Mutual antagonism of estrogen receptors alpha and beta and their preferred interactions with steroid receptor coactivators in human osteoblastic cell lines

Abstract

Estrogen is a major sex steroid that affects the growth, maintenance, and homeostasis of the skeleton. Two isoforms of the estrogen receptor (ERalpha and ERbeta) mediate the transcriptional effects of estrogen. Although both isoforms of ER are present and functional in some human osteoblast (OB) cell lines, there is minimal information on the differential regulation of transcription by ERalpha and ERbeta homo- or heterodimers. This report demonstrates that ERalpha and ERbeta coexpression decreases the transcriptional capacity (relative to each ER isoform alone) on an estrogen response element-dependent reporter gene in OBs but not in other non-osteoblastic cell lines. These data suggest that ERalpha and ERbeta coexpression can differentially influence the degree of transcriptional activation in certain cell types. Interestingly, the overexpression of the steroid hormone receptor coactivator-1 (SRC1) resulted in preferential transcriptional enhancement by ERbeta as well as coexpressed ERalpha and ERbeta, whereas SRC2 overexpression appeared to preferentially enhance ERalpha transactivation. SRC3 overexpression failed to enhance estrogen-dependent transcription of any ER combination in OBs. Similar overexpression experiments in COS7 cells exhibited preferential enhancement of ERalpha function with all SRCs, including SRC3. Our data also demonstrated that SRC3 mRNA is reduced in osteoblastic cells, suggesting that SRC3 may have only a minor role in these cells. These data suggest that the transactivation capacity of various ER isoforms is both SRC species and cell type dependent.

Full text

Mutual antagonism of estrogen receptors and and their
preferred interactions with steroid receptor coactivators in
human osteoblastic cell lines
D G Monroe, S A Johnsen, M Subramaniam, B J Getz, S Khosla
1
,
B L Riggs
1
and T C Spelsberg
Department of Molecular Biology and Biochemistry, Mayo Clinic and Mayo Foundation, 200 1st Avenue SW, Rochester, Minnesota 55905, USA
1
Endocrine Research Unit, Mayo Clinic and Mayo Foundation, Rochester, Minnesota 55905, USA
(Requests for offprints should be addressed to D G Monroe; Email: Monroe.David@mayo.edu)
Abstract
Estrogen is a major sex steroid that affects the growth,
maintenance, and homeostasis of the skeleton. Two iso-
forms of the estrogen receptor (ERand ER) mediate
the transcriptional effects of estrogen. Although both
isoforms of ER are present and functional in some human
osteoblast (OB) cell lines, there is minimal information on
the differential regulation of transcription by ERand
ERhomo- or heterodimers. This report demonstrates
that ERand ERcoexpression decreases the transcrip-
tional capacity (relative to each ER isoform alone) on an
estrogen response element-dependent reporter gene in
OBs but not in other non-osteoblastic cell lines. These
data suggest that ERand ERcoexpression can differ-
entially influence the degree of transcriptional activation in
certain cell types. Interestingly, the overexpression of the
steroid hormone receptor coactivator-1 (SRC1) resulted in
preferential transcriptional enhancement by ERas well
as coexpressed ERand ER, whereas SRC2 over-
expression appeared to preferentially enhance ER
transactivation. SRC3 overexpression failed to enhance
estrogen-dependent transcription of any ER combination
in OBs. Similar overexpression experiments in COS7 cells
exhibited preferential enhancement of ERfunction with
all SRCs, including SRC3. Our data also demonstrated
that SRC3 mRNA is reduced in osteoblastic cells, sug-
gesting that SRC3 may have only a minor role in these
cells. These data suggest that the transactivation capacity of
various ER isoforms is both SRC species and cell type
dependent.
Journal of Endocrinology (2003) 176, 349–357
Introduction
Estrogen is involved in mediating important physiological
processes in numerous target tissues including breast,
uterus, brain, and bone (reviewed in Rickard et al. 2000).
The effects of estrogen were initially considered to
be mediated by a single estrogen receptor (ER)
(Mangelsdorf et al. 1995); however, a second estrogen
receptor (ER) that increases the potential diversity of
estrogen responses was discovered (Kuiper et al. 1996,
Mosselman et al. 1996, Tremblay et al. 1997). The two
ER isoforms function by binding 17-estradiol (E2) with
high affinity. The ‘activated’ receptor isoforms subse-
quently dimerize with either themselves (e.g. forming
homodimers) or with the other ER isoform (e.g. forming
heterodimers) (Cowley et al. 1997). The receptor recog-
nizes specific cis-acting DNA elements (estrogen response
elements or EREs) located within the regulatory regions of
target genes and activates transcription through recruit-
ment of numerous coactivators and components of the
basal transcriptional machinery. ERand ERshare
significant homology within the DNA-binding domain
and ligand-binding domains (Rickard et al. 2000), sug-
gesting that both ER isoforms bind similar ligands and
EREs (Yi et al. 2002). However, significant differences in
tissue specificity and responses to certain antiestrogens
(Paech et al. 1997) demonstrate that the ER isoforms are
unique and may perform different functions in different
cell types.
It is well accepted that estrogen plays a pivotal role in
bone cell metabolism and overall skeletal homeostasis,
which involves the regulatory actions of the bone-forming
osteoblasts (OBs). The development of specific ER
knockouts in mice has proven useful in understanding
estrogen action on the skeleton. Homozygous deletion of
the ERgene in mice (ERKO) results in decreased
longitudinal bone growth, decreased cortical bone density,
as well as decreased bone formation (Korach et al. 1996,
Lindberg et al. 2002), suggesting that ERis of critical
importance in promoting overall bone growth. In contrast,
349
Journal of Endocrinology (2003) 176, 349–357
0022–0795/03/0176–349 2003 Society for Endocrinology Printed in Great Britain
Online version via http://www.endocrinology.org

homozygous deletion of the ERgene in female mice
(BERKO) results in increased longitudinal bone growth
and increased bone mineral density (Windahl et al.
