Sustained induction of epithelial to mesenchymal
transition activates DNA methylation of genes
silenced in basal-like breast cancers
Nancy Dumont, Matthew B. Wilson, Yongping G. Crawford*, Paul A. Reynolds†, Mahvash Sigaroudinia,
and Thea D. Tlsty‡
Department of Pathology and Comprehensive Cancer Center, University of California, San Francisco, CA 94143-0511
Communicated by Joan S. Brugge, Harvard Medical School, Boston, MA, July 31, 2008 (received for review March 4, 2008)
The active acquisition of epigenetic changes is a poorly understood
but important process in development, differentiation, and dis-
ease. Our work has shown that repression of the p16/pRb pathway
in human epithelial cells, a condition common to stem cells and
many tumor cells, induces dynamic epigenetic remodeling result-
ing in the targeted methylation of a selected group of CpG islands.
We hypothesized that cells in this epigenetically plastic state could
be programmed by the microenvironment to acquire epigenetic
changes associated with tumorigenesis. Here, we describe an in
vitro model system where epigenetically plastic cells were placed
in an environment that induced epithelial to mesenchymal transi-
tion (EMT) and led to a program of acquired de novo DNA meth-
ylation at targeted sites. In this model, we found that repression of
E-cadherin transcription preceded the subsequent acquisition of
methylated CpG sites. Furthermore, the induction of EMT was
accompanied by de novo methylation of several other gene pro-
moters, including those of the estrogen receptor and Twist. These
data demonstrate that signals from the microenvironment can
induce phenotypic and gene expression changes associated with
targeted de novo epigenetic alterations important in tumor pro-
gression, and that these alterations occur through a deterministic,
rather than stochastic, mechanism. Given the dynamic epigenetic
observed in human tumors may reflect the history of environmen-
tal exposures during the genesis of a tumor.
epigenetic remodeling ? human mammary epithelial cells ?
microenvironment ? ras
controlled by epigenetic modifications of proteins and DNA
sequences. We recently reported that the repression of p16INK4A
in primary human mammary epithelial cells (HMEC) activates
an E2F-mediated increase in proteins that remodel chromatin
and causes targeted de novo DNA methylation at a non-random
collection of loci (1). These studies show that cells can acquire
epigenetic plasticity by altering the p16/pRb pathway, and that
this program of acquired de novo methylation has a deterministic
(predictable) rather than stochastic (random) pattern. Further-
more, the coordinated set of de novo DNA methylation events
are preceded by, and dependent upon, the repression of gene
expression. Thus, during cancer progression, one may envision
that tumor cells can acquire epigenetic plasticity through repres-
sion of the p16/pRb pathway via mutations, deletions, or methyl-
ation (2), which then provides the potential for programming
epigenetic events. These observations are reminiscent of studies
that show the acquisition of promoter hypermethylation upon
modulation of estrogen or retinoic acid signaling (3, 4). In these
cell population-based studies it is unclear whether the non-
random hypermethylation events observed are due to induction
or selection. To explore this question further, we chose a
clinically relevant malignant phenotype and determined if re-
he heritable regulation of gene expression changes that are
pression of gene expression induced subsequent DNA methyl-
ation events or whether these occurred by selection.
chymal transition (EMT), play a critical role in tumor progres-
sion (5), have been implicated in tumor recurrence (6), and are
often associated with a poor prognosis in women with breast
cancer (7). Consistent with this, there is now evidence demon-
strating a link between EMT, basal-like tumors, the stem-cell
phenotype, and the acquisition of tumorigenic and metastatic
potential (8, 9). EMT is characterized by several molecular
changes that include the loss of epithelial markers such as
E-cadherin, and the induction of mesenchymal markers such as
N-cadherin, fibronectin, and Snail (5). Though alterations in
E-cadherin expression can occur through multiple mechanisms,
including loss of heterozygosity and mutational inactivation,
E-cadherin is frequently silenced through aberrant DNA hyper-
methylation of its promoter (10). Interestingly, when E-cadherin
is silenced through promoter DNA hypermethylation, mammary
cell lines often exhibit a mesenchymal morphology through the
coordinated induction of a set of genes involved in EMT (11). In
contrast, when E-cadherin is inactivated by mutation, the cells
not induced (11). This suggests that a program of molecular
alterations leading to EMT, invasion, and metastasis can be
EMT has been shown to be induced in murine cells by
oncogenic ras in cooperation with factors in serum (12). There
is also evidence that exposing cells to serum induces a gene
expression pattern that resembles that of a wounding response.
