Molecular Human Reproduction Vol.12, No.3 pp. 157–167, 2006Advance Access publication March 23, 2006 doi:10.1093/molehr/gal014

© The Author 2006. Published by Oxford University Press on behalf of the European Society of Human Reproduction and Embryology. All rights reserved. For

Permissions, please email: journals.permissions@oxfordjournals.org 157

Trimegestone differentially modulates the expression of matrix metalloproteinases in the endometrial stromal cell

M.Wahab1, A.H.Taylor2, J.H.Pringle3, J.Thompson4 and F.Al-Azzawi5,6

1George Eliot Hospital, Nuneaton, Warwickshire, 2Reproductive Sciences Section, 3Cell Signalling Section, Department of Cancer Studies and Molecular Medicine, 4Department of Health Sciences, Faculty of Medicine and Biological Sciences, University of Leicester and 5Gynaecology Research Unit, University Hospitals of Leicester, Leicester, UK6To whom correspondence should be addressed at: Gynaecology Research Unit, University Hospitals of Leicester, Victoria Building, Leicester Royal Infirmary, Leicester LE1 5WW, UK. E-mail: farookalazzawi@yahoo.co.uk

Matrix metalloproteinases (MMP) are considered to be of critical importance in the initiation of menstruation where MMPprotein levels are reciprocally modulated by the actions of the gonadal steroid hormones, estradiol (E2) and progesterone (P4),with P4 being considered the principal suppressor of endometrial MMP expression. Trimegestone (T) is a novel progestagen thattightly controls menstruation timing and duration through mechanisms that might involve MMP suppression. Endometrial stro-mal cells treated with 10–6 M E2, P4 or T in the presence and absence of 10–6M RU486 showed that both T and P4 suppressed theexpression of MMP-1 and MMP–3 transcripts and secreted protein, whereas MMP-9 was not produced in culture. The suppres-sive effect of T or P4 on MMP-1 and MMP-3 transcript levels was enhanced in the presence of E2 and attenuated in the presenceof RU486, although MMP-1 proteins were unaffected by the presence of RU486, which alone acted as a partial progesterone ago-nist in these cultures. Immunohistochemistry with MMP-1, MMP-3 and MMP-9-specific antibodies performed on endometrialbiopsies obtained from non-treated, LH-dated, normally cycling women and endometrial biopsies obtained from postmenopausalwomen treated with T-based HRT showed that immunoreactive MMP-1 and MMP-3 was higher in the menstrual phase, whilstMMP-9 expression was higher in the late luteal phase (P = 0.03) and T significantly inhibited the presence of MMP-9+ cells. Thesedata suggest that T acts in a similar manner to P4, but causes subtle differences in expression patterns of MMPs that may explainthe different clinical effect that this progestagen has on endometrial behaviour compared to P4.

Key words: endometrium/HRT/metalloproteinases/progesterone/trimegestone

IntroductionDuring the menstrual cycle, the gonadal steroid hormones estradioland progesterone control the ordered growth and differentiation ofuterine cells. This remodelling process is critical for implantation ofthe developing embryo, the formation of the placenta and the mainten-ance of pregnancy (Pilka and Hrachovec, 2003). To achieve thisremodelling process the extracellular matrix (ECM) needs to be mod-ulated to allow cell movement and proliferation (Tabibzadeh, 1996).Indeed, the ECM components themselves are known to not only playimportant roles in the maintenance of tissue morphology, but alsoinfluence cell proliferation, differentiation and local apoptosis (Wangand Passaniti, 1999; Sivridis and Giatromanolaki, 2004).

Limited degradation of ECM is important to these remodellingprocesses and a major controlling factor is the expression of differentendometrial metalloproteinases (MMPs). In the absence of blastocystimplantation, degradation of the ECM is considered to be of criticalimportance in the initiation of tissue breakdown that leads to menstru-ation (Marbaix et al., 1996; Salamonsen and Woolley, 1996).

During the proliferative phase of the natural cycle, MMP-1, MMP-3and MMP-9 are thought to be expressed at low levels in the stroma(Hulboy et al., 1997), presumably to allow limited cell movementwithin the ECM. The expression of MMPs then decline in the earlysecretory phase, when the most stable endometrial structure isrequired, and then increase during the late secretory phase in anticipation

of the next proliferative phase. These data are the reciprocal of ovar-ian follicular and luteal phase serum progesterone levels, which hasled to the suggestion that endometrial expression of MMPs is undergonadal steroid hormone control.

Critically, MMP-9 expression is highest in the menstrual phaseendometrium when tissue breakdown occurs and this is thought to bedue to the fact that MMP-9 is expressed mainly in polymorphs, monocyte-macrophages and eosinophils (Hulboy et al., 1997) that infiltrate theendometrium at this critical phase of the menstrual cycle. What role/effect MMP-1 and MMP-3 might have on MMP-9 expression in theendometrium is unknown although they are clearly implicated in theimplementation of the menstrual phase of the cycle.

Progesterone is considered to be the principal suppressor of MMPexpression in the endometrium, but the mechanism involved is notclearly understood, especially as the tissue expression of MMPs in thenormal menstrual cycle declines before the serum progesterone levelsreach physiological significance. Many theories have been proposedfor the suppressive effects of progesterone including the activation ofPAI-1 (Casslen et al., 1992), increased expression of TGF-β (Bruneret al., 1995) or the stimulation of TIMP-1 expression (Marbaix et al.,1992) all of which are modulated during the menstrual cycle. How-ever, progesterone may itself have a direct suppressive effect that doesnot involve these intermediary molecules. Thus, if the onset of men-strual bleeding is a function of differential MMP expression, then this

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

M.Wahab et al.

158

hypothesis could be tested using a model where menstrual bleedingpatterns differ. Postmenopausal sequential combined HRT with a pro-gestagen that has a different pharmacological profile to that of proges-terone would provide such a model.

Trimegestone (RU27987) is a novel progestin that has a 6-foldhigh-affinity for human PR than P4 (Robin-Jagerschmidt et al. 2000;Winneker et al., 2003) and has a relatively long biological half-life ofapproximately 17 h (range 7–37 h; Bouchoux et al., 1995). This com-pound when administered in a sequential, combined manner withestrogen to post-menopausal women results in predictable menstrua-tion and reduced menstrual periods, and as such is as good if not betterthan other similar regimens (Al-Azzawi et al., 1999). To understandwhy this might be so, we have undertaken a detailed examination ofthe expression patterns of MMP in endometrial samples from womentreated with either 0.05 mg or 0.5 mg sequentially administered tri-megestone, who demonstrated different bleeding patterns and com-pared that MMP expression pattern to the pattern of expressionobtained from the normal menstrual cycle (Al-Azzawi et al., 1999).We have also examined the MMP expression patterns of trimeges-tone- and progesterone-stimulated stromal cell cultures that mimicthis sequential combined treatment regimen and show differences inMMP-1 and MMP-3 expression profiles.

Materials and methods

MaterialsThe gonadal steroids 17β-estradiol (E2) and progesterone (P4) were from Sigma(Poole, Dorset, UK). Trimegestone (RU27987) and mifepristone (RU486) werekind gifts from Hoechst Marion Roussel (Romianville, France). The FirstResponse® Urinary LH kits were from Carter Wallace Limited (Folkestone, UK).

