Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Estrogens are some of the most commonly prescribed pharmacological agents, being used most frequently for hormone replacement therapy of postmenopausal women and as components of combination oral contraceptives, although they are also used for a variety of other purposes (Williams and Stancel, 1996). Most of the pharmacological actions of estrogens are thought to result from interactions with the classic nuclear estrogen receptor (ER), which is a ligand-activated transcription factor. The ER-ligand complex interacts with DNA binding sites, termed estrogen response elements (EREs), in target genes, and this complex recruits coactivators (or corepressors) and other regulatory proteins that form the active transcription complex (Shibata et al., 1997; White and Parker, 1998). A large number of steroidal and nonsteroidal compounds bind to this receptor and display hormone-like activity in humans and a number of experimental test systems (Anstead et al., 1997).

There are substantial differences in the potencies of different estrogens that result primarily from differences in their receptor-binding affinities and their pharmacokinetic properties, including absorption, first-pass metabolism, plasma protein binding, and elimination (Williams and Stancel, 1996). However, it was thought until recently that the response to all estrogens was basically the same if these differences in receptor affinities and pharmacokinetics were accounted for. All estrogens were thought to produce their responses by a similar activation of the ER, which led to the same basic pattern of gene expression in target tissues. Conversely, competitive hormone antagonists were thought to produce their effects by blocking hormone binding to the receptor and/or by favoring the interaction of the ER-ligand complex with factors that repress rather than activate transcription. In most early models of estrogen action, the receptor was thus viewed as an on/off switch that was activated by agonist binding, with any pharmacological differences between various estrogens due primarily to differences in potencies (Stancel et al., 1995; Katzenellenbogen et al., 1996; McDonnell and Norris, 1997).

Recently, results from a number of studies have called this unitary view of estrogen action into question (Katzenellenbogen et al., 1996; McDonnell and Norris; 1997; Stancel et al., 1995). At the biochemical level, protease digestion (McDonnell et al., 1995) and crystallographic studies (Brzozowski et al., 1997) have clearly established that the ER assumes very different conformations when occupied by different ligands, and functional studies have revealed that the pattern of gene expression produced by estrogens is both gene and context specific. For example, different ligands can produce distinct patterns of gene expression in cultured cells or in target tissues in vivo. The emerging paradigm of estrogen action is that different ligands cause the ER to assume different conformations that may selectively interact with different coactivators/corepressors, EREs, or other regulatory factors (Stancel et al., 1995; Katzenellenbogen et al., 1996; McDonnell and Norris, 1997). Thus, it can no longer be assumed that different estrogens produce identical responses in target tissues, especially considering that overall tissue responses to these hormones are likely to involve a primary activation of multiple genes.

These considerations prompted us to consider whether commonly used estrogenic drugs produce the same patterns of gene expression in target tissues as endogenous ovarian hormones. On searching the literature, we did not uncover any reports that directly addressed this question, which prompted us to compare the in vivo responses of the uterus to 17-ethinyl estradiol (17-EE) and 17-estradiol (17-E₂). We selected 17-EE for this study because it is the most frequently used synthetic estrogen (Williams and Stancel, 1996). The genes we selected to compare the response of the two estrogenic compounds are c-fos (Loose-Mitchell et al., 1988), c-jun (Chiappetta et al., 1992; Nephew et al., 1994), vascular endothelial growth factor (VEGF) (Hyder et al., 1996), and creatine kinase B (CKB) (Pentecost et al., 1990). These genes were chosen because they are well characterized primary responses of the normal uterus to estrogenic stimulation. This allowed us to compare the dose-response patterns of endogenous genes produced by the synthetic and naturally occurring estrogens in a physiologically relevant test system.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Animals. Immature (21 days old, 40-45 g) female Sprague-Dawley rats (Sasco, Omaha, NE) were ovariectomized under Chloropent anesthesia (Sigma, St. Louis, MO) 3 to 7 days before use. 17-E₂ and 17-EE (dissolved in 0.5 ml 5% ethanol/95% saline) were administered via s.c. injection at a dose of 40 µg/kg (except for the dose-response studies shown in Figs. 5 and 6). At the times indicated for individual experiments, uteri were removed under Chloropent anesthesia, and the animals were then sacrificed by decapitation while still anesthetized. Protocols for the care and use of these animals were approved by the University of Texas Animal Care and Use Committee, in accordance with the Guiding Principles for the Care and Use of Research Animals.

