Progestins prevent apoptosis in a rat endometrial cell line and increase the ratio of bcl-XL to bcl-XS.

Endometrial cell proliferation and cell death are regulated by ovarian hormones. The fall of ovarian progesterone in late secretory phase, or the artificial withdrawal of ovarian hormones during early pregnancy, are followed by programmed cell death of uterine epithelial cells. Aspects of this cell-specific response have been reproduced in a newly established rat endometrial cell line which expresses functional progesterone receptor. At low concentrations of serum and in the absence of glucocorticoids, these cells were dependent on progestins for survival. Removal of progesterone or addition of the antiprogestins RU38486 or ZK98299 led to a substantial increase of apoptotic cells indicated by the accumulation of internucleosomally degraded DNA. The hormonal control of cell proliferation and cell death correlated with the overall quantity and distribution of the different bcl-X transcripts. Progesterone administration not only increased total bcl-X mRNA level but also shifted the quantitative ratio between the different mRNA isoforms in favor for the apoptosis inhibiting form, bcl-XL, compared with the apoptosis promoting form, bcl-XS. These effects were rapid and could not be prevented by inhibitors of protein synthesis. As the low level of bcl-2 and bax mRNA was not influenced by progesterone treatment, the observed changes in total amount of bcl-X transcripts and spliced isoforms could represent the mechanism by which progesterone controls cell death in epithelial cells of the endometrium.

Sex steroid hormones control cell proliferation and cell differentiation in many organs, in particular in the genital tract and mammary gland. In the endometrium, growth of epithelial cells is dependent on estrogens and progesterone, and removal of ovarian hormones has been reported to cause cell death as exemplified in different species (1)(2)(3). Particularly, the degenerative response in the epithelium followed the characteristic morphological and biochemical features of programmed cell death, or apoptosis (4).
In the cycling human endometrium, apoptosis occurs in the glandular epithelium during the late secretory, the premenstrual, and the menstrual phases, and to a lesser extent during the proliferative phase (1). In the rabbit, estrogens induce proliferation of resting endometrial cells, whereas progesterone induces proliferation of already dividing cells and ultimately leads to differentiation and growth arrest. Ovariectomy of pseudopregnant rabbits, or administration of antiprogestins, cause intense apoptosis of endometrial epithelial cells (5,6). Attempts to reproduce this behavior in simplified in vitro cultivation systems have only partially been successful using primary cells (7,8).
The decision of cells to undergo apoptosis is controlled by external signals in combination with an autonomous genetic program. Cell death is regulated at several intracellular checkpoints by the action of members out of two gene families conferring opposite effects (9). The bcl-2 oncogene was first identified in human B-cell tumors with a chromosomal translocation placing it within the immunoglobulin locus. Bcl-2 was the founding member of a growing family of genes controlling apoptosis. These genes show homology clustered in two regions, BH1 and BH2 (10). Bcl-2 protects cells against some forms of apoptosis, but does not induce cell proliferation (9). Another member of the family is bax which promotes apoptosis and encodes a protein that can form homodimers or heterodimers with Bcl-2 (10). The ratio of Bcl-2 to Bax determines the sensitivity of some cells to apoptotic stimuli and could be the target of various factors and signals that influence cell survival. Another homologue of Bcl-2, Bcl-X, exists in two isoforms generated by alternative splicing. The large form, Bcl-X L , which contains BH1 and BH2, protects cells against cell death, whereas the short form, Bcl-X S , promotes cell death by inhibiting Bcl-2 and Bcl-X L function (11). Additional members of the bcl-2 family have been identified which influence the apoptotic pathway (12)(13)(14)(15). Interestingly, another checkpoint on the apoptotic pathway is provided by the balance of two alternatively sliced isoforms encoding different species of interleukin-1␤ converting enzyme (9).
The finding that bcl-2 is expressed in the terminal differentiated syncytial trophoblast and in endometrium suggested that Bcl-2 may be related to hormone-dependent apoptosis (16). In human epithelial endometrial cells Bcl-2 predominates at the end of the follicular phase and is low or absent in the secretory phase, when electron microscopic analysis shows apoptotic cells (17). These studies suggest that Bcl-2 may be involved in protection against apoptosis during the proliferative phase of the menstrual cycle, but does not seem to play a role in luteal phase (18). Presumingly bcl-2 could be a direct target for steroid hormones, as there are potential hormone responsive elements in the bcl-2 gene promoter (17).
