| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Tumor Biology |
Departments of Radiology [K. E. L., C. M. P., D. P-W.], Molecular Biology and Pharmacology [K. E. L., C. M. P., D. P-W.], and Pathology and Immunology [R. D. S.], Washington University Medical School, St. Louis, Missouri 63110
 |
ABSTRACT |
|---|
 Top ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(IRF9) is an IFN regulatory factor that
mediates signalingby type I IFNs (IFN
and IFNß). After single-step selection of
breastadenocarcinoma cells in paclitaxel, differential display and single
gene analysis demonstrated that transcriptional activation of IRF9 and other
IFN-responsive genes, independent of IFN, corresponded with resistance
to antimicrotubule agents. Transient overexpression of IRF9 reproduced
the drug-resistance phenotype and induced expression of IFN-responsive
genes. However, drug resistance was not induced by overexpression of
Stat1 or Stat2, or treatment with IFN per se. Using a
donor-matched array of cDNA prepared from human tumor and normal tissue
from a variety of organs, we observed overexpression of IRF9 in
approximately one-half of breast and uterine tumors, which indicated
that IRF9 may be important in signaling in these tumor types. These
data identify a novel IFN-independent role for IRF9 in the development
of resistance to antimicrotubule agents in breast tumor cells and may
link downstream mediators of IFN signaling to drug resistance in human
cancers. |
INTRODUCTION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Among antitumor agents directed against microtubules, paclitaxel is the most widely used, with clinical efficacy against a number of malignancies, including breast, ovarian, and lung carcinoma. Paclitaxel has been used extensively in studies to identify proteins and processes involved in cell survival after prolonged mitotic arrest (7 , 8) . Resistance to paclitaxel has been reported to be conferred by overexpression of bcl-2 (9) or p21/cip1 (10) . Conversely, paclitaxel cytotoxicity appears to be enhanced by p53 mutation (11) and expression of bcl-xs (12) . Induction of terminal proliferation arrest (SLP), rather than apoptosis, also has been shown to be positively regulated by p21 and p53 (13) . Overall, these studies have focused on mechanisms conferring cell survival as a measure of drug resistance. However, less is known about the proliferative capacity (clonogenicity) of surviving cells, a feature of drug resistance that may have greater clinical significance (14) .
In the present study, we investigated mechanisms of drug resistance in
sublines of MCF-7 breast tumor cells derived by single-step selection
with paclitaxel and clonogenic expansion. Using differential display, a
paclitaxel-resistant cell line was shown to constitutively overexpress
several proteins regulated by type I IFN in the absence of IFN. At the
center of this pathway was the IFN regulatory factor IRF9/p48/ISGF3
(IRF9), which we show can function independently of IFN as a positive
regulator of resistance to antimicrotubule agents.
|
MATERIALS AND METHODS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Reagents.
Paclitaxel, vinblastine, and doxorubicin were obtained from Sigma
Chemical Co. (St. Louis, MO), and stocks were prepared in DMSO. A stock
solution of GF120918 (gift of Glaxo-Wellcome, Research Triangle Park,
NC) was prepared in DMSO. IFN, derived from human leukocytes, was
obtained from Sigma Chemical Co. Plasmid pCMV-Ad-IRF9 was a gift from
D. E. Levy, New York University, New York, NY, and plasmids
RcCMV-Stat1 p91 and pcDNA3-Stat2 were gifts from J. Darnell,
Rockefeller University, New York, NY.
Fluctuation Analysis.
A fluctuation analysis-style selection of MCF-7 variants was performed
essentially as described previously (16)
. Cells were
expanded from 2000 to an average of 1.9 x 106 in 15 independent populations. Each
population was exposed to 72 nM paclitaxel for 7 days.
After 3 weeks of recovery in medium without paclitaxel, colonies were
counted.1 The variation rate (or mutation rate),
a, was calculated by the method of means, according the
following equation:
 |
Cytotoxicity Assays.
MCF-7 cells and variants from nonconfluent flasks were trypsinized and
passed through a 22-gauge needle prior to counting. Cells (5000/well)
were plated in 96-well microtiter plates and permitted to adhere. After
24 h, drugs were added to the medium at the indicated final
concentration (drug vehicle, 0.1% DMSO). Cell protein was quantified
72 h after drug addition with sulforhodamine B.
IC50 values were calculated by computer fit of
the data using SigmaPlot (19)
.
Clonogenic Assays.
MCF-7 cells and variants were seeded on day 0. After 24 h, drug
was added to the medium at the indicated final concentrations (day 1).
Medium containing drug was replaced on days 3 and 5. On day 8, cells
were washed with PBS and fed with fresh medium without drug. Medium was
replaced every 72–96 h for the remainder of the experiment. Cells were
monitored daily by light microscopy for small colonies of proliferating
cells. No evidence of proliferation was observed in MCF-7 cells or any
of the sublines tested until 
10–14 days after removal of
paclitaxel. Colonies were counted 3–4 weeks after the removal of drug
by staining with Coomassie Brilliant Blue. Cell clusters 
50 cells
were defined as colonies.
