Transcriptional regulation of the hepatocyte growth factor gene by the nuclear receptors chicken ovalbumin upstream promoter transcription factor and estrogen receptor.

Hepatocyte growth factor (HGF) is a multifunctional cytokine that controls the growth and differentiation of various tissues. Previously, we described the existence of a negative cis-acting regulatory element(s) within the −1- to −0.7-kilobase pair (kb) portion of the 5′-flanking region of the mouse HGF promoter. In the present study, we show that the repressor element is located at position −872 to −860 base pairs and comprises an imperfect estrogen-responsive element 5′-AGGTCAGAAAGACCA-3′. We demonstrate that chicken ovalbumin upstream promoter transcription factor (COUP-TF), a nuclear orphan receptor belonging to the steroid/thyroid hormone receptor superfamily, through binding to this site effectively silences the transcriptional activity of the HGF promoter. We show that estrogen receptor, on the other hand, relieves the repressive action of COUP-TF, resulting in the induction of the HGF promoter. Using mice transgenic for either 2.7 or 0.7 kb of the HGF promoter region linked to the chloramphenicol acetyltransferase reporter gene, we found that injection of estradiol stimulates HGF promoter activity in tissues such as the mammary gland and ovary of mice harboring 2.7 but not 0.7 kb of the mouse HGF promoter region. Potential involvement of the CCAAT/enhancer-binding protein (C/EBP) family of transcription factors in the regulation of HGF gene expression is also discussed.

Hepatocyte growth factor (HGF) is a multifunctional cytokine that controls the growth and differentiation of various tissues. Previously, we described the existence of a negative cis-acting regulatory element(s) within the ؊1to ؊0.7-kilobase pair (kb) portion of the 5-flanking region of the mouse HGF promoter. In the present study, we show that the repressor element is located at position ؊872 to ؊860 base pairs and comprises an imperfect estrogen-responsive element 5-AGGTCAGAAAGACCA-3. We demonstrate that chicken ovalbumin upstream promoter transcription factor (COUP-TF), a nuclear orphan receptor belonging to the steroid/thyroid hormone receptor superfamily, through binding to this site effectively silences the transcriptional activity of the HGF promoter. We show that estrogen receptor, on the other hand, relieves the repressive action of COUP-TF, resulting in the induction of the HGF promoter. Using mice transgenic for either 2.7 or 0.7 kb of the HGF promoter region linked to the chloramphenicol acetyltransferase reporter gene, we found that injection of estradiol stimulates HGF promoter activity in tissues such as the mammary gland and ovary of mice harboring 2.7 but not 0.7 kb of the mouse HGF promoter region. Potential involvement of the CCAAT/enhancer-binding protein (C/ EBP) family of transcription factors in the regulation of HGF gene expression is also discussed.
Hepatocyte growth factor (HGF) 1 was originally identified as the major mitogenic component of serum and platelets for normal adult rat hepatocytes (1,2). Further investigations revealed that HGF is a multifunctional cytokine with a broad range of target cells (3,4). HGF elicits its biological effects through binding to and activating a specific transmembrane tyrosine kinase (4 -6) known as Met (HGF receptor) (7). HGF gene expression is tightly regulated at the transcriptional level and is restricted to stromal cells such as fibroblasts, macrophages, leukocytes, and endothelial cells in various tissues. Numerous studies have shown that this growth factor is involved in the growth and regeneration of several organs (8 -10). In response to tissue loss, such as in the case of partial hepatectomy or hepatotoxin injury, the level of HGF mRNA dramatically increases in the liver as well as in other organs such as the lung and spleen (9,10). HGF mRNA also increases in the remaining kidney or lung after unilateral nephrectomy or pneumonectomy, respectively (9,10). Recent investigations in transgenic mice homozygous for a null mutation in the HGF gene (knock-out mice) have proved that HGF is essential for the development of various tissues such as liver, muscle, and placenta (11,12). HGF expression is modulated by several cytokines and hormones. For example, it has been shown that in MRC-5 fibroblasts and HL-60 cells HGF expression is down-regulated by transforming growth factor ␤1 and dexamethasone or by coculture with epithelial cells (13)(14)(15). In contrast, phorbol esters, interleukin-1, or tumor necrosis factor ␣ increase HGF mRNA level (16 -19). Overexpression of HGF mRNA in various neoplastic tissues has also been reported. To understand the transcriptional regulation of HGF, the sequences of the human, rat, and mouse HGF gene promoters have been determined and the major transcriptional start sites have been mapped (20 -23). Several potential regulatory sites including NF-IL6 (also known as C/EBP␤) were identified in the 5Ј-flanking region of the mouse HGF promoter by our laboratory (22) as well as by others (20 -24). We reported the identification and partial characterization of a transcriptional repressor in the promoter of the mouse HGF gene (25). We also localized a negative regulatory region in the 5Ј-flanking region of the mouse HGF promoter residing at position Ϫ1 to Ϫ0.7 kb (22). In the current study, we show that COUP-TF binds to an imperfect estrogen response element (ERE) in this region and represses HGF promoter activity. We show that high expression of the estrogen receptor counteracts the suppressive action of COUP-TF by competing for the same binding site.
