Estrogen Targets Genes Involved in Protein Processing, Calcium Homeostasis, and Wnt Signaling in the Mouse Uterus Independent of Estrogen Receptor-α and -β

Estrogen actions in target organs are normally mediated via activation of nuclear estrogen receptors (ERs). By using mRNA differential display technique, we show, herein, that estradiol-17β (E2) and its catechol metabolite 4-hydroxy-E2 (4OHE2) can modulate uterine gene expression in ERα(−/−) mice. Whereas administration of E2 or 4OHE2 rapidly up-regulated (4–8-fold) the expression of immunoglobulin heavy chain binding protein (Bip), calpactin I ( CalP), calmodulin (CalM), and Sik similar protein(Sik-SP) genes in ovariectomized wild-type or ERα(−/−) mice, the expression of secreted frizzled related protein-2 (SFRP-2) gene was down-regulated (4-fold). Bip, CalP, and CalM are calcium-binding proteins and implicated in calcium homeostasis, whereas SFRP-2 is a negative regulator of Wnt signaling. Bip and Sik-SP also possess chaperone-like functions. Administration of ICI-182,780 or cycloheximide failed to influence these estrogenic responses, demonstrating that these effects occur independent of ERα, ERβ, or protein synthesis. In situ hybridization showed differential cell-specific expression of these genes in wild-type and ERα(−/−) uteri. Although progesterone can antagonize or synergize estrogen actions, it had minimal effects on these estrogenic responses. Collectively, the results demonstrate that estrogens have a unique ability to influence specific genes in the uterus not involving classical nuclear ERs.

Estrogens regulate diverse physiological responses including normal functioning of the reproductive and cardiovascular systems and bone metabolism (1)(2)(3). The uterus is a primary target for various estrogenic responses during the cycle and pregnancy. In the mouse, estrogen induces uterine epithelial cell proliferation, and together with progesterone (P 4 ) 1 it di-rects stromal cell proliferation and epithelial cell differentiation. These coordinated estrogen and P 4 interactions prepare the uterus to the receptive state for implantation (reviewed in Ref. 4). The mechanism by which estrogen renders the P 4primed uterus receptive for implantation is not clearly understood.
Estrogen actions are primarily executed by its binding to nuclear estrogen receptors, ER␣ and/or ER␤, which are ligandinducible transcription factors (5,6). They modulate transcription of genes by virtue of their binding as hormone receptor complexes to specific DNA sequences (hormone response elements) in target promoters (5,6). Despite the classical estrogenic actions, there is increasing evidence that gene transactivation or modulation of cell functions by estrogens is also mediated independent of nuclear ERs (7)(8)(9)(10). In many cells, a myriad of estrogenic effects occurs rapidly within seconds or minutes. These responses do not require RNA or protein synthesis and are considered to be mediated by estrogen binding to the plasma membrane (10 -12). For example, increases in intracellular cAMP, calcium influx, inositol triphosphate, and release of prolactin are all attributed to membrane-mediated estrogen actions (10 -12). Although the presence of membrane ER has been claimed for more than two decades (13), the subject is still controversial. However, the identity of a membrane ER has recently been addressed by transfection studies in Chinese hamster ovary cells using cDNAs for ER␣ or ER␤ (14). It was shown that functionally active ER␣ or ER␤ is localized in the plasma membrane and in the nucleus originating from the same mRNA transcript. Furthermore, the existence of a membrane estrogen-binding protein, maxi-K channel, has also been reported (15). This channel consists of a regulatory subunit ␣ that confers higher Ca 2ϩ sensitivity and binds to estrogen for channel activation in the presence of a pore-forming ␤-subunit.
There is a general consensus that rapid actions of estrogens, especially in tissues lacking nuclear ERs, are the result of a novel mechanism that involves estrogen interaction with a yet unidentified receptor (7). By using ER␣-deficient mice and an ER antagonist, we have previously demonstrated that 4-hydroxyestradiol-17␤ (4OHE 2 ), a catechol metabolite of estradiol-17␤ (E 2 ), can induce the expression of lactoferrin (LF, an estrogen-responsive gene) in uteri of ER␣(Ϫ/Ϫ) mice (7). The result suggested that the response is not mediated by ER␣ or ER␤ and points toward a novel pathway of estrogen actions in the mouse uterus. To better understand the estrogen actions independent of ER␣ and/or ER␤, we sought to identify genes that are targets of E 2 and/or 4OHE 2 . We used mRNA differential display to identify estrogen-responsive genes in ER␣(Ϫ/Ϫ) uteri. Upon identification, the expression patterns of these genes were analyzed in wild-type and ER␣(Ϫ/Ϫ) uteri exposed to estrogens in the presence or absence of an ER antagonist. Although mRNA expression of four of the genes, glucose-regulated protein-78 kDa (GRP78)/immunoglobulin heavy chain binding protein (Bip), calpactin I (CalP), calmodulin (CalM), and SIK-similar protein (Sik-SP) was up-regulated, the expression of the secreted frizzled-related protein-2 (SFRP-2) was down-regulated in the uterus by E 2 or 4OHE 2 in both the wild-type and ER␣(Ϫ/Ϫ) mice. We also observed that these estrogenic responses are not influenced by an ER antagonist ICI-182,780 (ICI), a protein synthesis inhibitor cycloheximide (Cyhx), or progesterone, suggesting that these effects occur independent of ER␣, ER␤, protein synthesis, or progesterone receptor (PR).
