.JPG?ad=18503&adview=true)
[Cancer Research 60, 4130-4138, August 1, 2000]
© 2000 American Association for Cancer Research
Long-Term Hydroxytamoxifen Treatment of an MCF-7-derived Breast Cancer Cell Line Irreversibly Inhibits the Expression of Estrogenic Genes through Chromatin Remodeling1
Eric Badia2,
Marie-Josèphe Duchesne,
Abdelhabib Semlali,
Maryse Fuentes,
Claire Giamarchi,
Hélène Richard-Foy,
Jean-Claude Nicolas and
Michel Pons
Unité 439, Institut National de la Santé et de la Recherche Médicale, 34090 Montpellier [E. B., M. J. D., A. S., M. F., J-C. N., M. P.], and Laboratoire de Biologie Moléculaire Eucaryote, Centre National de la Recherche Scientifique, 31062 Toulouse Cedex [C. G., H. R-F.], France
 |
ABSTRACT
|
|---|
Antiestrogen resistance is frequently observed in patients after
long-term treatment with tamoxifen, a nonsteroidal antiestrogen widely
used for endocrine therapy of breast cancer. In vitro
studies in resistant cells showed that the expression of natural
estrogen-responsive genes is frequently altered. Using MVLN cells, an
MCF-7-derived cell model, we previously demonstrated that
4-hydroxytamoxifen (OHT) treatment irreversibly inactivated an
estrogen-regulated chimeric luciferase response by a direct effect of
the drug and not through a cell selection process (E. Badia et
al., Cancer Res., 54: 5860–5866, 1994). In the
present study, we present tamoxifen-resistant but still
estrogen-dependent clones isolated after long-term treatment of MVLN
cells with OHT and show that progesterone receptor (PR) expression was
irreversibly decreased in some of these clones, whereas the PRA:PRB
ratio of residual PR remained unchanged. The irreversible inactivation
of both chimeric luciferase gene and PR gene expression was associated
with the disappearance of DNase I-hypersensitive sites. In the case of
the chimeric gene, at least one of these sites was close to the
estrogen responsive element. Genomic sequencing analysis of a clone
with very low PR content did not reveal any methylation on CpG
dinucleotides or any mutation in the PR gene promoter region. In all of
the resistant clones tested and independently of their PR content,
estrogen receptor expression was only lowered by half and remained
functional, whereas pS2 expression was not modified. We
also observed that the residual luciferase activity level (1–2%) of
the MVLN clones, the luciferase expression of which had been
irreversibly inactivated, was raised 4-fold by trichostatin A
treatment. We conclude that long-term OHT treatment may modify the
chromatin structure and thus could contribute to differentially
silencing natural target genes.
|
INTRODUCTION
|
|---|
Long-term tamoxifen therapy is widely used in the management of
hormone-responsive breast cancer and is also being evaluated as a
preventive treatment for breast cancer in healthy women at risk for
breast cancer (1)
. Although curative tamoxifen treatment
is efficient in many patients, most of them experience further tumor
progression after several years of treatment, i.e.,
so-called acquired tamoxifen resistance. The exact molecular mechanisms
that account for acquired tamoxifen resistance remain unclear, but
there is an emerging body of evidence that multiple mechanisms could be
involved, concerning, for most of them, modifications in
trans-elements at different points along the
ER3
pathway: (a) alterations in the pharmacology and metabolism
of the AE (2)
; (b) ER mutations
(3)
; and (c) alterations in cross-talk between
signal transduction pathways (4
, 5)
or in the balance of
nuclear receptor coeffectors (6
, 7)
. Although each of
these mechanisms has been invoked to explain the tamoxifen resistance
process, none of them could account for the triggering of resistance
acquisition.
Acquired resistance to tamoxifen may be due either to a simple
selection process involving cells from an already heterogeneous
population or to a two-stage process first implying cell alteration by
the drug (here termed "direct effect") followed by the selection of
the new phenotype, provided that cell growth is no longer inhibited by
tamoxifen. In previous works (8
, 9)
on an MCF-7-derived
cell line, i.e., MVLN cells that express the luciferase gene
under the control of an ERE (10)
, we showed that OHT was
able to induce rapid and irreversible silencing of the luciferase gene.
The subsequent appearance of cell heterogeneity (cells expressing or
not expressing luciferase) was clearly incompatible with a selection
process, because the time of half inactivation was as short as 7 days.
We hypothesize that irreversibly affecting natural estrogen-responsive
gene(s) by an AE direct effect could be a primordial event that
triggers the resistance process; it then would lead to altered
expression of other genes involved in cell proliferation. In
vitro studies on tamoxifen- or OHT-resistant variant cell lines
showed that expressions of some natural estrogen-responsive genes are
frequently altered in these cells. The process leading to such
alterations has not yet been elucidated (11, 12, 13)
.
The aim of this paper is to show that long-term OHT treatment can
induce irreversible inhibition of the expression of estrogen-responsive
genes along with having a direct effect on chromatin structure. By
analyzing several OHT-resistant clones issued from OHT-treated MVLN
cells, we show that the AE irreversibly inactivated the expression of
the natural estrogen-controlled PR gene. This inactivation required a
longer treatment period than for the luciferase gene. Conversely, the
pS2 gene, another estrogen-regulated gene, was not
inactivated by such treatment. Inactivations of luciferase and PR genes
were associated with chromatin remodeling as revealed by the
disappearance of DHSSs, one of which was localized close to the ERE in
the chimeric gene.