To investigate the regulation of endometrial cell proliferation and differentiation by ovarian hormones, in particular by progesterone, we have recently established cell lines from rat endometrial epithelium which exhibit some of the properties of the corresponding differentiated tissue (19). Whereas some lines established by immortalization with simian virus 40 large T antigen lost most of their epithelial markers and developed a fibroblast-like phenotype, additional transformation by v-Haras partly restored the epithelial phenotype giving rise to the ancestor cell line RENTRO1 (20). Here we describe the stable gene transfer of recombinant progesterone receptor to this endometrial epithelial-like cell line to reconstitute progesterone receptor response to the formerly non-responsive line. One of the progeny cell line, RENTROP, was used to study the hormonal regulation of cell proliferation and cell death. In the presence of low serum, a fraction of the cell population undergoes apoptosis which leads to a progressive decrease in the overall cell number. Survival of these serum-starved cells can be restored by progesterone, as well as the synthetic glucocorticoid, dexamethasone, which lead to a decrease in the proportion of apoptotic cells. Consistently this effect can be antagonized by the antiprogestin RU486. Progestins and glucocorticoids also induce bcl-X transcripts. Induction of bcl-X mRNA is accompanied by a shift toward a relative increase of the bcl-X L encoding isoform over the bcl-X S isoform. As the cells express low level of bcl-2 mRNA and as their level of bax mRNA is not influenced by hormonal treatment, our results suggest that the effect of progesterone deprivation on the apoptosis in the endometrial epithelium is mediated by a reversal of the ratio between apoptosis-inducing and apoptosis-preventing isoforms of bcl-X.

EXPERIMENTAL PROCEDURES
Generation of Stable Cell Lines-An endometrial epithelial cell line that stably expresses the progesterone receptor was generated by applying two rounds of calcium phosphate precipitate-mediated gene transfer to the rat endometrial RENTRO-1 cell line (20). First tetracycline repressor-VP16 expression plasmid (21) was stably introduced into RENTRO-1 cells using the hygromycin-B phosphotransferase gene as selection marker (22). One of the selected clones, which is referred to as parental RENTRO-1, was stably transfected with a tetracyclineinducible vector for the rabbit progesterone receptor (23) (expression vector constructed and supplied by M. Gossen, Heidelberg) using puromycin-N-acetyltransferase as selection marker (24). Several puromycin-resistant clones were obtained, which expressed functional PR in the absence of tetracycline, one of which, RENTROP, was selected for use in this study.
Cell Culture and Treatments-Parental RENTRO-1 cells, as well as the progeny RENTROP cells, were cultured on polystyrene plastic dishes (Greiner) in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 5% fetal calf serum (FCS, 1 c.c. pro GmbH, Neustadt W., Germany) and 5% newborn calf serum (Life Technologies, Inc.) containing penicillin (100 IU/ml), streptomycin (100 g/ ml), and glutamine (2 mM) at 37°C under a humidified atmosphere with 5% CO 2 . RENTROP cells were maintained in medium supplemented with unstripped serum in the presence of tetracycline at 10 g/ml. Before the experiments, tetracycline was omitted for at least 1 week. Apoptosis was studied in cells cultured in Dulbecco's modified Eagle's medium supplemented with dextran-charcoal stripped FCS (CS-FCS) (25) containing phenol red. For serum depletion cells were grown in medium containing 10% CS-FCS for 9 h before the medium was adjusted to the desired serum concentration in the presence or absence of added hormones. Hormones were applied from 1000-fold stock solutions in ethanol. Unless stated otherwise the following hormone concentrations were used: the synthetic progestin R5020, 10 nM; progesterone (Sigma), 0.1 M; dexamethasone (Sigma), 10 nM; the antiprogestin RU38486 (Roussel-Uclaf, Romainville, France), 1 M; and the antiprogestin ZK98299 (Schering AG, Berlin, Germany), 0.1 M.
Viable cells were counted in a Neubauer hemocytometer by the trypan blue dye exclusion method. Trypsinized cells were counted within 5 min to avoid further damage.
Transient Transfection Protocol-Transient transfections were performed by the calcium phosphate precipitation procedure as described (26). As progesterone and glucocorticoid responsive reporter a plasmid pGRE 2 -tk-Luc (5 g) was used (constructed and kindly supplied by K. Joos, Marburg), containing a doublet of the canonical progesterone/ glucocorticoid responsive element (27) in front of the truncated (Ϫ109 to ϩ51) promoter of the thymidine kinase gene of herpes simplex virus driving the luciferase gene. The response of the reporter plasmid to glucocorticoids and progestins has been verified in various endometrial as well as non-endometrial cell lines (data not shown).