Bip, a member of the HSP70 (chaperone) family and a major protein of the endoplasmic reticulum lumen, is induced under a variety of stress situations (16). It is involved in the storage of rapidly exchanging Ca 2ϩ pool and correct folding of the newly synthesized proteins (17)(18)(19). CalP and CalM, two calciumbinding proteins, are expressed ubiquitously in eukaryotic cells and participate in the modulation of several Ca 2ϩ -dependent functions including protein kinase, adenylate cyclase, and cyclic nucleotide phosphodiesterase activities (20 -22). CalM can also regulate ER transcriptional activity by its direct association with ER and interact with myosin light chain kinase to control uterine muscle contraction (23)(24)(25)(26). CalP, a member of the annexin family, plays a role in immunomodulation (27). Sik-SP is conserved with a gene family nop5/sik1 that encodes components of small nucleolar ribonucleoprotein complexes. They have an essential role in rRNA processing and may also be involved in chaperone-like function (28). SFRP-2 is a modulator of Wnt signaling (29), which is involved in cell proliferation, differentiation, migration, polarity, and cell fate determination during development (29 -31). Wnts interact with cell surface frizzled receptors and were originally identified as regulators of tissue polarity in Drosophila (30 -32). SFRP-2 is a secreted frizzled, lacking the seven transmembrane and intracellular signaling domains (29,33). Secreted frizzled proteins are expressed in many cell types during embryogenesis (34,35) and participate in modulating Wnt-frizzled signaling (29) and apoptosis (36). Our present results showing the influence of estrogen on uterine expression of genes that are involved in three fundamental cellular functions, such as protein processing, calcium homeostasis, and Wnt signaling independent of the classical ER or PR pathway, suggest diverse mode of estrogen actions.

MATERIALS AND METHODS
Animals-Littermate wild-type and ER␣(Ϫ/Ϫ) mice of the same genetic background (129/J/C57BL/6J) were produced by crossing heterozygous females and males (37). Littermate wild-type and PR(Ϫ/Ϫ) mice of the same genetic background (129SvEv/C57BL/6) were produced by crossing homozygous males with heterozygous females (38). In all comparison studies, littermate wild-type mice were analyzed under similar conditions against ER␣(Ϫ/Ϫ) or PR(Ϫ/Ϫ) mice. ER␣ and PR mutant mice were originally obtained from Dennis B. Lubahn (University of Missouri, Columbia) and Bert O'Malley (Baylor College of Medicine, Houston), respectively. They were housed in the animal care facility at the University of Kansas Medical Center according to the National Institutes of Health and institutional guidelines for the care and use of laboratory animals. Mice were genotyped by PCR analysis of tail DNA. Adult (8 -10 weeks old) mice were ovariectomized and rested for 1 week before they received any injections.
Injection Schedule-ER␣(Ϫ/Ϫ) and littermate wild-type mice were given an injection of oil (0.1 ml), E 2 (250 ng/mouse), 4OHE 2 (250 ng/ mouse), ICI (500 g/mouse) or the same dose of ICI 30 min prior to steroid injections. Mice were killed 6 h after the last injection. For temporal studies, mice were killed at 0.5, 1, 2, 6, and 24 h after steroid injections. For protein synthesis inhibition studies, cycloheximide (Cyhx, 100 g/mouse) was used 30 min prior to the injection of steroids. PR(Ϫ/Ϫ) and littermate wild-type mice were given an injection of E 2 (250 ng/mouse) and/or P 4 (2 mg/mouse). They were killed 6 h after the last injection. All of the test agents were dissolved in sesame oil and injected (0.1 ml/mouse) subcutaneously.