COUP-TF, which belongs to the orphan nuclear receptor family, was originally found through its interaction with a response element in the chicken ovalbumin gene promoter (26,27) and has been shown to play both positive and negative roles in gene regulation upon binding to its cognate DNA regulatory element (28). The COUP-TF family (which consists of COUP-TF I and COUP-TF II) binds to the AGGTCA repeat sequence, which also is the DNA half-site for several other nuclear receptors such as the estrogen receptor (ER), vitamin D 3 receptor, thyroid hormone receptor, retinoic acid receptors (RAR and RXR), peroxisome proliferator-activated receptor, and hepatic nuclear factor 4 (HNF4) (28 -31). In previous studies we reported that estrogen stimulates HGF mRNA expression in mouse ovaries (32). Our current results implicate COUP-TF as a potent repressor of HGF gene transcription and provide new insights into the molecular mechanisms that regulate HGF gene expression.

EXPERIMENTAL PROCEDURES
Plasmid Construction--The 1.4 mouse HGF-CAT (Ϫ1386 to ϩ29), 1.0 mouse HGF-CAT (Ϫ964 to ϩ29), and 0.7 mouse HGF-CAT (Ϫ699 to ϩ29) constructs were prepared as described (22). The 0.85 mouse HGF-CAT construct was made by double digesting the 1.4 mouse HGF-CAT construct with HindIII/AflII. The 0.85 mouse DHGF-CAT construct harboring an internal deletion was generated by digesting the 1.0-kb HGF-CAT construct with AflII and BstXI. Blunt ends were created by either Klenow polymerase or T4 DNA polymerase, and the corresponding fragments were purified by agarose gel electrophoresis and then ligated. For heterologous promoter constructs, synthesized oligonucleotides HERE1 and HERE2 were inserted into the BglII site of the pCAT promoter (32).
Cell Culture, DNA Transfection, and CAT Assay-Human endometrial carcinoma RL95-2 cells, human hepatoma HepG2 cells, and mouse fibroblast NIH3T3 cells were obtained from American Type Culture Collection (Rockville, MD) and cultured as described previously (22). For transfection, the recipient cells were cultured in six-well plates for 24 h and then transfected with various mouse HGF promoter/CAT chimeric plasmids and heterologous promoter constructs using the DNA calcium phosphate method according to the instructions of the CellPhect transfection kit (Pharmacia Biotech Inc.). The ␤-galactosidase reference plasmid pCH110 (Pharmacia) was used as an internal standard for transfection efficiency. After cell transfection was carried out for 16 h, the cells were washed twice with serum-free medium. Complete medium containing 10% fetal calf serum was then added, and the cells were incubated for an additional 24 h before harvesting to test for CAT activity.
For co-transfection experiments of NIH3T3 cells with pHEO, the human estrogen receptor expression plasmid, and pRS-COUP, the human COUP-TF expression plasmid, phenol red-free medium and charcoal-stripped fetal calf serum were used. The amount of plasmid used in transfection was 5 g of one chimerical CAT construct, 1 g of pCH110, and different amounts of estrogen receptor or COUP-TF expression vectors as indicated. Protein concentration was used as a standard for transfection efficiency in the co-transfection experiments including those with the COUP-TF expression vector. CAT activity was determined by a modified assay as described previously (22). Transfection experiments were performed at least three separate times with two independent preparations of purified plasmid DNA.
