Do androgens have a direct effect on endometrial function? An in vitro study

Transcription

1 FERTILITY AND STERILITY VOL. 74, NO. 4, OCTOBER 2000 Copyright 2000 American Society for Reproductive Medicine Published by Elsevier Science Inc. Printed on acid-free paper in U.S.A. Do androgens have a direct effect on endometrial function? An in vitro study Elizabeth M. Tuckerman, M.Sc., a Marcus A. Okon, M.D., a Tin-Chiu Li, M.D., a and Susan M. Laird, Ph.D. b Jessop Hospital for Women and Sheffield Hallam University, Sheffield, United Kingdom Objective: To test the hypothesis that androgens have a direct effect on the function of endometrial epithelial cells. Design: In vitro study. Setting: Academic research center. Patient(s): Endometrial epithelial cells were prepared from biopsy samples obtained from normal fertile women. Interventions: Cells were incubated with androstenedione, testosterone, dihydrotestosterone, and DHEA. Main Outcome Measure(s): Secretion of glycodelin A into the culture fluid was used to assess secretory activity. Uptake of 3 H-thymidine and immunostaining for Ki67 was used to assess cell growth. The specific action of the androgens was confirmed by incubation with an antiandrogen, cyproterone acetate. Result(s): Androstenedione (10 6 M and 10 7 M) caused a dose-dependent decrease in glycodelin A secretion, uptake of 3 H-thymidine, and percentage of positive Ki67 cells in cultured human endometrial epithelial cells. Testosterone, dihydrotestosterone, and DHEA had no effect on glycodelin A secretion or 3 H-thymidine uptake. The direct effect of androgens on endometrial function were confirmed by demonstrating the presence of androgen receptors in cultured endometrial epithelial cells and showing that the direct effects of the androgens were not observed when cyproterone acetate was added to the cultures. Conclusion(s): The results suggest that androstenedione can inhibit human endometrial cell growth and secretory activity. Infertility and miscarriage associated with high androgen levels (e.g., that caused by the polycystic ovary syndrome) may be due to an adverse effect of high androgen levels on the endometrium. (Fertil Steril 2000;74: by American Society for Reproductive Medicine.) Key Words: Androgens, cell culture, cell proliferation, endometrium, glycodelin A Received January 4, 2000; accepted March 20, Supported by a research grant from the Special Trustees of the former United Sheffield Hospitals, Central Sheffield University Hospital, Sheffield, United Kingdom. Reprint requests: Susan Laird, Ph.D., Division of Biomedical Sciences, Sheffield Hallam University, City Campus, Sheffield S1 1WB, United Kingdom (FAX: ). a Biomedical Research Unit, Jessop Hospital for Women. b BRMC, Division of Biomedical Sciences, Sheffield Hallam University /00/$20.00 PII S (00) Several recent studies have shown that high plasma androgen levels are associated with adverse reproductive outcome, including infertility and increased incidence of miscarriage (1 4). However, it is not clear whether the relation is a causal or casual one. An earlier speculation was that high levels of LH, such as those seen in polycystic ovarian disease (and often associated with hyperandrogenemia), may adversely affect oocyte maturation and quality (2). A more recent hypothesis is that high androgen levels directly and adversely affect endometrial function and implantation, in turn leading to subfertility and recurrent miscarriage. To test such a hypothesis, androgens may be administered to volunteers and the impact of endometrial function can be studied. However, as in the case of estrogen and progestagen, the administration of a steroid hormone often affects ovarian function, and it is therefore impossible to determine whether any observed changes in endometrial function are a direct or indirect (through alteration of ovarian function) effect of the androgens. To eliminate the indirect effect of androgen on ovarian steroid hormones, two alternative strategies can be used. First, androgens may be administered in women with nonfunctioning ovaries along with standard, sequential hormone replacement therapy to mimic the natural cycle. Alternatively, the direct effects of the androgens can be examined in an in vitro endometrial cell culture system. Having considered the two strategies, we felt that an in vitro study should be conducted first to evaluate whether sufficient convincing 771