Materials. Estradiol was obtained from Steraloids (Wilton, NH) and 17-EE was purchased from Sigma. [³²P]UTP (800 Ci/mmol) was obtained from Amersham Radiochemicals (Arlington Heights, IL) and diluted to 400 Ci/mmol with radioinert UTP (Boehringer Mannheim, Indianapolis, IN) for synthesis of riboprobes. Guanidine isothiocyanate, CsCl, and formamide were obtained from International Biotechnologies (New Haven, CT). All other chemicals were obtained from Sigma and were the highest grade commercially available.

RNA Preparation and Analysis. RNA was prepared as described previously (Loose-Mitchell et al., 1988; Chiappetta et al., 1992). Briefly, uteri were removed from anesthetized animals and immediately homogenized in 5 M guanidinium isothiocyanate with a Polytron homogenizer set at half-maximal power for 60 s. Uteri from two or three animals were pooled for the preparation of a single RNA sample. RNA was pelleted through 5.7 M CsCl, extracted twice with phenol/chloroform (1:1), once with chloroform, and precipitated with ethanol.

In Situ Hybridization. In situ hybridization was performed with digoxigenin labeling kits obtained from Boehringer Mannheim, following the manufacturer's directions for the synthesis of digoxigenin riboprobes, with the c-fos (Boettger-Tong and Stancel, 1995) and VEGF (Hyder et al., 1996) sequences we have recently reported. Uteri were removed, fixed in 4% paraformaldehyde, embedded in paraffin, and used to prepare 5-µm sections that were mounted on subbed slides (Probe-On Plus from Fisher Scientific, Pittsburgh, PA), as described by Breitschopf and Suchanek (1996), and optimized for the VEGF and c-fos probes. Briefly, slides were dewaxed in xylene (2 × 10 min) followed by a graded series of 100 and 70% ethanol solutions (2 × 10 min each), rinsed in diethylpyrocarbonate water (1 × 5 min), and fixed again in 4% paraformaldehyde for 20 min. The slides were then rinsed in PBS, treated with 0.2 M HCl for 10 min, rinsed in PBS, and treated with 10 µg/ml proteinase K in Tris-EDTA buffer (pH 7.5) for 30 min at 37°C. Slides were then washed with Tris-buffered saline (pH 7.5), treated with 0.1 M triethanolamine for 5 min, and dipped in triethanolamine containing 0.3% acetic anhydride with gentle agitation. After a final rinse in Tris-buffered saline, the slides were dehydrated in 70% ethanol for 5 min followed by 100% ethanol (3 × 5 min) and dried at 40°C for 15 min.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Figure 1A illustrates the structures of 17-E₂, the major endogenous estrogen produced by the ovary, and 17-EE, which is the estrogen used most commonly in oral contraceptives. The structure of mestranol, the 3-O-methyl derivative of 17-EE, is also illustrated. Mestranol itself does not have estrogenic activity, but it is converted to 17-EE in vivo by hepatic biotransformation (Bolt, 1979). Mestranol was the first estrogen used in oral contraceptives and is still used as the estrogenic component of some combination oral contraceptives. The 17-ethinyl substitution on these drugs decreases their hepatic first-pass metabolism, which increases their oral bioavailability (Bolt, 1979). Figure 1B is a tabulation of the four estrogen-responsive genes we examined in this study and the cell types of the rodent uterus in which they are primarily expressed after estrogen treatment.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Structures of estrogens (A) and cellular localization of estrogen-induced transcripts (B) in the uterus.