Preparation of Nuclear Extracts-Human endometrial carcinoma RL95-2 cells, human hepatoma HepG2 cells, and mouse fibroblast NIH3T3 cells were originally obtained from ATCC. Cells were cultured under conditions described previously (22,32). The nuclear extracts were prepared as already reported (25). Briefly, cells growing to about 90% confluence were washed with cold phosphate-buffered saline and dislodged with a rubber policeman into phosphate-buffered saline. Cells were pelleted by low speed centrifugation, resuspended in 5 volumes of Buffer A containing protease inhibitors (10 mM HEPES (pH 7.9), 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM phenylmethylsulfonyl fluoride, 0.15 mM spermidine, 0.5 mM spermine, 0.5% Nonidet P-40, and 0.5 mM dithiothreitol, plus 1 g each of leupeptin and chymostatin per ml) and then placed on ice for 10 min. Immediately after lysis, the solution was transferred to a glass Dounce homogenizer and homogenized with 10 up and down strokes using a type B pestle. The nuclei were collected by centrifuging for 15 min at 3,300 ϫ g and resuspended in the same volume of Buffer B as Buffer A. Buffer B was the same as Buffer A except Nonidet P-40 was omitted. After the nuclei were recovered by centrifugation, Buffer C (20 mM HEPES (pH 7.9), 10% glycerol, 1.5 mM MgCl 2 , 10 mM KCl, 0.2 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, plus 1 g each of leupeptin and chymostatin per ml) was added to resuspend the nuclei. Nuclear protein was extracted with Buffer D (the same as Buffer C, but substituting 10 mM KCl for 1.6 M KCl) for 1 h on ice. The extracted nuclei were centrifuged for 45 min at 50,000 ϫ g at 4°C, and the supernatant was then collected and dialyzed against 60 mM KCl-TGM (10 mM Tris-HCl (pH 7.6), 10% glycerol, 3 mM MgCl 2 , 3 mM EGTA buffer using a mini-dialysis system (Life Technologies, Inc.). The insoluble materials were removed by centrifugation, and small aliquots of protein extract were quickly frozen and stored at Ϫ80°C after the protein concentration had been determined by protein assay (Bio-Rad).
Rat liver nonparenchymal cells were isolated by differential centrifugation of the collagenase-digested rat liver perfusate (provided by George Michalopoulos, Department of Pathology, University of Pittsburgh). Cells were washed twice with cold phosphate-buffered saline and then used to make nuclear extracts with the method described above.
Oligonucleotides-The following oligonucleotides were used in gel mobility shift assays: vitellogenin A2 ERE (VitA2ERE), 5Ј-GATCTAG-GTCACAGTGACCTA-3Ј (33); and COUP, 5Ј-GATCTTTCTATGGTGT-CAAAGGTCAAACT-3Ј (30). The following sequences were derived from the mouse HGF gene promoter (22) Gel Retardation Assays-A DNA fragment corresponding to nucleotide position Ϫ964 to Ϫ812 bp of the 5Ј-flanking region of the mouse HGF gene was isolated from the chimeric 1.4 mouse HGF-CAT construct (22) by restriction digestion with AatI/AflII. The 154-bp fragment was purified and labeled with [␣-32 P]dATP (3,000 Ci/mmol, Amersham Corp.) by filling in and blunt ending with Klenow enzyme (Life Technologies, Inc.). The double-stranded oligonucleotides used in gel shift assays were also labeled with 32 P by end labeling with T4 kinase. The labeled probes were then gel purified and used in gel mobility shift assays as described previously (25). Two micrograms of poly(dI-dC) (Pharmacia) were used as the nonspecific competitor in 10 l of reaction mixture. When antibodies were used in supershift experiments, they (1 l) were incubated with nuclear extracts for 15 min at room temperature before carrying out DNA binding shift assays. The binding reactions (20 l total volume/reaction) were carried out at room temperature for 20 min before loading on 5% nondenaturing polyacrylamide (19:1 acrylamide/bisacrylamide) gels. The concentration of nuclear protein extracts used in each reaction was 2 g and that of the labeled probe was between 0.2 and 0.4 ng. For competition experiments, a 100-fold molar excess of unlabeled DNA fragments or oligonucleotides was included in the reaction mixture. Gels were run in 0.5 ϫ TBE buffer (0.045 M Tris borate, 0.001 M EDTA) at a constant voltage of 190 V, dried, and autoradiographed using intensifying screens.
