Identification of an Inhibitor of the EWS-FLI1 Oncogenic Transcription Factor by High-Throughput Screening

Abstract

Background Chromosomal translocations generating oncogenic transcription factors are the hallmark of a variety of tumors, including
many sarcomas. Ewing sarcoma family of tumors (ESFTs) are characterized by the t(11;22)(q24;q12) translocation that generates
the Ewing sarcoma breakpoint region 1 and Friend leukemia virus integration 1 (EWS-FLI1) fusion transcription factor responsible
for the highly malignant phenotype of this tumor. Although continued expression of EWS-FLI1 is believed to be critical for
ESFT cell survival, a clinically effective small-molecule inhibitor remains elusive likely because EWS-FLI1 is a transcription
factor and therefore widely felt to be “undruggable.”

Full text

962 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
DOI: 10.1093/jnci/djr156 Published by Oxford University Press 2011.
Advance Access publication on June 8, 2011.
Oncogenic transcription factors generated by chromosomal trans-
locations are found in all types of cancer ranging from leukemias
to solid tumors of both epithelial and mesenchymal origin. In most
cases, these transcription factors are integral to malignant transfor-
mation and maintenance of the oncogenic phenotype suggesting
that these proteins would be ideal drug targets (1–3). However, it
has been challenging to identify small-molecule inhibitors of these
transcription factors and, in general, these proteins have been
described as “undruggable” targets.
The Ewing sarcoma family of tumors (ESFT) comprises a
group of highly malignant bone tumors of childhood. Approximately
85% of these tumors are characterized by the t(11;22)(q24;q12)
translocation, which generates the highly dysregulated EWS-FLI1
transcription factor that is believed to be responsible for malignant
transformation and progression (4–15). Multiple studies have
shown that knockdown of EWS-FLI1 with either small interfering
RNA (siRNA) or antisense DNA decreases viability as well as
tumorigenicity in orthotopic mouse models (9,10,16–18). This
finding has led to a variety of efforts to identify a small-molecule
inhibitor of the EWS-FLI1 transcription factor (19,20).
Here, we describe the development and implementation of a
high-throughput screening strategy to identify inhibitors of the
EWS-FLI1 transcription factor. We used a high-throughput
screen that used a promoter-based primary screen for luciferase
expression and a multiplex polymerase chain reaction (PCR) sec-
ondary screen of EWS-FLI1–induced downstream targets to
evaluate more than 50 000 compounds including many natural
products. The top candidate from this screen was mithramycin, a
ARTICLE
Identification of an Inhibitor of the EWS-FLI1 Oncogenic
Transcription Factor by High-Throughput Screening
Patrick J. Grohar, Girma M. Woldemichael, Laurie B. Griffin, Arnulfo Mendoza, Qing-Rong Chen, Choh Yeung, Duane G. Currier,
Sean Davis, Chand Khanna, Javed Khan, James B. McMahon, Lee J. Helman
Manuscript received August 11, 2010; revised March 14, 2011; accepted April 11, 2011.
Correspondence to: Patrick J. Grohar, MD, PhD, Pediatric Oncology Branch, National Cancer Institute, National Institutes of Health, 10 Center Dr-MSC
1104, 10 CRC 1W-3816, Bethesda, MD 20892-1104 (e-mail: groharp@mail.nih.gov).
Background Chromosomal translocations generating oncogenic transcription factors are the hallmark of a variety of tumors,
including many sarcomas. Ewing sarcoma family of tumors (ESFTs) are characterized by the t(11;22)(q24;q12)
translocation that generates the Ewing sarcoma breakpoint region 1 and Friend leukemia virus integration 1 (EWS-
FLI1) fusion transcription factor responsible for the highly malignant phenotype of this tumor. Although continued
expression of EWS-FLI1 is believed to be critical for ESFT cell survival, a clinically effective small-molecule inhibitor
remains elusive likely because EWS-FLI1 is a transcription factor and therefore widely felt to be “undruggable.”
Methods We developed a high-throughput screen to evaluate more than 50 000 compounds for inhibition of EWS-FLI1
activity in TC32 ESFT cells. We used a TC32 cell–based luciferase reporter screen using the EWS-FLI1 down-
stream target NR0B1 promoter and a gene signature secondary screen to sort and prioritize the compounds. We
characterized the lead compound, mithramycin, based on its ability to inhibit EWS-FLI1 activity in vitro using
microarray expression profiling, quantitative reverse transcription–polymerase chain reaction, and immunoblot
analysis, and in vivo using immunohistochemistry. We studied the impact of this inhibition on cell viability in
vitro and on tumor growth in ESFT xenograft models in vivo (n = 15–20 mice per group). All statistical tests
were two-sided.
Results Mithramycin inhibited expression of EWS-FLI1 downstream targets at the mRNA and protein levels and
decreased the growth of ESFT cells at half maximal inhibitory concentrations between 10 (95% confidence inter-
val [CI] = 8 to 13 nM) and 15 nM (95% CI = 13 to 19 nM). Mithramycin suppressed the growth of two different
ESFT xenograft tumors and prolonged the survival of ESFT xenograft–bearing mice by causing a decrease in
mean tumor volume. For example, in the TC32 xenograft model, on day 15 of treatment, the mean tumor
volume for the mithramycin-treated mice was approximately 3% of the tumor volume observed in the control
mice (mithramycin vs control: 69 vs 2388 mm3, difference = 2319 mm3, 95% CI = 1766 to 2872 mm3, P < .001).
Conclusion Mithramycin inhibits EWS-FLI1 activity and demonstrates ESFT antitumor activity both in vitro and in vivo.
J Natl Cancer Inst 2011;103:962–978

jnci.oxfordjournals.org JNCI |Articles 963
drug that binds GC-rich regions of the genome and regulates
the expression of specific genes including SRC, MYC, and
MDR1 often by inhibiting the SP1 family of transcription factors
(21–26). In this report, we studied the effects of mithramycin on
expression of EWS-FLI1 downstream targets in ESFT cells and on
tumor growth in vivo.
Materials and Methods
Cell Lines, Cell Culture, and Reagents
All cell lines are patient derived. The cells were maintained in
growth medium (RPMI 1640 [Invitrogen, Carlsbad, CA] with 10%
fetal bovine serum [Sigma-Aldrich, St Louis, MO], 100 U/mL
penicillin and 100 µg/mL streptomycin [Invitrogen], and 2 mM
L-glutamine [Invitrogen]). The cells were maintained at 80% or
less confluency at 37°C in an atmosphere of 5% CO. The osteo-
sarcoma cell lines MNNG-HOS and U20S have been described
previously (27,28). Breast carcinoma MD-MBA-231 and MCF7
cells were the kind gift of Dr Patricia Steeg (National Cancer
Institute, Bethesda, MD) and have been previously characterized
(29,30). Ovarian cancer SKOV-3 cells were obtained from
American Type Culture Collection (Manassas, VA). ESFT cell
lines TC32, TC71, A673, CHLA9, and TC167 were the gift of
Tim Triche (The Saban Research Hospital, Children’s Hospital of
Los Angeles, Los Angeles, CA) and were confirmed by short-
tandem repeat genotyping (http://strdb.cogcell.com/). EW8 ESFT
cells were the gift of Peter Houghton (Nationwide Children’s
Hospital, Columbus, OH) and have been characterized previously
(31). The identities of all ESFT cell lines were confirmed by
reverse transcription– PCR (RT-PCR) using primers specific for
the translocation. We did not independently confirm the identities
of the other cell lines. The expression of EWS-FLI1 was con-
firmed routinely by RT-PCR for all ESFT cell lines.
Chemical Library
To screen for inhibitors of the EWS-FLI1 transcription factor, we
used a chemical library containing more than 50 000 compounds
that were isolated from natural product extracts, purchased, or
acquired through collaborations by the Molecular Targets
Laboratory at the National Cancer Institute at Frederick
(Frederick, MD). Information about the source of the library and
the individual compounds is available online at http://dtp.nci.nih.
gov/branches/npb/repository.html. To develop and validate the high-
throughput screen, we used a test set comprising 6500 compounds
that were selected to reflect the diversity of the 50 000-compound
library to assess the consistency and reproducibility of the high-
throughput assay.
Luciferase Reporter Constructs
The NR0B1 promoter was amplified by PCR from genomic DNA
isolated from TC32 cells with primers containing recognition sites
for KpnI and NheI restriction enzymes (Supplementary Table 1,
available online) and ligated into a TOPO-TA shuttle vector
(Invitrogen). The resulting plasmid was amplified in DH5a
Escherichia coli, purified, and digested with KpnI and NheI (New
England Biolabs, Ipswich, MA) in buffer I containing bovine
serum albumin as per the manufacturer’s protocol. The NR0B1
CONTEXT AND CAVEATS
Prior knowledge
The Ewing sarcoma family of tumors (ESFTs) is characterized by a
chromosomal translocation that generates EWS-FLI1, an oncogenic
fusion transcription factor whose continued expression is believed
to be critical for ESFT cell survival.
Study design
A high-throughput promoter-based screen was conducted to iden-
tify compounds that inhibit EWS-FLI1 activity in ESFT TC32 cells.
A TC32 cell–based luciferase reporter screen using the EWS-FLI1
downstream target NR0B1 promoter and a gene signature sec-
ondary screen was used to sort and prioritize the compounds. The
lead compound identified in the screen—mithramycin—was char-
acterized by microarray expression profiling, quantitative reverse
transcriptase–polymerase chain reaction, immunoblot analysis,
immunohistochemistry, and in ESFT xenograft models.
Contribution
Mithramycin blocked expression of EWS-FLI1 downstream targets
in vitro at both the RNA and protein levels and suppressed protein
expression of a well-characterized downstream target, NR0B1, in
vivo. Mithramycin inhibited ESFT cell growth in vitro with IC50
values ranging from 10 to 15 nM and suppressed the growth of ESFT
xenografts in vivo.
Implications
Mithramycin inhibits EWS-FLI1 activity and demonstrates ESFT
antitumor activity both in vitro and in vivo.
Limitations
The screen and subsequent experiments were all conducted in cell
lines and xenografts of those cell lines. There was no independent
verification of the authenticity of the non-ESFT cell lines used in
this study. The investigators were not blinded to the treatment
groups in the mouse experiments. Only one assay for apoptosis
was performed.
From the Editors
promoter–containing fragment was purified and cloned into
pGL4.18 (Promega, Madison, WI), a reporter plasmid containing
the firefly luciferase gene. The resulting plasmid (NR0B1-Luc)
was linearized with SalI (New England Biolabs) and introduced
into TC32 cells in the presence of buffer R with the use of
program O-17 and an Amaxa Nucleofector System (Lonza, Basel,
Switzerland) according to the manufacturer’s instructions.
The cytomegalovirus (CMV) promoter was PCR amplified
from pGL4.75 (Promega) with primers containing the appropriate
In-Fusion cloning 2.0 (Clontech, Mountain View, CA) recombina-
tion sites, which were designed using the manufacturer’s online
software (Supplementary Table 1, available online). The PCR
product was purified and cloned into the NheI site of pGL4.18.
The resulting plasmid (CMV-Luc) was introduced into TC32 cells
by nucleofection using buffer R, program O-17, and the Amaxa
System (Lonza).
High-Throughput Screen Assay Development
We generated separate stable TC32 cell lines expressing the
NR0B1 reporter (NR0B1-Luc) or the CMV reporter (CMV-Luc)

