| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Regular Article |
a Infertility Clinic, Department of Gynaecology and Obstetrics, Geneva University Hospital, 1211 Geneva 14, Switzerland
b Fondation pour Recherches Médicales, University of Geneva, 1205 Geneva, Switzerland
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
corpus luteum, follicular development, granulosa cells, ovulation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The development of the ovarian follicle, its rupture at ovulation, and the subsequent formation of the corpus luteum constitute one of the most striking examples of tissue remodeling in adult mammals. Until ovulation, the inner follicle is isolated by a basement membrane and remains avascular. The ovulatory surge of LH triggers a rapid cascade of events resulting in follicular rupture and oocyte expulsion. Proteases from the plasmin-generating system and collagenases are activated, and acute inflammation takes place, involving vascular permeabilization with extravasation of plasma proteins into the follicular cavity where they form a fibrin clot [7]. Subsequently, granulosa and theca cells and blood vessels use the fibrin clot as a provisional matrix to fill the former antral cavity with highly vascularized luteal tissue [8]. Along this cycle, follicles failing to complete development and regressing corpora lutea are discarded by apoptosis [9]. Thus, multiple proteolytic events take place during the ovarian cycle, involving serine proteases of the plasminogen activation and coagulation cascades [10]. The importance of plasminogen activators for the breakdown of the follicular wall has been challenged by the fertility of transgenic mice deficient in various components of this system [11]. Ovulation was normal in mice deficient in urokinase and was only partly defective in mice deficient in both tissue-type and urokinase plasminogen activators [12, 13]. Whether components of the plasminogen activator system are involved in additional functions in the ovary remains to be investigated.
The activity of serine proteases of the plasminogen activation and coagulation cascades is controlled by inhibitors of the serpins superfamily, which are often modulated by heparin [14]. Serpin inhibitors such as plasminogen activator inhibitor-1 (PAI-1), protease nexin-1 (PN-1), and AT have been described in the ovary. PAI-1 and PN-1 synthesis and hormonal modulation have been documented in ovaries from gonadotropin-stimulated rats and mice, mostly at the mRNA level [15–18], and AT has been found in human follicular fluid [19, 20]. AT distribution in tissues is mainly restricted to vascular walls, where it is bound to aHSPGs [4, 21], but its distribution and synthesis in the ovary is unknown.
The expression of high levels of aHSPGs by gonadotropin-stimulated granulosa cells suggests an important function for these proteoglycans in the ovary, likely in the regulation of proteolysis through their interaction with follicular serpins. To test this hypothesis, we coordinately localized aHSPGs and the serpins AT, PN-1, and PAI-1 in the ovary during the reproductive cycle. We also manipulated the proteolytic balance at the time of ovulation by injecting an inactive form of AT, R393C-AT. This variant lacks thrombin inhibitory activity as a result of replacement of the specificity-determining arginine at position P1 by a cysteine, but it retains normal heparin binding [22] and thus can occupy AT-binding sites on aHSPGs.
Most of the ovarian physiology has been studied using gonadotropin-stimulated cycles in immature rodents [23–25]. However, commonly used protocols induce superovulation, which exaggerates the tissue response to excessive stimulation, and a better simulation of normal physiology is obtained by using naturally cycling animals [26, 27]. We used natural and gonadotropin-stimulated cycles as complementary models to determine the relative expression patterns of aHSPGs and serpins in the ovary. Because we predicted a function for aHSPGs in the ovulation process, we focused on healthy growing follicles from the ovulatory cohort, excluding atretic follicles.
This study illustrates the dynamic expression of aHSPGs and serpins in the ovary during the reproductive cycle. Their coexpression during the development of the ovarian follicle and the effect on ovulation of mutant inactive AT reveal how the proteolytic control of tissue remodeling in the ovary could involve the activation of heparin-binding serpins by granulosa cell aHSPGs.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Pure human AT was obtained from Cutter Biological (Berkeley, CA). 125I-Labeled AT was prepared as previously described, with a specific activity of 5 x 104 dpm/ng [28]. Porcine mucosal heparin was from Diosynth Inc. (Chicago, IL). Monoclonal anti-rat PN-1 antibody was kindly provided by Denis Monard (Friedrich Miescher Institut, Basel, Switzerland) [29], and polyclonal rabbit anti-rat AT antibody was a generous gift from Marc Schapira (Haematology Division, Lausanne University Hospital, Lausanne, Switzerland) [30]. Anti-rat PAI-1 antibody was purchased from American Diagnostica (Greenwich, CT). Rabbit anti-human fibrin(ogen) was purchased from DAKO Corp. (Carpintera, CA). Mouse and rabbit primary antibody isotype controls were purchased from Zymed Laboratories (South San Francisco, CA). Autoradiographic emulsion type NTB2 and Dektol developer were from Kodak (Rochester, NY). The TUNEL kit was obtained from Boehringer-Mannheim (Mannheim, Germany), and rat endothelial cell-specific monoclonal antibody RECA-1 [31] was purchased from Immuno Quality Products (Groningen, Netherlands). Human CG was purchased from Serono (Aubonne, Switzerland), and eCG was from Sigma (St. Louis, MO). All other chemicals were of the highest grade available.
Animals
Sprague-Dawley female rats were puchased from Iffa-Credo (L'Arbesle, France) and maintained on a 12L:12D light cycle. Ethical approval was obtained from veterinarian Geneva State authorities.