Stromal cell cultureStromal cells were isolated from late proliferative phase endometrium obtainedfrom hysterectomy specimens as described (Vigano et al., 1993; Hatayama et al.,1994). Briefly, finely minced endometrial tissue was incubated in 20ml of 0.25%bacterial collagenase type I (Gibco) in Hanks’ buffered salt solution (Gibco), for2 h at 37°C. Non-digested material was collected at unit gravity for 5 min, thesupernatant removed to a fresh tube and cells collected at 350 x g for 5 min. Thecell pellet was then washed six times in growth medium; phenol red-freeDMEM/F-12 medium (Gibco) supplemented with 10% charcoal-dextran treatedfetal calf serum (Hyclone, Logan, Utah, USA), antibiotics/antimycotic solution(Sigma; Penicillin 10,000 units/ml, streptomycin 10 mg/ml, Amphotericin B25μg/ml) and 1.25mg/ml sodium bicarbonate (Sigma). The cellular pellet wasresuspended in growth medium and seeded into a 25 mm2 flask (NUNC, FisherScientific, Loughborough, UK) and incubated at 37°C in 5% CO2 in humidifiedair until the culture reached confluence. The cells were then removed withtrypsin-EDTA solution (Gibco), washed twice with growth medium and platedinto six-well plates at a density of 0.8–1.0 × 105 cells/ml. Cells were also seededon polytetrafluoroethane-coated multispot glass slides (C.A.Hendley Ltd.,Loughton, UK) for vimentin immunocytochemical staining. After 24 h, non-adherent cells were removed by gentle washing with pre-warmed growthmedium and the adherent cells used in steroid-treatment experiments.

Immunocytochemistry for vimentinTo confirm that each culture was of stromal origin, cultures grown on glass cov-erslips were stained with mouse monoclonal anti-vimentin antibody (1 : 100;Novacastra, Vector Laboratories, Peterborough, UK) as previously described(Vigano et al., 1993; Laird et al., 1996; Shiokawa et al., 1996). In each culturemore than 95% of the stromal cells were positive for vimentin (Figure 1).

Effect of estradiol on progesterone receptor expressionTo confirm the steroid responsiveness of our stromal cell cultures, 0.8–1.0 x 105

cells were plated to six-well plates in triplicate and cultured for 4 days in thepresence of 0.1% ethanol until the cultures reached confluence. The culturemedium was then changed for one containing either 10–8M E2 or vehicle for

4 days in the presence or absence of 10μg/ml cycloheximide (Fisher Scientific).The medium was changed every 48 h. Cellular mRNA was then extractedusing Trizol (Invitrogen, Paisley, UK) according to the manufacturer’s instruc-tions and treated with DNase I and phenol–chloroform extraction prior toreverse transcription. Total cellular protein was produced using proceduresdescribed previously (Taylor et al., 2002) and pooled. After concentration ofproteins with 30,000 Dalton cut off concentrators (Pierce, Perbio Science,Cramlington, UK), protein concentration was determined by Bradford assay(Bradford, 1976; Bio-Rad Laboratories, Hemel Hempstead, UK). Proteins(100 μg) were resolved on 7.5% SDS-PAGE and analysed for the presence ofprogesterone receptor isoforms as described (Taylor et al., 2006).

Steroid treatment of stromal cell culturesTwo separate treatment protocols were used to mimic the normal menstrualcycle and trimegestone HRT treatment regimens (Figure 2). In one set of exper-iments, stromal cells cultured in triplicate wells were exposed to 10–6 M E2 dis-solved in growth medium or an equivalent amount of solvent (0.1% ethanol) for4 days, with the medium being changed on day 2. Next, medium containingeither 10–6 M progesterone alone, 10–6 M RU486 alone or a combination ofthese steroids in the presence of E2 was added to the cultures for an additional4 or 6 days with the medium being changed on days 6 and 8 (Figure 2). Inanother set of experiments, 10–6 M trimegestone was used instead of progester-one. At the end of the culture period, the medium was removed, centrifugedat 10,000 g for 30 s to remove cells and debris, and the supernatants stored at−80°C for MMP measurement by specific ELISA. The cells that remainedattached to the plasticware were then processed for cellular mRNA.

Extraction of mRNA for MMP analysisCellular mRNA was extracted and processed using oligo-dT-linked Dynabeads®(Dynal, Bromborough, UK), as described previously (Bicknell et al., 1996).Briefly, cells in each well were lysed in 500μl lysis/binding buffer (100 mMTris-HCl (pH 8), 0.5 M LiCl, 10 mM EDTA, 5 mM dithiothreitol, 1% lithiumdodecylsulphate (LiDS)) and incubated with 50 μg/ml proteinase K (RocheDiagnostics Ltd., Lewes, UK) for 1 h at 37°C. DNA was sheared by passing thepreparation through a 21G and then a 25G needle 5 times. The lysate was centri-fuged for 30 s at 10,000 g and the supernatant mixed with oligo-dT-linkedDynabeads® (Dynal). The mRNA was allowed to anneal to the Dynabeads®for 10 min at room temperature. After collecting the beads on a magnet,mRNA-linked Dynabeads® were washed twice in a buffer that consisted of10 mM Tris-HCl (pH 8), 150 mM LiCl, 1 mM EDTA, 0.1% LiDS (Dynal) andthen washed twice in the same buffer but without LiDS. The mRNA-linkedDynabeads® were finally resuspended in 30 μl of DEPC-treated de-ionizedwater and used immediately in the reverse transcriptase reaction.

Reverse transcriptase reaction (RT)The mRNA mixture was reverse transcribed at 42°C for 1 h as a 25μl mixture con-taining 10μl of mRNA-Dynabead® preparation or 1μg of total cellular RNA, 1μlDMSO, 2.5μl of 5 x AMV-RT buffer (Promega, Southampton, UK), 0.5μl 10 mMdNTPs (Roche), 0.25μl RNAse inhibitor (25U, Promega), 10.25μl DEPC-treated

Figure 1. Immunohistochemical staining of primary stromal cell cultures withanti-vimentin antibodies. Bar =100μm.

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

Trimegestone and endometrial MMPs

159

de-ionized water and 0.5μl AMV-RT (5U, in the +RT reaction, Promega). The–RT control for each sample was achieved by omitting the AMV-RT enzyme.

PCRPCR was performed using a hot-start procedure for GAPDH and MMP-1,MMP-3, MMP-9 and PR using mRNA-specific primers (Table I). Each PCR(50μl) contained 1μl of the +RT or –RT mixture, 42μl DEPC-treated de-ionized

water, 10pmol of the forward and reverse primers and 5μl of 10x AJ Buffer(450 mM Tris-HCl (pH 8.8), 110 mM NH4SO4, 45 mM MgCl2, 2 mM dNTPs,1.1 mg/ml acetylated BSA (Roche), 110 mM β-mercaptoethanol, 4.4 μMEDTA). After an initial denaturation step at 94°C for 2 min, the samples wereheated at 95°C for 30 s, and then held at the annealing temperature for the spe-cific mRNA target; GAPDH (60°C), MMP-1 (62°C), MMP-3 (56°C), MMP-9(62°C), PR (55°C) for 10 min. At this point, one unit of Taq polymerase(Promega) was added and the PCR cycles initiated. The PCR conditions were

Figure 2. Illustrations of treatments used in primary endometrial stromal cell culture. (A) Stromal cells seeded to six-well plates and were treated with either 17β-estradiol (E2) or ethanol (ETOH) control for 4 days. This was then followed by the addition of E2, progesterone (P4), trimegestone (T) or RU486 or a combinationof these steroids for an additional 4 days. (B) In a second experimental protocol, stromal cells were treated for 4 days with E2 or ETOH and then E2, P4, T or a com-bination of these steroids for an additional 6 days.

ETOH

ETOHETOH

P4

ETOH

E2

RU486

Day 4 Day 8

E2 + T

E2 + T+ RU486

E2 + RU486

E2 + P4

E2 + P4 + RU486

Day 0

ETOH

E2

E2

E2

E2

E2

A

Control

E2

P4

R

E2+P4

E2+R

E2+P4+R

E2+T

E2+T+R

B

ETOH

ETOHETOH

P4

ETOH

E2

T

Day 4 Day 10

E2 + T

E2 + P4

Day 0

ETOH

E2

E2Control

E2

P4

T

E2+P4

E2+T

Table I. Oligonucleotide primer sequences and expected amplicon sizes

aFrom Hall et al. (1998).bHirata et al. (2002).

mRNA species Oligonucleotide sequence Product size (bp)

Annealing temperature (°C)

Accession number

GAPDH5′ 671AGAACATCATCCCTGCCTC689 347 60 X01677a

3′ 1017GCCAAATTCGTTGTCATACC998

MMP-15′ 321AAGGTGATGAAGCAGCCCAGAT342 402 62 NM-002421.13′ 722ATGAGCCGCAACACGATGTAAG701

MMP-35′ 411TCTGAAAGTCTGGGAAGAGGTG432 287 56 J032093′ 697GTCACAACCGACTCACTTTCTC676

MMP-95′ 1328AATGGCATCCGGCACCTCTATG1349 373 65 J050703′ 1700GCCACTTGTCGGCGATAAGGAA1679

PR5′ 2592AGGAGTTTGTCAAGCTTCAA2611 284 55 M15176b

3′ 2875CTGCAGGGACTGGATAAATGTATTC2851

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

M.Wahab et al.