Before determining the dose-response curves for the induction of the four genes under study, it was important to define the time course for their induction in the uterus after administration of 17-EE or 17-E₂ to immature animals. Figure 2 illustrates representative results for VEGF and c-fos as assessed by Northern blot analysis. In both cases, transcript levels are very low in untreated animals, and induction is rapid after either treatment, with large increases in mRNA levels seen by 1 h. Transcript levels then remain elevated up to 3 h after administration of either estrogen and then return to baseline values after 12 to 24 h. Northern blot analysis was also performed on the same RNA samples, and the resultant blots were probed as described in the Experimental Procedures to analyze the time course for induction for CKB and c-jun transcripts by both estrogens (data not shown).

View larger version (53K): [in this window] [in a new window]   Fig. 2.   Time course of c-fos (top) and VEGF (bottom) induction by 17-EE and 17-E2 in the uterus. Groups of animals were treated with either compound (40 µg/kg) for the indicated times before sacrifice. RNA was then prepared as described under Experimental Procedures, and aliquots of the same RNA preparations were used for blot analyses of c-fos or VEGF. Each lane represents a separate determination with an RNA sample prepared from two or three pooled uteri. Arrows indicate the position of the 2.2-kb c-fos (top) and 3.7-kb VEGF (bottom) transcripts, and the 28S ribosomal RNA band used as a loading control.

The results from all four sets of blots were quantitated by laser densitometry, and the intensity of the bands for each transcript was normalized to the intensity of the 28S ribosomal band in the same lane as a loading control. Note that the induction of VEGF is maximal at 1 h, whereas the induction of the three other genes is maximal at 3 h. The reason for this more rapid induction of VEGF is unknown, but our results are similar to those seen in previous studies on the induction of VEGF (Hyder et al., 1996), CKB (Pentecost et al., 1990), and c-jun and c-fos (Chiappetta et al., 1992) by estradiol. The data shown in Fig. 3 establish that 17-EE also shows the same temporal patterns of gene expression as that seen after 17-E₂.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Temporal patterns of uterine gene expression after administration of 17-EE or 17-E2. Animals were treated with either 40 µg/kg of 17-E2 () or 17-EE () for the indicated periods. Uterine RNA was then prepared and analyzed for expression of c-fos, VEGF, CKB, or c-jun by blot analysis as described in Experimental Procedures. The intensity of each transcript band was then determined by laser densitometry and normalized for the level of 28S ribosomal RNA in the same lane. The data for each transcript/treatment group are presented as the percentage of the maximum level observed at 3 h for c-fos, CKB, and c-jun or at 1 h for VEGF, run on the same blot. Each point represents the mean of three determinations.

We and others have previously shown that expression of each of the genes under study displays a cell-specific pattern of induction in the uterus after 17-E₂ administration. For example, induction of c-fos by the endogenous hormone occurs primarily in epithelial cells (Bigsby and Li, 1994; Boettger-Tong and Stancel, 1995), whereas expression of VEGF is highly restricted to the stromal layer (Hyder et al., 1996). Because the RNA used for the Northern blots shown in Fig. 2 was prepared from the whole uterus, we next examined the induction of several transcripts by in situ hybridization to ensure that 17-EE and 17-E₂ were inducing the genes in the same uterine cell types.

Figures 4 and 5 represent a set of in situ hybridization studies illustrating the expression of c-fos and VEGF transcripts in uterine sections after administration of 17-EE or 17-E₂ for 3 h before sacrifice. As we have previously observed for c-fos (Boettger-Tong and Stancel, 1995) and VEGF (Hyder et al., 1996) and as expected on the basis of the Northern blots observed in Fig. 2, no signal was observed for either transcript in sections from vehicle-treated control animals (data not shown). However, 3 h after treatment with either estrogen, a marked increase in expression of both c-fos and VEGF is observed, as shown in Figs. 4 and 5, respectively. The fos proto-oncogene is induced primarily in epithelial cells by both 17-EE and the endogenous hormone when hybridized with an antisense probe, and very little signal is observed with a sense probe after either treatment (Fig. 4). On the other hand, the sections in Fig. 5 illustrate that both estrogenic ligands stimulate VEGF production only in the stromal layer, with virtually no induction in the epithelium. The VEGF signal after either treatment is also markedly diminished when sections are hybridized with a sense probe for this growth factor. The cellular pattern of expression for the two genes examined is thus similar for the both the synthetic and naturally occurring estrogens.