2 laboratory data exists to support a clinical study involving administration of different doses of androgens to human research subjects. In this study, we investigated the direct effect of four androgens androstenedione, testosterone, dihydrotestosterone (DHT), and DHEA on endometrial epithelial cell growth and secretory activity. In vitro uptake of 3 H-thymidine and immunostaining for Ki67 were used to assess cell growth. Secretion of glycodelin A was used to assess secretory activity because both in vitro and in vivo data suggest that glycodelin A is a reliable marker of endometrial function (5 8). MATERIALS AND METHODS Volunteers Endometrial tissue was obtained from 41 normal fertile volunteers. All tissue was obtained with informed consent from the volunteers, and institutional review board approval was received. The women were years of age and had regular menstrual cycles of days duration, evidence of normal ovulation, and no history of miscarriage. Women who had used any form of hormonal contraception or intrauterine device within the previous 3 months were excluded, as were any women with laparoscopic or ultrasonographic evidence of polycystic ovaries or endometriosis. Biopsy samples were dated according to the time of the last menstrual period and divided into proliferative (days 3 13; n 14), early secretory (days 14 21; n 9), and late secretory (days 22 30; n 18) phases. All culture media and chemicals were obtained from Sigma-Aldrich (Poole, Dorset, United Kingdom) unless otherwise stated. All plastics were obtained from Bibby Sterilin (Stone, Staffordshire, United Kingdom). Cell Culture Endometrial epithelial cells were prepared from endometrial biopsy material as described elsewhere (8, 9). The tissue was cut into small pieces and suspended in Dulbecco s modified Eagle medium (DMEM) containing 0.2% collagenase (type 1A) (DMEMC) and incubated at 37 C for 1 hour. The tissue suspension was gently pipetted every 30 minutes to disperse the cells. The epithelial cells, predominantly present as glands at this stage, were separated from the stromal cells by centrifugation at 100 g for 5 minutes. The pelleted epithelial cells were resuspended in 5 ml of DMEMC and incubated for another 1 hour. Epithelial cells were separated from stromal cells by centrifugation as before. The epithelial pellet was then resuspended in 2 ml of DMEM containing 10% fetal calf serum, 4 mmol/l glutamine, 100 g/ml streptomycin, and penicillin (CDMEM) and layered over 8 ml of CDMEM. The epithelial cells were allowed to sediment for 30 minutes, after which the bottom 2 ml of CDMEM containing the separated epithelial cells was retained for culture. Effect of Androgens on Glycodelin Secretion and Cell Proliferation Epithelial cells were grown in CDMEM in 96-well plates at cells/well at 37 C in 5% CO 2. When the cells were confluent (48 72 hours), the media in replicate (10 15) wells was replaced with media containing either no additions or androstenedione (10 biopsy samples), testosterone (16 samples), DHEA (4 samples), or DHT (5 samples) ( M). 3 H-thymidine (37 kbq/well) (Amersham Life Sciences, Little Chalfont, Buckinghamshire, United Kingdom) in CDMEM was added to all wells, and the cells were incubated for a further 72 hours. After incubation, the culture supernatants were removed and stored at 20 C for glycodelin A assay. The cells were washed in phosphate-buffered saline (PBS), incubated in cell-dissociation solution at 37 C for 15 minutes, and harvested onto filter paper by using a Titertek cell harvester (Flow Laboratories, Paisley, Scotland, U.K.). Cells corresponding to each well were placed in a scintillation tube with 2 ml of scintillation fluid (Wallac, Milton Keynes, Buckinghamshire, United Kingdom) and the tritium content was counted by using a -counter. To verify that the effects seen were due to androstenedione, cells prepared from an additional nine biopsy samples were incubated with androstenedione (10 6 and 10 8 M) in the presence and absence of the androgen antagonist cyproterone acetate (10 7 M). As previously, incubations contained 37 kbq 3 H-thymidine per well and the cells were incubated for 72 hours. The media was removed and stored at 20 C; the cells were harvested and the tritium content was determined as described above. Cell Culture for Immunocytochemical Analysis For some biopsy samples, additional cells