Estrogen-Dependent and Estrogen-Independent



Estrogen-Dependent and Estrogen-Independent
Published OnlineFirst on May 4, 2010 as 10.1158/0008-5472.CAN-09-4080
Tumor and Stem Cell Biology
Estrogen-Dependent and Estrogen-Independent Mechanisms
Contribute to AIB1-Mediated Tumor Formation
Maria I. Torres-Arzayus1, Jin Zhao1, Roderick Bronson2, and Myles Brown1
We have previously reported the oncogenic properties of the gene amplified in breast cancer 1 (AIB1), a
member of the p160 family of hormone receptor coactivators. In a transgenic mouse model, AIB1 overexpression resulted in a high incidence of tumors in various tissues, including mammary gland, uterus, lung, and
pituitary. To determine whether the AIB1 oncogenicity in this model depended on its function as an estrogen
receptor (ER) coactivator, we abolished ER signaling through two independent approaches, by performing
ovariectomy on AIB1 transgenic (AIB1-tg) mice to prevent gonadal estrogen production and by crossing
AIB1-tg mice with ERα-null mutant mice. Ovariectomized (ovx) mice, but not AIB1 × ERα−/− mice, still
developed mammary gland hyperplasia and ductal carcinoma in situ. Both approaches, however, completely
prevented the development of invasive mammary tumors, indicating that invasive mammary tumor formation
is strictly estrogen dependent. Once developed, AIB1-induced mammary tumors can subsequently lose their
dependence on estrogen: Injection of ERα(+) tumor cell lines derived from such tumors into ovx or untreated
wild-type mice resulted in a similar rate of tumor growth in both groups. Surprisingly, however, ovx mice had
an ∼4-fold higher rate of metastasis formation, suggesting that estrogen provided some protection from metastasis formation. Lastly, our experiments identified oncogenic functions of AIB1 that are independent of its
ER coactivation, as both approaches, ovariectomy and ER−/− crosses, still resulted in a high incidence of tumors in the lung and pituitary. We therefore conclude that AIB1 can exert its oncogenicity through tissuespecific estrogen-dependent and estrogen-independent functions. Cancer Res; 70(10); 4102–11. ©2010 AACR.
Breast cancer is the most prevalent form of cancer in women in the United States. The prognosis and response to therapy
largely depends on tumor phenotype. Targeted therapies are
available for estrogen receptor α (ERα)–positive tumors and
for those tumors harboring HER2 amplifications. However, tumors lacking hormone receptor expression or HER2 amplification, so-called triple-negative breast cancers, lack defined
targeted therapies and have a poor prognosis. Furthermore,
both de novo and acquired resistance to both endocrine and
HER2-targeted therapies remain a significant problem, and
once patients develop metastatic disease, cures remain extremely rare.
We have previously developed a mouse model for breast
cancer by overexpressing the human gene AIB1 in the mammary gland of FVB mice (1) to investigate mechanisms of
Authors' Affiliations: 1 Division of Molecular and Cellular Oncology,
Department of Medical Oncology, Dana-Farber Cancer Institute and
Harvard Medical School and 2 Department of Pathology, Harvard
Medical School, Boston, Massachusetts
Note: Supplementary data for this article are available at Cancer
Research Online (
Corresponding Author: Myles Brown, Dana-Farber Cancer Institute, 44
Binney Street, D730, Boston, MA 02115. Phone: 617-632-3948; Fax: 617632-5417; E-mail: [email protected]
doi: 10.1158/0008-5472.CAN-09-4080
©2010 American Association for Cancer Research.
Cancer Res; 70(10) May 15, 2010
breast cancer development, metastasis formation, and drug
resistance. AIB1 was discovered as a gene frequently amplified or overexpressed in breast cancer (2). It belongs to the
p160 family of hormone receptor coactivators, and consequently, one of its most studied functions is as an ERα coactivator (3). The AIB1 transgenic (AIB1-tg) mice in our model
developed invasive mammary gland tumors with a 75% incidence, as well as frequent tumors in other organs including
the uterus, pituitary, lung, and skin. In addition, some of the
mice harboring mammary gland tumors developed metastasis in bone and brain. These results showed for the first time
the ability of AIB1 to function as an oncogene.
The development of mammary and uterine tumors in
AIB1-tg mice was consistent with its role as an ER coactivator. However, the formation of tumors in the lung, skin, and
pituitary was surprising, as these tissues are not known to be
estrogen dependent. Moreover, about 20% of the mammary
tumors in AIB1-tg mice lacked expression of ERα. These observations raised the question whether AIB1 exerted its role
as an oncogene through ERα coactivation alone or whether it
had other functions as well.
In the present study, we used two strategies to address the
role of estrogen and ERα in AIB1-mediated tumor formation.