Natural cycle In the cycling rat, the synchronized wave of follicles developing during the estrous cycle is recruited during the preceding cycle, after ovulation. In diestrus, a cohort of small growing follicles is observed that develop through proestrus into estrus, when dominant follicles expel their oocyte. Estrus is followed by metestrus, in which the corpus luteum forms to become fully active in diestrus [9, 27, 32]. Regular cycles were documented in adult female rats by daily vaginal cytology over 10 days. Animals were killed between 0800 and 1000 h, and ovaries were recovered at each stage of the cycle (proestrus, estrus, metestrus, and diestrus).
Gonadotropin-stimulated cycle Immature 23-day-old female rats were treated with gonadotropins to induce ovulation by sequential injection of eCG (20 IU) and injection of hCG (10 IU) 48 h later [16]. The animals were killed, and ovaries were recovered before treatment (control immature), 48 h after eCG injection, and 6, 12, 24, or 72 h after hCG injection.
Injections in the ovarian bursa Injections in the ovarian bursa were performed according to the technique described by Tsafriri et al. and Bicksak et al. [23, 33]. Intrabursal injections of R393C-AT, diluted in 50 µl isotonic saline, were administered during the course of gonadotropin-stimulated cycles at the time of hCG injection, and the animals were killed 24 h later. The number of ovulated oocytes was quantified by flushing the oviducts, and the ovaries were embedded for cryosectioning. The number of ovulated oocytes in each ovary was not significantly different in animals injected with NaCl vehicle and in sham-operated animals or in animals injected with 140 µg of plasma AT (data not shown). Statistical analysis of the data was done with the Mann-Whitney nonparametric U-test, and differences were considered significant at P < 0.05.
R393C-AT
The baby hamster kidney cell line producing recombinant variant R393C-AT was a kind gift of Peter Gettins (University of Ilinois, Chicago, IL) [22]. Conditioned medium was used to purify the monomeric glycoform II of R393C-AT, which corresponds to the 
form of plasma AT [22]. Pure R393C-AT was stored at -80°C until used for intrabursal injections, at concentrations of 3.2 or 29.4 µM. For control injections, pure human plasma AT was used at a concentration of 60 µM.
Tissues
Ovaries were rapidly excised from animals, dissected free of the ovarian fat pad, embedded in Tissue Tek OCT compound (Sakura Finetek, Zoeterwoude, The Netherlands) and stored at -80°C. Cryosections (5 µm) were mounted on poly-L-lysine-coated slides and stored at -20°C until used. Duplicate serial sections were used for incubations with 125I-AT, histologic staining with hematoxylin and eosin, and immunohistochemistry. For RNA extraction, dissected ovaries were snap-frozen in liquid nitrogen and kept at -80°C until extraction. Localization of aHSPGs, immunohistochemistry, and Northern blot analysis were done in at least 3 independent experiments.
Localization of aHSPGs
Anticoagulant HSPGs were localized by 125I-AT binding on ovary cryosections followed by microscopic autoradiography. After 15 min of preincubation in PBS containing 50 µg/ml BSA, the sections were incubated with 125I-AT (15 000 cpm/µl) for 1 h at 4°C in humidified chambers. Excess unbound 125I-AT was removed by washing, and the sections were subsequently fixed in ethanol, dipped in autoradiographic NTB-2 emulsion, and developed after 6–10 days of exposure [34]. Control slides were incubated with 125I-AT in the presence of 100 µg/ml heparin, which prevented binding (Fig. 1f), whereas 100 µg/ml dextran sulfate did not (data not shown). The specificity of 125I-AT binding to aHSPGs also has been shown by preincubation of the cryosections with heparitinase, which abolished binding, whereas chondroitinase ABC did not affect binding [6].
|
Immunohistochemistry
AT labeling Endogenous AT was localized by immunoperoxidase labeling using a rabbit polyclonal anti-rat AT antibody [30]. Ovary cryosections were fixed in acetone and incubated with the primary antibody at a 1:4000 dilution. The bound antibody was identified with a biotin-conjugated goat anti-rabbit Ig secondary antibody and avidin-peroxidase-complex (ABC; Vector Laboratories, Burlingame, CA) and was revealed with diaminobenzidine (DAB substrate kit, Vector). Incubations were performed according to the manufacturer's instructions. No labeling was detected on control sections incubated with rabbit primary antibody isotype control substituting the anti-AT antibody.
PN-1 labeling Ovary cryosections were incubated with a monoclonal anti-rat PN-1 antibody at a 1:500 dilution, according to a previously published procedure [29] using the ABC kit, and labeling was revealed by incubation in DAB. In negative control sections, the primary antibody was replaced by an isotype control, and no labeling was observed.
PAI-1 labeling PAI-1 was localized with an anti-rat PAI-1 polyclonal antibody [35]. Cryosections were fixed in acetone, preincubated for 20 min in PBS containing 2% goat serum, and subsequently incubated for 3 h with anti-PAI-1 antibody, diluted 1:100 in PBS containing 5% goat serum. Bound antibodies were revealed using the ABC reagent and DAB according to the manufacturer's instructions. No labeling was detected on control sections incubated with nonimmune rabbit serum substituting for the anti-PAI-1 antibody.
