Article

Estrogen receptor α attenuates transforming growth factor-β signaling in breast cancer cells independent from agonistic and antagonistic ligands

Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany.
Breast Cancer Research and Treatment (Impact Factor: 3.94). 05/2009; 120(2):357-67. DOI: 10.1007/s10549-009-0393-2
Source: PubMed
ABSTRACT
To investigate a presumed crosstalk between estrogen receptor alpha (ERalpha) and the TGF-beta signaling pathway in breast cancer, we analyzed the TGF-beta-induced expression of the plasminogen activator inhibitor 1 (PAI-1) gene in ER-positive MCF-7 cells. After siRNA-mediated knock-down of endogenous ERalpha, the transcription level of PAI-1 was upregulated, pointing to an attenuation of TGF-beta signaling by the presence of ERalpha. We verified these findings by a vice versa approach using a primary ER-negative cell model transiently overexpressing either ERalpha or ERbeta. We found that ERalpha, but not ERbeta, led to a strong inhibition of the TGF-beta1 signal, monitored by TGF-beta reporter assays. This attenuation was completely independent of receptor stimulation by beta-estradiol (E2) or inhibition by the pure antagonist ICI 182.780 (ICI). Our results indicate a permanent repression of PAI-1 by ERalpha and suggest a ligand-independent crosstalk between ERalpha and TGF-beta signaling in breast cancer cells.

Full-text

Available from: Matthias B Stope
PRECLINICAL STUDY
Estrogen receptor a attenuates transforming growth factor-b
signaling in breast cancer cells independent from agonistic
and antagonistic ligands
Matthias B. Stope Æ Simone L. Popp Æ
Cornelius Knabbe Æ Miriam B. Buck
Received: 2 September 2008 / Accepted: 28 March 2009 / Published online: 16 April 2009
Ó Springer Science+Business Media, LLC. 2009
Abstract To investigate a presumed crosstalk between
estrogen receptor a (ERa) and the TGF-b signaling path-
way in breast cancer, we analyzed the TGF-b-induced
expression of the plasminogen activator inhibitor 1 (PAI-1)
gene in ER-positive MCF-7 cells. After siRNA-mediated
knock-down of endogenous ERa, the transcription level of
PAI-1 was upregulated, pointing to an attenuation of
TGF-b signaling by the presence of ERa. We verified these
findings by a vice versa approach using a primary
ER-negative cell model transiently overexpressing either
ERa or ERb. We found that ERa, but not ERb, led to a
strong inhibition of the TGF-b1 signal, monitored by
TGF-b reporter assays. This attenuation was completely
independent of receptor stimulation by b-estradiol (E2) or
inhibition by the pure antagonist ICI 182.780 (ICI). Our
results indicate a permanent repression of PAI-1 by ERa
and suggest a ligand-independent crosstalk between ERa
and TGF-b signaling in breast cancer cells.
Keywords Transforming growth factor b
Estrogen receptor Crosstalk
Introduction
Breast cancer is the most common cancer in women and is
characterized by several, highly variable oncogenic stadia
relating to clinical, pathological, and molecular parameters.
These different phases are defined by rearrangements of
therapeutic marker patterns as well as alterations in
response to chemo- and endocrine therapy [1, 2]. Repro-
ductive hormones, particularly estrogens, are major key
factors in breast cancer etiology and progression, conse-
quently resulting in estrogen receptors (ER) being an
important target for anti-cancer drug therapy. Moreover,
the ER status is a basic prognostic marker for primary
invasive breast cancer and an indicator for an individual
hormonal therapy [1]. The ER isoforms ERa and ERb are
members of the nuclear receptor superfamily and products
of distinct genes [3]. In breast cancer, ERa plays an
important role as a proliferative agent, thus determining
tumor progression. The later identified ERb fulfills over-
lapping but also unique tasks and, in contrast to the ERa
isoform, its function in breast cancer is not clear yet. The
classical mechanism of ER action starts with ligand rec-
ognition and leads to DNA binding of the receptor to
estrogen response elements (ERE) located in estrogen-
responsive genes. This DNA-protein interaction also
comprises the dimerization of ERs and an induction of
conformational changes of the molecules which allows co-
activator proteins to be recruited [4]. Besides, there exists
an alternative mode of action of nuclear receptors, referred
to as crosstalk. The underlying mechanism is predomi-
nantly based on protein–protein interactions, whereas DNA
binding appears to be secondary, as only one recognition
site in the target gene for one of the factors is sufficient.
This ER containing protein complex may then act as a
positive or negative regulator of transcription [
5].
