Regulation and Function of LEFTY-A/EBAF in the Human Endometrium

The human endometrium is a unique tissue that is periodically shed during menstruation. Although overall triggered by ovarian steroids withdrawal, menstrual induction of matrix metalloproteinases (MMPs) and resulting tissue breakdown are focal responses, pointing to additional local modulators. LEFTY-A, a novel member of the transforming growth factor-β family identified originally as an endometrialbleeding-associated factor (EBAF), is a candidate for this local control. We measuredLEFTY-A and β-ACTIN mRNA concentration during the menstrual cycle in vivo and found that their ratio was dramatically (∼100-fold) increased at the perimenstrual phase. A similar increase was seen when proliferative explants were cultured for 24 h in the absence of ovarian steroids; this was followed by spontaneous production of proMMP-1, -3, and -9. Both responses were inhibited by progesterone. Moreover, addition of recombinant LEFTY-A to proliferative explants was sufficient to stimulate the expression of proMMP-3 and -7; this response was also blocked by ovarian steroids. Collectively, these data indicate that LEFTY-A may provide a crucial signal for endometrial breakdown and bleeding by triggering expression of several MMPs. Progesterone appears to exert a dual block, upstream by inhibiting LEFTY-A expression and downstream by suppressing its stimulatory effect on MMPs.

Menstrual breakdown of the human endometrium offers a spectacular example of extracellular matrix (ECM) 1 remodeling that can be mimicked in cultured explants (1). Extensive lysis and renewal of this tissue occur every month under the control of ovarian steroids. Estrogens induce tissue proliferation during the proliferative phase; following ovulation, progesterone induces glandular secretion and stromal decidualization. In the absence of pregnancy, the fall of both ovarian steroids in the late secretory phase triggers a cascade of events leading to proteolytic breakdown, shedding of the functional layer, and menstrual bleeding. Irregular and unpredictable ECM breakdown and bleeding between menstruations are common disorders that occur in diverse endometrial pathologies or upon hormonal treatment (2,3).
In endometrium undergoing normal or abnormal bleeding, several matrix metalloproteinases (MMPs), including collagenase-1 (MMP-1) and stromelysin-1 (MMP-3), are produced selectively by foci of stromal cells in areas of ECM disruption (see Refs. 4 and 5, and for a review see Ref. 6). Expression of these MMPs and of gelatinase-A (MMP-2) progresses along with tissue breakdown, suggesting strongly that these proteinases play a crucial role in endometrial ECM degradation. Although the profile of gelatinase-B (MMP-9) expression along the menstrual cycle is still controversial, the presence of its active form seems restricted to the perimenstrual phase (7)(8)(9)(10). Expression of matrilysin-1 (MMP-7) by epithelial cells is also induced around menstruation (10 -12). Using an ex vivo explant system, MMPs were found to be responsible for the menstrual-like ECM degradation and to be repressed by physiological concentrations of progesterone (1,5,13). Thus, active MMPs are thought to degrade the endometrial ECM, including the basement membrane of the blood vessels, a prerequisite for the occurrence of menstrual and abnormal bleeding (5).
The focal nature of MMP production and initial ECM breakdown suggests strongly that a local response modulates the overall effect of ovarian steroid withdrawal (5,10). The following two factors likely contribute to the spatial heterogeneity of hormonal effects: (i) the unequal distribution of hormone receptors within the tissue; and (ii) the involvement of other molecules such as cytokines acting as local relays (14).
LEFTY-A, also called EBAF (endometrial bleeding-associated factor), is a recently discovered member of the transforming growth factor (TGF)-␤ family (15)(16)(17). Human LEFTY-A is secreted as a 42-kDa precursor susceptible to proteolytic cleavage, and LEFTY-A active forms are able to induce mitogenactivated protein kinase activity and to inhibit TGF-␤ signaling (18,19).
Tumors derived from transduced fibroblastic cells overexpressing LEFTY-A secrete abundant collagenolytic activity that was not detected in tumors derived from mock-transduced cells (20), suggesting that LEFTY-A may act as a physiological modulator of MMP expression. In human, LEFTY-A mRNA has been observed in several (but not all) tissues, including the endometrium (15,21). Using Northern and Western blotting, endometrial expression was always abundant in the perimenstrual phase (i.e. late secretory and menstrual endometrium), was detected at lower levels in some patients around ovulation, and was not detected during the other phases of the menstrual cycle. By in situ hybridization and immunohistochemistry, both LEFTY-A mRNA and protein were essentially found in stromal cells of the functional layer of the perimenstrual endometrium (15,17). Abnormal expression of LEFTY-A mRNA has also been observed in the non-perimenstrual endometrium of patients with irregular bleeding (15) and infertile women (22). Altogether, these observations suggest that LEFTY-A expression may be controlled by ovarian steroids and may correlate in time and space with bleeding-associated MMPs. LEFTY-A is therefore a good candidate to locally modulate progesterone action on these proteinases within the endometrium.