964 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
by expanding single-cell TC32 clones under selective pressure in
medium containing G418 (Sigma-Aldrich) at a concentration of
0.5 mg/mL. To ensure that these clones were suitable for high-
throughput screening, we selected cells that produced a strong
luciferase signal with minimal well-to-well variability. The cells
were optimized for cell seeding density, automation in cell plating
and liquid handling, length of incubation before treatment,
reporter half-life, and length of incubation after treatment for the
high-throughput screening assay. The assay was validated for
consistency and reproducibility by performing three separate
screens of the 6500-compound test set (see above), which yielded
a high correlation coefficient (median R2 = 0.9; range = 0.8720.91).
High-Throughput Screen
The 50 000-compound library was subjected to a high-throughput
screen to detect compounds that decreased expression of the
NR0B1-Luc reporter but had no effect on expression of the CMV-
Luc reporter or on cell viability in three separate screens that were
run in parallel. Test compounds were stored at various concentra-
tions in 100% dimethyl sulfoxide (DMSO) at 220°C in 384-well
plates, thawed just before use, and prepared in stock dilution plates
at a concentration of 100 µM in growth medium in a 384-well
dilution plate using a Biomek-FX liquid handler (Beckman
Coulter, Brea, CA). TC32 NROB1-Luc cells and TC32 CMV-
Luc cells were dispensed into opaque-bottom white 384-well assay
plates (PerkinElmer, Waltham, MA) at 5000 cells per well in 27 µL
growth medium using a µFill dispenser (BioTek, Winooski, VT)
for the luciferase reporter assay. TC32 NROB1-Luc cells were
similarly dispensed into clear 384-well assay plates (PerkinElmer)
for the cell viability assay. The cells were incubated for 4 hours at
37°C, and then 3 µL of each test compound from the 100-µM
stock dilution plate was transferred into assay plates (final concen-
tration of test compounds was 10 µM and final concentration
of DMSO was ≤1%). The cells were incubated for 24 hours at 37
°C. We added 30 µL of SteadyLite luciferase assay reagent
(PerkinElmer) to each well of an opaque-bottom white assay plate
and measured luciferase reporter activity with the use of a BMG
Labtech plate reader (Offenburg, Germany) in the luminescence
mode. Cell viability was assessed in parallel by adding XTT
reagent to the clear assay plate (10 µL per well; Developmental
Therapeutics Program, National Cancer Institute, Frederick,
MD) and measuring absorbance at 450 nm of the colored formazan
product of XTT metabolism with the use of a PerkinElmer Wallac
plate reader in the absorbance mode. All assay plates contained
16 wells each of the positive control compound [actinomycin D;
chosen as a general transcription inhibitor based on performance
in previous assays (32)] and the negative control compound
(DMSO). Routine measurement of assay performance using these
controls via the calculation of the Z-factor (33) for each screening
plate revealed a Z′ of 0.78–0.9, indicating that the screening assay
performed well during high-throughput screening. The following
criteria were used to define hits in the primary screen: 1) decrease
luciferase expression in TC32 NR0B1-Luc cells to less than 46%
of solvent control, 2) maintain greater than 80% of control
reporter activity in TC32 CMV-Luc cells, and 3) decrease cell
viability, as measured by an XTT assay, by less than 20%. Cut
points were chosen to yield a manageable number of compounds
for secondary screening with the premise of expanding the number
if an EWS-FLI1 inhibitor was not identified. A total of 150
compounds fulfilled all three criteria and, along with the 50 most
cytotoxic compounds as determined by the XTT cell viability
assay, were considered for secondary screening.
Compound Sort and Prioritization for the Secondary
Screen
The 200 compounds identified in the high-throughput screen
were subsequently evaluated for luciferase expression and cell
viability in the same manner as for the primary screen (described
in detail above) over six 10-fold dilutions ranging from 10 µM to
0.1 nM to prioritize them for further screening. The compounds
were sorted in descending order according to a score, X, which was
the sum of the log2 difference between the normalized luciferase
expression for the CMV-driven luciferase and the NR0B1-
normalized luciferase expression over all six concentrations:
2 2
[((log ( ) log ( 1 ))10
........ (log( ) log( 1 ))0.1 ]
Score CMV-Luc NR0B Luc nM
CMV-Luc NR0B Luc nM
X
= Σ − −
+ − −
We selected the 43 compounds with the highest scores for
secondary screening based on their ability to decrease luciferase
expression in TC32 NR0B1-Luc cells, while maintaining luciferase
expression in TC32 CMV-Luc cells when diluted 10-fold to a final
concentration of 1 µM.
Identification of EWS-FLI1 Downstream Targets
To compile a list of EWS-FLI1 downstream targets to develop the
secondary screen, we identified the following: 1) genes that were
highly expressed in tumor samples from ESFT patients, 2) genes
whose expression was induced with forced EWS-FLI1 expression
in mesenchymal progenitor cells [the presumed cell of origin of
Ewing sarcoma (7,34–37)], and 3) genes whose expression
decreased with siRNA knockdown of EWS-FLI1.
Highly expressed genes in ESFT tumor tissue were identified
from two pediatric cancer patient sample datasets (available at http://
home.ccr.cancer.gov/oncology/oncogenomics). We first selected
genes whose expression was twofold higher in ESFT tissue samples
compared with other tumor types and normal tissue samples (P < .001;
Student t test). We then retained only the genes whose expression
was induced twofold by EWS-FLI1 expression in mesenchymal pro-
genitor cells relative to a green fluorescent protein (GFP) control
(P < .01; Student t test). Finally, we eliminated genes whose expression
in ESFT cell lines did not decrease with siRNA knockdown of
EWS-FLI1 (38), yielding a list of 28 genes induced by EWS-FLI1
(Figure 1, B). We used a subset of these genes in the final multiplex
PCR assay of EWS-FLI1 downstream targets as described below.
Lentivirus Transduction of Human Mesenchymal
Progenitor Cells
Mesenchymal progenitor cells were isolated from bone marrow
samples from adult orthopedic patients in the laboratory of Dr
Rocky Tuan in the Cartilage Biology and Orthopedics branch of
the National Institute of Arthritis and Musculoskeletal and Skin
Diseases (Bethesda, MD) and transduced with a lentivirus contain-
ing EWS-FLI1 (generated by Betty Conde in the core facility
at National Cancer Institute-Frederick, Frederick, MD) at a

jnci.oxfordjournals.org JNCI |Articles 965
Figure 1. High-throughput luciferase-based primary screen and multi-
plex polymerase chain reaction secondary screen of a library of more
than 50 000 compounds. A) Primary screen design. A cell-based lucif-
erase primary screen included the EWS-FLI1 downstream target
NR0B1 promoter and a CMV promoter, which was used to control for
nonspecific cytotoxins and general transcription inhibitors. Downward
arrows indicate decreased activity of the luciferase reporter construct
and the left–right arrows represent no change in activity. B) List of
EWS-FLI1 downstream targets used for the secondary screen. Novel
list of EWS-FLI1–induced target genes comprising genes that are
highly expressed in Ewing sarcoma patient tissue, that suppress with
EWS-FLI1 siRNA knockdown, and that are induced when EWS-FLI1
is expressed in the presumed cell of origin. This list was used to
generate the gene list used in the multiplex polymerase chain
reaction assay. See Supplementary Table 1 (available online) for
full gene names. C) Summary of screen design. The summary of
the screen highlights the method used to identify mithramycin start-
ing from more than 50 000 pure compounds to the lead compound.
CMV = cytomegalovirus; ESFT = Ewing sarcoma family of tumors;
siRNA = small interfering RNA.