4-OH-tamoxifen and ICI 182.780 (ICI, Fulvestrant,
Faslodex), the most commonly used antiestrogens, block
estrogen-stimulated tumor growth and have demonstrated
efficacy for treatment and prevention of ER-positive breast
M. B. Stope S. L. Popp M. B. Buck
Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology,
Stuttgart, Germany
C. Knabbe (&)
Department of Clinical Chemistry, Robert Bosch Hospital,
Auerbachstrasse 110, 70376 Stuttgart, Germany
e-mail: cornelius.knabbe@rbk.de
123
Breast Cancer Res Treat (2010) 120:357–367
DOI 10.1007/s10549-009-0393-2
peer-00535358, version 1 - 11 Nov 2010
Author manuscript, published in "Breast Cancer Research and Treatment 120, 2 (2009) 357-367"
DOI : 10.1007/s10549-009-0393-2
Page 1
cancer [1]. In addition to the direct antagonistic effects of
antiestrogen treatment, we have formerly shown that anti-
estrogens also induce an intensified secretion and activa-
tion of TGF-b [6]. This multifunctional cytokine plays a
dual role in tumorigenesis, reflected by the two opposing
properties of growth inhibition and tumor promotion [7].
The antiproliferative and thereby tumor suppressive char-
acter is based on mechanisms involved in cell cycle regu-
lation, differentiation, cell proliferation, genome stability,
suppression of telomerases, senescence, and apoptosis [8].
In the later stages of breast cancer, TGF-b may lose this
potential and shift to a tumorigenic phenotype [9]. In this
state, the activities of TGF-b are mainly characterized by
growth stimulation, invasiveness, and metastasis. Even
though the cytokine is in the focus of numerous studies,
neither the initial trigger nor the underlying molecular
mechanisms of this change are well understood.
Previous studies of our group emphasized the crucial role
of TGF-b signaling in antiestrogen therapy and in the pro-
gression of breast cancer in general [6, 1013]. By Kaplan–
Meier analysis, we identified a correlation between the
expression of TGF-b receptor II and a highly reduced
overall survival in ER-negative breast cancer patients [14].
These findings indicate a coupling of both groups in a
functional network and strongly presume a crosstalk
between ERa and components of the TGF-b pathway in
breast cancer cells. In this work, we established a cell model
to investigate the putative crosstalk between ER and TGF-b
signaling. We could show a ligand-independent influence of
ERa on TGF-b1 signaling mediated by Smad3 and c-fos.
Materials and methods
Cell culture
The ER-positive human breast cancer line MCF-7 and the
well known ERa- and ERb-negative epithelial cancer line
MDA-MB-435 [15] were propagated in DMEM containing
4.5 g of glucose/l (Invitrogen, Carlsbad, CA) supplemented
with 1 mM sodium pyruvat (Invitrogen), 50 lg/ml genta-
micin (Invitrogen), and 10% fetal calf serum (FCS, Sigma–
Aldrich, Deisenhofen, Germany). Cells were passaged
twice per week. Before their use in experiments, cells were
maintained for one passage in the same medium as descri-
bed above but with 5% of steroid-depleted FCS (sulfatase
and charcoal-treated FCS). All media contained phenol red,
which is known to have a weak estrogenic effect [16].
Chemicals, plasmids, and siRNAs
b-estradiol (E2), ICI and purified human TGF-b1 were
obtained from Sigma–Aldrich, Tocris Bioscience (Bristol,
UK), and R&D Systems (Wiesbaden, Germany). ERa
specific and c-jun specific antibodies were from Cell Sig-
naling Technology (Danvers, MA), antibodies raised
against Smad4 and c-fos were from Santa Cruz Biotech-
nology (Heidelberg, Germany), a Smad2 specific antibody
from BD Biosciences (Erembodegem, Belgium), and an
antibody against Smad3 from Invitrogen. The TGF-b-spe-
cific reporter plasmids p3TP [17
] and p6SBE [18] were
kindly provided by Jens Wu
¨
rthner (Macclesfield, UK) and
Werner Hilgers (Paris, France). Expression constructs for
the human ER isoforms (pERa, pERb) were generously
provided by Francois Vignon (Montpellier, France) and
Jan-Ake Gustafsson (Huddinge, Sweden). An expression
vector for Smad4 (pSmad4) was a gift from Mark de
Caestecker (Nashville,TE). The empty vectors pGL3-basic
(Promega, Mannheim, Germany), phRL-TK (Promega),
pcDNA3.1(?) (Invitrogen), pSG5 (Stratagene, La Jolla,
CA), and pCMV5 [19] served as control plasmids.