Transgenic Mice-Transgenic mice harboring various lengths (2.7, 0.7, and 0.1 kb) of the HGF promoter region fused to the CAT reporter gene were recently generated and characterized in our laboratory. 2 Two different HGF-CAT transgenic mouse lines were used in the present study to evaluate HGF promoter activity in response to estradiol. The 2.7 HGF-CAT transgenic mouse lines (three different founders) harbor 2.7 kb of the HGF promoter that contains the HERE1 element. The 0.7 HGF-CAT transgenic lines (two different founders) carry 0.7 kb of the HGF promoter region that lacks the HERE1 element. Transgenic mice were maintained in animal facilities at the University of Pittsburgh Medical Center. All animal experiments were conducted in accordance with National Institutes of Health standards established by the "Guidelines for the Care and Use of Experimental Animals." Mice were injected subcutaneously with 100 mg of 17␤-estradiol (Sigma) per kg of body weight in a volume of 0.1 ml of buffered saline containing 0.2% ethanol. At 24 h after injection, groups of eight animals were killed by cervical dislocation, and the ovary and other organs were dissected and immediately frozen for CAT assays.

RESULTS
Identification of a cis-Acting Repressor in the HGF Promoter-We previously reported the existence of a negative regulatory region in the 5Ј-flanking region of the mouse HGF gene located between nucleotides Ϫ964 and Ϫ699 bp (22). To further define this region, two additional HGF-CAT chimeric constructs containing a 5Ј deletion (0.85 HGF-CAT) or an internal deletion (0.85 DHGF-CAT) were generated by using the restriction enzyme AflII (Fig. 1). Four different HGF-CAT chimeric constructs, 1.0 HGF-CAT, 0.85 HGF-CAT, 0.85 DHGF-CAT, and 0.7 HGF-CAT, were subsequently tested for promoter activity by transfection into human endometrial carcinoma RL95-2 cells, human hepatoma HepG2 cells, and mouse fibroblast NIH3T3 cells. Promoter activity of each construct as determined by CAT assays revealed that the 0.7 and 0.85 HGF-CAT constructs had consistently stronger promoter activity than the 0.85 DHGF-CAT and the 1.0 HGF-CAT constructs (Fig. 1). These results suggest that the region between nucleotides Ϫ964 and Ϫ812 bp contains a negative cis-acting element(s) and that a negative trans-acting factor(s) may interact with this 154-bp DNA fragment to repress HGF gene transcription.
trans-Acting Factors Interact with the cis-Acting Negative Element in a Distinct Binding Region-To examine whether trans-acting factors specifically bind to the 154-bp DNA fragment, we performed gel mobility band shift analysis using the end-labeled 154-bp fragment with nuclear protein extracts from NIH3T3 cells. Examination of the DNA-protein complexes by polyacrylamide gel electrophoresis under nondenaturing conditions revealed a prominent and specific band with a slow electrophoretic mobility. This complex was totally abrogated in the presence of 100-fold molar excess of the unlabeled "self" probe competitor (Fig. 2, lane 3). Since computer analysis of the nucleotide sequence of this DNA fragment identified consensus binding sites such as ERE and CRE, we carried out gel mobility band shift assays in the presence of synthetic oligonucleotides corresponding to the individual consensus sequences HERE1 and CRE and other potential binding sites that are present in the 5Ј-flanking region of the mouse HGF. As shown in Fig. 2, the binding was completely abolished when a 100-fold molar excess of the HERE1 oligonucleotide (Ϫ872 to Ϫ860 bp) was added to the reaction mixtures (Fig. 2, lane 4). On the other hand, the CRE oligonucleotide at a 100-fold molar excess did not compete for this binding site (Fig. 2, lane 6). Other oligonucleotides found in the mouse HGF promoter corresponding to GRE, TIE, AP1, IL-6RE, NF-IL6, and HERE2 (a mutated version of HERE1) also failed to compete with the 154-bp DNA fragment for binding to the nuclear protein present in NIH3T3 cells (Fig. 2, lanes 5 and 7-11). We confirmed these results by DNase I footprinting analysis using NIH3T3 cells and mouse liver tissue nuclear protein extracts. Taken together, these data suggest that the HERE1 (nucleotides Ϫ872 to Ϫ860, 5Ј-GGTCAGAAAGACC-3Ј) is involved in the formation of the protein-DNA complex. This binding sequence is similar to the estrogen response element (5Ј-AGGTCANNNTGACCT-3Ј, see also Fig. 3).