View larger version (118K): [in this window] [in a new window]   Fig. 4.   Cellular patterns of c-fos expression after 17-EE or 17-E2 administration. Animals were treated with estrogens (40 µg/kg) for 3 h before sacrifice, and preparation of sections for in situ hybridization was performed as described in Experimental Procedures. Sections from animals receiving 17-E2 were probed with sense (A) or antisense (B) c-fos probes. Sections from animals receiving 17-EE were probed with sense (C) or antisense (C) c-fos probes. LE, luminal epithelium; GE, glandular epithelium; S, stroma.

View larger version (107K): [in this window] [in a new window]   Fig. 5.   Cellular patterns of VEGF expression after 17-EE or 17-E2 administration. Animals were treated with estrogens (40 µg/kg) for 3 h before sacrifice, and preparation of sections for in situ hybridization was performed as described in Experimental Procedures. Sections from animals receiving 17-E2 were probed with sense (A) or antisense (B) VEGF probes. Sections from animals receiving 17-EE were probed with sense (C) or antisense (D) VEGF probes. LE, luminal epithelium; S, stroma.

Having established that the time course and tissue patterns of gene expression are similar for 17-EE and 17-E₂, we next investigated the dose-response curves for the four genes. In these experiments, animals were treated with increasing doses of the estrogens for 3 h. Uterine RNA was then prepared, and aliquots of the same RNA preparations were used to determine expression of all the genes under study by Northern blot analysis. Representative Northern blots illustrating the 2.0-kilobase (kb) CKB transcript levels are shown in Fig. 6, along with the 28S ribosomal RNA band for the same blots, which was used as an internal loading control.

View larger version (36K): [in this window] [in a new window]   Fig. 6.   Dose-dependent induction of CKB mRNA by 17-E2 and 17-EE. Animals were treated with the indicated doses of estrogens for 3 h, and uterine RNA was prepared and analyzed for expression of the 2.0-kb CKB transcript as described in Experimental Procedures. Each lane represents a separate determination with an RNA sample prepared from a pool of two or three uteri. Arrows indicate the positions of the 2.0-kb CKB transcripts and the 28S ribosomal RNA bands used as a loading control.

A series of similar Northern blots was performed with hybridization probes for the c-fos, c-jun, and VEGF transcripts. The amount of each transcript was then normalized to the level of 28S RNA in the same sample lane on the blot, and the resultant data are shown in Fig. 7. The transcript level for each dose is plotted as the percentage of the maximum induction for that transcript. Maximum induction in each case was taken as the average of the levels produced by the two highest doses of 17-E₂, which were not statistically different. The absolute level of induction (versus vehicle-treated controls) for the transcripts was 15.1-fold for CKB, 5.5-fold for c-jun, 28-fold for c-fos, and 9.9-fold for VEGF. The dose-response curves for induction of each of the genes by 17-E₂ and 17-EE are virtually identical. Note, however, that both estrogens begin to increase VEGF mRNA levels at a dose of 0.4 µg/kg but do not increase the levels of the other three transcripts until a dose of 4 µg/kg is administered. This suggests that induction of VEGF is a more sensitive marker for estrogen exposure than the other gene products examined.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Dose-response curves for induction of uterine genes by 17-E2 () and 17-EE (). Animals were treated with the indicated doses of estrogens for 3 h before sacrifice. RNA was then prepared as described in Experimental Procedures, and aliquots of each RNA sample were analyzed by Northern blotting for expression of the four indicated transcripts. The level of expression was quantitated by densitometric scanning of blots similar to those shown in Fig. 6. Each point represents the mean (and indicated S.E.) of five or six determinations, except for the 0.004-µg/kg-dose points, which represent the mean of two determinations.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

17-EE was originally synthesized by Inhoffen and Hohlweg (1938) as an orally active estrogen. The 17-substituent renders the D-ring of the steroid resistant to 16-hydroxylation in the liver, and it thus has a higher oral bioavailability and slower elimination than endogenous ovarian estrogens such as 17-E₂ (Bolt, 1979). Mestranol is a synthetic estrogen that lacks intrinsic estrogenic activity but is converted to 17-EE in liver microsomes (Bolt, 1979). 17-EE and mestranol are the estrogenic components of most combination oral contraceptives (Williams and Stancel, 1996), and millions of women worldwide have received 17-EE or its precursor in contraceptives and other preparations. Given the extent of usage of this drug, its pharmacology has been extensively studied in humans and experimental animals.

