Cell Stem Cell
Short Article
MEF Promotes Stemness
in the Pathogenesis of Gliomas
Elena Bazzoli,1,2,3,5,9 Teodoro Pulvirenti,4 Moritz C. Oberstadt,1,2,10 Fabiana Perna,5 Boyoung Wee,1,2 Nikolaus Schultz,6
Jason T. Huse,2,7 Elena I. Fomchenko,1,2 Francesca Voza,5 Viviane Tabar,8 Cameron W. Brennan,2,8 Lisa M. DeAngelis,2,3
Stephen D. Nimer,2,5 Eric C. Holland,1,2,8,* and Massimo Squatrito1,2,11,*
1Cancer Biology and Genetics Program
2Brain Tumor Center
3Department of Neurology
4Cell Biology Program
5Molecular Pharmacology and Chemistry Program
6Computational Biology Program
7Department of Pathology
8Department of Neurosurgery
Memorial Sloan Kettering Cancer Center, New York, NY 10021, USA
9Department of Neurological, Neuropsychological, Morphological and Motor Sciences, Azienda Ospedaliera Universitaria Integrata,
37134 Verona, Italy
10Department of Pharmacology, Ernst-Moritz-Arndt University of Greifswald, 17487 Greifswald, Germany
11Present address: FSB Brain Tumor Group, F-BBVA Cancer Cell Biology Programme, Centro Nacional de Investigaciones Oncolo´ gicas
(CNIO), Melchor Ferna´ ndez Almagro 3, Madrid E-28029, Spain
*Correspondence: hollande@mskcc.org (E.C.H.), msquatrito@cnio.es (M.S.)
http://dx.doi.org/10.1016/j.stem.2012.09.012
SUMMARY
High-grade gliomas are aggressive and uniformly
fatal tumors, composed of a heterogeneous population
of cells that include many with stem-cell-like
properties. The acquisition of stem-like traits might
contribute to glioma initiation, growth, and recurrence.
Here we investigated the role of the transcription
factor myeloid Elf-1 like factor (MEF, also known
as ELF4) in gliomas. We found that MEF is highly expressed
in both human and mouse glioblastomas
and its absence impairs gliomagenesis in a PDGFdriven
glioma mouse model. We show that modulation
of MEF levels in both mouse neural stem
cells and human glioblastoma cells has a signiﬁcant
impact on neurosphere formation. Moreover,
we identify Sox2 as a direct downstream target of
MEF. Taken together, our studies implicate MEF as
a previously unrecognized gatekeeper gene in gliomagenesis
that promotes stem cell characteristics
through Sox2 activation.
INTRODUCTION
Malignant gliomas represent the most prevalent primary brain
tumor in adults and inevitably have a poor prognosis. Despite
the implementation of new therapeutic strategies, the median
survival of patients with Glioblastoma multiforme (GBM),
the most aggressive glioma variant, is only 14–16 months
and these tumors remain rapidly and uniformly fatal (Wen and
Kesari, 2008).
GBMs are very heterogeneous tumors that contain both
neoplastic and nonneoplastic cells, including endothelial,
stromal, and inﬂammatory cells (Charles et al., 2011). A fraction
of cells within the tumor, identiﬁed as glioma stem-like cells
(GSCs), share some common features with normal neural stem
cells (NSCs); they are multipotent and have the property of
self-renewal. These cells may either derive from adult undifferentiated
stem and progenitor cells or could acquire stem-like
properties as a result of the genetic alterations that promote
the tumorigenic process (Holmberg et al., 2011). When implanted
into the brain of immunodeﬁcient animals, these GSCs
are capable of generating new tumors at high efﬁciency (Galli
et al., 2004). Moreover, GSCs are remarkably resistant to the
chemotherapy and radiotherapy (Dean et al., 2005) used as
standard ﬁrst-line treatment of patients with malignant gliomas.
Indeed, acquisition of stem-like characteristics likely contributes
to the malignant nature of high-grade gliomas and may be
responsible for the initiation, growth, and recurrence of these
tumors.
Myeloid Elf-1 like factor (MEF, also known as ELF4) is a
member of the ETS family of transcription factors, which
contains over 30 family members. Several ETS proteins can
function as oncogenes and show aberrant expression in solid
tumors as well as in hematological malignancies (Sashida
et al., 2010). While MEF has been proposed to function as
a tumor suppressor gene in some contexts, it could contribute
to tumor formation in mice as well as in humans (Mikkers et al.,
2002) (Sashida et al., 2010). Insight into its mechanism of action
has come from studies done in ﬁbroblasts demonstrating MEF’s
ability to induce transformation by stimulating Mdm2 expression,
thereby downregulating p53-dependent responses, and by
inhibiting activation of Ink4a, thereby allowing unrestrained
phosphorylation of the retinoblastoma (Rb) protein (Sashida
et al., 2009). Some of the ETS proteins are known to play
836 Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc.
a role in brain tumors (Uht et al., 2007), though so far no published
data are available regarding the role of MEF in normal
brain or in brain tumor biology. Here we show that MEF is highly
expressed in GBMs and it contributes to gliomagenesis by
promoting stem cell traits through direct activation of Sox2
expression.
RESULTS
MEF Is Highly Expressed in Human Gliomas and Its Loss
Impairs Glioma Formation in Mice
GBM represents a heterogeneous disease and recent genomic
analyses have quantiﬁed the expression level of a wide variety
of genes (Cancer Genome Atlas Research Network, 2008). We
analyzed the data set from The Cancer Genome Atlas (TCGA)
for GBMs and found MEF expression signiﬁcantly elevated in
the tumor samples (n = 195) as compared to nontumor brain
tissue (n = 10) (p < 0.0001; Student’s t test) (Figure 1A and Table
S1 available online), without evidence for ampliﬁcation at the
genomic level. Because MEF has previously been shown to
block the p53 pathway, we examined the p53 status of these
A B
C
D
E
Figure 1. MEF in Human and Mouse
Gliomas
(A) MEF expression in the TCGA data set obtained
from human GBM samples (n = 195) and nontumor
brain tissue (n = 10) (***p = 2.366eÀ09
, Student’s
t test).