Differential Display of mRNA-To examine the estrogenic responses on uterine gene expression independent of ER␣, we utilized ER␣(Ϫ/Ϫ) mice. By using differential display technique, we compared mRNA profiles of uterine samples obtained 6 h after single injections of either E 2 or 4OHE 2 with those of oil-treated controls. Differential display technique followed the protocol as described previously with some modifications (39,40). In brief, 1.0 g of DNA-free total RNA was used for reverse transcription (RT) reactions using three different one-base anchored primers as described earlier (39) with the following modifications: LHT11C (5Ј-TGCCGAAGCTTTTTTTTTTTC-3Ј), LHT11G (5Ј-TG-CCGAAGCTTTTTTTTTTTG-3Ј), and LHT11A (5Ј-TGCCGAAGCTTTT-TTTTTTTA-3Ј). The polymerase chain reaction (PCR) was performed in a reaction mixture containing 100 l of the RT product, 1ϫ PCR buffer (10 mM Tris-HCl, pH 8.3; 2.5 mM MgCl 2 , and 50 mM KCl), 600 M each of dATP, dTTP, dGTP, dCTP, and 500 Ci/ml 35 S-dATP (1200 Ci/mmol, NEN Life Science Products), 0.5 M of the respective primers LHT 11 C, LHT 11 G, or LHT 11 A, 0.5 M of the arbitrary primer, and 20 units/ml Ampli Taq TM DNA polymerase (Perkin-Elmer). The arbitrary primers were as described (39) but with the following modifications: LHAP1 (5Ј-TGCCGAAGCTTGATTGCC-3Ј), LHAP2 (5ЈTGCCGAAGCCTTCGA-CTGT-3Ј) or LHAP3 (5Ј-TGCCGAAGCTTTGGTCAG-3Ј). PCR was performed in a Perkin-Elmer 480 thermocycler using the following cycling parameters: first cycle at 94°C for 1 min, 40°C for 4 min, and 72°C for 1 min followed by 35 cycles at 94°C for 45 s, 60°C for 2 min, and 72°C for 1 min. The amplified cDNAs were separated on a 6% DNA sequencing gel. Differentially displayed bands of interest were reamplified by PCR using the appropriate primers and the reaction conditions as described above. The products were then cloned into the pCR-Script TM SK(ϩ) vector (Stratagene cloning systems, Stratagene, La Jolla, CA).
Sequencing of cDNA Subclones of the PCR Fragments-Doublestranded DNA sequencing was carried out with either T7 or T3 primers using the SequiTherm long-read cycle sequencing kit LC (Epicenter Technologies, Madison, WI). The nucleotide sequences were analyzed by the BLAST Sequence Similarity Searching Program (blastn) using the GenBank TM sequence data base of the National Center for Biotechnology Information, National Institutes of Health.
Hybridization Probes-For Northern hybridization, 32 P-labeled antisense cRNA probes were generated using either T7, T3, or SP6 RNA polymerases. For in situ hybridization, sense and antisense 35 S-labeled cRNA probes were generated. A 1.8-kilobase pair cDNA fragment (EcoRI/SacI) of a mouse cDNA clone for c-fos (41) was subcloned in pSP64 vector at EcoRI/SacI sites. The clone description for ribosomal protein L-7 (rpL7) cDNA was reported earlier (42).
Northern Blot Hybridization-For Northern blot hybridization, total RNA (6.0 g) was denatured and separated by formaldehyde/agarose gel electrophoresis, transferred to nylon membranes, and UV crosslinked. Northern blots were prehybridized, hybridized, and washed as described by us (40,42). Quantitation of hybridized bands was analyzed by densitometric scanning.
In Situ Hybridization-In situ hybridization was performed as described previously (42). Frozen uterine sections (10 m) were fixed in 4% paraformaldehyde in phosphate-buffered saline for 15 min at 4°C. Following fixation, sections were prehybridized and hybridized to 35 Slabeled antisense cRNA probes for 4 h at 45°C. As negative controls, uterine sections were hybridized with the 35 S-labeled sense probes. RNase A-resistant hybrids were detected within 3-7 days of autoradiography. The slides were post-stained with hematoxylin and eosin.
Competitive PCR-The quantitation of mRNAs by competitive PCR was described previously by us (7). In brief, the competitor templates were generated by introducing a nonspecific DNA fragment into a mouse target cDNA clone. Specifically, a 185-base pair blunt-ended fragment (SspI), obtained from pGEM7Zf(ϩ) vector, was inserted into the cDNA clones for CalP at the SmaI site, for CalM and SFRP-2 at the StuI site, and for Bip at the SspI site. These modified cDNA templates were used as competitors to carry out the competitive PCR for the respective genes. for Bip were used for Southern blot hybridization of the RT-PCR-amplified products. For rpL7, primers used for RT-PCR and Southern hybridization were same as described earlier (7). Quantitation of band intensity on the autoradiogram was achieved by densitometric analysis. The ratio of band intensities for the competitor and the target cDNA was calculated for each sample and plotted against the amounts of the competitor. The efficiency of the RT reaction was controlled by measuring the levels of rpL7 mRNA in each sample and were similar in all samples (ϳ4.0 ϫ 10 7 copies/g of total RNA).