Preparation of Cells for Electron Microscopy-Cells were fixed in 2.25% (w/v) formaldehyde, 1.25% (w/v) glutaraldehyde, and 0.025% (w/v) picric acid in 100 mM sodium cacodylate, pH 7.3, contrasted with 1% OsO 4 under reducing conditions, and soaked in 0.3% uranyl-acetate solution and 50 mM maleic acid as described (28). After dehydration the samples were mounted in Epon and slides were prepared using an Ultracut E microtome (Leica, Bensheim, Germany). Contrasting was intensified through incorporation of lead citrate and micrographs were taken employing an electron microscope EM9 (Zeiss, Oberkochen, Germany).
Fluorochrome DNA Nick End Labeling of Cells in Situ-Cells were fixed in situ in 4% phosphate-buffered saline-buffered formaldehyde. After equilibration in terminal deoxynucleotide transferase buffer (30 mM Tris, 140 mM sodium cacodylate, pH 7.2, 1 mM CoCl 2 ) for 30 min at room temperature, cells were incubated in terminal deoxynucleotide transferase buffer containing 40 M fluorescein-15-dATP (Boehringer, Mannheim, Germany) and 0.25 units/l of terminal deoxynucleotide transferase (Boehringer, Mannheim) in a humidified atmosphere at 37°C for 1 h. The reaction was terminated by incubating the cells in a solution containing 300 mM NaCl and 30 mM sodium citrate for 10 min at room temperature. Cells were counterstained with an aqueous solution of 4 g/ml propidium iodide for 15 min. Cells were rinsed with water, mounted in Mowiol (Höchst, Frankfurt) with 0.2 g/liter paraphenylenediamine, and examined by fluorescence microscopy (Leitz, Wetzlar).
Genomic DNA Analysis-For the detection of degraded DNA products, DNA from 3 ϫ 10 6 cells was isolated according to Ref. 29. Briefly, the cells were lysed in 0.1 ml of lysis buffer (50 mM Tris, pH 7.5, 20 mM EDTA, and 1% Nonidet P-40) for 30 s. After centrifugation 2,000 ϫ g for 5 min at room temperature apoptotic DNA remained in the supernatant while intact nuclear DNA was recovered in the pellet. The supernatants were treated at 56°C for 2 h with RNase A (5 mg/ml) and SDS (1%) followed by digestion with proteinase K (2.5 mg/ml) at 37°C for 18 h. DNA was precipitated and electrophoresed in 1.5% agarose gels containing 0.5 g/ml ethidium bromide and visualized under UV light. A 100-bp DNA ladder (Boehringer, Mannheim) was used as size marker. Quantitation of DNA was performed as described in the figure legends.
RNA Analysis-Total cellular RNA was isolated by the guanidinium thiocyanate-phenol-chloroform extraction method (30). RNase protection analysis was performed as described below (31). For preparing the bcl-X probe, plasmid pGLD3 was digested with HinfI and transcribed by T3 RNA polymerase. The full-length transcript size of the bcl-X riboprobe was 294 bp, and the protected fragments for bcl-X L and bcl-X S were 237 and 155 bp long, respectively. For preparing the bcl-2 probe, plasmid prBCL2 was digested with PvuII and transcribed by T3 RNA polymerase. The full-length transcipt size of the riboprobe was 416 bp, and the protected fragments for bcl-2␣ and bcl-2␤ were 371 and 234 bp long, respectively. The rat gapdh template pTRIGAPDH (Ambion, Austin, TX) was digested with BglI and transcribed with T3 RNA polymerase. The probe length was 204 bp and the size of the protected fragment was 145 bp. [␣-32 P]CTP (Amersham, Braunschweig) radiolabeled RNA probes were prepared using a kit according to the instructions of the manufacturer (Promega, Madison, WI). The probes were coprecipitated with RNA samples and dissolved in hybridization buffer, denatured at 95°C for 10 min, and hybridized at 52°C for 18 h. After digestion with RNases A and T 1 , followed by digestion with proteinase K, the samples were precipitated, denatured, and subjected to electrophoresis on a 5% denaturing acrylamide gel. Quantitation was performed with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA) using Image-Quant software. In all cases the quatitation was normalized against the gapdh signal.