Fibrin labeling For detection of cross-linked fibrin in rat ovaries, the animals were anesthetized by i.p. injection of ketalar:xylazine (2:1) (100 µl/100 g), and blood was removed by puncture of the abdominal aorta in a syringe containing sodium citrate as anticoagulant. Ovaries were then excised and frozen. Cross-linked fibrin was immunostained in ovary cryosections as described by Weiler-Guettler et al. [36]. The specificity of staining of cross-linked fibrin was assessed by parallel incubation of sections fixed in 10% buffered formalin, which retains fibrinogen and fibrin, and in 2% acetic acid (vol/vol)/10% buffered formalin to wash out non-cross-linked fibrinogen/fibrin. The fixed sections were incubated with a rabbit anti-human fibrin(ogen) antibody at a 1:16 000 dilution, and bound antibodies were revealed using the ABC reagent and DAB according to the manufacturer's instructions. No labeling was detected on control sections incubated with rabbit primary antibody isotype control substituting for the anti-fibrin antibody.
Observation and Photography
The micrographs shown in all of the figures were taken using a Zeiss Axiophot photomicroscope (Carl Zeiss, Zurich, Switzerland) equipped with Plan Neofluar 10/0.30 and Nikon E-plan 4/0 objectives (Nikon, Tokyo, Japan) and with a high-sensitivity Coolview color digital camera (Photonic Science, London, UK). Pictures were compiled by using PhotoShop Version 5.0 (Adobe System, Mountain View, CA) and printed with a digital Pictography 4000 printer (Fujifilm, Tokyo, Japan).
Serpin Probes
Probe for AT (ATIII) was a gift from P. Patston (University of Illinois, Chicago, IL). Probe ATIII is a 366-base pair (bp) HindIII-EcoRI fragment of rat AT cDNA inserted into pBluescript-KS (Stratagene, La Jolla, CA). Probe for PN-1 (rat Nexin probe 3.3) was kindly provided by D. Monard (Friedrich Miescher Institut). Probe rat Nexin 3.3 is a 320-bp PstI fragment of rat protease nexin-I cDNA inserted into pGEM-1 (Promega, Madison, WI). Probe for PAI-1 (prPAI106) was generously given by P. Sappino (Oncology Department, Geneva University Hospital). Probe prPAI106 is a 726-bp PstI-ApaI fragment of rat PAI-1 cDNA inserted into pBluescript-KS. Probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (pmGAPDH.FL) was a gift from Pierre-Alain Menoud (Institute of Histology, Fribourg, Switzerland). Probe pmGAPDH.FL is the complete cDNA of mouse GAPDH amplified by polymerase chain reation and inserted into pBluescript-KS.
RNA Extraction and Northern Analyses
RNA extraction Total RNA was isolated from homogenized rat tissues and cells by the guanidium isothiocyanate method as described by Chomczynski and Sacchi [37]. RNA was extracted with phenol:chloroform:isoamylalcohol (25:24:1), precipitated in isopropanol, and washed in 70% ethanol, as previously described [6]. Samples of total RNA (10 µg) were denatured in 10 mM sodium phosphate buffer (pH 6.8) containing 1 M glyoxal and 50% dimethyl sulfoxide and were resolved on 1.2% agarose gels. RNA was transferred to Hybond-N membranes (Amersham, Buckinghamshire, UK) by capillary blotting. The membranes were vacuum baked for 2 h at 80°C. RNA markers (Promega) were used as size and transfer efficiency markers. For quantitative comparisons of mRNA levels, hybridization signals were compared between the tested mRNA and the housekeeping gene GAPDH mRNA [38].
Northern blot analysis Prehybridization, hybridization, and washes were carried out according to published procedures for PN-1 and PAI-1 [38], and for AT these steps were performed at 68°C in 10 ml Quickhyb solution (Stratagene) following the manufacturer's instructions. Rat Nexin probe 3.3 plasmid was linearized by EcoRI, PAI-1 prPAI106 plasmid was linearized by BamHI, the cRNA were labeled with 32P-UTP using SP6 and T7 polymerases, respectively, and the probes were hybridized at 1 x 106 cpm/ml. The ATIII cDNA probe was labeled with 32P-dCTP (Hartmann, Zurich, Switzerland) with the Prime-a-Gene labeling system (Promega) and hybridized at 2 x 106 cpm/ml.
All radiolabeled probes were separated from unincorporated nucleotides using G-50 Sephadex 1-ml columns. All filters were exposed to Kodak XAR-5 film with 2 intensifying screens at -80°C for 24–48 h. Quantification of hybridization signal intensity was performed using a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To distinguish follicles bound to ovulate from degenerating follicles, ovary cryosections were labeled by TUNEL to identify apoptotic cells in atretic follicles [25]. Healthy developing follicles belonging to the ovulatory cohort were selected for the localization of aHSPGs and serpins based on their negative TUNEL staining on serial cryosections (data not shown).