160

30 s at 94°C, 30 s at the annealing temperature, 45 s at 70°C, followed by anextended extension period of 10 min at 70°C. The number of cycles were 28, 32,23, 35 and 35 for GAPDH, MMP-1, MMP-3, MMP-9 and PR, respectivelywhich were all within in the exponential phase of amplification (data not shown).

The PCR products were separated on 3% agarose gels, stained with ethidiumbromide (2μg/ml) and the images captured under UV transillumination with adigital camera (UVP, Cambridge, UK) using an UVP image store 5000 system(UVP, Cambridge, UK). Densitometry measurements of the amplicon densi-ties were performed using the Scion Image analysis software (version beta4.0.2, Scion Associates, Frederick, Maryland, USA) using KODAK neutraldensity step tablets to calibrate the densitometry images.

MMP-1 and MMP-3 ELISAMMP-1 and MMP-3 ELISA were performed using specific ELISA kits (totalMMP Biotrak® Assay, Amersham Biosciences, Chalfont St Giles, UK)according to the manufacturer’s instructions. The samples were analysed induplicate with the MMP-3 samples needing to be diluted 1 : 250 with samplediluent to fit on the standard curve. The data were corrected for cell numberand are presented as amount of MMP in ng/ml.

Patient groupsTrimegestone-treated groupThe Leicestershire Local Ethics Committee approved the clinical trial andresearch protocol for this part of this study and all patients signed informedconsent. The details of these patients are described in detail elsewhere(Al-Azzawi et al., 1999; Wahab et al., 1999a,b). Briefly, endometrial biopsieswere obtained using vabra curette on day 24 of the last treatment cycle. Of the 176women recruited in Leicester, those who received the highest (0.5 mg, n = 19) andthe lowest (0.05 mg, n = 21) doses of trimegestone were included in the presentstudy due to their marked contrast in bleeding patterns (Al-Azzawi et al., 1999).

Normal groupThe normal endometrial group consisted of 35 healthy, regularly menstruatingwomen aged 27–50 (38 ± 6.1, mean ± SD), undergoing laparoscopic steriliza-tion using sharp curette or hysterectomy for cervical intraepithelial neoplasia,premenstrual tension syndrome or benign ovarian cysts. None of these womenhad received any hormonal treatments for 2 months prior to specimen procure-ment. All women were given urinary LH surge detection kit tests (FirstResponse, Carter Wallace limited, Folkstone, UK), which were used during themonth preceding the endometrial biopsy or hysterectomy. The normal controlsamples were deep endometrial biopsies (which included functionalis andbasalis layers), obtained by opening the uterus by coronal section and by theslicing of the posterior wall of the uterine cavity vertically from fundus to isthmus,each slice being 1 cm wide. These were coded alphabetically A–E from centreto edge. Our study materials were from slice B.

The specimens were fixed in 10% formal-saline, embedded in paraffin wax,and 5 μm sections stained with haematoxylin and eosin for histological assess-ment of the phase of the menstrual cycle. These biopsies were then dated bothby LH surge data and the date of the last menstrual period, were examined bytwo independent pathologists who were blinded to the LH surge and menstrualdates and where all agreed, the specimen was included as a control sample.The histological diagnoses were proliferative (n = 8), early secretory (n = 8),late secretory (n = 11) and menstrual (n = 8).

MMP immunohistochemistrySections measuring 5 μm in thickness were dried to silane-coated slides for48–72 h before use. After dewaxing in xylene and re-hydration through gradeddoses of alcohol to water, antigens were retrieved using microwaving at 750Wpower in 10mM citrate buffer at pH 6.0, for 30 min. Endogenous peroxidaseactivity was then blocked by incubation with 6% hydrogen peroxide for10 min. Normal goat serum (NGS; DAKO, Glostrup, Denmark) diluted to 10%(v/v) in phosphate-buffered saline (PBS) was then applied to minimize non-specific reactivity. The sections were then incubated with the primary mousemonoclonal antibody (MMP-1; 2μg/ml, MMP-3 or MMP-9: 3μg/ml; Chemi-con International Ltd, Harrow, UK) diluted in 10% NGS for 18 h at 4°C. Afterwashing with PBS for 20 min, sections were then incubated with biotinylatedgoat anti-mouse immunoglobulins (1 : 400, DAKO), washed with PBS for

20 min, then with Vectastain ABC-linked horseradish peroxidase reagent(ABC Elite; Vector Laboratories, Peterborough, UK) for 30 min, washed againwith PBS for 20 min, incubated with 3,3´-diaminobenzidine (DAB substrate;Vector Laboratories, Peterborough, UK) for 5 min, lightly counterstained withMayer’s haematoxylin (Sigma) and permanently mounted in XAM mountant(BDH, Poole, UK). Controls included the use of mouse IgG (Sigma) in theplace of the primary antibody or the omission of the primary antibody. Thepositive controls (placenta and breast carcinoma) were stained with MMP-1,MMP–3 and MMP-9, and there was no background reaction, while the nega-tive control was devoid of any immunoreactivity (data not shown).

Assessment of MMP immunoreactivityAn observer blinded to the clinical data performed the qualitative and quantita-tive histomorphometric analyses of the positively stained cells. Histomorpho-metric analysis was performed on the stromal compartment by counting thenumber of positively stained stromal cells per field from 10 randomly selectedfields (Hamilton, 1995) per slide captured (x1000) under oil immersion from thefunctionalis layer. Images were captured using an Axioplan microscope (CarlZeiss, Welwyn Garden City, UK) and a colour video camera (Sony CCD/RGB).The cells counted were evaluated using the KS300 image analysis computer pro-gram (Kontron Imaging Systems, Thame, UK). The data presented are numberof positive cells per unit area and the corresponding areas of the images capturedat 1000 magnification were 0.0046 mm2 using a measurement graticule.

Statistical analysisPR transcript and protein levels were compared using one-way ANOVA withTukey’s HSD test. MMP transcript levels from primary stromal cell cultureswere analysed using a mixed model of the log of the ratios of MMP:GAPDHdensitometric values that incorporates both subjects and treatments. The Waldtest was used to compare treatment effect. Non-parametric Kruskal–Wallisone-way analysis of variance with Dunn’s post-test was used to compare treat-ment effect in the MMP protein ELISA measurements. The immunohisto-chemical data of endometrial sections did not fulfil the assumptions necessaryfor using analysis of variance and t-test; therefore, the non-parametricKruskal–Wallis and Mann–Whitney tests were used.

ResultsThe effect of P4 and T on MMP transcript levelsStromal cells in culture expressed transcripts for MMP-1 and MMP-3mRNA (Figure 3A) but not for MMP-9 (Figure 3B). Using experi-mental protocol number 1, where stromal cells were stimulated withE2 for a total of 8 days, E2 suppressed MMP-1 and MMP-3 mRNA

Figure 3. Ethidium bromide-stained agarose gels for MMPs and GAPDH. (A)The resolving cDNA bands of MMP-1, MMP-3 and GAPDH are shown forthree separate stromal cell culture extracts (lanes 1–3). The 100bp relative sizemarker is shown (lane bp) with sizes indicated on the left-hand side. (B) PCRgel indicating the presence of MMP-9 PCR products for proliferative (lane 1)and early secretory (lane 2) endometrial extracts, but absence of MMP-9 signalfor stromal cell cultures (lane 3).

