| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ARTICLES |
Tenovus Cancer Research Center, Welsh School of Pharmacy, Cardiff University, Cathays Park, Cardiff, United Kingdom CF10 3XF
Address all correspondence and requests for reprints to: Prof. R. I. Nicholson, Tenovus Cancer Research Center, Welsh School of Pharmacy, Cardiff University, Redwood Building, King Edward VII Avenue, Cathays Park, Cardiff, United Kingdom CF10 3XF. E-mail: mcclellandra{at}cardiff.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, and extracellular signal-regulated kinase 1/2 MAPK
activation levels are all increased. Unlike wild-type cells, FASMCF
cells are highly sensitive to growth inhibition by an EgfR-specific
tyrosine-kinase inhibitor (TKI), ZD1839 (Iressa), and an inhibitor of
the activation of MEK1 (MAPKK), PD098059.
Short-term (
3 weeks) withdrawal of cells from antiestrogen had no
effect on growth or phenotype, whereas longer withdrawal (>10 weeks)
appeared to partially reverse the cellular phenotype with increasing
estrogen receptor and decreasing EgfR levels.
In subsequent studies FASMCF cells were maintained in TKI, where their growth was again suppressed and secondary TKI resistance failed to develop within the 3-month period in which initial ICI 182,780 resistance arose. Furthermore, wild-type cells similarly maintained in combination ICI 182,780 and TKI treatment conditions remained growth arrested (>6 months), with notable cell loss through both reduced rates of cellular proliferation and increased cell death.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Among a number of mechanisms proposed to explain the acquisition of tamoxifen resistance, foremost is the recognition that tamoxifen may have partial estrogenicity (2, 3). Substituting for estradiol in its occupancy of the ER, the tamoxifen-receptor complex binds to specific recognition sequences [estrogen response elements (EREs)] in the DNA of responsive genes where ligand-dependent, activation factor-2 domain-mediated, gene transcription is inhibited. Perversely, however, tamoxifen appears capable of partial agonism of ligand-independent transcription from these same ERE-bearing, estrogen-responsive genes through trans-activation via the receptor’s activation factor-1 domain. Continued exposure to the drug may thus lead to an alteration in the role played in cellular growth processes by the gene products resulting from this tamoxifen-modified transcription, and this may ultimately result in the development of resistant disease.
Recently, a new class of endocrine agents, the pure antiestrogens, typified by ICI 182,780, have been introduced to the clinic (4, 5, 6, 7, 8, 9). These compounds differ significantly from tamoxifen in their mode of action, and ICI 182,780, for example, is reported to antagonize the nuclear/cytoplasm shuttling of ER, inhibiting its reentry to the nucleus and promoting its rapid degradation (10). It may also further reduce the effective levels of cellular ER by impeding receptor dimerization and DNA binding (11). It is significant that, unlike tamoxifen, these compounds show no estrogen agonist activity (4, 12).
Although clinical experience with these drugs is still in its infancy, initial clinical trials show promise. Indeed, early data suggest that ICI 182,780 may prove useful in inducing tumor remissions even within patients already heavily pretreated by antihormonal therapies, including a number with acquired tamoxifen-resistant disease (4, 13). To date, no clinical reports demonstrating the development of pure antiestrogen-resistant cancer have been published. However, in vitro studies demonstrate that the continuous long-term exposure of steroid hormone-responsive breast cancer cells that are initially growth inhibited by pure antiestrogens can lead to a loss of sensitivity to the compound, as defined by altered growth characteristics and an apparently resistant phenotype (14, 15). It is interesting to note that some of these studies report that long term pure antiestrogen exposure leads to the development of cells also seemingly cross-resistant to tamoxifen (15), but that the reverse situation does not necessarily occur (16).
A growing number of clinical studies have established an association between poor prognosis for breast cancer and the overexpression of members of the erbB receptor family (17, 18, 19). We have demonstrated a strong inverse relationship between ER and epidermal growth factor receptor (EgfR) protein expression, in both primary and recurrent breast cancer. Furthermore, we and others show that patients with EgfR-positive recurrent tumors are less likely to benefit from endocrine therapy and have reduced relapse-free and overall survival periods than EgfR-negative patients (17, 18).
Continuous culture of breast cancer cell lines in the presence of antiestrogen has led to the development of a number of resistant cell sublines. These represent potentially important models for study of the loss of endocrine sensitivity and the acquisition of the resistant phenotype. Although few of these lines are fully characterized, a number of significant differences in phenotype have been identified between, for example, MCF7 resistant sublines developed by similar methods.