For clonogenic assays in the presence of IFN, MCF-7 (7.5 x 106/T162) and 11C9 cells (1 x 106/T75) were seeded in the presence of
0, 1, 10, or 100 units/ml human IFN and maintained in IFN
throughout paclitaxel selection and recovery. Paclitaxel (100
nM) was added to the culture medium 24 h after
plating. Selection and colony counts were performed as above.
Clonogenic assays using transiently transfected cells were performed as above with the following exceptions. Cells (6 x 105/10-cm dish or 2 x 105/6-cm dish) were seeded and transfected after 24 h with 6 µg of pCMV-Ad-IRF9, RcCMV-Stat1 p91, pcDNA3-Stat2, or pcDNA6-V5-His-LacZ using FuGene-6 (Roche Molecular Biochemicals, Indianapolis, IN) according to manufacturer’s instructions. Transfection efficiency was 25–30% under these conditions as measured by transfection of cells with pcDNA6-V5-His-lacZ (Invitrogen, Carlsbad, CA) and staining with X-gal (5-bromo-4-chloro-3-indolyl ß-D-galactopyranoside). Paclitaxel (100 nM) or vinblastine (1 µM) were added 48 h after transfection.
Western Blots.
Western blots were prepared with whole cell lysates of subconfluent
cultures of cells, with the exception of Western blots for Pgp and
MRP1, which were performed using enriched membrane preparations as
described previously (20
, 21)
. Mouse monoclonal antibodies
against IRF9, Bcl-2, Bcl-x, MCH-3/caspase-7, CPP32/caspase-3,
Kip-1/p27, and Cip-1/Waf-1/p21 were obtained from BD Transduction
Laboratories, San Diego, CA. Rabbit polyclonal antibodies against IRF-1
and Stat2 (C-20) and mouse monoclonal antibodies against Stat1 p84/p91
(C-136) and p53 were obtained from Santa Cruz Biotechnology, Inc.,
Santa Cruz, CA. Mouse monoclonal antibodies against Pgp (C219) and MRP1
(QCLR-1) were obtained from Signet Laboratories, Dedham, MA. Binding of
primary antibodies against IRF9, Stat1, Stat2, and IRF1 was visualized
using antirabbit or antimouse IgG conjugated to alkaline phosphatase
(Santa Cruz Biotechnology) with nitro blue tetrazolium (NBT) and
5-bromo-4-chloro-3-indolyl phosphate (BCIP). Binding of all other
antibodies was visualized using antimouse IgG conjugated to horseradish
peroxidase (Amersham Pharmacia Biotech, Piscataway, NJ) and enhanced
chemiluminescence (Amersham Pharmacia Biotech).
Cellular Accumulation of Radiotracers.
Transport activity and modulation of MDR1 Pgp were assayed
with 99mTc-Sestamibi as described previously
(22)
. 99mTc-Sestamibi was prepared
with a one-step kit formulation (Cardiolite; DuPont Medical Products
Division, Wilmington, DE) as described previously (22)
.
Cells were seeded on 20-mm coverslips at a density of 1 x 106 per 10-cm dish and were allowed to
adhere for 24 h. Experiments were performed by equilibrating
coverslips with cells for 30 min in modified Earles’ balanced salt
solution transport buffer containing
99mTc-Sestamibi (10 pM; 7.5
pmol/mCi) in the presence or absence of GF120918 (300
nM). This concentration of GF120918 is sufficient
to maximally inhibit MDR1 Pgp in a number of cell lines
(23)
. Preparations were then washed, extracted, and
analyzed for protein content and activity (22)
. Data
are reported as fmol Tc-Sestamibi (mg protein)-1
(nMo)-1,
where nMo represents
total concentration of Tc-Sestamibi in the extracellular buffer.
Assays of [3H]paclitaxel accumulation were performed as above with the following exceptions. Coverslips were incubated in transport buffer containing 100 nM [3H]paclitaxel (Moravek Biochemicals, Brea, CA; 10 Ci/mmol) in 0.1% ethanol (final) or vehicle alone for various times. For quantitation of [3H]paclitaxel, cell lysates were added to scintillation mixture (Redi-Solv; Beckman Coulter, Inc., Fullerton, CA).
Flow Cytometry.
For 96-h time course studies of cell cycle arrest mediated by
paclitaxel, 600,000 cells per 10-cm dish were plated and allowed to
adhere for 24 h before the addition of 100 nM
paclitaxel. At the indicated times, medium was removed and retained,
adherent cells were trypsinized, and the suspended and adherent cells
were combined and washed with PBS and then fixed with 70% ethanol
prior to staining with PI. For each time point, 10,000 events were
recorded using a FacsCalibur flow cytometer (Becton Dickinson
Immunocytometry Systems, San Diego, CA).