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

Trimegestone and endometrial MMPs

161

levels compared to the vehicle control, but only MMP-1 transcriptlevels were significantly suppressed by E2 treatment (Figure 4). P4alone or in combination with E2 significantly suppressed MMP-1 andMMP-3 transcript levels (P<0.01 and P=0.001, respectively). Theantiprogestin, RU486, suppressed MMP-1 and MMP-3 expression by1.69- and 1.48-fold, respectively, but these data did not reach statist-ical significance. The combination of E2 and P4 caused a significant2.88-fold decrease in MMP-1 mRNA levels (P<0.01) and an 8.13-folddecrease in MMP-3 mRNA levels (P<0.01). When RU486 was com-bined with E2 and P4, RU486 completely ablated the E2/P4-regulatedsuppression of both MMP-1 and MMP-3 transcripts (Figure 4).

In this experiment, trimegestone alone had a significant suppressiveeffect on MMP-1 (P<0.01) and MMP-3 (P<0.05) transcript levels butwas not as effective as P4 alone (Figure 4). The combination of E2 andT further suppressed MMP-1 and MMP-3 mRNA levels from 1.51-fold to 4.46-fold for MMP-1 and 2.24-fold to 3.55-fold for MMP-3,and the suppression of MMP-1 was significantly more pronouncedthan the combination of E2 and P4 (P=0.011). The suppressive effectsof P4 or T when administered alone were not statistically different.The suppression of MMP-1 mRNA by E2 and T was significantly

reversed upon the addition of RU486 (P<0.001). However, this atten-uation was not complete for MMP-3 since mRNA levels were still sig-nificantly below that of the controls (P=0.020). By contrast theattenuation of the MMP-1 was complete and MMP-1 mRNA levelswere not different from that of the control cultures (P=0.36).

To examine whether extended stimulation of stromal cells with P4or T would lead to further suppression of MMP-1 or MMP-3 tran-script levels, cells were treated with steroid for 6 days (experimentalprotocol number 2; Figure 2). Stimulation with trimegestone or P4alone significantly suppressed both MMP-1 and MMP-3 expression(Figure 5) as was observed in experimental protocol number 1. Simi-larly, the combination of E2 with either T or P4 suppressed the levelsof both MMP transcripts further; however, in this extended stimula-tion protocol the suppression of MMP-3 mRNA transcripts were nowsignificantly (P<0.05) suppressed compared to P4 alone (Figure 5),whereas in the shorter protocol they were not (Figure 4). The extra 2days of stimulation also enhanced the suppressive effects of T in thepresence of E2, although these data did not reach significance(P=0.39).

The effect of P4 and T on secreted MMP protein levelsStromal cell cultures produced and released between 100- and 167-times more MMP-3 than MMP-1 (Figure 6). Progesterone, RU486and trimegestone reduced the secreted levels of MMP-1 in line withmRNA levels (Figures 4 and 5), but these protein data did not reachstatistical significance. However, MMP-1 protein levels were signifi-cantly suppressed by the E2/T combination (P<0.05) and the additionof RU486 to the cultures did not attenuate these effects with the levelsof MMP-1 remaining suppressed (Figure 6).

Similar observations were made with the measurement of secretedMMP-3 protein levels in that E2, P4, RU486 and T all suppressedMMP-3 protein levels but these data did not reach significance. How-ever, the combination of E2 and P4 significantly reduced MMP-3 pro-tein levels and that level was not reversed by the addition of RU486.Similarly, the addition of E2 to T-stimulated cultures caused a signi-ficant (P<0.05) decrease in MMP-3 levels, but in this case the additionof RU486 attenuated T actions and the levels of MMP-3 returned tocontrol values.

Figure 4. Log ratios of mRNAs encoding MMP-1 (A) or MMP-3 (B) toGAPDH for stromal cell cultures treated with steroids under experimental pro-tocol number 1. Data are presented as mean ± SE from six separate experi-ments performed in triplicate. C=control, E2=17β-estradiol, P4=progesterone,R=RU486, T=trimegestone. *P<0.05, **P<0.01, ***P<0.001 compared tocontrol. ψP<0.05 compared to E2.

θP<0.05 statistically significant reversal byRU486 compared to E2+P or E2+T.

C E2 P4 R E2+P4 E2+R E2+P4+R T E2+T E2+T+R

-1.0

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

Lo

g (

MM

P-1

:GA

PD

H)

***

***

ψ

ψ

θ θ

ψ ψ

θ

θ

*

**

C E2 P4 R E2+P4 E2+R E2+P4+R T E2+T E2+T+R

-1.0

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

Lo

g (

MM

P-3

:GA

PD

H)

**

*** **

*

A

B

Figure 5. Log ratios of mRNAs encoding MMP-1 (?) or MMP-3 (¦) toGAPDH for stromal cell cultures treated with steroids under experimental pro-tocol number 2. Data are presented as mean ± SE from six separate experi-ments performed in triplicate. C=control, E2=17β-estradiol, P4=progesterone,T=trimegestone. *P<0.05, **P<0.01, ***P<0.001 compared to control.ψP<0.05 compared to E2.

θP<0.05 compared to P4 or T alone.

C E2 P4 E2+P4 T E2+T

-1.0

-0.8

-0.6

-0.4

-0.2

0

0.2

Lo

g (

MM

P-1

or

MM

P-3

:GA

PD

H)

*

MMP-1

MMP-3

*

**

*

***

**

ψ

ψ ψ

ψ

θ

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

M.Wahab et al.

162

Effect of E2 on PR expressionStromal cell cultures responded to E2 stimulation by significantly(P<0.01) increasing PR transcript levels by 3.1-fold (Figure 7A and B)and significantly (P<0.05) increasing PR protein levels by 2.4-fold(Figure 7C and D). The addition of the protein synthesis inhibitorcycloheximide had no effect on either basal or E2-stimulated PR tran-script levels (Figure 7A and B) but had a minor non-significant effecton GAPDH transcript levels (Figure 7A).

Comparison of MMP-1 stainingMMP immunoreactivity was present in all samples assessed. Theimmunoreactive staining pattern for MMP-1 in endometrial stromalcells was found not to be uniform, with different cells scatteredthroughout the stroma showing nuclear and perinuclear immunoreac-tivity (Figure 8A). Although there were considerable differencesbetween individual patients within each phase of the menstrual cycle,MMP-1 immunoreactivity was not modulated by the phase of cycledespite an apparent nadir in MMP-1 immunoreactivity in the earlysecretory phase of the menstrual cycle (Figure 9A); this did not reachstatistical significance. There was no difference in the immunoreac-tive profile of stromal MMP-1 immunoreactivity in the 0.05 mg and0.5 mg trimegestone HRT groups although the mean number of positive

cells in the low-dose trimegestone group was significantly (P<0.05)different from the early secretory group of the normal menstrual cycle(Figure 9A). The glandular epithelial cells were 100% positive forMMP-1 throughout the menstrual cycle (data not shown). Unlike stro-mal cells, immunoreactivity was observed throughout the cytoplasm.

Comparison of MMP-3 stainingThe immunoreactive staining pattern for MMP-3 in endometrialstromal cells was scattered throughout the functionalis and in commonwith MMP-1 staining appeared to be perinuclear (Figure 7B). The cellcounts for MMP-3+ stromal cells in different phases of the normalmenstrual cycle and in the trimegestone treated groups were found tobe approximately 10-fold lower than either MMP-1+ or MMP-9+ cellnumbers (Figure 9). MMP-3 immunoreactivity was not modulated bythe phase of cycle despite a 20% increase in mean MMP-3+ cell countfrom the late secretory to menstrual phases of the cycle (Figure 9B).The immunoreactive profile of stromal MMP-3 immunoreactivity forthe 0.05 mg and 0.5 mg trimegestone HRT groups did not differ fromeach other or with the normal menstrual cycle samples, despite therebeing an apparent T-dependent dose-effect. MMP-3 immunoreactivitywas absent from glandular epithelial cells (Figure 8B).