An estrogen-specific reporter plasmid (pERE) was cloned
by insertion of the estrogen response element (ERE)
sequence into the luciferase vector pGL3-basic (Promega)
by standard techniques. A PCR product was generated using
self-hybridizing primers (restriction sites in lowercase, ERE-
specific sequences in uppercase, hybridization sequences are
underlined: ERE forward 5
0
-ctcgagAGGTCACAGTGAC
CTAGGTCACAGTGACCT-3
0
, ERE reverse 5
0
-aagcttTT
ATATACCCAGATCTAGGTCACTGTGACC-3
0
). Expres-
sion plasmids for Smad3, c-jun, and c-fos were constructed
by reverse transcription using an oligo-dT primer followed
by a specific PCR (Smad3 forward 5
0
-CCATGTCGTCC
ATCCTGCCTTT-3
0
, Smad3 reverse 5
0
-CTCGAGTTAAG
ACACACTGGAACAGCG-3
0
, c-jun forward 5
0
-CACGTG
AAGTGACGGACTGT-3
0
, c-jun reverse 5
0
-TTTTTCTCT
CCGTCGCAACT-3
0
, c-fos forward 5
0
-CCTACCCAGCT
CTGCTTCAC-3
0
, c-fos reverse 5
0
-CACAGCCTGGTGTG
TTTCAC-3
0
). The resulting PCR products were cloned into
the Invitrogen plasmids pcDNA3-flag (pSmad3),
pcDNA3.1(?) (pc-jun) and pcDNA3.1(-) (pc-fos), respec-
tively. For expression of Smad2, we subcloned the open
reading frame from the expression vector pCMV6-Smad2
(clone IMAGp998C076351, Imagenes, Berlin, Germany)
into the vector pcDNA3 (Invitrogen). For ERa knock-down
experiments, we used the ERa ShortCut siRNA Mix
(New England Biolabs, Mannheim, Germany) and a Control
siRNA (Qiagen).
Transfection experiments
One day before transfection, MCF-7 and MDA-MB-435
cells were seeded into 24-well culture plates at a density of
5 9 10
4
per well. Cells were transfected using Effectene
Transfection Reagent (Qiagen) with a total amount of 0.2
or 1 lg DNA. The 1 lg approach contained 200 ng pERE
358 Breast Cancer Res Treat (2010) 120:357–367
123
peer-00535358, version 1 - 11 Nov 2010
Page 2
[pcDNA3.1(-)] or p3TP (pGL3-basic) or p6SBE (pGL3-
basic) ? 1ngphRL-TK? 500 ng pERa (pCMV5) or pERb
(pSG5). The 0.2 lg approach consisted of 40 ng p3TP (pGL3-
basic) ? 5ng phRL-TK? 100 ng pERa (pCMV5) ?
60 ng pSmad2 (pcDNA3) or pSmad3 (pcDNA3) or pSmad4
(pcDNA3) or pc-jun [pcDNA3.1(?)] or pc-fos [pcDNA3.1
(-)]. The total amount of DNA was kept constant by addition
of the corresponding control vector DNA (control vectors are
given in brackets). For real-time RT-PCR analysis, cells were
seeded as described above. After 24 h, transfection was per-
formed with 500 ng pERa DNA using Effectene Transfection
Reagent (Qiagen). Knock-down experiments were carried out
in MCF-7 cells that were plated with 5 9 10
4
cells per well in
24-well plates and incubated for 24 h at 37°C. Afterwards,
cells were transfected using the RNAiFect Transfection
Reagent (Qiagen) and ERa-specific siRNA (ERa siRNA) or
control siRNA (Con siRNA) at a final concentration of 50 nM.
At the time of transfection, cells were also treated with various
substances as indicated in the results part.
Luciferase assay
After 48 h of transfection, cells were harvested. Luciferase
activity was measured on an AutoLumat Plus (Berthold
Technologies, Bad Wildbad, Germany) using the Dual
Luciferase Reporter Assay System (Promega) according to
the manufacturers instructions. Co-transfection of the
Renilla luciferase vector phRL-TK was used as an internal
control. Transfections were done in triplicates and repeated
at least thrice in independent experiments.
Western blot analysis
For verification of the protein composition, cells were lysed
in buffer containing 20 mM Tris, pH 7.5, 150 mM NaCl,
1 mM EDTA, 1 mM EGTA, 1% Triton X-100, and 10 ll/ml
Protease Inhibitor Cocktail Set III (Merck, Darmstadt,
Germany). Protein concentration was determined using
Bradford Reagent (Bio-Rad, Munich, Germany). Subse-
quently, equal amounts of protein were separated by SDS–
PAGE and transferred to a nitrocellulose membrane
(Whatman, Dassel, Germany). After blocking with 25 mM
Tris, pH 7.4, 137 mM NaCl, 3 mM KCl, 5% nonfat dry
milk, and 0.05% Tween 20, the transferred proteins were
incubated with a primary antibody over night, followed by
incubation with a peroxidase-conjugated secondary anti-
body for 1 h. Visualization of the proteins was carried out by
chemoluminescence using the Phototope-HRP Western Blot
Detection System (Cell Signaling).
Quantitative RT-PCR analysis
Quantification of target mRNAs from transfected cells was
performed by RT-PCR, monitoring the increase in fluores-
cence of the SYBR Green dye (Roche Applied Science,
Mannheim, Germany) on a Light Cycler (Roche Applied
Science) in real time. 24 h after transfection, cells were
harvested and total RNA was isolated with the RNeasy Mini
Kit (Qiagen) as described by the supplier. Reverse tran-
scription was performed with 500 ng of total RNA and the
SuperScript III First-Strand Synthesis System for RT-PCR
(Invitrogen) using an oligo-dT primer. Sequences of prim-
ers were as follows: PAI-1 forward 5
0
-TGCTGGTGAATG
CCCTCTACT-3
0
, PAI-1 reverse 5
0
-CGGTCATTCCCAGG
TTCTCTA-3
0
, ITGB5 forward 5
0
-CTGGAACAACGGTG
GAGATT-3
0
, ITGB5 reverse 5
0
-CCATCTTGGCAGGTAG
CAGT-3
0
, TIMP-1 forward 5
0
-CTGTTGTTGCTGTGGC
TGATA-3
0
, TIMP-1 reverse 5
0
-CCGTCCACAAGCAATG
AGT-3
0
, GAPDH forward 5
0
-CGGAGTCAACGGATTTG
GTCGTAT-3
0
, GAPDH reverse 5
0
-AGCCTTCTCCATGG
TGGTGAAGAC-3
0
.