were grown on coverslips and were used for immunocytochemical analysis of the androgen receptor. After 2 or 5 days in culture, coverslips were attached to a microscope slide by using petroleum jelly (two coverslips per slide), washed briefly in PBS, and fixed in 3.7% w/v formaldehyde in PBS for 15 minutes. Slides were then washed twice for 5 minutes in PBS and incubated in methanol ( 20 C) for 4 minutes and acetone ( 20 C) for 2 minutes. After two further washes in PBS (5 minutes each), slides were stored at 20 C in sugar storage solution (0.5 M sucrose in 50:50 v/v PBS and glycerol containing 7 mm MgCl 2 ) until immunocytochemical analysis. Cell Culture for Immunocytochemical Analysis of Ki67 Endometrial epithelial cells were also grown in eight-well chamber slides (Nunc; Life Technologies, Paisley, Scotland, United Kingdom) at cells/ml in CDMEM. At confluency (48 72 hours), the media was removed and replaced with media containing either no additions or andro- 772 Tuckerman et al. Androgens and endometrial function Vol. 74, No. 4, October 2000

3 stenedione, 10 6 M; cyproterone acetate, 10 7 M; or androstenedione, 10 6 M, plus cyproterone acetate, 10 7 M (two chambers per treatment). Slides were fixed and stored as for cells grown on coverslips. Glycodelin A Assay Glycodelin A was measured by radioimmunoassay using a method described elsewhere (5 8). Glycodelin A purified from term placenta was used as a standard. Glycodelin A was iodinated by using the chloramine-t method, and the resulting tracer was purified through a Con-A sepharose column. The polyclonal antibody used in the assay was raised in rabbits against purified glycodelin A. For the assay, 100 L of standards or sample were incubated at room temperature for 24 hours with 100 L antiserum, at a dilution to bind 45% of the added tracer. The antibody-bound tracer was separated from the unbound by using Amerlex MMT magnetic separating reagent (Amersham Life Sciences). The sensitivity of the assay was 3 ng/ml, and the intraassay and interassay coefficients of variation were less than 10%. Immunocytochemistry for Cytokeratin, Vimentin, and CD45 Immunocytochemistry using anticytokeratin, antivimentin, or anti-cd45 antibodies was performed to characterize the cultured cells. The cells were washed in Tris-buffered saline (ph 7.6) (TBS) for 5 minutes. The cells were incubated with anticytokeratin (Dako, Ely, Cambridgeshire, United Kingdom), antivimentin (Dako) (30 L of a 1:400 dilution in TBS) or anti-cd45 (Dako) (30 L of a 1:100 dilution in TBS containing 1% human serum) for 2 hours at room temperature. After this and all subsequent incubations, the cells were washed twice in TBS for 5 minutes. Cells were then incubated for 30 minutes with rabbit anti mouse IgG (30 L of a 1:20 dilution in TBS; Dako). After washing, the cells were incubated with alkaline phosphatase anti-alkaline phosphatase complex (30 L of a 1:20 dilution in TBS; Dako) for 30 minutes at room temperature, followed by another wash. The incubations with rabbit anti mouse IgG and alkaline phosphatase anti-alkaline phosphatase complex were repeated to amplify the reaction. Finally, the cells were incubated with the enzyme substrate Fast Red TR (1 mg/ml) for 15 minutes. The cells were then washed in running water and counterstained with hematoxylin (Shannon Ltd., Rotherham, United Kingdom). Immunocytochemistry for Androgen Receptor Cells grown on coverslips were washed twice in TBS and incubated in 3% hydrogen peroxide in methanol for 10 minutes to remove endogenous peroxidase activity. Cells were washed twice in TBS and blocked with 10% normal rabbit serum in TBS (TBSR) for 30 minutes. They were then incubated for 48 hours at 4 C with monoclonal anti human androgen receptor antibody NCL-AR-2F12 (Vector, Peterborough, United Kingdom) (diluted 1:25) or monoclonal anti human androgen receptor SC-7305 (Insight Biotechnology, Wembley, London, United Kingdom) (diluted 1:100) in TBSR. After washing, slides were incubated for 30 minutes in rabbit anti mouse IgG (1:25 in TBSR) at room temperature. This was followed by washing and incubation for 40 minutes with monoclonal mouse peroxidase antiperoxidase (Dako) (diluted 1:100 in TBS). Binding was amplified by repeating the incubations with rabbit anti mouse IgG and monoclonal mouse peroxidase antiperoxidase. Bound antibody was visualized by incubation with 3.3 -diaminobenzidene tetrachloride (DAB substrate solution; Vector) for 8 minutes. Cells were then washed in distilled water for 5 minutes and counterstained with hematoxylin for 10 minutes. Slides were washed in tap water, dehydrated through the alcohols, cleared in Histoclear (Farenheit, Rotherham, United Kingdom) and mounted in DPX. LnCap cells derived from human prostate carcinoma (ATCC designation CRL 1740) were used as a positive tissue control and a monoclonal mouse IgG 1 antibody (Dako control antibody) as a negative reagent control. Immunocytochemistry for Ki67 Staining for Ki67 was done by using the same procedure as for the androgen receptor except that the primary antibody was monoclonal mouse anti-ki67 (Vector) diluted 1:100 in TBSR. A subjective semiquantitative method was used to measure the number of cells that stained positive for Ki67. For each chamber, the number of positive and negative cells in 10 fields (magnification, 400) were counted independently by two observers and the percentage of positive cells per field was calculated. Statistical Analysis To assess the difference in glycodelin A production and 3 H-thymidine uptake in response to each androgen, the mean ( SE) was calculated for each set of replicated wells of cells prepared from each biopsy sample. To compare the results from one biopsy to another, the mean control value per plate was normalized to 100%, and the mean value per androgen concentration was expressed as a ratio of the control value. A one-tailed t-test was used to assess the overall effects of each concentration of androgen in cells prepared from multiple biopsies. The Student t-test was used in a similar manner to assess significant differences in the percent of positive nuclei staining for Ki67 for each test chamber compared with a control chamber. A P value.05 was considered significant. RESULTS Characterization of Cultured Endometrial Epithelial Cells Figure 1 shows immunocytochemical staining of human endometrial epithelial cells for cytokeratin (epithelial cell marker) after 5 days in culture. Intense red staining for FERTILITY & STERILITY 773

4 cytokeratin was seen in all the cultured cells. In contrast, none of the cells stained for vimentin or CD45 (data not shown). These results show that our epithelial cell cultures were essentially pure and were free of stromal or leukocyte contamination. Expression of Androgen Receptors in Cultured Cells Figure 1 shows immunocytochemical staining of androgen receptor in human endometrial epithelial cells after 2 days in culture. Similar staining was obtained in cells cultured for 5 days. Staining was carried out using two different antibodies. Positive brown staining in the epithelial cell nuclei was observed after incubation with both NCL-AR- 2F12 and SC-7305 androgen receptor antibodies. Incubation with SC-7305 resulted in only the nuclei staining positive, whereas with NCL-AR-2F12, positive cytoplasmic staining was also observed. The results with both antibodies show that the cells express androgen receptor and therefore should be capable of responding to androgens. Basal Glycodelin A Production Basal glycodelin A production by cells prepared from endometrial tissue obtained during the proliferative, early secretory, and late secretory phases of the menstrual cycle was measured. Significantly more glycodelin A was produced by cells prepared from late secretory endometrium ( ng/ml) than by cells prepared from early secretory endometrium (69 16 ng/ml) (P.001), and more glycodelin A was produced by cells prepared from late secretory endometrium than by cells prepared from proliferative endometrium ( ng/ml) (P.05). Effect of Androstenedione on Endometrial Epithelial Cell Growth and Glycodelin A Production Figure 2 shows the effect of androstenedione on uptake of 3 H-thymidine and secretion of glycodelin A by human endometrial epithelial cells. Androstenedione caused a dosedependent decrease in glycodelin A production, both when results from a single experiment and when the pooled results were considered. When results from all biopsies were considered, significant decreases were seen at androstenedione concentrations of 10 7 M(P.05) and 10 6 M(P.01) (Table 1). A similar decrease in glycodelin A production was seen in cells prepared from biopsy samples obtained from proliferative or secretory phase endometrium. Androstenedione also caused a dose-dependent decrease in 3 H-thymidine