Comparison of MMP-9 stainingThere was a statistically significant difference in MMP-9 expressionbetween the phases of the menstrual cycle (P=0.03, Figure 9C), withthe highest expression of MMP-9 in the late secretory phase comparedto the proliferative or menstrual phases (Figure 9C; P=0.02, P=0.03,respectively). By contrast, trimegestone-treated endometria had sig-nificantly lower levels of MMP-9+ cells compared to early secretoryand late secretory phase endometria, presumably indicating theabsence of large numbers of infiltrating plasma cells (Figure 8C). Incommon with MMP-1 staining, MMP-9 immunoreactivity in the glan-dular epithelial cells was 100% positive throughout the menstrualcycle and the T-treated endometria (data not shown). Again stainingwas observed throughout the cytoplasm and not confined to the peri-nuclear space as it is in the stromal cell (arrow in Figure 8C).

DiscussionIn a previous study (Al-Azzawi et al., 1999) we reported that patientstreated with 0.5mg trimegestone/day over a 14 day period, as part of asequential combined HRT, displayed better control of the day of onsetof withdrawal bleeding and a much lower incidence of breakthroughbleeding than women on low dose trimegestone or normally menstru-ating women. Additionally, the pattern of endometrial histomorpho-metric features exhibited by the trimegestone-treated groups differedto those obtained from normally menstruating women (Wahab et al.,1999b). These data suggested that trimegestone differed from proges-terone in its actions in the regulation of menstruation.

The mechanisms governing endometrial tissue shedding are notwell understood, but it is becoming clear that MMPs play a key role inthe cyclical breakdown of ECM that ultimately leads to menstruation.Because MMPs are able to digest matrix proteins under physiologicalconditions, activation and inhibition of MMP expression must betightly controlled to prevent inappropriate or untimely tissue destruc-tion. In this regard the ovarian steroid hormones appear to play a keyrole with focal steroid receptor expression being paramount (Kokrineet al., 1996).

In this study, we found that MMP-1 and MMP-3 mRNA and pro-tein were differentially regulated by P4 and T through an E2-dependentpathway. MMP-1 and MMP-3 transcript levels were suppressed instromal cells treated with P4 or T, and this suppression was more

Figure 6. Effect of steroid treatment on secreted MMP-1 (A) or MMP-3 (B)levels. Data are from the number of independent experiments indicated and arepresented as the mean ± SE. *P<0.05, **P<0.01 compared to control. ψP<0.05compared to E2 (MMP-1) and E2+T+R compared to E2+T (MMP-3).

A

B

control E2 P4 R E2+P4 E2+R E2+P4+R T E2+T E2+T+R

0

2

4

6

8

10

12

14

MM

P-1

(n

g/m

l)

* ψ ψ

12 6 12 12 6 12 6 6 6 6

*

control E2 P4 R E2+P4 E2+R E2+P4+R T E2+T E2+T+R

0

500

1000

1500

2000

2500

MM

P-3

(n

g/m

l)

**

**

ψ

15 3 15 15 3 15 3 12 12 12

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

Trimegestone and endometrial MMPs

163

profound when either steroid was combined with E2. The mechanisminvolved increased expression of stromal cell progesterone receptor(s)by E2 (Figure 7) as previously reported (Eckert and Katzenellenbogen,1981; Marbaix et al., 1995), but not de novo synthesis of factorsinvolved in PR mRNA synthesis as evidenced by a lack of effect by

cycloheximide on PR transcript levels. An interesting observation wasthat the levels of MMP-1 and MMP-3 mRNA and protein were con-sistently higher in the untreated samples suggesting that MMP-1 andMMP-3 are readily produced by stromal cells, either inherently in theabsence of gonadal steroids, or as a result of the culture conditions.

Figure 7. Effect of E2 on stromal cell progesterone receptor transcript and protein levels. Representative ethidium bromide-stained agarose gels (A) showing PRand GAPDH transcript levels obtained from cultures incubated with vehicle (lanes 1 and 2) or with 10–8M E2 for 4 days (lanes 3 and 4) either in the presence (lanes2 and 4) or absence of 10μg/ml cycloheximide (lanes 1 and 3). Densitometric analysis of the data shown in panel A was converted into PR:GAPDH ratios (B) andare vehicle alone (c), cycloheximide alone (cyclo), estradiol (E2) alone, or both E2 and cycloheximide (E2+cyclo). Representative immunoblots (C) for PR proteinobtained from (1) T47D breast carcinoma cell extracts, extracts from stromal cells incubated with 0.1% ethanol (2) or 10–8M E2 for 4 days (3). Relative molecularmass markers are indicated on the left and the identities of the different PR isoforms are shown on the right. Densitometric analysis of the data shown in panel C isshown in panel D. Data are the mean ± SD for three experiments performed in triplicate. *P<0.05, **P<0.01 compared to the untreated control.

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

M.Wahab et al.

164

Whatever the mechanism, it is clear from these observations that stro-mal cells in vivo are likely to be activated by E2 and P4 in the lutealphase of the normal cycle, suppressing MMP-1 and MMP-3 expres-sion. Indeed, these observations are supported by other studies whereother progestagens inhibited MMP expression (Salamonsen et al.,1997; Schatz et al., 1997; Hampton et al., 1999).

In experiments where stromal cells were subjected to T or P4 incombination with E2, the E2/T combination consistently led to agreater suppression of MMP-1 mRNA and protein levels when com-pared to that of the E2/P4 combination, and this is probably due to themuch higher relative binding affinity and protracted ligand-receptor

occupancy (Winneker et al., 2003) of trimegestone (330%) comparedto progesterone (50%) or promegestone (100%) to the progesteronereceptor (Wiegratz and Kuhl, 2004).

Similarly, levonorgestrel (LNG) and norethindrone (NET), whichare characterized by both high progesterone and androgen receptorbinding affinities (Sitruk-Ware, 2004b), suppressed MMP levels tothat of P4 (Hampton et al., 1999) indicative of a possible androgen

Figure 8. Immunohistochemical staining of endometrial sections for MMP-1,MMP-3 and MMP-9. Representative MMP-1 (A) MMP-3 (B) and MMP-9 (C)staining of a late secretory endometrium specimen; MMP9+ glandular epithe-lial cells are indicated by the arrow. Bar = 100μm. Figure 9. Histomorphometric analysis of the number of stromal cells positive

for MMP-1 (A), MMP-3 (B) and MMP-9 (C). P= Proliferative (n=8),ES=Early secretory (n=8), LS=Late secretory (n=11), M=Menstrual (n=8),0.05 and 0.5 = 0.05mg (n=21) and 0.5mg (n=19) of trimegestone, respectively.Data are mean ± SE, *P<0.05 compared to ES (MMP-1) or compared toP (MMP-9). **P<0.05, ψP<0.05 compared to ES and LS, respectively (MMP-9).