Expression of aHSPGs in Ovarian Follicles During the Natural Estrous Cycle
The dynamics of aHSPGs expression in ovarian follicles of naturally cycling rats was followed by localizing aHSPGs in ovaries taken at the different stages of the cycle (Fig. 1). Ovary cryosections were incubated with 125I-AT, and aHSPG-bound 125I-AT was revealed by microscopic autoradiography. Anticoagulant HSPGs were seen on granulosa cell layers of follicles and, as punctuated labeling, on vascular endothelial cells. At diestrus, the small secondary follicles beginning the next cycle were not labeled for aHSPGs, unlike adjacent corpora lutea (Fig. 1a). At proestrus, the granulosa cells of large preovulatory follicles were intensely labeled for aHSPGs, but theca cells remained negative (Fig. 1b), and whenever present, oocytes were always negative. At estrus, ovulation triggers remodeling of the follicular wall, and aHSPGs were less abundant and the labeling was less homogeneous on granulosa luteal cells (Fig. 1c). In the forming corpus luteum, at metestrus, aHSPG labeling was marked on invading capillaries and weak and diffuse on luteal cells (Fig. 1d). At diestrus, mature functional corpora lutea were strongly labeled for aHSPGs because of their intense vascularization (Fig. 1e). Weak but positive labeling of aHSPGs was consistently observed on luteal cells in central areas of young corpora lutea. These central areas were still devoid of capillaries, as confirmed by endothelial cell-specific immunohistochemistry using RECA-1 antibody (data not shown). The specificity of 125I-AT labeling of aHSPGs was verified on serial cryosection of the diestrus ovary incubated in the presence of heparin, which abolished all binding to the tissue (Fig. 1f).
Immunohistochemical Staining of AT, PN-1, and PAI-1 in the Ovary
To understand the relative distribution of aHSPGs and of serpins in the rat ovary, the heparin-binding serpins AT, PN-1, and PAI-1 were localized in ovary cryosections by immunohistochemistry. AT was present in the rat ovary (Fig. 2a) as was PN-1 (Fig. 2c) and PAI-1 (Fig. 2e). The specificity of labeling was visible on serial cryosections incubated with specific antibodies (Fig. 2, a, c, and e) or with isotype control antibodies (Fig. 2, b, d, and f). We localized PN-1 in rat ovaries during natural and gonadotropin-stimulated cycles. PN-1 was found exclusively on granulosa cells of developing follicles, where it accumulated until the onset of ovulation to quickly disappear thereafter, and it was absent from the corpus luteum and atretic follicles. PN-1 labeling was of similar intensity in natural and stimulated cycles (data not shown). The PN-1 protein expression pattern coincided with PN-1 mRNA distribution in the ovaries of gonadotropin-stimulated mice [17]. PAI-1 mRNA expression has been described in the ovary of gonadotropin-stimulated rats [15, 16]. Accordingly, PAI-1 protein was seen in outer theca layers of developing follicles in natural and gonadotropin-stimulated rats, corresponding to the mRNA expression in theca cells [16, 39], but the protein was strictly localized on basement membrane structures (Fig. 2e).
|
Localization of AT in the Rat Ovary During Natural and Gonadotropin-Stimulated Cycles
Figure 3 shows the localization of AT in the ovary during the estrous cycle in naturally cycling animals (Fig. 3, left column, a–e) and during the gonadotropin-induced cycle in immature animals (Fig. 3, right column, f–k). Follicles of the ovulatory cohort are shown side by side for each stage. Small growing follicles were observed at diestrus (Fig. 3a) and in immature animals (Fig. 3f), preovulatory follicles at proestrus (Fig. 3b), 48 h after eCG (Fig. 3g), and 6 h after hCG (Fig. 3h) stimulation, ovulatory luteinized follicles at estrus (Fig. 3c) and 12 h after hCG (Fig. 3i), forming corpus luteum at metestrus (Fig. 3d) and 24 h after hCG (Fig. 3j), and corpus luteum at diestrus (Fig. 3e) and 72 h after hCG (Fig. 3k). AT bound to endothelial cell aHSPGs was labeled in vessel walls of the ovarian vasculature at all stages of the cycle. In small developing follicles, labeling was weak or absent (Fig. 3, a and f) and AT appeared on granulosa cells from preovulatory follicles (Fig. 3, b, g, and h). The intensity of AT labeling on granulosa cells was stronger in stimulated than in naturally cycling rats, indicating that it depends on gonadotropin levels. At ovulation, AT was concentrated mainly on periantral and cumulus cells of ovulatory luteinized follicles (Fig. 3, c and i), and after ovulation the labeling was intense on invading capillaries and on luteal cells in the forming corpus luteum (Fig. 3, d, j, and k), particularly in the central area that was still devoid of capillaries. In contrast, in the mature corpus luteum, AT labeling was mainly seen on capillaries (Fig. 3e). In addition to the AT detected on granulosa cells of preovulatory follicles, AT was also present throughout the ovarian sections; it appeared around ovulation and remained elevated thereafter. Nonspecific AT labeling was excluded by the highly specific staining restricted to vascular walls observed in neighboring oviduct sections that were present on the same slides (data not shown). The widespread labeling seen in the ovary suggests that AT leaks from permeabilized vessels to bathe the ovarian tissue and concentrates on granulosa cell aHSPGs.
|
Expression of Serpins in the Rat Ovary During Natural and Gonadotropin-Stimulated Cycles
Having observed the presence of the heparin-binding serpins AT, PAI-1, and PN-1 proteins in rat ovarian follicles, we addressed the question of their biosynthesis in the naturally cycling ovary and in isolated granulosa cells.