Our results strongly suggest that (a) OHT can induce direct
and stable silencing of gene expression; and (b) this
inactivation process is associated with stable modification of the
chromatin structure. Whether a similar mechanism might be involved in
the silencing of genes leading to resistance acquisition of our
resistant cells and, more generally, of tumors primarily responding to
AE therapy is an open question.
|
MATERIALS AND METHODS
|
|---|
Materials
Materials for cell culture came from Life Technologies
(Cergy-Pontoise, France). Luciferin was synthesized by G. Auzou
(INSERM, Montpellier, France) according to the method of L. J. Bowie (14)
. OHT was a gift from Dr. A. E. Wakeling
(Imperial Chemical Industries, Macclesfield, England). Estradiol,
sodium bisulfite, hydroquinone, spermine, spermidine, proteinase K, and
NP40 were purchased from Sigma Chimie (Saint Quentin Fallavier,
France). DNase I (reference 6330) was from Worthington Biochemicals
(Freehold, NJ). Restriction endonucleases were from New England Biolabs
(Ozyme, Montigny le Bretonneux, France). The T-vector ligation kit,
Wizard DNA Clean-Up System, and competent JM 109 cells used for cloning
PCR products were from Promega (Charbonnières, France). The
sequencing kit, random labeling kits, nylon Hybond-N+ membrane, and the
ECL system were from Amersham Pharmacia Biotech (Les Ulis, France).
Detection of luciferase activity on cell-free extracts was performed
with an LKB-Wallac (Sundyberg, Sweden) 1251 luminometer. A
photon-counting camera (Argus 100) from Hamamatsu Photonics (Hamamatsu,
Japan) was used to detect luciferase activity in intact cells and to
analyze Western blots. The monoclonal anti-PR antibody (MA1–410) was
from Affinity-BioReagents (Interchim, Montluçon, France). A Fujix
BAS1000 PhosphorImager was used to analyze Northern blots.
Cell Line, Culture Conditions, and Luciferase Assay
The stable MVLN transfectant expressing the luciferase reporter
gene under estrogen control has been described elsewhere
(10)
. These MCF-7-derived cells express the firefly
luciferase gene Luc under control of the 5' flanking region
of the Xenopus vitellogenin A2 gene Vit
(15)
, inserted in front of the herpes simplex virus
promoter for tk. The stability of this expression was
checked over 70 passages in DMEM with phenol red, supplemented with 5%
FCS (FCS medium). For long-term treatments, cells were cultured in DMEM
without phenol red supplemented with 3% of a steroid-free,
dextran-coated charcoal-treated FCS (DCC medium). Medium was replaced
every other day. Luciferase activity was determined as described by
Badia et al. (8)
.
Isolation of Clones from OHT-treated MVLN Cells
After treatment with 10-7 M OHT in DCC
medium for various times, MVLN cells were dispersed at a density of 1
cell/cm2 to obtain separate clones in tissue
culture flasks. They were grown for 
1 month in FCS medium until the
clones were visible without magnification. Their NL or luminous
phenotype was determined using a photon-counting camera (after
supplementing the medium with 0.3 mM sterile luciferin).
Clones were individually harvested and grown for 1 extra month in FCS
medium. Clones NL 6 and NL 10 are NL clones obtained after a 12-day OHT
treatment, whereas clones CL 3.1–CL 3.35 and clones CL 6.1–CL
6.35 represent two NL clone series obtained after 3 and 6 months of
treatment, respectively. Clone CL 32inf was
obtained by culturing clone CL 6.32 in the presence of
10-7 M OHT for 12 extra months.
Cell Growth Assay
Each MVLN population or clone assayed was cultured for 5 days in
DCC medium. Cells were harvested, and 2 x 104 cells per well were seeded in 24-well tissue
culture cluster plates in the same medium. One day later, the medium
was replaced by fresh medium containing estradiol (0.1 nM),
OHT (100 nM), ICI 164.384 (100 nM), or vehicle
alone (0.1% ethanol). These media were renewed every 2 days. After an
8-day growth period with the different effectors, triplicate wells were
assayed for DNA content using 4',6-diamidino-2-phenylindole
(16)
.
PR Binding Assay and PR Isoform Analysis
Each MVLN population or clone was cultured in DCC medium for 5
days and then in DCC medium containing either 1 nM
estradiol or 100 nM OHT in 0.1% ethanol. After a 4-day
incubation, cells were harvested in PBS-EDTA, washed twice with PBS,
pelleted, and frozen at -20°C until assay. The pellets were thawed
in 1 ml of buffer (10 mM Tris-HCl, 1.5 mM EDTA,
and 1 mM DTT, pH 7.4), sonicated for 5 s at 0°C, and
then centrifuged at 180,000 x g for 40 min.
Cytosol aliquots (200 µl) were incubated at 0°C for 4 h in the
presence of either 10 nM
[3
H]promegestone or both 10
nM [3
H]promegestone and 2
µM [1H]promegestone.
Binding activity was measured by liquid scintillation counting after
dextran-coated charcoal treatment. Specific stimulated (with estradiol)
or unstimulated (with OHT) binding was expressed as the number of
femtomoles of [3
H]promegestone bound per
milligram of cytosol proteins, measured by the Bradford technique.
Relative expression of PRA and PRB isoforms was examined by Western
blotting on a 15–25-µl cytosol aliquot (i.e., 20 µg of
protein), using the ECL system after incubation with monoclonal mouse
anti-PR. The chemiluminescence level was quantified using a
photon-counting camera (Argus 100; Hamamatsu).
ER Binding Assay
Each tested clone was cultured in DCC medium for 5 days. Cells
were then harvested in PBS-EDTA, washed twice with PBS, pelleted, and
frozen at -20°C until assay. The pellets were treated as described
above for PR binding; cytosol aliquots (200 µl) were incubated at
0°C for 2 h with either 5 nM
[3
H]estradiol or both 5 nM
[3
H]estradiol and 1 µM
[1H]estradiol.