In the present study, we used competitive RT-PCR (RT-cPCR) (23) as a more quantitative and sensitive methodology than Northern blotting to measure endometrial LEFTY-A mRNA concentration, both along the normal menstrual cycle in vivo and in response to ovarian steroids in cultured explants ex vivo. We next investigated the temporal correlation between LEFTY-A mRNA expression and the production of several endometrial MMPs. Finally, we tested directly the ability of recombinant LEFTY-A to modulate MMP expression in explant culture.

EXPERIMENTAL PROCEDURES
Tissue Collection and Explant Culture-The study was approved by the Ethical Committee of the Université Catholique de Louvain, in accordance with the Declaration of Helsinki of the World Medical Association. Normal endometrial tissue was obtained from hysterectomy specimens (n ϭ 38) or biopsies (n ϭ 8) sampled at various phases of the menstrual cycle. An ideal menstrual cycle of 28 days was divided into the following phases: early proliferative (days 6 -8, n ϭ 11), midproliferative (days 9 -11, n ϭ 1), late proliferative (days 12-14, n ϭ 9), early secretory (days 15-18, n ϭ 6), mid-secretory (days 19 -22, n ϭ 6), late secretory (days 23-27, n ϭ 5), and perimenstrual phases (days 28 -5, n ϭ 8). All patients were premenopausal (range: 25 to 55 years old) and were not involved in hormonal treatment. For the hysterectomy specimens, endometrium that appeared normal to the naked eye was scraped gently from the surface of the uterine cavity with a sterile surgical blade by an experienced pathologist (E. M.) and put in ice-cold phosphate-buffered saline for further processing; biopsies were collected immediately in ice-cold phosphate-buffered saline. Part of the endometrial tissue used for the study was fixed in 4% formaldehyde for histological examination; dating was first estimated following established microscopic criteria (24) and was adjusted finely according to clinical information of the last menstrual period whenever available. Pathological examination of the operative specimens and biopsies showed no organic lesion (n ϭ 9), uterine leiomyoma(s) (n ϭ 31) or leiomyosarcoma (n ϭ 1), adenomyosis (n ϭ 10), endometrial polyp (n ϭ 3), cervical intraepithelial neoplasia (n ϭ 2), hydrosalpinx (n ϭ 2), ovarian dermoid cyst (n ϭ 1), or ovarian and vaginal endometriosis (n ϭ 1).
Non-cultured samples were frozen quickly at Ϫ80°C in lysis buffer (SV Total RNA Isolation System; Promega, Madison, WI). Alternatively, tissue explants were cultured as described (1). Briefly, tissue samples were cut in pieces of about 1-mm with a sterile surgical blade and placed in tissue culture inserts (Millipore, Bedford, MA; 24 explants/30-mm insert for RNA extraction; 6 explants/12-mm insert in triplicates for experiments addressing the effect of recombinant LEFTY-A). Dulbecco's modified Eagle's medium (Invitrogen), devoid of serum and phenol red, was placed in the lower chamber and renewed daily (300 l or 1.2 ml in 12-and 30-mm inserts, respectively). Medium was without hormonal addition (ϪH), or supplemented with 1 nM water-soluble 17␤-estradiol (ϩE), 100 nM progesterone (ϩP), or a combination of both (ϩEϩP; Sigma-Aldrich) and, when applicable, in the presence of the indicated concentration of recombinant LEFTY-A (R & D Systems). It was verified that the presence of 0.5 g/ml 2-hydroxypropyl-␤-cyclodextrin (Sigma-Aldrich) used as steroid vehicle did not affect the expression of LEFTY-A mRNA or ␤-ACTIN mRNA (data not shown). In addition, we have demonstrated previously that the inhibitory effect of progesterone on the expression of several MMPs was independent of this vehicle (1). After collection, all media were supplemented with 0.05 vol of 1 M Tris-HCl buffer, pH 7.5, 1% (v/v) Triton X-100, 0.1 M CaCl 2 , 60 mM NaN 3 and kept frozen at Ϫ20°C until biochemical analysis. At the end of the culture, explants were frozen at Ϫ80°C in lysis buffer until RNA extraction.