Figure 4A shows the results of Northern blot analysis of AT, PN-1, and PAI-1 mRNA expression in total RNA extracted from rat ovaries at the different stages of the natural cycle. The expression level of serpins was compared with that of control organs, and the amount of RNA present in each lane of the blot was verified using GAPDH expression. AT mRNA was detected at low levels throughout the natural cycle, demonstrating that AT is synthesized locally in the ovary. PN-1 mRNA was detected in the ovary at levels comparable to that seen in brain, and PN-1 expression was sustained throughout the cycle. Quantification of AT and PN-1 mRNA did not show consistent variations during the natural cycle. PAI-1 was expressed in the ovary at levels considerably lower than those in the placenta. In 5 independent experiments, we observed a transient 2-fold decrease in PAI-1 mRNA levels at metestrus, the stage of intense tissue remodeling after ovulation. PAI-1 mRNA expression has been previously documented in gonadotropin-stimulated ovary and granulosa cells with similar expression patterns [15–17].
|
We further analyzed the expression of AT and PN-1 in isolated granulosa cells, from gonadotropin-stimulated ovaries. Granulosa cells were taken at the preovulatory stage 48 h after eCG stimulation, and granulosa luteal cells were isolated just before ovulation 6 h after hCG treatment and at the postovulatory stage 24 h after hCG treatment (Fig. 4B). Although AT mRNA was readily detected in total ovary RNA, it was undetectable in granulosa cell RNA extracts. These data demonstrate that the expression of AT mRNA observed in the ovary cannot be accounted for by granulosa cells. The low AT mRNA expression seen in the ovary probably is due to endothelial cells. In contrast, a strong PN-1 expression was detected in isolated granulosa cells, suggesting that granulosa cells are a main source of PN-1 in the rat ovary. The signal observed in granulosa luteal cell mRNA 24 h after hCG administration was decreased by about 30%, consistent with the disappearance of PN-1 in ovulated follicles.
Coordinate Localization of aHSPGs and Serpins in Ovarian Follicles
To further document interactions between aHSPGs and serpins in the ovary, we localized their respective expression patterns to the level of single follicles. The coordinate localization of aHSPGs and the serpins AT, PN-1, and PAI-1 were determined in serial ovary cryosections taken during natural and gonadotropin-stimulated cycles. Figure 5 illustrates the estrous cycle in naturally cycling adult rats, and Figure 6 illustrates the gonadotropin-stimulated cycle in immature rats. The successive stages of the cycle are presented in rows, and the localization of aHSPGs and serpins are in columns.
|
|
In the natural cycle, at diestrus of the preceding cycle (first row) moderate levels of aHSPGs were present on granulosa cells of early antral follicles (Fig. 5a), with a strong signal for PN-1 (Fig. 5c); AT signal is low, and PAI-1 is absent from the inner follicle (Fig. 5, b and d). In the preovulatory follicle at proestrus (second row), aHSPG labeling of granulosa cells was intense (Fig. 5e), AT was present on granulosa cells, mostly on periantral layers (Fig. 5f), PN-1 was abundant in granulosa cells, particularly on periantral layers (Fig. 5g), and PAI-1 was present in basement membranes, with diffuse labeling spread to granulosa cells (Fig. 5h). At estrus, in ovulated follicles (third row) aHSPGs were most concentrated around the former antral cavity (Fig. 5i), a pattern reproduced by AT and PAI-1 (Fig. 5, j and l); PN-1 had mostly disappeared (Fig. 5k). In the forming corpus luteum at metestrus (fourth row), aHSPGs and AT had a similar distribution, with superimposed diffuse labeling on luteal cells and punctuated labeling on invading capillaries (Fig. 5, m and n). PAI-1 was also present on the whole surface of the young corpus luteum (Fig. 5p), and PN-1 had completely disappeared (Fig. 5o). At diestrus (fifth row), the mature corpus luteum was strongly labeled for aHSPGs (Fig. 5q), AT and PAI-1 were mainly restricted to capillaries, and PN-1 was absent (Fig. 5, r–t). Thus, in the natural cycle, aHSPGs coincided with PN-1 and AT in the inner follicle before ovulation and with AT and PAI-1 in the corpus luteum.
In gonadotropin-induced cycles (Fig. 6), 48 h after eCG induction of follicular growth (first row) aHSPGs were abundant on granulosa cells of a large antral follicle (Fig. 6a), AT and PN-1 were present and colocalized with aHSPGs, particularly in the periantral region (Fig. 6, b and c), and PAI-1 was weakly expressed in theca layers (Fig. 6d). Shortly before ovulation, 6 h after induction by hCG (second row), aHSPGs, AT, and PN-1 labeling was maximal, showing that AT and PN-1 are exactly colocalized with aHSPGs on the whole surface of the granulosa cell layer (Fig. 6, e–g), whereas PAI-1 was strictly restricted to basement membranes (Fig. 6h). In the ovulatory follicle seen 12 h after hCG priming (third row), aHSPGs and AT were still abundant and coordinately localized on granulosa luteal cells (Fig. 6, i and j). In contrast, PN-1 was markedly decreased and PAI-1 was still absent from the inner follicle (Fig. 6, k and l). After ovulation, 24 h after hCG (fourth row), a diffuse labeling was visible on the forming corpus luteum for aHSPGs, AT, and PAI-1 (Fig. 6, m, n, and p), whereas PN-1 had mostly disappeared (Fig. 6o). At 72 h after ovulation induction (fifth row), aHSPGs were visible on luteal cells in the center of the corpus luteum and in invading capillaries (Fig. 6q) with AT and PAI-1 (Fig. 6, r and t); PN-1 was absent (Fig. 6s). Thus, in the gonadotropin-stimulated cycle, aHSPGs were colocalized with AT at all stages, with PN-1 before ovulation, and with PAI-1 after ovulation.