Estradiol and Catecholestradiol Regulate Gene Expression in ER␣(Ϫ/Ϫ) Uteri-
We previously demonstrated that the expression of the LF gene, normally induced by E 2 in the mouse uterus, is up-regulated within 6 h after an injection of 4OHE 2 , but not E 2 , in ER␣(Ϫ/Ϫ) uteri. Furthermore, this response was not inhibited by prior administration of an ER antagonist, ICI, and was accompanied by early estrogenic responses, such as uterine water imbibition and macromolecular uptake (7). These results suggested that estrogens execute some uterine effects that are independent of ER␣ and/or ER␤ (7). To examine whether estrogen can also modulate other genes in the uterus in the absence of ER␣, we investigated the effects of E 2 or 4OHE 2 on uterine gene expression in ER␣(Ϫ/Ϫ) mice using the mRNA differential technique. We analyzed 26 PCR-amplified products that were displayed differentially in uterine RNA samples obtained from ovariectomized ER␣(Ϫ/Ϫ) mice 6 h after an injection of oil, E 2 , or 4OHE 2 . Cloning, sequencing, and expression studies led to the identification of five authentic cDNA clones whose corresponding genes showed either upregulation or down-regulation after estrogen treatments (Fig. 1). Among the five genes, the expression of Bip, CalP, CalM, and Sik-SP was up-regulated, whereas that of SFRP-2 was down-regulated by E 2 or 4OHE 2 . It is to be noted that after 4OHE 2 treatment, a band was detected on the differential display gel within close proximity but not of the same size of the SFRP-2 band as observed in the oil-treated sample. However, cloning, sequencing, and Northern blot hybridization revealed this band to be an artifact.

Differentially Displayed Genes Are Rapidly Modulated by
Estrogen in Wild-type or ER␣(Ϫ/Ϫ) Uteri and Are Unresponsive to Anti-estrogen Treatment-Although several genes were differentially displayed by uterine RNA samples of ovariectomized ER␣(Ϫ/Ϫ) mice after estrogen treatment, we wanted to confirm their authenticity, differential responses to estrogens, and an ER antagonist ICI. We examined the levels of Bip, CalP, CalM, Sik-SP, and SFRP-2 mRNAs in ovariectomized wildtype mice 6 after an injection of oil, E 2 , or 4OHE 2 with or without ICI by Northern hybridization (Fig. 2). We observed very low levels of uterine Bip, CalP, CalM, and Sik-SP mRNAs after an injection of oil. However, an injection of E 2 or 4OHE 2 increased the levels of these mRNAs 4 -8-fold by 6 h. Treatment of mice with ICI prior to the injections of estrogens failed to show any effects. In contrast, high levels of SFRP-2 mRNA were detected in oil-treated uterine samples, and these high levels were readily down-regulated (4-fold) by estrogen treatments. Again, ICI did not antagonize these effects.
To examine the temporal expression patterns of these genes by E 2 or 4OHE 2 , Northern blot hybridization was performed using uterine RNA samples isolated at different times (0.5, 1, 2, 6, and 24 h) after an injection of E 2 or 4OHE 2 in ovariectomized wild-type mice (Fig. 3). RNA samples from oil-treated uterine samples at 6 h served as controls. The effects of estrogens on the expression of these five genes were compared with that of c-fos, a known estrogen-dependent early-inducible gene in the rodent uterus (43). As expected, the levels of Bip, CalP, CalM, and Sik-SP mRNAs remained low in vehicle-treated uterine samples. However, an injection of E 2 (Fig. 3A) or 4OHE 2 (Fig. 3B) rapidly up-regulated the expression of these four genes and c-fos within 1-2 h. The levels of BIP mRNA peaked at 1 h and remained high through 6 h, whereas those of CalP and CalM reached highest levels at 6 h. As observed previously (43), the induction of c-fos mRNA by estrogen was very rapid and transient, reaching its peak at 1 h followed by a rapid decline. In general, the induction level of these genes by E 2 or 4OHE 2 was 4 -8-fold at 6 h. In sharp contrast, the levels of SFRP-2 mRNA were high in oil-treated ovariectomized uteri but declined rapidly after an E 2 or 4OHE 2 injection, reaching its lowest levels by 1 h.