RNA Extraction-Total RNA was extracted using the SV total RNA isolation system (Promega) according to the manufacturer's recommendations. The RNA concentration was quantified by spectrophotometry at 260 nm. RNA integrity was controlled by electrophoresis in 1% agarose gels containing ethidium bromide.
Reverse Transcription-Aliquots (100 or 200 ng) of total RNAs extracted from endometrial tissue were reverse-transcribed in a total volume of 20 l by using the Thermoscript™ RT-PCR system (Invitrogen). Oligo(dT) primers were compared with hexamers degenerated randomly, and the former ones were used in subsequent experiments, because stronger signals were obtained after PCR amplification (data not shown). The cDNA products were stored at 4°C until subsequent PCR analyses.
Oligonucleotide Primers Used for PCR Amplification-Specific oligonucleotide primers for human LEFTY-A and ␤-ACTIN (for standardization) were derived from their cDNA sequences (GenBank TM accession numbers AF081513 and X00351, respectively). Primers were designed to create amplicons crossing over exonic boundaries (Fig. 1), to ensure that the detected products resulted from specific amplification of the target sequence and not from contamination by genomic DNA. In addition, the LEFTY-A downstream primer was chosen in the 3Ј-untranslated region to ensure distinction from the related LEFTY-B sequence (GenBank TM accession number AF081512). Customized primers were FIG. 1. cDNA constructions. Competitors for LEFTY-A and ␤-AC-TIN cDNAs were constructed by deleting an 86-and a 101-bp fragment from the specific target sequence, respectively. The localization of the primer pairs used to perform the competitive PCR is also indicated. Position on cDNAs refers to GenBank TM sequences annotation (for accession numbers see text). Exon (Ex) boundaries can be found in GenBank TM files AF081508 -AF081511 (LEFTY-A gene) and M10277 (␤-ACTIN gene). UTR, untranslated region. obtained from Invitrogen, and their optimal hybridization temperatures (Table I) were determined by comparing different temperatures in parallel PCR reactions using a Biometra T-gradient thermocycler (Westburg, Leusden, The Netherlands).
Preparation of Competitors for LEFTY-A and ␤-ACTIN-Competitors for LEFTY-A and ␤-ACTIN cDNA PCR amplification were constructed by deleting, respectively, an 86-and a 101-bp fragment from the corresponding target cDNA. Truncated fragments were synthesized by PCR using a modified primer and the appropriate regular opposite primer (Table I). Mutated amplicons were purified from agarose gels with the QIAquick gel extraction kit (Qiagen) and inserted into pCR ® II-TOPO ® plasmids (Invitrogen) to be amplified in One Shot ® TOP10 chemically competent Escherichia coli (TOPO TA Cloning ® kit; Invitrogen). Purified plasmids were obtained using the QIAfilter Maxi Protocol (Qiagen). Competitor sequences were extracted from the vectors by EcoRI enzymatic digestion (Roche Molecular Biochemicals), purified from agarose gels, and quantified against a 100-bp DNA ladder (DNA Molecular Weight Marker XIV; Roche Molecular Biochemicals). Serial dilutions of competitors were prepared and stored at Ϫ20°C as aliquots (10 l) that were thawed only once and used immediately.
PCR-Aliquots (1 l) of the RT products were subjected to PCR in the PCR Supermix (Invitrogen) containing 22 mM Tris-HCl, pH 8.4, 55 mM KCl, 1.65 mM MgCl 2 , 220 M dNTPs, 22 units/ml Taq DNA polymerase with 0.5 M of adequate paired primers in a total volume of 20 l. A control PCR without template DNA was performed in each experiment. The PCR reactions were realized using Biometra thermocyclers as follows. After an initial denaturation at 95°C for 5 min, the DNA was amplified through 30 to 40 cycles of 30 s at 95°C, 30 s at the optimal annealing temperature of the primer pair (Table I), and 30 s at 72°C. The reaction was terminated at 72°C for 10 min. PCR products were stored at 4°C until use.
In competitive PCRs, RT products were coamplified with a defined amount of the corresponding competitor in a reaction mixture supplemented with 1 Ci [␣ 32 P]dCTP (Amersham Biosciences). For each sample, three or four different concentrations of the competitor were used for parallel PCR reactions. The concentrations of target cDNA were first estimated roughly and then measured precisely using closely adjusted concentrations of the diluted competitor. The amplicons (target and competitor) were resolved in a 5% polyacrylamide gel, and cpm were quantified by direct counting on defined areas of the dried gels using an InstantImager detector (Packard, Meriden, CT).
