelling takes place, clinically recognized as softening, effacementand dilatation. The role of oestrogens and their receptors (ER) in this process is not clear. ERα is a ligand-activatedtranscription factor involved in many physiological processes. The identification of a second oestrogen receptor,ERβ, has led to a re-evaluation of oestrogen signalling and physiology. The aim of this study was to monitor theexpression of the two ERs in the cervix from women at term pregnancy (TP) and after parturition (PP) comparedwith that of non-pregnant (NP). A solution hybridization assay showed that the level of ERα mRNA was significantlydecreased in the PP group, when compared with the NP and TP groups. In contrast the ERβ mRNA level wasincreased in the TP group compared with the NP and PP groups. These results were supported by reversetranscription–polymerase chain reaction (RT–PCR). Similar results were observed for the protein with immunohisto-chemistry. Intense ERβ immunostaining was observed in neutrophils and the endothelial cells of blood vessels. Inconclusion, this study supports a concept according to which oestrogen might be involved in the final remodellingof the cervix via the modulating effects of the two ERs. Furthermore, oestrogen may mediate some effects oncervical ripening via ERβ present in the invading neutrophils. Further studies are needed to elucidate this finding.

Key words: cervix/oestrogen/oestrogen receptors/parturition/pregnancy

IntroductionSteroid receptor regulation in the human cervix is similar tothat of other organs in the reproductive tract, with oestrogenstimulating, and progesterone antagonizing, oestrogen activity(Gorodeski, 1996). During pregnancy, the systemic concentra-tions of oestrogens and progesterone increase up to 100-folduntil parturition (Speroff et al., 1989). However, in contrast toother species, the serum concentrations of neither oestrogennor progesterone are abruptly changed immediately beforeparturition in humans (Csapo et al., 1971; Turnbull et al.,1974). The precise mechanisms of cervical ripening in humansare not known. The ripening process is associated with anincreased collagen turnover resulting in differently organizedcollagen fibrils (Uldbjerg et al., 1983). This structural changeis necessary for a normal onset and progress of parturition.

There has been some debate about whether or not oestrogensstimulate cervical ripening. A study showed an increase incervical ripening in a group locally treated with oestradiol(150 mg) in a viscous gel, when compared with a controlgroup receiving only the viscous gel (Gordon and Calder,1977). In another study, oestradiol (200 mg) had no effect onthe change of Bishop score and length of labour when given

© European Society of Human Reproduction and Embryology 293

in term pregnancy for induction of labour (Williams et al.,1988). Although oestrogen may not play an active part atparturition, it might influence the gradual cervical connectivetissue remodelling during pregnancy (Stjernholm et al., 1997).

Oestrogen action is mediated primarily via binding tospecific receptors (ER) in target cells (Gronemeyer, 1992). Inprevious studies, we have shown that the levels of bothERα and the progesterone receptor (PR) in the cervix aresignificantly lower in term pregnant and post-partal women incomparison with non-pregnant women (Stjernholm et al., 1996,1997). The decrease in receptor concentrations coincided withbiochemical changes in the cervical connective tissue, such asan increased collagen solubility and an altered proteoglycancomposition (Stjernholm et al., 1996). The mechanisms behindthe decreased levels of ERα and PR, and the concomitantbiochemical changes in the cervical connective tissue arenot known.

Since the second ER subtype (ERβ) was discovered in 1996(Kuiper et al., 1996), a re-evaluation of oestrogen signallingand physiology has been ongoing. It has been shown thatthe ERα and ERβ subtypes have opposite effects on genetranscription depending on the ligand and response elements

H.Wang et al.

(Paech et al., 1997). ERβ can act as a transdominant repressoron ERα transcriptional activity at subsaturating concentrationsof oestradiol (Hall and McDonnell, 1999). It is not knownwhether ERβ can regulate ERα in a similar way duringpregnancy when oestrogen concentrations are high. Recentlypublished results from ERβ knockout mice (βERKO) demon-strate that ERβ acts as a modulator of ERα-mediated genetranscription in the uterus, and is responsible for the down-regulation of PR in the luminal epithelium (Weihua et al., 2000).