Comparison of the data from natural and gonadotropin-stimulated cycles shows subtle differences in the distribution of serpins. Although the PN-1 expression pattern was identical in both cases, AT and PAI-1 fluctuations were less intense during the natural cycle than during the stimulated cycle. Labeling of granulosa cells by PAI-1 was more evident in the natural cycle proestrus stage than in the preovulatory follicles observed 6 h after hCG stimulation.
Altogether, these data demonstrate that aHSPGs are present in follicles with the heparin-binding serpins PN-1, AT, and PAI-1. Anticoagulant HSPGs are colocalized with PN-1 during follicular growth and with both PN-1 and AT in preovulatory stage. After ovulation, aHSPGs coincide with AT and PAI-1 during corpus luteum formation. These observations show that aHSPGs are adequately positioned in time and space to activate sequentially these 3 protease inhibitors.
Effect of Intrabursal Injection of Variant R393C-AT on Ovulation
The role of aHSPGs in the regulation of serpin activity in the ovary was further demonstrated by in vivo experiments of injections of R393C-AT in the ovarian bursa during gonadotropin-induced cycles in immature rats. R393C-AT (9 µg or 85 µg in 50 µl NaCl) was injected 48 h after eCG stimulation, at the time of hCG injection. Ovulation was scored 24 h later by counting oocytes present in oviducts (Table 1), and the deposition of cross-linked fibrin was visualized by immunohistochemistry in ovary cryosections (Fig. 7).
|
|
The number of oocytes ovulated by ovaries injected with 85 µg R393C-AT was significantly less than that in ovaries injected with NaCl vehicle. With a smaller dose of R393C-AT (9 µg), there was a smaller decrease in the number of ovulated oocytes, suggesting a dose-dependant effect of R393C-AT in the occupation of available aHSPGs AT-binding sites.
The histology and the deposition of cross-linked fibrin in ovaries injected with R393C-AT have been compared with those of control NaCl-injected ovaries. At 24 h after hCG ovulation induction, the histology was comparable for ovaries injected with R393C-AT and those injected with NaCl, with a high content in forming corpora lutea. However, several luteinized follicles in R393C-AT-injected ovaries still contained oocytes (Fig. 7b), whereas this finding was rare in control ovaries. Figure 7a shows heavy cross-linked fibrin deposits in R393C-AT-injected ovaries that are interspersed in the cell layers of the luteinizing follicle, covering much of the follicular cavity and embedding the oocyte. Fibrin deposits were also evident in granulosa cell layers of a small follicle that did not belong to the ovulatory cohort (Fig. 7, lower right, a and b). In comparison, NaCl-injected ovaries displayed moderate cross-linked fibrin deposits 24 h after hCG treatment, and in particular, small follicles were not stained (Fig. 7c). The discrete distribution of cross-linked fibrin is in contrast with the widespread localization of fibrinogen/monomeric fibrin in ovarian tissue (Fig. 7e), and the specificity of the immunodetection is further illustrated by the absence of signal when the primary antibody was replaced by an isotype control antibody (Fig. 7f).
The injection of R393C-AT in the ovarian bursa resulted in a significant decrease in the number of ovocytes ovulated, with increased fibrin deposition in the ovary. These results indicate that the serpin-activating potential of aHSPGs plays an active role in the control of the proteolytic balance in the ovary.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In naturally cycling animals, physiologic levels of gonadotropins ensure normal ovulation and corpus luteum function, whereas follicles that fail to respond to gonadotropin recruitment and regressing corpora lutea are discarded by apoptosis. In contrast, ovulation induction by high doses of gonadotropins results in superovulation of an oversized cohort of follicles because of the rescue action of FSH, which inhibits apoptosis in granulosa cells [40, 41]. After ovulation, large scale atresia occurs in follicles that fail to ovulate and maturation of the corpus luteum is delayed, as reported in hamsters and mice [26, 42]. Thus, the parallel examination of natural and gonadotropin-stimulated cycles allowed construction of a detailed dynamic picture of the inhibitory mechanisms available to control serine protease activities during the development, maturation, and luteinization of the ovarian follicle.
During the natural cycle, aHSPGs expression was intense on granulosa cells of preovulatory follicles. Similar labeling was observed in the periovulatory period during the gonadotropin-induced cycle [6], suggesting that physiologic levels of FSH are sufficient for inducing maximal levels of aHSPGs on granulosa cell layers. In addition, observations in cultured granulosa cells [5, 43] and in rat and human follicular fluid (unpublished observations) indicate that additional soluble aHSPGs are released by granulosa cells in follicular fluid. Such soluble aHSPGs are not seen in unfixed cryosections, where follicular fluid is lost during incubation. After ovulation, at metestrus, aHSPGs were transiently decreased in luteal cells of postovulatory follicles, and at diestrus a very intense labeling was restored on the mature corpus luteum because of superimposed aHSPGs signals in luteal and endothelial cells.
The functional interaction between granulosa cell aHSPGs and heparin-activated serpins has been shown in a purified system by their ability to accelerate the formation of inactive thrombin-AT complexes, with a specific activity comparable to that of high-affinity heparin [5]. Thus, aHSPGs could modulate the deposition of fibrin in the follicular cavity by activating thrombin-inhibiting serpins. To identify the serpins potentially interacting with aHSPGs in the ovary, we analyzed the distribution of AT, PN-1, and PAI-1 during natural and stimulated cycles.