This paper documents the development and subsequent characterization of a novel MCF7 breast cancer cell subline that has become capable of growth in the presence of ICI 182,780. Importantly, we show that although these cells grow in ICI 182,780-supplemented medium, they do not appear to be resistant to its cellular effects, but as a result of effective antiestrogen-mediated suppression of ER-directed cell signaling, they have adapted to more fully use at least one alternative growth-signaling pathway, as witnessed by their enhanced expression of EgfR. As these cells also demonstrate an increased sensitivity to an EgfR-selective tyrosine kinase inhibitor (TKI), ZD1839 (Iressa) (20, 21), vs. the parental MCF7 cell line, this paper suggests that these cells represent an important model of on-therapy breast cancer and proposes a novel therapeutic approach to the treatment of breast cancer.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
All cell counts were performed in triplicate on trypsinized cells resuspended from three replicate wells from a 24-well cell culture plate measuring each in a total volume of 10 ml using a Multisizer II cell counter (Beckman Coulter UK Ltd., High Wycombe, UK). 17ß-Estradiol was purchased from Sigma (Dorset, UK). TKI (ZD1839, Iressa) was a gift from Dr. Alan Wakeling, AstraZeneca. MEK inhibitor, PD98059, was purchased from Alexis Corp. (Nottingham, UK).
Immunocytochemistry
All immunocytochemistry was performed on cells seeded onto
sterile 3-aminopropyl-triethoxy-silane-coated glass coverslips during
log phase growth at 7.5 x 104
cells/coverslip in a 35-mm diameter dish with 2 ml medium and cultured
for 7 days before fixation (unless otherwise stated).
ER immunocytochemistry (ER-ICA). Evaluation of the cellular ER content was performed using both the ER ICA monoclonal antibody kit (Abbott Laboratories, North Chicago, IL; for procedure, see Refs. 22 and 23) and a mouse monoclonal ID-5 antiserum (24) (M7047, DAKO Corp., Ely, UK). The antiserum of ER ICA (H222) is reported to recognize an epitope situated within the hormone-binding domain of the ER protein, whereas ID-5 reveals an epitope toward the N-terminal of the receptor protein proximal to the DNA-binding domain (24). In paraffin-embedded breast tumor sections this antiserum is reported to be more sensitive than H222 (25, 26). Our assay for ID-5 requires coverslips to be fixed by the same procedures as those reported for ER ICA: sequentially using formaldehyde (4%, v/v) in PBS (12 min at room temperature), PBS (5 min), cold (-20 C) methanol (5 min) and cold (-20 C) acetone (3 min). After two 5-min PBS washes, coverslips were incubated with 10% normal goat serum for 10 min to inhibit nonspecific binding of antibody. Excess serum was removed, and ID-5 at 1:100 in PBS was added for 1 h at room temperature. Three PBS washes (of 4 min each) were followed by a 30-min incubation with secondary goat antimouse antibody (Z0420, DAKO Corp.) diluted 1:25 in PBS. After further PBS washes, a mouse peroxidase-antiperoxidase-conjugated antibody (P0850, DAKO Corp.) diluted 1:250 in PBS was added for 30 min. Staining was revealed using diaminobenzidine tetrahydrochloride (DAB) and hydrogen peroxide (H2O2), with methyl green as counterstain.
Progesterone receptor (PR) staining (PR ICA). The assay for PR staining was performed using the Abbott monoclonal kit (27). Apart from substitution of the rat antihuman PR primary antiserum, the procedures were the same as those for ER ICA (22, 23).
pS2 staining. After fixation as described for ER ICA, coverslips were washed twice in PBS (5 min), and 10% normal goat serum was added (10 min) to block nonspecific attachment of primary antiserum. Excess was removed, commercially obtained, prediluted primary mouse antihuman pS2 monoclonal antibody [CIS (UK) Ltd., High Wycombe, UK] was added, and coverslips were incubated overnight (17 h) in a humidified chamber. After extensive (three 5-min) PBS washes, secondary goat antimouse IgG diluted 1:25 in PBS (DAKO Corp.) was added for 30 min. Coverslips were then washed again, and mouse peroxidase-antiperoxidase-conjugated tertiary antiserum (DAKO Corp.) at 1:250 in PBS was added for an additional 30 min and then also washed off with PBS. The chromogen-substrate mixture, DAB-H2O2 (Abbott Laboratories), was applied for 10 min, and coverslips were then washed with distilled water. After counterstaining with 1% aqueous methyl green, coverslips were dehydrated and mounted inverted on glass slides with a xylene-soluble mountant.
EgfR staining (18, 28). Cell-bearing coverslips were fixed by initially air-drying them for 30 min at room temperature (at which point they were stored at -80 C). Immediately before assay they were further fixed by immersion in acetone/chloroform (1:1) at 4 C for 10 min and allowed to air-dry for 5 min before assay. Coverslips were then washed with PBS (four times, 2 min each time), and primary mouse monoclonal antihuman EgfR antiserum (NeoMarkers Ab-10, Stratech Scientific Ltd., Luton, UK) was added, diluted 1:150 in 1% BSA in PBS. Coverslips were incubated in primary antibody overnight (17 h) at 4 C. Coverslips were then washed (three times, 4 min each time) with PBS, and a secondary antimouse antiserum (SuperSensitive Monoclonal Link, BioGenex Laboratories, Inc., San Ramon, CA) was added, diluted 1:65 in 1% BSA in PBS for 20 min at room temperature. A tertiary peroxidase-conjugated antibody (SuperSensitive Monoclonal Label, BioGenex Laboratories, Inc.) was then added, also diluted 1:65 in 1% BSA in PBS for 20 min at room temperature. After additional washes, antigen localization was revealed using DAB/H2O2 as before. Coverslips were washed with distilled water and methyl green counterstained before dehydration and mounting in a xylene-soluble mountant.