Long-term response to paclitaxel was measured using subconfluent monolayers in T162 flasks treated with 100 nM paclitaxel. Medium containing drug was added on day 1 and replaced on days 3 and 5. On day 8, the cells were washed with PBS and fed fresh medium without drug. Medium was replaced every 3rd day for the remainder of the experiment. For detection of cells in S phase, BrdUrd (Amersham Pharmacia Biotech) was added to the medium at a final concentration of 10 nM 4 h prior to trypsinization (and 24 h after feeding). Cells were trypsinized at the indicated times, washed with PBS, and fixed with ethanol. Fixed cells were washed and stained with anti-BrdUrd antibody conjugated to FITC (Roche) and PI. For each point, 50,000 events were recorded.
Flow cytometric analysis of MHC class I cell surface expression was
performed according to the manufacturer’s recommendations using
FITC-conjugated anti-HLA-A,B,C (G46–2.6) mouse monoclonal antibody
recognizing MHC class I (BD PharMingen, San Diego, CA). Cells were
seeded and remained untreated or were either treated with 1000 units of
IFN/ml for 72 h or transfected with various plasmids 24 h
after seeding. In all cases, cells were stained 96 h after
seeding. Cells were trypsinized, washed in PBS (pH 7.5) with 1% serum,
and incubated for 20 min on ice in PBS containing 1% serum and
anti-HLA-A,B,C antibody. Cells were then washed in PBS with 1% serum
and analyzed by flow cytometry using a FacsCalibur flow cytometer.
Duplicate labeling reactions were prepared for each dish of cells in
addition to an unstained control. Fold increase in class I expression
was calculated as the ratio of mean fluorescence intensity of
treated cells:untreated parental cells in the same experiment.
IFN CPE Assays.
Assays of protection from viral infection were performed essentially as
described previously (24)
. WISH cells (7500/well) were
seeded in 96-well plates. On the day of seeding, human IFN
(0.1–1000 units/ml) or medium taken from 72 h cultures of MCF-7
or 11C9 cells (maximum concentration, 0.5x) was added to the wells.
After 24 h of incubation, cells were washed in warm PBS (pH 7.5),
and fresh prewarmed medium containing vesicular stomatitis virus (VSV)
was added. After 48–72 h, medium was removed, and live cells were
quantitated by crystal violet staining.
Differential Display.
Total RNA was isolated from MCF-7 cells and 11C9 cells (untreated and
after 7 days of 100 nM paclitaxel + 15-day
recovery in drug-free medium) by modified phenol/chloroform extraction
with Trizol (Life Technologies, Inc.). DNA was removed by DNase
digestion (MessageClean kit; GenHunter, Nashville, KY). Reverse
transcription was performed, and the resulting cDNA was used for
subsequent random primed PCR (RNAimage kits 1–10; GenHunter). PCR was
performed using 240 primer combinations using
[-33P]dATP. Samples were electrophoresed on
polyacrylamide gels (6%), which were dried and subjected to
autoradiography. Reactions producing differences in banding patterns
between MCF-7 and 11C9 samples were repeated, and 21 bands representing
reproduced differences were excised from the dried gels. DNA was
extracted with water from the excised bands and amplified using primers
specific to the originating reaction. The resulting amplified products
were electrophoresed on agarose gels and bands corresponding to the
predicted molecular weight were excised from the gels and ligated into
a PCR cloning plasmid, pNoTAT7 (PrimePCR cloning kit; Eppendorf
Scientific, Westbury, NY). The sequences of cDNA inserts in plasmids
from three colonies per ligation were compared with the GenBank
nonredundant and EST databases using the BLAST algorithm
(25)
.
Differences were first screened by reverse Northern dot blot.
Radiolabeled total cDNA was produced by reverse transcription with
[-32P]dCTP from the original total RNA
preparations for untreated MCF-7 and 11C9 cells (ReversePrime kit;
GenHunter). Duplicate dot blots were prepared from the clones of
interest and hybridized with each labeled cDNA according to the
manufacturer’s recommendations.
Northern blots were prepared by electrophoresing 5 µg of total
RNA in formaldehyde gels and blotting by capillary transfer. Probe DNA
was cut from the plasmid clones using suitable restriction enzymes in
the polylinker of the PCR cloning plasmid, and inserts were gel
purified and labeled with [-32P]dATP
using the random hexamer method (HotPrime kit; GenHunter). Radiolabeled
probes were hybridized with blots according to the manufacturer’s
directions. GenBank accession numbers for human cDNAs are: MxA (M30817), hepatitis C-associated microtubular aggregate protein p44
(SEG_HUMHCAMAP), IFI27 (NM_005532), and KIAA0069 (D31885).
Matched Tumor/Normal Expression Array.
