(Received for publication, January 8, 1997, and in revised form, February 5, 1997)
From the Women's Health Research Institute, Wyeth-Ayerst Research, Radnor, Pennsylvania 19087
The estrogen receptor (ER) belongs to a superfamily of ligand-inducible transcription factors. Functions of these proteins (dimerization, DNA binding, and interaction with other transcription factors) are modulated by binding of their corresponding ligands. It is, however, controversial whether various ER ligands affect the receptor's ability to bind its specific DNA element (ERE).
By using real time interaction analysis we have investigated the
kinetics of human (h)ER binding to DNA in the absence and presence of
17-estradiol, 17
-ethynyl estradiol, analogs of tamoxifen, raloxifene, and ICI-182,780. We show that ligand binding dramatically influences the kinetics of hER interaction with specific DNA. We have
found that binding of estradiol induces the rapid formation of a
relatively unstable ER·ERE complex, and binding of ICI-182,780 leads
to slow formation (ka is approximately 10 times lower) of a stable receptor-DNA complex (kd is
almost 2 orders of magnitude lower). Therefore, binding of estradiol accelerates the frequency of receptor-DNA complex formation more than
50-fold, compared with unliganded ER, and more than 1000-fold compared
with ER liganded with ICI-182,780. We hypothesize that a correlation
exists between the rate of gene transcription and the frequency of
receptor-DNA complex formation. We further show that a good correlation
exists between the kinetics of hER-ERE interaction induced by a ligand
and its biological effect.
Steroid hormones are widely distributed small, lipophilic molecules that participate in intracellular communication and control a wide spectrum of developmental and physiological processes. Their effects are mediated by specific intracellular receptors, a family of proteins that are characterized by a high affinity for the corresponding hormones and an ability to discriminate between structurally closely related ligands. These ligand-inducible receptors can modulate transcription of target genes by virtue of their binding to a specific sequence on DNA in target promoters known as hormone response elements. Although distinct proteins, these receptors are members of a large superfamily of steroid hormone receptors and share many common structural and functional features (1-4).
Binding of 17-estradiol
(E2)1 to estrogen receptor (ER)
is followed by a conformational change, leading to dissociation of the
receptor from the complex with the heat shock proteins hsp90 and p59
(5, 6), dimerization (7, 8), and activation of DNA binding. After DNA
binding the activated receptor can interact with basal transcription
factors (9). These interactions are thought to stabilize the
preinitiation complex at the promoter, allowing RNA polymerase to
initiate transcription (10). Recently a number of transcriptional
intermediary factors have been identified that can modify estrogen
responsiveness, and several of these proteins interact with the ligand
binding domain of the ER in a ligand-dependent manner
(11-13).
It is obvious that the ligand plays a key role in initiating this
cascade of events. However, it remains controversial as to whether
estrogen affects the receptor's ability to bind specific DNA (ERE).
Initially, in vitro analysis performed with hER (HEO) expressed in HeLa cells, Xenopus oocytes, yeast, or produced
by in vitro transcription/translation indicated that binding
of ER to an ERE was hormone-dependent and that
E2 induced the formation of receptor dimers (8). It was
subsequently found that this protein (HEO) has an artifactual mutation
(Gly-400 Val) that decreases hormone binding at 25 °C but not at
4 °C (14).
Subsequently it was reported that wild type hER (HEGO) binds DNA in absence of ligand (15, 16). Hormone-independent formation of ER·ERE complex was also reported with crude extracts (17) and purified ER from calf uterus (18), rat uterine extracts (19), mouse uterine extracts (20), transfected COS-1 cells (21, 22), and Sf9 cells infected with recombinant baculovirus (21). In contrast, ligand-induced DNA binding was reported with HEGO, produced by in vitro transcription/translation and in Sf9 cells (23). In vivo ligand-dependent and ligand-independent ERE association was reported for hER and Xenopus ER using promoter interference assay (24, 25). It was also shown that hormone may be required to promote DNA binding at low but not at high concentrations of ER (26). Genomic footprinting indicated that occupation of the ERE present in apoVLDLII promoter region is hormone-dependent (27) which suggests that the hormone is affecting ER interaction with ERE.