Fully activated (dually phosphorylated) mitogen-activated protein kinase (MAPK)-ICA for coverslips. Coverslips were fixed by the ER-ICA method. After a PBS wash, 20% normal human serum in PBS was added for 15 min to block nonspecific attachment of antigen. Excess was removed, and primary rabbit antihuman dually phosphorylated extracellular signal-regulated kinase 1 (erk1) and erk2 MAPK antibody (Promega Corp., Southampton, UK) diluted 1:100 in 5% normal human serum in PBS applied overnight in a humidified environment. Coverslips were then washed three times for 4 min each time with PBS, and secondary peroxidase antibody complex was added (A4914, goat antirabbit peroxidase conjugate, Sigma, Poole, UK) 1:40 in 5% normal human serum in PBS for 30 min. After further PBS washes, antigen localization was revealed by application of DAB/H2O2 substrate solution to coverslips for 10 min and, after a distilled water wash, was enhanced by incubation with 2% aqueous copper sulfate solution for 2 min. Counterstaining was performed with 1% aqueous methyl green.
Analysis of staining. Staining for all assays was assessed
by two of the authors (J.M.W.G. and R.A.M.), and estimates of
percentages of cells specifically stained and, where appropriate, of
staining intensity were recorded. An H-score (range, 0–300) was
calculated as previously reported (27); an H-score of 300
would describe strong staining of all (100%) tumor cells. In addition
to the 3 staining intensity categories (weak, moderate, and strong)
previously used, a category of very weak staining was incorporated into
the H-score calculation as shown below: H-score = 
(% very
weakly stained cells x 0.5) + (% weakly stained cells x 1)
+ (%moderately stained cells x 2) + (% strongly stained
cells x 3).
RT-PCR
Total RNA was isolated from cells cultured under the conditions
described for 72 h and then lysed using guanidinium isothiocyanate
as previously described (29). RT was performed by standard
methods using random hexamer primers. Twenty-five cycles of
simultaneous PCR amplification, using ER, PR, or EgfR primers, was
performed in combination with primers for actin. Primer sequences for
ER, PR, and actin and the optimized PCR conditions were described by
Knowlden et al. (29). Primer sequences for EgfR
derived from those of Ullrich et al. (30) are
given below. Optimized conditions proved to be the same as those for
ER-PCR. Products were revealed by horizontal electrophoresis on a 3%
(wt/vol) agarose gel, visualized using ethidium bromide under UV, and
the resultant photographic image was scanned and quantified by
densitometry (GS690 densitometer with Molecular Analyst software,
Bio-Rad Laboratories, Inc., Hercules, CA). The EgfR primer
sequences are: forward, 5'-AGC CAT GCC CGC ATT AGC TC-3'; and reverse,
5'-AAA GGA ATG CAA CTT CCC AA-3'.
Transient transfection and reporter gene product assay
The lipid-mediated gene transfection procedure (lipofection) was
previously described (31). Cells in log phase growth were
seeded at 250,000 cells/well of a 12-well plate in phenol-red free RPMI
medium with 5% steroid-stripped FCS, penicillin/streptomycin, and
ampicillin. After 24 h, cells were washed with PBS, and medium was
replaced with 0.5 ml/well of the lipofection mixture as described below
diluted in steroid-free phenol red-free DCCM-1 tissue culture medium
(Biological Industries, Kibbutz Beit Haemek, Israel)
supplemented only with glutamine.
The lipofection mixture per well was: 1) 3 µl of the transfection vehicle Lipofectin (Life Technologies, Inc.), a 1:1 (w/w) liposome preparation of the cationic lipid N-[1-(2,3-dioleyloxy)propyl]-n,n,n-triethylammonium chloride, and dioleoyl phosphotidylethanolamine previously equilibrated in the medium; 2) 1 µg ERE-firefly luciferase reporter plasmid DNA; 3) 400 ng Renilla luciferase plasmid DNA; and 4) 600 ng pcDNA3 carrier plasmid DNA. After 6-h incubation at 37 C the lipofection medium was removed from the cells and replaced with DCCM-1 containing 1% stripped FCS and glutamine and appropriate treatments as described. Twenty-four hours later, cells were washed with cold PBS and lysed by the addition, on ice, of 200 µl passive lysis buffer (Promega Corp.). The resultant lysate was scraped into sterile 1.5-ml microcentrifuge tubes, vortexed, and stored at -80 C until luciferase assay.
The luciferase reporter constructs used were: 1) a modified pGL2-firefly luciferase vector bearing a thymidine kinase (tk) promoter and single ERE sequence (gift from Prof. Malcolm Parker, Imperial Cancer Research Fund); and 2) a modified pGL2-Renilla luciferase vector with same tk promoter (Promega Corp.) for an internal control, enabling levels of nonspecific promoter trans-activation to be determined.