(B) Kaplan Meier survival curves of TCGA GBM
patients: low level of MEF (calculated as less than
one standard deviation from the mean of diploid
tumors) signiﬁcantly correlates with better overall
survival (***log-rank p value, p = 0.000386).
(C) Top panel: MEF mRNA level detected by RT
PCR is higher in PDGF-driven mouse GBMs (T)
compared to normal tissue (N). Data are normalized
to GAPDH expression. Results are presented
as mean ± SD (***p < 0.0001, Student’s t test).
Bottom panel: western blot showing that MEF
protein level is higher in PDGF-driven mouse
GBMs compared to normal tissue.
(D) Kaplan Meier survival curves of PDGF gliomas
generated in Ntv-a Mef+/+
versus MefÀ/À
mice.
MEF loss increases overall survival and reduces
gliomagenesis (log-rank p value, *p = 0.02).
(E) Left panel: mouse gliomas lacking MEF show
signiﬁcantly lower percentage of high-grade
versus low-grade tumors (Fisher’s exact test,
*p = 0.027); right panel: contingency table presenting
the number of high-grade (HG) tumors
versus low-grade (LG) tumors in the two different
genetic backgrounds.
See also Figure S1.
tumors and found no correlation between
MEF expression levels and p53 mutation
or deletion, suggesting that MEF may
function in a p53-independent manner in
glioma (Figure S1A available online).
A search conducted in the Oncomine
website (https://www.oncomine.org//) revealed
that in the Sun data set (Sun et al.,
2006), which includes 81 human GBMs (grade IV), 25 anaplastic
astrocytomas (grade III), and 50 oligodendrogliomas (grade II),
MEF expression is signiﬁcantly higher in GBMs than in lower
grade gliomas (p < 0.0001) (Figure S1B). To conﬁrm these data
we analyzed MEF expression by qPCR in 25 human glioma
surgical samples from Memorial Sloan-Kettering Cancer Center
(MSKCC) (six oligodendrogliomas, seven anaplastic oligodendrogliomas,
two anaplastic astrocytomas, and ten glioblastomas)
and found that MEF is signiﬁcantly elevated in high-grade
versus low-grade gliomas (p = 0.02, Student’s t test) (Figure
S1C). However, this difference might reﬂect a lower ratio of
tumor cells versus nontumor cells in the low-grade samples.
Lower levels of MEF were associated with a better prognosis
in a small cohort of acute myeloid leukemia patients (Fukushima
et al., 2003). We examined the TCGA GBM patient survival
data and found that low levels of MEF signiﬁcantly correlated
with better overall survival in GBMs as well (log rank test,
p = 0.0008; Figure 1B). Gene expression proﬁling studies have
identiﬁed four molecular subclasses of GBMs based on transcriptional
signatures: Classical, Mesenchymal, Proneural and
Neural. Each of these subtypes has been associated with
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc. 837
speciﬁc signaling alterations, such as EGFR, Ras, and PDGFR
pathway activation, respectively (Verhaak et al., 2010). When
the patients were stratiﬁed according to GBM subtypes, the
Proneural subclass showed the lowest level of MEF expression
(Figure S1D) and, within this subtype, lower levels of MEF also
correlated with better overall survival (Figure S1E). When we
looked at the IDH1 status, we found that the tumors with the
lowest MEF levels within the Proneural group were predominantly
IDH1 mutant tumors, whereas IDH1 wild-type Proneural
tumors had a range of MEF levels similar to the other tumor types
(data not shown).
To determine whether MEF plays an active role in gliomagenesis,
we used the RCAS/PDGF mouse glioma model, which
closely resembles the human Proneural GBM subtype. The
RCAS/tv-a system utilizes avian leukosis virus based vectors
(RCAS) to mediate gene transfer into somatic cells, engineered
to be transgenic for its receptor (tv-a). Speciﬁcally, we used Nestin
tv-a (Ntv-a) mice, where the tv-a receptor is under the control
of the Nestin promoter, a well-known marker of progenitor and
neural/glial cells. First, we evaluated MEF expression in an
A B
DC
E
0
1
2
3
4
5
6
7
0 1 3 5
#ofcellsx10
Days
Mef+/+
Mef-/-
0
10
20
30
40
50
0 1 3 5
Days
p53-/- Mef+/+
p53-/- Mef-/-
0
10
20
30
40
50
0 1 3 5 7
Days
Ctrl
MEF
p53-/- primary brain culture U87MG
0
50
100
150
200
250
0 1 3 5 7
Ctrl
MEF
0
20
40
60
80
0 1 3 5
Days
Ctrl
MEF
T98G
Days
**
*
*
**
***
**
**
***
***
***
**
*
**
**
**
*
4
#ofcellsx104
#ofcellsx104
#ofcellsx104
#ofcellsx104
Figure 2. Mef Promotes Proliferation of
both Primary Brain Cultures and Human
Glioma Cell Lines
(A and B) Growth curve of primary brain cultures
derived from MefÀ/À
, Mef+/+
(A) and p53À/À
/MefÀ/À
,
p53À/À
/Mef+/+
(B) newborn mice.
(C–E) Growth curve of primary brain cells derived
from p53À/À
newborn mice (C) and human U87MG
and T98G glioma cell lines (D and E).
Results are presented as mean ± SD from
a representative of three experiments performed
in triplicate. *p < 0.05; **p < 0.001; ***p < 0.0001
(Student’s t test). See also Figure S2.
Ink4a/Arf null background, in which
PDGF is able to uniformly generate highgrade
gliomas that share hallmark histological
features with human GBMs (Dai
et al., 2001). Using qPCR and Western
blot analysis, we found higher levels of
MEF mRNA and protein in these tumors
as compared to the contralateral normal
brain tissues (Figure 1C).