Statistical analysis
Data are given as mean ± SEM. Statistical comparisons
were performed by unpaired Student’s t test. A value of
P \ 0.05 was considered significant.
Results
Knock-down of ERa in ER-positive MCF-7 cells
by siRNA enhanced PAI-1 gene expression
According to our preclinical observations we postulated an
interference of ERa with components of the TGF-b sig-
naling pathway [14]. To verify this hypothesis, we carried
out knock down experiments using ER-positive breast
cancer cells MCF-7. Endogenously expressed ERa was
downregulated by ERa specific siRNA molecules and the
efficiency of the ERa knock-down was controlled by
western blotting. Transfection of MCF-7 cells with ERa
siRNA led to a clear reduction of the ERa protein level in
contrast to the cells transfected with a non-silencing control
siRNA (Fig. 1a).
The interference of ERa with the TGF-b signal was
measured by RT-PCR quantification of the plasminogen
activator inhibitor 1 (PAI-1) mRNA level, a TGF-b-
dependent protein. Analysis of mock transfected cells
treated with 10
-10
M TGF-b1 revealed a four-fold increase
of the PAI-1 mRNA level when compared with vehicle
treated cells (Fig. 1b) reflecting a predicted response of this
Breast Cancer Res Treat (2010) 120:357–367 359
123
peer-00535358, version 1 - 11 Nov 2010
Page 3
TGF-b target gene in ERa positive cells. After siRNA knock-
down of ERa, the stimulation of PAI-1 scaled up six-fold
when compared with BSA treated cells representing a sig-
nificant difference between ERa positive MCF-7 cells and
siRNA-mediated ERa knock-down cells (P = 0.0396).
These results strongly verified our hypothesis of an attenu-
ating effect of ERa on the TGF-b1 response. The house-
keeping gene GAPDH mRNA level was not influenced by
TGF-b1 stimulation and furthermore was unaffected by
transfection with siRNA (data not shown).
Generation and characterization of a model system
appropriate to study the ER/TGF-b crosstalk
To avoid well known difficulties of simultaneously trans-
fected siRNA and DNA molecules, we designed a model
cell system of primary ER-negativ cells transiently
overexpressing both ER isoforms. Cells were transfected
with either pERa or pERb for overexpression of the
receptors and transiently expressed ERa was detected by
Western Blotting (Fig. 5) and reporter assays (Fig. 2a, c).
The detection of ERb was restricted to the specific
response to the pERE reporter plasmid (Fig. 2b). The
biological activity of the ectopic expressed proteins was
measured by a co-transfected, ERE containing reporter
plasmid pERE (Fig. 2a), reflecting the pharmacological
response to E2. Both proteins exhibited full transcriptional
functionality and a weak basal activity which could be
stimulated in a dose dependent manner by E2. ERa activity
reached a plateau at 10
-11
M, ERb activity at 10
-9
ME2
(Fig. 2a, b). The respective E2 concentrations were used in
the following experiments to ensure full transcriptional
activity of each receptor. The basal activity of overex-
pressed ERa could be completely blocked by addition of
the antiestrogen ICI at a concentration of 10
-9
M (Fig. 2c).
This concentration was also used in the following experi-
ments to study effects of ER protein, independent of
ER-driven transcriptional activity. In all cases, mock
transfected cells showed no significant pERE activity.
Expression of ERa but not ERb diminished the TGF-b1
signal of a reporter plasmid
We investigated the effect of ERa and ERb expression on
TGF-b signaling by analyzing the activation of the two
established TGF-b dependent reporter vectors p3TP and
p6SBE. The plasmid p3TP was constructed using a part of
the promotor region of the TGF-b target gene PAI-1 [17].
Experiments were conducted in the presence of the above
mentioned E2 concentrations.
In MDA-MB-435 cells, the intrinsic p3TP-driven
luciferase activity was increased five-fold by treatment
with 10
-10
M TGF-b1. Transient overexpression of ERa
significantly reduced this activation to only two-fold
(P = 0.0328, Fig. 3a). Remarkably, the TGF-b1-mediated
activity in the presence of ERa was suppressed to a level
similar to the basal TGF-b1 response in the absence of the
receptor. The TGF-b1 response of unstimulated but ERa
transfected MDA-MB-435 cells, however, was slightly
lower than in control vector transfected cells.