Luciferase assay
This was performed using the dual luciferase reporter assay
system (Promega Corp.). Kit components were prepared
according to manufacturer’s instructions. One hundred microliters of
lysate were added to 100 µl Luciferase Assay Reagent II and gently
mixed, and firefly luciferase activity was measured for 10 sec using a
Lumat LB 9507 luminometer (E.G.&G Wallac Ltd., Milton
Keynes, UK). This reaction was terminated, and Renilla
luciferase activity was measured by the addition of 100 µl Stop &
Glo reagent (Promega UK, Southampton, UK). The effects of
treatment on transcription from the ERE-bearing firefly luciferase
reporters were determined after normalization for nonspecific
transcriptional activity based on constitutive Renilla
luciferase transcription were made. Estimates of the percentages of
cells transfected in each experiment were made by independent
transfection of a constitutive ß-galactosidase expression vector (a
gift from Dr. Gavin Wilkinson, Department of Medicine, University of
Wales College of Medicine, Cardiff, UK) into the cells by an identical
procedure as that described above. Subsequent histochemical staining
using
5-bromo-4-chloro-3indolyl-ß-D-galactopyranoside
(X-gal) allowed counts of transfected cells to be made. Briefly cells
were fixed in situ with 0.5% glutaraldehyde in PBS for 15
min, washed, and stained overnight at 37 C with X-gal in PBS
(125 µg/ml).
Statistical analysis
The two-way ANOVA test was used to assess the relative effects
of treatment on the overall growth or transcriptional activation of
cell lines. Direct comparisons of individual dose effects were
conducted using Student’s t test with Bonferroni
adjustments as appropriate. Significance in both analyses were assumed
at P < 0.05.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
shows
comparative phase contrast photomicrographs of the wild-type and FASMCF
cells in logarithmic phase growth in their respective basal medium. The
mean comparative growth curves of the wild-type and FASMCF cells
cultured in the presence of the antiestrogen are portrayed in Fig. 2 and illustrate the sustained elevated
growth of FASMCF. This basal FASMCF growth rate was unaltered by
coaddition of tamoxifen
(10-7 M) or its replacement of ICI 182,780 in the basal medium throughout equivalent growth curves. Dose-response experiments
further confirmed the capacity of FASMCFs to grow in the presence of
ICI 182,780. Thus, Fig. 3 illustrates a
typical example of these, confirming the relative insensitivity of
FASMCF cells to even
10-5-M
doses of the antiestrogen, whereas wild-type cells showed significantly
greater sensitivity to ICI 182,780 (P < 0.001).
Indeed, growth inhibition of wild-type cells was notable even at
10-9 M, where 50% growth inhibition on day 7
vs. that of controls was observed.
|
|
|
illustrates the
effects of estradiol on FASMCF growth while either maintaining ICI
182,780 or withdrawing cells from it. Thus, although estradiol had no
mitogenic effect when ICI 182,780 was maintained, its removal prompted
an increase in the cell growth rate. This estradiol-induced growth
appeared to be reversible upon subsequent readdition of ICI
182,780.
|
were subsequently repeated and gave comparable results (data not
shown).
Comparison of ER-mediated signaling between wild-type and FASMCF
cells
We previously demonstrated the sensitivity of our wild-type
parental MCF7 cells to both estrogen and antiestrogens and their
expression of steroid hormone receptors (6, 7). Optimal
stimulation of wild-type cell growth by 17ß-estradiol was achieved
with a dose of 10-9 M, whereas optimal specific growth inhibition was obtained
using hydroxytamoxifen or ICI182,780 at
10-7 M. The
relative expression levels of estrogen receptor messenger RNA (mRNA)
for wild-type and FASMCF cells after RT-PCR amplification are shown in
Table 1. Using a semiquantitative
densitometry-based assessment of ethidium bromide-stained products
isolated by electrophoresis on a 2% agarose gel and normalizing for
PCR-coamplified actin mRNA expression levels, wild-type cells growing
in basal medium were found to express markedly higher (6.6-fold)
ER mRNA than FASMCF.
|
, A and B. Thus, for wild-type cells significant levels of both mRNA
and protein were evident in basal (antiestrogen-free) conditions.
Immunocytochemistry using the H222-ER antiserum (Table 2A) revealed the presence of ER protein
in 80% of cells, with a semiquantitative H-score value, based on both
staining intensity and percentage of cells stained, of 82 (on a 0–300
scale). ER was markedly suppressed by ICI 182,780 treatment and was
down-regulated in response to estradiol (H-scores of 2 for both
treatments). Basal levels of ER protein in FASMCF cells growing in the
presence of ICI 182,780 were also low (H-score = 2), with only 2%
of cells being weakly or very weakly stained using the H222 antiserum.
Cellular expression of this was relatively unaffected by 7-day
withdrawal of ICI 182,780 (H-score = 4) or its replacement by
estradiol in the FASMCF basal medium (H-score = 2).
|
B). Using this technique, wild-type cell basal ER levels
were reported as an H-score of 146, and the percentage of cells stained
was 98%. Faslodex and estradiol treatments again were shown to very
strongly suppress ER (H-scores = 31 and 48, respectively, with
40–50% of the cells staining). In basal medium FASMCF cells were shown to express lower levels of ER protein than wild-type cells in basal medium. Thus, although most cells were seen to express some ER (82%), much of this was only at the very lowest staining category (H-score = 71). As with H222, little change in ER was noted after the removal of antiestrogen (H-score = 70) or its replacement with estradiol (H-score = 75). These ER levels appeared somewhat elevated relative to levels in ICI 182,780- or estradiol-treated wild-type cells, a feature that was not apparent using H222.