Next, to determine if lack of MEF
impacts on PDGF-induced gliomagenesis,
MefÀ/À
mice were crossed with
Ntv-a mice (which carry a wild-type
Ink4a/Arf locus) to generate Ntv-a Mef+/+
and Ntv-a MefÀ/À
mice. A survival analysis
of the two cohorts revealed that
loss of MEF signiﬁcantly impaired
PDGF-induced glioma formation, with
MEF null mice living an average of
129 days (n = 22) and wild-type mice
living 59 days (n = 24) (p < 0.02, log rank
test) (Figure 1D). The tumors were then
scored and graded by histological
features (see Experimental Procedures
for details). Tumors lacking MEF showed less aggressive
features, with signiﬁcantly fewer high-grade gliomas (p = 0.02,
Fisher’s exact test) (Figure 1E). Thus, MEF can affect both glioma
formation and progression.
MEF Promotes Proliferation of both Mouse Primary
Brain Cultures and Human Glioma Cell Lines
As MEF has been shown to promote the transition of cells
from G1 to S (Liu et al., 2006; Sashida et al., 2009) we evaluated
its effect on proliferation, using both mouse primary brain
cultures and human glioma cell lines. The primary cultures
were generated from the whole brain of newborn pups, and
to investigate potential p53-independent effects, we used cells
from Mef+/+
and MefÀ/À
mice and also from p53À/À
/ Mef+/+
and p53À/À
/MefÀ/À
mice. Cells lacking MEF grew more slowly
than the control cells (Figures 2A and 2B), while its overexpression
in p53À/À
cells (Figure S2A) increased proliferation
(Figure 2C). Similarly, the overexpression of MEF in the human
U87MG and T98G glioma cell lines (Figures S2B and S2C) resulted
in higher rates of cell proliferation (Figures 2D and 2E).
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
838 Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc.
These data conﬁrm the role of MEF in promoting the growth of
both mouse primary brain cultures and human glioma cell
lines.
MEF Promotes Stem Cell Characteristics in Mouse NSCs
and Human Glioma Cell Lines
In recent years, an increasing number of studies have investigated
the connection between malignancy and ‘‘stemness,’’
focusing on how stem/progenitors cells, as well as neoplastic
cells, change their properties during the process of malignant
transformation. Several transcription factors are known to
induce pluripotent stem cells from differentiated cells, as well
as to maintain multipotency of NSCs (Patel and Yang, 2010).
Because MEF contributes to glioma formation and aggressiveness
in vivo, we explored the possible role of MEF in promoting
stem cell characteristics.
NSCs and GSCs can be grown as spheres in culture, in the
absence of serum but in the presence of basic ﬁbroblast growth
factor (bFGF) and epidermal growth factor (EGF). Although
sphere-forming assays cannot be considered an exact readout
of in vivo stem cell activity, they are useful to measure the
in vitro potential of cells to exhibit stem-cell-like traits (Pastrana
et al., 2011); therefore, we investigated whether MEF could inﬂuence
the formation of neurospheres.
Freshly isolated MefÀ/À
and Mef+/+
cells from postnatal nonneoplastic
brains were plated at different cell densities (10, 5,
and1 cells/ml) in24-wellplatesandgrowninneurospheremedium
for 2 weeks to allow the generation of neurospheres. We then performed
serial passages over the course of 8 weeks, generating
secondary, tertiary, and quaternary spheres. Spheres were mechanically
dissociated every 2 weeks, and at each passage,
sphere number was assessed. We found that at a density of
1 cell/ml, lack of MEF signiﬁcantly decreased generation of
secondary, tertiary, and quaternary neurospheres (Student’s
t test, p < 0.0001, at each passage) (Figure 3A). Similar results
were obtained at higher cell concentrations (5 and 10 cells/ml)
(Figure S3A). Moreover, cells lacking MEF underwent morphological
changes over time, with some of them becoming attached to
the bottom of the wells and presenting short elongation
processes (Figure S3B). To understand whether the decreased
sphere forming ability seen in MefÀ/À
cells was accompanied by
increased differentiation, we grew cells from dissociated spheres
in 5% serum medium without FGF and EGF to evaluate their
differentiation potential. Phase contrast images revealed clear
differences in the morphology of the Mef+/+
and MefÀ/À
cells, suggesting
that differentiation is more profound in cells lacking MEF
(Figure 3B). Immunoﬂuorescence staining showed a signiﬁcant
reduction in Nestin+ cells (p = 0.0006), together with a signiﬁcant
increase in both Olig2 and GFAP+ cells (p < 000.1 and p = 0.0481,
respectively, Student’s t test) in MefÀ/À
cells (Figures 3C and 3D).
These differences were detected also by Western blot analysis
(Figure 3E). Interestingly, most of the GFAP-positive MefÀ/À
cells
showed the stellate morphology typical of astrocytes and
a brighter signal, which was conﬁrmed by increased GFAP
protein levels measured by Western blot (Figure 3E). We failed
to detect Tuj-1+ cells in three of the four MefÀ/À
samples, while
an average of 14.4% Tuj-1+ cells was found in four different
Mef+/+
samples (data not shown), which suggests that MEF might
also contribute to differentiation toward the neuronal lineage.
To determine whether lack of MEF could inﬂuence the stem
cell potential under more strict conditions, we generated primary
cell cultures obtained from the whole brain of newborn pups,
forced them to grow in serum for two passages, and switched
them to neurosphere medium. To evaluate possible p53-independent
effects, primary brain cultures were isolated from
different genetic backgrounds (Mef+/+
, MefÀ/À
, p53À/À
/Mef+/+
,
and p53À/À
/ MefÀ/À
). Four days after the switch to neurosphere
medium, cells lacking MEF failed to form spheres, while both the
Mef+/+
and the p53À/À
/Mef+/+
cells were able to do so (p = 0.0003
and p = 0.0001, respectively, Student’s t test) (Figure 3F). The
impaired neurosphere formation of the p53À/À
/MefÀ/À
cells
could be rescued by MEF re-expression (p = 0.00145, Student’s
t test) (Figure 3G and Figure S3C).
We then investigated whether MEF overexpression could
induce reprogramming signals under the same strict conditions,
inﬂuencing the sphere-forming ability of different primary and
glioma cells. We ﬁrst overexpressed MEF in primary brain
cultures obtained from p53 null pups, using retroviral vectors
(Figure S2A). After antibiotic selection, the cells were grown in
serum for two passages and then switched to neurosphere
medium. We found that cells overexpressing MEF formed
more neurospheres than the control cells (p = 0.0012 at 24 hr,
p = 0.0017 at 72 hr; Student’s t test) (Figure 3H). Similar differences
upon MEF overexpression were observed in the U87MG
and T98G human glioma cell lines (p < 000.1 and p < 000.1,
respectively, Student’s t test) (Figure 3I).