966 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
multiplicity of infection (MOI) of 10 in the presence of 6 µg/mL
polybrene to facilitate transduction. RNA was collected from the
transduced cells and subjected to microarray expression profiling as
described below.
Development of Multiplex PCR Assay of EWS-FLI1
Downstream Targets
EWS-FLI1 alters the expression of more than 500 genes (7).
However, our primary screen only evaluated the expression of one
downstream target via promoter-driven luciferase expression.
Therefore, to ensure that compounds identified in the primary
screen block EWS-FLI1 activity, we developed a multiplex PCR
assay that allowed simultaneous measurement of the expression of
multiple downstream targets of EWS-FLI1. Although we had
developed a list of 28 downstream targets of EWS-FLI1 as
described above, it was not possible to include all 28 genes in the
multiplex PCR assay. The assay uses capillary electrophoresis to
separate gene-specific PCR products. Therefore, each gene has to
yield a single clean PCR product that differs in length by at least
seven nucleotides from the PCR product of every other gene in the
assay and is between 130 and 300 nucleotides in total length.
These requirements limit the number of gene targets used in one
multiplex assay to, at most, 24 targets. Because we also wanted to
include four housekeeping genes—GAPDH, ACTB, EEF1A1, and
PPIA— as negative controls, a maximum of 20 target genes could
be included. We also wanted to include ID2 and CAV1, which are
strongly supported as EWS-FLI1 targets in the literature (39,40),
even though these genes did not appear in all four datasets used to
identify the 28 target genes. This further limited the number of
target genes that could be included to 18. We reevaluated each of
the genes in our original 28-gene list relative to the other genes on
the list and eliminated those with lower expression in primary
tumors (ie, FADS1, KCNE3, KIAA1462, VWA5A, NRN1, and
STK32B), those with less-pronounced induction in the mesen-
chymal progenitor cells (ie, GRK5 and GYG2), and those with
less-pronounced induction with siRNA knockdown of EWS-FLI1
(ie, FLRT2, KIAA1462, VWA5A, and OPN3). The remaining
17 genes were used for assay development. The nine genes that
produced the cleanest PCR product of the appropriate size were
included in the final assay, along with probes designed to detect
expression of EWS-FLI1, CAV1, ID2, and the four housekeeping
genes The final assay included 11 EWS-FLI1 target genes (ID2,
FCGRT, NR0B1, STEAP2, CAV1, LDB2, RCOR1, KDSR, IL1RAP,
CCK, and XG), a primer pair to amplify Kanr RNA (which was used
as an internal PCR control), a primer pair for EWS-FLI1, and the
four housekeeping genes (PPIA, GAPDH, ACTB, and EEF1A1).
The sequences of the gene-specific primer pairs are listed in
Supplementary Table 2 (available online).
Secondary Screen
The 43 compounds identified in the primary screen were evaluated
for their ability to suppress expression of EWS-FLI1 downstream
targets in the multiplex PCR assay without changing expression of
the housekeeping genes. The individual compounds were stored
frozen in 10% DMSO at a concentration of 100 µM. The com-
pounds were thawed immediately before use and diluted to a final
concentration of 100 nM in growth medium. TC32 NR0B1-Luc
cells (140 000 cells per well) were added to a 12-well plate and
incubated overnight. The medium was aspirated and replaced
with one of the 43 compounds at 100 nM. The cells were incu-
bated for 6 hours, at which point the medium was aspirated, and
the cells were washed and lysed in the plate with the addition of
RLT buffer (Qiagen, Germantown, MD). RNA was purified using
the RNeasy mini kit (Qiagen) and the automated QIAcube RNA
purification system (Qiagen) according to the manufacturer’s in-
structions. The RNA was quantitated with the use of a NanoDrop
2000 spectrophotometer (Thermo, Waltham, MA), and only
RNA with a ratio of the absorbance at 260 nm to that at 280 nm
greater than 2 was used in the assay. RNA was reverse transcribed
using gene-specific reverse primers (Supplementary Table 2,
available online) and a GeXP Start kit (Beckman Coulter) contain-
ing reverse transcriptase, reverse transcriptase buffer, Kanr reverse
transcriptase primer, and Kanr RNA, which was added to all
reactions as an internal PCR control. The reverse transcription
reaction was run in triplicate on a Veriti thermocycler (Applied
Biosystems, Norwalk, CT) with the following program: 48°C for
1 minute, 37°C for 5 minutes, 42°C for 60 minutes, and 95°C for
5 minutes. The reaction mixture (9.3 µL) was transferred to a
second tube for PCR amplification of the cDNA using Thermo-
start Taq DNA polymerase (Thermo), gene-specific forward
primers (Supplementary Table 2, available online), and the GeXP
PCR kit, which contained MgCl2 (final concentration 5 mM),
deoxynucleoside triphosphates, buffer, and fluorescently labeled
nested PCR primers. The PCR program was run on a Veriti ther-
mocycler (Applied Biosystems) with a standard “hot start” pro-
gram of 95°C for 10 minutes, 94°C for 30 seconds, 55°C for 30
seconds, and 70°C for 1 minute (35 cycles).
The PCR products were separated by capillary electrophoresis
with the use of a GenomeLab GeXP device (Beckman Coulter), and
the individual PCR products were quantitated based on their peak
height as measured by the fluorescent detector in the GeXP instru-
ment. Expression was quantitated by normalizing the peak height of
each gene to the Kanr internal PCR control and to the geometric
mean of the four housekeeping genes. This value was subsequently
fit to a standard curve using a third-order polynomial equation
while maintaining a correlation coefficient (R2) greater than .99 to
yield a final expression value. The standard curve was generated by
performing RT-PCR on RNA collected from untreated TC32 cells
using the same reagents and protocol that were used for cells treated
with the compounds over a range of RNA inputs from 500 to 1 ng.
The compounds were sorted based on their ability to suppress
the mRNA expression of each of the EWS-FLI1 downstream
targets in the assay. To quantitate the difference in expression for
treated vs solvent-treated control cells, we summed the difference
in expression for each EWS-FLI1 downstream target evaluated in
the assay for each compound using log-transformed data, which
yielded a score (Y) for each compound:
2 2
2 2
[((log ( 1 ) log ( 1 ))...
...(log ( 11 ) log ( 11 ))]
Score Gene control Gene treated
Gene control Gene treated
Y
= Σ −
+ −
The compounds were subsequently sorted in descending order
of Y score (data not shown); the highest scoring compound—
mithramycin—was selected for further study.

jnci.oxfordjournals.org JNCI |Articles 967
Quantitative RT-PCR
The goal of the secondary screen was to prioritize the compounds
for further study. Because mithramycin was the highest priority
compound, we first wanted to confirm that it was able to suppress
RNA expression of EWS-FLI1 downstream targets by using the
more generally accepted method of quantitative RT-PCR. TC32
cells (120 000 cells per well) were plated in triplicate in 12-well
plates and allowed to recover overnight. The growth medium was
subsequently removed and replaced with growth medium contain-
ing 100 nM mithramycin or solvent control (0.01% phosphate-
buffered saline [PBS] in growth medium). RNA was collected using
an RNeasy kit and a QIAcube (Qiagen) as described above. RNA
(500 ng) was reverse transcribed using a high-capacity reverse tran-
scriptase kit (Applied Biosystems), which contains 10× RT buffer, 25
× dNTPs, 10× random RT primers, and reverse transcriptase in a
Veriti thermocycler (Applied Biosystems) according to the fol-
lowing program: 25°C for 10 minutes, 37°C for 60 minutes, and
85°C for 5 minutes. The cDNA product (100 ng) was subsequently
PCR amplified using the primers developed for the multiplex assay,
which are known to produce one clean cDNA product of the target
genes (CCK, ID2, LDB2, NR0B1, RCOR1, and GAPDH) by quanti-
tative PCR using a Sybr green kit (Applied Biosystems) and the
CFX 96 Real Time System (Bio Rad, Hercules, CA) according
to the following program: 95°C for 10 minutes, 95°C for 15 mi-
nutes, 55°C for 15 minutes, and 72°C for 1 minute for 40 cycles.
Expression was determined by comparing the average threshold
cycle (Ct) for the three 100 nM mithramycin-treated replicates to
the average Ct for the three solvent control–treated replicates by
using the Pfaffl equation (36) to determine fold change in expression
for each gene. The average Ct was determined by normalizing the
Ct of CCK, ID2, LDB2, NR0B1, and RCOR1 to that of GAPDH (41).
The entire experiment was repeated two more times with similar
results, and the data are presented as the average for the three
independent experiments that included eight replicate samples.
Mithramycin
Mithramycin was purchased from Tocris Bioscience (Bristol, UK),
dissolved as a 1 mg/mL stock solution in PBS, and frozen in
aliquots for use in the in vitro and in vivo experiments. For in vivo
experiments, an aliquot of the 1 mg/mL stock solution was diluted
to the appropriate concentrations in PBS, and the dilutions were
frozen and stored at 280°C to avoid repeated freeze–thaw cycles.
For in vitro experiments, an aliquot of the 1 mg/mL stock solu-
tion was diluted with medium immediately before use.
Cell Proliferation Assays
The viability of all cell lines in the presence and absence of
mithramycin was determined with the use of the CellTiter 96 3-(4,
5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-
sulfophenyl)-2H-tetrazolium, inner salt (MTS) assay (Promega).
Cells were plated in a 96-well plate (5000 cells per well) and incu-
bated overnight. Mithramycin was diluted in a medium over a
range of concentrations from 500 to 0.5 nM and added to the cells.
The cells were incubated for 48 hours, and 20 µL of CellTiter
96 reagent (Promega) was added to each well and the cells were
incubated for 1 hour at 37°C. The color change produced by the
bioreduction of the MTS reagent was measured by reading the
percent transmittance at 490 nm with the use of a VersaMax
96-well plate reader (Molecular Devices, Sunnyvale, CA) and
plotted against a standard curve previously generated for each
individual cell line over a range of cell numbers to yield an estimate
of the cell number. The cell number was evaluated as a function of
drug concentration by nonlinear regression with the use of Prism
4.0 software (GraphPad Software Inc, La Jolla, CA) to determine
the concentration of half-maximal inhibition of viability (IC50).
Each experiment was performed with three to six replicates and
two to five independent experiments per cell line.
Immunoblot Analysis
TC32 and TC71 cells (1.5 × 106) were plated in 10 cm2 dishes and
incubated overnight. The medium was aspirated and the cells were
incubated with freshly diluted mithramycin at the times and con-
centrations specified for each experiment. The cells were scraped
into PBS using a cell lifter (Fisher Scientific, Waltham, MA) and
centrifuged (233g) for 2 minutes. The cells were washed with
an additional 1 mL of PBS, centrifuged (2000g) for 2 minutes, and
lysed in 4% LDS buffer (Sigma-Aldrich) containing Complete Mini
Protease Inhibitor Cocktail (one tablet per 10 mL LDS buffer;
Roche, Mannheim, Germany). The protein concentration of the
lysates was determined by the bicinchoninic acid (BCA) assay
(Pierce, Waltham, MA). Cell lysates (25 µg protein per sample)
were resolved on a 4%–12% Bis-Tris gel (Invitrogen) in 1× NuPage
3-(N-Morpholino)propanesulfonic acid (MOPS) sodium dodecyl
sulfate (SDS) buffer or 2-(N-Morpholino)ethanesulfonic acid
(MES) SDS buffer (Invitrogen) and transferred to nitrocellulose
membranes using 1× NuPage transfer buffer (Invitrogen) with 20%
methanol. The membranes were subsequently probed with the fol-
lowing antibodies: rabbit monoclonal anti-ID2 (1:1000 dilution;
Cell Signaling, Danvers, MA), mouse monoclonal anti-FLI1 (1.5
µg/mL; BD biosciences, Franklin Lakes, NJ), rabbit polyclonal anti-
NR0B1 (1:1000 dilution; Abcam, Cambridge, UK), rabbit poly-
clonal anti-b-actin (1:10 000 dilution; Abcam), mouse monoclonal
anti-cleaved PARP (1:1000 dilution; Cell Signaling), mouse mono-
clonal anti-phospho-histone H2A.X (ser 139) (gH2AX) (1:1000 di-
lution; Upstate, Billerica, MA), mouse monoclonal anti-p53 (1:1000
dilution; Cell Signaling), rabbit monoclonal anti-p21 (1:1000 dilu-
tion; Cell Signaling), and rabbit polyclonal anti-phos-P53 (ser 15)
(1:1000 dilution; Cell Signaling). The membranes were incubated with
horseradish peroxidase (HRP)–conjugated goat polyclonal anti-
mouse or anti-rabbit immunoglobulin (IgG) (1:2000 dilution; Pierce),
and bound antibody was visualized on film using an ECL Western
Blotting Analysis System (Amersham, Buckinghamshire, UK).
Confocal Microscopy
TC32 cells (20 000 cells) were plated in a Lab Tek II chamber slide
with cover (Nunc, Rochester, NY) and incubated overnight. The
medium was aspirated and the cells were treated with 100 nM
mithramycin for 24 hours, fixed in 4% paraformaldehyde in PBS,
washed, and permeabilized in 1% Triton X-100. The cells were
incubated with 10% goat serum to block nonspecific protein binding
and probed with a mouse monoclonal anti-phospho-histone H2A.X
(ser 139) antibody (1:200 dilution; Upstate). The cells were incu-
bated with Alexa488-labeled secondary antibody (Millipore, Billerica,
MA), mounted in VectaShield mounting medium containing DAPI