We hypothesized that ligand binding may affect the kinetics of ER interaction with DNA while having minimal effect on its affinity. It was previously reported that binding of hormone accelerates the kinetics of glucocorticoid and progesterone receptor binding to DNA (33), affects dimerization status and the kinetics of DNA binding of vitamin D3 receptor (28). In this case the discrepancies reported previously with the ligand effect on ER-ERE interactions may be due to the fact that the methodologies used could not detect the dynamics of the protein-DNA complex formation.
We have examined the ER-ERE interactions using surface plasmon resonance (SPR) methodology. Using this approach, we have found that ligand binding dramatically affects the kinetics of hER interaction with specific DNA. We show that binding of agonists or antagonists by hER has opposing effects on the kinetics of ER binding to ERE.
The BIAcore system, sensor chips CM
5 (certified), Tween 20, the amine coupling kit containing
N-hydroxysuccinimide,
N-ethyl-N-(3-diethylaminopropyl) carbodiimide
and ethanolamine hydrochloride were all obtained from Biacore Inc. The
buffer used for all experiments was 50 mM Tris-HCl, 150 mM NaCl, 10 mM MgCl2, 0.05% Tween
20, pH 7.5. 17-
-Estradiol, 17
-ethynyl estradiol,
4-hydroxytamoxifen, and tamoxifen were obtained from Sigma;
3-hydroxytamoxifen was obtained from RBI; raloxifene was synthesized by
Wyeth-Ayerst Medicinal Chemistry group. ICI-182,780 was provided by
Zeneca Pharmaceuticals. [3H]17-
-Estradiol was from
DuPont NEN.
Several potential EREs
were synthesized as self-annealing oligonucleotides that form hairpin
duplex upon heating and rapid cooling. A 75-bp oligonucleotide
(Vit.A2) containing a specific binding site for ER was
derived from the vitellogenin A2 gene of Xenopus
laevis response promoter. Its sequence is
5-AGCTCTTTGATCAGGTCACTGTGACCTGAACTTACTCCCCCCGAGCAAGTTCAGGTCACAGTGACCTGATCAAAG-3
. The second oligonucleotide (DR) was designed as two directly repeated half-sites. Its sequence is
5
-AGCTCTTTGATCTGACCTCTGTGACCTGAACTTACTCCCCCCGAGCAAGTTCAGGTCACAGAGGTCAGATCAAAG-3
. The third oligonucleotide (C3) was derived from mouse complement component C3 gene promoter (29). Its sequence is
5
-AGCTCTTTGATCTGACCTCTGTGACCTGAACTTACTCCCCCCGAGCAAGTTCAGGTCACAGAGGTCAGATCAAAG-3
. Following annealing, the oligonucleotides were biotinylated by incorporation of biotin-dATP with Klenow enzyme and purified from unincorporated biotinylated nucleotides by gel filtration on a Chromaspin 10 column (CLONTECH).
The BIAcoreTM- biosensor system (Pharmacia Biosensor, Uppsala, Sweden) permits the monitoring of macromolecular interactions in real time. The detection system uses SPR, a quantum mechanical phenomenon that detects changes in the refractive index at the surface of sensor chip (39). The binding of a soluble ligand to the immobilized one leads to an increase in the ligand concentration at the sensor surface, with a corresponding increase in the refractive index. This refractive index change alters the SPR which can be detected optically (40). BIAcore 2000 allows the simultaneously detection of the interaction events on four different spots, located in different flow cells (FC), on the sensor surface. The flow system allows the sample to be addressed to individual FCs or serially to predefined combinations of the four FCs.
To immobilize specific DNA, the surface of a CM 5 sensor chip
(certified) was first modified with streptavidin according to instructions from the manufacturer. The surface was activated by
injection of a solution containing 0.2 M
N-ethyl-N-(3-diethylaminopropyl) carbodiimide and 0.05 M
N-hydroxysuccinimide. Four different surfaces with 2918, 1922, 1363, and 908 resonance units (RU) of streptavidin were obtained
by injecting streptavidin at 10 µg/ml and varying a contact time from
1 to 6 min. Gradient surfaces with 1830, 1300, 900, and 600 RU of
Vit.A2-DNA were then obtained by injection of 50 µl of
biotinylated oligonucleotide solution at 33 ng/µl. After each protein
injection, the surface was regenerated with one 10-µl injection of
0.1% SDS solution. One cycle of regeneration was sufficient to remove
bound ER.