Limited information exists of ERβ expression in the cervix.In a previous study we have found that ERα and ERβ co-existin the rat cervix (Wang et al., 2000) and that there is asignificant difference in ERβ immunostaining in the stromabetween the oestrus and metoestrus stages during the oestrouscycle. Another study, using reverse transcription–polymerasechain reaction (RT–PCR), showed that both ERα and ERβmRNAs were expressed in the human cervix and in culturedcervical epithelial cells (Gorodeski and Pal, 2000). The roleof ERβ in cervical physiology as well as during pregnancy isunknown. Recently, one group (Wu et al., 2000) reported anincreased level of ERβ in myometrium of term pregnantwomen and a down-regulation in the expression of labour-associated genes in uterine smooth muscle cells, suggestingthat ERβ may play an important biological role during termpregnancy. The aim of this study was to monitor the expressionof the two ERs in the cervix from women at term pregnancyand immediately post-partum and to compare it with that ofthe non-pregnant woman in order to find possible changes andrelations to cervical ripening.

Materials and methods

Study participants

The non-pregnant group (NP) consisted of six regularly menstruatingwomen with a mean age of 42 years (range 32–49), and a meanparity of 2 (1–3). All underwent hysterectomies due to benigndisorders not involving the cervix. The term pregnant group (TP)consisted of eight primiparous women with a mean age of 32 years(range 28–36) and a mean gestational age of 38 weeks (range 37–40). All women had unripe cervices with a Bishop score �5 pointsand none of them were in labour. Biopsies were obtained duringelective Caesarean sections. The post-partal group (PP) consisted of10 primiparous women from whom biopsies were taken within 15 minafter spontaneous vaginal delivery. They had a mean age of 30 years(range 23–35) and a mean gestational age of 41 weeks (range 39–42). All women were healthy with uncomplicated pregnancies, andwere without medication. Ethical approval for the study was obtainedfrom the local ethics committee of the Karolinska Hospital (Ref no:99–099) and all women had given their informed consent.

Tissue collection

Cervical biopsies were obtained transvaginally from the anteriorcervical lip at the 12 o’clock position, from 10–20 mm depth. Thecervical biopsies were divided and one half was immersion-fixed in4% formaldehyde at 4°C overnight, stored at 4°C in 70% ethanoland thereafter embedded in paraffin. The other piece was immediatelyfrozen in liquid nitrogen and stored at –70°C until analysed.

Hybridization probes and solution hybridization

The probe used for the ERα mRNA determinations was derived frombcpel, a full-length cDNA containing the whole open reading frame

294

of the human ERα. The cDNA was inserted in a pGEM7zf vector.Restriction of this vector with BglII allows the synthesis of a probecorresponding to nucleotides 1470–2062 which encodes the C-terminal half of the steroid binding domain (E) and all of domain F.

The probe used for ERβ mRNA determinations was derived froma pBS plasmid with an insert of a 187 bp PVUII/EcoRI fragmentcorresponding to nucleotides 774–979 in the human ERβ gene.

Total nucleic acids (TNA) were prepared as described before(Sahlin et al., 1994). A solution hybridization analysis of specificmRNA was carried out as previously described (Sahlin et al., 1994;Wang et al., 1999).

RNA isolation and RT–PCR analyses

Frozen pieces of cervical tissue were processed for isolation of totalRNA according to the manual for the SV total RNA isolation system(Promega, Madison, WI, USA). Total RNA (1 µg) was reversetranscribed at 42°C for 45 minutes in a final volume of 40 µl with areaction mixture containing 50 mmol/l Tris–HCl pH 8.3, 75 mmol/lKCl, 3 mmol/l MgCl2, 7.5 mmol/l dithiothreitol, 0.5 mmol/l dNTPs,1 µg random hexamers and 400 IU of ML-MTV reverse transcriptase(GIBCO, BRL, Paisley, UK). Primer sequences (Arts J et al., 1997)and the predicted product sizes were for the human ERα gene:5�-AATTCAGATAATCGACGCCAG-3�; 5�-GTGTTTCAACATTC-TCCCTCCTC-3� and 340 bp; for the human ERβ gene: 5�-TAGT-GGTCCATCGCCAGTTAT-3�; 5�-GGGAGCCACACTTCACCAT-3�and 395 bp; and for the human ribosomal protein S28, which wasused as a control of the amount of RNA added: 5�-GTGCAGAT-CTTGGTGGTAGTAGC-3�; 5�-AGAGCCAATCCTTATCCCGAA-GTT-3� and 552 bp. For PCR amplification, the cDNA (correspondingto 50 ng RNA from the RT-PCR reaction) was added to the reactionmixture containing 20 mmol/l Tris–HCl pH 8.4, 50 mmol/l KCl,2.5 IU Taq DNA polymerase (Gibco BRL, Paisley, UK), 0.2 mmol/ldNTPs, 1.5 mmol/l MgCl2 and oligonucleotide primer pairs(50 pmol/pair), in a final volume of 50 µl. PCR was performed for18 cycles for 28S; 28 cycles for ERα and 32 cycles for ERβ, theamount of PCR products for ERα and ERβ increased linearly up to32 and 36 cycles respectively (data not shown). The PCR wasperformed using 94°C for 1 min (denaturing), 58°C for 1 min(annealing), and 72°C for 1 min (extension), with a final incubationat 72°C for 3 min in a thermal cycler (Perkin-Elmer, Norwalk, CT,USA). The products were subjected to electrophoresis in 2.5% agarosegels; data were analysed using a Fujifilm Las-1000 system (Fujifilm,Japan). The ERα and ERβ bands are shown together with a S28 bandto enable comparison of the amount of RNA loaded to each well.