Data were plotted on a logarithmic scale as (target cpm/competitor cpm) ratios for increasing concentrations of competitor ( Fig. 2A) (23). The concentration of the target sequence in the sample was defined as the intersection of the linear regression with the x axis (when ratio ϭ 1, log ϭ 0). In addition, for each experiment, the slope of the line was measured to control that the target and the competitor sequences were amplified with the same efficiencies. Data generating slopes deviating from the Ϫ0.6 to Ϫ1.2 range were discarded, and PCRs were repeated with redefined competitor concentrations. The sensitivity, linearity, and dynamic range of the method were evaluated by quantifying serial 10-fold dilutions (over 3 logs) of one sample (Fig. 2B). The lowest detectable amount measured in the present study was 22 LEFTY-A mRNA molecules in a reaction tube. To evaluate the reproducibility of the assay, LEFTY-A or ␤-ACTIN mRNA was measured twice by independent RT-cPCR in 26 samples of total RNA (Fig. 2C). In a comparative experiment using the same primers, similar results were obtained by competitive PCR assay and by real-time PCR (data not shown). ␤-ACTIN mRNA was used as a housekeeping gene to standardize the levels of LEFTY-A mRNA (25-27). The standardized ratio (LEFTY-A mRNA/␤-ACTIN mRNA) will hereafter be referred to as the relative amount of LEFTY-A mRNA.
Enzyme Assays-Conditioned media were assayed in solution for the release of spontaneous and total (i.e. after zymogen activation by aminophenylmercuric acetate, APMA) collagenase activities and by gelatin, casein, or reverse gelatin zymography for the identification of MMPs and their tissue inhibitors (TIMPs) (1).
Statistical Analysis-Statistical significance was tested using the Wilcoxon matched pairs test or the Wilcoxon two-sample test, as appropriate. Differences were interpreted as significant for p Ͻ 0.05.

Quantification of LEFTY-A mRNA Relative Amounts in the Cycling Human Endometrium-Previous studies based on
Northern blotting and in situ hybridization have shown that LEFTY-A mRNA levels vary in the cycling human endometrium, being detected only during perimenstrual phases and in some periovulatory tissues (15,21). To quantify the expression of LEFTY-A mRNA throughout the menstrual cycle, we analyzed by RT-cPCR 45 endometria sampled at different phases of the normal cycle. These values were expressed as relative amounts, by reference to ␤-ACTIN mRNA measured in parallel.
The relative amount of LEFTY-A mRNA varied considerably between patients, even for a given phase of the cycle, as shown in Fig. 3. We verified in two hysterectomy specimens that the variation was not due to heterogeneity of expression within the organ. Indeed, similar values were observed at three different sites of endometrium sampling (fundus, corpus, isthmus). Despite these individual differences, there was a significant increase of relative LEFTY-A mRNA amount during the perimenstrual phase (herein defined as the period from day 28 to day 5 of the next cycle; values Ն 2.0 ϫ 10 Ϫ4 in all patients; n ϭ 8), as compared with all other phases together (values Ͻ 2.0 ϫ 10 Ϫ4 in 28 of 37 non-perimenstrual endometria; p Ͻ 0.001). The median in the perimenstrual group (3.7 ϫ 10 Ϫ3 ) was 100-fold higher than in the non-perimenstrual group (3.5 ϫ 10 Ϫ5 ). Interestingly, two out the nine non-perimenstrual samples with higher LEFTY-A mRNA relative amounts showed superficial foci of menstrual-like tissue breakdown (two proliferative endometria represented by filled symbols in Fig. 3). When analyzing together all endometria with tissue breakdown regardless of the phase (n ϭ 10), the relative LEFTY-A mRNA amount was also significantly higher than in tissue without breakdown (p Ͻ 0.002); values in 28 of 35 endometria without signs of breakdown (Fig. 3, open symbols) were below the lowest level (2.0 ϫ 10 Ϫ4 ) found in endometria with tissue breakdown. The median in the group with tissue breakdown (3.0 ϫ 10 Ϫ3 ) was also 100-fold higher than in the group with intact tissue (2.8 ϫ 10 Ϫ5 ). Several, but not all, mid-and late secretory endometria also showed an increased expression of LEFTY-A mRNA, but none exhibited stromal breakdown. The relative levels of LEFTY-A mRNA in endometria sampled at the mid-secretory, late secretory, and perimenstrual phases of the cycle were significantly higher than in the endometria sampled during the other phases of the cycle (p Ͻ 0.005).