1999, Lindberg et al. 2002, Sims et al. 2002). Taken
together, these data suggest that ERmay exert a negative
effect on ER-mediated bone growth in mice. Although
this concept of ERantagonism has been suggested in
other cell systems (Hall & McDonnell 1999, Pettersson
et al. 2000), the physiological consequences of ER
coexpression with ERin bone cells, such as OBs, are
unknown. The adoption of similar principles of ER
function between the animal and human skeleton is
complicated by the observation that, in contrast to current
mouse models, a natural mutation of ERin man, which
inactivates the receptor, results in continued longitudinal
bone growth (Smith et al. 1994). This suggests that
estrogen has different effects on ER-mediated bone
growth in humans and mice. Thus, the understanding of
ER isoform actions and interactions in a human OB
system is critical to developing models for estrogen
action on the skeleton and developing new selective ER
modulators for the clinical treatment of osteoporosis.
Nuclear hormone receptor coactivators are involved in
enhancing the ligand-dependent transcriptional signal of
numerous nuclear hormone receptors, including ER. The
p160 family of coactivators includes steroid hormone
receptor coactivator (SRC)-1 (also called p160/
ERAP160) (Onate et al. 1995), SRC2 (also called GRIP1
and TIF2) (Voegel et al. 1996), and SRC3 (also called
TRAM1, ACTR, AIB1, RAC3, and p/CIP) (Li et al.
1997). Gene deletion experiments in mice demonstrate
that while homozygous deletion of SRC1 results in partial
hormone resistance (Xu et al. 1998, Weiss et al. 1999), the
phenotype of SRC1 null mice otherwise appears largely
normal. Interestingly, the SRC3 null mice demonstrate
dwarfism and a more severe reproductive phenotype (Xu
et al. 2000). However, the ultrastructure, densitometry,
and skeletal/bone growth patterns have not been exam-
ined in any SRC knockout model, nor have the coactiva-
tion properties of the SRCs been studied in a human OB
system.
This paper addresses several important and novel aspects
of ER function in OBs: the interactions between and
transcriptional regulation by coexpressed ERand ER,
the effects of SRC overexpression on ER-dependent
transcription, and the comparison of these effects between
osteoblastic and non-osteoblastic cells. As a first step to
understanding these aspects of ER function in human
OBs, it is important to first characterize these responses.
Materials and Methods
DNA constructs and reagents
Full-length cDNAs for ERand ERcontaining an
N-terminal FLAG epitope were constructed in the
pcDNA4/TO expression vector (Invitrogen, Carlsbad,
CA, USA) using PCR. Briefly, oligonucleotide primers
were designed to amplify the full-length ER(amino
acids 1–595) that contained BamHI and XhoI restriction
sites (5and 3respectively) for subcloning into pcDNA4/
TO. Full-length ER(amino acids 1–530) was con-
structed using primers containing HindIII and XbaI
restrictions sites (5and 3respectively) and subsequent
subcloning into pcDNA4/TO. A construct containing a
canonical ERE upstream of luciferase (ERE-TK-LUC)
was used as a reporter in transfection experiments.
pRL-TK (expressing Renilla luciferase) was used as a
transfection efficiency control (Promega, Madison, WI,
USA). The ER antagonist ICI 182,780 was generously
provided by Zeneca Pharmaceuticals (Macclesfield,
Cheshire, UK) and E2 was purchased from Sigma (St
Louis, MO, USA).
Cell culture and transfection
Human fetal osteoblasts (hFOB) (Harris et al. 1995)
were maintained in phenol-red free Dulbecco’s modi-
fied Eagles’s medium (DMEM)/F12 media contain-
ing 10% (v/v) fetal bovine serum (FBS) supplemented
with 300 mg/l G418 and 1antibiotic/antimycotic
(Invitrogen). MG63, COS7, Hs578T, U2OS and SaOS2
cells were maintained in the same media lacking the G418
selection. All cells were transfected at a density of 50% in
six-well dishes in serum-free media with the indicated
amount of ER expression plasmid, 1·0 µg ERE-TK-LUC,
and 1·0 µg pRL-TK using lipofectamine PLUS reagent
(Invitrogen). Three hours after transfection, the hFOB
cells were washed and treated with ICI 182,780 (10 nM)
in DMEM/F12 media containing 10% (v/v) charcoal-
stripped FBS for 24 h to eliminate basal estrogen-
independent activation of the reporter construct. All other
cell lines were maintained in similar media lacking ICI
182,780 during this period. Subsequently, the cells were
treated either with ethanol vehicle or 10 nM E2 in
DMEM/F12 media containing 10% (v/v) charcoal-
stripped FBS for an additional 24 h. It is important
to mention that following removal of the ICI 182,780
from the hFOB cell media, no subsequent estrogen-
independent activation of the reporter construct was
observed in the absence of E2 during the 24-h treatment
period (data not shown). The cells were harvested and
30 µl cell extract was assayed using the Dual Luciferase
Reporter System (Promega) with a Turner Designs
20/20 luminometer (Sunnyvale, CA, USA). Promoter
activity was quantified as a ratio of firefly luciferase
to Renilla luciferase (expressed as relative light units;
RLU). Experiments were repeated three times and a
representative experiment is shown.