Lastly, we evaluated the impact of MEF on stem-like properties
using human primary GSCs, with various genetic proﬁles.
Tumor samples, isolated from patients suffering from GBMs
and undergoing surgery at MSKCC, were dissociated and
initially grown in neurosphere medium as a monolayer on
plastic cell culture dishes coated with 10 ng/ml laminin. Speciﬁcally,
we used the following cells: GBM1 (++PDGFRA, +EGFR,
MET, CDK6, ÀPTEN, ÀCDKN2A), GBM3 (++EGFR, +MET,
CDK6, ÀCDKN2A,PTEN, carrying EGFRvIII mutation), and
GBM543 (++PDGFRA, ++CDK4, +(EGFR,MET,CDK6), ÀPTEN)
(Ozawa et al., 2010; Pulvirenti et al., 2011). Using lentiviral
vectors to express shRNAs, we silenced MEF expression (Figure
S3D) and found that decreased MEF levels lead to a signiﬁcant
decrease in neurosphere formation in all of these three GSC
lines (p < 0.0001 for both shRNA#1 and shRNA#2), as compared
to nontargeting shRNA control (Figure 3J). Moreover, when we
performed limiting dilution neurosphere assays, we conﬁrmed
that MEF knockdown leads to a reduced frequency of sphereforming
cells (p < 000.1 for both shRNAs in each GSC line,
ELDA software) (Figure 3K).
Taken together, these data suggest a role for MEF in
promoting stem-cell-like features in both primary brain cultures
and glioma cell lines.
Sox2 Gene Is a Direct Target of MEF and Rescues
Sphere-Forming Ability in MEF-Defective Cells
To investigate possible mechanisms behind the ability of MEF to
promote stem-cell traits, we measured the expression level of
several genes related to pluripotency, including Sox2, Oct4,
Nanog, Klf4, Hes1, and Hey1, in MEF-transduced p53À/À
primary brain cells. Among these genes, we found that Sox2
and Oct4 expression was signiﬁcantly increased by MEF. We
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc. 839
Mef+/+
Mef -/-
A
***
***
***
0.0
0.2
0.4
0.6
0.8
1.0
5
10
15
B C
Mef+/+
/HoechstNestin Merged
/HoechstNestin Merged
Mef+/+Mef-/GFAP/Hoechst
Olig2/Hoechst
GFAP/Hoechst Olig2/HoechstSecondary Tertiary Quaternary
Mef-/-
%ofpositivecells
Mef+/+
Mef -/-
D
Nestin Olig2 GFAP
***
Nestin
GFAP
Actin
Olig2
Actin
E
Mef+/+ Mef-/Mef+/+
Mef-/-
Mef+/+
Mef-/p53-/-
Mef+/+
p53-/- Mef-/F
G H
pBabectrl
24h
pBabeMEF
72h
pBabectrl
U87MG
pBabeMEF
T98G
I
-/-p53 Mef-/pBabe
ctrl
p53-/- Mef-/pBabe
MEF
30
20
10
0
**
Avg.#ofNS/
fieldofview
p53-/-
M
ef-/-
ctrl
p53-/-
M
ef-/-
M
EF
0
50
100
150
NSnumber/plate
J
**
**
**
5
10
0
Avg.#ofNS/
fieldofview
***
0
10
20
30
***
M
ef+/+
M
ef-/-
p53-/-
M
ef-/-
p53-/-
M
ef+/+
control vector
shRNA#1
shRNA#2
24h
MEFctrl
72h
ctrl ctrl MEF ctrl MEF
U87MG T98G
***
*
0
20
40
60
80
100
MEF
0
20
60
40
80
100
0
50
100
150
200
●
●
●
●
●
●
●
●
●
●
●
●
0 20 40 60 80 100
−1.5−1.0−0.50.0
number of cells / 100 l
rlogfaction
wellwithoutspheres
●
●
●
sh Ctrl
sh#1
sh#2
●
●
●
●
●
●
●
●
●
●
●
●
0 20 40 60 80 100
−1.5−1.0−0.50.0
●
●
●
●
●
●
●
●
●
●
●
●
0 20 40 60 80 100
−1.2−0.8−0.40.0
GBM1 GBM3 GBM543
**
**
***
***
number of cells / 100 lnumber of cells / 100 l
●
●
●
sh Ctrl
sh#1
sh#2
●
●
●
sh Ctrl
sh#1
sh#2
GBM1 GBM3 GBM543
K
fieldofview
Avg.#ofNS/
fieldofview
Avg.#ofNS/
Avg#ofNS/well
Figure 3. MEF Promotes Stem Cell Traits of both Primary Brain Cultures and Glioma Cell Lines
(A) Neurospheres isolated from newborn pups, with Mef+/+
and MefÀ/À
genetic background, respectively: MEF loss led to impaired neurosphere formation after
serial passages.
(B) Phase contrast images showing morphological changes between Mef+/+
and MefÀ/À
brain cells grown in 5% serum, without bFGF and EGF, for 5 days to favor
differentiation.
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
840 Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc.
observed a 3.6-fold increase in Sox2 gene expression and a 3fold
increase in Oct4 expression level (p = 0.02 and p = 0.013,
respectively, Student’s t test) (Figure 4A). Similarly, in U87MG
and T98G cells, MEF induced a 2.1 (p = 0.017) and 1.6 (p =
0.018) -fold increase in Sox2 levels and a 1.8 (p = 0.0025) and
1.6 (p = 0.0199) -fold increase in Oct4 levels (Figure 4B).