Northern Blot Analysis of pS2 Expression
MVLN cells that had grown for 3, 6, and 9 months in DCC medium
alone or in DCC medium containing 10-7
M OHT and clones isolated from MVLN cells treated with OHT
for 1, 3, and 6 months were cultured in 25-cm2
dishes and grown for 5 days in DCC medium and then in DCC medium
supplemented with 10-9 M estradiol
or 10-7 M OHT for 4 days. Total RNA
was isolated according to the method of Chomczynski and Sacchi
(17)
using RNAzol B reagent. Ten µg of total RNA were
electrophoretically separated on a 1% agarose denaturing gel and
transferred to a nylon membrane. The membrane was hybridized overnight
with 32P-labeled pS2 and 36B4 probes at 42°C in
50% formamide, as described previously (8)
. After
stringency washes, filters were first exposed to the PhosphorImager
screen to evaluate pS2 versus 36B4
expression and autoradiographed.
DHSS Analysis
The methodology was adapted from that of Richard-Foy et
al. (18)
as follows.
Nucleus Isolation.
Nuclei were isolated either from untreated MVLN cells or from clones CL
6.32, CL 6.32inf, and NL 6. Cells contained in 10
T150 flasks were brought near confluence; five flasks were then
incubated for 2 h in the presence of 10-9
M estradiol, and the other five were incubated in the
presence of 10-7 M OHT before
nucleus isolation. Cells were then carefully stripped, suspended in 40
ml of ice-cold PBS, which had been supplemented either with
10-9 M estradiol or with
10-7 M OHT (buffer H, buffer C, and
buffer W were similarly supplemented), and centrifuged for 5 min at
1000 rpm. The supernatant was discarded, and the pellet was resuspended
in 7 ml of ice-cold buffer H (15 mM NaCl, 60 mM
KCl, 0.15 mM spermine, 0.5 mM spermidine, 1
mM EDTA, 0.1 mM EGTA, 0.2% NP40, 5%
saccharose, and 10 mM Tris-HCl, pH 7.4). The cell
suspension was homogenized with 15 strokes of a cold Dounce
homogenizer. The state of the resulting nuclei was checked under a
microscope. The resulting slurry was centrifuged over 4 ml of ice-cold
buffer C (buffer H containing 10% saccharose) for 20 min at 3000 rpm.
The supernatant was discarded, and the pellet was washed twice with 13
ml of ice-cold buffer W (15 mM NaCl, 60 mM KCl,
0.15 mM spermine, 0.5 mM spermidine, and 10
mM Tris-HCl, pH 7.4). After a last centrifugation (3 min at
3000 rpm) the pellet was finally resuspended in 2 ml of buffer W.
DNase I Digestion.
Aliquots containing 500 µl of nuclei suspension (corresponding to
250–300 µg of DNA) were digested for 10 min at 37°C with various
amounts of DNase I (ranging from 1.6 to 8 units/500 µl) in the
presence of 0.5 mM CaCl2 and 1
mM MgCl2. DNase I digestion was
stopped by overnight incubation with 200 µl of proteinase K buffer
(25 mM EDTA, 2% SDS, and 200 µg/ml proteinase K).
Proteinase K was removed by phenol-chloroform extraction and ethanol
precipitation. The DNA was resuspended in 10 mM Tris-HCl
buffer (pH 7.4) containing 1 mM EDTA.
Southern Blot Experiments.
Approximately 40–60 µg of DNase Idigested DNA were cleaved by
StuI in 500 µl of the corresponding digestion buffer
(usually 3 units/µg DNA for 2 h; complete digestion was checked
by adding 0.5 µg of a known cleavable plasmid to a 20-µl aliquot of
the reaction solution). Digested DNA was then ethanol precipitated in
the presence of glycogen and 0.3 M NaOAc and resuspended in the sample
buffer. Forty µg of DNA were separated electrophoretically on a 0.8%
agarose gel and transferred to a nylon membrane. The membrane was
hybridized overnight with the appropriate
32P-labeled probes. Positions of the molecular
weight marker bands were measured before transfer onto the nylon
membrane.
|
Chemical Modification of DNA with Sodium Bisulfite
|
|---|
The sodium bisulfite reaction was performed as described by
Frommer et al. (19)
with slight modifications
partially reported by Raizis et al. (20)
.
Briefly, sodium bisulfite (1.9 g) was mixed with 2.5 ml of water. Then
0.7 ml of 2 M NaOH and 0.5 ml of 1 M hydroquinone were added. Template
DNA (5 µg) in 30–50 µl of water was denatured by a 10-min
incubation at room temperature after the addition of 0.1 volume of 2
M NaOH. DNA was resuspended in 1 ml of bisulfite
solution and then successively incubated for 4 h at 50°C, heated
at 95°C for 5 min, cooled to 50°C for 1 h, and then cooled to
4°C. DNA was purified with the Wizard DNA Clean-Up System and
dissolved in 100 µl of water. After neutralization (43 µl of 1
N NaOH) and precipitation, the pellet was
resuspended in 50 µl of water.
|
PCRs, Cloning, and Sequencing of the Products
|
|---|
Direct PCR Products.
The three pairs of oligonucleotides, D3 and D2, D8 and D6, and D4 and
D5 (see 
Fig. 4B
), generated 994-, 612-, and 617-bp-long
fragments, respectively. These fragments covered the entire promoter A
and part of promoter B as defined by Kastner et al.
(21)
. The PCR fragments were purified on an agarose gel,
cloned in the T-vector, and sequenced.
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 1. Growth of MVLN cells during long-term OHT treatment. MVLN
cells were seeded in a T25 flask and grown in DCC medium with
10-7 M OHT for 25 weeks. As soon as cells
reached confluence, one-tenth of them were plated in a new T25 flask.