A matched tumor/normal tissue expression array (Clontech Laboratories,
Inc., Palo Alto, CA) containing cDNA prepared from 68 untreated human
tumors with matched normal tissue from each individual was hybridized
according to the manufacturer’s directions with a full-length IRF9
probe labeled with [-32P]dATP as above.
Hybridization signals were quantified using a PhosphorImager (Amersham
Molecular Dynamics) with each spot interrogated by volume analysis and
background subtraction. The blot was stripped of bound radioactivity
according to manufacturer’s directions and hybridized with a ubiquitin
probe (Clontech) labeled with [-32P]dATP as
above. Bound radioactivity was again quantified using the
PhosphorImager. Autoradiographs were exposed for 72 h for IRF9 and
24 h for ubiquitin. Integrated phosphor signals are reported for
IRF9 and ubiquitin; data are presented as the ratio of radioactivity in
tumor:normal tissue blots (mean ± SE) for a given organ
of origin with or without normalization to ubiquitin. Additional
information including pathological characterization of tumors, sources
of cell lines, and preparation of positive and negative controls
appearing on the blot (catalogue number 7840-1, lot 9120890) can be
obtained from Clontech Laboratories, Inc. Detailed data sets are
available as supplemental material.1
|
RESULTS |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Clonal sublines were derived from surviving colonies in each of the 15
populations. Cytotoxicity assays (72 h) with paclitaxel demonstrated a
modest increase in IC50 averaging 2.2-fold
(range, 0.9- to 4.8-fold) in the 53 sublines (data for representative
variants are listed in Table 1
). Representative dose-response curves with paclitaxel for parental
MCF-7 cells and variant 11C9 are shown (Fig. 1)
. Modest cross-resistance to doxorubicin (2- to 5-fold) was observed in
4 (22%) of 18 variants, and cross-resistance to vinblastine was
observed in 2 (33%) of 6 variants tested (Table 1)
. There was no
correlation between resistance to paclitaxel and other agents in
short-term cytotoxicity assays. Interestingly, when resistance was
measured in clonogenic assays, we observed marked resistance to
paclitaxel in 2 of 6 variants (11C9 and 7H8; Table 1
; Fig. 1B
). 11C9, the most resistant variant, exhibited
cross-resistance to vinblastine (Fig. 1C)
, but not to
doxorubicin (<2-fold) in clonogenic assays and was chosen for further
analysis.
|
|
, implying a lack
of altered binding sites. In principle, selection with paclitaxel can
favor tubulin mutations that decrease the inherent stability of
microtubules. However, microtubule instability correlates with
collateral sensitivity to Vinca alkaloids, such as
vinblastine, which bind tubulin monomers (27)
. Because
variant 11C9 was cross-resistant to vinblastine in clonogenic assays,
paclitaxel-resistance in this variant was unlikely to be the result of
changes in tubulin. In addition, analysis of cell cycle response to
paclitaxel (below) indicated that 11C9 cells did not have alterations
in microtubular function, which would be expected to decrease the
effectiveness of paclitaxel in eliciting cell cycle arrest.
The lack of cell accumulation differences for
[3H]paclitaxel also suggested the absence of
MDR1 Pgp. To confirm, Pgp function was determined directly
using 99mTc-Sestamibi, a highly sensitive probe
of Pgp transport activity (28)
. Although modest
differences in tracer accumulation were observed among the variants,
there was no enhancement of radiotracer accumulation with GF120918, a
specific inhibitor of Pgp function (23)
. In addition, the
pattern of drug resistance observed in clonogenic assays with 11C9
cells also ruled out involvement of MDR1 Pgp, which would be
expected to confer resistance to doxorubicin as well as to paclitaxel
and vinblastine (29)
. Pgp and MRP1, a related protein that
also mediates multidrug resistance (30)
, were undetectable
by Western blot in parental MCF-7 cells and those variants tested (Fig. 2)
.
|
for correlation between paclitaxel resistance
and expression of proteins reported to mediate resistance to paclitaxel
(reviewed in Refs. 7
, 8
). Paclitaxel induces
phosphorylation of bcl-2 (31)
, and, in MCF-7 cells, bcl-2
and bcl-xL have been correlated with resistance
to paclitaxel (9
, 12
, 32)
. However, other studies suggest
that apoptotic response to paclitaxel may be independent of bcl-2
levels (33)
or phosphorylation state (34)
.
Although our untreated cell lines exhibited large differences in the
level of bcl-2 expression among the variants, neither bcl-2 nor bcl-x
levels correlated with paclitaxel resistance. Up-regulation of
cyclin-dependent kinase inhibitor p21/cip-1 via overexpression
of ErbB2 has been shown to mediate resistance to paclitaxel in
MDA-MB-435 breast cancer cells (35)
, and overexpression of
p21 has been reported to favor entry into a SLP rather than cell death
(13)
. However, we found no differences among the MCF-7
sublines in expression of p21 or a related cyclin-dependent kinase
inhibitor p27/kip. Although levels of the tumor suppressor p53 differed
somewhat among the MCF-7 variants tested, no correlation was evident
between levels of p53 and resistance to paclitaxel, consistent with
evidence that paclitaxel induces cell death through a p53-independent
pathway in cells arrested in G2-M
(36)
. Western blots confirmed the absence of caspase-3
(37
, 38)
in MCF-7 cells (data not shown) and showed equal
levels of caspase-7 in MCF-7 cells and sublines.