Each binding cycle was
performed with a constant flow of 50 mM Tris-HCl, 150 mM NaCl, 10 mM MgCl2, 0.05% Tween
20, pH 7.5, buffer of 5 µl/min. Samples of ER were injected across
the surface via a sample loop. Once the injection plug had passed the
surface, the formed complexes were washed with the buffer for an
additional 500-1000 s. All experiments were performed at 25 °C.
Data were collected at 1 Hz and analyzed using the BIAEvaluation
program 2.1 (Biacore Inc) on a Compaq PC. This program uses a nonlinear least squares analysis method for the determination of rate binding constants for macromolecular interactions. The dissociation kinetics of
the ER·ERE complexes can be described by a double exponential decay
as follows: AiBj = Ai + Bj. One of these complexes
dissociates from the surface with apparent rate of
kd 1 = 0.01 s1 which
corresponds to a complex half-life of several minutes and represents
only 5-10% of total bound protein. The second complex is much more
stable. Its apparent dissociation rate
kd 2 = 10
3 to
10
5 s
1, depending on the nature of ligand,
and it represents approximately 90-95% of total bound protein.
Association kinetics can be described by a model that uses two analytes
(A1 and A2) interacting
with the same binding site as follows: A1 + A2 + B = A1B + A2B (28).
The apparent equilibrium dissociation constants for the second complex, which represent the majority of bound material, were calculated from the two rate constants (KD = kd 2/ka 2).
Purification of ERPartially purified recombinant human estrogen receptor was obtained from PanVera Corp. (about 80% pure) and purified to homogeneity as assessed by visualization on a Coomassie-stained SDS gel by gel filtration on Superdex-200 column (Pharmacia) and chromatography on Ni2+ affinity column (Pharmacia). Receptor was eluted from Ni2+ affinity column with a linear gradient of imidazole from 0 to 0.5 M in 50 mM Tris-HCl, 500 mM NaCl, pH 7.5.
Gel Shift AssaysVitA2, DR, and C3
oligonucleotides were labeled by incorporation of
[-32P]dATP (6000 Ci/mmol) with Klenow enzyme and
purified from unincorporated nucleotides by gel filtration on a
Chromaspin 10 column (CLONTECH). Probes had a specific activity of
2.0 × 107 cpm/µg. DNA binding reactions were
carried out in buffer containing 2 ng of labeled DNA, 20 mM
Tris-HCl, pH 7.9, 0.1 mM EDTA, 5 mM MgCl2, 50 mM KCl, 0.5 µg/µl poly[d(I-C)],
10% glycerol, 1 mM dithiothreitol, and 0.5-3.0 µg of
purified ER in a total volume of 20 µl at 25 or 37 °C. The
protein-DNA complex was separated on 8% (75:1, acrylamide to
bisacrylamide) polyacrylamide gels. The gels were dried and subjected
to autoradiography. Gel quantitation was performed using PhosphorImager
SI (Molecular Dynamics).
Purified ER was prepared by boiling for 5 min in SDS-polyacrylamide gel electrophoresis sample buffer, separated in 12% SDS-polyacrylamide gels (34) using the PhastSystem (Pharmacia), and either stained with Coomassie Brilliant Blue G250 or transferred to nitrocellulose membrane (Hybond-ECL, Amersham Corp.) for Western blotting. After the transfer, the membranes were blocked, incubated with primary anti-hER antibody, and then developed with secondary antibody linked to hydrogen peroxide catalase (Amersham Corp.) as described (30). The anti-human polyclonal antibody against the AB region of the estrogen receptor was produced at Wyeth-Ayerst Research.
Gel Filtration ChromatographyA Superdex 200 (30:10) column
was obtained from Pharmacia. The column was equilibrated with 50 mM Tris-HCl, 150 mM NaCl, 10 mM
MgCl2, 0.05% Tween 20, pH 7.5, and calibrated with
thyroglobulin (669 kDa), ferritin (440 kDa), catalase (232 kDa),
aldolase (158 kDa), bovine serum albumin (67 kDa), ovalbumin (43 kDa),
and RNase (13.7 kDa), all at 1 mg/ml. To analyze ER·DNA complexes by
gel filtration, 35 µg of purified ER in 50 µl 50 mM
Tris-HCl, 150 mM NaCl, 10 mM MgCl2,
0.05% Tween 20, pH 7.5, were incubated with 20 ng of
32P-labeled oligonucleotide in presence or absence of
[3H]estradiol (106 M) at room
temperature for 30 min. The sample was then applied on to the Superdex
200 column and eluted in the buffer described above (flow rate 0.25 ml/min). Radioactivity in fractions (0.25 ml) from the column was
detected by liquid scintillation spectrometry.