PN-1 is strongly expressed in the mouse genital tract, predominantly in seminal vesicles and the ovary [18]. In the ovary, PN-1 mRNA was localized by in situ hybridization in mouse granulosa cells in the periovulatory stages of gonadotropin-stimulated cycles [17]. During the natural cycle, there was constant expression of rat PN-1 mRNA in the ovary. In isolated granulosa cells, the signal was higher than that in the total ovary, indicating that granulosa cells are the main source of PN-1. Strong PN-1 protein labeling was observed on granulosa cells of developing follicles from the secondary stage until ovulation. At the onset of ovulation, PN-1 quickly disappeared from ovulatory follicles, corresponding to the moderate decrease of PN-1 mRNA seen in isolated granulosa luteal cells after ovulation. These data are in agreement with the decrease in mRNA observed in situ in mouse ovulated follicles [17]. However, abundant PN-1 protein was detected in the developing follicles present at all stages of the cycle in similar amounts in the natural and stimulated cycles. Thus, PN-1 expression does not seem to depend directly on FSH levels but is likely regulated by other paracrine factors. PN-1 protein was strictly localized in the ovary to its site of synthesis on granulosa cells, suggesting that it is bound to extracellular matrix components.
PAI-1 mRNA is expressed in the gonadotropin-stimulated rat ovary [15, 16]. We observed differences in PAI-1 protein distribution in natural and stimulated ovaries. Before ovulation, immunolocalization revealed that PAI-1, synthesized by theca cells, accumulates in follicular basement membranes. In the periovulatory period, PAI-1 reportedly peaks just before ovulation and then sharply decreases at ovulation [16]. In the natural cycle, we observed a similar pattern for PAI-1 mRNA, but the intensity of PAI-1 protein labeling at ovulation was consistently higher at estrus than at 12 h after hCG-induced ovulation in stimulated cycles (data not shown). This observation suggests that the decrease in PAI-1 expression occurring at ovulation could be due to a dose-dependent inhibition by LH. Moreover, PAI-1 was more abundant in granulosa cell layers of ovulatory follicles in the natural cycle, in agreement with recent observations in human follicles [44]. In addition, in corpus luteum, there was a strong PAI-1 signal in luteal cells and in the abundant vasculature. PAI-1 mRNA expression has been shown in corpus luteum and is thought to modulate plasminogen activator activities during corpus luteum formation and regression [15, 45]. The presence of several corpora lutea at diestrus results in prominent PAI-1 labeling of most of the ovarian tissue (data not shown). PAI-1 circulates in blood complexed to vitronectin, and PAI-1-vitronectin complexes are bound to extracellular matrix, indicating the plasmatic origin of PAI-1 in vessel walls [46, 47]. Thus, although PAI-1 expression is tightly regulated at the mRNA level in ovarian follicles, the total PAI-1 protein detected in the ovary exceeded that predicted by mRNA determination because of the contribution of PAI-1 from the vasculature.
AT is an abundant plasma protein that is mainly synthesized in the liver and to a minor extent in the kidney and aorta [30, 48]. AT has been localized in various tissues, mainly bound to aHSPGs in vascular walls [4, 21]. The occurrence of AT in human follicular fluid [19] suggests that it might be present in the ovary, but its localization and synthesis in this organ has not been reported. AT is abundant in the ovary and in vessel walls but also in extravascular ovarian tissue. In the periovulatory period, AT concentrates on structures containing aHSPGs, such as the granulosa cells of preovulatory follicles and the central regions of forming corpora lutea. In addition, around ovulation an additional diffuse AT signal develops in the ovarian stroma and remains elevated thereafter. The analysis of AT mRNA expression revealed low level of AT synthesis in the ovary, but AT mRNA was not detected in isolated granulosa cells. These data suggest that the endogenous AT mRNA present in ovary extracts is synthesized by endothelial cells [30] and that the AT protein localized in the ovary is extravasated from plasma during the vascular permeabilization that takes place at ovulation [7, 49]. The observation that AT was more concentrated in gonadotropin-stimulated ovaries is consistent with increased inflammation and vascular permeabilization occurring during superovulation.
These data demonstrate overlapping localization of the serpins AT, PN-1, and PAI-1 during the follicular cycle. To reveal the scenario of interactions connecting aHSPGs to these serpins in the ovary, we determined the locations of these proteins in the same follicles along the cycle. Comparison between the patterns observed in natural and stimulated cycles resulted in the following model. During early gonadotropin-independent stages of follicular development, PN-1 is first expressed in preantral follicles, and at the time of recruitment of the ovulatory cohort by FSH, aHSPGs appear on granulosa cells of early antral follicles, where they are mainly colocalized with PN-1. Later, AT is released from vessels in the ovarian tissue and concentrates on aHSPGs in granulosa cell layers of large antral follicles together with PN-1. The onset of ovulation, triggered by the LH surge, coincides with the quick disappearance of PN-1 and with a less dramatic decrease in AT and aHSPGs in ovulatory follicles. After ovulation, aHSPGs, AT, and PAI-1 first colocalize in the central part of the forming corpus luteum, and then during the luteal phase, aHSPGs, AT, and PAI-1 are mainly colocalized on the capillary network of the mature corpus luteum.
To further investigate the functional role of aHSPGs in the ovulation process, we injected variant inactive R393C-AT in the ovarian bursa at the onset of hCG-triggered ovulation. Ovulation was impaired in R393C-AT-injected ovaries, with decreased numbers of ovulated oocytes and increased fibrin deposits in the ovary. R393C-AT has a disrupted thrombin-inhibition reactive site but has a normal heparin-binding site. Thus, R393C-AT could decrease the inhibitory potential of endogenous serpins by occupying aHSPGs binding sites in preovulatory follicles, thereby highlighting the functional role of aHSPGs in ovulation.