First, we surgically removed the ovaries of prepubertal AIB1-tg
mice to block gonadal estrogen production, which is the
main source of estrogen in female mice. Ovariectomized
(ovx) AIB1-tg mice did not develop any invasive mammary
Estrogen Dependence of AIB1-Induced Tumors
gland tumors. In sharp contrast, the incidence of pituitary, lung,
skin, and bone tumors was higher than in non-ovx AIB1-tg
mice, indicating that only a subset of tumors were affected
by the abolishment of gonadal estrogen production. Second,
we crossed AIB1-tg mice with ERα-null mutant mice. Lacking
ERα, these mice are unable to respond to estrogen through this
receptor, whether the estrogen is produced in the ovaries or in
other tissues via aromatization of adrenal androgens. Again,
these mice showed no signs of mammary tumors but developed tumors in lung, skin, and pituitary with the same high
incidence as their AIB1-tg counterparts expressing ERα.
Our results indicate that AIB1 can cause tumor formation
by estrogen-dependent and estrogen-independent mechanisms, depending on the organ or tissue affected. These
results further suggest that AIB1 exerts oncogenic activities
in cell signaling and/or gene regulation that are independent
of its function as an ER coactivator.
Materials and Methods
Mouse strains and treatments. Animal experiments were
compliant with the guidelines of the Dana-Farber Cancer Institute. The AIB1-tg mouse line (FVB-MMTV-AIB1) has been
previously described (1). FVB wild-type (wt) mice were purchased from Charles River Laboratories.
Ovaries from wt and AIB1-tg mice were surgically removed
before onset of puberty. Mammary gland development was
assessed after ovariectomy by whole-mount analysis and
H&E staining of cross sections (see below). Ovx animals were
also aged and monitored for tumor formation. Mice were
sacrificed when tumors developed or at 16 months of age,
depending on tumor status and overall health conditions.
Complete necropsies were then done. DNA, RNA, and protein
were extracted from various tissues for further analysis.
Heterozygous ERα deletion mutant mice (+/−) of the
C57BL/6 strain were kindly provided by Dr. Pierre Chambon
(Institute of Genetics and Molecular and Cellular Biology,
CNRS/INSERM/University of Strasbourg, Strasbourg, France;
ref. 4). These mice were crossed with the AIB1-tg mouse line
FVB-MMTV-AIB1 to obtain the double-mutant mouse line
AIB1-tg × ERα+/− and then inbred to obtain the homozygous
null mutant strain AIB1-tg × ERα−/−. Mammary gland development and tumor formation were assessed as described above.
Mammary gland whole mounts and immunostaining.
Mammary glands were dissected, mounted on glass slides,
and stained as described in
histological/Histology/index.htm1. Tissues were fixed overnight in 10% neutralized buffered formalin and embedded
in paraffin by standard procedures. Sections of 4-μm thickness were stained with H&E or processed for immunohistochemistry analysis following antigen retrieval, using
polyclonal ERα (Santa Cruz Biotechnology; diluted 1:3,000)
or monoclonal anti-AIB1 antibody (BD Biosciences; diluted
1:200) and hematoxylin counterstaining.
RNA isolation and quantitative PCR. RNA isolation from
tissue or tumors was done with Trizol (Invitrogen) according
to the manufacturer's protocol. Five micrograms of RNA
were reverse transcribed, using the ABI High-Capacity
cDNA Reverse Transcription Kit (Applied Biosystems).
The resulting cDNAs were used for quantitative PCR analysis
using an Applied Biosystem 7300 RT PCR device and ABI
SYBR Green master mix. The following primers were used:
AIB1-sense, 5′-GCGGCGAGTTTCCGATTTA-3′; antisense,
Protein extraction and Western blot analysis. The Western blot analysis protocol has been previously described (5).
Monoclonal anti-AIB1 antibody (BD Biosciences) and polyclonal anti-aromatase (Novus) were used for Western blots.
Polyclonal calnexin (Stressgen) or β-actin antibodies (Sigma)
were used as control for protein loading.
Tumor cell injection. Tumor cell lines were produced and
maintained as previously described (5). Immortalized SCP2
cells were used as controls (6). One thousand tumor cells or
1 million control cells suspended in PBS (100 μL) were injected
in the thoracic fat pad of gland no. 3. For each experiment, five
intact female mice, five ovx female mice, and five male mice
were used. The experiment was repeated three times. After injection, animals were routinely evaluated, tumor growth was
monitored daily, and tumor size was measured using a caliper.
Animals were sacrificed when tumor size reached 2 cm3 or the
animal showed signs of discomfort or the skin was compromised. Complete necropsies were done in all animals and
tumor weight and tumor volume were assessed.
Statistical analysis. Statistical analyses for PCR and tumor injection studies were done using unpaired two-tailed
Student's t tests. Statistical significance was considered to
be present at levels greater than 95% (P < 0.05). Means and
SDs have been estimated for continuous variables. Tumorfree survival was calculated using the product-limit method
of Kaplan-Meier (7). Comparisons between subgroups of
study mice and the significance of differences between survivals rates were ascertained using a two-sided log-rank test
(8). Differences between the two populations were considered
significant at confidence levels greater than 95% (P < 0.05).