Commercially available, partially purified recombinant baculovirus-infected Sf9 cell hER was used for this study. We further purified this protein to homogeneity using gel filtration and chromatography on a Ni2+-affinity column. As described previously (31), metal-affinity chromatography can be applied successfully for hER purification. hER binds to a Ni2+-resin without additional poly-His fusion. The affinity of this interaction is relatively high, and receptor can be eluted from the column with 100 mM imidazole. Purified ER is detected as one band on a Coomassie-stained SDS gel (data not shown). The apparent molecular mass of the purified receptor is 67 kDa. Western blot analyses demonstrates that this protein can be recognized by an anti-hER antibody (data not shown).
Purified hER Specifically Interacts with DNAIn
vitro, purified receptor binds with high affinity to a palindromic
inverted repeat element containing the sequence AGGTCA spaced by 3 bp
derived from the vitellogenin A2 gene (Vit.A2) promoter of X. laevis (32). In gel retardation assays bound complex typically yields a major band and a more slowly migrating minor
band (Fig. 1). Both of these bands are supershifted by
an anti-hER antibody (Fig. 1, lane 5) which implies that
they represent different ER·ERE complexes. Two other oligonucleotides
were also used in this work. One contains two directly repeated AGGTCA
elements spaced by 3 bp (DR) which represents a high affinity binding
site for the vitamin D3 receptor (35). The other
oligonucleotide, derived from the mouse complement component C3 gene
promoter (C3), which is regulated by estradiol (29). It contains a
perfect AGGTCA and one element, AGTCTA different from consensus,
positioned as an inverted repeat with a spacing of 3 bp. It has been
demonstrated that minimal activity is associated with a single copy of
the C3 enhancer element when this oligonucleotide is inserted 5 to a
thymidine kinase promoter and cotransfected with human ER in to
Ishikawa cells (36). Our results (Fig. 1) demonstrate that complexes of
the same size as detected with Vit.A2 can be formed by the
interaction between the hER and DR or C3 oligonucleotides in gel
retardation assays. Complex quantitation using PhosphorImager demonstrated that the amount of shifted DR or C3 is more than 100 times
lower than that for Vit.A2. This result implies that the
affinity of the ER interaction with these oligonucleotides is
significantly reduced compared with the Vit.A2.
Analysis of ER·ERE Complexes by Gel Filtration
To clarify the nature of the various ER·DNA complexes, we have examined them using analytical gel filtration chromatography. This technique is very useful for the evaluation of the apparent molecular weight and stoichiometry of different protein-DNA complexes, and this approach has been applied previously to the study of vitamin D3 receptor interaction with specific DNA (41).
Purified hER elutes from this column as a major peak with apparent mass of 140 kDa and several minor peaks with apparent mass of 70, 300, and 540 kDa (data not shown). This result suggests that the major molecular form of hER in solution is a dimer. The fact that other complexes besides monomers and dimers were found implies that hER has a tendency to aggregate. hER bound to [3H]estradiol elutes as two complexes, with apparent mass of 70 and 140 kDa (data not shown) with a dimer being the major molecular form of ER under these conditions.
To analyze ER·DNA complexes by gel filtration, the receptor was
incubated with [3H]estradiol and then mixed with
32P-labeled Vit.A2 oligonucleotide before this
mixture was applied on Superdex 200 column. For a liganded and
unliganded ER, major ER·ERE complex, approximately 95-98%, as
detected by peaks integration, elutes from this column as a peak with
apparent mass of 140 kDa (Fig. 2). A small amount,
approximately several percent of the total DNA loaded, was eluted as a
high molecular weight ER·ERE complex.
Analysis of the Ligand Effects on ER Binding to a Specific ERE by Gel Shift Assays (GSA)
Brown and Sharp (37) reported that in the presence of Mg2+ HEO, mutated hER (14), binds the ERE in hormone-dependent fashion at 20-37 °C. Later Metzger et al. (38) demonstrated that for a wild type hER (HEGO) in the presence of E2 or 4(OH)-tamoxifen, preincubation at 4 or at 37 °C had little effect on DNA binding, whereas in the absence of ligand, or in the presence of ICI-182,780, incubation at 37 °C resulted in greatly reduced DNA binding.