This wound-response signature is strongly predictive of future
invasive and metastatic behavior, both of which require EMT
(13). To determine whether HMEC with repressed p16INK4A
(vHMEC) could be programmed by the microenvironment to
acquire epigenetic changes associated with tumorigenesis, im-
mortalized vHMEC-expressing oncogenic ras (vHMEC-ras)
were exposed to serum. When cultured in serum-rich media,
vHMEC-ras cells underwent phenotypic changes indicative of
EMT and became more motile. This morphological transition
was accompanied by a program of directed de novo DNA
methylation of genes such as E-cadherin. Thus, signals from the
Author contributions: N.D. and T.D.T. designed research; N.D., M.B.W., Y.G.C., P.A.R., and
M.S. performed research; N.D. analyzed data; and N.D. and T.D.T. wrote the paper.
The authors declare no conflict of interest.
*Present address: Genentech, Inc., 1 DNA Way, South San Francisco, CA 94080.
†Present address: University of St. Andrews, Bute Medical School, St. Andrews KY16 9TS,
‡To whom correspondence should be addressed at: Department of Pathology, University
of California, San Francisco, 513 Parnassus Avenue, HSW 511, Box 0511, San Francisco,
CA 94143-0511. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
© 2008 by The National Academy of Sciences of the USA
September 30, 2008 ?
vol. 105 ?
no. 39 ?
microenvironment can induce phenotypic changes that modify
the epigenome in an active and deterministic manner. This
valuable model system will allow further study of phenotypic and
epigenetic modulation during malignant transformation, and
provide targets for therapeutic epigenetic reprogramming.
Immortalized HMEC Can Undergo EMT in a Serum-Rich Environment.
Because oncogenic ras has been shown to cooperate with factors
in serum to induce EMT and promote tumorigenesis in murine
ras0.5) or 10% (vHMEC-ras10) serum. The cells grew equally
well in both serum concentrations (Fig. 1A). In addition, we
found that both cell populations were capable of continued
proliferation after removal of serum [supporting information
(SI) Fig. S1A, vHMEC-ras0.5 3 0 and vHMEC-ras10 3 0].
However, unlike the cells grown in 0.5% serum, those grown in
10% serum underwent a striking change in morphology that was
associated with loss of E-cadherin expression, reorganization of
the actin cytoskeleton, and upregulation of the mesenchymal
markers N-cadherin and fibronectin (Fig. 1 B and C), demon-
strating that the cells were undergoing EMT. Because EMT is
associated with increased motility, we examined whether the
cells that had undergone EMT were more motile than the
vHMEC-vector, -ras, or -ras0.5 cells that maintained their
epithelial morphology, and indeed, they were (Fig. 1D), indi-
cating that the acquisition of this mesenchymal phenotype was
biologically and functionally relevant. Moreover, the mesen-
chymal phenotype did not require constitutive extracellular
serum stimulation, because it was maintained upon serum
withdrawal (Fig. 1B, 10 3 0). The epithelial morphology of
ras-expressing vHMEC grown in 0.5% serum and the mesen-
chymal morphology of the same cells grown in 10% serum was
manifested both on plastic (2D) and in matrigel (3D). When
cultured in matrigel, the epithelial-appearing vHMEC-ras0.5
cells organized into mammosphere-like structures, whereas the
mesenchymal-appearing vHMEC-ras10 cells retained their spin-
dle morphology (Fig. S1B).
TGF? Can Phenocopy the Morphological Effects Induced by a Serum-
Rich Environment. To gain insight into what factors in the serum-
rich environment were responsible for the EMT observed in
vHMEC-ras10 cells, we examined what signaling pathways were
differentially activated in vHMEC-ras cells that exhibit a mes-
enchymal morphology relative to those that exhibit an epithelial
morphology. Because TGF? plays an important role in mediat-
ing EMT (14), we first examined the phosphorylation status of
Smad2, one of the transcriptional mediators of TGF?, by
immunoblot analysis. As shown in Fig. 2A, basal Smad2 phos-
phorylation was increased in both early- (10E) and late- (10L)
passage vHMEC-ras10 cells that exhibit a mesenchymal mor-
phology, but was not detected in vHMEC-ras0.5 cells that exhibit
an epithelial morphology. In contrast, phosphorylation of MAP
kinase was elevated in all vHMEC-ras cells (independent of the
serum concentration in the growth media), consistent with the
activation of the ras signaling pathway in each of these cell
populations. These data indicate that TGF? signaling may be
required for the phenotypic alterations induced by 10% serum.