In MCF-7 cells with endogenous ERa expression, basal
activity of p3TP was three-fold lower than in ERa negative
MDA-MB-435 cells (Fig. 3b). p3TP was induced two-fold
by addition of TGF-b1. Transient overexpression of ERa
significantly reduced the basal activity (P = 0.0489) and
completely abrogated the TGF-b1 dependent induction of
p3TP (P = 0.0378). The reporter plasmid activities of both
unstimulated and TGF-b1 induced MCF-7 cells transiently
overexpressing ERa were clearly reduced when compared
with mock transfected cells.
Fig. 1 Effect of ERa knock-down on TGF-b1 signal transduction.
For knock-down experiments MCF-7 cells were transiently transfec-
ted with Con siRNA or ERa siRNA and treated with 10
-10
M
TGF-b1 or vehicle. a 24 h after transfection, cells were lysed and
obtained lysates were analyzed by Western Blotting to monitor the
silencing efficiency. b 24 h after transfection, cells were harvested for
RNA preparation followed by cDNA synthesis. The amount of PAI-1
mRNA was measured by real-time RT-PCR and normalized to PCR
standards with known DNA concentrations. The value of vehicle
treated cells was set to 100. All values represent means ± SEM of at
least three independent experiments. * P \ 0.05
360 Breast Cancer Res Treat (2010) 120:357–367
123
peer-00535358, version 1 - 11 Nov 2010
Page 4
The TGF-b reporter plasmid p6SBE was activated three-
fold in MDA-MB-435 cells by stimulation with TGF-b1
(Fig. 3c). A slight attenuation of this induction by
Fig. 2 Transcriptional activity of ERa and ERb in MDA-MB-435
cancer cells. ERa and ERb negative MDA-MB-435 cells were
transiently transfected with expression vectors for ERa (d)orERb
(j) or empty control vectors (s and h, respectively). Cells were
co-transfected with a reporter vector containing an ERE (pERE) and,
incubated with varying concentrations of E2 or ICI. 24 h after
transfection, cells were harvested and assayed for luciferase activity.
pERE activity is given in arbitrary units (firefly luciferase activity
normalized to Renilla luciferase activity). All values represent
means ± SEM of at least three independent experiments. a Tran-
siently ERa expressing MDA-435 cells treated with E2. b Transiently
ERb expressing MDA-MB-435 cells treated with E2. c Transiently
ERa expressing MDA-MB-435 cells treated with ICI
Fig. 3 Effect of ERa expression on TGF-b1 signal transduction.
MDA-MB-435 or MCF-7 cells were transiently transfected with p3TP
or p6SBE and treated with 10
-10
M TGF-b1 or vehicle for 24 h. For
analysis of ERa effects on TGF-b1 signal transduction, an expression
vector for ERa or control vector was co-transfected. For full ERa
transcriptional activity all experiments were carried out in the
presence of 10
-11
M E2. 24 h after transfection, cells were harvested
and assayed for luciferase activity. p3TP and p6SBE activity is given
in arbitrary units (firefly luciferase activity normalized to Renilla
luciferase activity). All values represent means ± SEM of at least
three independent experiments. * P \ 0.05. a p3TP activity in MDA-
MB-435 cells. b p3TP activity in MCF-7 cells. c p6SBE activity in
MDA-MB-435 cells
Breast Cancer Res Treat (2010) 120:357–367 361
123
peer-00535358, version 1 - 11 Nov 2010
Page 5
overexpressing ERa was statistically not significant.
Expressed in MCF-7 cells, p6SBE showed no induction by
TGF-b1, and thus, no effect of overexpressed ERa was
detectable (data not shown).
Overexpression of ERb had no effect on the TGF-b1
dependent induction of p3TP, neither in MDA-MB-435
cells, nor in MCF-7 cells (data not shown).
ERa-mediated down-regulation of the TGF-b1 signal
affects TGF-b target genes regulated by members of the
Smad and AP-1 transcription factor families
For a more detailed characterization of TGF-b1 signaling,
we studied the influence of transiently overexpressed ERa on
TGF-b target genes which are known to be co-activated by
different sets of transcription factors. PAI-1 [9], integrin b5
(ITGB5, [20]), and the tissue inhibitor of metalloproteinase 1
(TIMP-1, [21]) were shown to be regulated by TGF-b1,
whereas GAPDH was used as a TGF-b independent gene.
Real-time RT-PCR examinations of TGF-b1 stimulated
MDA-MB-435 cells overexpressing the receptor exhibited
an upregulation of all three TGF-b target genes, expressed
as relative activations (TGF-b1 activated mRNA level per
basal mRNA level). ITGB5 and TIMP-1 showed a 1.5- to
2-fold activation (Fig. 4b, c), whereas PAI-1 was induced
by a factor of 30 (Fig. 4a). As expected, the regulation of
GAPDH was independent from TGF-b1 and revealed no
induction (Fig. 4d).