The ␤-ACTIN housekeeping gene was expressed at similar steady-state levels throughout all phases of the menstrual cycle and during explant culture under various hormonal conditions, except at the mid-secretory phase during which a 3-to 6-fold decrease (p Ͻ 0.005) was observed compared with the other phases of the cycle (Table II). However, this relatively modest fall of ␤-ACTIN mRNA at the mid-secretory phase did not  influence the overall pattern of variation of LEFTY-A mRNA throughout the cycle.

LEFTY-A mRNA Concentration Increases in Explant Culture of Proliferative Endometria and Is Controlled by Combined
Ovarian Steroids-Because the menstrual phase is preceded by a fall in the concentrations of ovarian steroids, we next evaluated how LEFTY-A expression responds to culture in the absence of these hormones, by comparing its mRNA in noncultured endometria and in explants after 24 h of culture. This comparison (Fig. 4A, left panel) shows clearly that the relative amount of LEFTY-A mRNA increased after 1 day of culture in all proliferative samples tested (n ϭ 8; p Ͻ 0.005), reaching values similar to those observed in perimenstrual cycling endometrium and clustered within a narrow range (1.1 ϫ 10 Ϫ3 to 5.1 ϫ 10 Ϫ3 ). Values were up to 270 times higher than before culture, with a 70-fold increase when comparing the medians.
In contrast, the relative amounts of LEFTY-A mRNA molecules were significantly lower in explants from all proliferative endometria when cultured in the presence of 1 nM estradiol combined with 100 nM progesterone than in the absence of hormones (Fig. 4A, right panel, n ϭ 8; p Ͻ 0.005); values (range: 2.8 ϫ 10 Ϫ5 to 1.4 ϫ 10 Ϫ3 ) were up to 70 times lower in the presence of the ovarian steroids. There was no significant difference between non-cultured tissues and corresponding explants cultured with both ovarian steroids.

Progesterone Is Sufficient to Inhibit LEFTY-A mRNA Increase in Proliferative Endometrial
Explants-Because the combination of estradiol and progesterone inhibited the increase of LEFTY-A mRNA concentration in proliferative explants after 24 h of culture, these steroids were investigated separately. The relative amount of LEFTY-A mRNA molecules was measured in four proliferative endometria before culture and after 24 h of culture either without added hormones or in the presence of estradiol alone, progesterone alone, or both (Fig. 4B). These cultures demonstrated that progesterone alone was sufficient to account for the effects of the combined ovarian steroids, whereas estradiol alone was ineffective.

LEFTY-A mRNA Concentration Does Not Show Consistent
Changes during Culture of Secretory Endometria-In striking contrast to proliferative endometria (Fig. 4A, left panel), explants from secretory endometria did not show an increase of their relative amount of LEFTY-A mRNA to a clustered range of values when cultured in the absence of ovarian steroids (Fig.  4C, left panel). Although a moderate increase was observed in several secretory endometria, the trend was not significant, even when the culture was extended for 48 h to rule out a possible effect of residual endogenous hormones (not shown). In comparison to explants cultured without steroids, combined estradiol and progesterone decreased the relative amount of LEFTY-A mRNA in explants from all but one secretory endo- FIG. 4. Effects of estradiol and progesterone on LEFTY-A mRNA concentration in cultured explants from proliferative and secretory endometria. Explants from eight proliferative (A and B) and eight secretory (C) endometria were cultured for 24 h in presence or absence of 1 nM estradiol and/or 100 nM progesterone. The relative amount of LEFTY-A mRNA was measured in the explants and presented as in Fig. 3. Relative amounts of LEFTY-A mRNA are compared in explants cultured without steroids (ϪH) and in the corresponding noncultured endometria (NC; left panels in A and C), as well as between explants cultured without steroids and with combined estradiol and progesterone (ϩEϩP; right panels in A and C). In B, the effect of estradiol (ϩE) and progesterone (ϩP) alone are compared with the other culture conditions in four of the eight proliferative endometria presented in A. Numbers refer to patient identification in Fig. 3. NS, not significant (p Ͼ 0.05). metria cultured for 24 h (Fig. 4C, right panel). However, this trend was again not significant, even when the culture was extended for 48 h (not shown).

