Cellular Retinoic Acid-binding Protein II Gene Expression Is Directly Induced by Estrogen, but Not Retinoic Acid, in Rat Uterus*

It has been suggested that cellular retinoic acid-bind-ing protein (II) (CRABP(II)) may have a role in the movement of retinoic acid (RA) to its nuclear receptors, thereby enhancing the action of RA in the cells in which it is expressed. RA has also been shown to increase expression of CRABP ( II ). Previous work from our laboratory has shown that 17 (cid:1) -estradiol (E 2 ) administration to prepubertal female rats leads to acquisition of the ability of the lining epithelium to synthesize RA as well as to express CRABP(II). To determine whether this appearance of CRABP(II) was dependent on the production of RA, both E 2 and RA were administered to ovariecto- mized rats. E 2 administration induced expression of the CRABP ( II ) gene in the uterus within 4 h, and this induction was not inhibited by prior administration of puromycin, indicating that the induction was direct. In contrast, RA caused no change in CRABP(II) message level, even at times as late as 48 h after administration. Isolation and analysis of 4.5 kb of the 5 (cid:1) -flanking region of the gene revealed no apparent E 2 -response element. Using this portion of the gene to drive expression of the luciferase gene in transfected cells allowed identification of a region containing an imperfect estrogen-response element and estrogen-response element half-site, neces-sary for E 2 -driven induction. A possible Sp1 binding site in the 5 (cid:1) -flanking region of the CRABP

The vitamin A metabolite all-trans-retinoic acid (RA) 1 regulates multiple biological processes, including cell proliferation and differentiation, by virtue of its ability to modulate the rate of transcription of numerous target genes. The transcription activities of this hormone are mediated by a family of proteins, the retinoic acid receptors (RAR␣, ␤, and ␥ and isoforms). These receptors bind to specific response elements in the promoter regions of target genes and work as ligand-inducible transcription enhancers and repressors (1).
The numerous sites of action of retinoic acid, e.g. from reproductive organs to the respiratory system, suggest that there will be several different factors that regulate its synthesis. One such factor may be estrogen. Administration of E 2 to the prepubertal rat leads to a gain of the ability of the uterine lining epithelial cells to synthesize RA (2). Coincident with this gain of ability is the appearance of CRABP(II) within these cells. CRABP(II) is a member of a large family of small proteins that specifically bind lipophilic compounds such as fatty acids and retinoids (3).
Recent work has suggested that CRABP(II) may have a role in the movement of RA to the RARs, thereby enhancing the action of RA in the cells in which it is expressed (4 -7). Consistent with that idea, we have now observed coincident appearance of the ability to synthesize RA and expression of CRABP(II) in certain cell types in addition to the uterine epithelium: the developing granulosa cells of the ovarian follicle prior to ovulation and in stromal cells of the uterus undergoing the process of decidualization (2,8,9). Human mammary ductal epithelium also has the ability to synthesize RA, and those cells express CRABP(II) as well (10).
Increased expression of CRABP(II) has been noted after administration of RA to intact skin or to cells in culture (11,12). Analyses of the promoter regions of the human and murine genes have revealed the presence of RA-response elements (RAREs) (13,14). However, no prototypical estrogen-response element (ERE) was observed for the promoter region for the CRABP(II) gene of these two species. This suggests that the induction of expression of the native rat CRABP(II) gene that we have observed after E 2 administration may require the initiation of RA biosynthesis.
To address this question, we have examined expression of CRABP(II) after administration of either E 2 or RA to the ovariectomized rat. Only E 2 induced expression of CRABP(II). Further analysis of the rat CRABP(II) promoter region demonstrated that the E 2 is acting through both an imperfect ERE and a possible Sp1-response element.

EXPERIMENTAL PROCEDURES
Animals and Tissue Collection-Female ovariectomized Sprague-Dawley rats (180 -200 g) were purchased from Harlan Sprague-Dawley Inc. (Indianapolis, IN). Rats were housed in a temperature-and lightcontrolled room (Ϯ21°C; lights on 0700 -1900 h), fed rat chow (Ralston Purina Co., St. Louis, MO), provided with water ad libitum, and allowed to acclimate for 2 weeks before use. Rats were divided into six groups, each with at least three animals, and injected intraperitoneally with corn oil, puromycin (10 mg/rat), E 2 (10 g/rat), puromycin plus E 2 , RA (500 g/rat), or puromycin plus RA, respectively. Puromycin was injected 30 min before injection with E 2 or RA, for those experimental groups. After 4 h, uteri were harvested from the animals for RNA extraction. These studies were conducted in accordance with the Na-tional Institutes of Health Guide for the Care and Use of Laboratory Animals and with the oversight of veterinarian of our local institutional animal care and use committee.