For Northern blot analysis 25 g of total RNA were electrophoresed in a 1% agarose-formaldehyde gel and transferred to nylon membrane, type NY 13N (Schleicher & Schuell). Membranes were hybridized with 32 P-labeled 400-bp fragment of rat bax (32) for 16 h and washed in 2 ϫ SSC, 0.1% SDS at 55°C. Quantitation was done from densitometric scans of autoradiographs; normalization was calculated referring to the 28 S ribosomal RNA or the gapdh signal.
Incorporation of L-[ 35 S]Methionine-1.5 ϫ 10 5 cells were incubated in 6 wells plates in medium containing 10% CS-FCS for 12 h. The medium was removed and 1 ml of medium with 1% CS-FCS and 1 Ci of [ 35 S]methionine (Hartmann Analytic, Braunschweig, Germany, 37 TBq/mmol), and incubation was continued for 5 h in the presence or absence of cycloheximide (100 ng/ml). The incubation was stopped by harvesting the cells in phosphate-buffered saline containing 0.2% SDS. Cells were lysed by treatment with 10% trichloroacetic acid for 10 min on ice and the lysates were filtrated through glass microfiber filters GF/C (Whatman). Filters were washed twice with 5% trichloroacetic acid and once with ethanol and radioactivity on the filters was quantitated by liquid scintillation counting. The results are expressed as incorporated disintegrations/min/well and correspond to the average of two independent experiments performed by triplicates.
Western Blots-Protein extracts were prepared by lysing cells at 4°C for 1 h in 2 volumes of lysis buffer (50 mM Tris-Cl, pH 8.0, 150 mM NaCl, 0.02% sodium azide, 100 g/ml phenylmethylsulfonyl fluoride, 1 g/ml aprotinin, 1% Nonidet P-40). The lysate was centrifuged at 13,000 rpm and 4°C for 10 min, and the pellet discarded. Protein concentration in the supernatant was measured by Bradford assay (Bio-Rad). After adjusting to sample buffer and boiling for 5 min, 100 g of protein were applied on a 15% SDS-polyacrylamide gel, and electrophoresis was performed at 25 mA for 2 h. The resolved proteins were transferred to an Hybond ECL membrane (Amersham) by electroblotting. Antibody incubation was performed in blocking buffer (1% skim milk, 0.5% Tween) in Tris-buffered saline at 4°C. As primary antibodies were applied to rabbit polyclonal anti-Bax (N-20, Santa Cruz Biotechnology) and rabbit anti-actin (Sigma Immunochemicals). As secondary antibody, a peroxidase-labeled anti-rabbit antibody was used (Amersham). Proteins bands were detected using the ECL kit (Amersham).

RESULTS
Generation of the RENTROP Cell Line-The previously characterized RENTRO-1 endometrial cell line, as well as the parental clone used in these experiments, express glucocorticoid receptor but do not contain significant levels of progesterone receptor (PR), as demonstrated in steroid binding assays (20) and transient transfection (Table I). To obtain a progestinresponsive endometrial epithelial-like cell line the parental RENTRO-1 cells were stably transfected with an expression vector for the rabbit PR (23). Several clones, including RENT-ROP, were shown to express a functional PR, as demonstrated in transient transfection experiments using the progesterone inducible reporter plasmid pGRE 2 -tk-Luc (Table I). The level of induction with the synthetic progestin R5020 in RENTROP cells was comparable to values obtained in other PR expressing cell lines, as for example, mammary carcinoma-derived T47D (33) or diverse rabbit endometrial cell lines. 2 Note that in RENTROP cells the response to dexamethasone was about 5-fold higher than the response to R5020, suggesting a high level of constitutive glucocorticoid receptor expression. Note that the differences in absolute value of glucocorticoid induction between RENTRO-1 and RENTROP cells do not reflect significant differences in glucocorticoid receptor activity between the cell lines, as independently confirmed by another series of reporter transfection experiments (data not shown).
In the absence of hormone, RENTROP cells cultivated in medium supplemented with charcoal-stripped FCS exhibited a morphology which did not differ significantly from that of the parental RENTRO-1 cells. The cells were well attached to the plastic surface, polygonal or elongated in shape, had an enlarged cytoplasm and a regular cell surface (Fig. 1, left panel). Following treatment with progestins, RENTROP cells, but not parental RENTRO-1 cells (data not shown), changed morphology. While some cells showed a more elongated spindle-like appearance and exhibited dendritic protrusions, other cells rounded up and became only weakly attached to the plastic surface ( Fig. 1, right panel). In addition, some enlarged cells appeared exhibiting a larger cell nucleus (see also ultrastructure, Fig. 3). Concomitant treatment with the antiprogestin RU486 antagonized the morphological changes induced by progestins. Interestingly a similar morphological change was observed in response to the synthetic glucocorticoid dexamethasone both in RENTROP cells as well as the parental RENTRO-1 cell line (data not shown).
