(Received for publication, August 17, 1996, and in revised form, October 22, 1996)
From the Division of Cellular and Molecular Pathology, Department of Pathology, School of Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania 15261
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)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 (13-15). 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 5
-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 5
-flanking region of the mouse HGF
promoter residing at position
1 to
0.7 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 D3 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.
-The 1.4 mouse HGF-CAT (1386 to
+29), 1.0 mouse HGF-CAT (
964 to +29), and 0.7 mouse HGF-CAT (
699 to
+29) 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).
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 µg of one chimerical CAT construct, 1 µg 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 ExtractsHuman 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 MgCl2, 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 µg 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 × 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
MgCl2, 10 mM KCl, 0.2 mM
phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, plus
1 µg 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 × 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
MgCl2, 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 80 °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.
OligonucleotidesThe following oligonucleotides were used
in gel mobility shift assays: vitellogenin A2 ERE (VitA2ERE),
5-GATCTAGGTCACAGTGACCTA-3
(33); and COUP,
5
-GATCTTTCTATGGTGTCAAAGGTCAAACT-3
(30). The following sequences were
derived from the mouse HGF gene promoter (22): HERE1,
5
-GATCAAGGTCAGAAAGACCAT-3
; HERE2, 5
-GATCTGGTCAATCTAACCA-3
; AP1, 5
-GATCTGCCTTGACTTAGCGAG-3
; IL-6RE, 5
-GATCAGAGCTGGGATCTG-3
; CRE, 5
-GATCTTAAGACGTCATTTA-3
; NF-IL6, 5
-GATCCTTCTGAGGAAAGCTGC-3
; GRE, 5
-GATCTTCTGGTGCTATTTCTTCTCATT-3
; and TIE,
5
-GATCGGGCGAATTGGTGTTCTGC-3
.
A DNA fragment corresponding to
nucleotide position 964 to
812 bp of the 5
-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 [
-32P]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 32P 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 µl of reaction mixture. When antibodies were used in supershift
experiments, they (1 µl) were incubated with nuclear extracts for 15 min at room temperature before carrying out DNA binding shift assays.
The binding reactions (20 µl 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 µg 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 × 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 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.
We previously reported the existence of a negative
regulatory region in the 5-flanking region of the mouse HGF gene
located between nucleotides
964 and
699 bp (22). To further define this region, two additional HGF-CAT chimeric constructs containing a 5
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
964 and
812 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 5-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 (
872 to
860 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
872 to
860,
5
-GGTCAGAAAGACC-3
) is involved in the formation of the protein-DNA
complex. This binding sequence is similar to the estrogen response
element (5
-AGGTCANNNTGACCT-3
, see also Fig.
3).
To assess the functionality of this element, synthetic HERE1 and HERE2
(a mutated version of HERE1, 5-GGTCAATCTAACC-3
) 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).
To
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 872 bp in the 5
-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.
Our previous studies have shown that a negative regulatory element
is located within 964 to
699 of the 5
-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.
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
5-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 D3 receptor, thyroid hormone receptor, and RAR), thereby titrating RXR to a transcriptionally inactive complex (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 5
-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 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.
We are grateful to Drs. S. Y. Tsai and M.-J. Tsai for generously providing us with the anti-COUP-TF antibody and the COUP-TF expression vector pRS-COUP. We also thank Dr. M. C. DeFrances for critical review of the manuscript.