Cell number was determined on the basis of the dilution at each
passage. The logarithm of this cell number was plotted against the
number of treatment weeks (filled symbols). Cell
doubling time was then graphically determined. A theoretical model was
constructed, which simulates variations in growth rates of a population
of cells in which three cells/106/day raise their growth
rate instantaneously from 120- to 40-h doubling time (open
symbols).
|
|
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 2. PR content of MVLN and OHT-treated resistant cells and
PRA:PRB ratio analysis. PR content of MVLN cells and individual clones
was determined after 4-day stimulation with 1 nM estradiol.
PR content of untreated MVLN cells was taken as 100%.
A, MVLN cells were grown in DCC medium containing
10-7 M OHT for 1, 3, 6, and 9 months.
B and C, clones 1–35 were
isolated as described in "Materials and Methods" from MVLN cells
grown in DCC medium in the presence of 10-7 M
OHT for 3 months (B) and 6 months (C).
D, each clone was grown for 4 days with 1 nM
estradiol, after which 20 µg of cell extracts were submitted to 12%
SDS-PAGE, transblotted, and processed for Western analysis.
|
|
View larger version (80K):
[in this window]
[in a new window]
|
Fig. 3. Induction of pS2 in MVLN cells and clones
isolated from cells treated with OHT. A, MVLN cells that
had grown for 3, 6, and 9 months in DCC medium alone or in DCC medium
containing 10-7 M OHT were treated as
described in "Materials and Methods" with either 10-9
M estradiol (E2) or 10-7
M OHT. Total RNA was then isolated and processed for
Northern blot analysis using a pS2 probe as described in "Materials
and Methods." B, various clones were isolated from
MVLN cells treated with 10-7 M OHT for 1 month
(CL 1.4, CL 1.15, and CL 1.7), 3 months
(CL 3.23 and CL 3.24), and 6 months
(CL 6.7 and CL 6.27). Each of these
clones was analyzed as in A.
|
|
View larger version (36K):
[in this window]
[in a new window]
|
Fig. 4. Scheme of PR promoter and cartography of
Vit-tk-Luc copies integrated in MVLN cells. A, schematic
representation of the insert containing the pVit-tk-Luc copies
integrated in the MVLN cell line. Each of the three copies described
contained the entire gene unit (encompassed in the
XhoI-XhoI restriction fragment, here
denoted X) including the vitellogenin gene ERE, the tk
promoter, and the luciferase gene, as well as the polyadenylation site,
although the junction between two consecutive copies is not identical.
P1 represents a 665-bp probe edged by
BanI-BanII restriction sites, and
P2 is a 1654-bp probe edged by
XbaI-XbaI restriction sites.
B, schematic representation of the promoter part of
PR gene. Each vertical dash represents a CpG
dinucleotide. Filled triangles and
diamonds represent sites of transcription initiation for
PRA and PRB, respectively. The sequences and locations of the various
oligonucleotides used in the experiments are depicted.
|
|
PCR Products from Bisulfite-treated DNA.
PCR was performed on 2.5 µl of the reacted DNA with
oligonucleotides that matched the new sequence. The upper (coding)
strand of the PR gene promoter was amplified with the two
pairs of oligonucleotides, F1 and F2, and F3 and F4 described in Fig. 4B
. The sequences of oligonucleotides F1 and F3 were deduced
from the natural upper strand by changing C to T, and those of F2 and
F4 were deduced from the natural lower strand by changing G to A. The
natural sequences corresponding to these oligonucleotides are devoid of
methylatable CpGs; therefore, they fit perfectly with the gene sequence
modified after the bisulfite reaction. The PCR products were purified
on agarose gel, cloned in the T-vector, and sequenced.
|
RESULTS
|
|---|
OHT-treated MVLN Cells Acquired Growth Independent of This AE
Growth Rate Evaluated by Passage Frequency.
The MVLN cell line, a stably transfected MCF-7 cell line, is clonal. As
for the MCF-7 cell line, MVLN cell proliferation is stimulated by
estradiol and inhibited by AEs (22)
. At the beginning of
continuous 10-7 M OHT treatment, the
MVLN cell doubling time was as long as 120 h (Fig. 1
). However, after 6 weeks of culture, the MVLN doubling time shortened,
reaching a steady state of 40 h after 10 weeks. The experimental
results quite closely fit a theoretical model whereby an MVLN cell
population grown in the presence of OHT (doubling time, 120 h)
acquires AE resistance due to the modification of three cells per
million per day, the doubling time of which drops to 40 h.
Although speculative, this model suggests that as much as 95% of the
cells grown for 3 and 6 months in the presence of OHT could derive from
10 to 20 resistant cells modified in the first week of treatment.
Selection then became the driving force of cell proliferation.
Growth Rate Evaluated by Measuring DNA Content.
At the end of 3-, 6-, and 9-month OHT treatment times, 35 clones were
isolated from the culture. All of these clones grew in the presence of
OHT, which notably stimulated proliferation. Table 1
shows the proliferation of MVLN cells and eight clones isolated from
6-month treated cells (CL 6.1, CL 6.5, CL 6.7, CL 6.8, CL 6.20, CL
6.27, CL 6.28, and CL 6.32) after 8 days incubation of cells with
various effectors. In DCC medium, it was 2–3-fold higher than at
seeding (data not shown). The DNA content of
10-7 M OHT- and
10-9 M estradiol-treated resistant
clones was 2- and 4-fold higher, respectively, than that of cells
grown in DCC, whereas it was 2-fold lower in
10-7 M ICI 164384-treated cells.