To assess the functionality of this element, synthetic HERE1 and HERE2 (a mutated version of HERE1, 5Ј-GGTCAATCTA-ACC-3Ј) oligonucleotides were linked to the heterologous pCAT promoter reporter gene (Promega), which contains the SV40 promoter, and the promoter activity in three different cell lines using transient transfection and CAT assays was determined. The results are summarized in Fig. 3. HERE1 efficiently repressed the SV40 promoter activity in HepG2, RL95-2, and NIH3T3 cells, whereas HERE2 (which failed to specifically bind to the nuclear protein, Fig. 2, lane 11) did not have any function. These results indicate that HERE1 negatively modulates HGF promoter activity and that this element functions not only in the context of the autologous promoter ( Fig. 1) but also in the context of a heterologous promoter (Fig. 3).
Identification of the Binding Protein(s) as COUP-TF- To   FIG. 1. Localization of a cis-acting repressor element in the 5-upstream flanking region of the mouse HGF promoter. Mouse HGF-CAT constructs consisting of a series of 5Ј deletions or an internal deletion were transiently transfected into NIH3T3, HepG2, and RL95-2 cells. Graphical representation of relative CAT activity after normalization against ␤-galactosidase is shown. The results presented here are the average from at least three independent experiments performed in duplicate. The transfected plasmids were from two independent preparations. Numbers indicate the insert sizes (kb) of the mouse HGF promoter region 5Ј to the CAT reporter gene. DNA fragments were inserted in correct (5Ј to 3Ј) orientation relative to the transcription initiation site, and the nucleotide boundaries were confirmed by sequencing.

FIG. 2. Specific interaction of the 5-flanking region of the mouse HGF gene with a negative regulatory activity in nuclear protein extracts.
The 154-bp DNA fragment that exerted negative regulatory function was labeled with 32 P and incubated with nuclear protein extract (NPE) from NIH3T3 cells. The binding reactions were performed at room temperature for 20 min in the presence of the nonspecific competitor poly(dI-dC). Several double-stranded synthetic oligonucleotide sequences including HERE1, GRE, CRE, TIE, AP1, IL-6RE, NF-IL6, and HERE2 were added to the reaction mixture to compete for binding as indicated. C, binding complex; F, free 32 P-labeled DNA probe. identify the HERE1 binding protein(s), labeled HERE1 was used as a probe in gel mobility shift assays with the nuclear extracts from NIH3T3 and RL95-2 cells. In NIH3T3 nuclear extract, the specific complex was abolished by unlabeled HERE1, VitA2ERE (wild type ERE), and COUP binding sites (Fig. 4A, lanes 3-5) but not by HERE2 and GRE (Fig. 4A, lanes  6 and 7). We used COUP binding site because it is well known that COUP-TF binds to the AGGTCA motif regardless of its configuration (single half-site, direct as well as indirect repeats, and various spacings) (29). Although RL95-2 cell nuclear extract formed two different bands with the HERE1 probe, these bands were totally abolished by 100-fold molar excess of unlabeled HERE1, VitA2ERE, and COUP oligonucleotides (Fig. 4A, lanes 9 -11) but not by HERE2 and GRE (Fig. 4A,  lanes 12 and 13). It is known that RL95-2 and NIH3T3 cells express small amounts of ER, and that the major function of ER in the transcriptional regulation of target genes is induction (30,32). Given the fact that COUP-TF mainly exerts repression on its target genes through binding to VitA2ERE (30,35,36), we opted to determine if the protein complex formed by HERE1 is COUP-TF. To test this possibility, we performed supershift assays using antibody against COUP-TF, estrogen receptor, or nonimmune serum using the labeled HERE1 as probe. As shown in Fig. 4, anti-COUP-TF antibody totally abrogated complex formation between the HERE1 probe and nuclear proteins from different sources such as NIH3T3, HepG2, RL95-2 (Fig. 4B), and nonparenchymal liver cells (Fig.  4C) that are known to express the HGF gene. The anti-ER antibody and nonimmune control serum did not show any effects (Fig. 4, B and C). These results demonstrate that the orphan nuclear receptor COUP-TF binds to the HERE1 site of the HGF promoter region. It is of interest to note that HERE1 and HERE2 also bind to the estrogen receptor if this receptor is highly expressed (32). We previously demonstrated this by using extract from RL95-2 cells that were transfected with an ER expression vector (32). Taken together, our findings indicate that HERE1, like VitA2ERE, interacts with both ER and COUP-TF.