Correlation between LEFTY-A mRNA Concentration and Release of MMPs and TIMPs by Cultured Explants-
The dramatic increase in LEFTY-A mRNA expression during the perimenstrual phase suggests that LEFTY-A could either be induced by endometrial breakdown or, conversely, act as a local regulator triggering this process by causing the induction and/or activation of MMPs. To investigate a possible relationship between LEFTY-A mRNA expression and MMP activity, we first analyzed by enzymatic assay and zymography the release profile of interstitial collagenase (MMP-1), gelatinase-A (MMP-2) and -B (MMP-9), stromelysin-1 (MMP-3), as well as of their inhibitors, TIMP-1 and TIMP-2, in conditioned media obtained from seven proliferative and eight secretory endometria cultured up to 48 h.
Representative examples of the hormonal control of MMP production and activation are shown at Fig. 5. During the first day of culture, no spontaneous collagenase activity (Fig. 5A, ϪAPMA) or active forms of MMP-1, -3 (Fig. 5C), and -9 (Fig.  5D) were detected in the conditioned media, irrespectively of the presence of ovarian steroids, except for the presence of active MMP-3 and -9 in one proliferative sample (not shown). This is consistent with the fact that appearance of menstrual-like breakdown upon culture of explant from non-menstrual specimens without these hormones always requires more than 24 h (1). In addition, although active MMP-2 was found readily in media conditioned by secretory endometria (Fig. 5D, right  panel), it was barely detectable in those of proliferative endometria. Thus, no MMP activity or ECM degradation could be detected in cultured proliferative explants after 1 day of culture in the absence of hormones, despite the fact that LEFTY-A mRNA was increased (Fig. 4A). Because endometrial ECM breakdown does not precede, it cannot trigger LEFTY-A mRNA expression.
In contrast, during the second day of culture in the absence of hormones, the medium conditioned by proliferative explants showed high total (ϩAPMA) and spontaneous (ϪAPMA) collagenase activities, enhanced release and activation of MMP-1, -2, -3, and -9, and decreased TIMP-1 release (Fig. 5, left panel). The high levels of LEFTY-A mRNA found in these explants after 1 day of culture therefore suggest that LEFTY-A could have induced the MMP/TIMP response during the second day of culture. Indeed, explants cultured with estradiol and progesterone contained lower amounts of LEFTY-A mRNA after 1 day and did not show such dramatic changes in the media conditioned during the second day.
However, no consistent relationship between LEFTY-A mRNA and MMPs or TIMPs was found in the secretory endo- metria. The profile of MMP or TIMP release and pro-MMP activation and its variation in response to estradiol and progesterone were similar to those of proliferative endometria, despite the absence of a consistent trend in changes of LEFTY-A mRNA levels, as exemplified by the interesting exception that is illustrated (compare patient 11 in Fig. 4C with Fig. 5, right panel).

Recombinant LEFTY-A Enhances proMMP-3 and -7 Expression in Cultured Proliferative
Explants-To test directly for an induction of MMP expression by LEFTY-A in proliferative endometria, recombinant LEFTY-A was added to cultured explants (Fig. 6). In the absence of added ovarian steroids, the release of both proMMP-3 and -7 (two MMPs that are induced selectively in the perimenstrual phase) was enhanced significantly by LEFTY-A addition after 1 and 2 days of culture and declined during the third day (not shown). The dose-response curve was biphasic, with a progressive induction peaking at 5 ng/ml (p Ͻ 0.05) and a decline back to the level of untreated samples at 125 ng/ml. Such a bell-shaped-like profile has been described for other agents, including TGF-␤1 effect on MMP-2 and -9 expression and activation (29). Moreover, a combination of estradiol and progesterone inhibited not only the spontaneous release of these proenzymes but also its stimulation by LEFTY-A (Fig. 6C). These data demonstrate that LEFTY-A can enhance the expression of proMMPs in the human endometrium and that stimulation of LEFTY-A expression and its effects can both be overcome by progesterone combined with estradiol. DISCUSSION This manuscript demonstrates that LEFTY-A mRNA is expressed in the human endometrium in vivo throughout the menstrual cycle, with a dramatic increase (about 100-fold) during the perimenstrual phase. A similar increase in LEFTY-A mRNA occurs in explants from proliferative endometria cultured in the absence of ovarian steroids. Using this ex vivo culture system, we also provide the first direct evidence that recombinant LEFTY-A increases the expression of proMMP-3 and -7, which normally appear during menstruation, and that progesterone exerts a dual block on LEFTY-A action, by preventing the increase of LEFTY-A mRNA concentration and by inhibiting LEFTY-A-induced MMP expression.