Although the results of differential display suggested estrogen modulation of these five genes in the ER␣(Ϫ/Ϫ) uterus with very low levels of ER␤ (44), the extent of their responsiveness to estrogen or the participation of ER␤ in these estrogenic responses could not be ascertained. We used a quantitative RT-PCR technique to address these questions, because of the limited availability of uterine RNA from ER␣(Ϫ/Ϫ) mice. This technique uses gene-specific competitive templates to measure FIG. 1. Differential display of uterine mRNAs in ER␣(؊/؊) mice after injections of oil, E 2 , or 4OHE 2 . Three different uterine total RNA samples isolated from ovariectomized ER␣(Ϫ/Ϫ) mice 6 h after injections of oil, E 2 (250 ng/mouse), or 4OHE 2 (250 ng/mouse) were compared by differential display. Reverse transcription reaction was performed using 5.0 g of total RNA in the presence of one-base anchored modified primers LHT11G, LHT11C, or LHT11A. Each primer-driven RT products were PCR-amplified using the corresponding RT primer together with an arbitrary primer LHAP1, LHAP2m or LHAP3 (39). The PCR-amplified cDNA fragments were obtained by a pair of primers as follows: (a) LHAP3/LHT11G, (b) LHAP3/LHT11A, (c) LHAP2/LHT11C, and (d) LHAP3/LHT11A. Arrows indicate cDNA bands displayed differentially. These experiments were repeated twice with independent RNA samples, and similar results were obtained. mRNA levels of choice and was used to measure the mRNA levels of differentially displayed genes in ER␣(Ϫ/Ϫ) uteri after exposure to estrogens. As shown in Tables I-III, an injection of E 2 or 4OHE 2 significantly increased the uterine levels of Bip (Ϸ3-16-fold), CalP (Ϸ5-7-fold), and CalM (Ϸ4 -6-fold) mRNAs in ovariectomized ER␣(Ϫ/Ϫ) mice within 6 h. In contrast, similar treatments with estrogens drastically reduced the levels (8 -10-fold) of uterine SFRP-2 mRNA (Table IV). ICI, which neutralizes ER␣ and ER␤ functions (45), failed to antagonize these estrogenic responses (Tables I-IV), suggesting that ER␤ is also not involved in these responses. These results suggest that estrogens can modulate expression of certain genes in the mouse uterus independent of the classical ERs.
Effects of E 2 or 4OHE 2 on Uterine Gene Expression Are Independent of Protein Synthesis-The rapid responses of these genes to estrogens independent of ER␣ and ER␤ led us to examine whether these responses required new protein synthesis. Uterine RNA was analyzed by Northern hybridization. As shown in Fig. 4, the levels of Bip, CalP, CalM and Sik-SP mRNAs remained low after an injection of oil or Cyhx alone. Although a single injection of E 2 , as expected, up-regulated the mRNA levels of these genes, a prior treatment with Cyhx failed to alter the estrogen-induced responses. Similarly, the downregulation of SFRP-2 mRNA levels by estrogen was also not affected by Cyhx pretreatment. The effects of Cyhx on 4OHE 2induced modulation of these various genes were similar to those of E 2 (data not shown). An administration of the same dose of Cyhx (100 g) 30 min prior to an injection of estrogen was shown to block uterine amino acid incorporation into protein in the rat (46) or uterine c-myc expression in the mouse during the early phase (47).
Differentially Displayed Genes Are Spatially Expressed by E 2 and 4OHE 2 in Wild-type and ER␣(Ϫ/Ϫ) Uteri-To determine whether estrogen modulates uterine gene expression in a cell type-specific manner, in situ hybridization was performed on uterine sections obtained from ovariectomized wild-type or ER␣(Ϫ/Ϫ) mice 6 h after receiving an injection of oil, E 2 , or 4OHE 2 with or without ICI. The accumulation of Bip, CalP, CalM, and Sik-SP mRNAs was low to undetectable in wild-type or ER␣(Ϫ/Ϫ) uteri after an injection of oil (Figs. 5 and 6).

FIG. 2. E 2 or 4OHE 2 regulates uterine expression of Bip, CalP, CalM, Sik-SP, and SFRP-2 mRNAs in ovariectomized wild-type mice, and this expression was unresponsive to ICI.
Adult ovariectomized mice were given an injection of oil, E 2 (250 ng/mouse), 4-OH-E 2 (250 ng/mouse), ICI (500 g/kg), or ICI 30 min before an injection of E 2 or 4OHE 2 and killed 6 h after the last injection. Total uterine RNA (6 g) pooled from 5 to 7 mice was used for each group. Autoradiographic exposures were 6 h for SFRP-2 and Sik-SP, 3 h for Bip, CalP, and CalM, and 2 h for rpL7. These experiments were repeated twice with independent RNA samples, and average values with range of responses from two experiments are shown in histograms. Fold changes in mRNA levels were calculated with respect to oil after normalization with rpL7 mRNA levels.