Conversely, in parental MVLN cells, OHT as well as ICI 164384 inhibited
cell proliferation by half. These results were confirmed by experiments
in which cell growth was evaluated by the
[3
H]thymidine incorporation method (results not
shown).
View this table:
[in this window]
[in a new window]
|
Table 1 Growth rate and PR and ER content of MVLN cells and clones isolated
from 6-month OHT-treated cells
Among the clones issued from 6-month OHT-treated MVLN cells, eight were
selected for their different PR contents, i.e., very low for
CL 6.7 and CL 6.32, low for CL 6.1 and CL 6.20, average for CL 6.5 and
CL 6.8, and high (similar to PR content in untreated cells) for CL 6.27
and CL 6.28. Results (mean ± SD) are given as
femtomoles per milligram of mg proteins. The numbers of experiments
(each performed in duplicate) are given in parentheses for PR content;
the number was three for ER content. The proliferation was studied in
various culture media (DCC or in the presence of estradiol (E2), OHT,
and ICI 164384). Growth rates are presented as a percentage of DNA
content (mean ± SD for three determinations) of cells
grown in the presence of estradiol.
|
|
|
Estrogen-stimulated PR Expression of MVLN Cells Is Irreversibly
Decreased as a Function of OHT Treatment Time
|
|---|
In a previous paper (9)
, it was shown that the MVLN
cell expression of firefly luciferase was irreversibly inactivated by a
short OHT treatment (<3 weeks), and we showed that this inactivation
resulted from a 50-fold inactivation of individual cells. We describe
here the irreversible inactivation of PR expression, a natural
estrogenic response. The estrogen-stimulated PR content of MVLN cell
cultures decreased with OHT treatment time during the first 6 months of
treatment (Fig. 2A
). Because this PR content was only a mean value, we also
measured the PR content of individual clones issued from 3- and 6-month
OHT-treated MVLN cells, respectively, (Fig. 2 and C
) and observed that PR content (mean ± SD)
was 60 ± 26 and 25 ± 23%, respectively
(100% is the MVLN cell PR content, i.e., 648 ± 110 fmol/mg of proteins). This indicated that (a)
the mean PR content decreased as the treatment time increased; and
(b) the PR contents of individual clones varied markedly as
suggested in each case by the high SD value (see above: 26 and 23%),
reflecting the progressive appearance of heterogeneity and the
randomness of this event. After 3 or 6 months of OHT treatment, clones
were grown for 2 months in FCS medium (an estrogenic culture condition)
before harvesting; hence the fact that the PR content was still low
indicated that the loss of PR expression was irreversible. Northern
blot experiments suggested that this inactivation occurred at the
transcription level (results not shown). As a control, the
estrogen-stimulated PR content of 12 clones issued from the parental
MVLN cells was 108 ± 20%.
Approximately 100 clones issued from MVLN cells treated for 3, 6, or 9
months with OHT were analyzed by Western blotting for their respective
PRA and PRB contents. On seven clones obtained after 3 months of
treatment and expressing various PR levels, Fig. 2D
shows
that both PR isoforms disappeared proportionally regardless of the PR
content (wells were loaded with an equal cytosol protein amount).
PRA:PRB ratios remained stable around the 3–4 value obtained with MVLN
cells. This suggests that both promoters were equally modified by the
OHT treatment. Although the band intensities in many clones issued from
cells treated for 6 and 9 months (data not shown) were faint, no
differences in the PRA:PRB ratios were noted in the clones analyzed. We
also observed (Fig. 2D
) that MVLN cells expressed
significantly less PR than T47D cells, used as a control in these
experiments with a smaller PRA:PRB ratio.
|
Estradiol Receptor Expression Is Decreased by Half in OHT-treated
MVLN Cells
|
|---|
The eight clones (CL 6.1, CL 6.5, CL 6.7, CL 6.8, CL 6.20, CL
6.27, CL 6.28, and CL 6.32) expressing various amounts of PR were
analyzed for their ER content (Table 1)
. They exhibited nearly 50% of
the MVLN cell ER content (292 fmol/mg of protein). Very low PR levels
in CL 6.7 (2.6%) and CL 6.32 (3.5%) or high PR levels in CL 6.27
(117%) and CL 6.28 (88.5%) were thus associated with comparable ER
levels (36, 40, 48, and 68%, respectively), suggesting that the PR
expression level is not directly related to the ER level.
|
pS2 Expression Is Not Modified in OHT-treated MVLN
Cells
|
|---|
Fig. 3A
shows the results of the analysis of the pS2
transcription level of MVLN cell cultures treated for various times
with OHT. Estradiol-stimulated and basal (in the presence of OHT)
pS2 mRNA levels, compared with those of the 36B4
mRNA standard, were not significantly modified. The same results were
obtained with clones issued from these treated MVLN cells (Fig. 3B
). It is noteworthy that pS2 expression was
constant regardless of the PR expression levels (Fig. 2
). For example,
CL 3.23 and CL 6.27 on one hand and CL 3.24 and CL 6.7 on the other
expressed a normal PR level and a very low PR level, respectively. As
discussed below, this latter result shows that the different natural
estrogen-controlled responses may have different degrees of sensitivity
to long-term OHT treatment. In addition, when administered with
estradiol, OHT was still able to counteract pS2 induction in
cells pretreated for 6 months with OHT, indicating that its
antiestrogenic effect was still efficient on this estrogen-controlled
response of resistant cells (results not shown). These results also
show that the ER and the basal transcription machinery functions were
not affected by prolonged exposure to OHT and were still able to elicit
some of the estrogenic responses.