RT-cPCR largely outpassed Northern blotting and in situ hybridization to quantitate LEFTY-A mRNA variation in the human endometrium (15,17). Because of its sensitivity and linearity over a wide range of target concentrations (5 logs in our experiments), RT-cPCR permits detection of LEFTY-A mRNA in samples containing but a few copies per assay and requires much smaller amounts of material; 200 ng of total RNAs was sufficient to perform RT and 10 subsequent cPCR measurements in duplicate, whereas 20 g of total RNAs was required to detect LEFTY-A mRNA by Northern blotting (15). 2 In one comparative experiment, whereas a relative LEFTY-A mRNA amount of 4.0 ϫ 10 Ϫ3 , measured by RT-cPCR from 10 ng of total RNAs, was detectable by Northern blotting using 20 g of total RNAs, the latter method failed to detect a relative amount of 3.2 ϫ 10 Ϫ5 in the same conditions. Moreover, the use of homologous competitors (i.e. derived from the target sequence) allowed validation of the measurements in each experiment by controlling that both DNA species, target and competitor, displayed similar amplification rates.
This RT-cPCR analysis not only shows that LEFTY-A mRNA remains expressed during the entire cycle but also highlights the striking increase (2 logs between the medians, 4 logs between measurable extremes) in all samples collected during the perimenstrual phase compared with the rest of the cycle, with LEFTY-A mRNA levels reaching up to 4% of the very abundant ␤-ACTIN mRNA. The considerable intragroup variation (over 3 logs in non-perimenstrual samples, over 2 logs in perimenstrual samples) underscores the advantage of the explant system to study the regulation of LEFTY-A expression in different conditions on tissue derived from the same patient, so as to avoid interpatient variation.
Several other cytokines present in the endometrium show cyclical expression. Among them, interleukins (IL)-1␤, -6, and -8, leukemia inhibitory factor, TGF-␤1, insulin-like growth factor-II, and endothelin also display enhanced mRNA concentration at the end of the cycle (30 -32). However, by comparing with reported PCR-based measurements of the variations of leukemia inhibitory factor (33) IL-1␤ and IL-6 (34) mRNA concentrations, expression of LEFTY-A mRNA shows the largest range of variation between cycle phases. ECM breakdown is a major event leading to menstruation and is reproduced upon culture of non-menstrual endometrial explants in the absence of ovarian steroids (1). Because of reciprocal signaling between cells and their matrix, ECM remodeling can influence deeply cell behavior by altering gene expression (35,36). Comparison between the profile of MMP production and the variations in LEFTY-A mRNA in endometrial explants shows that the detection of active MMP-1, -3, and -9 and of tissue breakdown does not precede, but actually follows, the up-regulation of LEFTY-A mRNA in cultured proliferative samples. Therefore, increased LEFTY-A mRNA expression is not triggered by matrix degradation but could induce it. Indeed, addition of recombinant LEFTY-A to cultured explants is sufficient to stimulate the production of proMMP-3 and -7, compatible with a role of LEFTY-A in the induction of matrix degradation in vivo. This is consistent with the abundant collagenolytic activity observed in tumors derived from cells overexpressing this cytokine (20). The hypothesis that up-regulation of LEFTY-A mRNA expression starts in the midand late secretory endometrium and precedes the occurrence of stromal breakdown is supported by the increased LEFTY-A mRNA amounts in several mid-and late secretory endometria that do not show any sign of stromal breakdown. Indeed, previous in situ hybridization studies revealed a strong expression of LEFTY-A mRNA in the stroma of late secretory endometria that did not show any tissue breakdown (15,17,21). Moreover, it is not disproved by the persistence of high LEFTY-A mRNA levels in perimenstrual endometrium, because matrix breakdown starts in foci and spreads progressively throughout the surrounding tissue. LEFTY-A mRNA could be expressed in the preserved areas even though it disappeared in the broken down foci.