Modulation of HGF Promoter Activity by COUP-TF and ER-To test whether the estrogen receptor relieves the repression of the HGF promoter activity mediated by COUP-TF, we co-transfected various HGF-CAT chimeric constructs with an estrogen expression vector into NIH3T3 cells. As shown in Fig.  5, the promoter activity of the 0.85 DHGF-CAT and 1.0 HGF-CAT constructs that contain HERE1 increased in the presence of ER in a dose-dependent manner. However, the activity of the 0.85 HGF-CAT and 0.7 HGF-CAT constructs were not affected by ER since they do not contain the HERE1 region (Fig. 5). Similar results were obtained when HERE1 was tested for its responsiveness to ER in the context of the SV40 heterologous promoter. CAT assays of NIH3T3 cell extracts co-transfected with the chimeric constructs HERE1-SV40-CAT and HERE2-SV40-CAT revealed that addition of increasing amounts of the estrogen receptor expression vector not only overcame the suppressive function conferred to the SV40 promoter by COUP-TF bound to HERE1 but also highly stimulated SV40 promoter activity (Fig. 6A). Interestingly, HERE2 did not suppress the SV40 promoter (since it does not bind COUP-TF; see Fig. 2, lane 11) but it bound ER with low affinity as we reported previously (32). Furthermore, overexpression of COUP-TF (pRS-COUP) in NIH3T3 cells counteracted the stimulatory function of ER on HERE1 but had no effect on the HERE2-SV40-CAT construct (Fig. 6B). These results demonstrate that the mouse HGF gene is transcriptionally suppressed by COUP-TF via binding to the HERE1 element at nucleotide Ϫ872 bp in the 5Ј-flanking region of the mouse HGF gene. High expression of ER counteracts the suppressive function of COUP-TF by competing with ER for binding to HERE1.
Stimulation of the Mouse HGF Promoter in Transgenic Mice by Estradiol-We previously showed that injection of estradiol into mice dramatically stimulated HGF mRNA expression in mouse ovary (32). Our present data reveal that the molecular mechanism by which estradiol stimulates the HGF promoter is through the counteraction by ER of COUP-TF, which directly down-regulates the mouse HGF promoter. Only high levels of ER counteract the suppression caused by COUP-TF bound to HERE1. To further confirm that HERE1 is functional and responsible for the stimulation of HGF mRNA by estradiol in vivo, we analyzed transgenic mice that harbor 2.7 and 0.7 kb of the mouse HGF promoter linked to the CAT reporter gene. The activity of the 2.7 HGF-CAT promoter construct in transgenic mice that carries the HERE1 element was up-regulated in the mammary tissue and ovary after injection of estradiol (Fig. 7). On the other hand, the activity of the 0.7 HGF-CAT construct that does not contain the HERE1 element did not change upon treatment with estradiol (data not shown). This result strongly suggests that in vivo high expression of ER counteracts COUP-TF and stimulates HGF mRNA expression.

DISCUSSION
Our previous studies have shown that a negative regulatory element is located within Ϫ964 to Ϫ699 of the 5Ј-flanking region of the mouse HGF gene (22). In this region, an estrogen response element (HERE1) was identified that binds to ER and is responsible for the estradiol stimulation of the mouse HGF promoter in RL95-2 cells co-transfected with an ER expression vector (32). We now show that this same element binds with high affinity to the nuclear orphan receptor COUP-TF, which belongs to the steroid/thyroid receptor superfamily, and silences the activity of the HGF promoter. We demonstrate that the estrogen receptor competes with COUP-TF for binding to this site, thereby relieving the repressive action of COUP-TF.

FIG. 3. Synthetic mouse HERE1 sequence confers repression on a heterologous promoter.
Shown on the left is the diagrammatic representation of the chimeric CAT constructs. Synthetic oligonucleotides corresponding to the mouse HERE1 and HERE2 were cloned into a position upstream of the SV40 promoter (SV). These constructs were transiently transfected into NIH3T3 cells followed by CAT assay as described under "Experimental Procedures." Depicted on the right is a graphic representation of relative promoter activity as determined by CAT assays. Each bar represents the mean of three independent transfection experiments. The nucleotide sequences of wild type ERE (VitA2ERE), mouse HERE1, and mouse HERE2 are shown for comparison.