968 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
(Vector Labs, Burlingame, CA), and viewed with the use of a Zeiss
510 confocal microscope.
Xenograft Experiments
First, we performed an in vivo dose–response experiment to deter-
mine the effective dose of mithramycin (ie, the maximum dose of
mithramycin that was well tolerated without appreciable toxicity).
Five cohorts of SCID-Bg mice (20 mice per group, 4–6 weeks old;
Charles River Laboratories, Wilmington, MA) were injected with
2 × 106 TC71 cells per mouse into the left gastrocnemius muscle.
Once a tumor with a minimum diameter of 0.5 cm was established,
the mice were randomly assigned to receive no treatment or treat-
ment with mithramycin at a dose of 1, 0.5, 0.1, or 0.05 mg/kg body
weight on an Monday-Wednesday-Friday (M-W-F) schedule.
Two mice in each group were killed by CO2 asphyxiation on days
2 and 4 after initiation of treatment and their tumor tissue was
collected immediately and frozen for later analysis of protein ex-
pression by either immunoblotting or immunohistochemistry.
The dose–response analysis was performed by plotting the tumor
volume for every mouse as a function of dose on day 11 and select-
ing the most effective dose where minimal toxicity was observed.
The lower-dose mithramycin cohorts had fewer mice because
some were killed when their tumors reached a diameter of 2 cm.
Once the maximum effective dose without appreciable toxicity
was determined to be 1 mg/kg body weight per dose on a M-W-F
schedule, three cohorts of female SCID-Bg mice (30 mice per
group, 4–6 weeks old; Charles River Laboratories) were injected
with either TC32, TC71, or MNNG-HOS cells (2 × 106 cells per
mouse) into the left gastrocnemius muscle. Once every mouse that
was injected with a particular cell line had established a tumor with
a minimum diameter of 0.5 cm, all mice in that group (either
bearing TC32, TC71, or MNNG-HOS) were randomly assigned
to treatment (n = 15; mithramycin) or control (n = 15; no treat-
ment). Mice in the treated cohorts were subsequently treated
by intraperitoneal injection with 1 mg/kg body weight per dose
mithramycin or no treatment on a M-W-F schedule and killed by
CO2 asphyxiation when the tumor reached a diameter of 2 cm.
Survival analysis includes 12 mice at risk because three mice in
each cohort were collected for tumor tissue analysis 12 days after the
start of treatment and therefore were excluded from the analysis.
Each tumor was measured three times per week, and tumor
volume was determined using the equation (D × d2)/6 × 3.12, where
D is the maximum diameter and d is the minimum diameter.
Tumor volumes for all mice in each xenograft-treatment group
(excluding the mice that were killed on the predetermined schedule)
were averaged to yield the mean tumor volume for the corresponding
group. The experiments were approved by the Animal Care and
Use Committee of the National Cancer Institute. Investigators
were not blinded to the treatment groups.
Immunohistochemistry of Xenograft Tumor Tissue
The tumor tissue was fixed in 10% formalin, sectioned (10-µm-thick
sections), and stained with hematoxylin–eosin (American Histo
Labs, Gaithersburg, MD). Additional 10-µm sections were
obtained for immunohistochemistry and deparaffinized with
xylene, 100% ethanol, and 95% ethanol. Antigen retrieval was
performed with citrate buffer (DakoCytomation, Glostrup,
Denmark). The sections were incubated with 3% hydrogen peroxide
to quench peroxidase activity, followed by incubation with 5%
goat serum to block nonspecific protein binding and incubation
overnight with rabbit polyclonal anti-NR0B1 (6 µg/mL; Abcam).
Sections were subsequently labeled using biotinylated goat
anti-rabbit secondary IgG antibody and streptavidin-conjugated
horseradish peroxidase (Dako). The sections were stained with
hematoxylin, washed, and mounted with mounting medium
(Dako). Images were obtained in the National Institutes of Health
core imaging facility.
Microarray Analysis of EWS-FLI1 Downstream Targets
TC32 and TC71 cells were plated in triplicate (1.5 × 106 cells per
replicate) and incubated overnight. The cells were treated with
100 nM mithramycin for 6 hours, washed, and lysed in 1 mL of
TRIzol reagent (Invitrogen). The lysates were subjected to chloro-
form extraction and centrifuged at 16 000g at 4°C. The aqueous
layer was collected, mixed with an equal volume of 70% ethanol, and
total RNA was purified using the RNeasy protocol (Qiagen) as per
the manufacturer’s instructions. RNA integrity was determined
with the use of a Bioanalyzer (Agilent, Santa Clara, CA) according
to the manufacturer’s instructions. Gene expression profiling was
performed using the GeneChip Human Genome U133 Plus 2.0
microarray (Affymetrix, Santa Clara, CA). RNA was in vitro tran-
scribed, fragmented, hydridized, and stained using the appropriate
GeneChip kits (Affymetrix) as per the manufacturer’s instructions.
The CEL files were exported from Affymetrix GCOS software and
normalized with RMA-sketch from Affymetrix Power Tools.
Statistical Methods
Data are presented as mean values with 95% confidence intervals
(CIs). The Student t test was performed to determine the statistical
significance of differences between the treated and the control
groups for the PCR results and the tumor volumes in the mouse
studies. The survival analysis for the mouse experiment was per-
formed using Kaplan–Meier analysis, log-rank test, and the Prism
4.0 software package (GraphPad Software Inc). A P value less than
.05 was considered statistically significant. All statistical tests were
two-sided. Expression profiling was analyzed using GCOS soft-
ware and normalized with RMA-sketch from Affymetrix Power
Tools (42). The gene set enrichment analysis (GSEA) method
(http://www.broad.mit.edu/gsea/) was applied to investigate the
enrichment of the list of EWS-FLI1–induced downstream targets.
GSEA analysis was completed with a weighted enrichment statistic
and genes were ranked using the log2 ratio of expression for
mithramycin-treated cells to control cells. The limma bioconductor
package was used to construct a two-factor linear model between
control and treatment groups. Enrichment of the EWS-FLI1 gene
list was tested using the geneSetTest function from the limma
package (43).
Results
The Primary and Secondary Screens
The goal of this study was to identify inhibitors of the EWS-FLI1
transcription factor. We first conducted a primary screen of
50 000 compounds to identify those that could suppress the

jnci.oxfordjournals.org JNCI |Articles 969
expression of a well-characterized EWS-FLI1 downstream target
luciferase reporter construct. In the primary screen, we evaluated
the library of compounds by using an EWS-FLI1 downstream
target NR0B1 promoter luciferase construct in parallel with a
constitutively active CMV promoter construct to control for non-
specific transcription inhibitors and a cell viability assay to control
for nonspecific cytotoxins (Figure 1, A). This screen yielded com-
pounds that inhibited expression of the target reporter construct
but had no effect on the CMV reporter construct or on cell
viability.
We then conducted a secondary screen of the 200 compounds
to identify those that could directly suppress expression of multiple
downstream targets of EWS-FLI1. We first generated a novel list
of EWS-FLI1 downstream targets from four datasets comprising
two clinical patient sample datasets and two experimental datasets.
Genes that were highly expressed in ESFT patient tissue samples
compared with normal tissue and other tumor types were selected.
We subsequently included only those genes whose expression
increased with EWS-FLI1 expression in the presumed cell of
origin, mesenchymal progenitor cells (7,34–37). Finally, we
included only the genes whose expression decreased with siRNA
knockdown of EWS-FLI1 from a dataset in the literature (38). We
selected the remaining genes that were in all four sets to give a
preliminary list of 28 genes (Figure 1, B).
To generate the final list of genes used in the multiplex PCR
assay, we selected nine of the targets from the preliminary list (see
“Materials and Methods” for details); included two additional well-
characterized targets, ID2 and CAV1, from the literature (39,40);
a primer pair for EWS-FLI1 itself; and four different house-
keeping genes to generate the final list of genes to use in the sec-
ondary screen (Supplementary Figure 2, available online).
Once the assays were designed and validated, we evaluated and
prioritized 50 000 compounds based on the inhibition of the
EWS-FLI1–specific promoter while maintaining expression of the
CMV promoter yielding the top 200 compounds (see “Materials
and Methods”). We eliminated all compounds that lost activity or
selectivity with a 10-fold dilution and screened the remaining top
43 compounds by using the multiplex PCR EWS-FLI1 gene
signature assay (summarized in Figure 1, C). We sorted these
compounds mathematically using log-transformed data and plotted
the data in a heat map format for visualization purposes only
(Supplementary Figure 1, available online). The compound that
showed the best suppression of the EWS-FLI1 gene signature,
mithramycin, was selected for further study.
Effect of Mithramycin Treatment on the NR0B1- and
CMV-Driven Luciferase Reporters
To examine the dose-dependent effect of mithramycin on expres-
sion of downstream targets at the promoter level, we treated TC32
cells containing the NR0B1- and CMV-driven luciferase reporter
constructs for 6 hours over a range of concentrations from 200 to
5 nM and measured the effect on bioluminescence. Mithramycin
decreased NR0B1 promoter activity in a dose-dependent fashion
at concentrations between 5 and 200 nM at 6 hours without
affecting the activity of the constitutively active CMV promoter.
For example, treatment of TC32 cells with 50 nM mithramycin for
6 hours reduced luciferase expression to only 61% of control (95%
CI = 57% to 65%) (P < .001) (Figure 2, A). At 6 hours, these effects
could not be accounted for by a general reduction in transcription
because there was only minimal change in CMV-driven luciferase
activity or by cell death because no change in cell number was
observed at this time point (data not shown). Similar results were
obtained with replicate experiments at 6, 12, 24, and 48 hours
of incubation (data not shown). However, the suppression of
luciferase expression at the longer incubation times was confounded
by an increase in cell death and therefore only the 6-hour time
point is shown.
Effect of Mithramycin on EWS-FLI1 Gene Signatures
We confirmed the effect of mithramycin treatment on the novel
EWS-FLI1 gene signature by incubating TC32 cells with 100 nM
mithramycin for 6 hours and measuring the effect on the expression
of EWS-FLI1 target genes using RT multiplex PCR (Figure 2, B).
Mithramycin decreased the expression of 10 of the 11 EWS-FLI1
downstream target genes (all but CCK) but did not suppress
expression of GAPDH. We confirmed the result of the multiplex
PCR by performing quantitative RT-PCR for five genes: CCK,
which did not suppress in the multiplex PCR; NR0B1 and ID2 (for
which antibodies are available); and RCOR1 and LDB2 because
their expression was decreased the most by mithramycin compared
with control in the multiplex PCR assay. The data confirmed the
results of the multiplex PCR: Mithramycin treatment did not
change CCK expression but suppressed it for the other four targets
relative to control (P = .008) (gene expression ratio for mithramycin-
treated to control, ID2: 0.7, 95% CI = 0.56 to 0.90 [P = .09]; LDB2:
0.7, 95% CI = 0.49 to 0.84 [P = .04]; NR0B1: 0.4, 95% CI = 0.36
to 0.49 [P = .01]; RCOR1: 0.2, 95% CI = 0.08 to 0.31 [P = .01])
(Figure 2, C).
Next, to demonstrate suppression of an independent EWS-
FLI1 gene signature, we evaluated a list of EWS-FLI1–induced
targets from the literature (45). We performed microarray gene
expression analysis on TC32 and TC71 cells treated with mithra-
mycin or control and used GSEA (44) to investigate the enrich-
ment of EWS-FLI1 target genes in mithramycin-treated cells.
EWS-FLI1–induced target genes were statistically significantly
enriched in the list of genes whose expression decreased with
mithramycin treatment (P < .001) with a core set of 30 genes found
in the leading edge subset that accounts for the enrichment signal
(Figure 2, D). Finally, we used the geneSetTest function from
the limma package as an alternative statistical method to confirm
that mithramycin treatment suppressed expression of the list of
EWS-FLI1 targets (P < .001) (43).
Effect of Mithramycin on Protein Expression of NR0B1
and ID2
Next, we evaluated the effect of mithramycin treatment on the
protein expression of NR0B1 and ID2, which are generally
accepted as EWS-FLI1 downstream targets (38,39,46) and for
which reliable antibodies exist. In TC32 cells treated with 100 nM
mithramycin, expression of both targets decreased between 6 and
12 hours of treatment (Figure 3, A). This time frame is consistent
with the time frame of mRNA suppression detected by quantita-
tive PCR (Figure 2, C) and occurs at the same time or just before
the induction of apoptosis as determined by the cleavage of PARP.