P ES LS M 0.05 0.5

0

2500

5000

7500

10000

No

. o

f p

os

itiv

e c

ell

s/m

m2

MMP-1*

Trimegestone

P ES LS M 0.05 0.5

0

200

400

600

800

1000

No

. o

f p

os

itiv

e c

ell

s/m

m2

MMP-3

Trimegestone

P ES LS M 0.05 0.5

0

2000

4000

6000

8000

No

. o

f p

os

itiv

e c

ell

s/m

m2

MMP-9*

ψ **

Trimegestone

A

B

C

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

Trimegestone and endometrial MMPs

165

receptor (AR) mediated effect. To test this idea, the potent type I anti-progestin RU486 was used to mimic the effect of progesteronewithdrawal by competition for, and conformational alteration of, pro-gesterone receptor binding sites (Baulieu et al., 1987). The data indi-cated that RU486 blocked the actions of both P4- or T-inducedsuppression of MMP-1 and MMP-3 expression by increasing bothMMP-1 and MMP-3 mRNA and protein levels in agreement with thepublished literature (Schatz et al., 1997). However, RU486 did notreturn protein and mRNA levels back to baseline values indicatingthat equimolar concentrations (10–6 M) of RU486 was insufficient toprevent P4- and T-induced MMP-1 and MMP-3 repression, suggest-ing that a type II antiprogestin would have been more useful, or thatsome of the effects on stromal cell MMP-1 and MMP-3 repression aremediated through glucocorticoid or androgen receptors. Alternatively,since RU486 can act as a partial agonist/antagonist in several tissues(Gompel et al., 1986; Robbins and Spitz, 1996; Dai et al., 2003),including the endometrial glandular epithelial cell (Taylor et al.,1998), then part of the observed MMP-1 and MMP-3 repression couldbe directly related to RU486 agonist activity. Indeed, RU486 alonepartially inhibited MMP-1 and MMP-3 transcript levels and produceddifferential effects on T-induced MMP-1 and MMP-3 protein levels,where it blocked T-induced MMP-3 suppression, but not T-inducedMMP-1 suppression, suggesting other non-PR mediated pathways for T.The differential action of RU486 on T-induced MMP-3 suppressiveaction but not MMP-1 suppression might be due to the differentialinhibitory actions of P4 and T, or binding potentials at different PR-isoforms. Alternatively, RU486 has recently been documented as hav-ing inhibitory effects through the androgen receptor (Agoulnik et al.,2003; Song et al., 2004), and as RU486 binds to the androgen receptorwhereas progesterone and trimegestone do not, despite trimegestonehaving significant anti-androgenic activity (Winneker et al., 2003);perhaps this may explain the discrepancy between P4 and T in thepresence of RU486? Alternatively, the discrepancy may be due to therelative stabilities of the MMP-1 and MMP-3 proteins in culture, as isshown for MMP-9 (Singer et al., 1999).

The differential regulation of MMP-1 and MMP-3 are unlikely tobe due to the phase of cycle chosen to produce the stromal cell cul-tures because basal MMP expression levels are not affected by phaseof the menstrual cycle from which the cultures arise (Rawdanowiczet al., 1994). Additionally, vimentin immunocytochemical staining(Kamelle et al., 2002) for stromal cells indicated that the cultures were>95% vimentin positive demonstrating that contaminating endome-trial epithelial cells were not the source of the MMP-1 and MMP-3differential expression patterns.

Our primary stromal cell cultures did not express MMP-9 mRNA(Figure 3) indicating that stromal cells in culture lack the ability toexpress MMP-9, because MMP-9 is considered to be confined to infil-trating leukocytes and epithelial cells that were removed in the pro-cess of generating the cell cultures. Indeed, stromal cells in culture canonly transiently express MMP-9 transcripts after tumour necrosis fac-tor alpha-α (TNF-α) stimulation and only at the 6 h time-point (Singeret al., 1999). The clinical or physiological relevance of such transientexpression is not clear since there was a clear dissociation from tran-script levels at the 6 h period and protein levels that were sustained forup to 48 h (Singer et al., 1999).

To apply this new knowledge to the in vivo situation, we examinedthe MMP expression patterns in normal and T-treated endometria byimmunohistochemistry. The data showed a high degree of inter-patient variability within a phase of the cycle and within the trimeges-tone treatment groups. Variability of the results between patient sam-ples is a common feature of clinical material (Hampton et al., 1999),when obtained from women who demonstrate different functionalcapacities to activate nuclear receptors, as recently demonstrated for

ERα (Kitawaki et al., 2001) and for PR (Treloar et al., 2005) in theprogression of endometriosis.

Given that no statistically significant difference in the cycle modu-lation of MMP-1 and MMP-3 protein expression was noted, weelected to examine the potential effect of a 10-fold dose differential ofwomen treated with 0.05 mg trimegestone dose to 0.5 mg. No dose–response effect of trimegestone on the staining patterns of MMP-1 orMMP-3 was observed, although the number of MMP-1+ stromal cellswas observed to be significantly higher in the 0.05mg trimegestonegroup when compared to the early secretory group of the normal men-strual cycle. Therefore, there appears to be no correlation between thepattern of bleeding in the trimegestone-treated women and the expres-sion of MMP-1, MMP-3 in their endometrial biopsies. This is in gen-eral agreement with other reports on the effect of differentprogestagens (Vincent et al., 1999; Hickey and Fraser, 2001; Perrot-Applanat, 2004). Hence, the expression of these MMP-1 and MMP-3,per se, does not explain the different bleeding pattern with differentdoses of progestagen (Oliveira-Ribeiro et al., 2004).

In the endometrium of the normal menstrual cycle, MMP-1 andMMP-3 proteins were expressed in all phases and were highest in themenstrual phase, as previously reported (Goldberg et al., 1986; Smith,1996; Hampton et al., 1999; Vincent and Salamonsen, 2000; Vincentet al., 2000); however, they are also at variance to those studies sinceour data did not show statistically significant differences. The reasonfor the discrepancy between the present study and others is not readilyapparent although differences in methodology could account for this.For example, we used different antigen retrieval steps and antibodies,which may have identified more MMP+ cells in relation to those pre-vious studies.

Two studies (Skinner et al., 1999; Vincent and Salamonsen, 2000)reported a differential effect of LNG on the MMP protein expressioncompared to the normal menstrual cycle, while in our study, there wasno difference in the immunoreactivity of MMP-1, MMP-3 and MMP-9between the T-treated endometrium and that of the normal menstrualcycle, suggesting a fundamental difference between T and otherprogestagens, such as LNG. For example, T has minimal androgenreceptor affinity compared to that of LNG and androgen receptorexpression changes markedly just prior to menstruation (Taylor et al.,2005). Alternatively, these progestagens may exhibit differential bind-ing for PR isoforms (Sitruk-Ware, 2004a; Sitruk-Ware, 2004b) induc-ing the differential MMP expression patterns demonstrated in thepresent study.

MMP-9 immunoreactivity, by contrast, was expressed in theendometrial stroma in all phases of the normal menstrual cycle withhighest expression in the late secretory phase and a decrease in themenstrual phase, agreeing with others (Skinner et al., 1999; Cornetet al., 2005). However, this maximal stromal expression of MMP-9was detected only 2 days before the onset of and during menstruation,and is in association with endometrial leukocytes using dual immu-nolocalization (Jeziorska et al., 1996). Similarly, the number ofMMP-3+ stromal cells was much lower in all the phases of the men-strual cycle and in the T-treated endometria compared to that ofMMP-1+ or MMP-9+ stromal cells, agreeing with a previous report(Vincent and Salamonsen, 2000). It is not clear whether the lowercount is due to a limited role of this MMP in the process of endome-trial degradation, or is because MMP-3 acts indirectly through activa-tion of other MMPs such as MMP-7 (Osteen et al., 2005), which inturn activates MMP-9 (Vincent and Salamonsen, 2000). This studysupports a role for a subset of stromal cells producing large quantitiesof MMP-3 from a small number of cells (Figure 8).

Co-ordinated expression of the MMPs, 1, 3 and 9 are thus importantfor endometrial stability at crucial points in the normal menstrualcycle such that a pliable proliferative phase endometrium allows cell

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

M.Wahab et al.

166

migration and expansion due to proliferation, whereas a secretoryphase endometrium remains stable under the influence of P4; hencethe progestagenic inhibition of MMP-1 and MMP-3 expressionobserved in this study. During pregnancy increased serum P4 levelsstabilize the endometrial ECM (Seval et al., 2004) and allows for lim-ited expansion of the decidualized stromal cell. Failure to implant theblastocyst leads to luteolysis of the corpus luteum and progesteronewithdrawal leading to restoration of proliferative phase MMP-1 andMMP-3 levels allowing MMP-9+ leukocytes to infiltrate theendometrium and cause the tissue instability that leads to menstrua-tion. Because of the longer receptor occupancy of T compared to P4(Winneker et al., 2003), MMP-3 levels remain slightly elevated pre-venting the recruitment of MMP-9+ cells to the endometrium. Thiscould explain the reduced and less severe menstrual flows experi-enced by post-menopausal women on the sequential combined tri-megestone HRT regimen (Grubb et al., 2003; Al-Azzawi et al., 2004).

Although there was no dose–response effect of T on the expressionof MMP-1, MMP-3 and MMP-9, the lower number of MMP-9+ cellsin the endometrial stroma of T-treated women suggest that MMP-9 iscritical in the control of menstruation, although it does not explain thedifferent bleeding patterns experienced by women on the differentdoses of T-based HRT (Grubb et al., 2003; Al-Azzawi et al., 2004).