Consistent with this, in our analysis of several markers of EMT,
including E47 and Slug, which remained unchanged (Fig. S2A),
we observed an upregulation of smad-interacting protein 1
(SIP1), a protein that has been implicated in TGF?-mediated
EMT (15), in the vHMEC-ras10 cells that had undergone EMT
to induce Snail (16), a transcription factor that is a central
mediator of EMT (17). Snail was also upregulated in the
vHMEC-ras10 cells that had undergone EMT (Fig. 2C). In
addition, the negative regulator of Snail, MTA3, a protein that
forms part of chromatin complexes (18), was downregulated in
vHMEC-ras10 cells (Fig. 2C). These data demonstrate that
upregulated in vHMEC-ras cells grown in a serum-rich environ-
ment, and suggest that TGF? may be involved in mediating
If TGF? is one of the factors present in a serum-rich envi-
ronment that cooperates with oncogenic ras to induce EMT, we
reasoned that TGF? could be used in lieu of 10% serum to
induce EMT in vHMEC-ras0.5 cells that exhibit an epithelial
morphology. To test this, we treated vHMEC-ras0.5 cells with
TGF? and assessed the expression of molecular markers asso-
ciated with EMT by immunofluorescence and immunoblot
analysis. Within 48 h of treatment with TGF?, the cells showed
signs of a morphological change, which were clearly manifested
by 72 h. This morphological change was associated with a
diminution in pancytokeratin expression, reorganization of the
actin cytoskeleton, and disruption of both cell–cell and cell-
matrix contacts, as evidenced by the loss of E-cadherin and
?1-integrin expression, respectively (Fig. 2D). Coincident with
the loss of epithelial marker expression was an upregulation in
the expression of the mesenchymal markers N-cadherin and
fibronectin (Fig. 2E). In addition, vHMEC-ras0.5 cells stimu-
lated with TGF? were more motile than their unstimulated
counterparts in transwell migration assays, consistent with the
expected increase in motility associated with EMT (Fig. 2F).
Days in culture
Motility (arbitrary units x103)
vector in the absence or presence of 0.5% or 10% serum. Arrows indicate time at which serum was added. (B) Phase contrast (10?) and immunofluorescence
photomicrographs (63?) of vHMEC-ras cells grown in their original concentration of serum (0.5% and 10%, first and second columns, respectively) or after they
of cells toward MEGM ? 10% FBS as a chemoattractant for 48 h.
Immortalized HMEC expressing oncogenic ras undergo EMT in a serum-rich environment. (A) Growth curves of vHMEC expressing Ha-rasV12 or control
www.pnas.org?cgi?doi?10.1073?pnas.0807146105Dumont et al.
These data suggest that TGF? is one of the factors present in
serum that cooperates with ras to induce EMT. However, unlike
the serum-induced EMT, the TGF?-induced EMT was rever-
sible upon withdrawal of TGF? (Fig. S2B), as previously
observed in other cell types (19).
Serum-Induced EMT Is Accompanied by Repression of E-cadherin
Expression and Subsequent DNA Hypermethylation at the E-cadherin
Promoter. Because serum-induced EMT was irreversible and
heritable (it was maintained upon the withdrawal of serum and
transmitted to daughter cells), we examined whether epigenetic
alterations were involved in permanently repressing E-cadherin
expression in these cells. Epigenetic modifications in the pro-
moter region of E-cadherin were assessed using methylation-
specific PCR. Consistent with the heritable and morphological
appearance of the cells, DNA methylation of the E-cadherin
promoter was observed in the mesenchymal-appearing vHMEC-
ras10 cells, but not in the epithelial-appearing vHMEC-ras0.5
cells (Fig. 3A, lanes 1–4). This methylation pattern was also
maintained after the cells were switched to no (0.5 3 0) or low
(10 3 0.5) serum growth conditions (Fig. 3A, lanes 5–8),
consistent with the maintenance of their morphology under
those conditions. In contrast, and as expected, we did not
observe methylation of the E-cadherin promoter after TGF?-
induced EMT, which is reversible (Fig. S2C). These data suggest
that TGF? is sufficient to induce the phenotypic alterations
observed in a serum-rich environment, but that additional
factors (or time) are required for the epigenetic modifications
that render EMT heritable to occur.