FIG. 3. Temporal effects of E 2 (A) or 4OHE 2 (B) on uterine expression of Bip, CalP, CalM, SFRP-2, Sik-SP, c-fos, and rpL7
mRNAs in ovariectomized wild-type mice. Adult ovariectomized mice were given a single injection of E 2 (250 ng/mouse) or 4OHE 2 (250 ng/mouse) and killed at the times indicated. Mice injected with oil and killed 6 h later served as a control. Total uterine RNA (6 g) pooled from 5 to 7 mice was used for each group. Autoradiographic exposures were 6 h for SFRP-2, Sik-SP, and c-fos, 3 h for Bip, CalP, and CalM, and 2 h for rpL7. These experiments were repeated two times with independent RNA samples, and average values with range of responses from two experiments are shown in histograms. Fold changes in mRNA levels were calculated with respect to oil and were normalized against rpL7 mRNA levels. However, an injection of E 2 or 4OHE 2 in wild-type mice showed increased accumulation of these mRNAs predominantly in luminal and glandular epithelia with low levels in the stroma (Figs. 5 and 6). In contrast, similar treatments induced these genes differentially in ER␣(Ϫ/Ϫ) uteri (Figs. 5 and 6). For example, E 2 or 4OHE 2 primarily induced the expression of Bip mRNA in stromal cells (Fig. 5), CalP in epithelial cells (Fig. 5), and CalM in both epithelial and stromal cells (Fig. 6). Interestingly, Sik-SP mRNA accumulation was primarily detected in epithelial cells by E 2 but in both stromal and epithelial cells by 4OHE 2 (Fig. 6). In contrast, distinct accumulation of SFRP-2 mRNA was observed in stromal cells of ovariectomized oil-treated wild-type and ER␣(Ϫ/Ϫ) uteri (Fig. 7), whereas an injection of E 2 or 4OHE 2 dramatically down-regulated its expression in these cells (Fig. 7). Pretreatment of mice with ICI did not influence the levels or the pattern of expression for all of these genes in response to estrogens either in wild-type or ER␣(Ϫ/Ϫ) mice (Figs. 5-7). Furthermore, the responses to ICI alone in the wild-type and ER(Ϫ/Ϫ) mice were similar to those of vehicle-treated controls (Figs. 5-7). The expression of these genes was specific, since hybridization with corresponding sense cRNA probes did not show any positive signals (data not shown).
Estrogen-dependent Modulation of Uterine Gene Expression Is Not Altered by Progesterone-Since estrogen interacts with P 4 synergistically or antagonistically, we surmised that the estrogenic effects on these genes could be modulated by P 4 . Thus, we compared the effects of P 4 on uterine expression of these genes in wild-type mice with those in PR(Ϫ/Ϫ) mice. Ovariectomized wild-type or PR(Ϫ/Ϫ) mice received an injection of oil, P 4 , or P 4 plus E 2 . Mice were killed 6 h later, and uterine RNA was analyzed by Northern hybridization. As shown in Fig. 8, levels of Bip and Sik-SP mRNAs were low in vehicle-treated uteri, whereas levels of SFRP-2 were high in both the wild-type and PR(Ϫ/Ϫ) mice. As expected, an injection of E 2 up-regulated the levels of Bip and Sik-SP mRNAs and down-regulated the levels of SFRP-2 mRNA in these mice. In contrast, treatment with P 4 alone or with E 2 failed to show any noticeable effects on the levels of Sik-SP and SFRP-2 mRNAs in wild-type or PR(Ϫ/Ϫ) uteri. However, uterine Bip expression was modestly up-regulated by P 4 alone in wild-type but not in PR(Ϫ/Ϫ) mice, although P 4 did not antagonize or synergize E 2 -induced Bip expression. These results suggest that P 4 alone can influence this gene via activation of PR but does not influence its responsiveness to estrogen. Our initial studies also showed that P 4 is ineffective in influencing the expression of CalP or CalM (data not shown). DISCUSSION Many of the diverse biological functions of estrogens are the result of their direct interactions with nuclear ERs. There is now evidence for specific functions and gene expression in the target organs elicited by estrogens independent of ER␣ and/or ER␤ (7,44). For example, 4OHE 2 , but not E 2 , can induce LF

FIG. 4. Effects of Cyhx on uterine expression of Bip, CalP,
CalM, Sik-SP, and SFRP-2 mRNAs in response to E 2 or 4OHE 2 in ovariectomized wild-type mice. Adult ovariectomized mice were given a single injection of oil, E 2 (250 ng/mouse), 4-OH-E 2 (250 ng/ mouse), Cyhx (100 g/mouse), or the same dose of Cyhx 30 min before the injection of E 2 or 4OHE 2 and killed 6 h after the last injection. Total uterine RNA (6 g) pooled from 5 to 7 mice was used for each group. Autoradiographic exposures were 6 h for SFRP-2 and Sik-SP, 3 h for Bip, CalP, and CalM, and 2 h for rpL7. These experiments were repeated twice with independent RNA samples, and similar results were obtained. expression, water imbibition, and macromolecular uptake in ER␣(Ϫ/Ϫ) uteri, and these responses are not neutralized by ICI (7). The signaling system involved in these responses is not yet understood. The present investigation demonstrates that not only 4OHE 2 but also E 2 can modulate a group of genes in the uterus that are involved in protein processing, calcium homeostasis, and