Recent crystallographic data suggest possible functions for aHSPGs in tissues. Like heparin, aHSPGs possess a pentasaccharide of defined sequence that specifically binds and activates AT, stabilizing its conformation with a fully exposed reactive loop [50, 51]. The serpins AT, PN-1, and PAI-1 share a common binding site for heparin on their D helix [52]. The structural requirements on heparin for activation of PAI-1 and PN-1 are less restrictive than those for AT [53, 54], but their binding sites on heparin might overlap the AT-activating pentasaccharide [55]. Moreover, heparan sulfates (HS) are formed of alternating domains with low sulfate and high sulfate content [56], and several AT-binding sites can be clustered in highly sulfated domains [57]. Thus, it is likely that anticoagulant HS (aHS) contain adequate structures to accommodate the activation of PN-1 and PAI-1. Therefore, we postulate that the localization of aHS in the ovary by 125I-AT binding identifies aHSPG types also able to activate PN-1 and PAI-1.
We analyzed the effects that follicular aHSPGs could exert on PN-1, PAI-1, and AT during the reproductive cycle. PN-1 is a potent thrombin inhibitor; it inhibits plasminogen activators at lower reaction rates, and its reactivity toward these enzymes is increased by heparin. In addition, collagen type IV specifically binds PN-1 in extracellular matrix and decreases its reaction rate toward plasminogen activator [58, 59]. Granulosa cells are epithelial cells that produce extracellular matrix, including various isoforms of collagen type IV [32]. Thus, before ovulation PN-1 is likely bound to collagen IV during early follicular development, but as ovulation approaches PN-1 could also bind aHSPGs as they appear on granulosa cells. This latter interaction should activate PN-1 reactivity toward both thrombin and plasminogen activators. The quick disappearance of PN-1 at the onset of ovulation precludes its involvement in the regulation of subsequent proteolysis and suggests that it could be prohibitive for breakdown of the follicle wall. PAI-1 is localized until ovulation in theca cells that are devoid of aHSPGs, and PAI-1 is presumably associated with follicular extracellular matrix through vitronectin binding, which stabilizes it in active conformation [46, 60]. After ovulation, PAI-1 is found with aHSPGs in the corpus luteum, where it could bind to both vitronectin and aHSPGs. PAI-1 primary targets are the plasminogen activators, but in the presence of heparin its reaction rate toward thrombin is markedly increased [61], suggesting that when colocalized with aHSPGs, PAI-1 could control thrombin activity. Before ovulation, plasma AT extravasates from capillaries and is concentrated on granulosa cells of large antral follicles. After ovulation, AT follows central localization of aHSPGs in the forming corpus luteum. AT activation by aHSPGs increases its reactivity toward coagulation enzymes, mainly factor Xa and thrombin [3]. Outside of the vascular bed, AT was suggested to be involved in the modulation of fibrin formation [62]. AT circulates in a low activity conformation with minimal anticoagulant activity [51], and because it leaks from the vasculature into the ovary stroma where no aHS are present, fibrin deposition occurs [63]. In contrast, the abundant expression of aHSPGs on granulosa cells of developing follicles could stabilize AT in a fully active conformation and prevent fibrin formation in the inner follicle until ovulation. Thus, AT is critically located on granulosa cell aHSPGs to control thrombin activity in the follicle. Moreover, both PN-1 and PAI-1 could play dual roles in the regulation of serine protease activities in the follicle and could be involved in controlling plasminogen activators activity, until their interaction with aHSPGs redirects them toward the control of thrombin activity at ovulation. According to this model, aHSPGs increase greatly the inhibitory potential of PN-1 and AT in the inner follicle before ovulation. At ovulation, this inhibition is released because of the rapid disappearance of PN-1 and the decrease in aHSPGs, which greatly reduce the thrombin inhibition potential in the follicle. As a result, a fibrin clot can form in the former antral cavity and serve as a provisional matrix for invading luteal cells. Altogether, the thrombin-inhibitory potential successively present in the follicle during the cycle is impressive, with the redundant presence of 3 potent thrombin inhibitors, AT, PN-1, and PAI-1, that are likely activated by aHSPGs. This model was further supported by injections of AT variant R393C-AT in the ovarian bursa, which resulted in decreased ovulation success and the observation of luteinized follicles with entrapped oocytes, reminiscent of luteinized unruptured follicle syndrome in humans [64].
Anticoagulant HSPGs are critically expressed in the ovary to promote thrombin inhibition in the follicle and therefore to maintain the fluidity of the oocyte environment at ovulation. Studies are underway in our laboratory to further investigate the biologic functions of aHSPGs in reproduction.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
1 This work was supported by Swiss National Fund for Scientific Research grants 32-39587.93 and 32-49646.96. S.H. was the recipient of a Swiss Federal Fellowship. M.P. was supported in part by fellowships from the Sir Jules Thorn Trust and from the Fondation pour Recherches Médicales, Geneva. 
2 Correspondence: Ariane de Agostini, Fondation pour Recherches Médicales, 64 avenue de la Roseraie, 1205 Geneva, Switzerland. FAX: 41 22 347 59 79; ariane.deagostini{at}medecine.unige.ch
Accepted: August 28, 2001.
Received: December 7, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