970 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
In addition, we observed a clear dose-dependent inhibition of ID2
protein expression in both the TC32 and TC71 cell lines at
24 hours of treatment with 100 nM and 10 nM mithramycin
(Figure 3, B). There was no change in expression of EWS-FLI1 in
TC32 or TC71 cells treated with mithramycin at any concentration
for any time.
Effects of Mithramycin on DNA Damage
Mithramycin is known to bind the minor groove of DNA (47). We
therefore examined the effects of mithramycin treatment on the
integrity of the DNA by treating TC32 cells with mithramycin and
measuring the effect on p53 activation and on gH2AX formation.
The phosphorylation of histone H2AX at serine 139 (ie, the
generation of gH2AX) is known to occur at sites of DNA double-
strand breaks and therefore marks DNA damage (48). We found
that there is a secondary DNA damaging effect of mithramycin
treatment that occurs after inhibiting the downstream target
expression as measured by immunoblot analysis showing the
activation and subsequent degradation of p53 and the phosphory-
lation of H2AX. At 12–18 hours of exposure, mithramycin caused
the induction of phosphorylation at serine 15 and subsequent
degradation of p53, which was complete by 48 hours (Figure 4, A).
To confirm that the generation of gH2AX observed was not simply
the result of an apoptotic effect, we performed confocal micros-
copy following mithramycin treatment. The gH2AX foci appeared
in a diffuse nuclear pattern consistent with the accumulation of
DNA double-strand breaks (48) in the mithramycin-treated cells but
not the control cells at 24 hours following treatment (Figure 4, B).
Finally, it is notable that p21 induction occurred before the
phosphorylation of serine 15 of p53 and that no further increase in
p21 expression occurred after activation of p53.
Effects of Mithramycin on Viability of a Panel of Cell Lines
To evaluate the effect of EWS-FLI1 suppression on cell viability,
we treated a panel of cell lines for 48 hours with concentrations of
mithramycin that ranged from 500 to 0.5 nM and measured the
(continued)

jnci.oxfordjournals.org JNCI |Articles 971
Figure 2. Effect of mithramycin on EWS-FLI1
downstream target expression. A) Luciferase
reporter assay. TC32 cells bearing the NR0B1-
driven luciferase reporter or the CMV
(cytomegalovirus)-driven control promoter
were incubated with mithramycin at concen-
trations ranging from 5 to 200 nM for 6 hours,
and the bioluminescence was read after the
addition of luciferin reagent. Data are normal-
ized to the solvent control and represent the
average of six replicates from a single experi-
ment; error bars represent 95% confidence
intervals. The Student t test was used to com-
pare the statistical significance of the differ-
ence in mean expression between the
CMV-driven luciferase and the NR0B1-driven
luciferase in cells treated with 50 nM mithra-
mycin (n = 6) (P < .001). B) EWS-FLI1 down-
stream target expression. TC32 cells bearing
the NR0B1-driven luciferase reporter were
treated with 100 nM mithramycin or solvent
(control) for 6 hours, and expression of the
downstream targets was assessed with the
multiplex polymerase chain reaction (PCR)
assay and normalized to an internal PCR con-
trol (Kanr RNA). The data represent the av-
erage of three replicates from one experiment;
error bars represent 95% confidence intervals.
The housekeeping gene GAPDH is shown as
an internal control. C) Quantitative PCR verifi-
cation. RNA was collected from TC32 cells
treated with 100 nM mithramycin for 6 hours
at 37°C and analyzed by reverse transcription–
quantitative PCR. Expression of each gene
was evaluated as a function of a change in the
expression of the GAPDH reference gene and
plotted as the ratio of expression of the treated
to control target gene. The data represent
the average of eight replicates from three
independent experiments. Error bars indicate
the 95% confidence intervals. The P value rep-
resents the statistical significance of the differ-
ence between the treated and solvent control
groups and was determined by the Student
t test (two-sided). CCK was not included in the
t test because there was no difference in CCK
expression between the mithramycin-treated
and solvent control samples. D) Enrichment
plot of EWS-FLI1 target genes. TC32 cells were
treated with 100 nM mithramycin for 6 hours.
RNA was collected and gene expression was
profiled using Affymetrix microarrays.
Expression of EWS-FLI1–induced target genes
was statistically significantly suppressed by
mithramycin treatment compared with control
(P < .001). Gene set enrichment analysis (44)
was completed with weighted enrichment
statistics and genes ranked using the log2
ratio of gene expression in mithramycin-
treated and control cells as described in the
“Material and Methods” section. The black
vertical lines mark the hits of EWS-FLI1 target
genes. The red vertical line specifies the max-
imum enrichment score, which reflects the
degree to which a gene set is overrepresented
at the bottom of a ranked list of genes. The genes listed under the plot are the leading-edge subset, which is a subset of EWS-FLI1–induced target
genes that contribute most to the maximum enrichment score.
IC50 values. Mithramycin was a potent cytotoxic agent in a panel of
ESFT cell lines, with IC50 values that ranged from 10 nM (95%
CI = 8 to 13 nM) to 15 nM (95% CI = 13 to 19 nM) (Table 1). We
also evaluated other solid tumor cell lines, including osteosarcoma,
breast carcinoma, and ovarian carcinoma and found somewhat
higher IC50 values. It is notable that all ESFT cell lines tested were
extremely sensitive to the drug, and they were slightly more sensi-
tive to mithramycin treatment than the other cell lines tested.

972 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
Effect of Mithramycin on Growth of ESFT and
Osteosarcoma Xenografts
Next, we examined the effect of mithramycin on tumor growth in
vivo using two different Ewing sarcoma xenografts and one osteo-
sarcoma xenograft. In the first experiment, we defined the effective
dose that did not result in substantial toxicity. Mice with TC71
xenografts were treated with mithramycin at four different doses
every Monday, Wednesday, and Friday for 11 days; control mice
received no treatment (n = 14–16 mice per group). The mice that
were treated with the two highest doses of mithramycin—0.5 and
1.0 mg/kg body weight—had statistically significantly reduced
growth of the TC71 xenografts compared with control (P < .001
for both). On day 10 of treatment, mean tumor volume was
reduced by 47% in the 0.5 mg/kg dose group (control vs 0.5 mg/kg
mithramycin: 2696 vs 1259 mm3, difference = 1437 mm3, 95%
CI = 787 to 2087 mm3) and by 75% in the 1.0 mg/kg dose
group (control vs 1.0 mg/kg mithramycin: 2860 vs 686 mm3,
difference = 2174 mm3, 95% CI = 1589 to 2770 mm3) relative to
the control tumor volume (Figure 5, A).
We also observed a statistically significant decrease in
growth of the TC32 and TC71 xenografts over time in mice
treated thrice weekly with mithramycin at 1.0 mg/kg body
weight (n = 15 mice per group) (Figure 5, B). The TC32 xeno-
grafts showed profound tumor growth inhibition with a decrease
in the mean tumor volume of the entire cohort after a single
dose of mithramycin to almost no appreciable tumor in any of
the mice that persisted with continued treatment. For example,
on day 15 of treatment, whereas the mean tumor volume for the
mithramycin-treated mice was only slightly less than mean
volume at the start of treatment (69 vs 76 mm3, difference = 7 mm3,
95% CI = 213 to 27 mm3, P = .57), it was approximately 3% of
the tumor volume observed in the control mice (mithramycin vs
control: 69 vs 2388 mm3, difference = 2319 mm3, 95% CI = 1766
to 2872 mm3, P < .001). Only three of the 12 evaluable mithramycin-
treated mice bearing TC32 xenografts had any appreciable tumor
at day 19 of treatment, when all of the control mice had to be
killed due to tumor progression to 2 cm (Figure 5, B). The TC71
xenografts in mice treated with mithramycin also showed a statis-
tically significant suppression of tumor growth compared with
control (mean tumor volume on day 15 of treatment for mithra-
mycin vs control: 762 vs 3502 mm3, difference = 2740 mm3, 95%
CI = 2329 to 3157 mm3, P < .001), a volume roughly 22%
of control, which translated into a near doubling in median sur-
vival from 15 to 26 days for mice bearing the TC71 xenograft
(Figure 5, B and C).
As a control for nonspecific effects of mithramycin, we
implanted a cohort of mice with the osteosarcoma MNNG-HOS
cells, which lack the EWS-FLI1 translocation, and treated them by
intraperitoneal injection with mithramycin at 1 mg/kg body
weight per day, three times per week. Despite the relatively similar
in vitro IC50 values for mithramycin in MNNG-HOS, TC32, and
TC71 cells, mithramycin had less of an impact on the growth of
the MNNG-HOS xenografts than it had on the EWS-FLI1–
expressing TC32 and TC71 xenografts and no impact on the
survival of the mice at the time when the surviving mice had to be
euthanized (Figure 5, D).
Figure 3. Immunoblot analysis of the EWS-
FLI1 downstream targets NR0B1 and ID2. A)
Time course of NR0B1 and ID2 protein expres-
sion and PARP cleavage. TC32 cells were in-
cubated with 100 nM mithramycin. Protein
lysates were made at various times of incuba-
tion and subjected to immunoblot analysis.
PARP cleavage was used to indicate induction
of apoptosis, and b-actin was used as a con-
trol for equal protein loading. Two repeat time
course experiments verified these results. B)
Dose-dependent effect of mithramycin treat-
ment. TC32 (left) and TC71 (right) ESFT cells
were treated with 100, 10, or 1 nM mithramy-
cin for 12 or 24 hours. Protein lysates were
collected and subjected to immunoblot
analysis. Two repeat experiments yielded
similar results in both TC71 and TC32 cells.
S = solvent control; conc = concentration.