To rule out the unlikely possibility that the mesenchymal
features in these cells were due to the selection and expansion
of a rare contaminating population of fibroblasts, we analyzed
the promoter gene region for the protein 14–3-3?, which is a
mammary epithelial-specific marker that is methylated in fibro-
blasts but unmethylated in epithelial cells (20). This locus
remained unmethylated in all our cells, including those that
origin (Fig. 3A). vHMEC-ras10 cells that exhibit a mesenchymal
phenotype can also be distinguished from fibroblasts on the basis
of their E-cadherin methylation as the E-cadherin gene promoter
is repressed and unmethylated in human mammary fibroblasts
(Fig. S2C, lane 2), but repressed and methylated in mesenchymal-
appearing vHMEC-ras10 cells (Fig. 3). These data demonstrate
that the acquisition of mesenchymal features by vHMEC-ras10
cells is the consequence of an active molecular process that
occurs in epithelial cells.
To gain insight into the kinetics of E-cadherin silencing and
methylation during serum-induced EMT, we examined E-
cadherin mRNA expression and methylation at multiple
passages after exposure to serum. This analysis indicated that
E-cadherin gene expression was dramatically reduced after eight
passages in serum (?50-fold less than ras0.5), and was even
further repressed before methylation, which was detected at
3B). This demonstrates that the repression of E-cadherin and
emergence of mesenchymal characteristics precede methylation
of the E-cadherin promoter.
To confirm that DNA methylation events were causal (and not
simply correlative), for the long-term repression of E-cadherin
expression, late-passage vHMEC-ras10 cells exhibiting hyper-
methylated E-cadherin promoter sequences were exposed to the
DNA methyltransferase (DNMT) inhibitor 5-aza-2?deoxycyti-
dine (AZA). As shown in Fig. S3, inhibition of DNA methylation
restored E-cadherin expression, but only modestly. Because
Snail expression was upregulated in vHMEC-ras10 cells that
exhibit a mesenchymal morphology, and Snail can mediate
E-cadherin repression by the recruitment of histone deacetylases
(HDAC) (21), we treated the cells with the HDAC inhibitor
trichostatin A (TSA) to determine whether deacetylation was
also involved in E-cadherin repression. Inhibition of HDACs
resulted in an increase in E-cadherin expression, which was even
more robust when the cells were treated with both inhibitors.
0.5 10E 10L
0.5 10E 10L
Motility (arbitrary units)
Immunoblot analysis of phospho-Smad2 and phospho-MAP kinase (A) or Snail and MTA3 (C) on cell lysates prepared from vHMEC-ras0.5 (0.5) and vHMEC-ras10
early- (10E) and late- (10L) passage cells. (B) Quantitative real-time PCR (qPCR) analysis of SIP1 expression. (D) Immunofluorescence analysis of vHMEC-ras0.5
untreated (?) or treated (?) with 2 ng/ml TGF? for 72 h and immunostained as indicated. (E) Immunoblot analysis of fibronectin (Fn) and N-cadherin (N-cad)
on cell lysates prepared from vHMEC-ras0.5 either untreated (?) or treated (?) as in (D). Cell lysates prepared from fibroblasts were used as a positive control
(C). (F) Transwell motility assay depicting vHMEC-ras0.5 cell migration toward media not supplemented (?) or supplemented (?) with 10 ng/ml TGF? as a
chemoattractant for 48 h.
TGF? signaling is activated in vHMEC-ras cells with a mesenchymal morphology and can induce EMT in vHMEC-ras cells with an epithelial morphology.
Dumont et al.
September 30, 2008 ?
vol. 105 ?
no. 39 ?
This suggests that deacetylation and DNA hypermethylation at
the E-cadherin promoter both contribute to the mechanistic
basis of long-term silencing of E-cadherin expression.
To determine whether the mechanism by which E-cadherin is
initially repressed before permanent silencing through DNA
methylation also involved deacetylation, early-passage vHMEC-
ras10 cells were exposed TSA and E-cadherin expression was
examined. The inhibition of HDACs increased expression of
E-cadherin (Fig. S3). These data suggest that deacetylation is
implicated in repressing E-cadherin expression before long-term
silencing by DNA methylation.