uptake by human endometrial epithelial cells. As for glycodelin A production, significant decreases were seen at androstenedione concentrations of 10 7 M (P.05) and 10 6 M(P.01). A similar decrease in 3 H- thymidine uptake was seen in cells prepared from biopsy samples obtained during either the proliferative or secretory phase of the cycle. TABLE 1 Effect of androstenedione on glycodelin A production and 3 H-thymidine uptake by human endometrial epithelial cells. Treatment Glycodelin A production (%) 3 H-Thymidine uptake (%) Control Androstenedione, 10 8 M Androstenedione, 10 7 M a 81 7 a Androstenedione, 10 6 M b b Note: Values are expressed as means ( SEM). Cells were prepared from 10 biopsy samples obtained throughout the cycle. a P.05 compared with controls. b P.01 compared with controls. Effect of Testosterone, DHEA, and DHT on Endometrial Epithelial Cell Growth and Glycodelin A Production Addition of testosterone, DHEA, or DHT to cultured endometrial epithelial cells had no effect on uptake of 3 H- thymidine or production of glycodelin A by these cells. Effect of Cyproterone Acetate on Androstenedione-Stimulated Epithelial Cell Growth and Glycodelin Production To confirm the specificity of the results obtained with androstenedione, epithelial cells prepared from nine additional endometrial biopsy samples obtained throughout the menstrual cycle were incubated with androstenedione in the presence and absence of the androgen antagonist cyproterone acetate. Figure 3 shows that as in the original experiments, androstenedione caused a dose-dependent decrease TABLE 2 Effect of androstenedione and cyproterone acetate (10 7 M) on glycodeline A production and 3 H-thymidine uptake. Treatment Glycodeline A production (%) 3 H-Thymidine update (%) Control Androstenedione, 10 8 M a b Androstenedione, 10 6 M c b Cyproterone acetate Androstenedione 10 8 M, plus cyproterone acetate Androstenedione, 10 6 M, plus cyproterone acetate Note: Values are expressed as means ( SEM). Cells were prepared from 9 biopsy samples obtained throughout the cycle. a P.05 compared with controls. b P.001 compared with controls. c P.01 compared with controls. 774 Tuckerman et al. Androgens and endometrial function Vol. 74, No. 4, October 2000

5 FIGURE 1 Immunostaining for (A) cytokeratin (original magnification, 190) and (B) androgen receptor (original magnification, 760) in endometrial epithelial cells after 2 days in culture. in glycodelin A production and 3 H-thymidine uptake. When the results from all biopsies were considered together, significant decreases in both values were seen at 10 8 (P.05 and.001) and 10 6 M(P.01 and.001). However, in the presence of cyproterone acetate (10 7 M), no inhibitory effect of androstendione was seen (Table 2). The effect of androstenedione and cyproterone acetate on cell growth was also assessed by immunocytochemical anal- FERTILITY & STERILITY 775

6 FIGURE 2 Effect of androstenedione on (A) glycodelin A production and (B) 3 H-thymidine uptake by human endometrial epithelial cells in culture. Graphs show representative data from a single biopsy; see Table 1 for data pooled from all 10 biopsy samples obtained throughout the cycle. Values are means SEM (n 5). *P.05 compared with controls; **P.01 compared with controls; ***P.001 compared with controls. ysis of the cell proliferation marker Ki67. In these experiments, the mean number of nuclei in 10 fields that stained positive for Ki67 was 75% 9.3% (P.05) of controls in the presence of 10 6 M androstenedione, 105% 12.3% of control values with cyproterone acetate alone, and 100% 8.8% of the control value with androstenedione plus cyproterone acetate. These results confirmed that cyproterone acetate antagonized the effect of androstenedione on cell growth. DISCUSSION An association between increased androgen levels and adverse reproductive outcome was reported in several recent studies (2 4). The expression of androgen receptors in the human endometrium has also been demonstrated. Positive immunostaining of the androgen receptor has been described in cells of both the stromal and epithelial compartments through the menstrual cycle (10, 11), with more intense 776 Tuckerman et al. Androgens and endometrial function Vol. 74, No. 4, October 2000