Hormone Withdrawal Induces Apoptosis-The growth be-  b The variation in dexamethasone induction between the two cell lines is solely due to difference in transfection efficiency as verified in an independent experiment applying a different hormone-dependent reporter (data not shown).
havior of RENTROP cells was influenced by the concentration of CS-FCS and by the presence of steroid hormones in the medium. At 10% CS-FCS the cell number was increased by addition of the synthetic progestin R5020 (Fig. 2, upper panel) or the synthetic glucocorticoid dexamethasone and this effect was reverted by the antagonist RU486 (data not shown). In the presence of 1% CS-FCS and under the influence of glucocorticoids or progestins, RENTROP cells maintained their number in culture. In the absence of hormone or in the presence of antiprogestional RU486 the cell number decreased due to a loss of dying cells (Fig. 2, middle panel). At 0.1% CS-FCS even addition of hormones was not sufficient to prevent cell death (Fig. 2, lower panel).
An electron microscopical analysis of cells cultivated at low serum in the absence of progestins or glucocorticoids (Fig. 3A), or in the presence of the antihormone RU486 (Fig. 3B), demonstrated apoptotic cells with condensed cell nucleus, shrunk cytoplasm, but with intact organelles. Furthermore, the cells exhibited membrane blebs at the cell surface, a characteristic feature of apoptotically dying cells. In the presence of either R5020 (Fig. 3C) or dexamethasone (data not shown), the cells showed an even surface, and some of them were enlarged and exhibited an extensively developed ergastoplasm, indicative of secretory cells (see higher magnification in Fig. 3D).  After 24 h total DNA was extracted and 5 g were analyzed by electrophoresis in a 1.5% agarose gel. The gel was stained with ethidium bromide and photographed (inset). The negative was scanned with a densitometer and the density plotted against the migration distance in centimeters. The area corresponding to DNA fragments smaller than 2000 bp was quantified. A scan of the 100-bp ladder marker DNA (lane 1) is shown at the bottom.
The fraction of RENTROP cells that undergo apoptosis when cultivated under serum starvation (1% FCS) was determined using a cytochemical assay that detects DNA nicks (34). A representative example of the results is shown in Fig. 4. Quantitation of the stained cells grown in the absence of ligand (Fig.  4, upper left panel) showed that 74% (Ϯ6) of the cells were positive for nicked DNA. In the presence of progesterone (Fig.  4, upper right panel) the proportion of apoptotic cells decreased to 39% (Ϯ9), and addition of the antagonist RU486 counteracted the effect of progesterone (Fig. 4, lower right panel), leaving 72% (Ϯ3) of apoptotic cells. Dexamethasone treatment also reduced the proportion of apoptotic cells to 44% (data not shown). The effects of glucocorticoids and progestins were additive (Fig. 4, lower left panel), leaving only 22% (Ϯ7) of apoptotic cells.
A similar result was obtained when the genomic DNA of the cells was analyzed in agarose gel electrophoresis for the occurrence of the characteristic nucleosomal ladder found in apoptotic cells. Using an assay which selectively solubilizes nuclease cleaved DNA (29), a nucleosomal ladder was found with serum-starved cells (1% CS-FCS) grown in the absence of hormone (Fig. 5A, left panel). The proportion of degraded DNA was estimated after incubation with various hormones (Fig. 5B). The numbers obtained from the densitometric scans confirmed the values found with the cytochemical assay: both dexamethasone and R5020 reduced the proportion of fragmented DNA as observed in the absence of hormone or in the presence of RU486.
We conclude that the effect of R5020 on the cell population at low serum as detected in the growth kinetics of RENTROP cells (Fig. 2), is due to a protection of cells otherwise undergoing programmed cell death.
Hormonal Effects on bcl-X Transcripts-To analyze the molecular mechanism by which hormones prevent apoptosis of RENTROP cells we first measured their effects on the expression of bcl-2 family members which are known to control apoptosis in many other cell types. Using RNase protection assays we could not detect expression of the two bcl-2 transcripts, which should correspond to the rat homologues of the human ␣ and ␤ isoforms of bcl-2 mRNA (35) (Fig. 6A). We conclude that Bcl-2 does not play an essential role in hormonal regulation of apoptosis in RENTROP cells.