The current hypothesis of endometrial remodeling is that expression of menstruation-associated MMPs is overall turned on by the fall of ovarian steroids and finely tuned by a local network of cytokines. We demonstrated previously in primary co-cultures of endometrial cells that epithelium-derived IL-1␣ was a key inducer of MMP-1 production by stromal cells and that progesterone acted at two levels to silence MMP-1 expression, upstream of the cytokine by inhibiting IL-1␣ release and downstream of the cytokine by suppressing IL-1␣-stimulated MMP-1 expression (14). It is also known that MMP-7 expression is induced in the endometrial epithelium around menstruation (10), and it was suggested that progesterone blocks MMP-7 expression in the secretory endometrium through the paracrine action of a member of the TGF-␤ family released by the stromal cells (37,38). It can thus be proposed that LEFTY-A triggers menstrual ECM degradation in vivo by directly up-regulating the expression of MMPs and/or indirectly by inhibiting their TGF-␤-mediated block, because LEFTY-A is able to inhibit the signaling cascade downstream of TGF-␤ receptors (19).
Several observations suggest further that LEFTY-A mRNA expression is controlled not only by inhibitory signals including progesterone but also by stimulatory signals that remain to be identified and that LEFTY-A is not sufficient but participates in a complex network of local factors that regulate focal matrix degradation in the perimenstrual endometrium. First, LEFTY-A mRNA was low in non-cultured proliferative endometria, suggesting that the absence of progesterone is not sufficient to increase LEFTY-A mRNA expression. Second, the increase of LEFTY-A mRNA in explants from the same proliferative endometria cultured without progesterone was observed irrespectively of the presence of estradiol, arguing strongly against inhibition by this steroid in non-cultured endometria. Maintenance of low LEFTY-A mRNA concentration in non-cultured proliferative endometria could therefore result from a lack of LEFTY-A inducer and/or from the presence of (an) inhibitory agent(s) other than ovarian steroids. A putative retinoic acid response element is present in the promoter region of the human LEFTY-A gene (16), and the murine orthologous sequence is functional in vitro (39). However, because stromal cells derived from human endometrium synthesize retinoic acid exclusively if collected during the secretory phase (40), an increase of retinoic acid level is unlikely to explain the observed increase of LEFTY-A mRNA in proliferative endometria in culture without addition of ovarian steroids. Third, the unequal sensitivity of secretory endometria to ovarian steroid withdrawal, as well as the increased expression of LEFTY-A mRNA in mid-secretory endometria sampled at the time when progesterone concentrations are at a maximal level, suggest that additional modulators of LEFTY-A mRNA expression are expressed in the secretory endometrium in vivo. From a functional perspective, these mechanisms would protect the endometrium from fluctuations in progesterone levels, especially around the implantation window. The sequestration of such modulators and their slow release from the extracellular matrix could account for the "progesterone memory" of secretory explants cultured without ovarian steroids. Alternatively, LEFTY-A modulators could be expressed by immune cells (such as large granular lymphocytes) that are sensitive to progesterone (41) and preferentially colonize the late secretory endometrium, although with a likely interpatient variation. Further-more, it is possible that secretory endometrium does not respond to steroid hormone withdrawal, and perhaps to LEFTY-A addition, because the level of LEFTY-A or of an inducer of LEFTY-A is already sufficient for a maximal response. For instance, a maximal level of LEFTY-A mRNA expression could already be present in endometrium 14 before culture, preventing any further increase upon culture without steroids (see Fig. 4C). The unequal global response to steroid withdrawal may also reflect (i) focal responses limited to areas where the background for progesterone inhibition of LEFTY-A expression would fall below a threshold level or (ii) heterogeneity between stromal and epithelial cells in the expression of progesterone receptors or in changes between the A and B receptor isoforms, because LEFTY-A mRNA was shown to be expressed by both cell types (17). Fourth, there was no consistent correlation between LEFTY-A mRNA levels and the presence of ovarian steroids, ECM breakdown, or MMP expression in secretory endometria in vivo or in culture. In particular, the secretory endometrium represented in Fig. 5 (patient 11) showed the expected up-regulation of the production and/or activation of MMPs and the decreased production of TIMP-1, despite decreased expression of LEFTY-A mRNA during the first day of culture without hormones. Although this endometrium was the only one showing an inversed response of LEFTY-A mRNA expression to the ovarian steroids or their withdrawal, these observations suggest that, besides progesterone, a network of local regulators participates in the control of endometrial ECM remodeling.
In conclusion, this report demonstrates that LEFTY-A stimulates the production of human MMPs that are induced selectively during menstruation and that progesterone controls both the expression of LEFTY-A and its effect on these MMPs when combined to estradiol. This may explain the dramatic increase in LEFTY-A mRNA during the perimenstrual phase and suggests that LEFTY-A is a key local regulator accounting for the focal ECM breakdown in the cycling human endometrium.