To further examine if hER demonstrates a ligand dependence for ER·ERE complex formation detectable by GSA, the receptor was incubated at 25 or at 37 °C in the absence or presence of E2, 4-hydroxytamoxifen, raloxifene, or ICI-182,780, and then labeled DNA was added and incubation was continued at the same temperature. Complexes formed were analyzed by GSA (data not shown). Our results demonstrate that E2 binding had little or no effect on the amount of ER·ERE complex formed at 25 °C, as detected by GSA. A similar result was obtained with 4-hydroxytamoxifen, raloxifene, and ICI-182,780. If, however, preincubation was performed at 37 °C, significant reduction of ER·ERE complex formation was observed in the absence of ligand or in presence of ICI-182,780. Therefore, the purified hER exhibits the same ligand dependence for DNA binding as wild type hER (HEGO) (38).
Analysis of ER-ERE Interaction by SPR Detector, BIAcoreAbsence of detectable differences in the amount of formed ER·ERE complex by liganded or unliganded ER at 25 °C may also reflect the inability of GSA to detect differences in affinity of ligand-induced ER-ERE interaction. On the other hand, it may be due to the fact that ligand binding affects the kinetics of ER-ERE interaction while having a minimal effect on affinity. To address these possibilities we have used real time analysis with BIAcore to study ligand effects on the kinetics of ER·ERE complex formation at 25 °C.
Experiments with the BIAcore usually start with the equilibration of
the surface with running buffer (Fig. 3). There is an "injection jump" at the beginning and end of each injection due to
the difference in the refractive index between the running and sample
buffers (presumably due to small changes in salt concentration). Protein is injected during the "wash-in" phase. During the
"wash-out" phase, the running buffer is injected across the flow
cells. After approximately 15 s from the start or the end of
injection, the data generated can be used for the kinetic analysis.
We approached our study by immobilization of 1094 RU of DR- oligonucleotide in FC 1, 943 RU of Vit.A2 oligonucleotides in FC 2, and 1063 RU of C3 oligonucleotides in FC 3 on the surface of a sensor chip (see "Experimental Procedures"). Human ER bound with E2 was injected over the surfaces coated with these different oligonucleotides (Fig. 3). High affinity interaction was detected in the flow cell with Vit.A2 immobilized. This is indicated by a fast increase of the refractive index, detectable after the injection jump in the beginning of injection (see Fig. 3, Vit.A2). At the end of injection, a significant amount of ER is still bound to the surface, which can be detected as a difference between the refractive index of the injection over the surface with immobilized Vit.A2 and the base line. At the same time binding of hER to immobilized streptavidin (data not shown), DR, or C3 surfaces was very low. Thus, the mass increase on the surface of the sensor chip in response to the injection of hER is due to specific receptor binding to DNA. Injection of ER over the surface with no immobilized DNA results in no detectable binding (data not shown).
To study how different ligands affect ER-ERE interactions, a surface
with a gradient of immobilized Vit.A2 was created. First a
gradient of immobilized streptavidin in different flow cells was
obtained by varying the streptavidin injection time as illustrated in
Fig. 4A. As a result, surface with 2918, 1922, 1363, and 908 RU of streptavidin in FCs 1-4, respectively, was
obtained. Injection of 50 µl of biotinylated Vit.A2 at 33 ng/µl resulted in a surface with 1817, 1290, 917, and 605 RU of
immobilized Vit.A2 in FCs 1-4, respectively (Fig.
4B).
Serial injections of unliganded ER and ER incubated overnight with
106 M of E2, 4(OH)-tamoxifen,
raloxifene, and ICI-182,780 at protein concentrations ranging from 35 to 270 nM were run over the sensor chip with immobilized
gradient of the Vit.A2. Fig. 5 demonstrates overlaid sensograms of the injections of the hER at different protein
concentrations, liganded with E2, over the gradient of immobilized
Vit.A2. Clear correlation between the change of refractive index and the level of immobilized DNA was obtained. Saturation of
immobilized ERE with injected hER was reached for the hER liganded with
4(OH)-tamoxifen and E2 at 270 nM hER. We found
that there is a linear relationship between the surface density (amount
of immobilized DNA) and the saturation response (refractive index at
the end of injection) obtained with 270 nM injection of hER in the interval of the DNA densities used for this experiment (data not
shown). Calculated stoichiometry for the formed ER·ERE complex (42)
was 1:1.92 ± 0.11. This result confirms that hER binds specific
oligonucleotide-Vit.A2 as a homodimer.