Immunohistochemistry

Paraffin sections (5 µm) were used and a standard immunohistochemi-cal technique (avidin–biotin–peroxidase) was carried out as describedpreviously (Wang et al., 1999), to visualize ERα immunostainingintensity and distribution. A monoclonal mouse anti-human antibodywas used for detection of ERα (08–1149; Zymed Laboratories, Inc.,San Francisco, CA, USA). This antibody recognizes the N-terminaldomain (A/B region) of the ERα. Replacement of the primaryantibody with non-immune serum was used as negative control.

A polyclonal chicken anti human ERβ (503) antibody was usedfor the ERβ immunostaining. The preparation of this antibody hasbeen described (Saji et al., 2000). The ERβ antibody pre-absorbedwith an ERβ protein (Panvera, Madison, WI, USA) 1:50 (V/V)overnight at 4°C was used to demonstrate antigen specificity. Tissuesections were incubated with 1:500 dilution of ERβ antibody overnightat 4°C in phosphate-buffered saline (PBS) with 3% bovine serumalbumin (BSA). For negative controls, incubations were done withthe absorbed antibody. After washing, sections were incubated with

Oestrogen receptors α and β in human cervix

peroxidase-conjugated rabbit anti-chicken immunoglobulin G (IgG)for 1 h at room temperature. The peroxidase substrate diaminobenzid-ine (DAB) was used to visualize the reaction (SK 4100; VectorLaboratory, Burlingame, CA, USA). Thereafter the procedure was asdescribed before (Wang et al., 1999).

Another polyclonal rabbit anti-human ERβ antibody (PA1-313,Affinity Bioreagents, Inc., Golden, CO, USA) which corresponds tothe C-terminal amino acid residues 467–485 of human ERβ, was alsoused for detection of ERβ. A standard immunohistochemical technique(avidin–biotin–peroxidase) was used for this antibody immuno-staining. The protocol has previously been described (Wang et al.,1999). The washing buffer was Tris-buffered saline (TBS,0.05 mol/l, pH 7.5) used for ERβ (PA1-313) and PBS (0.1 mol/l,pH 7.5) for ERα and ERβ (503).

We found corresponding immunostaining patterns between the ERβantibodies (PA1 313 versus ERβ 503), although PA1 313 displayedmore background in the cytoplasm (data not shown). Therefore, theERβ 503 antibody was used for this study.

Image analysis

A Leica microscope and Sony video camera (Park Ridge, NJ, USA)connected to a computer with an image analysis system (Leica ImagingSystem Ltd, Cambridge, UK) was used to assess quantitative valuesfrom immunohistochemistry. The quantification of immunostaining wasperformed as described previously (Wang et al., 1999). In short, byusing colour discrimination software the total area of positively stainednuclei was measured, and expressed as a ratio of the total area of cellnuclei. The intensity of positive staining was also semi-quantitativelymeasured by three different colour discriminations: strong (���),moderate (��) or faint (�) brown reaction.

Statistical analysis

Statistical calculations for the solution hybridization results were per-formed with one way analysis of variance (ANOVA), and evaluationfor significance was done with Scheffe’s test (Scheffe, 1959). Statisticalcalculations for the immunohistochemistry results by image analysiswere performed by ANOVA on ranks (Kruskal–Wallis test) and signific-ances were evaluated by Dunn’s test. Values with different letterdesignations are significantly different (P � 0.05).