jnci.oxfordjournals.org JNCI |Articles 973
To confirm that the effects mithramycin on tumor growth were
related to suppression of EWS-FLI1 activity, we used immunohis-
tochemistry to examine the expression of NR0B1 in the TC71
xenograft. Hematoxylin- and eosin-stained sections of tumors
resected on day 4 of treatment from both treated and control mice
showed viable small round cells characteristic of ESFT (Figure 5, E).
However, the tumors from control mice displayed nuclear and
cytoplasmic staining with an antibody against NR0B1, whereas the
tumors from mithramycin-treated mice showed almost no such
staining (Figure 5, E). Similar to the results obtained in vitro,
immunoblot analysis of tumor lysates revealed induction of
apoptosis, as measured by the cleavage of PARP, and phosphorylation
of H2AX by day 11 in the treated mice but not the control mice
(Figure 5, F).
The mice tolerated mithramycin very well on the dose and
schedule used in this study; we observed only minimal myelosup-
pression, liver enzyme elevations, and minor electrolyte abnormal-
ities, which were not life threatening.
Figure 4. Effect of mithramycin on DNA damage response. A)
Immunoblot analysis. TC32 cells were treated with 100 nM mithramy-
cin. Cells were collected at the indicated times during incubation and
used to make protein lysates, which were subjected to immunoblot
analysis to assess the induction of the DNA damage response as
reflected by the phosphorylation of H2AX (detected with a phosphory-
lation-specific antibody) and phosphorylation of serine 15 of p53 (ph
-p53) and the subsequent degradation of p53. Similar results were
obtained in repeat experiments. B) Confocal microscopy. TC32 cells
were treated with 100 nM mithramycin or phosphate-buffered saline
(control) for 24 hours and immunostained with an antibody that rec-
ognizes phosphorylated H2AX (gH2AX; green), counterstained with
DAPI (blue) to image nuclei, and analyzed by confocal microscopy.
Scale bars = 20 µm.
Table 1. IC50 values for mithramycin in human cancer cell lines*
Cell line Cell type Mean IC50, nM (95% CI)
TC71 ESFT 10 (8 to 13)
A673 ESFT 10 (7 to 12)
TC167 ESFT 11 (11 to 11)
TC32 ESFT 15 (13 to 19)
EW8 ESFT 15 (14 to 19)
CHLA9 ESFT 15 (12 to 15)
MNNG-HOS Osteosarcoma 21 (11 to 40)
U2OS Osteosarcoma 35 (12 to 95)
MD-MBA-231 Breast cancer 20 (12 to 35)
MCF7 Breast cancer 70 (42 to 100)
SKOV3 Ovarian carcinoma >500 (UTD)
* IC50 experiments were performed in 96-well plates using an MTS assay.
Each cell line was assayed in 2–5 independent experiments. CI = confidence
interval; ESFT = Ewing sarcoma family of tumors; IC50 = concentration that
caused a 50% loss of cell viability compared with untreated control cells;
UTD = unable to determine (because the cells did not achieve <1% viability
even at concentrations >500 nM).
Discussion
In this study, we have used a novel screening method to identify
mithramycin as an inhibitor of the EWS-FLI1 transcription
factor. We showed that this drug blocks expression of EWS-FLI1
downstream targets in vitro at both the RNA and protein level and
suppresses protein expression of a well-characterized downstream
target, NR0B1, in vivo. Mithramycin inhibited ESFT cell growth
in vitro with IC50 values ranging from 10 to 15 nM and suppressed
the growth of ESFT xenografts in vivo to 3% of control in the
more sensitive TC32 xenograft model of ESFT.
Ewing sarcoma is a malignant bone tumor of childhood with
an extremely poor prognosis, particularly for high-risk patients.
To improve patient outcomes, recent efforts have been aimed at
developing therapies targeting the EWS-FLI1 oncogenic tran-
scription factor that characterizes this disease (19,20). This goal
stems from the widely validated dependence of this tumor on the
continued expression of EWS-FLI1 to maintain the malignant
phenotype. Most notably, a gene signature approach has been
used to identify cytosine arabinoside as an inhibitor of EWS-FLI1
(19). Another investigation characterized an interaction between
EWS-FLI1 and RNA helicase A and used high-throughput
screening to identify a small molecule (YK-4-279) that blocks this
protein–protein interaction to disrupt EWS-FLI1 activity (20).
To date, neither cytosine arabinoside nor YK-4-279 has shown
activity in the clinic.
The idea of targeting oncogenic transcription factors with
small-molecule inhibitors is attractive from both a theoretical and
a practical perspective. Aside from the likely dependence of a
variety of carcinomas, sarcomas, and leukemia on specific fusion
transcription factors, the success of all-trans retinoic acid and
arsenic for the treatment of high-risk acute promyelocytic leukemia
and ET-743 for myxoid liposarcoma offers further evidence for
optimism regarding transcription factor–targeted small molecules
(49–55). In both cases, the dramatic responses that have been
observed in the clinic are likely due to inhibition of the oncogenic
fusion transcription factor responsible for oncogenesis and pro-
gression for the individual tumor (56–58). However, it was only
after these drugs were found to have activity in the clinic that the

974 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
mechanism was determined to be interference with the activity of
the oncogenic fusion transcription factor.
We have developed a novel method to screen for transcription
factor inhibitors with the goal of inhibiting EWS-FLI1 in ESFT.
The method combines an EWS-FLI1 downstream target pro-
moter reporter cell-based screen with a gene signature multiplex
PCR approach. The assays proved to be robust enough to evaluate
a large library of more than 50 000 compounds and yet specific
enough to allow us to successfully identify and characterize the
compound mithramycin as an inhibitor of EWS-FLI1. The ultimate
validation of both the method and the drug is the observed
response of the TC32 xenograft to drug treatment.
Like most natural products, mithramycin appears to work by
multiple mechanisms in ESFT cells. First and foremost, mithramy-
cin inhibits the transcriptional activity of EWS-FLI1 as measured
by suppression of two different gene signatures and specific
well-established downstream targets both in vitro and in vivo.
The mechanism of this transcriptional interference likely occurs at
the promoter level. Mithramycin is known to bind DNA and sup-
press the activity of specific transcription factors, most notably
SP1. It is known that SP1 cooperates with members of the E26
transformation-specific (ETS) family of transcription factors, in-
cluding FLI1, to both activate and suppress expression of down-
stream targets (59–63). In addition, SP1 and EWS-FLI1 have been
shown to cooperate to activate the VEGF and CCND1 promoters
(64). This cooperative activity is only partially blocked by mutating
the SP1-binding sites in the CCND1 promoter, which suggests
that the activity of both EWS-FLI1 and SP1 is important for the
transcription of CCND1 (64). Currently, it is unclear whether the
EWS-FLI1 suppression reported here results from the direct
blockage of EWS-FLI1 binding, SP1 binding, or both, and ongoing
experiments are aimed at answering this question.
(continued)

jnci.oxfordjournals.org JNCI |Articles 975
We also demonstrated that mithramycin induces DNA damage
in ESFT cells as measured by the phosphorylation and degradation
of p53 and the generation of gH2AX foci. To our knowledge, this
is the first report of mithramycin inducing DNA damage in a cell.
In ESFT cells, the relative contribution of this DNA damage to
the apoptotic process is not clear, given that it has been shown
by chromatin immunoprecipitation that mithramycin blocks the
activation of the downstream effectors of p53, namely PUMA, Bak,
and p21, via an SP1-dependent mechanism (65). The effects of
mithramycin on p21 expression are particularly notable because
Figure 5. Effect of mithramycin on mice harboring TC32 and TC71 ESFT
and MNNG-HOS osteosarcoma xenografts. A) Dose–response effect in
mice bearing TC71 xenografts on day 11. Mice were treated with the
mithramycin at the dose specified on the x-axis on a Monday-
Wednesday-Friday (M-W-F) schedule for 11 days and killed when the
tumor reached 2 cm. Control mice were untreated (n = 20 mice per
group). Each dot represents the tumor of an individual mouse of the
possible 16 in each cohort treated with mithramycin (four mice from
each group were killed according to a predefined schedule to collect
tissue for biochemical analysis). The horizontal lines represent mean
values. The number of mice remaining in each cohort is specified above
each column. There were fewer mice in control and 0.5 mg/kg groups
because the tumors in some of those mice reached the allowable max-
imum size, necessitating killing. The asterisk indicates statistically sig-
nificant difference in mean tumor volume for the 1 mg/kg body weight
per dose cohort vs the control group (P < .001, two-sided Student t test).
B) Growth of ESFT xenografts. Mice were injected into the left gastroc-
nemius muscle with TC32 or TC71 ESFT cells and tumors were allowed
to establish. Mice were subsequently randomly assigned to the treat-
ment (n = 15) or control (n = 15) group; treated mice received mithra-
mycin by intraperitoneal injection at a dose of 1 mg/kg body weight
dose on a M-W-F schedule, and control mice were untreated. The
curves represent the mean tumor volume for mice bearing the TC32
(left) or TC71 (right) xenografts and error bars represent 95% confi-
dence intervals. The asterisks indicate a statistically significant differ-
ence in tumor volumes on day 15 between the treatment and control
groups (P < .001 for both; two-sided Student t test). C) Kaplan–Meier
survival curves of mice bearing ESFT xenografts. The survival of the
mice treated as described above is shown for the mithramycin-treated
mice bearing either the TC32 (left) or TC71 (right) xenografts compared
with the control cohort. D) Growth of MNNG-HOS osteosarcoma xeno-
grafts and survival of mice bearing these tumors. Mice were implanted
with MNNG-HOS xenografts and, once tumors were established, ran-
domly assigned to receive mithramycin by intraperitoneal injection at a
dose of 1 mg/kg body weight dose on a M-W-F schedule or no treat-
ment (control) (n = 15 mice per group). Mean tumor volumes (left) and
the survival curves (right) are shown; error bars represent 95% confi-
dence intervals. The asterisk indicates a statistically significant differ-
ence in tumor volumes on day 30 between the treatment and control
groups (P < .001; two-sided Student t test). E) Immunohistochemical
analysis of TC71 xenograft tumors. Mice bearing TC71 xenograft
tumors were treated with mithramycin at 1 mg/kg per dose on day 1 or
not treated (control) and tumor tissue collected on day 4. Tumor sec-
tions were stained with hematoxylin–eosin (upper panels) and immu-
nostained for NR0B1 (brown, lower panels). The 20× and 40× fields
shown are representative of the entire section. Similar results were
obtained by staining another section. F) Immunoblot analysis of TC71
xenograft tumor lysates. Tumors were collected from control (C) and
mithramycin-treated mice (M), 11 days after intraperitoneal treatment
with mithramycin on an M-W-F schedule was initiated. The lysates
shown were collected from two different control mice and two different
mice in the 1 mg/kg per dose cohort.