Methylation of the E-cadherin Promoter Occurs de Novo. To deter-
mine if methylation of the E-cadherin promoter was a de novo
event in vHMEC-ras10 cells, or the result of the selection of a
pre-existing population of epithelial cells with a methylated
of vHMEC-ras0.5 cells that exhibit an epithelial morphology and
exposed them to media containing 10% serum. Early-passage
clones (C1-E and C2-E) retained their epithelial morphology
after short-term exposure to 10% serum (Fig. 4A) and expressed
E-cadherin (Fig. 4B). Upon continued exposure to media con-
taining 10% serum, both clones gradually acquired a mesen-
chymal morphology (C1-L and C2-L), E-cadherin expression
was lost, the mesenchymal markers N-cadherin and fibronectin
were upregulated (Fig. 4B), and the cells became more motile
(Fig. 4C). Subsequent to the induction of EMT, their E-cadherin
promoter became hypermethylated (Fig. 4D). These data dem-
onstrate that the changes in cellular morphology and DNA
methylation observed when vHMEC-ras cells are exposed to
10% serum occur de novo.
Methylation of the E-cadherin Promoter Is Part of a Nonrandom
Program of Targeted Epigenetic Changes. The de novo DNA
hypermethylation observed at the E-cadherin locus could be due
to a generalized increase in DNMT function that results in a
generalized hypermethylation of CpG islands, or a program of
more directed events. Exposing vHMEC-ras cells to 10% serum
led to an increase in DNMT3a expression (Fig. S4); however,
only selected CpG islands subsequently acquired DNA hyper-
methylation. Previous studies have shown that the E-cadherin
and estrogen receptor (ER) promoter sequences are often
hypermethylated in the same samples (22, 23). Additionally, a
number of genes have been reported to be more frequently
hypermethylated in invasive breast carcinomas (24, 25) and in
basal-like breast cancers that have a poor prognosis (26). Be-
cause EMT is associated with invasive behavior, and has been
correlated with the basal-like phenotype (8), we investigated
whether genes that have been reported to be methylated in
ated promoter sequences in vHMEC-ras10 cells that exhibit a
mesenchymal morphology. Several genes, including those that
encode ER and Twist, which have been reported to be methyl-
ated in basal-like and/or invasive breast cancers (25, 26), were
hypermethylated in vHMEC-ras10 cells that exhibit a mesen-
chymal morphology, but not in vHMEC-ras or vHMEC-ras0.5
cells that exhibit an epithelial morphology (Table 1 and Fig.
S4B). The promoter region for CST6, another locus that is
hypermethylated in basal-like tumors, was also hypermethylated
in cells that had acquired a mesenchymal phenotype, but its
methylation occurred before methylation of the E-cadherin
promoter. In contrast, other loci, known to be frequently hyper-
(MSP) analysis of E-cadherin and 14–3-3? on bisulfite-treated DNA isolated
from vHMEC-ras cells grown in 0.5, 10, 0.5 3 0, or 10 3 0.5% serum using
DNA. Unmethylated (?) and methylated (?) control templates. (B) qPCR
analysis of E-cadherin expression in vHMEC-ras10 cells at the indicated pas-
sages (Top); MSP analysis of the same cells (Bottom).
EMT is accompanied by repression of E-cadherin expression and
Motility (arbitrary units x103)
C2-L) passage clones. (B) Immunoblot analysis of fibronectin (Fn), N-cadherin (N-cad), and E-cadherin (E-cad) on cell lysates prepared from each clone. (C)
Transwell motility assay depicting migration of clones toward MEGM ? 10% FBS as a chemoattractant for 48 h. (D) MSP analysis of E-cadherin and 14–3-3?.
EMT and epigenetic modifications at the E-cadherin locus occur de novo. (A) Photomicrographs (10?) of two early- (C1-E and C2-E) and late- (C1-L and
www.pnas.org?cgi?doi?10.1073?pnas.0807146105Dumont et al.
methylated in breast tumors, such as BRCA1, GATA3, TIMP3,
and DKK3 did not exhibit DNA promoter hypermethylation in
these cells, thus confirming a nonrandom rather than general-
ized process (Table 1 and Fig. S4B). These data suggest that
there is a sequential and progressive increase in promoter gene
methylation along the progression to malignancy that reflects
the effect of signals from the surrounding environment.