|
Irreversible Inhibition of Vit-tk-Luc Gene Expression
in MVLN Cells Is Associated with Complete Disappearance of DHSSs
|
|---|
In a previous study (9)
, Southern blotting
experiments suggested that three copies of the Vit-tk-Luc plasmid
placed in a head-to-tail configuration were integrated in the MVLN
cells (Fig. 4A
). It was shown that the XhoI-XhoI
restriction fragment, which contains the complete gene unit composed of
the luciferase gene, the regulatory and promoter elements, and the
polyadenylation site, was entire in each copy, and the copies were not
rearranged by OHT treatment. In the present study, digestion with
StuI, which does not cleave pVit-tk-Luc, gave only one band
with a length (21 kbp; Fig. 4A
) compatible with that of an
insert containing the three plasmid copies flanked by short genomic
sequences. In this StuI-StuI fragment, DHSSs were
analyzed in cells with luciferase that was either expressed or
irreversibly inactivated. In MVLN cells, at least four DHSSs were
visible, as indicated by the arrows in Fig. 5
, all of them being inducible by estradiol stimulation. In clones NL 6,
CL 6.32, and CL 32inf, the luciferase expression
of which was irreversibly inactivated, DHSSs 2–4 completely
disappeared, whereas site 1 was dramatically decreased.
View larger version (51K):
[in this window]
[in a new window]
|
Fig. 5. DNase I hypersensitivity sites of the insert containing
pVit-tk-Luc copies in MVLN, NL 6, CL 6.32, and CL 32infcells. MVLN cells and clones NL 6, CL 6.32, and CL
32inf were cultured in FCS medium. Two h before nucleus
isolation, they were incubated with 10-9 M
estradiol (E2) or 10-7 M OHT,
as indicated below the autoradiograph.
Nucleus isolation and DNase I digestion were performed as described in
"Materials and Methods." DNase I amounts (units/500 µl) are
indicated above each lane on the autoradiograph. The
Southern blotted DNA was hybridized with P2 containing the luciferase
part of the Vit-tk-Luc plasmid. Arrows, DNase I
hypersensitivity sites.
|
|
|
Localization of DHSSs in Integrated pVit-tk-Luc Copies
|
|---|
To localize DHSSs in the StuI-StuI fragment
of the pVit-tk-Luc integrated copies, DNA from estradiol-stimulated
MVLN cells was restricted by ApaLI and BanII
(Fig. 4A
) and gave a fragment encompassing the ERE near its
5' end. The pattern obtained after Southern blotting and hybridization
with probe P1, a 665-bp BanI-BanII fragment
strictly included in the ApaLI-BanII fragment,
gave, as expected, a single 1950-bp-long band for DNA undigested by
DNase I (Fig. 6
, line A). After digestion with DNase I, the
ApaLI-BanII DNA fragment was partially cleaved,
and a single extra band of 1400 bp was generated (Fig. 6
, line
B). Because the ERE is located 1450 bp upstream of the
BanII restriction site, the cartographic analysis revealed
that at least one of the four above-mentioned DHSSs was located close
the ERE of one pVit-tk-Luc copy.
View larger version (47K):
[in this window]
[in a new window]
|
Fig. 6. Localization of the DNase I hypersensitivity sites of the
insert containing pVit-tk-Luc copies in MVLN cells. Forty µg of DNA
obtained from MVLN cell nuclei stimulated by estradiol were digested or
not with 8 units/500 µl of DNase I, as indicated above
the autoradiograph, and then extracted by
phenol-chloroform and cleaved by ApaLI + BanII restriction endonucleases (3 units/µg of DNA).
The digestion products were analyzed by Southern blotting on a 0.8%
agarose gel and probed with P1.
|
|
|
Irreversible Inhibition of PR Gene Expression in MVLN Cells Is
Associated with Partial Disappearance of a DHSS
|
|---|
Hypersensitive sites were studied in the 5' region of the
PR gene in untreated MVLN cells and in CL 6.32, a 6-month
OHT-treated clone with very low PR expression. At least two DHSSs could
be distinguished on the autoradiograph: site 1 located around position
-570 bp and site 2 (faint) around position +530 (Fig. 7
). Site 3 around position +1130 of the PR gene promoter also
present in undigested DNA could not be considered a DHSS. Although
these sites were not obviously inducible by 2-h estradiol stimulation
(results not shown), the intensity of bands corresponding to sites 1
and 2 was much lower in clone CL 6.32 than in MVLN cells.
|
DHSS Disappearance in the PR Gene Is Not Associated
with Methylation or with Mutations in Its Promoter Part
|
|---|
Fig. 4B
shows the location of the CpG dinucleotides
belonging to the promoter part of the human PR gene
(21)
. They are densely associated in promoter A and in the
3' end of promoter B. Using the bisulfite reaction, an exhaustive
analysis of the methylation of this region was performed on genomic DNA
from clone CL 6.7, which exhibits a very low PR expression level. The
PCR products obtained with primer pairs F1 and F2, and F3 and F4 were
cloned, and individual molecules belonging to 10 different "PCR
clones" were sequenced (to ensure that both alleles would be
analyzed). On a total of 53 CpGs present between positions -164 and
+1044, no methylation was observed in any of the 10 tested PCR
clones. Examples of such sequences are shown in Fig. 8 and D
.
View larger version (90K):
[in this window]
[in a new window]
|
Fig. 8. Genomic sequencing of CL 6.7 bisulfite-treated DNA.
A and B, control reaction. DNA from the
pVit-tk-Luc plasmid (30 pg mixed with 10 µg of DNA extracted from
MCF-7 cells, which do not contain pVit-tk-Luc) was methylated
(A) or not (B) with SssI
CpG methylase. Each DNA was then subjected to the bisulfite reaction.