Western blot analysis
hFOB cells were transfected with 1·0 µg ERor ER
expression plasmid using lipofectamine PLUS reagent.
D G MONROE and others · ER isoforms and interactions with SRCs350
www.endocrinology.orgJournal of Endocrinology (2003) 176, 349–357

Total protein extracts were prepared in RIPA buffer (1%
(v/v) NP40, 0·5% (v/v) sodium deoxycholate, 0·1%
(v/v), SDS, and 1protease inhibitor cocktail (Roche
Diagnostics Corporation, Indianapolis, IN, USA) in
1phosphate-buffered saline) and 75 µg of the protein
extract was subjected to Western blot analysis and detected
using an -FLAG-M2 antibody (Sigma) followed by an
-mouse IgG secondary antibody (Sigma). The proteins
were visualized using an enhanced chemiluminescence kit
(Amersham Pharmacia, Piscataway, NJ, USA).
Real-time PCR analysis
Total cellular RNA from triplicate plates of COS7,
MCF7, hFOB, MG63, U2OS, and SaOS2 cells were
harvested using Trizol Reagent (Invitrogen). Four micro-
grams of total RNA were heat denatured at 68 Cfor
15 min in a reverse transcription reaction buffer (11st
strand buffer (50 mM Tris–HCl, 75 mM KCl, 3 mM
MgCl
2
), 50 mM dithiothreitol, 1 µM dNTPs, and 500 ng
oligo-dT primer). Following heat denaturation, 1 unit
MMLV-RT (Invitrogen) was added and the mixture
incubated at 37 C for 45 min followed by a 68 C
incubation for an additional 15 min. The resultant cDNA
products were diluted to 50 µl with PCR grade water and
2 µl was used in a real-time PCR reaction (1PCR
buffer (20 mM Tris–HCl, 50 mM KCl), 3 mM MgCl
2
,
300 nM of both 5and 3primer, 1Sybr Green (Mol-
ecular Probes, Eugene, OR, USA), and 1 unit Taq
Polymerase (Promega)). The primer sequences used for
real time PCR are as follows: SRC1 5primer (5-TGC
CTC CGG GTA TCA GTC ACC AG), SRC1 3primer
(5-AGG CGT GGG CTG GTT CTG GAC AG), SRC2
5primer (5-GTG GTA TGC CAG CAA CTA TGA
GC), SRC2 3primer (5-TGG ATC AGG TTG CTG
ACT TAT TCC G), SRC3 5primer (5-ACA ACC
AGA TCC AGC CTT TGG TC), SRC3 3primer
(5-TGG ATG CAG CCT GCG GGT GTT GC),
-actin 5primer (5-TCA CCC ACA CTG TGC CCA
TCT ACG A), and -actin 3primer (5-CAG CGG
AAC CGC TCA TTG CCA ATG G). The reactions
were amplified using the I-Cycler (BioRad, Hercules,
CA, USA) using the following thermal protocol: 1 cycle at
94 C (1 min), 35 cycles at 94 C (30 s), 65 C (30 s), and
72 C (30 s). The C
T
measurement is defined at the
fractional cycle number at which the amount of amplified
target reaches a fixed threshold above background Sybr
Green fluorescence. The amount of target in the cDNA
sample relative to -actin was calculated by the formula:
2
C
T
, where C
T
is the difference between the
average C
T
value for the target (SRC1, SRC2, or SRC3)
and -actin.
Results
Development of FLAG-tagged ER expression constructs and
determination of transfection conditions
For this study, N-terminal FLAG epitope-tagged ERand
ERexpression constructs were developed (see Materials
and Methods and Fig. 1A). Figure 1B demonstrates that
transfection of equimolar amounts of ERand ER
expression construct in hFOB cells produces proteins of
66 kDa and 54 kDa respectively, which are expressed at
approximately equal levels. To demonstrate that these
receptors are functional, either the ER–FLAG or ER–
FLAG construct was transiently co-transfected with an
ERE–reporter construct into hFOB. Figure 1C demon-
strates that these receptors are functional, i.e. displays
E2-inducible transcription of an ERE-dependent reporter
gene when the cells were treated with E2. Under these
conditions, each isoform exhibits a comparable activity. In
order to determine the optimal amount of expression
construct to achieve the maximal transcriptional response,
a titration of ER–FLAG and ER–FLAG expression
constructs was performed in hFOB cells. Interestingly,
Fig. 2 demonstrates that cells containing the lowest
Figure 1 Development of FLAG epitope-tagged (TAG) ERand
ERexpression constructs. (A) Expression constructs containing an
N-terminal FLAG epitope were constructed using PCR and
subcloned into the pcDNA4/TO expression vector. (B) One
microgram of each ER–FLAG construct was transfected into hFOB
cells. Following 24 h to allow for expression, the cells were
harvested and 75 g total cellular extract was separated by PAGE
and transferred to a nitrocellulose membrane. The membrane was
probed with an anti-FLAG-M2 antibody and visualized using
enhanced chemiluminescence. (C) Ten nanograms of each
ER–FLAG construct was cotransfected with 1 g ERE-TK-LUC and
1gRenilla luciferase construct into hFOB cells. Following a 24-h
E2 treatment (10 nM E2), the cells were harvested and assayed for
luciferase activity. The bars represent fold induction by E2S.D.
and the asterisks represent significance at the P<0·001 level
(ANOVA) compared with the vehicle (EtOH control).
ER isoforms and interactions with SRCs ·D G MONROE and others 351
www.endocrinology.org Journal of Endocrinology (2003) 176, 349–357

amounts of the transfected ER constructs (10 ng) display
the greatest fold of E2-inducible transcription (approxi-
mately fourfold) for both ER isoforms. Other experiments,
using 5 ng ER, demonstrated similar levels of induction as
the 10 ng dose (data not shown). Therefore, 10 ng total
ER expression construct(s) was used throughout this study.