7 FIGURE 3 Effect of cyproterone acetate (10 7 M) and androstenedione on (A) glycodelin A production and (B) 3 H-thymidine uptake by human endometrial epithelial cells in culture. Graphs show representative data from a single biopsy; see Table 2 for data pooled from 9 biopsy samples obtained throughout the cycle. Values are means SE (n 5). *P.05 compared with controls (Con); ***P.001 compared with controls; #P.05 compared with androstenedione (A) alone; ###P.001 compared with androstenedione alone. CA cyproterone acetate. staining in the stroma than in the epithelium. The presence of androgen receptors within the normal human endometrium and the observation that androgens have been shown to increase prolactin secretion by human stromal cells in vitro (12) suggest that androgens may directly affect the endometrium. A recent report from our laboratory have shown a negative correlation between androstenedione and testosterone levels in the proliferative phase of the menstrual cycle and levels of PP14 (glycodelin A) in uterine flushings collected in the secretory phase of the cycle, providing further indirect evidence of an effect of androgens on endometrial secretory activity (1). In this study, we tested the hypothesis that androgens FERTILITY & STERILITY 777

8 have a direct adverse effect on endometrial cell function. We conducted an in vitro study to examine the impact of three different doses of four androgens on endometrial epithelial cell growth and secretory activity by using an established endometrial epithelial cell culture system (8, 9). Immunocytochemical staining of the cells by using anticytokeratin (an epithelial cell marker), antivimentin (a stromal cell marker) and anti-cd45 (a leukocyte marker) agreed with previous reports showing that after 5 days in culture, the epithelial cells are essentially free of stromal cell and leukocyte contamination (9, 13). In vitro cell culture studies such as ours often attract criticism based on the fact that the environment is not a physiologic one. The various receptors present may change after the cells have been kept in culture for some time. Moreover, the cellular response of epithelial cells may change in the absence of stromal cells. However, when properly conducted, cell culture experiments permit examination of the action of a specific hormone or its antagonist to provide direction for in vivo studies. In our study, secretion of glycodelin A was used as an indicator of endometrial cell secretory activity. Glycodelin A is a major secretory protein of the epithelial endometrium in vivo, and previous studies have shown that measurement of this substance is a reliable marker for endometrial function (6, 7, 14, 15). Our previous work suggested that the level of glycodelin A secretion by human endometrial epithelial cells cultured by the method used in the current study is dependent on the time in the cycle at which the endometrial biopsy is obtained (8). In the current study, basal glycodelin A production by cultured epithelial cells showed a similar pattern of secretion as previously described, with more glycodelin A produced by cells prepared from late secretory endometrium than by cells prepared from proliferative or early secretory endometrium. An additional finding was that more glycodelin A was produced by epithelial cells from proliferative endometrium than was produced by cells from early secretory endometrium. This observation seems to agree with previous clinical observations that serum levels of glycodelin are lowest around the periovulatory phase (including the early proliferative phase) (14) and that the concentration of glycodelin in uterine flushings is negligible in the early proliferative phase (9). Our finding that androgens have a direct inhibitory effect on endometrial epithelial cells are strengthened by two further observations. First, immunohistochemical staining of endometrial epithelial cells showed positive nuclear staining for the androgen receptor in the cells at the time of androgen addition. No reports of androgen receptor expression in cultured human endometrial epithelial cells have been published, although its expression in cultured human endometrial stromal cells has been documented (12, 16). The presence of the androgen receptor in cultured human endometrial epithelial cells is consistent with the hypothesis that androgens have a direct action on these cells. Second, the inhibitory effects of androstenedione were reversed when the androgen receptor antagonist cyproterone acetate was added along with