The accumulation of aberrant DNA methylation events that
arise during tumorigenesis could occur through random epi-
genetic changes within the tumor cell population that are then
selected for expansion because of a selective advantage or,
alternatively, they could occur through more deterministic
events and be programmed by specific repression of targeted
genes. The data in this study provide a mechanism by which
premalignant cells can acquire de novo DNA methylation at
biologically relevant sites early in the carcinogenic process in a
deterministic manner. Individual clones with an epithelial
phenotype were reprogrammed by exogenous signals to acquire
a mesenchymal phenotype associated with increased motility
and predictable de novo DNA methylation events. Conceptually,
our data suggest that the profile of methylation events in tumors
is the sum of environmental exposures during their inception.
Different environmental conditions can be expected to elicit
different programs of epigenetic changes that create a memory
of processes that drive carcinogenesis in each particular tumor.
Thus, these cells represent a valuable model in which to study
phenotypic and epigenetic plasticity, and the acquisition of
directed de novo DNA methylation events.
The association between E-cadherin and ER promoter methyl-
ation we report in our in vitro system has been observed previously
in human breast tumors and correlates with clinical parameters.
Examination of several methylated gene loci for associations with
clinicopathological features revealed that methylation at the E-
at diagnosis (27). Parrella et al. (23) examined the frequency of
methylation at these loci in benign and malignant tissue, and
reported an association between E-cadherin and ER methylation
that increased with progression. Likewise, Nass et al. (22) reported
that E-cadherin and ER promoter methylation as individual events
were detectable in early lesions such as DCIS, were more frequent
in invasive lesions, and that the coincident methylation of both
promoters occurred in both types of lesions. Our in vitro studies
provide a rationale for the association between E-cadherin and ER
promoter methylation. Under specific conditions, both genes are
for epigenetic remodeling via an initial repression of gene expres-
sion and the subsequent recruitment of DNA methyltransferases.
The clinical ramifications of these methylation events are
critical. Many human breast cancers and breast cancer cell lines
display features of EMT (11, 28), and recent studies have shown
that this phenotype is prominently featured in human basal-like
breast tumors (8). Consistent with this, studies have reported
that loss of E-cadherin expression is positively correlated with
advanced histological grade, metastasis, and decreased survival
(29, 30), characteristics that are typically associated with the
basal-like subtype. Thus, methylation of the E-cadherin pro-
moter leading to loss of E-cadherin expression and increased
motility in vHMEC-ras10 cells may be an important event in the
transformation of these cells. Coleman and coworkers (26)
reported a methylation signature that exists in basal-like breast
tumors. This panel of six genes is frequently hypermethylated in
basal-like tumors and distinguishes them from other breast
tumor subtypes. Of these six genes, the three that we examined—
E-cadherin, ER, and CST6—were hypermethylated in the cells
that underwent EMT.
Our data also suggest that there are consequences to the
silencing of gene loci by DNA hypermethylation when compared
with transcriptional repression via deacetylase activity. In this
report we show that the repression of E-cadherin in early-
passage vHMEC-ras10 cells (that have undergone EMT, but do
not exhibit methylation of the E-cadherin promoter) is depen-
dent on deacetylation, as inhibition of HDACs by TSA resulted
in an increase in E-cadherin expression in these cells. In contrast,
late-passage vHMEC-ras10 cells (that have undergone EMT as
well as E-cadherin promoter methylation) appear to repress
E-cadherin through the dual action of histone deacetylation and
DNA promoter hypermethylation. The consequences of differ-
entially repressing E-cadherin are twofold. First, early-passage
(P8) vHMEC-ras10 cells exhibit a higher level of E-cadherin
expression than late-passage vHMEC-ras10 cells, and though
they are more motile than vHMEC-ras cells that exhibit an
epithelial morphology, they are less motile than vHMEC-ras10
cells that exhibit both a mesenchymal morphology and E-
cadherin promoter methylation (Fig. 3 and data not shown,
of DNA methylation events ensures that subsequent generations
with sustained induction of EMT even in the absence of inducing
Finally, this in vitro model system provides unique insights into
the timing and induction of methylation events that may occur
during tumor progression. In this system, there is an ordering to
vHMEC to 10% serum or TGF? in the absence of ras overex-
pression evokes a pronounced cell cycle arrest (unpublished
cells used in this study. This suggests that EMT is a process that
occurs later in the transformation process and that an increase
in genomic instability as well as additional molecular alterations
are required before invasive phenotypes such as EMT can occur.