Overexpression of ERa caused a significant reduction of
the PAI-1 mRNA level when compared with TGF-b1 stim-
ulated but non-transfected MDA-MB-435 cells
(P = 0.0489, Fig. 4a). These findings are in accordance with
the previously shown experiments (Figs. 1b, 3a, b). Regu-
lation of PAI-1 gene expression after TGF-b1 stimulation is
mediated by Smad proteins and members of the AP-1 family
(reviewed in [22]). In contrast, TGF-b-dependent tran-
scription of the ITGB5 gene needs a collaboration of the
transcription factors Smad and Sp1 [
20]. This transcriptional
system, however, was not influenced by ERa overexpression
after TGF-b stimulation (Fig. 4b). TIMP-1 is transcribed by
an interaction of AP-1 proteins and additional but non-Smad
proteins [23], and was also not affected by the overexpres-
sion of ERa (Fig. 4c). The house-keeping gene GAPDH is
known to be poorly regulated and is not regulated by TGF-b
[24]. Consequently, the GAPDH mRNA level was not
influenced by overexpression of ERa (Fig. 4d).
ERa-mediated down-regulation of the TGF-b1 signal
depended on regulatory sequences specific for Smad3
and c-fos
The observations above indicated that the inhibitory effect
of ERa might be dependent on members of the Smad and
AP-1 transcription factor families. Until now, there was no
evidence for an upregulation of TGF-b signaling proteins
by ERa except for c-fos [25]. Thus, we examined the
influence of overexpressed ERa on endogenously synthe-
sized Smad and AP-1 proteins in the absence and presence
of TGF-b1 in MDA-MB-435 cells. Western Blot analysis
showed no significant impact on Smad2, Smad3, Smad4
and c-jun expression, but an E2-driven upregulation of the
c-fos protein level due to the transient expression of ERa
(Fig. 5).
Subsequently, we transfected MDA-MB-435 cells to
transiently express one of the Smad or AP-1 proteins
beside the luciferase assay plasmids and the ERa expres-
sion vector. These co-transfection experiments were car-
ried out using cells pre-incubated with E2 or ICI to
determine a conceivable influence of the pharmacological
activity of ERa. The overexpression of the transcription
factors was confirmed by Western Blotting (data not
shown).
Fig. 4 Effect of ERa expression on TGF-b target gene expression.
MDA-MB-435 cells were transiently transfected with an expression
vector for ERa or control vector and treated with 10
-10
M TGF-b1or
vehicle. 24 h after transfection, cells were harvested for RNA
preparation and cDNA synthesis. mRNA levels of the genes were
detected by real-time RT-PCR and related to DNA standards of
known concentrations. All values are expressed as relativ units
(RU: mRNA after TGF-b1 stimulation/basal mRNA) and represent
means ± SEM of at least three independent experiments. * P \ 0.05.
a mRNA induction of PAI-1. b mRNA induction of ITGB5. c mRNA
induction of TIMP-1. d mRNA induction of GAPDH
362 Breast Cancer Res Treat (2010) 120:357–367
123
peer-00535358, version 1 - 11 Nov 2010
Page 6
In cells treated with 10
-11
M E2, overexpression of
Smad2 (Fig. 6a, column 5–8) showed no significant influ-
ence on TGF-b1 response compared to control samples
without overexpressed transcription factor (Fig. 6a, column
1–4). Neither the level of TGF-b1 induction nor the effect
of overexpressed ERa was significantly affected by tran-
siently expressed Smad2. In contrast, after co-transfection
of a Smad3 expression vector, the basal p3TP activity was
significantly increased (three-fold, P = 0.0016, Fig. 6a,
columns 1 and 9) and so was the activity after TGF-b1
stimulation (four-fold, P = 0.0010, Fig. 6a, columns 2 and
10). By accessory overexpression of ERa, the TGF-b
response was strongly reduced compared with the p3TP
activity of ERa free cells (P = 0.0060, Fig. 6a, column 11
and 9). This activity was equal to the level of ERa, TGF-b1
and Smad3 negative control cells (Fig. 6a, column 11 and
1). The Smad3 overexpressing TGF-b induced sample also
showed an activity reduction comparable to the activity of
untreated cells (Fig. 6a, columns 12 and 9). Briefly
depicted, after co-transfection of a Smad3 expression
vector, the TGF-b response showed almost the same dis-
tribution of p3TP activity as the mock transfected cells but
on an approximately three-fold higher level (Fig. 6a, col-
umns 9–12 compared to 1–4). As already shown for
Smad2, co-transfection of a Smad4 expression vector
revealed no statistically significant change of p3TP activity
(Fig. 6a, columns 13–16 compared to 1–4).
Similar results were obtained after E2 and TGF-b1
co-stimulation of AP-1 transfected cells. Overexpression of
c-jun did not result in any significant activity shift in
comparison with the control cells (Fig. 6c, columns 1–8).
After co-transfection of a c-fos expression plasmid, the
basal activity of the p3TP reporter was enhanced (three-
fold, P = 0.0133, Fig 6c, columns 1 and 9) and so was the
TGF-b1 induced activity (3.5-fold, P = 0.0128, Fig. 6c,
columns 2 and 10). This increase was completely abolished
after co-expression of ERa. The value without TGF-
b1
stimulation decreased to approximately one-third of the
corresponding sample without ERa expression (P =
0.0106, Fig 6c, columns 11 and 9) and the TGF-b induced
activity of cells with overexpressed ERa was also reduced
to the level of the unstimulated, ERa-free, c-fos expressing
cells (Fig. 6c, columns 12 and 9).