androstenedione to the culture medium. Two antiandrogens are currently in clinical use: cyproterone acetate (Schering Health, Burgess, West Sussex, United Kingdom) and flutamide (Boehringer Ingelheim, Bracknell, Berkshire, United Kingdom). Although some have claimed that flutamide is a more specific antiandrogen, we chose to use cyproterone acetate because it is more often used in clinical practice in the United Kingdom in the treatment of hirsutism. Cyproterone acetate is known to have some progestogenic effects; however, these effects are unlikely to have influenced our study results, as addition of cyproterone acetate reversed both the proliferative and secretory activities of the endometrial epithelial cells. Progestogens would be expected to enhance secretory activity but inhibit proliferative activity. Uptake of 3 H-thymidine is a traditional method for measurement of cell proliferation, although its accuracy in some tissue culture systems has been questioned (17). In particular, it has been suggested that during extended incubation times, the cells might degrade labeled thymidine and spread the label in its general metabolism. The relation between uptake of 3 H-thymidine and cell proliferation in the tissue culture system used in this study was confirmed by immunostaining with an antibody to Ki67, an antigen that is present in the nucleolus during late G1, S, G2, and M phases. Because a similar result was obtained by both methods, it is likely that 3 H-thymidine uptake is a true measurement of cell proliferation. The concentrations of androstenedione that we used are substantially higher than the physiologic concentration of androgens in serum, which are in the nmol/l (10 9 mol/l) range. However, other studies on the effects of androgens on cultured endometrial stromal cells have reported effects of androgens on EGF receptor expression and prolactin production only at concentrations similar to or higher than those used in our study (12, 18). The need for increased concentrations of steroid in vitro may be due to delivery of steroid. In vitro steroids are delivered as a single pulse rather than continuously supplied through the blood. In vitro steroids may be sequestered by binding proteins present in fetal calf serum and may also be broken down by the cells in culture to inactive metabolites. These metabolites are not removed, nor is the steroid supply replenished, as occurs in vivo. Our finding of dose-dependent effects and suppression of these effects by the androgen receptor antagonist cyproterone acetate supports the view that androstenedione can inhibit the growth and secretory activity of human endometrial epithelial cells. Although our findings support the hypothesis that andro- 778 Tuckerman et al. Androgens and endometrial function Vol. 74, No. 4, October 2000

9 gens have a direct inhibitory effect on endometrial epithelial cell function, we were unable to confirm significant effects of testosterone and DHT. The latter finding is contrary to expectations, as it is generally accepted that testosterone and DHT are more potent androgens. One possible explanation is the differential bioavailability of the different androgens in the in vitro system. For example, steroid binding by proteins in fetal calf serum (which was used in our system) could preferentially reduce the uptake of testosterone and DHT. To explore such a possibility requires a separate experiment using defined, growth factor supplemented media in the absence of serum. Another possible explanation relates to the intrinsic weakness of the in vitro system when only epithelial cells were cultured, because it does not take into consideration the possible interaction of epithelial and stromal cells as occurs in vivo. In a previous study in which epithelial cells and stromal cells were cultured together as in a whole biopsy, the effects of testosterone were observed to be more profound and it inhibited the uptake of 3 H-thymidine (19). On the other hand, two recent clinical reports suggested that recurrent miscarriage is associated with high follicular phase levels of circulating androstenedione but not testosterone (1, 4). In conclusion, we show that androstenedione can inhibit human endometrial epithelial cell growth and secretory activity in vitro. Our findings are consistent with the hypothesis that adverse reproductive outcome in women with hyperandrogenemia may be