Cell Cycle Response to Paclitaxel.
To determine whether resistance to paclitaxel in clonogenic assays
might arise from a difference in cell cycle response to paclitaxel, we
measured the cell cycle distribution of adherent MCF-7 and 11C9 cells
at intervals throughout a clonogenic assay (Fig. 3)
. Untreated MCF-7 and 11C9 cells (time 0) had similar cell cycle
profiles (Fig. 3A)
. On addition of paclitaxel [100
nM (Fig. 3A
) or 10
nM (data not shown)], cell cycle arrest in MCF-7
and 11C9 was essentially identical over the first 24 h. Prolonged
exposure to paclitaxel (100 nM x 7 days)
and subsequent culture in drug-free medium (Fig. 3
; days 0–28)
also demonstrated little difference between MCF-7 and 11C9 cells in the
2n, S phase, and the 4n populations at all
points (Fig. 3, B–D)
. 11C9 cells were more likely to
endoreduplicate and enter a polyploid state (Fig. 3, E and F)
, suggesting a potential defect in the checkpoint
regulating G1 to S phase transition. In both cell
lines, emergence of a diploid S-phase population and a slight increase
in the fraction of 2n cells was noted concurrent with
microscopic observation of colony formation (days 25 and 28). On the
basis of biochemical and morphological markers, cell death in response
to paclitaxel was necrotic rather than apoptotic in both MCF-7 and 11C9
cells (data not shown), taking place gradually over the several weeks
in drug-free medium. 11C9 cells were resistant to cell death throughout
the recovery phase, having a greater fraction of adherent cells with
intact DNA (2n; Fig. 3G
) as well as more
adherent cells that excluded trypan blue (2- to 3-fold more on day 21;
data not shown). Although the fraction of live diploid cells entering
S-phase is the same for MCF-7 and 11C9 cells throughout the experiment,
the absolute number of live 11C9 cells is considerably greater at the
onset of colony formation, resulting in a greater number of
proliferating diploid 11C9 cells. Overall, these results suggest
resistance to cell death, not increased ability to resume
proliferation, as the main cause of increased clonogenicity of 11C9
cells after exposure to paclitaxel.
|
). RNA was prepared from (a) untreated
asynchronous cultures of each and (b) paclitaxel-treated
cultures of each 15 days after removal of paclitaxel (day 23 of
protocol). The latter samples were taken at a time when cell
proliferation was first observed by light microscopy and cells were
primarily in a 4n state with 10% of 11C9 cells
progressing through abnormal S phase to 8n (as observed in
Fig. 3
and related experiments). Only 21 reproducible differences were
noted between 11C9 and parental MCF-7 cells using 240 primer
combinations (designed to theoretically sample 24,000 transcripts),
which suggested very limited genetic variation between these cell
lines. Of the five differences in gene expression between 11C9 and
MCF-7 cells confirmed by Northern blots (Fig. 4, B–G)
, four
represented overexpression of a transcript in 11C9 cells, and one was
the result of a COOH-terminal truncation of a transcript (predicted
transmembrane protein KIAA0069) through mutation to produce a stop
codon. All of the differences confirmed by Northern blots were equally
evident in asynchronous and paclitaxel-treated cultures (Fig. 4
A and data not shown). Of the four overexpressed
transcripts, three matched entries in GenBank: MxA protein (MxA),
hepatitis C-associated microtubular aggregate protein (p44), and
IFN-inducible p27 (IFI27). Among these proteins only MxA has a known
function, which is to confer cellular resistance to viral infection.
All three, however, have been shown to be inducible by type I IFNs
(40, 41, 42)
.
|
to be overexpressed in 11C9 cells (2.5-fold) relative to
MCF-7 cells. IFN increased expression of class I in both cell lines
to a level 2.5-fold greater than that in untreated 11C9 cells. However,
receptor-mediated activation of type I IFN signaling was ruled out as
the source of paclitaxel resistance in 11C9 cells based on the absence
of IFN in 11C9 culture supernatants (as determined by CPE assay; data
not show) and the inability of IFN to confer paclitaxel resistance
on MCF-7 cells (Fig. 5B)
. Type I IFN signaling was intact in
MCF-7 and 11C9 cells, as evidenced by the ability of IFN to protect
MCF-7 and 11C9 cells from virus-mediated cell death in CPE assays (data
not shown).