976 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
EWS-FLI1 has also been shown to suppress p21 expression by
binding to the promoter (65,66). We showed that treatment of
ESFT cells with mithramycin-induced expression of p21 within
1–3 hours, consistent with a release of EWS-FLI1–mediated
repression of p21. However, no increase in expression of p21
occurred with p53 activation as would be expected with DNA
damage, suggesting that these cells have either a mutated p53 or a
downstream mithramycin-mediated block in expression of p21 in
the setting of p53 activation.
We found that mithramycin is highly cytotoxic to ESFT cells
in vitro, with IC50 values of 10–15 nM in a panel of ESFT cell lines.
We found that mithramycin was slightly more cytotoxic to the
ESFT cell lines than the other cell lines tested particularly when
compared with the carcinoma cell lines. It is notable that the
TC32, TC71, and MNNG-HOS osteosarcoma cells had relatively
similar IC50 values in vitro with a one- to twofold increase in
cytotoxicity observed in the ESFT cell lines compared with the
osteosarcoma MNNG-HOS cell line.
On the other hand, the selective cytotoxicity of mithramycin
for ESFT cell lines was more pronounced in vivo. For example, the
MNNG-HOS osteosarcoma xenograft displayed limited tumor
growth suppression and no improvement in survival for the mice
treated with mithramycin. By contrast, both ESFT xenografts
showed a pronounced and sustained suppression of tumor growth
with mithramycin treatment that translated into prolonged sur-
vival. We believe the in vivo selectivity of mithramycin for ESFT
comes not from the drug but from the dependence of this tumor
on the target, EWS-FLI1. The reasons for the observed differ-
ences in tumor growth between the two ESFT xenografts treated
with mithramycin are of interest and may reflect the marked tumor
heterogeneity that exists both between genetically similar tumors
as well as within a single tumor.
Mithramycin was originally evaluated as an antitumor agent in
the clinic during the 1960s. Although mithramycin showed some
promise in patients with testicular cancer, it was not pursued likely
due to the development of other successful treatment regimens as
well as limitations in supportive care during that era that hindered
management of the side effects of mithramycin (67). Mithramycin
was also tested for the treatment of Ewing sarcoma (68,69). At least
two of five Ewing sarcoma patients treated with mithramycin
achieved widespread clinical responses, and one patient with
metastatic disease achieved a durable complete response lasting at
least 7 years following a course of mithramycin as a single agent
(69). Despite these impressive early clinical results, it is unclear
why the use of mithramycin for Ewing sarcoma was abandoned.
Nevertheless, the strong dependence of ESFT cells on EWS-FLI1
coupled with the evidence in this report that mithramycin inhibits
expression of EWS-FLI1 downstream targets, when considered in
light of these clinical case reports of patients responding to treat-
ment, strongly suggest a potential role for mithramycin in the
treatment of Ewing sarcoma. We are therefore currently obtaining
clinical grade mithramycin to open a clinical trial and reinvestigate
this drug for the treatment of ESFT.
The major limitation of this study is that the screen and subse-
quent experiments were all conducted in cell lines and xenografts of
those cell lines. Furthermore, we did not verify the authenticity of the
non-ESFT cell lines used in this study. In addition, the investigators
were not blinded to the treatment groups in the mouse experiments,
and only one assay (PARP cleavage) was performed to detect apo-
ptosis. Finally, the multiplex PCR was repeated only one time and the
findings were instead verified by alternative methods; the quantati-
tive PCR and microarray experiments both of which confirmed the
results in three independent experiments each.
References
1. Helman LJ, Meltzer P. Mechanisms of sarcoma development. Nat Rev
Cancer. 2003;3(9):685–694.
2. Martens JH, Stunnenberg HG. The molecular signature of oncofusion
proteins in acute myeloid leukemia. FEBS Lett. 2010;584(12):2662–2669.
3. Prensner JR, Chinnaiyan AM. Oncogenic gene fusions in epithelial carci-
nomas. Curr Opin Genet Dev. 2009;19(1):82–91.
4. Delattre O, Zucman J, Plougastel B, et al. Gene fusion with an ETS
DNA-binding domain caused by chromosome translocation in human
tumours. Nature. 1992;359(6391):162–165.
5. Delattre O, Zucman J, Melot T, et al. The Ewing family of tumors—a
subgroup of small-round-cell tumors defined by specific chimeric tran-
scripts. N Engl J Med. 1994;331(5):294–299.
6. Bailly RA, Bosselut R, Zucman J, et al. DNA-binding and transcriptional
activation properties of the EWS-FLI-1 fusion protein resulting from the
t(11;22) translocation in Ewing sarcoma. Mol Cell Biol. 1994;14(5):
3230–3241.
7. Kauer M, Ban J, Kofler R, et al. A molecular function map of Ewing’s
sarcoma. PLoS One. 2009;4(4):e5415.
8. May WA, Gishizky ML, Lessnick SL, et al. Ewing sarcoma 11;22 translo-
cation produces a chimeric transcription factor that requires the DNA-
binding domain encoded by FLI1 for transformation. Proc Natl Acad Sci
U S A. 1993;90(12):5752–5756.
9. Tanaka K, Iwakuma T, Harimaya K, Sato H, Iwamoto Y. EWS-Fli1
antisense oligodeoxynucleotide inhibits proliferation of human Ewing’s
sarcoma and primitive neuroectodermal tumor cells. J Clin Invest. 1997;
99(2):239–247.
10. Maksimenko A, Malvy C. Oncogene-targeted antisense oligonucleotides
for the treatment of Ewing sarcoma. Expert Opin Ther Targets. 2005;9(4):
825–830.
11. Truong AH, Ben-David Y. The role of Fli-1 in normal cell function and
malignant transformation. Oncogene. 2000;19(55):6482–6489.
12. May WA, Arvand A, Thompson AD, Braun BS, Wright M, Denny CT.
EWS/FLI1-induced manic fringe renders NIH 3T3 cells tumorigenic.
Nat Genet. 1997;17(4):495–497.
13. Eliazer S, Spencer J, Ye D, Olson E, Ilaria RL Jr. Alteration of mesodermal
cell differentiation by EWS/FLI-1, the oncogene implicated in Ewing’s
sarcoma. Mol Cell Biol. 2003;23(2):482–492.
14. Richter GH, Plehm S, Fasan A, et al. EZH2 is a mediator of EWS/FLI1
driven tumor growth and metastasis blocking endothelial and neuro-
ectodermal differentiation. Proc Natl Acad Sci U S A. 2009;106(13):
5324–5329.
15. Toretsky JA, Kalebic T, Blakesley V, LeRoith D, Helman LJ. The insulin-
like growth factor-I receptor is required for EWS/FLI-1 transformation
of fibroblasts. J Biol Chem. 1997;272(49):30822–30827.
16. Kovar H, Aryee DN, Jug G, et al. EWS/FLI-1 antagonists induce growth
inhibition of Ewing tumor cells in vitro. Cell Growth Differ. 1996;7(4):
429–437.
17. Dohjima T, Lee NS, Li H, Ohno T, Rossi JJ. Small interfering RNAs
expressed from a Pol III promoter suppress the EWS/Fli-1 transcript in
an Ewing sarcoma cell line. Mol Ther. 2003;7(6):811–816.
18. Toretsky JA, Connell Y, Neckers L, Bhat NK. Inhibition of EWS-FLI-1
fusion protein with antisense oligodeoxynucleotides. J Neurooncol.
1997;31(1–2):9–16.
19. Stegmaier K, Wong JS, Ross KN, et al. Signature-based small molecule
screening identifies cytosine arabinoside as an EWS/FLI modulator in
Ewing sarcoma. PLoS Med. 2007;4(4):e122.
20. Erkizan HV, Kong Y, Merchant M, et al. A small molecule blocking
oncogenic protein EWS-FLI1 interaction with RNA helicase A inhibits
growth of Ewing’s sarcoma. Nat Med. 2009;15(7):750–756.