ERand ERcoexpression exhibits lowered transactivation
capacity in OBs but not in non-OBs
Gene deletion experiments in mice suggest that ERand
ERhave opposing actions on bone growth (Korach et al.
1996, Windahl et al. 1999). Therefore, since ERand
ERare coexpressed in OBs (Arts et al. 1997, Rickard
et al. 2000), we tested the hypothesis that coexpression of
ERand ERhas an antagonistic effect on ERE-
dependent (E2-inducible) reporter activity. Figure 3A
demonstrates that expression of ERor ERalone (10 ng
each) results in equivalent E2 induction of ERE-
dependent transcription in hFOB cells. Interestingly,
coexpression of equimolar amounts of ERand ER
(10 ng total ER) results in a statistically significant 31%
decrease in E2-inducible transcription in hFOB cells (Fig.
3A). Similar results are shown in MG63 osteosarcomas,
which also lack detectable endogenous ER expression
(Lambertini et al. 2002), where a statistically significant
47% decrease in E2-inducible transcription was observed
(Fig. 3B). This demonstrates that various OB cell lines,
coexpressing both ERand ERin equimolar ratios,
exhibit diminished E2-inducible responses. Since it is well
established in the literature that coexpression of equimolar
amounts of ERand ERexpression construct results in
primarily heterodimer formation (Cowley et al. 1997,
Tremblay et al. 1999), these studies reflect the transcrip-
tional effects of the ER/heterodimer and ERand
ERhomodimers on ERE-dependent promoter activity.
Furthermore, these data suggest that the ER heterodimer
has a unique transcriptional function since its E2-inducible
activity is significantly lower than either ER isoform alone.
To examine the cell type specificity of this ER isoform
antagonism, similar experiments were conducted in the
monkey kidney cell line, COS7, and the human breast
carcinoma, Hs578T, both of which lack detectable
endogenous ER expression (Gopalakrishna et al. 1999,
Kahlert et al. 2000). Figure 4A demonstrates that the
coexpression of ERand ERin COS7 cells results in no
difference in E2-inducible transcription compared with
Figure 2 The effect of cotransfecting different ER amounts into
hFOB cells on the ERE reporter gene activity under E2 stimulation.
Increasing amounts of ERor ERexpression construct (10, 20,
40, 60, 80 ng) were cotransfected with 1 g ERE-TK-LUC and 1 g
Renilla luciferase into hFOB cells. Following a 24-h E2 treatment
(10 nM), the cells were harvested and assayed for luciferase
activity. The bars represent fold induction by E2S.D. and the
asterisks represent significance at the P<0·01 level (ANOVA)
compared with the 10 ng data point for each ERand ER
titration.
Figure 3 ERand ERcoexpression results in transcriptional
antagonism on ERE-reporter activity in hFOB cells and MG63
osteosarcomas. (A) ER(10 ng), ER(10 ng), or ER(5 ng) and
ER(5 ng) together (ER/) were cotransfected with 1 g
ERE-TK-LUC and 1 gRenilla luciferase into hFOB cells. Following
a 24-h E2 treatment (10 nM E2), the cells were harvested and
assayed for luciferase activity. (B) This panel represents the
identical experimental procedure as in (A) except that the
transfected cells were MG63 osteosarcoma cells. Each
experimental condition was performed in triplicate and a
representative experiment is shown. Luciferase values represent
the meansS.D. and the asterisks represent significance at the
P<0·001 level (ANOVA).
Figure 4 ERand ERcoexpression does not result in
transcriptional antagonism on ERE-reporter activity in COS7 and
Hs578T cells. (A) A total of 10 ng ER,ER,orERand ER
together (ER/) were cotransfected with 1 g ERE-TK-LUC and
1gRenilla luciferase into COS7 cells. Following a 24-h E2
treatment (10 nM E2), the cells were harvested and assayed for
luciferase activity. Luciferase values represent the meansS.D. (B)
This panel represents the identical experimental procedure as in
(A) except that the transfected cells were Hs578T breast
carcinoma cells. Each experimental condition was performed in
triplicate and a representative experiment is shown.
D G MONROE and others · ER isoforms and interactions with SRCs352
www.endocrinology.orgJournal of Endocrinology (2003) 176, 349–357

either isoform alone. In contrast, in Hs578T cells, ER
had weaker transcriptional activation capability compared
with ERand coexpression of ERand ERresulted
in an activity intermediate to either ER isoform alone
(Fig. 4B). Importantly, coexpression of ERand ERin
Hs578T cells does not exhibit a similar type of transcrip-
tional antagonism as seen in OBs, where coexpressed ER
and ERresulted in lowered E2-inducible activity than
either ER isoform alone (Fig. 3). Notably, the observation
that ERis a weaker activator than ERin Hs578T cells
has been described in other non-OB cell types (i.e.
Hep2 G cells) (Hall & McDonnell 1999), demonstrating
fundamental differences between the ERand ER
homodimer activities when expressed in breast carcinoma
and OB cells.
Enhancement of ER signaling by SRCs is both SRC and cell
type dependent
Transcriptional responses elicited by the ER are mediated
by a group of nuclear transcription factors termed ‘steroid
receptor coactivators’ or SRCs (McKenna et al. 1999).