The Sox2 promoter is known to contain ETS binding sites
(Wiebe et al., 2000), though they have not been identiﬁed in the
Oct4 promoter; therefore, we focused on Sox2 as a possible
direct downstream target of MEF. We analyzed Sox2
expression in Mef+/+
versus MefÀ/À
and p53À/À
/Mef+/+
versus
p53À/À
/MefÀ/À
primary brain cells and found a signiﬁcant reduction
in Sox2 levels in the absence of MEF (p < 0.0001, Student’s
t test) (Figures 4C and 4D). To exclude the possibility that developmental
adaptation to the lack of MEF resulted in decreased
Sox2 expression, we acutely knocked down MEF in p53 null
primary brain cultures ($60%) and found a signiﬁcant reduction
in Sox2 expression (p = 0.015 and p = 0.0059, respectively,
Student’s t test) (Figure 4E). Thus, both the acute loss and
chronic absence of MEF substantially decreases Sox2
expression.
To test whether MEF can activate the Sox2 promoter, we
transiently transfected a Sox2 promoter-driven luciferase
reporter plasmid (Kuwabara et al., 2004) into p53 null primary
brain cells, together with MEF or an empty vector control. As
shown in Figure 4F, MEF induced a greater than 3-fold increase
in luciferase activity from the Sox2 promoter (p < 0.0001,
Student’s t test) while having no effect on Hes1 promoterdriven
luciferase activity. We also investigated whether MEF
activates the Sox2 promoter in human glioma cell lines and
found a similar upregulation (p < 0.0001, Student’s t test)
(Figures 4G and 4H).
Having identiﬁed Sox2 as a transcriptional target of MEF, we
investigated whether Sox2 overexpression could rescue the
decrease in neurosphere formation seen in cells lacking MEF.
We overexpressed Sox2 in p53À/À
/MefÀ/À
primary brain cultures
(Figure S4B) and grew them in neurosphere conditions. The
absence of neurospheres seen in cells lacking MEF was rescued
by Sox2 overexpression (p = 0.00018, Student’s t test), indicating
that Sox2 acts downstream of MEF to promote neurosphere
formation (Figure 4I).
To conﬁrm that Sox2 gene regulation by MEF was direct, we
performed Chromatin Immunoprecipitation (ChIP) assays using
T98G glioma cells that were transduced with a pBabe retroviral
vector expressing a FLAG-tagged MEF cDNA or a control vector.
Using an anti-FLAG ChIP assay, we found that in human glioma
cells, MEF protein is recruited to speciﬁc regions of the Sox2
gene (which covered À1.964, À1.672, and À1.324 kb from the
transcription starting site [TSS]; p values: 0.0009, 0.003,
0.0008, respectively). No direct binding was observed at the
farthest region (À3.841 kb) (Figure 4J).
DISCUSSION
The regulation of stemness is of interest to many disciplines,
including developmental biology, regenerative medicine, degenerative
disease, and cancer. Normal stem cells are known to
play critical roles in tissue development, differentiation, and
organogenesis (Vierbuchen et al., 2010). Several transcription
factors, such as Notch, Id1, Sox2, Oct4, and others, have been
reported to control different aspects of stemness, inﬂuencing
the delicate balance between stem cell maintenance and the
promotion of stem cell differentiation. In recent years, many
studies have shown that genes involved in normal stem cell
biology also play a relevant role in tumorigenesis (Nam and Benezra,
2009) (Venere et al., 2011).
Malignant gliomas are highly aggressive cancers composed of
a heterogeneous cell population, a fraction of which exhibits
stem-like characteristics (Vescovi et al., 2006). Whether GSCs
arise from normal NSCs or from neoplastic cells that have
acquired stem cell traits is still unclear; however, it has been
proposed that the acquisition of stem cell characteristics
through the activation of stem cell gene signatures can confer
malignant potential to gliomas (Holmberg et al., 2011).
We have shown that MEF, a member of the ETS family of transcription
factors, contributes to gliomagenesis and promotes
stem-like characteristics. MEF is highly expressed in both
human and mouse GBMs, and GBM patients with low levels of
MEF show a signiﬁcantly better overall survival. This ﬁnding
was replicated in the RCAS/PDGF model where gliomas that
lack MEF had less aggressive histological features and better
overall survival.
Genes overexpressed in cancer can affect many different
biological processes including growth, stress response, block
of apoptosis, and stemness. Uncontrolled proliferation is a key
aspect of cancer cell behavior and we found that MEF promotes
the proliferation of both mouse primary brain cultures and human
glioma cell lines, contributing to the pathogenesis of gliomas.
(C and D) Immunoﬂuorescence staining and quantiﬁcation of Mef+/+
and MefÀ/À
brain cells grown in the above condition to favor differentiation. Positive staining
is quantiﬁed as percentages of total Hoechst+ cells per image (10 ﬁelds of images were taken from four different cell lines for each genotype, MefÀ/À
and Mef+/+
);
scale bars indicate 50 mm.
(E) Western blot comparing MefÀ/À
to Mef+/+
brain cells: lack of MEF lead to decreased Nestin and increased Olig2 and GFAP protein levels.
(F) Mef+/+
, MefÀ/À
, p53À/À
/Mef+/+
, and p53À/À
/MefÀ/À
primary brain cultures grown in neurosphere conditions. Scale bars indicate 100 mm.
(G) Re-expression of MEF in p53À/À
MefÀ/À
cells rescues their ability to form neurospheres when cultured in neurosphere conditions.
(H) Micrograph showing p53À/À
primary brain cells transfected with pBabe-empty and pBabe-MEF retroviral vectors, cultured in neurosphere conditions. Bar
graphs on the bottom show average neurosphere number per ﬁeld of view.
(I) MEF overexpressing human U87MG and T98G cell lines, cultured in neurosphere conditions, form more neurospheres compared to control. Bar graphs on the
bottom show average neurosphere number per ﬁeld of view.
(J) Graph bar showing decreased neurosphere formation after MEF knockdown in patient-derived human primary GBMs (GBM1, GBM3, and GBM543).
(K) Graph representing limiting dilution neurosphere assays in patient-derived human primary GBMs (GBM1, GBM3, and GBM543) infected with pGipz control
vector and pGipz MEF shRNAs #1 and #2.
Results are presented as mean ± SD from a representative of three experiments performed in triplicate, unless otherwise speciﬁed. *p < 0.05; **p < 0.005;
***p < 0.0001 (Student’s t test). See also Figure S3.
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc. 841
Another key feature of glioma malignancy is the activation of
neural and pluripotent stem cell gene expression signatures.