Under basal conditions wild-type cells expressed low, but detectable,
levels of PR mRNA (Table 1) and protein (Table 2C), with only 2% of
cell nuclei staining weakly (H-score = 2). Protein expression was
markedly induced by treatment of cells for 7 days with estradiol, with
all cells expressing PR protein, many at the strongest intensity
(H-score = 160). Addition of ICI 182,780 to the wild-type basal
medium resulted in no immunodetection of PR protein.
No PCR-amplified PR mRNA or protein was found in FASMCF cells cultured in the presence of ICI 182,780. Furthermore, no PR protein was detected after removal of antiestrogen for 7 days. Interestingly and in marked contrast to wild-type cells, replacement of ICI 182,780 by estradiol for 7 days also failed to induce any significant increase in PR levels (H-score = 2).
An immunocytochemical analysis of the estrogen-inducible gene product
pS2 was also undertaken. The promoter for this gene is complex and
contains both steroid hormone and growth factor response elements
(32). Table 2D illustrates that pS2 expression in
wild-type cells grown in basal medium was high (H-score = 105),
but was markedly suppressed after exposure to the pure antiestrogen
(H-score = 45) and was induced by estradiol treatment
(H-score = 185).
The basal pS2 staining intensity of FASMCF cells maintained in ICI 182,780 was less intense than that in wild-type cells (H-score = 80), although it was present in a similar proportion of cells (65% vs. 70%) Staining was unaltered by removal of ICI 182,780 for 7 days. Replacement of ICI 182,780 by estradiol in the culture medium for 7 days resulted in an elevation in staining intensity similar to that witnessed with wild-type cells (H-score = 180).
The potential for estrogen response in the FASMCF cells
maintained in ICI 182,780 was further addressed through comparing
transient transfections of an ERE-bearing reporter gene plasmid
construct into the wild-type and FASMCF cell lines. Treatment effects
on the transcription mediated via the ERE were determined by assay for
luciferase expression. Similar levels of transfection were observed
with the two cell lines, with approximately 40% of cells transfected
with a ß-galactosidase expression vector in all experiments. After
normalization of data for equivalence in nonspecific promoter
activation using a constitutive Renilla luciferase control
expression vector (Fig. 5), levels of
ERE-mediated firefly luciferase activity were calculated. Similar low
levels of firefly luciferase expression were observed between the cell
lines maintained in the presence of ICI 182,780. The data illustrated
in Fig. 5 show relative changes to these levels of expression after
treatment. Slight and equivalent increases were observed if medium were
further supplemented with estradiol for 24 h. Removal of ICI
182,780 led to rapid reporter transcription in wild-type cells
(20-fold vs. ICI 182,780-treated controls), but
significantly less up-regulation in FASMCFs (3-fold). Removal of the
antiestrogen and supplementation with estrogen induced transcription
approximately 10-fold in FASMCFs and 30-fold in wild-type cells. The
results suggest the presence of functional wild-type ER protein in both
sublines, which can, upon removal of ICI 182,780, mediate a rapid
response to estradiol, although this is less marked in the FASMCF.
ER-ERE signaling, however, is very effectively suppressed by ICI
182,780 and is unlikely to be the significant growth regulatory
pathway in FASMCF cells while they are maintained in the
antiestrogen.
|
). Immunostaining of both cytoplasmic and
membrane-specific EgfR was also very low in wild-type cells grown in
basal medium. Table 2E illustrates the results of membrane-associated
EgfR staining, which we have previously shown to better reflect the
endocrine-insensitive phenotype than cytoplasmic staining. Thus, using
a highly sensitive antiserum, less than 20% of cells were deemed
positive (H-score = 45). Addition of ICI 182,780 for 7 days had no
appreciable effect on wild-type MCF7 EgfR expression. In contrast, EgfR
mRNA levels of FASMCF cells cultured in basal medium and harvested 7
days after subculture were approximately 9-fold higher than those in
wild-type cells as determined using a semiquantitative
densitometry-based estimation of ethidium bromide-stained
RT-PCR-amplified products (Table 1). Similarly, ICA-determined protein
levels were much higher in FASMCF cells, both in those maintained in
ICI 182,780 (80%; H-score = 125) and where the antiestrogen had
been removed for 7 days (H-score = 125). Figure 6 illustrates the rapid induction and
sustained elevation of EgfR mRNA levels expressed by FASMCF
vs. wild-type cells as determined by RT-PCR of material
extracted from cells cultured in basal medium throughout a short time
course.
|
(Tgf; 10 ng/ml). When
administered to our wild-type MCF7 cells Tgf is not a significant
mitogen and frequently produces no elevation of growth above basal
levels (6, 7). However, the growth of FASMCF cells was
often, but not inevitably, enhanced by supplementation of the medium
with Tgf, although effects were modest compared, for example, to the
mitogenicity of estradiol given to wild-type cells. Confirmation that
this effect was predominantly mediated via the elevated levels of EgfR
protein was obtained by treating FASMCF and wild-type cells with an
EgfR-tk-specific inhibitor, TKI, ZD1839 (20, 21) (Fig. 7). The dose-response curves show TKI to
have little inhibitory effect on the growth of wild-type cells across a
range of doses, whereas in contrast, marked growth inhibition was
observed (P < 0.001) with FASMCF cells, where a
significant reduction in cell numbers was observed even with a
1-µM concentration (P =
0.001).