Products were PCR amplified using two pairs of oligonucleotides
surrounding the tk promoter, which amplify the coding strand only,
cloned in the T-vector, and sequenced with the primer located on the 3'
end sequence. The sequence is thus read as the original DNA in which
all of the guanine residues (facing a reacted cytosine) have been
converted to adenines except those facing a methylated cytosine. In the
native nonmethylated plasmid (B), all guanines were read
as adenines, and no bands appeared in the G track. In
the plasmid methylated with SssI CpG methylase
(A), guanines were read as adenines except for those
facing a CpG motif cytosine and, therefore, appeared as bands in the
G track. C and D,
PR promoter methylation status of a OHT-treated clone. DNA
from clone CL6.7 was reacted with bisulfite, and the upper (coding)
strand of the reaction product was amplified using F1 and F2 primers
(Fig. 4
B). PCR products were sequenced with the primer
located on the 3' end (C) and 5' end (D);
the sequences are thus read as the original DNA, in which all of the
guanine residues have been converted to adenines except those facing a
methylated cytosine (C) and the cytosine residues have
been converted to thymines except those that were methylated
(D).
|
|
The PR gene promoter part (unreacted with bisulfite) was
also amplified using three pairs of primers, D3 and D2, D8 and D6, and
D4 and D5 (Fig. 4B
). PCR products were cloned, and
individual molecules belonging to five different PCR clones were
sequenced. No mutation was observed in this part of the promoter, which
mainly controls PRA expression (21)
.
|
Trichostatin A Experiments
|
|---|
A histone deacetylase inhibitor such as trichostatin A increases
the acetylated histone level in many cell types and thus was expected
to enhance the expression of some repressed genes. The effect of
trichostatin A was investigated on cells from the parental MVLN cell
line and from NL 6 and CL 32inf clones (Fig. 9
). These two clones displayed a residual luciferase activity that could
be estradiol induced to 2% at most of the maximal value reached by the
parental MVLN cell line. We observed a small but reproducible increase
(by 4-fold) of the estradiol-induced luciferase expression on cell
lines in which luciferase was irreversibly inactivated but not in MVLN
cells. No significant effect of TSA (used at 200 nM for
48 h) was observed either in cell lines cultured in DCC (control,
results not shown) or in the presence of OHT. Conversely, this
treatment had no effect on PR induction in CL 6.32 (results not shown).
View larger version (25K):
[in this window]
[in a new window]
|
Fig. 9. Effect of trichostatin A on the luciferase expression of
OHT-treated clones. MVLN cells, clone NL 6, and clone CL
32inf were cultured either in FCS medium or in DCC medium
for 24 h. FCS and DCC media were then changed and supplemented
with 1 nM estradiol (E2) with or without 200
nM trichostatin (TSA) or 10-7
M OHT with or without 200 nM trichostatin
(TSA), respectively, for 48 h.
|
|
|
DISCUSSION
|
|---|
Apart from the well-known beneficial antiproliferative or
antitumoral effect of the drug, studies on the direct side effects of
tamoxifen have mainly been restricted to investigating the possible
formation of drug adducts with DNA, prompting mutation events
(23, 24, 25)
. However, it has not been clearly proven that
these mutations could lead to gene inactivation. The present study,
performed on MVLN cells derived from MCF-7 cells, was designed to show
that tamoxifen could also alter the expression of natural and chimeric
genes in breast cancer cells (PR and Vit-tk-Luc
genes) through epigenetic modifications of the chromatin template. In
the case of the chimeric Vit-tk-Luc gene, the high gene
inactivation rate (t1/2, 7 days) is clearly
incompatible with a selection process (8)
. This does not
a priori seem to be as clear for natural genes and, in
particular, the PR gene, because a few months of OHT treatment are
required to irreversibly inhibit its expression in most cells. A
homogeneous (tagged with the chimeric luciferase gene) MVLN cell line
was isolated and used for the long-term OHT treatment experiments.
During the first 3 months of treatment, the antiproliferative effect of
OHT promoted the selection of resistant clones, and many but not all
resistant clones irreversibly lost PR expression. In this regard,
Graham et al. (26)
already observed that
tamoxifen treatment could alter PR content in T47-D cells, leading to
mixed subpopulations of cells. Once resistant cells had emerged and the
growth rate was stabilized (after 3 months of OHT treatment), the mean
level of PR expression continued to decrease (see Fig. 2
after 6 months
of treatment). Therefore, no synchronism between OHT resistance
acquisition and PR inactivation was observed, showing that PR gene
inactivation is not a prerequisite for the emergence of resistant
cells.
Although it cannot be definitively excluded that drug-induced
mutations could be responsible for the inactivation of PR or luciferase
genes, no mutation was observed in their promoter parts, suggesting
that these inactivation processes might rather involve a change in
chromatin structure. The human PR gene promoter has been
extensively studied by Kastner et al. (21
, 27)
,
who found that PR expression is controlled by two promoters, one
located between -711 and +31 (promoter B) and one located between +464
and +1105 (promoter A), that direct the synthesis of mRNA transcripts
originating from two clusters of transcription start sites and coding
for PRB and PRA proteins, respectively. The expression of PRA and PRB
differs in a cell type-, promoter-, or ligand-specific manner: for
example, Graham et al. (28)
showed that
estradiol induced preferential stimulation of PRB expression in human
T47-D breast cancer cells. This suggests that if both promoters can be
differentially activated, their inactivation might also occur
independently. We investigated this point by Western blotting
experiments. Clearly, PR expression was irreversibly decreased, whereas
the PRA:PRB ratio of residual PR remained unchanged,
suggesting that the inhibition process did not affect a limited part of
the gene but involved a large portion of the chromatin template.