In cell-free systems with ER from rats, humans, and other sources, it has been established that 17-EE and 17-E₂ have essentially identical affinities for the receptor (see Anstead et al., 1997 and references therein), and more recently it has been shown that the DNA binding properties of the ER liganded with the two compounds are also very similar (Cheskis et al., 1997). In vivo studies have established that administration of 17-EE leads to occupancy of nuclear ERs in both the uterus (Bowden et al., 1986; Tetsuo et al., 1989) and liver of rats (Marr et al., 1980a,b; Aten and Eisenfeld, 1982; Lax et al., 1983). The literature also contains numerous reports that the two compounds elicit qualitatively similar responses, including the expression of target genes and transfected reporters, in a variety of cultured cells and tissues. However, few of these studies compared the dose response curves for the compounds, and to our knowledge none have ever sought to quantitatively compare the transcriptional responses of multiple genes in a target tissue in a physiological setting. This lack of data on patterns of gene expression coupled with the emerging realization that different estrogenic ligands produce distinct conformational states of the ER that are expected to produce different biological responses (Stancel et al., 1995; Katzenellenbogen et al., 1996; McDonnell and Norris, 1997) prompted us to perform the studies reported here.

We selected the rodent uterus for these experiments because it is a well characterized experimental system for the study of estrogen action, and because uterine toxicities (e.g., bleeding, hyperplasia, and endometrial carcinoma) are major concerns that accompany the pharmacological use of estrogens (Williams and Stancel, 1996). Estradiol very rapidly increases mRNA levels of the four genes we have examined and the increases in transcript levels are blocked by inhibitors of RNA synthesis, but NOT inhibitors of protein synthesis, indicating that induction is a primary hormonal response in all cases (Loose-Mitchell et al., 1988; Pentecost et al., 1990; Hyder et al., 1996). Hormone mediated increases in expression of at least three of these genes [c-fos (Hyder et al., 1997), VEGF (Hyder et al., 1997), and CKB (Castro-Rivera and Safe, 1998)] are blocked by the pure anti-estrogen ICI 182,780 indicating that their induction involves the classical ER. This is further supported by the identification of functional estrogen response elements in the c-fos (Hyder et al., 1992), c-jun (Hyder et al., 1995), and CKB (Wu-Peng et al., 1992) genes, and we have recently identified several candidate EREs in the VEGF gene that bind purified ER in band shift assays (Hyder et al., 1999). Collectively these observations suggest that the initial patterns of uterine gene expression produced by 17-EE and 17-E₂ are primary pharmacological responses produced when the nuclear ER is occupied by the two steroids.

We found that the dose-response curves for 17-EE and 17-E₂ are virtually identical (Fig. 7) for induction of all four genes examined, and we established that both compounds induced all four genes with a similar time course. Note that the four genes we have studied are preferentially induced by 17-E₂ in different uterine cell types, c-fos and CKB in the epithelial cell compartment (Bergen et al., 1993; Bigsby and Li, 1994; Boettger-Tong and Stancel, 1995), VEGF in the stromal layer (Hyder et al., 1996), and c-jun and CKB in the smooth muscle cells of the myometrium (Nephew et al., 1994); and we have confirmed that 17-EE induces the same cellular pattern for at least two of the genes examined (Figs. 4 and 5). This indicates that the dose-response curves for the two estrogens are identical in three different uterine cell types, all of which are likely to contain different complements of coactivators, corepressors, and other regulatory proteins because they are highly differentiated. The fact that multiple interactions with these factors are thought to be involved in transcriptional regulation by steroid hormone receptors (Katzenellenbogen et al., 1996; Shibata et al., 1997; White and Parker, 1998) suggests that the conformations of the ER produced by binding 17-EE and 17-E₂ are similar. This is also consistent with recent modeling studies of the ligand-binding pocket of the receptor that suggest the protein contains a flexible subsite capable of accommodating small substituents in the 17-position of the D-ring in estrogenic steroids (Anstead et al., 1997).