Wnt signaling without involving classical ERs, PR, and nascent protein synthesis. These unique uterine estrogenic FIG. 5. In situ hybridization of Bip and CalP genes in uteri of ovariectomized wild-type and ER␣(؊/؊) mice after exposure to E 2 , E 2 plus ICI, 4OHE 2 , or 4OHE 2 plus ICI. Adult ovariectomized mice rested for 2 weeks were used. Mice were given a single injection with oil, ICI (20 mg/kg), E 2 (250 ng/ mouse), 4OHE 2 (250 ng/mouse), or the same dose of ICI 30 min before the injection of E 2 or 4OHE 2 , and they were killed 6 h after the last injection. Frozen sections (10 m), fixed in paraformaldehyde, were mounted onto glass slides, prehybridized, and hybridized with 35 S-labeled sense (data not shown) or antisense cRNA probes. RNase A-resistant hybrids were detected after 2-7 days of autoradiography. Dark field photomicrographs of uterine sections are shown at ϫ 100. le, luminal epithelium; ge, glandular epithelium; s, stroma; and myo, myometrium. These experiments were repeated three times with 3-4 mice in each group, and similar results were obtained.
FIG. 6. In situ hybridization of CalM and Sik-SP genes in uteri of ovariectomized wild-type and ER␣(؊/؊) mice after exposure to E 2 , E 2 plus ICI, 4OHE 2 , or 4OHE 2 plus ICI. Injection schedules and the doses of various agents were same as described in Fig. 5 legend. Frozen sections (10 m), fixed in paraformaldehyde, were mounted onto glass slides, prehybridized, and hybridized with 35 S-labeled sense (data not shown) or antisense cRNA probes. RNase A-resistant hybrids were detected after 2-7 days of autoradiography. Dark field photomicrographs of uterine sections are shown at ϫ 100. le, luminal epithelium; ge, glandular epithelium; s, stroma; and myo, myometrium. These experiments were repeated three times with 3-4 mice in each group, and similar results were obtained.
responses point toward the concept that certain fundamental estrogenic functions, such as protein processing, calcium homeostasis, and Wnt signaling in the target organ are retained in the absence of classical ERs. Whether orphan or yet unidentified nuclear receptors are involved in these responses remains unknown. Nonetheless, our present results are intriguing and likely to stimulate further research in identifying the signaling mechanism for these responses.
Although Although estrogen induction of these genes is independent of ERs, their differential cell-specific expression between the wild-type and ER␣(Ϫ/Ϫ) uteri suggests an interaction between this novel pathway and classical ERs in specifying cell-specific expression. Epithelial-mesenchymal "cross-talk" is important for normal uterine functions and gene expression (48). It is possible that this cross-talk is impaired or absent in ER␣(Ϫ/Ϫ) uteri causing differential cell-specific gene expression.
In adult mice, estrogens produce a biphasic uterine response (49,50). The immediate early responses occur within 6 h of estrogen administration, and water imbibition and macromolecular uptake are two predominant characteristics. The late or growth responses occur by 18 -30 h and are characterized by hyperplasia and hypertrophy. Our present observation of rapid modulation of genes after injection of estrogens suggests that specific early estrogenic responses are independent of classical ERs or new protein synthesis. However, these early responses could be important for the onset of the late growth phase that is ER␣-dependent. The manifestation of these early responses with the absence of the growth phase in ER␣(Ϫ/Ϫ) mice suggests the lack of the machinery for the growth phase. The induction of immediate early genes (c-fos, c-jun, and c-myc) by short-acting estrogens is not adequate to stimulate DNA synthesis in the rodent uterus (51). Thus, it appears that the mitogenic stimulation requires further changes that depend on prolonged estrogen action. A cross-talk between the non-classical and classical actions of steroid hormones is described by Katzenellenbogen (52). For example, protein kinase activators enhance the ER␣ transcriptional activity. There is also evidence that IGF-1 and agents that raise intracellular cAMP also stimulate ER phosphorylation and activation (53). Estrogen activation of the traditional "genomic" pathway involves mRNA and protein synthesis, whereas rapid estrogenic responses occurring via a non-traditional pathway are believed to be mediated via membrane receptor and do not require new protein synthesis. However, the identity of the putative membrane receptor is still controversial. Our observations of rapid estrogenic modulation of uterine gene expression independent of protein synthesis and classical ERs are also characteristics of an early response. Defining the signaling mechanism of the early estrogenic responses may have clinical significance in distinguishing the beneficial effects (cardiovascular and neurological protections) of estrogens from their long term detrimental (carcinogenic consequences) effects.