ResultsThe level of ERα mRNA measured by solution hybridizationwas decreased in the PP group as compared with the NP andTP groups (Figure 1A). This was in contrast to the ERβ mRNAlevel, which was significantly increased in the TP group,compared with the NP and PP groups (Figure 1B).

Representative gels from RT–PCR analyses are shown inFigure 2. The ERα mRNA bands of the NP and TP groupsappeared to be stronger compared with the PP group (Figure 2,top). The ERβ mRNA expression seemed to be higher in mostwomen of the TP group as compared with the NP and PP groups(Figure 2, middle). S28 was used as a standard to enablecomparison of the amount of RNA loaded to each well (Figure 2,bottom). Thus, the RT–PCR results were in agreement with thesolution hybridization results.

Immunohistochemistry showed that the ERα protein waspresent in the nuclei of cells in the different compartments ofthe cervix (Figure 3A,C,E and Figure 4). Of the cell nuclei inthe epithelium, �70% were ERα positive, but no significantdifferences were found between the groups (Figure 3A and

295

Figure 1. Cervical levels of (A) oestrogen receptor (ER)α mRNAand (B) ERβ mRNA measured by solution hybridization. Non-pregnant women (NP), n � 6; term pregnant women (TP), n � 8 andpost-partal women (PP), n � 10. Bars with different letterdesignations are significantly different (P � 0.05).

4a–c). In the stroma of the NP group, ~50% of the nuclei stainedERα positive (Figure 3C). ERα immunoreactivity declined inthe cervical stroma of the TP group (Figure 3C; 4e and k)and was significantly decreased in the PP group (Figure 3C;4f and l) compared with the NP group (Figure 3C; 4d and j).When studying only the more intense immunostaining (��/���), both the TP and PP groups were decreased comparedwith the NP group (Figure 3E). In the glandular epithelium,ERα immunostaining was intensely expressed in the NP group(Figure 4g), but was less strong in the TP group (Figure 4h)and only faint staining was observed in some glandular epithelialcells in the PP group (Figure 4i). Image analysis was notperformed for immunostaining of glandular epithelium, since aninsufficient number of biopsies contained cervical glands. Theimmunoreactivity of ERα in smooth muscle cells seemed lessin the cervix from the TP (Figure 4n) and PP (Figure 4o) groupsas compared with the NP group (Figure 4m), indicating a similarpattern to that seen in the stromal connective tissue (Figure 4d–f). ERα immunostaining in the vascular endothelial cells wasnegative (Figure 4j–l, black arrows), although some smoothmuscle cells were positively stained in all groups. No ERαpositive staining could be observed in the neutrophils (whitearrows) in the blood vessels (Figure 4j–l). No positive stainingwas observed in the negative control sections (Figure 4p).

Positive nuclear ERβ immunostaining was found in thecervical epithelium and stroma (Figure 3B,D,F and Figure 5).The results from image analysis (Figure 3B) showed that 60–

H.Wang et al.

Figure 2. The levels of oestrogen receptor (ER)α and ERβ mRNAs in the human cervix as detected by reverse transcription–polymerase chainreaction (RT–PCR). Lanes 1–5 � non-pregnant women (NP); lanes 6–10 � term pregnant women (TP); and lanes 11–15 � post-partal women(PP), n � 5 in each group. S28 was used as a standard to enable comparison of the amount of RNA loaded to each well. M � DNA sizemarker. B � blank, no reverse transcriptase added.

Figure 3. Image analysis score of positive oestrogen receptor (ER)immunoreactivity in squamous epithelium (A) ERα; (B) ERβ andcervical stroma (C) ERα and (D) ERβ. Image analysis scores on themore intense immunostaining (��/���) in the stroma are shownfor (E) ERα and (F) ERβ. Non-pregnant women (NP), n � 6; termpregnant women (TP), n � 8 and post-partal women (PP), n � 10.Box and whisker plots representing the median value with 50% of alldata falling within the box. The whiskers extend to the 5th and 95thpercentiles. Values with different letter designations are significantlydifferent.