The specific modulation of SRC expression levels or their
coactivation properties with ER may explain the transcrip-
tional differences of ER function in different cell types
(Figs 3 and 4). Although SRC-dependent coactivation of
both ER isoforms has been demonstrated in some cell
types, the effects of SRC coexpression on ER function in
OBs have not been explored. Therefore, in order to
understand the contribution of SRC1, SRC2, and SRC3
to E2 responses in hFOBs, titration of these SRC mol-
ecules with either ER isoform alone, or coexpressed ER
and ER, was performed. Titration of the SRC1 expres-
sion construct ranging from 0 to 80 ng resulted in only a
slight enhancement of ER-dependent transcription
on the ERE-luciferase construct (Fig. 5A). However,
overexpression of SRC1 resulted in a dose-dependent
enhancement of ER-dependent transcription (Fig. 5A).
The results support a preferential enhancement of ER
function. The transcriptional enhancement of coexpressed
ERand ER, reported in the literature to create pri-
marily heterodimers (Cowley et al. 1997, Tremblay et al.
1999), appeared similar to that of ERat higher SRC1
concentrations. Titration of SRC2 resulted in a dose-
dependent enhancement in cells containing ER,ER,or
Figure 5 Overexpression of SRC1, SRC2, or SRC3 results in an
SRC-specific and ER isoform-specific enhancement of
E2-dependent transcription in hFOB cells. Ten nanograms of ER,
ER,orERand ER(ER/) was cotransfected with 1 g
ERE-TK-LUC, 1 gRenilla luciferase, and increasing concentrations
(0, 20, 40, 60, 80 ng) of (A) SRC1, (B), SRC2, or (C) SRC3
expression constructs into hFOB cells. Following a 24-h E2
treatment (10 nM), the cells were harvested and assayed for
luciferase activity. Luciferase values represent the meansS.D.The
asterisks in (A) denote statistical significance between ERand
ERor ER/at the P<0·001 level (ANOVA). The asterisks in (B)
denote statistical significance among all ER combinations at the
P<0·01 level (ANOVA). Since estrogen-independent effects of SRC
overexpression were not observed, only the data points
representing estrogen treatment are shown. Each experimental
condition was performed in triplicate and a representative
experiment is shown.
ER isoforms and interactions with SRCs ·D G MONROE and others 353
www.endocrinology.org Journal of Endocrinology (2003) 176, 349–357

coexpressed ERand ER; however, SRC2 consistently
enhanced ER-dependent transcription to a greater
extent than ER(Fig. 5B). The coexpression of ERand
ERresults in an intermediate transcriptional activity of
either ER isoform alone. Surprisingly, overexpression of
SRC3 had no statistically significant effect on transcrip-
tional enhancement with any ER isoform combination
tested (Fig. 5C).
To examine the cell type specificity of these SRC-
dependent responses, similar studies were conducted in
COS7 cells. Figure 6A–C demonstrates that ERfunction
is enhanced to a greater extent than that of ERby all the
SRCs tested, and that the transcriptional enhancement
of coexpressed ERand ERappears intermediate to
either ER isoform alone in the presence of all the SRCs.
Thus, the preferential enhancement of SRC-dependent
coactivation, as observed in Fig. 5 with hFOB cells,
was not observed in COS7 cells. This demonstrates an
apparent cell type preference of SRC function with the
various ER dimers.
SRC3 mRNA is reduced in osteoblastic cell lines when
compared with non-osteoblastic cell lines
A plausible explanation for the lack of transcriptional
enhancement of SRC3 with any ER combination in the
hFOB cells is that sufficient SRC3 is already present in
these cells and that additional (e.g. transfected) SRC3 will
not increase the ER-mediated transcriptional response.
Therefore, we examined the relative levels of endogen-
ous SRC1, SRC2, and SRC3 mRNAs in hFOB and
COS7 cells using real-time PCR. Figure 7 demonstrates
that no statistically significant difference was observed in
SRC1 and SRC2 mRNA levels between the hFOB and
COS7 cells. However, a 20-fold decrease in SRC3
mRNA was observed in hFOB cells when compared with
COS7 cells, strengthening the notion that sufficient SRC3
is not present in hFOB cells to increase ER transactivation.
To further explore this phenomenon, SRC3 mRNA
levels were examined in another non-OB cell line, MCF7,
and three additional human osteoblastic cell lines (MG63,
U2OS, and SaOS2). Figure 8 demonstrates that all the
osteoblastic cell lines tested have significantly reduced
SRC3 mRNA levels when compared with either COS7
or MCF7 cells.
Figure 6 Overexpression of SRC1, SRC2, or SRC3 results in
preferential enhancement of E2-dependent transcription by ERin
COS7 cells. Ten nanograms of ER,ER,orERand ER(ER/)
were cotransfected with 1 g ERE-TK-LUC, 1 gRenilla luciferase,
and increasing concentrations (0, 20, 40, 60, 80 ng) of (A) SRC1,
(B), SRC2, or (C) SRC3 expression constructs into COS7 cells.
Following a 24-h E2 treatment (10 nM), the cells were harvested
and assayed for luciferase activity. Luciferase values represent the
meansS.D. All asterisks denote statistical significance among all
ER combinations at the P<0·001 level (ANOVA). Since
estrogen-independent effects of SRC overexpression were not
observed, only the data points representing estrogen treatment are
shown. Each experimental condition was performed in triplicate
and a representative experiment is shown.
D G MONROE and others · ER isoforms and interactions with SRCs354
www.endocrinology.orgJournal of Endocrinology (2003) 176, 349–357

Discussion
This report addresses two important and largely unex-
plored aspects of ER function in OBs; the function of
ERand ERcoexpression and the effects of SRC
overexpression on E2-dependent transcription. We have
demonstrated that dual expression of ERand ERin
either hFOB or MG63 osteosarcoma results in an
estrogen-dependent transcriptional response that is 31%
and 47% lower respectively than either ER isoform acting
alone. Similar experiments in two non-OB cell lines,
COS7 and Hs578T, failed to exhibit the same transcrip-
tional antagonism when ERand ERwere coexpressed,
suggesting that transcriptional antagonism elicited by ER
and ERcoexpression is a cell type-specific phenomenon.