Subsequently, we carried out the same set of experi-
ments, this time inhibiting overexpressed ERa transcrip-
tional activity by 10
-9
M ICI. In Smad3 overexpressing
cells, a significant increase of p3TP activity was detected
compared with control cells (2.5-fold, Fig. 6b, basal
activity P = 0.0165, columns 1 and 9, TGF-b1 activation
P = 0.0337, columns 2 and 10) as well as a reduction by
ERa (Fig. 6b, untreated reduction P = 0.0107, columns 9
and 11, TGF-b1 reduction to the range of ERa free activity,
columns 10 and 12). Also in c-fos overexpressing cells, the
increase of the TGF-b1 response and the ERa-mediated
switch of this TGF-b1 signal could be observed (Fig. 6d,
basal activity P = 0.0027, three-fold, columns 1 and 9,
TGF-b1 activation P = 0.0038, 3.5-fold, columns 2 and
10, untreated reduction P = 0.0010, columns 9 and 11,
TGF-b1 reduction to the range of ERa free activity, col-
umns 10 and 12). Smad2 (Fig. 6b, columns 5–8), Smad4
(Fig. 6b, columns 13–16), and c-jun (Fig. 6d, columns 5–8)
overexpression did not lead to significant alterations of the
TGF-b response in cells transfected or not transfected with
the ERa expression vector. Strikingly, the effect of phar-
macologically inactive ERa was nearly the same as shown
for E2-activated ERa, pointing to a mechanism indepen-
dent of the formation of a ligand–receptor complex.
Discussion
To analyze a potential correlation between ERa and the
TGF-b pathway, we transfected ERa-specific siRNA into
ERa-positive MCF-7 cells and monitored the effect on the
regulation of the TGF-b responsive gene PAI-1 [17]. Due
to the decreased expression of ERa the PAI-1 gene was
significantly upregulated. These data provide strong evi-
dence that constitutively expressed ER
a leads to a contin-
uous repression of PAI-1.
Fig. 5 Effect of ERa expression on Smad and AP-1 protein levels.
MDA-MB-435 cells were mock transfected or transiently transfected
with ERa and stimulated with 10
-10
M TGF-b1 or vehicle in the
presence of 10
-9
M E2 for 24 h. For protein analysis, cells were lysed
and equal amounts of total protein were subjected to Western Blot
analysis
Breast Cancer Res Treat (2010) 120:357–367 363
123
peer-00535358, version 1 - 11 Nov 2010
Page 7
We additionally confirmed this putative crosstalk with a
vice versa approach using a model system of ER-negative
cancer cells, transfected to transiently overexpress the ER
isoform ERa or ERb. These transfected MDA-MB-435
cells were completely sensitive to agonist and antagonist
treatment and maximal effects on ligand-dependent
receptor activation or inactivation were achieved in con-
centration ranges similar to those determined in prior
stimulation experiments using ER-positive MCF-7 breast
cancer cells [13, 26]. Interestingly, these concentrations
(10
-11
M E2, 10
-9
M ICI) also maximized the effects on
growth induction or growth inhibition of untransfected
MCF-7 cells [27, 28]. Our data showed an ER response of
the cellular model system quite similar to established breast
cancer cell lines naturally expressing ERs.
To study our previous observations in more detail, we
used the p3TP reporter plasmid, which was derived from a
PAI-1 TGF-b responsive sequence, and the reporter
p6SBE. We have shown an interference of ERa with
signaling components of the TGF-b system. While over-
expressed ERa strongly reduced the TGF-b1 signal in
ER-negative MDA-MB-435 cells, ERb had no effect.
Interestingly, Burdette and Woodruff [29] described a very
similar effect of ERa protein to the activin signaling
pathway by using the same TGF-b-sensitive reporter p3TP.
Still, the question remains why our reporter assays dis-
played ERa-sensitive signals using the p3TP construct while
the p6SBE plasmid showed only a weak or actually no signal
in MDA-MB-435 cells and MCF-7 cells, respectively. Both
reporter plasmids contain different TGF-b-sensitive binding
Fig. 6 Effect of ERa and transcription factor co-expression on
TGF-b1 signal transduction. MDA-MB-435 cells were transiently co-
transfected with expression vectors for ERa or control vector and the
transcription factors Smad2, Smad3, Smad4, c-jun, c-fos, or the
corresponding control vectors. Transfected cells were treated with
10
-11
ME2or10
-9
M ICI. TGF-b1 treatment and the luciferase
assay were performed as specified in Fig. 2. All values represent
means ± SEM of at least three independent experiments. * P \ 0.05.