Rapid responsiveness of uterine Bip and Sik-SP to estrogens could be physiologically important. The late estrogen action primarily involves uterine growth which requires correct fold- FIG. 7. In situ hybridization of SFRP-2 gene in uteri of ovariectomized wild-type and ER␣(؊/؊) mice after exposure to E 2 , E 2 plus ICI, 4OHE 2 , or 4OHE 2 plus ICI. Injection schedules and the doses of various agents were the same as described in Fig. 5 legend. Frozen sections (10 m), fixed in paraformaldehyde, were mounted onto glass slides, prehybridized, and hybridized with 35 S-labeled sense (data not shown) or antisense cRNA probes. RNase A-resistant hybrids were detected after 2-7 days of autoradiography. Dark field photomicrographs of uterine sections are shown at ϫ 100. le, luminal epithelium; ge, glandular epithelium; s, stroma; and myo, myometrium. These experiments were repeated three times with 3-4 mice in each group, and similar results were obtained.

FIG. 8. Effects of P 4 on estrogen-induced uterine expression of
Bip, Sik-SP, and SFRP-2 in wild-type and PR(؊/؊) mice. Adult ovariectomized mice were given an injection of oil, E 2 (250 ng/mouse), P 4 (2 mg/mouse), or the same doses of E 2 plus P 4 . Total uterine RNA (6 g) pooled from 5 to 7 mice was used for each group. Autoradiographic exposures were 6 h for SFRP-2 and Sik-SP, 3 h for Bip, and 2 h for rpL7. These experiments were repeated twice with independent RNA samples, and similar results were obtained.
ing and functioning of a variety of newly synthesized proteins. Because of the involvement of Bip in folding and translocation of nascent proteins within the endoplasmic reticulum, one of the early functions of estrogen could be to prepare the uterine environment for protein processing for the late phase. A chaperone-like role for Bip was recently reported in the rat uterus during decidualization (54). Sik-SP could also be involved in similar functions, because of its chaperone-like functions. We suggest that genes regulated by estrogen independent of nuclear ERs could be linked with the ER-dependent late estrogenic effects.
Calcium plays a major role in mediating estrogen signaling (55,56), and it can act as a second messenger to induce Bip in monocytes (57). Cellular calcium homeostasis depends on the concerted efforts of calcium-binding proteins. Since Bip, CalP, and CalM all bind calcium and are regulated in the uterus by estrogen, it is possible that calcium is involved in regulating these genes. The spatiotemporal regulation of uterine Bip, CalP, and CalM by estrogen suggests that these genes function in a coordinated manner. In rodents, uterine CalM levels increase during pregnancy and after estrogen treatment (58). Furthermore, an intrauterine injection of CalM antagonist (chlorpromazine) inhibits implantation in the rat (59), suggesting its role in this process. Since estrogen is an absolute requirement for implantation in mice, it is possible that one of the actions of estrogen in implantation is to induce CalM via a non-ER pathway. CalP is localized in the syncytiotrophoblast cells in the developing human placenta and possesses Fc gamma receptor activity, suggesting its role in immunomodulation (60). Uterine CalP expression by estrogen may have a role in local immunomodulation.
The uterine regulation of SFRP-2 is an interesting observation, since very few genes are known to be down-regulated by estrogen (61)(62)(63)(64)(65). To our knowledge, this is a gene that is abundantly expressed in quiescent uterine stromal cells but is down-regulated by estrogens. Since SFRP-2 negatively regulates Wnt functions, it is envisioned that its down-regulation by estrogen allows Wnt-frizzled signaling to execute specific uterine functions. Wnt ligands participate in mesenchymal-epithelial interactions (66), and uterine expression of Wnt ligands (Wnt-4, Wnt-5a, and Wnt-7) is tightly regulated during the estrous cycle by estrogen and/or P 4 (67,68). Since Wnts regulate cellular proliferation, differentiation, and/or reorganization, we suggest that they act as estrogen-mediated transducers of these events in the uterus. Bip could also be a part of this system, since Wnt-1 interacts with Bip for its secretion from the cell (69). Wnts are involved in two signaling pathways. They can activate ␤-catenin that modulates transcription of specific target genes in the nucleus. They can also stimulate increases in intracellular Ca 2ϩ or protein kinase C activity via activation of pertussis toxin-sensitive G-proteins . Whether these Wnt signaling pathways are operative in the uterus remains to be examined.