In summary, T behaviour in primary stromal cell culture was simi-lar to P4 in suppressing the expression of MMP-1 and MMP-3 mRNAand protein, which appears to be mediated through binding to PR,since this action was partially mimicked and reversed by the additionof RU486 and was enhanced in the presence of E2, which increasedPR expression. The differential effect of T and P4 on MMP-1 proteinexpression in the presence of RU486 suggests that T and P4 bind to oractivate different PR isoforms or that these two progestagens differen-tially regulate the levels of MMPs through a mechanism involving dif-ferential activation of the AR. The data suggest that T acts in a similarmanner to P4, but causes subtle differences in MMP expression pat-terns that go towards explaining the tighter control that T exerts onmenstruation compared to P4.

AcknowledgementsThe authors wish to thank Angie Gillies and Lindsay Primrose for excellenttechnical assistance. We also thank Hoechst Marion Roussel, Romainville,France, for supplying the trimegestone, RU486 and the trimegestone-treatedendometrial biopsy specimens used in this study.

ReferencesAgoulnik IU, Krause WC, Bingman WE III, Rahman HT, Amrikachi M, Ayala GE

and Weigel NL (2003) Repressors of androgen and progesterone receptoraction. J Biol Chem 278,31136–31148.

Al-Azzawi F, Wabab M, Sami S, Proudler AJ, Thompson J and Stevenson J(2004) Randomized trial of effects of estradiol in combination with eithernorethisterone acetate or trimegestone on lipids and lipoproteins in postmen-opausal women. Climacteric 7,292–300.

Al-Azzawi F, Wahab M, Thompson J, Whitehead M and Thompson W (1999)Acceptability and patterns of uterine bleeding in sequential trimegestone-based hormone replacement therapy: a dose-ranging study. Hum Reprod14,636–641.

Baulieu EE, Ulmann A and Philibert D (1987) Contragestion by antiprogestinRU 486: a review. Arch Gynecol Obstet 241,73–85.

Bicknell GR, Shaw JA, Pringle JH and Furness PN (1996) Amplification ofspecific mRNA from a single human renal glomerulus, with an approach tothe separation of epithelial cell mRNA. J Pathol 180,188–193.

Bouchoux F, Cerede E and Philibert D (1995) Measurement of the relativebinding affinity of RU27987 to recombinant human steroid receptors:progestogen, glucocorticoid, androgen and oestrogen determination ofthe binding parameters of RU27987 and progesterone for the recom-binant human progesterone receptor. 1995; 93/5353/PH data on file atRoussel–Uclaf.

Bradford MM (1976) A rapid and sensitive method for quantitation of micro-gram quantities of protein utilizing the principle of protein-dye binding.Anal Biochem 72,248–254.

Bruner KL, Rodgers WH, Gold LI, Korc M, Hargrove JT, Matrisian LM andOsteen KG (1995) Transforming growth factor beta mediates the progester-one suppression of an epithelial metalloproteinase by adjacent stroma in thehuman endometrium. Proc Natl Acad Sci USA 92,7362–7366.

Casslen B, Urano S and Ny T (1992) Progesterone regulation of plasminogenactivator inhibitor 1 (PAI-1) antigen and mRNA levels in human endome-trial stromal cells. Thromb Res 66,75–87.

Cornet PB, Galant C, Eeckhout Y, Courtoy PJ, Marbaix E and Henriet P(2005) Regulation of matrix metalloproteinase-9/gelatinase B expressionand activation by ovarian steroids and LEFTY-A/endometrial bleeding-associated factor in the human endometrium. J Clin Endocrinol Metab90,1001–1011.

Dai B, Cao Y, Liu W, Li S, Yang Y, Chen D and Duan E (2003) Dual roles ofprogesterone in embryo implantation in mouse. Endocrine 21,123–132.

Eckert RL and Katzenellenbogen BS (1981) Human endometrial cells in primarytissue culture: modulation of the progesterone receptor level by natural andsynthetic estrogens in vitro. J Clin Endocrinol Metab 52,699–708.

Goldberg GI, Wilhelm SM, Kronberger A, Bauer EA, Grant GA and Eisen AZ(1986) Human fibroblast collagenase. Complete primary structure andhomology to an oncogene transformation-induced rat protein. J Biol Chem261,6600–6005.

Gompel A, Malet C, Spritzer P, Lalardrie JP, Kuttenn F and Mauvais-Jarvis P(1986) Progestin effect on cell proliferation and 17 beta-hydroxysteroiddehydrogenase activity in normal human breast cells in culture. J Clin Endo-crinol Metab 63,1174–1180.

Grubb G, Spielmann D, Pickar J and Constantine G (2003) Clinical experiencewith trimegestone as a new progestin in HRT. Steroids 68,921–926.

Hall LL, Bicknell GR, Primrose L, Pringle JH, Shaw JA and Furness PN(1998) Reproducibility in the quantification of mRNA levels by RT-PCR-ELISA and RT competitive-PCR-ELISA. Biotechniques 24,652–658.

Hamilton PW (1995) Designing a morphometric study. In Allen DC (ed.),Quantitative Clinical Pathology. Blackwell Science, Oxford, pp. 311–315.

Hampton AL, Nie G and Salamonsen LA (1999) Progesterone analogues similarlymodulate endometrial matrix metalloproteinase-1 and matrix metalloproteinase-3and their inhibitor in a model for long-term contraceptive effects. Mol HumReprod 5,365–371.

Hatayama H, Kanzaki H, Iwai M, Fujimoto M, Higuchi T, Kojima K,Nakayama H, Mori T and Fujita J (1994) Progesterone enhances macro-phage colony-stimulating factor production in human endometrial cells inculture. Endocrinology 135,1921–1927.

Hickey M and Fraser IS (2001) The causes and management of endometrialbreakthrough bleeding. Reprod Med Rev 9,153–171.

Hirata S, Shoda T, Kato J and Hoshi K (2002) Novel isoforms of the mRNAfor human female sex steroid hormone receptors. J Steroid Biochem MolBiol 83,25–30.

Hulboy DL, Rudolph LA and Matrisian LM (1997) Matrix metalloproteinasesas mediators of reproductive function. Mol Hum Reprod 3,27–45.

Jeziorska M, Nagase H, Salamonsen LA and Woolley DE (1996) Immunolo-calization of the matrix metalloproteinases gelatinase B and stromelysin 1 inhuman endometrium throughout the menstrual cycle. J Reprod Fertil107,43–51.

Kamelle S, Sienko A and Benbrook DM (2002) Retinoids and steroids regulatemenstrual phase histological features in human endometrial organotypic cul-tures. Fertil Steril 78,596–602.

Kitawaki J, Obayashi H, Ishihara H, Koshiba H, Kusuki I, Kado N, Tsukamoto K,Hasegawa G, Nakamura N and Honjo H (2001) Oestrogen receptor-alphagene polymorphism is associated with endometriosis, adenomyosis and leio-myomata. Hum Reprod 16,51–55.

Kokrine L, Marbaix E, Henriet P, Donnez J, Eeckhout Y and Couirtoy PJ(1996) Focal cellular origin and regulation of interstitial collagenase (Matrixmetalloproteinase-1) are related to menstrual breakdown in the humanendometrium. J Cell Sci 109,2151–2160.

Laird SM, Tuckerman EM, Saravelos H and Li TC (1996) The production oftumour necrosis factor alpha (TNF-alpha) by human endometrial cells inculture. Hum Reprod 11,1318–1323.

Marbaix E, Donnez J, Courtoy PJ and Eeckhout Y (1992) Progesterone regu-lates the activity of collagenase and related gelatinases A and B in humanendometrial explants. Proc Natl Acad Sci USA 89,11789–11793.

Marbaix E, Kokorine I, Donnez J, Eeckhout Y and Courtoy PJ (1996) Regula-tion and restricted expression of interstitial collagenase suggest a pivotalrole in the initiation of menstruation. Hum Reprod 11,134–143.