RNA Extraction and RNase Protection Assays (RPA)-Total RNA was extracted from individual rat uterus using TRIzol reagent (Invitrogen) and quantified by spectrophotometry. The antisense riboprobe for rat CRABP(II) and cyclophilin was transcribed using the MAXIscript kit (Ambion Inc., Austin, TX) and [␣-32 P]UTP (10 Ci/ml; PerkinElmer Life Sciences). The RPAs were carried out using the RPA III kit (Ambion Inc., Austin, TX) according to the user's manual. Briefly, samples of total RNA were hybridized for 15-18 h at 50°C with excess radiolabeled antisense riboprobe (n ϭ 3 individual samples/time point) and digested by RNase at 37°C for 30 min. The hybridized products were submitted to electrophoresis on 6% acrylamide gels containing 8% urea. Gels were exposed to BioMAX MR film (Eastman Kodak Co.) with intensifying screens for up to 3 days. Loading variation between samples was standardized by including cyclophilin riboprobe in all hybridization reactions.
Cloning, Southern Blotting, and Preparation of DNA Constructs-A portion of the rat CRABP(II) gene was cloned from a P1 phage library (Incyte Genomics Inc., Palo Alto, CA). The library was screened using PCR primers that amplified a portion of exon 1. The positions of the primers were: 5Ј primer, ϩ4 to ϩ84, and 3Ј primer, ϩ162 to ϩ182, numbered according to the transcription start site. One positive clone was identified. It was digested with EcoRI or HindIII, and subsequent Southern blotting analysis identified a 3.5-kb EcoRI fragment and a 14-kb HindIII fragment by hybridizing with the 32 P-labeled probe, obtained by PCR labeling using the above primers for exon 1. The 14-kb fragment was purified and cloned. Restriction enzyme mapping and sequencing revealed that it contained 6.6 kb of the 5Ј-flanking region of the rat CRABP(II) gene. This 14-kb fragment was further digested with kpnI restriction enzyme to obtain a 4466-base fragment that contained the basic promoter region. This was ligated to the pGL3 basic luciferase reporter vector (Promega Co., Madison, WI). Further deletions of the Ϫ1354, Ϫ1211, Ϫ1182, Ϫ504 fragments were obtained by PCR amplification using the same reverse primer-added KpnI restriction enzyme site, (55/78) 5Ј-GGTACCGTACCTTGCTGTCCCCTT-3Ј and the forward primer-added KpnI restriction enzyme site, (Ϫ1354/Ϫ1337), 5Ј-GGTAC-CGCGATCCAGAAGCCCTTC-3Ј (Ϫ1210/Ϫ1193), 5Ј-GGTACCCAGTT-CGACCCTCCACC-3Ј (Ϫ1177/Ϫ1160), 5Ј-GGTACCCACCAAAGCCTG-TCAGTC-3Ј (Ϫ504/Ϫ487), 5Ј-GGTACCGAACAGAGCGACACCTCC-3Ј, respectively. Each PCR product was digested by KpnI and cloned into the corresponding sites of the pGL3 basic vector. The right orientation was identified by PCR screening. Constructs 6 and 7 were generated by ligating the synthetic fragment of Ϫ1211/Ϫ1178 to the fragments of Ϫ799 or Ϫ754, respectively (Fig. 4). All the constructs created by PCR amplification were verified by sequencing.
Cell Culture, Transient Transfection, and Luciferase Assay-MCF-7 cells obtained from the American Type Culture Collection (Manassas, VA) were routinely maintained in a humidified atmosphere containing 5% CO 2 . Cells were transiently transfected when they approached 70% confluence in 24-well plate using the SuperFect transfection reagent (Qiagen Inc., Valencia, CA) according to the manufacturer's instructions. The cells were cotransfected with 1 g of estrogen receptor (ER) expression plasmid (ATCC) as the endogenous level of ER was low, and with 10 ng of pRL-TK vector to normalize the differences in transfection efficiencies. Following transfection, cells were changed to phenol redfree minimum Eagle's medium (Invitrogen) containing 10% dextrancoated charcoal-stripped fetal bovine serum, and cells were incubated with 100 nM E 2 or vehicle for 2 days before harvesting for the luciferase assay. Cells were lysed using 1ϫ passive lysis buffer, and 20 l of lysate was assayed using Dual-Luciferase reporter assay system (Promega).