Because gene expression is usually associated with the induction or at
least the presence of DHSSs (29)
, we wished to determine
whether the DHSS pattern of the two inactivated genes would be coherent
with that observation. Such was the case, because (a) the
four estrogen-inducible DHSSs in the luciferase gene were absent in two
clones with luciferase expression that was irreversibly inactivated;
and (b) the intensities of DHSS 1 and DHSS 2 bands belonging
to the PR promoter were much weaker in the PR-negative clone
CL 6.32 than in untreated MVLN cells, although the sites were not
estradiol inducible. This could be compared with a study on the mouse
uterus PR gene (30)
in which three DHSSs were
found in the 5' region, and no clear inducibility of any of these sites
was observed. These results highly suggested that both luciferase and
PR genes were irreversibly inhibited along with a parallel closure of
the promoter chromatin template.
To date, the epigenetic long-term silencing of genes has been
shown to be involved in situations as diverse as position-effect
variegation in Drosophila, telomeric position effect in
yeast, X chromosome inactivation, control of homeotic gene clusters
during the development and imprinting in mammals (31, 32, 33, 34)
,
as well as some silencing of reporter transgenes (35)
. DNA
methylation could be involved in permanent silencing (36)
through either direct interference of methylation with the binding of
transcription factors or the binding of specific repressors such as
MeCP1 or MeCP2 to methyl-CpG (37)
. Our results suggest
that DNA methylation was not involved in PR gene silencing, because no
methylated cytosine was found in the part of PR gene
promoter that contains densely associated CpGs, and because the
strongest DHSS1 was located in a part of promoter B that contains very
few CpGs. With regard to clones in which luciferase expression is
inhibited, one CpG methylation was previously observed (9)
and strictly correlated with gene expression disappearance; albeit this
methylation site belongs to one of the two NotI restriction
sites of the reporter gene polylinker and, therefore, outside of the
gene promoter part. The luciferase expression inhibition was
furthermore not reversed by a 5-azacytidine treatment (9)
.
CpG methylation therefore does not seem to be the main mechanism
leading to the inactivation of PR and luciferase expressions in
OHT-treated MVLN cells. In a recent work, Ferguson et al.
(38)
showed that a few CpGs located downstream of promoter
A of PR were methylated in ER-and PR-negative MDA-MB-231
cells but that these methylations per se cannot prevent PR
gene induction by transfected ER. This result again suggests that the
methylation process is not responsible for the PR-negative phenotype.
It was recently shown that histone deacetylation could also be
involved in gene silencing (39, 40, 41)
, whereas histone
acetylation was found to be involved in transcriptional activation
mediated by nuclear receptors (42, 43, 44)
. However, the
influence of histone acetylation on steroid hormone gene expression is
not clear, because both inhibitory (45)
and stimulatory
(46)
effects have been reported. Treatment of
OHT-inactivated clones with trichostatin A, a deacetylase inhibitor,
increased luciferase expression 4-fold at most but was ineffective on
PR gene expression. No synergy between trichostatin A and azacytidine
was observed (results not shown) such as that reported in a recent
review (47)
.
Differences observed concerning DNA methylation and histone
deacetylation of PR and luciferase genes may reflect a
difference in mechanisms between the rapid Vit-tk-Luc
inactivation and the much slower PR inactivation.
Another gene-silencing mechanism could involve the formation of
"inactive heterochromatin-like" condensed structures initiated by
appropriate factors (48, 49, 50)
, suggesting that mechanisms
other than methylation or acetylation should be investigated. Recently,
a functional link between these mechanisms and those involving nuclear
receptors began to emerge (51, 52, 53, 54)
. These findings
substantiated a mechanism that would involve a direct effect of nuclear
receptors in chromatin structure remodeling. It is not yet known
whether such a mechanism is involved in OHT-induced silencing.
In conclusion, the present study demonstrated that, besides its
reported genotoxic effects, tamoxifen could also directly and
irreversibly alter estrogen-dependent gene expression in cultured
breast cancer cells, through epigenetic modifications of the chromatin
template. This is the first documented example of long-term gene
silencing induced by an antihormone. In our resistant clones, this
phenomenon differentially affects hormone-responsive genes, because we
observed that it modified the expression of two estrogen-induced genes
at different rates and that the expression of a third one,
i.e., the pS2 gene, was not modified. It should
still be determined whether such irreversible inhibition of gene
expression could be involved in OHT resistance acquisition by
shutting off the expression of putative growth suppressors. Besides the
classical ways mentioned in "Introduction" that could explain the
resistance process, we think that estrogen-controlled gene alteration
at the chromatin structure level deserves special attention, because it
may be the consequence of a direct and primordial effect of the AE.
Using the OHT-resistant cells described in this paper, we are presently
addressing this question by an exhaustive search for irreversibly
inactivated genes.
|
ACKNOWLEDGMENTS
|
|---|
We thank David Manley for correcting the English in the
manuscript.
|
FOOTNOTES
|
|---|
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.
1 This work was supported by the Institut National
de la Santé et de la Recherche Médicale and Association
pour la Recherche sur le Cancer Grant 5002. 
2 To whom requests for reprints should be
addressed, at Unité 439, Institut National de la Santé et
de la Recherche Médicale, 70 rue de Navacelles, 34090
Montpellier, France.
3 The abbreviations used are: ER, estrogen
receptor; AE, antiestrogen; ERE, estrogen response element; OHT,
4-hydroxytamoxifen Z; PR, progesterone receptor; DHSS, DNase
I-hypersensitive site; tk, thymidine kinase; NL, nonluminous; H,
homogenization; C, centrifugation cushion; W, washing.
Received 12/13/99.
Accepted 5/30/00.
|
REFERENCES
|
|---|
Top
ABSTRACT
INTRODUCTION