|
F). Staining specificity was confirmed after its reduction in
FASMCF cells following their growth in the presence of a specific
inhibitor of MAP kinase kinase (MEK1, PD098059, 50
µM), a compound shown previously to inhibit the
phosphorylation and activation of erk1 and erk2 (33) (data
not presented). Importantly, PD098059 was also found to be a highly
effective inhibitor of the growth of FASMCF cells (Fig. 8), producing a virtual arrest of cell
proliferation over a 14-day period.
|
To further determine the significance of the EgfR pathway in FASMCF cells, we investigated the long-term growth effects of addition of TKI to the maintenance medium. Growth was rapidly and markedly inhibited, and significantly, no regrowth of the cells, indicating the development of TKI resistance, was observed after 3 months, a period longer than that required for the original ICI 182,780 resistance to arise.
We also administered a combination of ICI 182,780 (10-7 M) and TKI (1 µM) to wild-type MCF7 cells previously unexposed to these drugs. Importantly, throughout the exposure period, which was longer than 6 months, a sustained inhibition of cell growth was observed. Indeed, significant cell loss occurred, and there was no evidence of a resistant phenotype developing. Early attempts to determine whether this observed cell loss is due to cell death or reduced cell adhesion suggest the former. Thus, these cells show increased numbers of early stage apoptotic cells, more late stage apoptotic and necrotic cells, and fewer proliferating cells as witnessed by cell cycle-specific immunomarkers. Further characterization of both of these interesting sublines is now being undertaken.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
ER expression in ICI 182,780- or estradiol-treated wild-type MCF7 parental cells, as determined using the now conventional H222 antiserum, was almost undetectable while basal levels in the absence of antiestrogen were moderate. In FASMCF cells, H222-ER levels were at the lowest levels of detection. Despite these apparently low levels, removal of antiestrogen or addition of estradiol was able to initiate appreciable ERE-mediated gene transcription. The use of an alternative ER antiserum ID5 in a more sensitive immunocytochemical procedure confirmed the presence of low levels of ER protein in the majority of cells in both cell lines. Although results were broadly similar to those of H222 (although staining was stronger with ID5), some anomalous results relative to H222 were observed. In particular, the levels of ER expressed by FASMCF cells, although low and unaffected by removal of ICI 182,780 or addition of estradiol, are somewhat higher than those expressed by the Faslodex- or estradiol-suppressed wild-type cells, which was not apparent after the H222 assay. In regard to this it is interesting to note that the two antisera reportedly recognize different epitopes on the wild-type ER molecule. H222 is believed to localize a region proximal to the ligand-binding domain, whereas the recognition site for ID5 antiserum is nearer the DNA-binding domain (24).
A number of studies have questioned whether the acquisition of antiestrogen resistance is associated with an increase in significant quantities of mutant or splice variant species of the ER that are capable of ligand-independent gene trans-activation (34, 35). Therefore, we considered whether the ID5 antiserum was detecting a hormone-binding domain-deficient species of ER protein in the FASMCF cells such as would arise, for example, from an exon 5 deletion variant (36). A preliminary PCR analysis of ER mRNA expression using primers capable of differentiating wild-type and exon 5 deletion-derived forms showed only low levels of the truncated mRNA to be transcribed in FASMCF cells relative to wild-type ER mRNA transcripts. These two forms were also found in parental MCF7 cells and in similar proportions (data not shown).
As the increased staining levels observed using ID5 relative to those seen with H222 were common to both wild-type and FASMCF cells (and are also very frequently observed in comparably stained clinical specimens), it is most likely that the differences in detected ER are due to differences in the sensitivity of the two assay procedures or in antigen preservation. It is feasible, for example, that the two differing epitopes recognized by the primary antisera are differentially sensitive to the formalin-based procedure, rendering some sites more susceptible to degradation (e.g. H222 epitope) and thus reduced staining. Interestingly, short-term ICI182,780 or estradiol given to wild-type cells seems, by ID5-ICA at least, to suppress ER to levels below the basal FASMCF levels. The reasons for this, and, indeed, establishment that these slight and as yet only immunocytochemically determined differences are of biological significance are currently unresolved.
Basal PR levels in both wild-type and FASMCF cells were very low, but although wild-type cell PR was markedly induced by estradiol, only a minimal effect was observed with FASMCF cells. In contrast, significant amounts of pS2 protein, another estrogen-inducible gene product in wild-type cells, were detected in the basal medium of both wild-type and FASMCF cells, and levels were similarly elevated by estradiol treatment. This apparent dichotomy in response of two estrogen-responsive genes in FASMCF cells has a number of possible explanations. Firstly, it is feasible that the transcriptional regulation of PR by estradiol has become irreversibly altered in these cells or that pS2 trans-activation may be more sensitive and more rapidly inducible than PR. Badia et al. recently demonstrated the loss of PR gene expression in long-term hydroxytamoxifen-treated MCF7 cells was associated with the disappearance of deoxyribonuclease 1-hypersensitive sites (37).