Fig. 6 shows overlaid sensograms obtained by injections
of 50 µl of unliganded hER and hER liganded with E2,
4(OH)-tamoxifen, raloxifene, and ICI-182,780 at receptor concentration
of 89.5 nM over the FC with 900 RU of VitA2 immobilized. It
can be seen that different ligands significantly affect the ER-ERE
interaction. Relative to unliganded ER or ER liganded with ICI-182,780,
much more of the ER·ERE complex is formed when ER is liganded with E2 or 4(OH)-tamoxifen. It can also be seen that this
complex is less stable than the complex induced by ICI-182,780 and
raloxifene.
Sensograms of the injections of liganded and unliganded hER over the
gradient surface with immobilized Vit.A2 were analyzed using the
BIAEvaluation 2.1 program, as described (28). Values of apparent
dissociation, association, and affinity rate constants determined for
ER-ERE interaction in the absence and presence of different ligands are
summarized in Table I. It is interesting that
E2 binding (pure agonist) is inducing fast formation
(ka = 9.62 × 104
M1 s
1) of unstable ER·ERE
complex (kd = 1.86 × 10
3
s
1), at the same time ICI-182,780 binding (pure
antagonist) is inducing slow formation (ka = 1.48 × 103 M
1
s
1) of a very stable ER·ERE complex
(kd = 6.35 × 10
5
s
1).
|
We were also interested in examining if derivatives of the same
compound may have a different effect on ER-ERE interactions. hER was
preincubated with the analogs of 4(OH)-tamoxifen at 106
M (tamoxifen, 3(OH)-tamoxifen, 4(OH)-tamoxifen,
zuclomiphen, and toremifene) overnight and then used for serial
injections at different protein concentrations over the surface of a
sensor chip carrying immobilized Vit.A2. The effect of
these compounds on the kinetics of hER interaction with the ERE was
examined as previously described (28). Values of the apparent
association and dissociation rates are summarized in Table I. Fig.
7 presents overlaid sensograms of hER liganded with the
analogs of 4(OH)-tamoxifen injected over a surface with ERE
immobilized. It is clear that even a small chemical modification of the
ligand (position of OH group in the molecule of tamoxifen), presumably
modulating receptor's conformation, may affect receptor-DNA
interactions. It can be seen that stability of the ER·ERE complex
induced by the 4(OH)-tamoxifen is much lower (see Table I) than for the complexes induced by 3(OH)-tamoxifen and tamoxifen. It is interesting that on consensus ERE, 4(OH)-tamoxifen is a more potent agonist than
other analogs of tamoxifen used in this study.
We have used the SPR technology to study the DNA binding by the hER, in the absence and presence of estradiol, analogs of tamoxifen, raloxifene, and ICI-182,780. This approach allows detection of the receptor-DNA complex formation in real time. This information can then be fit into a mathematical model describing this interaction to obtain kinetic and thermodynamic rates of the receptor-DNA association and dissociation. Therefore, it was important to use pure hER and to know the nature of the complexes that could be formed by ER-ERE interaction.
Commercially available, partially purified recombinant baculovirus-infected Sf9 cells hER were used for this study. We further purified this protein to homogeneity. Purified hER specifically binds to the Xenopus vitellogenin A2 gene palindromic ERE in vitro and can be recognized by a specific anti-hER antibody. Analytical gel filtration analysis showed that the predominant molecular form of this hER preparation, in the presence or absence of estrogen, at protein concentrations ranging from 40 to 300 nM is a homodimer. Monomeric hER and complexes with greater than 2 orders of oligomerization were also identified, and this result is in agreement with previous studies (38, 43, 44). In the presence of specific labeled oligonucleotide, derived from the vitellogenin A2 gene of X. laevis response promoter, homodimer-(hER)2ERE complex was detected. Some of the labeled DNA, approximately 3-5%, was eluted close to a column void volume, which suggests that oligomeric hER·ERE complexes could also be formed.