Next we measured the expression of bax, as in many cells Bax promotes apoptosis by opposing the action of Bcl-2 (36). In total RNA from RENTROP cells bax was detectable by Northern blot analysis. After correction for RNA loading, based on quantitative estimation of 28 S ribosomal RNA from densitometric scans and confirmed by quantitation of the gapdh signal, no change in the expression level was detected after various hormonal treatments (Fig. 6B and Table II). Western blots performed with anti-Bax antibodies did not detect changes in the levels of the Bax protein following hormone treatment (Fig.  6C). These results suggest that the ratio of Bcl-2 to Bax is not likely involved in mediating apoptosis following hormone withdrawal.
An alternative possibility to control apoptosis would be to influence the ratio of the two Bcl-X isoforms. Whereas the large isoform, Bcl-X L , which contains all regions of homology to Bcl-2, protects against several forms of apoptosis, the short isoform, Bcl-X S , which lacks two of the homology regions, BH1 and BH2, promotes apoptosis by counteracting Bcl-2 and Bcl-X L function (36). Therefore, the ratio between the two isoforms could decide whether a given stimulus will cause apoptosis. In RNA from thymus and RENTROP cells both forms of bcl-X mRNA were detected by RNase mapping (Fig. 7). This result was also confirmed by semi-quantitative reverse transcriptase-polymerase chain reaction (data not shown).
Both in thymus and in RENTROP cells, transcripts of the large isoform were considerably more abundant than tran- 6. Expression of bcl-2 and bax transcripts. A, detection of bcl-2 transcripts. RNase protection assay of total RNA harvested from RENTROP cells cultivated in 10% CS-FCS for 9 h and further incubated in 1% CS-FCS for 2 h (lanes 1, 3, 5, and 7) and 5 h (lanes 2, 4, 6, and 8), without hormone (ethanol control, lanes 1 and 2), with RU486 ( lanes 3 and 4), with dexamethasone (lanes 5 and 6), or with R5020 (lanes 7 and 8). The protected bands of both isoforms are indicated by arrows. Lane 9, RNA from rat thymus as positive control. Lane 10, RNA from rat liver as negative control. All samples were hybridized with the gapdh riboprobe for normalization. B, detection of bax transcripts.  9 and 10). Lane 1, t-RNA as negative control; lane 11, rat thymus RNA as positive control. Lower panel, rRNA was stained with ethidium bromide after transfer to the nitrocellulose membrane as control of recovery. RNA was quantified by scanning the autoradiograph and data were normalized to the 28 S rRNA band. C, immunoblotting of Bax protein. Bax protein was measured by Western blot analysis of total extract harvested from RENTROP cells cultivated in 10% CS-FCS for 9 h and further incubated in 1% CS-FCS for 2 h (lanes 2, 4, 6, and 8) and 7 h (lanes 3, 5, 6, and 9), without hormone (ethanol control, lanes 2 and 3), with dexamethasone (lanes 4 and 5), with RU486 ( lanes 6 and 7), or with R5020 (lanes 8 and 9). Lane 1, protein size marker. Bax protein was quantified by scanning the autoradiograph and data were normalized against actin as control. Normalized values are shown at the bottom of the figure. scripts of the short isoform (Fig. 7A, lanes 2 and 3). Following hormonal treatment, transcripts for both isoforms accumulated rapidly, the effect of dexamethasone being more pronounced than that of R5020, but the proportion of bcl-X L mRNA increased more markedly (Fig. 7A, lanes 8 -11). As a consequence the ratio of short to large isoforms decreased by 2-fold, from 0.08 to 0.04, following treatment with either glucocorticoids or progestins (Fig. 7B, lower panel). The hormonal effects on the total level of bcl-X transcripts and on the ratio of both isoforms was almost maximal after 2 h of hormone treatment (Fig. 7B). Both effects were inhibited by the antihormone RU486 which changed the ratio of bcl-X S to bcl-X L from 0.08 to 0.16 ( Fig. 7B and Table III). In the parental RENTRO-1 cells only glucocorticoids but not progestins increased bcl-X transcripts and decreased the ratio of short to long isoforms (Table III). Thus, the effects of various hormonal treatments on the ratio of bcl-X isoforms correlated with their effects on the proportion of apoptotic cells.