70% of the epithelial cells were ERβ-positive. There was asignificant difference between the TP and PP groups (Figure 3Band 5b-c). In the stroma, the number of ERβ positive cells wassignificantly increased in the TP group (~50% of the cells)(Figure 3D, 5e and k) compared with the NP (Figure 3D, 5dand j) and PP (Figure 3D, 5f and l) groups (~25% of the cells

296

stained). Similar results were obtained when studying only themore intense ERβ staining (��/���) (Figure 3F), althoughfewer cells were intensely stained. In the glandular epithelium,only faint ERβ immunostaining was observed in some epithelialcells in all the groups (Figure 5g–i). The immunoreactivity ofERβ in smooth muscle cells showed a similar pattern to that of thestroma in the three groups (Figure 5m–o). The immunoreactivityseemed to be more intense in smooth muscle cells from the TPgroup (Figure 5n). ERβ immunostaining was observed in theendothelial cells (Figure 5j–l, black arrows) and some smoothmuscle cells of the vessels in all groups. Neutrophils (definedby the morphology) stained positive for ERβ (Figure 5j–l, whitearrows), and were more obvious in the TP and PP groups (Figure5k–l) since the samples contained a higher number of neutrophils.No specific nuclear staining was found in the negative controlsections after incubation with a peptide corresponding to theepitope of the ERβ antibody (Figure 5p).

DiscussionCervical ripening during pregnancy and parturition occurs indistinct stages, including a gradual softening during pregnancy,a latency phase and an active phase of dilatation and cervicalprotraction at parturition, allowing for expulsion of the foetus(Calder, 1994). It is known that progesterone withdrawal initiatescervical softening (Frydman et al., 1992) even though there isno abrupt fall in peripheral serum progesterone before cervicalremodelling.

To further study the potential roles for gonadal steroids incervical ripening, we measured the two subtypes of ER in thehuman cervix and found that the ERβ mRNA level wasincreased at term pregnancy, although the ERα mRNA levelwas unchanged. After parturition, ERβ expression was decreasedagain to the level of the non-pregnant group, whereas the ERαmRNA level was decreased to a level less than half of that inthe non-pregnant group. The protein profile of both ERα andERβ showed similar patterns in the stroma as the respectivemRNA, suggesting that the regulation of the ERs is primarilyat the transcriptional level.

In contrast to spontaneous ripening, antiprogestin treatment atterm has been shown to be followed by an increased cervical

Oestrogen receptors α and β in human cervix

Figure 4. Immunohistochemical localization of oestrogen receptor (ER)α in the cervix from non-pregnant (NP), term pregnant (TP) and post-partal (PP) women. Nuclear ERα immunostaining in the squamous epithelium (Epi) and stromal cells (Str) of the (a, d) NP, (b, e) TP and (c, f)PP groups. (g) High intensity of nuclear ERα immunostaining was present in the glandular epithelium (GE) of the NP group. (h) Less intenseimmunostaining was present in the GE of the cervix from the TP group and (i) faint immunostaining was observed in the GE from the PPgroup. (j–l) No ERα immunostaining was found in the endothelial cells of the vessels (black arrows) in any of the groups. (j–l) No positivestaining was present in the neutrophils (white arrows). (m–o) A decreasing level of immunoreactivity was observed in the smooth muscle cellsfrom the NP to the PP group. (p) Negative control for ERα incubated with mouse immunoglobulin G. Original magnification �200, scale bar30 µm (a–i and m–p) and �400, scale bar 20 µm (j–l).

297

H.Wang et al.

Figure 5. Immunohistochemical localization of oestrogen receptor (ER)β in the cervix from non-pregnant (NP), term pregnant (TP) and post-partal (PP) women. Nuclear ERβ immunostaining in the squamous epithelium (Epi) and stromal cells (Str) in the (a, d) NP, (b, e) TP and (c, f)PP groups. (g–i) Faint ERβ immunostaining was present in the glandular epithelium (GE) of all groups. (j–l) High intensity of ERβimmunostaining was present in the endothelial cells (black arrows) and neutrophils (white arrows). ERβ immunostaining was present in some ofthe smooth muscle cells of the (m) NP, (n) TP and (o) PP groups. (p) Negative control for ERβ was the antibody pre-absorbed overnight with asynthetic ERβ peptide. Original magnification �200, scale bar 30 µm (a–i and m–p) and �400, scale bar � 20 µm (j–l).