We found that mouse primary brain cells that lack MEF and
silencing of MEF in human primary glioblastoma stem-like cells
each lead to an impaired capacity to generate spheres, together
with an increased differentiation potential. Conversely, MEF
overexpression increases neurosphere formation of both mouse
primary brain cells and glioma cell lines. Collectively, our data
suggest that MEF plays a role in promoting stem-like traits,
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
Sox2 Oct4 Nanog Klf4 HES1 HEY1
relativemRNAexprenoiss
p53-/- ctrl
p53 MEF-/-
*
*
ns
nsnsns
A B
0
5
4
3
2
1
7
6
realtevimRNAexprenoiss
ctrl
MEF
***
**
*
* ***
U87MG T98G U87MG U87MGT98G T98G
MEF Oct4
0
0.5
1
1.5
0
0.5
1
1.5
0
1
0.8
0.6
0.4
0.2
relative2xoSexprenoiss
realtevi2xoSexprenoiss
realtevi2xoSexprenoiss
C D E
*** ***
** *
F G H
Mef +/+
Mef -/-
/M
ef+/+
p53-/-
/M
ef
p53-/-
-/MEF
Sox2
Sox2
p53-/-
ctrl
p53
shRNA MEF
-/-
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
ctrl MEF ctrl MEF
0
5
4
3
2
1
ctrl MEF ctrl MEF
0
1
3
4
5
6
2
ctrl MEF ctrl MEF
Sox2
promoter
Hes-1
promoter
Sox2
promoter
Hes-1
promoter
Sox2
promoter
Hes-1
promoter
***
*** ***
ns
ns
ns
p53-/- T98G U87
I
p53-/- /Mef -/- ctrl p53-/-/Mef-/- Sox2
0
10
20
30
40
relativeluciferaseactivity
relativeluciferaseactivity
relativeluciferaseactivity
p53-/-/M
ef-/-ctrlp53-/-/M
ef-/-Sox2
Avg#ofNS/
fieldofview
***
normalizedby10%input
0
2
1.5
1
0.5
IgG
MEF-Flag
**
*
**
Reg1 Reg2 Reg3 Reg4
-4Kb -3Kb -2Kb -1Kb
Reg1 Reg2 Reg3 Reg4
TSS
J
Figure 4. Sox2 Gene Is Regulated by MEF
(A and B) MEF overexpression in p53À/À
primary
brain cells as well as in U87MG and T98G human
glioma cell lines led to increased Sox2 expression.
Data are normalized to GAPDH expression.
(C and D) Lack of MEF decreases Sox2 expression.
Data are normalized to GAPDH expression.
(E) Acute knockdown of MEF in p53À/À
primary
brain cells led to decreased Sox2 expression. Data
are normalized to GAPDH expression.
(F–H) Dual luciferase assays showing that MEF
activates Sox2 promoter in p53À/À
primary brain
cells as well as T98G and U87MG human glioma
cell lines.
(I) Sox2 overexpression in p53À/À
/MefÀ/À
primary
brain cultures grown in neurosphere conditions
rescues their ability to form neurospheres. Scale
bar indicates 100 mm. Bar graph shows average
neurosphere number per ﬁeld of view.
(J) ChIP experiments were performed on T98G
glioma cells using antibody to Flag. Plotted values
are relative enrichments (y axis) to 10% input and
measured for sites in the Sox2 promoter (x axis).
Antibody against IgG was used as a nonspeciﬁc
control.
Results are presented as mean ± SD from
a representative of three experiments performed in
triplicate. *p < 0.05; **p < 0.001; ***p < 0.0001
(Student’s t test). See also Figure S4.
which might reﬂect changes in the stem
cell signature of both neoplastic and nonneoplastic
cells.
We have further implicated MEF in
promoting stem cell traits through direct
activation of Sox2 expression. The HMGbox
transcription factor Sox2 is known to
play an important role in maintaining
stem cell self-renewal within the central
nervous system (CNS) and this activity is
present in gliomas as well (Pevny and Nicolis,
2010). Sox2 has been described to
act as an oncogene in different human
cancers including gliomas. While the role
of Sox2 in normal brain and glioma cells
has been well documented, the transcription
factors that activate its expression in
these contexts are less clear (Denysenko
et al., 2010; Ikushima et al., 2009). We
identiﬁed Sox2 as a direct target of MEF
in mouse primary brain cells and human
glioma cell lines, because both acute
and chronic modulation of MEF levels affected Sox2 gene
expression and promoter activation. MEF directly binds to Sox2
promoter in T98G glioma cells. We observed that overexpression
of Sox2 in p53À/À
/MefÀ/À
cells was sufﬁcient to rescue the
impaired ability of these cells to form neurospheres. Thus, our
data suggest that Sox2 is downstream of MEF and may be
responsible for its ability to modulate stem-like characteristics.
Given the cellular heterogeneity and molecular complexity of
gliomas, it is unclear in which contexts the stem cell property
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
842 Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc.
of self-renewal might represent a feature associated with
increased tumorigenic potential (Barrett et al., 2012). Additionally,
the so called ‘‘glioma stem-like cells’’ within different tumors
may vary in terms of proliferation rates, which might impact
their aggressiveness regardless of their self-renewal properties.
To date, it is often difﬁcult to distinguish between the contributions
of proliferation versus self-renewal to the process of
gliomagenesis; nonetheless, self-renewal inevitably entails
proliferative events. Adding to the intricacy, acquisition of
reprogramming signals generated by alterations in speciﬁc
genes or pathways (Notch, Id1, Sox2, Oct4, etc.) might differentially
impact on these two diverse aspects.
Our work deﬁnes the role of MEF in promoting stem cell
features in both primary mouse brain cells and human glioblastoma
cells via direct regulation of Sox2 expression, providing
insights into the multifaceted regulation of stemness. Further
work is necessary to better deﬁne the MEF/Sox2-mediated
acquisition of stem cell traits and whether it modulates the
response to chemotherapy and radiotherapy.