To test whether the hormonal influence on the levels of the bcl-X transcripts are mediated by induction of an intermediary factor, we investigated the influence of inhibitors of protein synthesis. For these experiments, progesterone was used instead of R5020, but the effects on bcl-X transcripts were similar. Cycloheximide, at concentrations which blocked protein synthesis by 85% (the incorporation of [ 35 S]methionine without cycloheximide was 7.26 ϫ 10 3 Ϯ 0.6 ϫ 10 3 dmp/well, and in the presence of 100 ng/ml cycloheximide, 1.5 10 3 Ϯ 0.3 10 3 dpm/ well) did not influence the levels of bcl-X transcripts nor the ratio of short to large isoforms (Fig. 8). As in the previous experiments, glucocorticoids and progestins increased the level of transcripts of the bcl-X gene by 2-3-fold and this effect was not prevented by cycloheximide (Fig. 8B, top panel). The hormonally induced decrease in the ratio of bcl-X S to bcl-X L was also not prevented by treatment with cycloheximide (Fig. 8B, bottom panel) demonstrating that ongoing protein synthesis is probably not essential for this hormonal effect. DISCUSSION RENTROP Cells Reproduce the Apoptosis Response of Endometrial Tissue-The appearance of apoptosis in epithelial cells of the endometrium following ovariectomy or treatment with the antiprogestin RU486 has been described in pseudopregnant rabbits (5, 6) and in primary cell culture (7,8). Here we describe for the first time a hormone-dependent endometrial cell line. These rat endometrial cells, called RENTROP, as well as the parental RENTRO-1 cells, undergo spontaneous apoptosis by an intrinsic default mechanism when cultured under limiting concentrations of stripped serum, probably due to the lack of growth factors. The identification of apoptotic cells was based on light and electron microscopic morphology as well as on biochemical evidence for internucleosomal DNA degradation. In the presence of physiological concentrations of steroid hormones, namely glucocorticoids and progestins, apoptosis is prevented and the cell number is maintained, both effects being reverted by the antagonistic ligand RU486. Whereas the parental RENTRO-1 cell line lacks PR and only responds to dexamethasone, RENTROP cells express constitutively PR and respond to progestins as well. The behavior of RENTROP cells does not seem to represent a clonal exception as suggested by the observation of other independent cell clones derived from RENTRO-1 by stable transfection with PR which similarly respond to progestins. Moreover, a rabbit endometrial cell line, RBE7 (37), when transfected stably with an expression vector for the progesterone receptor, similarly exhibits apoptotic cell  death in response to progesterone withdrawal. 3 The effects of progestins and glucocorticoids are additive in RENTROP cells, probably reflecting the fact that the cellular concentration of each hormone receptor is limiting 3 (38). When comparing the two hormones, the effect of dexamethasone is more pronounced in agreement with a higher concentration of glucocorticoid receptor, as demonstrated in transient transfection assays with a hormone responsive reporter (Table I). Even in the presence of saturating amounts of progestins, glucocorticoids, or both hormones, a small fraction of RENTROP cells still undergoes apoptosis when cultured under conditions of serum starvation. Therefore, either the receptor concentration is still limiting, or the steroid hormones cannot completely replace other survival factors present in serum. The availability of the cell culture system described here should facilitate the identification of these factors.
Molecular Mechanism of Hormonal Regulation of Endometrial Apoptosis-Bcl-2 is known to prevent apoptosis induced by a wide range of agents, suggesting that multiple pathways to cell death converge in a step that can be regulated by Bcl-2. Bcl-2 is mainly found in cell populations with a long life and/or proliferating ability, or in cells which undergo a transition from undifferentiated stem cells to committed precursor cells. The finding of Bcl-2 in some tissues responsive to steroid hormones, such as endometrium and myometrium, suggested that bcl-2 gene expression may be related to hormone-dependent apoptosis (16). In the endometrium, Bcl-2 predominates in glandular epithelium at the end of the follicular phase, decreases at the onset of the secretory phase, and eventually disappears in the late secretory phase, when apoptotic cells accumulate (39). There are indications that bcl-2 is regulated by steroid hormones and vitamin D 3 (40,41), and potential hormone response element have been found in the bcl-2 promoter (17,18). However, the absence of significant bcl-2 expression in the endometrium of women treated with ovarian hormones to maintain the luteal phase makes it highly unlikely that Bcl-2 is important in prolonging endometrial cell survival in the secretory phase of the menstrual cycle (18). This idea is further supported by the observation that the levels of Bcl-2 in rat ovary do not change during gonadotropin-induced follicle growth (32).