298

Oestrogen receptors α and β in human cervix

ERα concentration (Stjernholm et al., 1999). These findingsindicate that an increase in ERα, or a withdrawal of receptor-mediated progesterone inhibition, does not explain the eventsbehind the physiological cervical ripening at parturition. Therole of ERβ in the cervical ripening during pregnancy andparturition is not known. Results from studies in βERKO miceshowed that ERβ could be responsible for the down-regulationof PR in the uterine luminal epithelium (Weihua et al., 2000).The present study suggests a switch from ERα to ERβ expressionin the cervix during pregnancy as could be seen in the TP group.When this occurs during pregnancy is not known. Thus, theincreased ERβ level could lead to decreased levels of PR in thecervix, leading to decreased progesterone activity similar toantiprogestin treatment. A dramatic switch from ERα to ERβexpression in the myometrium during pregnancy has been shownpreviously (Wu et al., 2000). They suggest that ERβ inhibits theactivity of the transcription factor AP-1 during gestation, thusblocking labour-associated genes, e.g. connexin 43. The AP-1binding site is activated by 17β-oestradiol only via ERα, whereasoestradiol via ERβ inhibits transcription (Paech et al., 1997).The collagenase gene has an AP-1 site in its promoter region(Jonat et al., 1992). Thus, the increased ERβ mRNA level inthe TP group could indicate a block of AP-1 activity and therebyproduction of metalloproteinases (MMPs, e.g. collagenase), thuspreventing the occurrence of the final cervical ripening. In thePP group, the ERβ expression was decreased again to the levelof the NP group, indicating a released inhibition. In addition,ERα expression had declined compared with the TP group,implying decreased ERα activity at the AP-1 site.

Several groups have reported increased concentrations ofdihydroepiandrosterone sulphate (DHEAS), a precursor to oestro-gen, in women with ripe cervices, compared with those withunripe cervices (Zuidema et al. 1986; Liapis et al. 1993), whileserum oestradiol showed no difference between the groups.Administration i.v. of DHEAS to women in late pregnancy wasalso shown to ripen the cervix (Mochizuki and Maruo, 1985).In-vitro studies on human cervical tissues showed increasedcollagenase activity in tissue cultures treated with DHEAS, whilefree steroids had no effect (Yoshida et al., 1993). These resultsimply that oestrogen, also via its precursor DHEAs, could inducecervical remodelling by regulating collagenase activity.

The endocervical mucosa has been found to contain thehighest, while the stroma and the squamous epithelium respect-ively contain lower and the lowest amounts of ERα (Sanbornet al., 1976). The present immunohistochemistry data showedthat both ERα and ERβ were expressed in the squamousepithelium and stroma. Both density and cell numbers of ERαand ERβ immunostaining were significantly changed in thestroma at term pregnancy and post-partum, whereas in theepithelium only ERβ immunoreactivity was decreased afterparturition and ERα remained unchanged. To our knowledge,cervical squamous epithelium does not play any important rolein the cervical ripening process. At present it is difficult to explainthe biological significance of the decreased ERβ expression inthe epithelium after spontaneous delivery. The present study andprevious data on the changed local endocrine environment withdecreased ERα and PR concentrations (Stjernholm et al., 1996,1997) but increased ERβ in term pregnancy with a following

299

decrease in ERβ after parturition, suggest that gonadal steroidsmay exert their main receptor mediated biological effects in thehuman cervix just before parturition.

It is known that the biochemical events that occur in thecervix during the cervical ripening are similar to those observedin tissue inflammation (Junqueira et al., 1980). Oestrogen hasbeen shown to have a role in leukocyte migration via regulationof granulocyte-macrophage colony-stimulating factor (GM-CSF)in uterine epithelial cells in mice (Robertson et al., 1996). Wefound that ERβ, but not ERα, showed intense immunostainingin neutrophils and the endothelial cells of the vessels. It istempting to speculate that neutrophils have an ability to respondto oestrogen in the cervix regulated via ERβ. Cervical collagenasehas been found to originate mainly from neutrophils accumulatingin the cervical stroma during parturition (Osmers et al., 1992).Extravasation of neutrophils, mast cells and monocytes withsubsequent release of proteolytic enzymes and cytokines mayrapidly accelerate the ripening process (Uldbjerg et al., 1983;Kelly et al., 1994). It has been suggested that the oestradiol-

regulated, and cytokine-mediated, up-regulation of the adhes-iveness of cervical vascular endothelium followed by extravasa-tion and degranulation of neutrophils, may play a crucial role inthe initiation of human parturition at term (Winkler et al., 1999).Our data, together to those of others, imply that ERβ might beinvolved in neutrophil invasion of the cervical stroma, andthereby directly or indirectly regulate tissue remodelling at term.