Electrophoretic Mobility Shift Assays (EMSA)-All the double strand oligonucleotides used for the EMSAs were prepared and high pressure liquid chromatography-purified by Qiagen Operon (Alameda, CA). The sequence of each sense oligonucleotide is shown in Table I. Oligonucleotides were labeled with [␥-32 P]ATP (10 Ci/ml) and T4 polynucleotide kinase. The EMSAs were performed as described previously (15). Briefly, 0.1-0.3 g of the purified active form of ER protein (PanVera Co., Madison, WI) was first incubated in 20 l of binding buffer containing 3 mM EDTA, 1 mM dithiothreitol, 2 g of poly(dI-dC), and 4% Ficoll 400 at room temperature for 10 min, and then 1 ng of probe was added, and the sample was incubated another 15 min. For the competition experiments and supershift assay, 100 -800-fold excess of competitor or 1-4 g of ER antibody (PanVera Co., Madison, WI) was added to the reaction mixture, and the incubation was further continued at room temperature for 20 min. The reaction mixtures were submitted to electrophoresis on 4% polyacrylamide native gels. After running, the gel was dried and exposed to the Kodak film overnight.

Expression of CRABP(II) in Rat Uterus Was Induced
Directly by E 2 but Not by RA-Ovariectomized rats were administered either E 2 or RA or vehicle alone and uteri were collected after 4 h. Uteri were also collected from ovariectomized rats that had received puromycin 30 min prior to administration of E 2 , RA, or vehicle. Analysis of the RNA obtained from these uteri by RPA revealed a potent induction (16 -18-fold) of CRABP(II) message by E 2 administration (Fig. 1). Puromycin had no effect on this increase, suggesting that the response to E 2 did not require new protein synthesis and was sufficiently rapid to be a direct response.
Surprisingly, no increase in CRABP(II) message in the uterus was observed for animals administered RA, in contrast to observations for both the human and mouse CRABP(II) gene, albeit in different systems (11,16). No increase was noted even at times as late as 48 h after RA administration (data not shown). To confirm that RA had indeed reached the uterus, the   (17), was determined and found to have increased about 5-fold in the animals receiving RA (Fig. 1). Thus, the lack of response of the rat CRABP(II) gene to RA could not be ascribed to delivery failure. It was considered possible that an RA-induced expression might require exposure of the uterus to E 2 . To test this possibility, we injected ovariectomized rats with E 2 and RA simultaneously and also provided RA 24 h after E 2 treatment. When message levels were determined 4 h later, no effect of RA was observed over that of E 2 alone (results not shown).
Analysis of the 5Ј-flanking Region of the Rat CRABP(II) Gene-Previous analyses of the 5Ј-flanking regions of the mouse and human CRABP(II) gene had not revealed any indication of an ERE. To see whether there might be a species difference in estrogen response, we isolated this region for the rat gene, as described under "Experimental Procedures." A 3.5-kb EcoRI fragment and a 14-kb HindIII fragment were identified ( Fig. 2A). Restriction enzyme mapping analysis revealed that the 14-kb fragment contained about 6.6 kb of the 5Ј-flanking region (Fig. 2B). Digestion with KpnI produced a fragment that contained the TATA box and transcription start site but excluded the coding region.
The complete sequence of this 4.5-kb fragment was analyzed for known DNA binding protein recognition sites. The palindromic sequences of the canonical ERE (18) were not found. Putative regulatory elements for ubiquitous transcription factors such as AP1 (19), AP2 (20), and Sp1 (21) were evident in the proximal region of this fragment (Fig. 3). The proximal region shown corresponds to the published sequence of the mouse CRABP(II) 5Ј region, with which it has 85% sequence identity (13). Although we found no response of the CRABP(II)

FIG. 2. Isolation of the 5-flanking region of the CRABP(II) gene.