Another candidate for the regulation of apoptosis is Bax, a 21-kDa protein which can associate with Bcl-2 to which it exhibits extensive homology over the two conserved regions. Overexpression of Bax accelerates apoptotic death induced by cytokine deprivation, and counteracts Bcl-2 (10). However, we could not find indications for hormonal control of bax mRNA or Bax protein in endometrial cells.
Bcl-X is also related to Bcl-2, and immunoreactivity against Bcl-X has been detected in a wide variety of cells (42). Among them were a variety of cells in the reproductive organs, including the mammary epithelium, the secretory epithelial and basal cells of the prostate, and the secretory cells of the uterine endometrium. In many cases, the patterns of bcl-X expression are strikingly different from those reported previously for bcl-2, suggesting that these two genes regulate cell survival at different stages of cell differentiation or in different cells (42).
Alternative splicing generates two isoforms of Bcl-X: a long form, Bcl-X L , similar in size to Bcl-2, which inhibits cell death (43)(44)(45)(46), and a short form, Bcl-X S , that inhibits Bcl-2 function and promotes apoptosis (47). The latter form is highly expressed in proliferating cells, whereas the long form is found in long-living cells, such as brain neurons (11). However, Bcl-2 and Bcl-X may not be redundant, since in a B cell line Bcl-X L blocked apoptosis induced by immunosuppressants whereas Bcl-2 was uneffective (48).
Our RNase mapping results indicate that in the presence of 1% serum, hormonal treatment with dexamethasone or R5020 for 2 h increases bcl-X expression 3-and 2-fold, respectively. The levels of expression of the short isoform are very low compared with those of the long isoform. The ratio bcl-X S / bcl-X L is 0.08 in cells cultivated with 10% CS-FCS and it is not affected when the cells are grown at low serum concentration. However, analysis of the ratio bcl-X S /bcl-X L after hormone treatment shows that it decreases significantly, from 0.08 in the control cells to 0.03 and 0.05 in cells treated with dexamethasone and R5020, respectively. The presence of the antihormone RU486 does not affect the total level of bcl-X message, but it increases the bcl-X S /bcl-X L ratio from 0.08 to 0.18. RU486 can antagonize completely both hormone effects: the quantity of total bcl-X message and the bcl-X S /bcl-X L ratio, demonstrating the specific action of the hormone.
The effects of progestins on bcl-X transcript levels are likely to be direct as they are rapid and cannot be affected by an inhibitor of protein synthesis such as cycloheximide. However, as the concentration of cycloheximide tolerated by these cells only blocks 85% of the total protein synthesis, we cannot exclude that the residual protein synthesis is sufficient to produce a protein involved in bcl-X transcription and/or splicing. It is also possible that the hormonal effect is mediated by a non-transcriptional effect on the activity of one or more components of the pathway leading to bcl-X transcription and processing.
There are indications for a direct effect of steroid hormones on the expression of the bcl-X gene in other cell types. For instance, in untreated myeloid cells bcl-X S transcripts could not be detected, but bcl-X L was up-regulated upon dexamethasone treatment (40). This situation is similar to that found in endometrial cells in which the bcl-X S mRNA is hardly detect-  and 7), or progesterone (Prog.) (lanes 5 and 8), and cycloheximide (Cyclo.) (100 ng/ml) (lanes 6 -8), using the bcl-X riboprobe. The protected bands of both isoforms are indicated with arrows. Lane 1, t-RNA (negative control); lane 2, time 0, cells cultivated in 10% CS-FCS containing medium. All samples were also hybridized with the gapdh riboprobe as a control of recovery. B, quantification of bcl-X expression. Top, fold induction of total bcl-X mRNA. Bottom, ratio between bcl-X S and bcl-X L . The relative numbers listed were calculated relative to the value obtained in the ethanol control (equal 1.0). Data represent average and standard deviation of three independent experiments. able, and the hormonal treatment leads to accumulation of bcl-X L transcripts. A final elucidation of the molecular mechanisms underlying these hormonal effects awaits the cloning and functional analysis of the bcl-X promoter and the characterization of the regulated splicing of bcl-X transcripts. Unfortunately, Bcl-X deficient mice will not be useful for elucidating the role of Bcl-X in the endometrium, as the homozygous bcl-X Ϫ/Ϫ mice die around embryonic day 13 due to extensive postmitotic neuronal death, apoptosis in the hematopoietic cells of the liver, and defective maturation of lymphocytes (49).