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from

Trimegestone and endometrial MMPs

167

Marbaix E, Kokorine L, Henriet P, Donnez J, Courtoy PJ and Eeckhout Y(1995) The expression of the interstitial collagenase in human endometriumis controlled by progesterone and estradiol and is related to menstruation.Biochem J 305,1027–1030.

Oliveira-Ribeiro M, Petta CA, De Angelo Andrade LAL, Bahamondez L andHidalgo MM (2004) Correlation between endometrial histology, microvas-cular density and calibre, matrix metalloproteinase-3 and bleeding pattern inwomen using a levonorgestrel-releasing intrauterine system. Hum Reprod19,1778–1784.

Osteen KG, Bruner-Tran KL and Eisenberg E (2005) Reduced progesteroneaction during endometrial maturation: a potential risk factor for the develop-ment of endometriosis. Fertil Steril 83,529–537.

Perrot-Applanat M (2004) Immunolocalization of endometrial matrix metal-loproteases and their implication in bleeding. Mt Endocrinol Reprod6,173–177.

Pilka R and Hrachovec P (2003) Metalloproteinases matrix and menstruationreview. Ceska Gynekol 68,36–40.

Rawdanowicz TJ, Hampton AL, Nagase H, Woolley DE and Salamonsen LA(1994) Matrix metalloproteinase production by cultured human endometrialstromal cells: identification of interstitial collagenase, gelatinase-A, gelati-nase-B, and stromelysin-1 and their differential regulation by interleukin-1alpha and tumor necrosis factor-alpha. J Clin Endocrinol Metab 79,530–536.

Robin-Jagerschmidt C, Wurtz J-M, Guillot B, Gofflo D, Benhamou B,Vergezac A, Ossart C, Moras D and Philibert D (2000) Residues in theligand binding domain that confer progestin or glucocorticoid specificityand modulate the receptor transactivation capacity. Mol Endocrinol14,1028–1037.

Robbins A and Spitz IM (1996) Mifepristone: clinical pharmacology. ClinObstet Gynecol 39,436–450.

Salamonsen LA, Butt AR, Hammond FR, Garcia S and Zhang J (1997)Production of endometrial matrix metalloproteinase, but not their tissueinhibitors, is modulated by progesterone withdrawal in an in vitro model formenstruation. J Clin Endocrinol Metab 82,1409–1415.

Salamonsen LA and Woolley DE (1996) Matrix metalloproteinases as media-tors of reproductive function. Hum Reprod 11,124–133.

Schatz F, Papp C, Aigner S, Krikun G, Hausknecht V and Lockwood CJ(1997) Biological mechanisms underlying the clinical effects of RU 486:modulation of cultured endometrial stromal cell stromelysin-1 and prolactinexpression. J Clin Endocrinol Metab 82,188–193.

Seval Y, Akkoyunlu G, Demir R and Asar M (2004) Distribution patterns ofmatrix metalloproteinase (MMP) -2 and – 9 and their inhibitors (TIMP-1and TIMP-2) in the human decidua during early pregnancy. Acta Histochem106,353–362.

Shiokawa S, Yoshimura Y, Nagamatsu S, Sawa H, Hanashi H, Oda T,Katsuma Y, Koyama N and Nakamura Y (1996) Expression of beta 1integrins in human endometrial stromal and decidual cells. J Clin EndocrinolMetabol 81,1533–1540.

Singer CF, Marbaix E, Lemoine P, Courtoy PJ and Eeckhout Y (1999) Localcytokines induce differential expression of matrix metalloproteinases butnot their tissue inhibitors in human endometrial fibroblasts. Eur J Biochem359,40–45.

Sitruk-Ware R (2004a) New progestogens: a review of their effects in peri-menopausal and postmenopausal women. Drug Aging 21,865–883.

Sitruk-Ware R (2004b) Pharmacological profile of progestins. Maturitas47,277–283.

Sivridis E and Giatromanolaki A (2004) New insights into the normal men-strual cycle-regulatory molecules. Histol Histopathol 19,511–516.

Skinner JL, Riley SC, Gebbie AE, Glasier AF and Critchley HO (1999) Regu-lation of matrix metalloproteinase-9 in endometrium during the menstrual

cycle and following administration of intrauterine levonorgestrel. HumReprod 14,793–799.

Smith SK (1996) Vascular endothelial growth factor and the endometrium.Hum Reprod 11,56–61.

Song LN, Coghlan M and Gelmann EP (2004) Antiandrogen effects of mife-pristone on coactivator and corepressor interactions with the androgenreceptor. Mol Endocrinol 18,70–85.

Tabibzadeh S (1996) The signals and molecular pathways involved in humanmenstruation, a unique process of tissue destruction and remodelling. MolHum Reprod 2,77–92.

Taylor AH, Al-Azzawi F, Pringle JH and Bell SC (2002) Inhibition ofendometrial carcinoma cell growth using antisense estrogen receptor oligo-deoxyribonucleotides. Anticancer Res 22,3993–4004.

Taylor AH, Guzail M, Wahab M, Thompson JR and Al-Azzawi F (2005)Quantitative histomorphometric analysis of gonadal steroid receptor distri-bution in the normal human endometrium through the menstrual cycle.Histochem Cell Biol 123,463–474.

Taylor AH, McParland PC, Taylor DJ and Bell SC (2006) The progesteronereceptor in human term amniochorion and placenta is isoform C. Endo-crinology 147,687–693.

Taylor RN, Savouret J-F, Vaisse C, Vigne J-L, Ryan I, Hornung D, Seppala Mand Migrom E (1998) Promegestone (R5020) and mifepristone (RU486)both function as progestastional agonists of human glycodelin gene expres-sion in isolated epithelial cells. J Clin Endocrinol Metab 83,4006–4012.

Treloar SA, Zhao ZZ, Armitage T, Duffy DL, Wicks J, O’Connor DTO,Martin NG and Montgomery GW (2005) Association between polymor-phisms in the progesterone receptor gene and endometriosis. Mol HumReprod 11,641–647.

Vigano P, Di Blasio A, Dell Antonio G and Vignali M (1993) Culture ofhuman endometrial cells: a new simple technique to completely separateepithelial cells. Acta Obstet Gynecol Scand 72,87–92.

Vincent AJ, Malakooti N, Zhang J, Rogers PA, Affandi B and Salamonsen LA(1999) Endometrial breakdown in women using Norplant is associated withmigratory cells expressing matrix metalloproteinase-9 (gelatinase B). HumReprod 14,807–815.

Vincent AJ and Salamonsen LA (2000) The role of matrix metalloproteinasesand leukocytes in abnormal uterine bleeding associated with progestin-onlycontraceptives. Hum Reprod 15,135–143.

Vincent AJ, Zhang J, Ostor A, Rogers PA, Affandi B, Kovacs G and SalamonsenLA (2000) Matrix metalloproteinase-1 and – 3 and mast cells are present inthe endometrium of women using progestin-only contraceptives. HumReprod 15,123–130.

Wahab M, Thompson J and Al-Azzawi F (1999a) The distribution of endome-trial leukocytes and their proliferation markers in trimegestone-treated post-menopausal women compared to the endometrium of the natural cycle: adose-ranging study. Hum Reprod 14,1201–1206.

Wahab M, Thompson J, Hamid B, Deen S and Al-Azzawi F (1999b) Endome-trial histomorphometry of trimegestone-based sequential hormone replace-ment therapy: a weighted comparison with the endometrium of the naturalcycle. Hum Reprod 14,2609–2618.

Wang W and Passaniti A (1999) Extracellular matrix inhibits apoptosis andenhances endothelial cell differentiation by a NFkappaB-dependent mecha-nism. J Cell Biochem 73,321–331.

Wiegratz I and Kuhl H (2004) Progestogen therapies: Differences in clinicaleffects? Trends Endocrinol Metab 15,277–285.

Winneker RC, Bitran D and Zhang Z (2003) The preclinical biology of a newpotent and selective progestin: Trimegestone. Steroids 68,915–920.

Submitted on December 1, 2005; accepted on January 9, 2006

by guest on June 3, 2013http://m

olehr.oxfordjournals.org/D

ownloaded from