After a positive clone was digested with either EcoRI or HindIII, the digestion mixtures were separated on a 0.8% agarose gel. After transfer to membrane, a 3.5-and 14-kb fragment were identified from these two restriction enzymes, respectively, by using a partial sequence of exon I as probe (A). From further restriction enzyme mapping of the 14-kb fragment (shown in B), the fragment generated by kpnI was selected for further study. gene to RA in the uterus of the intact rat, there is a region corresponding to the mouse RA-response element mRARE2 but no correspondence to the reported mRARE1 (Fig. 3).
Promoter Activity of the 5Ј-flanking Region of the Rat CRABP(II) Gene-To determine whether the cloned 5Ј-flanking region of the rat CRABP(II) gene contained unidentified response elements allowing transcriptional activation by E 2 , various 5Ј end deletions were ligated into the pGL3-basic luciferase reporter vector (Fig. 4). The cell line chosen for testing these constructs was the MCF-7 human mammary carcinoma cell line as it expresses both CRABP(II) and the ER and, consequently, should contain any cell-specific factors involved in CRABP(II) expression. However, preliminary studies indicated that a better response was obtained if an ER expression vector was cotransfected and all experiments included that vector.
The longest fragment tested contained, with respect to the transcription start site, the region from Ϫ4411 to ϩ77, ending 72 nucleotides prior to the initiation codon for translation (construct 1). After treatment with E 2 for 48 h, the transcription activity of this fragment was about 13-fold higher than the basal promoter level (Fig. 4). Treatment with 1 M RA gave no increase in transcription (data not shown). Various deletions from the 5Ј end of this fragment were examined, and induction remained at this level until Ϫ1210 (constructs 2 and 3). Deletion constructs shorter than Ϫ1210 were substantially lower, falling to a 2-3-fold increase over basal levels (constructs 4 and 5), indicating that the Ϫ1210 to Ϫ1178 region was critical for E 2 response. The sequence of this region contained an imperfect consensus ERE, GCTCANNNCGACC, and an ERE halfsite, TGTCA (Table I).
To examine whether the GC-rich regions, putative Sp1 binding sites, might also contribute to the ability of E 2 to induce transcription, as has been observed for other genes, the Ϫ1210 to Ϫ1178 sequence was ligated to the proximal promoter region at position Ϫ799, just proximal to the second GC-rich sequence at Ϫ781 to Ϫ755 (construct 6). Induction was similar to that seen with the full-length construct (ϳ8-fold versus ϳ12-fold.). However, ligation of the Ϫ1210 to Ϫ1178 fragment to the proximal region at position Ϫ754 reduced induction to the 2-3-fold increase seen when the Ϫ1210 to Ϫ1178 fragment was not present (construct 7; compare with construct 4). This suggested that the fragment Ϫ1210 to Ϫ1178 and the region Ϫ799 to Ϫ754 are both required for the E 2 -induced transcription of the rat CRABP(II) gene. The possible additional requirement of the more proximal GC-rich region was not examined.
We further examined the response of this promoter region to E 2 in the rat hepatoma cell line H-4-II-E, which does not express CRABP(II) or the ER. The results were essentially the same for each construct as found in the MCF-7 cell line (data not shown), which suggested that no factor specific to a cell type was required in this in vitro deletion assay.
Test of the Putative ERE by ER Binding Studies-Known EREs are able to bind activated ER in vitro, as shown by electrophoretic mobility shift assays (EMSA). The Ϫ1210 to Ϫ1178 fragment, designated EREc (Table I), was tested for the ability to bind purified, activated ER. As a control, a sequence containing a canonical ERE was also used (wtERE).
When ER was incubated with 32 P-labeled oligonucleotide EREc, a protein⅐DNA complex was formed, as revealed by EMSA (Fig. 5A, lane 1). This protein⅐DNA complex was confirmed to be an ER⅐DNA complex by the ability of ER antibody to supershift the complex in a dose-dependent manner (Fig. 5A,  lanes 2-5).