This is of particular significance because though EMT has been
observed in a few rodent cell lines in culture, a survey of 14
human epithelial cell lines and two primary human cells revealed
that none of the human cells could undergo EMT in response to
48 h of TGF? treatment (31). Thus, our data provide previously
undescribed evidence that extracellular cues (leading to immor-
talization) can cooperate with TGF? signaling and oncogenic
stress during transformation to induce EMT in human cells.
These data suggest that genetic and epigenetic events, provoked
by ras overexpression, set the stage for the program of gene
repression and DNA hypermethylation that can subsequently
occur. Translating these observations to a biological setting, we
postulate that damaging agents, such as ionizing radiation, which
activate TGF?, would typically induce a proliferative arrest in
the bulk of breast epithelial tissue, but activate targeted epi-
genetic reprogramming in foci of cells that had already acquired
preparatory molecular events such as ras overexpression and
Table 1. MSP analysis of genes hypermethylated in breast cancer
U, unmethylated; M, methylated.
Dumont et al.
September 30, 2008 ?
vol. 105 ?
no. 39 ?
immortalization. Understanding these processes may provide
avenues to effective intervention and prevention.
Materials and Methods
pLXSP3 vector (N.D., et al., unpublished data) were grown in MEGM (Lonza)
supplemented with either 0.5% or 10% FBS and 2 ?g/ml puromycin, as
previously described (1). vHMEC-ras0.5 clones were isolated using standard
were conducted between 5 and 15 passages following exposure to serum;
experiments with late passage clones were conducted 40 or more passages
following exposure to serum. For 3D cultures, 5 ? 103cells were resuspended
in media containing 2% Matrigel (BD Biosciences) and seeded into eight-well
glass chamber slides on top of a thin layer of Matrigel.
Immunofluorescence Analysis. Cells were fixed with 100% methanol for 7 min
at ?20°C (for E-cadherin, ?1-integrin, and cytokeratin staining) or with 4%
paraformaldehyde/PBS for 20 min at room temperature (RT) (for F-actin
staining). Paraformaldehyde-fixed cells were permeabilized with 0.1% Triton
X-100/PBS for 10 min at RT and washed with PBS. Nonspecific binding sites
were blocked and the cells were then incubated with primary antibodies
RT, or incubated with Alexa Fluor 488-Phalloidin (Molecular Probes, A12379)
for 1 h at RT and appropriately washed. Images were captured using a Nikon
epifluorescence inverted microscope and a cooled CCD digital camera
Motility Assay. Motility assays were conducted using Chemicon’s QCM Fluoro-
Methylation-Specific PCR. Genomic DNA was isolated from cells using the
Wizard Genomic DNA Isolation Kit (Promega). Approximately 750 ng of DNA
was bisulfite treated with the EZ DNA Methylation-Gold Kit according to the
performed on bisulfite-modified DNA using previously described primer pairs
and PCR cycle conditions (see SI Text for additional details on this and other
TGF? and Dr. Stephen Baylin for GATA3 MSP primers. This research was
(to N.D.), the National Cancer Institute Institutional Training Grant T32
CA009043 (to M.B.W.), the Department of Defense Breast Cancer Research
Program Concept Award BC023982 (to T.D.T.), and National Institutes of
Health National Cancer Institute Grants CA097214–01A1 and CA122024–01
1. Reynolds PA, et al. (2006) Tumor suppressor p16INK4A regulates polycomb-mediated
DNA hypermethylation in human mammary epithelial cells. J Biol Chem 281:24790–
2. Sherr CJ, McCormick F (2002) The RB and p53 pathways in cancer. Cancer Cell 2:103–112.
3. Leu YW, et al. (2004) Loss of estrogen receptor signaling triggers epigenetic silencing
of downstream targets in breast cancer. Cancer Res 64:8184–8192.
4. Ren M, et al. (2005) Impaired retinoic acid (RA) signal leads to RARbeta2 epigenetic
silencing and RA resistance. Mol Cell Biol 25:10591–10603.
5. Lee JM, Dedhar S, Kalluri R, Thompson EW (2006) The epithelial-mesenchymal transi-
tion: New insights in signaling, development, and disease. J Cell Biol 172:973–981.
6. Moody SE, et al. (2005) The transcriptional repressor Snail promotes mammary tumor
recurrence. Cancer Cell 8:197–209.
in breast cancer. Anticancer Res 22:3415–3419.