Earlier studies demonstrated that the ratio of ER:ER
changes significantly throughout OB differentiation (Arts
et al. 1997). Our data demonstrating lowered activation
potential when ERand ERare coexpressed suggest that
altering ERand ERratios (which presumably shifts the
balance of ER homo- and heterodimers) during OB
differentiation may potentially have important implications
in the overall magnitude of estrogen responses. These data
suggest that factors other than ER/heterodimerization
itself, presumably cell type-specific factors, are responsible
for determining the degree of E2-dependent transcription
at a classical ERE in various cell types. We are currently
investigating how dual expression of both ER isoforms
would affect different classes of E2-responsive elements
(i.e. YERE/SP1 and AP1 elements) in this human OB
system.
A potential mechanism explaining the differences in ER
activity in divergent cell types (OB versus non-OB) may
involve the differential expression or recruitment of tran-
scriptional accessory factors, such as the SRCs. Therefore,
we overexpressed SRC1, SRC2, and SRC3 with various
ER isoform combinations to test this hypothesis. We have
demonstrated that in hFOB cells, the overexpression of
SRC1 preferentially enhances ER-induced transcription
whereas SRC2 enhances ER-induced transcription to a
greater extent than either ERor coexpressed ERand
ER. This finding may have fundamentally important
implications on the activation of estrogen-responsive genes
that are dependent on a classical ERE for activation. The
activation of certain genes would be theoretically depen-
dent on the particular ER dimer bound at the promoter,
the concentration of each coactivator in the OB nucleus,
and possibly the promoter element involved. Our hypoth-
esis is that the modulation of ER dimers (e.g. homo- or
heterodimers), as occurs in in vivo osteoblastic differentia-
tion (Arts et al. 1997), acts to ‘select’ a particular set of
accessory factors (e.g. SRCs) for transcriptional activation.
Interestingly, recent biochemical purification of SRC1
and SRC3 from HeLa nuclear extracts revealed that these
coactivators reside in unique complexes containing unique
enzymatic activities (Wu et al. 2002). Therefore, specific
ER dimers may determine the affinity for a particular
coactivator complex and elicit specific coactivation func-
tions based on which coactivator complex is recruited.
Further delineation of the specific SRC complexes
involved in the regulation of estrogen-responsive genes
and of how the ER dimers differentially interact with
these coactivator complexes would significantly aid in
testing this hypothesis.
Our data also suggest that the patterns of SRC-
dependent coactivation of transcription with a particular
ER dimer is a tissue-specific phenomenon, as SRC1,
SRC2, or SRC3 overexpression enhanced ER-induced
transcription to higher levels than either ERor coex-
pressed ERand ERin COS7 cells. This agrees with a
previous observation of SRC1 enhancement of ER-
induced transcription in COS1 cells (Sheppard et al.
2001). More specifically, our data extend this observation
to demonstrate that SRC2 and SRC3 overexpression also
preferentially enhances ER-induced transcription over
that of either ERor coexpressed ERand ER. Taken
Figure 7 Comparison of SRC1, SRC2, and SRC3 mRNA levels
demonstrates reduced SRC3 mRNA in hFOB cells. mRNA from
COS7 and hFOB cells were subjected to RT-PCR using real-time
PCR technology as described in the Materials and Methods
section. The data are plotted relative to -actin and normalized to
the COS7 data (arbitrarily set at 100). Values represent the
meansS.D. (n=3) and the asterisk denotes statistical significance
at the P<0·001 level (ANOVA).
Figure 8 SRC3 mRNA levels are reduced in various osteoblastic
cells compared with non-osteoblastic cells. mRNA from COS7,
MCF7, hFOB, MG63, U2OS, and SaOS2 cells were subjected to
RT-PCR using real-time PCR technology as described in the
Materials and Methods section. The data are plotted relative to
-actin and normalized to the COS7 data (arbitrarily set at 100).
Values represent the meansS.D. (n=3) and the asterisks denote
statistical significance at the P<0·001 level (ANOVA).
ER isoforms and interactions with SRCs ·D G MONROE and others 355
www.endocrinology.org Journal of Endocrinology (2003) 176, 349–357

together with the observation of SRC overexpression in
hFOBs, these data suggest that fundamental differences
exist in the ability of SRCs to enhance ER isoform
function that is critically dependent on the cell type. This
also demonstrates that the lack of ER-induced transcrip-
tional enhancement in hFOB cells with SRC3 is a
biologically relevant phenomenon, and not due to techni-
cal issues with the SRC3 construct, since SRC3 enhanced
ER-dependent transcription in the COS7 cell line.
A plausible mechanism to explain the differences in the
responses of various ER dimers to a particular SRC within
a specific cell type may lie in the physical interaction
interface between the ER and SRC molecules. Previous
studies demonstrated the stoichiometry for ER–SRC
interactions are one SRC molecule per ER dimer
(Kalkhoven et al. 1998, Tremblay et al. 1999). Therefore,
the SRC interaction interface is unique depending on
whether ERhomodimers, ERhomodimers, or ER/
heterodimers are present at the ERE. Our data suggest
that these unique interaction surfaces may be critically
important in determining the strength of coactivation
depending on which SRC molecule is present. Recent
reports have demonstrated that variations in ERE
sequences influence the magnitude of SRC coactivation of
ER function (Hall et al. 2002) which, in turn, is also
affected by various estrogenic compounds (Bramlett et al.