due to a direct detrimental effect of androgens on the endometrium. To confirm these in vitro findings, an in vivo study of endometrium after administration of defined doses of androgens in artificial cycles of women with nonfunctioning ovaries is required. Because there is little information on endometrial function of women with hyperandrogenemia, such as those with the polycystic ovary syndrome, studies should be undertaken in the near future. Acknowledgements: The authors thank staff nurse Barbara Anstie for help in recruiting patients; Dr. Elizabeth Anderson and Angela Cramer, Tumour Biochemistry Laboratory, Christie Hospital, Manchester, United Kingdom, for invaluable advice on immunostaining; Dr. D. Grey, Statistical Services Unit, Sheffield University, for statistical advice; and the Special Trustees for the Former United Sheffield Hospitals Charitable Funds for financial support. References 1. Okon MA, Laird SM, Tuckerman EM, Li TC. Serum androgen levels in women who suffer recurrent miscarriage and their correlation with markers of endometrial function. Fertil Steril 1998;69: Tulppala M, Stenman UH, Cacciatore B, Ylikorkala O. Polycystic ovaries and levels of gonadotrophins and androgens in recurrent miscarriage: prospective study in 50 women. Br J Obstet Gynaecol 1993; 100: Homburg R, Eshel A, Kilborn J. Combined luteinizing hormone releasing hormone analogue and exogenous gonadotrophins for the treatment of infertility associated with polycystic ovaries. Hum Reprod 1990;5: Bussen S, Sütterlin M, Steck T. Endocrine abnormalities during the follicular phase in women with recurrent spontaneous abortion. Hum Reprod 1999;14: Dalton CF, Laird SM, Serle E, Saravelos H, Warren MA, Li TC, et al. The measurement of CA 125 and placental protein 14 in uterine flushings in women with recurrent miscarriage: relation to endometrial morphology. Hum Reprod 1995;10: Mackenna A, Li TC, Dalton C, Bolton AE, Cooke ID. Placental protein 14 levels in uterine flushing and plasma of women with unexplained infertility. Fertil Steril 1993;59: Li TC, Dalton C, Hunjan KS, Bolton AE, Warren MA. The correlation of endometrial PP14 concentrations in uterine flushing and endometrial morphology in the peri-implantation period. Hum Reprod 1993;8: Laird SM, Li TC, Bolton AE. The production of placental protein 14 and interleukin 6 by human endometrial cells in culture. Hum Reprod 1993;8: Laird SM, Tuckerman E, Li TC, Bolton AE. Stimulation of endometrial epithelial cell interleukin 6 (IL6) production by interleukin 1 and placental protein 14. Hum Reprod 1994;9: Horie K, Takakura KK, Imai K, Liao S, Mori T. Immunohistochemical localisation of androgen receptor in the human endometrium, decidua, and placenta and pathological conditions of the endometrium. Hum Reprod 1992;7: Mertens HJ, Heineman MJ, Koudstaal J, Theunissen P, Evers JL. Androgen receptor content in human endometrium. Eur J Obstet Gynaecol Reprod Biol 1996;70: Narukawa S, Kanzaki H, Inoue T, Imai K, Higuchi T, Hatayama H, et al. Androgens induce prolactin production by human endometrial stromal cells in vitro. J Clin Endocrinol Metab 1994;78: Laird SM, Tuckerman EM, Li T C, Saravelos H. The production of tumour necrosis factor alpha (TNF-alpha) by human endometrial cells in culture. Hum Reprod 1996;11: Seppala M, Koistinen R, Rutanen EM. Uterine endocrinology and paracrinology: insulin growth factor binding protein-1 and placental protein 14 revisited. Hum Reprod 1994;9: Klentzeris LD, Bulmer JN, Seppala M, Li TC, Warren MA, Cooke ID. Placental protein 14 in cycles with normal and retarded endometrial differentiation. Hum Reprod 1994;9: Iwai M, Kanzaki H, Fujimoto M, Kojima K, Hatayama H, Inoue T, et al. Regulation of sex steroid receptor gene expression by progesterone and testosterone in cultured human endometrial stromal cells. J Clin Endocrinol Metab 1995;80: Xiao CW, Goff AK. Differential effects of oestradiol and progesterone on proliferation and morphology of cultured bovine uterine epithelial and stromal cells. J Reprod Fertil 1998;112: Watson H, Franks S, Bonney RC. Regulation of epidermal growth factor receptor by androgens in human endometrial cells in culture. Hum Reprod 1998;13; Rose GL, White JO, Dowsett M. The inhibitory effects of danazol, danazol metabolites, gestrinone, and testosterone on the growth of human endometrial cells in vitro. Fertil Steril 1998;49: FERTILITY & STERILITY 779