Gel shift analysis of hER interaction with DNA clearly demonstrated ligand dependence at 37 °C. When incubated at 25 °C, there were no differences found between amounts of hER·ERE complex formed in the presence and absence of different ligands as detected by gel shift analysis. These observations are in keeping with a previous study of Metzger et al. (38) who reported that wild type hER when incubated at 25 °C binds DNA irrespective of the presence or absence of estrogen and that in absence of ligand, or in the presence of ICI-182,780, incubation at 37 °C resulted in greatly reduced DNA binding.
We have hypothesized that ligand binding may affect the kinetics of ER interaction with DNA while having a minimal effect on its affinity. It was previously reported that binding of hormone accelerates the kinetics of glucocorticoid and progesterone receptor binding to DNA (33) and that binding of 1,25-(OH)2D3 affects the dimerization status and kinetic of DNA binding by the vitamin D3 receptor (28).
To address our hypothesis, we have used the surface plasmon resonance detector BIAcore to measure the effect of various ligands on the hER interaction with DNA. Previous studies (28, 45-48) have demonstrated that surface plasmon resonance-based methodology can be effectively used to study protein-DNA interactions. We found that this approach gives adequate information about hER interaction with DNA. High affinity binding was found to the Xenopus vitellogenin A2 gene palindromic ERE. Low levels of interaction were detected using oligonucleotides designed as two directly repeated AGGTCA elements spaced by 3 bp (DR) and an oligonucleotide that contains a perfect AGGTCA and one element AGTCTA different from consensus, positioned as an inverted repeat with a spacing of 3 bp (C3). Similar results were obtained using gel shift analysis.
To obtain kinetic rates of ER-ERE interaction, gradient surfaces of immobilized DNA were used. We show that immobilized DNA can be saturated with hER which implies that the interaction studied is specific. Based on the signal at equilibrium we calculated the stoichiometry between ER and ERE. This result reconfirms that ER binds DNA as a homodimer.
Using real time interaction analysis, we found that ligand binding
dramatically affects kinetics of hER interaction with the Vit.A2 response element. We also found that binding of
estradiol induces rapid formation of a unstable ER·ERE complex and,
furthermore, binding of "pure" antagonist such as ICI-182,780
results in a slow formation (ka is approximately 100 times lower) and a very stable receptor-DNA complex
(kd is almost 100 times lower). Most importantly, we
demonstrate that there is a good correlation between the kinetics of
hER-ERE interaction induced by a hormone and its biological effect. For
example, the stability of ER·Vit.A2 complex, which can be
characterized by its kd is decreasing from
E2 (pure agonist) 17
-ethynyl estradiol > 4(OH)-tamoxifen (partial agonist) > raloxifene (partial antagonist) > ICI-182,780 (pure antagonist). It is interesting that this order
corresponds to the increase in the antagonistic activity of these
compounds on consensus ERE (49).2
Currently, we do not know the precise mechanism whereby differences in kinetics of receptor-DNA interaction induced by the binding of the ligand could be related to the observed behavior of the estrogen receptor in vivo. It is clear, however, that binding of estradiol accelerates the ER turnover more than 50-fold, compared with unliganded ER, and more than 1000-fold compared with ER liganded with ICI-182,780. Therefore, ligand binding inducing conformational changes is not just modulating receptor's interaction with other transcriptional factors (11-13). Our data suggest that these changes are affecting the kinetics of receptor-DNA binding which regulates the frequency of receptor-DNA complex formation. We hypothesize that a correlation exists between the rate of gene transcription and the frequency of receptor-DNA complex formation.
Various manifestations of the ligand binding including dissociation from hsp90 and p59, modulation of receptor dimerization status, modification of the kinetics of receptor-DNA interaction, and effects on receptor interaction with transcriptional intermediary factors represent different levels at which ligands of steroid receptors may affect transcriptional regulation. We have investigated the role of the estrogen receptor ligands in receptor's interaction with the consensus ERE. The pattern that we found may, however, be different for other DNA elements or it may be modified by other transcription factors interacting with estrogen receptors, which probably can explain the tissue-specific effect of different ligands of estrogen receptor.
We thank Dr. Chen-Shian Suen for providing the anti-human polyclonal antibody against the AB region of the estrogen receptor. We also thank Drs. B. Komm and S. Lundeen for critically reading the manuscript.