There are two imperfect EREs found in the mouse HGF promoter region that we have called HERE1 and HERE2 (22,32). HERE1 is localized in the 5Ј-flanking region and HERE2 is in the first intron of the HGF gene. By using gel mobility band shift and supershift assays, we demonstrated that HERE1 is similar to the ERE found in the vitellogenin (vit) gene (37) and that it binds not only to ER but also to COUP-TF (Fig. 4) (30  and 32). Co-transfection of ER or COUP-TF expression vectors in NIH3T3 cells revealed that HERE1 confers the repressive function of COUP-TF to the heterologous SV40 promoter and that not only did ER overcome the suppression of COUP-TF but it also stimulated the SV40 promoter (Fig. 6). In the homologous mouse HGF promoter, ER only counteracted the negative effects of COUP-TF (Fig. 5).
The molecular mechanisms of COUP-TF-mediated transcriptional repression are partially understood. COUP-TF acts via several different mechanisms to inhibit target gene transcription including competition with other nuclear receptors for the occupancy of DNA binding sites, active repression of basal transcription, active repression of transactivator-dependent transcription, quenching of transactivator-dependent transcription, transrepression of activated transcription, and titration of the common coregulator RXR (31,38). Our data shown here reveal that COUP-TF may down-regulate the mouse HGF promoter by active repression of basal transcription because HERE1 suppresses the basal SV40 promoter and that ER may activate HGF transcriptional activity by replacing COUP-TF and relieving the promoter from repression (Fig. 5). COUP-TF belongs to the orphan (thus far, no ligand has been found that can bind and directly regulate the activity of this factor) nuclear receptor family and resembles the thyroid hormone receptor/steroid hormone receptor superfamily (31,36). The mode by which it represses its target gene involves the binding of COUP-TF to the AGGTCA motif arranged in various configurations (i.e. direct repeats (DR) or inverted repeats (IR) with different numbers of spacers ranging from 0 to 8 nucleotides) (31). Interestingly, the AGGTCA half-site depending on its configuration forms the response elements for other nuclear receptors such as the estrogen receptor (which binds to IR3), thyroid hormone receptor (which binds to IR0 or DR4), retinoid receptors (RAR, which binds to IR0 and DR2, and RXR, which binds to DR5 and DR1), vitamin D receptor (which binds to DR3), HNF4, and peroxisome proliferator-activated receptor (31). It is believed that COUP-TF directly competes with these receptors for binding to the AGGTCA sequence and thus inhibits their transactivating functions. Suppression by COUP-TF is also achieved in part through heterodimerization of COUP-TF with RXR (which is the essential transactivating partner for vitamin D 3 receptor, thyroid hormone receptor, and RAR), thereby titrating RXR to a transcriptionally inactive complex FIG. 4. Identification of the protein(s) binding to the HERE1 element as COUP-TF. A, the HERE1 element was labeled with 32 P and incubated with nuclear protein extracts from NIH3T3 or RL95-2 cells. Double-stranded oligonucleotides including self, COUP, VitA2ERE, HERE2, and GRE were added to the reaction mixture to compete for binding as indicated (100ϫ). B and C, supershift of HERE1 bound to nuclear protein(s) from NIH3T3, HepG2, RL95-2, and rat liver nonparenchymal cell (NPC) nuclear extracts by anti-COUP-TF antibody is shown. Anti-ER antibody and normal rabbit serum were also used for comparison. The binding assays were done as described under "Experimental Procedures." C, specific complex; NS, nonspecific complex; F, free 32 P-labeled DNA probe; Ab, antibody; S, normal rabbit serum. (31,38). It is tempting to speculate that in addition to ER other members of the nuclear receptor family mentioned above may bind to the COUP-TF binding site in the HGF promoter HERE1 and thus activate HGF transcription in a tissue-specific manner. The COUP-TF site in the HGF promoter, however, is composed of an estrogen response element (IR3) and apparently binds only to the estrogen receptor.