a Transient co-expression of ERa and the transcription factor Smad2,
Smad3, or Smad4 and treatment with 10
-11
M E2 and 10
-10
M
TGF-b1. b Transient co-expression of ERa and the transcription
factor Smad2, Smad3, or Smad4 and treatment with 10
-9
M ICI and
10
-10
M TGF-b1. c Transient co-expression of ERa and the
transcription factor c-jun or c-fos and treatment with 10
-11
ME2
and 10
-10
M TGF-b1. d Transient co-expression of ERa and the
transcription factor c-jun or c-fos and treatment with 10
-9
M ICI and
10
-10
M TGF-b1
364 Breast Cancer Res Treat (2010) 120:357–367
123
peer-00535358, version 1 - 11 Nov 2010
Page 8
sites. p3TP contains three 12-O-tetradecanoylphorbol
13-acetate (TPA) responsive elements (TRE) and a TGFb-
specific promotor region of the PAI-1 gene [17], whereas the
p6SBE reporter includes six sequential Smad binding ele-
ments (SBE) controlling a SV40 promotor [18]. We presume
relevant differences in the transcription factor binding site
setting of both constructs, leading to a more AP-1 controlled
activation in case of the p3TP plasmid and a more Smad
induced activation for the p6SBE construct. Depending on
the cellular context these differences may draw a distinction
in the cellular response after stimulation.
Subsequently, we could identify some players contrib-
uting to the ERa-TGF-b1 interference. mRNA analysis of
different TGF-b responsive genes showed a combination of
Smad proteins, the primary signal transducers of TGF-b
signaling [7], and AP-1 transcription factors being
involved. More precisely, we identified the transcription
factors Smad3 and c-fos as switch points of the TGF-b
pathway. Both proteins led to an enhanced TGF-b1 activity
after overexpression, whereas Smad2, Smad4, and c-jun
had no effect. Nevertheless, the stimulatory effect of
Smad3 and c-fos diminished by ERa co-expression, dem-
onstrating an interaction between the receptor and TGF-b1
signal processing components.
Two types of interactions are conceivable. A molecular
interaction between ERa and proteins involved in TGF-b
signaling or regulation of gene expression by ERa. In the
latter case one has to distinguish between a direct effect by
downregulation of TGF-b signaling proteins or an indirect
effect by upregulation of TGF-b inhibitory factors. Thus, a
set of experiments was designed to differentiate between
these two possible modes of action. Even though the
recombinant ERa protein was transcriptionally competent,
a significant role of the transcriptional activity of the
receptor could be excluded. This was shown by the fact
that the suppressor function of ERa overexpressed in our
model MDA-MB-435 cells was insensitive to agonist and
antagonist stimulation.
Several lines of evidence from other groups also suggest
that a molecular interaction is responsible for this ERa
function in MCF-7 cells. Qi and co-workers [30] found
some hints on binding of ERa and c-jun as a suppressor of
stress-induced cell death. Another group detected an in
vitro binding of immobilized c-jun to recombinant ERa,
simultaneously negating an interaction of ERa and c-fos
[31]. Moreover, Matsuda et al. [32] demonstrated that ERa
suppresses TGF-b signaling in the presence of estrogen by
complex formation with Smad3, and Wu et al. [33] showed
an ERa-Smad4 interaction. Nevertheless, there are some
reports dealing with a repressor or co-repressor function of
ERa on specific DNA sequences. Green and co-workers
[34] postulate a mechanism of this ERa activity depending
on antiestrogen binding and mediated by HDAC. This
explicit model is not consistent with our data, because the
effect described in our study is independent of an anties-
trogen impact. Several additional data on an ERa-mediated
repressor function exist, but all of them are based on a
ligand-bound state of the receptor (reviewed in [4]). Fur-
ther experiments designed to investigate potential binding
partners of ERa are necessary, taking into account both
types of models, protein–protein as well as protein–DNA
binding.
What is the potential role of both, ERa and TGF-b,
embedded into the complex processes of breast cancer
incidence and progression? With regard to breast cancer
progression, PAI-1 as well as the transcription factors
Smad3 and c-fos are known as pro-oncogenic regulators of
invasive cell behavior and tumor metastasis [35, 36]. Our
data provide strong evidence that constitutively expressed
ERa leads to a continuous repression of PAI-1. This
assumption is supported by retrospective studies showing a
definite correlation between high amounts of ERa and low
PAI-1 expression [35, 37]. The pro-invasive capacity of
Smad3 is extensively reviewed by Roberts [38], and there
is also evidence for a pro-metastatic potential of c-fos,
demonstrated in hormone receptor negative breast cancer
cells [39]. Both oncogenic proteins were strongly dimin-
ished in the presence of ERa. Thus, the anti-metastatic
effect of ligand-free ERa might be mediated by knock-
down of components of the TGF-b signaling pathway,
namely by inhibition of the pro-invasive proteins PAI-1,
Smad3, and c-fos. This mechanism also explains the role of
TGF-b in ERa-positive and ERa-negative patients
according to the overall survival, which we studied in
2004. The expression of TGF-b receptor type II in ERa-
negative patients is correlated with highly reduced overall
survival, whereas simultaneous loss of both ERa and TGF-b
receptor type II is comparable with<