EXPERIMENTAL PROCEDURES
TCGA Analysis
TCGA data was downloaded from the TCGA Data Portal (https://tcga-data.nci.
nih.gov/) or the cBio Cancer Genomics Portal (Cerami et al., 2012) (http://
cbioportal.org/). Subtype information was retrieved from Verhaak et al.
(2010). Low MEF expression in the TCGA data set was deﬁned using the
median expression data (median of three platforms, see Verhaak et al.,
2010) as one standard deviation lower than the mean of all tumors. All other
analyses were performed using data from the Affymetrix U133 microarray
platform. Expression data was available for 195 tumors and 10 normal brain
samples (Table S1).
Generation of Primary Brain Cultures and Primary Brain
Neurosphere and Neurosphere Formation Assay
The Mef+/+
, MefÀ/À
, p53À/À
, and p53À/À
MefÀ/À
primary brain cultures were
prepared with mechanical dissociation of the whole brain from newborn
mice. The cells were then ﬁltered through a 70 mm Nylon strainer and plated
in 10 cm culture dishes (with the procedure derived with minor modiﬁcations
from a previously described experimental method) (Dai et al., 2001). Mouse
primary brain cultures, as well as T98G and U87MG human glioma cell lines,
were plated at 20 cells/ml in neurosphere medium. Mouse primary brain neurospheres
were isolated as previously described (Bleau et al., 2009) and plated at
different concentrations (10, 5, and 1 cell/ml). Human primary glioblastoma
neurospheres were freshly isolated from patients suffering from GBMs and
undergoing surgery at MSKCC. Primary GBM samples were dissociated,
plated at different concentrations (1, 0.5, 0.2, and 0.1 cells/ml) and grown in
neurosphere medium. Neurosphere medium consisted of NSC Basal Medium,
NSC proliferation supplements, 10 ng/ml EGF, 20 ng/ml basic-FGF, and 1 mg/
ml Heparin (Stem Cell Technologies, Vancouver, Canada). Neurosphere
number was established by counting the number of neurospheres per well
or, for U87MG and T98G, by taking pictures of the central part of the plates
and then counting the neurosphere number per ﬁeld of view.
SUPPLEMENTAL INFORMATION
Supplemental Information for this article includes four ﬁgures, Supplemental
Experimental Procedures, and one table and can be found with this article online
at http://dx.doi.org/10.1016/j.stem.2012.09.012.
ACKNOWLEDGMENTS
We thank J. Finney and Q. Zhang for technical assistance. We extend thanks to
Dr. L. Abrey as well as to all members of Dr. E. Holland’s and Dr. S. Nimer’s
laboratories for their helpful comments and suggestions. We sincerely thank
Prof. F.H. Gage for providing the Sox2-promoter-driven luciferase construct,
Dr. E.P. Papapetrou for providing the human pLM-YS SOX2 vector, and G.
Sashida for providing the pBabe-Puro-lentivirus MEF vector and the
pLKO.1-Puro-lentivirus vector containing mouse Mef-speciﬁc shRNA. This
work was supported by a grant from National Brain Tumor Foundation,
RO1-DK52208 grant (S.D.N.), and U54CAI43798, UO1CA141502 and
RO1CA100688 grants (E.C.H). E.B. has been supported by Brain Tumor
Center of Memorial Sloan-Kettering Cancer Center. M.C.O. has been supported
by the Rottendorf Foundation, Ennigerloh, Germany, and the Gerhard
Domagk Scholarship Program, Greifswald, Germany. This work is dedicated
to the memory of Dr. Mark A. Zatzkis, a dear friend of S.D.N., who died of
a GBM in 2007.
Received: October 6, 2011
Revised: July 2, 2012
Accepted: September 13, 2012
Published: December 7, 2012
REFERENCES
Barrett, L.E., Granot, Z., Coker, C., Iavarone, A., Hambardzumyan, D., Holland,
E.C., Nam, H.S., and Benezra, R. (2012). Self-renewal does not predict tumor
growth potential in mouse models of high-grade glioma. Cancer Cell 21,
11–24.
Bleau, A.M., Hambardzumyan, D., Ozawa, T., Fomchenko, E.I., Huse, J.T.,
Brennan, C.W., and Holland, E.C. (2009). PTEN/PI3K/Akt pathway regulates
the side population phenotype and ABCG2 activity in glioma tumor stem-like
cells. Cell Stem Cell 4, 226–235.
Cancer Genome Atlas Research Network. (2008). Comprehensive genomic
characterization deﬁnes human glioblastoma genes and core pathways.
Nature 455, 1061–1068.
Cerami, E., Gao, J., Dogrusoz, U., Gross, B.E., Sumer, S.O., Aksoy, B.A.,
Jacobsen, A., Byrne, C.J., Heuer, M.L., Larsson, E., et al. (2012). The cBio
cancer genomics portal: an open platform for exploring multidimensional
cancer genomics data. Cancer Discov. 2, 401–404.
Charles, N.A., Holland, E.C., Gilbertson, R., Glass, R., and Kettenmann, H.
(2011). The brain tumor microenvironment. Glia 59, 1169–1180.
Dai, C., Celestino, J.C., Okada, Y., Louis, D.N., Fuller, G.N., and Holland, E.C.
(2001). PDGF autocrine stimulation dedifferentiates cultured astrocytes and
induces oligodendrogliomas and oligoastrocytomas from neural progenitors
and astrocytes in vivo. Genes Dev. 15, 1913–1925.
Dean, M., Fojo, T., and Bates, S. (2005). Tumour stem cells and drug resistance.
Nat. Rev. Cancer 5, 275–284.
Denysenko, T., Gennero, L., Roos, M.A., Melcarne, A., Juenemann, C.,
Faccani, G., Morra, I., Cavallo, G., Reguzzi, S., Pescarmona, G., and
Ponzetto, A. (2010). Glioblastoma cancer stem cells: heterogeneity, microenvironment
and related therapeutic strategies. Cell Biochem. Funct. 28,
343–351.