AcknowledgementsThe ERα cDNA was a generous gift from Donald McDonnell,Department of Pharmacology, Duke University Medical Center, Durham,North Carolina, USA. We gratefully acknowledge Eva Enmark forproviding the human ERβ cDNA and the ERβ 503 antibody was kindlyprovided by Jan-Åke Gustafson, both from Dept. of Medical Nutrition,Novum, Huddinge, Sweden. The present study received financial supportfrom Magn. Bergvalls stiftelse, the Swedish Medical Research Council(grants 03X-3972 and 04X-9508), Karolinska Institutet and SwedishSociety for Medical Research (HW).

ReferencesArts, J., Kuiper, G.G.J.M., Janssen, J.M.M.F. et al. (1997) Differential expression

of estrogen receptors α and β mRNA during differentiation of humanosteoblast SV-HFO cells. Endocrinology, 138, 5067–5070.

Calder, A.A. (1994) The cervix during pregnancy. In Chard, T. and Grudzinskas,J.G. (eds), The Uterus. Ch.13. p288–307.

Csapo, A.L., Knobil, E., van der Molen, H. et al. (1971) Peripheral plasmaprogesterone levels during human pregnancy and labour. Am. J. Obstet.Gynecol., 110, 630–633.

Frydman, L., Lelaidier, C., Baton, C. et al. (1992) Labor induction in womenat term with mifepristone (RU486): A double-blind, randomised, placebo-controlled study. Obstet. Gynecol., 80, 972–975.

Gordon, A.J. and Calder, A.A. (1977) Oestradiol applied locally to ripen theunfavourable cervix. Lancet, ii, 1319–1321.

Gorodeski, G.I. (1996) The cervical cycle. In Adashi, E.Y., Rock, J.A. andRosenwaks, Z. (eds), Reproductive Endocrinology, Surgery, and Technology.Vol. 1. Lippincott-Raven publishers, pp. 302–324.

Gorodeski, G.J. and Pal, D. (2000) Involvement of estrogen receptors α and βin the regulation of cervical permeability. Am. J. Physiol. Cell Physiol., 278,689–696.

Gronemeyer, H. (1992) Control of transcription activation by steroid hormonereceptors. FASEB J., 6, 2524–2529.

H.Wang et al.

Hall, J.M. and McDonnell, D.P. (1999) The estrogen receptor β-isoform (ERβ)of the human estrogen receptor modulates ERα transcriptional activity and isa key regulator of the cellular response to estrogens and antiestrogens.Endocrinology, 140, 5566–5578.

Junqueira, L.C.U., Zugaib, M., Montes, G.S. et al. (1980) Morphologic andhistochemical evidence for the occurrence of collagenolysis and for the roleof neutrophilic polymorphonuclear leukocytes during cervical dilation. Am.J. Obstet. Gynecol., 138, 273–281.

Jonat, C., Stein, B., Ponta, H. et al. (1992) Positive and negative regulation ofthe collagenase gene expression. Matrix Supplement, 1, 145–155.

Kelly, R.W., Illingworth, P., Baldie, G. et al. (1994) Progesterone control ofinterleukin-8 production in endometrium and choriodecidual cells underlinesthe role of the neutrophil in menstruation and parturition. Hum. Reprod., 9,253–258.

Kuiper, G.G.J.M., Enmark, E., Pelto-Huikko, M. et al. (1996) Cloning of anovel estrogen receptor expressed in rat prostate and ovary. Proc. Natl. Acad.Sci. USA., 93, 5925–5930.

Liapis, A., Hassiakos, D., Sarantakou, A. et al. (1993) The role of steroidhormones in cervical ripening. Clin. Exp. Obstet. Gynecol., 20, 163–166.

Mochizuki, M. and Maruo, T. (1985) Effect of dehydroepiandrosterone sulphateon uterine cervical ripening in late pregnancy. Acta Physiol. Hung., 65,267–274.

Osmers, R., Rath, W., Adelmann-Grill, B.C. et al. (1992) Origin of cervicalcollagenase during parturition. Am. J. Obstet. Gynecol., 166, 1455–1460.

Paech, K., Webb, P., Kuiper, G.G.J.M. et al. (1997) Differential ligand activationof estrogen receptors ERα and ERβ at AP1 sites. Science, 277, 1508–1510.