jnci.oxfordjournals.org JNCI |Articles 977
21. Lombo F, Menendez N, Salas JA, Mendez C. The aureolic acid family of
antitumor compounds: structure, mode of action, biosynthesis, and novel
derivatives. Appl Microbiol Biotechnol. 2006;73(1):1–14.
22. Ward DC, Reich E, Goldberg IH. Base specificity in the interaction of
polynucleotides with antibiotic drugs. Science. 1965;149(689):1259–1263.
23. Remsing LL, Bahadori HR, Carbone GM, McGuffie EM, Catapano CV,
Rohr J. Inhibition of c-src transcription by mithramycin: structure-activity
relationships of biosynthetically produced mithramycin analogues using
the c-src promoter as target. Biochemistry. 2003;42(27):8313–8324.
24. Baker VV, Shingleton HM, Hatch KD, Miller DM. Selective inhibition of
c-myc expression by the ribonucleic acid synthesis inhibitor mithramycin.
Am J Obstet Gynecol. 1988;158(4):762–767.
25. Tagashira M, Kitagawa T, Nozato N, et al. Two novel C-glycosides of
aureolic acid repress transcription of the MDR1 gene. Chem Pharm Bull
(Tokyo). 2000;48(4):575–578.
26. Snyder RC, Ray R, Blume S, Miller DM. Mithramycin blocks transcrip-
tional initiation of the c-myc P1 and P2 promoters. Biochemistry. 1991;
30(17):4290–4297.
27. Van Valen F. Human Cell Culture. London, UK: Kluwer; 1999.
28. Kim SY, Lee CH, Midura BV, et al. Inhibition of the CXCR4/CXCL12
chemokine pathway reduces the development of murine pulmonary
metastases. Clin Exp Metastasis. 2008;25(3):201–211.
29. Palmieri D, Halverson DO, Ouatas T, et al. Medroxyprogesterone acetate
elevation of Nm23-H1 metastasis suppressor expression in hormone
receptor-negative breast cancer. J Natl Cancer Inst. 2005;97(9):632–642.
30. Thompson EW, Brunner N, Torri J, et al. The invasive and metastatic
properties of hormone-independent but hormone-responsive variants
of MCF-7 human breast cancer cells. Clin Exp Metastasis. 1993;11(1):
15–26.
31. Smith MA, Morton CL, Phelps D, Girtman K, Neale G, Houghton PJ.
SK-NEP-1 and Rh1 are Ewing family tumor lines. Pediatr Blood Cancer.
2008;50(3):703–706.
32. Woldemichael GM, Vasselli JR, Gardella RS, McKee TC, Linehan WM,
McMahon JB. Development of a cell-based reporter assay for screening of
inhibitors of hypoxia-inducible factor 2-induced gene expression. J Biomol
Screen. 2006;11(6):678–687.
33. Zhang JH, Chung TD, Oldenburg KR. A simple statistical parameter
for use in evaluation and validation of high throughput screening assays.
J Biomol Screen. 1999;4(2):67–73.
34. Riggi N, Suva ML, Suva D, et al. EWS-FLI-1 expression triggers a
Ewing’s sarcoma initiation program in primary human mesenchymal stem
cells. Cancer Res. 2008;68(7):2176–2185.
35. Riggi N, Cironi L, Provero P, et al. Development of Ewing’s sarcoma
from primary bone marrow-derived mesenchymal progenitor cells. Cancer
Res. 2005;65(24):11459–11468.
36. Castillero-Trejo Y, Eliazer S, Xiang L, Richardson JA, Ilaria RL Jr.
Expression of the EWS/FLI-1 oncogene in murine primary bone-derived
cells results in EWS/FLI-1-dependent, ewing sarcoma-like tumors.
Cancer Res. 2005;65(19):8698–8705.
37. Tirode F, Laud-Duval K, Prieur A, Delorme B, Charbord P, Delattre O.
Mesenchymal stem cell features of Ewing tumors. Cancer Cell. 2007;11(5):
421–429.
38. Kinsey M, Smith R, Lessnick SL. NR0B1 is required for the oncogenic
phenotype mediated by EWS/FLI in Ewing’s sarcoma. Mol Cancer Res.
2006;4(11):851–859.
39. Nishimori H, Sasaki Y, Yoshida K, et al. The Id2 gene is a novel target of
transcriptional activation by EWS-ETS fusion proteins in Ewing family
tumors. Oncogene. 2002;21(54):8302–8309.
40. Tirado OM, Mateo-Lozano S, Villar J, et al. Caveolin-1 (CAV1) is a target
of EWS/FLI-1 and a key determinant of the oncogenic phenotype and
tumorigenicity of Ewing’s sarcoma cells. Cancer Res. 2006;66(20):
9937–9947.
41. Pfaffl MW. A new mathematical model for relative quantification in real-
time RT-PCR. Nucleic Acids Res. 2001;29(9):e45.
42. Gautier L, Cope L, Bolstad BM, Irizarry RA. affy—analysis of Affymetrix
GeneChip data at the probe level. Bioinformatics. 2004;20(3):307–315.
43. Smyth G. Limma: linear models for microarray data. In: Gentleman R,
Carey V, Dudoit S, Irizarry R, Huber W, eds. Bioinformatics and Computational
Biology Solutions Using R and Bioconductor. New York, NY: Springer; 2005:
397–420.
44. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment
analysis: a knowledge-based approach for interpreting genome-wide
expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545–15550.
45. Hancock JD, Lessnick SL. A transcriptional profiling meta-analysis
reveals a core EWS-FLI gene expression signature. Cell Cycle. 2008;7(2):
250–256.
46. Mendiola M, Carrillo J, Garcia E, et al. The orphan nuclear receptor
DAX1 is up-regulated by the EWS/FLI1 oncoprotein and is highly
expressed in Ewing tumors. Int J Cancer. 2006;118(6):1381–1389.
47. Banville DL, Keniry MA, Shafer RH. NMR investigation of mithramycin
A binding to d(ATGCAT)2: a comparative study with chromomycin A3.
Biochemistry. 1990;29(39):9294–9304.
48. Bonner WM, Redon CE, Dickey JS, et al. GammaH2AX and cancer. Nat
Rev Cancer. 2008;8(12):957–967.
49. Hu J, Liu YF, Wu CF, et al. Long-term efficacy and safety of all-
trans retinoic acid/arsenic trioxide-based therapy in newly diagnosed
acute promyelocytic leukemia. Proc Natl Acad Sci U S A. 2009;106(9):
3342–3347.
50. Tallman MS. Treatment of relapsed or refractory acute promyelocytic
leukemia. Best Pract Res Clin Haematol. 2007;20(1):57–65.
51. Casali PG, Sanfilippo R, D’Incalci M. Trabectedin therapy for sarcomas.
Curr Opin Oncol. 2010;22(4):342–346.
52. Garcia-Carbonero R, Supko JG, Manola J, et al. Phase II and pharmaco-
kinetic study of ecteinascidin 743 in patients with progressive sarcomas of
soft tissues refractory to chemotherapy. J Clin Oncol. 2004;22(8):1480–1490.
53. Grosso F, Jones RL, Demetri GD, et al. Efficacy of trabectedin
(ecteinascidin-743) in advanced pretreated myxoid liposarcomas: a
retrospective study. Lancet Oncol. 2007;8(7):595–602.
54. Le Cesne A, Blay JY, Judson I, et al. Phase II study of ET-743 in advanced
soft tissue sarcomas: a European Organisation for the Research and
Treatment of Cancer (EORTC) soft tissue and bone sarcoma group trial.
J Clin Oncol. 2005;23(3):576–584.
55. Delaloge S, Yovine A, Taamma A, et al. Ecteinascidin-743: a marine-
derived compound in advanced, pretreated sarcoma patients–preliminary
evidence of activity. J Clin Oncol. 2001;19(5):1248–1255.
56. Zhu J, Gianni M, Kopf E, et al. Retinoic acid induces proteasome-dependent
degradation of retinoic acid receptor alpha (RARalpha) and oncogenic
RARalpha fusion proteins. Proc Natl Acad Sci U S A. 1999;96(26):
14807–14812.
57. Zhu J, Koken MH, Quignon F, et al. Arsenic-induced PML targeting
onto nuclear bodies: implications for the treatment of acute promyelocytic
leukemia. Proc Natl Acad Sci U S A. 1997;94(8):3978–3983.
58. Forni C, Minuzzo M, Virdis E, et al. Trabectedin (ET-743) promotes
differentiation in myxoid liposarcoma tumors. Mol Cancer Ther. 2009;
8(2):449–457.
59. Turner EC, Kinsella BT. Transcriptional regulation of the human prostacy-
clin receptor gene is dependent on Sp1, PU.1 and Oct-1 in megakaryocytes
and endothelial cells. J Mol Biol. 2009;386(3):579–597.
60. Hempel N, Wang H, LeCluyse EL, McManus ME, Negishi M. The
human sulfotransferase SULT1A1 gene is regulated in a synergistic
manner by Sp1 and GA binding protein. Mol Pharmacol. 2004;66(6):
1690–1701.
61. Lu N, Heuchel R, Barczyk M, Zhang WM, Gullberg D. Tandem Sp1/
Sp3 sites together with an Ets-1 site cooperate to mediate alpha11
integrin chain expression in mesenchymal cells. Matrix Biol. 2006;25(2):
118–129.
62. Jinnin M, Ihn H, Asano Y, Yamane K, Trojanowska M, Tamaki K.
Tenascin-C upregulation by transforming growth factor-beta in human
dermal fibroblasts involves Smad3, Sp1, and Ets1. Oncogene. 2004;23(9):
1656–1667.
63. Ahn J, Ko M, Lee K, Oh J, Jeon SH, Seong RH. Expression of SRG3, a
core component of mouse SWI/SNF chromatin-remodeling complex, is
regulated by cooperative interactions between Sp1/Sp3 and Ets transcription
factors. Biochem Biophys Res Commun. 2005;338(3):1435–1446.
64. Fuchs B, Inwards CY, Janknecht R. Vascular endothelial growth factor
expression is up-regulated by EWS-ETS oncoproteins and Sp1 and may

978 Articles |JNCI Vol. 103, Issue 12 | June 22, 2011
represent an independent predictor of survival in Ewing’s sarcoma. Clin
Cancer Res. 2004;10(4):1344–1353.
65. Koutsodontis G, Kardassis D. Inhibition of p53-mediated transcriptional
responses by mithramycin A. Oncogene. 2004;23(57):9190–9200.
66. Nakatani F, Tanaka K, Sakimura R, et al. Identification of p21WAF1/
CIP1 as a direct target of EWS-Fli1 oncogenic fusion protein. J Biol
Chem. 2003;278(17):15105–15115.
67. Kennedy BJ, Torkelson JL. Long-term follow-up of stage III testicular
carcinoma treated with mithramycin (plicamycin). Med Pediatr Oncol.
1995;24(5):327–328.
68. Kofman S, Medrek TJ, Alexander RW. Mithramycin in the treatment of
embryonal cancer. Cancer. 1964;17(7):938–948.
69. Kofman S, Perlia CP, Economou SG. Mithramycin in the treatment of
metastatic Ewing’s sarcoma. Cancer. 1973;31(4):889–893.
Funding
This project has been funded in whole or in part with federal funds from
the National Cancer Institute, National Institutes of Health, under con-
tract N01-CO-12400. This research was supported in part by the Intramural
Research Program of the NIH, National Cancer Institute, Center for Cancer
Research. Grant support received from the Sarcoma Alliance for Research
through Collaboration Career Development Award (to P.J.G.).
Notes
We would like to thank Susan Garfield for help with the confocal micros-
copy experiments, Xiaolin Wu for hybridizing and staining the microarrays,
Carly Smith and Su Young Kim for help with the immunohistochemistry, Rick
Dreyfuss for obtaining the immunohistochemistry images, Yves Pommier for
helpful discussions, and Kathryn Sciabica for assistance developing the mul-
tiplex PCR assay. The content of this publication does not necessarily reflect
the views or policies of the Department of Health and Human Services, nor
does mention of trade names, commercial products, or organizations imply en-
dorsement by the US Government. The funders had no role in the design of
the study; the collection, analysis, and interpretation of the data; the decision to
submit the article for publication; or the writing of the article.
Affiliations of authors: Molecular Oncology Section (PJG, LBG, CY, LJH),
Tumor and Metastasis Biology Section (AM, CK), Oncogenomics Section
(Q-RC, JK), Pediatric Oncology Branch, Center for Cancer Research, National
Cancer Institute, National Institutes of Health, Bethesda, MD; Molecular
Targets Laboratory, SAIC-Frederick Inc, National Cancer Institute, Frederick,
MD (GMW); Department of Physiology, Johns Hopkins University School of
Medicine, Baltimore, MD (DGC); Genetics Branch, Center for Cancer
Research, National Cancer Institute, National Institutes of Health, Bethesda,
MD (SD); Molecular Targets Laboratory, National Cancer Institute, Center for
Cancer Research, Frederick, MD (JBM).