All statistical analyses were carried out using GraphPad
Prism version 5.0b for Mac OS X (GraphPad Software).
Tumors formed in AIB1-tg mice lack a correlation
between AIB1 and ER expression levels. We have previously reported that AIB1-tg mice display a high incidence
of mammary tumors. Whereas many of these mammary tumors are ERα(+), a significant proportion of them do not express detectable levels of ERα (1). This finding was somewhat
surprising as ERα coactivation is thought to be the primary
function of AIB1. This finding raised the question of whether
AIB1-induced formation of ERα(−) mammary tumors is ER
independent. Alternatively, formation of ERα(−) tumors
Cancer Res; 70(10) May 15, 2010
Torres-Arzayus et al.
Figure 1. Lack of correlation of
ERα and AIB1 expression in
mammary tumor samples from
AIB1-tg mice. A, relative mRNA
expression levels of AIB1 and ERα
were determined by quantitative
PCR in the tumor samples
indicated. B, for each tumor
sample, the relative expression
levels of AIB1 and ERα were
plotted against each other. Linear
regression analysis was done
using GraphPad Prism version
5.0b for Mac OS X.
initially requires AIB1 coactivation of ER, but the tumors
later become ER independent and lose ER expression.
Before we determined whether ERα and estrogen were
necessary for AIB1-mediated tumorigenesis, we evaluated
the correlation between AIB1 and ERα expression in mammary gland tumors derived from AIB1-tg mice. We compared
AIB1 and ERα mRNA and protein levels in a large number of
tumor samples, using quantitative PCR (Fig. 1A) and Western
blot (Supplementary Fig. S1). Notably, several tumors with
high AIB1 expression levels had undetectable or only barely
detectable levels of ERα. A statistical analysis of the expression levels revealed the lack of a correlation between AIB1
and ERα mRNA levels (Fig. 1B).
Cancer Res; 70(10) May 15, 2010
Ovariectomy impairs mammary gland development in
AIB1-tg mice. To address the causal relationship between
ER function, AIB1 expression, and mammary tumor formation, we abolished ovarian estrogen production in 25 AIB1tg FVB mice and 25 wt FVB mice by performing ovariectomy
before onset of puberty. First, we analyzed mammary gland
development in these mice and compared it to that in untreated AIB1-tg and wt mice. Whole-mount analysis of
20-week-old ovx mice revealed severe impairment of mammary gland development, consistent with previous reports showing the requirement of estrogen for normal mammary
gland development (9–11). Whereas mammary glands of
AIB1 and wt mice with intact ovaries showed complete ductal
Cancer Research
Estrogen Dependence of AIB1-Induced Tumors
Figure 2. Effect of ovariectomy on mammary gland development in AIB1-tg mice. Wt and AIB1-tg mice were ovariectomized at 3 wk of age. At 20 wk,
controls and ovx mice were sacrificed and mammary gland whole mounts or cross sections were prepared. A, whole mounts, top: left, untreated wt;
right, untreated AIB1; bottom: left, ovx wt; right, ovx AIB1 mouse (×20 magnification). Arrow, example of branching that is absent in ovx wt mice.
B, H&E-stained cross sections of mammary glands (×100). Ovariectomy results in impaired mammary gland development in wt and AIB1-tg mice. AIB1-tg
mice display hyperplasia and disorganized mammary duct architecture, even after ovariectomy. Inset, detail of a hyperplastic mammary duct.
Right, example of a DCIS observed in some of the ovx AIB1-tg mice. C, effect of ovariectomy on ERα expression. Mammary gland cross sections of
untreated and ovx wt and AIB1-tg mice were probed with ERα antibody. All mice showed comparable ERα staining. D, ERα expression of DCIS.
Cancer Res; 70(10) May 15, 2010
Torres-Arzayus et al.
elongation that fully branched out and penetrated the mammary fat pad (Fig. 2A), duct formation after ovariectomy was
only rudimentary and ductal elongation and side branching
were limited or absent in both AIB1-tg and wt mice (Fig. 2A).
We, however, observed a difference in overall mammary
gland development between AIB1-tg and wt mice after ovariectomy. Whereas there was no detectable ductal branching
in wt mice following ovariectomy, mammary glands of
AIB1-tg mice showed limited but clearly present ductal elongation and branching (Fig. 2A, arrow). These results confirm
that ovarian estrogen production is required for full ductal
development. AIB1 overexpression alone cannot compensate
for the lack of gonadal estrogen during mammary gland development, although it increases ductal elongation and
branching to some extent. Serum estrogen levels were undetectable in the ovx mice that developed uterine atrophy and
increased in body weight (not shown), in agreement with
previous studies (9, 12–14).