|
(IRF9; Ref. 45
). Stat1
and Stat2 are phosphorylated by Janus kinases Jak1 and Tyk2 upon
association of IFN with the type I IFN receptor (IFNAR). IRF9 contains
an N-terminal bipartite nuclear retention signal (46)
, and
is the primary DNA binding protein of ISGF3, recognizing the
IFN-responsive promoter element (ISRE) in type I IFN regulated genes
(47)
. Western blot analysis demonstrated that, relative to
parental MCF-7 cells, 11C9 cells overexpressed IRF9 and Stat1 (both
6–10 fold) as well as Stat2 (2-fold; Fig. 5C
). MCF-7 cells treated
with IFN exhibited a pattern of expression similar to untreated 11C9
cells, whereas 11C9 cells treated with IFN showed a slightly
amplified pattern. In contrast to IRF9, IRF1, a related IFN regulatory
factor (48)
, was not overexpressed under any of these
conditions.
Transfection with IRF9 Reproduces Gene Expression Pattern and Drug
Resistance.
Because IRF9 is the primary DNA binding component of ISGF3
(47)
, overexpression of IRF9 might induce expression of a
subset of IFN-inducible genes, thereby conferring the drug-resistant
phenotype. To test this hypothesis, MCF-7 cells were transfected with
wild-type IRF9. Transient transfection with IRF9 was found to induce
expression of Stat1, Stat2 (Fig. 5C
, Lane 7), and MHC class
I (Fig. 5A)
, but not IRF1 (Fig. 5C)
or IFN
(determined by CPE assay; data not shown). Transient transfection of
MCF-7 cells with IRF9 also produced 13-fold resistance to paclitaxel
(100 nM; Fig. 5D
) and 3.3-fold
resistance to vinblastine (1.0 µM), but no
resistance to doxorubicin, recapitulating the resistance phenotype of
11C9 cells. Transient transfection of MCF-7 cells with Stat1 p91 or
Stat2, did not confer resistance to paclitaxel (Fig. 5D)
,
although these transfections with Stat1 or Stat2 produced 2.6-fold
increases in MHC class I expression (Fig. 5A)
. These
transfection data provide evidence that the multidrug-resistance
phenotype observed in variant 11C9 is linked to IRF9 overexpression.
Overexpression of IRF9 in Breast and Uterine Tumors.
To begin to characterize IRF9 expression in clinical specimens, we
compared IRF9 expression levels in donor-matched pairs of untreated
human tumors and normal organ tissues (Fig. 6)
4
using a commercial tissue blot. As a group, breast
tumors overexpressed IRF9 3-fold relative to donor-matched normal
breast tissue, and a subset of tumors highly overexpressed IRF9 [IRF9
expression: tumor:normal tissue ratio, 2.83 ± 0.53
(mean ± SE); range, 1.27 to 5.64; ratio > 2.0 in six of nine tumors; IRF9 expression normalized to
ubiquitin: tumor:normal tissue ratio, 2.83 ± 0.96
(mean ± SE); range, 0.97–10.48; ratio > 2.0 in three of nine tumors]. Uterine tumors overall also
exhibited significant IRF9 overexpression with a subset highly
overexpressing the factor [IRF9 expression: tumor:normal tissue ratio,
2.61 ± 0.99 (mean ± SE); range,
1.04–5.81; ratio > 2.0 in three of seven tumors; IRF9
expression normalized to ubiquitin: tumor/normal tissue ratio,
2.88 ± 1.09 (mean ± SE); range,
0.7–5.03; ratio > 2.0 in five of seven tumors].
Although breast tumors in this data set were primarily derived from
epithelial cell types (six infiltrating ductal carcinomas, one lobular
carcinoma, one mucinous adenocarcinoma, and one medullary carcinoma),
and the uterine "tumors" represented various abnormal cell types
(three benign tumors, two squamous cell carcinomas, and two
adenocarcinomas), IRF9 overexpression did not segregate into any
specific cell types in the data set. Breast and uterine tumors were the
only tested tumor types to show systematic overexpression of IRF9.
|
|
DISCUSSION |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
did not confer drug resistance,
despite induction of a pattern of gene expression similar to that
produced by overexpression of IRF9. These results demonstrate a new
role for IRF9 in conferring resistance to microtubule-directed
antitumor agents.
IRF9 Overexpression Confers Resistance to Antimicrotubule Agents.