Fukushima, T., Miyazaki, Y., Tsushima, H., Tsutsumi, C., Taguchi, J., Yoshida,
S., Kuriyama, K., Scadden, D., Nimer, S., and Tomonaga, M. (2003). The level
of MEF but not ELF-1 correlates with FAB subtype of acute myeloid leukemia
and is low in good prognosis cases. Leuk. Res. 27, 387–392.
Galli, R., Binda, E., Orfanelli, U., Cipelletti, B., Gritti, A., De Vitis, S., Fiocco, R.,
Foroni, C., Dimeco, F., and Vescovi, A. (2004). Isolation and characterization of
tumorigenic, stem-like neural precursors from human glioblastoma. Cancer
Res. 64, 7011–7021.
Holmberg, J., He, X., Peredo, I., Orrego, A., Hesselager, G., Ericsson, C.,
Hovatta, O., Oba-Shinjo, S.M., Marie, S.K., Niste´ r, M., and Muhr, J. (2011).
Activation of neural and pluripotent stem cell signatures correlates with
increased malignancy in human glioma. PLoS ONE 6, e18454.
Ikushima, H., Todo, T., Ino, Y., Takahashi, M., Miyazawa, K., and Miyazono, K.
(2009). Autocrine TGF-beta signaling maintains tumorigenicity of glioma-initiating
cells through Sry-related HMG-box factors. Cell Stem Cell 5, 504–514.
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc. 843
Kuwabara, T., Hsieh, J., Nakashima, K., Taira, K., and Gage, F.H. (2004). A
small modulatory dsRNA speciﬁes the fate of adult neural stem cells. Cell
116, 779–793.
Liu, G., Yuan, X., Zeng, Z., Tunici, P., Ng, H., Abdulkadir, I.R., Lu, L., Irvin, D.,
Black, K.L., and Yu, J.S. (2006). Analysis of gene expression and chemoresistance
of CD133+ cancer stem cells in glioblastoma. Mol. Cancer 5, 67.
Mikkers, H., Allen, J., Knipscheer, P., Romeijn, L., Hart, A., Vink, E., and Berns,
A. (2002). High-throughput retroviral tagging to identify components of speciﬁc
signaling pathways in cancer. Nat. Genet. 32, 153–159.
Nam, H.S., and Benezra, R. (2009). High levels of Id1 expression deﬁne B1 type
adult neural stem cells. Cell Stem Cell 5, 515–526.
Ozawa, T., Brennan, C.W., Wang, L., Squatrito, M., Sasayama, T., Nakada, M.,
Huse, J.T., Pedraza, A., Utsuki, S., Yasui, Y., et al. (2010). PDGFRA gene rearrangements
are frequent genetic events in PDGFRA-ampliﬁed glioblastomas.
Genes Dev. 24, 2205–2218.
Pastrana, E., Silva-Vargas, V., and Doetsch, F. (2011). Eyes wide open: a critical
review of sphere-formation as an assay for stem cells. Cell Stem Cell 8,
486–498.
Patel, M., and Yang, S. (2010). Advances in reprogramming somatic cells to
induced pluripotent stem cells. Stem Cell Rev. 6, 367–380.
Pevny, L.H., and Nicolis, S.K. (2010). Sox2 roles in neural stem cells. Int. J.
Biochem. Cell Biol. 42, 421–424.
Pulvirenti, T., Van Der Heijden, M., Droms, L.A., Huse, J.T., Tabar, V., and Hall,
A. (2011). Dishevelled 2 signaling promotes self-renewal and tumorigenicity in
human gliomas. Cancer Res. 71, 7280–7290.
Sashida, G., Liu, Y., Elf, S., Miyata, Y., Ohyashiki, K., Izumi, M., Menendez, S.,
and Nimer, S.D. (2009). ELF4/MEF activates MDM2 expression and blocks
oncogene-induced p16 activation to promote transformation. Mol. Cell. Biol.
29, 3687–3699.
Sashida, G., Bazzoli, E., Menendez, S., Liu, Y., and Nimer, S.D. (2010). The
oncogenic role of the ETS transcription factors MEF and ERG. Cell Cycle 9,
3457–3459.
Sun, L., Hui, A.M., Su, Q., Vortmeyer, A., Kotliarov, Y., Pastorino, S., Passaniti,
A., Menon, J., Walling, J., Bailey, R., et al. (2006). Neuronal and glioma-derived
stem cell factor induces angiogenesis within the brain. Cancer Cell 9, 287–300.
Uht, R.M., Amos, S., Martin, P.M., Riggan, A.E., and Hussaini, I.M. (2007). The
protein kinase C-eta isoform induces proliferation in glioblastoma cell lines
through an ERK/Elk-1 pathway. Oncogene 26, 2885–2893.
Venere, M., Fine, H.A., Dirks, P.B., and Rich, J.N. (2011). Cancer stem cells in
gliomas: identifying and understanding the apex cell in cancer’s hierarchy. Glia
59, 1148–1154.
Verhaak, R.G., Hoadley, K.A., Purdom, E., Wang, V., Qi, Y., Wilkerson, M.D.,
Miller, C.R., Ding, L., Golub, T., Mesirov, J.P., et al.; Cancer Genome Atlas
Research Network. (2010). Integrated genomic analysis identiﬁes clinically
relevant subtypes of glioblastoma characterized by abnormalities in
PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17, 98–110.
Vescovi, A.L., Galli, R., and Reynolds, B.A. (2006). Brain tumour stem cells.
Nat. Rev. Cancer 6, 425–436.
Vierbuchen, T., Ostermeier, A., Pang, Z.P., Kokubu, Y., Su¨ dhof, T.C., and
Wernig, M. (2010). Direct conversion of ﬁbroblasts to functional neurons by
deﬁned factors. Nature 463, 1035–1041.
Wen, P.Y., and Kesari, S. (2008). Malignant gliomas in adults. N. Engl. J. Med.
359, 492–507.
Wiebe, M.S., Wilder, P.J., Kelly, D., and Rizzino, A. (2000). Isolation, characterization,
and differential expression of the murine Sox-2 promoter. Gene 246,
383–393.
Cell Stem Cell
MEF Role in Pathogenesis of Gliomas
844 Cell Stem Cell 11, 836–844, December 7, 2012 ª2012 Elsevier Inc.