Robertson, S.A., Mayrhofer, G. and Semark, R.F. (1996) Ovarian steroidhormones regulate granulocyte-marophage colony-stimulating factor synthesisby uterine epithelial cells in the mouse. Biol. Reprod., 54, 183–196.

Sahlin, L., Norstedt, G. and Eriksson, H. (1994) Estrogen regulation of theestrogen receptor and insulin-like growth factor-I in the rat uterus: a potentialcoupling between effects of estrogen and IGF-I. Steroids, 59, 421–430.

Saji, S., Jensen, E.V., Nilsson, S. et al. (2000) Estrogen receptors α and β inthe rodent mammary gland. Proc. Natl Acad. Sci. USA, 97, 337–342.

Sanborn, B.M., Held, B. and Kuo, H.S. (1976) Hormonal action in humancervix – II Specific progestogen binding proteins in human cervix. J. SteroidBiochem., 7, 665–672.

Scheffe, H. (1959) The Analysis of Variance. John Wiley. New York, USA.Speroff, L., Glass, R. and Kase, N. (eds) (1989). Clinical Gynecologic

Endocrinology and Infertility. 4th edn. Williams & Wilkins, USA, pp. 317–350.Stjernholm, Y., Sahlin, L., Åkerberg, S. et al. (1996) Cervical ripening in

300

humans: potential roles of estrogen, progesterone, and insulin-like growthfactor-1. Am. J. Obset. Gynecol., 174, 1065–1071.

Stjernholm, Y., Sahlin, L., Malmstrom, A. et al. (1997) Potential roles forgonadal steroids and insulin-like growth factor I during final cervical ripening.Obstet. Gynecol., 90, 375–380.

Stjernholm, Y.M., Sahlin, L., Eriksson, H.A. et al. (1999) Cervical ripeningafter treatment with prostaglandin E2 or antiprogestin (RU486). Possiblemechanisms in relation to gonadal steroids. Eur. J. Obstet. Gynecol. Reprod.Biol., 84, 83–88.

Turnbull, A.C., Pattern, P., Flint, A. et al. (1974) Significant fall in progestroneand rise on oestradiol levels in human peripheal plasma before onset oflabour. Lancet, i, 101–104.

Uldbjerg, N., Ekman, G., Malmstrom, A. et al. (1983) Ripening of the humanuterine cervix related to changes in collagen, glycosaminoglycans andcollagenolytic activity. Am. J. Obstet. Gynecol., 147, 662–666.

Wang, H., Masironi, B., Eriksson, H. et al. (1999) The expression of estrogenreceptors α and β in the uterus of ovariectomized and steroid hormone treatedrats. Biol. Reprod., 61, 955–964.

Wang, H., Eriksson, H. and Sahlin, L. (2000) Estrogen receptors α and β in thefemale reproductive tract of the rat during the estrous cycle. Biol. Reprod.,63, 1331–1340.

Weihua, Z., Saji, S., Makinen S. et al. (2000) Estrogen receptor (ER) β, amodulator of ERα in the uterus. Proc. Natl Acad. Sci. USA, 97, 5936–5941.

Williams, J.K., Lewis, M.L., Cohen, G.R. et al. (1988) The sequential use ofestradiol and prostaglandin E2 topical gels for cervical ripening in high-riskterm pregnancies requiring induction of labor. Am. J. Obstet. Gynecol., 158,55–58.

Winkler, M. Kemp, B and Rath W. (1999) Abstract, the 5th internationalconference on the extracellular matrix of the female reproductive tract. June5–8, 1999. Stockholm, Sweden. Karolinska Institute.

Wu, J.J., Geimonen, E. and Anderson, J. (2000) Increased expression of estrogenreceptor β in human uterine smooth muscle at term. Eur. J. Endocrinol., 142,92–99.

Yoshida, K., Tahara, R., Nakayama, T. et al. (1993) Effect of dehydro-epiandrosterone sulphate, oestrogens and prostaglandins on collagenmetabolism in human cervical tissue in relation to cervical ripening. J. Int.Med. Res., 21, 26–35.

Zuidema, L.J., Khan-Dawood, F., Dawood, M.Y. et al. (1986) Hormones andcervical ripening: dehydroepiandrosterone sulphate, estradiol, estriol, andprogesterone. Am. J. Obstet. Gynecol., 155, 1252–1254.

Received on August 29, 2000; accepted on December 20, 2000