Alternatively, as basal pS2 activity is apparent at relatively high levels in FASMCF cells despite the continued presence of ICI 182,780, it is possible that pS2 transcription is not exclusively or even predominantly ERE driven, but uses different transcription factor-activated response elements within the promoter of the gene. It is well established that the pS2 promoter comprises a complex series of elements, including a growth factor-activated activator protein-1 (AP-1) recognition sequence (AP-1-RE), thus presenting the possibility of EgfR-driven pS2 expression in the resistant cells (32). Indeed, we have noted that treatment of wild-type MCF7 cells with the TKI, although not affecting PR protein, causes some down-regulation of pS2 staining (data not presented). Furthermore, we have previously shown (38) that the AP-1-RE-inducing phorbol ester TPA is capable of causing elevation of pS2 mRNA expression in wild-type MCF7 cells, but had a down-regulatory effect on other estrogen-regulated genes (e.g. PR, pLIV1). Interestingly, Paech et al. suggest that antiestrogen-ERß complexes can bind to AP-1 protein and activate estrogen-responsive genes at AP-1-RE sites (39). It is important to note that the addition of estradiol to the FASMCF cells induced pS2 protein. Whether this activation was via ERE-mediated trans-activation of the pS2 gene or interactions between the steroid and AP-1 is unknown; however, we observed only minor estradiol-induced transcription from a transfected ERE-reporter while FASMCF cells were maintained in ICI182,780.
A number of studies have investigated the effects of estrogen withdrawal on steroid hormone signaling and cellular proliferation in MCF7 cells, and a number have created sublines capable of growth in estrogen-deprived conditions (40, 41, 42). Similar studies have been performed using other, differentially estrogen-responsive breast cancer cells, e.g. ZR 75–1 and T-47-D, with often contrasting results (43, 44). In general, long-term estrogen deprivation of MCF7 cells through culture in steroid-stripped, phenol red-free medium causes an initial growth inhibition followed by a resumption at an often accelerated basal rate some 1–3 months later. Cells growth frequently appears unresponsive to the readdition of estradiol, but proliferation is usually inhibited by antiestrogens. In some cases the sublines generated by long-term estrogen deprivation have shown elevated levels of ER and low basal PR that is markedly elevated by addition of estradiol. Much of these data contrast with the phenotype we describe here (e.g. retained estrogen inducibility, very low ER, noninducible PR) and are possibly the result of combined long-term use of a pure antiestrogen and steroid-depleted medium to achieve essentially estrogen-free conditions.
Our most significant observation with FASMCF cells is that in their
adaptation to growth in the steroid-deprived environment and the
resultant reduction of effective ERE-signaling mechanisms, they appear
to have acquired increased expression of a number of components
involved in the EgfR/MAPK signaling pathway, a phenomenon we have also
recognized in a newly developed tamoxifen-resistant cell subline
derived from the same parental MCF7 cells (unpublished observations).
Thus, FASMCF cells show elevated levels of both EgfR protein and mRNA
vs. wild type cells and some increased sensitivity to the
growth promotory properties of the EgfR-ligand Tgf. Importantly, and
unlike wild-type cells, FASMCFs are also strongly growth inhibited by
physiological doses of an EgfR-specific tyrosine kinase inhibitor.
Furthermore, the cells show elevated levels of immunolocalized
phosphorylated MAP kinases ERK1/ERK2, and their growth is more
effectively inhibited by treatment with a specific inhibitor of the
upstream kinase, MEK1. Surprisingly, despite the increased levels of
EgfR and the marked growth inhibition achieved with the TKI and MEK
inhibitor, growth stimulation by exogenous Tgf was inconsistent.
Experiments assessing the roles of other ligands and other members of
the erb family of receptors are ongoing, but as yet no
significant differences between c-erbB2, c-erbB3,
or c-erbB4 mRNA levels of wild-type and FASMCF cells have
been identified by RT-PCR. The possibility that increased autocrine
production of EgfR ligands by FASMCF is stimulating growth while
masking the exogenous treatment experiments is being addressed.
The suggestion that an increased role for the EgfR/MAPK signaling pathway in ER-positive breast tumor cell growth might arise as a consequence of effective suppression of ER-mediated signaling after treatment with antiestrogens has some support in the literature. The work of Yarden and colleagues (45, 46, 47) in particular has demonstrated clearly that EgfR mRNA and protein levels can be elevated in breast cancer cells in vitro by estrogen deprivation or antiestrogen treatments and that the regulation of this is clearly linked with estradiol and the ER. We did not observe a significant change in wild-type cell EgfR protein in response to ICI 182,780, although there was a modest rise in activated erk1/2 MAPKs levels.