Despite impaired mammary gland development, ovx
AIB1-tg mice display mammary gland hyperplasia and
ductal carcinoma in situ. We have previously reported that
AIB1-tg mice develop mammary epithelial hyperplasia, as
evidenced by increased cell numbers and a disorganized pattern of cells surrounding the ducts (1). We now find that despite ovariectomy, the mammary glands of AIB1-tg mice also
show an increase in cell number and disorganized structures
as compared with wt mice (Fig. 2B). Thus, despite impaired
mammary gland development, ovx AIB1-tg mice develop
hyperplasia and, in some cases, develop ductal carcinoma
in situ (DCIS; Fig. 2B, right). These results indicate that the
apparent proliferative effects of the AIB1 gene on mammary
gland epithelial cells are exerted even in the absence of gonadal estrogen.
To determine the effect of estrogen removal on ERα expression levels, we performed immunohistochemistry. The
pattern of ERα expression in luminal epithelial cells was
not significantly different with or without ovariectomy, notwithstanding the smaller number of ducts and branches
(Fig. 2C), indicating that estrogen deficiency does not influence ER expression levels or distribution. The DCIS lesions
were also ERα(+) (Fig. 2D).
Ovx AIB1-tg mice display a reduced incidence of tumors
in hormone-dependent tissues but an increased incidence
of tumors in hormone-independent tissues. Twenty-five
AIB1-tg mice treated by ovariectomy and 27 control AIB1-tg
mice were aged and monitored for tumor formation. When
animals presented evidence of disease or, at the latest, at the
end of the study, they were sacrificed and a full pathologic
examination was conducted. Data are presented as KaplanMeier disease-free survival curves. Three categories of diseases
were scored: invasive mammary gland tumors; noninvasive
DCIS (both shown in Fig. 3A); and nonmammary gland tumors, such as lung, skin, bone, or pituitary tumors (Fig. 3B).
DCIS does not lead to visible disease and, consequently, only
was diagnosed when the animals suffered from other
symptoms or at the end point of the study at 16 months.
Most strikingly, none of the 25 ovx AIB1 mice developed invasive mammary gland tumors (Fig. 3A) or uterine tumors
Cancer Res; 70(10) May 15, 2010
(not shown). Thus, the incidence of invasive tumors in these
hormone-dependent tissues was significantly reduced as
compared with non-ovx AIB1 mice. In sharp contrast, 13 of
25 (52%) ovx AIB1 mice developed tumors in hormoneindependent tissues, including lung, skin, pituitary, and bone.
The incident of these tumors was even significantly higher
than in untreated AIB1 mice, where 6 of 27 (22%) of mice developed such tumors (Fig. 3B). These data show that AIB1
overexpression can give rise to two distinct groups of tumors:
those that are prevented by ovariectomy and thus are estrogen
dependent (Fig. 3A) and those that remain unaffected or even
increase in frequency on ovariectomy and thus are independent of gonadal estrogen production (Fig. 3B). Representative
sections of tumors developing in ovx AIB1-tg mice are shown
in Supplementary Fig. S2. They include bone sarcoma, skin
papilloma, lung carcinoma, and pituitary adenoma.
Lack of ovarian production of estrogen leads to
increased aromatase expression in the mammary gland.
An essential step of estrogen biosynthesis is catalyzed by
the enzyme aromatase. Aromatase is responsible not only
for the estrogen produced in the ovaries but also for extragonadal estrogen production, for example, in adipose tissue
Figure 3. Comparison of tumor incidence in AIB1-tg mice with and
without treatment by ovariectomy. A, Kaplan-Meier disease-free survival
curves comparing AIB1-tg mice that developed invasive mammary gland
tumors with ovx AIB1-tg mice that only developed DCIS. B, comparison
of the incidence of nonmammary gland tumors (lung, skin, bone, and
pituitary tumors) in AIB1-tg mice versus ovx AIB1-tg mice.
Cancer Research
Estrogen Dependence of AIB1-Induced Tumors
and skin (15–17). It has been suggested that aromatase expression is increased on reduction of estrogen levels in an
attempt by the organism to increase estrogen production
to compensate for reduced estrogen signaling (18, 19). To
determine whether abolishment of gonadal estrogen production leads to increased aromatase production in our system,
we analyzed aromatase mRNA expression levels by quantitative PCR in whole mammary glands from mature ovx and
non-ovx AIB1-tg and wt mice. Ovariectomy led to a 3- to
4-fold increase in aromatase levels in both AIB1-tg and wt
mice (Supplementary Fig. S3). Interestingly, AIB1-tg mice
had about 2- to 3-fold lower aromatase expression as compared
with wt mice. This difference was apparent in treated and
untreated AIB1 mice, indicating that there might also be
a negative feedback loop by which increased ER activity
(due to increased AIB1 coactivation) down-regulates estrogen production by aromatase.