As a further test of EREc as a potential ERE, gel mobility competition assays were performed with 100 -400-fold molar excess of unlabeled wtERE over 32 P-labeled EREc (Fig. 5B) and 200 -800-fold molar excess of unlabeled EREc over 32 P-labeled wtERE (Fig. 5C). A 400-fold excess of wtERE competed for all detectable binding of labeled EREc, whereas an 800-fold excess of EREc was required to compete for all detectable binding of wtERE. This suggested that the affinities of the ER for these sequences were similar but that the wtERE appears to bind more tightly to the ER. As a final test of specificity of binding of the ER to the EREc sequence, competition was tested for a 3-base mutant, EREm, with two base changes in the imperfect ERE region and one in the half-site (Table I). No apparent competition was observed for either a 400-or 800-fold excess. These results established that the Ϫ1120 to Ϫ1178 contained an imperfect ERE that can bind specifically to ER in vitro.

DISCUSSION
Previous work from our laboratory has shown that the expression of CRABP(II) in certain cells of the rat uterus and ovary correlates with the production of RA by those cells (2,22). Recent studies by others have demonstrated an ability of the RA⅐CRABP(II) complex to translocate to the nucleus and mediate a direct transfer of RA to RARs, an ability not shared by the closely related protein CRABP (6,7). The association of CRABP(II) with the nuclear receptor complex has been demonstrated to enhance the expression of RA-responsive genes (23).
In the work presented here, the demonstration that E 2 directly induced the expression of CRABP(II) indicates that the effects of E 2 on a particular tissue/cell may well result in the induction or modulation of expression of RA-responsive genes, in addition to E 2 -responsive genes. This would appear to greatly increase the number of genes accessible by the E 2 signal and provides a direct link between the action of a steroid hormone and the action of RA.
Interestingly, we saw no change in expression of the endogenous CRABP(II) gene or of the reporter constructs when RA was provided. Previously, we had observed that E 2 administration to the prepubertal female rat led to acquisition of the ability of the lining epithelium to synthesize RA as well as to express CRABP(II). That expression of CRABP(II) would then appear to be independent of RA synthesis by those cells. This result is different from other studies in which CRABP(II) gene expression was strongly increased by RA either in human skin in vivo or in cultured human skin fibroblasts in vitro (24). RA also was shown to increase expression of CRABP(II) in the F9 murine cell line (16). These differences might be explained by species or cell-type differences. However, it was observed that this increase of expression in both intact human skin and in the F9 cell line was blocked by inhibition of protein synthesis, suggesting that the effect was indirect. Nuclear run-on experiments suggested that the increase was controlled by a posttranscriptional mechanism (25). Still, we observed no increase in CRABP(II) expression at times up to 48 h, suggesting that even an indirect regulation by RA was not occurring in the rat uterus under the conditions examined here.
Analysis of the 5Ј-flanking region of the rat CRABP(II) gene showed 85% sequence identity to the proximal 1 kb of the corresponding sequence reported for the mouse gene, including an identical DR1 mRARE2 repeated motif (13) and some ubiquitous transcription factor binding sites, such as Sp1, AP1, and AP2 ( fig. 3), but there was no sequence corresponding to an unidentified functional DR2 mRARE1. It should be noted that this region of the mouse promoter did lead to RA-stimulated transcription of a reporter gene in transfected cells, whereas we did not observe such an increase. This may explain why RA has no effect on rat CRABP(II) gene expression in vivo or when transfected in MCF-7 cells in vitro and may suggest a species difference in regulation by RA. However, it should be stressed that the actual direct regulation of the native gene in both human and murine cells by RA remains in question.
No obvious ERE was noted in the analysis of the 4.5-kb fragment, but it is well established that E 2 -directed gene regulation can be quite complex. Many naturally occurring copies of the 13-base pair consensus ERE are imperfect (26 -31). Also, several genes activated by E 2 involve both ERE and GC-rich Sp1 binding sites, such as creatine kinase B (32), bcl-2 (33), uteroglobin (34), insulin-like growth factor-binding protein-4 (35), transforming growth factor ␣ (36), RAR␣ (37), and LDL receptor (38), or ERE interaction with USF-1 and USF-2 as in the cathepsin D promoter (39). That proved to be the case here as well. The identified region that bound the ER contained three imperfect half-palindromic EREs, not obvious from sequence analysis alone. In addition, a possible Sp1 binding site was required to obtain the E 2 response in vitro. Thus, this appears to be another version of a complex E 2 -responsive promoter.
Our preliminary studies 2 indicate that E 2 also directly regulates expression of the enzymology of RA synthesis. It will be of interest to examine whether this regulation shares features observed for the CRABP(II) gene.