2001, Hall et al. 2002). However, the novelty of our data
lies in the demonstration that SRCs can affect estrogen
signaling at the same ERE with the same ligand (e.g. E2)
in different ways dependent on the ER dimer and cell
type.
An extreme example of this cell-specific coactivation is
illustrated with the overexpression of SRC3 having no
effect on ER-induced transcription in OBs compared with
the statistically significant effects of SRC3 overexpression
in COS7 cells. Thus, it appears that osteoblastic cells are
somehow impaired in their capability to utilize SRC3 in
estrogenic responses, possibly due to factors specific to
cell types other than OBs. Our data also indicate that
comparatively low levels of endogenous SRC3 exist in
multiple osteoblastic cell lines. Collectively, these obser-
vations suggest that SRC3 has little function in mediating
estrogenic responses in osteoblastic cells and that these
responses are most likely dependent on SRC1 and/or
SRC2 coactivation of the ER.
In summary, the present study serves to address two
fundamental aspects of ER signaling that occurs in OBs.
First, the notion that ERand ERcoexpression results in
transcriptional antagonism in OBs, as suggested by the
bone phenotypes of the ERKO (Couse & Korach 1999)
and BERKO (Windahl et al. 1999) gene disruption
mouse models (ERand ERrespectively). Secondly, to
characterize the transcriptional responses of the ER iso-
forms (homo- and heterodimers) to the nuclear receptor
coactivators, SRC1, SRC2, and SRC3. The responses of
ER homo- and heterodimers to SRC overexpression
illuminate an alternative form of transcriptional regulation
that is dependent on the ER dimerization status, the
specific SRC present at the promoter, and the cell type.
Acknowledgements
We thank Dr Bert O’Malley for providing the SRC1
expression construct (Onate et al. 1995), Dr Pierre
Chambon for the TIF2 (SRC2) expression construct
(Voegel et al. 1996), and Dr Paul Meltzer for the AIB-1
(SRC3) expression construct (Anzick et al. 1997). This
work was supported by an NIH grant (PO1-AG04875),
NIH training grant CA09441, and the Mayo Foundation.
References
Anzick SL, Kononen J, Walker RL, Azorsa DO, Tanner MM, Guan
XY, Sauter G, Kallioniemi OP, Trent JM & Meltzer PS 1997
AIB1, a steroid receptor coactivator amplified in breast and ovarian
cancer. Science 277 965–968.
Arts J, Kuiper GGJM, Janssen JMMF, Gustafsson J-A, Lowik CWGM,
Pols HAP & Van Leeuwen PTM 1997 Differential expression of
estrogen receptors and mRNA during differentiation of human
osteoblasts SV-HFO cells. Endocrinology 138 5067–5070.
Bramlett KS, Wu Y & Burris TP 2001 Ligands specific coactivator
nuclear receptor (NR) box affinity for estrogen receptor subtypes.
Molecular Endocrinology 15 909–922.
Couse JF & Korach KS 1999 Estrogen receptor null mice: what have
we learned and where will they lead us? Endocrine Reviews 20
358–417.
Cowley SM, Hoare S, MosselmanS&Parker MG 1997 Estrogen
receptors alpha and beta form heterodimers on DNA. Journal of
Biological Chemistry 272 19858–19862.
Gopalakrishna R, Gundimeda U, Fontana JA & Clarke R 1999
Differential distribution of protein phosphatase 2A in human breast
carcinoma cell lines and its relation to estrogen receptor status.
Cancer Letters 136 143–151.
Hall JM & McDonnell DP 1999 The estrogen receptor isoform
(ER) of the human estrogen receptor modulates ER
transcriptional activity and is a key regulator of cellular response to
estrogens and antiestrogens. Endocrinology 140 5566–5578.
Hall JM, McDonnell DP & Korach KS 2002 Allosteric regulation of
estrogen receptor structure, function, and coactivator recruitment by
different estrogen response elements. Molecular Endocrinology 16
469–486.
Harris SA, Enger RJ, Riggs BL & Spelsberg TC 1995 Development
and characterization of a conditionally immortalized human fetal
osteoblastic cell line. Journal of Bone and Mineral Research 10
178–186.
Kahlert S, Nuedling S, van Eickels M, Vetter H, MeyerR&Grohe
C 2000 Estrogen receptor alpha rapidly activates the IGF-1 receptor
pathway. Journal of Biological Chemistry 275 18447–18453.
Kalkhoven E, Valentine JE, Heery DM & Parker MG 1998 Isoforms
of steroid receptor co-activator 1 differ in their ability to potentiate
transcription by the oestrogen receptor. EMBO Journal 17 232–243.
Korach KS, Couse JF, Curtis SW, Washburn TF, Lindzey J, Kimbro
KS, Eddy EM, Migliaccio S, Snedeker SM, Lubahn DB,
Schomberg DW & Smith EP 1996 Estrogen receptor gene
disruption: molecular characterization and experimental and clinical
phenotypes. Recent Progress in Hormone Research 51 159–186.
Kuiper GGJM, Enmark E, Pelto-Huikko M, NilssonS&Gustafsson
J-Å 1996 Cloning of a novel estrogen receptor expressed in rat
prostate and ovary. PNAS 93 5925–5930.
D G MONROE and others · ER isoforms and interactions with SRCs356
www.endocrinology.orgJournal of Endocrinology (2003) 176, 349–357

Lambertini E, Penolazzi L, Aguiari G, del Senno L, Pezzetti F,
SollazzoV&PivaR2002 Osteoblastic differentiation induced by
transcription factor decoy against estrogen receptor alpha gene.