11C9 cells demonstrated overexpression of proteins associated with type
I IFN signaling, including all three subunits of the ISGF3
transcriptional activator (IRF9, Stat1, and Stat2) as well as
IFN-inducible MxA, IFI27, p44, and MHC class I proteins. Because IRF9
is the major DNA-binding protein in the ISGF3 complex
(47)
, we predicted that IRF9 overexpression might activate
IFN-responsive genes through increased occupancy of binding sites in
promoter elements. Enforced overexpression of IRF9 in parental MCF-7
cells indeed produced a pattern of gene expression mimicking that
observed in 11C9 cells and IFN-treated MCF-7 cells. However, in
MCF-7 cells, transfection only with IRF9, but not Stat1 or Stat2,
reproduced the drug-resistance phenotype observed in 11C9 cells. Our
results suggested that IRF9 within untreated 11C9 cells and
IRF9-transfected MCF-7 cells did not produce its effect through the
ISGF3 heterotrimer typical of type I IFN signaling. Most notably,
treatment of MCF-7 cells with IFN at low concentrations produced no
effect on resistance to paclitaxel and, at high concentrations,
produced the expected cytostatic effect (45)
, rather than
inducing paclitaxel resistance.
It is unclear whether IRF9 acts alone or as a complex with other proteins such as Stats to activate transcription of IFN-inducible genes in MCF-7 cells. IRF9 can access the nucleus in the absence of IFN by virtue of an NH2-terminal bipartite nuclear retention signal (46) . Stats 1 and 2 are cytosolic, and nuclear localization of IRF9 is reduced in cells overexpressing Stat2 (46) . Previous studies have shown that overexpression of IRF9 by transfection results in the induction of several IFN-inducible genes (49) , which suggests that IRF9-induced transcription may be mediated by IRF9 alone. However, the binding of IRF9 alone to IFN-responsive promoter elements has been shown to be weak relative to the binding of ISGF3 (50) . Experiments are in progress to determine the mechanism of activation of IFN-inducible genes by IRF9 in MCF-7 cells.
Our data indicate that IFN signaling has been activated in 11C9 cells in the absence of IFN. Several other mechanisms of induction of IFN-responsive genes have been described. Retinoids and interleukin-6 have been reported to induce overexpression of IRF9 (51 , 52) . In addition, decreased expression of IRF2, a repressor of transcription of many IFN-responsive genes, has been shown to result in increased expression of IFN-responsive genes in IRF2-/- mice (53) . Overexpression of IFN-inducible genes also has been reported in confluent and senescent cultured human mammary epithelial (HME) cells (but not normally-proliferating HME cells), again suggesting that signals other than IFN may induce expression of IFN-responsive genes (54) .
Surprisingly, many of the characteristics of the resistance phenotype conferred by IRF9 in MCF-7 cells appear essentially opposite to the typical effects of IFNs. IFNs have primarily been associated with antiproliferative function and initiation of apoptosis (45) . Although many IFN-inducible proteins, including MxA, inhibit cell death from viral infection (55) , their primary action appears to be inhibition of virus replication, rather than prevention of apoptosis or necrosis. Hepatitis C-associated microtubular aggregate protein p44 has potential to contribute to protection from antimicrotubule agents by virtue of its association with microtubules (41) . However, the function of p44 is unknown, as is the function of IFI27. Understanding the mechanism of resistance to paclitaxel conferred by IRF9 will require a detailed examination of the effect of IRF9 overexpression on proteins mediating arrest and cell death in response to paclitaxel, as well as the identification of other IFN-inducible genes up-regulated by IRF9.
IFN-responsive Genes, Drug Resistance, and Breast Cancer.
Using a donor-matched array of cDNA prepared from untreated human tumor
and normal organ tissue from a variety of organs, we observed
overexpression of IRF9 in a subset of breast and uterine tumors. These
data strengthen an emerging pattern of overexpression of IFN-responsive
genes in breast cancer. IRF9, Stat1, hepatitis C-associated
microtubular aggregate protein p44, IFI27, and MHC class I, among
others, have been reported to be overexpressed in approximately
one-half of untreated breast tumors examined using cDNA microarrays
(54
, 56)
. Although cDNA prepared from whole tissue can be
criticized for reflecting a variety of cell types, several studies have
directly identified overexpression of IFN-responsive genes such as
Stat1 and IFI27 in breast tumor cells when examined by
immunohistochemistry (42
, 54
, 57
, 58)
. It is possible that
IRF9 overexpression in tumors in vivo may be associated with
resistance to antimicrotubule agents; however, no outcome data were
available for the samples included in the commercial blot. Direct
correlation of tumor IRF9 levels with outcome in patients treated with
antimicrotubule agents remains to be done to prove whether IRF9
overexpression is associated with treatment failure or recurrence of
disease clinically.
Overall, we conclude that overexpression of IRF9 is involved in regulating downstream IFN-responsive genes in human breast tumor cells independent of IFN. Our data suggest that IRF9 may be associated with drug resistance and has the potential to be a surrogate marker of response.
|
FOOTNOTES |
|---|
1 Supplementary data for this article is available
at Cancer Research Online
(http://cancerres.aacrjournals.org). 
2 Supported by grants from the United States
Department of Energy (DE-FG02-94ER61885) and NIH (P20 CA86251).
K. E. L. was recipient of a 2001 AFLAC Travel Award to present
an abstract of this work at the 92nd Annual Meeting of the AACR.