8. Sarrio D, et al. (2008) Epithelial-mesenchymal transition in breast cancer relates to the
basal-like phenotype. Cancer Res 68:989–997.
9. Mani SA, et al. (2008) The epithelial-mesenchymal transition generates cells with
properties of stem cells. Cell 133:704–715.
10. Droufakou S, et al. (2001) Multiple ways of silencing E-cadherin gene expression in
lobular carcinoma of the breast. Int J Cancer 92:404–408.
11. Lombaerts M, et al. (2006) E-cadherin transcriptional downregulation by promoter
methylation but not mutation is related to epithelial-to-mesenchymal transition in
breast cancer cell lines. Br J Cancer 94:661–671.
12. Oft M, et al. (1996) TGF-beta1 and Ha-Ras collaborate in modulating the phenotypic
plasticity and invasiveness of epithelial tumor cells. Genes Dev 10:2462–2477.
13. Chang HY, et al. (2004) Gene expression signature of fibroblast serum response
predicts human cancer progression: Similarities between tumors and wounds. PLoS
14. Thiery JP, Sleeman JP (2006) Complex networks orchestrate epithelial-mesenchymal
transitions. Nat Rev Mol Cell Biol 7:131–142.
15. Shirakihara T, Saitoh M, Miyazono K (2007) Differential regulation of epithelial and
mesenchymal markers by deltaEF1 proteins in epithelial mesenchymal transition in-
duced by TGF-beta. Mol Biol Cell 18:3533–3544.
16. Peinado H, Quintanilla M, Cano A (2003) Transforming growth factor beta-1 induces
snail transcription factor in epithelial cell lines: Mechanisms for epithelial mesenchy-
mal transitions. J Biol Chem 278:21113–21123.
17. Davidson NE, Sukumar S (2005) Of Snail, mice, and women. Cancer Cell 8:173–174.
18. Dhasarathy A, Kajita M, Wade PA (2007) The transcription factor snail mediates
epithelial to mesenchymal transitions by repression of estrogen receptor-alpha. Mol
19. Oft M, Heider KH, Beug H (1998) TGFbeta signaling is necessary for carcinoma cell
invasiveness and metastasis. Curr Biol 8:1243–1252.
20. Moreira JM, Ohlsson G, Rank FE, Celis JE (2005) Down-regulation of the tumor sup-
pressor protein 14–3-3sigma is a sporadic event in cancer of the breast. Mol Cell
21. Peinado H, Ballestar E, Esteller M, Cano A (2004) Snail mediates E-cadherin repression
by the recruitment of the Sin3A/histone deacetylase 1 (HDAC1)/HDAC2 complex. Mol
Cell Biol 24:306–319.
22. Nass SJ, et al. (2000) Aberrant methylation of the estrogen receptor and E-cadherin 5?
CpG islands increases with malignant progression in human breast cancer. Cancer Res
23. Parrella P, et al. (2004) Nonrandom distribution of aberrant promoter methylation of
cancer-related genes in sporadic breast tumors. Clin Cancer Res 10:5349–5354.
24. Lehmann U, et al. (2002) Quantitative assessment of promoter hypermethylation
during breast cancer development. Am J Pathol 160:605–612.
25. Fackler MJ, et al. (2003) DNA methylation of RASSF1A, HIN-1, RAR-beta, Cyclin D2 and
Twist in in situ and invasive lobular breast carcinoma. Int J Cancer 107:970–975.
26. Roll JD, Rivenbark AG, Jones WD, Coleman WB (2008) DNMT3b overexpression con-
27. Li S, Rong M, Iacopetta B (2006) DNA hypermethylation in breast cancer and its
association with clinicopathological features. Cancer Lett 237:272–280.
in tumor metastasis. Cell 117:927–939.
beta-catenin expression during breast carcinogenesis and tumour progression: A
comparative study with CD44. Histopathology 34:25–34.
30. Park D, et al. (2007) Expression pattern of adhesion molecules (E-cadherin, alpha-,
beta-, gamma-catenin and claudin-7), their influence on survival in primary breast
carcinoma, and their corresponding axillary lymph node metastasis. Apmis 115:52–65.
mesenchymal transition is a rare event in vitro. Breast Cancer Res 6:R215–231.
www.pnas.org?cgi?doi?10.1073?pnas.0807146105 Dumont et al.