Whereas our primary objective was to compare the uterine responses to 17-EE and 17-E₂, note that both compounds induce VEGF somewhat differently than the other three genes studied. The data in Fig. 3 illustrate that VEGF mRNA is induced more rapidly than the other transcripts, and this is consistent with previous reports on induction of the four genes we have studied in this work (Pentecost et al., 1990; Chiappetta et al., 1992; and Hyder et al., 1996). In addition, the data in Fig. 7 illustrate that VEGF induction by both 17-EE and 17-E₂ occurs at a slightly lower dose than induction of the other transcripts. This is the converse of results from a previous study in which we observed that higher doses of the pure antiestrogen ICI 182,780 are required to block the induction of VEGF by estradiol than to block the induction of c-fos by the hormone in the uterus (Hyder et al., 1997). Collectively, these findings indicate that induction of VEGF by estrogens may be a more rapid and sensitive marker of estrogen action than other transcriptional responses.

Taken together, these observations suggest that 17-EE and 17-E₂ are likely to have similar effects on many primary estrogen-responsive genes. This is consistent with studies in experimental animals showing that both compounds produce generally similar effects on many endpoints such as gonadotropin levels, bone density, and uterine weight (Papadaki et al., 1979). Also note that 17-EE induces c-fos and c-jun in the hamster kidney (Li et al., 1998), c-jun in the rat liver (Hallstrom et al., 1996), and CKB in the rat uterus (Bowden et al., 1986), although neither dose-response curves nor comparisons to 17-E₂ were performed in those studies. We are unaware of other published reports demonstrating that 17-EE regulates expression of VEGF.

Whereas 17-EE and 17-E₂ appear to produce similar effects in most tissues, they may have different effects on the liver. For example, 17-EE is more potent than 17-E₂ in the production of certain hepatic effects (Krattenmacher et al., 1994), it more readily occupies nuclear ER sites in the liver (Marr et al., 1980a; Aten and Eisenfeld, 1982), and it remains bound to hepatic receptors longer than the endogenous estrogen (Marr et al., 1980b). Such differences probably result from reduced metabolism of 17-EE (relative to 17-E₂) and are thus expected to be observed primarily in tissues such as the liver that contain high levels of steroid metabolizing enzymes.

17-EE and 17-E₂ have also been reported to have different carcinogenic actions in some experimental models of liver and kidney cancer. For example, 17-EE produces greater effects than the natural hormone in a two-stage model of liver carcinogenesis in the rat (Yager et al., 1984), and the two compounds display differences in their ability to induce neoplasms in hamster kidney and liver (Li et al., 1998). In theory, such differences could occur if the two estrogens produced different patterns of gene expression, one of which caused an "imbalanced" production of growth regulatory signals (Stancel et al., 1995) in a given system. Our results do not favor such a possibility, however, and suggest other potential mechanisms may account for these differences, e.g., the production of reactive metabolites specific to the ethinyl group that form DNA adducts or cause damage to other critical cellular components (Yager and Liehr, 1996).

Our results show that the widely used synthetic estrogen 17-EE and the predominant endogenous estrogen 17-E₂ produce a similar in vivo pattern of uterine gene expression for four primary hormone-response genes. These effects are likely to be mediated by the classic ER-, because this is the predominant ER in the uterus and because it is thought to be the receptor subtype responsible for the uterotrophic actions of estrogens (Korach et al., 1996). Whether other primary estrogen-response genes show similar responses to the two estrogens in other tissues remains to be determined. This will be an interesting area for further study because the relative levels of ER- and the newly discovered ER- may vary from tissue to tissue. Similarly, our studies do not address the issue of whether 17-EE and 17-E₂ will produce similar or different responses when acting by nongenomic mechanisms that have been proposed to mediate certain effects of steroid hormones (Wehling, 1997).