Previous observations in our laboratory showed that administration of 17␤-estradiol to immature female mice caused a significant induction of HGF mRNA in the ovaries (32). The tissue-and cell type-specific expression of the mouse HGF gene by estrogen may be due to different amounts of hormone receptors in these tissues, as shown by other studies (37). Our present studies show that high expression of estrogen receptor in NIH3T3 cells is sufficient to counteract the suppression of COUP-TF on the mouse HGF promoter (Fig. 5). HGF promoter activity or lack thereof in transgenic mice also supports this hypothesis because estradiol stimulated the HGF promoter in mice carrying the 2.7 HGF-CAT construct that contains HERE1, particularly in tissues that have high expression of ER (Fig. 7). In this regard, it is important to understand how HGF gene expression is relieved from COUP-TF repression in other tissues in which ER may not be operational. The factors that are known to activate HGF expression include cytokines such as IL-1, IL-6, and tumor necrosis factor ␣ and polypeptide growth factors such as epidermal growth factor. These agents do not exert their gene-regulatory function directly through the AGGTCA motif, the DNA binding site for COUP-TF. Although the molecular mechanisms by which these agents induce their target genes are not fully understood, it is well established that they utilize members of the AP1, C/EBP, nuclear factor B (NFKB), and signal transducer and activator of transcription (STAT) family of transcription factors. Indeed, in our previous report (22) we demonstrated that several NF-IL6 (C/EBP␤) response elements exist in the 5Ј-flanking region of the HGF promoter region and that IL-6 induces HGF promoter activity in a stromal (fibroblast) cell line stably transfected with an HGF-CAT chimeric construct. In our very recent investigations using gel shift, supershift, and CAT assays, we have found that functional C/EBP binding sites exist in the HGF promoter. We also found that co-transfection of expression vectors encoding C/EBP␤ or C/EBP␣ with HGF-CAT vectors dramatically (more than 20-fold) stimulates the transcriptional activity of various HGF-CAT promoter constructs (even those constructs that do not contain COUP-TF binding sites, thus indicating that promoter activation is not due to interference with COUP-TF binding). 3 These observations provide an explanation for how the HGF promoter is released from repression exerted by negative acting factors such as COUP-TF. The C/EBP family of transcription factors is known to consist of regulators of important cellular functions ranging from metabolism to cell growth to differentiation.
The physiological significance of the transcriptional regulation of the HGF gene by ER and COUP-TF also remains to be elucidated. Nevertheless, it has been shown that both HGF and COUP-TF play an important role during embryonic development. COUP-TFs are differentially expressed in a restricted manner during organogenesis (40). Both COUP-TF I and COUP-TF II genes are essential since COUP-TF I mutants die prenatally and COUP-TF II mutants die in utero (41). Additionally, HGF is believed to be one of the major mediators of stromal-epithelial interactions controlling growth and morphogenesis of various tissues such as breast (39). Studies on the 3 J.-G. Jiang and R. Zarnegar, manuscript in preparation. FIG. 6. COUP-TF and estrogen receptor counteract one another's effects through binding to HERE1. A, chimeric SV40-CAT plasmids containing HERE1 and HERE2 were co-transfected with increasing amounts of ER expression vector into the NIH3T3 cells. The estrogen response of the HERE1-SV40-CAT and HERE2-SV40-CAT constructs was then examined by CAT assay. SV40-CAT plasmid was included as a control. The activity of each construct was normalized against ␤-galactosidase activity, and relative CAT activities are presented as the averages from three independent experiments. B, overexpression of COUP-TF repressed the ER-dependent, estrogen-stimulated response of HERE1. Chimeric plasmids (same as in A) were co-transfected with the estrogen receptor expression vector pHEO (0.2 g/well) and various concentrations of pRS-COUP, the COUP-TF expression vector, in NIH3T3 cells. CAT activity was normalized against protein concentration, and the relative CAT activities are presented as the averages from at least three independent experiments. Estradiol concentration was 10 Ϫ7 M. embryological development of such tissues have revealed that steroid hormone-induced ductal morphogenesis and growth is accomplished via the interactions between the stromal and epithelial compartments of these tissues (34). Therefore, the functions of COUP-TF and ER during development may be mediated in part by their ability to regulate HGF gene expression. Our data shown here shed light on the significant interplay among transcription factors such as ER, COUP-TF, and C/EBP and polypeptide growth factors such as HGF. FIG. 7. Estradiol stimulates HGF promoter activity in transgenic mice. Cell extracts for CAT assays were prepared from mammary gland, adjacent skin, and ovary from 2.7 HGF-CAT transgenic mice that harbor 2.7 kb of the HGF promoter region and from nontransgenic mice after injection of 17␤-estradiol (100 mg/kg of body weight) or control saline solution. Fifty micrograms of total protein extract from each tissue were subjected to CAT assay. A graphic representation of relative CAT activity after normalization for protein concentration is shown. The results presented are the average from at least three groups of mice. TG, transgenic mouse; NTG, nontransgenic mouse; AS, adjacent skin of mammary gland; MG, mammary gland; OV, ovary tissue; E2, 17␤-estradiol.