Biochemical and Biophysical Research Communications 292 761–770.
Li H, Gomes PJ & Chen JD 1997 RAC3, a steroid/nuclear
receptor-associated coactivator that is related to SRC-1 and TIF2.
PNAS 94 8479–8484.
Lindberg MK, Moverare S, Skrtic S, Alatalo S, Halleen J, Mohan S,
Gustafsson J-Å & Ohlsson C 2002 Two different pathways for the
maintenance of trabecular bone in adult male mice. Journal of Bone
and Mineral Research 17 555–562.
Mangelsdorf DJ, Thummel C, Beato M, Herrlich P, Schutz G,
Umesono K, Blumberg B, Kaster P, Mark M, Chambon P &
Evans RM 1995 The nuclear receptor superfamily: the second
decade. Cell 83 835–839.
McKenna NJ, Lanz RB & O’Malley BW 1999 Nuclear receptor
coregulators: cellular and molecular biology. Endocrine Reviews 20
321–344.
Mosselman S, PolmanJ&Dijkema R 1996 ER: identification and
characterization of a novel human estrogen receptor. FEBS Letters
392 49–53.
Onate SA, Tsai SY, Tsai MJ & O’Malley BW 1995 Sequence and
characterization of a coactivator for the steroid hormone receptor
superfamily. Science 270 1354–1357.
Paech K, Webb P, Kuiper GGJM, Nilsson S, Gustafsson J-Å, Kushner
PJ & Scanlan TS 1997 Differential ligand activation of estrogen
receptors ERand ERat AP1 sites. Science 277 1508–1510.
Pettersson K, Delaunay F & Gustafsson J-Å 2000 Estrogen receptor 
acts as a dominant regulator of estrogen signaling. Oncogene 19
4970–4978.
Rickard D, Harris SA, Turner R, Khosla S & Spelsberg TC 2000
Estrogen and progestins. In Principles of Bone Biology, edn 2,
pp 655–676. Eds P Bilezikian, LG Raisz & GA Rodan. San Diego:
Academic Press.
Sheppard HM, Harries JC, Hussain S, BevanC&Heery DM 2001
Analysis of the steroid receptor coactivator 1 (SRC1)–CREB
binding protein interaction interface and its importance for the
function of SRC1. Molecular and Cellular Biology 21 39–50.
Sims NA, Dupont S, Krust A, Clement-Lacroix P, Minet D,
Resche-Rigon M, Gaillard-KellyM&BaronR2002 Deletion of
estrogen receptors reveals a regulatory role for estrogen receptor-b
in bone remodeling in females but not in males. Bone 30 18–25.
Smith EP, Boyd J, Frank GR, Takahashi H, Cohen RM, Specker B,
Williams TC, Lubahn DB & Korach KS 1994 Estrogen resistance
caused by a mutation in the estrogen receptor gene in a man.
New England Journal of Medicine 331 1056–1061.
Tremblay GB, Tremblay A, Copeland NG, Gilbert DJ, Jenkins NA,
LabrieF&GiguereV1997Cloning, chromosomal location and
functional analysis of the murine estrogen receptor .Molecular
Endocrinology 11 353–365.
Tremblay GB, Tremblay A, LabrieF&GiguereV1999Dominant
activity of activation function 1 (AF1) and differential stoichiometric
requirements for AF-1 and -2 in the estrogen receptor –
heterodimeric complex. Molecular and Cellular Biology 19
1919–1927.
Voegel JJ, Heine MJ, Zechel C, ChambonP&Gronemeyer H 1996
TIF2, a 160 kDa transcriptional mediator for the ligand-dependent
activation function AF-2 of nuclear receptors. EMBO Journal 15
3667–3675.
Weiss RE, Xu J, Ning G, Pohlenz J, O’Malley BW & RefetoffS
1999 Mice deficient in the steroid receptor co-activator 1 (SRC1)
are resistant to thyroid hormone. EMBO Journal 18 1900–1904.
Windahl S, Vidal O, Andersson G, Gustafsson J-Å & Ohlsson C 1999
Increased cortical bone mineral content but unchanged trabecular
bone mineral density in adult female estrogen receptor
/mice. Journal of Clinical Investigation 104 895–901.
Wu R-C, Qin J, Hashimoto Y, Wong J, Xu J, Tsai SY, Tsai MJ &
O’Malley BW 2002 Regulation of SRC-3 (pCIP/ACTR/AIB-1/
RAC-3/TRAM-1) coactivator activity by IB kinase.
Molecular and Cellular Biology 22 3549–3561.
Xu J, Qiu Y, DeMayo FJ, Tsai SY, Tsai MJ & O’Malley BW 1998
Partial hormone resistance in mice with disruption of the steroid
receptor coactivator-1 (SRC-1) gene. Science 279 1922–1925.
Xu J, Liao L, Ning G, Yoshida-Komiya H, DengC&O’Malley BW
2000 The steroid receptor coactivator SRC3 (pCIP/RAC3/AIB1/
ACTR/TRAM-1) is required for normal growth. PNAS 97
6379–6384.
Yi P, Driscoll MD, Huang J, Bhagat S, Hilf R, Bambara RA &
Muyan M 2002 The effects of estrogen-responsive element- and
ligand-induced structural changes on the recruitment of cofactors
and transcriptional responses by ERand ER.Molecular
Endocrinology 16 674–693.
Received in final form 5 November 2002
Accepted 7 November 2002
ER isoforms and interactions with SRCs ·D G MONROE and others 357
www.endocrinology.org Journal of Endocrinology (2003) 176, 349–357