AIB1 × ERα−/− double-mutant mice show impaired
mammary gland development and no mammary tumors,
but a high incidence of nonmammary tumors. We next investigated whether extragonadal estrogen was responsible
for mediating the phenotype of AIB1 mice after ovariectomy
by abolishing all estrogen signaling through removal of
ERα. We did this by generating crosses of AIB1-tg mice with
ERα-null mutant mice. This approach should allow us to unequivocally dissect ERα-dependent and ERα-independent
functions of AIB1-mediated tumorigenesis. Double-mutant
mice of the AIB1-tg × ERα−/− genotype were aged and monitored for tumor formation. When tumors were detected or
at the end point of the study, mice were sacrificed and dissected for pathologic evaluation. As expected, all 16 of the
double-mutant mice displayed impaired mammary gland
development. Specifically, mammary glands did not show
signs of ductal elongation or side branching (Fig. 4), nor
did they develop terminal end buds (not shown). A similar
mammary gland phenotype was detected in control ERα−/−
mice, in agreement with previous reports showing that
ERα−/− mice only develop a truncated ductal rudiment, demonstrating that normal mammary gland development requires
ERα signaling (20–24). As expected, no mammary or uterine
tumors were found during 16 months of aging. However,
7 of the 16 mice displayed lung or pituitary tumors, a comparable incidence to the 12 of 25 AIB1-tg mice showing the same
kinds of tumors after ovariectomy. The lung and pituitary
tumors were of similar type and showed similar onsets than
the tumors developing in the single-mutant AIB1-tg mice
(not shown). These results show that lung and pituitary tumors are indeed arising in an ERα- and estrogen-independent
Lack of estrogen dependence of tumor growth in normal
mice after injection with an ERα(+) AIB1 tumor cell line.
Thus far, we have shown that the formation of mammary
tumors in AIB1-tg mice is strictly estrogen dependent. However, that does not necessarily mean that once mammary
tumors have formed, they continue to depend on estrogen.
In fact, when we derived ERα(+) cell lines from AIB1-tg mammary tumors and injected those cell lines into normal mice,
they only showed a limited response to tamoxifen, as we previously reported (5). On injection into normal, syngeneic
mice, these mammary tumor cell lines have been shown to
be particularly aggressive, leading to uniform tumor formation in 100% of injected animals within 2 weeks (5).
We wanted to determine whether tumor growth on injection of these ERα(+) AIB1 tumor cell lines still required the
presence of gonadal estrogen. Therefore, groups of five
female wt FVB mice were ovariectomized before onset of puberty. At 20 weeks of age, mice were injected with 1,000 cells
of the ER(+) AIB1-tg tumor cell line. As control, five agematched, intact female wt mice were injected, and five wt
FVB male mice were treated with the same tumor cell line
as the females. After injection, the volume of developing tumors was determined at regular intervals. Mice were sacrificed after 24 days or when tumor volumes reached 2 cm3.
Tumors developed in 100% of the animals injected, regardless of gender or prior treatment with ovariectomy. The
group of mice injected with the tumor cell line showed no
difference in tumor growth rate (Fig. 5A). No tumors were
obtained when we injected the nontumorigenic mammary
Figure 4. Whole mounts depicting the effect of ERα removal. Left, 10-wk-old wt mouse; middle, 10-wk-old double-mutant ERα−/− x AIB1 mouse;
right, 10-wk-old ERα−/− mutant mouse.
Cancer Res; 70(10) May 15, 2010
Torres-Arzayus et al.
tasis (80–90% versus 20–30% for non-ovx mice; Fig. 5B).
Furthermore, the rate of metastasis in male mice was also increased over female untreated mice (40–50% versus 20–30%),
suggesting that estrogen signaling may play a role in the prevention of metastasis.
Figure 5. A, estrogen dependence of in vivo tumor growth potential
of a mammary tumor cell line derived from AIB1-tg mice. Groups of
wt ovx and control female mice and of wt male mice were injected with
the tumor cell line J110, derived from a mammary tumor of an AIB1-tg
mouse. The cell line SCP2 was injected as a control. Tumor growth
was monitored by palpation and caliper measurements. Final tumor size
and metastasis formation were determined by pathologic examination.
There was no significant difference in tumor growth between the groups
of mice injected with AIB1 tumor cells. B, rates of tumor and metastasis
formation in above mice. Female mice after ovariectomy and male mice
showed a significant increase in metastasis formation as compared with
non-ovx female mice.
epithelial cell line SCP2 (6) into ovx or untreated female mice
or into male mice (Fig. 5A). These results indicate that an ER
(+) tumor cell line derived from an AIB1-induced mammary
tumor can form subsequent tumors, in vivo, in the absence of
circulating gonadal estrogen.
In addition to tumor formation and growth, we also assessed metastasis formation by dissection and full pathologic
analysis, as we have previously shown that this cell line can
give rise to metastasis in various organs (5). As expected
from those studies, we observed a considerable number of
metastasis in the injected animals. Surprisingly, however,
ovx mice showed a dramatically larger proportion of metas-
Cancer Res; 70(10) May 15, 2010
We have examined the dependency of AIB1-mediated tumor formation on estrogen in a mouse model to differentiate
the AIB1 function as an ER coactivator from other reported
functions (25). We followed a dual strategy: In one set of experiments, we removed the capacity for gonadal estrogen
production by performing ovariectomies. In a second set of
experiments, we removed the capacity for ERα signaling by
creating AIB1-tg × ERα−/− double-mutant mice. A comparison of the results from these two approaches allowed us to
distinguish the effects of gonadal estrogen and ERα on
normal and premalignant mammary gland phenotype, mammary tumorigenesis, and tumor formation in other hormonedependent and hormone-independent tissues.