A number of previous publications (17, 18) have shown a clear inverse relationship to exist in both clinical and in vitro studies between ER and EgfR expression. Indeed, we ourselves have demonstrated this relationship to exist even at the cellular level (28). Elevated EgfR expression has also been shown in recurrent breast cancer to correlate with poor levels of responsiveness to endocrine therapy and in a poorer prognosis in both time to further relapse and overall survival (17, 18). Furthermore, a number of reports are strongly suggestive of a link between increased growth factor signaling and the acquisition of antiestrogen resistance. The stable transfection of either MCF7 or ZR 75–1 breast cancer cells with the EgfR gene and the subsequent increased expression of EgfR protein (48, 49) or, indeed, of Her-2/neu (50), for example, has been shown to result in a loss in hormone responsiveness of the cells. Dumont et al. (51) and Johnson et al. (52) both reported elevated AP-1 DNA binding in their antiestrogen-resistant MCF7 cells and clinical specimens, respectively. Smith et al. (53) produced a hormone-resistant phenotype in MCF7 cells by overexpressing the AP-1 component c-Jun, whereas deFazio et al. (54) showed that antisense ER RNA inhibition can increase EgfR gene expression in MCF7 cells. More recently, Shim et al. (41) report on an MCF7 cell subline generated through long-term estrogen deprivation, which when transplanted as a xenograft into nude mice exhibits estrogen hypersensitivity and elevated levels of activated MAPK. Coutts and Murphy (55) report a similar activated MAPK elevation in their long-term estrogen-deprived breast cancer subline, whereas Kurokawa et al. (56) found higher levels of MAPK in a stably transfected MCF7 line with forced overexpression of HER2. Interestingly, with these latter cells tamoxifen resistance developed that was reversible by inhibitors of HER2 or MAPK. The cells did not, however, show any alterations in EgfR expression.
Larsen et al. (57) report that neither their
tamoxifen- or pure antiestrogen-resistant cell lines acquired
resistance through an alteration in the expression of EgfR mRNA or
through changes in the expression of other members of the
erbB family. Thus, although their ICI 182,780-resistant cell
lines (57, 58) are similar to our own in a number of
respects (e.g. lowered ER, no PR, retention of estradiol
growth response) they do not appear to be the same; FASMCF cells,
unlike theirs, also appear to be cross-resistant to tamoxifen. Indeed,
in this respect, although not in others, our cells show similarities to
those of Brunner et al. (15). The differences
in the ICI 182,780-resistant cell lines reported to date and the
contrasts with our own are interesting and may result from variations
in the protocols employed to generate them or in the characteristics of
the parental MCF7 cells used. In this light, we have, for example,
previously reported on a lack of overall change in EgfR or Tgf
protein expression between pre- and posttreatment samples from a group
of 21 clinical breast cancer specimens after short-term (7-day) ICI
182,780 treatment. Within this population, however, elevations in EgfR
and Tgf levels were, in fact, observed in single examples of
posttreatment tissue (9).
Subsequent to our studies characterizing the FASMCF cells we continuously coexposed a subcultured flask of these to both ICI 182,780 and TKI in an attempt to induce the development of a dually resistant line. However, the combination of antiestrogen and antigrowth factor resulted in very strong inhibition of cell growth with marked cell loss. Furthermore, regrowth of a secondarily TKI-resistant phenotype had not occurred within 3 months, the time-frame within which the apparent ICI 182,780-resistant phenotype had originally arisen. This may prove to be an important finding and supportive of the concept of using combination therapies for advanced breast cancer involving a pure antiestrogen to fully suppress estrogen-signaling pathways and an anti-EgfR strategy involving a TKI. Indeed, our combination of ICI182,780 and ZD1939 administered to previously unexposed wild-type MCF7 cells over a period in excess of 6 months resulted in sustained inhibition of cell growth and subsequent cell loss and no evidence of the resistant phenotypes developing.
In summary, we report the development of a subline of wild-type MCF7 cells that, after an extended period of exposure to ICI 182,780, demonstrate an increased reliance upon elements of the EgfR/MAPK intracellular signaling pathway and a mirrored decline in ER-ERE-mediated gene transcription. It remains probable that components of other signaling pathways in these cells have also been altered in response to ICI 182,780 exposure and that the identified changes in the growth and phenotype of these cells may in part reflect these changes also. Therefore, an examination of the roles of these alternative pathways and their interactions with ER/ERE and EgfR/MAPK signaling is currently in progress.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received December 15, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and ERß at AP1 sites. Science 277:1508–1510
) and FASMCF (
) cells grown in steroid-depleted medium and in
the presence of 10-7 M ICI
182,780. Data portrayed for each time point represent mean triple cell
counts from triplicate wells from each of three separate experiments.
Results are expressed as a percentage of cell number recorded on day 1.
The mean ± SD are shown.
) and FASMCF
(
) cells exposed to increasing doses of ICI 182,780. The data
portrayed are a typical example of a twice performed experiment in
which mean triplicate cell number counts from triplicate wells were
made for each time point after 7 days of treatment. For illustration,
data are expressed as a percentage of mean cell numbers for day 1. By
overall two-way ANOVA, P < 0.001. *,
P < 0.05, by t test for individual
treatment pairs. Error bars = 1 SD.
, ICI 182,780 withdrawn from day 1, estradiol
(10-9 M) maintained throughout,
ICI 182,780 (10-7 M) readded on
day 8. The data shown are from a typical example of an experiment
performed three times. The mean of three cell counts of triplicate
wells counted for each time point ± SD are shown.