3 To whom requests for reprints should be
addressed, at Molecular Imaging Center, Mallinckrodt Institute of
Radiology, Washington University Medical School, 510 South Kingshighway
Boulevard, Box 8225, St. Louis, MO 63110. Phone: (314) 362-9356; Fax:
(314) 362-0152; E-mail: piwnica-wormsd{at}mir.wustl.edu
4 The abbreviations used are: SLP, senescence-like
phenotype; BrdUrd, bromodeoxyuridine; Pgp, P-glycoprotein; PI,
propidium iodide; MDR1, multidrug resistance gene-1; MRP1, multidrug
resistance-associated protein-1; CPE, cytopathic effect.
Received 1/23/01. Accepted 7/ 3/01.
|
REFERENCES |
|---|
|
TopABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
interferon and interferon requires transcriptionally active Stat1 protein. J. Virol., 70: 647-650, 1996.[Abstract]
/ß.. Eur. J. Biochem., 224: 877-883, 1994.[Medline]
-inducible gene (p27) on human chromosome 14q32 and its expression in breast carcinoma.. Cancer Res., 53: 4096-4101, 1993.-Interferon-induced transcription of HLA and metallothionein genes containing homologous upstream sequences.. Nature (Lond.), 314: 637-639, 1985.[Medline]
that mediate protein-DNA and protein-protein interactions during transcription factor assembly contribute to DNA-binding specificity. Mol. Cell. Biol., 13: 196-206, 1993. gene using CCAAT/enhancer binding protein-ß (C/EBP-ß).. Biochim. Biophys. Acta, 1492: 163-171, 2000.[Medline]
/ß signaling.. Immunity, 13: 643-655, 2000.[Medline]
This article has been cited by other articles:
|
|
 |
|
  K. E. Luker, M. Gupta, and G. D. Luker Imaging chemokine receptor dimerization with firefly luciferase complementation FASEB J, March 1, 2009; 23(3): 823 - 834. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. C. P. Smith, K. E. Luker, J. R. Garbow, J. L. Prior, E. Jackson, D. Piwnica-Worms, and G. D. Luker CXCR4 Regulates Growth of Both Primary and Metastatic Breast Cancer Cancer Res., December 1, 2004; 64(23): 8604 - 8612. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. R. Bani, M. I. Nicoletti, N. W. Alkharouf, C. Ghilardi, D. Petersen, E. Erba, E. A. Sausville, E. T. Liu, and R. Giavazzi Gene expression correlating with response to paclitaxel in ovarian carcinoma xenografts Mol. Cancer Ther., February 1, 2004; 3(2): 111 - 121. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Klampfer, J. Huang, G. Corner, J. Mariadason, D. Arango, T. Sasazuki, S. Shirasawa, and L. Augenlicht Oncogenic Ki-Ras Inhibits the Expression of Interferon-responsive Genes through Inhibition of STAT1 and STAT2 Expression J. Biol. Chem., November 21, 2003; 278(47): 46278 - 46287. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. D. Luker, V. Sharma, C. M. Pica, J. L. Prior, W. Li, and D. Piwnica-Worms Molecular Imaging of Protein-Protein Interactions: Controlled Expression of p53 and Large T-Antigen Fusion Proteins in Vivo Cancer Res., April 15, 2003; 63(8): 1780 - 1788. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. A. Kraus, J. F. Lau, J.-P. Parisien, and C. M. Horvath A Hybrid IRF9-STAT2 Protein Recapitulates Interferon-stimulated Gene Expression and Antiviral Response J. Biol. Chem., April 4, 2003; 278(15): 13033 - 13038. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Wittig, M. Nessling, R. D. Will, J. Mollenhauer, R. Salowsky, E. Munstermann, M. Schick, H. Helmbach, B. Gschwendt, B. Korn, et al. Candidate Genes for Cross-Resistance against DNA-damaging Drugs Cancer Res., November 15, 2002; 62(22): 6698 - 6705. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. D. Luker, V. Sharma, C. M. Pica, J. L. Dahlheimer, W. Li, J. Ochesky, C. E. Ryan, H. Piwnica-Worms, and D. Piwnica-Worms Noninvasive imaging of protein-protein interactions in living animals PNAS, May 14, 2002; 99(10): 6961 - 6966. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
) and variant 11C9 cells (•);
n = 3; error bars, SD when
larger than symbol). B and C,
characterization of drug resistance by colony formation assays. Number
of colonies/106 cells seeded for parental cells (
) and
variant 11C9 cells (
) treated 7 days with (B)
paclitaxel or (C) vinblastine, followed by 21 days of
recovery in medium without drug. Plating efficiency was 85–90% for
both lines. Values represent mean of triplicate determinations (except
0.1 µM paclitaxel, for which n = 9, representing 3 experiments); error bars, SE;
*, average of <1 colony observed.
) and variant 11C9 cells (