Consistent with previous reports (9–11, 20–22), ovariectomy and deletion of the ERα gene both resulted in severely
impaired mammary gland development. However, wt and
AIB1-tg mice revealed different degrees of impairment after
ovariectomy, as mammary glands of AIB1-tg mice showed
limited but clearly present ductal elongation and branching.
These features were absent in AIB1 × ERα−/− mice, indicating that they are strictly dependent on ERα signaling.
Unexpectedly, despite a general impairment of mammary
gland development, ovariectomy did not prevent the formation of AIB1-mediated mammary gland abnormalities, including disorganized mammary gland architecture, the
development of epithelial cell hyperplasia, and the formation of ductal carcinoma in situ, indicative of a premalignant phenotype (26, 27). This observation raised the
question whether these abnormalities were caused by
AIB1 through an ERα-dependent mechanism. We were able
to establish the requirement for ERα by creating AIB1 ×
ERα−/− mice and examining their phenotype. As these mice
did not show any evidence of the above AIB1-mediated mammary gland abnormalities, we conclude that these features
are strictly ER dependent. Thus, the premalignant mammary
phenotype still present after ovariectomy must be the result
of extragonadal estrogen production or ligand-independent
activation of ERα, or both. We expect extragonadal estrogen
levels to be somewhat increased after ovariectomy due to the
increased levels of aromatase mRNA and protein, which is
consistent with previous reports of increased aromatase levels after ovariectomy in mice or rats (18, 19). However, we
conclude that the levels of extragonadal estrogen were not
sufficient either for normal mammary gland development
or for the generation of AIB1-induced mammary tumors.
Our findings establish that the development of mammary
gland tumors and uterine tumors in the AIB1 model is
strictly dependent on the production of gonadal estrogen,
as ovx mice showed a 0% incidence of these tumors, as
compared with a 75% incidence of mammary tumors
Cancer Research
Estrogen Dependence of AIB1-Induced Tumors
Figure 6. Models of
estrogen-dependent and
estrogen-independent AIB1
actions. A, role of estrogen and
ERα in mammary gland
development and the formation of
a premalignant phenotype and
invasive tumors. B, estrogen
dependence of tumor progression
and metastasis formation.
(and 20% incidence of uterine tumors) for untreated AIB1-tg
mice. These findings argue that AIB1 exerts it oncogenicity in
the mammary gland and uterus by acting as a coactivator for
estrogen-bound ERα, leading to the expression of genes involved in proliferation and survival. Unexpectedly, we found
that ovariectomy suppressed not only ER(+) but also ER(−)
mammary gland tumors, which developed in untreated
AIB1-tg mice with an ∼15% incidence. This finding suggests
that the formation of ER(−) tumors in AIB1-tg mice requires
estrogen at some stage during tumorigenesis, but that this
requirement is lost at a later stage of tumor progression.
This hypothesis is further supported by the results from an
ER(+) mammary tumor cell line derived from an AIB1 mammary tumor that seems to represent an intermediate step on
the path to estrogen independence. As shown in Fig. 5A, injection of this tumor cell line can give rise to solid tumors in
ovx female mice or in male mice. Thus, this particular tumor
cell line is no longer dependent on gonadal estrogen for proliferation and survival, in vivo. It remains possible that this
cell line uses extragonadal sources of estrogen or possibly
produces estrogen in an autocrine manner. However, we consider this possibility unlikely, as the cell line has also ceased
Cancer Res; 70(10) May 15, 2010
Torres-Arzayus et al.
to respond to estrogen in vitro and is not affected by treatment with the ER antagonist ICI 182780 (data not shown).
Therefore, this cell line seems indeed to be estrogen independent despite continued expression of ER (also see Fig. 6A).
A totally unexpected finding from these injection experiments was that estrogen seems to have a protective effect
from the development of metastasis, as ovx mice have a significantly increased rate of metastasis following injection of
the ER(+) tumor cell line. The increase in metastasis formation after ovariectomy seems indeed to be due to reduced
estrogen levels rather than to the effect of a prior surgery,
as male mice (not treated by any surgery) showed a similar
increase in metastasis over female untreated mice. Thus, although this cell line does not seem to require ER signaling for
proliferation and survival, ER must still provide some other
signals that result in the protection from metastasis, most
likely by regulating transcription of genes involved in migration or homing (see in Fig. 6B). These findings indicate that
loss of ER may present additional, thus far unanticipated
risks in the progression of breast cancer.
In sharp contrast to mammary tumors and uterine tumors,
which were completely abolished by ovariectomy or deletion
of the ERα gene, tumors in other organs, such as lung, pituitary, and skin, still persisted following these treatments. This
indicates that AIB1 does not function as an ER coactivator in
the formation of these latter tumors and thus must have
functions other than ER coactivation that can result in oncogenesis. AIB1 has been reported to serve as a coactivator for
several other transcription factors, besides the ER, including
other hormone receptors such as progesterone receptor, thyroid hormone receptor, glucocorticoid receptor, and retinoic
acid receptor, when overexpressed in the presence of ligand
(28–30). AIB1 has also been shown to function in a hormone
receptor–independent fashion to coactivate other p300/
CBP-associated transcription factors, such as STAT and
AP-1 (31). Furthermore, we and others have implicated AIB1
in the regulation of several signaling pathways, including the
phosphoinositide 3-kinase/Akt/mammalian target of rapamycin pathway and the NF-κB pathway (1, 32). At this point,
we do not know whether any of these functions, an entirely
new function, or a combination of these is responsible for
tumorigenesis in lung, pituitary, and skin mediated by AIB1.
It is interesting to note that in the absence of estrogen or
ER, the incidence of nonmammary gland tumors increases
rather than remaining unchanged (see Fig. 3B). This
indicates that there may be a connection between the
estrogen-dependent and estrogen-independent functions of
AIB1. For example, various pathways or molecules (including
ER) might compete for AIB1, such that AIB1 might be the
limiting factor in deciding which pathway is being activated.
This result thus highlights a tight regulation of the various
functions of AIB1.
Our findings that AIB1 can mediate its oncogenicity in the
lung in an estrogen-independent manner have some immediate medical relevance, as AIB1 overexpression and amplification have recently been shown to correlate with poor
prognosis in non–small-cell lung cancer in humans (33).
However, such correlation studies cannot address cause
and effect, and therefore, it remains unclear whether in humans AIB1 overexpression is causal to lung cancer formation
and whether AIB1 functions as an ERα coactivator in the
lung. The data from our mouse model clearly show that
AIB1 overexpression by itself is sufficient to cause lung cancer
and does so in an estrogen-independent manner. Our findings
may thus have important implications in the search for drugs
and drug targets in lung cancer treatment. Taken together, our
present study shows that AIB1 can exert its oncogenicity
through at least two distinct, tissue-dependent mechanisms,
one that is ER dependent and one that is ER independent.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
We thank Dr. Pierre Chambon for providing the ER−/− mice and Dr. Stefan
Doerre for critical reading of the manuscript.
Grant Support
NIH grants P01 CA8011105 and P50C89393 and the DF/HCC Breast Cancer
Spore grant. M.I. Torres-Arzayus received research funding from the Building
Interdisciplinary Research in Women's Health (K12 HD051959, supported by
the National Institute of Mental Health, the National Institute of Allergy
and Infectious Diseases, the National Institute of Child Health and Human
Development, and the National Institutes of Health Office of the Director).
The costs of publication of this article were defrayed in part by the payment
of page charges. This article must therefore be hereby marked advertisement in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Received 11/09/2009; revised 02/12/2010; accepted 03/03/2010; published
OnlineFirst 05/04/2010.
Torres-Arzayus MI, Font de Mora J, Yuan J, et al. High tumor incidence and activation of the PI3K/AKT pathway in transgenic mice
define AIB1 as an oncogene. Cancer Cell 2004;6:263–74.
Anzick SL, Kononen J, Walker RL, et al. AIB1, a steroid receptor
coactivator amplified in breast and ovarian cancer. Science 1997;
Li H, Gomes PJ, Chen JD. RAC3, a steroid/nuclear receptorassociated coactivator that is related to SRC-1 and TIF2. Proc Natl
Acad Sci U S A 1997;94:8479–84.
Dupont S, Krust A, Gansmuller A, Dierich A, Chambon P, Mark M.
Effect of single and compound knockouts of estrogen receptors α
Cancer Res; 70(10) May 15, 2010
(ERα) and β (ERβ) on mouse reproductive phenotypes. Development
Torres-Arzayus MI, Yuan J, DellaGatta JL, Lane H, Kung AL, Brown
M. Targeting the AIB1 oncogene through mammalian target of
rapamycin inhibition in the mammary gland. Cancer Res 2006;66:
Desprez PY, Hara E, Bissell MJ, Campisi J. Suppression of mammary
epithelial cell differentiation by the helix-loop-helix protein Id-1. Mol
Cell Biol 1995;15:3398–404.
Kaplan E, Meier P. Non-parametric estimation for incomplete observation. J Am Stat Assoc 1958;53:457–71.
Cancer Research
Estrogen Dependence of AIB1-Induced Tumors
Peto R, Peto J. Asymptotically efficient rank invariant test procedures (with discussion). J R Stat Soc A 1972;135:185–207.
Westerlind KC, Gibson KJ, Malone P, Evans GL, Turner RT. Differential effects of estrogen metabolites on bone and reproductive tissues
of ovariectomized rats. J Bone Miner Res 1998;13:1023–31.
Daniel CW, Silberstein GB, Strickland P. Direct action of 17β-estradiol
on mouse mammary ducts analyzed by sustained release implants
and steroid autoradiography. Cancer Res 1987;47:6052–7.
Silberstein GB, Van Horn K, Shyamala G, Daniel CW. Essential role
of endogenous estrogen in directly stimulating mammary growth
demonstrated by implants containing pure antiestrogens. Endocrinology 1994;134:84–90.
Buchanan DL, Kurita T, Taylor JA, Lubahn DB, Cunha GR, Cooke PS.
Role of stromal and epithelial estrogen receptors in vaginal epithelial
proliferation, stratification, and cornification. Endocrinology 1998;
Lotinun S, Westerlind KC, Kennedy AM, Turner RT. Comparative effects of long-term continuous release of 16 α-hydroxyestrone and
17β-estradiol on bone, uterus, and serum cholesterol in ovariectomized adult rats. Bone 2003;33:124–31.
Kurita T, Lee KJ, Cooke PS, Taylor JA, Lubahn DB, Cunha GR.
Paracrine regulation of epithelial progesterone receptor by estradiol in the mouse female reproductive tract. Biol Reprod 2000;62:
Grodin JM, Siiteri PK, MacDonald PC. Source of estrogen production in postmenopausal women. J Clin Endocrinol Metab 1973;36:
Simpson ER, Clyne C, Rubin G, et al. Aromatase—a brief overview.
Annu Rev Physiol 2002;64:93–127.
Bulun SE, Chen D, Lu M, et al. Aromatase excess in cancers of
breast, endometrium and ovary. J Steroid Biochem Mol Biol 2007;
Zhao H, Tian Z, Cheng L, Chen B. Electroacupuncture enhances
extragonadal aromatization in ovariectomized rats. Reprod Biol
Endocrinol 2004;2:18.
Thordarson G, Lee AV, McCarty M, et al. Growth and characterization of N-methyl-N-nitrosourea-induced mammary tumors in intact
and ovariectomized rats. Carcinogenesis 2001;22:2039–47.
Bocchinfuso WP, Korach KS. Mammary gland development and
tumorigenesis in estrogen receptor knockout mice. J Mammary
Gland Biol Neoplasia 1997;2:323–34.
21. Cunha GR, Young P, Hom YK, Cooke PS, Taylor JA, Lubahn DB.
Elucidation of a role for stromal steroid hormone receptors in
mammary gland growth and development using tissue recombinants. J Mammary Gland Biol Neoplasia 1997;2:393–402.
22. Mallepell S, Krust A, Chambon P, Brisken C. Paracrine signaling
through the epithelial estrogen receptor α is required for proliferation
and morphogenesis in the mammary gland. Proc Natl Acad Sci U S A
23. Mueller SO, Clark JA, Myers PH, Korach KS. Mammary gland development in adult mice requires epithelial and stromal estrogen receptor α. Endocrinology 2002;143:2357–65.
24. Bocchinfuso WP, Lindzey JK, Hewitt SC, et al. Induction of mammary
gland development in estrogen receptor-α knockout mice. Endocrinology 2000;141:2982–94.
25. Liao L, Kuang SQ, Yuan Y, Gonzalez SM, O'Malley BW, Xu J.
Molecular structure and biological function of the cancer-amplified
nuclear receptor coactivator SRC-3/AIB1. J Steroid Biochem Mol
Biol 2002;83:3–14.
26. Cardiff RD, Moghanaki D, Jensen RA. Genetically engineered mouse
models of mammary intraepithelial neoplasia. J Mammary Gland Biol
Neoplasia 2000;5:421–37.
27. Cardiff RD, Anver MR, Gusterson BA, et al. The mammary pathology of genetically engineered mice: the consensus report and
recommendations from the Annapolis meeting. Oncogene 2000;
28. Glass CK, Rosenfeld MG. The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev 2000;14:121–41.
29. McKenna NJ, Lanz RB, O'Malley BW. Nuclear receptor coregulators:
cellular and molecular biology. Endocr Rev 1999;20:321–44.
30. Brown K, Chen Y, Underhill TM, Mymryk JS, Torchia J. The coactivator p/CIP/SRC-3 facilitates retinoic acid receptor signaling via
recruitment of GCN5. J Biol Chem 2003;278:39402–12.
31. Torchia J, Rose DW, Inostroza J, et al. The transcriptional coactivator p/CIP binds CBP and mediates nuclear-receptor function.
Nature 1997;387:677–84.
32. Wu RC, Qin J, Hashimoto Y, et al. Regulation of SRC-3 (pCIP/ACTR/
AIB-1/RAC-3/TRAM-1) coactivator activity by IκB kinase. Mol Cell
Biol 2002;22:3549–61.
33. He LR ZH, Li BK, Zhang LJ, et al. Overexpression of AIB1 negatively
affects survival of surgically resected non-small-cell lung cancer
patients. Ann Oncol. Epub 2010.
Cancer Res; 70(10) May 15, 2010

Similar documents